From 9d97170d41e465b598065ebfed2acffda3153377 Mon Sep 17 00:00:00 2001 From: "Documenter.jl" Date: Thu, 29 Feb 2024 14:20:58 +0000 Subject: [PATCH] build based on 3f2e907 --- previews/PR1624/.documenter-siteinfo.json | 2 +- previews/PR1624/authors/index.html | 2 +- previews/PR1624/callbacks/index.html | 4 +- previews/PR1624/code_of_conduct/index.html | 2 +- previews/PR1624/contributing/index.html | 4 +- previews/PR1624/conventions/index.html | 2 +- previews/PR1624/development/index.html | 4 +- previews/PR1624/github-git/index.html | 4 +- previews/PR1624/index.html | 4 +- previews/PR1624/license/index.html | 2 +- .../PR1624/meshes/dgmulti_mesh/index.html | 4 +- previews/PR1624/meshes/p4est_mesh/index.html | 224 +- .../PR1624/meshes/structured_mesh/index.html | 2 +- previews/PR1624/meshes/tree_mesh/index.html | 2 +- .../meshes/unstructured_quad_mesh/index.html | 4 +- .../PR1624/multi-physics_coupling/index.html | 2 + previews/PR1624/overview/index.html | 2 +- previews/PR1624/parallelization/index.html | 4 +- previews/PR1624/performance/index.html | 6 +- previews/PR1624/reference-trixi/index.html | 402 +- .../PR1624/reference-trixi2vtk/index.html | 4 +- .../PR1624/reference-trixibase/index.html | 10 + previews/PR1624/restart/index.html | 4 +- previews/PR1624/search_index.js | 2 +- previews/PR1624/styleguide/index.html | 2 +- previews/PR1624/testing/index.html | 4 +- previews/PR1624/time_integration/index.html | 2 +- previews/PR1624/troubleshooting/index.html | 4 +- .../DGMulti_1/{fbd159b8.svg => 0064e13f.svg} | 6334 ++++---- .../DGMulti_1/{fadf51f7.svg => 10abe6bf.svg} | 12350 ++++++++-------- .../DGMulti_1/{61e49aa9.svg => 208d42b7.svg} | 2756 ++-- .../DGMulti_1/{9bd05ce4.svg => 5d9b08f6.svg} | 2998 ++-- .../DGMulti_1/{f8f84043.svg => 95b64f7c.svg} | 8254 +++++------ .../DGMulti_1/{165f977b.svg => fba4076c.svg} | 3650 ++--- .../PR1624/tutorials/DGMulti_1/index.html | 104 +- .../PR1624/tutorials/DGMulti_2/index.html | 6 +- .../{464dc32e.svg => 081aafce.svg} | 230 +- .../{7294c625.svg => b2d0b36c.svg} | 230 +- .../tutorials/DGSEM_FluxDiff/index.html | 38 +- .../{1bd6887d.svg => 52aac2a9.svg} | 878 +- .../adaptive_mesh_refinement/index.html | 10 +- .../{ce6bbcfb.svg => b74794b6.svg} | 62 +- .../adding_new_parabolic_terms/index.html | 10 +- .../{e507ae71.svg => 81b075c6.svg} | 64 +- .../{8e74c7d4.svg => 9022340d.svg} | 56 +- .../{ffb38f98.svg => bff52588.svg} | 56 +- .../{31378d7b.svg => e3183096.svg} | 56 +- .../{a03f54bf.svg => ef0ee0e1.svg} | 58 +- .../adding_new_scalar_equations/index.html | 18 +- .../{76254cf8.svg => ad596dd0.svg} | 122 +- .../index.html | 10 +- .../f4d8e049.svg | 406 + .../index.html | 49 + .../{f6530f04.svg => 3937d78a.svg} | 68 +- .../{86f51692.svg => 3b4928d2.svg} | 72 +- .../{9ae14439.svg => 5b969adf.svg} | 72 +- .../{5b2f690f.svg => 609468ff.svg} | 72 +- .../{e8340c1c.svg => 6edf9480.svg} | 64 +- .../custom_semidiscretization/index.html | 160 +- .../{40aa74a7.svg => 03e57ed3.svg} | 64 +- .../{aaa20a0c.svg => 1b3a890e.svg} | 586 +- .../{0f9c8ab7.svg => 2f3e8a35.svg} | 3136 ++-- .../{2b2487ee.svg => 4287a5af.svg} | 2126 +-- .../{08ad012c.svg => 6a3d8c62.svg} | 1588 +- .../{cc21e91c.svg => 8d91dd5e.svg} | 1594 +- .../{781540f1.svg => ac36cec2.svg} | 64 +- .../{23db3480.svg => de5f17b3.svg} | 4172 +++--- .../differentiable_programming/index.html | 24 +- .../first_steps/changing_trixi/index.html | 12 + .../create_first_setup/2419cef9.svg | 1142 ++ .../create_first_setup/69e2f11d.svg | 896 ++ .../first_steps/create_first_setup/index.html | 307 + .../first_steps/getting_started/04672d8e.svg | 4101 +++++ .../first_steps/getting_started/6e85c410.svg | 705 + .../first_steps/getting_started/index.html | 227 + .../tutorials/hohqmesh_tutorial/index.html | 10 +- .../PR1624/tutorials/introduction/index.html | 2 +- .../{763e5217.svg => 18af4fab.svg} | 72 +- .../non_periodic_boundaries/index.html | 10 +- .../tutorials/notebooks/DGMulti_1.ipynb | 2 +- .../tutorials/notebooks/DGMulti_2.ipynb | 2 +- .../tutorials/notebooks/DGSEM_FluxDiff.ipynb | 2 +- .../notebooks/adaptive_mesh_refinement.ipynb | 2 +- .../adding_new_parabolic_terms.ipynb | 2 +- .../adding_new_scalar_equations.ipynb | 2 +- .../adding_nonconservative_equation.ipynb | 2 +- .../behind_the_scenes_simulation_setup.ipynb | 628 + .../notebooks/custom_semidiscretization.ipynb | 2 +- .../differentiable_programming.ipynb | 2 +- .../first_steps/changing_trixi.ipynb | 156 + .../first_steps/create_first_setup.ipynb | 597 + .../first_steps/getting_started.ipynb | 435 + .../notebooks/hohqmesh_tutorial.ipynb | 2 +- .../notebooks/non_periodic_boundaries.ipynb | 6 +- .../tutorials/notebooks/p4est_from_gmsh.ipynb | 559 + .../tutorials/notebooks/parabolic_terms.ipynb | 2 +- .../scalar_linear_advection_1d.ipynb | 2 +- .../tutorials/notebooks/shock_capturing.ipynb | 2 +- .../notebooks/structured_mesh_mapping.ipynb | 2 +- .../tutorials/notebooks/time_stepping.ipynb | 2 +- .../tutorials/notebooks/upwind_fdsbp.ipynb | 2 +- .../PR1624/tutorials/out/restart_000050.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/restart_000100.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/restart_000150.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/restart_000180.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000.pvd | 26 +- .../PR1624/tutorials/out/solution_000000.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000010.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000010.vtu | Bin 18962 -> 18969 bytes .../PR1624/tutorials/out/solution_000020.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000020.vtu | Bin 19007 -> 18994 bytes .../PR1624/tutorials/out/solution_000030.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000030.vtu | Bin 19040 -> 19043 bytes .../PR1624/tutorials/out/solution_000040.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000040.vtu | Bin 19074 -> 19079 bytes .../PR1624/tutorials/out/solution_000050.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000050.vtu | Bin 19114 -> 19107 bytes .../PR1624/tutorials/out/solution_000060.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000060.vtu | Bin 19135 -> 19136 bytes .../PR1624/tutorials/out/solution_000065.h5 | Bin 205696 -> 205696 bytes .../PR1624/tutorials/out/solution_000070.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000070.vtu | Bin 19162 -> 19146 bytes .../PR1624/tutorials/out/solution_000080.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000080.vtu | Bin 19161 -> 19157 bytes .../PR1624/tutorials/out/solution_000090.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000090.vtu | Bin 19180 -> 19175 bytes .../PR1624/tutorials/out/solution_000100.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000100.vtu | Bin 19199 -> 19199 bytes .../PR1624/tutorials/out/solution_000110.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000110.vtu | Bin 19193 -> 19209 bytes .../PR1624/tutorials/out/solution_000120.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000120.vtu | Bin 19152 -> 19146 bytes .../PR1624/tutorials/out/solution_000130.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000130.vtu | Bin 19166 -> 19174 bytes .../PR1624/tutorials/out/solution_000140.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000140.vtu | Bin 19172 -> 19174 bytes .../PR1624/tutorials/out/solution_000150.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000150.vtu | Bin 19164 -> 19169 bytes .../PR1624/tutorials/out/solution_000160.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000160.vtu | Bin 19183 -> 19168 bytes .../PR1624/tutorials/out/solution_000170.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000170.vtu | Bin 19194 -> 19193 bytes .../PR1624/tutorials/out/solution_000180.h5 | Bin 11976 -> 11976 bytes .../PR1624/tutorials/out/solution_000180.vtu | Bin 19171 -> 19173 bytes .../tutorials/out/solution_000_celldata.pvd | 26 +- .../tutorials/p4est_from_gmsh/index.html | 366 + .../{bed141da.svg => cda863ac.svg} | 62 +- .../tutorials/parabolic_terms/index.html | 10 +- .../{7bbbc1a1.svg => 1b78a1fd.svg} | 80 +- .../{00c0ec1b.svg => 51a3929a.svg} | 64 +- .../{4c20d287.svg => 5dbf4fa6.svg} | 64 +- .../{0cdc6371.svg => bdf59697.svg} | 68 +- .../{6c8c5e59.svg => e8522b82.svg} | 68 +- .../{b9ccf219.svg => fe5b2ffe.svg} | 64 +- .../scalar_linear_advection_1d/index.html | 20 +- .../{d2d35149.svg => 8985776c.svg} | 230 +- .../tutorials/shock_capturing/index.html | 10 +- .../structured_mesh_mapping/300da432.svg | 9477 ------------ .../{fb53b05b.svg => 5839e9cb.svg} | 246 +- .../structured_mesh_mapping/60cf2592.svg | 9749 ++++++++++++ .../{e3dbbc35.svg => d7d4dc96.svg} | 2114 +-- .../structured_mesh_mapping/index.html | 48 +- .../PR1624/tutorials/time_stepping/index.html | 6 +- .../PR1624/tutorials/upwind_fdsbp/index.html | 6 +- previews/PR1624/visualization/index.html | 4 +- 165 files changed, 48554 insertions(+), 37532 deletions(-) create mode 100644 previews/PR1624/multi-physics_coupling/index.html create mode 100644 previews/PR1624/reference-trixibase/index.html rename previews/PR1624/tutorials/DGMulti_1/{fbd159b8.svg => 0064e13f.svg} (58%) rename previews/PR1624/tutorials/DGMulti_1/{fadf51f7.svg => 10abe6bf.svg} (90%) rename previews/PR1624/tutorials/DGMulti_1/{61e49aa9.svg => 208d42b7.svg} (77%) rename previews/PR1624/tutorials/DGMulti_1/{9bd05ce4.svg => 5d9b08f6.svg} (76%) rename previews/PR1624/tutorials/DGMulti_1/{f8f84043.svg => 95b64f7c.svg} (90%) rename previews/PR1624/tutorials/DGMulti_1/{165f977b.svg => fba4076c.svg} (89%) rename previews/PR1624/tutorials/DGSEM_FluxDiff/{464dc32e.svg => 081aafce.svg} (94%) rename previews/PR1624/tutorials/DGSEM_FluxDiff/{7294c625.svg => b2d0b36c.svg} (95%) rename previews/PR1624/tutorials/adaptive_mesh_refinement/{1bd6887d.svg => 52aac2a9.svg} (85%) rename previews/PR1624/tutorials/adding_new_parabolic_terms/{ce6bbcfb.svg => b74794b6.svg} (93%) rename previews/PR1624/tutorials/adding_new_scalar_equations/{e507ae71.svg => 81b075c6.svg} (91%) rename previews/PR1624/tutorials/adding_new_scalar_equations/{8e74c7d4.svg => 9022340d.svg} (84%) rename previews/PR1624/tutorials/adding_new_scalar_equations/{ffb38f98.svg => bff52588.svg} (92%) rename previews/PR1624/tutorials/adding_new_scalar_equations/{31378d7b.svg => e3183096.svg} (92%) rename previews/PR1624/tutorials/adding_new_scalar_equations/{a03f54bf.svg => ef0ee0e1.svg} (85%) rename previews/PR1624/tutorials/adding_nonconservative_equation/{76254cf8.svg => ad596dd0.svg} (84%) create mode 100644 previews/PR1624/tutorials/behind_the_scenes_simulation_setup/f4d8e049.svg create mode 100644 previews/PR1624/tutorials/behind_the_scenes_simulation_setup/index.html rename previews/PR1624/tutorials/custom_semidiscretization/{f6530f04.svg => 3937d78a.svg} (86%) rename previews/PR1624/tutorials/custom_semidiscretization/{86f51692.svg => 3b4928d2.svg} (86%) rename previews/PR1624/tutorials/custom_semidiscretization/{9ae14439.svg => 5b969adf.svg} (86%) rename previews/PR1624/tutorials/custom_semidiscretization/{5b2f690f.svg => 609468ff.svg} (86%) rename previews/PR1624/tutorials/custom_semidiscretization/{e8340c1c.svg => 6edf9480.svg} (85%) rename previews/PR1624/tutorials/differentiable_programming/{40aa74a7.svg => 03e57ed3.svg} (84%) rename previews/PR1624/tutorials/differentiable_programming/{aaa20a0c.svg => 1b3a890e.svg} (69%) rename previews/PR1624/tutorials/differentiable_programming/{0f9c8ab7.svg => 2f3e8a35.svg} (67%) rename previews/PR1624/tutorials/differentiable_programming/{2b2487ee.svg => 4287a5af.svg} (68%) rename previews/PR1624/tutorials/differentiable_programming/{08ad012c.svg => 6a3d8c62.svg} (65%) rename previews/PR1624/tutorials/differentiable_programming/{cc21e91c.svg => 8d91dd5e.svg} (68%) rename previews/PR1624/tutorials/differentiable_programming/{781540f1.svg => ac36cec2.svg} (85%) rename previews/PR1624/tutorials/differentiable_programming/{23db3480.svg => de5f17b3.svg} (66%) create mode 100644 previews/PR1624/tutorials/first_steps/changing_trixi/index.html create mode 100644 previews/PR1624/tutorials/first_steps/create_first_setup/2419cef9.svg create mode 100644 previews/PR1624/tutorials/first_steps/create_first_setup/69e2f11d.svg create mode 100644 previews/PR1624/tutorials/first_steps/create_first_setup/index.html create mode 100644 previews/PR1624/tutorials/first_steps/getting_started/04672d8e.svg create mode 100644 previews/PR1624/tutorials/first_steps/getting_started/6e85c410.svg create mode 100644 previews/PR1624/tutorials/first_steps/getting_started/index.html rename previews/PR1624/tutorials/non_periodic_boundaries/{763e5217.svg => 18af4fab.svg} (81%) create mode 100644 previews/PR1624/tutorials/notebooks/behind_the_scenes_simulation_setup.ipynb create mode 100644 previews/PR1624/tutorials/notebooks/first_steps/changing_trixi.ipynb create mode 100644 previews/PR1624/tutorials/notebooks/first_steps/create_first_setup.ipynb create mode 100644 previews/PR1624/tutorials/notebooks/first_steps/getting_started.ipynb create mode 100644 previews/PR1624/tutorials/notebooks/p4est_from_gmsh.ipynb create mode 100644 previews/PR1624/tutorials/p4est_from_gmsh/index.html rename previews/PR1624/tutorials/parabolic_terms/{bed141da.svg => cda863ac.svg} (93%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{7bbbc1a1.svg => 1b78a1fd.svg} (79%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{00c0ec1b.svg => 51a3929a.svg} (83%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{4c20d287.svg => 5dbf4fa6.svg} (82%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{0cdc6371.svg => bdf59697.svg} (85%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{6c8c5e59.svg => e8522b82.svg} (85%) rename previews/PR1624/tutorials/scalar_linear_advection_1d/{b9ccf219.svg => fe5b2ffe.svg} (83%) rename previews/PR1624/tutorials/shock_capturing/{d2d35149.svg => 8985776c.svg} (96%) delete mode 100644 previews/PR1624/tutorials/structured_mesh_mapping/300da432.svg rename previews/PR1624/tutorials/structured_mesh_mapping/{fb53b05b.svg => 5839e9cb.svg} (96%) create mode 100644 previews/PR1624/tutorials/structured_mesh_mapping/60cf2592.svg rename previews/PR1624/tutorials/structured_mesh_mapping/{e3dbbc35.svg => d7d4dc96.svg} (86%) diff --git a/previews/PR1624/.documenter-siteinfo.json b/previews/PR1624/.documenter-siteinfo.json index e2d832f4dfd..e4bc38eca7a 100644 --- a/previews/PR1624/.documenter-siteinfo.json +++ b/previews/PR1624/.documenter-siteinfo.json @@ -1 +1 @@ -{"documenter":{"julia_version":"1.9.4","generation_timestamp":"2024-01-11T16:12:41","documenter_version":"1.2.1"}} \ No newline at end of file +{"documenter":{"julia_version":"1.9.4","generation_timestamp":"2024-02-29T14:20:09","documenter_version":"1.2.1"}} \ No newline at end of file diff --git a/previews/PR1624/authors/index.html b/previews/PR1624/authors/index.html index d7a387bc430..5fc080da4ab 100644 --- a/previews/PR1624/authors/index.html +++ b/previews/PR1624/authors/index.html @@ -1,2 +1,2 @@ -Authors · Trixi.jl
GitHub

Authors

Trixi.jl's development is coordinated by a group of principal developers, who are also its main contributors and who can be contacted in case of questions about Trixi.jl. In addition, there are contributors who have provided substantial additions or modifications. Together, these two groups form "The Trixi.jl Authors" as mentioned under License.

Principal Developers

Contributors

The following people contributed major additions or modifications to Trixi.jl and are listed in alphabetical order:

  • Maximilian D. Bertrand
  • Benjamin Bolm
  • Simon Candelaresi
  • Jesse Chan
  • Lars Christmann
  • Christof Czernik
  • Daniel Doehring
  • Patrick Ersing
  • Erik Faulhaber
  • Gregor Gassner
  • Lucas Gemein
  • Sven Goldberg
  • Joshua Lampert
  • Julia Odenthal
  • Sigrun Ortleb
  • Hendrik Ranocha
  • Andrés M. Rueda-Ramírez
  • Felipe Santillan
  • Michael Schlottke-Lakemper
  • Toskan Theine
  • Andrew Winters
+Authors · Trixi.jl
GitHub

Authors

Trixi.jl's development is coordinated by a group of principal developers, who are also its main contributors and who can be contacted in case of questions about Trixi.jl. In addition, there are contributors who have provided substantial additions or modifications. Together, these two groups form "The Trixi.jl Authors" as mentioned under License.

Principal Developers

Contributors

The following people contributed major additions or modifications to Trixi.jl and are listed in alphabetical order:

  • Maximilian D. Bertrand
  • Benjamin Bolm
  • Simon Candelaresi
  • Jesse Chan
  • Lars Christmann
  • Christof Czernik
  • Daniel Doehring
  • Patrick Ersing
  • Erik Faulhaber
  • Gregor Gassner
  • Lucas Gemein
  • Sven Goldberg
  • Joshua Lampert
  • Julia Odenthal
  • Sigrun Ortleb
  • Hendrik Ranocha
  • Andrés M. Rueda-Ramírez
  • Felipe Santillan
  • Michael Schlottke-Lakemper
  • Toskan Theine
  • Andrew Winters
diff --git a/previews/PR1624/callbacks/index.html b/previews/PR1624/callbacks/index.html index 091f14ba5dd..08f53700f9a 100644 --- a/previews/PR1624/callbacks/index.html +++ b/previews/PR1624/callbacks/index.html @@ -1,5 +1,5 @@ -Callbacks · Trixi.jl
GitHub

Callbacks

Many of the advanced features of Trixi.jl, such as adaptive mesh refinement, are implemented as callbacks. A callback is an algorithmic entity that gets passed to the ODE solver and is called at specific points during execution to perform certain tasks. Callbacks in Trixi.jl are either called after each time step (step callbacks) or after each stage of the ODE solver (stage callbacks).

callbacks_illustration

The advantage of callbacks over hard-coding all features is that it allows to extend Trixi.jl without modifying the internal source code. Trixi.jl provides callbacks for time step control, adaptive mesh refinement, I/O, and more.

Step callbacks

CFL-based time step control

Time step control can be performed with a StepsizeCallback. An example making use of this can be found at examples/tree_2d_dgsem/elixir_advection_basic.jl

Adaptive mesh refinement

Trixi.jl uses a hierarchical Cartesian mesh which can be locally refined in a solution-adaptive way. This can be used to speed up simulations with minimal loss in overall accuracy. Adaptive mesh refinement (AMR) can be used by passing an AMRCallback to the ODE solver. The AMRCallback requires a controller such as ControllerThreeLevel or ControllerThreeLevelCombined to tell the AMR algorithm which cells to refine/coarsen.

An example elixir using AMR can be found at examples/tree_2d_dgsem/elixir_advection_amr.jl.

Analyzing the numerical solution

The AnalysisCallback can be used to analyze the numerical solution, e.g. calculate errors or user-specified integrals, and print the results to the screen. The results can also be saved in a file. An example can be found at examples/tree_2d_dgsem/elixir_euler_vortex.jl. Note that the errors (e.g. L2 error or Linf error) are computed with respect to the initial condition. The percentage of the simulation time refers to the ratio of the current time and the final time, i.e. it does not consider the maximal number of iterations. So the simulation could finish before 100% are reached. Note that, e.g., due to AMR or smaller time step sizes, the simulation can actually take longer than the percentage indicates. In Performance metrics of the AnalysisCallback you can find a detailed description of the different performance metrics the AnalysisCallback computes.

I/O

Solution and restart files

To save the solution in regular intervals you can use a SaveSolutionCallback. It is also possible to create restart files using the SaveRestartCallback. An example making use of these can be found at examples/tree_2d_dgsem/elixir_advection_extended.jl. An example showing how to restart a simulation from a restart file can be found at examples/tree_2d_dgsem/elixir_advection_restart.jl.

Time series

Sometimes it is useful to record the evaluations of state variables over time at a given set of points. This can be achieved by the TimeSeriesCallback, which is used, e.g., in examples/tree_2d_dgsem/elixir_acoustics_gaussian_source.jl. The TimeSeriesCallback constructor expects a semidiscretization and a list of points at which the solution should be recorded in regular time step intervals. After the last time step, the entire record is stored in an HDF5 file.

For the points, two different input formats are supported: You can either provide them as a list of tuples, which is handy if you specify them by hand on the REPL. Alternatively, you can provide them as a two-dimensional array, where the first dimension is the point number and the second dimension is the coordinate dimension. This is especially useful when reading them from a file.

For example, to record the primitive variables at the points (0.0, 0.0) and (-1.0, 0.5) every five timesteps and storing the collected data in the file tseries.h5, you can create the TimeSeriesCallback as

time_series = TimeSeriesCallback(semi, [(0.0, 0.0), (-1.0, 0.5)];
+Callbacks · Trixi.jl
GitHub
Callbacks
Many of the advanced features of Trixi.jl, such as adaptive mesh refinement, are implemented as callbacks. A callback is an algorithmic entity that gets passed to the ODE solver and is called at specific points during execution to perform certain tasks. Callbacks in Trixi.jl are either called after each time step (step callbacks) or after each stage of the ODE solver (stage callbacks).
callbacks_illustration
The advantage of callbacks over hard-coding all features is that it allows to extend Trixi.jl without modifying the internal source code. Trixi.jl provides callbacks for time step control, adaptive mesh refinement, I/O, and more.
Step callbacks
CFL-based time step control
Time step control can be performed with a StepsizeCallback. An example making use of this can be found at examples/tree_2d_dgsem/elixir_advection_basic.jl
Adaptive mesh refinement
Trixi.jl uses a hierarchical Cartesian mesh which can be locally refined in a solution-adaptive way. This can be used to speed up simulations with minimal loss in overall accuracy. Adaptive mesh refinement (AMR) can be used by passing an AMRCallback to the ODE solver. The AMRCallback requires a controller such as ControllerThreeLevel or ControllerThreeLevelCombined to tell the AMR algorithm which cells to refine/coarsen.
An example elixir using AMR can be found at examples/tree_2d_dgsem/elixir_advection_amr.jl.
Analyzing the numerical solution
The AnalysisCallback can be used to analyze the numerical solution, e.g. calculate errors or user-specified integrals, and print the results to the screen. The results can also be saved in a file. An example can be found at examples/tree_2d_dgsem/elixir_euler_vortex.jl. Note that the errors (e.g. L2 error or Linf error) are computed with respect to the initial condition. The percentage of the simulation time refers to the ratio of the current time and the final time, i.e. it does not consider the maximal number of iterations. So the simulation could finish before 100% are reached. Note that, e.g., due to AMR or smaller time step sizes, the simulation can actually take longer than the percentage indicates. In Performance metrics of the AnalysisCallback you can find a detailed description of the different performance metrics the AnalysisCallback computes.
I/O
Solution and restart files
To save the solution in regular intervals you can use a SaveSolutionCallback. It is also possible to create restart files using the SaveRestartCallback. An example making use of these can be found at examples/tree_2d_dgsem/elixir_advection_extended.jl. An example showing how to restart a simulation from a restart file can be found at examples/tree_2d_dgsem/elixir_advection_restart.jl.
Time series
Sometimes it is useful to record the evaluations of state variables over time at a given set of points. This can be achieved by the TimeSeriesCallback, which is used, e.g., in examples/tree_2d_dgsem/elixir_acoustics_gaussian_source.jl. The TimeSeriesCallback constructor expects a semidiscretization and a list of points at which the solution should be recorded in regular time step intervals. After the last time step, the entire record is stored in an HDF5 file.
For the points, two different input formats are supported: You can either provide them as a list of tuples, which is handy if you specify them by hand on the REPL. Alternatively, you can provide them as a two-dimensional array, where the first dimension is the point number and the second dimension is the coordinate dimension. This is especially useful when reading them from a file.
For example, to record the primitive variables at the points (0.0, 0.0) and (-1.0, 0.5) every five timesteps and storing the collected data in the file tseries.h5, you can create the TimeSeriesCallback as
time_series = TimeSeriesCallback(semi, [(0.0, 0.0), (-1.0, 0.5)];
                                  interval=5,
                                  solution_variables=cons2prim,
                                  filename="tseries.h5")
For a full list of possible arguments, please check the documentation for the TimeSeriesCallback. As an alternative to specifying the point coordinates directly in the elixir or on the REPL, you can read them from a file. For instance, with a text file points.dat with content
 0.0 0.0
@@ -12,4 +12,4 @@
 
 julia> pd2 = PlotData1D(time_series, 2)
 
-julia> plot(pd1["p_prime"]); plot!(pd2["p_prime"], xguide="t")
will yield the following plot:
image
Miscellaneous
Equation-specific callbacks
Some callbacks provided by Trixi.jl implement specific features for certain equations:
Usage of step callbacks
Step callbacks are passed to the solve method from the ODE solver via the keyword argument callback. If you want to use a single callback cb, pass it as callback=cb. When using two or more callbacks, you need to turn them into a CallbackSet first by calling callbacks = CallbackSet(cb1, cb2) and passing it as callback=callbacks.
Note
There are some restrictions regarding the order of callbacks in a CallbackSet.
The callbacks are called after each time step but some callbacks actually belong to the next time step. Therefore, the callbacks should be ordered in the following way:
  • Callbacks that belong to the current time step:
    • SummaryCallback controls, among other things, timers and should thus be first
    • SteadyStateCallback may mark a time step as the last one
    • AnalysisCallback may do some checks that mark a time step as the last one
    • AliveCallback should be nearby AnalysisCallback
    • SaveSolutionCallback/SaveRestartCallback should save the current solution before it is degraded by AMR
    • VisualizationCallback should be called before the mesh is adapted
  • Callbacks that belong to the next time step:
    • AMRCallback
    • StepsizeCallback must be called after AMRCallback to accommodate potential changes to the mesh
    • GlmSpeedCallback must be called after StepsizeCallback because the step size affects the value of c_h
    • LBMCollisionCallback is already part of the calculations of the next time step and should therefore be called after StepsizeCallback
Stage callbacks
PositivityPreservingLimiterZhangShu is a positivity-preserving limiter, used to enforce physical constraints. An example elixir using this feature can be found at examples/tree_2d_dgsem/elixir_euler_positivity.jl.
Implementing new callbacks
Since Trixi.jl is compatible with OrdinaryDiffEq.jl, both packages share the same callback interface. A detailed description of it can be found in the OrdinaryDiffEq.jl documentation. Step callbacks are just called callbacks. Stage callbacks are called stage_limiter!.
An example elixir showing how to implement a new simple stage callback and a new simple step callback can be found at examples/tree_2d_dgsem/elixir_advection_callbacks.jl.

+julia> plot(pd1["p_prime"]); plot!(pd2["p_prime"], xguide="t")

will yield the following plot:

image

Miscellaneous

Equation-specific callbacks

Some callbacks provided by Trixi.jl implement specific features for certain equations:

Usage of step callbacks

Step callbacks are passed to the solve method from the ODE solver via the keyword argument callback. If you want to use a single callback cb, pass it as callback=cb. When using two or more callbacks, you need to turn them into a CallbackSet first by calling callbacks = CallbackSet(cb1, cb2) and passing it as callback=callbacks.

Note

There are some restrictions regarding the order of callbacks in a CallbackSet.

The callbacks are called after each time step but some callbacks actually belong to the next time step. Therefore, the callbacks should be ordered in the following way:

  • Callbacks that belong to the current time step:
    • SummaryCallback controls, among other things, timers and should thus be first
    • SteadyStateCallback may mark a time step as the last one
    • AnalysisCallback may do some checks that mark a time step as the last one
    • AliveCallback should be nearby AnalysisCallback
    • SaveSolutionCallback/SaveRestartCallback should save the current solution before it is degraded by AMR
    • VisualizationCallback should be called before the mesh is adapted
  • Callbacks that belong to the next time step:
    • AMRCallback
    • StepsizeCallback must be called after AMRCallback to accommodate potential changes to the mesh
    • GlmSpeedCallback must be called after StepsizeCallback because the step size affects the value of c_h
    • LBMCollisionCallback is already part of the calculations of the next time step and should therefore be called after StepsizeCallback

Stage callbacks

PositivityPreservingLimiterZhangShu is a positivity-preserving limiter, used to enforce physical constraints. An example elixir using this feature can be found at examples/tree_2d_dgsem/elixir_euler_positivity.jl.

Implementing new callbacks

Since Trixi.jl is compatible with OrdinaryDiffEq.jl, both packages share the same callback interface. A detailed description of it can be found in the OrdinaryDiffEq.jl documentation. Step callbacks are just called callbacks. Stage callbacks are called stage_limiter!.

An example elixir showing how to implement a new simple stage callback and a new simple step callback can be found at examples/tree_2d_dgsem/elixir_advection_callbacks.jl.

diff --git a/previews/PR1624/code_of_conduct/index.html b/previews/PR1624/code_of_conduct/index.html index f5150f59e36..52aa14084d3 100644 --- a/previews/PR1624/code_of_conduct/index.html +++ b/previews/PR1624/code_of_conduct/index.html @@ -1,2 +1,2 @@ -Code of Conduct · Trixi.jl
GitHub

Code of Conduct

Contributor Covenant Code of Conduct

Our Pledge

We as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation.

We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community.

Our Standards

Examples of behavior that contributes to a positive environment for our community include:

  • Demonstrating empathy and kindness toward other people
  • Being respectful of differing opinions, viewpoints, and experiences
  • Giving and gracefully accepting constructive feedback
  • Accepting responsibility and apologizing to those affected by our mistakes, and learning from the experience
  • Focusing on what is best not just for us as individuals, but for the overall community

Examples of unacceptable behavior include:

  • The use of sexualized language or imagery, and sexual attention or advances of any kind
  • Trolling, insulting or derogatory comments, and personal or political attacks
  • Public or private harassment
  • Publishing others' private information, such as a physical or email address, without their explicit permission
  • Other conduct which could reasonably be considered inappropriate in a professional setting

Enforcement Responsibilities

Community leaders are responsible for clarifying and enforcing our standards of acceptable behavior and will take appropriate and fair corrective action in response to any behavior that they deem inappropriate, threatening, offensive, or harmful.

Community leaders have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, and will communicate reasons for moderation decisions when appropriate.

Scope

This Code of Conduct applies within all community spaces, and also applies when an individual is officially representing the community in public spaces. Examples of representing our community include using an official e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event.

Enforcement

Instances of abusive, harassing, or otherwise unacceptable behavior may be reported to Michael Schlottke-Lakemper, Hendrik Ranocha, or any other of the principal developers responsible for enforcement listed in Authors. All complaints will be reviewed and investigated promptly and fairly.

All community leaders are obligated to respect the privacy and security of the reporter of any incident.

Enforcement Guidelines

Community leaders will follow these Community Impact Guidelines in determining the consequences for any action they deem in violation of this Code of Conduct:

1. Correction

Community Impact: Use of inappropriate language or other behavior deemed unprofessional or unwelcome in the community.

Consequence: A private, written warning from community leaders, providing clarity around the nature of the violation and an explanation of why the behavior was inappropriate. A public apology may be requested.

2. Warning

Community Impact: A violation through a single incident or series of actions.

Consequence: A warning with consequences for continued behavior. No interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, for a specified period of time. This includes avoiding interactions in community spaces as well as external channels like social media. Violating these terms may lead to a temporary or permanent ban.

3. Temporary Ban

Community Impact: A serious violation of community standards, including sustained inappropriate behavior.

Consequence: A temporary ban from any sort of interaction or public communication with the community for a specified period of time. No public or private interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, is allowed during this period. Violating these terms may lead to a permanent ban.

4. Permanent Ban

Community Impact: Demonstrating a pattern of violation of community standards, including sustained inappropriate behavior, harassment of an individual, or aggression toward or disparagement of classes of individuals.

Consequence: A permanent ban from any sort of public interaction within the community.

Attribution

This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 2.0, available at https://www.contributor-covenant.org/version/2/0/codeofconduct.html.

Community Impact Guidelines were inspired by Mozilla's code of conduct enforcement ladder.

[homepage]: https://www.contributor-covenant.org

For answers to common questions about this code of conduct, see the FAQ at https://www.contributor-covenant.org/faq. Translations are available at https://www.contributor-covenant.org/translations.

+Code of Conduct · Trixi.jl
GitHub

Code of Conduct

Contributor Covenant Code of Conduct

Our Pledge

We as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation.

We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community.

Our Standards

Examples of behavior that contributes to a positive environment for our community include:

  • Demonstrating empathy and kindness toward other people
  • Being respectful of differing opinions, viewpoints, and experiences
  • Giving and gracefully accepting constructive feedback
  • Accepting responsibility and apologizing to those affected by our mistakes, and learning from the experience
  • Focusing on what is best not just for us as individuals, but for the overall community

Examples of unacceptable behavior include:

  • The use of sexualized language or imagery, and sexual attention or advances of any kind
  • Trolling, insulting or derogatory comments, and personal or political attacks
  • Public or private harassment
  • Publishing others' private information, such as a physical or email address, without their explicit permission
  • Other conduct which could reasonably be considered inappropriate in a professional setting

Enforcement Responsibilities

Community leaders are responsible for clarifying and enforcing our standards of acceptable behavior and will take appropriate and fair corrective action in response to any behavior that they deem inappropriate, threatening, offensive, or harmful.

Community leaders have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, and will communicate reasons for moderation decisions when appropriate.

Scope

This Code of Conduct applies within all community spaces, and also applies when an individual is officially representing the community in public spaces. Examples of representing our community include using an official e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event.

Enforcement

Instances of abusive, harassing, or otherwise unacceptable behavior may be reported to Michael Schlottke-Lakemper, Hendrik Ranocha, or any other of the principal developers responsible for enforcement listed in Authors. All complaints will be reviewed and investigated promptly and fairly.

All community leaders are obligated to respect the privacy and security of the reporter of any incident.

Enforcement Guidelines

Community leaders will follow these Community Impact Guidelines in determining the consequences for any action they deem in violation of this Code of Conduct:

1. Correction

Community Impact: Use of inappropriate language or other behavior deemed unprofessional or unwelcome in the community.

Consequence: A private, written warning from community leaders, providing clarity around the nature of the violation and an explanation of why the behavior was inappropriate. A public apology may be requested.

2. Warning

Community Impact: A violation through a single incident or series of actions.

Consequence: A warning with consequences for continued behavior. No interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, for a specified period of time. This includes avoiding interactions in community spaces as well as external channels like social media. Violating these terms may lead to a temporary or permanent ban.

3. Temporary Ban

Community Impact: A serious violation of community standards, including sustained inappropriate behavior.

Consequence: A temporary ban from any sort of interaction or public communication with the community for a specified period of time. No public or private interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, is allowed during this period. Violating these terms may lead to a permanent ban.

4. Permanent Ban

Community Impact: Demonstrating a pattern of violation of community standards, including sustained inappropriate behavior, harassment of an individual, or aggression toward or disparagement of classes of individuals.

Consequence: A permanent ban from any sort of public interaction within the community.

Attribution

This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 2.0, available at https://www.contributor-covenant.org/version/2/0/codeofconduct.html.

Community Impact Guidelines were inspired by Mozilla's code of conduct enforcement ladder.

[homepage]: https://www.contributor-covenant.org

For answers to common questions about this code of conduct, see the FAQ at https://www.contributor-covenant.org/faq. Translations are available at https://www.contributor-covenant.org/translations.

diff --git a/previews/PR1624/contributing/index.html b/previews/PR1624/contributing/index.html index ccfff267848..c446e761375 100644 --- a/previews/PR1624/contributing/index.html +++ b/previews/PR1624/contributing/index.html @@ -1,5 +1,5 @@ -Contributing · Trixi.jl
GitHub

Contributing

Trixi.jl is an open-source project and we are very happy to accept contributions from the community. Please feel free to open issues or submit patches (preferably as merge requests) any time. For planned larger contributions, it is often beneficial to get in contact with one of the principal developers first (see Authors).

Trixi.jl and its contributions are licensed under the MIT license (see License). As a contributor, you certify that all your contributions are in conformance with the Developer Certificate of Origin (Version 1.1), which is reproduced below.

Developer Certificate of Origin (Version 1.1)

The following text was taken from https://developercertificate.org:

Developer Certificate of Origin
+Contributing · Trixi.jl
GitHub
Contributing
Trixi.jl is an open-source project and we are very happy to accept contributions from the community. Please feel free to open issues or submit patches (preferably as merge requests) any time. For planned larger contributions, it is often beneficial to get in contact with one of the principal developers first (see Authors).
Trixi.jl and its contributions are licensed under the MIT license (see License). As a contributor, you certify that all your contributions are in conformance with the Developer Certificate of Origin (Version 1.1), which is reproduced below.
Developer Certificate of Origin (Version 1.1)
The following text was taken from https://developercertificate.org:
Developer Certificate of Origin
 Version 1.1
 
 Copyright (C) 2004, 2006 The Linux Foundation and its contributors.
@@ -35,4 +35,4 @@
     are public and that a record of the contribution (including all
     personal information I submit with it, including my sign-off) is
     maintained indefinitely and may be redistributed consistent with
-    this project or the open source license(s) involved.

+    this project or the open source license(s) involved.
diff --git a/previews/PR1624/conventions/index.html b/previews/PR1624/conventions/index.html index 2dd46883b75..759514671ee 100644 --- a/previews/PR1624/conventions/index.html +++ b/previews/PR1624/conventions/index.html @@ -1,2 +1,2 @@ -Conventions · Trixi.jl
GitHub

Conventions

Spatial dimensions and directions

We use the following numbering schemes on Cartesian or curved structured meshes.

  • The orientations are numbered as 1 => x, 2 => y, 3 => z. For example, numerical fluxes such as flux_central(u_ll, u_rr, orientation, equations::AbstractEquations) use the orientation in this way.
  • The directions are numbered as 1 => -x, 2 => +x, 3 => -y, 4 => +y, 5 => -z, 6 => +z. For example, the boundary_conditions are ordered in this way when a Tuple of boundary conditions per direction is passed to the constructor of a SemidiscretizationHyperbolic.
  • For structured and unstructured curved meshes the concept of direction is generalized via the variable normal_direction. This variable points in the normal direction at a given, curved surface. For the computation of boundary fluxes the normal_direction is normalized to be a normal_vector used, for example, in FluxRotated.

Cells vs. elements vs. nodes

To uniquely distinguish between different components of the discretization, we use the following naming conventions:

  • The computational domain is discretized by a mesh, which is made up of individual cells. In general, neither the mesh nor the cells should be aware of any solver-specific knowledge, i.e., they should not store anything that goes beyond the geometrical information and the connectivity.
  • The numerical solvers do not directly store their information inside the mesh, but use own data structures. Specifically, for each cell on which a solver wants to operate, the solver creates an element to store solver-specific data.
  • For discretization schemes such as the discontinuous Galerkin or the finite element method, inside each element multiple nodes may be defined, which hold nodal information. The nodes are again a solver-specific component, just like the elements.
  • We often identify elements, nodes, etc. with their (local or global) integer index. Convenience iterators such as eachelement, eachnode use these indices.

Keywords in elixirs

Trixi.jl is distributed with several examples in the form of elixirs, small Julia scripts containing everything to set up and run a simulation. Working interactively from the Julia REPL with these scripts can be quite convenient while for exploratory research and development of Trixi.jl. For example, you can use the convenience function trixi_include to include an elixir with some modified arguments. To enable this, it is helpful to use a consistent naming scheme in elixirs, since trixi_include can only perform simple replacements. Some standard variables names are

  • polydeg for the polynomial degree of a solver
  • surface_flux for the numerical flux at surfaces
  • volume_flux for the numerical flux used in flux differencing volume terms

Moreover, convergence_test requires that the spatial resolution is set via the keywords

Variable names

  • Use descriptive names (using snake_case for variables/functions and CamelCase for types)
  • Use a suffix _ as in name_ for local variables that would otherwise hide existing symbols.
  • Use a prefix _ as in _name to indicate internal methods/data that are "fragile" in the sense that there's no guarantee that they might get changed without notice. These are also not documented with a docstring (but maybe with comments using #).

Array types and wrapping

To allow adaptive mesh refinement efficiently when using time integrators from OrdinaryDiffEq, Trixi.jl allows to represent numerical solutions in two different ways. Some discussion can be found online and in form of comments describing Trixi.wrap_array and Trixi.wrap_array_native in the source code of Trixi.jl. The flexibility introduced by this possible wrapping enables additional performance optimizations. However, it comes at the cost of some additional abstractions (and needs to be used with caution, as described in the source code of Trixi.jl). Thus, we use the following conventions to distinguish between arrays visible to the time integrator and wrapped arrays mainly used internally.

  • Arrays visible to the time integrator have a suffix _ode, e.g., du_ode, u_ode.
  • Wrapped arrays do not have a suffix, e.g., du, u.

Methods either accept arrays visible to the time integrator or wrapped arrays based on the following rules.

  • When some solution is passed together with a semidiscretization semi, the solution must be a u_ode that needs to be wrapped via wrap_array(u_ode, semi) (or wrap_array_native(u_ode, semi)) for further processing.
  • When some solution is passed together with the mesh, equations, solver, cache, ..., it is already wrapped via wrap_array (or wrap_array_native).
  • Exceptions of this rule are possible, e.g. for AMR, but must be documented in the code.
  • wrap_array should be used as default option. wrap_array_native should only be used when necessary, e.g., to avoid additional overhead when interfacing with external C libraries such as HDF5, MPI, or visualization.
+Conventions · Trixi.jl
GitHub

Conventions

Spatial dimensions and directions

We use the following numbering schemes on Cartesian or curved structured meshes.

  • The orientations are numbered as 1 => x, 2 => y, 3 => z. For example, numerical fluxes such as flux_central(u_ll, u_rr, orientation, equations::AbstractEquations) use the orientation in this way.
  • The directions are numbered as 1 => -x, 2 => +x, 3 => -y, 4 => +y, 5 => -z, 6 => +z. For example, the boundary_conditions are ordered in this way when a Tuple of boundary conditions per direction is passed to the constructor of a SemidiscretizationHyperbolic.
  • For structured and unstructured curved meshes the concept of direction is generalized via the variable normal_direction. This variable points in the normal direction at a given, curved surface. For the computation of boundary fluxes the normal_direction is normalized to be a normal_vector used, for example, in FluxRotated.

Cells vs. elements vs. nodes

To uniquely distinguish between different components of the discretization, we use the following naming conventions:

  • The computational domain is discretized by a mesh, which is made up of individual cells. In general, neither the mesh nor the cells should be aware of any solver-specific knowledge, i.e., they should not store anything that goes beyond the geometrical information and the connectivity.
  • The numerical solvers do not directly store their information inside the mesh, but use own data structures. Specifically, for each cell on which a solver wants to operate, the solver creates an element to store solver-specific data.
  • For discretization schemes such as the discontinuous Galerkin or the finite element method, inside each element multiple nodes may be defined, which hold nodal information. The nodes are again a solver-specific component, just like the elements.
  • We often identify elements, nodes, etc. with their (local or global) integer index. Convenience iterators such as eachelement, eachnode use these indices.

Keywords in elixirs

Trixi.jl is distributed with several examples in the form of elixirs, small Julia scripts containing everything to set up and run a simulation. Working interactively from the Julia REPL with these scripts can be quite convenient while for exploratory research and development of Trixi.jl. For example, you can use the convenience function trixi_include to include an elixir with some modified arguments. To enable this, it is helpful to use a consistent naming scheme in elixirs, since trixi_include can only perform simple replacements. Some standard variables names are

  • polydeg for the polynomial degree of a solver
  • surface_flux for the numerical flux at surfaces
  • volume_flux for the numerical flux used in flux differencing volume terms

Moreover, convergence_test requires that the spatial resolution is set via the keywords

Variable names

  • Use descriptive names (using snake_case for variables/functions and CamelCase for types)
  • Use a suffix _ as in name_ for local variables that would otherwise hide existing symbols.
  • Use a prefix _ as in _name to indicate internal methods/data that are "fragile" in the sense that there's no guarantee that they might get changed without notice. These are also not documented with a docstring (but maybe with comments using #).

Array types and wrapping

To allow adaptive mesh refinement efficiently when using time integrators from OrdinaryDiffEq, Trixi.jl allows to represent numerical solutions in two different ways. Some discussion can be found online and in form of comments describing Trixi.wrap_array and Trixi.wrap_array_native in the source code of Trixi.jl. The flexibility introduced by this possible wrapping enables additional performance optimizations. However, it comes at the cost of some additional abstractions (and needs to be used with caution, as described in the source code of Trixi.jl). Thus, we use the following conventions to distinguish between arrays visible to the time integrator and wrapped arrays mainly used internally.

  • Arrays visible to the time integrator have a suffix _ode, e.g., du_ode, u_ode.
  • Wrapped arrays do not have a suffix, e.g., du, u.

Methods either accept arrays visible to the time integrator or wrapped arrays based on the following rules.

  • When some solution is passed together with a semidiscretization semi, the solution must be a u_ode that needs to be wrapped via wrap_array(u_ode, semi) (or wrap_array_native(u_ode, semi)) for further processing.
  • When some solution is passed together with the mesh, equations, solver, cache, ..., it is already wrapped via wrap_array (or wrap_array_native).
  • Exceptions of this rule are possible, e.g. for AMR, but must be documented in the code.
  • wrap_array should be used as default option. wrap_array_native should only be used when necessary, e.g., to avoid additional overhead when interfacing with external C libraries such as HDF5, MPI, or visualization.
diff --git a/previews/PR1624/development/index.html b/previews/PR1624/development/index.html index 66ffaa007e4..73a3c7f8836 100644 --- a/previews/PR1624/development/index.html +++ b/previews/PR1624/development/index.html @@ -1,5 +1,5 @@ -Development · Trixi.jl
GitHub

Development

Interactive use of Julia

When a Julia program is executed, Julia first loads and parses all code. Then, the just-in-time compiler has to compile all functions at their first use, which incurs an overhead each time a program is run. For proper packages and commands executed in the REPL (= "return-eval-print loop", which is what the Julia community calls the interactive command-line prompt that opens when executing julia without any files as arguments), however, the previously compiled functions are cached. Therefore, Trixi.jl should generally always be used interactively from the REPL without closing Julia during development, as it allows much faster turnaround times.

If you naively run Trixi.jl from the REPL, you will not be able to change your Trixi.jl source files and then run the changed code without restarting the REPL, which destroys any potential benefits from caching. However, restarting Julia can be avoided by using the Revise.jl package, which tracks changed files and re-loads them automatically. Therefore, it is highly recommended to first install Revise with the following command in Julia: To enter the package REPL mode, press ] in the standard Julia REPL mode. Then, execute

(@v1.9) pkg> add Revise

Now you are able to run Trixi.jl from the REPL, change Trixi.jl code between runs, and enjoy the advantages of the compilation cache! Before you start using Revise regularly, please be aware of some of the Pitfalls when using Revise.

Another recommended package for working from the REPL is OhMyREPL.jl. It can be installed by running

(@v1.9) pkg> add OhMyREPL

and adds syntax highlighting, bracket highlighting, and other helpful improvements for using Julia interactively. To automatically use OhMyREPL when starting the REPL, follow the instructions given in the official documentation.

Running Trixi.jl interactively in the global environment

If you've installed Trixi.jl and Revise in your default environment, begin by executing:

julia

This will start the Julia REPL. Then, run

julia> using Revise; using Trixi

You can run a simulation by executing

julia> trixi_include(default_example())

Together, all of these commands can take some time, roughly half a minute on a modern workstation. Most of the time is spent on compilation of Julia code etc. If you execute the last command again in the same REPL, it will finish within a few milliseconds (maybe ~45 on a modern workstation). This demonstrates the second reason for using the REPL: the compilation cache. That is, those parts of the code that do not change between two Trixi.jl runs do not need to be recompiled and thus execute much faster after the first run.

Manually starting Trixi.jl in the local environment

If you followed the installation instructions for developers, execute Julia with the project directory set to the run directory of the program/tool you want to use. For example, to run Trixi.jl this way, you need to start the REPL with

julia --project=path/to/Trixi.jl/run

and execute

julia> using Revise; using Trixi

to load Revise and Trixi.jl. You can then proceed with the usual commands and run Trixi.jl as in the example above. The --project flag is required such that Julia can properly load Trixi.jl and all dependencies if Trixi.jl is not installed in the global environment. The same procedure also applies should you opt to install the postprocessing tool Trixi2Vtk manually such that you can modify their implementations.

Pitfalls when using Revise

While Revise is a great help for developing Julia code, there are a few situations to watch out for when using Revise. The following list of potential issues is based on personal experiences of the Trixi.jl developers and probably incomplete. Further information on limitations and possible issues with Revise can be found in the official documentation.

If in doubt, restart the REPL

Oftentimes, it is possible to recover from issues with Revise by fixing the offending code. Sometimes, however, this is not possible or you might have troubles finding out what exactly caused the problems. Therefore, in these cases, or if in doubt, restart the REPL to get a fresh start.

Syntax errors are easy to miss

Revise does not stop on syntax errors, e.g., when you accidentally write a[i) instead of a[i]. In this case, Revise reports an error but continues to use the old version of your files! This is especially dangerous for syntax errors, as they are detected while Revise reloads changed code, which happens in the beginning of a new execution. Thus, the syntax error message quickly disappears from the terminal once Trixi.jl starts writing output to the screen and you might not even have noticed that an error occurred at all.

Therefore, when you are deep in a coding/debugging session and wonder why your code modifications do not seem to have any effect, scroll up in your terminal to check if you missed earlier syntax errors, or - if in doubt - restart your REPL.

Files are not tracked after changing branches

Sometimes, Revise stops tracking files when changing the Git branch. That is, modifications to Trixi.jl's source files will not be reloaded by Revise and thus have no effect of a currently running REPL session. This issue is particularly annoying for a developer, since it does not come with any warning! Therefore, it is good practice to always restart the REPL after changing branches.

Changes to type definitions are not allowed

Revise cannot handle changes to type definitions, e.g., when modifying the fields in a struct. In this case, Revise reports an error and refuses to run your code unless you undo the modifications. Once you undo the changes, Revise will usually continue to work as expected again. However, if you want to keep your type modifications, you need to restart the REPL.

Using the Julia REPL effectively

The Julia manual is an excellent resource to learn Julia. Here, we list some helpful commands than can increase your productivity in the Julia REPL.

  • Use the REPL help mode entered by typing ?.

    julia> using Trixi
+Development · Trixi.jl
    GitHub
    Development
    Interactive use of Julia
    When a Julia program is executed, Julia first loads and parses all code. Then, the just-in-time compiler has to compile all functions at their first use, which incurs an overhead each time a program is run. For proper packages and commands executed in the REPL (= "return-eval-print loop", which is what the Julia community calls the interactive command-line prompt that opens when executing julia without any files as arguments), however, the previously compiled functions are cached. Therefore, Trixi.jl should generally always be used interactively from the REPL without closing Julia during development, as it allows much faster turnaround times.
    If you naively run Trixi.jl from the REPL, you will not be able to change your Trixi.jl source files and then run the changed code without restarting the REPL, which destroys any potential benefits from caching. However, restarting Julia can be avoided by using the Revise.jl package, which tracks changed files and re-loads them automatically. Therefore, it is highly recommended to first install Revise with the following command in Julia: To enter the package REPL mode, press ] in the standard Julia REPL mode. Then, execute
    (@v1.9) pkg> add Revise
    Now you are able to run Trixi.jl from the REPL, change Trixi.jl code between runs, and enjoy the advantages of the compilation cache! Before you start using Revise regularly, please be aware of some of the Pitfalls when using Revise.
    Another recommended package for working from the REPL is OhMyREPL.jl. It can be installed by running
    (@v1.9) pkg> add OhMyREPL
    and adds syntax highlighting, bracket highlighting, and other helpful improvements for using Julia interactively. To automatically use OhMyREPL when starting the REPL, follow the instructions given in the official documentation.
    Running Trixi.jl interactively in the global environment
    If you've installed Trixi.jl and Revise in your default environment, begin by executing:
    julia
    This will start the Julia REPL. Then, run
    julia> using Revise; using Trixi
    You can run a simulation by executing
    julia> trixi_include(default_example())
    Together, all of these commands can take some time, roughly half a minute on a modern workstation. Most of the time is spent on compilation of Julia code etc. If you execute the last command again in the same REPL, it will finish within a few milliseconds (maybe ~45 on a modern workstation).  This demonstrates the second reason for using the REPL: the compilation cache.  That is, those parts of the code that do not change between two Trixi.jl runs do not need to be recompiled and thus execute much faster after the first run.
    Manually starting Trixi.jl in the local environment
    If you followed the installation instructions for developers, execute Julia with the project directory set to the run directory of the program/tool you want to use. For example, to run Trixi.jl this way, you need to start the REPL with
    julia --project=path/to/Trixi.jl/run
    and execute
    julia> using Revise; using Trixi
    to load Revise and Trixi.jl. You can then proceed with the usual commands and run Trixi.jl as in the example above. The --project flag is required such that Julia can properly load Trixi.jl and all dependencies if Trixi.jl is not installed in the global environment. The same procedure also applies should you opt to install the postprocessing tool Trixi2Vtk manually such that you can modify their implementations.
    Pitfalls when using Revise
    While Revise is a great help for developing Julia code, there are a few situations to watch out for when using Revise. The following list of potential issues is based on personal experiences of the Trixi.jl developers and probably incomplete.  Further information on limitations and possible issues with Revise can be found in the official documentation.
    If in doubt, restart the REPL
    Oftentimes, it is possible to recover from issues with Revise by fixing the offending code. Sometimes, however, this is not possible or you might have troubles finding out what exactly caused the problems. Therefore, in these cases, or if in doubt, restart the REPL to get a fresh start.
    Syntax errors are easy to miss
    Revise does not stop on syntax errors, e.g., when you accidentally write a[i) instead of a[i].  In this case, Revise reports an error but continues to use the old version of your files! This is especially dangerous for syntax errors, as they are detected while Revise reloads changed code, which happens in the beginning of a new execution. Thus, the syntax error message quickly disappears from the terminal once Trixi.jl starts writing output to the screen and you might not even have noticed that an error occurred at all.
    Therefore, when you are deep in a coding/debugging session and wonder why your code modifications do not seem to have any effect, scroll up in your terminal to check if you missed earlier syntax errors, or - if in doubt - restart your REPL.
    Files are not tracked after changing branches
    Sometimes, Revise stops tracking files when changing the Git branch. That is, modifications to Trixi.jl's source files will not be reloaded by Revise and thus have no effect of a currently running REPL session. This issue is particularly annoying for a developer, since it does not come with any warning!  Therefore, it is good practice to always restart the REPL after changing branches.
    Changes to type definitions are not allowed
    Revise cannot handle changes to type definitions, e.g., when modifying the fields in a struct. In this case, Revise reports an error and refuses to run your code unless you undo the modifications. Once you undo the changes, Revise will usually continue to work as expected again. However, if you want to keep your type modifications, you need to restart the REPL.
    Using the Julia REPL effectively
    The Julia manual is an excellent resource to learn Julia. Here, we list some helpful commands than can increase your productivity in the Julia REPL.
    • Use the REPL help mode entered by typing ?.
      julia> using Trixi
 
 help?> trixi_include
 search: trixi_include
@@ -39,4 +39,4 @@
 Trixi.IdealGlmMhdEquations2D
 Trixi.IdealGlmMhdMulticomponentEquations1D
 [...]
    Text editors
    When writing code, the choice of text editor can have a significant impact on productivity and developer satisfaction. While using the default text editor of the operating system has its own benefits (specifically the lack of an explicit installation procure), usually it makes sense to switch to a more programming-friendly tool. In the following, a few of the many options are listed and discussed:
    VS Code
    Visual Studio Code is a modern open source editor with good support for Julia. While Juno had some better support in the past, the developers of Juno and the Julia VS Code plugin are joining forces and concentrating on VS Code since support of Atom has been suspended. Basically, all comments on Juno below also apply to VS Code.
    Juno
    If you are new to programming or do not have a preference for a text editor yet, Juno is a good choice for developing Julia code. It is based on Atom, a sophisticated and widely used editor for software developers, and is enhanced with several Julia-specific features. Furthermore and especially helpful for novice programmers, it has a MATLAB-like appearance with easy and interactive access to the current variables, the help system, and a debugger.
    Vim or Emacs
    Vim and Emacs are both very popular editors that work great with Julia. One of their advantages is that they are text editors without a GUI and as such are available for almost any operating system. They also are preinstalled on virtually all Unix-like systems.  However, Vim and Emacs come with their own, steep learning curve if they have never been used before. Therefore, if in doubt, it is probably easier to get started with a classic GUI-based text editor (like Juno). If you decide to use Vim or Emacs, make sure that you install the corresponding Vim plugin julia-vim or Emacs major mode julia-emacs.
    Debugging
    Julia offers several options for debugging. A classical debugger is available with the Debugger.jl package or in the Julia extension for VS Code. However, it can be quite slow and, at the time of writing (January 2023), currently does not work properly with Trixi.jl. The Infiltrator.jl package on the other hand does not offer all features of a full debugger, but is a fast and simple tool that allows users to set breakpoints to open a local REPL session and access the call stack and variables.
    Infiltrator
    The Infiltrator package provides fast, interactive breakpoints using the @infiltrate command, which drops the user into a local REPL session. From there, it is possible to access local variables, see the call stack, and execute statements.
    The package can be installed in the Julia REPL by executing
    (@v1.9) pkg> add Infiltrator
    To load the package in the Julia REPL execute
    julia> using Infiltrator
    Breakpoints can be set by adding a line with the @infiltrate macro at the respective position in the code. Use Revise if you want to set and delete breakpoints in your package without having to restart Julia.
    Use `@autoinfiltrate` when debugging Trixi.jl
    When running Julia inside a package environment, e.g., inside the source code of Trixi.jl itself, the @infiltrate macro only works if Infiltrator has been added to the package dependencies. To avoid this, you can use the (non-exported) @autoinfiltrate macro in Trixi.jl, which only requires Infiltrator.jl to be available in the current environment stack and will auto-load it for you.
    Triggering the breakpoint starts a REPL session where it is possible to interact with the current local scope. Possible commands are:
    • @locals: Print the local variables.
    • @exfiltrate: Save the local variables to a global storage, which can be accessed with the safehouse variable outside the Infiltrator session.
    • @trace: Print the current stack trace.
    • Execute other arbitrary statements
    • ?: Print a help list with all options
    To finish a debugging session, either use @continue to continue and eventually stop at the next breakpoint or @exit to skip further breakpoints. After the code has finished, local variables saved with @exfiltrate can be accessed in the REPL using the safehouse variable.
    Limitations of using Infiltrator.jl are that local variables cannot be changed, and that it is not possible to step into further calls or access other function scopes.
    Releasing a new version of Trixi.jl, Trixi2Vtk
    • Check whether everything is okay, tests pass etc.
    • Set the new version number in Project.toml according to the Julian version of semver. Commit and push.
    • Comment @JuliaRegistrator register on the commit setting the version number.
    • JuliaRegistrator will create a PR with the new version in the General registry. Wait for it to be merged.
    • Increment the version number in Project.toml again with suffix -pre. For example, if you have released version v0.2.0, use v0.2.1-pre as new version number.
    • When a new version of Trixi.jl was released, check whether the [compat] entries in test/Project.toml in Trixi2Vtk should be updated. When a new version of Trixi2Vtk was released, check whether the [compat] entries in docs/Project.toml in Trixi.jl should be updated.
      These entries will also be checked regularly by CompatHelper (once a day). Hence, if everything was released correctly, you should only need to do these checks manually if new minor versions with changes in the docs of Trixi2Vtk were released but no new version of Trixi.jl was released afterwards.
    Preview of the documentation
    You can build the documentation of Trixi.jl locally by running
    julia --project=docs -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd())); Pkg.instantiate()'
-julia --project=docs --color=yes docs/make.jl
    from the Trixi.jl main directory. Then, you can look at the html files generated in docs/build. For PRs triggered from branches inside the Trixi.jl main repository previews of the new documentation are generated at https://trixi-framework.github.io/Trixi.jl/previews/PRXXX, where XXX is the number of the PR. This does not work for PRs from forks for security reasons (since anyone could otherwise push arbitrary stuff to the Trixi.jl website, including malicious code).
    Developing Trixi2Vtk
    Trixi2Vtk has Trixi.jl as dependency and uses Trixi.jl's implementation to, e.g., load mesh files. When developing Trixi2Vtk, one may want to change functions in Trixi.jl to allow them to be reused in Trixi2Vtk. To use a locally modified Trixi.jl clone instead of a Trixi.jl release, one can tell Pkg to use the local source code of Trixi.jl instead of a registered version by running
    (@v1.9) pkg> develop path/to/Trixi.jl
    
+julia --project=docs --color=yes docs/make.jl

    from the Trixi.jl main directory. Then, you can look at the html files generated in docs/build. For PRs triggered from branches inside the Trixi.jl main repository previews of the new documentation are generated at https://trixi-framework.github.io/Trixi.jl/previews/PRXXX, where XXX is the number of the PR. This does not work for PRs from forks for security reasons (since anyone could otherwise push arbitrary stuff to the Trixi.jl website, including malicious code).

    Developing Trixi2Vtk

    Trixi2Vtk has Trixi.jl as dependency and uses Trixi.jl's implementation to, e.g., load mesh files. When developing Trixi2Vtk, one may want to change functions in Trixi.jl to allow them to be reused in Trixi2Vtk. To use a locally modified Trixi.jl clone instead of a Trixi.jl release, one can tell Pkg to use the local source code of Trixi.jl instead of a registered version by running

    (@v1.9) pkg> develop path/to/Trixi.jl
diff --git a/previews/PR1624/github-git/index.html b/previews/PR1624/github-git/index.html index 7e0b614c4c7..2a71d7e47c1 100644 --- a/previews/PR1624/github-git/index.html +++ b/previews/PR1624/github-git/index.html @@ -1,5 +1,5 @@ -GitHub & Git · Trixi.jl
GitHub

GitHub & Git

This page contains information on how to use GitHub and Git when developing Trixi.jl.

Development workflow

For adding modifications to Trixi.jl, we generally follow these steps:

Create an issue (optional)

In many cases it makes sense to start by creating an issue on GitHub. For example, if the implementation approach for a new feature is not yet clear or if there should be a discussion about the desired outcome, it is good practice to first get a consensus on what is the expected result of this modification. A GitHub issue is the place to lead this discussion, as it preserves it in the project and - together with the actual code changes - allows in the future to revisit the reasons for a particular choice of implementation or feature.

Create a branch and immediately create a pull request

All feature development, bug fixes etc. should be developed in a branch and not directly on main. If you do not have write access to the main repository on GitHub, first create a fork of the Trixi.jl repository and clone the fork to your machine. Then, create a branch locally by executing git checkout -b yourbranch, push it to the repository, and create a pull request (PR).

If you have already cloned Trixi.jl from the main repo to your local machine, you can also work in that clone. You just need to add your fork as additional remote repository and push your new branch there.

git remote add myfork git@github.com:YOUR_NAME/Trixi.jl.git
+GitHub & Git · Trixi.jl
GitHub
GitHub & Git
This page contains information on how to use GitHub and Git when developing Trixi.jl.
Development workflow
For adding modifications to Trixi.jl, we generally follow these steps:
Create an issue (optional)
In many cases it makes sense to start by creating an issue on GitHub. For example, if the implementation approach for a new feature is not yet clear or if there should be a discussion about the desired outcome, it is good practice to first get a consensus on what is the expected result of this modification. A GitHub issue is the place to lead this discussion, as it preserves it in the project and - together with the actual code changes - allows in the future to revisit the reasons for a particular choice of implementation or feature.
Create a branch and immediately create a pull request
All feature development, bug fixes etc. should be developed in a branch and not directly on main. If you do not have write access to the main repository on GitHub, first create a fork of the Trixi.jl repository and clone the fork to your machine. Then, create a branch locally by executing git checkout -b yourbranch, push it to the repository, and create a pull request (PR).
If you have already cloned Trixi.jl from the main repo to your local machine, you can also work in that clone. You just need to add your fork as additional remote repository and push your new branch there.
git remote add myfork git@github.com:YOUR_NAME/Trixi.jl.git
 # get latest main from the main repo
 git checkout main
 git pull
@@ -41,4 +41,4 @@
 git rebase
 
 # Clean reflog and force garbage collection
-git reflog expire --expire=now --all && git gc --prune=now --aggressive
IMPORTANT: You need to do a git rebase instead of a git pull when updating the fixed branch.

+git reflog expire --expire=now --all && git gc --prune=now --aggressive

IMPORTANT: You need to do a git rebase instead of a git pull when updating the fixed branch.

diff --git a/previews/PR1624/index.html b/previews/PR1624/index.html index 61411a042d0..79e342b959a 100644 --- a/previews/PR1624/index.html +++ b/previews/PR1624/index.html @@ -1,5 +1,5 @@ -Home · Trixi.jl
GitHub

Trixi.jl

Docs-stable Docs-dev Slack Youtube Build Status Codecov Coveralls Aqua QA License: MIT DOI

Trixi.jl is a numerical simulation framework for conservation laws written in Julia. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures. Its features include:

  • 1D, 2D, and 3D simulations on line/quad/hex/simplex meshes
    • Cartesian and curvilinear meshes
    • Conforming and non-conforming meshes
    • Structured and unstructured meshes
    • Hierarchical quadtree/octree grid with adaptive mesh refinement
    • Forests of quadtrees/octrees with p4est via P4est.jl
  • High-order accuracy in space and time
  • Discontinuous Galerkin methods
  • Compatible with the SciML ecosystem for ordinary differential equations
  • Native support for differentiable programming
  • Periodic and weakly-enforced boundary conditions
  • Multiple governing equations:
    • Compressible Euler equations
    • Compressible Navier-Stokes equations
    • Magnetohydrodynamics (MHD) equations
    • Multi-component compressible Euler and MHD equations
    • Linearized Euler and acoustic perturbation equations
    • Hyperbolic diffusion equations for elliptic problems
    • Lattice-Boltzmann equations (D2Q9 and D3Q27 schemes)
    • Shallow water equations
    • Scalar advection
  • Multi-physics simulations
  • Shared-memory parallelization via multithreading
  • Multi-node parallelization via MPI
  • Visualization and postprocessing of the results
    • In-situ and a posteriori visualization with Plots.jl
    • Interactive visualization with Makie.jl
    • Postprocessing with ParaView/VisIt via Trixi2Vtk

Installation

If you have not yet installed Julia, please follow the instructions for your operating system. Trixi.jl works with Julia v1.8 and newer. We recommend using the latest stable release of Julia.

For users

Trixi.jl and its related tools are registered Julia packages. Hence, you can install Trixi.jl, the visualization tool Trixi2Vtk, OrdinaryDiffEq.jl, and Plots.jl by executing the following commands in the Julia REPL:

julia> using Pkg
+Home · Trixi.jl
GitHub
Trixi.jl
Docs-stable Docs-dev Slack Youtube Build Status Codecov Coveralls Aqua QA License: MIT DOI
Trixi.jl is a numerical simulation framework for conservation laws written in Julia. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures. Its features include:
  • 1D, 2D, and 3D simulations on line/quad/hex/simplex meshes
    • Cartesian and curvilinear meshes
    • Conforming and non-conforming meshes
    • Structured and unstructured meshes
    • Hierarchical quadtree/octree grid with adaptive mesh refinement
    • Forests of quadtrees/octrees with p4est via P4est.jl
  • High-order accuracy in space and time
  • Discontinuous Galerkin methods
  • Compatible with the SciML ecosystem for ordinary differential equations
  • Native support for differentiable programming
  • Periodic and weakly-enforced boundary conditions
  • Multiple governing equations:
    • Compressible Euler equations
    • Compressible Navier-Stokes equations
    • Magnetohydrodynamics (MHD) equations
    • Multi-component compressible Euler and MHD equations
    • Linearized Euler and acoustic perturbation equations
    • Hyperbolic diffusion equations for elliptic problems
    • Lattice-Boltzmann equations (D2Q9 and D3Q27 schemes)
    • Shallow water equations
    • Scalar advection
  • Multi-physics simulations
  • Shared-memory parallelization via multithreading
  • Multi-node parallelization via MPI
  • Visualization and postprocessing of the results
    • In-situ and a posteriori visualization with Plots.jl
    • Interactive visualization with Makie.jl
    • Postprocessing with ParaView/VisIt via Trixi2Vtk
Installation
If you have not yet installed Julia, please follow the instructions for your operating system. Trixi.jl works with Julia v1.8 and newer. We recommend using the latest stable release of Julia.
For users
Trixi.jl and its related tools are registered Julia packages. Hence, you can install Trixi.jl, the visualization tool Trixi2Vtk, OrdinaryDiffEq.jl, and Plots.jl by executing the following commands in the Julia REPL:
julia> using Pkg
 
 julia> Pkg.add(["Trixi", "Trixi2Vtk", "OrdinaryDiffEq", "Plots"])
You can copy and paste all commands to the REPL including the leading julia> prompts - they will automatically be stripped away by Julia. The package OrdinaryDiffEq.jl provides time integration schemes used by Trixi.jl, while Plots.jl can be used to directly visualize Trixi.jl's results from the REPL.
Note on package versions: If some of the examples for how to use Trixi.jl do not work, verify that you are using a recent Trixi.jl release by comparing the installed Trixi.jl version from
julia> using Pkg; Pkg.update("Trixi"); Pkg.status("Trixi")
to the latest release. If the installed version does not match the current release, please check the Troubleshooting section.
The commands above can also be used to update Trixi.jl. A brief list of notable changes to Trixi.jl is available in NEWS.md.
For developers
If you plan on editing Trixi.jl itself, you can download Trixi.jl to a local folder and use the code from the cloned directory:
git clone git@github.com:trixi-framework/Trixi.jl.git
 cd Trixi.jl
@@ -111,4 +111,4 @@
     NumFOCUS -->
   

-
This project has benefited from funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through the following grants:
  • Excellence Strategy EXC 2044-390685587, Mathematics Münster: Dynamics-Geometry-Structure.
  • Research unit FOR 5409 "Structure-Preserving Numerical Methods for Bulk- and Interface Coupling of Heterogeneous Models (SNuBIC)" (project number 463312734).
  • Individual grant no. 528753982.
This project has benefited from funding from the European Research Council through the ERC Starting Grant "An Exascale aware and Un-crashable Space-Time-Adaptive Discontinuous Spectral Element Solver for Non-Linear Conservation Laws" (Extreme), ERC grant agreement no. 714487.
This project has benefited from funding from Vetenskapsrådet (VR, Swedish Research Council), Sweden through the VR Starting Grant "Shallow water flows including sediment transport and morphodynamics", VR grant agreement 2020-03642 VR.
This project has benefited from funding from the United States National Science Foundation (NSF) under awards DMS-1719818 and DMS-1943186.
This project has benefited from funding from the German Federal Ministry of Education and Research (BMBF) through the project grant "Adaptive earth system modeling with significantly reduced computation time for exascale supercomputers (ADAPTEX)" (funding id: 16ME0668K).
This project has benefited from funding by the Daimler und Benz Stiftung (Daimler and Benz Foundation) through grant no. 32-10/22.
Trixi.jl is supported by NumFOCUS as an Affiliated Project.
+

This project has benefited from funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through the following grants:

  • Excellence Strategy EXC 2044-390685587, Mathematics Münster: Dynamics-Geometry-Structure.
  • Research unit FOR 5409 "Structure-Preserving Numerical Methods for Bulk- and Interface Coupling of Heterogeneous Models (SNuBIC)" (project number 463312734).
  • Individual grant no. 528753982.

This project has benefited from funding from the European Research Council through the ERC Starting Grant "An Exascale aware and Un-crashable Space-Time-Adaptive Discontinuous Spectral Element Solver for Non-Linear Conservation Laws" (Extreme), ERC grant agreement no. 714487.

This project has benefited from funding from Vetenskapsrådet (VR, Swedish Research Council), Sweden through the VR Starting Grant "Shallow water flows including sediment transport and morphodynamics", VR grant agreement 2020-03642 VR.

This project has benefited from funding from the United States National Science Foundation (NSF) under awards DMS-1719818 and DMS-1943186.

This project has benefited from funding from the German Federal Ministry of Education and Research (BMBF) through the project grant "Adaptive earth system modeling with significantly reduced computation time for exascale supercomputers (ADAPTEX)" (funding id: 16ME0668K).

This project has benefited from funding by the Daimler und Benz Stiftung (Daimler and Benz Foundation) through grant no. 32-10/22.

Trixi.jl is supported by NumFOCUS as an Affiliated Project.

diff --git a/previews/PR1624/license/index.html b/previews/PR1624/license/index.html index da454aaf2ef..b32b2b951f8 100644 --- a/previews/PR1624/license/index.html +++ b/previews/PR1624/license/index.html @@ -1,2 +1,2 @@ -License · Trixi.jl
GitHub

License

MIT License

Copyright (c) 2020-present The Trixi.jl Authors (see Authors)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+License · Trixi.jl
GitHub

License

MIT License

Copyright (c) 2020-present The Trixi.jl Authors (see Authors)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

diff --git a/previews/PR1624/meshes/dgmulti_mesh/index.html b/previews/PR1624/meshes/dgmulti_mesh/index.html index 3ef82ae37ef..3453e2e8873 100644 --- a/previews/PR1624/meshes/dgmulti_mesh/index.html +++ b/previews/PR1624/meshes/dgmulti_mesh/index.html @@ -1,8 +1,8 @@ -DGMulti mesh · Trixi.jl
GitHub

Unstructured meshes and the DGMulti solver

Trixi.jl includes support for simplicial and tensor product meshes via the DGMulti solver type, which is based on the StartUpDG.jl package. DGMulti solvers also provide support for quadrilateral and hexahedral meshes, though this feature is currently restricted to Cartesian grids. On these line/quad/hex meshes, the DGMulti solver also allows to use all (finite domain) SBP derivative operators provided by SummationByPartsOperators.jl, including several finite difference SBP methods.

We make a few simplifying assumptions about supported meshes:

  • meshes consist of a single type of element
  • meshes are conforming (e.g., each face of an element is shared with at most one other element).
  • the geometric mapping from reference to physical elements is polynomial (currently, only affine mappings are supported).

StartUpDG.jl includes both simple uniform meshes via uniform_mesh, as well as support for triangular meshes constructed using Triangulate.jl, a wrapper around Jonathan Shewchuk's Triangle package.

The DGMulti solver type

Trixi.jl solvers on simplicial meshes use the DGMulti solver type, which allows users to specify element_type and approximation_type in addition to polydeg, surface_flux, surface_integral, and volume_integral.

DGMulti(; polydeg::Integer,
+DGMulti mesh · Trixi.jl
GitHub
Unstructured meshes and the DGMulti solver
Trixi.jl includes support for simplicial and tensor product meshes via the DGMulti solver type, which is based on the StartUpDG.jl package. DGMulti solvers also provide support for quadrilateral and hexahedral meshes, though this feature is currently restricted to Cartesian grids. On these line/quad/hex meshes, the DGMulti solver also allows to use all (finite domain) SBP derivative operators provided by SummationByPartsOperators.jl, including several finite difference SBP methods.
We make a few simplifying assumptions about supported meshes:
  • meshes consist of a single type of element
  • meshes are conforming (e.g., each face of an element is shared with at most one other element).
  • the geometric mapping from reference to physical elements is polynomial (currently, only affine mappings are supported).
StartUpDG.jl includes both simple uniform meshes via uniform_mesh, as well as support for triangular meshes constructed using Triangulate.jl, a wrapper around Jonathan Shewchuk's Triangle package.
The DGMulti solver type
Trixi.jl solvers on simplicial meshes use the DGMulti solver type, which allows users to specify element_type and approximation_type in addition to polydeg, surface_flux, surface_integral, and volume_integral.
DGMulti(; polydeg::Integer,
           element_type::AbstractElemShape,
           approximation_type=Polynomial(),
           surface_flux=flux_central,
           surface_integral=SurfaceIntegralWeakForm(surface_flux),
           volume_integral=VolumeIntegralWeakForm(),
-          RefElemData_kwargs...)
Here, element_type can be Tri(), Quad(), Tet(), or Hex(), and approximation_type can be
  • Polynomial(), which specifies a DG discretization using a polynomial basis using quadrature rules which are exact for degree 2 * polydeg integrands, or
  • SBP(), which specifies a DG discretization using multi-dimensional SBP operators. Types of SBP discretizations available include: SBP{Kubatko{LobattoFaceNodes}}() (the default choice), SBP{Kubatko{LegendreFaceNodes}}(), and SBP{Hicken}(). For polydeg = 1, ..., 4, the SBP{Kubatko{LegendreFaceNodes}}() SBP nodes are identical to the SBP nodes of Chen and Shu. More detailed descriptions of each SBP node set can be found in the StartUpDG.jl docs. Trixi.jl will also specialize certain parts of the solver based on the SBP approximation type.
  • a (periodic or non-periodic) derivative operator from SummationByPartsOperators.jl, usually constructed as D = derivative_operator(...). In this case, you do not need to pass a polydeg. Periodic derivative operators will only work with single-element meshes constructed using DGMultiMesh.
Additional options can also be specified through RefElemData_kwargs:
  • quad_rule_vol = quad_nodes(Tri(), Nq) will substitute in a volume quadrature rule of degree Nq instead of the default (which is a quadrature rule of degree polydeg). Here, a degree Nq rule will be exact for at least degree 2*Nq integrands (such that the mass matrix is integrated exactly). Quadrature rules of which exactly integrate degree Nq integrands may also be specified (for example, quad_rule_vol = StartUpDG.quad_nodes_tri(Nq) on triangles).
  • quad_rule_face = quad_nodes(Line(), Nq)) will use a face quadrature rule of degree Nq rather than the default. This rule is also exact for at least degree 2*Nq integrands.
The GaussSBP() approximation type on Quad() and Hex() meshes
When using VolumeIntegralFluxDifferencing on Quad() and Hex() meshes, one can also specify approximation_type = GaussSBP() to use an entropy stable Gauss collocation scheme. Here, GaussSBP() refers to "generalized" summation-by-parts operators (see for example Ranocha 2018 or Fernandez and Zingg 2015).
Unlike traditional SBP operators, generalized SBP operators are constructed from nodes which do not include boundary nodes (i.e., Gauss quadrature nodes as opposed to Gauss-Lobatto quadrature nodes). This makes the computation of interface fluxes slightly more expensive, but also usually results in a more accurate solution. Roughly speaking, an entropy stable Gauss collocation scheme will yield results similar to a modal entropy stable scheme using a Polynomial() approximation type, but will be more efficient at high orders of approximation.
Trixi.jl elixirs on simplicial and tensor product element meshes
Example elixirs with triangular, quadrilateral, and tetrahedral meshes can be found in the examples/dgmulti_2d/ and examples/dgmulti_3d/ folders. Some key elixirs to look at:
For developers
DGMultiMesh wrapper type
DGMulti meshes in Trixi.jl are represented using a DGMultiMesh{NDIMS, ...} type. This mesh type is assumed to have fields md::MeshData, which contains geometric terms derived from the mapping between the reference and physical elements, and boundary_faces, which contains a Dict of boundary segment names (symbols) and list of faces which lie on that boundary segment.
A DGMultiMesh can be constructed in several ways. For example, DGMultiMesh(dg::DGMulti) will return a Cartesian mesh on $[-1, 1]^d$ with element types specified by dg. DGMulti meshes can also be constructed by specifying a list of vertex coordinates vertex_coordinates_x, vertex_coordinates_y, vertex_coordinates_z and a connectivity matrix EToV where EToV[e,:] gives the vertices which correspond to element e. These quantities are available from most unstructured mesh generators.
Initial support for curved DGMultiMeshes is available for flux differencing solvers using SBP and GaussSBP approximation types on quadrilateral and hexahedral meshes. These can be called by specifying mesh = DGMultiMesh(dg, cells_per_dimension, mapping), where mapping is a function which specifies the warping of the mesh (e.g., mapping(xi, eta) = SVector{2}(xi, eta) is the identity mapping) similar to the mapping argument used by StructuredMesh.
Variable naming conventions
We use the convention that coordinates on the reference element are $r$ in 1D, $r, s$ in 2D, or $r, s, t$ in 3D. Physical coordinates use the standard conventions $x$ (1D), $x, y$ (2D), and $x, y, z$ (3D).
"Ref-to-physical mapping"
Derivatives of reference coordinates with respect to physical coordinates are abbreviated, e.g., $\frac{\partial r}{\partial x} = r_x$. Additionally, $J$ is used to denote the determinant of the Jacobian of the reference-to-physical mapping.
Variable meanings and conventions in StartUpDG.jl
StartUpDG.jl exports structs RefElemData{NDIMS, ElemShape, ...} (which contains data associated with the reference element, such as interpolation points, quadrature rules, face nodes, normals, and differentiation/interpolation/projection matrices) and MeshData{NDIMS} (which contains geometric data associated with a mesh). These are currently used for evaluating DG formulations in a matrix-free fashion. These structs contain fields similar (but not identical) to those in Globals1D, Globals2D, Globals3D in the Matlab codes from "Nodal Discontinuous Galerkin Methods" by Hesthaven and Warburton (2007).
In general, we use the following code conventions:
  • variables such as r, s,... and x, y,... correspond to values at nodal interpolation points.
  • variables ending in q (e.g., rq, sq,... and xq, yq,...) correspond to values at volume quadrature points.
  • variables ending in f (e.g., rf, sf,... and xf, yf,...) correspond to values at face quadrature points.
  • variables ending in p (e.g., rp, sp,...) correspond to "plotting" points, which are usually a fine grid of equispaced points.
  • V matrices correspond to interpolation matrices from nodal interpolation points, e.g., Vq interpolates to volume quadrature points, Vf interpolates to face quadrature points.
  • geometric quantities in MeshData are stored as matrices of dimension $\text{number of points per element} \times \text{number of elements}$.
Quantities in rd::RefElemData:
  • rd.Np, rd.Nq, rd.Nf: the number of nodal interpolation points, volume quadrature points, and face quadrature points on the reference element, respectively.
  • rd.Vq: interpolation matrices from values at nodal interpolation points to volume quadrature points
  • rd.wq: volume quadrature weights on the reference element
  • rd.Vf: interpolation matrices from values at nodal interpolation points to face quadrature points
  • rd.wf: a vector containing face quadrature weights on the reference element
  • rd.M: the quadrature-based mass matrix, computed via rd.Vq' * diagm(rd.wq) * rd.Vq.
  • rd.Pq: a quadrature-based $L^2$ projection matrix rd.Pq = rd.M \ rd.Vq' * diagm(rd.wq) which maps between values at quadrature points and values at nodal points.
  • Dr, Ds, Dt matrices are nodal differentiation matrices with respect to the $r,s,t$ coordinates, e.g., Dr*f.(r,s) approximates the derivative of $f(r,s)$ at nodal points.
Quantities in md::MeshData:
  • md.xyz is a tuple of matrices md.x, md.y, md.z, where column e contains coordinates of physical interpolation points.
  • md.xyzq is a tuple of matrices md.xq, md.yq, md.zq, where column e contains coordinates of physical quadrature points.
  • md.rxJ, md.sxJ, ... are matrices where column e contains values of $J\frac{\partial r}{\partial x}$, $J\frac{\partial s}{\partial x}$, etc. at nodal interpolation points on the element e.
  • md.J is a matrix where column e contains values of the Jacobian $J$ at nodal interpolation points.
  • md.Jf is a matrix where column e contains values of the face Jacobian (e.g., determinant of the geometric mapping between a physical face and a reference face) at face quadrature points.
  • md.nxJ, md.nyJ, ... are matrices where column e contains values of components of the unit normal scaled by the face Jacobian md.Jf at face quadrature points.
For more details, please see the StartUpDG.jl docs.

+          RefElemData_kwargs...)

Here, element_type can be Tri(), Quad(), Tet(), or Hex(), and approximation_type can be

  • Polynomial(), which specifies a DG discretization using a polynomial basis using quadrature rules which are exact for degree 2 * polydeg integrands, or
  • SBP(), which specifies a DG discretization using multi-dimensional SBP operators. Types of SBP discretizations available include: SBP{Kubatko{LobattoFaceNodes}}() (the default choice), SBP{Kubatko{LegendreFaceNodes}}(), and SBP{Hicken}(). For polydeg = 1, ..., 4, the SBP{Kubatko{LegendreFaceNodes}}() SBP nodes are identical to the SBP nodes of Chen and Shu. More detailed descriptions of each SBP node set can be found in the StartUpDG.jl docs. Trixi.jl will also specialize certain parts of the solver based on the SBP approximation type.
  • a (periodic or non-periodic) derivative operator from SummationByPartsOperators.jl, usually constructed as D = derivative_operator(...). In this case, you do not need to pass a polydeg. Periodic derivative operators will only work with single-element meshes constructed using DGMultiMesh.

Additional options can also be specified through RefElemData_kwargs:

  • quad_rule_vol = quad_nodes(Tri(), Nq) will substitute in a volume quadrature rule of degree Nq instead of the default (which is a quadrature rule of degree polydeg). Here, a degree Nq rule will be exact for at least degree 2*Nq integrands (such that the mass matrix is integrated exactly). Quadrature rules of which exactly integrate degree Nq integrands may also be specified (for example, quad_rule_vol = StartUpDG.quad_nodes_tri(Nq) on triangles).
  • quad_rule_face = quad_nodes(Line(), Nq)) will use a face quadrature rule of degree Nq rather than the default. This rule is also exact for at least degree 2*Nq integrands.

The GaussSBP() approximation type on Quad() and Hex() meshes

When using VolumeIntegralFluxDifferencing on Quad() and Hex() meshes, one can also specify approximation_type = GaussSBP() to use an entropy stable Gauss collocation scheme. Here, GaussSBP() refers to "generalized" summation-by-parts operators (see for example Ranocha 2018 or Fernandez and Zingg 2015).

Unlike traditional SBP operators, generalized SBP operators are constructed from nodes which do not include boundary nodes (i.e., Gauss quadrature nodes as opposed to Gauss-Lobatto quadrature nodes). This makes the computation of interface fluxes slightly more expensive, but also usually results in a more accurate solution. Roughly speaking, an entropy stable Gauss collocation scheme will yield results similar to a modal entropy stable scheme using a Polynomial() approximation type, but will be more efficient at high orders of approximation.

Trixi.jl elixirs on simplicial and tensor product element meshes

Example elixirs with triangular, quadrilateral, and tetrahedral meshes can be found in the examples/dgmulti_2d/ and examples/dgmulti_3d/ folders. Some key elixirs to look at:

For developers

DGMultiMesh wrapper type

DGMulti meshes in Trixi.jl are represented using a DGMultiMesh{NDIMS, ...} type. This mesh type is assumed to have fields md::MeshData, which contains geometric terms derived from the mapping between the reference and physical elements, and boundary_faces, which contains a Dict of boundary segment names (symbols) and list of faces which lie on that boundary segment.

A DGMultiMesh can be constructed in several ways. For example, DGMultiMesh(dg::DGMulti) will return a Cartesian mesh on $[-1, 1]^d$ with element types specified by dg. DGMulti meshes can also be constructed by specifying a list of vertex coordinates vertex_coordinates_x, vertex_coordinates_y, vertex_coordinates_z and a connectivity matrix EToV where EToV[e,:] gives the vertices which correspond to element e. These quantities are available from most unstructured mesh generators.

Initial support for curved DGMultiMeshes is available for flux differencing solvers using SBP and GaussSBP approximation types on quadrilateral and hexahedral meshes. These can be called by specifying mesh = DGMultiMesh(dg, cells_per_dimension, mapping), where mapping is a function which specifies the warping of the mesh (e.g., mapping(xi, eta) = SVector{2}(xi, eta) is the identity mapping) similar to the mapping argument used by StructuredMesh.

Variable naming conventions

We use the convention that coordinates on the reference element are $r$ in 1D, $r, s$ in 2D, or $r, s, t$ in 3D. Physical coordinates use the standard conventions $x$ (1D), $x, y$ (2D), and $x, y, z$ (3D).

"Ref-to-physical mapping"

Derivatives of reference coordinates with respect to physical coordinates are abbreviated, e.g., $\frac{\partial r}{\partial x} = r_x$. Additionally, $J$ is used to denote the determinant of the Jacobian of the reference-to-physical mapping.

Variable meanings and conventions in StartUpDG.jl

StartUpDG.jl exports structs RefElemData{NDIMS, ElemShape, ...} (which contains data associated with the reference element, such as interpolation points, quadrature rules, face nodes, normals, and differentiation/interpolation/projection matrices) and MeshData{NDIMS} (which contains geometric data associated with a mesh). These are currently used for evaluating DG formulations in a matrix-free fashion. These structs contain fields similar (but not identical) to those in Globals1D, Globals2D, Globals3D in the Matlab codes from "Nodal Discontinuous Galerkin Methods" by Hesthaven and Warburton (2007).

In general, we use the following code conventions:

  • variables such as r, s,... and x, y,... correspond to values at nodal interpolation points.
  • variables ending in q (e.g., rq, sq,... and xq, yq,...) correspond to values at volume quadrature points.
  • variables ending in f (e.g., rf, sf,... and xf, yf,...) correspond to values at face quadrature points.
  • variables ending in p (e.g., rp, sp,...) correspond to "plotting" points, which are usually a fine grid of equispaced points.
  • V matrices correspond to interpolation matrices from nodal interpolation points, e.g., Vq interpolates to volume quadrature points, Vf interpolates to face quadrature points.
  • geometric quantities in MeshData are stored as matrices of dimension $\text{number of points per element} \times \text{number of elements}$.

Quantities in rd::RefElemData:

  • rd.Np, rd.Nq, rd.Nf: the number of nodal interpolation points, volume quadrature points, and face quadrature points on the reference element, respectively.
  • rd.Vq: interpolation matrices from values at nodal interpolation points to volume quadrature points
  • rd.wq: volume quadrature weights on the reference element
  • rd.Vf: interpolation matrices from values at nodal interpolation points to face quadrature points
  • rd.wf: a vector containing face quadrature weights on the reference element
  • rd.M: the quadrature-based mass matrix, computed via rd.Vq' * diagm(rd.wq) * rd.Vq.
  • rd.Pq: a quadrature-based $L^2$ projection matrix rd.Pq = rd.M \ rd.Vq' * diagm(rd.wq) which maps between values at quadrature points and values at nodal points.
  • Dr, Ds, Dt matrices are nodal differentiation matrices with respect to the $r,s,t$ coordinates, e.g., Dr*f.(r,s) approximates the derivative of $f(r,s)$ at nodal points.

Quantities in md::MeshData:

  • md.xyz is a tuple of matrices md.x, md.y, md.z, where column e contains coordinates of physical interpolation points.
  • md.xyzq is a tuple of matrices md.xq, md.yq, md.zq, where column e contains coordinates of physical quadrature points.
  • md.rxJ, md.sxJ, ... are matrices where column e contains values of $J\frac{\partial r}{\partial x}$, $J\frac{\partial s}{\partial x}$, etc. at nodal interpolation points on the element e.
  • md.J is a matrix where column e contains values of the Jacobian $J$ at nodal interpolation points.
  • md.Jf is a matrix where column e contains values of the face Jacobian (e.g., determinant of the geometric mapping between a physical face and a reference face) at face quadrature points.
  • md.nxJ, md.nyJ, ... are matrices where column e contains values of components of the unit normal scaled by the face Jacobian md.Jf at face quadrature points.

For more details, please see the StartUpDG.jl docs.

diff --git a/previews/PR1624/meshes/p4est_mesh/index.html b/previews/PR1624/meshes/p4est_mesh/index.html index 87cdc2a3226..4290cce80ee 100644 --- a/previews/PR1624/meshes/p4est_mesh/index.html +++ b/previews/PR1624/meshes/p4est_mesh/index.html @@ -1,16 +1,16 @@ -P4est-based mesh · Trixi.jl
GitHub

P4est-based mesh

The P4estMesh is an unstructured, curvilinear, nonconforming mesh type for quadrilateral (2D) and hexahedral (3D) cells. It supports quadtree/octree-based adaptive mesh refinement (AMR) via the C library p4est. See AMRCallback for further information.

Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.

Construction of a P4estMesh from an Abaqus file

One available option to construct a P4estMesh is to read in an Abaqus (.inp) mesh file. We briefly describe the structure of this file, the conventions it uses, and how the mesh file is parsed to create an initial unstructured, curvilinear, and conforming mesh.

Mesh in two spatial dimensions

For this discussion we use the following two-dimensional unstructured curved mesh with three elements:

abaqus-2dmesh-docs

We note that the corner and element connectivity information parsed from the Abaqus file creates a straight sided (linear) mesh. From this linear mesh there are two strategies available to make the mesh curvilinear:

  1. Apply a mapping function to describe a transformation of the linear mesh to another physical domain. The mapping is approximated using interpolation polynomial of a user specified polynomial degree. The default value of this polynomial degree is 1 that corresponds to an uncurved geometry.
  2. High-order boundary information is available in the .inp mesh file because it was created with the HOHQMesh mesh generator, which is available via the Julia package HOHQMesh.jl. This information is used to create appropriate transfinite mappings during the mesh construction.

We divide our discussion into two parts. The first part discusses the standard corner and element information contained in the .inp mesh file. The second part specifically deals with the mesh file parsing of an Abaqus file created by HOHQMesh.jl.

Mesh file header

An Abaqus .inp mesh file typically begins with a *Heading. Though optional, the *Heading is helpful to give users some information about the mesh described by the mesh file. In particular, a .inp mesh file created with HOHQMesh will contain the header

*Heading
- File created by HOHQMesh

This heading is used to indicate to the mesh constructor which of the above mapping strategies to apply in order to create a curvilinear mesh. If the Abaqus file header is not present then the P4estMesh is created with the first strategy above.

List of corner nodes

Next, prefaced with *NODE, comes a list of the physical (x,y,z) coordinates of all the corners. The first integer in the list of the corners provides its id number. Thus, for the two-dimensional example mesh this block of corner information is

*NODE
+P4est-based mesh · Trixi.jl
GitHub
P4est-based mesh
The P4estMesh is an unstructured, curvilinear, nonconforming mesh type for quadrilateral (2D) and hexahedral (3D) cells. It supports quadtree/octree-based adaptive mesh refinement (AMR) via the C library p4est. See AMRCallback for further information.
Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.
Construction of a P4estMesh from an Abaqus file
One available option to construct a P4estMesh is to read in an Abaqus (.inp) mesh file. We briefly describe the structure of this file, the conventions it uses, and how the mesh file is parsed to create an initial unstructured, curvilinear, and conforming mesh.
Mesh in two spatial dimensions
For this discussion we use the following two-dimensional unstructured curved mesh with three elements:
abaqus-2dmesh-docs
We note that the corner and element connectivity information parsed from the Abaqus file creates a straight sided (linear) mesh. From this linear mesh there are two strategies available to make the mesh curvilinear:
  1. Apply a mapping function to describe a transformation of the linear mesh to another physical domain. The mapping is approximated using interpolation polynomial of a user specified polynomial degree. The default value of this polynomial degree is 1 that corresponds to an uncurved geometry.
  2. High-order boundary information is available in the .inp mesh file because it was created with the HOHQMesh mesh generator, which is available via the Julia package HOHQMesh.jl. This information is used to create appropriate transfinite mappings during the mesh construction.
We divide our discussion into two parts. The first part discusses the standard corner and element information contained in the .inp mesh file. The second part specifically deals with the mesh file parsing of an Abaqus file created by HOHQMesh.jl.
Mesh file header
An Abaqus .inp mesh file typically begins with a *Heading. Though optional, the *Heading is helpful to give users some information about the mesh described by the mesh file. In particular, a .inp mesh file created with HOHQMesh will contain the header
*Heading
+ File created by HOHQMesh
This heading is used to indicate to the mesh constructor which of the above mapping strategies to apply in order to create a curvilinear mesh. If the Abaqus file header is not present then the P4estMesh is created with the first strategy above.
List of corner nodes
Next, prefaced with *NODE, comes a list of the physical (x,y,z) coordinates of all the corners. The first integer in the list of the corners provides its id number. Thus, for the two-dimensional example mesh this block of corner information is
*NODE
 1, 1.0, -1.0, 0.0
 2, 3.0,  0.0, 0.0
 3, 1.0,  1.0, 0.0
 4, 2.0,  0.0, 0.0
 5, 0.0,  0.0, 0.0
 6, 3.0,  1.0, 0.0
-7, 3.0, -1.0, 0.0
List of elements
The element connectivity is given after the list of corners. The header for this information block is
*ELEMENT, type=CPS4, ELSET=Surface1
The Abaqus element type CPS4 corresponds to a quadrilateral element. Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. The elements connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion; making the element right-handed. This element handedness is indicated using the circular arrow in the figure above. Just as with the corner list, the first integer in the element connectivity list indicates the element id number. Thus, the element connectivity list for the three element example mesh is
*ELEMENT, type=CPS4, ELSET=Surface1
+7, 3.0, -1.0, 0.0
List of elements
The element connectivity is given after the list of corners. The header for this information block is
*ELEMENT, type=CPS4, ELSET=Surface1
The Abaqus element type CPS4 corresponds to a quadrilateral element. Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. The elements connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion; making the element right-handed. This element handedness is indicated using the circular arrow in the figure above. Just as with the corner list, the first integer in the element connectivity list indicates the element id number. Thus, the element connectivity list for the three element example mesh is
*ELEMENT, type=CPS4, ELSET=Surface1
 1, 5, 1, 4, 3
 2, 4, 2, 6, 3
-3, 7, 2, 4, 1
Element neighbor connectivity
The construction of the element neighbor ids and identifying physical boundary surfaces is done using functionality directly from the p4est library. For example, the neighbor connectivity is created in the mesh constructor using the wrapper read_inp_p4est function.
HOHQMesh boundary information
If present, any additional information in the mesh file that was created by HOHQMesh is prefaced with ** to make it an Abaqus comment. This ensures that the read in of the file by a standard Abaqus file parser, as done in the read_inp_p4est function, is done correctly.
The high-order, curved boundary information and labels of the physical boundary created by HOHQMesh is found below the comment line
** ***** HOHQMesh boundary information ***** **
Next comes the polynomial degree that the mesh will use to represent any curved sides
** mesh polynomial degree = 8
The mesh file then, again, provides the element connectivity as well as information for curved surfaces either interior to the domain or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (-y, +x, +y, or -x) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y,z) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y,z) data will be given in the direction of the local coordinate system. Given below is the element curvature information for the example mesh:
**  5 1 4 3
+3, 7, 2, 4, 1
Element neighbor connectivity
The construction of the element neighbor ids and identifying physical boundary surfaces is done using functionality directly from the p4est library. For example, the neighbor connectivity is created in the mesh constructor using the wrapper read_inp_p4est function.
Encoding of boundaries
HOHQMesh boundary information
If present, any additional information in the mesh file that was created by HOHQMesh is prefaced with ** to make it an Abaqus comment. This ensures that the read in of the file by a standard Abaqus file parser, as done in the read_inp_p4est function, is done correctly.
The high-order, curved boundary information and labels of the physical boundary created by HOHQMesh is found below the comment line
** ***** HOHQMesh boundary information ***** **
Next comes the polynomial degree that the mesh will use to represent any curved sides
** mesh polynomial degree = 8
The mesh file then, again, provides the element connectivity as well as information for curved surfaces either interior to the domain or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (-y, +x, +y, or -x) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y,z) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y,z) data will be given in the direction of the local coordinate system. Given below is the element curvature information for the example mesh:
**  5 1 4 3
 **  0 0 1 1
 **   1.000000000000000   1.000000000000000   0.0
 **   1.024948365654583   0.934461926834452   0.0
@@ -95,7 +95,10 @@
 **  0 0 0 0
 **  Bezier --- Slant ---
 **  --- Right --- Top
-**  Bottom --- Right ---
Mesh in three spatial dimensions
The 3D Abaqus file format with high-order boundary information from HOHQMesh is very similar to the 2D version discussed above. There are only three changes:
  1. The element connectivity would be given in terms of the eight corners that define a hexahedron. The corners are numbered as shown in the figure below. The header of the element list changes to be
    *ELEMENT, type=C3D8, ELSET=Volume1
    where C3D8 corresponds to a Abaqus hexahedral element.
  2. There are six check digits included directly below the eight corner indexes to indicate whether the local face within the element is straight sided, 0, or is curved, 1. For curved faces (x,y,z) coordinate values are available in order to construct an face interpolant with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes.
  3. The boundary labels are given in the following order according to the local surface index -x +x -y +y -z +z as required by the p4est library.
For completeness, we also give a short description and derivation of the three-dimensional transfinite mapping formulas used to compute the physical coordinates $\mathbf{x}=(x,y,z)$ of a (possibly curved) hexahedral element give the reference coordinates $\boldsymbol{\xi} = (\xi, \eta, \zeta)$ which lie in $[-1,1]^3$. That is, we will create an expression $\mathbf{x}= \mathbf{X}(\boldsymbol{\xi})$.
Below we provide a sketch of a single hexahedral element with curved faces. This is done to introduce the numbering conventions for corners, edges, and faces of the element.
abaqus-3dmesh-docs
When the hexahedron is a straight sided (linear) element we compute the transfinite mapping directly from the element corner points according to
\[\begin{aligned}
+**  Bottom --- Right ---
Standard Abaqus format boundary information

As an alternative to an Abaqus mesh generated by HOHQMesh, .inp files with boundary information encoded as nodesets *NSET,NSET= can be used to construct a p4est mesh. This is especially useful for usage of existing meshes (consisting of bilinear elements) which could stem from the popular gmsh meshing software.

In addition to the list of nodes and elements given above, there are nodesets of the form 

*NSET,NSET=PhysicalLine1
+1, 4, 52, 53, 54, 55, 56, 57, 58,

present which are used to associate the edges defined through their corner nodes with a label. In this case it is called PhysicalLine1. By looping over every element and its associated edges, consisting of two nodes, we query the read in NSETs if the current node pair is present.

To prevent that every nodeset following *NSET,NSET= is treated as a boundary, the user must supply a boundary_symbols keyword to the P4estMesh constructor:

boundary_symbols = [:PhysicalLine1]
+
+mesh = P4estMesh{2}(mesh_file, polydeg = polydeg, boundary_symbols = boundary_symbols)

By doing so, only nodesets with a label present in boundary_symbols are treated as physical boundaries. Other nodesets that could be used for diagnostics are not treated as external boundaries. Note that there is a leading colon : compared to the label in the .inp mesh file. This is required to turn the label into a Symbol.

A 2D example for this mesh, which is read-in for an unstructured mesh file created with gmsh, is presented in examples/p4est_2d_dgsem/elixir_euler_NACA6412airfoil_mach2.jl.

Mesh in three spatial dimensions

HOHQMesh-Extended Abaqus format

The 3D Abaqus file format with high-order boundary information from HOHQMesh is very similar to the 2D version discussed above. There are only three changes:

  1. The element connectivity would be given in terms of the eight corners that define a hexahedron. The corners are numbered as shown in the figure below. The header of the element list changes to be
    *ELEMENT, type=C3D8, ELSET=Volume1
    where C3D8 corresponds to a Abaqus hexahedral element.
  2. There are six check digits included directly below the eight corner indexes to indicate whether the local face within the element is straight sided, 0, or is curved, 1. For curved faces (x,y,z) coordinate values are available in order to construct an face interpolant with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes.
  3. The boundary labels are given in the following order according to the local surface index -x +x -y +y -z +z as required by the p4est library.

For completeness, we also give a short description and derivation of the three-dimensional transfinite mapping formulas used to compute the physical coordinates $\mathbf{x}=(x,y,z)$ of a (possibly curved) hexahedral element give the reference coordinates $\boldsymbol{\xi} = (\xi, \eta, \zeta)$ which lie in $[-1,1]^3$. That is, we will create an expression $\mathbf{x}= \mathbf{X}(\boldsymbol{\xi})$.

Below we provide a sketch of a single hexahedral element with curved faces. This is done to introduce the numbering conventions for corners, edges, and faces of the element.

abaqus-3dmesh-docs

When the hexahedron is a straight sided (linear) element we compute the transfinite mapping directly from the element corner points according to

\[\begin{aligned} \mathbf{X}_{linear}(\boldsymbol{\xi}) &= \frac{1}{8}[\quad\, \mathbf{x}_1(1-\xi)(1-\eta)(1-\zeta) + \mathbf{x}_2(1+\xi)(1-\eta)(1-\zeta)\\[-0.15cm] & \qquad\; + \mathbf{x}_3(1+\xi)(1+\eta)(1-\zeta) @@ -133,4 +136,213 @@ & \qquad\; + (1-\xi)(1+\zeta)\texttt{edge}_{12}\quad]. \end{aligned}\]

However, subtracting the edge correction terms $\mathcal{C}_{\texttt{edge}}(\boldsymbol{\xi})$ from $\boldsymbol\Sigma(\boldsymbol{\xi})$ removes the interior element contributions twice. Thus, to complete the construction of the transfinite mapping $\mathbf{X}(\boldsymbol{\xi})$ we add the transfinite map of the straight sided hexahedral element to find

\[\mathbf{X}(\boldsymbol{\xi}) = \boldsymbol\Sigma(\boldsymbol{\xi}) - \mathcal{C}_{\texttt{edge}}(\boldsymbol{\xi}) - + \mathbf{X}_{linear}(\boldsymbol{\xi}).\]

+ + \mathbf{X}_{linear}(\boldsymbol{\xi}).\]

Construction from standard Abaqus

Also for a mesh in standard Abaqus format there are no qualitative changes when going from 2D to 3D. The most notable difference is that boundaries are formed in 3D by faces defined by four nodes while in 2D boundaries are edges consisting of two elements. A simple mesh file, which is used also in examples/p4est_3d_dgsem/elixir_euler_free_stream_boundaries.jl, is given below:

*Heading
+<SOMETHING DIFFERENT FROM "File created by HOHQMesh">
+*NODE
+1, -2, 0, 0
+2, -1, 0, 0
+3, -1, 1, 0
+4, -2, 1, 0
+5, -2, 0, 1
+6, -1, 0, 1
+7, -1, 1, 1
+8, -2, 1, 1
+9, -1.75, 1, 0
+10, -1.5, 1, 0
+11, -1.25, 1, 0
+12, -1, 0.75000000000035, 0
+13, -1, 0.50000000000206, 0
+14, -1, 0.25000000000104, 0
+15, -1.25, 0, 0
+16, -1.5, 0, 0
+17, -1.75, 0, 0
+18, -2, 0.24999999999941, 0
+19, -2, 0.49999999999869, 0
+20, -2, 0.74999999999934, 0
+21, -1.75, 0, 1
+22, -1.5, 0, 1
+23, -1.25, 0, 1
+24, -1, 0.24999999999941, 1
+25, -1, 0.49999999999869, 1
+26, -1, 0.74999999999934, 1
+27, -1.25, 1, 1
+28, -1.5, 1, 1
+29, -1.75, 1, 1
+30, -2, 0.75000000000035, 1
+31, -2, 0.50000000000206, 1
+32, -2, 0.25000000000104, 1
+33, -2, 0, 0.24999999999941
+34, -2, 0, 0.49999999999869
+35, -2, 0, 0.74999999999934
+36, -2, 1, 0.24999999999941
+37, -2, 1, 0.49999999999869
+38, -2, 1, 0.74999999999934
+39, -1, 0, 0.24999999999941
+40, -1, 0, 0.49999999999869
+41, -1, 0, 0.74999999999934
+42, -1, 1, 0.24999999999941
+43, -1, 1, 0.49999999999869
+44, -1, 1, 0.74999999999934
+45, -1.25, 0.25000000000063, 0
+46, -1.25, 0.50000000000122, 0
+47, -1.25, 0.7500000000001, 0
+48, -1.5, 0.25000000000023, 0
+49, -1.5, 0.50000000000038, 0
+50, -1.5, 0.74999999999984, 0
+51, -1.75, 0.24999999999982, 0
+52, -1.75, 0.49999999999953, 0
+53, -1.75, 0.74999999999959, 0
+54, -1.75, 0.25000000000063, 1
+55, -1.75, 0.50000000000122, 1
+56, -1.75, 0.7500000000001, 1
+57, -1.5, 0.25000000000023, 1
+58, -1.5, 0.50000000000038, 1
+59, -1.5, 0.74999999999984, 1
+60, -1.25, 0.24999999999982, 1
+61, -1.25, 0.49999999999953, 1
+62, -1.25, 0.74999999999959, 1
+63, -2, 0.24999999999982, 0.24999999999941
+64, -2, 0.49999999999953, 0.24999999999941
+65, -2, 0.74999999999959, 0.24999999999941
+66, -2, 0.25000000000023, 0.49999999999869
+67, -2, 0.50000000000038, 0.49999999999869
+68, -2, 0.74999999999984, 0.49999999999869
+69, -2, 0.25000000000063, 0.74999999999934
+70, -2, 0.50000000000122, 0.74999999999934
+71, -2, 0.7500000000001, 0.74999999999934
+72, -1.25, 1, 0.74999999999934
+73, -1.25, 1, 0.49999999999869
+74, -1.25, 1, 0.24999999999941
+75, -1.5, 1, 0.74999999999934
+76, -1.5, 1, 0.49999999999869
+77, -1.5, 1, 0.24999999999941
+78, -1.75, 1, 0.74999999999934
+79, -1.75, 1, 0.49999999999869
+80, -1.75, 1, 0.24999999999941
+81, -1, 0.25000000000063, 0.24999999999941
+82, -1, 0.50000000000122, 0.24999999999941
+83, -1, 0.7500000000001, 0.24999999999941
+84, -1, 0.25000000000023, 0.49999999999869
+85, -1, 0.50000000000038, 0.49999999999869
+86, -1, 0.74999999999984, 0.49999999999869
+87, -1, 0.24999999999982, 0.74999999999934
+88, -1, 0.49999999999953, 0.74999999999934
+89, -1, 0.74999999999959, 0.74999999999934
+90, -1.75, 0, 0.74999999999934
+91, -1.75, 0, 0.49999999999869
+92, -1.75, 0, 0.24999999999941
+93, -1.5, 0, 0.74999999999934
+94, -1.5, 0, 0.49999999999869
+95, -1.5, 0, 0.24999999999941
+96, -1.25, 0, 0.74999999999934
+97, -1.25, 0, 0.49999999999869
+98, -1.25, 0, 0.24999999999941
+99, -1.75, 0.25000000000043, 0.74999999999934
+100, -1.75, 0.25000000000023, 0.49999999999869
+101, -1.75, 0.25000000000002, 0.24999999999941
+102, -1.75, 0.5000000000008, 0.74999999999934
+103, -1.75, 0.50000000000038, 0.49999999999869
+104, -1.75, 0.49999999999995, 0.24999999999941
+105, -1.75, 0.74999999999997, 0.74999999999934
+106, -1.75, 0.74999999999984, 0.49999999999869
+107, -1.75, 0.74999999999972, 0.24999999999941
+108, -1.5, 0.25000000000023, 0.74999999999934
+109, -1.5, 0.25000000000023, 0.49999999999869
+110, -1.5, 0.25000000000023, 0.24999999999941
+111, -1.5, 0.50000000000038, 0.74999999999934
+112, -1.5, 0.50000000000038, 0.49999999999869
+113, -1.5, 0.50000000000038, 0.24999999999941
+114, -1.5, 0.74999999999984, 0.74999999999934
+115, -1.5, 0.74999999999984, 0.49999999999869
+116, -1.5, 0.74999999999984, 0.24999999999941
+117, -1.25, 0.25000000000002, 0.74999999999934
+118, -1.25, 0.25000000000023, 0.49999999999869
+119, -1.25, 0.25000000000043, 0.24999999999941
+120, -1.25, 0.49999999999995, 0.74999999999934
+121, -1.25, 0.50000000000038, 0.49999999999869
+122, -1.25, 0.5000000000008, 0.24999999999941
+123, -1.25, 0.74999999999972, 0.74999999999934
+124, -1.25, 0.74999999999984, 0.49999999999869
+125, -1.25, 0.74999999999997, 0.24999999999941
+******* E L E M E N T S *************
+*ELEMENT, type=C3D8, ELSET=Volume1
+153, 54, 21, 5, 32, 99, 90, 35, 69
+154, 99, 90, 35, 69, 100, 91, 34, 66
+155, 100, 91, 34, 66, 101, 92, 33, 63
+156, 101, 92, 33, 63, 51, 17, 1, 18
+157, 55, 54, 32, 31, 102, 99, 69, 70
+158, 102, 99, 69, 70, 103, 100, 66, 67
+159, 103, 100, 66, 67, 104, 101, 63, 64
+160, 104, 101, 63, 64, 52, 51, 18, 19
+161, 56, 55, 31, 30, 105, 102, 70, 71
+162, 105, 102, 70, 71, 106, 103, 67, 68
+163, 106, 103, 67, 68, 107, 104, 64, 65
+164, 107, 104, 64, 65, 53, 52, 19, 20
+165, 29, 56, 30, 8, 78, 105, 71, 38
+166, 78, 105, 71, 38, 79, 106, 68, 37
+167, 79, 106, 68, 37, 80, 107, 65, 36
+168, 80, 107, 65, 36, 9, 53, 20, 4
+169, 57, 22, 21, 54, 108, 93, 90, 99
+170, 108, 93, 90, 99, 109, 94, 91, 100
+171, 109, 94, 91, 100, 110, 95, 92, 101
+172, 110, 95, 92, 101, 48, 16, 17, 51
+173, 58, 57, 54, 55, 111, 108, 99, 102
+174, 111, 108, 99, 102, 112, 109, 100, 103
+175, 112, 109, 100, 103, 113, 110, 101, 104
+176, 113, 110, 101, 104, 49, 48, 51, 52
+177, 59, 58, 55, 56, 114, 111, 102, 105
+178, 114, 111, 102, 105, 115, 112, 103, 106
+179, 115, 112, 103, 106, 116, 113, 104, 107
+180, 116, 113, 104, 107, 50, 49, 52, 53
+181, 28, 59, 56, 29, 75, 114, 105, 78
+182, 75, 114, 105, 78, 76, 115, 106, 79
+183, 76, 115, 106, 79, 77, 116, 107, 80
+184, 77, 116, 107, 80, 10, 50, 53, 9
+185, 60, 23, 22, 57, 117, 96, 93, 108
+186, 117, 96, 93, 108, 118, 97, 94, 109
+187, 118, 97, 94, 109, 119, 98, 95, 110
+188, 119, 98, 95, 110, 45, 15, 16, 48
+189, 61, 60, 57, 58, 120, 117, 108, 111
+190, 120, 117, 108, 111, 121, 118, 109, 112
+191, 121, 118, 109, 112, 122, 119, 110, 113
+192, 122, 119, 110, 113, 46, 45, 48, 49
+193, 62, 61, 58, 59, 123, 120, 111, 114
+194, 123, 120, 111, 114, 124, 121, 112, 115
+195, 124, 121, 112, 115, 125, 122, 113, 116
+196, 125, 122, 113, 116, 47, 46, 49, 50
+197, 27, 62, 59, 28, 72, 123, 114, 75
+198, 72, 123, 114, 75, 73, 124, 115, 76
+199, 73, 124, 115, 76, 74, 125, 116, 77
+200, 74, 125, 116, 77, 11, 47, 50, 10
+201, 24, 6, 23, 60, 87, 41, 96, 117
+202, 87, 41, 96, 117, 84, 40, 97, 118
+203, 84, 40, 97, 118, 81, 39, 98, 119
+204, 81, 39, 98, 119, 14, 2, 15, 45
+205, 25, 24, 60, 61, 88, 87, 117, 120
+206, 88, 87, 117, 120, 85, 84, 118, 121
+207, 85, 84, 118, 121, 82, 81, 119, 122
+208, 82, 81, 119, 122, 13, 14, 45, 46
+209, 26, 25, 61, 62, 89, 88, 120, 123
+210, 89, 88, 120, 123, 86, 85, 121, 124
+211, 86, 85, 121, 124, 83, 82, 122, 125
+212, 83, 82, 122, 125, 12, 13, 46, 47
+213, 7, 26, 62, 27, 44, 89, 123, 72
+214, 44, 89, 123, 72, 43, 86, 124, 73
+215, 43, 86, 124, 73, 42, 83, 125, 74
+216, 42, 83, 125, 74, 3, 12, 47, 11
+*NSET,NSET=PhysicalSurface1
+1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 
+11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 
+21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 
+31, 32, 33, 34, 35, 36, 37, 38, 45, 46, 
+47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 
+57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 
+67, 68, 69, 70, 71, 
+*NSET,NSET=PhysicalSurface2
+1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 
+11, 12, 13, 14, 15, 16, 17, 21, 22, 23, 
+24, 25, 26, 27, 28, 29, 33, 34, 35, 36, 
+37, 38, 39, 40, 41, 42, 43, 44, 72, 73, 
+74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 
+84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 
+94, 95, 96, 97, 98,
diff --git a/previews/PR1624/meshes/structured_mesh/index.html b/previews/PR1624/meshes/structured_mesh/index.html index 8fca2d48d6e..05c09800044 100644 --- a/previews/PR1624/meshes/structured_mesh/index.html +++ b/previews/PR1624/meshes/structured_mesh/index.html @@ -1,2 +1,2 @@ -Structured mesh · Trixi.jl
GitHub

Structured mesh

The StructuredMesh is a structured, curvilinear, conforming mesh type available for one-, two-, and three-dimensional simulations.

Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.

+Structured mesh · Trixi.jl
GitHub

Structured mesh

The StructuredMesh is a structured, curvilinear, conforming mesh type available for one-, two-, and three-dimensional simulations.

Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.

diff --git a/previews/PR1624/meshes/tree_mesh/index.html b/previews/PR1624/meshes/tree_mesh/index.html index 47e701c1a7b..280d85d4a1a 100644 --- a/previews/PR1624/meshes/tree_mesh/index.html +++ b/previews/PR1624/meshes/tree_mesh/index.html @@ -1,2 +1,2 @@ -Tree mesh · Trixi.jl
GitHub

Tree mesh

The TreeMesh is a Cartesian, $h$-non-conforming mesh type used in many parts of Trixi.jl. Often, the support for this mesh type is developed best since it was the first mesh type in Trixi.jl, and it is available in one, two, and three space dimensions.

It is limited to hypercube domains but supports AMR via the AMRCallback. Due to its Cartesian nature, (numerical) fluxes need to implement methods dispatching on the orientation::Integer as described in the conventions.

+Tree mesh · Trixi.jl
GitHub

Tree mesh

The TreeMesh is a Cartesian, $h$-non-conforming mesh type used in many parts of Trixi.jl. Often, the support for this mesh type is developed best since it was the first mesh type in Trixi.jl, and it is available in one, two, and three space dimensions.

It is limited to hypercube domains but supports AMR via the AMRCallback. Due to its Cartesian nature, (numerical) fluxes need to implement methods dispatching on the orientation::Integer as described in the conventions.

diff --git a/previews/PR1624/meshes/unstructured_quad_mesh/index.html b/previews/PR1624/meshes/unstructured_quad_mesh/index.html index a769ffb128c..82b589dcfc6 100644 --- a/previews/PR1624/meshes/unstructured_quad_mesh/index.html +++ b/previews/PR1624/meshes/unstructured_quad_mesh/index.html @@ -1,5 +1,5 @@ -Unstructured mesh · Trixi.jl
GitHub

Unstructured quadrilateral mesh

The UnstructuredMesh2D is an unstructured, curvilinear, conforming mesh type in two space dimensions.

Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.

Next, we describe the conventions taken in the implementation for two-dimensional unstructured quadrilateral meshes. Principally, this relates to how a file with the extension .mesh encodes information about the numbering and orientation of elements in an unstructured quadrilateral mesh with possibly curved boundaries.

We use the following unstructured mesh with three elements for this discussion:

example-mesh

Further, a simulation using Trixi.jl on this example unstructured mesh is provided in examples/unstructured_2d_dgsem/elixir_euler_basic.jl.

Mesh file header

The first two lines of the mesh file lists the mesh file type as well as the total number of corners, surfaces, elements, and the polynomial degree that the mesh will use to represent any curved sides. For the example mesh these quantities are

ISM-V2
+Unstructured mesh · Trixi.jl
GitHub
Unstructured quadrilateral mesh
The UnstructuredMesh2D is an unstructured, curvilinear, conforming mesh type in two space dimensions.
Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.
Next, we describe the conventions taken in the implementation for two-dimensional unstructured quadrilateral meshes. Principally, this relates to how a file with the extension .mesh encodes information about the numbering and orientation of elements in an unstructured quadrilateral mesh with possibly curved boundaries.
We use the following unstructured mesh with three elements for this discussion:
example-mesh
Further, a simulation using Trixi.jl on this example unstructured mesh is provided in examples/unstructured_2d_dgsem/elixir_euler_basic.jl.
Mesh file header
The first two lines of the mesh file lists the mesh file type as well as the total number of corners, surfaces, elements, and the polynomial degree that the mesh will use to represent any curved sides. For the example mesh these quantities are
ISM-V2
     7    9    3    8
corresponding to seven corners, nine surfaces, and three elements. The mesh polynomial degree of eight is taken only for illustrative purposes. In practice, this mesh polynomial degree depends on the particular application for which the curved, unstructured mesh is required.
List of corner nodes
After these global counts that prescribe information about the mesh skeleton, the mesh file give a list of the physical (x,y) coordinates of all the corners. The corner nodes are listed in the order prescribed by mesh generator. Thus, for the example mesh this node list would be
 1.0    -1.0
  3.0    0.0
  1.0    1.0
@@ -76,4 +76,4 @@
   * Upload to YouTube
   * Obtain responsive code by inserting link on https://embedresponsively.com
   -->
-  

+
diff --git a/previews/PR1624/multi-physics_coupling/index.html b/previews/PR1624/multi-physics_coupling/index.html new file mode 100644 index 00000000000..74b8600a116 --- /dev/null +++ b/previews/PR1624/multi-physics_coupling/index.html @@ -0,0 +1,2 @@ + +Coupling · Trixi.jl
GitHub

Multi-physics coupling

A complex simulation can consist of different spatial domains in which different equations are being solved, different numerical methods being used or the grid structure is different. One example would be a fluid in a tank and an extended hot plate attached to it. We would then like to solve the Navier-Stokes equations in the fluid domain and the heat conduction equations in the plate. The coupling would happen at the interface through the exchange of thermal energy.

Converter coupling

It may happen that the two systems to be coupled do not share any variables, but share some of the physics. In such a situation, the same physics is just represented in a different form and with a different set of variables. This is the case, for instance assuming two domains, if there is a fluid system in one domain and a Vlasov system in the other domain. In that case we would have variables representing distribution functions of the Vlasov system on one side and variables representing the mechanical quantities, like density, of the fluid system. To translate the fields from one description to the other one needs to use converter functions. These functions need to be hand tailored by the user in the elixir file where each pair of coupled systems requires two coupling functions, one for each direction.

In the general case, we have a system $A$ with $m$ variables $u_{A,i}, \: i = 1, \dots, m$ and another system $B$ with $n$ variables $u_{B,j}, \: j = 1, \dots, n$. We then define two coupling functions, one that transforms $u_A$ into $u_B$ and one that goes the other way.

In their minimal form they take the position vector $x$, state vector $u$ and the equations of the two coupled systems and return the transformed variables. By passing the equations we can make use of their parameters, if they are required. Examples can be seen in examples/structured_2d_dgsem/elixir_advection_coupled.jl.

Warning about binary compatibility

Currently the coordinate values on the nodes can differ by machine precision when simulating the mesh and when splitting the mesh in multiple domains. This is an issue coming from the coordinate interpolation on the nodes. As a result, running a simulation in a single system and in two coupled domains may result in a difference of the order of the machine precision. While this is not an issue for most practical problems, it is best to keep this in mind when comparing test runs.

diff --git a/previews/PR1624/overview/index.html b/previews/PR1624/overview/index.html index ee1d547a0e1..f27c1e5dd0d 100644 --- a/previews/PR1624/overview/index.html +++ b/previews/PR1624/overview/index.html @@ -1,2 +1,2 @@ -Overview · Trixi.jl
GitHub

Overview of the structure of Trixi.jl

Trixi.jl is designed as a library of components for discretizations of hyperbolic conservation laws. Thus, it is not a monolithic PDE solver that is configured at runtime via parameter files, as it is often found in classical numerical simulation codes. Instead, each simulation is configured by pure Julia code. Many examples of such simulation setups, called elixirs in Trixi.jl, are provided in the examples/ folder.

Trixi.jl uses the method of lines, i.e., the full space-time discretization is separated into two steps; the spatial semidiscretization is performed at first and the resulting ODE system is solved numerically using a suitable time integration method. Thus, the main ingredients of an elixir designed to solve a PDE numerically are the spatial semidiscretization and the time integration scheme.

Semidiscretizations

Semidiscretizations are high-level descriptions of spatial discretizations specialized for certain PDEs. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Such semidiscretizations are usually named semi in Trixi.jl

semidiscretization_overview

The basic building blocks of a semidiscretization are

  • a mesh describing the geometry of the domain
  • a set of equations describing the physical model
  • a solver describing the numerical approach

In addition, a semidiscretization bundles initial and boundary conditions, and possible source terms. These different ingredients of a semidiscretization can be configured individually and combined together. When a semidiscretization is constructed, it will create an internal cache, i.e., a collection of setup-specific data structures, that is usually passed to all lower level functions.

Due to Trixi.jl's modular nature using Julia's multiple dispatch features, new ingredients can be created specifically for a certain combination of other ingredients. For example, a new mesh type can be created and implemented at first only for a specific solver. Thus, there is no need to consider all possible combinations of meshes, equations, and solvers when implementing new features. This allows rapid prototyping of new ideas and is one of the main design goals behind Trixi.jl. Below is a brief overview of the availability of different features on different mesh types.

FeatureTreeMeshStructuredMeshUnstructuredMesh2DP4estMeshDGMultiMeshFurther reading
Spatial dimension1D, 2D, 3D1D, 2D, 3D2D2D, 3D1D, 2D, 3D
CoordinatesCartesiancurvilinearcurvilinearcurvilinearcurvilinear
Connectivityh-nonconformingconformingconformingh-nonconformingconforming
Element typeline, square, cubeline, quadᵃ, hexᵃquadᵃquadᵃ, hexᵃsimplex, quadᵃ, hexᵃ
Adaptive mesh refinement✅❌❌✅❌AMRCallback
Solver typeDGSEMDGSEMDGSEMDGSEMDGMulti
Domainhypercubemapped hypercubearbitraryarbitraryarbitrary
Weak form✅✅✅✅✅VolumeIntegralWeakForm
Flux differencing✅✅✅✅✅VolumeIntegralFluxDifferencing
Shock capturing✅✅✅✅❌VolumeIntegralShockCapturingHG
Nonconservative equations✅✅✅✅✅e.g., GLM MHD or shallow water equations
Parabolic terms✅✅❌✅✅e.g., CompressibleNavierStokesDiffusion2D

ᵃ: quad = quadrilateral, hex = hexahedron

Time integration methods

Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem is a wrapper of Trixi.rhs!(du_ode, u_ode, semi, t), which gets called in ODE solvers. Further information can be found in the section on time integration methods.

Next steps

We explicitly encourage people interested in Trixi.jl to have a look at the examples/ bundled with Trixi.jl to get an impression of what is possible and the general look and feel of Trixi.jl. Before doing that, it is usually good to get an idea of how to visualize numerical results.

If you like learning by doing, looking at the tutorials and trying to mix your own elixirs based thereon is probably a good next step. Otherwise, you can further dig into the documentation by looking at Trixi.jl's basic building blocks.

+Overview · Trixi.jl
GitHub

Overview of the structure of Trixi.jl

Trixi.jl is designed as a library of components for discretizations of hyperbolic conservation laws. Thus, it is not a monolithic PDE solver that is configured at runtime via parameter files, as it is often found in classical numerical simulation codes. Instead, each simulation is configured by pure Julia code. Many examples of such simulation setups, called elixirs in Trixi.jl, are provided in the examples/ folder.

Trixi.jl uses the method of lines, i.e., the full space-time discretization is separated into two steps; the spatial semidiscretization is performed at first and the resulting ODE system is solved numerically using a suitable time integration method. Thus, the main ingredients of an elixir designed to solve a PDE numerically are the spatial semidiscretization and the time integration scheme.

Semidiscretizations

Semidiscretizations are high-level descriptions of spatial discretizations specialized for certain PDEs. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Such semidiscretizations are usually named semi in Trixi.jl

semidiscretization_overview

The basic building blocks of a semidiscretization are

  • a mesh describing the geometry of the domain
  • a set of equations describing the physical model
  • a solver describing the numerical approach

In addition, a semidiscretization bundles initial and boundary conditions, and possible source terms. These different ingredients of a semidiscretization can be configured individually and combined together. When a semidiscretization is constructed, it will create an internal cache, i.e., a collection of setup-specific data structures, that is usually passed to all lower level functions.

Due to Trixi.jl's modular nature using Julia's multiple dispatch features, new ingredients can be created specifically for a certain combination of other ingredients. For example, a new mesh type can be created and implemented at first only for a specific solver. Thus, there is no need to consider all possible combinations of meshes, equations, and solvers when implementing new features. This allows rapid prototyping of new ideas and is one of the main design goals behind Trixi.jl. Below is a brief overview of the availability of different features on different mesh types.

FeatureTreeMeshStructuredMeshUnstructuredMesh2DP4estMeshDGMultiMeshFurther reading
Spatial dimension1D, 2D, 3D1D, 2D, 3D2D2D, 3D1D, 2D, 3D
CoordinatesCartesiancurvilinearcurvilinearcurvilinearcurvilinear
Connectivityh-nonconformingconformingconformingh-nonconformingconforming
Element typeline, square, cubeline, quadᵃ, hexᵃquadᵃquadᵃ, hexᵃsimplex, quadᵃ, hexᵃ
Adaptive mesh refinement✅❌❌✅❌AMRCallback
Solver typeDGSEMDGSEMDGSEMDGSEMDGMulti
Domainhypercubemapped hypercubearbitraryarbitraryarbitrary
Weak form✅✅✅✅✅VolumeIntegralWeakForm
Flux differencing✅✅✅✅✅VolumeIntegralFluxDifferencing
Shock capturing✅✅✅✅❌VolumeIntegralShockCapturingHG
Nonconservative equations✅✅✅✅✅e.g., GLM MHD or shallow water equations
Parabolic terms✅✅❌✅✅e.g., CompressibleNavierStokesDiffusion2D

ᵃ: quad = quadrilateral, hex = hexahedron

Time integration methods

Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem is a wrapper of Trixi.rhs!(du_ode, u_ode, semi, t), which gets called in ODE solvers. Further information can be found in the section on time integration methods.

Next steps

We explicitly encourage people interested in Trixi.jl to have a look at the examples/ bundled with Trixi.jl to get an impression of what is possible and the general look and feel of Trixi.jl. Before doing that, it is usually good to get an idea of how to visualize numerical results.

If you like learning by doing, looking at the tutorials and trying to mix your own elixirs based thereon is probably a good next step. Otherwise, you can further dig into the documentation by looking at Trixi.jl's basic building blocks.

diff --git a/previews/PR1624/parallelization/index.html b/previews/PR1624/parallelization/index.html index 519118a1dd9..168d250a478 100644 --- a/previews/PR1624/parallelization/index.html +++ b/previews/PR1624/parallelization/index.html @@ -1,5 +1,5 @@ -Parallelization · Trixi.jl
GitHub

Parallelization

Shared-memory parallelization with threads

Many compute-intensive loops in Trixi.jl are parallelized using the multi-threading support provided by Julia. You can recognize those loops by the @threaded macro prefixed to them, e.g.,

@threaded for element in eachelement(dg, cache)
+Parallelization · Trixi.jl
GitHub
Parallelization
Shared-memory parallelization with threads
Many compute-intensive loops in Trixi.jl are parallelized using the multi-threading support provided by Julia. You can recognize those loops by the @threaded macro prefixed to them, e.g.,
@threaded for element in eachelement(dg, cache)
   ...
 end
This will statically assign an equal iteration count to each available thread.
To use multi-threading, you need to tell Julia at startup how many threads you want to use by either setting the environment variable JULIA_NUM_THREADS or by providing the -t/--threads command line argument. For example, to start Julia with four threads, start Julia with
julia --threads=4
If both the environment variable and the command line argument are specified at the same time, the latter takes precedence.
If you use time integration methods from OrdinaryDiffEq.jl and want to use multiple threads therein, you need to set the keyword argument thread=OrdinaryDiffEq.True() of the algorithms, as described in the section on time integration methods.
Warning
Not everything is parallelized yet and there are likely opportunities to improve scalability. Multi-threading isn't considered part of the public API of Trixi.jl yet.
Distributed computing with MPI
In addition to the shared memory parallelization with multi-threading, Trixi.jl supports distributed parallelism via MPI.jl, which leverages the Message Passing Interface (MPI). MPI.jl comes with its own MPI library binaries such that there is no need to install MPI yourself. However, it is also possible to instead use an existing MPI installation, which is recommended if you are running MPI programs on a cluster or supercomputer (see the MPI.jl docs to find out how to select the employed MPI library). Additional notes on how to use a system-provided MPI installation with Trixi.jl can be found in the following subsection.
Work in progress
MPI-based parallelization is work in progress and not finished yet. Nothing related to MPI is part of the official API of Trixi.jl yet.
Using a system-provided MPI installation
When using Trixi.jl with a system-provided MPI backend, the underlying p4est, t8code and HDF5 libraries need to be compiled with the same MPI installation. If you want to use p4est (via the P4estMesh) or t8code (via the T8codeMesh) from Trixi.jl, you also need to use system-provided p4est or t8code installations (for notes on how to install p4est and t8code see, e.g., here and here, use the configure option --enable-mpi). Otherwise, there will be warnings that no preference is set for P4est.jl and T8code.jl that can be ignored if you do not use these libraries from Trixi.jl. Note that t8code already comes with a p4est installation, so it suffices to install t8code. In order to use system-provided p4est and t8code installations, P4est.jl and T8code.jl need to be configured to use the custom installations. Follow the steps described here and here for the configuration. The paths that point to libp4est.so (and potentially to libsc.so) need to be the same for P4est.jl and T8code.jl. This could, e.g., be libp4est.so that usually can be found in lib/ or local/lib/ in the installation directory of t8code. Note that the T8codeMesh, however, does not support MPI yet. The preferences for HDF5.jl always need to be set, even if you do not want to use HDF5 from Trixi.jl, see also issue #1079 in HDF5.jl. To set the preferences for HDF5.jl, follow the instructions described here.
In total, in your active Julia project you should have a LocalPreferences.toml file with sections [MPIPreferences], [T8code] (only needed if T8codeMesh is used), [P4est] (only needed if P4estMesh is used), and [HDF5] as well as an entry MPIPreferences in your Project.toml to use a custom MPI installation. A LocalPreferences.toml file  created as described above might look something like the following:
[HDF5]
 libhdf5 = "/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so"
@@ -44,4 +44,4 @@
            UUID("f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f"), # UUID of HDF5.jl
            "libhdf5" => "/path/to/your/libhdf5.so",
            "libhdf5_hl" => "/path/to/your/libhdf5_hl.so", force = true)
Alternatively, with HDF5.jl v0.17.1 or higher you can use
julia> using HDF5
-julia> HDF5.API.set_libraries!("/path/to/your/libhdf5.so", "/path/to/your/libhdf5_hl.so")
For more information see also the documentation of HDF5.jl. In total, you should have a file called LocalPreferences.toml in the project directory that contains a section [MPIPreferences], a section [HDF5] with entries libhdf5 and libhdf5_hl, a section [P4est] with the entry libp4est as well as a section [T8code] with the entries libt8, libp4est and libsc. If you use HDF5.jl v0.16 or older, instead of setting the preferences for HDF5.jl, you need to set the environment variable JULIA_HDF5_PATH to the path, where the HDF5 binaries are located and then call ]build HDF5 from Julia.
If HDF5 is not MPI-enabled, Trixi.jl will fall back on a less efficient I/O mechanism. In that case, all disk I/O is performed only on rank zero and data is distributed to/gathered from the other ranks using regular MPI communication.

+julia> HDF5.API.set_libraries!("/path/to/your/libhdf5.so", "/path/to/your/libhdf5_hl.so")

For more information see also the documentation of HDF5.jl. In total, you should have a file called LocalPreferences.toml in the project directory that contains a section [MPIPreferences], a section [HDF5] with entries libhdf5 and libhdf5_hl, a section [P4est] with the entry libp4est as well as a section [T8code] with the entries libt8, libp4est and libsc. If you use HDF5.jl v0.16 or older, instead of setting the preferences for HDF5.jl, you need to set the environment variable JULIA_HDF5_PATH to the path, where the HDF5 binaries are located and then call ]build HDF5 from Julia.

If HDF5 is not MPI-enabled, Trixi.jl will fall back on a less efficient I/O mechanism. In that case, all disk I/O is performed only on rank zero and data is distributed to/gathered from the other ranks using regular MPI communication.

diff --git a/previews/PR1624/performance/index.html b/previews/PR1624/performance/index.html index c118ce3cdc6..890b3467579 100644 --- a/previews/PR1624/performance/index.html +++ b/previews/PR1624/performance/index.html @@ -1,5 +1,5 @@ -Performance · Trixi.jl
GitHub

Performance

Trixi.jl is designed to balance performance and readability. Since Julia provides a lot of zero-cost abstractions, it is often possible to optimize both goals simultaneously.

The usual development workflow in Julia is

  1. Make it work.
  2. Make it nice.
  3. Make it fast.

To achieve the third step, you should be familiar with (at least) the section on performance tips in the Julia manual. Here, we just list some important aspects you should consider when developing Trixi.jl.

  • Consider using @views/view(...) when using array slices, except on the left-side of an assignment (further details).
  • Functions are essentially for free, since they are usually automatically inlined where it makes sense (using @inline can be used as an additional hint to the compiler) (further details).
  • Function barriers can improve performance due to type stability (further details).
  • Look for type instabilities using @code_warntype. Consider using @descend from Cthulhu.jl to investigate deeper call chains.

Manual benchmarking

If you modify some internal parts of Trixi.jl, you should check the impact on performance. Hence, you should at least investigate the performance roughly by comparing the reported timings of several elixirs. Deeper investigations and micro-benchmarks should usually use BenchmarkTools.jl. For example, the following steps were used to benchmark the changes introduced in PR #256.

  1. git checkout e7ebf3846b3fd62ee1d0042e130afb50d7fe8e48 (new version)

  2. Start julia --threads=1 --check-bounds=no.

  3. Execute the following code in the REPL to benchmark the rhs! call at the final state.

    julia> using BenchmarkTools, Revise; using Trixi
+Performance · Trixi.jl
    GitHub
    Performance
    Trixi.jl is designed to balance performance and readability. Since Julia provides a lot of zero-cost abstractions, it is often possible to optimize both goals simultaneously.
    The usual development workflow in Julia is
    1. Make it work.
    2. Make it nice.
    3. Make it fast.
    To achieve the third step, you should be familiar with (at least) the section on performance tips in the Julia manual. Here, we just list some important aspects you should consider when developing Trixi.jl.
    • Consider using @views/view(...) when using array slices, except on the left-side of an assignment (further details).
    • Functions are essentially for free, since they are usually automatically inlined where it makes sense (using @inline can be used as an additional hint to the compiler) (further details).
    • Function barriers can improve performance due to type stability (further details).
    • Look for type instabilities using @code_warntype. Consider using @descend from Cthulhu.jl to investigate deeper call chains.
    Manual benchmarking
    If you modify some internal parts of Trixi.jl, you should check the impact on performance. Hence, you should at least investigate the performance roughly by comparing the reported timings of several elixirs. Deeper investigations and micro-benchmarks should usually use BenchmarkTools.jl. For example, the following steps were used to benchmark the changes introduced in PR #256.
    1. git checkout e7ebf3846b3fd62ee1d0042e130afb50d7fe8e48 (new version)
    2. Start julia --threads=1 --check-bounds=no.
    3. Execute the following code in the REPL to benchmark the rhs! call at the final state.
      julia> using BenchmarkTools, Revise; using Trixi
 
 julia> trixi_include("examples/2d/elixir_euler_sedov_blast_wave.jl")
 
@@ -41,7 +41,7 @@
  maximum time:     5.880 ms (0.00% GC)
  --------------
  samples:          1070
- evals/sample:     1
      Run the @benchmark ... commands multiple times to see whether there are any significant fluctuations.
    Follow these steps for both commits you want to compare. The relevant benchmark results you should typically be looking at are the median and mean values of the runtime and the memory/allocs estimate. In this example, the differences of the runtimes are of the order of the fluctuations one gets when running the benchmarks multiple times. Since the memory/allocs are (roughly) the same, there doesn't seem to be a significant performance regression here.
    You can also make it more detailed by benchmarking only, e.g., the calculation of the volume terms, but whether that's necessary depends on the modifications you made and their (potential) impact.
    Some more detailed description of manual profiling and benchmarking as well as resulting performance improvements of Trixi.jl are given in the following blog posts.
    Automated benchmarking
    We use PkgBenchmark.jl to provide a standard set of benchmarks for Trixi.jl. The relevant benchmark script is benchmark/benchmarks.jl. You can run a standard set of benchmarks via
    julia> using PkgBenchmark, Trixi
+ evals/sample:     1
    Run the @benchmark ... commands multiple times to see whether there are any significant fluctuations.

Follow these steps for both commits you want to compare. The relevant benchmark results you should typically be looking at are the median and mean values of the runtime and the memory/allocs estimate. In this example, the differences of the runtimes are of the order of the fluctuations one gets when running the benchmarks multiple times. Since the memory/allocs are (roughly) the same, there doesn't seem to be a significant performance regression here.

You can also make it more detailed by benchmarking only, e.g., the calculation of the volume terms, but whether that's necessary depends on the modifications you made and their (potential) impact.

Some more detailed description of manual profiling and benchmarking as well as resulting performance improvements of Trixi.jl are given in the following blog posts.

Automated benchmarking

We use PkgBenchmark.jl to provide a standard set of benchmarks for Trixi.jl. The relevant benchmark script is benchmark/benchmarks.jl. To benchmark the changes made in a PR, please proceed as follows:

  1. Check out the latest main branch of your Trixi.jl development repository.
  2. Check out the latest development branch of your PR.
  3. Change your working directory to the benchmark directory of Trixi.jl.
  4. Execute julia run_benchmarks.jl.

This will take some hours to complete and requires at least 8 GiB of RAM. When everything is finished, some output files will be created in the benchmark directory of Trixi.jl.

Warning

Please note that the benchmark scripts use --check-bounds=no at the moment. Thus, they will not work in any useful way for Julia v1.10 (and newer?), see Julia issue #50985.

You can also run a standard set of benchmarks manually via

julia> using PkgBenchmark, Trixi
 
 julia> results = benchmarkpkg(Trixi, BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`))
 
@@ -52,4 +52,4 @@
              BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`, id="main") # baseline
        )
 
-julia> export_markdown(pkgdir(Trixi, "benchmark", "results.md"), results)

By default, the target is the current state of the repository. Remember that you need to be in a clean state (commit or stash your changes) to run this successfully. You can also run this comparison and an additional one using two threads via

julia> include("benchmark/run_benchmarks.jl")

Then, markdown files including the results are saved in benchmark/. This example result was obtained using a GitHub action for the PR #535. Note that GitHub actions run on in the cloud in a virtual machine. Hence, we do not really have control over it and performance results must be taken with a grain of salt. Nevertheless, significant runtime differences and differences of memory allocations should be robust indicators of performance changes.

Runtime performance vs. latency aka using @nospecialize selectively

Usually, Julia will compile specialized versions of each method, using as much information from the types of function arguments as possible (based on some heuristics). The compiler will generate code that is as efficient as comparable code written in a low-level language such as C or Fortran. However, there are cases where the runtime performance does not really matter but the time needed to compile specializations becomes significant. This is related to latency or the time-to-first-plot problem, well-known in the Julia community. In such a case, it can be useful to remove some burden from the compiler by avoiding specialization on every possible argument types using the macro @nospecialize. A prime example of such a case is pretty printing of structs in the Julia REPL, see the associated PR for further discussions.

As a rule of thumb:

  • Do not use @nospecialize in performance-critical parts, in particular not for methods involved in computing Trixi.rhs!.
  • Consider using @nospecialize for methods like custom implementations of Base.show.

Performance metrics of the AnalysisCallback

The AnalysisCallback computes two performance indicators that you can use to evaluate the serial and parallel performance of Trixi.jl. They represent measured run times that are normalized by the number of rhs! evaluations and the number of degrees of freedom of the problem setup. The normalization ensures that we can compare different measurements for each type of indicator independent of the number of time steps or mesh size. All indicators have in common that they are still in units of time, thus lower is better for each of them.

Here, the term "degrees of freedom" (DOFs) refers to the number of independent state vectors that are used to represent the numerical solution. For example, if you use a DGSEM-type scheme in 2D on a mesh with 8 elements and with 5-by-5 Gauss-Lobatto nodes in each element (i.e., a polynomial degree of 4), the total number of DOFs would be

\[n_\text{DOFs,DGSEM} = \{\text{number of elements}\} \cdot \{\text{number of nodes per element}\} = 8 \cdot (5 \cdot 5) = 200.\]

In contrast, for a finite volume-type scheme on a mesh with 8 elements, the total number of DOFs would be (independent of the number of spatial dimensions)

\[n_\text{DOFs,FV} = \{\text{number of elements}\} = 8,\]

since for standard finite volume methods you store a single state vector in each element. Note that we specifically count the number of state vectors and not the number of state variables for the DOFs. That is, in the previous example $n_\text{DOFs,FV}$ is equal to 8 independent of whether this is a compressible Euler setup with 5 state variables or a linear scalar advection setup with one state variable.

For each indicator, the measurements are always since the last invocation of the AnalysisCallback. That is, if the analysis callback is called multiple times, the indicators are repeatedly computed and can thus also be used to track the performance over the course of a longer simulation, e.g., to analyze setups with varying performance characteristics. Note that the time spent in the AnalysisCallback itself is always excluded, i.e., the performance measurements are not distorted by potentially expensive solution analysis computations. All other parts of a Trixi.jl simulation are included, however, thus make sure that you disable everything you do not want to be measured (such as I/O callbacks, visualization etc.).

Performance indicators and adaptive mesh refinement

Currently it is not possible to compute meaningful performance indicators for a simulation with arbitrary adaptive mesh refinement, since this would require to explicitly keep track of the number of DOF updates due to the mesh size changing repeatedly. The only way to do this at the moment is by setting the analysis interval to the same value as the AMR interval.

Local, rhs!-only indicator

The local, rhs!-only indicator is computed as

\[\text{time/DOF/rhs!} = \frac{t_\text{\texttt{rhs!}}}{n_\text{DOFs,local} \cdot n_\text{calls,\texttt{rhs!}}},\]

where $t_\text{\texttt{rhs!}}$ is the accumulated time spent in rhs!, $n_\text{DOFs,local}$ is the local number of DOFs (i.e., on the current MPI rank; if doing a serial run, you can just think of this as the number of DOFs), and $n_\text{calls,\texttt{rhs!}}$ is the number of times the rhs! function has been evaluated. Note that for this indicator, we measure only the time spent in rhs!, i.e., by definition all computations outside of rhs! - specifically all other callbacks and the time integration method - are not taken into account.

The local, rhs!-only indicator is usually most useful if you do serial measurements and are interested in the performance of the implementation of your core numerical methods (e.g., when doing performance tuning).

Performance index (PID)

The performance index (PID) is computed as

\[\text{PID} = \frac{t_\text{wall} \cdot n_\text{ranks,MPI}}{n_\text{DOFs,global} \cdot n_\text{calls,\texttt{rhs!}}},\]

where $t_\text{wall}$ is the walltime since the last call to the AnalysisCallback, $n_\text{ranks,MPI}$ is the number of MPI ranks used, $n_\text{DOFs,global}$ is the global number of DOFs (i.e., the sum of DOFs over all MPI ranks; if doing a serial run, you can just think of this as the number of DOFs), and $n_\text{calls,\texttt{rhs!}}$ is the number of times the rhs! function has been evaluated since the last call to the AnalysisCallback. The PID measures everything except the time spent in the AnalysisCallback itself - specifically, all other callbacks and the time integration method itself are included.

The PID is usually most useful if you would like to compare the parallel performance of your code to its serial performance. Specifically, it allows you to evaluate the parallelization overhead of your code by giving you a measure of the resources that are necessary to solve a given simulation setup. In a sense, it mimics the "core hours" metric that is often used by supercomputer centers to measure how many resources a particular compute job requires. It can thus be seen as a proxy for "energy used" and, as an extension, "monetary cost".

Initialization overhead in measurements

When using one of the integration schemes from OrdinaryDiffEq.jl, their implementation will initialize some OrdinaryDiffEq.jl-specific information during the first time step. Among other things, one additional call to rhs! is performed. Therefore, make sure that for performance measurements using the PID either the number of timesteps or the workload per rhs! call is large enough to make the initialization overhead negligible. Note that the extra call to rhs! is properly accounted for in both the number of calls and the measured time, so you do not need to worry about it being expensive. If you want a perfect timing result, you need to set the analysis interval such that the AnalysisCallback is invoked at least once during the course of the simulation and discard the first PID value.

+julia> export_markdown(pkgdir(Trixi, "benchmark", "results.md"), results)

By default, the target is the current state of the repository. Remember that you need to be in a clean state (commit or stash your changes) to run this successfully. You can also run this comparison and an additional one using two threads via

julia> include("benchmark/run_benchmarks.jl")

Then, markdown files including the results are saved in benchmark/. This example result was obtained using a GitHub action for the PR #535. Note that GitHub actions run on in the cloud in a virtual machine. Hence, we do not really have control over it and performance results must be taken with a grain of salt. Nevertheless, significant runtime differences and differences of memory allocations should be robust indicators of performance changes.

Runtime performance vs. latency aka using @nospecialize selectively

Usually, Julia will compile specialized versions of each method, using as much information from the types of function arguments as possible (based on some heuristics). The compiler will generate code that is as efficient as comparable code written in a low-level language such as C or Fortran. However, there are cases where the runtime performance does not really matter but the time needed to compile specializations becomes significant. This is related to latency or the time-to-first-plot problem, well-known in the Julia community. In such a case, it can be useful to remove some burden from the compiler by avoiding specialization on every possible argument types using the macro @nospecialize. A prime example of such a case is pretty printing of structs in the Julia REPL, see the associated PR for further discussions.

As a rule of thumb:

  • Do not use @nospecialize in performance-critical parts, in particular not for methods involved in computing Trixi.rhs!.
  • Consider using @nospecialize for methods like custom implementations of Base.show.

Performance metrics of the AnalysisCallback

The AnalysisCallback computes two performance indicators that you can use to evaluate the serial and parallel performance of Trixi.jl. They represent measured run times that are normalized by the number of rhs! evaluations and the number of degrees of freedom of the problem setup. The normalization ensures that we can compare different measurements for each type of indicator independent of the number of time steps or mesh size. All indicators have in common that they are still in units of time, thus lower is better for each of them.

Here, the term "degrees of freedom" (DOFs) refers to the number of independent state vectors that are used to represent the numerical solution. For example, if you use a DGSEM-type scheme in 2D on a mesh with 8 elements and with 5-by-5 Gauss-Lobatto nodes in each element (i.e., a polynomial degree of 4), the total number of DOFs would be

\[n_\text{DOFs,DGSEM} = \{\text{number of elements}\} \cdot \{\text{number of nodes per element}\} = 8 \cdot (5 \cdot 5) = 200.\]

In contrast, for a finite volume-type scheme on a mesh with 8 elements, the total number of DOFs would be (independent of the number of spatial dimensions)

\[n_\text{DOFs,FV} = \{\text{number of elements}\} = 8,\]

since for standard finite volume methods you store a single state vector in each element. Note that we specifically count the number of state vectors and not the number of state variables for the DOFs. That is, in the previous example $n_\text{DOFs,FV}$ is equal to 8 independent of whether this is a compressible Euler setup with 5 state variables or a linear scalar advection setup with one state variable.

For each indicator, the measurements are always since the last invocation of the AnalysisCallback. That is, if the analysis callback is called multiple times, the indicators are repeatedly computed and can thus also be used to track the performance over the course of a longer simulation, e.g., to analyze setups with varying performance characteristics. Note that the time spent in the AnalysisCallback itself is always excluded, i.e., the performance measurements are not distorted by potentially expensive solution analysis computations. All other parts of a Trixi.jl simulation are included, however, thus make sure that you disable everything you do not want to be measured (such as I/O callbacks, visualization etc.).

Performance indicators and adaptive mesh refinement

Currently it is not possible to compute meaningful performance indicators for a simulation with arbitrary adaptive mesh refinement, since this would require to explicitly keep track of the number of DOF updates due to the mesh size changing repeatedly. The only way to do this at the moment is by setting the analysis interval to the same value as the AMR interval.

Local, rhs!-only indicator

The local, rhs!-only indicator is computed as

\[\text{time/DOF/rhs!} = \frac{t_\text{\texttt{rhs!}}}{n_\text{DOFs,local} \cdot n_\text{calls,\texttt{rhs!}}},\]

where $t_\text{\texttt{rhs!}}$ is the accumulated time spent in rhs!, $n_\text{DOFs,local}$ is the local number of DOFs (i.e., on the current MPI rank; if doing a serial run, you can just think of this as the number of DOFs), and $n_\text{calls,\texttt{rhs!}}$ is the number of times the rhs! function has been evaluated. Note that for this indicator, we measure only the time spent in rhs!, i.e., by definition all computations outside of rhs! - specifically all other callbacks and the time integration method - are not taken into account.

The local, rhs!-only indicator is usually most useful if you do serial measurements and are interested in the performance of the implementation of your core numerical methods (e.g., when doing performance tuning).

Performance index (PID)

The performance index (PID) is computed as

\[\text{PID} = \frac{t_\text{wall} \cdot n_\text{ranks,MPI}}{n_\text{DOFs,global} \cdot n_\text{calls,\texttt{rhs!}}},\]

where $t_\text{wall}$ is the walltime since the last call to the AnalysisCallback, $n_\text{ranks,MPI}$ is the number of MPI ranks used, $n_\text{DOFs,global}$ is the global number of DOFs (i.e., the sum of DOFs over all MPI ranks; if doing a serial run, you can just think of this as the number of DOFs), and $n_\text{calls,\texttt{rhs!}}$ is the number of times the rhs! function has been evaluated since the last call to the AnalysisCallback. The PID measures everything except the time spent in the AnalysisCallback itself - specifically, all other callbacks and the time integration method itself are included.

The PID is usually most useful if you would like to compare the parallel performance of your code to its serial performance. Specifically, it allows you to evaluate the parallelization overhead of your code by giving you a measure of the resources that are necessary to solve a given simulation setup. In a sense, it mimics the "core hours" metric that is often used by supercomputer centers to measure how many resources a particular compute job requires. It can thus be seen as a proxy for "energy used" and, as an extension, "monetary cost".

Initialization overhead in measurements

When using one of the integration schemes from OrdinaryDiffEq.jl, their implementation will initialize some OrdinaryDiffEq.jl-specific information during the first time step. Among other things, one additional call to rhs! is performed. Therefore, make sure that for performance measurements using the PID either the number of timesteps or the workload per rhs! call is large enough to make the initialization overhead negligible. Note that the extra call to rhs! is properly accounted for in both the number of calls and the measured time, so you do not need to worry about it being expensive. If you want a perfect timing result, you need to set the analysis interval such that the AnalysisCallback is invoked at least once during the course of the simulation and discard the first PID value.

Performance issues with multi-threaded reductions

False sharing is a known performance issue for systems with distributed caches. It also occurred for the implementation of a thread parallel bounds checking routine for the subcell IDP limiting in PR #1736. After some testing and discussion, it turned out that initializing a vector of length n * Threads.nthreads() and only using every n-th entry instead of a vector of length Threads.nthreads() fixes the problem. Since there are no processors with caches over 128B, we use n = 128B / size(uEltype). Now, the bounds checking routine of the IDP limiting scales as hoped.

diff --git a/previews/PR1624/reference-trixi/index.html b/previews/PR1624/reference-trixi/index.html index 74cb6dada30..bf7863515a4 100644 --- a/previews/PR1624/reference-trixi/index.html +++ b/previews/PR1624/reference-trixi/index.html @@ -1,9 +1,9 @@ -Trixi.jl · Trixi.jl
GitHub

Trixi.jl API

Trixi.Trixi — Module
Trixi

Trixi.jl is a numerical simulation framework for hyperbolic conservation laws. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures.

To get started, run your first simulation with Trixi.jl using

trixi_include(default_example())

See also: trixi-framework/Trixi.jl

source
Trixi.flux_hll_chen_noelle — Constant
flux_hll_chen_noelle = FluxHLL(min_max_speed_chen_noelle)

An instance of FluxHLL specific to the shallow water equations that uses the wave speed estimates from min_max_speed_chen_noelle. This HLL flux is guaranteed to have zero numerical mass flux out of a "dry" element, maintain positivity of the water height, and satisfy an entropy inequality.

For complete details see Section 2.4 of the following reference

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI: 10.1137/15M1053074
source
Trixi.AMRCallback — Type
AMRCallback(semi, controller [,adaptor=AdaptorAMR(semi)];
+Trixi.jl · Trixi.jl
GitHub
Trixi.jl API
Trixi.Trixi — Module
Trixi
Trixi.jl is a numerical simulation framework for hyperbolic conservation laws. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures.
To get started, run your first simulation with Trixi.jl using
trixi_include(default_example())
See also: trixi-framework/Trixi.jl
source
Trixi.AMRCallback — Type
AMRCallback(semi, controller [,adaptor=AdaptorAMR(semi)];
             interval,
             adapt_initial_condition=true,
             adapt_initial_condition_only_refine=true,
-            dynamic_load_balancing=true)
Performs adaptive mesh refinement (AMR) every interval time steps for a given semidiscretization semi using the chosen controller.
source
Trixi.AbstractEquations — Type
AbstractEquations{NDIMS, NVARS}
An abstract supertype of specific equations such as the compressible Euler equations. The type parameters encode the number of spatial dimensions (NDIMS) and the number of primary variables (NVARS) of the physics model.
source
Trixi.AbstractMesh — Type
AbstractMesh{NDIMS}
An abstract supertype of specific mesh types such as TreeMesh or StructuredMesh. The type parameters encode the number of spatial dimensions (NDIMS).
source
Trixi.AcousticPerturbationEquations2D — Type
AcousticPerturbationEquations2D(v_mean_global, c_mean_global, rho_mean_global)
Acoustic perturbation equations (APE) in two space dimensions. The equations are given by
\[\begin{aligned}
+            dynamic_load_balancing=true)

Performs adaptive mesh refinement (AMR) every interval time steps for a given semidiscretization semi using the chosen controller.

source
Trixi.AbstractEquations — Type
AbstractEquations{NDIMS, NVARS}

An abstract supertype of specific equations such as the compressible Euler equations. The type parameters encode the number of spatial dimensions (NDIMS) and the number of primary variables (NVARS) of the physics model.

source
Trixi.AbstractMesh — Type
AbstractMesh{NDIMS}

An abstract supertype of specific mesh types such as TreeMesh or StructuredMesh. The type parameters encode the number of spatial dimensions (NDIMS).

source
Trixi.AcousticPerturbationEquations2D — Type
AcousticPerturbationEquations2D(v_mean_global, c_mean_global, rho_mean_global)

Acoustic perturbation equations (APE) in two space dimensions. The equations are given by

\[\begin{aligned} \frac{\partial\mathbf{v'}}{\partial t} + \nabla (\bar{\mathbf{v}}\cdot\mathbf{v'}) + \nabla\left( \frac{\bar{c}^2 \tilde{p}'}{\bar{\rho}} \right) &= 0 \\ \frac{\partial \tilde{p}'}{\partial t} + @@ -11,21 +11,21 @@ \end{aligned}\]

The bar $\bar{(\cdot)}$ indicates time-averaged quantities. The unknowns of the APE are the perturbed velocities $\mathbf{v'} = (v_1', v_2')^T$ and the scaled perturbed pressure $\tilde{p}' = \frac{p'}{\bar{c}^2}$, where $p'$ denotes the perturbed pressure and the perturbed variables are defined by $\phi' = \phi - \bar{\phi}$.

In addition to the unknowns, Trixi.jl currently stores the mean values in the state vector, i.e. the state vector used internally is given by

\[\mathbf{u} = \begin{pmatrix} v_1' \\ v_2' \\ \tilde{p}' \\ \bar{v}_1 \\ \bar{v}_2 \\ \bar{c} \\ \bar{\rho} - \end{pmatrix}.\]

This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the mean values must be zero.
  • The mean values have to be considered when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes these variables too.
  • Trixi.jl's visualization tools will visualize the mean values by default.

The constructor accepts a 2-tuple v_mean_global and scalars c_mean_global and rho_mean_global which can be used to make the definition of initial conditions for problems with constant mean flow more flexible. These values are ignored if the mean values are defined internally in an initial condition.

The equations are based on the APE-4 system introduced in the following paper:

source
Trixi.Adiabatic — Type
struct Adiabatic

Used to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_normal_flux_function should be a function with signature boundary_value_normal_flux_function(x, t, equations) and return a scalar value for the normal heat flux at point x and time t.

source
Trixi.AliveCallback — Type
AliveCallback(analysis_interval=0, alive_interval=analysis_interval÷10)

Inexpensive callback showing that a simulation is still running by printing some information such as the current time to the screen every alive_interval time steps. If analysis_interval ≂̸ 0, the output is omitted every analysis_interval time steps.

source
Trixi.AnalysisCallback — Type
AnalysisCallback(semi; interval=0,
+  \end{pmatrix}.\]
This affects the implementation and use of these equations in various ways:
  • The flux values corresponding to the mean values must be zero.
  • The mean values have to be considered when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes these variables too.
  • Trixi.jl's visualization tools will visualize the mean values by default.
The constructor accepts a 2-tuple v_mean_global and scalars c_mean_global and rho_mean_global which can be used to make the definition of initial conditions for problems with constant mean flow more flexible. These values are ignored if the mean values are defined internally in an initial condition.
The equations are based on the APE-4 system introduced in the following paper:
source
Trixi.Adiabatic — Type
struct Adiabatic

Used to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_normal_flux_function should be a function with signature boundary_value_normal_flux_function(x, t, equations) and return a scalar value for the normal heat flux at point x and time t.

source
Trixi.AliveCallback — Type
AliveCallback(analysis_interval=0, alive_interval=analysis_interval÷10)

Inexpensive callback showing that a simulation is still running by printing some information such as the current time to the screen every alive_interval time steps. If analysis_interval ≂̸ 0, the output is omitted every analysis_interval time steps.

source
Trixi.AnalysisCallback — Type
AnalysisCallback(semi; interval=0,
                        save_analysis=false,
                        output_directory="out",
                        analysis_filename="analysis.dat",
                        extra_analysis_errors=Symbol[],
-                       extra_analysis_integrals=())

Analyze a numerical solution every interval time steps and print the results to the screen. If save_analysis, the results are also saved in joinpath(output_directory, analysis_filename).

Additional errors can be computed, e.g. by passing extra_analysis_errors = (:l2_error_primitive, :linf_error_primitive) or extra_analysis_errors = (:conservation_error,).

If you want to omit the computation (to safe compute-time) of the default_analysis_errors, specify analysis_errors = Symbol[]. Note: default_analysis_errors are :l2_error and :linf_error for all equations. If you want to compute extra_analysis_errors such as :conservation_error solely, i.e., without :l2_error, :linf_error you need to specify analysis_errors = [:conservation_error] instead of extra_analysis_errors = [:conservation_error].

Further scalar functions func in extra_analysis_integrals are applied to the numerical solution and integrated over the computational domain. Some examples for this are entropy, energy_kinetic, energy_internal, and energy_total. You can also write your own function with the same signature as the examples listed above and pass it via extra_analysis_integrals. See the developer comments about Trixi.analyze, Trixi.pretty_form_utf, and Trixi.pretty_form_ascii for further information on how to create custom analysis quantities.

In addition, the analysis callback records and outputs a number of quantities that are useful for evaluating the computational performance, such as the total runtime, the performance index (time/DOF/rhs!), the time spent in garbage collection (GC), or the current memory usage (alloc'd memory).

source
Trixi.AnalysisCallbackCoupled — Type
AnalysisCallbackCoupled(semi, callbacks...)

Combine multiple analysis callbacks for coupled simulations with a SemidiscretizationCoupled. For each coupled system, an indididual AnalysisCallback must be created and passed to the AnalysisCallbackCoupled in order, i.e., in the same sequence as the indidvidual semidiscretizations are stored in the SemidiscretizationCoupled.

Experimental code

This is an experimental feature and can change any time.

source
Trixi.AveragingCallback — Type
AveragingCallback(semi::SemidiscretizationHyperbolic, tspan; output_directory="out",
-                  filename="averaging.h5")
Experimental code

This callback is experimental and may change in any future release.

A callback that averages the flow field described by semi which must be a semidiscretization of the compressible Euler equations in two dimensions. The callback records the mean velocity, mean speed of sound, mean density, and mean vorticity for each node over the time interval given by tspan and stores the results in an HDF5 file filename in the directory output_directory. Note that this callback does not support adaptive mesh refinement (AMRCallback).

source
Trixi.BoundaryConditionCoupled — Type
BoundaryConditionCoupled(other_semi_index, indices, uEltype)

Boundary condition to glue two meshes together. Solution values at the boundary of another mesh will be used as boundary values. This requires the use of SemidiscretizationCoupled. The other mesh is specified by other_semi_index, which is the index of the mesh in the tuple of semidiscretizations.

Note that the elements and nodes of the two meshes at the coupled boundary must coincide. This is currently only implemented for StructuredMesh.

Arguments

  • other_semi_index: the index in SemidiscretizationCoupled of the semidiscretization from which the values are copied
  • indices::Tuple: node/cell indices at the boundary of the mesh in the other semidiscretization. See examples below.
  • uEltype::Type: element type of solution

Examples

# Connect the left boundary of mesh 2 to our boundary such that our positive
+                       extra_analysis_integrals=())

Analyze a numerical solution every interval time steps and print the results to the screen. If save_analysis, the results are also saved in joinpath(output_directory, analysis_filename).

Additional errors can be computed, e.g. by passing extra_analysis_errors = (:l2_error_primitive, :linf_error_primitive) or extra_analysis_errors = (:conservation_error,).

If you want to omit the computation (to safe compute-time) of the default_analysis_errors, specify analysis_errors = Symbol[]. Note: default_analysis_errors are :l2_error and :linf_error for all equations. If you want to compute extra_analysis_errors such as :conservation_error solely, i.e., without :l2_error, :linf_error you need to specify analysis_errors = [:conservation_error] instead of extra_analysis_errors = [:conservation_error].

Further scalar functions func in extra_analysis_integrals are applied to the numerical solution and integrated over the computational domain. Some examples for this are entropy, energy_kinetic, energy_internal, and energy_total. You can also write your own function with the same signature as the examples listed above and pass it via extra_analysis_integrals. See the developer comments about Trixi.analyze, Trixi.pretty_form_utf, and Trixi.pretty_form_ascii for further information on how to create custom analysis quantities.

In addition, the analysis callback records and outputs a number of quantities that are useful for evaluating the computational performance, such as the total runtime, the performance index (time/DOF/rhs!), the time spent in garbage collection (GC), or the current memory usage (alloc'd memory).

source
Trixi.AnalysisCallbackCoupled — Type
AnalysisCallbackCoupled(semi, callbacks...)

Combine multiple analysis callbacks for coupled simulations with a SemidiscretizationCoupled. For each coupled system, an indididual AnalysisCallback must be created and passed to the AnalysisCallbackCoupled in order, i.e., in the same sequence as the indidvidual semidiscretizations are stored in the SemidiscretizationCoupled.

Experimental code

This is an experimental feature and can change any time.

source
Trixi.AveragingCallback — Type
AveragingCallback(semi::SemidiscretizationHyperbolic, tspan; output_directory="out",
+                  filename="averaging.h5")
Experimental code

This callback is experimental and may change in any future release.

A callback that averages the flow field described by semi which must be a semidiscretization of the compressible Euler equations in two dimensions. The callback records the mean velocity, mean speed of sound, mean density, and mean vorticity for each node over the time interval given by tspan and stores the results in an HDF5 file filename in the directory output_directory. Note that this callback does not support adaptive mesh refinement (AMRCallback).

source
Trixi.BoundaryConditionCoupled — Type
BoundaryConditionCoupled(other_semi_index, indices, uEltype, coupling_converter)

Boundary condition to glue two meshes together. Solution values at the boundary of another mesh will be used as boundary values. This requires the use of SemidiscretizationCoupled. The other mesh is specified by other_semi_index, which is the index of the mesh in the tuple of semidiscretizations.

Note that the elements and nodes of the two meshes at the coupled boundary must coincide. This is currently only implemented for StructuredMesh.

Arguments

  • other_semi_index: the index in SemidiscretizationCoupled of the semidiscretization from which the values are copied
  • indices::Tuple: node/cell indices at the boundary of the mesh in the other semidiscretization. See examples below.
  • uEltype::Type: element type of solution
  • coupling_converter::CouplingConverter: function to call for converting the solution state of one system to the other system

Examples

# Connect the left boundary of mesh 2 to our boundary such that our positive
 # boundary direction will match the positive y direction of the other boundary
-BoundaryConditionCoupled(2, (:begin, :i), Float64)
+BoundaryConditionCoupled(2, (:begin, :i), Float64, fun)
 
 # Connect the same two boundaries oppositely oriented
-BoundaryConditionCoupled(2, (:begin, :i_backwards), Float64)
+BoundaryConditionCoupled(2, (:begin, :i_backwards), Float64, fun)
 
 # Using this as y_neg boundary will connect `our_cells[i, 1, j]` to `other_cells[j, end-i, end]`
-BoundaryConditionCoupled(2, (:j, :i_backwards, :end), Float64)
Experimental code

This is an experimental feature and can change any time.

source
Trixi.BoundaryConditionDirichlet — Type
BoundaryConditionDirichlet(boundary_value_function)

Create a Dirichlet boundary condition that uses the function boundary_value_function to specify the values at the boundary. This can be used to create a boundary condition that specifies exact boundary values by passing the exact solution of the equation. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as

boundary_value_function(x, t, equations)

where x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.

Examples

julia> BoundaryConditionDirichlet(initial_condition_convergence_test)
source
Trixi.BoundaryConditionNavierStokesWall — Type
struct BoundaryConditionNavierStokesWall

Creates a wall-type boundary conditions for the compressible Navier-Stokes equations. The fields boundary_condition_velocity and boundary_condition_heat_flux are intended to be boundary condition types such as the NoSlip velocity boundary condition and the Adiabatic or Isothermal heat boundary condition.

source
Trixi.BoundaryConditionNeumann — Type
BoundaryConditionNeumann(boundary_normal_flux_function)

Similar to BoundaryConditionDirichlet, but creates a Neumann boundary condition for parabolic equations that uses the function boundary_normal_flux_function to specify the values of the normal flux at the boundary. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as

boundary_normal_flux_function(x, t, equations)

where x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.

source
Trixi.BoundsCheckCallback — Type
BoundsCheckCallback(; output_directory="out", save_errors=false, interval=1)

Subcell limiting techniques with SubcellLimiterIDP are constructed to adhere certain local or global bounds. To make sure that these bounds are actually met, this callback calculates the maximum deviation from the bounds. The maximum deviation per applied bound is printed to the screen at the end of the simulation. For more insights, when setting save_errors=true the occurring errors are exported every interval time steps during the simulation. Then, the maximum deviations since the last export are saved in "output_directory/deviations.txt". The BoundsCheckCallback has to be applied as a stage callback for the SSPRK time integration scheme.

Note

For SubcellLimiterIDP, the solution is corrected in the a posteriori correction stage SubcellLimiterIDPCorrection. So, to check the final solution, this bounds check callback must be called after the correction stage.

source
Trixi.CarpenterKennedy2N54 — Type
CarpenterKennedy2N54()

The following structures and methods provide a minimal implementation of the low-storage explicit Runge-Kutta method of

Carpenter, Kennedy (1994) Fourth order 2N storage RK schemes, Solution 3

using the same interface as OrdinaryDiffEq.jl.

source
Trixi.CompressibleEulerEquations1D — Type
CompressibleEulerEquations1D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} +BoundaryConditionCoupled(2, (:j, :i_backwards, :end), Float64, fun)

Experimental code

This is an experimental feature and can change any time.

source
Trixi.BoundaryConditionDirichlet — Type
BoundaryConditionDirichlet(boundary_value_function)

Create a Dirichlet boundary condition that uses the function boundary_value_function to specify the values at the boundary. This can be used to create a boundary condition that specifies exact boundary values by passing the exact solution of the equation. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as

boundary_value_function(x, t, equations)

where x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.

Examples

julia> BoundaryConditionDirichlet(initial_condition_convergence_test)
source
Trixi.BoundaryConditionNavierStokesWall — Type
struct BoundaryConditionNavierStokesWall

Creates a wall-type boundary conditions for the compressible Navier-Stokes equations. The fields boundary_condition_velocity and boundary_condition_heat_flux are intended to be boundary condition types such as the NoSlip velocity boundary condition and the Adiabatic or Isothermal heat boundary condition.

source
Trixi.BoundaryConditionNeumann — Type
BoundaryConditionNeumann(boundary_normal_flux_function)

Similar to BoundaryConditionDirichlet, but creates a Neumann boundary condition for parabolic equations that uses the function boundary_normal_flux_function to specify the values of the normal flux at the boundary. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as

boundary_normal_flux_function(x, t, equations)

where x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.

source
Trixi.BoundsCheckCallback — Type
BoundsCheckCallback(; output_directory="out", save_errors=false, interval=1)

Subcell limiting techniques with SubcellLimiterIDP are constructed to adhere certain local or global bounds. To make sure that these bounds are actually met, this callback calculates the maximum deviation from the bounds. The maximum deviation per applied bound is printed to the screen at the end of the simulation. For more insights, when setting save_errors=true the occurring errors are exported every interval time steps during the simulation. Then, the maximum deviations since the last export are saved in "output_directory/deviations.txt". The BoundsCheckCallback has to be applied as a stage callback for the SSPRK time integration scheme.

Note

For SubcellLimiterIDP, the solution is corrected in the a posteriori correction stage SubcellLimiterIDPCorrection. So, to check the final solution, this bounds check callback must be called after the correction stage.

source
Trixi.CarpenterKennedy2N54 — Type
CarpenterKennedy2N54()

The following structures and methods provide a minimal implementation of the low-storage explicit Runge-Kutta method of

Carpenter, Kennedy (1994) Fourth order 2N storage RK schemes, Solution 3

using the same interface as OrdinaryDiffEq.jl.

source
Trixi.CompressibleEulerEquations1D — Type
CompressibleEulerEquations1D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho v_1 \\ \rho e \end{pmatrix} @@ -37,7 +37,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 -\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in one space dimension. Here, $\rho$ is the density, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure.

source
Trixi.CompressibleEulerEquations2D — Type
CompressibleEulerEquations2D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} +\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in one space dimension. Here, $\rho$ is the density, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure.

source
Trixi.CompressibleEulerEquations2D — Type
CompressibleEulerEquations2D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho v_1 \\ \rho v_2 \\ \rho e \end{pmatrix} @@ -54,7 +54,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 -\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in two space dimensions. Here, $\rho$ is the density, $v_1$, $v_2$ the velocities, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2) \right)\]

the pressure.

source
Trixi.CompressibleEulerEquations3D — Type
CompressibleEulerEquations3D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} +\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in two space dimensions. Here, $\rho$ is the density, $v_1$, $v_2$ the velocities, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2) \right)\]

the pressure.

source
Trixi.CompressibleEulerEquations3D — Type
CompressibleEulerEquations3D(gamma)

The compressible Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho v_1 \\ \rho v_2 \\ \rho v_3 \\ \rho e \end{pmatrix} @@ -76,7 +76,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 \\ 0 -\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in three space dimensions. Here, $\rho$ is the density, $v_1$, $v_2$, $v_3$ the velocities, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2+v_3^2) \right)\]

the pressure.

source
Trixi.CompressibleEulerEquationsQuasi1D — Type
CompressibleEulerEquationsQuasi1D(gamma)

The quasi-1d compressible Euler equations (see Chan et al. DOI: 10.48550/arXiv.2307.12089 for details)

\[\frac{\partial}{\partial t} +\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in three space dimensions. Here, $\rho$ is the density, $v_1$, $v_2$, $v_3$ the velocities, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2+v_3^2) \right)\]

the pressure.

source
Trixi.CompressibleEulerEquationsQuasi1D — Type
CompressibleEulerEquationsQuasi1D(gamma)

The quasi-1d compressible Euler equations (see Chan et al. DOI: 10.48550/arXiv.2307.12089 for details)

\[\frac{\partial}{\partial t} \begin{pmatrix} a \rho \\ a \rho v_1 \\ a e \end{pmatrix} @@ -93,7 +93,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 -\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in one space dimension. Here, $\rho$ is the density, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, $a$ the (possibly) variable nozzle width, and

\[p = (\gamma - 1) \left( e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure.

The nozzle width function $a(x)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the compressible Euler equations one can set the nozzle width variable $a$ to one.

In addition to the unknowns, Trixi.jl currently stores the nozzle width values at the approximation points despite being fixed in time. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the nozzle width must be zero.
  • The nozzle width values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the nozzle width by default.
source
Trixi.CompressibleEulerMulticomponentEquations1D — Type
CompressibleEulerMulticomponentEquations1D(; gammas, gas_constants)

Multicomponent version of the compressible Euler equations

\[\frac{\partial}{\partial t} +\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in one space dimension. Here, $\rho$ is the density, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, $a$ the (possibly) variable nozzle width, and

\[p = (\gamma - 1) \left( e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure.

The nozzle width function $a(x)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the compressible Euler equations one can set the nozzle width variable $a$ to one.

In addition to the unknowns, Trixi.jl currently stores the nozzle width values at the approximation points despite being fixed in time. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the nozzle width must be zero.
  • The nozzle width values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the nozzle width by default.
source
Trixi.CompressibleEulerMulticomponentEquations1D — Type
CompressibleEulerMulticomponentEquations1D(; gammas, gas_constants)

Multicomponent version of the compressible Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho v_1 \\ \rho e \\ \rho_1 \\ \rho_2 \\ \vdots \\ \rho_{n} \end{pmatrix} @@ -106,7 +106,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 \\ \vdots \\ 0 -\end{pmatrix}\]

for calorically perfect gas in one space dimension. Here, $\rho_i$ is the density of component $i$, $\rho=\sum_{i=1}^n\rho_i$ the sum of the individual $\rho_i$, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure,

\[\gamma=\frac{\sum_{i=1}^n\rho_i C_{v,i}\gamma_i}{\sum_{i=1}^n\rho_i C_{v,i}}\]

total heat capacity ratio, $\gamma_i$ heat capacity ratio of component $i$,

\[C_{v,i}=\frac{R}{\gamma_i-1}\]

specific heat capacity at constant volume of component $i$.

In case of more than one component, the specific heat ratios gammas and the gas constants gas_constants should be passed as tuples, e.g., gammas=(1.4, 1.667).

The remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.

source
Trixi.CompressibleEulerMulticomponentEquations2D — Type
CompressibleEulerMulticomponentEquations2D(; gammas, gas_constants)

Multicomponent version of the compressible Euler equations

\[\frac{\partial}{\partial t} +\end{pmatrix}\]

for calorically perfect gas in one space dimension. Here, $\rho_i$ is the density of component $i$, $\rho=\sum_{i=1}^n\rho_i$ the sum of the individual $\rho_i$, $v_1$ the velocity, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v_1^2 \right)\]

the pressure,

\[\gamma=\frac{\sum_{i=1}^n\rho_i C_{v,i}\gamma_i}{\sum_{i=1}^n\rho_i C_{v,i}}\]

total heat capacity ratio, $\gamma_i$ heat capacity ratio of component $i$,

\[C_{v,i}=\frac{R}{\gamma_i-1}\]

specific heat capacity at constant volume of component $i$.

In case of more than one component, the specific heat ratios gammas and the gas constants gas_constants should be passed as tuples, e.g., gammas=(1.4, 1.667).

The remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.

source
Trixi.CompressibleEulerMulticomponentEquations2D — Type
CompressibleEulerMulticomponentEquations2D(; gammas, gas_constants)

Multicomponent version of the compressible Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho v_1 \\ \rho v_2 \\ \rho e \\ \rho_1 \\ \rho_2 \\ \vdots \\ \rho_{n} \end{pmatrix} @@ -123,7 +123,7 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 \\ 0 \\ \vdots \\ 0 -\end{pmatrix}\]

for calorically perfect gas in two space dimensions. Here, $\rho_i$ is the density of component $i$, $\rho=\sum_{i=1}^n\rho_i$ the sum of the individual $\rho_i$, $v_1$, $v_2$ the velocities, $e$ the specific total energy rather than specific internal energy, and

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2 + v_2^2) \right)\]

the pressure,

\[\gamma=\frac{\sum_{i=1}^n\rho_i C_{v,i}\gamma_i}{\sum_{i=1}^n\rho_i C_{v,i}}\]

total heat capacity ratio, $\gamma_i$ heat capacity ratio of component $i$,

\[C_{v,i}=\frac{R}{\gamma_i-1}\]

specific heat capacity at constant volume of component $i$.

In case of more than one component, the specific heat ratios gammas and the gas constants gas_constants in [kJ/(kg*K)] should be passed as tuples, e.g., gammas=(1.4, 1.667).

The remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.

source
Trixi.CompressibleNavierStokesDiffusion1D — Type
CompressibleNavierStokesDiffusion1D(equations; mu, Pr,
+\end{pmatrix}\]
for calorically perfect gas in two space dimensions. Here, $\rho_i$ is the density of component $i$, $\rho=\sum_{i=1}^n\rho_i$ the sum of the individual $\rho_i$, $v_1$, $v_2$ the velocities, $e$ the specific total energy rather than specific internal energy, and
\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2 + v_2^2) \right)\]
the pressure,
\[\gamma=\frac{\sum_{i=1}^n\rho_i C_{v,i}\gamma_i}{\sum_{i=1}^n\rho_i C_{v,i}}\]
total heat capacity ratio, $\gamma_i$ heat capacity ratio of component $i$,
\[C_{v,i}=\frac{R}{\gamma_i-1}\]
specific heat capacity at constant volume of component $i$.
In case of more than one component, the specific heat ratios gammas and the gas constants gas_constants in [kJ/(kg*K)] should be passed as tuples, e.g., gammas=(1.4, 1.667).
The remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.
source
Trixi.CompressibleNavierStokesDiffusion1D — Type
CompressibleNavierStokesDiffusion1D(equations; mu, Pr,
                                     gradient_variables=GradientVariablesPrimitive())

Contains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations1D.

  • equations: instance of the CompressibleEulerEquations1D
  • mu: dynamic viscosity,
  • Pr: Prandtl number,
  • gradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().

Fluid properties such as the dynamic viscosity $\mu$ can be provided in any consistent unit system, e.g., [$\mu$] = kg m⁻¹ s⁻¹.

The particular form of the compressible Navier-Stokes implemented is

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho v \\ \rho e @@ -137,7 +137,7 @@ \frac{\partial}{\partial x} \begin{pmatrix} 0 \\ \tau \\ \tau v - q -\end{pmatrix}\]

where the system is closed with the ideal gas assumption giving

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v^2 \right)\]

as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations1D. The terms on the right hand side of the system above are built from the viscous stress

\[\tau = \mu \frac{\partial}{\partial x} v\]

where the heat flux is

\[q = -\kappa \frac{\partial}{\partial x} \left(T\right),\quad T = \frac{p}{R\rho}\]

where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is

\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]

From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes

\[q = -\kappa \frac{\partial}{\partial x} \left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}} \frac{\partial}{\partial x} \left(\frac{p}{\rho}\right)\]

which is the form implemented below in the flux function.

In one spatial dimensions we require gradients for two quantities, e.g., primitive quantities

\[\frac{\partial}{\partial x} v,\, \frac{\partial}{\partial x} T\]

or the entropy variables

\[\frac{\partial}{\partial x} w_2,\, \frac{\partial}{\partial x} w_3\]

where

\[w_2 = \frac{\rho v1}{p},\, w_3 = -\frac{\rho}{p}\]

source
Trixi.CompressibleNavierStokesDiffusion2D — Type
CompressibleNavierStokesDiffusion2D(equations; mu, Pr,
+\end{pmatrix}\]
where the system is closed with the ideal gas assumption giving
\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho v^2 \right)\]
as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations1D. The terms on the right hand side of the system above are built from the viscous stress
\[\tau = \mu \frac{\partial}{\partial x} v\]
where the heat flux is
\[q = -\kappa \frac{\partial}{\partial x} \left(T\right),\quad T = \frac{p}{R\rho}\]
where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is
\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]
From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes
\[q = -\kappa \frac{\partial}{\partial x} \left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}} \frac{\partial}{\partial x} \left(\frac{p}{\rho}\right)\]
which is the form implemented below in the flux function.
In one spatial dimensions we require gradients for two quantities, e.g., primitive quantities
\[\frac{\partial}{\partial x} v,\, \frac{\partial}{\partial x} T\]
or the entropy variables
\[\frac{\partial}{\partial x} w_2,\, \frac{\partial}{\partial x} w_3\]
where
\[w_2 = \frac{\rho v1}{p},\, w_3 = -\frac{\rho}{p}\]
source
Trixi.CompressibleNavierStokesDiffusion2D — Type
CompressibleNavierStokesDiffusion2D(equations; mu, Pr,
                                     gradient_variables=GradientVariablesPrimitive())

Contains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations2D.

  • equations: instance of the CompressibleEulerEquations2D
  • mu: dynamic viscosity,
  • Pr: Prandtl number,
  • gradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().

Fluid properties such as the dynamic viscosity $\mu$ can be provided in any consistent unit system, e.g., [$\mu$] = kg m⁻¹ s⁻¹.

The particular form of the compressible Navier-Stokes implemented is

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho \mathbf{v} \\ \rho e @@ -151,7 +151,7 @@ \nabla \cdot \begin{pmatrix} 0 \\ \underline{\tau} \\ \underline{\tau}\mathbf{v} - \mathbf{q} -\end{pmatrix}\]

where the system is closed with the ideal gas assumption giving

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2) \right)\]

as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor

\[\underline{\tau} = \mu \left(\nabla\mathbf{v} + \left(\nabla\mathbf{v}\right)^T\right) - \frac{2}{3} \mu \left(\nabla\cdot\mathbf{v}\right)\underline{I}\]

where $\underline{I}$ is the $2\times 2$ identity matrix and the heat flux is

\[\mathbf{q} = -\kappa\nabla\left(T\right),\quad T = \frac{p}{R\rho}\]

where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is

\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]

From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes

\[\mathbf{q} = -\kappa\nabla\left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}}\nabla\left(\frac{p}{\rho}\right)\]

which is the form implemented below in the flux function.

In two spatial dimensions we require gradients for three quantities, e.g., primitive quantities

\[\nabla v_1,\, \nabla v_2,\, \nabla T\]

or the entropy variables

\[\nabla w_2,\, \nabla w_3,\, \nabla w_4\]

where

\[w_2 = \frac{\rho v_1}{p},\, w_3 = \frac{\rho v_2}{p},\, w_4 = -\frac{\rho}{p}\]

source
Trixi.CompressibleNavierStokesDiffusion3D — Type
CompressibleNavierStokesDiffusion3D(equations; mu, Pr,
+\end{pmatrix}\]
where the system is closed with the ideal gas assumption giving
\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2) \right)\]
as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor
\[\underline{\tau} = \mu \left(\nabla\mathbf{v} + \left(\nabla\mathbf{v}\right)^T\right) - \frac{2}{3} \mu \left(\nabla\cdot\mathbf{v}\right)\underline{I}\]
where $\underline{I}$ is the $2\times 2$ identity matrix and the heat flux is
\[\mathbf{q} = -\kappa\nabla\left(T\right),\quad T = \frac{p}{R\rho}\]
where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is
\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]
From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes
\[\mathbf{q} = -\kappa\nabla\left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}}\nabla\left(\frac{p}{\rho}\right)\]
which is the form implemented below in the flux function.
In two spatial dimensions we require gradients for three quantities, e.g., primitive quantities
\[\nabla v_1,\, \nabla v_2,\, \nabla T\]
or the entropy variables
\[\nabla w_2,\, \nabla w_3,\, \nabla w_4\]
where
\[w_2 = \frac{\rho v_1}{p},\, w_3 = \frac{\rho v_2}{p},\, w_4 = -\frac{\rho}{p}\]
source
Trixi.CompressibleNavierStokesDiffusion3D — Type
CompressibleNavierStokesDiffusion3D(equations; mu, Pr,
                                     gradient_variables=GradientVariablesPrimitive())

Contains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations3D.

  • equations: instance of the CompressibleEulerEquations3D
  • mu: dynamic viscosity,
  • Pr: Prandtl number,
  • gradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().

Fluid properties such as the dynamic viscosity $\mu$ can be provided in any consistent unit system, e.g., [$\mu$] = kg m⁻¹ s⁻¹.

The particular form of the compressible Navier-Stokes implemented is

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho \mathbf{v} \\ \rho e @@ -165,40 +165,41 @@ \nabla \cdot \begin{pmatrix} 0 \\ \underline{\tau} \\ \underline{\tau}\mathbf{v} - \mathbf{q} -\end{pmatrix}\]

where the system is closed with the ideal gas assumption giving

\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2+v_3^2) \right)\]

as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor

\[\underline{\tau} = \mu \left(\nabla\mathbf{v} + \left(\nabla\mathbf{v}\right)^T\right) - \frac{2}{3} \mu \left(\nabla\cdot\mathbf{v}\right)\underline{I}\]

where $\underline{I}$ is the $3\times 3$ identity matrix and the heat flux is

\[\mathbf{q} = -\kappa\nabla\left(T\right),\quad T = \frac{p}{R\rho}\]

where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is

\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]

From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes

\[\mathbf{q} = -\kappa\nabla\left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}}\nabla\left(\frac{p}{\rho}\right)\]

which is the form implemented below in the flux function.

In two spatial dimensions we require gradients for three quantities, e.g., primitive quantities

\[\nabla v_1,\, \nabla v_2,\, \nabla v_3,\, \nabla T\]

or the entropy variables

\[\nabla w_2,\, \nabla w_3,\, \nabla w_4\, \nabla w_5\]

where

\[w_2 = \frac{\rho v_1}{p},\, w_3 = \frac{\rho v_2}{p},\, w_4 = \frac{\rho v_3}{p},\, w_5 = -\frac{\rho}{p}\]

source
Trixi.ControllerThreeLevel — Type
ControllerThreeLevel(semi, indicator; base_level=1,
+\end{pmatrix}\]
where the system is closed with the ideal gas assumption giving
\[p = (\gamma - 1) \left( \rho e - \frac{1}{2} \rho (v_1^2+v_2^2+v_3^2) \right)\]
as the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor
\[\underline{\tau} = \mu \left(\nabla\mathbf{v} + \left(\nabla\mathbf{v}\right)^T\right) - \frac{2}{3} \mu \left(\nabla\cdot\mathbf{v}\right)\underline{I}\]
where $\underline{I}$ is the $3\times 3$ identity matrix and the heat flux is
\[\mathbf{q} = -\kappa\nabla\left(T\right),\quad T = \frac{p}{R\rho}\]
where $T$ is the temperature and $\kappa$ is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is
\[\kappa = \frac{\gamma \mu R}{(\gamma - 1)\textrm{Pr}}.\]
From this combination of temperature $T$ and thermal conductivity $\kappa$ we see that the gas constant R cancels and the heat flux becomes
\[\mathbf{q} = -\kappa\nabla\left(T\right) = -\frac{\gamma \mu}{(\gamma - 1)\textrm{Pr}}\nabla\left(\frac{p}{\rho}\right)\]
which is the form implemented below in the flux function.
In two spatial dimensions we require gradients for three quantities, e.g., primitive quantities
\[\nabla v_1,\, \nabla v_2,\, \nabla v_3,\, \nabla T\]
or the entropy variables
\[\nabla w_2,\, \nabla w_3,\, \nabla w_4\, \nabla w_5\]
where
\[w_2 = \frac{\rho v_1}{p},\, w_3 = \frac{\rho v_2}{p},\, w_4 = \frac{\rho v_3}{p},\, w_5 = -\frac{\rho}{p}\]
source
Trixi.ControllerThreeLevel — Type
ControllerThreeLevel(semi, indicator; base_level=1,
                                       med_level=base_level, med_threshold=0.0,
-                                      max_level=base_level, max_threshold=1.0)

An AMR controller based on three levels (in descending order of precedence):

  • set the target level to max_level if indicator > max_threshold
  • set the target level to med_level if indicator > med_threshold; if med_level < 0, set the target level to the current level
  • set the target level to base_level otherwise
source
Trixi.ControllerThreeLevelCombined — Type
ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;
+                                      max_level=base_level, max_threshold=1.0)

An AMR controller based on three levels (in descending order of precedence):

  • set the target level to max_level if indicator > max_threshold
  • set the target level to med_level if indicator > med_threshold; if med_level < 0, set the target level to the current level
  • set the target level to base_level otherwise
source
Trixi.ControllerThreeLevelCombined — Type
ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;
                              base_level=1,
                              med_level=base_level, med_threshold=0.0,
                              max_level=base_level, max_threshold=1.0,
-                             max_threshold_secondary=1.0)

An AMR controller based on three levels (in descending order of precedence):

  • set the target level to max_level if indicator_primary > max_threshold
  • set the target level to med_level if indicator_primary > med_threshold; if med_level < 0, set the target level to the current level
  • set the target level to base_level otherwise

If indicator_secondary >= max_threshold_secondary, set the target level to max_level.

source
Trixi.DGMulti — Method
DGMulti(approximation_type::AbstractDerivativeOperator;
+                             max_threshold_secondary=1.0)

An AMR controller based on three levels (in descending order of precedence):

  • set the target level to max_level if indicator_primary > max_threshold
  • set the target level to med_level if indicator_primary > med_threshold; if med_level < 0, set the target level to the current level
  • set the target level to base_level otherwise

If indicator_secondary >= max_threshold_secondary, set the target level to max_level.

source
Trixi.DGMulti — Method
DGMulti(approximation_type::AbstractDerivativeOperator;
         element_type::AbstractElemShape,
         surface_flux=flux_central,
         surface_integral=SurfaceIntegralWeakForm(surface_flux),
         volume_integral=VolumeIntegralWeakForm(),
-        kwargs...)

Create a summation by parts (SBP) discretization on the given element_type using a tensor product structure based on the 1D SBP derivative operator passed as approximation_type.

For more info, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.

source
Trixi.DGMulti — Method
DGMulti(; polydeg::Integer,
+        kwargs...)

Create a summation by parts (SBP) discretization on the given element_type using a tensor product structure based on the 1D SBP derivative operator passed as approximation_type.

For more info, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.

source
Trixi.DGMulti — Method
DGMulti(; polydeg::Integer,
           element_type::AbstractElemShape,
           approximation_type=Polynomial(),
           surface_flux=flux_central,
           surface_integral=SurfaceIntegralWeakForm(surface_flux),
           volume_integral=VolumeIntegralWeakForm(),
-          RefElemData_kwargs...)

Create a discontinuous Galerkin method which uses

  • approximations of polynomial degree polydeg
  • element type element_type (Tri(), Quad(), Tet(), and Hex() currently supported)

Optional:

  • approximation_type (default is Polynomial(); SBP() also supported for Tri(), Quad(), and Hex() element types).
  • RefElemData_kwargs are additional keyword arguments for RefElemData, such as quad_rule_vol. For more info, see the StartUpDG.jl docs.
source
Trixi.DGMultiMesh — Type
DGMultiMesh{NDIMS, ...}

DGMultiMesh describes a mesh type which wraps StartUpDG.MeshData and boundary_faces in a dispatchable type. This is intended to store geometric data and connectivities for any type of mesh (Cartesian, affine, curved, structured/unstructured).

source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{2, Tri}, triangulateIO, boundary_dict::Dict{Symbol, Int})
  • dg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • triangulateIO is a TriangulateIO mesh representation
  • boundary_dict is a Dict{Symbol, Int} which associates each integer TriangulateIO boundary tag with a Symbol.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti)

Constructs a single-element DGMultiMesh for a single periodic element given a DGMulti with approximation_type set to a periodic (finite difference) SBP operator from SummationByPartsOperators.jl.

source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{NDIMS}, vertex_coordinates, EToV;
+          RefElemData_kwargs...)

Create a discontinuous Galerkin method which uses

  • approximations of polynomial degree polydeg
  • element type element_type (Tri(), Quad(), Tet(), and Hex() currently supported)

Optional:

  • approximation_type (default is Polynomial(); SBP() also supported for Tri(), Quad(), and Hex() element types).
  • RefElemData_kwargs are additional keyword arguments for RefElemData, such as quad_rule_vol. For more info, see the StartUpDG.jl docs.
source
Trixi.DGMultiMesh — Type
DGMultiMesh{NDIMS, ...}

DGMultiMesh describes a mesh type which wraps StartUpDG.MeshData and boundary_faces in a dispatchable type. This is intended to store geometric data and connectivities for any type of mesh (Cartesian, affine, curved, structured/unstructured).

source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{2, Tri}, triangulateIO, boundary_dict::Dict{Symbol, Int})
  • dg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • triangulateIO is a TriangulateIO mesh representation
  • boundary_dict is a Dict{Symbol, Int} which associates each integer TriangulateIO boundary tag with a Symbol.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti)

Constructs a single-element DGMultiMesh for a single periodic element given a DGMulti with approximation_type set to a periodic (finite difference) SBP operator from SummationByPartsOperators.jl.

source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{NDIMS}, vertex_coordinates, EToV;
             is_on_boundary=nothing,
-            periodicity=ntuple(_->false, NDIMS)) where {NDIMS}
  • dg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • vertex_coordinates is a tuple of vectors containing x,y,... components of the vertex coordinates
  • EToV is a 2D array containing element-to-vertex connectivities for each element
  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of booleans specifying if the domain is periodic true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{NDIMS}, cells_per_dimension, mapping;
+            periodicity=ntuple(_->false, NDIMS)) where {NDIMS}
  • dg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • vertex_coordinates is a tuple of vectors containing x,y,... components of the vertex coordinates
  • EToV is a 2D array containing element-to-vertex connectivities for each element
  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of booleans specifying if the domain is periodic true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti{NDIMS}, cells_per_dimension, mapping;
             is_on_boundary=nothing,
-            periodicity=ntuple(_ -> false, NDIMS), kwargs...) where {NDIMS}

Constructs a Curved() DGMultiMesh with element type dg.basis.element_type.

  • mapping is a function which maps from a reference [-1, 1]^NDIMS domain to a mapped domain, e.g., xy = mapping(x, y) in 2D.
  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti, cells_per_dimension;
+            periodicity=ntuple(_ -> false, NDIMS), kwargs...) where {NDIMS}

Constructs a Curved() DGMultiMesh with element type dg.basis.element_type.

  • mapping is a function which maps from a reference [-1, 1]^NDIMS domain to a mapped domain, e.g., xy = mapping(x, y) in 2D.
  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti, cells_per_dimension;
             coordinates_min=(-1.0, -1.0), coordinates_max=(1.0, 1.0),
             is_on_boundary=nothing,
-            periodicity=ntuple(_ -> false, NDIMS))

Constructs a Cartesian DGMultiMesh with element type dg.basis.element_type. The domain is the tensor product of the intervals [coordinates_min[i], coordinates_max[i]].

  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti, filename::String)
  • dg::DGMulti contains information associated with the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • filename is a path specifying a .mesh file generated by HOHQMesh.
source
Trixi.DGSEM — Type
DGSEM(; RealT=Float64, polydeg::Integer,
+            periodicity=ntuple(_ -> false, NDIMS))

Constructs a Cartesian DGMultiMesh with element type dg.basis.element_type. The domain is the tensor product of the intervals [coordinates_min[i], coordinates_max[i]].

  • is_on_boundary specifies boundary using a Dict{Symbol, <:Function}
  • periodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.
source
Trixi.DGMultiMesh — Method
DGMultiMesh(dg::DGMulti, filename::String)
  • dg::DGMulti contains information associated with the reference element (e.g., quadrature, basis evaluation, differentiation, etc).
  • filename is a path specifying a .mesh file generated by HOHQMesh.
source
Trixi.DGSEM — Type
DGSEM(; RealT=Float64, polydeg::Integer,
         surface_flux=flux_central,
         surface_integral=SurfaceIntegralWeakForm(surface_flux),
         volume_integral=VolumeIntegralWeakForm(),
-        mortar=MortarL2(basis))

Create a discontinuous Galerkin spectral element method (DGSEM) using a LobattoLegendreBasis with polynomials of degree polydeg.

source
Trixi.DissipationLocalLaxFriedrichs — Type
DissipationLocalLaxFriedrichs(max_abs_speed=max_abs_speed_naive)

Create a local Lax-Friedrichs dissipation operator where the maximum absolute wave speed is estimated as max_abs_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to max_abs_speed_naive.

source
Trixi.EulerAcousticsCouplingCallback — Type
EulerAcousticsCouplingCallback
Experimental code

This callback is experimental and may change in any future release.

A callback that couples the acoustic perturbation equations and compressible Euler equations. Must be used in conjunction with SemidiscretizationEulerAcoustics. This callback manages the flow solver - which is always one time step ahead of the acoustics solver - and calculates the acoustic source term after each time step. The linearized Lamb vector is used as the source term, i.e.

\[\mathbf{s} = -(\mathbf{\omega'} \times \bar{\mathbf{v}} - + \bar{\mathbf{\omega}} \times \mathbf{v'}),\]

where $\mathbf{v}$ denotes the velocity, $\mathbf{\omega}$ denotes the vorticity, the bar $\bar{(\cdot)}$ indicates time-averaged quantities (see AveragingCallback) and prime $(\cdot)'$ denotes perturbed quantities defined by $\phi' = \phi - \bar{\phi}$. Note that the perturbed quantities here are based entirely on the pure flow solution and should not be confused with the state variables of the acoustic perturbation equations.

In addition, this callback manages the time step size for both solvers and initializes the mean values of the acoustic perturbation equations using results obtained with the AveragingCallback.

source
Trixi.EulerAcousticsCouplingCallback — Method
EulerAcousticsCouplingCallback(ode_euler, averaging_file::AbstractString, alg,
-                               cfl_acoustics::Real, cfl_euler::Real; kwargs...)
Experimental code

This callback is experimental and may change in any future release.

Creates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_file (see AveragingCallback).

source
Trixi.DissipationLocalLaxFriedrichs — Type
DissipationLocalLaxFriedrichs(max_abs_speed=max_abs_speed_naive)

Create a local Lax-Friedrichs dissipation operator where the maximum absolute wave speed is estimated as max_abs_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to max_abs_speed_naive.

source
Trixi.EulerAcousticsCouplingCallback — Type
EulerAcousticsCouplingCallback
Experimental code

This callback is experimental and may change in any future release.

A callback that couples the acoustic perturbation equations and compressible Euler equations. Must be used in conjunction with SemidiscretizationEulerAcoustics. This callback manages the flow solver - which is always one time step ahead of the acoustics solver - and calculates the acoustic source term after each time step. The linearized Lamb vector is used as the source term, i.e.

\[\mathbf{s} = -(\mathbf{\omega'} \times \bar{\mathbf{v}} + + \bar{\mathbf{\omega}} \times \mathbf{v'}),\]

where $\mathbf{v}$ denotes the velocity, $\mathbf{\omega}$ denotes the vorticity, the bar $\bar{(\cdot)}$ indicates time-averaged quantities (see AveragingCallback) and prime $(\cdot)'$ denotes perturbed quantities defined by $\phi' = \phi - \bar{\phi}$. Note that the perturbed quantities here are based entirely on the pure flow solution and should not be confused with the state variables of the acoustic perturbation equations.

In addition, this callback manages the time step size for both solvers and initializes the mean values of the acoustic perturbation equations using results obtained with the AveragingCallback.

source
Trixi.EulerAcousticsCouplingCallback — Method
EulerAcousticsCouplingCallback(ode_euler, averaging_file::AbstractString, alg,
+                               cfl_acoustics::Real, cfl_euler::Real; kwargs...)
Experimental code

This callback is experimental and may change in any future release.

Creates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_file (see AveragingCallback).

source
Trixi.EulerAcousticsCouplingCallback — Method
EulerAcousticsCouplingCallback(ode_euler,
                                averaging_callback::DiscreteCallback{<:Any, <:AveragingCallback},
-                               alg, cfl_acoustics::Real, cfl_euler::Real; kwargs...)
Experimental code

This callback is experimental and may change in any future release.

Creates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_callback (see AveragingCallback).

source
Trixi.FDSBP — Type
FDSBP(D_SBP; surface_integral, volume_integral)

Specialization of DG methods that uses general summation by parts (SBP) operators from SummationByPartsOperators.jl. In particular, this includes classical finite difference (FD) SBP methods. These methods have the same structure as classical DG methods - local operations on elements with connectivity through interfaces without imposing any continuity constraints.

D_SBP is an SBP derivative operator from SummationByPartsOperators.jl. The other arguments have the same meaning as in DG or DGSEM.

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.FluxHLL — Type
FluxHLL(min_max_speed=min_max_speed_naive)

Create an HLL (Harten, Lax, van Leer) numerical flux where the minimum and maximum wave speeds are estimated as λ_min, λ_max = min_max_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to min_max_speed_naive. Original paper:

  • Amiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002
source
Trixi.FluxHydrostaticReconstruction — Type
FluxHydrostaticReconstruction(numerical_flux, hydrostatic_reconstruction)
Experimental code

This numerical flux is experimental and may change in any future release.

Allow for some kind of hydrostatic reconstruction of the solution state prior to the surface flux computation. This is a particular strategy to ensure that the method remains well-balanced for the shallow water equations, see ShallowWaterEquations1D or ShallowWaterEquations2D.

For example, the hydrostatic reconstruction from Audusse et al. is implemented in one and two spatial dimensions, see hydrostatic_reconstruction_audusse_etal or the original paper

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090

Other hydrostatic reconstruction techniques are available, particularly to handle wet / dry fronts. A good overview of the development and application of hydrostatic reconstruction can be found in

  • Guoxian Chen and Sebastian Noelle A unified surface-gradient and hydrostatic reconstruction scheme for the shallow water equations (2021) RWTH Aachen preprint
  • Andreas Buttinger-Kreuzhuber, Zsolt Horváth, Sebastian Noelle, Günter Blöschl and Jürgen Waser (2019) A fast second-order shallow water scheme on two-dimensional structured grids over abrupt topography DOI: 10.1016/j.advwatres.2019.03.010
source
Trixi.FluxLMARS — Type
FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,
-             equations::CompressibleEulerEquations3D)

Low Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.

References:

  • Xi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1
source
Trixi.FluxPlusDissipation — Type
FluxPlusDissipation(numerical_flux, dissipation)

Combine a numerical_flux with a dissipation operator to create a new numerical flux.

source
Trixi.FluxRotated — Type
FluxRotated(numerical_flux)

Compute a numerical_flux flux in direction of a normal vector by rotating the solution, computing the numerical flux in x-direction, and rotating the calculated flux back.

Requires a rotationally invariant equation with equation-specific functions rotate_to_x and rotate_from_x.

source
Trixi.FluxUpwind — Type
FluxUpwind(splitting)

A numerical flux f(u_left, u_right) = f⁺(u_left) + f⁻(u_right) based on flux vector splitting.

The SurfaceIntegralUpwind with a given splitting is equivalent to the SurfaceIntegralStrongForm with FluxUpwind(splitting) as numerical flux (up to floating point differences).

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.GlmSpeedCallback — Type
GlmSpeedCallback(; glm_scale=0.5, cfl)

Update the divergence cleaning wave speed c_h according to the time step computed in StepsizeCallback for the ideal GLM-MHD equations. The cfl number should be set to the same value as for the time step size calculation. The glm_scale ensures that the GLM wave speed is lower than the fastest physical waves in the MHD solution and should thus be set to a value within the interval [0,1]. Note that glm_scale = 0 deactivates the divergence cleaning.

source
Trixi.GradientVariablesPrimitive — Type

GradientVariablesPrimitive and GradientVariablesEntropy are gradient variable type parameters for CompressibleNavierStokesDiffusion1D. By default, the gradient variables are set to be GradientVariablesPrimitive. Specifying GradientVariablesEntropy instead uses the entropy variable formulation from

  • Hughes, Mallet, Franca (1986) A new finite element formulation for computational fluid dynamics: I. Symmetric forms of the compressible Euler and Navier-Stokes equations and the second law of thermodynamics. https://doi.org/10.1016/0045-7825(86)90127-1

Under GradientVariablesEntropy, the Navier-Stokes discretization is provably entropy stable.

source
Trixi.HypDiffN3Erk3Sstar52 — Type
HypDiffN3Erk3Sstar52()

Five stage, second-order accurate explicit Runge-Kutta scheme with stability region optimized for the hyperbolic diffusion equation with LLF flux and polynomials of degree polydeg=3.

source
Trixi.HyperbolicDiffusionEquations1D — Type
HyperbolicDiffusionEquations1D

The linear hyperbolic diffusion equations in one space dimension. A description of this system can be found in Sec. 2.5 of the book

Further analysis can be found in the paper

source
Trixi.IdealGlmMhdEquations1D — Type
IdealGlmMhdEquations1D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in one space dimension.

Note

There is no divergence cleaning variable psi because the divergence-free constraint is satisfied trivially in one spatial dimension.

source
Trixi.IdealGlmMhdEquations2D — Type
IdealGlmMhdEquations2D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in two space dimensions.

source
Trixi.IdealGlmMhdEquations3D — Type
IdealGlmMhdEquations3D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in three space dimensions.

source
Trixi.IndicatorHennemannGassner — Type
IndicatorHennemannGassner(equations::AbstractEquations, basis;
+                               alg, cfl_acoustics::Real, cfl_euler::Real; kwargs...)
Experimental code

This callback is experimental and may change in any future release.

Creates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_callback (see AveragingCallback).

source
Trixi.FDSBP — Type
FDSBP(D_SBP; surface_integral, volume_integral)

Specialization of DG methods that uses general summation by parts (SBP) operators from SummationByPartsOperators.jl. In particular, this includes classical finite difference (FD) SBP methods. These methods have the same structure as classical DG methods - local operations on elements with connectivity through interfaces without imposing any continuity constraints.

D_SBP is an SBP derivative operator from SummationByPartsOperators.jl. The other arguments have the same meaning as in DG or DGSEM.

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.FluxHLL — Type
FluxHLL(min_max_speed=min_max_speed_davis)

Create an HLL (Harten, Lax, van Leer) numerical flux where the minimum and maximum wave speeds are estimated as λ_min, λ_max = min_max_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to min_max_speed_davis. Original paper:

  • Amiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002
source
Trixi.FluxHydrostaticReconstruction — Type
FluxHydrostaticReconstruction(numerical_flux, hydrostatic_reconstruction)
Experimental code

This numerical flux is experimental and may change in any future release.

Allow for some kind of hydrostatic reconstruction of the solution state prior to the surface flux computation. This is a particular strategy to ensure that the method remains well-balanced for the shallow water equations, see ShallowWaterEquations1D or ShallowWaterEquations2D.

For example, the hydrostatic reconstruction from Audusse et al. is implemented in one and two spatial dimensions, see hydrostatic_reconstruction_audusse_etal or the original paper

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090

Other hydrostatic reconstruction techniques are available, particularly to handle wet / dry fronts. A good overview of the development and application of hydrostatic reconstruction can be found in

  • Guoxian Chen and Sebastian Noelle A unified surface-gradient and hydrostatic reconstruction scheme for the shallow water equations (2021) RWTH Aachen preprint
  • Andreas Buttinger-Kreuzhuber, Zsolt Horváth, Sebastian Noelle, Günter Blöschl and Jürgen Waser (2019) A fast second-order shallow water scheme on two-dimensional structured grids over abrupt topography DOI: 10.1016/j.advwatres.2019.03.010
source
Trixi.FluxLMARS — Type
FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerEquations2D)

Low Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.

References:

  • Xi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1
source
Trixi.FluxLMARS — Method
FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerEquations3D)

Low Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.

References:

  • Xi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1
source
Trixi.FluxPlusDissipation — Type
FluxPlusDissipation(numerical_flux, dissipation)

Combine a numerical_flux with a dissipation operator to create a new numerical flux.

source
Trixi.FluxRotated — Type
FluxRotated(numerical_flux)

Compute a numerical_flux flux in direction of a normal vector by rotating the solution, computing the numerical flux in x-direction, and rotating the calculated flux back.

Requires a rotationally invariant equation with equation-specific functions rotate_to_x and rotate_from_x.

source
Trixi.FluxUpwind — Type
FluxUpwind(splitting)

A numerical flux f(u_left, u_right) = f⁺(u_left) + f⁻(u_right) based on flux vector splitting.

The SurfaceIntegralUpwind with a given splitting is equivalent to the SurfaceIntegralStrongForm with FluxUpwind(splitting) as numerical flux (up to floating point differences).

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.GlmSpeedCallback — Type
GlmSpeedCallback(; glm_scale=0.5, cfl)

Update the divergence cleaning wave speed c_h according to the time step computed in StepsizeCallback for the ideal GLM-MHD equations. The cfl number should be set to the same value as for the time step size calculation. The glm_scale ensures that the GLM wave speed is lower than the fastest physical waves in the MHD solution and should thus be set to a value within the interval [0,1]. Note that glm_scale = 0 deactivates the divergence cleaning.

source
Trixi.GradientVariablesPrimitive — Type

GradientVariablesPrimitive and GradientVariablesEntropy are gradient variable type parameters for CompressibleNavierStokesDiffusion1D. By default, the gradient variables are set to be GradientVariablesPrimitive. Specifying GradientVariablesEntropy instead uses the entropy variable formulation from

  • Hughes, Mallet, Franca (1986) A new finite element formulation for computational fluid dynamics: I. Symmetric forms of the compressible Euler and Navier-Stokes equations and the second law of thermodynamics. https://doi.org/10.1016/0045-7825(86)90127-1

Under GradientVariablesEntropy, the Navier-Stokes discretization is provably entropy stable.

source
Trixi.HypDiffN3Erk3Sstar52 — Type
HypDiffN3Erk3Sstar52()

Five stage, second-order accurate explicit Runge-Kutta scheme with stability region optimized for the hyperbolic diffusion equation with LLF flux and polynomials of degree polydeg=3.

source
Trixi.HyperbolicDiffusionEquations1D — Type
HyperbolicDiffusionEquations1D

The linear hyperbolic diffusion equations in one space dimension. A description of this system can be found in Sec. 2.5 of the book

Further analysis can be found in the paper

source
Trixi.IdealGlmMhdEquations1D — Type
IdealGlmMhdEquations1D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in one space dimension.

Note

There is no divergence cleaning variable psi because the divergence-free constraint is satisfied trivially in one spatial dimension.

source
Trixi.IdealGlmMhdEquations2D — Type
IdealGlmMhdEquations2D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in two space dimensions.

source
Trixi.IdealGlmMhdEquations3D — Type
IdealGlmMhdEquations3D(gamma)

The ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in three space dimensions.

source
Trixi.IndicatorHennemannGassner — Type
IndicatorHennemannGassner(equations::AbstractEquations, basis;
                           alpha_max=0.5,
                           alpha_min=0.001,
                           alpha_smooth=true,
@@ -207,24 +208,20 @@
                           alpha_max=0.5,
                           alpha_min=0.001,
                           alpha_smooth=true,
-                          variable)

Indicator used for shock-capturing (when passing the equations and the basis) or adaptive mesh refinement (AMR, when passing the semi).

See also VolumeIntegralShockCapturingHG.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.IndicatorHennemannGassnerShallowWater — Type
IndicatorHennemannGassnerShallowWater(equations::AbstractEquations, basis;
-                                      alpha_max=0.5,
-                                      alpha_min=0.001,
-                                      alpha_smooth=true,
-                                      variable)

Modified version of the IndicatorHennemannGassner indicator used for shock-capturing for shallow water equations. After the element-wise values for the blending factors are computed an additional check is made to see if the element is partially wet. In this case, partially wet elements are set to use the pure finite volume scheme that is guaranteed to be well-balanced for this wet/dry transition state of the flow regime.

See also VolumeIntegralShockCapturingHG.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.IndicatorLöhner — Type
IndicatorLöhner (equivalent to IndicatorLoehner)
+                          variable)

Indicator used for shock-capturing (when passing the equations and the basis) or adaptive mesh refinement (AMR, when passing the semi).

See also VolumeIntegralShockCapturingHG.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.IndicatorLöhner — Type
IndicatorLöhner (equivalent to IndicatorLoehner)
 
 IndicatorLöhner(equations::AbstractEquations, basis;
                 f_wave=0.2, variable)
 IndicatorLöhner(semi::AbstractSemidiscretization;
-                f_wave=0.2, variable)

AMR indicator adapted from a FEM indicator by Löhner (1987), also used in the FLASH code as standard AMR indicator. The indicator estimates a weighted second derivative of a specified variable locally.

When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.

References

source
Trixi.IndicatorMax — Type
IndicatorMax(equations::AbstractEquations, basis; variable)
-IndicatorMax(semi::AbstractSemidiscretization; variable)

A simple indicator returning the maximum of variable in an element. When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.

source
Trixi.Isothermal — Type
struct Isothermal

Used to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and return a scalar value for the temperature at point x and time t.

source
Trixi.LaplaceDiffusion1D — Type
LaplaceDiffusion1D(diffusivity, equations)

LaplaceDiffusion1D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LaplaceDiffusion2D — Type
LaplaceDiffusion2D(diffusivity, equations)

LaplaceDiffusion2D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LaplaceDiffusion3D — Type
LaplaceDiffusion3D(diffusivity, equations)

LaplaceDiffusion3D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LatticeBoltzmannEquations2D — Type
LatticeBoltzmannEquations2D(; Ma, Re, collision_op=collision_bgk,
+                f_wave=0.2, variable)

AMR indicator adapted from a FEM indicator by Löhner (1987), also used in the FLASH code as standard AMR indicator. The indicator estimates a weighted second derivative of a specified variable locally.

When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.

References

source
Trixi.IndicatorMax — Type
IndicatorMax(equations::AbstractEquations, basis; variable)
+IndicatorMax(semi::AbstractSemidiscretization; variable)

A simple indicator returning the maximum of variable in an element. When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.

source
Trixi.Isothermal — Type
struct Isothermal

Used to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and return a scalar value for the temperature at point x and time t.

source
Trixi.LaplaceDiffusion1D — Type
LaplaceDiffusion1D(diffusivity, equations)

LaplaceDiffusion1D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LaplaceDiffusion2D — Type
LaplaceDiffusion2D(diffusivity, equations)

LaplaceDiffusion2D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LaplaceDiffusion3D — Type
LaplaceDiffusion3D(diffusivity, equations)

LaplaceDiffusion3D represents a scalar diffusion term $\nabla \cdot (\kappa\nabla u))$ with diffusivity $\kappa$ applied to each solution component defined by equations.

source
Trixi.LatticeBoltzmannEquations2D — Type
LatticeBoltzmannEquations2D(; Ma, Re, collision_op=collision_bgk,
                            c=1, L=1, rho0=1, u0=nothing, nu=nothing)

The Lattice-Boltzmann equations

\[\partial_t u_\alpha + v_{\alpha,1} \partial_1 u_\alpha + v_{\alpha,2} \partial_2 u_\alpha = 0\]

in two space dimensions for the D2Q9 scheme.

The characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).

The nine discrete velocity directions of the D2Q9 scheme are sorted as follows [4]:

  6     2     5       y
     ┌───┼───┐         │
     │       │         │
   3 ┼   9   ┼ 1        ──── x
     │       │        ╱
     └───┼───┘       ╱
-  7     4     8    z

Note that usually the velocities are numbered from 0 to 8, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 9, where 1 through 8 correspond to the velocity directions in [4] and 9 is the zero velocity.

The corresponding opposite directions are:

  • 1 ←→ 3
  • 2 ←→ 4
  • 3 ←→ 1
  • 4 ←→ 2
  • 5 ←→ 7
  • 6 ←→ 8
  • 7 ←→ 5
  • 8 ←→ 6
  • 9 ←→ 9

The main sources for the base implementation were

  1. Misun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024
  2. Karsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.
  3. Dieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0
  4. Dieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3
source
Trixi.LatticeBoltzmannEquations3D — Type
LatticeBoltzmannEquations3D(; Ma, Re, collision_op=collision_bgk,
+  7     4     8    z

Note that usually the velocities are numbered from 0 to 8, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 9, where 1 through 8 correspond to the velocity directions in [4] and 9 is the zero velocity.

The corresponding opposite directions are:

  • 1 ←→ 3
  • 2 ←→ 4
  • 3 ←→ 1
  • 4 ←→ 2
  • 5 ←→ 7
  • 6 ←→ 8
  • 7 ←→ 5
  • 8 ←→ 6
  • 9 ←→ 9

The main sources for the base implementation were

  1. Misun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024
  2. Karsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.
  3. Dieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0
  4. Dieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3
source
Trixi.LatticeBoltzmannEquations3D — Type
LatticeBoltzmannEquations3D(; Ma, Re, collision_op=collision_bgk,
                            c=1, L=1, rho0=1, u0=nothing, nu=nothing)

The Lattice-Boltzmann equations

\[\partial_t u_\alpha + v_{\alpha,1} \partial_1 u_\alpha + v_{\alpha,2} \partial_2 u_\alpha + v_{\alpha,3} \partial_3 u_\alpha = 0\]

in three space dimensions for the D3Q27 scheme.

The characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).

The twenty-seven discrete velocity directions of the D3Q27 scheme are sorted as follows [4]:

  • plane at z = -1:
      24    17     21       y
      ┌───┼───┐          │
      │       │          │
@@ -243,7 +240,7 @@
   16 ┼   5   ┼ 9         ──── x
      │       │         ╱
      └───┼───┘        ╱
-  22    18     23    z

Note that usually the velocities are numbered from 0 to 26, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 27, where 1 through 26 correspond to the velocity directions in [4] and 27 is the zero velocity.

The corresponding opposite directions are:

  • 1 ←→ 2
  • 2 ←→ 1
  • 3 ←→ 4
  • 4 ←→ 3
  • 5 ←→ 6
  • 6 ←→ 5
  • 7 ←→ 8
  • 8 ←→ 7
  • 9 ←→ 10
  • 10 ←→ 9
  • 11 ←→ 12
  • 12 ←→ 11
  • 13 ←→ 14
  • 14 ←→ 13
  • 15 ←→ 16
  • 16 ←→ 15
  • 17 ←→ 18
  • 18 ←→ 17
  • 19 ←→ 20
  • 20 ←→ 19
  • 21 ←→ 22
  • 22 ←→ 21
  • 23 ←→ 24
  • 24 ←→ 23
  • 25 ←→ 26
  • 26 ←→ 25
  • 27 ←→ 27

The main sources for the base implementation were

  1. Misun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024
  2. Karsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.
  3. Dieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0
  4. Dieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3
source
Trixi.LinearScalarAdvectionEquation2D — Type
LinearScalarAdvectionEquation2D

The linear scalar advection equation

\[\partial_t u + a_1 \partial_1 u + a_2 \partial_2 u = 0\]

in two space dimensions with constant velocity a.

source
Trixi.LinearScalarAdvectionEquation3D — Type
LinearScalarAdvectionEquation3D

The linear scalar advection equation

\[\partial_t u + a_1 \partial_1 u + a_2 \partial_2 u + a_3 \partial_3 u = 0\]

in three space dimensions with constant velocity a.

source
Trixi.LinearizedEulerEquations2D — Type
LinearizedEulerEquations2D(v_mean_global, c_mean_global, rho_mean_global)

Linearized euler equations in two space dimensions. The equations are given by

\[\partial_t + 22 18 23 z

Note that usually the velocities are numbered from 0 to 26, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 27, where 1 through 26 correspond to the velocity directions in [4] and 27 is the zero velocity.

The corresponding opposite directions are:

  • 1 ←→ 2
  • 2 ←→ 1
  • 3 ←→ 4
  • 4 ←→ 3
  • 5 ←→ 6
  • 6 ←→ 5
  • 7 ←→ 8
  • 8 ←→ 7
  • 9 ←→ 10
  • 10 ←→ 9
  • 11 ←→ 12
  • 12 ←→ 11
  • 13 ←→ 14
  • 14 ←→ 13
  • 15 ←→ 16
  • 16 ←→ 15
  • 17 ←→ 18
  • 18 ←→ 17
  • 19 ←→ 20
  • 20 ←→ 19
  • 21 ←→ 22
  • 22 ←→ 21
  • 23 ←→ 24
  • 24 ←→ 23
  • 25 ←→ 26
  • 26 ←→ 25
  • 27 ←→ 27

The main sources for the base implementation were

  1. Misun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024
  2. Karsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.
  3. Dieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0
  4. Dieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3
source
Trixi.LinearScalarAdvectionEquation2D — Type
LinearScalarAdvectionEquation2D

The linear scalar advection equation

\[\partial_t u + a_1 \partial_1 u + a_2 \partial_2 u = 0\]

in two space dimensions with constant velocity a.

source
Trixi.LinearScalarAdvectionEquation3D — Type
LinearScalarAdvectionEquation3D

The linear scalar advection equation

\[\partial_t u + a_1 \partial_1 u + a_2 \partial_2 u + a_3 \partial_3 u = 0\]

in three space dimensions with constant velocity a.

source
Trixi.LinearizedEulerEquations2D — Type
LinearizedEulerEquations2D(v_mean_global, c_mean_global, rho_mean_global)

Linearized euler equations in two space dimensions. The equations are given by

\[\partial_t \begin{pmatrix} \rho' \\ v_1' \\ v_2' \\ p' \end{pmatrix} @@ -260,19 +257,20 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 -\end{pmatrix}\]

The bar $\bar{(\cdot)}$ indicates uniform mean flow variables and c is the speed of sound. The unknowns are the acoustic velocities $v' = (v_1', v_2')$, the pressure $p'$ and the density $\rho'$.

source
Trixi.LobattoLegendreBasis — Type
LobattoLegendreBasis([RealT=Float64,] polydeg::Integer)

Create a nodal Lobatto-Legendre basis for polynomials of degree polydeg.

For the special case polydeg=0 the DG method reduces to a finite volume method. Therefore, this function sets the center point of the cell as single node.

source
Trixi.NoSlip — Type
struct NoSlip

Use to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and should return a SVector{NDIMS} whose entries are the velocity vector at a point x and time t.

source
Trixi.NonConservativeLocal — Type
NonConservativeLocal()

Struct used for multiple dispatch on non-conservative flux functions in the format of "local * symmetric". When the argument nonconservative_type is of type NonConservativeLocal, the function returns the local part of the non-conservative term.

source
Trixi.NonConservativeSymmetric — Type
NonConservativeSymmetric()

Struct used for multiple dispatch on non-conservative flux functions in the format of "local * symmetric". When the argument nonconservative_type is of type NonConservativeSymmetric, the function returns the symmetric part of the non-conservative term.

source
Trixi.P4estMesh — Type
P4estMesh{NDIMS} <: AbstractMesh{NDIMS}

An unstructured curved mesh based on trees that uses the C library p4est to manage trees and mesh refinement.

source
Trixi.P4estMesh — Method
P4estMesh(trees_per_dimension; polydeg,
+\end{pmatrix}\]
The bar $\bar{(\cdot)}$ indicates uniform mean flow variables and c is the speed of sound. The unknowns are the acoustic velocities $v' = (v_1', v_2')$, the pressure $p'$ and the density $\rho'$.
source
Trixi.LobattoLegendreBasis — Type
LobattoLegendreBasis([RealT=Float64,] polydeg::Integer)

Create a nodal Lobatto-Legendre basis for polynomials of degree polydeg.

For the special case polydeg=0 the DG method reduces to a finite volume method. Therefore, this function sets the center point of the cell as single node.

source
Trixi.NoSlip — Type
struct NoSlip

Use to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and should return a SVector{NDIMS} whose entries are the velocity vector at a point x and time t.

source
Trixi.NonConservativeLocal — Type
NonConservativeLocal()

Struct used for multiple dispatch on non-conservative flux functions in the format of "local * symmetric". When the argument nonconservative_type is of type NonConservativeLocal, the function returns the local part of the non-conservative term.

source
Trixi.NonConservativeSymmetric — Type
NonConservativeSymmetric()

Struct used for multiple dispatch on non-conservative flux functions in the format of "local * symmetric". When the argument nonconservative_type is of type NonConservativeSymmetric, the function returns the symmetric part of the non-conservative term.

source
Trixi.P4estMesh — Type
P4estMesh{NDIMS} <: AbstractMesh{NDIMS}

An unstructured curved mesh based on trees that uses the C library p4est to manage trees and mesh refinement.

source
Trixi.P4estMesh — Method
P4estMesh(trees_per_dimension; polydeg,
           mapping=nothing, faces=nothing, coordinates_min=nothing, coordinates_max=nothing,
           RealT=Float64, initial_refinement_level=0, periodicity=true, unsaved_changes=true,
-          p4est_partition_allow_for_coarsening=true)

Create a structured curved P4estMesh of the specified size.

There are three ways to map the mesh to the physical domain.

  1. Define a mapping that maps the hypercube [-1, 1]^n.
  2. Specify a Tuple faces of functions that parametrize each face.
  3. Create a rectangular mesh by specifying coordinates_min and coordinates_max.

Non-periodic boundaries will be called :x_neg, :x_pos, :y_neg, :y_pos, :z_neg, :z_pos.

Arguments

  • trees_per_dimension::NTupleE{NDIMS, Int}: the number of trees in each dimension.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • faces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
source
Trixi.P4estMesh — Method
P4estMesh{NDIMS}(meshfile::String;
+          p4est_partition_allow_for_coarsening=true)

Create a structured curved P4estMesh of the specified size.

There are three ways to map the mesh to the physical domain.

  1. Define a mapping that maps the hypercube [-1, 1]^n.
  2. Specify a Tuple faces of functions that parametrize each face.
  3. Create a rectangular mesh by specifying coordinates_min and coordinates_max.

Non-periodic boundaries will be called :x_neg, :x_pos, :y_neg, :y_pos, :z_neg, :z_pos.

Arguments

  • trees_per_dimension::NTupleE{NDIMS, Int}: the number of trees in each dimension.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • faces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
source
Trixi.P4estMesh — Method
P4estMesh{NDIMS}(meshfile::String;
                  mapping=nothing, polydeg=1, RealT=Float64,
                  initial_refinement_level=0, unsaved_changes=true,
-                 p4est_partition_allow_for_coarsening=true)

Main mesh constructor for the P4estMesh that imports an unstructured, conforming mesh from an Abaqus mesh file (.inp). Each element of the conforming mesh parsed from the meshfile is created as a p4est tree datatype.

To create a curved unstructured mesh P4estMesh two strategies are available:

  • p4est_mesh_from_hohqmesh_abaqus: High-order, curved boundary information created by HOHQMesh.jl is available in the meshfile. The mesh polynomial degree polydeg of the boundaries is provided from the meshfile. The computation of the mapped tree coordinates is done with transfinite interpolation with linear blending similar to UnstructuredMesh2D. Boundary name information is also parsed from the meshfile such that different boundary conditions can be set at each named boundary on a given tree.
  • p4est_mesh_from_standard_abaqus: By default, with mapping=nothing and polydeg=1, this creates a straight-sided from the information parsed from the meshfile. If a mapping function is specified then it computes the mapped tree coordinates via polynomial interpolants with degree polydeg. The mesh created by this function will only have one boundary :all, as distinguishing different physical boundaries is non-trivial.

Note that the mapping and polydeg keyword arguments are only used by the p4est_mesh_from_standard_abaqus function. The p4est_mesh_from_hohqmesh_abaqus function obtains the mesh polydeg directly from the meshfile and constructs the transfinite mapping internally.

The particular strategy is selected according to the header present in the meshfile where the constructor checks whether or not the meshfile was created with HOHQMesh.jl. If the Abaqus file header is not present in the meshfile then the P4estMesh is created with the function p4est_mesh_from_standard_abaqus.

The default keyword argument initial_refinement_level=0 corresponds to a forest where the number of trees is the same as the number of elements in the original meshfile. Increasing the initial_refinement_level allows one to uniformly refine the base mesh given in the meshfile to create a forest with more trees before the simulation begins. For example, if a two-dimensional base mesh contains 25 elements then setting initial_refinement_level=1 creates an initial forest of 2^2 * 25 = 100 trees.

Arguments

  • meshfile::String: an uncurved Abaqus mesh file that can be imported by p4est.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
source
Trixi.ParametersEulerGravity — Type
ParametersEulerGravity(; background_density=0.0,
+                 p4est_partition_allow_for_coarsening=true,
+                 boundary_symbols = nothing)

Main mesh constructor for the P4estMesh that imports an unstructured, conforming mesh from an Abaqus mesh file (.inp). Each element of the conforming mesh parsed from the meshfile is created as a p4est tree datatype.

To create a curved unstructured mesh P4estMesh two strategies are available:

  • p4est_mesh_from_hohqmesh_abaqus: High-order, curved boundary information created by HOHQMesh.jl is available in the meshfile. The mesh polynomial degree polydeg of the boundaries is provided from the meshfile. The computation of the mapped tree coordinates is done with transfinite interpolation with linear blending similar to UnstructuredMesh2D. Boundary name information is also parsed from the meshfile such that different boundary conditions can be set at each named boundary on a given tree.
  • p4est_mesh_from_standard_abaqus: By default, with mapping=nothing and polydeg=1, this creates a straight-sided from the information parsed from the meshfile. If a mapping function is specified then it computes the mapped tree coordinates via polynomial interpolants with degree polydeg. The mesh created by this function will only have one boundary :all if boundary_symbols is not specified. If boundary_symbols is specified the mesh file will be parsed for nodesets defining the boundary nodes from which boundary edges (2D) and faces (3D) will be assigned.

Note that the mapping and polydeg keyword arguments are only used by the p4est_mesh_from_standard_abaqus function. The p4est_mesh_from_hohqmesh_abaqus function obtains the mesh polydeg directly from the meshfile and constructs the transfinite mapping internally.

The particular strategy is selected according to the header present in the meshfile where the constructor checks whether or not the meshfile was created with HOHQMesh.jl. If the Abaqus file header is not present in the meshfile then the P4estMesh is created with the function p4est_mesh_from_standard_abaqus.

The default keyword argument initial_refinement_level=0 corresponds to a forest where the number of trees is the same as the number of elements in the original meshfile. Increasing the initial_refinement_level allows one to uniformly refine the base mesh given in the meshfile to create a forest with more trees before the simulation begins. For example, if a two-dimensional base mesh contains 25 elements then setting initial_refinement_level=1 creates an initial forest of 2^2 * 25 = 100 trees.

Arguments

  • meshfile::String: an uncurved Abaqus mesh file that can be imported by p4est.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
  • boundary_symbols::Vector{Symbol}: A vector of symbols that correspond to the boundary names in the meshfile. If nothing is passed then all boundaries are named :all.
source
Trixi.PerformanceCounter — Type
PerformanceCounter()

A PerformanceCounter can be used to track the runtime performance of some calls. Add a new runtime measurement via put!(counter, runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.

source
Trixi.PerformanceCounterList — Type
PerformanceCounterList{N}()

A PerformanceCounterList{N} can be used to track the runtime performance of calls to multiple functions, adding them up. Add a new runtime measurement via put!(counter.counters[i], runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.

source
Trixi.PlotData1D — Type
PlotData1D

Holds all relevant data for creating 1D plots of multiple solution variables and to visualize the mesh.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData1D — Method
PlotData1D(u, semi [or mesh, equations, solver, cache];
-           solution_variables=nothing, nvisnodes=nothing)

Create a new PlotData1D object that can be used for visualizing 1D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.

nvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.

When visualizing data from a two-dimensional simulation, a 1D slice is extracted for plotting. slice specifies the axis along which the slice is extracted and may be :x, or :y. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0). Both of these values are ignored when visualizing 1D data. This applies analogously to three-dimensional simulations, where slice may be :xy, :xz, or :yz.

Another way to visualize 2D/3D data is by creating a plot along a given curve. This is done with the keyword argument curve. It can be set to a list of 2D/3D points which define the curve. When using curve any other input from slice or point will be ignored.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData1D — Method
PlotData1D(sol; kwargs...)

Create a PlotData1D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData2DCartesian — Type
PlotData2D

Holds all relevant data for creating 2D plots of multiple solution variables and to visualize the mesh.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PerformanceCounter — Type
PerformanceCounter()

A PerformanceCounter can be used to track the runtime performance of some calls. Add a new runtime measurement via put!(counter, runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.

source
Trixi.PerformanceCounterList — Type
PerformanceCounterList{N}()

A PerformanceCounterList{N} can be used to track the runtime performance of calls to multiple functions, adding them up. Add a new runtime measurement via put!(counter.counters[i], runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.

source
Trixi.PlotData1D — Type
PlotData1D

Holds all relevant data for creating 1D plots of multiple solution variables and to visualize the mesh.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData1D — Method
PlotData1D(u, semi [or mesh, equations, solver, cache];
+           solution_variables=nothing, nvisnodes=nothing)

Create a new PlotData1D object that can be used for visualizing 1D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.

nvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.

When visualizing data from a two-dimensional simulation, a 1D slice is extracted for plotting. slice specifies the axis along which the slice is extracted and may be :x, or :y. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0). Both of these values are ignored when visualizing 1D data. This applies analogously to three-dimensional simulations, where slice may be :xy, :xz, or :yz.

Another way to visualize 2D/3D data is by creating a plot along a given curve. This is done with the keyword argument curve. It can be set to a list of 2D/3D points which define the curve. When using curve any other input from slice or point will be ignored.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData1D — Method
PlotData1D(sol; kwargs...)

Create a PlotData1D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PlotData2DCartesian — Type
PlotData2D

Holds all relevant data for creating 2D plots of multiple solution variables and to visualize the mesh.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.PolytropicEulerEquations2D — Type
PolytropicEulerEquations2D(gamma, kappa)

The polytropic Euler equations

\[\frac{\partial}{\partial t} \begin{pmatrix} \rho \\ \rho v_1 \\ \rho v_2 \end{pmatrix} @@ -289,103 +287,65 @@ = \begin{pmatrix} 0 \\ 0 \\ 0 -\end{pmatrix}\]

for an ideal gas with ratio of specific heats gamma in two space dimensions. Here, $\rho$ is the density and $v_1$ andv_2 the velocities and

\[p = \kappa\rho^\gamma\]

the pressure, which we replaced using this relation.

source
Trixi.PositivityPreservingLimiterShallowWater — Type
PositivityPreservingLimiterShallowWater(; variables)

The limiter is specifically designed for the shallow water equations. It is applied to all scalar variables in their given order using the defined threshold_limiter from the ShallowWaterEquations1D struct or the ShallowWaterEquations2D struct to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.

As opposed to the standard version of the PositivityPreservingLimiterZhangShu, nodes with a water height below the threshold_limiter are treated in a special way. To avoid numerical problems caused by velocities close to zero, the velocity is cut off, such that the node can be identified as "dry". The special feature of the ShallowWaterEquations used here is that the bottom topography is stored as an additional quantity in the solution vector u. However, the value of the bottom topography should not be changed. That is why, it is not limited.

After the limiting process is applied to all degrees of freedom, for safety reasons, the threshold_limiter is applied again on all the DG nodes in order to avoid water height below. In the case where the cell mean value is below the threshold before applying the limiter, there could still be dry nodes afterwards due to the logic of the limiter.

This fully-discrete positivity-preserving limiter is based on the work of

  • Zhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153
source
Trixi.PositivityPreservingLimiterZhangShu — Type
PositivityPreservingLimiterZhangShu(; threshold, variables)

The fully-discrete positivity-preserving limiter of

  • Zhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153

The limiter is applied to all scalar variables in their given order using the associated thresholds to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.

source
Trixi.SaveRestartCallback — Type
SaveRestartCallback(; interval=0,
+\end{pmatrix}\]
for an ideal gas with ratio of specific heats gamma in two space dimensions. Here, $\rho$ is the density and $v_1$ andv_2 the velocities and
\[p = \kappa\rho^\gamma\]
the pressure, which we replaced using this relation.
source
Trixi.PositivityPreservingLimiterZhangShu — Type
PositivityPreservingLimiterZhangShu(; threshold, variables)

The fully-discrete positivity-preserving limiter of

  • Zhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153

The limiter is applied to all scalar variables in their given order using the associated thresholds to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.

source
Trixi.SaveRestartCallback — Type
SaveRestartCallback(; interval=0,
                       save_final_restart=true,
-                      output_directory="out")

Save the current numerical solution in a restart file every interval time steps.

source
Trixi.SaveSolutionCallback — Type
SaveSolutionCallback(; interval::Integer=0,
+                      output_directory="out")

Save the current numerical solution in a restart file every interval time steps.

source
Trixi.SaveSolutionCallback — Type
SaveSolutionCallback(; interval::Integer=0,
                        dt=nothing,
                        save_initial_solution=true,
                        save_final_solution=true,
                        output_directory="out",
-                       solution_variables=cons2prim)

Save the current numerical solution in regular intervals. Either pass interval to save every interval time steps or pass dt to save in intervals of dt in terms of integration time by adding additional (shortened) time steps where necessary (note that this may change the solution). solution_variables can be any callable that converts the conservative variables at a single point to a set of solution variables. The first parameter passed to solution_variables will be the set of conservative variables and the second parameter is the equation struct.

source
Trixi.SemidiscretizationCoupled — Type
SemidiscretizationCoupled

A struct used to bundle multiple semidiscretizations. semidiscretize will return an ODEProblem that synchronizes time steps between the semidiscretizations. Each call of rhs! will call rhs! for each semidiscretization individually. The semidiscretizations can be coupled by gluing meshes together using BoundaryConditionCoupled.

Experimental code

This is an experimental feature and can change any time.

source
Trixi.SemidiscretizationEulerAcoustics — Type
SemidiscretizationEulerAcoustics(semi_acoustics::SemiAcoustics, semi_euler::SemiEuler;
-                                 source_region=x->true, weights=x->1.0)
Experimental code

This semidiscretization is experimental and may change in any future release.

Construct a semidiscretization of the acoustic perturbation equations that is coupled with the compressible Euler equations via source terms for the perturbed velocity. Both semidiscretizations have to use the same mesh and solvers with a shared basis. The coupling region is described by a function source_region that maps the coordinates of a single node to true or false depending on whether the point lies within the coupling region or not. A weighting function weights that maps coordinates to weights is applied to the acoustic source terms. Note that this semidiscretization should be used in conjunction with EulerAcousticsCouplingCallback and only works in two dimensions.

source
Trixi.SemidiscretizationEulerGravity — Type
SemidiscretizationEulerGravity

A struct containing everything needed to describe a spatial semidiscretization of a the compressible Euler equations with self-gravity, reformulating the Poisson equation for the gravitational potential as steady-state problem of the hyperblic diffusion equations.

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) "A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics" arXiv: 2008.10593
source
Trixi.SemidiscretizationHyperbolic — Method
SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;
+                       solution_variables=cons2prim)

Save the current numerical solution in regular intervals. Either pass interval to save every interval time steps or pass dt to save in intervals of dt in terms of integration time by adding additional (shortened) time steps where necessary (note that this may change the solution). solution_variables can be any callable that converts the conservative variables at a single point to a set of solution variables. The first parameter passed to solution_variables will be the set of conservative variables and the second parameter is the equation struct.

source
Trixi.SemidiscretizationCoupled — Type
SemidiscretizationCoupled

A struct used to bundle multiple semidiscretizations. semidiscretize will return an ODEProblem that synchronizes time steps between the semidiscretizations. Each call of rhs! will call rhs! for each semidiscretization individually. The semidiscretizations can be coupled by gluing meshes together using BoundaryConditionCoupled.

Experimental code

This is an experimental feature and can change any time.

source
Trixi.SemidiscretizationEulerAcoustics — Type
SemidiscretizationEulerAcoustics(semi_acoustics::SemiAcoustics, semi_euler::SemiEuler;
+                                 source_region=x->true, weights=x->1.0)
Experimental code

This semidiscretization is experimental and may change in any future release.

Construct a semidiscretization of the acoustic perturbation equations that is coupled with the compressible Euler equations via source terms for the perturbed velocity. Both semidiscretizations have to use the same mesh and solvers with a shared basis. The coupling region is described by a function source_region that maps the coordinates of a single node to true or false depending on whether the point lies within the coupling region or not. A weighting function weights that maps coordinates to weights is applied to the acoustic source terms. Note that this semidiscretization should be used in conjunction with EulerAcousticsCouplingCallback and only works in two dimensions.

source
Trixi.SemidiscretizationEulerGravity — Type
SemidiscretizationEulerGravity

A struct containing everything needed to describe a spatial semidiscretization of a the compressible Euler equations with self-gravity, reformulating the Poisson equation for the gravitational potential as steady-state problem of the hyperblic diffusion equations.

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) "A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics" arXiv: 2008.10593
source
Trixi.SemidiscretizationHyperbolic — Method
SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;
                              source_terms=nothing,
                              boundary_conditions=boundary_condition_periodic,
                              RealT=real(solver),
                              uEltype=RealT,
-                             initial_cache=NamedTuple())

Construct a semidiscretization of a hyperbolic PDE.

source
Trixi.SemidiscretizationHyperbolicParabolic — Method
SemidiscretizationHyperbolicParabolic(mesh, both_equations, initial_condition, solver;
                                       solver_parabolic=default_parabolic_solver(),
                                       source_terms=nothing,
                                       both_boundary_conditions=(boundary_condition_periodic, boundary_condition_periodic),
                                       RealT=real(solver),
                                       uEltype=RealT,
-                                      both_initial_caches=(NamedTuple(), NamedTuple()))

Construct a semidiscretization of a hyperbolic-parabolic PDE.

source
Trixi.ShallowWaterEquations1D — Type
ShallowWaterEquations1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)

Shallow water equations (SWE) in one space dimension. The equations are given by

\[\begin{aligned} + both_initial_caches=(NamedTuple(), NamedTuple()))

Construct a semidiscretization of a hyperbolic-parabolic PDE.

source
Trixi.ShallowWaterEquations1D — Type
ShallowWaterEquations1D(; gravity, H0 = 0)

Shallow water equations (SWE) in one space dimension. The equations are given by

\[\begin{aligned} \frac{\partial h}{\partial t} + \frac{\partial}{\partial x}(h v) &= 0 \\ \frac{\partial}{\partial t}(h v) + \frac{\partial}{\partial x}\left(h v^2 + \frac{g}{2}h^2\right) + g h \frac{\partial b}{\partial x} &= 0 -\end{aligned}\]

The unknown quantities of the SWE are the water height $h$ and the velocity $v$. The gravitational constant is denoted by g and the (possibly) variable bottom topography function $b(x)$. Conservative variable water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

Also, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is "wet" before calculating the numerical flux.

The bottom topography function $b(x)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.

In addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography by default.

References for the SWE are many but a good introduction is available in Chapter 13 of the book:

source
Trixi.ShallowWaterEquations2D — Type
ShallowWaterEquations2D(; gravity, H0 = 0, threshold_limiter = nothing, threshold_wet = nothing)

Shallow water equations (SWE) in two space dimensions. The equations are given by

\[\begin{aligned} +\end{aligned}\]

The unknown quantities of the SWE are the water height $h$ and the velocity $v$. The gravitational constant is denoted by g and the (possibly) variable bottom topography function $b(x)$. Conservative variable water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

The bottom topography function $b(x)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.

In addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography by default.

References for the SWE are many but a good introduction is available in Chapter 13 of the book:

source
Trixi.ShallowWaterEquations2D — Type
ShallowWaterEquations2D(; gravity, H0 = 0)

Shallow water equations (SWE) in two space dimensions. The equations are given by

\[\begin{aligned} \frac{\partial h}{\partial t} + \frac{\partial}{\partial x}(h v_1) + \frac{\partial}{\partial y}(h v_2) &= 0 \\ \frac{\partial}{\partial t}(h v_1) + \frac{\partial}{\partial x}\left(h v_1^2 + \frac{g}{2}h^2\right) + \frac{\partial}{\partial y}(h v_1 v_2) + g h \frac{\partial b}{\partial x} &= 0 \\ \frac{\partial}{\partial t}(h v_2) + \frac{\partial}{\partial x}(h v_1 v_2) + \frac{\partial}{\partial y}\left(h v_2^2 + \frac{g}{2}h^2\right) + g h \frac{\partial b}{\partial y} &= 0. -\end{aligned}\]

The unknown quantities of the SWE are the water height $h$ and the velocities $\mathbf{v} = (v_1, v_2)^T$. The gravitational constant is denoted by g and the (possibly) variable bottom topography function $b(x,y)$. Conservative variable water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

Also, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is "wet" before calculating the numerical flux.

The bottom topography function $b(x,y)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.

In addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography by default.

References for the SWE are many but a good introduction is available in Chapter 13 of the book:

source
Trixi.ShallowWaterEquationsQuasi1D — Type
ShallowWaterEquationsQuasi1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)

The quasi-1D shallow water equations (SWE). The equations are given by

\[\begin{aligned} +\end{aligned}\]

The unknown quantities of the SWE are the water height $h$ and the velocities $\mathbf{v} = (v_1, v_2)^T$. The gravitational constant is denoted by g and the (possibly) variable bottom topography function $b(x,y)$. Conservative variable water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

The bottom topography function $b(x,y)$ is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.

In addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography by default.

References for the SWE are many but a good introduction is available in Chapter 13 of the book:

source
Trixi.ShallowWaterEquationsQuasi1D — Type
ShallowWaterEquationsQuasi1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)

The quasi-1D shallow water equations (SWE). The equations are given by

\[\begin{aligned} \frac{\partial}{\partial t}(a h) + \frac{\partial}{\partial x}(a h v) &= 0 \\ \frac{\partial}{\partial t}(a h v) + \frac{\partial}{\partial x}(a h v^2) + g a h \frac{\partial}{\partial x}(h + b) &= 0 -\end{aligned}\]

The unknown quantities of the Quasi-1D SWE are the water height $h$ and the scaled velocity $v$. The gravitational constant is denoted by g, the (possibly) variable bottom topography function $b(x)$, and (possibly) variable channel width $a(x)$. The water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

Also, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is "wet" before calculating the numerical flux.

The bottom topography function $b(x)$ and channel width $a(x)$ are set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero and $a$ to one.

In addition to the unknowns, Trixi.jl currently stores the bottom topography and channel width values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography and channel width must be zero.
  • The bottom topography and channel width values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography and channel width by default.
source
Trixi.ShallowWaterTwoLayerEquations1D — Type
ShallowWaterTwoLayerEquations1D(gravity, H0, rho_upper, rho_lower)

Two-Layer Shallow Water equations (2LSWE) in one space dimension. The equations are given by

\[\begin{alignat*}{4} -&\frac{\partial}{\partial t}h_{upper} -&&+ \frac{\partial}{\partial x}\left(h_{upper} v_{1,upper}\right) -&&= 0 \\ -&\frac{\partial}{\partial t}\left(h_{upper}v_{1,upper}\right) -&&+ \frac{\partial}{\partial x}\left(h_{upper}v_{1,upper}^2 + \dfrac{gh_{upper}^2}{2}\right) -&&= -gh_{upper}\frac{\partial}{\partial x}\left(b+h_{lower}\right)\\ -&\frac{\partial}{\partial t}h_{lower} -&&+ \frac{\partial}{\partial x}\left(h_{lower}v_{1,lower}\right) -&&= 0 \\ -&\frac{\partial}{\partial t}\left(h_{lower}v_{1,lower}\right) -&&+ \frac{\partial}{\partial x}\left(h_{lower}v_{1,lower}^2 + \dfrac{gh_{lower}^2}{2}\right) -&&= -gh_{lower}\frac{\partial}{\partial x}\left(b+\dfrac{\rho_{upper}}{\rho_{lower}}h_{upper}\right). -\end{alignat*}\]

The unknown quantities of the 2LSWE are the water heights of the {lower} layer $h_{lower}$ and the {upper} layer $h_{upper}$ with respective velocities $v_{1,upper}$ and $v_{1,lower}$. The gravitational constant is denoted by g, the layer densitites by $\rho_{upper}$and $\rho_{lower}$ and the (possibly) variable bottom topography function $b(x)$. The conservative variable water height $h_{lower}$ is measured from the bottom topography $b$ and $h_{upper}$ relative to $h_{lower}$, therefore one also defines the total water heights as $H_{upper} = h_{upper} + h_{upper} + b$ and $H_{lower} = h_{lower} + b$.

The densities must be chosen such that $\rho_{upper} < \rho_{lower}$, to make sure that the heavier fluid $\rho_{lower}$ is in the bottom layer and the lighter fluid $\rho_{upper}$ in the {upper} layer.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

The bottom topography function $b(x)$ is set inside the initial condition routine for a particular problem setup.

In addition to the unknowns, Trixi currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi's visualization tools will visualize the bottom topography by default.

A good introduction for the 2LSWE is available in Chapter 12 of the book:

source
Trixi.ShallowWaterTwoLayerEquations2D — Type
ShallowWaterTwoLayerEquations2D(gravity, H0, rho_upper, rho_lower)

Two-Layer Shallow water equations (2LSWE) in two space dimension. The equations are given by

\[\begin{alignat*}{8} -&\frac{\partial}{\partial t}h_{upper} -&&+ \frac{\partial}{\partial x}\left(h_{upper} v_{1,upper}\right) -&&+ \frac{\partial}{\partial y}\left(h_{upper} v_{2,upper}\right) \quad -&&= \quad 0 \\ -&\frac{\partial}{\partial t}\left(h_{upper} v_{1,upper}\right) -&&+ \frac{\partial}{\partial x}\left(h_{upper} v_{1,upper}^2 + \frac{gh_{upper}^2}{2}\right) -&&+ \frac{\partial}{\partial y}\left(h_{upper} v_{1,upper} v_{2,upper}\right) \quad -&&= -gh_{upper}\frac{\partial}{\partial x}\left(b+h_{lower}\right) \\ -&\frac{\partial}{\partial t}\left(h_{upper} v_{2,upper}\right) -&&+ \frac{\partial}{\partial x}\left(h_{upper} v_{1,upper} v_{2,upper}\right) -&&+ \frac{\partial}{\partial y}\left(h_{upper} v_{2,upper}^2 + \frac{gh_{upper}^2}{2}\right) -&&= -gh_{upper}\frac{\partial}{\partial y}\left(b+h_{lower}\right)\\ -&\frac{\partial}{\partial t}h_{lower} -&&+ \frac{\partial}{\partial x}\left(h_{lower} v_{1,lower}\right) -&&+ \frac{\partial}{\partial y}\left(h_{lower} v_{2,lower}\right) -&&= \quad 0 \\ -&\frac{\partial}{\partial t}\left(h_{lower} v_{1,lower}\right) -&&+ \frac{\partial}{\partial x}\left(h_{lower} v_{1,lower}^2 + \frac{gh_{lower}^2}{2}\right) -&&+ \frac{\partial}{\partial y}\left(h_{lower} v_{1,lower} v_{2,lower}\right) -&&= -gh_{lower}\frac{\partial}{\partial x}\left(b+\frac{\rho_{upper}}{\rho_{lower}} h_{upper}\right)\\ -&\frac{\partial}{\partial t}\left(h_{lower} v_{2,lower}\right) -&&+ \frac{\partial}{\partial x}\left(h_{lower} v_{1,lower} v_{2,lower}\right) -&&+ \frac{\partial}{\partial y}\left(h_{lower} v_{2,lower}^2 + \frac{gh_{lower}^2}{2}\right) -&&= -gh_{lower}\frac{\partial}{\partial y}\left(b+\frac{\rho_{upper}}{\rho_{lower}} h_{upper}\right) -\end{alignat*}\]

The unknown quantities of the 2LSWE are the water heights of the lower layer $h_{lower}$ and the upper layer $h_{upper}$ and the respective velocities in x-direction $v_{1,lower}$ and $v_{1,upper}$ and in y-direction $v_{2,lower}$ and $v_{2,upper}$. The gravitational constant is denoted by g, the layer densitites by $\rho_{upper}$and $\rho_{lower}$ and the (possibly) variable bottom topography function by $b(x)$. Conservative variable water height $h_{lower}$ is measured from the bottom topography $b$ and $h_{upper}$ relative to $h_{lower}$, therefore one also defines the total water heights as $H_{lower} = h_{lower} + b$ and $H_{upper} = h_{upper} + h_{lower} + b$.

The densities must be chosen such that $\rho_{upper} < \rho_{lower}$, to make sure that the heavier fluid $\rho_{lower}$ is in the bottom layer and the lighter fluid $\rho_{upper}$ in the upper layer.

The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.

The bottom topography function $b(x)$ is set inside the initial condition routine for a particular problem setup.

In addition to the unknowns, Trixi currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:

  • The flux values corresponding to the bottom topography must be zero.
  • The bottom topography values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi's visualization tools will visualize the bottom topography by default.

A good introduction for the 2LSWE is available in Chapter 12 of the book:

source
Trixi.SimpleSSPRK33 — Type
SimpleSSPRK33(; stage_callbacks=())

The third-order SSP Runge-Kutta method of Shu and Osher.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.StepsizeCallback — Type
StepsizeCallback(; cfl=1.0)

Set the time step size according to a CFL condition with CFL number cfl if the time integration method isn't adaptive itself.

source
Trixi.StructuredMesh — Type
StructuredMesh{NDIMS} <: AbstractMesh{NDIMS}

A structured curved mesh.

Different numbers of cells per dimension are possible and arbitrary functions can be used as domain faces.

source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, coordinates_min, coordinates_max; periodicity=true)

Create a StructuredMesh that represents a uncurved structured mesh with a rectangular domain.

Arguments

  • cells_per_dimension::NTuple{NDIMS, Int}: the number of cells in each dimension.
  • coordinates_min::NTuple{NDIMS, RealT}: coordinate of the corner in the negative direction of each dimension.
  • coordinates_max::NTuple{NDIMS, RealT}: coordinate of the corner in the positive direction of each dimension.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, mapping; RealT=Float64, unsaved_changes=true, mapping_as_string=mapping2string(mapping, length(cells_per_dimension)))

Create a StructuredMesh of the given size and shape that uses RealT as coordinate type.

Arguments

  • cells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.
  • mapping: a function of NDIMS variables to describe the mapping, which transforms the reference mesh to the physical domain. If no mapping_as_string is defined, this function must be defined with the name mapping to allow for restarts. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.
  • RealT::Type: the type that should be used for coordinates.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • mapping_as_string::String: the code that defines the mapping. If CodeTracking can't find the function definition, it can be passed directly here. The code string must define the mapping function with the name mapping. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.
source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, faces; RealT=Float64, unsaved_changes=true, faces_as_string=faces2string(faces))

Create a StructuredMesh of the given size and shape that uses RealT as coordinate type.

Arguments

  • cells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.
  • faces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D).
  • RealT::Type: the type that should be used for coordinates.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.SubcellLimiterIDP — Type
SubcellLimiterIDP(equations::AbstractEquations, basis;
+\end{aligned}\]
The unknown quantities of the Quasi-1D SWE are the water height $h$ and the scaled velocity $v$. The gravitational constant is denoted by g, the (possibly) variable bottom topography function $b(x)$, and (possibly) variable channel width $a(x)$. The water height $h$ is measured from the bottom topography $b$, therefore one also defines the total water height as $H = h + b$.
The additional quantity $H_0$ is also available to store a reference value for the total water height that is useful to set initial conditions or test the "lake-at-rest" well-balancedness.
The bottom topography function $b(x)$ and channel width $a(x)$ are set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero and $a$ to one. 
In addition to the unknowns, Trixi.jl currently stores the bottom topography and channel width values at the approximation points  despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height $H$ or the entropy variables. This affects the implementation and use of these equations in various ways:
  • The flux values corresponding to the bottom topography and channel width must be zero.
  • The bottom topography and channel width values must be included when defining initial conditions, boundary conditions or source terms.
  • AnalysisCallback analyzes this variable.
  • Trixi.jl's visualization tools will visualize the bottom topography and channel width by default.
source
Trixi.SimpleSSPRK33 — Type
SimpleSSPRK33(; stage_callbacks=())

The third-order SSP Runge-Kutta method of Shu and Osher.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.StepsizeCallback — Type
StepsizeCallback(; cfl=1.0)

Set the time step size according to a CFL condition with CFL number cfl if the time integration method isn't adaptive itself.

source
Trixi.StructuredMesh — Type
StructuredMesh{NDIMS} <: AbstractMesh{NDIMS}

A structured curved mesh.

Different numbers of cells per dimension are possible and arbitrary functions can be used as domain faces.

source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, coordinates_min, coordinates_max; periodicity=true)

Create a StructuredMesh that represents a uncurved structured mesh with a rectangular domain.

Arguments

  • cells_per_dimension::NTuple{NDIMS, Int}: the number of cells in each dimension.
  • coordinates_min::NTuple{NDIMS, RealT}: coordinate of the corner in the negative direction of each dimension.
  • coordinates_max::NTuple{NDIMS, RealT}: coordinate of the corner in the positive direction of each dimension.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, mapping; RealT=Float64, unsaved_changes=true, mapping_as_string=mapping2string(mapping, length(cells_per_dimension)))

Create a StructuredMesh of the given size and shape that uses RealT as coordinate type.

Arguments

  • cells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.
  • mapping: a function of NDIMS variables to describe the mapping, which transforms the reference mesh to the physical domain. If no mapping_as_string is defined, this function must be defined with the name mapping to allow for restarts. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.
  • RealT::Type: the type that should be used for coordinates.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • mapping_as_string::String: the code that defines the mapping. If CodeTracking can't find the function definition, it can be passed directly here. The code string must define the mapping function with the name mapping. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.
source
Trixi.StructuredMesh — Method
StructuredMesh(cells_per_dimension, faces; RealT=Float64, unsaved_changes=true, faces_as_string=faces2string(faces))

Create a StructuredMesh of the given size and shape that uses RealT as coordinate type.

Arguments

  • cells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.
  • faces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D).
  • RealT::Type: the type that should be used for coordinates.
  • periodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.SubcellLimiterIDP — Type
SubcellLimiterIDP(equations::AbstractEquations, basis;
                   local_minmax_variables_cons = String[],
                   positivity_variables_cons = String[],
-                  positivity_correction_factor = 0.1)

Subcell invariant domain preserving (IDP) limiting used with VolumeIntegralSubcellLimiting including:

  • Local maximum/minimum Zalesak-type limiting for conservative variables (local_minmax_variables_cons)
  • Positivity limiting for conservative variables (positivity_variables_cons)

Conservative variables to be limited are passed as a vector of strings, e.g. local_minmax_variables_cons = ["rho"] and positivity_variables_cons = ["rho"].

The bounds are calculated using the low-order FV solution. The positivity limiter uses positivity_correction_factor such that u^new >= positivity_correction_factor * u^FV.

Note

This limiter and the correction callback SubcellLimiterIDPCorrection only work together. Without the callback, no correction takes place, leading to a standard low-order FV scheme.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.SubcellLimiterIDPCorrection — Type
SubcellLimiterIDPCorrection()

Perform antidiffusive correction stage for the a posteriori IDP limiter SubcellLimiterIDP called with VolumeIntegralSubcellLimiting.

Note

This callback and the actual limiter SubcellLimiterIDP only work together. This is not a replacement but a necessary addition.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.SurfaceIntegralWeakForm — Type
SurfaceIntegralWeakForm(surface_flux=flux_central)

The classical weak form surface integral type for DG methods as explained in standard textbooks.

See also VolumeIntegralWeakForm.

References

source
Trixi.T8codeMesh — Type
T8codeMesh{NDIMS} <: AbstractMesh{NDIMS}

An unstructured curved mesh based on trees that uses the C library 't8code' to manage trees and mesh refinement.

source
Trixi.T8codeMesh — Method
T8codeMesh(trees_per_dimension; polydeg, mapping=identity,
-           RealT=Float64, initial_refinement_level=0, periodicity=true)

Create a structured potentially curved 'T8codeMesh' of the specified size.

Non-periodic boundaries will be called ':xneg', ':xpos', ':yneg', ':ypos', ':zneg', ':zpos'.

Arguments

  • 'treesperdimension::NTupleE{NDIMS, Int}': the number of trees in each dimension.
  • 'polydeg::Integer': polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • 'mapping': a function of 'NDIMS' variables to describe the mapping that transforms the reference mesh ('[-1, 1]^n') to the physical domain.
  • 'RealT::Type': the type that should be used for coordinates.
  • 'initialrefinementlevel::Integer': refine the mesh uniformly to this level before the simulation starts.
  • 'periodicity': either a 'Bool' deciding if all of the boundaries are periodic or an 'NTuple{NDIMS, Bool}' deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.T8codeMesh — Method
T8codeMesh{NDIMS}(conn::Ptr{p4est_connectivity},
-                  mapping=nothing, polydeg=1, RealT=Float64,
-                  initial_refinement_level=0)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.

Arguments

  • conn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.T8codeMesh — Method
T8codeMesh{NDIMS}(cmesh::Ptr{t8_cmesh},
-                 mapping=nothing, polydeg=1, RealT=Float64,
-                 initial_refinement_level=0)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a t8_cmesh data structure.

Arguments

  • cmesh::Ptr{t8_cmesh}: Pointer to a cmesh object.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.T8codeMesh — Method
T8codeMesh{NDIMS}(meshfile::String;
-                 mapping=nothing, polydeg=1, RealT=Float64,
-                 initial_refinement_level=0)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a Gmsh mesh file (.msh).

Arguments

  • meshfile::String: path to a Gmsh mesh file.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.TimeSeriesCallback — Type
TimeSeriesCallback(semi, point_coordinates;
+                  positivity_variables_nonlinear = [],
+                  positivity_correction_factor = 0.1,
+                  max_iterations_newton = 10,
+                  newton_tolerances = (1.0e-12, 1.0e-14),
+                  gamma_constant_newton = 2 * ndims(equations))

Subcell invariant domain preserving (IDP) limiting used with VolumeIntegralSubcellLimiting including:

  • Local maximum/minimum Zalesak-type limiting for conservative variables (local_minmax_variables_cons)
  • Positivity limiting for conservative variables (positivity_variables_cons) and nonlinear variables

(positivity_variables_nonlinear)

Conservative variables to be limited are passed as a vector of strings, e.g. local_minmax_variables_cons = ["rho"] and positivity_variables_cons = ["rho"]. For nonlinear variables the specific functions are passed in a vector, e.g. positivity_variables_nonlinear = [pressure].

The bounds are calculated using the low-order FV solution. The positivity limiter uses positivity_correction_factor such that u^new >= positivity_correction_factor * u^FV. The limiting of nonlinear variables uses a Newton-bisection method with a maximum of max_iterations_newton iterations, relative and absolute tolerances of newton_tolerances and a provisional update constant gamma_constant_newton (gamma_constant_newton>=2*d, where d = #dimensions). See equation (20) of Pazner (2020) and equation (30) of Rueda-Ramírez et al. (2022).

Note

This limiter and the correction callback SubcellLimiterIDPCorrection only work together. Without the callback, no correction takes place, leading to a standard low-order FV scheme.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.SubcellLimiterIDPCorrection — Type
SubcellLimiterIDPCorrection()

Perform antidiffusive correction stage for the a posteriori IDP limiter SubcellLimiterIDP called with VolumeIntegralSubcellLimiting.

Note

This callback and the actual limiter SubcellLimiterIDP only work together. This is not a replacement but a necessary addition.

References

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.SurfaceIntegralWeakForm — Type
SurfaceIntegralWeakForm(surface_flux=flux_central)

The classical weak form surface integral type for DG methods as explained in standard textbooks.

See also VolumeIntegralWeakForm.

References

source
Trixi.T8codeMesh — Type
T8codeMesh{NDIMS} <: AbstractMesh{NDIMS}

An unstructured curved mesh based on trees that uses the C library 't8code' to manage trees and mesh refinement.

source
Trixi.T8codeMesh — Method
T8codeMesh(trees_per_dimension; polydeg, mapping=identity,
+           RealT=Float64, initial_refinement_level=0, periodicity=true)

Create a structured potentially curved 'T8codeMesh' of the specified size.

Non-periodic boundaries will be called ':xneg', ':xpos', ':yneg', ':ypos', ':zneg', ':zpos'.

Arguments

  • 'treesperdimension::NTupleE{NDIMS, Int}': the number of trees in each dimension.
  • 'polydeg::Integer': polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • faces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • coordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.
  • 'RealT::Type': the type that should be used for coordinates.
  • 'initialrefinementlevel::Integer': refine the mesh uniformly to this level before the simulation starts.
  • 'periodicity': either a 'Bool' deciding if all of the boundaries are periodic or an 'NTuple{NDIMS, Bool}' deciding for each dimension if the boundaries in this dimension are periodic.
source
Trixi.T8codeMesh — Method
T8codeMesh(conn::Ptr{p4est_connectivity}; kwargs...)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.

Arguments

  • conn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.T8codeMesh — Method
T8codeMesh(conn::Ptr{p8est_connectivity}; kwargs...)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.

Arguments

  • conn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.T8codeMesh — Method
T8codeMesh(cmesh::Ptr{t8_cmesh},
+           mapping=nothing, polydeg=1, RealT=Float64,
+           initial_refinement_level=0)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a t8_cmesh data structure.

Arguments

  • cmesh::Ptr{t8_cmesh}: Pointer to a cmesh object.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.T8codeMesh — Method
T8codeMesh(meshfile::String, ndims; kwargs...)

Main mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a Gmsh mesh file (.msh).

Arguments

  • meshfile::String: path to a Gmsh mesh file.
  • ndims: Mesh file dimension: 2 or 3.
  • mapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
source
Trixi.TimeSeriesCallback — Type
TimeSeriesCallback(semi, point_coordinates;
                    interval=1, solution_variables=cons2cons,
                    output_directory="out", filename="time_series.h5",
-                   RealT=real(solver), uEltype=eltype(cache.elements))

Create a callback that records point-wise data at points given in point_coordinates every interval time steps. The point coordinates are to be specified either as a vector of coordinate tuples or as a two-dimensional array where the first dimension is the point number and the second dimension is the coordinate dimension. By default, the conservative variables are recorded, but this can be controlled by passing a different conversion function to solution_variables.

After the last time step, the results are stored in an HDF5 file filename in directory output_directory.

The real data type RealT and data type for solution variables uEltype default to the respective types used in the solver and the cache.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.TreeMesh — Type
TreeMesh{NDIMS} <: AbstractMesh{NDIMS}

A Cartesian mesh based on trees of hypercubes to support adaptive mesh refinement.

source
Trixi.UnstructuredMesh2D — Type
UnstructuredMesh2D <: AbstractMesh{2}

An unstructured (possibly curved) quadrilateral mesh.

UnstructuredMesh2D(filename; RealT=Float64, periodicity=false)

All mesh information, neighbour coupling, and boundary curve information is read in from a mesh file filename.

source
Trixi.UnstructuredSortedBoundaryTypes — Type
UnstructuredSortedBoundaryTypes

General container to sort the boundary conditions by type for some unstructured meshes/solvers. It stores a set of global indices for each boundary condition type to expedite computation during the call to calc_boundary_flux!. The original dictionary form of the boundary conditions set by the user in the elixir file is also stored for printing.

source
Trixi.ViscousFormulationLocalDG — Type
ViscousFormulationLocalDG(penalty_parameter)

The local DG (LDG) flux from "The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems" by Cockburn and Shu (1998).

Note that, since this implementation does not involve the parabolic "upwinding" vector, the LDG solver is equivalent to ViscousFormulationBassiRebay1 with an LDG-type penalization.

source
Trixi.VisualizationCallback — Method
VisualizationCallback(; interval=0,
+                   RealT=real(solver), uEltype=eltype(cache.elements))

Create a callback that records point-wise data at points given in point_coordinates every interval time steps. The point coordinates are to be specified either as a vector of coordinate tuples or as a two-dimensional array where the first dimension is the point number and the second dimension is the coordinate dimension. By default, the conservative variables are recorded, but this can be controlled by passing a different conversion function to solution_variables.

After the last time step, the results are stored in an HDF5 file filename in directory output_directory.

The real data type RealT and data type for solution variables uEltype default to the respective types used in the solver and the cache.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.TrafficFlowLWREquations1D — Type
TrafficFlowLWREquations1D

The classic Lighthill-Witham Richards (LWR) model for 1D traffic flow. The car density is denoted by $u \in [0, 1]$ and the maximum possible speed (e.g. due to speed limits) is $v_{\text{max}}$.

\[\partial_t u + v_{\text{max}} \partial_1 [u (1 - u)] = 0\]

For more details see e.g. Section 11.1 of

  • Randall LeVeque (2002)

Finite Volume Methods for Hyperbolic Problems [DOI: 10.1017/CBO9780511791253]https://doi.org/10.1017/CBO9780511791253

source
Trixi.TreeMesh — Type
TreeMesh{NDIMS} <: AbstractMesh{NDIMS}

A Cartesian mesh based on trees of hypercubes to support adaptive mesh refinement.

source
Trixi.UnstructuredMesh2D — Type
UnstructuredMesh2D <: AbstractMesh{2}

An unstructured (possibly curved) quadrilateral mesh.

UnstructuredMesh2D(filename; RealT=Float64, periodicity=false)

All mesh information, neighbour coupling, and boundary curve information is read in from a mesh file filename.

source
Trixi.UnstructuredSortedBoundaryTypes — Type
UnstructuredSortedBoundaryTypes

General container to sort the boundary conditions by type for some unstructured meshes/solvers. It stores a set of global indices for each boundary condition type to expedite computation during the call to calc_boundary_flux!. The original dictionary form of the boundary conditions set by the user in the elixir file is also stored for printing.

source
Trixi.ViscousFormulationLocalDG — Type
ViscousFormulationLocalDG(penalty_parameter)

The local DG (LDG) flux from "The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems" by Cockburn and Shu (1998).

Note that, since this implementation does not involve the parabolic "upwinding" vector, the LDG solver is equivalent to ViscousFormulationBassiRebay1 with an LDG-type penalization.

source
Trixi.VisualizationCallback — Method
VisualizationCallback(; interval=0,
                         solution_variables=cons2prim,
                         variable_names=[],
                         show_mesh=false,
                         plot_data_creator=PlotData2D,
                         plot_creator=show_plot,
-                        plot_arguments...)

Create a callback that visualizes results during a simulation, also known as in-situ visualization.

Experimental implementation

This is an experimental feature and may change in any future releases.

The interval specifies the number of time step iterations after which a new plot is generated. The available variables to plot are configured with the solution_variables parameter, which acts the same way as for the SaveSolutionCallback. The variables to be actually plotted can be selected by providing a single string or a list of strings to variable_names, and if show_mesh is true, an additional plot with the mesh will be generated.

To customize the generated figure, plot_data_creator allows to use different plot data types. With plot_creator you can further specify an own function to visualize results, which must support the same interface as the default implementation show_plot. All remaining keyword arguments are collected and passed as additional arguments to the plotting command.

source
Trixi.VolumeIntegralFluxDifferencing — Type
VolumeIntegralFluxDifferencing(volume_flux)

Volume integral type for DG methods based on SBP operators and flux differencing using a symmetric two-point volume_flux. This volume_flux needs to satisfy the interface of numerical fluxes in Trixi.jl.

References

source
Trixi.VolumeIntegralPureLGLFiniteVolume — Type
VolumeIntegralPureLGLFiniteVolume(volume_flux_fv)

A volume integral that only uses the subcell finite volume schemes of the VolumeIntegralShockCapturingHG.

This gives a formally O(1)-accurate finite volume scheme on an LGL-type subcell mesh (LGL = Legendre-Gauss-Lobatto).

Experimental implementation

This is an experimental feature and may change in future releases.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.VolumeIntegralShockCapturingHG — Type
VolumeIntegralShockCapturingHG(indicator; volume_flux_dg=flux_central,
-                                          volume_flux_fv=flux_lax_friedrichs)

Shock-capturing volume integral type for DG methods using a convex blending of the finite volume method with numerical flux volume_flux_fv and the VolumeIntegralFluxDifferencing with volume flux volume_flux_dg. The amount of blending is determined by the indicator, e.g., IndicatorHennemannGassner.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.VolumeIntegralSubcellLimiting — Type
VolumeIntegralSubcellLimiting(limiter;
-                              volume_flux_dg, volume_flux_fv)

A subcell limiting volume integral type for DG methods based on subcell blending approaches with a low-order FV method. Used with limiter SubcellLimiterIDP.

Note

Subcell limiting methods are not fully functional on non-conforming meshes. This is mainly because the implementation assumes that low- and high-order schemes have the same surface terms, which is not guaranteed for non-conforming meshes. The low-order scheme with a high-order mortar is not invariant domain preserving.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.VolumeIntegralUpwind — Type
VolumeIntegralUpwind(splitting)

Specialized volume integral for finite difference summation-by-parts (FDSBP) solvers. Can be used together with the upwind SBP operators of Mattsson (2017) implemented in SummationByPartsOperators.jl. The splitting controls the discretization.

See also splitting_steger_warming, splitting_lax_friedrichs, splitting_vanleer_haenel.

References

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.VolumeIntegralWeakForm — Type
VolumeIntegralWeakForm()

The classical weak form volume integral type for DG methods as explained in standard textbooks.

References

VolumeIntegralWeakForm() is only implemented for conserved terms as non-conservative terms should always be discretized in conjunction with a flux-splitting scheme, see VolumeIntegralFluxDifferencing. This treatment is required to achieve, e.g., entropy-stability or well-balancedness.

source
Base.getindex — Method
Base.getindex(pd::AbstractPlotData, variable_name)

Extract a single variable variable_name from pd for plotting with Plots.plot.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Base.resize! — Method
resize!(c::AbstractContainer, new_length) -> AbstractContainer

Resize c to contain new_length elements. If new_length is smaller than the current container length, the first new_length elements will be retained. If new_length is larger, the new elements are invalidated.

source
PolynomialBases.compute_coefficients — Method
compute_coefficients(func, t, semi::AbstractSemidiscretization)

Compute the discrete coefficients of the continuous function func at time t associated with the semidiscretization semi. For example, the discrete coefficients of func for a discontinuous Galerkin spectral element method (DGSEM) are the values of func at the Lobatto-Legendre nodes. Similarly, a classical finite difference method will use the values of func at the nodes of the grid assoociated with the semidiscretization semi.

For semidiscretizations semi associated with an initial condition, func can be omitted to use the given initial condition at time t.

source
PolynomialBases.integrate — Method
integrate(f, u, basis::LobattoLegendreBasis)

Map the function f to the coefficients u and integrate with respect to the quadrature rule given by basis.

source
PolynomialBases.integrate — Method
integrate([func=(u_node,equations)->u_node,] u_ode, semi::AbstractSemidiscretization; normalize=true)

Call func(u_node, equations) for each vector of nodal variables u_node in u_ode and integrate the result using a quadrature associated with the semidiscretization semi.

If normalize is true, the result is divided by the total volume of the computational domain.

source
SciMLBase.add_tstop! — Method
add_tstop!(integrator::SimpleIntegratorSSP, t)

Add a time stop during the time integration process. This function is called after the periodic SaveSolutionCallback to specify the next stop to save the solution.

source
SummationByPartsOperators.semidiscretize — Method
semidiscretize(semi::SemidiscretizationHyperbolicParabolic, tspan)

Wrap the semidiscretization semi as a split ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The parabolic right-hand side is the first function of the split ODE problem and will be used by default by the implicit part of IMEX methods from the SciML ecosystem.

source
SummationByPartsOperators.semidiscretize — Method
semidiscretize(semi::AbstractSemidiscretization, tspan, restart_file::AbstractString)

Wrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The initial condition etc. is taken from the restart_file.

source
Trixi.DGMultiBasis — Method
DGMultiBasis(element_type, polydeg; approximation_type = Polynomial(), kwargs...)

Constructs a basis for DGMulti solvers. Returns a "StartUpDG.RefElemData" object. The kwargs arguments are additional keyword arguments for RefElemData, such as quad_rule_vol. These are the same as the RefElemData_kwargs used in DGMulti. For more info, see the StartUpDG.jl docs.

source
Trixi.P4estMeshCubedSphere — Method
P4estMeshCubedSphere(trees_per_face_dimension, layers, inner_radius, thickness;
+                        plot_arguments...)

Create a callback that visualizes results during a simulation, also known as in-situ visualization.

Experimental implementation

This is an experimental feature and may change in any future releases.

The interval specifies the number of time step iterations after which a new plot is generated. The available variables to plot are configured with the solution_variables parameter, which acts the same way as for the SaveSolutionCallback. The variables to be actually plotted can be selected by providing a single string or a list of strings to variable_names, and if show_mesh is true, an additional plot with the mesh will be generated.

To customize the generated figure, plot_data_creator allows to use different plot data types. With plot_creator you can further specify an own function to visualize results, which must support the same interface as the default implementation show_plot. All remaining keyword arguments are collected and passed as additional arguments to the plotting command.

source
Trixi.VolumeIntegralFluxDifferencing — Type
VolumeIntegralFluxDifferencing(volume_flux)

Volume integral type for DG methods based on SBP operators and flux differencing using a symmetric two-point volume_flux. This volume_flux needs to satisfy the interface of numerical fluxes in Trixi.jl.

References

source
Trixi.VolumeIntegralPureLGLFiniteVolume — Type
VolumeIntegralPureLGLFiniteVolume(volume_flux_fv)

A volume integral that only uses the subcell finite volume schemes of the VolumeIntegralShockCapturingHG.

This gives a formally O(1)-accurate finite volume scheme on an LGL-type subcell mesh (LGL = Legendre-Gauss-Lobatto).

Experimental implementation

This is an experimental feature and may change in future releases.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.VolumeIntegralShockCapturingHG — Type
VolumeIntegralShockCapturingHG(indicator; volume_flux_dg=flux_central,
+                                          volume_flux_fv=flux_lax_friedrichs)

Shock-capturing volume integral type for DG methods using a convex blending of the finite volume method with numerical flux volume_flux_fv and the VolumeIntegralFluxDifferencing with volume flux volume_flux_dg. The amount of blending is determined by the indicator, e.g., IndicatorHennemannGassner.

References

  • Hennemann, Gassner (2020) "A provably entropy stable subcell shock capturing approach for high order split form DG" arXiv: 2008.12044
source
Trixi.VolumeIntegralSubcellLimiting — Type
VolumeIntegralSubcellLimiting(limiter;
+                              volume_flux_dg, volume_flux_fv)

A subcell limiting volume integral type for DG methods based on subcell blending approaches with a low-order FV method. Used with limiter SubcellLimiterIDP.

Note

Subcell limiting methods are not fully functional on non-conforming meshes. This is mainly because the implementation assumes that low- and high-order schemes have the same surface terms, which is not guaranteed for non-conforming meshes. The low-order scheme with a high-order mortar is not invariant domain preserving.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.VolumeIntegralUpwind — Type
VolumeIntegralUpwind(splitting)

Specialized volume integral for finite difference summation-by-parts (FDSBP) solvers. Can be used together with the upwind SBP operators of Mattsson (2017) implemented in SummationByPartsOperators.jl. The splitting controls the discretization.

See also splitting_steger_warming, splitting_lax_friedrichs, splitting_vanleer_haenel.

References

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.VolumeIntegralWeakForm — Type
VolumeIntegralWeakForm()

The classical weak form volume integral type for DG methods as explained in standard textbooks.

References

VolumeIntegralWeakForm() is only implemented for conserved terms as non-conservative terms should always be discretized in conjunction with a flux-splitting scheme, see VolumeIntegralFluxDifferencing. This treatment is required to achieve, e.g., entropy-stability or well-balancedness.

source
Base.getindex — Method
Base.getindex(pd::AbstractPlotData, variable_name)

Extract a single variable variable_name from pd for plotting with Plots.plot.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Base.resize! — Method
resize!(c::AbstractContainer, new_length) -> AbstractContainer

Resize c to contain new_length elements. If new_length is smaller than the current container length, the first new_length elements will be retained. If new_length is larger, the new elements are invalidated.

source
PolynomialBases.compute_coefficients — Method
compute_coefficients(func, t, semi::AbstractSemidiscretization)

Compute the discrete coefficients of the continuous function func at time t associated with the semidiscretization semi. For example, the discrete coefficients of func for a discontinuous Galerkin spectral element method (DGSEM) are the values of func at the Lobatto-Legendre nodes. Similarly, a classical finite difference method will use the values of func at the nodes of the grid assoociated with the semidiscretization semi.

For semidiscretizations semi associated with an initial condition, func can be omitted to use the given initial condition at time t.

source
PolynomialBases.integrate — Method
integrate(f, u, basis::LobattoLegendreBasis)

Map the function f to the coefficients u and integrate with respect to the quadrature rule given by basis.

source
PolynomialBases.integrate — Method
integrate([func=(u_node,equations)->u_node,] u_ode, semi::AbstractSemidiscretization; normalize=true)

Call func(u_node, equations) for each vector of nodal variables u_node in u_ode and integrate the result using a quadrature associated with the semidiscretization semi.

If normalize is true, the result is divided by the total volume of the computational domain.

source
SciMLBase.add_tstop! — Method
add_tstop!(integrator::SimpleIntegratorSSP, t)

Add a time stop during the time integration process. This function is called after the periodic SaveSolutionCallback to specify the next stop to save the solution.

source
SummationByPartsOperators.semidiscretize — Method
semidiscretize(semi::SemidiscretizationHyperbolicParabolic, tspan)

Wrap the semidiscretization semi as a split ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The parabolic right-hand side is the first function of the split ODE problem and will be used by default by the implicit part of IMEX methods from the SciML ecosystem.

source
SummationByPartsOperators.semidiscretize — Method
semidiscretize(semi::AbstractSemidiscretization, tspan, restart_file::AbstractString)

Wrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The initial condition etc. is taken from the restart_file.

source
Trixi.DGMultiBasis — Method
DGMultiBasis(element_type, polydeg; approximation_type = Polynomial(), kwargs...)

Constructs a basis for DGMulti solvers. Returns a "StartUpDG.RefElemData" object. The kwargs arguments are additional keyword arguments for RefElemData, such as quad_rule_vol. These are the same as the RefElemData_kwargs used in DGMulti. For more info, see the StartUpDG.jl docs.

source
Trixi.P4estMeshCubedSphere — Method
P4estMeshCubedSphere(trees_per_face_dimension, layers, inner_radius, thickness;
                      polydeg, RealT=Float64,
                      initial_refinement_level=0, unsaved_changes=true,
-                     p4est_partition_allow_for_coarsening=true)

Build a "Cubed Sphere" mesh as P4estMesh with 6 * trees_per_face_dimension^2 * layers trees.

The mesh will have two boundaries, :inside and :outside.

Arguments

  • trees_per_face_dimension::Integer: the number of trees in the first two local dimensions of each face.
  • layers::Integer: the number of trees in the third local dimension of each face, i.e., the number of layers of the sphere.
  • inner_radius::Integer: the inner radius of the sphere.
  • thickness::Integer: the thickness of the sphere. The outer radius will be inner_radius + thickness.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
source
Trixi.PlotData2D — Method
PlotData2D(u, semi [or mesh, equations, solver, cache];
+                     p4est_partition_allow_for_coarsening=true)

Build a "Cubed Sphere" mesh as P4estMesh with 6 * trees_per_face_dimension^2 * layers trees.

The mesh will have two boundaries, :inside and :outside.

Arguments

  • trees_per_face_dimension::Integer: the number of trees in the first two local dimensions of each face.
  • layers::Integer: the number of trees in the third local dimension of each face, i.e., the number of layers of the sphere.
  • inner_radius::Integer: the inner radius of the sphere.
  • thickness::Integer: the thickness of the sphere. The outer radius will be inner_radius + thickness.
  • polydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.
  • RealT::Type: the type that should be used for coordinates.
  • initial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.
  • unsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.
  • p4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.
source
Trixi.PlotData2D — Method
PlotData2D(u, semi [or mesh, equations, solver, cache];
            solution_variables=nothing,
            grid_lines=true, max_supported_level=11, nvisnodes=nothing,
            slice=:xy, point=(0.0, 0.0, 0.0))

Create a new PlotData2D object that can be used for visualizing 2D/3D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.

If grid_lines is true, also extract grid vertices for visualizing the mesh. The output resolution is indirectly set via max_supported_level: all data is interpolated to 2^max_supported_level uniformly distributed points in each spatial direction, also setting the maximum allowed refinement level in the solution. nvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.

When visualizing data from a three-dimensional simulation, a 2D slice is extracted for plotting. slice specifies the plane that is being sliced and may be :xy, :xz, or :yz. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0, 0.0). Both of these values are ignored when visualizing 2D data.

Experimental implementation

This is an experimental feature and may change in future releases.

Examples

julia> using Trixi, Plots
@@ -400,220 +360,198 @@
 
 julia> plot(pd["scalar"]) # To plot only a single variable
 
-julia> plot!(getmesh(pd)) # To add grid lines to the plot
source
Trixi.PlotData2D — Method
PlotData2D(sol; kwargs...)

Create a PlotData2D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.ScalarPlotData2D — Method
ScalarPlotData2D(u, semi::AbstractSemidiscretization; kwargs...)

Returns an PlotData2DTriangulated object which is used to visualize a single scalar field. u should be an array whose entries correspond to values of the scalar field at nodal points.

source
Trixi.SummaryCallback — Function
SummaryCallback()

Create and return a callback that prints a human-readable summary of the simulation setup at the beginning of a simulation and then resets the timer. When the returned callback is executed directly, the current timer values are shown.

source
Trixi.adapt_to_mesh_level! — Method
adapt_to_mesh_level!(u_ode, semi, level)
-adapt_to_mesh_level!(sol::Trixi.TrixiODESolution, level)

Like adapt_to_mesh_level, but modifies the solution and parts of the semidiscretization (mesh and caches) in place.

source
Trixi.adapt_to_mesh_level — Method
adapt_to_mesh_level(u_ode, semi, level)
-adapt_to_mesh_level(sol::Trixi.TrixiODESolution, level)

Use the regular adaptive mesh refinement routines to adaptively refine/coarsen the solution u_ode with semidiscretization semi towards a uniformly refined grid with refinement level level. The solution and semidiscretization are copied such that the original objects remain unaltered.

A convenience method accepts an ODE solution object, from which solution and semidiscretization are extracted as needed.

See also: adapt_to_mesh_level!

source
Trixi.PlotData2D — Method
PlotData2D(sol; kwargs...)

Create a PlotData2D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.ScalarPlotData2D — Method
ScalarPlotData2D(u, semi::AbstractSemidiscretization; kwargs...)

Returns an PlotData2DTriangulated object which is used to visualize a single scalar field. u should be an array whose entries correspond to values of the scalar field at nodal points.

source
Trixi.SummaryCallback — Function
SummaryCallback()

Create and return a callback that prints a human-readable summary of the simulation setup at the beginning of a simulation and then resets the timer. When the returned callback is executed directly, the current timer values are shown.

source
Trixi.adapt! — Method
Trixi.adapt!(mesh::T8codeMesh, adapt_callback; kwargs...)

Adapt a T8codeMesh according to a user-defined adapt_callback.

Arguments

  • mesh::T8codeMesh: Initialized mesh object.

  • adapt_callback: A user-defined callback which tells the adaption routines if an element should be refined, coarsened or stay unchanged.

    The expected callback signature is as follows:

    `adapt_callback(forest, ltreeid, eclass_scheme, lelemntid, elements, is_family, user_data)`
+  # Arguments
+  - `forest`: Pointer to the analyzed forest.
+  - `ltreeid`: Local index of the current tree where the analyzed elements are part of.
+  - `eclass_scheme`: Element class of `elements`.
+  - `lelemntid`: Local index of the first element in `elements`.
+  - `elements`: Array of elements. If consecutive elements form a family
+                they are passed together, otherwise `elements` consists of just one element.
+  - `is_family`: Boolean signifying if `elements` represents a family or not.
+  - `user_data`: Void pointer to some arbitrary user data. Default value is `C_NULL`.
+  # Returns
+    -1 : Coarsen family of elements.
+     0 : Stay unchanged.
+     1 : Refine element.

  • kwargs:

    • recursive = true: Adapt the forest recursively. If true the caller must ensure that the callback returns 0 for every analyzed element at some point to stop the recursion.
    • balance = true: Make sure the adapted forest is 2^(NDIMS-1):1 balanced.
    • partition = true: Partition the forest to redistribute elements evenly among MPI ranks.
    • ghost = true: Create a ghost layer for MPI data exchange.
    • user_data = C_NULL: Pointer to some arbitrary user-defined data.
source
Trixi.adapt_to_mesh_level! — Method
adapt_to_mesh_level!(u_ode, semi, level)
+adapt_to_mesh_level!(sol::Trixi.TrixiODESolution, level)

Like adapt_to_mesh_level, but modifies the solution and parts of the semidiscretization (mesh and caches) in place.

source
Trixi.adapt_to_mesh_level — Method
adapt_to_mesh_level(u_ode, semi, level)
+adapt_to_mesh_level(sol::Trixi.TrixiODESolution, level)

Use the regular adaptive mesh refinement routines to adaptively refine/coarsen the solution u_ode with semidiscretization semi towards a uniformly refined grid with refinement level level. The solution and semidiscretization are copied such that the original objects remain unaltered.

A convenience method accepts an ODE solution object, from which solution and semidiscretization are extracted as needed.

See also: adapt_to_mesh_level!

source
Trixi.balance! — Method
Trixi.balance!(mesh::T8codeMesh)

Balance a T8codeMesh to ensure 2^(NDIMS-1):1 face neighbors.

source
Trixi.boundary_condition_noslip_wall — Method
boundary_condition_noslip_wall(u_inner, orientation, direction, x, t,
                                surface_flux_function,
-                               equations::LatticeBoltzmannEquations2D)

No-slip wall boundary condition using the bounce-back approach.

source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
-                             equations::AcousticPerturbationEquations2D)

Use an orthogonal projection of the perturbed velocities to zero out the normal velocity while retaining the possibility of a tangential velocity in the boundary state. Further details are available in the paper:

  • Marcus Bauer, Jürgen Dierke and Roland Ewert (2011) Application of a discontinuous Galerkin method to discretize acoustic perturbation equations DOI: 10.2514/1.J050333
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
-                             equations::CompressibleEulerEquations2D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

Details about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761

Should be used together with UnstructuredMesh2D.

source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
-                             equations::CompressibleEulerEquations3D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

Details about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
-                             equations::ShallowWaterEquations2D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value. For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
-                             equations::ShallowWaterTwoLayerEquations2D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.

For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, orientation, direction, x, t,
-                             surface_flux_function, equations::CompressibleEulerEquations1D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

    Should be used together with TreeMesh.

source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,
-                              equations::ShallowWaterEquations1D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.

For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,
-                             equations::ShallowWaterTwoLayerEquations1D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.

For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_wall — Method
boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,
-                        equations::AcousticPerturbationEquations2D)

Boundary conditions for a solid wall.

source
Trixi.boundary_condition_wall — Method
boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,
-                            equations::LinearizedEulerEquations2D)

Boundary conditions for a solid wall.

source
Trixi.calc_error_norms — Method
calc_error_norms([func=(u_node,equations)->u_node,] u_ode, t, analyzer, semi::AbstractSemidiscretization, cache_analysis)

Calculate discrete L2 and L∞ error norms of func applied to each nodal variable u_node in u_ode. If no exact solution is available, "errors" are calculated using some reference state and can be useful for regression tests.

source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, direction, equations::IdealGlmMhdEquations1D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations2D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_wavespeed_roe — Method
calc_wavespeed_roe(u_ll, u_rr, direction::Integer,
-                   equations::ShallowWaterEquations1D)

Calculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} See for instance equation (62) in

  • Paul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044

Or equation (9.17) in this lecture notes.

source
Trixi.calc_wavespeed_roe — Method
calc_wavespeed_roe(u_ll, u_rr, direction::Integer,
-                   equations::ShallowWaterEquations2D)

Calculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} depending on direction. See for instance equation (62) in

  • Paul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044

Or this slides, slides 8 and 9.

source
Trixi.collision_bgk — Method
collision_bgk(u, dt, equations::LatticeBoltzmannEquations2D)

Collision operator for the Bhatnagar, Gross, and Krook (BGK) model.

source
Trixi.collision_bgk — Method
collision_bgk(u, dt, equations::LatticeBoltzmannEquations3D)

Collision operator for the Bhatnagar, Gross, and Krook (BGK) model.

source
Trixi.cons2cons — Method
cons2cons(u, equations)

Return the conserved variables u. While this function is as trivial as identity, it is also as useful.

source
Trixi.cons2entropy — Function
cons2entropy(u, equations)

Convert the conserved variables u to the entropy variables for a given set of equations with chosen standard entropy.

u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by entropy2cons.

source
Trixi.cons2prim — Function
cons2prim(u, equations)

Convert the conserved variables u to the primitive variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by prim2cons.

source
Trixi.convergence_test — Method
convergence_test([mod::Module=Main,] elixir::AbstractString, iterations; kwargs...)

Run iterations Trixi.jl simulations using the setup given in elixir and compute the experimental order of convergence (EOC) in the $L^2$ and $L^\infty$ norm. In each iteration, the resolution of the respective mesh will be doubled. Additional keyword arguments kwargs... and the optional module mod are passed directly to trixi_include.

This function assumes that the spatial resolution is set via the keywords initial_refinement_level (an integer) or cells_per_dimension (a tuple of integers, one per spatial dimension).

source
Trixi.default_example_unstructured — Method
default_example_unstructured()

Return the path to an example elixir that can be used to quickly see Trixi.jl in action on an UnstructuredMesh2D. This simulation is run on the example curved, unstructured mesh given in the Trixi.jl documentation regarding unstructured meshes.

source
Trixi.density — Method
density(p::Real, equations::LatticeBoltzmannEquations2D)
-density(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic density from the pressure p or the particle distribution functions u.

source
Trixi.density — Method
density(p::Real, equations::LatticeBoltzmannEquations3D)
-density(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic density from the pressure p or the particle distribution functions u.

source
Trixi.each_dof_global — Method
each_dof_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the degrees of freedom (DOF) in dg. In particular, not the DOFs themselves are returned.

source
Trixi.each_face_node — Method
each_face_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the face nodes in dg. In particular, not the face_nodes themselves are returned.

source
Trixi.each_face_node_global — Method
each_face_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the face nodes in mesh. In particular, not the face nodes themselves are returned.

source
Trixi.each_quad_node — Method
each_quad_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the quadrature nodes in dg. In particular, not the quadrature nodes themselves are returned.

source
Trixi.each_quad_node_global — Method
each_quad_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the global quadrature nodes in mesh. In particular, not the quadrature nodes themselves are returned.

source
Trixi.eachboundary — Method
eachboundary(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the boundaries in cache. In particular, not the boundaries themselves are returned.

source
Trixi.eachcomponent — Method
eachcomponent(equations::AbstractCompressibleEulerMulticomponentEquations)

Return an iterator over the indices that specify the location in relevant data structures for the components in AbstractCompressibleEulerMulticomponentEquations. In particular, not the components themselves are returned.

source
Trixi.eachcomponent — Method
eachcomponent(equations::AbstractIdealGlmMhdMulticomponentEquations)

Return an iterator over the indices that specify the location in relevant data structures for the components in AbstractIdealGlmMhdMulticomponentEquations. In particular, not the components themselves are returned.

source
Trixi.eachdim — Method
eachdim(mesh)

Return an iterator over the indices that specify the location in relevant data structures for the dimensions in AbstractTree. In particular, not the dimensions themselves are returned.

source
Trixi.eachdirection — Method
eachdirection(tree::AbstractTree)

Return an iterator over the indices that specify the location in relevant data structures for the directions in AbstractTree. In particular, not the directions themselves are returned.

source
Trixi.eachelement — Method
eachelement(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the elements in cache. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the elements in mesh. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer1D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer2D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer3D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::UnstructuredElementContainer2D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachinterface — Method
eachinterface(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the interfaces in cache. In particular, not the interfaces themselves are returned.

source
Trixi.eachmortar — Method
eachmortar(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the mortars in cache. In particular, not the mortars themselves are returned.

source
Trixi.eachmpiinterface — Method
eachmpiinterface(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the MPI interfaces in cache. In particular, not the interfaces themselves are returned.

source
Trixi.eachmpimortar — Method
eachmpimortar(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the MPI mortars in cache. In particular, not the mortars themselves are returned.

source
Trixi.eachnode — Method
eachnode(dg::DG)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in dg. In particular, not the nodes themselves are returned.

source
Trixi.eachnode — Method
eachnode(basis::LobattoLegendreBasis)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in basis. In particular, not the nodes themselves are returned.

source
Trixi.eachnode — Method
eachnode(analyzer::LobattoLegendreAnalyzer)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in analyzer. In particular, not the nodes themselves are returned.

source
Trixi.eachvariable — Method
eachvariable(equations::AbstractEquations)

Return an iterator over the indices that specify the location in relevant data structures for the variables in equations. In particular, not the variables themselves are returned.

source
Trixi.energy_internal — Function
energy_internal(u, equations)

Return the internal energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.energy_kinetic — Function
energy_kinetic(u, equations)

Return the kinetic energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.energy_total — Function
energy_total(u, equations)

Return the total energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.entropy — Function
entropy(u, equations)

Return the chosen entropy of the conserved variables u for a given set of equations.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.entropy2cons — Function
entropy2cons(w, equations)

Convert the entropy variables w based on a standard entropy to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2entropy.

source
Trixi.equilibrium_distribution — Method
equilibrium_distribution(alpha, rho, v1, v2, v3, equations::LatticeBoltzmannEquations3D)

Calculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2, v3.

source
Trixi.equilibrium_distribution — Method
equilibrium_distribution(alpha, rho, v1, v2, equations::LatticeBoltzmannEquations2D)

Calculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2.

source
Trixi.examples_dir — Method
examples_dir()

Return the directory where the example files provided with Trixi.jl are located. If Trixi.jl is installed as a regular package (with ]add Trixi), these files are read-only and should not be modified. To find out which files are available, use, e.g., readdir:

Examples

readdir(examples_dir())
source
Trixi.flux — Function
flux(u, orientation_or_normal, equations)

Given the conservative variables u, calculate the (physical) flux in Cartesian direction orientation::Integer or in arbitrary direction normal::AbstractVector for the corresponding set of governing equations. orientation is 1, 2, and 3 for the x-, y-, and z-directions, respectively.

source
Trixi.flux — Method
flux(u, normal_direction::AbstractVector, equations::AbstractEquations{1})

Enables calling flux with a non-integer argument normal_direction for one-dimensional equations. Returns the value of flux(u, 1, equations) scaled by normal_direction[1].

source
Trixi.flux_central — Method
flux_central(u_ll, u_rr, orientation_or_normal_direction, equations::AbstractEquations)

The classical central numerical flux f((u_ll) + f(u_rr)) / 2. When this flux is used as volume flux, the discretization is equivalent to the classical weak form DG method (except floating point errors).

source
Trixi.flux_chan_etal — Method

@inline function fluxchanetal(ull, urr, orientation::Integer, equations::CompressibleEulerEquationsQuasi1D)

Conservative (symmetric) part of the entropy conservative flux for quasi 1D compressible Euler equations split form. This flux is a generalization of flux_ranocha for CompressibleEulerEquations1D. Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_chan_etal — Method
flux_chan_etal(u_ll, u_rr, orientation,
-               equations::ShallowWaterEquationsQuasi1D)

Total energy conservative (mathematical entropy for quasi 1D shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., FluxPlusDissipation(flux_chan_etal, DissipationLocalLaxFriedrichs()).

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations1D)

Entropy conserving two-point flux by

  • Ayoub Gouasmi, Karthik Duraisamy (2020) "Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations" arXiv:1904.00972v3 [math.NA] 4 Feb 2020
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations2D)

Adaption of the entropy conserving two-point flux by

  • Ayoub Gouasmi, Karthik Duraisamy (2020) "Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations" arXiv:1904.00972v3 [math.NA] 4 Feb 2020
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations2D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations3D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)

Entropy conserving two-point flux adapted by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdMulticomponentEquations2D)

Entropy conserving two-point flux adapted by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_es_ersing_etal — Method
flux_es_ersing_etal(u_ll, u_rr, orientation_or_normal_direction,
-                    equations::ShallowWaterTwoLayerEquations2D)

Entropy stable surface flux for the two-layer shallow water equations. Uses the entropy conservative flux_wintermeyer_etal and adds a Lax-Friedrichs type dissipation dependent on the jump of entropy variables.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_es_ersing_etal — Method
flux_es_ersing_etal(u_ll, u_rr, orientation_or_normal_direction,
-                    equations::ShallowWaterTwoLayerEquations1D)

Entropy stable surface flux for the two-layer shallow water equations. Uses the entropy conservative flux_wintermeyer_etal and adds a Lax-Friedrichs type dissipation dependent on the jump of entropy variables.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_fjordholm_etal — Method
flux_fjordholm_etal(u_ll, u_rr, orientation,
-                    equations::ShallowWaterEquations1D)

Total energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.

Details are available in Eq. (4.1) in the paper:

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042
source
Trixi.flux_fjordholm_etal — Method
flux_fjordholm_etal(u_ll, u_rr, orientation_or_normal_direction,
-                    equations::ShallowWaterEquations2D)

Total energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.

Details are available in Eq. (4.1) in the paper:

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042
source
Trixi.flux_godunov — Method
flux_godunov(u_ll, u_rr, orientation_or_normal_direction,
-             equations::LinearizedEulerEquations2D)

An upwind flux for the linearized Euler equations based on diagonalization of the physical flux matrix. Given the physical flux $Au$, $A=T \Lambda T^{-1}$ with $\Lambda$ being a diagonal matrix that holds the eigenvalues of $A$, decompose $\Lambda = \Lambda^+ + \Lambda^-$ where $\Lambda^+$ and $\Lambda^-$ are diagonal matrices holding the positive and negative eigenvalues of $A$, respectively. Then for left and right states $u_L, u_R$, the numerical flux calculated by this function is given by $A^+ u_L + A^- u_R$ where $A^{\pm} = T \Lambda^{\pm} T^{-1}$.

The diagonalization of the flux matrix can be found in

source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
-                        equations::IdealGlmMhdEquations1D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
-                        equations::IdealGlmMhdEquations2D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
-                        equations::IdealGlmMhdEquations3D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
-                        equations::IdealGlmMhdMulticomponentEquations1D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
-                        equations::IdealGlmMhdMulticomponentEquations2D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,
-                    equations::CompressibleEulerEquations2D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,
-                    equations::CompressibleEulerEquations3D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_nonconservative_audusse_etal — Method
flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,
-                                  equations::ShallowWaterEquations1D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.

This hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations1D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.

Further details on the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
+                             equations::AcousticPerturbationEquations2D)

Use an orthogonal projection of the perturbed velocities to zero out the normal velocity while retaining the possibility of a tangential velocity in the boundary state. Further details are available in the paper:

  • Marcus Bauer, Jürgen Dierke and Roland Ewert (2011) Application of a discontinuous Galerkin method to discretize acoustic perturbation equations DOI: 10.2514/1.J050333
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
+                             equations::CompressibleEulerEquations2D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

Details about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761

Should be used together with UnstructuredMesh2D.

source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
+                             equations::CompressibleEulerEquations3D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

Details about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,
+                             equations::ShallowWaterEquations2D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value. For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, orientation, direction, x, t,
+                             surface_flux_function, equations::CompressibleEulerEquations1D)

Determine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:

  • J. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF

    Should be used together with TreeMesh.

source
Trixi.boundary_condition_slip_wall — Method
boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,
+                              equations::ShallowWaterEquations1D)

Create a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.

For details see Section 9.2.5 of the book:

  • Eleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662
source
Trixi.boundary_condition_wall — Method
boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,
+                        equations::AcousticPerturbationEquations2D)

Boundary conditions for a solid wall.

source
Trixi.boundary_condition_wall — Method
boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,
+                            equations::LinearizedEulerEquations2D)

Boundary conditions for a solid wall.

source
Trixi.calc_error_norms — Method
calc_error_norms([func=(u_node,equations)->u_node,] u_ode, t, analyzer, semi::AbstractSemidiscretization, cache_analysis)

Calculate discrete L2 and L∞ error norms of func applied to each nodal variable u_node in u_ode. If no exact solution is available, "errors" are calculated using some reference state and can be useful for regression tests.

source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, direction, equations::IdealGlmMhdEquations1D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations2D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_fast_wavespeed_roe — Method
calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)

Compute the fast magnetoacoustic wave speed using Roe averages as given by

  • Cargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773
source
Trixi.calc_wavespeed_roe — Method
calc_wavespeed_roe(u_ll, u_rr, direction::Integer,
+                   equations::ShallowWaterEquations1D)

Calculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} See for instance equation (62) in

  • Paul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044

Or equation (9.17) in this lecture notes.

source
Trixi.calc_wavespeed_roe — Method
calc_wavespeed_roe(u_ll, u_rr, direction::Integer,
+                   equations::ShallowWaterEquations2D)

Calculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} depending on direction. See for instance equation (62) in

  • Paul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044

Or this slides, slides 8 and 9.

source
Trixi.collision_bgk — Method
collision_bgk(u, dt, equations::LatticeBoltzmannEquations2D)

Collision operator for the Bhatnagar, Gross, and Krook (BGK) model.

source
Trixi.collision_bgk — Method
collision_bgk(u, dt, equations::LatticeBoltzmannEquations3D)

Collision operator for the Bhatnagar, Gross, and Krook (BGK) model.

source
Trixi.cons2cons — Method
cons2cons(u, equations)

Return the conserved variables u. While this function is as trivial as identity, it is also as useful.

source
Trixi.cons2entropy — Function
cons2entropy(u, equations)

Convert the conserved variables u to the entropy variables for a given set of equations with chosen standard entropy.

u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by entropy2cons.

source
Trixi.cons2prim — Function
cons2prim(u, equations)

Convert the conserved variables u to the primitive variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by prim2cons.

source
Trixi.convergence_test — Method
convergence_test([mod::Module=Main,] elixir::AbstractString, iterations; kwargs...)

Run iterations Trixi.jl simulations using the setup given in elixir and compute the experimental order of convergence (EOC) in the $L^2$ and $L^\infty$ norm. In each iteration, the resolution of the respective mesh will be doubled. Additional keyword arguments kwargs... and the optional module mod are passed directly to trixi_include.

This function assumes that the spatial resolution is set via the keywords initial_refinement_level (an integer) or cells_per_dimension (a tuple of integers, one per spatial dimension).

source
Trixi.default_example_unstructured — Method
default_example_unstructured()

Return the path to an example elixir that can be used to quickly see Trixi.jl in action on an UnstructuredMesh2D. This simulation is run on the example curved, unstructured mesh given in the Trixi.jl documentation regarding unstructured meshes.

source
Trixi.density — Method
density(p::Real, equations::LatticeBoltzmannEquations2D)
+density(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic density from the pressure p or the particle distribution functions u.

source
Trixi.density — Method
density(p::Real, equations::LatticeBoltzmannEquations3D)
+density(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic density from the pressure p or the particle distribution functions u.

source
Trixi.download — Method
Trixi.download(src_url, file_path)

Download a file from given src_url to given file_path if file_path is not already a file. This function just returns file_path. This is a small wrapper of Downloads.download(src_url, file_path) that avoids race conditions when multiple MPI ranks are used.

source
Trixi.each_dof_global — Method
each_dof_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the degrees of freedom (DOF) in dg. In particular, not the DOFs themselves are returned.

source
Trixi.each_face_node — Method
each_face_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the face nodes in dg. In particular, not the face_nodes themselves are returned.

source
Trixi.each_face_node_global — Method
each_face_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the face nodes in mesh. In particular, not the face nodes themselves are returned.

source
Trixi.each_quad_node — Method
each_quad_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the quadrature nodes in dg. In particular, not the quadrature nodes themselves are returned.

source
Trixi.each_quad_node_global — Method
each_quad_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the global quadrature nodes in mesh. In particular, not the quadrature nodes themselves are returned.

source
Trixi.eachboundary — Method
eachboundary(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the boundaries in cache. In particular, not the boundaries themselves are returned.

source
Trixi.eachcomponent — Method
eachcomponent(equations::AbstractCompressibleEulerMulticomponentEquations)

Return an iterator over the indices that specify the location in relevant data structures for the components in AbstractCompressibleEulerMulticomponentEquations. In particular, not the components themselves are returned.

source
Trixi.eachcomponent — Method
eachcomponent(equations::AbstractIdealGlmMhdMulticomponentEquations)

Return an iterator over the indices that specify the location in relevant data structures for the components in AbstractIdealGlmMhdMulticomponentEquations. In particular, not the components themselves are returned.

source
Trixi.eachdim — Method
eachdim(mesh)

Return an iterator over the indices that specify the location in relevant data structures for the dimensions in AbstractTree. In particular, not the dimensions themselves are returned.

source
Trixi.eachdirection — Method
eachdirection(tree::AbstractTree)

Return an iterator over the indices that specify the location in relevant data structures for the directions in AbstractTree. In particular, not the directions themselves are returned.

source
Trixi.eachelement — Method
eachelement(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the elements in cache. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(mesh::DGMultiMesh, dg::DGMulti, other_args...)

Return an iterator over the indices that specify the location in relevant data structures for the elements in mesh. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer1D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer2D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::ElementContainer3D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachelement — Method
eachelement(elements::UnstructuredElementContainer2D)

Return an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.

source
Trixi.eachinterface — Method
eachinterface(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the interfaces in cache. In particular, not the interfaces themselves are returned.

source
Trixi.eachmortar — Method
eachmortar(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the mortars in cache. In particular, not the mortars themselves are returned.

source
Trixi.eachmpiinterface — Method
eachmpiinterface(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the MPI interfaces in cache. In particular, not the interfaces themselves are returned.

source
Trixi.eachmpimortar — Method
eachmpimortar(dg::DG, cache)

Return an iterator over the indices that specify the location in relevant data structures for the MPI mortars in cache. In particular, not the mortars themselves are returned.

source
Trixi.eachnode — Method
eachnode(dg::DG)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in dg. In particular, not the nodes themselves are returned.

source
Trixi.eachnode — Method
eachnode(basis::LobattoLegendreBasis)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in basis. In particular, not the nodes themselves are returned.

source
Trixi.eachnode — Method
eachnode(analyzer::LobattoLegendreAnalyzer)

Return an iterator over the indices that specify the location in relevant data structures for the nodes in analyzer. In particular, not the nodes themselves are returned.

source
Trixi.eachvariable — Method
eachvariable(equations::AbstractEquations)

Return an iterator over the indices that specify the location in relevant data structures for the variables in equations. In particular, not the variables themselves are returned.

source
Trixi.energy_internal — Function
energy_internal(u, equations)

Return the internal energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.energy_kinetic — Function
energy_kinetic(u, equations)

Return the kinetic energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.energy_total — Function
energy_total(u, equations)

Return the total energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.entropy — Function
entropy(u, equations)

Return the chosen entropy of the conserved variables u for a given set of equations.

u is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).

source
Trixi.entropy2cons — Function
entropy2cons(w, equations)

Convert the entropy variables w based on a standard entropy to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2entropy.

source
Trixi.equilibrium_distribution — Method
equilibrium_distribution(alpha, rho, v1, v2, v3, equations::LatticeBoltzmannEquations3D)

Calculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2, v3.

source
Trixi.equilibrium_distribution — Method
equilibrium_distribution(alpha, rho, v1, v2, equations::LatticeBoltzmannEquations2D)

Calculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2.

source
Trixi.examples_dir — Method
examples_dir()

Return the directory where the example files provided with Trixi.jl are located. If Trixi.jl is installed as a regular package (with ]add Trixi), these files are read-only and should not be modified. To find out which files are available, use, e.g., readdir:

Examples

readdir(examples_dir())
source
Trixi.flux — Function
flux(u, orientation_or_normal, equations)

Given the conservative variables u, calculate the (physical) flux in Cartesian direction orientation::Integer or in arbitrary direction normal::AbstractVector for the corresponding set of governing equations. orientation is 1, 2, and 3 for the x-, y-, and z-directions, respectively.

source
Trixi.flux — Method
flux(u, normal_direction::AbstractVector, equations::AbstractEquations{1})

Enables calling flux with a non-integer argument normal_direction for one-dimensional equations. Returns the value of flux(u, 1, equations) scaled by normal_direction[1].

source
Trixi.flux_central — Method
flux_central(u_ll, u_rr, orientation_or_normal_direction, equations::AbstractEquations)

The classical central numerical flux f((u_ll) + f(u_rr)) / 2. When this flux is used as volume flux, the discretization is equivalent to the classical weak form DG method (except floating point errors).

source
Trixi.flux_chan_etal — Method

@inline function fluxchanetal(ull, urr, orientation::Integer, equations::CompressibleEulerEquationsQuasi1D)

Conservative (symmetric) part of the entropy conservative flux for quasi 1D compressible Euler equations split form. This flux is a generalization of flux_ranocha for CompressibleEulerEquations1D. Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_chan_etal — Method
flux_chan_etal(u_ll, u_rr, orientation,
+               equations::ShallowWaterEquationsQuasi1D)

Total energy conservative (mathematical entropy for quasi 1D shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., FluxPlusDissipation(flux_chan_etal, DissipationLocalLaxFriedrichs()).

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)

Entropy conserving two-point flux by

  • Chandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations1D)

Entropy conserving two-point flux by

  • Ayoub Gouasmi, Karthik Duraisamy (2020) "Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations" arXiv:1904.00972v3 [math.NA] 4 Feb 2020
source
Trixi.flux_chandrashekar — Method
flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations2D)

Adaption of the entropy conserving two-point flux by

  • Ayoub Gouasmi, Karthik Duraisamy (2020) "Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations" arXiv:1904.00972v3 [math.NA] 4 Feb 2020
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations2D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations3D)

Entropy conserving two-point flux by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)

Entropy conserving two-point flux adapted by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_derigs_etal — Method
flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdMulticomponentEquations2D)

Entropy conserving two-point flux adapted by

  • Derigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002
source
Trixi.flux_fjordholm_etal — Method
flux_fjordholm_etal(u_ll, u_rr, orientation,
+                    equations::ShallowWaterEquations1D)

Total energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.

Details are available in Eq. (4.1) in the paper:

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042
source
Trixi.flux_fjordholm_etal — Method
flux_fjordholm_etal(u_ll, u_rr, orientation_or_normal_direction,
+                    equations::ShallowWaterEquations2D)

Total energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.

Details are available in Eq. (4.1) in the paper:

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042
source
Trixi.flux_godunov — Method
flux_godunov(u_ll, u_rr, orientation_or_normal_direction,
+             equations::LinearizedEulerEquations2D)

An upwind flux for the linearized Euler equations based on diagonalization of the physical flux matrix. Given the physical flux $Au$, $A=T \Lambda T^{-1}$ with $\Lambda$ being a diagonal matrix that holds the eigenvalues of $A$, decompose $\Lambda = \Lambda^+ + \Lambda^-$ where $\Lambda^+$ and $\Lambda^-$ are diagonal matrices holding the positive and negative eigenvalues of $A$, respectively. Then for left and right states $u_L, u_R$, the numerical flux calculated by this function is given by $A^+ u_L + A^- u_R$ where $A^{\pm} = T \Lambda^{\pm} T^{-1}$.

The diagonalization of the flux matrix can be found in

source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
+                        equations::IdealGlmMhdEquations1D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
+                        equations::IdealGlmMhdEquations2D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
+                        equations::IdealGlmMhdEquations3D)

Entropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
+                        equations::IdealGlmMhdMulticomponentEquations1D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_hindenlang_gassner — Method
flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,
+                        equations::IdealGlmMhdMulticomponentEquations2D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.

References

  • Florian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications
  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig
  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,
+                    equations::CompressibleEulerEquations2D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_kennedy_gruber — Method
flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,
+                    equations::CompressibleEulerEquations3D)

Kinetic energy preserving two-point flux by

  • Kennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020
source
Trixi.flux_nonconservative_audusse_etal — Method
flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,
+                                  equations::ShallowWaterEquations1D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.

This hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations1D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.

Further details on the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.flux_nonconservative_audusse_etal — Method
flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,
                                   equations::ShallowWaterEquations2D)
 flux_nonconservative_audusse_etal(u_ll, u_rr,
                                   normal_direction_ll     ::AbstractVector,
                                   normal_direction_average::AbstractVector,
-                                  equations::ShallowWaterEquations2D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.

This hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations2D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.

Further details for the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.flux_nonconservative_chan_etal — Method
flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,
+                                  equations::ShallowWaterEquations2D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.

This hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations2D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.

Further details for the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.flux_nonconservative_chan_etal — Method
flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,
                                equations::CompressibleEulerEquationsQuasi1D)
 flux_nonconservative_chan_etal(u_ll, u_rr, normal_direction, 
                                equations::CompressibleEulerEquationsQuasi1D)
 flux_nonconservative_chan_etal(u_ll, u_rr, normal_ll, normal_rr,
-                               equations::CompressibleEulerEquationsQuasi1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the pressure CompressibleEulerEquationsQuasi1D and the nozzle width.

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_nonconservative_chan_etal — Method
flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,
+                               equations::CompressibleEulerEquationsQuasi1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the pressure CompressibleEulerEquationsQuasi1D and the nozzle width.

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_nonconservative_chan_etal — Method
flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,
                                equations::ShallowWaterEquationsQuasi1D)
 flux_nonconservative_chan_etal(u_ll, u_rr, normal_direction::AbstractVector,
                                equations::ShallowWaterEquationsQuasi1D)    
 flux_nonconservative_chan_etal(u_ll, u_rr, 
                                normal_ll::AbstractVector, normal_rr::AbstractVector,
-                               equations::ShallowWaterEquationsQuasi1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquationsQuasi1D and the channel width.

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_nonconservative_chen_noelle — Method
flux_nonconservative_chen_noelle(u_ll, u_rr,
-                                 orientation::Integer,
-                                 equations::ShallowWaterEquations1D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_chen_noelle on the conservative variables.

Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_chen_noelle in the surface flux to ensure consistency.

Further details on the hydrostatic reconstruction and its motivation can be found in

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.flux_nonconservative_chen_noelle — Method
flux_nonconservative_chen_noelle(u_ll, u_rr,
-                                 orientation::Integer,
-                                 equations::ShallowWaterEquations2D)
-flux_nonconservative_chen_noelle(u_ll, u_rr,
-                                 normal_direction_ll      ::AbstractVector,
-                                 normal_direction_average ::AbstractVector,
-                                 equations::ShallowWaterEquations2D)

Non-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_chen_noelle on the conservative variables.

Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_chen_noelle in the surface flux to ensure consistency.

Further details on the hydrostatic reconstruction and its motivation can be found in

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
-                                 equations::ShallowWaterEquations1D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
+                               equations::ShallowWaterEquationsQuasi1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquationsQuasi1D and the channel width.

Further details are available in the paper:

  • Jesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
+                                 equations::ShallowWaterEquations1D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
                                  equations::ShallowWaterEquations2D)
 flux_nonconservative_ersing_etal(u_ll, u_rr,
                                  normal_direction_ll::AbstractVector,
                                  normal_direction_average::AbstractVector,
-                                 equations::ShallowWaterEquations2D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
-                                 equations::ShallowWaterTwoLayerEquations1D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterTwoLayerEquations1D and an additional term that couples the momentum of both layers.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_ersing_etal — Method
flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,
-                                 equations::ShallowWaterTwoLayerEquations2D)
-flux_nonconservative_ersing_etal(u_ll, u_rr,
-                                 normal_direction_ll::AbstractVector,
-                                 normal_direction_average::AbstractVector,
-                                 equations::ShallowWaterTwoLayerEquations2D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterTwoLayerEquations2D and an additional term that couples the momentum of both layers.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_fjordholm_etal — Method
flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,
-                                    equations::ShallowWaterEquations1D)

Non-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations1D.

This contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.

Further details for the original finite volume formulation are available in

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042

and for curvilinear 2D case in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_fjordholm_etal — Method
flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,
+                                 equations::ShallowWaterEquations2D)
Experimental code

This numerical flux is experimental and may change in any future release.

Non-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

This is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.

For further details see:

  • Patrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699
source
Trixi.flux_nonconservative_fjordholm_etal — Method
flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,
+                                    equations::ShallowWaterEquations1D)

Non-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations1D.

This contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.

Further details for the original finite volume formulation are available in

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042

and for curvilinear 2D case in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_fjordholm_etal — Method
flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,
                                     equations::ShallowWaterEquations2D)
 flux_nonconservative_fjordholm_etal(u_ll, u_rr,
                                     normal_direction_ll     ::AbstractVector,
                                     normal_direction_average::AbstractVector,
-                                    equations::ShallowWaterEquations2D)

Non-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

This contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.

Further details for the original finite volume formulation are available in

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042

and for curvilinear 2D case in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
+                                    equations::ShallowWaterEquations2D)

Non-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

This contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.

Further details for the original finite volume formulation are available in

  • Ulrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042

and for curvilinear 2D case in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
                             equations::IdealGlmMhdEquations2D)
 flux_nonconservative_powell(u_ll, u_rr,
                             normal_direction_ll     ::AbstractVector,
                             normal_direction_average::AbstractVector,
-                            equations::IdealGlmMhdEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
+                            equations::IdealGlmMhdEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
                             equations::IdealGlmMhdEquations3D)
 flux_nonconservative_powell(u_ll, u_rr,
                             normal_direction_ll     ::AbstractVector,
                             normal_direction_average::AbstractVector,
-                            equations::IdealGlmMhdEquations3D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations3D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
-                            equations::IdealGlmMhdMulticomponentEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdMulticomponentEquations2D.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,
+                            equations::IdealGlmMhdEquations3D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations3D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell — Method
flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,
+                            equations::IdealGlmMhdMulticomponentEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdMulticomponentEquations2D.

References

  • Marvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027
source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,
                                             equations::IdealGlmMhdEquations2D,
                                             nonconservative_type::NonConservativeSymmetric,
-                                            nonconservative_term::Integer)

Symmetric part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered "fluxes" for subcell limiting. See, e.g.,

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.

This function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.

source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, u_rr,
+                                            nonconservative_term::Integer)

Symmetric part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered "fluxes" for subcell limiting. See, e.g.,

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.

This function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.

source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, u_rr,
                                             orientation::Integer,
-                                            equations::IdealGlmMhdEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.

This implementation uses a non-conservative term that can be written as the product of local and symmetric parts. It is equivalent to the non-conservative flux of Bohm et al. (flux_nonconservative_powell) for conforming meshes but it yields different results on non-conforming meshes(!).

The two other flux functions with the same name return either the local or symmetric portion of the non-conservative flux based on the type of the nonconservativetype argument, employing multiple dispatch. They are used to compute the subcell fluxes in dg2dsubcelllimiters.jl.

References

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.
source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,
+                                            equations::IdealGlmMhdEquations2D)

Non-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.

This implementation uses a non-conservative term that can be written as the product of local and symmetric parts. It is equivalent to the non-conservative flux of Bohm et al. (flux_nonconservative_powell) for conforming meshes but it yields different results on non-conforming meshes(!).

The two other flux functions with the same name return either the local or symmetric portion of the non-conservative flux based on the type of the nonconservativetype argument, employing multiple dispatch. They are used to compute the subcell fluxes in dg2dsubcelllimiters.jl.

References

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.
source
Trixi.flux_nonconservative_powell_local_symmetric — Method
flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,
                                             equations::IdealGlmMhdEquations2D,
                                             nonconservative_type::NonConservativeLocal,
-                                            nonconservative_term::Integer)

Local part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered "fluxes" for subcell limiting. See, e.g.,

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.

This function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.

source
Trixi.flux_nonconservative_wintermeyer_etal — Method
flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,
-                                      equations::ShallowWaterEquations1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.

Further details are available in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_wintermeyer_etal — Method
flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,
+                                            nonconservative_term::Integer)

Local part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered "fluxes" for subcell limiting. See, e.g.,

  • Rueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.

This function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.

source
Trixi.flux_nonconservative_wintermeyer_etal — Method
flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,
+                                      equations::ShallowWaterEquations1D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.

Further details are available in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_nonconservative_wintermeyer_etal — Method
flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,
                                       equations::ShallowWaterEquations2D)
 flux_nonconservative_wintermeyer_etal(u_ll, u_rr,
                                       normal_direction_ll     ::AbstractVector,
                                       normal_direction_average::AbstractVector,
-                                      equations::ShallowWaterEquations2D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

Further details are available in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations1D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
-             equations::CompressibleEulerEquations2D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
-             equations::CompressibleEulerEquations3D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
-             equations::CompressibleEulerMulticomponentEquations1D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
-             equations::CompressibleEulerMulticomponentEquations2D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,
-                equations::CompressibleEulerEquations2D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,
-                equations::CompressibleEulerEquations3D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation,
-                      equations::ShallowWaterEquations1D)

Total energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation_or_normal_direction,
-                      equations::ShallowWaterEquations2D)

Total energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation,
-                      equations::ShallowWaterTwoLayerEquations1D)

Total energy conservative (mathematical entropy for two-layer shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used with the nonconservative flux_nonconservative_ersing_etal. To obtain the flux for the two-layer shallow water equations the flux that is described in the paper for the normal shallow water equations is used within each layer.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation,
-                      equations::ShallowWaterTwoLayerEquations2D)
-flux_wintermeyer_etal(u_ll, u_rr,
-                      normal_direction::AbstractVector,
-                      equations::ShallowWaterTwoLayerEquations2D)

Total energy conservative (mathematical entropy for two-layer shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used with the nonconservative flux_nonconservative_ersing_etal. To obtain the flux for the two-layer shallow water equations the flux that is described in the paper for the normal shallow water equations is used within each layer.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_winters_etal — Method
flux_winters_etal(u_ll, u_rr, orientation_or_normal_direction,
-                  equations::PolytropicEulerEquations2D)

Entropy conserving two-point flux for isothermal or polytropic gases. Requires a special weighted Stolarsky mean for the evaluation of the density denoted here as stolarsky_mean. Note, for isothermal gases where gamma = 1 this stolarsky_mean becomes the ln_mean.

For details see Section 3.2 of the following reference

  • Andrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w
source
Trixi.get_boundary_outer_state — Method
get_boundary_outer_state(boundary_condition::BoundaryConditionDirichlet,
-                         cache, t, equations, dg, indices...)

For subcell limiting, the calculation of local bounds for non-periodic domains require the boundary outer state. This function returns the boundary value at time t and for node with spatial indices indices.

source
Trixi.get_name — Method
get_name(x)

Returns a name of x ready for pretty printing. By default, return string(y) if x isa Val{y} and return string(x) otherwise.

Examples

julia> Trixi.get_name("test")
+                                      equations::ShallowWaterEquations2D)

Non-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.

On curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.

Further details are available in the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations1D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerEquations2D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerEquations3D)

Entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerMulticomponentEquations1D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_ranocha — Method
flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,
+             equations::CompressibleEulerMulticomponentEquations2D)

Adaption of the entropy conserving and kinetic energy preserving two-point flux by

  • Hendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig

See also

  • Hendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,
+                equations::CompressibleEulerEquations2D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_shima_etal — Method
flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,
+                equations::CompressibleEulerEquations3D)

This flux is is a modification of the original kinetic energy preserving two-point flux by

  • Yuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058

The modification is in the energy flux to guarantee pressure equilibrium and was developed by

  • Nao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation,
+                      equations::ShallowWaterEquations1D)

Total energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_wintermeyer_etal — Method
flux_wintermeyer_etal(u_ll, u_rr, orientation_or_normal_direction,
+                      equations::ShallowWaterEquations2D)

Total energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.

Further details are available in Theorem 1 of the paper:

  • Niklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036
source
Trixi.flux_winters_etal — Method
flux_winters_etal(u_ll, u_rr, orientation_or_normal_direction,
+                  equations::PolytropicEulerEquations2D)

Entropy conserving two-point flux for isothermal or polytropic gases. Requires a special weighted Stolarsky mean for the evaluation of the density denoted here as stolarsky_mean. Note, for isothermal gases where gamma = 1 this stolarsky_mean becomes the ln_mean.

For details see Section 3.2 of the following reference

  • Andrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w
source
Trixi.get_boundary_outer_state — Method
get_boundary_outer_state(boundary_condition::BoundaryConditionDirichlet,
+                         cache, t, equations, dg, indices...)

For subcell limiting, the calculation of local bounds for non-periodic domains require the boundary outer state. This function returns the boundary value at time t and for node with spatial indices indices.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.get_name — Method
get_name(x)

Returns a name of x ready for pretty printing. By default, return string(y) if x isa Val{y} and return string(x) otherwise.

Examples

julia> Trixi.get_name("test")
 "test"
 
 julia> Trixi.get_name(Val(:test))
-"test"
source
Trixi.get_name — Method
get_name(equations::AbstractEquations)

Returns the canonical, human-readable name for the given system of equations.

Examples

julia> Trixi.get_name(CompressibleEulerEquations1D(1.4))
-"CompressibleEulerEquations1D"
source
Trixi.getmesh — Method
getmesh(pd::AbstractPlotData)

Extract grid lines from pd for plotting with Plots.plot.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.global_mean_vars — Method
global_mean_vars(equations::AcousticPerturbationEquations2D)

Returns the global mean variables stored in equations. This makes it easier to define flexible initial conditions for problems with constant mean flow.

source
Trixi.have_nonconservative_terms — Method
have_nonconservative_terms(equations)

Trait function determining whether equations represent a conservation law with or without nonconservative terms. Classical conservation laws such as the CompressibleEulerEquations2D do not have nonconservative terms. The ShallowWaterEquations2D with non-constant bottom topography are an example of equations with nonconservative terms. The return value will be True() or False() to allow dispatching on the return type.

source
Trixi.hydrostatic_reconstruction_audusse_etal — Method
hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation::Integer,
-                                        equations::ShallowWaterEquations1D)

A particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details on this hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.hydrostatic_reconstruction_audusse_etal — Method
hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation_or_normal_direction,
-                                        equations::ShallowWaterEquations2D)

A particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details for the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.hydrostatic_reconstruction_chen_noelle — Method
hydrostatic_reconstruction_chen_noelle(u_ll, u_rr, orientation::Integer,
-                                       equations::ShallowWaterEquations1D)

A particular type of hydrostatic reconstruction of the water height to guarantee well-balancedness for a general bottom topography of the ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. The key idea is a linear reconstruction of the bottom and water height at the interfaces using subcells. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details on this hydrostatic reconstruction and its motivation can be found in

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.hydrostatic_reconstruction_chen_noelle — Method
hydrostatic_reconstruction_chen_noelle(u_ll, u_rr, orientation::Integer,
-                                       equations::ShallowWaterEquations2D)

A particular type of hydrostatic reconstruction of the water height to guarantee well-balancedness for a general bottom topography of the ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. The key idea is a linear reconstruction of the bottom and water height at the interfaces using subcells. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details on this hydrostatic reconstruction and its motivation can be found in

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.init_mpi — Method
init_mpi()

Initialize MPI by calling MPI.Initialized(). The function will check if MPI is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.init_p4est — Method
init_p4est()

Initialize p4est by calling p4est_init and setting the log level to SC_LP_ERROR. This function will check if p4est is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.init_t8code — Method
init_t8code()

Initialize t8code by calling sc_init, p4est_init, and t8_init while setting the log level to SC_LP_ERROR. This function will check if t8code is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.initial_condition_constant — Method
initial_condition_constant(x, t, equations::AcousticPerturbationEquations2D)

A constant initial condition where the state variables are zero and the mean flow is constant. Uses the global mean values from equations.

source
Trixi.initial_condition_density_wave — Method
initial_condition_density_wave(x, t, equations::CompressibleEulerEquations1D)

A sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper

  • Gregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026

with the following parameters

  • domain [-1, 1]
  • mesh = 4x4
  • polydeg = 5
source
Trixi.initial_condition_density_wave — Method
initial_condition_density_wave(x, t, equations::CompressibleEulerEquations2D)

A sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper

  • Gregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026

with the following parameters

  • domain [-1, 1]
  • mesh = 4x4
  • polydeg = 5
source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations1D)

One dimensional variant of the setup used for convergence tests of the Euler equations with self-gravity from

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593
Note

There is no additional source term necessary for the manufactured solution in one spatial dimension. Thus, source_terms_eoc_test_coupled_euler_gravity is not present there.

source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations2D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.

source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.

source
Trixi.initial_condition_gauss — Method
initial_condition_gauss(x, t, equations::AcousticPerturbationEquations2D)

A Gaussian pulse in a constant mean flow. Uses the global mean values from equations.

source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations1D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations2D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations3D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations1D)

A for multicomponent adapted weak blast wave adapted to multicomponent and taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations2D)

A for multicomponent adapted weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations1D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations3D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations1D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::PolytropicEulerEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations1D)

A weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.

source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations2D)

A weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.

source
Trixi.integrate_via_indices — Method
integrate_via_indices(func, u_ode, semi::AbstractSemidiscretization, args...; normalize=true)

Call func(u, i..., element, equations, solver, args...) for all nodal indices i..., element and integrate the result using a quadrature associated with the semidiscretization semi.

If normalize is true, the result is divided by the total volume of the computational domain.

source
Trixi.inv_ln_mean — Method
inv_ln_mean(x, y)

Compute the inverse 1 / ln_mean(x, y) of the logarithmic mean ln_mean.

This function may be used to increase performance where the inverse of the logarithmic mean is needed, by replacing a (slow) division by a (fast) multiplication.

source
Trixi.get_name — Method
get_name(equations::AbstractEquations)

Returns the canonical, human-readable name for the given system of equations.

Examples

julia> Trixi.get_name(CompressibleEulerEquations1D(1.4))
+"CompressibleEulerEquations1D"
source
Trixi.getmesh — Method
getmesh(pd::AbstractPlotData)

Extract grid lines from pd for plotting with Plots.plot.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.global_mean_vars — Method
global_mean_vars(equations::AcousticPerturbationEquations2D)

Returns the global mean variables stored in equations. This makes it easier to define flexible initial conditions for problems with constant mean flow.

source
Trixi.have_nonconservative_terms — Method
have_nonconservative_terms(equations)

Trait function determining whether equations represent a conservation law with or without nonconservative terms. Classical conservation laws such as the CompressibleEulerEquations2D do not have nonconservative terms. The ShallowWaterEquations2D with non-constant bottom topography are an example of equations with nonconservative terms. The return value will be True() or False() to allow dispatching on the return type.

source
Trixi.hydrostatic_reconstruction_audusse_etal — Method
hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation::Integer,
+                                        equations::ShallowWaterEquations1D)

A particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details on this hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.hydrostatic_reconstruction_audusse_etal — Method
hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation_or_normal_direction,
+                                        equations::ShallowWaterEquations2D)

A particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.

Further details for the hydrostatic reconstruction and its motivation can be found in

  • Emmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090
source
Trixi.init_mpi — Method
init_mpi()

Initialize MPI by calling MPI.Initialized(). The function will check if MPI is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.init_p4est — Method
init_p4est()

Initialize p4est by calling p4est_init and setting the log level to SC_LP_ERROR. This function will check if p4est is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.init_t8code — Method
init_t8code()

Initialize t8code by calling sc_init, p4est_init, and t8_init while setting the log level to SC_LP_ERROR. This function will check if t8code is already initialized and if yes, do nothing, thus it is safe to call it multiple times.

source
Trixi.initial_condition_constant — Method
initial_condition_constant(x, t, equations::AcousticPerturbationEquations2D)

A constant initial condition where the state variables are zero and the mean flow is constant. Uses the global mean values from equations.

source
Trixi.initial_condition_density_wave — Method
initial_condition_density_wave(x, t, equations::CompressibleEulerEquations1D)

A sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper

  • Gregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026

with the following parameters

  • domain [-1, 1]
  • mesh = 4x4
  • polydeg = 5
source
Trixi.initial_condition_density_wave — Method
initial_condition_density_wave(x, t, equations::CompressibleEulerEquations2D)

A sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper

  • Gregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026

with the following parameters

  • domain [-1, 1]
  • mesh = 4x4
  • polydeg = 5
source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations1D)

One dimensional variant of the setup used for convergence tests of the Euler equations with self-gravity from

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593
Note

There is no additional source term necessary for the manufactured solution in one spatial dimension. Thus, source_terms_eoc_test_coupled_euler_gravity is not present there.

source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations2D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.

source
Trixi.initial_condition_eoc_test_coupled_euler_gravity — Method
initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.

source
Trixi.initial_condition_gauss — Method
initial_condition_gauss(x, t, equations::AcousticPerturbationEquations2D)

A Gaussian pulse in a constant mean flow. Uses the global mean values from equations.

source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations1D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations2D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations3D)

A weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations1D)

A for multicomponent adapted weak blast wave adapted to multicomponent and taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations2D)

A for multicomponent adapted weak blast wave taken from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations1D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations3D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations1D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::PolytropicEulerEquations2D)

A weak blast wave adapted from

  • Sebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044
source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations1D)

A weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.

source
Trixi.initial_condition_weak_blast_wave — Method
initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations2D)

A weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.

source
Trixi.integrate_via_indices — Method
integrate_via_indices(func, u_ode, semi::AbstractSemidiscretization, args...; normalize=true)

Call func(u, i..., element, equations, solver, args...) for all nodal indices i..., element and integrate the result using a quadrature associated with the semidiscretization semi.

If normalize is true, the result is divided by the total volume of the computational domain.

source
Trixi.inv_ln_mean — Method
inv_ln_mean(x, y)

Compute the inverse 1 / ln_mean(x, y) of the logarithmic mean ln_mean.

This function may be used to increase performance where the inverse of the logarithmic mean is needed, by replacing a (slow) division by a (fast) multiplication.

source
Trixi.jacobian_ad_forward — Method
jacobian_ad_forward(semi::AbstractSemidiscretization;
                     t0=zero(real(semi)),
-                    u0_ode=compute_coefficients(t0, semi))

Uses the right-hand side operator of the semidiscretization semi and forward mode automatic differentiation to compute the Jacobian J of the semidiscretization semi at state u0_ode.

source
Trixi.jacobian_fd — Method
jacobian_fd(semi::AbstractSemidiscretization;
+                    u0_ode=compute_coefficients(t0, semi))

Uses the right-hand side operator of the semidiscretization semi and forward mode automatic differentiation to compute the Jacobian J of the semidiscretization semi at state u0_ode.

source
Trixi.jacobian_fd — Method
jacobian_fd(semi::AbstractSemidiscretization;
             t0=zero(real(semi)),
-            u0_ode=compute_coefficients(t0, semi))

Uses the right-hand side operator of the semidiscretization semi and simple second order finite difference to compute the Jacobian J of the semidiscretization semi at state u0_ode.

source
Trixi.linear_structure — Method
linear_structure(semi::AbstractSemidiscretization;
-                 t0=zero(real(semi)))

Wraps the right-hand side operator of the semidiscretization semi at time t0 as an affine-linear operator given by a linear operator A and a vector b.

source
Trixi.ln_mean — Method
ln_mean(x, y)

Compute the logarithmic mean

ln_mean(x, y) = (y - x) / (log(y) - log(x)) = (y - x) / log(y / x)

Problem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Ismail and Roe (2009). Set ξ = y / x. Then, we have

(y - x) / log(y / x) = (x + y) / log(ξ) * (ξ - 1) / (ξ + 1)

Set f = (ξ - 1) / (ξ + 1) = (y - x) / (x + y). Then, we use the expansion

log(ξ) = 2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7) + O(ξ^9)

Inserting the first few terms of this expansion yields

(y - x) / log(ξ) ≈ (x + y) * f / (2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7))
+            u0_ode=compute_coefficients(t0, semi))

Uses the right-hand side operator of the semidiscretization semi and simple second order finite difference to compute the Jacobian J of the semidiscretization semi at state u0_ode.

source
Trixi.linear_structure — Method
linear_structure(semi::AbstractSemidiscretization;
+                 t0=zero(real(semi)))

Wraps the right-hand side operator of the semidiscretization semi at time t0 as an affine-linear operator given by a linear operator A and a vector b.

source
Trixi.ln_mean — Method
ln_mean(x, y)

Compute the logarithmic mean

ln_mean(x, y) = (y - x) / (log(y) - log(x)) = (y - x) / log(y / x)

Problem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Ismail and Roe (2009). Set ξ = y / x. Then, we have

(y - x) / log(y / x) = (x + y) / log(ξ) * (ξ - 1) / (ξ + 1)

Set f = (ξ - 1) / (ξ + 1) = (y - x) / (x + y). Then, we use the expansion

log(ξ) = 2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7) + O(ξ^9)

Inserting the first few terms of this expansion yields

(y - x) / log(ξ) ≈ (x + y) * f / (2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7))
                  = (x + y) / (2 + 2/3 * f^2 + 2/5 * f^4 + 2/7 * f^6)

Since divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use

f^2 = (y - x)^2 / (x + y)^2
     = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)

Given ε = 1.0e-4, we use the following algorithm.

if f^2 < ε
   # use the expansion above
 else
   # use the direct formula (y - x) / log(y / x)
-end

References

source
Trixi.load_adaptive_time_integrator! — Method
load_adaptive_time_integrator!(integrator, restart_file::AbstractString)

Load the context information for time integrators with error-based step size control saved in a restart_file.

source
Trixi.load_dt — Method
load_dt(restart_file::AbstractString)

Load the time step size (dt in OrdinaryDiffEq.jl) saved in a restart_file.

source
Trixi.load_mesh — Method
load_mesh(restart_file::AbstractString; n_cells_max)

Load the mesh from the restart_file.

source
Trixi.load_time — Method
load_time(restart_file::AbstractString)

Load the time saved in a restart_file.

source
Trixi.load_timestep! — Method
load_timestep!(integrator, restart_file::AbstractString)

Load the time step number saved in a restart_file and assign it to both the time step number and and the number of accepted steps (iter and stats.naccept in OrdinaryDiffEq.jl, respectively) in integrator.

source
Trixi.load_timestep — Method
load_timestep(restart_file::AbstractString)

Load the time step number (iter in OrdinaryDiffEq.jl) saved in a restart_file.

source
Trixi.max — Method
max(x, y, ...)

Return the maximum of the arguments. See also the maximum function to take the maximum element from a collection.

This version in Trixi.jl is semantically equivalent to Base.max but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).

Examples

julia> max(2, 5, 1)
-5
source
Trixi.max_abs_speed_naive — Function
max_abs_speed_naive(u_ll, u_rr, orientation::Integer,   equations)
-max_abs_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimate of the maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, based only on the local wave speeds associated to u_ll and u_rr.

For non-integer arguments normal_direction in one dimension, max_abs_speed_naive returns abs(normal_direction[1]) * max_abs_speed_naive(u_ll, u_rr, 1, equations).

source
Trixi.min — Method
min(x, y, ...)

Return the minimum of the arguments. See also the minimum function to take the minimum element from a collection.

This version in Trixi.jl is semantically equivalent to Base.min but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).

Examples

julia> min(2, 5, 1)
-1
source
Trixi.min_max_speed_chen_noelle — Method
min_max_speed_chen_noelle(u_ll, u_rr, orientation::Integer,
-                          equations::ShallowWaterEquations1D)

The approximated speeds for the HLL type numerical flux used by Chen and Noelle for their hydrostatic reconstruction. As they state in the paper, these speeds are chosen for the numerical flux to ensure positivity and to satisfy an entropy inequality.

Further details on this hydrostatic reconstruction and its motivation can be found in

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.min_max_speed_chen_noelle — Method
min_max_speed_chen_noelle(u_ll, u_rr, orientation::Integer,
-                          equations::ShallowWaterEquations2D)
-min_max_speed_chen_noelle(u_ll, u_rr, normal_direction::AbstractVector,
-                          equations::ShallowWaterEquations2D)

Special estimate of the minimal and maximal wave speed of the shallow water equations for the left and right states u_ll, u_rr. These approximate speeds are used for the HLL-type numerical flux flux_hll_chen_noelle. These wave speed estimates together with a particular hydrostatic reconstruction technique guarantee that the numerical flux is positive and satisfies an entropy inequality.

Further details on this hydrostatic reconstruction and its motivation can be found in the reference below. The definition of the wave speeds are given in Equation (2.20).

  • Guoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074
source
Trixi.min_max_speed_davis — Function
min_max_speed_davis(u_ll, u_rr, orientation::Integer, equations)
-min_max_speed_davis(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimates of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.

See also FluxHLL, min_max_speed_naive, min_max_speed_einfeldt.

source
Trixi.min_max_speed_einfeldt — Function
min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations)
-min_max_speed_einfeldt(u_ll, u_rr, normal_direction::AbstractVector, equations)

More advanced mininmal and maximal wave speed computation based on

originally developed for the compressible Euler equations. A compact representation can be found in this lecture notes, eq. (9.28).

See also FluxHLL, min_max_speed_naive, min_max_speed_davis.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations2D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations3D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

Original publication:

Compactly summarized:

  • Siddhartha Mishra, Ulrik Skre Fjordholm and Rémi Abgrall Numerical methods for conservation laws and related equations. Link
source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_naive — Function
min_max_speed_naive(u_ll, u_rr, orientation::Integer, equations)
-min_max_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimate(!) of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.

  • Amiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002

See also FluxHLL, min_max_speed_davis, min_max_speed_einfeldt.

source
Trixi.modify_dt_for_tstops! — Method
modify_dt_for_tstops!(integrator::SimpleIntegratorSSP)

Modify the time-step size to match the time stops specified in integrator.opts.tstops. To avoid adding OrdinaryDiffEq to Trixi's dependencies, this routine is a copy of https://github.com/SciML/OrdinaryDiffEq.jl/blob/d76335281c540ee5a6d1bd8bb634713e004f62ee/src/integrators/integrator_utils.jl#L38-L54

source
Trixi.multiply_dimensionwise — Method
multiply_dimensionwise(matrix::AbstractMatrix, data_in::AbstractArray{<:Any, NDIMS+1})

Multiply the array data_in by matrix in each coordinate direction, where data_in is assumed to have the first coordinate for the number of variables and the remaining coordinates are multiplied by matrix.

source
Trixi.n_nonconservative_terms — Function
n_nonconservative_terms(equations)

Number of nonconservative terms in the form local * symmetric for a particular equation. This function needs to be specialized only if equations with nonconservative terms are combined with certain solvers (e.g., subcell limiting).

source
Trixi.ndofs — Method
ndofs(semi::AbstractSemidiscretization)

Return the number of degrees of freedom associated with each scalar variable.

source
Trixi.negative_part — Method
negative_part(x)

Return x if x is negative, else zero. In other words, return (x - abs(x)) / 2 for real numbers x.

source
Trixi.ode_default_options — Method
ode_default_options()

Return the default options for OrdinaryDiffEq's solve. Pass ode_default_options()... to solve to only return the solution at the final time and enable MPI aware error-based step size control, whenever MPI is used. For example, use solve(ode, alg; ode_default_options()...).

source
Trixi.ode_norm — Method
ode_norm(u, t)

Implementation of the weighted L2 norm of Hairer and Wanner used for error-based step size control in OrdinaryDiffEq.jl. This function is aware of MPI and uses global MPI communication when running in parallel.

You must pass this function as a keyword argument internalnorm=ode_norm to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.

See the "Advanced Adaptive Stepsize Control" section of the documentation.

source
Trixi.ode_unstable_check — Method
ode_unstable_check(dt, u, semi, t)

Implementation of the basic check for instability used in OrdinaryDiffEq.jl. Instead of checking something like any(isnan, u), this function just checks isnan(dt). This helps when using MPI parallelization, since no additional global communication is required and all ranks will return the same result.

You should pass this function as a keyword argument unstable_check=ode_unstable_check to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.

See the "Miscellaneous" section of the documentation.

source
Trixi.partition! — Method
partition!(mesh::ParallelTreeMesh, allow_coarsening=true)

Partition mesh using a static domain decomposition algorithm based on leaf cell count and tree structure. If allow_coarsening is true, the algorithm will keep leaf cells together on one rank when needed for local coarsening (i.e. when all children of a cell are leaves).

source
Trixi.positive_part — Method
positive_part(x)

Return x if x is positive, else zero. In other words, return (x + abs(x)) / 2 for real numbers x.

source
Trixi.pressure — Method
pressure(rho::Real, equations::LatticeBoltzmannEquations2D)
-pressure(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic pressure from the density rho or the particle distribution functions u.

source
Trixi.pressure — Method
pressure(rho::Real, equations::LatticeBoltzmannEquations3D)
-pressure(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic pressure from the density rho or the particle distribution functions u.

source
Trixi.prim2cons — Function
prim2cons(u, equations)

Convert the primitive variables u to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2prim.

source
Trixi.rotate_from_x — Function
rotate_from_x(u, normal, equations)

Apply the rotation that maps the x-axis onto normal to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.

See also: rotate_to_x

source
Trixi.rotate_to_x — Function
rotate_to_x(u, normal, equations)

Apply the rotation that maps normal onto the x-axis to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.

See also: rotate_from_x

source
Trixi.load_adaptive_time_integrator! — Method
load_adaptive_time_integrator!(integrator, restart_file::AbstractString)

Load the context information for time integrators with error-based step size control saved in a restart_file.

source
Trixi.load_dt — Method
load_dt(restart_file::AbstractString)

Load the time step size (dt in OrdinaryDiffEq.jl) saved in a restart_file.

source
Trixi.load_mesh — Method
load_mesh(restart_file::AbstractString; n_cells_max)

Load the mesh from the restart_file.

source
Trixi.load_time — Method
load_time(restart_file::AbstractString)

Load the time saved in a restart_file.

source
Trixi.load_timestep! — Method
load_timestep!(integrator, restart_file::AbstractString)

Load the time step number saved in a restart_file and assign it to both the time step number and and the number of accepted steps (iter and stats.naccept in OrdinaryDiffEq.jl, respectively) in integrator.

source
Trixi.load_timestep — Method
load_timestep(restart_file::AbstractString)

Load the time step number (iter in OrdinaryDiffEq.jl) saved in a restart_file.

source
Trixi.max — Method
max(x, y, ...)

Return the maximum of the arguments. See also the maximum function to take the maximum element from a collection.

This version in Trixi.jl is semantically equivalent to Base.max but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).

Examples

julia> max(2, 5, 1)
+5
source
Trixi.max_abs_speed_naive — Function
max_abs_speed_naive(u_ll, u_rr, orientation::Integer,   equations)
+max_abs_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimate of the maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, based only on the local wave speeds associated to u_ll and u_rr.

For non-integer arguments normal_direction in one dimension, max_abs_speed_naive returns abs(normal_direction[1]) * max_abs_speed_naive(u_ll, u_rr, 1, equations).

source
Trixi.min — Method
min(x, y, ...)

Return the minimum of the arguments. See also the minimum function to take the minimum element from a collection.

This version in Trixi.jl is semantically equivalent to Base.min but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).

Examples

julia> min(2, 5, 1)
+1
source
Trixi.min_max_speed_davis — Function
min_max_speed_davis(u_ll, u_rr, orientation::Integer, equations)
+min_max_speed_davis(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimates of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.

See eq. (10.38) from

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction DOI: 10.1007/b79761

See also FluxHLL, min_max_speed_naive, min_max_speed_einfeldt.

source
Trixi.min_max_speed_einfeldt — Function
min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations)
+min_max_speed_einfeldt(u_ll, u_rr, normal_direction::AbstractVector, equations)

More advanced mininmal and maximal wave speed computation based on

originally developed for the compressible Euler equations. A compact representation can be found in this lecture notes, eq. (9.28).

See also FluxHLL, min_max_speed_naive, min_max_speed_davis.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations2D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations3D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

Original publication:

Compactly summarized:

  • Siddhartha Mishra, Ulrik Skre Fjordholm and Rémi Abgrall Numerical methods for conservation laws and related equations. Link
source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_einfeldt — Method
min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)

Computes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.

source
Trixi.min_max_speed_naive — Function
min_max_speed_naive(u_ll, u_rr, orientation::Integer, equations)
+min_max_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)

Simple and fast estimate(!) of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr. Slightly more diffusive than min_max_speed_davis.

  • Amiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002

See eq. (10.37) from

  • Eleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction DOI: 10.1007/b79761

See also FluxHLL, min_max_speed_davis, min_max_speed_einfeldt.

source
Trixi.modify_dt_for_tstops! — Method
modify_dt_for_tstops!(integrator::SimpleIntegratorSSP)

Modify the time-step size to match the time stops specified in integrator.opts.tstops. To avoid adding OrdinaryDiffEq to Trixi's dependencies, this routine is a copy of https://github.com/SciML/OrdinaryDiffEq.jl/blob/d76335281c540ee5a6d1bd8bb634713e004f62ee/src/integrators/integrator_utils.jl#L38-L54

source
Trixi.multiply_dimensionwise — Method
multiply_dimensionwise(matrix::AbstractMatrix, data_in::AbstractArray{<:Any, NDIMS+1})

Multiply the array data_in by matrix in each coordinate direction, where data_in is assumed to have the first coordinate for the number of variables and the remaining coordinates are multiplied by matrix.

source
Trixi.n_nonconservative_terms — Function
n_nonconservative_terms(equations)

Number of nonconservative terms in the form local * symmetric for a particular equation. This function needs to be specialized only if equations with nonconservative terms are combined with certain solvers (e.g., subcell limiting).

source
Trixi.ndofs — Method
ndofs(semi::AbstractSemidiscretization)

Return the number of degrees of freedom associated with each scalar variable.

source
Trixi.negative_part — Method
negative_part(x)

Return x if x is negative, else zero. In other words, return (x - abs(x)) / 2 for real numbers x.

source
Trixi.ode_default_options — Method
ode_default_options()

Return the default options for OrdinaryDiffEq's solve. Pass ode_default_options()... to solve to only return the solution at the final time and enable MPI aware error-based step size control, whenever MPI is used. For example, use solve(ode, alg; ode_default_options()...).

source
Trixi.ode_norm — Method
ode_norm(u, t)

Implementation of the weighted L2 norm of Hairer and Wanner used for error-based step size control in OrdinaryDiffEq.jl. This function is aware of MPI and uses global MPI communication when running in parallel.

You must pass this function as a keyword argument internalnorm=ode_norm to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.

See the "Advanced Adaptive Stepsize Control" section of the documentation.

source
Trixi.ode_unstable_check — Method
ode_unstable_check(dt, u, semi, t)

Implementation of the basic check for instability used in OrdinaryDiffEq.jl. Instead of checking something like any(isnan, u), this function just checks isnan(dt). This helps when using MPI parallelization, since no additional global communication is required and all ranks will return the same result.

You should pass this function as a keyword argument unstable_check=ode_unstable_check to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.

See the "Miscellaneous" section of the documentation.

source
Trixi.partition! — Method
Trixi.partition!(mesh::T8codeMesh)

Partition a T8codeMesh in order to redistribute elements evenly among MPI ranks.

Arguments

  • mesh::T8codeMesh: Initialized mesh object.
source
Trixi.partition! — Method
partition!(mesh::ParallelTreeMesh, allow_coarsening=true)

Partition mesh using a static domain decomposition algorithm based on leaf cell count and tree structure. If allow_coarsening is true, the algorithm will keep leaf cells together on one rank when needed for local coarsening (i.e. when all children of a cell are leaves).

source
Trixi.positive_part — Method
positive_part(x)

Return x if x is positive, else zero. In other words, return (x + abs(x)) / 2 for real numbers x.

source
Trixi.pressure — Method
pressure(rho::Real, equations::LatticeBoltzmannEquations2D)
+pressure(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic pressure from the density rho or the particle distribution functions u.

source
Trixi.pressure — Method
pressure(rho::Real, equations::LatticeBoltzmannEquations3D)
+pressure(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic pressure from the density rho or the particle distribution functions u.

source
Trixi.prim2cons — Function
prim2cons(u, equations)

Convert the primitive variables u to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2prim.

source
Trixi.rotate_from_x — Function
rotate_from_x(u, normal, equations)

Apply the rotation that maps the x-axis onto normal to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.

See also: rotate_to_x

source
Trixi.rotate_to_x — Function
rotate_to_x(u, normal, equations)

Apply the rotation that maps normal onto the x-axis to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.

See also: rotate_from_x

source
Trixi.save_plot — Method
save_plot(plot_data, variable_names;
           show_mesh=true, plot_arguments=Dict{Symbol,Any}(),
-          time=nothing, timestep=nothing)

Visualize the plot data object provided in plot_data and save result as a PNG file in the out directory, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.

The timestep is used in the filename. time is currently unused by this function.

Experimental implementation

This is an experimental feature and may change in future releases.

See also: VisualizationCallback, show_plot

source
Trixi.show_plot — Method
show_plot(plot_data, variable_names;
+          time=nothing, timestep=nothing)

Visualize the plot data object provided in plot_data and save result as a PNG file in the out directory, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.

The timestep is used in the filename. time is currently unused by this function.

Experimental implementation

This is an experimental feature and may change in future releases.

See also: VisualizationCallback, show_plot

source
Trixi.set_log_type — Method
Trixi.set_log_type(type; force = true)

Set the type of the (natural) log function to be used in Trixi.jl. The default is "sqrt_Trixi_NaN" which returns NaN for negative arguments instead of throwing an error. Alternatively, you can set type to "sqrt_Base" to use the Julia built-in sqrt function which provides a stack-trace of the error which might come in handy when debugging code.

source
Trixi.set_sqrt_type — Method
Trixi.set_sqrt_type(type; force = true)

Set the type of the square root function to be used in Trixi.jl. The default is "sqrt_Trixi_NaN" which returns NaN for negative arguments instead of throwing an error. Alternatively, you can set type to "sqrt_Base" to use the Julia built-in sqrt function which provides a stack-trace of the error which might come in handy when debugging code.

source
Trixi.show_plot — Method
show_plot(plot_data, variable_names;
           show_mesh=true, plot_arguments=Dict{Symbol,Any}(),
-          time=nothing, timestep=nothing)

Visualize the plot data object provided in plot_data and display result, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.

This function is the default plot_creator argument for the VisualizationCallback. time and timestep are currently unused by this function.

Experimental implementation

This is an experimental feature and may change in future releases.

See also: VisualizationCallback, save_plot

source
Trixi.solve — Function
solve(ode, alg; dt, callbacks, kwargs...)

The following structures and methods provide the infrastructure for SSP Runge-Kutta methods of type SimpleAlgorithmSSP.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.source_terms_convergence_test — Method
source_terms_convergence_test(u, x, t, equations::ShallowWaterEquations2D)

Source terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).

This manufactured solution source term is specifically designed for the bottom topography function b(x,y) = 2 + 0.5 * sin(sqrt(2)*pi*x) + 0.5 * sin(sqrt(2)*pi*y) as defined in initial_condition_convergence_test.

source
Trixi.source_terms_convergence_test — Method
source_terms_convergence_test(u, x, t, equations::ShallowWaterEquationsQuasi1D)

Source terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).

This manufactured solution source term is specifically designed for the bottom topography function b(x) = 0.2 - 0.05 * sin(sqrt(2) * pi *x[1]) and channel width 'a(x)= 1 + 0.1 * cos(sqrt(2) * pi * x[1])' as defined in initial_condition_convergence_test.

source
Trixi.source_terms_eoc_test_coupled_euler_gravity — Method
source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations2D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

source
Trixi.source_terms_eoc_test_coupled_euler_gravity — Method
source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

source
Trixi.source_terms_eoc_test_euler — Method
source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

Note

This method is to be used for testing pure Euler simulations with analytic self-gravity. If you intend to do coupled Euler-gravity simulations, you need to use source_terms_eoc_test_coupled_euler_gravity instead.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations1D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations2D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations3D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.splitting_coirier_vanleer — Method
splitting_coirier_vanleer(u, orientation::Integer,
+          time=nothing, timestep=nothing)

Visualize the plot data object provided in plot_data and display result, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.

This function is the default plot_creator argument for the VisualizationCallback. time and timestep are currently unused by this function.

Experimental implementation

This is an experimental feature and may change in future releases.

See also: VisualizationCallback, save_plot

source
Trixi.solve — Function
solve(ode, alg; dt, callbacks, kwargs...)

The following structures and methods provide the infrastructure for SSP Runge-Kutta methods of type SimpleAlgorithmSSP.

Experimental implementation

This is an experimental feature and may change in future releases.

source
Trixi.source_terms_convergence_test — Method
source_terms_convergence_test(u, x, t, equations::ShallowWaterEquations2D)

Source terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).

This manufactured solution source term is specifically designed for the bottom topography function b(x,y) = 2 + 0.5 * sin(sqrt(2)*pi*x) + 0.5 * sin(sqrt(2)*pi*y) as defined in initial_condition_convergence_test.

source
Trixi.source_terms_convergence_test — Method
source_terms_convergence_test(u, x, t, equations::ShallowWaterEquationsQuasi1D)

Source terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).

This manufactured solution source term is specifically designed for the bottom topography function b(x) = 0.2 - 0.05 * sin(sqrt(2) * pi *x[1]) and channel width 'a(x)= 1 + 0.1 * cos(sqrt(2) * pi * x[1])' as defined in initial_condition_convergence_test.

source
Trixi.source_terms_eoc_test_coupled_euler_gravity — Method
source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations2D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

source
Trixi.source_terms_eoc_test_coupled_euler_gravity — Method
source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

source
Trixi.source_terms_eoc_test_euler — Method
source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations3D)

Setup used for convergence tests of the Euler equations with self-gravity used in

  • Michael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593

in combination with initial_condition_eoc_test_coupled_euler_gravity.

Note

This method is to be used for testing pure Euler simulations with analytic self-gravity. If you intend to do coupled Euler-gravity simulations, you need to use source_terms_eoc_test_coupled_euler_gravity instead.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations1D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations2D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.source_terms_harmonic — Method
source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations3D)

Source term that only includes the forcing from the hyperbolic diffusion system.

source
Trixi.splitting_coirier_vanleer — Method
splitting_coirier_vanleer(u, orientation::Integer,
                           equations::CompressibleEulerEquations1D)
 splitting_coirier_vanleer(u, which::Union{Val{:minus}, Val{:plus}}
                           orientation::Integer,
-                          equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux from Coirier and van Leer. The splitting has correction terms in the pressure splitting as well as the mass and energy flux components. The motivation for these corrections are to handle flows at the low Mach number limit.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • William Coirier and Bram van Leer (1991) Numerical flux formulas for the Euler and Navier-Stokes equations. II - Progress in flux-vector splitting DOI: 10.2514/6.1991-1566
source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
+                          equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux from Coirier and van Leer. The splitting has correction terms in the pressure splitting as well as the mass and energy flux components. The motivation for these corrections are to handle flows at the low Mach number limit.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • William Coirier and Bram van Leer (1991) Numerical flux formulas for the Euler and Navier-Stokes equations. II - Progress in flux-vector splitting DOI: 10.2514/6.1991-1566
source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
                          equations::CompressibleEulerEquations2D)
 splitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations2D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) similar to a flux splitting one would apply, e.g., to Burgers' equation.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
+                         equations::CompressibleEulerEquations2D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) similar to a flux splitting one would apply, e.g., to Burgers' equation.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
                          equations::InviscidBurgersEquation1D)
 splitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::InviscidBurgersEquation1D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ = abs(u).

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
+                         equations::InviscidBurgersEquation1D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ = abs(u).

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_lax_friedrichs — Method
splitting_lax_friedrichs(u, orientation::Integer,
                          equations::LinearScalarAdvectionEquation1D)
 splitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::LinearScalarAdvectionEquation1D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ is the absolute value of the advection velocity.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
+                         equations::LinearScalarAdvectionEquation1D)

Naive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ is the absolute value of the advection velocity.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
                          equations::CompressibleEulerEquations1D)
 splitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
+                         equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
                          equations::CompressibleEulerEquations2D)
 splitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations2D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
+                         equations::CompressibleEulerEquations2D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_steger_warming — Method
splitting_steger_warming(u, orientation::Integer,
                          equations::CompressibleEulerEquations3D)
 splitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations3D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_vanleer_haenel — Method
splitting_vanleer_haenel(u, orientation::Integer,
+                         equations::CompressibleEulerEquations3D)

Splitting of the compressible Euler flux of Steger and Warming.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

  • Joseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum
source
Trixi.splitting_vanleer_haenel — Method
splitting_vanleer_haenel(u, orientation::Integer,
                          equations::CompressibleEulerEquations1D)
 splitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

source
Trixi.splitting_vanleer_haenel — Method
splitting_vanleer_haenel(u, orientation::Integer,
+                         equations::CompressibleEulerEquations1D)

Splitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

source
Trixi.splitting_vanleer_haenel — Method
splitting_vanleer_haenel(u, orientation::Integer,
                          equations::CompressibleEulerEquations2D)
 splitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}
                          orientation::Integer,
-                         equations::CompressibleEulerEquations2D)

Splitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

source
Trixi.stolarsky_mean — Method
stolarsky_mean(x, y, gamma)

Compute an instance of a weighted Stolarsky mean of the form

stolarsky_mean(x, y, gamma) = (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))

where gamma > 1.

Problem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Winters et al. (2020). Set f = (y - x) / (y + x) and g = gamma (for compact notation). Then, we use the expansions

((1+f)^g - (1-f)^g) / g = 2*f + (g-1)(g-2)/3 * f^3 + (g-1)(g-2)(g-3)(g-4)/60 * f^5 + O(f^7)

and

((1+f)^(g-1) - (1-f)^(g-1)) / (g-1) = 2*f + (g-2)(g-3)/3 * f^3 + (g-2)(g-3)(g-4)(g-5)/60 * f^5 + O(f^7)

Inserting the first few terms of these expansions and performing polynomial long division we find that

stolarsky_mean(x, y, gamma) ≈ (y + x) / 2 * (1 + (g-2)/3 * f^2 - (g+1)(g-2)(g-3)/45 * f^4 + (g+1)(g-2)(g-3)(2g(g-2)-9)/945 * f^6)

Since divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use

f^2 = (y - x)^2 / (x + y)^2
+                         equations::CompressibleEulerEquations2D)

Splitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.

Returns a tuple of the fluxes "minus" (associated with waves going into the negative axis direction) and "plus" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().

Experimental implementation (upwind SBP)

This is an experimental feature and may change in future releases.

References

source
Trixi.stolarsky_mean — Method
stolarsky_mean(x, y, gamma)

Compute an instance of a weighted Stolarsky mean of the form

stolarsky_mean(x, y, gamma) = (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))

where gamma > 1.

Problem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Winters et al. (2020). Set f = (y - x) / (y + x) and g = gamma (for compact notation). Then, we use the expansions

((1+f)^g - (1-f)^g) / g = 2*f + (g-1)(g-2)/3 * f^3 + (g-1)(g-2)(g-3)(g-4)/60 * f^5 + O(f^7)

and

((1+f)^(g-1) - (1-f)^(g-1)) / (g-1) = 2*f + (g-2)(g-3)/3 * f^3 + (g-2)(g-3)(g-4)(g-5)/60 * f^5 + O(f^7)

Inserting the first few terms of these expansions and performing polynomial long division we find that

stolarsky_mean(x, y, gamma) ≈ (y + x) / 2 * (1 + (g-2)/3 * f^2 - (g+1)(g-2)(g-3)/45 * f^4 + (g+1)(g-2)(g-3)(2g(g-2)-9)/945 * f^6)

Since divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use

f^2 = (y - x)^2 / (x + y)^2
     = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)

Given ε = 1.0e-4, we use the following algorithm.

if f^2 < ε
   # use the expansion above
 else
   # use the direct formula (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))
-end

References

source
Trixi.totalgamma — Method
totalgamma(u, equations::CompressibleEulerMulticomponentEquations1D)

Function that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.

source
Trixi.totalgamma — Method
totalgamma(u, equations::CompressibleEulerMulticomponentEquations2D)

Function that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.

source
Trixi.trixi_include — Method
trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)

include the file elixir and evaluate its content in the global scope of module mod. You can override specific assignments in elixir by supplying keyword arguments. It's basic purpose is to make it easier to modify some parameters while running Trixi.jl from the REPL. Additionally, this is used in tests to reduce the computational burden for CI while still providing examples with sensible default values for users.

Before replacing assignments in elixir, the keyword argument maxiters is inserted into calls to solve and Trixi.solve with it's default value used in the SciML ecosystem for ODEs, see the "Miscellaneous" section of the documentation.

Examples

julia> redirect_stdout(devnull) do
-         trixi_include(@__MODULE__, joinpath(examples_dir(), "tree_1d_dgsem", "elixir_advection_extended.jl"),
-                       tspan=(0.0, 0.1))
-         sol.t[end]
-       end
-[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.
-0.1
source
Trixi.varnames — Function
varnames(conversion_function, equations)

Return the list of variable names when applying conversion_function to the conserved variables associated to equations. This function is mainly used internally to determine output to screen and for IO, e.g., for the AnalysisCallback and the SaveSolutionCallback. Common choices of the conversion_function are cons2cons and cons2prim.

source
Trixi.velocity — Method
velocity(u, orientation, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y) from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, orientation, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y, 3 -> z) from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic velocity vector from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic velocity vector from the particle distribution functions u.

source
Trixi.@autoinfiltrate — Macro
@autoinfiltrate
-@autoinfiltrate condition::Bool

Invoke the @infiltrate macro of the package Infiltrator.jl to create a breakpoint for ad-hoc interactive debugging in the REPL. If the optional argument condition is given, the breakpoint is only enabled if condition evaluates to true.

As opposed to using Infiltrator.@infiltrate directly, this macro does not require Infiltrator.jl to be added as a dependency to Trixi.jl. As a bonus, the macro will also attempt to load the Infiltrator module if it has not yet been loaded manually.

Note: For this macro to work, the Infiltrator.jl package needs to be installed in your current Julia environment stack.

See also: Infiltrator.jl

Internal use only

Please note that this macro is intended for internal use only. It is not part of the public API of Trixi.jl, and it thus can altered (or be removed) at any time without it being considered a breaking change.

source
Trixi.@threaded — Macro
@threaded for ... end

Semantically the same as Threads.@threads when iterating over a AbstractUnitRange but without guarantee that the underlying implementation uses Threads.@threads or works for more general for loops. In particular, there may be an additional check whether only one thread is used to reduce the overhead of serial execution or the underlying threading capabilities might be provided by other packages such as Polyester.jl.

Warn

This macro does not necessarily work for general for loops. For example, it does not necessarily support general iterables such as eachline(filename).

Some discussion can be found at https://discourse.julialang.org/t/overhead-of-threads-threads/53964 and https://discourse.julialang.org/t/threads-threads-with-one-thread-how-to-remove-the-overhead/58435.

source
+end

References

source
Trixi.totalgamma — Method
totalgamma(u, equations::CompressibleEulerMulticomponentEquations1D)

Function that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.

source
Trixi.totalgamma — Method
totalgamma(u, equations::CompressibleEulerMulticomponentEquations2D)

Function that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.

source
Trixi.varnames — Function
varnames(conversion_function, equations)

Return the list of variable names when applying conversion_function to the conserved variables associated to equations. This function is mainly used internally to determine output to screen and for IO, e.g., for the AnalysisCallback and the SaveSolutionCallback. Common choices of the conversion_function are cons2cons and cons2prim.

source
Trixi.velocity — Method
velocity(u, orientation, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y) from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, orientation, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y, 3 -> z) from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, equations::LatticeBoltzmannEquations2D)

Calculate the macroscopic velocity vector from the particle distribution functions u.

source
Trixi.velocity — Method
velocity(u, equations::LatticeBoltzmannEquations3D)

Calculate the macroscopic velocity vector from the particle distribution functions u.

source
Trixi.@autoinfiltrate — Macro
@autoinfiltrate
+@autoinfiltrate condition::Bool

Invoke the @infiltrate macro of the package Infiltrator.jl to create a breakpoint for ad-hoc interactive debugging in the REPL. If the optional argument condition is given, the breakpoint is only enabled if condition evaluates to true.

As opposed to using Infiltrator.@infiltrate directly, this macro does not require Infiltrator.jl to be added as a dependency to Trixi.jl. As a bonus, the macro will also attempt to load the Infiltrator module if it has not yet been loaded manually.

Note: For this macro to work, the Infiltrator.jl package needs to be installed in your current Julia environment stack.

See also: Infiltrator.jl

Internal use only

Please note that this macro is intended for internal use only. It is not part of the public API of Trixi.jl, and it thus can altered (or be removed) at any time without it being considered a breaking change.

source
Trixi.@threaded — Macro
@threaded for ... end

Semantically the same as Threads.@threads when iterating over a AbstractUnitRange but without guarantee that the underlying implementation uses Threads.@threads or works for more general for loops. In particular, there may be an additional check whether only one thread is used to reduce the overhead of serial execution or the underlying threading capabilities might be provided by other packages such as Polyester.jl.

Warn

This macro does not necessarily work for general for loops. For example, it does not necessarily support general iterables such as eachline(filename).

Some discussion can be found at https://discourse.julialang.org/t/overhead-of-threads-threads/53964 and https://discourse.julialang.org/t/threads-threads-with-one-thread-how-to-remove-the-overhead/58435.

source
diff --git a/previews/PR1624/reference-trixi2vtk/index.html b/previews/PR1624/reference-trixi2vtk/index.html index c3f2226aa30..d26ac0dbb66 100644 --- a/previews/PR1624/reference-trixi2vtk/index.html +++ b/previews/PR1624/reference-trixi2vtk/index.html @@ -1,6 +1,6 @@ -Trixi2Vtk.jl · Trixi.jl
GitHub

Trixi2Vtk.jl API

Trixi2Vtk.trixi2vtk — Method
trixi2vtk(filename::AbstractString...;
+Trixi2Vtk.jl · Trixi.jl
GitHub
Trixi2Vtk.jl API
Trixi2Vtk.trixi2vtk — Method
trixi2vtk(filename::AbstractString...;
           format=:vtu, verbose=false, hide_progress=false, pvd=nothing,
           output_directory=".", nvisnodes=nothing, save_celldata=true,
           reinterpolate=true, data_is_uniform=false)
Convert Trixi-generated output files to VTK files (VTU or VTI).
Arguments
  • filename: One or more Trixi solution/restart/mesh files to convert to a VTK file.             Filenames support file globbing, e.g., "solution*" to match all files starting             with solution.
  • format: Output format for solution/restart files. Can be 'vtu' or 'vti'.
  • verbose: Set to true to enable verbose output.
  • hide_progress: Hide progress bar (will be hidden automatically if verbose is true).
  • pvd: Use this filename to store PVD file (instead of auto-detecting name). Note that        only the name will be used (directory and file extension are ignored).
  • output_directory: Output directory where generated files are stored.
  • nvisnodes: Number of visualization nodes per element.              (default: number of DG nodes for StructuredMesh or UnstructuredMesh2D,                        twice the number of DG nodes for TreeMesh).              A value of 0 (zero) uses the number of nodes in the DG elements.
  • save_celldata: Boolean value to determine if cell-based data should be saved.                  (default: true)
  • reinterpolate: Boolean value to determine if data should be reinterpolated                  onto uniform points. When false the raw data at the compute nodes                  is copied into the appropriate format.                  (default: true)
  • data_is_uniform: Boolean to indicate if the data to be converted is from a finite difference                    method on a uniform grid of points.                    (default: false)
Examples
julia> trixi2vtk("out/solution_000*.h5")
-[...]
source

+[...]
source
diff --git a/previews/PR1624/reference-trixibase/index.html b/previews/PR1624/reference-trixibase/index.html new file mode 100644 index 00000000000..e1f841b1ed0 --- /dev/null +++ b/previews/PR1624/reference-trixibase/index.html @@ -0,0 +1,10 @@ + +TrixiBase.jl · Trixi.jl
GitHub

TrixiBase.jl API

TrixiBase.trixi_include — Method
trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)

include the file elixir and evaluate its content in the global scope of module mod. You can override specific assignments in elixir by supplying keyword arguments. Its basic purpose is to make it easier to modify some parameters while running simulations from the REPL. Additionally, this is used in tests to reduce the computational burden for CI while still providing examples with sensible default values for users.

Before replacing assignments in elixir, the keyword argument maxiters is inserted into calls to solve with it's default value used in the SciML ecosystem for ODEs, see the "Miscellaneous" section of the documentation.

Examples

julia> using TrixiBase, Trixi
+
+julia> redirect_stdout(devnull) do
+         trixi_include(@__MODULE__, joinpath(examples_dir(), "tree_1d_dgsem", "elixir_advection_extended.jl"),
+                       tspan=(0.0, 0.1))
+         sol.t[end]
+       end
+[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.
+0.1
source
diff --git a/previews/PR1624/restart/index.html b/previews/PR1624/restart/index.html index 36760431dbf..e415bc183c4 100644 --- a/previews/PR1624/restart/index.html +++ b/previews/PR1624/restart/index.html @@ -1,6 +1,6 @@ -Restart simulation · Trixi.jl
GitHub

Restart simulation

You can continue running an already finished simulation by first preparing the simulation for the restart and then performing the restart. Here we suppose that in the first run your simulation stops at time 1.0 and then you want it to run further to time 2.0.

Prepare the simulation for a restart

In you original elixir you need to specify to write out restart files. Those will later be read for the restart of your simulation. This is done almost the same way as writing the snapshots using the SaveSolutionCallback callback. For the restart files it is called SaveRestartCallback:

save_restart = SaveRestartCallback(interval=100,
+Restart simulation · Trixi.jl
GitHub
Restart simulation
You can continue running an already finished simulation by first preparing the simulation for the restart and then performing the restart. Here we suppose that in the first run your simulation stops at time 1.0 and then you want it to run further to time 2.0.
Prepare the simulation for a restart
In you original elixir you need to specify to write out restart files. Those will later be read for the restart of your simulation. This is done almost the same way as writing the snapshots using the SaveSolutionCallback callback. For the restart files it is called SaveRestartCallback:
save_restart = SaveRestartCallback(interval=100,
                                    save_final_restart=true)
Make this part of your CallbackSet.
An example is examples/examples/structured_2d_dgsem/elixir_advection_extended.jl.
Perform the simulation restart
Since all of the information about the simulation can be obtained from the last snapshot, the restart can be done with relatively few lines in an extra elixir file. However, some might prefer to keep everything in one elixir and conditionals like if restart with a boolean variable restart that is user defined.
First we need to define from which file we want to restart, e.g.
restart_file = "restart_000021.h5"
 restart_filename = joinpath("out", restart_file)
Then we load the mesh file:
mesh = load_mesh(restart_filename)
This is then needed for the semidiscretization:
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)
We then define a new time span for the simulation that takes as starting time the one form the snapshot:
tspan = (load_time(restart_filename), 2.0)
We now also take the last dt, so that our solver does not need to first find one to fulfill the CFL condition:
dt = load_dt(restart_filename)
The ODE that we will pass to the solver is now:
ode = semidiscretize(semi, tspan, restart_filename)
You should now define a SaveSolutionCallback similar to the original simulation, but with save_initial_solution=false, otherwise our initial snapshot will be overwritten. If you are using one file for the original simulation and the restart you can reuse your SaveSolutionCallback, but need to set
save_solution.condition.save_initial_solution = false
Before we compute the solution using  OrdinaryDiffEq.jl we need to set the integrator and its time step number, e.g.:
integrator = init(ode, CarpenterKennedy2N54(williamson_condition=false),
                   dt=dt, save_everystep=false, callback=callbacks);
-load_timestep!(integrator, restart_filename)
Now we can compute the solution:
sol = solve!(integrator)
An example is in examples/structured_2d_dgsem/elixir_advection_restart.jl.

+load_timestep!(integrator, restart_filename)

Now we can compute the solution:

sol = solve!(integrator)

An example is in examples/structured_2d_dgsem/elixir_advection_restart.jl.

diff --git a/previews/PR1624/search_index.js b/previews/PR1624/search_index.js index dacfab49cb2..51ac875b472 100644 --- a/previews/PR1624/search_index.js +++ b/previews/PR1624/search_index.js @@ -1,3 +1,3 @@ var documenterSearchIndex = {"docs": -[{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"EditURL = \"../../literate/src/files/adaptive_mesh_refinement.jl\"","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#adaptive_mesh_refinement","page":"12 Adaptive mesh refinement","title":"12: Adaptive mesh refinement","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Adaptive mesh refinement (AMR) is a method of adapting the resolution of the numerical method to the solution features such as turbulent regions or shocks. In those critical regions of the domain, we want the simulation to use elements with smaller mesh sizes compared to other regions. This should be automatically and dynamically adapted during the run of the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Implementation-in-Trixi.jl","page":"12 Adaptive mesh refinement","title":"Implementation in Trixi.jl","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"In Trixi.jl, AMR is possible for the mesh types TreeMesh and P4estMesh. Both meshes are organized in a tree structure and therefore, each element can be refined independently. In Trixi.jl, AMR is restricted to a 2:1 refinement ratio between neighbor elements. This means that the maximum resolution difference of neighboring elements is a factor of two.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"The implementation of AMR is divided into different steps. The basic refinement setting contains an indicator and a controller. These are added to the simulation by using an AMR callback.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Indicators","page":"12 Adaptive mesh refinement","title":"Indicators","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"An indicator estimates the current accuracy of the numerical approximation. It indicates which regions of the domain need finer or coarser resolutions. In Trixi.jl, you can use for instance IndicatorLöhner and IndicatorHennemannGassner.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"IndicatorLöhner (also callable with IndicatorLoehner) is an interpretation and adaptation of a FEM indicator by Löhner (1987) and estimates a weighted second derivative of a specified variable locally.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_indicator = IndicatorLöhner(semi, variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"All indicators have the parameter variable which is used to specify the variable for the indicator calculation. You can use for instance density, pressure or density_pressure for the compressible Euler equations. Moreover, you have the option to use simply the first conservation variable with first for any equations. This might be a good choice for a starting example.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"IndicatorHennemannGassner, also used as a shock-capturing indicator, was developed by Hennemann et al. (2021) and is explained in detail in the tutorial about shock-capturing. It can be constructed as follows.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_indicator = IndicatorHennemannGassner(semi,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Another indicator is the very basic IndicatorMax. It indicates the maximal value of a variable and is therefore mostly used for verification and testing. But it might be useful for the basic understanding of the implementation of indicators and AMR in Trixi.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_indicator = IndicatorMax(semi, variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Controllers","page":"12 Adaptive mesh refinement","title":"Controllers","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"The spatial discretization into elements is tree-based for both AMR supporting mesh types TreeMesh and P4estMesh. Thus, the higher the level in the tree the higher the level of refinement. For instance, a mesh element of level 3 has double resolution in each direction compared to another element with level 2.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"To map specific indicator values to a desired level of refinement, Trixi.jl uses controllers. They are build in three levels: There is a base level of refinement base_level, which is the minimum allowed refinement level. Then, there is a medium level med_level, which corresponds to the initial level of refinement, for indicator values above the threshold med_threshold and equally, a maximal level max_level for values above max_threshold. This variant of controller is called ControllerThreeLevel in Trixi.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevel(semi, amr_indicator;\n base_level=4,\n med_level=5, med_threshold=0.1,\n max_level=6, max_threshold=0.6)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"You can also set med_level=0 to use the current level as target, see the docstring of ControllerThreeLevel.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"An extension is ControllerThreeLevelCombined, which uses two different indicators. The primary indicator works the same as the single indicator for ControllerThreeLevel. The second indicator with its own maximum threshold adds the property, that the target level is set to max_level additionally if this indicator's value is greater than max_threshold_secondary. This is for instance used to assure that a shock has always the maximum refinement level.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;\n base_level=2,\n med_level=6, med_threshold=0.0003,\n max_level=8, max_threshold=0.003,\n max_threshold_secondary=0.3)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"This controller is for instance used in elixir_euler_astro_jet_amr.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Callback","page":"12 Adaptive mesh refinement","title":"Callback","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"The AMR indicator and controller are added to the simulation through the callback AMRCallback. It contains a semidiscretization semi, the controller amr_controller and the parameters interval, adapt_initial_condition, and adapt_initial_condition_only_refine.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Adaptive mesh refinement will be performed every interval time steps. adapt_initial_condition indicates whether the initial condition already should be adapted before the first time step. And with adapt_initial_condition_only_refine=true the mesh is only refined at the beginning but not coarsened.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Exemplary-simulation","page":"12 Adaptive mesh refinement","title":"Exemplary simulation","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Here, we want to implement a simple AMR simulation of the 2D linear advection equation for a Gaussian pulse.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"using OrdinaryDiffEq\nusing Trixi\n\nadvection_velocity = (0.2, -0.7)\nequations = LinearScalarAdvectionEquation2D(advection_velocity)\n\ninitial_condition = initial_condition_gauss\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\ncoordinates_min = (-5.0, -5.0)\ncoordinates_max = ( 5.0, 5.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n n_cells_max=30_000)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\n\ntspan = (0.0, 10.0)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"For the best understanding about indicators and controllers, we use the simple AMR indicator IndicatorMax. As described before, it returns the maximal value of the specified variable (here the only conserved variable). Therefore, regions with a high maximum are refined. This is not really useful numerical application, but a nice demonstration example.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_indicator = IndicatorMax(semi, variable=first)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"These values are transferred to a refinement level with the ControllerThreeLevel, such that every element with maximal value greater than 0.1 is refined once and elements with maximum above 0.6 are refined twice.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevel(semi, amr_indicator,\n base_level=4,\n med_level=5, med_threshold=0.1,\n max_level=6, max_threshold=0.6)\n\namr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(amr_callback, stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Running the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"We plot the solution and add the refined mesh at the end of the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#More-examples","page":"12 Adaptive mesh refinement","title":"More examples","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Trixi.jl provides many elixirs using AMR. We want to give some examples for different mesh types:","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"elixir_euler_blast_wave_amr.jl for TreeMesh and P4estMesh\nelixir_euler_kelvin_helmholtz_instability_amr.jl for TreeMesh\nelixir_euler_double_mach_amr.jl for P4estMesh","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Animations of more interesting and complicated AMR simulations can be found below and on Trixi.jl's youtube channel \"Trixi Framework\".","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"First, we give a purely hyperbolic simulation of a Sedov blast wave with self-gravity. This simulation uses the mesh type TreeMesh as we did and the AMR indicator IndicatorHennemannGassner.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":" \n
","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Source: Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"The next example is a numerical simulation of an ideal MHD rotor on an unstructured AMR mesh. The used mesh type is a P4estMesh.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":" \n
","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"Source: Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"For more information, please have a look at the respective links.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Package-versions","page":"12 Adaptive mesh refinement","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"12 Adaptive mesh refinement","title":"12 Adaptive mesh refinement","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"EditURL = \"../../literate/src/files/structured_mesh_mapping.jl\"","category":"page"},{"location":"tutorials/structured_mesh_mapping/#structured_mesh_mapping","page":"13 Structured mesh with curvilinear mapping","title":"13: Structured mesh with curvilinear mapping","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Here, we want to introduce another mesh type of Trixi.jl. More precisely, this tutorial is about the curved mesh type StructuredMesh supporting curved meshes.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Creating-a-curved-mesh","page":"13 Structured mesh with curvilinear mapping","title":"Creating a curved mesh","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"There are two basic options to define a curved StructuredMesh in Trixi.jl. You can implement curves for the domain boundaries, or alternatively, set up directly the complete transformation mapping. We now present one short example each.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Mesh-defined-by-domain-boundary-curves","page":"13 Structured mesh with curvilinear mapping","title":"Mesh defined by domain boundary curves","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Both examples are based on a semdiscretization of the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"using OrdinaryDiffEq\nusing Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"We start with a pressure perturbation at (xs, 0.0) as initial condition.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"function initial_condition_pressure_perturbation(x, t, equations::CompressibleEulerEquations2D)\n xs = 1.5 # location of the initial disturbance on the x axis\n w = 1/8 # half width\n p = exp(-log(2) * ((x[1]-xs)^2 + x[2]^2)/w^2) + 1.0\n v1 = 0.0\n v2 = 0.0\n rho = 1.0\n\n return prim2cons(SVector(rho, v1, v2, p), equations)\nend\ninitial_condition = initial_condition_pressure_perturbation","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Initialize every boundary as a boundary_condition_slip_wall.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"boundary_conditions = boundary_condition_slip_wall","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"The approximation setup is an entropy-stable split-form DG method with polydeg=4. We are using the two fluxes flux_ranocha and flux_lax_friedrichs.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"solver = DGSEM(polydeg=4, surface_flux=flux_lax_friedrichs,\n volume_integral=VolumeIntegralFluxDifferencing(flux_ranocha))","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"We want to define a circular cylinder as physical domain. It contains an inner semicircle with radius r0 and an outer semicircle of radius r1.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"(Image: )","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"The domain boundary curves with curve parameter in -11 are sorted as shown in the sketch. They always are orientated from negative to positive coordinate, such that the corners have to fit like this f_1(+1) = f_4(-1), f_3(+1) = f_2(-1), etc.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"In our case we can define the domain boundary curves as follows:","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"r0 = 0.5 # inner radius\nr1 = 5.0 # outer radius\nf1(xi) = SVector( r0 + 0.5 * (r1 - r0) * (xi + 1), 0.0) # right line\nf2(xi) = SVector(-r0 - 0.5 * (r1 - r0) * (xi + 1), 0.0) # left line\nf3(eta) = SVector(r0 * cos(0.5 * pi * (eta + 1)), r0 * sin(0.5 * pi * (eta + 1))) # inner circle\nf4(eta) = SVector(r1 * cos(0.5 * pi * (eta + 1)), r1 * sin(0.5 * pi * (eta + 1))) # outer circle","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"We create a curved mesh with 16 x 16 elements. The defined domain boundary curves are passed as a tuple.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"cells_per_dimension = (16, 16)\nmesh = StructuredMesh(cells_per_dimension, (f1, f2, f3, f4), periodicity=false)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Then, we define the simulation with endtime T=3 with semi, ode and callbacks.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_conditions)\n\ntspan = (0.0, 3.0)\node = semidiscretize(semi, tspan)\n\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi, interval=analysis_interval)\n\nalive_callback = AliveCallback(analysis_interval=analysis_interval)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(analysis_callback,\n alive_callback,\n stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Running the simulation","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"pd = PlotData2D(sol)\nplot(pd[\"p\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Mesh-directly-defined-by-the-transformation-mapping","page":"13 Structured mesh with curvilinear mapping","title":"Mesh directly defined by the transformation mapping","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"As mentioned before, you can also define the domain for a StructuredMesh by directly setting up a transformation mapping. Here, we want to present a nice mapping, which is often used to test free-stream preservation. Exact free-stream preservation is a crucial property of any numerical method on curvilinear grids. The mapping is a reduced 2D version of the mapping described in Rueda-Ramírez et al. (2021), p.18.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"using OrdinaryDiffEq\nusing Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"As mentioned, this mapping is used for testing free-stream preservation. So, we use a constant initial condition.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"initial_condition = initial_condition_constant\n\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"We define the transformation mapping with variables in -1 1 as described in Rueda-Ramírez et al. (2021), p.18 (reduced to 2D):","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"function mapping(xi_, eta_)\n # Transform input variables between -1 and 1 onto [0,3]\n xi = 1.5 * xi_ + 1.5\n eta = 1.5 * eta_ + 1.5\n\n y = eta + 3/8 * (cos(1.5 * pi * (2 * xi - 3)/3) *\n cos(0.5 * pi * (2 * eta - 3)/3))\n\n x = xi + 3/8 * (cos(0.5 * pi * (2 * xi - 3)/3) *\n cos(2 * pi * (2 * y - 3)/3))\n\n return SVector(x, y)\nend","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Instead of a tuple of boundary functions, the mesh now has the mapping as its parameter.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"cells_per_dimension = (16, 16)\nmesh = StructuredMesh(cells_per_dimension, mapping)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan)\n\nanalysis_callback = AnalysisCallback(semi, interval=250)\n\nstepsize_callback = StepsizeCallback(cfl=0.8)\n\ncallbacks = CallbackSet(analysis_callback,\n stepsize_callback)\n\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Now, we want to verify the free-stream preservation property and plot the mesh. For the verification, we calculate the absolute difference of the first conservation variable density u[1] and 1.0. To plot this error and the mesh, we are using the visualization feature ScalarPlotData2D, explained in visualization.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"error_density = let u = Trixi.wrap_array(sol.u[end], semi)\n abs.(u[1, :, :, :] .- 1.0) # density, x, y, elements\nend\npd = ScalarPlotData2D(error_density, semi)\n\nusing Plots\nplot(pd, title=\"Error in density\")\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"We observe that the errors in the variable density are at the level of machine accuracy. Moreover, the plot shows the mesh structure resulting from our transformation mapping.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"Of course, you can also use other mappings as for instance shifts by (x y)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"mapping(xi, eta) = SVector(xi + x, eta + y)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"or rotations with a rotation matrix T","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"mapping(xi, eta) = T * SVector(xi, eta).","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"For more curved mesh mappings, please have a look at some elixirs for StructuredMesh. For another curved mesh type, there is a tutorial about Trixi.jl's unstructured mesh type [UnstructuredMesh2D] and its use of the High-Order Hex-Quad Mesh (HOHQMesh) generator, created and developed by David Kopriva.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Package-versions","page":"13 Structured mesh with curvilinear mapping","title":"Package versions","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"13 Structured mesh with curvilinear mapping","title":"13 Structured mesh with curvilinear mapping","text":"This page was generated using Literate.jl.","category":"page"},{"location":"github-git/#GitHub-and-Git","page":"GitHub & Git","title":"GitHub & Git","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This page contains information on how to use GitHub and Git when developing Trixi.jl.","category":"page"},{"location":"github-git/#Development-workflow","page":"GitHub & Git","title":"Development workflow","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"For adding modifications to Trixi.jl, we generally follow these steps:","category":"page"},{"location":"github-git/#Create-an-issue-(optional)","page":"GitHub & Git","title":"Create an issue (optional)","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"In many cases it makes sense to start by creating an issue on GitHub. For example, if the implementation approach for a new feature is not yet clear or if there should be a discussion about the desired outcome, it is good practice to first get a consensus on what is the expected result of this modification. A GitHub issue is the place to lead this discussion, as it preserves it in the project and - together with the actual code changes - allows in the future to revisit the reasons for a particular choice of implementation or feature.","category":"page"},{"location":"github-git/#Create-a-branch-and-*immediately*-create-a-pull-request","page":"GitHub & Git","title":"Create a branch and immediately create a pull request","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"All feature development, bug fixes etc. should be developed in a branch and not directly on main. If you do not have write access to the main repository on GitHub, first create a fork of the Trixi.jl repository and clone the fork to your machine. Then, create a branch locally by executing git checkout -b yourbranch, push it to the repository, and create a pull request (PR).","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you have already cloned Trixi.jl from the main repo to your local machine, you can also work in that clone. You just need to add your fork as additional remote repository and push your new branch there.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"git remote add myfork git@github.com:YOUR_NAME/Trixi.jl.git\n# get latest main from the main repo\ngit checkout main\ngit pull\n# create a new branch for a cool new feature, bug fix, ...\ngit checkout -b YOUR_BRANCH_NAME\n# do some work and push it to your fork\ngit push -u myfork\n# go to https://github.com/trixi-framework/Trixi.jl/pull\n# and create a PR from your new branch","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"info: Why using pull requests?\nImmediately creating a PR for your branch has the benefit that all code discussions can now be held directly next to the corresponding code. Also, the PR allows to easily compare your branch to the upstream branch (usually main) to see what you have changed. Moreover, tests will run automatically.","category":"page"},{"location":"github-git/#Make-changes","page":"GitHub & Git","title":"Make changes","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"With a branch and PR in place, you can now write your code and commit it to your branch. If you request feedback from someone else, make sure to push your branch to the repository such that the others can easily review your changes or dive in and change something themselves.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"warning: Avoid committing unwanted files\nWhen you use git add . or similar catch-all versions, make sure you do not accidentally commit unwanted files (e.g., Trixi.jl output files, images or videos etc.). If it happens anyways, you can undo the last commit (also multiple times) by running git reset HEAD~ (see also Undo last commit). However, this strategy only works if you have not yet pushed your changes. If you did push your changes, please talk to one of the core developers on how to proceed.","category":"page"},{"location":"github-git/#Keep-your-branch-in-sync-with-main","page":"GitHub & Git","title":"Keep your branch in sync with main","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"For larger features with longer-living branches, it may make sense to synchronize your branch with the current main, e.g., if there was a bug fix in main that is relevant for you. In this case, perform the following steps to merge the current main to your branch:","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Commit all your local changes to your branch and push it. This allows you to delete your clone in case you make a mistake and need to abort the merge.\nExecute git fetch to get the latest changes from the repository.\nMake sure you are in the correct branch by checking the output of git status or by running git checkout yourbranch.\nMerge main using git merge main. If there were no conflicts, hooray!, you are done. Otherwise you need to resolve your merge conflicts and commit the changes afterwards. A good guide for resolving merge conflicts can be found here.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"In general, always use git merge and not git rebase to get the latest changes from main. It is less error-prone and does not create problems on branches that are worked on collaboratively.","category":"page"},{"location":"github-git/#Prepare-for-review","page":"GitHub & Git","title":"Prepare for review","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you feel like your branch is ready to be merged to main, prepare it for review. That is, you should","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"merge the current main to your branch\nrun tests if available, but at least ensure that you did not accidentally change the results for one of the existing example elixirs\nproperly comment your code\ndelete old/unused code, especially commented lines (unless they contain helpful code, in which case you should add a comment on why you keep this around)\nremove debug statements\nadd a elixir_xxx.jl that uses your feature (only relevant for new features)\nmake sure your code formatting adheres to the Style guide\ndescribe changes in NEWS.md if appropriate","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"After you are confident that your branch is cleaned up properly, commit all changes and push them to the repository.","category":"page"},{"location":"github-git/#Get-reviewed","page":"GitHub & Git","title":"Get reviewed","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Ask one of the core developers to review your code. Sometimes this will be done directly, either face-to-face or via a video call. Other times a review will be conducted asynchronously, with the reviewer leaving comments and annotations. In some cases it will be necessary to do multiple rounds of reviews, especially if there are larger changes to be added. Just commit and push your changes to your branch, and the corresponding pull request will be updated automatically.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Please note that a review has nothing to do with the lack of experience of the person developing changes: We try to review all code before it gets added to main, even from the most experienced developers. This is good practice and helps to keep the error rate low while ensuring that the code is developed in a consistent fashion. Furthermore, do not take criticism of your code personally - we just try to keep Trixi.jl as accessible and easy to use for everyone.","category":"page"},{"location":"github-git/#Merge-branch","page":"GitHub & Git","title":"Merge branch","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Once your branch is reviewed and declared ready for merging by the reviewer, make sure that all the latest changes have been pushed. Then, one of the developers will merge your PR. If you are one of the developers, you can also go to the pull request page on GitHub and and click on Merge pull request. Voilà, you are done! Your branch will have been merged to main and the source branch will have been deleted in the GitHub repository (if you are not working in your own fork).","category":"page"},{"location":"github-git/#Update-your-working-copy","page":"GitHub & Git","title":"Update your working copy","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Once you have merged your branch by accepting the PR on GitHub, you should clean up your local working copy of the repository by performing the following steps:","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Update your clone by running git fetch.\nCheck out main using git checkout main.\nDelete merged branch locally with git branch -d yourbranch.\nRemove local references to deleted remote branch by executing git remote prune origin.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"You can now proceed with your next changes by starting again at the top.","category":"page"},{"location":"github-git/#Using-Git","page":"GitHub & Git","title":"Using Git","text":"","category":"section"},{"location":"github-git/#Resources-for-learning-Git","page":"GitHub & Git","title":"Resources for learning Git","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Here are a few resources for learning do use Git that at least one of us found helpful in the past (roughly ordered from novice to advanced to expert):","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Git Handbook by GitHub\nLearn Git Branching","category":"page"},{"location":"github-git/#Tips-and-tricks","page":"GitHub & Git","title":"Tips and tricks","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This is an unordered collection of different tips and tricks that can be helpful while working with Git. As usual, your mileage might vary.","category":"page"},{"location":"github-git/#Undo-last-commit","page":"GitHub & Git","title":"Undo last commit","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you made a mistake in your last commit, e.g., by committing an unwanted file, you can undo the latest commit by running","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"git reset HEAD~","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This only works if you have not yet pushed your branch to the GitHub repository. In this case, please talk to one of the core developers on how to proceed. Especially when you accidentally committed a large file (image, or video), please let us know as fast as possible, since the effort to fix the repository grows considerably over time.","category":"page"},{"location":"github-git/#Remove-large-file-from-repository","page":"GitHub & Git","title":"Remove large file from repository","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If a large file was accidentally committed and pushed to the Trixi.jl repository, please talk to one of the core developers as soon as possible so that they can fix it.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"danger: Large files\nYou should never try to fix this yourself, as it potentially disrupts/destroys the work of others!","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Based on the instructions found here and here, the following steps need to be taken (as documented for GitLab in issue #33):","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Tell everyone to commit and push their changes to the repository.\nFix the branch in which the file was committed by removing it and committing the removal. This is especially important on main.\nPerform the following steps to clean up the Git repository:\ncd /tmp\n\n# Download bfg-1.13.0.jar from https://rtyley.github.io/bfg-repo-cleaner/\n\n# Get fresh clone of repo (so you can throw it away in case there is a problem)\ngit clone --mirror git@github.com:trixi-framework/Trixi.jl.git\n\n# Clean up repo of all files larger than 10M\njava -jar bfg-1.13.0.jar --strip-blobs-bigger-than 10M Trixi.jl.git\n\n# Enter repo\ncd Trixi.jl.git\n\n# Clean up reflog and force aggressive garbage collection\ngit reflog expire --expire=now --all && git gc --prune=now --aggressive\n\n# Push changes\ngit push\n\n# Delete clone\nrm -rf Trixi.jl.git\nTell everyone to clean up their local working copies by performing the following steps (also do this yourself):\n# Enter repo\ncd Trixi.jl\n\n# Get current changes\ngit fetch\n\n# Check out the fixed branch\ngit checkout branchname\n\n# IMPORTANT: Do a rebase instead of a pull!\ngit rebase\n\n# Clean reflog and force garbage collection\ngit reflog expire --expire=now --all && git gc --prune=now --aggressive\nIMPORTANT: You need to do a git rebase instead of a git pull when updating the fixed branch.","category":"page"},{"location":"testing/#Testing","page":"Testing","title":"Testing","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"During the development of Trixi.jl, we rely on continuous testing to ensure that modifications or new features do not break existing functionality or add other errors. In the main Trixi.jl repository (and the repositories for the visualization tool Trixi2Vtk), this is facilitated by GitHub Actions, which allows to run tests automatically upon certain events. When, how, and what is tested by GitHub Actions is controlled by the workflow file .github/workflows/ci.yml. In Trixi.jl and its related repositories, tests are triggered by","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"each git push to main and\neach git push to any pull request.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"Besides checking functionality, we also analyse the Test coverage to ensure that we do not miss important parts during testing.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Test and coverage requirements\nBefore merging a pull request (PR) to main, we require thatthe code passes all functional tests\ncode coverage does not decrease.","category":"page"},{"location":"testing/#Testing-setup","page":"Testing","title":"Testing setup","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"The entry point for all testing is the file test/runtests.jl, which is run by the automated tests and which can be triggered manually by executing","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"julia> using Pkg; Pkg.test(\"Trixi\")","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"in the REPL. Since there already exist many tests, we have split them up into multiple files in the test directory to allow for faster testing of individual parts of the code. Thus in addition to performing all tests, you can also just include one of the files named test_xxx.jl to run only a specific subset, e.g.,","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"julia> # Run all 2D tests on the P4estMesh\n include(joinpath(\"test\", \"test_p4est_2d.jl\"))\n\njulia> # Run all 1D tests for the Euler equations on the TreeMesh\n include(joinpath(\"test\", \"test_tree_1d_euler.jl\"))","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"For the automated tests with GitHub Actions, we run multiple jobs in parallel to reduce the waiting time until all tests are finished. You can see the different components that are run as jobs by looking at the TRIXI_TEST variable in test/runtests.jl.","category":"page"},{"location":"testing/#Adding-new-tests","page":"Testing","title":"Adding new tests","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"We use Julia's built-in unit testing capabilities to configure tests. In general, newly added code must be covered by at least one test, and all new elixirs added to the examples/ directory must be used at least once during testing. New tests should be added to the corresponding test/test_xxx.jl file, e.g., a test involving the 3D linear advection equation on the TreeMesh would go into test/test_tree_3d_advection.jl. Please study one of the existing tests and stay consistent to the current style when creating new tests.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"Since we want to test as much as possible, we have a lot of tests and frequently create new ones. Naturally, this increases the time to wait for all tests to pass with each novel feature added to Trixi.jl. Therefore, new tests should be as short as reasonably possible, i.e., without being too insensitive to pick up changes or errors in the code.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"When you add new tests, please check whether all CI jobs still take approximately the same time. If the job where you added new tests takes much longer than everything else, please consider moving some tests from one job to another (or report this incident and ask the main developers for help).","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Test duration\nAs a general rule, tests should last no more than 10 seconds when run with a single thread and after compilation (i.e., excluding the first run).","category":"page"},{"location":"testing/#Test-coverage","page":"Testing","title":"Test coverage","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"In addition to ensuring that the code produces the expected results, the automated tests also record the code coverage. The resulting coverage reports, i.e., which lines of code were executed by at least one test and are thus considered \"covered\" by testing, are automatically uploaded to Coveralls for easy analysis. Typically, you see a number of Coveralls results at the bottom of each pull request: One for each parallel job (see Testing setup), which can usually be ignored since they only cover parts of the code by definition, and a cumulative coverage result named coverage/coveralls. The \"Details\" link takes you to a detailed report on which lines of code are covered by tests, which ones are missed, and especially which new lines the pull requests adds to Trixi.jl's code base that are not yet covered by testing.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Coverage requirements\nIn general, we require pull requests to not decrease the overall test coverage percentage in main, with a hard lower bound of 97%.","category":"page"},{"location":"authors/#Authors","page":"Authors","title":"Authors","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"Trixi.jl's development is coordinated by a group of principal developers, who are also its main contributors and who can be contacted in case of questions about Trixi.jl. In addition, there are contributors who have provided substantial additions or modifications. Together, these two groups form \"The Trixi.jl Authors\" as mentioned under License.","category":"page"},{"location":"authors/#Principal-Developers","page":"Authors","title":"Principal Developers","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"Michael Schlottke-Lakemper, RWTH Aachen University/High-Performance Computing Center Stuttgart (HLRS), Germany\nGregor Gassner, University of Cologne, Germany\nHendrik Ranocha, Johannes Gutenberg University Mainz, Germany\nAndrew Winters, Linköping University, Sweden\nJesse Chan, Rice University, US","category":"page"},{"location":"authors/#Contributors","page":"Authors","title":"Contributors","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"The following people contributed major additions or modifications to Trixi.jl and are listed in alphabetical order:","category":"page"},{"location":"authors/","page":"Authors","title":"Authors","text":"Maximilian D. Bertrand\nBenjamin Bolm\nSimon Candelaresi\nJesse Chan\nLars Christmann\nChristof Czernik\nDaniel Doehring\nPatrick Ersing\nErik Faulhaber\nGregor Gassner\nLucas Gemein\nSven Goldberg\nJoshua Lampert\nJulia Odenthal\nSigrun Ortleb\nHendrik Ranocha\nAndrés M. Rueda-Ramírez\nFelipe Santillan\nMichael Schlottke-Lakemper\nToskan Theine\nAndrew Winters","category":"page"},{"location":"license/#License","page":"License","title":"License","text":"","category":"section"},{"location":"license/","page":"License","title":"License","text":"MIT LicenseCopyright (c) 2020-present The Trixi.jl Authors (see Authors)Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the \"Software\"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.","category":"page"},{"location":"restart/#restart","page":"Restart simulation","title":"Restart simulation","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"You can continue running an already finished simulation by first preparing the simulation for the restart and then performing the restart. Here we suppose that in the first run your simulation stops at time 1.0 and then you want it to run further to time 2.0.","category":"page"},{"location":"restart/#restart_preparation","page":"Restart simulation","title":"Prepare the simulation for a restart","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"In you original elixir you need to specify to write out restart files. Those will later be read for the restart of your simulation. This is done almost the same way as writing the snapshots using the SaveSolutionCallback callback. For the restart files it is called SaveRestartCallback:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"save_restart = SaveRestartCallback(interval=100,\n save_final_restart=true)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Make this part of your CallbackSet.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"An example is examples/examples/structured_2d_dgsem/elixir_advection_extended.jl.","category":"page"},{"location":"restart/#restart_perform","page":"Restart simulation","title":"Perform the simulation restart","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Since all of the information about the simulation can be obtained from the last snapshot, the restart can be done with relatively few lines in an extra elixir file. However, some might prefer to keep everything in one elixir and conditionals like if restart with a boolean variable restart that is user defined.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"First we need to define from which file we want to restart, e.g.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"restart_file = \"restart_000021.h5\"\nrestart_filename = joinpath(\"out\", restart_file)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Then we load the mesh file:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"mesh = load_mesh(restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"This is then needed for the semidiscretization:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"We then define a new time span for the simulation that takes as starting time the one form the snapshot:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"tspan = (load_time(restart_filename), 2.0)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"We now also take the last dt, so that our solver does not need to first find one to fulfill the CFL condition:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"dt = load_dt(restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"The ODE that we will pass to the solver is now:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"ode = semidiscretize(semi, tspan, restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"You should now define a SaveSolutionCallback similar to the original simulation, but with save_initial_solution=false, otherwise our initial snapshot will be overwritten. If you are using one file for the original simulation and the restart you can reuse your SaveSolutionCallback, but need to set","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"save_solution.condition.save_initial_solution = false","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Before we compute the solution using OrdinaryDiffEq.jl we need to set the integrator and its time step number, e.g.:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"integrator = init(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=dt, save_everystep=false, callback=callbacks);\nload_timestep!(integrator, restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Now we can compute the solution:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"sol = solve!(integrator)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"An example is in examples/structured_2d_dgsem/elixir_advection_restart.jl.","category":"page"},{"location":"meshes/structured_mesh/#Structured-mesh","page":"Structured mesh","title":"Structured mesh","text":"","category":"section"},{"location":"meshes/structured_mesh/","page":"Structured mesh","title":"Structured mesh","text":"The StructuredMesh is a structured, curvilinear, conforming mesh type available for one-, two-, and three-dimensional simulations.","category":"page"},{"location":"meshes/structured_mesh/","page":"Structured mesh","title":"Structured mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"development/#Development","page":"Development","title":"Development","text":"","category":"section"},{"location":"development/#interactive-use-of-julia","page":"Development","title":"Interactive use of Julia","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"When a Julia program is executed, Julia first loads and parses all code. Then, the just-in-time compiler has to compile all functions at their first use, which incurs an overhead each time a program is run. For proper packages and commands executed in the REPL (= \"return-eval-print loop\", which is what the Julia community calls the interactive command-line prompt that opens when executing julia without any files as arguments), however, the previously compiled functions are cached. Therefore, Trixi.jl should generally always be used interactively from the REPL without closing Julia during development, as it allows much faster turnaround times.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"If you naively run Trixi.jl from the REPL, you will not be able to change your Trixi.jl source files and then run the changed code without restarting the REPL, which destroys any potential benefits from caching. However, restarting Julia can be avoided by using the Revise.jl package, which tracks changed files and re-loads them automatically. Therefore, it is highly recommended to first install Revise with the following command in Julia: To enter the package REPL mode, press ] in the standard Julia REPL mode. Then, execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add Revise","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Now you are able to run Trixi.jl from the REPL, change Trixi.jl code between runs, and enjoy the advantages of the compilation cache! Before you start using Revise regularly, please be aware of some of the Pitfalls when using Revise.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Another recommended package for working from the REPL is OhMyREPL.jl. It can be installed by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add OhMyREPL","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"and adds syntax highlighting, bracket highlighting, and other helpful improvements for using Julia interactively. To automatically use OhMyREPL when starting the REPL, follow the instructions given in the official documentation.","category":"page"},{"location":"development/#Running-Trixi.jl-interactively-in-the-global-environment","page":"Development","title":"Running Trixi.jl interactively in the global environment","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you've installed Trixi.jl and Revise in your default environment, begin by executing:","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"This will start the Julia REPL. Then, run","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Revise; using Trixi","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"You can run a simulation by executing","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> trixi_include(default_example())","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Together, all of these commands can take some time, roughly half a minute on a modern workstation. Most of the time is spent on compilation of Julia code etc. If you execute the last command again in the same REPL, it will finish within a few milliseconds (maybe ~45 on a modern workstation). This demonstrates the second reason for using the REPL: the compilation cache. That is, those parts of the code that do not change between two Trixi.jl runs do not need to be recompiled and thus execute much faster after the first run.","category":"page"},{"location":"development/#Manually-starting-Trixi.jl-in-the-local-environment","page":"Development","title":"Manually starting Trixi.jl in the local environment","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you followed the installation instructions for developers, execute Julia with the project directory set to the run directory of the program/tool you want to use. For example, to run Trixi.jl this way, you need to start the REPL with","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia --project=path/to/Trixi.jl/run","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"and execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Revise; using Trixi","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"to load Revise and Trixi.jl. You can then proceed with the usual commands and run Trixi.jl as in the example above. The --project flag is required such that Julia can properly load Trixi.jl and all dependencies if Trixi.jl is not installed in the global environment. The same procedure also applies should you opt to install the postprocessing tool Trixi2Vtk manually such that you can modify their implementations.","category":"page"},{"location":"development/#Pitfalls-when-using-Revise","page":"Development","title":"Pitfalls when using Revise","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"While Revise is a great help for developing Julia code, there are a few situations to watch out for when using Revise. The following list of potential issues is based on personal experiences of the Trixi.jl developers and probably incomplete. Further information on limitations and possible issues with Revise can be found in the official documentation.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"tip: If in doubt, restart the REPL\nOftentimes, it is possible to recover from issues with Revise by fixing the offending code. Sometimes, however, this is not possible or you might have troubles finding out what exactly caused the problems. Therefore, in these cases, or if in doubt, restart the REPL to get a fresh start.","category":"page"},{"location":"development/#Syntax-errors-are-easy-to-miss","page":"Development","title":"Syntax errors are easy to miss","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Revise does not stop on syntax errors, e.g., when you accidentally write a[i) instead of a[i]. In this case, Revise reports an error but continues to use the old version of your files! This is especially dangerous for syntax errors, as they are detected while Revise reloads changed code, which happens in the beginning of a new execution. Thus, the syntax error message quickly disappears from the terminal once Trixi.jl starts writing output to the screen and you might not even have noticed that an error occurred at all.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Therefore, when you are deep in a coding/debugging session and wonder why your code modifications do not seem to have any effect, scroll up in your terminal to check if you missed earlier syntax errors, or - if in doubt - restart your REPL.","category":"page"},{"location":"development/#Files-are-not-tracked-after-changing-branches","page":"Development","title":"Files are not tracked after changing branches","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Sometimes, Revise stops tracking files when changing the Git branch. That is, modifications to Trixi.jl's source files will not be reloaded by Revise and thus have no effect of a currently running REPL session. This issue is particularly annoying for a developer, since it does not come with any warning! Therefore, it is good practice to always restart the REPL after changing branches.","category":"page"},{"location":"development/#Changes-to-type-definitions-are-not-allowed","page":"Development","title":"Changes to type definitions are not allowed","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Revise cannot handle changes to type definitions, e.g., when modifying the fields in a struct. In this case, Revise reports an error and refuses to run your code unless you undo the modifications. Once you undo the changes, Revise will usually continue to work as expected again. However, if you want to keep your type modifications, you need to restart the REPL.","category":"page"},{"location":"development/#Using-the-Julia-REPL-effectively","page":"Development","title":"Using the Julia REPL effectively","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"The Julia manual is an excellent resource to learn Julia. Here, we list some helpful commands than can increase your productivity in the Julia REPL.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Use the REPL help mode entered by typing ?.\njulia> using Trixi\n\nhelp?> trixi_include\nsearch: trixi_include\n\n trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)\n\n include the file elixir and evaluate its content in the global scope of module mod. You can override specific\n assignments in elixir by supplying keyword arguments. It's basic purpose is to make it easier to modify some\n parameters while running Trixi.jl from the REPL. Additionally, this is used in tests to reduce the computational\n burden for CI while still providing examples with sensible default values for users.\n\n Examples\n ≡≡≡≡≡≡≡≡≡≡\n\n julia> trixi_include(@__MODULE__, default_example(), tspan=(0.0, 0.1))\n [...]\n\n julia> sol.t[end]\n 0.1\nYou can copy and paste REPL history including julia> prompts into the REPL.\nUse tab completion in the REPL, both for names of functions/types/variables and for function arguments.\njulia> flux_ranocha( # and TAB\nflux_ranocha(u_ll, u_rr, orientation::Integer, equations::CompressibleEulerEquations1D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_1d.jl:390\nflux_ranocha(u_ll, u_rr, orientation::Integer, equations::CompressibleEulerEquations2D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_2d.jl:839\n[...]\nUse methodswith to discover methods associated to a given type etc.\njulia> methodswith(CompressibleEulerEquations2D)\n[1] initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations2D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_2d.jl:51\n[...]\nUse @which (or @edit) for method calls.\njulia> @which trixi_include(default_example())\ntrixi_include(elixir::AbstractString; kwargs...) in Trixi at ~/.julia/dev/Trixi/src/auxiliary/special_elixirs.jl:36\nUse apropos to search through the documentation and docstrings.\njulia> apropos(\"MHD\")\nTrixi.IdealGlmMhdEquations3D\nTrixi.IdealGlmMhdMulticomponentEquations2D\nTrixi.calc_fast_wavespeed_roe\nTrixi.IdealGlmMhdEquations1D\nTrixi.initial_condition_constant\nTrixi.flux_nonconservative_powell\nTrixi.GlmSpeedCallback\nTrixi.flux_derigs_etal\nTrixi.flux_hindenlang_gassner\nTrixi.initial_condition_convergence_test\nTrixi.min_max_speed_naive\nTrixi.IdealGlmMhdEquations2D\nTrixi.IdealGlmMhdMulticomponentEquations1D\n[...]","category":"page"},{"location":"development/#Text-editors","page":"Development","title":"Text editors","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"When writing code, the choice of text editor can have a significant impact on productivity and developer satisfaction. While using the default text editor of the operating system has its own benefits (specifically the lack of an explicit installation procure), usually it makes sense to switch to a more programming-friendly tool. In the following, a few of the many options are listed and discussed:","category":"page"},{"location":"development/#VS-Code","page":"Development","title":"VS Code","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Visual Studio Code is a modern open source editor with good support for Julia. While Juno had some better support in the past, the developers of Juno and the Julia VS Code plugin are joining forces and concentrating on VS Code since support of Atom has been suspended. Basically, all comments on Juno below also apply to VS Code.","category":"page"},{"location":"development/#Juno","page":"Development","title":"Juno","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you are new to programming or do not have a preference for a text editor yet, Juno is a good choice for developing Julia code. It is based on Atom, a sophisticated and widely used editor for software developers, and is enhanced with several Julia-specific features. Furthermore and especially helpful for novice programmers, it has a MATLAB-like appearance with easy and interactive access to the current variables, the help system, and a debugger.","category":"page"},{"location":"development/#Vim-or-Emacs","page":"Development","title":"Vim or Emacs","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Vim and Emacs are both very popular editors that work great with Julia. One of their advantages is that they are text editors without a GUI and as such are available for almost any operating system. They also are preinstalled on virtually all Unix-like systems. However, Vim and Emacs come with their own, steep learning curve if they have never been used before. Therefore, if in doubt, it is probably easier to get started with a classic GUI-based text editor (like Juno). If you decide to use Vim or Emacs, make sure that you install the corresponding Vim plugin julia-vim or Emacs major mode julia-emacs.","category":"page"},{"location":"development/#Debugging","page":"Development","title":"Debugging","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Julia offers several options for debugging. A classical debugger is available with the Debugger.jl package or in the Julia extension for VS Code. However, it can be quite slow and, at the time of writing (January 2023), currently does not work properly with Trixi.jl. The Infiltrator.jl package on the other hand does not offer all features of a full debugger, but is a fast and simple tool that allows users to set breakpoints to open a local REPL session and access the call stack and variables.","category":"page"},{"location":"development/#Infiltrator","page":"Development","title":"Infiltrator","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"The Infiltrator package provides fast, interactive breakpoints using the @infiltrate command, which drops the user into a local REPL session. From there, it is possible to access local variables, see the call stack, and execute statements.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"The package can be installed in the Julia REPL by executing","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add Infiltrator","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"To load the package in the Julia REPL execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Infiltrator","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Breakpoints can be set by adding a line with the @infiltrate macro at the respective position in the code. Use Revise if you want to set and delete breakpoints in your package without having to restart Julia.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"note: Use `@autoinfiltrate` when debugging Trixi.jl\nWhen running Julia inside a package environment, e.g., inside the source code of Trixi.jl itself, the @infiltrate macro only works if Infiltrator has been added to the package dependencies. To avoid this, you can use the (non-exported) @autoinfiltrate macro in Trixi.jl, which only requires Infiltrator.jl to be available in the current environment stack and will auto-load it for you.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Triggering the breakpoint starts a REPL session where it is possible to interact with the current local scope. Possible commands are:","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"@locals: Print the local variables.\n@exfiltrate: Save the local variables to a global storage, which can be accessed with the safehouse variable outside the Infiltrator session.\n@trace: Print the current stack trace.\nExecute other arbitrary statements\n?: Print a help list with all options","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"To finish a debugging session, either use @continue to continue and eventually stop at the next breakpoint or @exit to skip further breakpoints. After the code has finished, local variables saved with @exfiltrate can be accessed in the REPL using the safehouse variable.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Limitations of using Infiltrator.jl are that local variables cannot be changed, and that it is not possible to step into further calls or access other function scopes.","category":"page"},{"location":"development/#Releasing-a-new-version-of-Trixi.jl,-Trixi2Vtk","page":"Development","title":"Releasing a new version of Trixi.jl, Trixi2Vtk","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Check whether everything is okay, tests pass etc.\nSet the new version number in Project.toml according to the Julian version of semver. Commit and push.\nComment @JuliaRegistrator register on the commit setting the version number.\nJuliaRegistrator will create a PR with the new version in the General registry. Wait for it to be merged.\nIncrement the version number in Project.toml again with suffix -pre. For example, if you have released version v0.2.0, use v0.2.1-pre as new version number.\nWhen a new version of Trixi.jl was released, check whether the [compat] entries in test/Project.toml in Trixi2Vtk should be updated. When a new version of Trixi2Vtk was released, check whether the [compat] entries in docs/Project.toml in Trixi.jl should be updated.\nThese entries will also be checked regularly by CompatHelper (once a day). Hence, if everything was released correctly, you should only need to do these checks manually if new minor versions with changes in the docs of Trixi2Vtk were released but no new version of Trixi.jl was released afterwards.","category":"page"},{"location":"development/#Preview-of-the-documentation","page":"Development","title":"Preview of the documentation","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"You can build the documentation of Trixi.jl locally by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia --project=docs -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd())); Pkg.instantiate()'\njulia --project=docs --color=yes docs/make.jl","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"from the Trixi.jl main directory. Then, you can look at the html files generated in docs/build. For PRs triggered from branches inside the Trixi.jl main repository previews of the new documentation are generated at https://trixi-framework.github.io/Trixi.jl/previews/PRXXX, where XXX is the number of the PR. This does not work for PRs from forks for security reasons (since anyone could otherwise push arbitrary stuff to the Trixi.jl website, including malicious code).","category":"page"},{"location":"development/#trixi2vtk-dev","page":"Development","title":"Developing Trixi2Vtk","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Trixi2Vtk has Trixi.jl as dependency and uses Trixi.jl's implementation to, e.g., load mesh files. When developing Trixi2Vtk, one may want to change functions in Trixi.jl to allow them to be reused in Trixi2Vtk. To use a locally modified Trixi.jl clone instead of a Trixi.jl release, one can tell Pkg to use the local source code of Trixi.jl instead of a registered version by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> develop path/to/Trixi.jl","category":"page"},{"location":"visualization/#visualization","page":"Visualization","title":"Visualization","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"There are two possible approaches to visualize results from Trixi.jl: either directly from the REPL using Plots.jl or with ParaView/VisIt by postprocessing Trixi.jl's output files with Trixi2Vtk.","category":"page"},{"location":"visualization/#Plots.jl","page":"Visualization","title":"Plots.jl [experimental]","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By far the easiest and most convenient plotting approach is to use the powerful Plots.jl package to directly visualize Trixi.jl's results from the REPL. In the following, you will find more information on a number of topics for how to use Plots.jl together with Trixi.jl:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Getting started\nCustomizing plot results via a PlotData2D object\nPlotting a 3D solution as a 2D plot\nCreating a 1D plot\nPlotting a 2D or 3D solutions as a 1D plot\nVisualizing results during a simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nPlotting via Plots.jl is still considered an experimental feature and might change in any future releases.","category":"page"},{"location":"visualization/#getting-started-plots-jl","page":"Visualization","title":"Getting started","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"After running a simulation with Trixi.jl in the REPL, load the Plots package with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Plots","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize the solution, execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Here we assume that sol holds the return value of the solve(...) method (with type SciMLBase.ODESolution), which is the default variable name when you use one of the example elixirs. This will generate a grid layout with one subplot for each solution variable, convenient for getting an overview of the current solution:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You can save the resulting file as a PNG image file by calling savefig(...) with an output file name that ends in .png, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> savefig(\"solution-overview.png\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In Trixi.jl, two plot types are available: 2D heatmap plots and 1D line plots. If you use plot(sol), Trixi.jl will automatically choose the plot type that fits the dimensions of the sol input: 2D/3D data will be visualized as a heatmap, 1D data as a line plot. For more fine-grained control over what to plot, you can create such an object yourself, which can either be a PlotData2D or a PlotData1D object. For further details on both of these see below:","category":"page"},{"location":"visualization/#Customizing-plot-results-via-a-PlotData2D-object","page":"Visualization","title":"Customizing plot results via a PlotData2D object","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For more fine-grained control over what to plot, first create a PlotData2D object by executing","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol)\njulia> pd = PlotData2D(u, semi)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"where u is an array containing the solution and semi is the semidiscretization. For example, if PlotData2D(sol.u[2], semi) is specified, this will create a PlotData2D instance from the 2nd saved time-step. If PlotData2D(sol(0.5), semi) is specified, it will construct a PlotData2D instance using OrdinaryDiffEq.jl's interpolation to evaluate the solution at time t=0.5.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This takes the results generated by Trixi.jl and stores them in a data format that can be understood by the Plots package, and pd holds all data relevant for plotting sol. You can pass variable names as strings to pd using a dictionary-like syntax, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd[\"rho\"])","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will create a single 2D heatmap plot of the variable rho:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The default plot type and style can be overridden by passing any additional arguments that are understood by the Plots package. For example, to change the color scheme and add names to the axes, modify the previous command to","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd[\"rho\"], seriescolor = :heat, xguide=\"x\", yguide=\"y\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to yield","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-modified)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For more details on the various format options for plot, please consult the Plots documentation.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In addition, you can plot the mesh lines on top of the solution variables by calling the getmesh(...) function on the PlotData2D object","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot!(getmesh(pd)) # here we use `plot!` with an `!` to add to the previous plot","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"which modifies the previous plot to","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-modified-mesh)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default, PlotData2D will convert the conserved variables to primitive variables, but this can be changed by passing an appropriate conversion function in the solution_variables keyword argument, similar to the behavior of the SaveSolutionCallback:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol; solution_variables=cons2cons) # Plot conservative variables","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"There are several other keyword arguments that influence how the solution data is processed for visualization with the Plots package. A detailed explanation can be found in the docstring of the PlotData2D constructor.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Another way to change the appearance of a plot is to convert the solution to a uniformly refined mesh before plotting. This can be helpful, e.g., when trying different settings for a simulation with adaptive mesh refinement, where one would like to ignore the mesh changes when comparing solutions. This is achieved with adapt_to_mesh_level, which uses the mesh adaptation routines to adapt the solution to a uniform grid. For example, the AMR solution from above could be preprocessed with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(adapt_to_mesh_level(sol, 4)...)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When plotted together with the mesh, this will yield the following visualization:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-uniform-mesh)","category":"page"},{"location":"visualization/#Plotting-a-user-defined-scalar-field","page":"Visualization","title":"Plotting a user-defined scalar field","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To plot a scalar quantity, one can call plot(ScalarPlotData2D(u, semi)), where u is an array of nodal values of the scalar field to plot. The layout of u should match the layout of the x and y nodal coordinates of the respective solver. For example, after running trixi_include(joinpath(\"examples\", \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\")), the following can be used to plot the function f(x, y) = x * y:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"x = view(semi.cache.elements.node_coordinates, 1, :, :, :)\ny = view(semi.cache.elements.node_coordinates, 2, :, :, :)\nplot(ScalarPlotData2D(x .* y, semi))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This produces the following plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: scalar-plotting-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This routine can be used to visualize scalar quantities which depend on the solution, such as the norm of a velocity vector or two-dimensional vorticity. For example, we can visualize vorticity for a compressible version of the Brown-Minion vortex problem:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi, Plots\n\njulia> redirect_stdout(devnull) do\n # runs the elixir without any output from callbacks etc.\n trixi_include(@__MODULE__,\n joinpath(examples_dir(), \"dgmulti_2d\", \"elixir_euler_brown_minion_vortex.jl\"))\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n\njulia> function compute_vorticity(velocity, mesh, equations::CompressibleEulerEquations2D,\n dg::DGMulti, cache)\n rd = dg.basis\n md = mesh.md\n @unpack Dr, Ds = rd\n @unpack rxJ, sxJ, ryJ, syJ, J = md\n v1, v2 = velocity\n dv1dy = ryJ .* (Dr * v1) + syJ .* (Ds * v1)\n dv2dx = rxJ .* (Dr * v2) + sxJ .* (Ds * v2)\n return dv2dx - dv1dy\n end;\n\njulia> compute_vorticity(velocity, semi) =\n compute_vorticity(velocity, Trixi.mesh_equations_solver_cache(semi)...);\n\njulia> function get_velocity(sol)\n rho, rhou, rhov, E = StructArrays.components(sol.u[end])\n v1 = rhou ./ rho\n v2 = rhov ./ rho\n return v1, v2\n end;\n\njulia> vorticity = compute_vorticity(get_velocity(sol), semi);\n\njulia> plot(ScalarPlotData2D(vorticity, semi;\n variable_name = \"Vorticity at t = $(sol.prob.tspan[end])\"))\nPlot{Plots.GRBackend() n=1}","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This produces the following plot of vorticity.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: vorticity-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Since the mesh is fairly coarse, we observe numerical artifacts due to the low resolution. These errors vanish under mesh refinement; for example, doubling the mesh resolution by running","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(examples_dir(), \"dgmulti_2d\", \"elixir_euler_BM_vortex.jl\"), cells_per_dimension = 32)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"yields the following plot of vorticity:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: vorticity-example-refined)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nWhen visualizing a scalar field, the plotted solution is reinterpolated using a high order polynomial approximation. Thus, small discrepancies may be observed when the underlying data is highly non-smooth or under-resolved.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"ScalarPlotData2D objects can also be used with Makie through iplot. For example, the following code plots two surfaces:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi, CairoMakie\n\njulia> redirect_stdout(devnull) do\n # runs the elixir without any output from callbacks etc.\n trixi_include(@__MODULE__,\n joinpath(examples_dir(), \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\"))\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n\njulia> x = view(semi.cache.elements.node_coordinates, 1, :, :, :); # extracts the node x coordinates\n\njulia> y = view(semi.cache.elements.node_coordinates, 2, :, :, :); # extracts the node y coordinates\n\njulia> fig_ax_plt = iplot(ScalarPlotData2D((@. 1 - .25*(x^2 + y^2)), semi), plot_mesh=true, colormap=:viridis);\n\njulia> fig_ax_plt2 = iplot!(fig_ax_plt, ScalarPlotData2D((@. .125*(x+y)), semi), plot_mesh=true, colormap=:blues)\nFigureAxisPlot()","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This creates the following plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: ScalarPlotData2D_example)","category":"page"},{"location":"visualization/#Plotting-a-3D-solution-as-a-2D-plot","page":"Visualization","title":"Plotting a 3D solution as a 2D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"It is possible to plot 2D slices from 3D simulation data using the TreeMesh with the same commands as above:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol) # `sol` is from a 3D simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default, plotting sol or creating a PlotData2D object from a 3D simulation will create a 2D slice of the solution in the xy-plane. You can customize this behavior by explicitly creating a PlotData2D object and passing appropriate keyword arguments:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"slice specifies the plane which is being sliced and can be :xy, :xz, or :yz (default: :xy)\npoint specifies a three-dimensional point. The sliced plane is then created such that it lies on the point (default: (0.0, 0.0, 0.0)).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"All other attributes for PlotData2D objects apply here as well.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For example, to plot the velocity field orthogonal to the yz-plane at different x-axis locations, you can execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(examples_dir(), \"tree_3d_dgsem\", \"elixir_euler_taylor_green_vortex.jl\"), tspan=(0.0, 1.0))\n[...]\n\njulia> plots = []\nAny[]\n\njulia> for x in range(0, stop=pi/2, length=6)\n pd = PlotData2D(sol, slice=:yz, point=(x, 0.0, 0.0))\n push!(plots, plot(pd[\"v1\"], clims=(-1,1), title=\"x = \"*string(round(x, digits=2))))\n end\n\njulia> plot(plots..., layout=(2, 3), size=(750,350))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"which results in a 2x3 grid of slices of the yz-plane:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-v1-0.0-to-0.5pi)","category":"page"},{"location":"visualization/#Creating-a-1D-plot","page":"Visualization","title":"Creating a 1D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When plotting a 1D solution with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol) # `sol` is from a 1D simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi.jl automatically creates a PlotData1D object and visualizes it as a line plot: (Image: 1d-plot)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To customize your 1D plot, you can create a PlotData1D object manually as follows:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol)\njulia> pd = PlotData1D(u, semi)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The behavior is analogous to the PlotData2D behavior.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In a very similar fashion to PlotData2D, you can customize your plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"plot(pd) creates the same plot as in plot(sol).\nplot(pd[\"rho\", \"p\"]) only plots specific variables. In this case rho and p.\nplot!(getmesh(pd)) adds mesh lines after creating a plot.\nAny attributes from Plots can be used, e.g., plot(pd, yguide=:temperature).\npd = PlotData1D(adapt_to_mesh_level(sol, 4)...) adapts the mesh before plotting (in this example to a mesh with refinement level 4).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You can also customize the PlotData1D object itself by passing attributes to the PlotData1D constructor:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"solution_variables specifies the variables to be plotted.\nnvisnodes sets the amount of nodes per element which the solution then is interpolated on.","category":"page"},{"location":"visualization/#Plotting-a-2D-or-3D-solutions-as-a-1D-plot","page":"Visualization","title":"Plotting a 2D or 3D solutions as a 1D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"It is possible to extract a straight, axis-parallel line from a 2D or 3D solution and visualize it as a 1D plot. This is done by creating a PlotData1D object with a 2D/3D solution sol as input:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot is then created with:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default the x-axis is extracted, which can be changed with following attributes:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"slice specifies the axis which is being extracted and can be :x, :y or :z (:z is only for 3D input and default is :x)\npoint specifies a two or three dimensional point. The sliced axis is then created in such a way, that it lies on the point. (default: (0.0, 0.0) or (0.0, 0.0, 0.0))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"All other attributes for PlotData1D objects apply here as well.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In the following, is an example for a 2D simulation of the linear scalar advection equation. First, we have the regular 2D heatmap plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 2d-plot-for-slice)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"From this, we can extract a line plot parallel to the y-axis going through the point (1.0, 0.0) with the following commands:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol, slice=:y, point=(1.0, 0.0))\njulia> plot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 1d-plot-for-slice)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This convenient method of slicing is limited to axis-parallel slices, but for 2D/3D solutions it is also possible to create a plot along any curve you want. To do so, you first need to create a list of 2D/3D points that define your curve. Then you can create a PlotData1D with the keyword argument curve set to your list.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Let's give an example of this with the basic advection equation from above by creating a plot along the circle marked in green:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 2d-plot-along-circle)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"We can write a function like this, that outputs a list of points on a circle:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"function circle(radius, center, n_points)\n coordinates = zeros(2, n_points)\n for i in 1:n_points\n coordinates[:,i] = radius*[cospi(2*i/n_points), sinpi(2*i/n_points)] .+ center\n end\n return coordinates\nend","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Then create and plot a PlotData1D object along a circle with radius one, center at (1,1), and 100 points:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"pd = PlotData1D(sol, curve=circle(1.0, (1.0, 1.0), 100))\nplot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This gives you the following plot: (Image: 1d-plot-along-circle)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Creating a plot like this has its downsides. For one, it is unclear what to put on the abscissa of the plot. By default, the arc length of the given curve is used. Also, with this way of plotting you lose the ability to use a mesh plot from getmesh.","category":"page"},{"location":"visualization/#Visualizing-results-during-a-simulation","page":"Visualization","title":"Visualizing results during a simulation","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize solutions while a simulation is still running (also known as in-situ visualization), you can use the VisualizationCallback. It is created as a regular callback and accepts upon creation a number of keyword arguments that allow, e.g., to control the visualization interval, to specify the variables to plot, or to customize the plotting style.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"During the simulation, the visualization callback creates and displays visualizations of the current solution in regular intervals. This can be useful to, e.g., monitor the validity of a long-running simulation or for illustrative purposes. An example for how to create a VisualizationCallback can be found in examples/tree_2d_dgsem/elixir_advection_amr_visualization.jl:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"[...]\n\n# Enable in-situ visualization with a new plot generated every 20 time steps\n# and additional plotting options passed as keyword arguments\nvisualization = VisualizationCallback(interval=20; clims=(0,1))\n\n[...]","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The resulting output of the referenced elixir can be seen in the embedded video below:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":" \n
","category":"page"},{"location":"visualization/#Trixi2Vtk","page":"Visualization","title":"Trixi2Vtk","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi2Vtk converts Trixi.jl's .h5 output files to VTK files, which can be read by ParaView, VisIt, and other visualization tools. It automatically interpolates solution data from the original quadrature node locations to equidistant visualization nodes at a higher resolution, to make up for the loss of accuracy from going from a high-order polynomial representation to a piecewise constant representation in VTK.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In the Julia REPL, first load the package Trixi2Vtk","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi2Vtk","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To process an HDF5 file generated by Trixi.jl, execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(joinpath(\"out\", \"solution_000000.h5\"), output_directory=\"out\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will create two unstructured VTK files in the out subdirectory that can be opened with ParaView or VisIt: solution_000000.vtu contains the discontinuous Galerkin solution data while solution_000000_celldata.vtu holds any cell-based values such as the current AMR indicator or the cell refinement level.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: \"solution_000000_scalar_mesh\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This allows you to generate VTK files for solution, restart and mesh files. By default, Trixi2Vtk generates .vtu (unstructured VTK) files for both cell/element data (e.g., cell ids, element ids) and node data (e.g., solution variables). This format visualizes each cell with the same number of nodes, independent of its size. Alternatively, you can provide format=:vti as a keyword argument to trixi2vtk, which causes Trixi2Vtk to generate .vti (image data VTK) files for the solution files, while still using .vtu files for cell-/element-based data. In .vti files, a uniform resolution is used throughout the entire domain, resulting in different number of visualization nodes for each element. This can be advantageous to create publication-quality images, but increases the file size.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"If you want to convert multiple solution/restart files at once, you can just supply multiple input files as the positional arguments to trixi2vtk, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/solution_000000.h5\", \"out/solution_000040.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You may also use file globbing to select a range of files based on filename patterns, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/solution_*.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to convert all solution files in the out/ directory or","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/restart_00[0-9]000.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to convert every one-thousandth restart file (out/restart_000000.h5, out/restart_001000.h5 etc.).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When multiple solution/restart files are provided, Trixi2Vtk will also generate a .pvd file, which allows ParaView to read all .vtu/.vti files at once and which uses the time attribute in solution/restart files to inform ParaView about the solution time. A comprehensive list of all possible arguments for trixi2vtk can be found in the Trixi2Vtk.jl API.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Further information regarding the development of Trixi2Vtk can be found in the development section.","category":"page"},{"location":"visualization/#Makie.jl","page":"Visualization","title":"Makie.jl [experimental]","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In addition to Plots.jl support, Trixi.jl includes visualization utilities through Makie.jl. Trixi.jl provides Makie-based visualization options both for heatmap-type plots (similar to the Plots.jl recipes) as well as for interactive surface plots. Support is currently limited to the UnstructuredMesh2D type.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nPlotting via Makie.jl is still considered an experimental feature and might change in any future releases.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"A Makie plot can be created as follows: after running a simulation with Trixi.jl in the REPL, load a Makie backend (for example, GLMakie or CairoMakie).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using GLMakie","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize the solution and mesh with a heatmap-type plot, simply run","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nBoth Makie.jl and Plots.jl export plot, so if you load both libraries, you will have to specify which plot function to call via Plots.plot or Makie.plot.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"As with Plots.jl recipes, one can view individual solution components by creating a PlotData2D object and indexing into it with the desired variable name","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol)\njulia> plot(pd[\"rho\"])","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Unlike the Plots.jl recipe, mesh plotting is controlled using the keyword argument plot_mesh = false, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol; plot_mesh=false)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot command also returns figure and axis handles, which can be used to edit plot titles or labels:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> fig, axes = plot(sol)\njulia> axes[1,1].title = \"New title for subplot (1,1)\"","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi.jl also supports interactive surface plots using iplot. After executing","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(\"examples\", \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\"))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"we can run","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> iplot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will open up an interactive visualization window:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: makie-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot can be rotated (click and hold), zoomed in and out (scroll up and down), and panned (hold right click and drag). Two toggle buttons control whether mesh lines are visible on top of and below the solution.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Both plot and iplot use colormap = :inferno by default. A different colormap can be selected by providing an appropriate keyword argument. For example, plot(sol, colormap=:blues) and iplot(sol, colormap=:blues) produce the following figures:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: makie-plot-example) (Image: makie-iplot-example)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"EditURL = \"../../literate/src/files/adding_new_parabolic_terms.jl\"","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#adding_new_parabolic_terms","page":"11 Adding new parabolic terms","title":"11: Adding new parabolic terms","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"This demo illustrates the steps involved in adding new parabolic terms for the scalar advection equation. In particular, we will add an anisotropic diffusion. We begin by defining the hyperbolic (advection) part of the advection-diffusion equation.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"using OrdinaryDiffEq\nusing Trixi\n\n\nadvection_velocity = (1.0, 1.0)\nequations_hyperbolic = LinearScalarAdvectionEquation2D(advection_velocity);\nnothing #hide","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Define-a-new-parabolic-equation-type","page":"11 Adding new parabolic terms","title":"Define a new parabolic equation type","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Next, we define a 2D parabolic diffusion term type. This is similar to LaplaceDiffusion2D except that the diffusivity field refers to a spatially constant diffusivity matrix now. Note that ConstantAnisotropicDiffusion2D has a field for equations_hyperbolic. It is useful to have information about the hyperbolic system available to the parabolic part so that we can reuse functions defined for hyperbolic equations (such as varnames).","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"The abstract type Trixi.AbstractEquationsParabolic has three parameters: NDIMS (the spatial dimension, e.g., 1D, 2D, or 3D), NVARS (the number of variables), and GradientVariable, which we set as GradientVariablesConservative. This indicates that the gradient should be taken with respect to the conservative variables (e.g., the same variables used in equations_hyperbolic). Users can also take the gradient with respect to a different set of variables; see, for example, the implementation of CompressibleNavierStokesDiffusion2D, which can utilize either \"primitive\" or \"entropy\" variables.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"struct ConstantAnisotropicDiffusion2D{E, T} <: Trixi.AbstractEquationsParabolic{2, 1, GradientVariablesConservative}\n diffusivity::T\n equations_hyperbolic::E\nend\n\nvarnames(variable_mapping, equations_parabolic::ConstantAnisotropicDiffusion2D) =\n varnames(variable_mapping, equations_parabolic.equations_hyperbolic)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Next, we define the viscous flux function. We assume that the mixed hyperbolic-parabolic system is of the form","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"partial_t u(tx) + partial_x (f_1(u) - g_1(u nabla u))\n + partial_y (f_2(u) - g_2(u nabla u)) = 0","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"where f_1(u), f_2(u) are the hyperbolic fluxes and g_1(u nabla u), g_2(u nabla u) denote the viscous fluxes. For anisotropic diffusion, the viscous fluxes are the first and second components of the matrix-vector product involving diffusivity and the gradient vector.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Here, we specialize the flux to our new parabolic equation type ConstantAnisotropicDiffusion2D.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"function Trixi.flux(u, gradients, orientation::Integer, equations_parabolic::ConstantAnisotropicDiffusion2D)\n @unpack diffusivity = equations_parabolic\n dudx, dudy = gradients\n if orientation == 1\n return SVector(diffusivity[1, 1] * dudx + diffusivity[1, 2] * dudy)\n else # if orientation == 2\n return SVector(diffusivity[2, 1] * dudx + diffusivity[2, 2] * dudy)\n end\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Defining-boundary-conditions","page":"11 Adding new parabolic terms","title":"Defining boundary conditions","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Trixi.jl's implementation of parabolic terms discretizes both the gradient and divergence using weak formulation. In other words, we discretize the system","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"beginaligned\nbmq = nabla u \nbmsigma = beginpmatrix g_1(u bmq) g_2(u bmq) endpmatrix \ntextviscous contribution = nabla cdot bmsigma\nendaligned","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Boundary data must be specified for all spatial derivatives, e.g., for both the gradient equation bmq = nabla u and the divergence of the viscous flux nabla cdot bmsigma. We account for this by introducing internal Gradient and Divergence types which are used to dispatch on each type of boundary condition.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"As an example, let us introduce a Dirichlet boundary condition with constant boundary data.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"struct BoundaryConditionConstantDirichlet{T <: Real}\n boundary_value::T\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"This boundary condition contains only the field boundary_value, which we assume to be some real-valued constant which we will impose as the Dirichlet data on the boundary.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Boundary conditions have generally been defined as \"callable structs\" (also known as \"functors\"). For each boundary condition, we need to specify the appropriate boundary data to return for both the Gradient and Divergence. Since the gradient is operating on the solution u, the boundary data should be the value of u, and we can directly impose Dirichlet data.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"@inline function (boundary_condition::BoundaryConditionConstantDirichlet)(flux_inner, u_inner, normal::AbstractVector,\n x, t, operator_type::Trixi.Gradient,\n equations_parabolic::ConstantAnisotropicDiffusion2D)\n return boundary_condition.boundary_value\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"While the gradient acts on the solution u, the divergence acts on the viscous flux bmsigma. Thus, we have to supply boundary data for the Divergence operator that corresponds to bmsigma. However, we've already imposed boundary data on u for a Dirichlet boundary condition, and imposing boundary data for bmsigma might overconstrain our problem.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Thus, for the Divergence boundary data under a Dirichlet boundary condition, we simply return flux_inner, which is boundary data for bmsigma computed using the \"inner\" or interior solution. This way, we supply boundary data for the divergence operation without imposing any additional conditions.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"@inline function (boundary_condition::BoundaryConditionConstantDirichlet)(flux_inner, u_inner, normal::AbstractVector,\n x, t, operator_type::Trixi.Divergence,\n equations_parabolic::ConstantAnisotropicDiffusion2D)\n return flux_inner\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#A-note-on-the-choice-of-gradient-variables","page":"11 Adding new parabolic terms","title":"A note on the choice of gradient variables","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"It is often simpler to transform the solution variables (and solution gradients) to another set of variables prior to computing the viscous fluxes (see CompressibleNavierStokesDiffusion2D for an example of this). If this is done, then the boundary condition for the Gradient operator should be modified accordingly as well.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Putting-things-together","page":"11 Adding new parabolic terms","title":"Putting things together","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"Finally, we can instantiate our new parabolic equation type, define boundary conditions, and run a simulation. The specific anisotropic diffusion matrix we use produces more dissipation in the direction (1 -1) as an isotropic diffusion.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"For boundary conditions, we impose that u=1 on the left wall, u=2 on the bottom wall, and u = 0 on the outflow walls. The initial condition is taken to be u = 0. Note that we use BoundaryConditionConstantDirichlet only for the parabolic boundary conditions, since we have not defined its behavior for the hyperbolic part.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"using Trixi: SMatrix\ndiffusivity = 5.0e-2 * SMatrix{2, 2}([2 -1; -1 2])\nequations_parabolic = ConstantAnisotropicDiffusion2D(diffusivity, equations_hyperbolic);\n\nboundary_conditions_hyperbolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1.0)),\n y_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(2.0)),\n y_pos = boundary_condition_do_nothing,\n x_pos = boundary_condition_do_nothing)\n\nboundary_conditions_parabolic = (; x_neg = BoundaryConditionConstantDirichlet(1.0),\n y_neg = BoundaryConditionConstantDirichlet(2.0),\n y_pos = BoundaryConditionConstantDirichlet(0.0),\n x_pos = BoundaryConditionConstantDirichlet(0.0));\n\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\ncoordinates_min = (-1.0, -1.0) # minimum coordinates (min(x), min(y))\ncoordinates_max = ( 1.0, 1.0) # maximum coordinates (max(x), max(y))\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n periodicity=false, n_cells_max=30_000) # set maximum capacity of tree data structure\n\ninitial_condition = (x, t, equations) -> SVector(0.0)\n\nsemi = SemidiscretizationHyperbolicParabolic(mesh,\n (equations_hyperbolic, equations_parabolic),\n initial_condition, solver;\n boundary_conditions=(boundary_conditions_hyperbolic,\n boundary_conditions_parabolic))\n\ntspan = (0.0, 2.0)\node = semidiscretize(semi, tspan)\ncallbacks = CallbackSet(SummaryCallback())\ntime_int_tol = 1.0e-6\nsol = solve(ode, RDPK3SpFSAL49(); abstol=time_int_tol, reltol=time_int_tol,\n ode_default_options()..., callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Package-versions","page":"11 Adding new parabolic terms","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"11 Adding new parabolic terms","title":"11 Adding new parabolic terms","text":"This page was generated using Literate.jl.","category":"page"},{"location":"code_of_conduct/","page":"Code of Conduct","title":"Code of Conduct","text":"EditURL = \"https://github.com/trixi-framework/Trixi.jl/blob/main/CODE_OF_CONDUCT.md\"","category":"page"},{"location":"code_of_conduct/#code-of-conduct","page":"Code of Conduct","title":"Code of Conduct","text":"","category":"section"},{"location":"code_of_conduct/","page":"Code of Conduct","title":"Code of Conduct","text":"Contributor Covenant Code of ConductOur PledgeWe as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation.We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community.Our StandardsExamples of behavior that contributes to a positive environment for our community include:Demonstrating empathy and kindness toward other people\nBeing respectful of differing opinions, viewpoints, and experiences\nGiving and gracefully accepting constructive feedback\nAccepting responsibility and apologizing to those affected by our mistakes, and learning from the experience\nFocusing on what is best not just for us as individuals, but for the overall communityExamples of unacceptable behavior include:The use of sexualized language or imagery, and sexual attention or advances of any kind\nTrolling, insulting or derogatory comments, and personal or political attacks\nPublic or private harassment\nPublishing others' private information, such as a physical or email address, without their explicit permission\nOther conduct which could reasonably be considered inappropriate in a professional settingEnforcement ResponsibilitiesCommunity leaders are responsible for clarifying and enforcing our standards of acceptable behavior and will take appropriate and fair corrective action in response to any behavior that they deem inappropriate, threatening, offensive, or harmful.Community leaders have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, and will communicate reasons for moderation decisions when appropriate.ScopeThis Code of Conduct applies within all community spaces, and also applies when an individual is officially representing the community in public spaces. Examples of representing our community include using an official e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event.EnforcementInstances of abusive, harassing, or otherwise unacceptable behavior may be reported to Michael Schlottke-Lakemper, Hendrik Ranocha, or any other of the principal developers responsible for enforcement listed in Authors. All complaints will be reviewed and investigated promptly and fairly.All community leaders are obligated to respect the privacy and security of the reporter of any incident.Enforcement GuidelinesCommunity leaders will follow these Community Impact Guidelines in determining the consequences for any action they deem in violation of this Code of Conduct:1. CorrectionCommunity Impact: Use of inappropriate language or other behavior deemed unprofessional or unwelcome in the community.Consequence: A private, written warning from community leaders, providing clarity around the nature of the violation and an explanation of why the behavior was inappropriate. A public apology may be requested.2. WarningCommunity Impact: A violation through a single incident or series of actions.Consequence: A warning with consequences for continued behavior. No interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, for a specified period of time. This includes avoiding interactions in community spaces as well as external channels like social media. Violating these terms may lead to a temporary or permanent ban.3. Temporary BanCommunity Impact: A serious violation of community standards, including sustained inappropriate behavior.Consequence: A temporary ban from any sort of interaction or public communication with the community for a specified period of time. No public or private interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, is allowed during this period. Violating these terms may lead to a permanent ban.4. Permanent BanCommunity Impact: Demonstrating a pattern of violation of community standards, including sustained inappropriate behavior, harassment of an individual, or aggression toward or disparagement of classes of individuals.Consequence: A permanent ban from any sort of public interaction within the community.AttributionThis Code of Conduct is adapted from the [Contributor Covenant][homepage], version 2.0, available at https://www.contributor-covenant.org/version/2/0/codeofconduct.html.Community Impact Guidelines were inspired by Mozilla's code of conduct enforcement ladder.[homepage]: https://www.contributor-covenant.orgFor answers to common questions about this code of conduct, see the FAQ at https://www.contributor-covenant.org/faq. Translations are available at https://www.contributor-covenant.org/translations.","category":"page"},{"location":"conventions/#conventions","page":"Conventions","title":"Conventions","text":"","category":"section"},{"location":"conventions/#Spatial-dimensions-and-directions","page":"Conventions","title":"Spatial dimensions and directions","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"We use the following numbering schemes on Cartesian or curved structured meshes.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"The orientations are numbered as 1 => x, 2 => y, 3 => z. For example, numerical fluxes such as flux_central(u_ll, u_rr, orientation, equations::AbstractEquations) use the orientation in this way.\nThe directions are numbered as 1 => -x, 2 => +x, 3 => -y, 4 => +y, 5 => -z, 6 => +z. For example, the boundary_conditions are ordered in this way when a Tuple of boundary conditions per direction is passed to the constructor of a SemidiscretizationHyperbolic.\nFor structured and unstructured curved meshes the concept of direction is generalized via the variable normal_direction. This variable points in the normal direction at a given, curved surface. For the computation of boundary fluxes the normal_direction is normalized to be a normal_vector used, for example, in FluxRotated.","category":"page"},{"location":"conventions/#Cells-vs.-elements-vs.-nodes","page":"Conventions","title":"Cells vs. elements vs. nodes","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"To uniquely distinguish between different components of the discretization, we use the following naming conventions:","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"The computational domain is discretized by a mesh, which is made up of individual cells. In general, neither the mesh nor the cells should be aware of any solver-specific knowledge, i.e., they should not store anything that goes beyond the geometrical information and the connectivity.\nThe numerical solvers do not directly store their information inside the mesh, but use own data structures. Specifically, for each cell on which a solver wants to operate, the solver creates an element to store solver-specific data.\nFor discretization schemes such as the discontinuous Galerkin or the finite element method, inside each element multiple nodes may be defined, which hold nodal information. The nodes are again a solver-specific component, just like the elements.\nWe often identify elements, nodes, etc. with their (local or global) integer index. Convenience iterators such as eachelement, eachnode use these indices.","category":"page"},{"location":"conventions/#Keywords-in-elixirs","page":"Conventions","title":"Keywords in elixirs","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Trixi.jl is distributed with several examples in the form of elixirs, small Julia scripts containing everything to set up and run a simulation. Working interactively from the Julia REPL with these scripts can be quite convenient while for exploratory research and development of Trixi.jl. For example, you can use the convenience function trixi_include to include an elixir with some modified arguments. To enable this, it is helpful to use a consistent naming scheme in elixirs, since trixi_include can only perform simple replacements. Some standard variables names are","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"polydeg for the polynomial degree of a solver\nsurface_flux for the numerical flux at surfaces\nvolume_flux for the numerical flux used in flux differencing volume terms","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Moreover, convergence_test requires that the spatial resolution is set via the keywords","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"initial_refinement_level (an integer, e.g. for the TreeMesh and the P4estMesh) or\ncells_per_dimension (a tuple of integers, one per spatial dimension, e.g. for the StructuredMesh and the DGMultiMesh).","category":"page"},{"location":"conventions/#Variable-names","page":"Conventions","title":"Variable names","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Use descriptive names (using snake_case for variables/functions and CamelCase for types)\nUse a suffix _ as in name_ for local variables that would otherwise hide existing symbols.\nUse a prefix _ as in _name to indicate internal methods/data that are \"fragile\" in the sense that there's no guarantee that they might get changed without notice. These are also not documented with a docstring (but maybe with comments using #).","category":"page"},{"location":"conventions/#Array-types-and-wrapping","page":"Conventions","title":"Array types and wrapping","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"To allow adaptive mesh refinement efficiently when using time integrators from OrdinaryDiffEq, Trixi.jl allows to represent numerical solutions in two different ways. Some discussion can be found online and in form of comments describing Trixi.wrap_array and Trixi.wrap_array_native in the source code of Trixi.jl. The flexibility introduced by this possible wrapping enables additional performance optimizations. However, it comes at the cost of some additional abstractions (and needs to be used with caution, as described in the source code of Trixi.jl). Thus, we use the following conventions to distinguish between arrays visible to the time integrator and wrapped arrays mainly used internally.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Arrays visible to the time integrator have a suffix _ode, e.g., du_ode, u_ode.\nWrapped arrays do not have a suffix, e.g., du, u.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Methods either accept arrays visible to the time integrator or wrapped arrays based on the following rules.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"When some solution is passed together with a semidiscretization semi, the solution must be a u_ode that needs to be wrapped via wrap_array(u_ode, semi) (or wrap_array_native(u_ode, semi)) for further processing.\nWhen some solution is passed together with the mesh, equations, solver, cache, ..., it is already wrapped via wrap_array (or wrap_array_native).\nExceptions of this rule are possible, e.g. for AMR, but must be documented in the code.\nwrap_array should be used as default option. wrap_array_native should only be used when necessary, e.g., to avoid additional overhead when interfacing with external C libraries such as HDF5, MPI, or visualization.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"EditURL = \"../../literate/src/files/non_periodic_boundaries.jl\"","category":"page"},{"location":"tutorials/non_periodic_boundaries/#non_periodic_boundaries","page":"4 Non-periodic boundaries","title":"4: Non-periodic boundaries","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Dirichlet-boundary-condition","page":"4 Non-periodic boundaries","title":"Dirichlet boundary condition","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"First, let's look at the Dirichlet boundary condition BoundaryConditionDirichlet.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"BoundaryConditionDirichlet(boundary_value_function)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"In Trixi.jl, this creates a Dirichlet boundary condition where the function boundary_value_function is used to set the values at the boundary. It can be used to create a boundary condition that sets exact boundary values by passing the exact solution of the equation.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"It is important to note that standard Dirichlet boundary conditions for hyperbolic PDEs do not make sense in most cases. However, we are using a special weak form of the Dirichlet boundary condition, based on the application of the numerical surface flux. The numerical surface flux takes the solution value from inside the domain and the prescribed value of the outer boundary state as arguments, and solves an approximate Riemann problem to introduce dissipation (and hence stabilization) at the boundary. Hence, the performance of the Dirichlet BC depends on the fidelity of the numerical surface flux.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"The passed boundary value function is called with the same arguments as an initial condition function, i.e.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"boundary_value_function(x, t, equations)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"where x specifies the spatial coordinates, t is the current time, and equations is the corresponding system of equations.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"We want to give a short example for a simulation with such a Dirichlet BC.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Consider the one-dimensional linear advection equation with domain Omega=0 2 and a constant zero initial condition.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"using OrdinaryDiffEq, Trixi\n\nadvection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)\n\ninitial_condition_zero(x, t, equation::LinearScalarAdvectionEquation1D) = SVector(0.0)\ninitial_condition = initial_condition_zero\n\nusing Plots\nplot(x -> sum(initial_condition(x, 0.0, equations)), label=\"initial condition\", ylim=(-1.5, 1.5))","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Using an advection velocity of 1.0 and the (local) Lax-Friedrichs/Rusanov flux FluxLaxFriedrichs as a numerical surface flux, we are able to create an inflow boundary on the left and an outflow boundary on the right, as the Lax-Friedrichs flux is in this case an exact characteristics Riemann solver. We note that for more complex PDEs different strategies for inflow/outflow boundaries are necessary. To define the inflow values, we initialize a boundary_value_function.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"function boundary_condition_sine_sector(x, t, equation::LinearScalarAdvectionEquation1D)\n if 1 <= t <= 3\n scalar = sin(2 * pi * sum(t - 1))\n else\n scalar = zero(t)\n end\n return SVector(scalar)\nend\nboundary_condition = boundary_condition_sine_sector","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"We set the BC in negative and positive x-direction.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"boundary_conditions = (x_neg=BoundaryConditionDirichlet(boundary_condition),\n x_pos=BoundaryConditionDirichlet(boundary_condition))","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\ncoordinates_min = (0.0,)\ncoordinates_max = (2.0,)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"For the mesh type TreeMesh the parameter periodicity must be set to false in the corresponding direction.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"mesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n n_cells_max=10_000,\n periodicity=false)\n\n\nsemi = SemidiscretizationHyperbolic(mesh, equations,\n initial_condition,\n solver,\n boundary_conditions=boundary_conditions)\n\ntspan = (0.0, 6.0)\node = semidiscretize(semi, tspan)\n\nanalysis_callback = AnalysisCallback(semi, interval=100,)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(analysis_callback,\n stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"We define some equidistant nodes for the visualization","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"visnodes = range(tspan[1], tspan[2], length=300)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"and run the simulation.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, saveat=visnodes, callback=callbacks);\n\nusing Plots\n@gif for step in 1:length(sol.u)\n plot(sol.u[step], semi, ylim=(-1.5, 1.5), legend=true, label=\"approximation\", title=\"time t=$(round(sol.t[step], digits=5))\")\n scatter!([0.0], [sum(boundary_condition(SVector(0.0), sol.t[step], equations))], label=\"boundary condition\")\nend","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Other-available-example-elixirs-with-non-trivial-BC","page":"4 Non-periodic boundaries","title":"Other available example elixirs with non-trivial BC","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Moreover, there are other boundary conditions in Trixi.jl. For instance, you can use the slip wall boundary condition boundary_condition_slip_wall.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Trixi.jl provides some interesting examples with different combinations of boundary conditions, e.g. using boundary_condition_slip_wall and other self-defined boundary conditions using BoundaryConditionDirichlet.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"For instance, there is a 2D compressible Euler setup for a Mach 3 wind tunnel flow with a forward facing step in the elixir elixir_euler_forward_step_amr.jl discretized with a P4estMesh using adaptive mesh refinement (AMR).","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"A double Mach reflection problem for the 2D compressible Euler equations elixir_euler_double_mach_amr.jl exercises a special boundary conditions along the bottom of the domain that is a mixture of Dirichlet and slip wall.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"A channel flow around a cylinder at Mach 3 elixir_euler_supersonic_cylinder.jl contains supersonic Mach 3 inflow at the left portion of the domain and supersonic outflow at the right portion of the domain. The top and bottom of the channel as well as the cylinder are treated as Euler slip wall boundaries.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Package-versions","page":"4 Non-periodic boundaries","title":"Package versions","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"4 Non-periodic boundaries","title":"4 Non-periodic boundaries","text":"This page was generated using Literate.jl.","category":"page"},{"location":"parallelization/#Parallelization","page":"Parallelization","title":"Parallelization","text":"","category":"section"},{"location":"parallelization/#Shared-memory-parallelization-with-threads","page":"Parallelization","title":"Shared-memory parallelization with threads","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Many compute-intensive loops in Trixi.jl are parallelized using the multi-threading support provided by Julia. You can recognize those loops by the @threaded macro prefixed to them, e.g.,","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"@threaded for element in eachelement(dg, cache)\n ...\nend","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"This will statically assign an equal iteration count to each available thread.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"To use multi-threading, you need to tell Julia at startup how many threads you want to use by either setting the environment variable JULIA_NUM_THREADS or by providing the -t/--threads command line argument. For example, to start Julia with four threads, start Julia with","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia --threads=4","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If both the environment variable and the command line argument are specified at the same time, the latter takes precedence.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If you use time integration methods from OrdinaryDiffEq.jl and want to use multiple threads therein, you need to set the keyword argument thread=OrdinaryDiffEq.True() of the algorithms, as described in the section on time integration methods.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"warning: Warning\nNot everything is parallelized yet and there are likely opportunities to improve scalability. Multi-threading isn't considered part of the public API of Trixi.jl yet.","category":"page"},{"location":"parallelization/#Distributed-computing-with-MPI","page":"Parallelization","title":"Distributed computing with MPI","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"In addition to the shared memory parallelization with multi-threading, Trixi.jl supports distributed parallelism via MPI.jl, which leverages the Message Passing Interface (MPI). MPI.jl comes with its own MPI library binaries such that there is no need to install MPI yourself. However, it is also possible to instead use an existing MPI installation, which is recommended if you are running MPI programs on a cluster or supercomputer (see the MPI.jl docs to find out how to select the employed MPI library). Additional notes on how to use a system-provided MPI installation with Trixi.jl can be found in the following subsection.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"warning: Work in progress\nMPI-based parallelization is work in progress and not finished yet. Nothing related to MPI is part of the official API of Trixi.jl yet.","category":"page"},{"location":"parallelization/#parallel_system_MPI","page":"Parallelization","title":"Using a system-provided MPI installation","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"When using Trixi.jl with a system-provided MPI backend, the underlying p4est, t8code and HDF5 libraries need to be compiled with the same MPI installation. If you want to use p4est (via the P4estMesh) or t8code (via the T8codeMesh) from Trixi.jl, you also need to use system-provided p4est or t8code installations (for notes on how to install p4est and t8code see, e.g., here and here, use the configure option --enable-mpi). Otherwise, there will be warnings that no preference is set for P4est.jl and T8code.jl that can be ignored if you do not use these libraries from Trixi.jl. Note that t8code already comes with a p4est installation, so it suffices to install t8code. In order to use system-provided p4est and t8code installations, P4est.jl and T8code.jl need to be configured to use the custom installations. Follow the steps described here and here for the configuration. The paths that point to libp4est.so (and potentially to libsc.so) need to be the same for P4est.jl and T8code.jl. This could, e.g., be libp4est.so that usually can be found in lib/ or local/lib/ in the installation directory of t8code. Note that the T8codeMesh, however, does not support MPI yet. The preferences for HDF5.jl always need to be set, even if you do not want to use HDF5 from Trixi.jl, see also issue #1079 in HDF5.jl. To set the preferences for HDF5.jl, follow the instructions described here.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"In total, in your active Julia project you should have a LocalPreferences.toml file with sections [MPIPreferences], [T8code] (only needed if T8codeMesh is used), [P4est] (only needed if P4estMesh is used), and [HDF5] as well as an entry MPIPreferences in your Project.toml to use a custom MPI installation. A LocalPreferences.toml file created as described above might look something like the following:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"[HDF5]\nlibhdf5 = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\"\nlibhdf5_hl = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\"\n\n[HDF5_jll]\nlibhdf5_hl_path = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\"\nlibhdf5_path = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\"\n\n[MPIPreferences]\n__clear__ = [\"preloads_env_switch\"]\n_format = \"1.0\"\nabi = \"OpenMPI\"\nbinary = \"system\"\ncclibs = []\nlibmpi = \"/lib/x86_64-linux-gnu/libmpi.so\"\nmpiexec = \"mpiexec\"\npreloads = []\n\n[P4est]\nlibp4est = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\"\nlibsc = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\"\n\n[T8code]\nlibp4est = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\"\nlibsc = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\"\nlibt8 = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libt8.so\"","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"This file is created with the following sequence of commands:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using MPIPreferences\njulia> MPIPreferences.use_system_binary()","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Restart the Julia REPL","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using P4est\njulia> P4est.set_library_p4est!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\")\njulia> P4est.set_library_sc!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\")\njulia> using T8code\njulia> T8code.set_libraries_path!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/\")\njulia> using HDF5\njulia> HDF5.API.set_libraries!(\"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\", \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\")","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"After the preferences are set, restart the Julia REPL again.","category":"page"},{"location":"parallelization/#parallel_usage","page":"Parallelization","title":"Usage","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"To start Trixi.jl in parallel with MPI, there are three options:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Run from the REPL with mpiexec(): You can start a parallel execution directly from the REPL by executing\njulia> using MPI\n\njulia> mpiexec() do cmd\n run(`$cmd -n 3 $(Base.julia_cmd()) --threads=1 --project=@. -e 'using Trixi; trixi_include(default_example())'`)\n end\nThe parameter -n 3 specifies that Trixi.jl should run with three processes (or ranks in MPI parlance) and should be adapted to your available computing resources and problem size. The $(Base.julia_cmd()) argument ensures that Julia is executed in parallel with the same optimization level etc. as you used for the REPL; if this is unnecessary or undesired, you can also just use julia. Further, if you are not running Trixi.jl from a local clone but have installed it as a package, you need to omit the --project=@..\nRun from the command line with mpiexecjl: Alternatively, you can use the mpiexecjl script provided by MPI.jl, which allows you to start Trixi.jl in parallel directly from the command line. As a preparation, you need to install the script once by running\njulia> using MPI\n\njulia> MPI.install_mpiexecjl(destdir=\"/somewhere/in/your/PATH\")\nThen, to execute Trixi.jl in parallel, execute the following command from your command line:\nmpiexecjl -n 3 julia --threads=1 --project=@. -e 'using Trixi; trixi_include(default_example())'\nRun interactively with tmpi (Linux/MacOS only): If you are on a Linux/macOS system, you have a third option which lets you run Julia in parallel interactively from the REPL. This comes in handy especially during development, as in contrast to the first two options, it allows to reuse the compilation cache and thus facilitates much faster startup times after the first execution. It requires tmux and the OpenMPI library to be installed before, both of which are usually available through a package manager. Once you have installed both tools, you need to configure MPI.jl to use the OpenMPI for your system, which is explained here. Then, you can download and install the tmpi script by executing\ncurl https://raw.githubusercontent.com/Azrael3000/tmpi/master/tmpi -o /somewhere/in/your/PATH/tmpi\nFinally, you can start and control multiple Julia REPLs simultaneously by running\ntmpi 3 julia --threads=1 --project=@.\nThis will start Julia inside tmux three times and multiplexes all commands you enter in one REPL to all other REPLs (try for yourself to understand what it means). If you have no prior experience with tmux, handling the REPL this way feels slightly weird in the beginning. However, there is a lot of documentation for tmux available and once you get the hang of it, developing Trixi.jl in parallel becomes much smoother this way. Some helpful commands are the following. To close a single pane you can press Ctrl+b and then x followed by y to confirm. To quit the whole session you press Ctrl+b followed by :kill-session. Often you would like to scroll up. You can do that by pressing Ctrl+b and then [, which allows you to use the arrow keys to scroll up and down. To leave the scroll mode you press q. Switching between panes can be done by Ctrl+b followed by o. As of March 2022, newer versions of tmpi also support mpich, which is the default backend of MPI.jl (via MPICH_Jll.jl). To use this setup, you need to install mpiexecjl as described in the documentation of MPI.jl and make it available as mpirun, e.g., via a symlink of the form\nln -s ~/.julia/bin/mpiexecjl /somewhere/in/your/path/mpirun\n(assuming default installations).","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"note: Hybrid parallelism\nIt is possible to combine MPI with shared memory parallelism via threads by starting Julia with more than one thread, e.g. by passing the command line argument julia --threads=2 instead of julia --threads=1 used in the examples above. In that case, you should make sure that your system supports the number of processes/threads that you try to start.","category":"page"},{"location":"parallelization/#parallel_performance","page":"Parallelization","title":"Performance","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"For information on how to evaluate the parallel performance of Trixi.jl, please have a look at the Performance metrics of the AnalysisCallback section, specifically at the descriptions of the performance index (PID).","category":"page"},{"location":"parallelization/#Using-error-based-step-size-control-with-MPI","page":"Parallelization","title":"Using error-based step size control with MPI","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If you use error-based step size control (see also the section on error-based adaptive step sizes) together with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.","category":"page"},{"location":"parallelization/#Using-parallel-input-and-output","page":"Parallelization","title":"Using parallel input and output","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Trixi.jl allows parallel I/O using MPI by leveraging parallel HDF5.jl. On most systems, this is enabled by default. Additionally, you can also use a local installation of the HDF5 library (with MPI support). For this, you first need to use a system-provided MPI library, see also here and you need to tell HDF5.jl to use this library. To do so with HDF5.jl v0.17 and newer, set the preferences libhdf5 and libhdf5_hl to the local paths of the libraries libhdf5 and libhdf5_hl, which can be done by","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using Preferences, UUIDs\njulia> set_preferences!(\n UUID(\"f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f\"), # UUID of HDF5.jl\n \"libhdf5\" => \"/path/to/your/libhdf5.so\",\n \"libhdf5_hl\" => \"/path/to/your/libhdf5_hl.so\", force = true)","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Alternatively, with HDF5.jl v0.17.1 or higher you can use","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using HDF5\njulia> HDF5.API.set_libraries!(\"/path/to/your/libhdf5.so\", \"/path/to/your/libhdf5_hl.so\")","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"For more information see also the documentation of HDF5.jl. In total, you should have a file called LocalPreferences.toml in the project directory that contains a section [MPIPreferences], a section [HDF5] with entries libhdf5 and libhdf5_hl, a section [P4est] with the entry libp4est as well as a section [T8code] with the entries libt8, libp4est and libsc. If you use HDF5.jl v0.16 or older, instead of setting the preferences for HDF5.jl, you need to set the environment variable JULIA_HDF5_PATH to the path, where the HDF5 binaries are located and then call ]build HDF5 from Julia.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If HDF5 is not MPI-enabled, Trixi.jl will fall back on a less efficient I/O mechanism. In that case, all disk I/O is performed only on rank zero and data is distributed to/gathered from the other ranks using regular MPI communication.","category":"page"},{"location":"overview/#Overview-of-the-structure-of-Trixi.jl","page":"Overview","title":"Overview of the structure of Trixi.jl","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl is designed as a library of components for discretizations of hyperbolic conservation laws. Thus, it is not a monolithic PDE solver that is configured at runtime via parameter files, as it is often found in classical numerical simulation codes. Instead, each simulation is configured by pure Julia code. Many examples of such simulation setups, called elixirs in Trixi.jl, are provided in the examples/ folder.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl uses the method of lines, i.e., the full space-time discretization is separated into two steps; the spatial semidiscretization is performed at first and the resulting ODE system is solved numerically using a suitable time integration method. Thus, the main ingredients of an elixir designed to solve a PDE numerically are the spatial semidiscretization and the time integration scheme.","category":"page"},{"location":"overview/#overview-semidiscretizations","page":"Overview","title":"Semidiscretizations","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Semidiscretizations are high-level descriptions of spatial discretizations specialized for certain PDEs. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Such semidiscretizations are usually named semi in Trixi.jl","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"(Image: semidiscretization_overview)","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"The basic building blocks of a semidiscretization are","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"a mesh describing the geometry of the domain\na set of equations describing the physical model\na solver describing the numerical approach","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"In addition, a semidiscretization bundles initial and boundary conditions, and possible source terms. These different ingredients of a semidiscretization can be configured individually and combined together. When a semidiscretization is constructed, it will create an internal cache, i.e., a collection of setup-specific data structures, that is usually passed to all lower level functions.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Due to Trixi.jl's modular nature using Julia's multiple dispatch features, new ingredients can be created specifically for a certain combination of other ingredients. For example, a new mesh type can be created and implemented at first only for a specific solver. Thus, there is no need to consider all possible combinations of meshes, equations, and solvers when implementing new features. This allows rapid prototyping of new ideas and is one of the main design goals behind Trixi.jl. Below is a brief overview of the availability of different features on different mesh types.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Feature TreeMesh StructuredMesh UnstructuredMesh2D P4estMesh DGMultiMesh Further reading\nSpatial dimension 1D, 2D, 3D 1D, 2D, 3D 2D 2D, 3D 1D, 2D, 3D \nCoordinates Cartesian curvilinear curvilinear curvilinear curvilinear \nConnectivity h-nonconforming conforming conforming h-nonconforming conforming \nElement type line, square, cube line, quadᵃ, hexᵃ quadᵃ quadᵃ, hexᵃ simplex, quadᵃ, hexᵃ \nAdaptive mesh refinement ✅ ❌ ❌ ✅ ❌ AMRCallback\nSolver type DGSEM DGSEM DGSEM DGSEM DGMulti \nDomain hypercube mapped hypercube arbitrary arbitrary arbitrary \nWeak form ✅ ✅ ✅ ✅ ✅ VolumeIntegralWeakForm\nFlux differencing ✅ ✅ ✅ ✅ ✅ VolumeIntegralFluxDifferencing\nShock capturing ✅ ✅ ✅ ✅ ❌ VolumeIntegralShockCapturingHG\nNonconservative equations ✅ ✅ ✅ ✅ ✅ e.g., GLM MHD or shallow water equations\nParabolic terms ✅ ✅ ❌ ✅ ✅ e.g., CompressibleNavierStokesDiffusion2D","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"ᵃ: quad = quadrilateral, hex = hexahedron","category":"page"},{"location":"overview/#Time-integration-methods","page":"Overview","title":"Time integration methods","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem is a wrapper of Trixi.rhs!(du_ode, u_ode, semi, t), which gets called in ODE solvers. Further information can be found in the section on time integration methods.","category":"page"},{"location":"overview/#Next-steps","page":"Overview","title":"Next steps","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"We explicitly encourage people interested in Trixi.jl to have a look at the examples/ bundled with Trixi.jl to get an impression of what is possible and the general look and feel of Trixi.jl. Before doing that, it is usually good to get an idea of how to visualize numerical results.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"If you like learning by doing, looking at the tutorials and trying to mix your own elixirs based thereon is probably a good next step. Otherwise, you can further dig into the documentation by looking at Trixi.jl's basic building blocks.","category":"page"},{"location":"callbacks/#callbacks-id","page":"Callbacks","title":"Callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Many of the advanced features of Trixi.jl, such as adaptive mesh refinement, are implemented as callbacks. A callback is an algorithmic entity that gets passed to the ODE solver and is called at specific points during execution to perform certain tasks. Callbacks in Trixi.jl are either called after each time step (step callbacks) or after each stage of the ODE solver (stage callbacks).","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"(Image: callbacks_illustration)","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The advantage of callbacks over hard-coding all features is that it allows to extend Trixi.jl without modifying the internal source code. Trixi.jl provides callbacks for time step control, adaptive mesh refinement, I/O, and more.","category":"page"},{"location":"callbacks/#Step-callbacks","page":"Callbacks","title":"Step callbacks","text":"","category":"section"},{"location":"callbacks/#CFL-based-time-step-control","page":"Callbacks","title":"CFL-based time step control","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Time step control can be performed with a StepsizeCallback. An example making use of this can be found at examples/tree_2d_dgsem/elixir_advection_basic.jl","category":"page"},{"location":"callbacks/#Adaptive-mesh-refinement","page":"Callbacks","title":"Adaptive mesh refinement","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Trixi.jl uses a hierarchical Cartesian mesh which can be locally refined in a solution-adaptive way. This can be used to speed up simulations with minimal loss in overall accuracy. Adaptive mesh refinement (AMR) can be used by passing an AMRCallback to the ODE solver. The AMRCallback requires a controller such as ControllerThreeLevel or ControllerThreeLevelCombined to tell the AMR algorithm which cells to refine/coarsen.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"An example elixir using AMR can be found at examples/tree_2d_dgsem/elixir_advection_amr.jl.","category":"page"},{"location":"callbacks/#Analyzing-the-numerical-solution","page":"Callbacks","title":"Analyzing the numerical solution","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The AnalysisCallback can be used to analyze the numerical solution, e.g. calculate errors or user-specified integrals, and print the results to the screen. The results can also be saved in a file. An example can be found at examples/tree_2d_dgsem/elixir_euler_vortex.jl. Note that the errors (e.g. L2 error or Linf error) are computed with respect to the initial condition. The percentage of the simulation time refers to the ratio of the current time and the final time, i.e. it does not consider the maximal number of iterations. So the simulation could finish before 100% are reached. Note that, e.g., due to AMR or smaller time step sizes, the simulation can actually take longer than the percentage indicates. In Performance metrics of the AnalysisCallback you can find a detailed description of the different performance metrics the AnalysisCallback computes.","category":"page"},{"location":"callbacks/#I/O","page":"Callbacks","title":"I/O","text":"","category":"section"},{"location":"callbacks/#Solution-and-restart-files","page":"Callbacks","title":"Solution and restart files","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"To save the solution in regular intervals you can use a SaveSolutionCallback. It is also possible to create restart files using the SaveRestartCallback. An example making use of these can be found at examples/tree_2d_dgsem/elixir_advection_extended.jl. An example showing how to restart a simulation from a restart file can be found at examples/tree_2d_dgsem/elixir_advection_restart.jl.","category":"page"},{"location":"callbacks/#Time-series","page":"Callbacks","title":"Time series","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Sometimes it is useful to record the evaluations of state variables over time at a given set of points. This can be achieved by the TimeSeriesCallback, which is used, e.g., in examples/tree_2d_dgsem/elixir_acoustics_gaussian_source.jl. The TimeSeriesCallback constructor expects a semidiscretization and a list of points at which the solution should be recorded in regular time step intervals. After the last time step, the entire record is stored in an HDF5 file.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For the points, two different input formats are supported: You can either provide them as a list of tuples, which is handy if you specify them by hand on the REPL. Alternatively, you can provide them as a two-dimensional array, where the first dimension is the point number and the second dimension is the coordinate dimension. This is especially useful when reading them from a file.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For example, to record the primitive variables at the points (0.0, 0.0) and (-1.0, 0.5) every five timesteps and storing the collected data in the file tseries.h5, you can create the TimeSeriesCallback as","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"time_series = TimeSeriesCallback(semi, [(0.0, 0.0), (-1.0, 0.5)];\n interval=5,\n solution_variables=cons2prim,\n filename=\"tseries.h5\")","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For a full list of possible arguments, please check the documentation for the TimeSeriesCallback. As an alternative to specifying the point coordinates directly in the elixir or on the REPL, you can read them from a file. For instance, with a text file points.dat with content","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":" 0.0 0.0\n-1.0 0.5","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"you can create a time series callback with","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"using DelimitedFiles: readdlm\ntime_series = TimeSeriesCallback(semi, readdlm(\"points.dat\"))","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"To plot the individual point data series over time, you can create a PlotData1D from the TimeSeriesCallback and a given point ID. For example, executing","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"julia> using Trixi, Plots\n\njulia> trixi_include(joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_acoustics_gaussian_source.jl\"))\n\njulia> pd1 = PlotData1D(time_series, 1)\n\njulia> pd2 = PlotData1D(time_series, 2)\n\njulia> plot(pd1[\"p_prime\"]); plot!(pd2[\"p_prime\"], xguide=\"t\")","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"will yield the following plot:","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"(Image: image)","category":"page"},{"location":"callbacks/#Miscellaneous","page":"Callbacks","title":"Miscellaneous","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The AliveCallback prints some information to the screen to show that a simulation is still running.\nThe SummaryCallback prints a human-readable summary of the simulation setup and controls the automated performance measurements, including an output of the recorded timers after a simulation.\nThe VisualizationCallback can be used for in-situ visualization. See Visualizing results during a simulation.\nThe TrivialCallback does nothing and can be used to easily disable some callbacks via trixi_include.","category":"page"},{"location":"callbacks/#Equation-specific-callbacks","page":"Callbacks","title":"Equation-specific callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Some callbacks provided by Trixi.jl implement specific features for certain equations:","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The LBMCollisionCallback implements the Lattice-Boltzmann method (LBM) collision operator and should only be used when solving the Lattice-Boltzmann equations. See e.g. examples/tree_2d_dgsem/elixir_lbm_constant.jl\nThe SteadyStateCallback terminates the time integration when the residual steady state falls below a certain threshold. This checks the convergence of the potential phi for hyperbolic diffusion. See e.g. examples/tree_2d_dgsem/elixir_hypdiff_nonperiodic.jl.\nThe GlmSpeedCallback updates the divergence cleaning wave speed c_h for the ideal GLM-MHD equations. See e.g. examples/tree_2d_dgsem/elixir_mhd_alfven_wave.jl.","category":"page"},{"location":"callbacks/#Usage-of-step-callbacks","page":"Callbacks","title":"Usage of step callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Step callbacks are passed to the solve method from the ODE solver via the keyword argument callback. If you want to use a single callback cb, pass it as callback=cb. When using two or more callbacks, you need to turn them into a CallbackSet first by calling callbacks = CallbackSet(cb1, cb2) and passing it as callback=callbacks.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"note: Note\nThere are some restrictions regarding the order of callbacks in a CallbackSet.The callbacks are called after each time step but some callbacks actually belong to the next time step. Therefore, the callbacks should be ordered in the following way:Callbacks that belong to the current time step:\nSummaryCallback controls, among other things, timers and should thus be first\nSteadyStateCallback may mark a time step as the last one\nAnalysisCallback may do some checks that mark a time step as the last one\nAliveCallback should be nearby AnalysisCallback\nSaveSolutionCallback/SaveRestartCallback should save the current solution before it is degraded by AMR\nVisualizationCallback should be called before the mesh is adapted\nCallbacks that belong to the next time step:\nAMRCallback\nStepsizeCallback must be called after AMRCallback to accommodate potential changes to the mesh\nGlmSpeedCallback must be called after StepsizeCallback because the step size affects the value of c_h\nLBMCollisionCallback is already part of the calculations of the next time step and should therefore be called after StepsizeCallback","category":"page"},{"location":"callbacks/#Stage-callbacks","page":"Callbacks","title":"Stage callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"PositivityPreservingLimiterZhangShu is a positivity-preserving limiter, used to enforce physical constraints. An example elixir using this feature can be found at examples/tree_2d_dgsem/elixir_euler_positivity.jl.","category":"page"},{"location":"callbacks/#Implementing-new-callbacks","page":"Callbacks","title":"Implementing new callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Since Trixi.jl is compatible with OrdinaryDiffEq.jl, both packages share the same callback interface. A detailed description of it can be found in the OrdinaryDiffEq.jl documentation. Step callbacks are just called callbacks. Stage callbacks are called stage_limiter!.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"An example elixir showing how to implement a new simple stage callback and a new simple step callback can be found at examples/tree_2d_dgsem/elixir_advection_callbacks.jl.","category":"page"},{"location":"#Trixi.jl","page":"Home","title":"Trixi.jl","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"(Image: Docs-stable) (Image: Docs-dev) (Image: Slack) (Image: Youtube) (Image: Build Status) (Image: Codecov) (Image: Coveralls) (Image: Aqua QA) (Image: License: MIT) (Image: DOI)","category":"page"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is a numerical simulation framework for conservation laws written in Julia. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures. Its features include:","category":"page"},{"location":"","page":"Home","title":"Home","text":"1D, 2D, and 3D simulations on line/quad/hex/simplex meshes\nCartesian and curvilinear meshes\nConforming and non-conforming meshes\nStructured and unstructured meshes\nHierarchical quadtree/octree grid with adaptive mesh refinement\nForests of quadtrees/octrees with p4est via P4est.jl\nHigh-order accuracy in space and time\nDiscontinuous Galerkin methods\nKinetic energy-preserving and entropy-stable methods based on flux differencing\nEntropy-stable shock capturing\nPositivity-preserving limiting\nFinite difference summation by parts (SBP) methods\nCompatible with the SciML ecosystem for ordinary differential equations\nExplicit low-storage Runge-Kutta time integration\nStrong stability preserving methods\nCFL-based and error-based time step control\nNative support for differentiable programming\nForward mode automatic differentiation via ForwardDiff.jl\nPeriodic and weakly-enforced boundary conditions\nMultiple governing equations:\nCompressible Euler equations\nCompressible Navier-Stokes equations\nMagnetohydrodynamics (MHD) equations\nMulti-component compressible Euler and MHD equations\nLinearized Euler and acoustic perturbation equations\nHyperbolic diffusion equations for elliptic problems\nLattice-Boltzmann equations (D2Q9 and D3Q27 schemes)\nShallow water equations\nScalar advection\nMulti-physics simulations\nSelf-gravitating gas dynamics\nShared-memory parallelization via multithreading\nMulti-node parallelization via MPI\nVisualization and postprocessing of the results\nIn-situ and a posteriori visualization with Plots.jl\nInteractive visualization with Makie.jl\nPostprocessing with ParaView/VisIt via Trixi2Vtk","category":"page"},{"location":"#Installation","page":"Home","title":"Installation","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you have not yet installed Julia, please follow the instructions for your operating system. Trixi.jl works with Julia v1.8 and newer. We recommend using the latest stable release of Julia.","category":"page"},{"location":"#For-users","page":"Home","title":"For users","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl and its related tools are registered Julia packages. Hence, you can install Trixi.jl, the visualization tool Trixi2Vtk, OrdinaryDiffEq.jl, and Plots.jl by executing the following commands in the Julia REPL:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg\n\njulia> Pkg.add([\"Trixi\", \"Trixi2Vtk\", \"OrdinaryDiffEq\", \"Plots\"])","category":"page"},{"location":"","page":"Home","title":"Home","text":"You can copy and paste all commands to the REPL including the leading julia> prompts - they will automatically be stripped away by Julia. The package OrdinaryDiffEq.jl provides time integration schemes used by Trixi.jl, while Plots.jl can be used to directly visualize Trixi.jl's results from the REPL.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note on package versions: If some of the examples for how to use Trixi.jl do not work, verify that you are using a recent Trixi.jl release by comparing the installed Trixi.jl version from","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg; Pkg.update(\"Trixi\"); Pkg.status(\"Trixi\")","category":"page"},{"location":"","page":"Home","title":"Home","text":"to the latest release. If the installed version does not match the current release, please check the Troubleshooting section.","category":"page"},{"location":"","page":"Home","title":"Home","text":"The commands above can also be used to update Trixi.jl. A brief list of notable changes to Trixi.jl is available in NEWS.md.","category":"page"},{"location":"#for-developers","page":"Home","title":"For developers","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you plan on editing Trixi.jl itself, you can download Trixi.jl to a local folder and use the code from the cloned directory:","category":"page"},{"location":"","page":"Home","title":"Home","text":"git clone git@github.com:trixi-framework/Trixi.jl.git\ncd Trixi.jl\nmkdir run\ncd run\njulia --project=. -e 'using Pkg; Pkg.develop(PackageSpec(path=\"..\"))'","category":"page"},{"location":"","page":"Home","title":"Home","text":"If you installed Trixi.jl this way, you always have to start Julia with the --project flag set to your run directory, e.g.,","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia --project=.","category":"page"},{"location":"","page":"Home","title":"Home","text":"if already inside the run directory.","category":"page"},{"location":"","page":"Home","title":"Home","text":"The advantage of using a separate run directory is that you can also add other related packages (see below, e.g., for time integration or visualization) to the project in the run folder and always have a reproducible environment at hand to share with others.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Since the postprocessing tool Trixi2Vtk.jl typically does not need to be modified, it is recommended to install it as a normal package. Likewise, you can install OrdinaryDiffEq.jl and Plots.jl as ordinary packages. To achieve this, use the following REPL commands:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg\n\njulia> Pkg.add([\"OrdinaryDiffEq\", \"Trixi2Vtk\", \"Plots\"])","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note that the postprocessing tools Trixi2Vtk.jl and Plots.jl are optional and can be omitted.","category":"page"},{"location":"#Example:-Installing-Trixi.jl-as-a-package","page":"Home","title":"Example: Installing Trixi.jl as a package","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":" ","category":"page"},{"location":"","page":"Home","title":"Home","text":"Please note that the playback speed is set to 3x, thus the entire installation procedure lasts around 45 seconds in real time (depending on the performance of your computer and on how many dependencies had already been installed before).","category":"page"},{"location":"#Usage","page":"Home","title":"Usage","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"In the Julia REPL, first load the package Trixi.jl","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Trixi","category":"page"},{"location":"","page":"Home","title":"Home","text":"Then start a simulation by executing","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> trixi_include(default_example())","category":"page"},{"location":"","page":"Home","title":"Home","text":"Please be patient since Julia will compile the code just before running it. To visualize the results, load the package Plots","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Plots","category":"page"},{"location":"","page":"Home","title":"Home","text":"and generate a heatmap plot of the results with","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> plot(sol) # No trailing semicolon, otherwise no plot is shown","category":"page"},{"location":"","page":"Home","title":"Home","text":"This will open a new window with a 2D visualization of the final solution:","category":"page"},{"location":"","page":"Home","title":"Home","text":"(Image: image)","category":"page"},{"location":"","page":"Home","title":"Home","text":"The method trixi_include(...) expects a single string argument with the path to a Trixi.jl elixir, i.e., a text file containing Julia code necessary to set up and run a simulation. To quickly see Trixi.jl in action, default_example() returns the path to an example elixir with a short, two-dimensional problem setup. A list of all example elixirs packaged with Trixi.jl can be obtained by running get_examples(). Alternatively, you can also browse the examples/ subdirectory. If you want to modify one of the elixirs to set up your own simulation, download it to your machine, edit the configuration, and pass the file path to trixi_include(...).","category":"page"},{"location":"#Example:-Running-a-simulation-with-Trixi.jl","page":"Home","title":"Example: Running a simulation with Trixi.jl","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":" ","category":"page"},{"location":"","page":"Home","title":"Home","text":"If this produces weird symbols or question marks in the terminal on your system, you are probably using Mac OS with problematic fonts. In that case, please check the Troubleshooting section.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note on performance: Julia uses just-in-time compilation to transform its source code to native, optimized machine code at the time of execution and caches the compiled methods for further use. That means that the first execution of a Julia method is typically slow, with subsequent runs being much faster. For instance, in the example above the first execution of trixi_include takes about 20 seconds, while subsequent runs require less than 60 milliseconds.","category":"page"},{"location":"#Performing-a-convergence-analysis","page":"Home","title":"Performing a convergence analysis","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"To automatically determine the experimental order of convergence (EOC) for a given setup, execute","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> convergence_test(default_example(), 4)","category":"page"},{"location":"","page":"Home","title":"Home","text":"This will run a convergence test with the elixir default_example(), using four iterations with different initial refinement levels. The initial iteration will use the elixir unchanged, while for each subsequent iteration the initial_refinement_level parameter is incremented by one. Finally, the measured l^2 and l^infty errors and the determined EOCs will be displayed like this:","category":"page"},{"location":"","page":"Home","title":"Home","text":"[...]\nl2\nscalar\nerror EOC\n9.14e-06 -\n5.69e-07 4.01\n3.55e-08 4.00\n2.22e-09 4.00\n\nmean 4.00\n--------------------------------------------------------------------------------\nlinf\nscalar\nerror EOC\n6.44e-05 -\n4.11e-06 3.97\n2.58e-07 3.99\n1.62e-08 4.00\n\nmean 3.99\n--------------------------------------------------------------------------------","category":"page"},{"location":"","page":"Home","title":"Home","text":"An example with multiple variables looks like this:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> convergence_test(joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_euler_source_terms.jl\"), 3)","category":"page"},{"location":"","page":"Home","title":"Home","text":"[...]\nl2\nrho rho_v1 rho_v2 rho_e\nerror EOC error EOC error EOC error EOC\n9.32e-07 - 1.42e-06 - 1.42e-06 - 4.82e-06 -\n7.03e-08 3.73 9.53e-08 3.90 9.53e-08 3.90 3.30e-07 3.87\n4.65e-09 3.92 6.09e-09 3.97 6.09e-09 3.97 2.12e-08 3.96\n\nmean 3.82 mean 3.93 mean 3.93 mean 3.91\n--------------------------------------------------------------------------------\nlinf\nrho rho_v1 rho_v2 rho_e\nerror EOC error EOC error EOC error EOC\n9.58e-06 - 1.17e-05 - 1.17e-05 - 4.89e-05 -\n6.23e-07 3.94 7.48e-07 3.97 7.48e-07 3.97 3.22e-06 3.92\n4.05e-08 3.94 4.97e-08 3.91 4.97e-08 3.91 2.10e-07 3.94\n\nmean 3.94 mean 3.94 mean 3.94 mean 3.93\n--------------------------------------------------------------------------------","category":"page"},{"location":"#Showcase-of-advanced-features","page":"Home","title":"Showcase of advanced features","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"The presentation From Mesh Generation to Adaptive Simulation: A Journey in Julia, originally given as part of JuliaCon 2022, outlines how to use Trixi.jl for an adaptive simulation of the compressible Euler equations in two spatial dimensions on a complex domain. More details as well as code to run the simulation presented can be found at the reproducibility repository for the presentation.","category":"page"},{"location":"#Referencing","page":"Home","title":"Referencing","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you use Trixi.jl in your own research or write a paper using results obtained with the help of Trixi.jl, please cite the following articles:","category":"page"},{"location":"","page":"Home","title":"Home","text":"@article{ranocha2022adaptive,\n title={Adaptive numerical simulations with {T}rixi.jl:\n {A} case study of {J}ulia for scientific computing},\n author={Ranocha, Hendrik and Schlottke-Lakemper, Michael and Winters, Andrew Ross\n and Faulhaber, Erik and Chan, Jesse and Gassner, Gregor},\n journal={Proceedings of the JuliaCon Conferences},\n volume={1},\n number={1},\n pages={77},\n year={2022},\n doi={10.21105/jcon.00077},\n eprint={2108.06476},\n eprinttype={arXiv},\n eprintclass={cs.MS}\n}\n\n@article{schlottkelakemper2021purely,\n title={A purely hyperbolic discontinuous {G}alerkin approach for\n self-gravitating gas dynamics},\n author={Schlottke-Lakemper, Michael and Winters, Andrew R and\n Ranocha, Hendrik and Gassner, Gregor J},\n journal={Journal of Computational Physics},\n pages={110467},\n year={2021},\n month={06},\n volume={442},\n publisher={Elsevier},\n doi={10.1016/j.jcp.2021.110467},\n eprint={2008.10593},\n eprinttype={arXiv},\n eprintclass={math.NA}\n}","category":"page"},{"location":"","page":"Home","title":"Home","text":"In addition, you can also refer to Trixi.jl directly as","category":"page"},{"location":"","page":"Home","title":"Home","text":"@misc{schlottkelakemper2020trixi,\n title={{T}rixi.jl: {A}daptive high-order numerical simulations\n of hyperbolic {PDE}s in {J}ulia},\n author={Schlottke-Lakemper, Michael and Gassner, Gregor J and\n Ranocha, Hendrik and Winters, Andrew R and Chan, Jesse},\n year={2021},\n month={09},\n howpublished={\\url{https://github.com/trixi-framework/Trixi.jl}},\n doi={10.5281/zenodo.3996439}\n}","category":"page"},{"location":"#authors-index-md","page":"Home","title":"Authors","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl was initiated by Michael Schlottke-Lakemper (RWTH Aachen University/High-Performance Computing Center Stuttgart (HLRS), Germany) and Gregor Gassner (University of Cologne, Germany). Together with Hendrik Ranocha (Johannes Gutenberg University Mainz, Germany) and Andrew Winters (Linköping University, Sweden), and Jesse Chan (Rice University, US), they are the principal developers of Trixi.jl. The full list of contributors can be found under Authors.","category":"page"},{"location":"#License-and-contributing","page":"Home","title":"License and contributing","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is licensed under the MIT license (see License). Since Trixi.jl is an open-source project, we are very happy to accept contributions from the community. Please refer to Contributing for more details. Note that we strive to be a friendly, inclusive open-source community and ask all members of our community to adhere to our Code of Conduct. To get in touch with the developers, join us on Slack or create an issue.","category":"page"},{"location":"#Acknowledgments","page":"Home","title":"Acknowledgments","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"
\n
\n
\n
\n
\n
\n
\n
","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through the following grants:","category":"page"},{"location":"","page":"Home","title":"Home","text":"Excellence Strategy EXC 2044-390685587, Mathematics Münster: Dynamics-Geometry-Structure.\nResearch unit FOR 5409 \"Structure-Preserving Numerical Methods for Bulk- and Interface Coupling of Heterogeneous Models (SNuBIC)\" (project number 463312734).\nIndividual grant no. 528753982.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the European Research Council through the ERC Starting Grant \"An Exascale aware and Un-crashable Space-Time-Adaptive Discontinuous Spectral Element Solver for Non-Linear Conservation Laws\" (Extreme), ERC grant agreement no. 714487.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from Vetenskapsrådet (VR, Swedish Research Council), Sweden through the VR Starting Grant \"Shallow water flows including sediment transport and morphodynamics\", VR grant agreement 2020-03642 VR.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the United States National Science Foundation (NSF) under awards DMS-1719818 and DMS-1943186.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the German Federal Ministry of Education and Research (BMBF) through the project grant \"Adaptive earth system modeling with significantly reduced computation time for exascale supercomputers (ADAPTEX)\" (funding id: 16ME0668K).","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding by the Daimler und Benz Stiftung (Daimler and Benz Foundation) through grant no. 32-10/22.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is supported by NumFOCUS as an Affiliated Project.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"EditURL = \"../../literate/src/files/adding_nonconservative_equation.jl\"","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#adding_nonconservative_equation","page":"9 Adding a non-conservative equation","title":"9: Adding a non-conservative equation","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"If you want to use Trixi.jl for your own research, you might be interested in a new physics model that is not present in Trixi.jl. In this tutorial, we will implement the nonconservative linear advection equation in a periodic domain","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"left\nbeginalignedpartial_t u(tx) + a(x) partial_x u(tx) = 0 \nu(0x)=sin(x) \nu(t-pi)=u(tpi)\nendaligned\nright","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"where a(x) = 2 + cos(x). The analytic solution is","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"u(tx)=-sin left(2 tan ^-1left(sqrt3 tan left(fracsqrt3 t2-tan ^-1left(frac1sqrt3tan left(fracx2right)right)right)right)right)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"In Trixi.jl, such a mathematical model is encoded as a subtype of Trixi.AbstractEquations.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Basic-setup","page":"9 Adding a non-conservative equation","title":"Basic setup","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"Since there is no native support for variable coefficients, we need to transform the PDE to the following system:","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"left\nbeginalignedpartial_t beginpmatrixu(tx)a(tx) endpmatrix +beginpmatrix a(tx) partial_x u(tx) 0 endpmatrix = 0 \nu(0x)=sin(x) \na(0x)=2+cos(x) \nu(t-pi)=u(tpi)\nendaligned\nright","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"# Define new physics\nusing Trixi\nusing Trixi: AbstractEquations, get_node_vars\nimport Trixi: varnames, default_analysis_integrals, flux, max_abs_speed_naive,\n have_nonconservative_terms\n\n# Since there is no native support for variable coefficients, we use two\n# variables: one for the basic unknown `u` and another one for the coefficient `a`\nstruct NonconservativeLinearAdvectionEquation <: AbstractEquations{1 #= spatial dimension =#,\n 2 #= two variables (u,a) =#}\nend\n\nvarnames(::typeof(cons2cons), ::NonconservativeLinearAdvectionEquation) = (\"scalar\", \"advection_velocity\")\n\ndefault_analysis_integrals(::NonconservativeLinearAdvectionEquation) = ()\n\n\n# The conservative part of the flux is zero\nflux(u, orientation, equation::NonconservativeLinearAdvectionEquation) = zero(u)\n\n# Calculate maximum wave speed for local Lax-Friedrichs-type dissipation\nfunction max_abs_speed_naive(u_ll, u_rr, orientation::Integer, ::NonconservativeLinearAdvectionEquation)\n _, advection_velocity_ll = u_ll\n _, advection_velocity_rr = u_rr\n\n return max(abs(advection_velocity_ll), abs(advection_velocity_rr))\nend\n\n\n# We use nonconservative terms\nhave_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()\n\n# This \"nonconservative numerical flux\" implements the nonconservative terms.\n# In general, nonconservative terms can be written in the form\n# g(u) ∂ₓ h(u)\n# Thus, a discrete difference approximation of this nonconservative term needs\n# - `u mine`: the value of `u` at the current position (for g(u))\n# - `u_other`: the values of `u` in a neighborhood of the current position (for ∂ₓ h(u))\nfunction flux_nonconservative(u_mine, u_other, orientation,\n equations::NonconservativeLinearAdvectionEquation)\n _, advection_velocity = u_mine\n scalar, _ = u_other\n\n return SVector(advection_velocity * scalar, zero(scalar))\nend","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"The implementation of nonconservative terms uses a single \"nonconservative flux\" function flux_nonconservative. It will basically be applied in a loop of the form","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"du_m(D, u) = sum(D[m, l] * flux_nonconservative(u[m], u[l], 1, equations)) # orientation 1: x","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"where D is the derivative matrix and u contains the nodal solution values.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"Now, we can run a simple simulation using a DGSEM discretization.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"# Create a simulation setup\nusing Trixi\nusing OrdinaryDiffEq\n\nequation = NonconservativeLinearAdvectionEquation()\n\n# You can derive the exact solution for this setup using the method of\n# characteristics\nfunction initial_condition_sine(x, t, equation::NonconservativeLinearAdvectionEquation)\n x0 = -2 * atan(sqrt(3) * tan(sqrt(3) / 2 * t - atan(tan(x[1] / 2) / sqrt(3))))\n scalar = sin(x0)\n advection_velocity = 2 + cos(x[1])\n SVector(scalar, advection_velocity)\nend\n\n# Create a uniform mesh in 1D in the interval [-π, π] with periodic boundaries\nmesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=4, n_cells_max=10^4)\n\n# Create a DGSEM solver with polynomials of degree `polydeg`\n# Remember to pass a tuple of the form `(conservative_flux, nonconservative_flux)`\n# as `surface_flux` and `volume_flux` when working with nonconservative terms\nvolume_flux = (flux_central, flux_nonconservative)\nsurface_flux = (flux_lax_friedrichs, flux_nonconservative)\nsolver = DGSEM(polydeg=3, surface_flux=surface_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# Setup the spatial semidiscretization containing all ingredients\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create an ODE problem with given time span\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan)\n\n# Set up some standard callbacks summarizing the simulation setup and computing\n# errors of the numerical solution\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes\n# the passed callbacks\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false, callback=callbacks)\n\n# Print the timer summary\nsummary_callback()\n\n# Plot the numerical solution at the final time\nusing Plots: plot\nplot(sol)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"You see a plot of the final solution.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"We can check whether everything fits together by refining the grid and comparing the numerical errors. First, we look at the error using the grid resolution above.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"error_1 = analysis_callback(sol).l2 |> first","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"Next, we increase the grid resolution by one refinement level and run the simulation again.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"mesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=5, n_cells_max=10^4)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\n\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback);\n\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false, callback=callbacks);\nsummary_callback()\n\nerror_2 = analysis_callback(sol).l2 |> first","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"error_1 / error_2","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"As expected, the new error is roughly reduced by a factor of 16, corresponding to an experimental order of convergence of 4 (for polynomials of degree 3).","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Summary-of-the-code","page":"9 Adding a non-conservative equation","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"Here is the complete code that we used (without the callbacks since these create a lot of unnecessary output in the doctests of this tutorial). In addition, we create the struct inside the new module NonconservativeLinearAdvection. That ensures that we can re-create structs defined therein without having to restart Julia.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"Define new physics","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"module NonconservativeLinearAdvection\n\nusing Trixi\nusing Trixi: AbstractEquations, get_node_vars\nimport Trixi: varnames, default_analysis_integrals, flux, max_abs_speed_naive,\n have_nonconservative_terms\n\n# Since there is not yet native support for variable coefficients, we use two\n# variables: one for the basic unknown `u` and another one for the coefficient `a`\nstruct NonconservativeLinearAdvectionEquation <: AbstractEquations{1 #= spatial dimension =#,\n 2 #= two variables (u,a) =#}\nend\n\nvarnames(::typeof(cons2cons), ::NonconservativeLinearAdvectionEquation) = (\"scalar\", \"advection_velocity\")\n\ndefault_analysis_integrals(::NonconservativeLinearAdvectionEquation) = ()\n\n\n# The conservative part of the flux is zero\nflux(u, orientation, equation::NonconservativeLinearAdvectionEquation) = zero(u)\n\n# Calculate maximum wave speed for local Lax-Friedrichs-type dissipation\nfunction max_abs_speed_naive(u_ll, u_rr, orientation::Integer, ::NonconservativeLinearAdvectionEquation)\n _, advection_velocity_ll = u_ll\n _, advection_velocity_rr = u_rr\n\n return max(abs(advection_velocity_ll), abs(advection_velocity_rr))\nend\n\n\n# We use nonconservative terms\nhave_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()\n\n# This \"nonconservative numerical flux\" implements the nonconservative terms.\n# In general, nonconservative terms can be written in the form\n# g(u) ∂ₓ h(u)\n# Thus, a discrete difference approximation of this nonconservative term needs\n# - `u mine`: the value of `u` at the current position (for g(u))\n# - `u_other`: the values of `u` in a neighborhood of the current position (for ∂ₓ h(u))\nfunction flux_nonconservative(u_mine, u_other, orientation,\n equations::NonconservativeLinearAdvectionEquation)\n _, advection_velocity = u_mine\n scalar, _ = u_other\n\n return SVector(advection_velocity * scalar, zero(scalar))\nend\n\nend # module\n\n\n\n# Create a simulation setup\nimport .NonconservativeLinearAdvection\nusing Trixi\nusing OrdinaryDiffEq\n\nequation = NonconservativeLinearAdvection.NonconservativeLinearAdvectionEquation()\n\n# You can derive the exact solution for this setup using the method of\n# characteristics\nfunction initial_condition_sine(x, t, equation::NonconservativeLinearAdvection.NonconservativeLinearAdvectionEquation)\n x0 = -2 * atan(sqrt(3) * tan(sqrt(3) / 2 * t - atan(tan(x[1] / 2) / sqrt(3))))\n scalar = sin(x0)\n advection_velocity = 2 + cos(x[1])\n SVector(scalar, advection_velocity)\nend\n\n# Create a uniform mesh in 1D in the interval [-π, π] with periodic boundaries\nmesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=4, n_cells_max=10^4)\n\n# Create a DGSEM solver with polynomials of degree `polydeg`\n# Remember to pass a tuple of the form `(conservative_flux, nonconservative_flux)`\n# as `surface_flux` and `volume_flux` when working with nonconservative terms\nvolume_flux = (flux_central, NonconservativeLinearAdvection.flux_nonconservative)\nsurface_flux = (flux_lax_friedrichs, NonconservativeLinearAdvection.flux_nonconservative)\nsolver = DGSEM(polydeg=3, surface_flux=surface_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# Setup the spatial semidiscretization containing all ingredients\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create an ODE problem with given time span\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\n\n# Set up some standard callbacks summarizing the simulation setup and computing\n# errors of the numerical solution\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback);\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes\n# the passed callbacks\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false);\n\n# Plot the numerical solution at the final time\nusing Plots: plot\nplot(sol);\nnothing #hide","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Package-versions","page":"9 Adding a non-conservative equation","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"9 Adding a non-conservative equation","title":"9 Adding a non-conservative equation","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"EditURL = \"../../literate/src/files/differentiable_programming.jl\"","category":"page"},{"location":"tutorials/differentiable_programming/#differentiable_programming","page":"16 Differentiable programming","title":"16: Differentiable programming","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Julia and its ecosystem provide some tools for differentiable programming. Trixi.jl is designed to be flexible, extendable, and composable with Julia's growing ecosystem for scientific computing and machine learning. Thus, the ultimate goal is to have fast implementations that allow automatic differentiation (AD) without too much hassle for users. If some parts do not meet these requirements, please feel free to open an issue or propose a fix in a PR.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"In the following, we will walk through some examples demonstrating how to differentiate through Trixi.jl.","category":"page"},{"location":"tutorials/differentiable_programming/#Forward-mode-automatic-differentiation","page":"16 Differentiable programming","title":"Forward mode automatic differentiation","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Trixi.jl integrates well with ForwardDiff.jl for forward mode AD.","category":"page"},{"location":"tutorials/differentiable_programming/#Computing-the-Jacobian","page":"16 Differentiable programming","title":"Computing the Jacobian","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"The high-level interface to compute the Jacobian this way is jacobian_ad_forward. First, we load the required packages and compute the Jacobian of a semidiscretization of the compressible Euler equations, a system of nonlinear conservation laws.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, LinearAlgebra, Plots\n\nequations = CompressibleEulerEquations2D(1.4)\n\nsolver = DGSEM(3, flux_central)\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi);\nsize(J)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Next, we compute the eigenvalues of the Jacobian.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ = eigvals(J)\nscatter(real.(λ), imag.(λ), label=\"central flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"As you can see here, the maximal real part is close to zero.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Interestingly, if we add dissipation by switching to the flux_lax_friedrichs at the interfaces, the maximal real part of the eigenvalues increases.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"solver = DGSEM(3, flux_lax_friedrichs)\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\nλ = eigvals(J)\n\nscatter!(real.(λ), imag.(λ), label=\"Lax-Friedrichs flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Although the maximal real part is still somewhat small, it's larger than for the purely central discretization.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"However, we should be careful when using this analysis, since the eigenvectors are not necessarily well-conditioned.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ, V = eigen(J)\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"In one space dimension, the situation is a bit different.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"equations = CompressibleEulerEquations1D(1.4)\n\nsolver = DGSEM(3, flux_central)\nmesh = TreeMesh((-1.0,), (1.0,), initial_refinement_level=6, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\n\nλ = eigvals(J)\n\nscatter(real.(λ), imag.(λ), label=\"central flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Here, the maximal real part is basically zero to machine accuracy.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Moreover, the eigenvectors are not as ill-conditioned as in 2D.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ, V = eigen(J)\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"If we add dissipation, the maximal real part is still approximately zero.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"solver = DGSEM(3, flux_lax_friedrichs)\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\nλ = eigvals(J)\n\nscatter!(real.(λ), imag.(λ), label=\"Lax-Friedrichs flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"As you can see from the plot generated above, the maximal real part is still basically zero to machine precision.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Let's check the condition number of the eigenvectors.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ, V = eigen(J)\n\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Note that the condition number of the eigenvector matrix increases but is still smaller than for the example in 2D.","category":"page"},{"location":"tutorials/differentiable_programming/#Computing-other-derivatives","page":"16 Differentiable programming","title":"Computing other derivatives","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"It is also possible to compute derivatives of other dependencies using AD in Trixi.jl. For example, you can compute the gradient of an entropy-dissipative semidiscretization with respect to the ideal gas constant of the compressible Euler equations as described in the following. This example is also available as the elixir examples/special_elixirs/elixir_euler_ad.jl","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"First, we create a semidiscretization of the compressible Euler equations.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, LinearAlgebra, ForwardDiff\n\nequations = CompressibleEulerEquations2D(1.4)\n\n\"\"\"\n initial_condition_isentropic_vortex(x, t, equations::CompressibleEulerEquations2D)\n\nThe classical isentropic vortex test case of\n- Chi-Wang Shu (1997)\n Essentially Non-Oscillatory and Weighted Essentially Non-Oscillatory\n Schemes for Hyperbolic Conservation Laws\n [NASA/CR-97-206253](https://ntrs.nasa.gov/citations/19980007543)\n\"\"\"\nfunction initial_condition_isentropic_vortex(x, t, equations::CompressibleEulerEquations2D)\n inicenter = SVector(0.0, 0.0) # initial center of the vortex\n iniamplitude = 5.0 # size and strength of the vortex\n\n rho = 1.0 # base flow\n v1 = 1.0\n v2 = 1.0\n vel = SVector(v1, v2)\n p = 25.0\n\n rt = p / rho # ideal gas equation\n t_loc = 0.0\n\n cent = inicenter + vel*t_loc # shift advection of center to handle periodic BC, but only for v1 = v2 = 1.0\n cent = x - cent # distance to center point\n cent = SVector(-cent[2], cent[1])\n\n r2 = cent[1]^2 + cent[2]^2\n du = iniamplitude / (2*π) * exp(0.5 * (1 - r2)) # vel. perturbation\n dtemp = -(equations.gamma - 1) / (2 * equations.gamma * rt) * du^2 # isentropic\n\n rho = rho * (1 + dtemp)^(1 / (equations.gamma - 1))\n vel = vel + du * cent\n v1, v2 = vel\n p = p * (1 + dtemp)^(equations.gamma / (equations.gamma - 1))\n\n prim = SVector(rho, v1, v2, p)\n return prim2cons(prim, equations)\nend\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsolver = DGSEM(3, flux_lax_friedrichs, VolumeIntegralFluxDifferencing(flux_ranocha))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_isentropic_vortex, solver)\n\nu0_ode = Trixi.compute_coefficients(0.0, semi)\nsize(u0_ode)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Next, we compute the Jacobian using ForwardDiff.jacobian.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"J = ForwardDiff.jacobian((du_ode, γ) -> begin\n equations_inner = CompressibleEulerEquations2D(first(γ))\n semi_inner = Trixi.remake(semi, equations=equations_inner, uEltype=eltype(γ))\n Trixi.rhs!(du_ode, u0_ode, semi_inner, 0.0)\nend, similar(u0_ode), [1.4]); # γ needs to be an `AbstractArray`\n\nround.(extrema(J), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Note that we create a semidiscretization semi at first to determine the state u0_ode around which we want to perform the linearization. Next, we wrap the RHS evaluation inside a closure and pass that to ForwardDiff.jacobian. There, we need to make sure that the internal caches are able to store dual numbers from ForwardDiff.jl by setting uEltype appropriately. A similar approach is used by jacobian_ad_forward.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Note that the ideal gas constant does not influence the semidiscrete rate of change of the density, as demonstrated by","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"norm(J[1:4:end])","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Here, we used some knowledge about the internal memory layout of Trixi.jl, an array of structs with the conserved variables as fastest-varying index in memory.","category":"page"},{"location":"tutorials/differentiable_programming/#Differentiating-through-a-complete-simulation","page":"16 Differentiable programming","title":"Differentiating through a complete simulation","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"It is also possible to differentiate through a complete simulation. As an example, let's differentiate the total energy of a simulation using the linear scalar advection equation with respect to the wave number (frequency) of the initial data.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, OrdinaryDiffEq, ForwardDiff, Plots\n\nfunction energy_at_final_time(k) # k is the wave number of the initial condition\n equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\n solver = DGSEM(3, flux_lax_friedrichs)\n initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\n end\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k))\n ode = semidiscretize(semi, (0.0, 1.0))\n sol = solve(ode, BS3(), save_everystep=false)\n Trixi.integrate(energy_total, sol.u[end], semi)\nend\n\nk_values = range(0.9, 1.1, length=101)\n\nplot(k_values, energy_at_final_time.(k_values), label=\"Energy\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"You see a plot of a curve that resembles a parabola with local maximum around k = 1.0. Why's that? Well, the domain is fixed but the wave number changes. Thus, if the wave number is not chosen as an integer, the initial condition will not be a smooth periodic function in the given domain. Hence, the dissipative surface flux (flux_lax_friedrichs in this example) will introduce more dissipation. In particular, it will introduce more dissipation for \"less smooth\" initial data, corresponding to wave numbers k further away from integers.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"We can compute the discrete derivative of the energy at the final time with respect to the wave number k as follows.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"round(ForwardDiff.derivative(energy_at_final_time, 1.0), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"This is rather small and we can treat it as zero in comparison to the value of this derivative at other wave numbers k.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"dk_values = ForwardDiff.derivative.((energy_at_final_time,), k_values);\n\nplot(k_values, dk_values, label=\"Derivative\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"If you remember basic calculus, a sufficient condition for a local maximum is that the first derivative vanishes and the second derivative is negative. We can also check this discretely.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"second_derivative = round(ForwardDiff.derivative(\n k -> Trixi.ForwardDiff.derivative(energy_at_final_time, k), 1.0),\n sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Having seen this application, let's break down what happens step by step.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"function energy_at_final_time(k) # k is the wave number of the initial condition\n equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\n solver = DGSEM(3, flux_lax_friedrichs)\n initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\n end\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k))\n ode = semidiscretize(semi, (0.0, 1.0))\n sol = solve(ode, BS3(), save_everystep=false)\n Trixi.integrate(energy_total, sol.u[end], semi)\nend\n\nk = 1.0\nround(ForwardDiff.derivative(energy_at_final_time, k), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"When calling ForwardDiff.derivative(energy_at_final_time, k) with k=1.0, ForwardDiff.jl will basically use the chain rule and known derivatives of existing basic functions to calculate the derivative of the energy at the final time with respect to the wave number k at k0 = 1.0. To do this, ForwardDiff.jl uses dual numbers, which basically store the result and its derivative w.r.t. a specified parameter at the same time. Thus, we need to make sure that we can treat these ForwardDiff.Dual numbers everywhere during the computation. Fortunately, generic Julia code usually supports these operations. The most basic problem for a developer is to ensure that all types are generic enough, in particular the ones of internal caches.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"The first step in this example creates some basic ingredients of our simulation.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\nsolver = DGSEM(3, flux_lax_friedrichs);\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"These do not have internal caches storing intermediate values of the numerical solution, so we do not need to adapt them. In fact, we could also define them outside of energy_at_final_time (but would need to take care of globals or wrap everything in another function).","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Next, we define the initial condition","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\nend;\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"as a closure capturing the wave number k passed to energy_at_final_time. If you call energy_at_final_time(1.0), k will be a Float64. Thus, the return values of initial_condition will be SVectors of Float64s. When calculating the ForwardDiff.derivative, k will be a ForwardDiff.Dual number. Hence, the initial_condition will return SVectors of ForwardDiff.Dual numbers.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"The semidiscretization semi uses some internal caches to avoid repeated allocations and speed up the computations, e.g. for numerical fluxes at interfaces. Thus, we need to tell Trixi.jl to allow ForwardDiff.Dual numbers in these caches. That's what the keyword argument uEltype=typeof(k) in","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k));\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"does. This is basically the only part where you need to modify your standard Trixi.jl code to enable automatic differentiation. From there on, the remaining steps","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"ode = semidiscretize(semi, (0.0, 1.0))\nsol = solve(ode, BS3(), save_everystep=false)\nround(Trixi.integrate(energy_total, sol.u[end], semi), sigdigits=5)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"do not need any modifications since they are sufficiently generic (and enough effort has been spend to allow general types inside these calls).","category":"page"},{"location":"tutorials/differentiable_programming/#Propagating-errors-using-Measurements.jl","page":"16 Differentiable programming","title":"Propagating errors using Measurements.jl","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"(Image: Error bars by Randall Munroe) \"Error bars\" by Randall Munroe, linked from https://xkcd.com/2110","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Similar to AD, Trixi.jl also allows propagating uncertainties using linear error propagation theory via Measurements.jl. As an example, let's create a system representing the linear advection equation in 1D with an uncertain velocity. Then, we create a semidiscretization using a sine wave as initial condition, solve the ODE, and plot the resulting uncertainties in the primitive variables.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, OrdinaryDiffEq, Measurements, Plots, LaTeXStrings\n\nequations = LinearScalarAdvectionEquation1D(1.0 ± 0.1)\n\nmesh = TreeMesh((-1.0,), (1.0,), n_cells_max=10^5, initial_refinement_level=5)\n\nsolver = DGSEM(3)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test,\n solver, uEltype=Measurement{Float64})\n\node = semidiscretize(semi, (0.0, 1.5))\n\nsol = solve(ode, BS3(), save_everystep=false);\n\nplot(sol)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"You should see a plot where small error bars are shown around the extrema and larger error bars are shown in the remaining parts. This result is in accordance with expectations. Indeed, the uncertain propagation speed will affect the extrema less since the local variation of the solution is relatively small there. In contrast, the local variation of the solution is large around the turning points of the sine wave, so the uncertainties will be relatively large there.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"All this is possible due to allowing generic types and having good abstractions in Julia that allow packages to work together seamlessly.","category":"page"},{"location":"tutorials/differentiable_programming/#Finite-difference-approximations","page":"16 Differentiable programming","title":"Finite difference approximations","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Trixi.jl provides the convenience function jacobian_fd to approximate the Jacobian via central finite differences.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, LinearAlgebra\n\nequations = CompressibleEulerEquations2D(1.4)\n\nsolver = DGSEM(3, flux_central)\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ_fd = jacobian_fd(semi)\n\nJ_ad = jacobian_ad_forward(semi)\n\nrelative_difference = norm(J_fd - J_ad) / size(J_fd, 1)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"This discrepancy is of the expected order of magnitude for central finite difference approximations.","category":"page"},{"location":"tutorials/differentiable_programming/#Linear-systems","page":"16 Differentiable programming","title":"Linear systems","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"When a linear PDE is discretized using a linear scheme such as a standard DG method, the resulting semidiscretization yields an affine ODE of the form","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"partial_t u(t) = A u(t) + b","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"where A is a linear operator (\"matrix\") and b is a vector. Trixi.jl allows you to obtain this linear structure in a matrix-free way by using linear_structure. The resulting operator A can be used in multiplication, e.g. mul! from LinearAlgebra, converted to a sparse matrix using sparse from SparseArrays, or converted to a dense matrix using Matrix for detailed eigenvalue analyses. For example,","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using Trixi, LinearAlgebra, Plots\n\nequations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n\nsolver = DGSEM(3, flux_lax_friedrichs)\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test, solver)\n\nA, b = linear_structure(semi)\n\nsize(A), size(b)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"Next, we compute the eigenvalues of the linear operator.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ = eigvals(Matrix(A))\n\nscatter(real.(λ), imag.(λ))","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"As you can see here, the maximal real part is close to machine precision.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"λ = eigvals(Matrix(A))\nrelative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/#Package-versions","page":"16 Differentiable programming","title":"Package versions","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\", \"ForwardDiff\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"","category":"page"},{"location":"tutorials/differentiable_programming/","page":"16 Differentiable programming","title":"16 Differentiable programming","text":"This page was generated using Literate.jl.","category":"page"},{"location":"meshes/tree_mesh/#Tree-mesh","page":"Tree mesh","title":"Tree mesh","text":"","category":"section"},{"location":"meshes/tree_mesh/","page":"Tree mesh","title":"Tree mesh","text":"The TreeMesh is a Cartesian, h-non-conforming mesh type used in many parts of Trixi.jl. Often, the support for this mesh type is developed best since it was the first mesh type in Trixi.jl, and it is available in one, two, and three space dimensions.","category":"page"},{"location":"meshes/tree_mesh/","page":"Tree mesh","title":"Tree mesh","text":"It is limited to hypercube domains but supports AMR via the AMRCallback. Due to its Cartesian nature, (numerical) fluxes need to implement methods dispatching on the orientation::Integer as described in the conventions.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Unstructured-quadrilateral-mesh","page":"Unstructured mesh","title":"Unstructured quadrilateral mesh","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The UnstructuredMesh2D is an unstructured, curvilinear, conforming mesh type in two space dimensions.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Next, we describe the conventions taken in the implementation for two-dimensional unstructured quadrilateral meshes. Principally, this relates to how a file with the extension .mesh encodes information about the numbering and orientation of elements in an unstructured quadrilateral mesh with possibly curved boundaries.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We use the following unstructured mesh with three elements for this discussion:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"(Image: example-mesh)","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Further, a simulation using Trixi.jl on this example unstructured mesh is provided in examples/unstructured_2d_dgsem/elixir_euler_basic.jl.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Mesh-file-header","page":"Unstructured mesh","title":"Mesh file header","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The first two lines of the mesh file lists the mesh file type as well as the total number of corners, surfaces, elements, and the polynomial degree that the mesh will use to represent any curved sides. For the example mesh these quantities are","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"ISM-V2\n 7 9 3 8","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"corresponding to seven corners, nine surfaces, and three elements. The mesh polynomial degree of eight is taken only for illustrative purposes. In practice, this mesh polynomial degree depends on the particular application for which the curved, unstructured mesh is required.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-corner-nodes","page":"Unstructured mesh","title":"List of corner nodes","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"After these global counts that prescribe information about the mesh skeleton, the mesh file give a list of the physical (x,y) coordinates of all the corners. The corner nodes are listed in the order prescribed by mesh generator. Thus, for the example mesh this node list would be","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 1.0 -1.0\n 3.0 0.0\n 1.0 1.0\n 2.0 0.0\n 0.0 0.0\n 3.0 1.0\n 3.0 -1.0","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The corner nodes are internally referenced by their position in the list above. For example, here the node at (1.0, -1.0) would have node id 1, node id 2 would be at (3.0, 0.0) etc.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-neighbor-connectivity","page":"Unstructured mesh","title":"List of neighbor connectivity","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"After the corner list comes the neighbor connectivity along each surface in the mesh. This includes local indexing and orientation information necessary to compute the coupling between elements in the mesh. In 2D each surface is defined by connecting two nodes indexed as with the numbering above. We adopt the convention that node id 1 < node id 2.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Each surface will have two neighbors where the element on the left locally as one \"walks\" from node id 1 to node id 2 is taken to be the primary element and the element locally on the right is taken to be the secondary element. If, however, there is no secondary element index, then the surface lies along a physical boundary. In this case the only available element index is considered to be primary and the secondary index is set to zero.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The final two index numbers within the neighbor information list are used to identify the local surface within each element. The first local surface index (on the primary element) will always be positive whereas the second local surface index (on the primary element) can be positive or negative. If the second local surface index is positive, then the local coordinate systems in the primary element and secondary element match, i.e., the indexing on either side runs from 1:polydeg+1. However, if the local surface index of the secondary element is negative in the mesh file, then the coordinate system in the secondary element is flipped with respect to the primary element. Therefore, care must be taken in the implementation to ensure that the primary element indexing runs from 1:polydeg+1 whereas the secondary element indexing must run in reverse from polydeg+1:-1:1. Finally, if the secondary element index is zero, then so will be the local surface index because the surface is on a physical boundary. Also, there is no flipping of coordinate indexing required at the physical boundary because only the primary element's coordinate system exists.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Three-examples:-One-along-a-physical-boundary-and-two-along-interior-surfaces.","page":"Unstructured mesh","title":"Three examples: One along a physical boundary and two along interior surfaces.","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {8} we connect node (2) to node (7) and are along a physical boundary in element 3 with the local surface index 1 and the neighbor information:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 7 3 0 1 0","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {1} we connect node (2) to node (4) such that the primary element is 3 with local surface index 2 and the secondary element is 2 with local surface index 1. Furthermore, we see that coordinate system in the secondary element 2 is flipped with respect to the primary element's coordinate system such that the sign of the local surface index in the secondary element flips. This gives the following neighbor information:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 4 3 2 2 -1","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {4} we connect node (1) to node (4) such that the primary element is 1 with local surface index 2 and the secondary element is 3 with local surface index 3. The coordinate systems in both elements match and no sign change is required on the local surface index in the secondary element:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 1 4 1 3 2 3","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We collect the complete neighbor information for the example mesh above:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 4 3 2 2 -1\n 3 5 1 0 4 0\n 1 5 1 0 1 0\n 1 4 1 3 2 3\n 2 6 2 0 2 0\n 1 7 3 0 4 0\n 3 6 2 0 3 0\n 2 7 3 0 1 0\n 3 4 2 1 4 -3","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-elements","page":"Unstructured mesh","title":"List of elements","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. We connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion thus making the element right-handed indicated using the circular arrow in the figure above. In turn, this right-handedness defines the local surface indexing (i.e. the four local sides) and the local (xi eta) coordinate system. For example, the four corners for element 1 would be listed as","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The mesh file also encodes information for curved surfaces either interior to the domain (as surface {9} above) or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (1, 2, 3, or 4) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y) data will be given in the direction of the local coordinate system.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The last piece of information provided by the mesh file are labels for the different surfaces of an element. These labels are useful to set boundary conditions along physical surfaces. The labels can be short descriptive words. The label --- indicates an internal surface where no boundary condition is required.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"As an example, the complete information for element 1 in the example mesh would be","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3\n 0 0 1 1\n 1.000000000000000 1.000000000000000\n 1.024948365654583 0.934461926834452\n 1.116583018200151 0.777350964621867\n 1.295753434047077 0.606254343587194\n 1.537500000000000 0.462500000000000\n 1.768263070247418 0.329729152118310\n 1.920916981799849 0.185149035378133\n 1.986035130050921 0.054554577460044\n 2.000000000000000 0.0\n 0.0 0.0\n 0.035513826946206 0.105291711848750\n 0.148591270347399 0.317731556850611\n 0.340010713990041 0.452219430075470\n 0.575000000000000 0.462500000000000\n 0.788022294598950 0.483764065630034\n 0.926408729652601 0.644768443149389\n 0.986453164464803 0.883724792445746\n 1.000000000000000 1.000000000000000\n Slant --- --- Bezier","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"where the curved boundary information is encoded \"back to back\". That is, the first nine (x,y) nodes in the list above correspond to the interior boundary curve along local side 3 in element 1 and the next nine (x,y) nodes denote the curved physical boundary named Bezier along local side 4.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We collect the complete set of element information for the example mesh","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3\n 0 0 1 1\n 1.000000000000000 1.000000000000000\n 1.024948365654583 0.934461926834452\n 1.116583018200151 0.777350964621867\n 1.295753434047077 0.606254343587194\n 1.537500000000000 0.462500000000000\n 1.768263070247418 0.329729152118310\n 1.920916981799849 0.185149035378133\n 1.986035130050921 0.054554577460044\n 2.000000000000000 0.0\n 0.0 0.0\n 0.035513826946206 0.105291711848750\n 0.148591270347399 0.317731556850611\n 0.340010713990041 0.452219430075470\n 0.575000000000000 0.462500000000000\n 0.788022294598950 0.483764065630034\n 0.926408729652601 0.644768443149389\n 0.986453164464803 0.883724792445746\n 1.000000000000000 1.000000000000000\n Slant --- --- Bezier\n 4 2 6 3\n 0 0 0 1\n 2.000000000000000 0.0\n 1.986035130050921 0.054554577460044\n 1.920916981799849 0.185149035378133\n 1.768263070247418 0.329729152118310\n 1.537500000000000 0.462500000000000\n 1.295753434047077 0.606254343587194\n 1.116583018200151 0.777350964621867\n 1.024948365654583 0.934461926834452\n 1.000000000000000 1.000000000000000\n --- Right Top ---\n 7 2 4 1\n 0 0 0 0\n Right --- --- Bottom","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Trixi.jl-on-an-unstructured-quadrilateral-mesh","page":"Unstructured mesh","title":"Trixi.jl on an unstructured quadrilateral mesh","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We provide an example simulation on an unstructured quadrilateral mesh by executing","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"julia> trixi_include(default_example_unstructured())","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Note this may download a copy of the mesh file described above for the three element unstructured mesh. This elixir provides the solution for the compressible Euler equations in two spatial dimensions for a smooth propagating wave solution. Below we provide the time evolution of the pressure wave for this example created with the Trixi2Vtk tool and visualized using ParaView.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" \n
","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"EditURL = \"../../literate/src/files/time_stepping.jl\"","category":"page"},{"location":"tutorials/time_stepping/#time_stepping","page":"15 Explicit time stepping","title":"15: Explicit time stepping","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"For the time integration, Trixi.jl uses the package OrdinaryDiffEq.jl from the SciML ecosystem. The interface to this package is the solve(...) function. It always requires an ODE problem and a time integration algorithm as input parameters.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"solve(ode, alg; kwargs...)","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"In Trixi.jl, the ODE problem is created by semidiscretize(semi, tspan) for a semidiscretization semi and the time span tspan. In particular, semidiscretize returns an ODEProblem used by OrdinaryDiffEq.jl.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"OrdinaryDiffEq.jl provides many integration algorithms, which are summarized in the documentation. Particularly interesting for Trixi.jl are their strong stability preserving (SSP) methods and low-storage methods. There are some differences regarding the choice of the used time step.","category":"page"},{"location":"tutorials/time_stepping/#adaptive_step_sizes","page":"15 Explicit time stepping","title":"Error-based adaptive step sizes","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"First, we treat time integration algorithms with adaptive step sizes, such as SSPRK43. It is used in some elixirs, like elixir_euler_colliding_flow.jl or elixir_euler_astro_jet_amr.jl.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"Other error-based adaptive integration algorithms are for instance RDPK3SpFSAL35, RDPK3Sp35, RDPK3SpFSAL49, RDPK3Sp49, RDPK3SpFSAL510, RDPK3Sp510.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"They already contain an error-based adaptive step size control and heuristics to guess a starting step size. If this heuristic fails in your case, you can specify an appropriately small initial step size as keyword argument dt=... of solve.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"If you run Trixi in parallel with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to enable MPI aware error-based adaptive step size control. These keyword arguments are also included in ode_default_options.","category":"page"},{"location":"tutorials/time_stepping/#CFL-based-step-size-control","page":"15 Explicit time stepping","title":"CFL-based step size control","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"The SciML ecosystem also provides time integration algorithms without adaptive time stepping on their own, such as CarpenterKennedy2N54. Moreover, you also can deactivate the automatic adaptivity of adaptive integration algorithms by passing adaptive=false in the solve function.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"These algorithms require another way of setting the step size. You have to pass dt=... in the solve function. Without other settings, the simulation uses this fixed time step.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"For hyperbolic PDEs, it is natural to use an adaptive CFL-based step size control. Here, the time step is proportional to a ratio of the local measure of mesh spacing Delta x_i for an element i and the maximum (local) wave speed lambda_max related to the largest-magnitude eigenvalue of the flux Jacobian of the hyperbolic system.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"Delta t_n = textCFL * min_i fracDelta x_ilambda_max(u_i^n)","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"We compute Delta x_i by scaling the element size by a factor of 1(N+1), cf. Gassner and Kopriva (2011), Section 5.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"Trixi.jl provides such a CFL-based step size control. It is implemented as the callback StepsizeCallback.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"stepsize_callback = StepsizeCallback(; cfl=1.0)","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"A suitable CFL number depends on many parameters such as the chosen grid, the integration algorithm and the polynomial degree of the spatial DG discretization. So, the optimal number for an example is mostly determined experimentally.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"You can add this CFL-based step size control to your simulation like any other callback.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"callbacks = CallbackSet(stepsize_callback)\nalg = CarpenterKennedy2N54(williamson_condition=false)\nsolve(ode, alg;\n dt=1.0 # solve needs some value here but it will be overwritten by the stepsize_callback\n callback=callbacks)","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"You can find simple examples with a CFL-based step size control for instance in the elixirs elixir_advection_basic.jl or elixir_euler_source_terms.jl.","category":"page"},{"location":"tutorials/time_stepping/#Package-versions","page":"15 Explicit time stepping","title":"Package versions","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"","category":"page"},{"location":"tutorials/time_stepping/","page":"15 Explicit time stepping","title":"15 Explicit time stepping","text":"This page was generated using Literate.jl.","category":"page"},{"location":"contributing/#Contributing","page":"Contributing","title":"Contributing","text":"","category":"section"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Trixi.jl is an open-source project and we are very happy to accept contributions from the community. Please feel free to open issues or submit patches (preferably as merge requests) any time. For planned larger contributions, it is often beneficial to get in contact with one of the principal developers first (see Authors).","category":"page"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Trixi.jl and its contributions are licensed under the MIT license (see License). As a contributor, you certify that all your contributions are in conformance with the Developer Certificate of Origin (Version 1.1), which is reproduced below.","category":"page"},{"location":"contributing/#Developer-Certificate-of-Origin-(Version-1.1)","page":"Contributing","title":"Developer Certificate of Origin (Version 1.1)","text":"","category":"section"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"The following text was taken from https://developercertificate.org:","category":"page"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Developer Certificate of Origin\nVersion 1.1\n\nCopyright (C) 2004, 2006 The Linux Foundation and its contributors.\n1 Letterman Drive\nSuite D4700\nSan Francisco, CA, 94129\n\nEveryone is permitted to copy and distribute verbatim copies of this\nlicense document, but changing it is not allowed.\n\n\nDeveloper's Certificate of Origin 1.1\n\nBy making a contribution to this project, I certify that:\n\n(a) The contribution was created in whole or in part by me and I\n have the right to submit it under the open source license\n indicated in the file; or\n\n(b) The contribution is based upon previous work that, to the best\n of my knowledge, is covered under an appropriate open source\n license and I have the right under that license to submit that\n work with modifications, whether created in whole or in part\n by me, under the same open source license (unless I am\n permitted to submit under a different license), as indicated\n in the file; or\n\n(c) The contribution was provided directly to me by some other\n person who certified (a), (b) or (c) and I have not modified\n it.\n\n(d) I understand and agree that this project and the contribution\n are public and that a record of the contribution (including all\n personal information I submit with it, including my sign-off) is\n maintained indefinitely and may be redistributed consistent with\n this project or the open source license(s) involved.","category":"page"},{"location":"performance/#Performance","page":"Performance","title":"Performance","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"Trixi.jl is designed to balance performance and readability. Since Julia provides a lot of zero-cost abstractions, it is often possible to optimize both goals simultaneously.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The usual development workflow in Julia is","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Make it work.\nMake it nice.\nMake it fast.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"To achieve the third step, you should be familiar with (at least) the section on performance tips in the Julia manual. Here, we just list some important aspects you should consider when developing Trixi.jl.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Consider using @views/view(...) when using array slices, except on the left-side of an assignment (further details).\nFunctions are essentially for free, since they are usually automatically inlined where it makes sense (using @inline can be used as an additional hint to the compiler) (further details).\nFunction barriers can improve performance due to type stability (further details).\nLook for type instabilities using @code_warntype. Consider using @descend from Cthulhu.jl to investigate deeper call chains.","category":"page"},{"location":"performance/#Manual-benchmarking","page":"Performance","title":"Manual benchmarking","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"If you modify some internal parts of Trixi.jl, you should check the impact on performance. Hence, you should at least investigate the performance roughly by comparing the reported timings of several elixirs. Deeper investigations and micro-benchmarks should usually use BenchmarkTools.jl. For example, the following steps were used to benchmark the changes introduced in PR #256.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"git checkout e7ebf3846b3fd62ee1d0042e130afb50d7fe8e48 (new version)\nStart julia --threads=1 --check-bounds=no.\nExecute the following code in the REPL to benchmark the rhs! call at the final state.\njulia> using BenchmarkTools, Revise; using Trixi\n\njulia> trixi_include(\"examples/2d/elixir_euler_sedov_blast_wave.jl\")\n\njulia> du_test = copy(sol.u[end]); u_test = copy(sol.u[end]);\n\njulia> @benchmark Trixi.rhs!(\n $(du_test),\n $(u_test),\n $(semi),\n $(sol.t[end]))\nBenchmarkTools.Trial:\n memory estimate: 10.48 KiB\n allocs estimate: 67\n --------------\n minimum time: 4.510 ms (0.00% GC)\n median time: 4.646 ms (0.00% GC)\n mean time: 4.699 ms (0.00% GC)\n maximum time: 7.183 ms (0.00% GC)\n --------------\n samples: 1065\n evals/sample: 1\n\nshell> git checkout 222241ff54f8a4ca9876cc1fc25ae262416a4ea0\n\njulia> trixi_include(\"examples/2d/elixir_euler_sedov_blast_wave.jl\")\n\njulia> @benchmark Trixi.rhs!(\n $(du_test),\n $(u_test),\n $(semi),\n $(sol.t[end]))\nBenchmarkTools.Trial:\n memory estimate: 10.36 KiB\n allocs estimate: 67\n --------------\n minimum time: 4.500 ms (0.00% GC)\n median time: 4.635 ms (0.00% GC)\n mean time: 4.676 ms (0.00% GC)\n maximum time: 5.880 ms (0.00% GC)\n --------------\n samples: 1070\n evals/sample: 1\nRun the @benchmark ... commands multiple times to see whether there are any significant fluctuations.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Follow these steps for both commits you want to compare. The relevant benchmark results you should typically be looking at are the median and mean values of the runtime and the memory/allocs estimate. In this example, the differences of the runtimes are of the order of the fluctuations one gets when running the benchmarks multiple times. Since the memory/allocs are (roughly) the same, there doesn't seem to be a significant performance regression here.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"You can also make it more detailed by benchmarking only, e.g., the calculation of the volume terms, but whether that's necessary depends on the modifications you made and their (potential) impact.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Some more detailed description of manual profiling and benchmarking as well as resulting performance improvements of Trixi.jl are given in the following blog posts.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Improving performance of AMR with p4est, cf. #638\nImproving performance of EC methods, cf. #643","category":"page"},{"location":"performance/#Automated-benchmarking","page":"Performance","title":"Automated benchmarking","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"We use PkgBenchmark.jl to provide a standard set of benchmarks for Trixi.jl. The relevant benchmark script is benchmark/benchmarks.jl. You can run a standard set of benchmarks via","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> using PkgBenchmark, Trixi\n\njulia> results = benchmarkpkg(Trixi, BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`))\n\njulia> export_markdown(pkgdir(Trixi, \"benchmark\", \"single_benchmark.md\"), results)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"This will save a markdown file with a summary of the benchmark results similar to this example. Note that this will take quite some time. Additional options are described in the docs of PkgBenchmark.jl. A particularly useful option is to specify a BenchmarkConfig including Julia command line options affecting the performance such as disabling bounds-checking and setting the number of threads.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"A useful feature when developing Trixi.jl is to compare the performance of Trixi.jl's current state vs. the main branch. This can be achieved by executing","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> using PkgBenchmark, Trixi\n\njulia> results = judge(Trixi,\n BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`), # target\n BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`, id=\"main\") # baseline\n )\n\njulia> export_markdown(pkgdir(Trixi, \"benchmark\", \"results.md\"), results)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"By default, the target is the current state of the repository. Remember that you need to be in a clean state (commit or stash your changes) to run this successfully. You can also run this comparison and an additional one using two threads via","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> include(\"benchmark/run_benchmarks.jl\")","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Then, markdown files including the results are saved in benchmark/. This example result was obtained using a GitHub action for the PR #535. Note that GitHub actions run on in the cloud in a virtual machine. Hence, we do not really have control over it and performance results must be taken with a grain of salt. Nevertheless, significant runtime differences and differences of memory allocations should be robust indicators of performance changes.","category":"page"},{"location":"performance/#Runtime-performance-vs.-latency-aka-using-@nospecialize-selectively","page":"Performance","title":"Runtime performance vs. latency aka using @nospecialize selectively","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"Usually, Julia will compile specialized versions of each method, using as much information from the types of function arguments as possible (based on some heuristics). The compiler will generate code that is as efficient as comparable code written in a low-level language such as C or Fortran. However, there are cases where the runtime performance does not really matter but the time needed to compile specializations becomes significant. This is related to latency or the time-to-first-plot problem, well-known in the Julia community. In such a case, it can be useful to remove some burden from the compiler by avoiding specialization on every possible argument types using the macro @nospecialize. A prime example of such a case is pretty printing of structs in the Julia REPL, see the associated PR for further discussions.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"As a rule of thumb:","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Do not use @nospecialize in performance-critical parts, in particular not for methods involved in computing Trixi.rhs!.\nConsider using @nospecialize for methods like custom implementations of Base.show.","category":"page"},{"location":"performance/#performance-metrics","page":"Performance","title":"Performance metrics of the AnalysisCallback","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The AnalysisCallback computes two performance indicators that you can use to evaluate the serial and parallel performance of Trixi.jl. They represent measured run times that are normalized by the number of rhs! evaluations and the number of degrees of freedom of the problem setup. The normalization ensures that we can compare different measurements for each type of indicator independent of the number of time steps or mesh size. All indicators have in common that they are still in units of time, thus lower is better for each of them.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Here, the term \"degrees of freedom\" (DOFs) refers to the number of independent state vectors that are used to represent the numerical solution. For example, if you use a DGSEM-type scheme in 2D on a mesh with 8 elements and with 5-by-5 Gauss-Lobatto nodes in each element (i.e., a polynomial degree of 4), the total number of DOFs would be","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"n_textDOFsDGSEM = textnumber of elements cdot textnumber of nodes per element = 8 cdot (5 cdot 5) = 200","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"In contrast, for a finite volume-type scheme on a mesh with 8 elements, the total number of DOFs would be (independent of the number of spatial dimensions)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"n_textDOFsFV = textnumber of elements = 8","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"since for standard finite volume methods you store a single state vector in each element. Note that we specifically count the number of state vectors and not the number of state variables for the DOFs. That is, in the previous example n_textDOFsFV is equal to 8 independent of whether this is a compressible Euler setup with 5 state variables or a linear scalar advection setup with one state variable.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"For each indicator, the measurements are always since the last invocation of the AnalysisCallback. That is, if the analysis callback is called multiple times, the indicators are repeatedly computed and can thus also be used to track the performance over the course of a longer simulation, e.g., to analyze setups with varying performance characteristics. Note that the time spent in the AnalysisCallback itself is always excluded, i.e., the performance measurements are not distorted by potentially expensive solution analysis computations. All other parts of a Trixi.jl simulation are included, however, thus make sure that you disable everything you do not want to be measured (such as I/O callbacks, visualization etc.).","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"note: Performance indicators and adaptive mesh refinement\nCurrently it is not possible to compute meaningful performance indicators for a simulation with arbitrary adaptive mesh refinement, since this would require to explicitly keep track of the number of DOF updates due to the mesh size changing repeatedly. The only way to do this at the moment is by setting the analysis interval to the same value as the AMR interval.","category":"page"},{"location":"performance/#Local,-rhs!-only-indicator","page":"Performance","title":"Local, rhs!-only indicator","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The local, rhs!-only indicator is computed as","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"texttimeDOFrhs = fract_texttextttrhsn_textDOFslocal cdot n_textcallstextttrhs","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"where t_texttextttrhs is the accumulated time spent in rhs!, n_textDOFslocal is the local number of DOFs (i.e., on the current MPI rank; if doing a serial run, you can just think of this as the number of DOFs), and n_textcallstextttrhs is the number of times the rhs! function has been evaluated. Note that for this indicator, we measure only the time spent in rhs!, i.e., by definition all computations outside of rhs! - specifically all other callbacks and the time integration method - are not taken into account.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The local, rhs!-only indicator is usually most useful if you do serial measurements and are interested in the performance of the implementation of your core numerical methods (e.g., when doing performance tuning).","category":"page"},{"location":"performance/#Performance-index-(PID)","page":"Performance","title":"Performance index (PID)","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The performance index (PID) is computed as","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"textPID = fract_textwall cdot n_textranksMPIn_textDOFsglobal cdot n_textcallstextttrhs","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"where t_textwall is the walltime since the last call to the AnalysisCallback, n_textranksMPI is the number of MPI ranks used, n_textDOFsglobal is the global number of DOFs (i.e., the sum of DOFs over all MPI ranks; if doing a serial run, you can just think of this as the number of DOFs), and n_textcallstextttrhs is the number of times the rhs! function has been evaluated since the last call to the AnalysisCallback. The PID measures everything except the time spent in the AnalysisCallback itself - specifically, all other callbacks and the time integration method itself are included.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The PID is usually most useful if you would like to compare the parallel performance of your code to its serial performance. Specifically, it allows you to evaluate the parallelization overhead of your code by giving you a measure of the resources that are necessary to solve a given simulation setup. In a sense, it mimics the \"core hours\" metric that is often used by supercomputer centers to measure how many resources a particular compute job requires. It can thus be seen as a proxy for \"energy used\" and, as an extension, \"monetary cost\".","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"note: Initialization overhead in measurements\nWhen using one of the integration schemes from OrdinaryDiffEq.jl, their implementation will initialize some OrdinaryDiffEq.jl-specific information during the first time step. Among other things, one additional call to rhs! is performed. Therefore, make sure that for performance measurements using the PID either the number of timesteps or the workload per rhs! call is large enough to make the initialization overhead negligible. Note that the extra call to rhs! is properly accounted for in both the number of calls and the measured time, so you do not need to worry about it being expensive. If you want a perfect timing result, you need to set the analysis interval such that the AnalysisCallback is invoked at least once during the course of the simulation and discard the first PID value.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"EditURL = \"../../literate/src/files/custom_semidiscretization.jl\"","category":"page"},{"location":"tutorials/custom_semidiscretization/#custom_semidiscretization","page":"17 Custom semidiscretizations","title":"17: Custom semidiscretizations","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"As described in the overview section, semidiscretizations are high-level descriptions of spatial discretizations in Trixi.jl. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Hyperbolic-parabolic problems based on the advection-diffusion equation or the compressible Navier-Stokes equations can be represented in a SemidiscretizationHyperbolicParabolic. This is described in the basic tutorial on parabolic terms and its extension to custom parabolic terms. In this tutorial, we will describe how these semidiscretizations work and how they can be used to create custom semidiscretizations involving also other tasks.","category":"page"},{"location":"tutorials/custom_semidiscretization/#Overview-of-the-right-hand-side-evaluation","page":"17 Custom semidiscretizations","title":"Overview of the right-hand side evaluation","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"The semidiscretizations provided by Trixi.jl are set up to create ODEProblems from the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem bundles an initial condition, a right-hand side (RHS) function, the time span, and possible parameters. The ODEProblems created by Trixi.jl use the semidiscretization passed to semidiscretize as a parameter. For a SemidiscretizationHyperbolic, the ODEProblem wraps Trixi.rhs! as ODE RHS. For a SemidiscretizationHyperbolicParabolic, Trixi.jl uses a SplitODEProblem combining Trixi.rhs_parabolic! for the (potentially) stiff part and Trixi.rhs! for the other part.","category":"page"},{"location":"tutorials/custom_semidiscretization/#Standard-Trixi.jl-setup","page":"17 Custom semidiscretizations","title":"Standard Trixi.jl setup","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"In this tutorial, we will consider the linear advection equation with source term","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"partial_t u(tx) + partial_x u(tx) = -exp(-t) sinbigl(pi (x - t) bigr)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"with periodic boundary conditions in the domain [-1, 1] as a model problem. The initial condition is","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"u(0x) = sin(pi x)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"The source term results in some damping and the analytical solution","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"u(tx) = exp(-t) sinbigl(pi (x - t) bigr)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"First, we discretize this equation using the standard functionality of Trixi.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"using Trixi, OrdinaryDiffEq, Plots","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"The linear scalar advection equation is already implemented in Trixi.jl as LinearScalarAdvectionEquation1D. We construct it with an advection velocity 1.0.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"equations = LinearScalarAdvectionEquation1D(1.0)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Next, we use a standard DGSEM solver.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"solver = DGSEM(polydeg = 3)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"We create a simple TreeMesh in 1D.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"coordinates_min = (-1.0,)\ncoordinates_max = (+1.0,)\nmesh = TreeMesh(coordinates_min, coordinates_max;\n initial_refinement_level = 4,\n n_cells_max = 10^4,\n periodicity = true)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"We wrap everything in in a semidiscretization and pass the source terms as a standard Julia function. Please note that Trixi.jl uses SVectors from StaticArrays.jl to store the conserved variables u. Thus, the return value of the source terms must be wrapped in an SVector - even if we consider just a scalar problem.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"function initial_condition(x, t, equations)\n return SVector(exp(-t) * sinpi(x[1] - t))\nend\n\nfunction source_terms_standard(u, x, t, equations)\n return -initial_condition(x, t, equations)\nend\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition,\n solver;\n source_terms = source_terms_standard)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Now, we can create the ODEProblem, solve the resulting ODE using a time integration method from OrdinaryDiffEq.jl, and visualize the numerical solution at the final time using Plots.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"tspan = (0.0, 3.0)\node = semidiscretize(semi, tspan)\n\nsol = solve(ode, RDPK3SpFSAL49(); ode_default_options()...)\n\nplot(sol; label = \"numerical sol.\", legend = :topright)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"We can also plot the analytical solution for comparison. Since Trixi.jl uses SVectors for the variables, we take their first (and only) component to get the scalar value for manual plotting.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"let\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topright)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"We can also add the initial condition to the plot.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"plot!(sol.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"You can of course also use some callbacks provided by Trixi.jl as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi; interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/#Using-a-custom-ODE-right-hand-side-function","page":"17 Custom semidiscretizations","title":"Using a custom ODE right-hand side function","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Next, we will solve the same problem but use our own ODE RHS function. To demonstrate this, we will artificially create a global variable containing the current time of the simulation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"const GLOBAL_TIME = Ref(0.0)\n\nfunction source_terms_custom(u, x, t, equations)\n t = GLOBAL_TIME[]\n return -initial_condition(x, t, equations)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Next, we create our own RHS function to update the global time of the simulation before calling the RHS function from Trixi.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"function rhs_source_custom!(du_ode, u_ode, semi, t)\n GLOBAL_TIME[] = t\n Trixi.rhs!(du_ode, u_ode, semi, t)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Next, we create an ODEProblem manually copying over the data from the one we got from semidiscretize earlier.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"ode_source_custom = ODEProblem(rhs_source_custom!,\n ode.u0,\n ode.tspan,\n ode.p #= semi =#)\nsol_source_custom = solve(ode_source_custom, RDPK3SpFSAL49();\n ode_default_options()...)\n\nplot(sol_source_custom; label = \"numerical sol.\")\nlet\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol_source_custom.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topleft)\nend\nplot!(sol_source_custom.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"This also works with callbacks as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi; interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode_source_custom, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/#Setting-up-a-custom-semidiscretization","page":"17 Custom semidiscretizations","title":"Setting up a custom semidiscretization","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Using a global constant is of course not really nice from a software engineering point of view. Thus, it can often be useful to collect additional data in the parameters of the ODEProblem. Thus, it is time to create our own semidiscretization. Here, we create a small wrapper of a standard semidiscretization of Trixi.jl and the current global time of the simulation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"struct CustomSemidiscretization{Semi, T} <: Trixi.AbstractSemidiscretization\n semi::Semi\n t::T\nend\n\nsemi_custom = CustomSemidiscretization(semi, Ref(0.0))","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"To get pretty printing in the REPL, you can consider specializing","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Base.show(io::IO, parameters::CustomSemidiscretization)\nBase.show(io::IO, ::MIME\"text/plain\", parameters::CustomSemidiscretization)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"for your custom semidiscretiation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Next, we create our own source terms that use the global time stored in the custom semidiscretiation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"source_terms_custom_semi = let semi_custom = semi_custom\n function source_terms_custom_semi(u, x, t, equations)\n t = semi_custom.t[]\n return -initial_condition(x, t, equations)\n end\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"We also create a custom ODE RHS to update the current global time stored in the custom semidiscretization. We unpack the standard semidiscretization created by Trixi.jl and pass it to Trixi.rhs!.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"function rhs_semi_custom!(du_ode, u_ode, semi_custom, t)\n semi_custom.t[] = t\n Trixi.rhs!(du_ode, u_ode, semi_custom.semi, t)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Finally, we set up an ODEProblem and solve it numerically.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"ode_semi_custom = ODEProblem(rhs_semi_custom!,\n ode.u0,\n ode.tspan,\n semi_custom)\nsol_semi_custom = solve(ode_semi_custom, RDPK3SpFSAL49();\n ode_default_options()...)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"If we want to make use of additional functionality provided by Trixi.jl, e.g., for plotting, we need to implement a few additional specializations. In this case, we forward everything to the standard semidiscretization provided by Trixi.jl wrapped in our custom semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Base.ndims(semi::CustomSemidiscretization) = ndims(semi.semi)\nfunction Trixi.mesh_equations_solver_cache(semi::CustomSemidiscretization)\n Trixi.mesh_equations_solver_cache(semi.semi)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Now, we can plot the numerical solution as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"plot(sol_semi_custom; label = \"numerical sol.\")\nlet\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol_semi_custom.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topleft)\nend\nplot!(sol_semi_custom.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"This also works with many callbacks as usual. However, the AnalysisCallback requires some special handling since it makes use of a performance counter contained in the standard semidiscretizations of Trixi.jl to report some performance metrics. Here, we forward all accesses to the performance counter to the wrapped semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"function Base.getproperty(semi::CustomSemidiscretization, s::Symbol)\n if s === :performance_counter\n wrapped_semi = getfield(semi, :semi)\n wrapped_semi.performance_counter\n else\n getfield(semi, s)\n end\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Moreover, the AnalysisCallback also performs some error calculations. We also need to forward them to the wrapped semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"function Trixi.calc_error_norms(func, u, t, analyzer,\n semi::CustomSemidiscretization,\n cache_analysis)\n Trixi.calc_error_norms(func, u, t, analyzer,\n semi.semi,\n cache_analysis)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"Now, we can work with the callbacks used before as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi_custom;\n interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode_semi_custom, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"For even more advanced usage of custom semidiscretizations, you may look at the source code of the ones contained in Trixi.jl, e.g.,","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"SemidiscretizationHyperbolicParabolic\nSemidiscretizationEulerGravity\nSemidiscretizationEulerAcoustics\nSemidiscretizationCoupled","category":"page"},{"location":"tutorials/custom_semidiscretization/#Package-versions","page":"17 Custom semidiscretizations","title":"Package versions","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"17 Custom semidiscretizations","title":"17 Custom semidiscretizations","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"EditURL = \"../../literate/src/files/DGSEM_FluxDiff.jl\"","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#DGSEM_FluxDiff","page":"2 DGSEM with flux differencing","title":"2: DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This tutorial starts with a presentation of the weak formulation of the discontinuous Galerkin spectral element method (DGSEM) in order to fix the notation of the used operators. Then, the DGSEM formulation with flux differencing (split form DGSEM) and its implementation in Trixi.jl is shown.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We start with the one-dimensional conservation law","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"u_t + f(u)_x = 0 qquad tin mathbbR^+ xinOmega","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"with the physical flux f.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We split the domain Omega into elements K with center x_K and size Delta x. With the transformation mapping x(xi)=x_K + fracDelta x2 xi we can transform the reference element -11 to every physical element. So, the equation can be restricted to the reference element using the determinant of the Jacobian matrix of the transformation mapping J=fracpartial xpartial xi=fracDelta x2.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"J u_t + f(u)_xi = 0 qquad tin mathbbR^+ xiin -11","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#The-weak-form-of-the-DGSEM","page":"2 DGSEM with flux differencing","title":"The weak form of the DGSEM","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We consider the so-called discontinuous Galerkin spectral element method (DGSEM) with collocation. It results from choosing a nodal DG ansatz using N+1 Gauss-Lobatto nodes xi_i in -11 with matching interpolation weights w_i, which are used for numerical integration and interpolation with the Lagrange polynomial basis l_i of degree N. The Lagrange functions are created with those nodes and hence fulfil a Kronecker property at the GL nodes. The weak formulation of the DGSEM for one element is","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"J underlinedotu(t) = - M^-1 B underlinef^* + M^-1 D^T M underlinef","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"where underlineu=(u_0 u_1 dots u_N)^TinmathbbR^N+1 is the collected pointwise evaluation of u at the discretization nodes and dotu = partial u partial t = u_t is the temporal derivative. The nodal values of the flux function f results with collocation in underlinef, since underlinef_j=f(underlineu_j). Moreover, we got the numerical flux f^*=f^*(u^- u^+).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We will now have a short overview over the operators we used.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The derivative matrix DinmathbbR^(N+1)times (N+1) mimics a spatial derivation on a discrete level with underlinef_x approx D underlinef. It is defined by D_ij = l_j(xi_i).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The diagonal mass matrix M is defined by M_ij=langle l_j l_irangle_N with the numerical scalar product langle cdot cdotrangle_N defined for functions f and g by","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"langle f grangle_N = int_-1 N^1 f(xi) g(xi) dxi = sum_k=0^N f(xi_k) g(xi_k) w_k","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The multiplication by M matches a discrete integration","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":" int_-1^1 f(xi) underlinel(xi) dxi approx M underlinef","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The boundary matrix B=textdiag(-1 0 0 1) represents an evaluation of a function at the boundaries xi_0=-1 and xi_N=1.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"For these operators the following property holds:","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":" M D + (M D)^T = B","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This is called the summation-by-parts (SBP) property since it mimics integration by parts on a discrete level (Gassner (2013)).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The explicit definitions of the operators and the construction of the 1D algorithm can be found for instance in the tutorial introduction to DG methods or in more detail in Kopriva (2009).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This property shows the equivalence between the weak form and the following strong formulation of the DGSEM.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"beginalign*\nJ underlinedotu(t)\n= - M^-1 B underlinef^* + M^-1 D^T M underlinef5pt\n= - M^-1 B underlinef^* + M^-1 (B - MD) underlinef5pt\n= - M^-1 B (underlinef^* - underlinef) - D underlinef\nendalign*","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"More information about the equivalence you can find in Kopriva, Gassner (2010).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#DGSEM-with-flux-differencing","page":"2 DGSEM with flux differencing","title":"DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"When using the diagonal SBP property it is possible to rewrite the application of the derivative operator D in the calculation of the volume integral into a subcell based finite volume type differencing formulation (Fisher, Carpenter (2013)). Generalizing","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"(D underlinef)_i = sum_j D_ij underlinef_j\n= 2sum_j frac12 D_ij (underlinef_j + underlinef_i)\neqqcolon 2sum_j D_ij f_textcentral(u_i u_j)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"we replace D underlinef in the strong form by 2D underlinef_vol(u^- u^+) with the consistent two-point volume flux f_vol and receive the DGSEM formulation with flux differencing (split form DGSEM) (Gassner, Winters, Kopriva (2016)).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"beginalign*\nJ underlinedotu(t) = - M^-1 B (underlinef^* - underlinef) - 2D underlinef_vol(u^- u^+)5pt\n= - M^-1 B (underlinef^* - underlinef_vol(underlineu underlineu)) - 2D underlinef_vol(u^- u^+)5pt\n= - M^-1 B underlinef_surface^* - (2D - M^-1 B) underlinef_vol5pt\n= - M^-1 B underlinef_surface^* - D_split underlinef_vol\nendalign*","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This formulation is in a weak form type formulation and can be implemented by using the derivative split matrix D_split=(2D-M^-1B) and two different fluxes. We divide between the surface flux f=f_surface used for the numerical flux f_surface^* and the already mentioned volume flux f_vol especially for this formulation.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This formulation creates a more stable version of DGSEM, because it fulfils entropy stability. Moreover it allows the construction of entropy conserving discretizations without relying on exact integration. This is achieved when using a two-point entropy conserving flux function as volume flux in the volume flux differencing formulation. Then, the numerical surface flux can be used to control the dissipation of the discretization and to guarantee decreasing entropy, i.e. entropy stability.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#fluxDiffExample","page":"2 DGSEM with flux differencing","title":"Implementation in Trixi.jl","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"Now, we have a look at the implementation of DGSEM with flux differencing with Trixi.jl.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"using OrdinaryDiffEq, Trixi","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We implement a simulation for the compressible Euler equations in 2D","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"partial_t beginpmatrix rho rho v_1 rho v_2 rho e endpmatrix\n+ partial_x beginpmatrix rho v_1 rho v_1^2 + p rho v_1 v_2 (rho e +p) v_1 endpmatrix\n+ partial_y beginpmatrix rho v_2 rho v_1 v_2 rho v_2^2 + p (rho e +p) v_2 endpmatrix\n= beginpmatrix 0 0 0 0 endpmatrix","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"for an ideal gas with ratio of specific heats gamma=14. Here, rho is the density, v_1, v_2 the velocities, e the specific total energy and","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"p = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"the pressure.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"gamma = 1.4\nequations = CompressibleEulerEquations2D(gamma)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"As our initial condition we will use a weak blast wave from Hennemann, Gassner (2020). The primitive variables are defined by","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"beginpmatrix rho v_1 v_2 p endpmatrix\n= beginpmatrix 10 00 00 10 endpmatrix textif x_2 05\ntextand beginpmatrix rho v_1 v_2 p endpmatrix\n= beginpmatrix 11691 01882 * cos(phi) 01882 * sin(phi) 1245 endpmatrix textelse","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"with phi = tan^-1(fracx_2x_1).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This initial condition is implemented in Trixi.jl under the name initial_condition_weak_blast_wave.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"initial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"In Trixi.jl, flux differencing for the volume integral can be implemented with VolumeIntegralFluxDifferencing using symmetric two-point volume fluxes. First, we set up a simulation with the entropy conserving and kinetic energy preserving flux flux_ranocha by Hendrik Ranocha (2018) as surface and volume flux.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We will confirm the entropy conservation property numerically.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"volume_flux = flux_ranocha # = f_vol\nsolver = DGSEM(polydeg=3, surface_flux=volume_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"Now, we implement Trixi.jl's mesh, semi and ode in a simple framework. For more information please have a look at the documentation, the basic tutorial introduction to DG methods or some basic elixirs.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"coordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=5,\n n_cells_max=10_000,\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_condition_periodic)\n\n# ODE solvers\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"To analyse the entropy conservation of the approximation, we will use the analysis calllback implemented in Trixi. It provides some information about the approximation including the entropy change.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"analysis_callback = AnalysisCallback(semi, interval=100);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We now run the simulation using flux_ranocha for both surface and volume flux.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=analysis_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"A look at the change in entropy sum partial Spartial U cdot U_t in the analysis callback confirms that the flux is entropy conserving since the change is about machine precision.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We can plot the approximated solution at the time t=0.4.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"Now, we can use for instance the dissipative flux flux_lax_friedrichs as surface flux to get an entropy stable method.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"using OrdinaryDiffEq, Trixi\n\ngamma = 1.4\nequations = CompressibleEulerEquations2D(gamma)\n\ninitial_condition = initial_condition_weak_blast_wave\n\nvolume_flux = flux_ranocha # = f_vol\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\ncoordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=5,\n n_cells_max=10_000,\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_condition_periodic)\n\n# ODE solvers\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan);\n\nanalysis_callback = AnalysisCallback(semi, interval=100);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"We now run the simulation using the volume flux flux_ranocha and surface flux flux_lax_friedrichs.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=analysis_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"The change in entropy confirms the expected entropy stability.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"Of course, you can use more than these two fluxes in Trixi. Here, we will give a short list of possible fluxes for the compressible Euler equations. For the volume flux Trixi.jl provides for example flux_ranocha, flux_shima_etal, flux_chandrashekar, flux_kennedy_gruber. As surface flux you can use all volume fluxes and additionally for instance flux_lax_friedrichs, flux_hll, flux_hllc.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#Package-versions","page":"2 DGSEM with flux differencing","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"2 DGSEM with flux differencing","title":"2 DGSEM with flux differencing","text":"This page was generated using Literate.jl.","category":"page"},{"location":"meshes/dgmulti_mesh/#DGMulti","page":"DGMulti mesh","title":"Unstructured meshes and the DGMulti solver","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Trixi.jl includes support for simplicial and tensor product meshes via the DGMulti solver type, which is based on the StartUpDG.jl package. DGMulti solvers also provide support for quadrilateral and hexahedral meshes, though this feature is currently restricted to Cartesian grids. On these line/quad/hex meshes, the DGMulti solver also allows to use all (finite domain) SBP derivative operators provided by SummationByPartsOperators.jl, including several finite difference SBP methods.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"We make a few simplifying assumptions about supported meshes:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"meshes consist of a single type of element\nmeshes are conforming (e.g., each face of an element is shared with at most one other element).\nthe geometric mapping from reference to physical elements is polynomial (currently, only affine mappings are supported).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"StartUpDG.jl includes both simple uniform meshes via uniform_mesh, as well as support for triangular meshes constructed using Triangulate.jl, a wrapper around Jonathan Shewchuk's Triangle package.","category":"page"},{"location":"meshes/dgmulti_mesh/#The-DGMulti-solver-type","page":"DGMulti mesh","title":"The DGMulti solver type","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Trixi.jl solvers on simplicial meshes use the DGMulti solver type, which allows users to specify element_type and approximation_type in addition to polydeg, surface_flux, surface_integral, and volume_integral.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"DGMulti(; polydeg::Integer,\n element_type::AbstractElemShape,\n approximation_type=Polynomial(),\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n RefElemData_kwargs...)","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Here, element_type can be Tri(), Quad(), Tet(), or Hex(), and approximation_type can be","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Polynomial(), which specifies a DG discretization using a polynomial basis using quadrature rules which are exact for degree 2 * polydeg integrands, or\nSBP(), which specifies a DG discretization using multi-dimensional SBP operators. Types of SBP discretizations available include: SBP{Kubatko{LobattoFaceNodes}}() (the default choice), SBP{Kubatko{LegendreFaceNodes}}(), and SBP{Hicken}(). For polydeg = 1, ..., 4, the SBP{Kubatko{LegendreFaceNodes}}() SBP nodes are identical to the SBP nodes of Chen and Shu. More detailed descriptions of each SBP node set can be found in the StartUpDG.jl docs. Trixi.jl will also specialize certain parts of the solver based on the SBP approximation type.\na (periodic or non-periodic) derivative operator from SummationByPartsOperators.jl, usually constructed as D = derivative_operator(...). In this case, you do not need to pass a polydeg. Periodic derivative operators will only work with single-element meshes constructed using DGMultiMesh.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Additional options can also be specified through RefElemData_kwargs:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"quad_rule_vol = quad_nodes(Tri(), Nq) will substitute in a volume quadrature rule of degree Nq instead of the default (which is a quadrature rule of degree polydeg). Here, a degree Nq rule will be exact for at least degree 2*Nq integrands (such that the mass matrix is integrated exactly). Quadrature rules of which exactly integrate degree Nq integrands may also be specified (for example, quad_rule_vol = StartUpDG.quad_nodes_tri(Nq) on triangles).\nquad_rule_face = quad_nodes(Line(), Nq)) will use a face quadrature rule of degree Nq rather than the default. This rule is also exact for at least degree 2*Nq integrands.","category":"page"},{"location":"meshes/dgmulti_mesh/#The-GaussSBP()-approximation-type-on-Quad()-and-Hex()-meshes","page":"DGMulti mesh","title":"The GaussSBP() approximation type on Quad() and Hex() meshes","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"When using VolumeIntegralFluxDifferencing on Quad() and Hex() meshes, one can also specify approximation_type = GaussSBP() to use an entropy stable Gauss collocation scheme. Here, GaussSBP() refers to \"generalized\" summation-by-parts operators (see for example Ranocha 2018 or Fernandez and Zingg 2015).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Unlike traditional SBP operators, generalized SBP operators are constructed from nodes which do not include boundary nodes (i.e., Gauss quadrature nodes as opposed to Gauss-Lobatto quadrature nodes). This makes the computation of interface fluxes slightly more expensive, but also usually results in a more accurate solution. Roughly speaking, an entropy stable Gauss collocation scheme will yield results similar to a modal entropy stable scheme using a Polynomial() approximation type, but will be more efficient at high orders of approximation.","category":"page"},{"location":"meshes/dgmulti_mesh/#Trixi.jl-elixirs-on-simplicial-and-tensor-product-element-meshes","page":"DGMulti mesh","title":"Trixi.jl elixirs on simplicial and tensor product element meshes","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Example elixirs with triangular, quadrilateral, and tetrahedral meshes can be found in the examples/dgmulti_2d/ and examples/dgmulti_3d/ folders. Some key elixirs to look at:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"examples/dgmulti_2d/elixir_euler_weakform.jl: basic weak form DG discretization on a uniform triangular mesh. Changing element_type = Quad() or approximation_type = SBP() will switch to a quadrilateral mesh or an SBP-type discretization. Changing surface_integral = SurfaceIntegralWeakForm(flux_ec) and volume_integral = VolumeIntegralFluxDifferencing(flux_ec) for some entropy conservative flux (e.g., flux_chandrashekar or flux_ranocha) will switch to an entropy conservative formulation.\nexamples/dgmulti_2d/elixir_euler_triangulate_pkg_mesh.jl: uses an unstructured mesh generated by Triangulate.jl.\nexamples/dgmulti_3d/elixir_euler_weakform.jl: ´basic weak form DG discretization on a uniform tetrahedral mesh. Changing element_type = Hex() will switch to a hexahedral mesh. Changing surface_integral = SurfaceIntegralWeakForm(flux_ec) and volume_integral = VolumeIntegralFluxDifferencing(flux_ec) for some entropy conservative flux (e.g., flux_chandrashekar or flux_ranocha) will switch to an entropy conservative formulation.","category":"page"},{"location":"meshes/dgmulti_mesh/#For-developers","page":"DGMulti mesh","title":"For developers","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/#DGMultiMesh-wrapper-type","page":"DGMulti mesh","title":"DGMultiMesh wrapper type","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"DGMulti meshes in Trixi.jl are represented using a DGMultiMesh{NDIMS, ...} type. This mesh type is assumed to have fields md::MeshData, which contains geometric terms derived from the mapping between the reference and physical elements, and boundary_faces, which contains a Dict of boundary segment names (symbols) and list of faces which lie on that boundary segment.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"A DGMultiMesh can be constructed in several ways. For example, DGMultiMesh(dg::DGMulti) will return a Cartesian mesh on -1 1^d with element types specified by dg. DGMulti meshes can also be constructed by specifying a list of vertex coordinates vertex_coordinates_x, vertex_coordinates_y, vertex_coordinates_z and a connectivity matrix EToV where EToV[e,:] gives the vertices which correspond to element e. These quantities are available from most unstructured mesh generators.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Initial support for curved DGMultiMeshes is available for flux differencing solvers using SBP and GaussSBP approximation types on quadrilateral and hexahedral meshes. These can be called by specifying mesh = DGMultiMesh(dg, cells_per_dimension, mapping), where mapping is a function which specifies the warping of the mesh (e.g., mapping(xi, eta) = SVector{2}(xi, eta) is the identity mapping) similar to the mapping argument used by StructuredMesh.","category":"page"},{"location":"meshes/dgmulti_mesh/#Variable-naming-conventions","page":"DGMulti mesh","title":"Variable naming conventions","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"We use the convention that coordinates on the reference element are r in 1D, r s in 2D, or r s t in 3D. Physical coordinates use the standard conventions x (1D), x y (2D), and x y z (3D).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"(Image: \"Ref-to-physical mapping\")","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Derivatives of reference coordinates with respect to physical coordinates are abbreviated, e.g., fracpartial rpartial x = r_x. Additionally, J is used to denote the determinant of the Jacobian of the reference-to-physical mapping.","category":"page"},{"location":"meshes/dgmulti_mesh/#Variable-meanings-and-conventions-in-StartUpDG.jl","page":"DGMulti mesh","title":"Variable meanings and conventions in StartUpDG.jl","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"StartUpDG.jl exports structs RefElemData{NDIMS, ElemShape, ...} (which contains data associated with the reference element, such as interpolation points, quadrature rules, face nodes, normals, and differentiation/interpolation/projection matrices) and MeshData{NDIMS} (which contains geometric data associated with a mesh). These are currently used for evaluating DG formulations in a matrix-free fashion. These structs contain fields similar (but not identical) to those in Globals1D, Globals2D, Globals3D in the Matlab codes from \"Nodal Discontinuous Galerkin Methods\" by Hesthaven and Warburton (2007).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"In general, we use the following code conventions:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"variables such as r, s,... and x, y,... correspond to values at nodal interpolation points.\nvariables ending in q (e.g., rq, sq,... and xq, yq,...) correspond to values at volume quadrature points.\nvariables ending in f (e.g., rf, sf,... and xf, yf,...) correspond to values at face quadrature points.\nvariables ending in p (e.g., rp, sp,...) correspond to \"plotting\" points, which are usually a fine grid of equispaced points.\nV matrices correspond to interpolation matrices from nodal interpolation points, e.g., Vq interpolates to volume quadrature points, Vf interpolates to face quadrature points.\ngeometric quantities in MeshData are stored as matrices of dimension textnumber of points per element times textnumber of elements.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Quantities in rd::RefElemData:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"rd.Np, rd.Nq, rd.Nf: the number of nodal interpolation points, volume quadrature points, and face quadrature points on the reference element, respectively.\nrd.Vq: interpolation matrices from values at nodal interpolation points to volume quadrature points\nrd.wq: volume quadrature weights on the reference element\nrd.Vf: interpolation matrices from values at nodal interpolation points to face quadrature points\nrd.wf: a vector containing face quadrature weights on the reference element\nrd.M: the quadrature-based mass matrix, computed via rd.Vq' * diagm(rd.wq) * rd.Vq.\nrd.Pq: a quadrature-based L^2 projection matrix rd.Pq = rd.M \\ rd.Vq' * diagm(rd.wq) which maps between values at quadrature points and values at nodal points.\nDr, Ds, Dt matrices are nodal differentiation matrices with respect to the rst coordinates, e.g., Dr*f.(r,s) approximates the derivative of f(rs) at nodal points.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Quantities in md::MeshData:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"md.xyz is a tuple of matrices md.x, md.y, md.z, where column e contains coordinates of physical interpolation points.\nmd.xyzq is a tuple of matrices md.xq, md.yq, md.zq, where column e contains coordinates of physical quadrature points.\nmd.rxJ, md.sxJ, ... are matrices where column e contains values of Jfracpartial rpartial x, Jfracpartial spartial x, etc. at nodal interpolation points on the element e.\nmd.J is a matrix where column e contains values of the Jacobian J at nodal interpolation points.\nmd.Jf is a matrix where column e contains values of the face Jacobian (e.g., determinant of the geometric mapping between a physical face and a reference face) at face quadrature points.\nmd.nxJ, md.nyJ, ... are matrices where column e contains values of components of the unit normal scaled by the face Jacobian md.Jf at face quadrature points.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"For more details, please see the StartUpDG.jl docs.","category":"page"},{"location":"meshes/p4est_mesh/#P4est-based-mesh","page":"P4est-based mesh","title":"P4est-based mesh","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The P4estMesh is an unstructured, curvilinear, nonconforming mesh type for quadrilateral (2D) and hexahedral (3D) cells. It supports quadtree/octree-based adaptive mesh refinement (AMR) via the C library p4est. See AMRCallback for further information.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"meshes/p4est_mesh/#Construction-of-a-P4estMesh-from-an-Abaqus-file","page":"P4est-based mesh","title":"Construction of a P4estMesh from an Abaqus file","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"One available option to construct a P4estMesh is to read in an Abaqus (.inp) mesh file. We briefly describe the structure of this file, the conventions it uses, and how the mesh file is parsed to create an initial unstructured, curvilinear, and conforming mesh.","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-in-two-spatial-dimensions","page":"P4est-based mesh","title":"Mesh in two spatial dimensions","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For this discussion we use the following two-dimensional unstructured curved mesh with three elements:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"(Image: abaqus-2dmesh-docs)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"We note that the corner and element connectivity information parsed from the Abaqus file creates a straight sided (linear) mesh. From this linear mesh there are two strategies available to make the mesh curvilinear:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Apply a mapping function to describe a transformation of the linear mesh to another physical domain. The mapping is approximated using interpolation polynomial of a user specified polynomial degree. The default value of this polynomial degree is 1 that corresponds to an uncurved geometry.\nHigh-order boundary information is available in the .inp mesh file because it was created with the HOHQMesh mesh generator, which is available via the Julia package HOHQMesh.jl. This information is used to create appropriate transfinite mappings during the mesh construction.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"We divide our discussion into two parts. The first part discusses the standard corner and element information contained in the .inp mesh file. The second part specifically deals with the mesh file parsing of an Abaqus file created by HOHQMesh.jl.","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-file-header","page":"P4est-based mesh","title":"Mesh file header","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"An Abaqus .inp mesh file typically begins with a *Heading. Though optional, the *Heading is helpful to give users some information about the mesh described by the mesh file. In particular, a .inp mesh file created with HOHQMesh will contain the header","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*Heading\n File created by HOHQMesh","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"This heading is used to indicate to the mesh constructor which of the above mapping strategies to apply in order to create a curvilinear mesh. If the Abaqus file header is not present then the P4estMesh is created with the first strategy above.","category":"page"},{"location":"meshes/p4est_mesh/#List-of-corner-nodes","page":"P4est-based mesh","title":"List of corner nodes","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next, prefaced with *NODE, comes a list of the physical (x,y,z) coordinates of all the corners. The first integer in the list of the corners provides its id number. Thus, for the two-dimensional example mesh this block of corner information is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*NODE\n1, 1.0, -1.0, 0.0\n2, 3.0, 0.0, 0.0\n3, 1.0, 1.0, 0.0\n4, 2.0, 0.0, 0.0\n5, 0.0, 0.0, 0.0\n6, 3.0, 1.0, 0.0\n7, 3.0, -1.0, 0.0","category":"page"},{"location":"meshes/p4est_mesh/#List-of-elements","page":"P4est-based mesh","title":"List of elements","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The element connectivity is given after the list of corners. The header for this information block is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*ELEMENT, type=CPS4, ELSET=Surface1","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The Abaqus element type CPS4 corresponds to a quadrilateral element. Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. The elements connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion; making the element right-handed. This element handedness is indicated using the circular arrow in the figure above. Just as with the corner list, the first integer in the element connectivity list indicates the element id number. Thus, the element connectivity list for the three element example mesh is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*ELEMENT, type=CPS4, ELSET=Surface1\n1, 5, 1, 4, 3\n2, 4, 2, 6, 3\n3, 7, 2, 4, 1","category":"page"},{"location":"meshes/p4est_mesh/#Element-neighbor-connectivity","page":"P4est-based mesh","title":"Element neighbor connectivity","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The construction of the element neighbor ids and identifying physical boundary surfaces is done using functionality directly from the p4est library. For example, the neighbor connectivity is created in the mesh constructor using the wrapper read_inp_p4est function.","category":"page"},{"location":"meshes/p4est_mesh/#HOHQMesh-boundary-information","page":"P4est-based mesh","title":"HOHQMesh boundary information","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"If present, any additional information in the mesh file that was created by HOHQMesh is prefaced with ** to make it an Abaqus comment. This ensures that the read in of the file by a standard Abaqus file parser, as done in the read_inp_p4est function, is done correctly.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The high-order, curved boundary information and labels of the physical boundary created by HOHQMesh is found below the comment line","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** ***** HOHQMesh boundary information ***** **","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next comes the polynomial degree that the mesh will use to represent any curved sides","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** mesh polynomial degree = 8","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The mesh file then, again, provides the element connectivity as well as information for curved surfaces either interior to the domain or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (-y, +x, +y, or -x) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y,z) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y,z) data will be given in the direction of the local coordinate system. Given below is the element curvature information for the example mesh:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** 5 1 4 3\n** 0 0 1 1\n** 1.000000000000000 1.000000000000000 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 2.000000000000000 0.0 0.0\n** 0.0 0.0 0.0\n** 0.035513826946206 0.105291711848750 0.0\n** 0.148591270347399 0.317731556850611 0.0\n** 0.340010713990041 0.452219430075470 0.0\n** 0.575000000000000 0.462500000000000 0.0\n** 0.788022294598950 0.483764065630034 0.0\n** 0.926408729652601 0.644768443149389 0.0\n** 0.986453164464803 0.883724792445746 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 4 2 6 3\n** 0 0 0 1\n** 2.000000000000000 0.0 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 7 2 4 1\n** 0 0 0 0","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The last piece of information provided by HOHQMesh are labels for the different surfaces of an element. These labels are useful to set boundary conditions along physical surfaces. The labels can be short descriptive words up to 32 characters in length. The label --- indicates an internal surface where no boundary condition is required.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"It is important to note that these labels are given in the following order according to the local surface index -x +x -y +y as required by the p4est library.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** Bezier --- Slant ---\n** --- Right --- Top\n** Bottom --- Right ---","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For completeness, we provide the entire Abaqus mesh file for the example mesh in the figure above:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*Heading\n File created by HOHQMesh\n*NODE\n1, 1.0, -1.0, 0.0\n2, 3.0, 0.0, 0.0\n3, 1.0, 1.0, 0.0\n4, 2.0, 0.0, 0.0\n5, 0.0, 0.0, 0.0\n6, 3.0, 1.0, 0.0\n7, 3.0, -1.0, 0.0\n*ELEMENT, type=CPS4, ELSET=Surface1\n1, 5, 1, 4, 3\n2, 4, 2, 6, 3\n3, 7, 2, 4, 1\n** ***** HOHQMesh boundary information ***** **\n** mesh polynomial degree = 8\n** 5 1 4 3\n** 0 0 1 1\n** 1.000000000000000 1.000000000000000 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 2.000000000000000 0.0 0.0\n** 0.0 0.0 0.0\n** 0.035513826946206 0.105291711848750 0.0\n** 0.148591270347399 0.317731556850611 0.0\n** 0.340010713990041 0.452219430075470 0.0\n** 0.575000000000000 0.462500000000000 0.0\n** 0.788022294598950 0.483764065630034 0.0\n** 0.926408729652601 0.644768443149389 0.0\n** 0.986453164464803 0.883724792445746 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 4 2 6 3\n** 0 0 0 1\n** 2.000000000000000 0.0 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 7 2 4 1\n** 0 0 0 0\n** Bezier --- Slant ---\n** --- Right --- Top\n** Bottom --- Right ---","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-in-three-spatial-dimensions","page":"P4est-based mesh","title":"Mesh in three spatial dimensions","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The 3D Abaqus file format with high-order boundary information from HOHQMesh is very similar to the 2D version discussed above. There are only three changes:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The element connectivity would be given in terms of the eight corners that define a hexahedron. The corners are numbered as shown in the figure below. The header of the element list changes to be\n*ELEMENT, type=C3D8, ELSET=Volume1\nwhere C3D8 corresponds to a Abaqus hexahedral element.\nThere are six check digits included directly below the eight corner indexes to indicate whether the local face within the element is straight sided, 0, or is curved, 1. For curved faces (x,y,z) coordinate values are available in order to construct an face interpolant with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes.\nThe boundary labels are given in the following order according to the local surface index -x +x -y +y -z +z as required by the p4est library.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For completeness, we also give a short description and derivation of the three-dimensional transfinite mapping formulas used to compute the physical coordinates mathbfx=(xyz) of a (possibly curved) hexahedral element give the reference coordinates boldsymbolxi = (xi eta zeta) which lie in -11^3. That is, we will create an expression mathbfx= mathbfX(boldsymbolxi).","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Below we provide a sketch of a single hexahedral element with curved faces. This is done to introduce the numbering conventions for corners, edges, and faces of the element.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"(Image: abaqus-3dmesh-docs)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"When the hexahedron is a straight sided (linear) element we compute the transfinite mapping directly from the element corner points according to","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\nmathbfX_linear(boldsymbolxi) = frac18quad mathbfx_1(1-xi)(1-eta)(1-zeta)\n + mathbfx_2(1+xi)(1-eta)(1-zeta)-015cm\n qquad + mathbfx_3(1+xi)(1+eta)(1-zeta)\n + mathbfx_4(1-xi)(1+eta)(1-zeta) \n qquad + mathbfx_5(1-xi)(1-eta)(1+zeta)\n + mathbfx_6(1+xi)(1-eta)(1+zeta) \n qquad + mathbfx_7(1+xi)(1+eta)(1+zeta)\n + mathbfx_8(1-xi)(1+eta)(1+zeta)quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next, we create a transfinite mapping function, mathbfX(boldsymbolxi), for a hexahedron that has one or more curved faces. For this we assume that have a set of six interpolating polynomials Gamma_i_i=1^6 that approximate the faces. The interpolating polynomial for any curved faces is provided by the information in a HOHQMesh Abaqus mesh file or is constructed on the fly via a bi-linear interpolation routine for any linear faces. Explicitly, these six face interpolation polynomials depend on the computational coordinates boldsymbolxi as follows","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":" beginaligned\n Gamma_1(xi zeta) quad quad Gamma_3(xi eta) quad quad Gamma_4(eta zeta)01cm\n Gamma_2(xi zeta) quad quad Gamma_5(xi eta) quad quad Gamma_6(eta zeta)\n endaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"To determine the form of the mapping we first create linear interpolations between two opposing faces, e.g., Gamma_3 and Gamma_5 and sum them together to have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\n boldsymbolSigma(boldsymbolxi) = frac12quad(1-xi)Gamma_6(etazeta) + (1+xi)Gamma_4(etazeta) -015cm\n qquad+ (1-eta)Gamma_1(xizeta) + (1+eta)Gamma_2(xizeta) -015cm\n qquad +(1-zeta)Gamma_3(xieta) + (1+zeta)Gamma_5(xieta)quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Unfortunately, the linear interpolations boldsymbolSigma(boldsymbolxi) no longer match at the faces, e.g., evaluating at eta = -1 we have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"boldsymbolSigma(xi-1zeta) = Gamma_1(xizeta) + frac12(1-xi)Gamma_6(-1zeta) + (1+xi)Gamma_4(-1zeta)\n +(1-zeta)Gamma_3(xi-1) + (1+zeta)Gamma_5(xi-1)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"which is the desired face Gamma_1(xizeta) plus four edge error terms. Analogous edge error terms occur at the other faces evaluating boldsymbolSigma(boldsymbolxi) at eta=1, xi=pm 1, and zeta=pm 1. In order to match the faces, we subtract a linear interpolant in the xi, eta, and zeta directions of the edge error terms, e.g., the terms in braces in the above equation. So, continuing the example above, the correction term to be subtracted for face Gamma_1 to match would be","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"left(frac1-eta2right) bigg frac12 (1-xi)Gamma_6(-1zeta) + (1+xi)Gamma_4(-1zeta)+(1-zeta)Gamma_3(xi-1)\n + (1+zeta)Gamma_5(xi-1) bigg","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For clarity, and to allow an easier comparison to the implementation, we introduce auxiliary notation for the 12 edge values present in the complete correction term. That is, for given values of xi, eta, and zeta we have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":" beginaligned\n textttedge_1 = Gamma_1(xi -1) quad quad textttedge_5 = Gamma_2(xi -1) quad quad textttedge_9 = Gamma_6(eta -1)01cm\n textttedge_2 = Gamma_1(1 zeta) quad quadtextttedge_6 = Gamma_2(1 zeta) quad quad textttedge_10 = Gamma_4(eta -1)01cm\n textttedge_3 = Gamma_1(xi 1) quad quad textttedge_7 = Gamma_2(xi 1) quad quad textttedge_11 = Gamma_4(eta 1)01cm\n textttedge_4 = Gamma_1(-1 zeta) quad quad textttedge_8 = Gamma_2(-1 zeta) quad quad textttedge_12 = Gamma_6(eta 1)\n endaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"With this notation for the edge terms (and after some algebraic manipulation) we write the complete edge correction term, mathcalC_textttedge(boldsymbolxi), as","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\nmathcalC_textttedge(boldsymbolxi) = frac14quad (1-eta)(1-zeta)textttedge_1-015cm\n qquad + (1+xi)(1-eta)textttedge_2 \n qquad + (1-eta)(1+zeta)textttedge_3 \n qquad + (1-xi)(1-eta)textttedge_4 \n qquad + (1+eta)(1-zeta)textttedge_5 \n qquad + (1+xi)(1+eta)textttedge_6 \n qquad + (1+eta)(1+zeta)textttedge_7 \n qquad + (1-xi)(1+eta)textttedge_8 \n qquad + (1-xi)(1-zeta)textttedge_9 \n qquad + (1+xi)(1-zeta)textttedge_10 \n qquad + (1+xi)(1+zeta)textttedge_11 \n qquad + (1-xi)(1+zeta)textttedge_12quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"However, subtracting the edge correction terms mathcalC_textttedge(boldsymbolxi) from boldsymbolSigma(boldsymbolxi) removes the interior element contributions twice. Thus, to complete the construction of the transfinite mapping mathbfX(boldsymbolxi) we add the transfinite map of the straight sided hexahedral element to find","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"mathbfX(boldsymbolxi) = boldsymbolSigma(boldsymbolxi)\n - mathcalC_textttedge(boldsymbolxi)\n + mathbfX_linear(boldsymbolxi)","category":"page"},{"location":"troubleshooting/#Troubleshooting-and-FAQ","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In general, Trixi.jl works best with the newest Julia release and up-to-date dependencies. If something does not work as expected, try updating your installed Julia packages via the package manager, e.g., by running","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> import Pkg; Pkg.update()","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"If you do not use the latest stable release of Julia from the official website, consider updating your Julia installation.","category":"page"},{"location":"troubleshooting/#old-release","page":"Troubleshooting and FAQ","title":"Installing Trixi.jl as a package only provides an older release","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Trixi.jl requires fairly recent versions of several of its dependencies, which sometimes causes issues when other installed packages have conflicting version requirements. In this case, Julia's package manager Pkg will try to handle this gracefully by going back in history until it finds a Trixi.jl release whose version requirements can be met, resulting in an older - and usually outdated - version of Trixi.jl being installed.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The following example illustrates this issue:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The current Trixi.jl release v0.3.6 requires package Foo with a minimum version of v0.2.\nAn older Trixi.jl release v0.2.1 requires package Foo only with a minimum version of v0.1.\nA user has already installed package Bar, which itself requires Foo with a maximum version of v0.1.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In this case, installing Trixi.jl via Pkg will result in version v0.2.1 to be installed instead of the current release v0.3.6. That is, a specific release of Trixi.jl may not be installable if it has a dependency with a higher minimum version that at the same time is restricted to a lower maximum version by another installed package.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"You can check whether an outdated version of Trixi.jl is installed by executing","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> import Pkg; Pkg.update(\"Trixi\"); Pkg.status(\"Trixi\")","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"in the REPL and comparing the reported Trixi.jl version with the version of the latest release. If the versions differ, you can confirm that it is due to a version conflict by forcing Pkg to install the latest Trixi.jl release, where version is the current release:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> Pkg.add(name=\"Trixi\", version=\"0.3.6\")","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In case of a conflict, the command above will produce an error that informs you about the offending packages, similar to the following:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":" Updating registry at `~/.julia/registries/General`\n Resolving package versions...\nERROR: Unsatisfiable requirements detected for package DataStructures [864edb3b]:\n DataStructures [864edb3b] log:\n ├─possible versions are: [0.9.0, 0.10.0, 0.11.0-0.11.1, 0.12.0, 0.13.0, 0.14.0-0.14.1, 0.15.0, 0.16.1, 0.17.0-0.17.20, 0.18.0-0.18.8] or uninstalled\n ├─restricted by compatibility requirements with DiffEqCallbacks [459566f4] to versions: 0.18.0-0.18.8\n │ └─DiffEqCallbacks [459566f4] log:\n │ ├─possible versions are: [2.0.0, 2.1.0, 2.2.0, 2.3.0, 2.4.0, 2.5.0-2.5.2, 2.6.0, 2.7.0, 2.8.0, 2.9.0, 2.10.0, 2.11.0, 2.12.0-2.12.1, 2.13.0-2.13.5, 2.14.0-2.14.1, 2.15.0] or uninstalled\n │ ├─restricted by compatibility requirements with Trixi [a7f1ee26] to versions: [2.14.0-2.14.1, 2.15.0]\n │ │ └─Trixi [a7f1ee26] log:\n │ │ ├─possible versions are: [0.1.0-0.1.2, 0.2.0-0.2.6, 0.3.0-0.3.6] or uninstalled\n │ │ └─restricted to versions 0.3.6 by an explicit requirement, leaving only versions 0.3.6\n │ └─restricted by compatibility requirements with StaticArrays [90137ffa] to versions: 2.15.0 or uninstalled, leaving only versions: 2.15.0\n │ └─StaticArrays [90137ffa] log:\n │ ├─possible versions are: [0.8.0-0.8.3, 0.9.0-0.9.2, 0.10.0, 0.10.2-0.10.3, 0.11.0-0.11.1, 0.12.0-0.12.5, 1.0.0-1.0.1] or uninstalled\n │ └─restricted by compatibility requirements with Trixi [a7f1ee26] to versions: 1.0.0-1.0.1\n │ └─Trixi [a7f1ee26] log: see above\n └─restricted by compatibility requirements with JLD2 [033835bb] to versions: [0.9.0, 0.10.0, 0.11.0-0.11.1, 0.12.0, 0.13.0, 0.14.0-0.14.1, 0.15.0, 0.16.1, 0.17.0-0.17.20] — no versions left\n └─JLD2 [033835bb] log:\n ├─possible versions are: [0.1.0-0.1.14, 0.2.0-0.2.4, 0.3.0-0.3.1] or uninstalled\n └─restricted by compatibility requirements with BinaryBuilder [12aac903] to versions: 0.1.0-0.1.14\n └─BinaryBuilder [12aac903] log:\n ├─possible versions are: [0.1.0-0.1.2, 0.1.4, 0.2.0-0.2.6] or uninstalled\n └─restricted to versions * by an explicit requirement, leaving only versions [0.1.0-0.1.2, 0.1.4, 0.2.0-0.2.6]","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"From the error message, we can see that ultimately BinaryBuilder is the problem here: It restricts the package DataStructures to version v0.17 (via its dependency JLD2), while Trixi.jl requires at least v0.18 (via its dependency DiffEqCallbacks). Following the official Pkg documentation, there are a number of things you can try to fix such errors:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Try updating all packages with julia -e 'using Pkg; Pkg.update()'. A newer version of the problematic package may exist that has updated version requirements.\nRemove the offending package. Running\njulia> import Pkg; Pkg.rm(\"BinaryBuilder\"); Pkg.update(); Pkg.status()\nin the REPL will remove BinaryBuilder and (hopefully) update Trixi.jl to the latest version.\nReport the versioning issue to us and/or the development repository of the conflicting package. Maybe it is possible to lift the version restrictions such that both packages can live side by side.\nInstead of installing Trixi.jl and conflicting packages in the same (default) environment, consider creating new environments/projects and install only packages required for the specific tasks, as explained in the official Pkg documentation. For example, if you use Trixi.jl for a research project (Bachelor/Master thesis or a paper), you should create a new Julia project/environment for that research and add Trixi.jl as a dependency. If you track all your code and the Project.toml, Manifest.toml files (generated by Pkg) in a version control system such as git, you can make your research easily reproducible (if you also record the version of Julia you are using and leave some comments for others who do not know what you are trying to do, including your future self 😉).","category":"page"},{"location":"troubleshooting/#font-issues","page":"Troubleshooting and FAQ","title":"There are many questions marks and weird symbols in the output of Trixi.jl","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This probably means that the default font used by your operating system does not support enough Unicode symbols. Try installing a modern font with decent unicode support, e.g. JuliaMono. Detailed installation instructions are available there.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This problems affects users of Mac OS particularly often. At the time of writing, installing JuliaMono is as simple as","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"$ brew tap homebrew/cask-fonts\n$ brew install --cask font-juliamono","category":"page"},{"location":"troubleshooting/#There-are-no-timing-results-of-the-initial-mesh-creation","page":"Troubleshooting and FAQ","title":"There are no timing results of the initial mesh creation","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"By default, the SummaryCallback resets the timer used internally by Trixi.jl when it is initialized (when solve is called). If this step needs to be timed, e.g. to debug performance problems, explicit timings can be used as follows.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"using Trixi\n\nbegin\n Trixi.reset_timer!(Trixi.timer())\n\n equations = LinearScalarAdvectionEquation2D(0.2, -0.7)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), n_cells_max=10^5, initial_refinement_level=5)\n solver = DGSEM(3)\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test, solver)\n\n Trixi.print_timer(Trixi.timer())\nend","category":"page"},{"location":"troubleshooting/#MPI-ranks-are-assigned-zero-cells-in-[P4estMesh](@ref)-even-though-there-are-enough-cells","page":"Troubleshooting and FAQ","title":"MPI ranks are assigned zero cells in P4estMesh even though there are enough cells","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The P4estMesh allows one to coarsen the mesh by default. When Trixi.jl is parallelized with multiple MPI ranks, this has the consequence that sibling cells (i.e., child cells with the same parent cell) are kept on the same MPI rank to be able to coarsen them easily. This might cause an unbalanced distribution of cells on different ranks. For 2D meshes, this also means that initially each rank will at least own 4 cells, and for 3D meshes, initially each rank will at least own 8 cells. See issue #1329.","category":"page"},{"location":"troubleshooting/#Installing-and-updating-everything-takes-a-lot-of-time","page":"Troubleshooting and FAQ","title":"Installing and updating everything takes a lot of time","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Julia compiles code to get good (C/Fortran-like) performance. At the same time, Julia provides a dynamic environment and usually compiles code just before using it. Over time, Julia has improved its caching infrastructure, allowing to store and reuse more results from (pre-)compilation. This often results in an increased time to install/update packages, in particular when updating to Julia v1.8 or v1.9 from older versions.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Some packages used together with Trixi.jl provide options to configure the amount of precompilation. For example, OrdinaryDiffEq.jl precompiles many ODE solvers for a good runtime experience of average users. Currently, Trixi.jl does not use all of the available solvers. Thus, you can save some time at every update by setting their precompilation options.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"At the time of writing, this could look as follows. First, you need to activate the environment where you have installed OrdinaryDiffEq.jl. Then, you need to execute the following Julia code.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"using Preferences, UUIDs\nlet uuid = UUID(\"1dea7af3-3e70-54e6-95c3-0bf5283fa5ed\")\n set_preferences!(uuid, \"PrecompileAutoSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileAutoSwitch\" => false)\n set_preferences!(uuid, \"PrecompileDefaultSpecialize\" => true)\n set_preferences!(uuid, \"PrecompileFunctionWrapperSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileLowStorage\" => true)\n set_preferences!(uuid, \"PrecompileNoSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileNonStiff\" => true)\n set_preferences!(uuid, \"PrecompileStiff\" => false)\nend","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This disables precompilation of all implicit methods. This should usually not affect the runtime latency with Trixi.jl since most setups use explicit time integration methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"EditURL = \"../../literate/src/files/shock_capturing.jl\"","category":"page"},{"location":"tutorials/shock_capturing/#shock_capturing","page":"3 Shock capturing with flux differencing and stage limiter","title":"3: Shock capturing with flux differencing and stage limiter","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"This tutorial contains a short summary of the idea of shock capturing for DGSEM with flux differencing and its implementation in Trixi.jl. In the second part, an implementation of a positivity preserving limiter is added to the simulation.","category":"page"},{"location":"tutorials/shock_capturing/#Shock-capturing-with-flux-differencing","page":"3 Shock capturing with flux differencing and stage limiter","title":"Shock capturing with flux differencing","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"The following rough explanation is on a very basic level. More information about an entropy stable shock-capturing strategy for DGSEM discretizations of advection dominated problems, such as the compressible Euler equations or the compressible Navier-Stokes equations, can be found in Hennemann et al. (2021). In Rueda-Ramírez et al. (2021) you find the extension to the systems with non-conservative terms, such as the compressible magnetohydrodynamics (MHD) equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"The strategy for a shock-capturing method presented by Hennemann et al. is based on a hybrid blending of a high-order DG method with a low-order variant. The low-order subcell finite volume (FV) method is created directly with the Legendre-Gauss-Lobatto (LGL) nodes already used for the high-order DGSEM. Then, the final method is a convex combination with regulating indicator alpha of these two methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Since the surface integral is equal for both the DG and the subcell FV method, only the volume integral divides between the two methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"This strategy for the volume integral is implemented in Trixi.jl under the name of VolumeIntegralShockCapturingHG with the three parameters of the indicator and the volume fluxes for the DG and the subcell FV method.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Note, that the DG method is based on the flux differencing formulation. Hence, you have to use a two-point flux, such as flux_ranocha, flux_shima_etal, flux_chandrashekar or flux_kennedy_gruber, for the DG volume flux. We would recommend to use the entropy conserving flux flux_ranocha by Ranocha (2018) for the compressible Euler equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;\n volume_flux_dg=volume_flux_dg,\n volume_flux_fv=volume_flux_fv)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We now focus on a choice of the shock capturing indicator indicator_sc. A possible indicator is alpha_HG presented by Hennemann et al. (p.10), which depends on the current approximation with modal coefficients m_j_j=0^N of a given variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"The indicator is calculated for every DG element by itself. First, we calculate a smooth alpha by","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"alpha = frac11+exp(-frac-smathbbT(mathbbE-mathbbT))","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"with the total energy mathbbE=maxbig(fracm_N^2sum_j=0^N m_j^2 fracm_N-1^2sum_j=0^N-1 m_j^2big), threshold mathbbT= 05 * 10^-18*(N+1)^14 and parameter s=lnbig(frac1-0000100001big)approx 921024.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"For computational efficiency, alpha_min is introduced and used for","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"tildealpha = begincases\n0 textif alphaalpha_min\nalpha textif alpha_minleq alpha leq 1- alpha_min\n1 textif 1-alpha_minalpha\nendcases","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Moreover, the parameter alpha_max sets a maximal value for alpha by","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"alpha = mintildealpha alpha_max","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"This allows to control the maximal dissipation.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"To remove numerical artifact the final indicator is smoothed with all the neighboring elements' indicators. This is activated with alpha_smooth=true.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"alpha_HG = max_E alpha 05 * alpha_E","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"where E are all elements sharing a face with the current element.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Furthermore, you can specify the variable used for the calculation. For instance you can choose density, pressure or both with density_pressure for the compressible Euler equations. For every equation there is also the option to use the first conservation variable with first.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"This indicator is implemented in Trixi.jl and called IndicatorHennemannGassner with the parameters equations, basis, alpha_max, alpha_min, alpha_smooth and variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"indicator_sc = IndicatorHennemannGassner(equations, basis,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=variable)","category":"page"},{"location":"tutorials/shock_capturing/#Positivity-preserving-limiter","page":"3 Shock capturing with flux differencing and stage limiter","title":"Positivity preserving limiter","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Some numerical solutions are physically meaningless, for instance negative values of pressure or density for the compressible Euler equations. This often results in crashed simulations since the calculation of numerical fluxes or stable time steps uses mathematical operations like roots or logarithms. One option to avoid these cases are a-posteriori positivity preserving limiters. Trixi.jl provides the fully-discrete positivity-preserving limiter of Zhang, Shu (2011).","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"It works the following way. For every passed (scalar) variable and for every DG element we calculate the minimal value value_min. If this value falls below the given threshold varepsilon, the approximation is slightly adapted such that the minimal value of the relevant variable lies now above the threshold.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"underlineu^new = theta * underlineu + (1-theta) * u_mean","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"where underlineu are the collected pointwise evaluation coefficients in element e and u_mean the integral mean of the quantity in e. The new coefficients are a convex combination of these two values with factor","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"theta = fracvalue_mean - varepsilonvalue_mean - value_min","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"where value_mean is the relevant variable evaluated for the mean value u_mean.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"The adapted approximation keeps the exact same mean value, but the relevant variable is now greater or equal the threshold varepsilon at every node in every element.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We specify the variables the way we did before for the shock capturing variables. For the compressible Euler equations density, pressure or the combined variable density_pressure are a reasonable choice.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"You can implement the limiter in Trixi.jl using PositivityPreservingLimiterZhangShu with parameters threshold and variables.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"stage_limiter! = PositivityPreservingLimiterZhangShu(thresholds=thresholds,\n variables=variables)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Then, the limiter is added to the time integration method in the solve function. For instance, like","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"CarpenterKennedy2N54(stage_limiter!, williamson_condition=false)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"or","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"SSPRK43(stage_limiter!).","category":"page"},{"location":"tutorials/shock_capturing/#Simulation-with-shock-capturing-and-positivity-preserving","page":"3 Shock capturing with flux differencing and stage limiter","title":"Simulation with shock capturing and positivity preserving","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Now, we can run a simulation using the described methods of shock capturing and positivity preserving limiters. We want to give an example for the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"using OrdinaryDiffEq, Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"As our initial condition we use the Sedov blast wave setup.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"function initial_condition_sedov_blast_wave(x, t, equations::CompressibleEulerEquations2D)\n # Set up polar coordinates\n inicenter = SVector(0.0, 0.0)\n x_norm = x[1] - inicenter[1]\n y_norm = x[2] - inicenter[2]\n r = sqrt(x_norm^2 + y_norm^2)\n\n r0 = 0.21875 # = 3.5 * smallest dx (for domain length=4 and max-ref=6)\n # r0 = 0.5 # = more reasonable setup\n E = 1.0\n p0_inner = 3 * (equations.gamma - 1) * E / (3 * pi * r0^2)\n p0_outer = 1.0e-5 # = true Sedov setup\n # p0_outer = 1.0e-3 # = more reasonable setup\n\n # Calculate primitive variables\n rho = 1.0\n v1 = 0.0\n v2 = 0.0\n p = r > r0 ? p0_outer : p0_inner\n\n return prim2cons(SVector(rho, v1, v2, p), equations)\nend\ninitial_condition = initial_condition_sedov_blast_wave","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"basis = LobattoLegendreBasis(3)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We set the numerical fluxes and divide between the surface flux and the two volume fluxes for the DG and FV method. Here, we are using flux_lax_friedrichs and flux_ranocha.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"surface_flux = flux_lax_friedrichs\nvolume_flux = flux_ranocha","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Now, we specify the shock capturing indicator alpha.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We implement the described indicator of Hennemann, Gassner as explained above with parameters equations, basis, alpha_max, alpha_min, alpha_smooth and variable. Since density and pressure are the critical variables in this example, we use density_pressure = density * pressure = rho * p as indicator variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"indicator_sc = IndicatorHennemannGassner(equations, basis,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=density_pressure)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"Now, we can use the defined fluxes and the indicator to implement the volume integral using shock capturing.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;\n volume_flux_dg=volume_flux,\n volume_flux_fv=surface_flux)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We finalize the discretization by implementing Trixi.jl's solver, mesh, semi and ode, while solver now has the extra parameter volume_integral.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"solver = DGSEM(basis, surface_flux, volume_integral)\n\ncoordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=6,\n n_cells_max=10_000)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We add some callbacks to get an solution analysis and use a CFL-based time step size calculation.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"analysis_callback = AnalysisCallback(semi, interval=100)\n\nstepsize_callback = StepsizeCallback(cfl=0.8)\n\ncallbacks = CallbackSet(analysis_callback, stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"We now run the simulation using the positivity preserving limiter of Zhang and Shu for the variables density and pressure.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"stage_limiter! = PositivityPreservingLimiterZhangShu(thresholds=(5.0e-6, 5.0e-6),\n variables=(Trixi.density, pressure))\n\nsol = solve(ode, CarpenterKennedy2N54(stage_limiter!, williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/shock_capturing/#Package-versions","page":"3 Shock capturing with flux differencing and stage limiter","title":"Package versions","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"","category":"page"},{"location":"tutorials/shock_capturing/","page":"3 Shock capturing with flux differencing and stage limiter","title":"3 Shock capturing with flux differencing and stage limiter","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"EditURL = \"../../literate/src/files/DGMulti_1.jl\"","category":"page"},{"location":"tutorials/DGMulti_1/#DGMulti_1","page":"5 DG schemes via DGMulti solver","title":"5: DG schemes via DGMulti solver","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"DGMulti is a DG solver that allows meshes with simplex elements. The basic idea and implementation of this solver is explained in section \"Meshes\". Here, we want to give some examples and a quick overview about the options with DGMulti.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"We start with a simple example we already used in the tutorial about flux differencing. There, we implemented a simulation with initial_condition_weak_blast_wave for the 2D compressible Euler equations CompressibleEulerEquations2D and used the DG formulation with flux differencing using volume flux flux_ranocha and surface flux flux_lax_friedrichs.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"Here, we want to implement the equivalent example, only now using the DGMulti solver instead of DGSEM.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\n\nequations = CompressibleEulerEquations2D(1.4)\n\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"To use the Gauss-Lobatto nodes again, we choose approximation_type=SBP() in the solver. Since we want to start with a Cartesian domain discretized with squares, we use the element type Quad().","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Quad(),\n approximation_type = SBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"Run the simulation with the same time integration algorithm as before.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"sol = solve(ode, RDPK3SpFSAL49(), abstol=1.0e-6, reltol=1.0e-6,\n callback=callbacks, save_everystep=false);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"plot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"This simulation is not as fast as the equivalent with TreeMesh since no special optimizations for quads (for instance tensor product structure) have been implemented. Figure 4 in \"Efficient implementation of modern entropy stable and kinetic energy preserving discontinuous Galerkin methods for conservation laws\" (2021) provides a nice runtime comparison between the different mesh types. On the other hand, the functions are more general and thus we have more option we can choose from.","category":"page"},{"location":"tutorials/DGMulti_1/#Simulation-with-Gauss-nodes","page":"5 DG schemes via DGMulti solver","title":"Simulation with Gauss nodes","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"For instance, we can change the approximation type of our simulation.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\nequations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"We now use Gauss nodes instead of Gauss-Lobatto nodes which can be done for the element types Quad() and Hex(). Therefore, we set approximation_type=GaussSBP(). Alternatively, we can use a modal approach using the approximation type Polynomial().","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Quad(),\n approximation_type = GaussSBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/#Simulation-with-triangular-elements","page":"5 DG schemes via DGMulti solver","title":"Simulation with triangular elements","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"Also, we can set another element type. We want to use triangles now.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\nequations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"Since there is no direct equivalent to Gauss-Lobatto nodes on triangles, the approximation type SBP() now uses Gauss-Lobatto nodes on faces and special nodes in the interior of the triangular. More details can be found in the documentation of StartUpDG.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Tri(),\n approximation_type = SBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"plot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/#Triangular-meshes-on-non-Cartesian-domains","page":"5 DG schemes via DGMulti solver","title":"Triangular meshes on non-Cartesian domains","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"To use triangular meshes on a non-Cartesian domain, Trixi.jl uses the package StartUpDG.jl. The following example is based on elixir_euler_triangulate_pkg_mesh.jl and uses a pre-defined mesh from StartUpDG.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"We want to simulate the smooth initial condition initial_condition_convergence_test with source terms source_terms_convergence_test for the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"equations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_convergence_test\nsource_terms = source_terms_convergence_test","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"We create the solver DGMulti with triangular elements (Tri()) as before.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3, element_type = Tri(),\n approximation_type=Polynomial(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"StartUpDG.jl provides for instance a pre-defined Triangulate geometry for a rectangular domain with hole RectangularDomainWithHole. Other pre-defined Triangulate geometries are e.g., SquareDomain, Scramjet, and CircularDomain.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"meshIO = StartUpDG.triangulate_domain(StartUpDG.RectangularDomainWithHole());\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"The pre-defined Triangulate geometry in StartUpDG has integer boundary tags. With DGMultiMesh we assign boundary faces based on these integer boundary tags and create a mesh compatible with Trixi.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"mesh = DGMultiMesh(dg, meshIO, Dict(:outer_boundary=>1, :inner_boundary=>2))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"boundary_condition_convergence_test = BoundaryConditionDirichlet(initial_condition)\nboundary_conditions = (; :outer_boundary => boundary_condition_convergence_test,\n :inner_boundary => boundary_condition_convergence_test)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n source_terms = source_terms,\n boundary_conditions = boundary_conditions)\n\ntspan = (0.0, 0.2)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=20)\nanalysis_callback = AnalysisCallback(semi, interval=200, uEltype=real(dg))\ncallbacks = CallbackSet(alive_callback, analysis_callback);\n\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt = 0.5 * estimate_dt(mesh, dg), save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"For more information, please have a look in the StartUpDG.jl documentation.","category":"page"},{"location":"tutorials/DGMulti_1/#Package-versions","page":"5 DG schemes via DGMulti solver","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"StartUpDG\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"","category":"page"},{"location":"tutorials/DGMulti_1/","page":"5 DG schemes via DGMulti solver","title":"5 DG schemes via DGMulti solver","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"EditURL = \"../../literate/src/files/adding_new_scalar_equations.jl\"","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#adding_new_scalar_equations","page":"8 Adding a new scalar conservation law","title":"8: Adding a new scalar conservation law","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"If you want to use Trixi.jl for your own research, you might be interested in a new physics model that's not already included in Trixi.jl. In this tutorial, we will implement the cubic conservation law","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"partial_t u(tx) + partial_x u(tx)^3 = 0","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"in a periodic domain in one space dimension. In Trixi.jl, such a mathematical model is encoded as a subtype of Trixi.AbstractEquations.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Basic-setup","page":"8 Adding a new scalar conservation law","title":"Basic setup","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"using Trixi\n\nstruct CubicEquation <: Trixi.AbstractEquations{1 #= number of spatial dimensions =#,\n 1 #= number of primary variables, i.e. scalar =#};\nend","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"We create CubicEquation as an empty struct since we do not use any parameters for this equation. Other models could bundle arbitrary parameters, e.g., the ideal gas constant for the compressible Euler equations.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Next, we define the physical flux f(u) = u^3 using the calling structure used in Trixi.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Trixi.flux(u, orientation, equation::CubicEquation) = u.^3\nTrixi.varnames(_, ::CubicEquation) = (\"scalar\",)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"In Trixi.jl, the conserved variables u are usually passed as SVectors of variables at a single physical location. Hence, we must use u.^3 instead of the scalar operation u^3.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"That's already enough to run a simple simulation with a standard DGSEM discretization using the non-dissipative central flux at interfaces.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"using OrdinaryDiffEq\n\n# Create a simulation setup\nequation = CubicEquation()\n\ninitial_condition_sine(x, t, equation::CubicEquation) = SVector(sinpi(x[1]))\n\nmesh = TreeMesh(-1.0, 1.0, # min/max coordinates\n initial_refinement_level=4,\n n_cells_max=10^4)\n\nsolver = DGSEM(3 #= polynomial degree =#, flux_central)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"We wrap the return value of the initial_condition_sine inside an SVector since that's the approach used in Trixi.jl also for systems of equations. We need to index the spatial coordinate x[1], since it is an SVector with one component. In multiple space dimensions, all spatial coordinates are passed together.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Next, we create an ODEProblem from the SciML/DifferentialEquations ecosystem. We can solve this ODE numerically using any time integration method, e.g., SSPRK43 from OrdinaryDiffEq.jl. Before, we set up a callback to summarize the simulation setup.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"# Create ODE problem with given time span\ntspan = (0.0, 0.09)\node = semidiscretize(semi, tspan)\n\nsummary_callback = SummaryCallback()\ncallbacks = CallbackSet(summary_callback)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes the passed callbacks\nsol = solve(ode, SSPRK43();\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"That's it, you ran your first simulation using your new equation with Trixi.jl! Now, we can plot the solution at the final time using Plots.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"You can already see that discontinuities will develop and oscillations start to occur around steep parts of the wave. That's expected from our central discretization. To avoid these issues, we need to use dissipative numerical fluxes (approximate Riemann solvers) at interfaces.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Advanced-setup","page":"8 Adding a new scalar conservation law","title":"Advanced setup","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Thus, we add a Godunov's flux for our cubic equation. That is easy for this equation since the wave speed f'(u) = 3u^2 is always non-negative.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"@inline Trixi.flux_godunov(u_ll, u_rr, orientation, equation::CubicEquation) = flux(u_ll, orientation, equation)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Let's run the example again but with a dissipative numerical flux at interfaces. remake will recreate the semidiscretization we used before and only change selected parameters, in this case the solver.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"# A new setup with dissipation\nsemi = remake(semi, solver=DGSEM(3, flux_godunov))\node = semidiscretize(semi, tspan)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot!(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"You can see that there are fewer oscillations, in particular around steep edges. Now let's increase the final time (and also the spatial resolution).","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"# A larger final time: Nonclassical shocks develop (you can even increase the refinement to 12)\nsemi = remake(semi, mesh=TreeMesh(-1.0, 1.0, initial_refinement_level=8, n_cells_max=10^5))\node = semidiscretize(semi, (0.0, 0.5) #= tspan =#)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"You can observe that nonclassical shocks develop and are stable under grid refinement, e.g. for initial_refinement_level=12. In this case, these nonclassical shocks can be avoided by using an entropy-dissipative semidiscretization. Thus, we need to define an entropy-conservative numerical flux","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"@inline function Trixi.flux_ec(u_ll, u_rr, orientation, equation::CubicEquation)\n return SVector(0.25 * (u_ll[1]^3 + u_ll[1]^2 * u_rr[1] + u_ll[1] * u_rr[1]^2 + u_rr[1]^3))\nend","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"and use a VolumeIntegralFluxDifferencing instead of the standard VolumeIntegralWeakForm in the DGSEM.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"# Let's use a provably entropy-dissipative semidiscretization\nsemi = remake(semi, solver=DGSEM(3, flux_godunov, VolumeIntegralFluxDifferencing(flux_ec)))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...);\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"Possible next steps could be","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"to define Trixi.max_abs_speeds(u, equations::CubicEquation) = 3 * u[1]^2 to use CFL-based time step control via a StepsizeCallback\nto define quantities of interest like Trixi.entropy(u, equations::CubicEquation) = u[1]^2 and integrate them in a simulation using the AnalysisCallback\nto experiment with shock-capturing volume integrals VolumeIntegralShockCapturingHG and adaptive mesh refinement AMRCallback","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"For further reading, Trixi.jl provides another example on adding a scalar equation. In the elixir about the KPP problem, the 2D scalar \"KPP equation\" from Kurganov, Petrova, Popov (2007) is implemented.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Summary-of-the-code","page":"8 Adding a new scalar conservation law","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"To sum up, here is the complete code that we used (without the callbacks since these create a lot of unnecessary output in the doctests of this tutorial). In addition, we create the struct inside the new module CubicConservationLaw. That ensures that we can re-create structs defined therein without having to restart Julia.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"# Define new physics\nmodule CubicConservationLaw\n\nusing Trixi\n\nstruct CubicEquation <: Trixi.AbstractEquations{1 #= number of spatial dimensions =#,\n 1 #= number of primary variables, i.e. scalar =#}\nend\n\n@inline Trixi.flux(u, orientation, equation::CubicEquation) = u.^3\nTrixi.varnames(_, ::CubicEquation) = (\"scalar\",)\n\n@inline Trixi.flux_godunov(u_ll, u_rr, orientation, equation::CubicEquation) = flux(u_ll, orientation, equation)\n@inline function Trixi.flux_ec(u_ll, u_rr, orientation, equation::CubicEquation)\n return SVector(0.25 * (u_ll[1]^3 + u_ll[1]^2 * u_rr[1] + u_ll[1] * u_rr[1]^2 + u_rr[1]^3))\nend\n\nend # module\n\n\n# Create a simulation setup\nimport .CubicConservationLaw\nusing Trixi\nusing OrdinaryDiffEq\nusing Plots\n\nequation = CubicConservationLaw.CubicEquation()\n\ninitial_condition_sine(x, t, equation::CubicConservationLaw.CubicEquation) = SVector(sinpi(x[1]))\n\nmesh = TreeMesh(-1.0, 1.0, # min/max coordinates\n initial_refinement_level=4,\n n_cells_max=10^4)\n\nsolver = DGSEM(3 #= polynomial degree =#, flux_central)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create ODE problem with given time span\ntspan = (0.0, 0.1)\node = semidiscretize(semi, tspan)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes the passed callbacks\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)\n\n\n# A new setup with dissipation\nsemi = remake(semi, solver=DGSEM(3, flux_godunov))\node = semidiscretize(semi, tspan)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot!(sol)\n\n\n# A larger final time: Nonclassical shocks develop (you can even increase the refinement to 12)\nsemi = remake(semi, mesh=TreeMesh(-1.0, 1.0, initial_refinement_level=8, n_cells_max=10^5))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)\n\n\n# Let's use a provably entropy-dissipative semidiscretization\nsemi = remake(semi, solver=DGSEM(3, flux_godunov, VolumeIntegralFluxDifferencing(flux_ec)))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Package-versions","page":"8 Adding a new scalar conservation law","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"8 Adding a new scalar conservation law","title":"8 Adding a new scalar conservation law","text":"This page was generated using Literate.jl.","category":"page"},{"location":"time_integration/#time-integration","page":"Time integration","title":"Time integration methods","text":"","category":"section"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, explicit Runge-Kutta methods from OrdinaryDiffEq.jl are tested extensively. Interesting classes of time integration schemes are","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Explicit low-storage Runge-Kutta methods\nStrong stability preserving methods","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Some common options for solve from OrdinaryDiffEq.jl are the following. Further documentation can be found in the SciML docs.","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"If you use a fixed time step method like CarpenterKennedy2N54, you need to pass a time step as dt=.... If you use a StepsizeCallback, the value passed as dt=... is irrelevant since it will be overwritten by the StepsizeCallback. If you want to use an adaptive time step method such as SSPRK43 or RDPK3SpFSAL49 and still want to use CFL-based step size control via the StepsizeCallback, you need to pass the keyword argument adaptive=false to solve.\nYou should usually set save_everystep=false. Otherwise, OrdinaryDiffEq.jl will (try to) save the numerical solution after every time step in RAM (until you run out of memory or start to swap).\nYou can set the maximal number of time steps via maxiters=....\nSSP methods and many low-storage methods from OrdinaryDiffEq.jl support stage_limiter!s and step_limiter!s, e.g., PositivityPreservingLimiterZhangShu from Trixi.jl.\nIf you start Julia with multiple threads and want to use them also in the time integration method from OrdinaryDiffEq.jl, you need to pass the keyword argument thread=OrdinaryDiffEq.True() to the algorithm, e.g., RDPK3SpFSAL49(thread=OrdinaryDiffEq.True()) or CarpenterKennedy2N54(thread=OrdinaryDiffEq.True(), williamson_condition=false). For more information on using thread-based parallelism in Trixi.jl, please refer to Shared-memory parallelization with threads.\nIf you use error-based step size control (see also the section on error-based adaptive step sizes together with MPI, you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"note: Number of `rhs!` calls\nIf you use explicit Runge-Kutta methods from OrdinaryDiffEq.jl, the total number of rhs! calls can be (slightly) bigger than the number of steps times the number of stages, e.g. to allow for interpolation (dense output), root-finding for continuous callbacks, and error-based time step control. In general, you often should not need to worry about this if you use Trixi.jl.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"EditURL = \"../../literate/src/files/upwind_fdsbp.jl\"","category":"page"},{"location":"tutorials/upwind_fdsbp/#upwind_fdsbp","page":"7 Upwind FD SBP schemes","title":"7: Upwind FD SBP schemes","text":"","category":"section"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"General tensor product SBP methods are supported via the DGMulti solver in a reasonably complete way, see the previous tutorial. Nevertheless, there is also experimental support for SBP methods with other solver and mesh types.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"The first step is to set up an SBP operator. A classical (central) SBP operator can be created as follows.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"using Trixi\nD_SBP = derivative_operator(SummationByPartsOperators.MattssonNordström2004(),\n derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Instead of prefixing the source of coefficients MattssonNordström2004(), you can also load the package SummationByPartsOperators.jl. Either way, this yields an object representing the operator efficiently. If you want to compare it to coefficients presented in the literature, you can convert it to a matrix.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Matrix(D_SBP)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Upwind SBP operators are a concept introduced in 2017 by Ken Mattsson. You can create them as follows.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"D_upw = upwind_operators(SummationByPartsOperators.Mattsson2017,\n derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Upwind operators are derivative operators biased towards one direction. The \"minus\" variants has a bias towards the left side, i.e., it uses values from more nodes to the left than from the right to compute the discrete derivative approximation at a given node (in the interior of the domain). In matrix form, this means more non-zero entries are left from the diagonal.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Matrix(D_upw.minus)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Analogously, the \"plus\" variant has a bias towards the right side.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Matrix(D_upw.plus)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"For more information on upwind SBP operators, please refer to the documentation of SummationByPartsOperators.jl and references cited there.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"The basic idea of upwind SBP schemes is to apply a flux vector splitting and use appropriate upwind operators for both parts of the flux. In 1D, this means to split the flux","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"f(u) = f^-(u) + f^+(u)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"such that f^-(u) is associated with left-going waves and f^+(u) with right-going waves. Then, we apply upwind SBP operators D^- D^+ with an appropriate upwind bias, resulting in","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"partial_x f(u) approx D^+ f^-(u) + D^- f^+(u)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Note that the established notations of upwind operators D^pm and flux splittings f^pm clash. The right-going waves from f^+ need an operator biased towards their upwind side, i.e., the left side. This upwind bias is provided by the operator D^-.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Many classical flux vector splittings have been developed for finite volume methods and are described in the book \"Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction\" of Eleuterio F. Toro (2009), DOI: 10.1007/b79761. One such a well-known splitting provided by Trixi.jl is splitting_steger_warming.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"Trixi.jl comes with several example setups using upwind SBP methods with flux vector splitting, e.g.,","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"elixir_euler_vortex.jl\nelixir_euler_taylor_green_vortex.jl","category":"page"},{"location":"tutorials/upwind_fdsbp/#Package-versions","page":"7 Upwind FD SBP schemes","title":"Package versions","text":"","category":"section"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"SummationByPartsOperators\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"7 Upwind FD SBP schemes","title":"7 Upwind FD SBP schemes","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"EditURL = \"../../literate/src/files/hohqmesh_tutorial.jl\"","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#hohqmesh_tutorial","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14: Unstructured meshes with HOHQMesh.jl","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Trixi.jl supports numerical approximations on unstructured quadrilateral meshes with the UnstructuredMesh2D mesh type.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The purpose of this tutorial is to demonstrate how to use the UnstructuredMesh2D functionality of Trixi.jl. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will demonstrate how to conceptualize a problem with curved boundaries, generate a curvilinear mesh using the available software in the Trixi.jl ecosystem, and then run a simulation using Trixi.jl on said mesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Unstructured quadrilateral meshes can be made with the High-Order Hex-Quad Mesh (HOHQMesh) generator created and developed by David Kopriva. HOHQMesh is a mesh generator specifically designed for spectral element methods. It provides high-order boundary curve information (needed to accurately set boundary conditions) and elements can be larger (due to the high accuracy of the spatial approximation) compared to traditional finite element mesh generators. For more information about the design and features of HOHQMesh one can refer to its official documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"HOHQMesh is incorporated into the Trixi.jl framework via the registered Julia package HOHQMesh.jl. This package provides a Julia wrapper for the HOHQMesh generator that allows users to easily create mesh files without the need to build HOHQMesh from source. To install the HOHQMesh package execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"import Pkg; Pkg.add(\"HOHQMesh\")","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Now we are ready to generate an unstructured quadrilateral mesh that can be used by Trixi.jl.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Running-and-visualizing-an-unstructured-simulation","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Running and visualizing an unstructured simulation","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Trixi.jl supports solving hyperbolic problems on several mesh types. There is a default example for this mesh type that can be executed by","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using Trixi\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi_include(default_example_unstructured())\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"This will compute a smooth, manufactured solution test case for the 2D compressible Euler equations on the curved quadrilateral mesh described in the Trixi.jl documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Apart from the usual error and timing output provided by the Trixi.jl run, it is useful to visualize and inspect the solution. One option available in the Trixi.jl framework to visualize the solution on unstructured quadrilateral meshes is post-processing the Trixi.jl output file(s) with the Trixi2Vtk tool and plotting them with ParaView.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"To convert the HDF5-formatted .h5 output file(s) from Trixi.jl into VTK format execute the following","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using Trixi2Vtk\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000180.h5\", output_directory=\"out\")\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Note this step takes about 15-30 seconds as the package Trixi2Vtk must be precompiled and executed for the first time in your REPL session. The trixi2vtk command above will convert the solution file at the final time into a .vtu file which can be read in and visualized with ParaView. Optional arguments for trixi2vtk are: (1) Pointing to the output_directory where the new files will be saved; it defaults to the current directory. (2) Specifying a higher number of visualization nodes. For instance, if we want to use 12 uniformly spaced nodes for visualization we can execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"redirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000180.h5\", output_directory=\"out\", nvisnodes=12)\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"By default trixi2vtk sets nvisnodes to be the same as the number of nodes specified in the elixir file used to run the simulation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Finally, if you want to convert all the solution files to VTK execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"redirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000*.h5\", output_directory=\"out\", nvisnodes=12)\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"then it is possible to open the .pvd file with ParaView and create a video of the simulation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Creating-a-mesh-using-HOHQMesh","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Creating a mesh using HOHQMesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The creation of an unstructured quadrilateral mesh using HOHQMesh.jl is driven by a control file. In this file the user dictates the domain to be meshed, prescribes any desired boundary curvature, the polynomial order of said boundaries, etc. In this tutorial we cover several basic features of the possible control inputs. For a complete discussion on this topic see the HOHQMesh control file documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"To begin, we provide a complete control file in this tutorial. After this we give a breakdown of the control file components to explain the chosen parameters.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Suppose we want to create a mesh of a domain with straight sided outer boundaries and a curvilinear \"ice cream cone\" shaped object at its center.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_boundary_cartoon)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The associated ice_cream_straight_sides.control file is created below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"open(\"out/ice_cream_straight_sides.control\", \"w\") do io\n println(io, raw\"\"\"\n\\begin{CONTROL_INPUT}\n \\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_straight_sides.mesh\n plot file name = ice_cream_straight_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n \\end{RUN_PARAMETERS}\n\n \\begin{BACKGROUND_GRID}\n x0 = [-8.0, -8.0, 0.0]\n dx = [1.0, 1.0, 0.0]\n N = [16,16,1]\n \\end{BACKGROUND_GRID}\n\n \\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n \\end{SPRING_SMOOTHER}\n\n\\end{CONTROL_INPUT}\n\n\\begin{MODEL}\n\n \\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n \\end{INNER_BOUNDARIES}\n\n\\end{MODEL}\n\\end{FILE}\n\"\"\")\nend","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The first three blocks of information are wrapped within a CONTROL_INPUT environment block as they define the core components of the quadrilateral mesh that will be generated.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The first block of information in RUN_PARAMETERS is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"\\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_straight_sides.mesh\n plot file name = ice_cream_straight_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n\\end{RUN_PARAMETERS}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The mesh and plot file names will be the files created by HOHQMesh once successfully executed. The stats file name is available if you wish to also save a collection of mesh statistics. For this example it is deactivated. These file names given within RUN_PARAMETERS should match that of the control file, and although this is not required by HOHQMesh, it is a useful style convention. The mesh file format ISM-v2 in the format currently required by Trixi.jl. The polynomial order prescribes the order of an interpolant constructed on the Chebyshev-Gauss-Lobatto nodes that is used to represent any curved boundaries on a particular element. The plot file format of skeleton means that visualizing the plot file will only draw the element boundaries (and no internal nodes). Alternatively, the format can be set to sem to visualize the interior nodes of the approximation as well.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The second block of information in BACKGROUND_GRID is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"\\begin{BACKGROUND_GRID}\n x0 = [-8.0, -8.0, 0.0]\n dx = [1.0, 1.0, 0.0]\n N = [16,16,1]\n\\end{BACKGROUND_GRID}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"This lays a grid of Cartesian elements for the domain beginning at the point x0 as its bottom-left corner. The value of dx, which could differ in each direction if desired, controls the step size taken in each Cartesian direction. The values in N set how many Cartesian box elements are set in each coordinate direction. The above parameters define a 16times 16 element square mesh on -88^2. Further, this sets up four outer boundaries of the domain that are given the default names: Top, Left, Bottom, Right.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The third block of information in SPRING_SMOOTHER is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"\\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n\\end{SPRING_SMOOTHER}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Once HOHQMesh generates the mesh, a spring-mass-dashpot model is created to smooth the mesh and create \"nicer\" quadrilateral elements. The default parameters of Hooke's law for the spring-mass-dashpot model have been selected after a fair amount of experimentation across many meshes. If you wish to deactivate this feature you can set smoothing = OFF (or remove this block from the control file).","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"After the CONTROL_INPUT environment block comes the MODEL environment block. It is here where the user prescribes curved boundary information with either:","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"An OUTER_BOUNDARY (covered in the next section of this tutorial).\nOne or more INNER_BOUNDARIES.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"There are several options to describe the boundary curve data to HOHQMesh like splines or parametric curves.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"For the example ice_cream_straight_sides.control we define three internal boundaries; two straight-sided and one as a circular arc. Within the HOHQMesh control input each curve must be assigned to a CHAIN as shown below in the complete INNER_BOUNDARIES block.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"\\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n\\end{INNER_BOUNDARIES}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"It is important to note there are two name quantities one for the CHAIN and one for the PARAMETRIC_EQUATION_CURVE. The name for the CHAIN is used internally by HOHQMesh, so if you have multiple CHAINs they must be given a unique name. The name for the PARAMETRIC_EQUATION_CURVE will be printed to the appropriate boundaries within the .mesh file produced by HOHQMesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We create the mesh file ice_cream_straight_sides.mesh and its associated file for plotting ice_cream_straight_sides.tec by using HOHQMesh.jl's function generate_mesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using HOHQMesh\ncontrol_file = joinpath(\"out\", \"ice_cream_straight_sides.control\")\noutput = generate_mesh(control_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The mesh file ice_cream_straight_sides.mesh and its associated file for plotting ice_cream_straight_sides.tec are placed in the out folder. The resulting mesh generated by HOHQMesh.jl is given in the following figure.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_straight_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We note that Trixi.jl uses the boundary name information from the control file to assign boundary conditions in an elixir file. Therefore, the name should start with a letter and consist only of alphanumeric characters and underscores. Please note that the name will be treated as case sensitive.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Example-simulation-on-ice_cream_straight_sides.mesh","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Example simulation on ice_cream_straight_sides.mesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"With this newly generated mesh we are ready to run a Trixi.jl simulation on an unstructured quadrilateral mesh. For this we must create a new elixir file.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The elixir file given below creates an initial condition for a uniform background flow state with a free stream Mach number of 0.3. A focus for this part of the tutorial is to specify the boundary conditions and to construct the new mesh from the file that was generated in the previous exercise.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"It is straightforward to set the different boundary condition types in an elixir by assigning a particular function to a boundary name inside a Julia dictionary, Dict, variable. Observe that the names of these boundaries match those provided by HOHQMesh either by default, e.g. Bottom, or user assigned, e.g. IceCream. For this problem setup use","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Freestream boundary conditions on the four box edges.\nFree slip wall boundary condition on the interior curved boundaries.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"To construct the unstructured quadrilateral mesh from the HOHQMesh file we point to the appropriate location with the variable mesh_file and then feed this into the constructor for the UnstructuredMesh2D type in Trixi.jl","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"# create the unstructured mesh from your mesh file\nusing Trixi\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file);","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The complete elixir file for this simulation example is given below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using OrdinaryDiffEq, Trixi\n\nequations = CompressibleEulerEquations2D(1.4) # set gas gamma = 1.4\n\n# freestream flow state with Ma_inf = 0.3\n@inline function uniform_flow_state(x, t, equations::CompressibleEulerEquations2D)\n\n # set the freestream flow parameters\n rho_freestream = 1.0\n u_freestream = 0.3\n p_freestream = inv(equations.gamma)\n\n theta = 0.0 # zero angle of attack\n si, co = sincos(theta)\n v1 = u_freestream * co\n v2 = u_freestream * si\n\n prim = SVector(rho_freestream, v1, v2, p_freestream)\n return prim2cons(prim, equations)\nend\n\n# initial condition\ninitial_condition = uniform_flow_state\n\n# boundary condition types\nboundary_condition_uniform_flow = BoundaryConditionDirichlet(uniform_flow_state)\n\n# boundary condition dictionary\nboundary_conditions = Dict( :Bottom => boundary_condition_uniform_flow,\n :Top => boundary_condition_uniform_flow,\n :Right => boundary_condition_uniform_flow,\n :Left => boundary_condition_uniform_flow,\n :LeftSlant => boundary_condition_slip_wall,\n :RightSlant => boundary_condition_slip_wall,\n :IceCream => boundary_condition_slip_wall );\n\n# DGSEM solver.\n# 1) polydeg must be >= the polynomial order set in the HOHQMesh control file to guarantee\n# freestream preservation. As a extra task try setting polydeg=3\n# 2) VolumeIntegralFluxDifferencing with central volume flux is activated\n# for dealiasing\nvolume_flux = flux_ranocha\nsolver = DGSEM(polydeg=4, surface_flux=flux_hll,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# create the unstructured mesh from your mesh file\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file)\n\n# Create semidiscretization with all spatial discretization-related components\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_conditions)\n\n# Create ODE problem from semidiscretization with time span from 0.0 to 2.0\ntspan = (0.0, 2.0)\node = semidiscretize(semi, tspan)\n\n\n# Create the callbacks to output solution files and adapt the time step\nsummary_callback = SummaryCallback()\nsave_solution = SaveSolutionCallback(interval=10,\n save_initial_solution=true,\n save_final_solution=true)\nstepsize_callback = StepsizeCallback(cfl=1.0)\n\ncallbacks = CallbackSet(summary_callback, save_solution, stepsize_callback)\n\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\n# Evolve ODE problem in time using `solve` from OrdinaryDiffEq\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n# print the timer summary\nsummary_callback()\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Visualization of the solution is carried out in a similar way as above. That is, one converts the .h5 output files with trixi2vtk and then plot the solution in ParaView. An example plot of the pressure at the final time is shown below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_straight_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Making-a-mesh-with-a-curved-outer-boundary","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Making a mesh with a curved outer boundary","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Let us modify the mesh from the previous task and place a circular outer boundary instead of straight-sided outer boundaries. Note, the \"ice cream cone\" shape is still placed at the center of the domain.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We create the new control file ice_cream_curved_sides.control file below and will then highlight the major differences compared to ice_cream_straight_sides.control.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"open(\"out/ice_cream_curved_sides.control\", \"w\") do io\n println(io, raw\"\"\"\n\\begin{CONTROL_INPUT}\n \\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_curved_sides.mesh\n plot file name = ice_cream_curved_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n \\end{RUN_PARAMETERS}\n\n \\begin{BACKGROUND_GRID}\n background grid size = [1.0, 1.0, 0.0]\n \\end{BACKGROUND_GRID}\n\n \\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n \\end{SPRING_SMOOTHER}\n\n\\end{CONTROL_INPUT}\n\n\\begin{MODEL}\n\n \\begin{OUTER_BOUNDARY}\n \\begin{PARAMETRIC_EQUATION_CURVE}\n name = OuterCircle\n xEqn = x(t) = 8.0*sin(2.0*pi*t)\n yEqn = y(t) = 8.0*cos(2.0*pi*t)\n zEqn = z(t) = 0.0\n \\end{PARAMETRIC_EQUATION_CURVE}\n\n \\end{OUTER_BOUNDARY}\n\n \\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n \\end{INNER_BOUNDARIES}\n\n\\end{MODEL}\n\\end{FILE}\n\"\"\")\nend","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The first alteration is that we have altered the second block of information BACKGROUND_GRID within the CONTROL_INPUT to be","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"\\begin{BACKGROUND_GRID}\n background grid size = [1.0, 1.0, 0.0]\n\\end{BACKGROUND_GRID}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"This mesh control file has an outer boundary that determines the extent of the domain to be meshed. Therefore, we only need to supply the background grid size to the BACKGROUND_GRID control input.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The second alteration is that the MODEL now contains information for an OUTER_BOUNDARY. In this case it is a circle of radius 8 centered at [0.0, 0.0, 0.0] written as a set of PARAMETRIC_EQUATION_CURVEs.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":" \\begin{OUTER_BOUNDARY}\n\n \\begin{PARAMETRIC_EQUATION_CURVE}\n name = OuterCircle\n xEqn = x(t) = 8.0*sin(2.0*pi*t)\n yEqn = y(t) = 8.0*cos(2.0*pi*t)\n zEqn = z(t) = 0.0\n \\end{PARAMETRIC_EQUATION_CURVE}\n\n \\end{OUTER_BOUNDARY}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Just as with the inner boundary curves, we must assign a name to the OUTER_BOUNDARY. It will be included in the generated .mesh file and is used within the Trixi.jl elixir file to set boundary conditions.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Again, we create the .mesh and .tec files with HOHQMesh.jl's function generate_mesh","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"control_file = joinpath(\"out\", \"ice_cream_curved_sides.control\")\noutput = generate_mesh(control_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The files are placed in the out folder.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The resulting mesh generated by HOHQMesh.jl is given in the following figure.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_curved_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Running-Trixi.jl-on-ice_cream_curved_sides.mesh","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Running Trixi.jl on ice_cream_curved_sides.mesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We can reuse much of the elixir file to setup the uniform flow over an ice cream cone from the previous part of this tutorial. The only component of the elixir file that must be changed is the boundary condition dictionary because we now have a boundary named OuterCircle instead of four edges of a bounding box.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"# boundary condition dictionary\nboundary_conditions = Dict( :OuterCircle => boundary_condition_uniform_flow,\n :LeftSlant => boundary_condition_slip_wall,\n :RightSlant => boundary_condition_slip_wall,\n :IceCream => boundary_condition_slip_wall );\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Also, we must update the construction of the mesh from our new mesh file ice_cream_curved_sides.mesh that is located in the out folder.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"# create the unstructured mesh from your mesh file\nmesh_file = joinpath(\"out\", \"ice_cream_curved_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We can then post-process the solution file at the final time on the new mesh with Trixi2Vtk and visualize with ParaView.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_curved_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Setting-up-a-simulation-with-AMR-via-P4estMesh","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Setting up a simulation with AMR via P4estMesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"The above explained mesh file format of ISM-V2 only works with UnstructuredMesh2D and so does not support AMR. On the other hand, the mesh type P4estMesh allows AMR. The mesh constructor for the P4estMesh imports an unstructured, conforming mesh from an Abaqus mesh file (.inp).","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"As described above, the first block of the HOHQMesh control file contains the parameter mesh file format. If you set mesh file format = ABAQUS instead of ISM-V2, HOHQMesh.jl's function generate_mesh creates an Abaqus mesh file .inp.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using HOHQMesh\ncontrol_file = joinpath(\"out\", \"ice_cream_straight_sides.control\")\noutput = generate_mesh(control_file);","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Now, you can create a P4estMesh from your mesh file. It is described in detail in the P4est-based mesh part of the Trixi.jl docs.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using Trixi\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.inp\")\nmesh = P4estMesh{2}(mesh_file)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"Since P4estMesh supports AMR, we just have to extend the setup from the first example by the standard AMR procedure. For more information about AMR in Trixi.jl, see the matching tutorial.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"amr_indicator = IndicatorLöhner(semi, variable=density)\n\namr_controller = ControllerThreeLevel(semi, amr_indicator,\n base_level=0,\n med_level =1, med_threshold=0.05,\n max_level =3, max_threshold=0.1)\n\namr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)\n\ncallbacks = CallbackSet(..., amr_callback)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"We can then post-process the solution file at the final time on the new mesh with Trixi2Vtk and visualize with ParaView, see the appropriate visualization section for details.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_straight_sides_p4est_amr)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Package-versions","page":"14 Unstructured meshes with HOHQMesh.jl","title":"Package versions","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\", \"Trixi2Vtk\", \"HOHQMesh\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"14 Unstructured meshes with HOHQMesh.jl","title":"14 Unstructured meshes with HOHQMesh.jl","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"EditURL = \"../../literate/src/files/DGMulti_2.jl\"","category":"page"},{"location":"tutorials/DGMulti_2/#DGMulti_2","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6: Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"section"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"For a tutorial about DG schemes via the DGMulti solver please visit the previous tutorial. The DGMulti solver also supports other methods than DG. The important property a method has to fulfill is the summation-by-parts (SBP) property. The package SummationByPartsOperators.jl provides such methods, like a finite difference SBP (FD SBP) scheme. To do this, you need to create an SBP derivative operator and pass that as approximation_type to the DGMulti constructor. For example, the classical second-order FD SBP operator can be created as","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly\nD = derivative_operator(MattssonNordström2004(), derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"Here, the arguments xmin and xmax do not matter beyond setting the real type used for the operator - they just set a reference element and are rescaled on the physical elements. The parameter N determines the number of finite difference nodes. Then, D can be used as approximation_type like SBP() in a multi-block fashion. In multiple dimensions, such a 1D SBP operator will be used in a tensor product fashion, i.e., in each coordinate direction. In particular, you can use them only on 1D, 2D Quad(), and 3D Hex() elements.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"You can also use fully periodic single-block FD methods by creating a periodic SBP operator. For example, a fully periodic FD operator can be constructed as","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"D = periodic_derivative_operator(derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"An example using such an FD method is implemented in elixir_euler_fdsbp_periodic.jl. For all parameters and other calling options, please have a look in the documentation of SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"Another possible method is for instance a continuous Galerkin (CGSEM) method. You can use such a method with polynomial degree of 3 (N=4 Legendre Lobatto nodes on [0, 1]) coupled continuously on a uniform mesh with Nx=10 elements by setting approximation_type to","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly\nD = couple_continuously(legendre_derivative_operator(xmin=0.0, xmax=1.0, N=4),\n UniformPeriodicMesh1D(xmin=-1.0, xmax=1.0, Nx=10))","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"To choose a discontinuous coupling (DGSEM), use couple_discontinuously() instead of couple_continuously().","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"For more information and other SBP operators, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/DGMulti_2/#Package-versions","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"StartUpDG\", \"SummationByPartsOperators\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"page"},{"location":"tutorials/DGMulti_2/","page":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"EditURL = \"../../literate/src/files/parabolic_terms.jl\"","category":"page"},{"location":"tutorials/parabolic_terms/#parabolic_terms","page":"10 Parabolic terms","title":"10: Parabolic terms","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"Experimental support for parabolic diffusion terms is available in Trixi.jl. This demo illustrates parabolic terms for the advection-diffusion equation.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"using OrdinaryDiffEq\nusing Trixi","category":"page"},{"location":"tutorials/parabolic_terms/#Splitting-a-system-into-hyperbolic-and-parabolic-parts.","page":"10 Parabolic terms","title":"Splitting a system into hyperbolic and parabolic parts.","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"For a mixed hyperbolic-parabolic system, we represent the hyperbolic and parabolic parts of the system separately. We first define the hyperbolic (advection) part of the advection-diffusion equation.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"advection_velocity = (1.5, 1.0)\nequations_hyperbolic = LinearScalarAdvectionEquation2D(advection_velocity);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"Next, we define the parabolic diffusion term. The constructor requires knowledge of equations_hyperbolic to be passed in because the LaplaceDiffusion2D applies diffusion to every variable of the hyperbolic system.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"diffusivity = 5.0e-2\nequations_parabolic = LaplaceDiffusion2D(diffusivity, equations_hyperbolic);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Boundary-conditions","page":"10 Parabolic terms","title":"Boundary conditions","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"As with the equations, we define boundary conditions separately for the hyperbolic and parabolic part of the system. For this example, we impose inflow BCs for the hyperbolic system (no condition is imposed on the outflow), and we impose Dirichlet boundary conditions for the parabolic equations. Both BoundaryConditionDirichlet and BoundaryConditionNeumann are defined for LaplaceDiffusion2D.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"The hyperbolic and parabolic boundary conditions are assumed to be consistent with each other.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"boundary_condition_zero_dirichlet = BoundaryConditionDirichlet((x, t, equations) -> SVector(0.0))\n\nboundary_conditions_hyperbolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1 + 0.5 * x[2])),\n y_neg = boundary_condition_zero_dirichlet,\n y_pos = boundary_condition_do_nothing,\n x_pos = boundary_condition_do_nothing)\n\nboundary_conditions_parabolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1 + 0.5 * x[2])),\n y_neg = boundary_condition_zero_dirichlet,\n y_pos = boundary_condition_zero_dirichlet,\n x_pos = boundary_condition_zero_dirichlet);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Defining-the-solver-and-mesh","page":"10 Parabolic terms","title":"Defining the solver and mesh","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"The process of creating the DG solver and mesh is the same as for a purely hyperbolic system of equations.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\ncoordinates_min = (-1.0, -1.0) # minimum coordinates (min(x), min(y))\ncoordinates_max = ( 1.0, 1.0) # maximum coordinates (max(x), max(y))\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n periodicity=false, n_cells_max=30_000) # set maximum capacity of tree data structure\n\ninitial_condition = (x, t, equations) -> SVector(0.0);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Semidiscretizing-and-solving","page":"10 Parabolic terms","title":"Semidiscretizing and solving","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"To semidiscretize a hyperbolic-parabolic system, we create a SemidiscretizationHyperbolicParabolic. This differs from a SemidiscretizationHyperbolic in that we pass in a Tuple containing both the hyperbolic and parabolic equation, as well as a Tuple containing the hyperbolic and parabolic boundary conditions.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"semi = SemidiscretizationHyperbolicParabolic(mesh,\n (equations_hyperbolic, equations_parabolic),\n initial_condition, solver;\n boundary_conditions=(boundary_conditions_hyperbolic,\n boundary_conditions_parabolic))","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"The rest of the code is identical to the hyperbolic case. We create a system of ODEs through semidiscretize, defining callbacks, and then passing the system to OrdinaryDiffEq.jl.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"tspan = (0.0, 1.5)\node = semidiscretize(semi, tspan)\ncallbacks = CallbackSet(SummaryCallback())\ntime_int_tol = 1.0e-6\nsol = solve(ode, RDPK3SpFSAL49(); abstol=time_int_tol, reltol=time_int_tol,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"We can now visualize the solution, which develops a boundary layer at the outflow boundaries.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/parabolic_terms/#Package-versions","page":"10 Parabolic terms","title":"Package versions","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"","category":"page"},{"location":"tutorials/parabolic_terms/","page":"10 Parabolic terms","title":"10 Parabolic terms","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"EditURL = \"../../literate/src/files/index.jl\"","category":"page"},{"location":"tutorials/introduction/#Tutorials-for-Trixi.jl","page":"Introduction","title":"Tutorials for Trixi.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"The tutorial section for Trixi.jl also contains interactive step-by-step explanations via Binder.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Right now, you are using the classic documentation. The corresponding interactive notebooks can be opened in Binder and viewed in nbviewer via the icons (Image: ) and (Image: ) in the respective tutorial. You can also open the raw notebook files via (Image: ).","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Note: To improve responsiveness via caching, the notebooks are updated only once a week. They are only available for the latest stable release of Trixi.jl at the time of caching.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"There are tutorials for the following topics:","category":"page"},{"location":"tutorials/introduction/#[1-Introduction-to-DG-methods](@ref-scalar_linear_advection_1d)","page":"Introduction","title":"1 Introduction to DG methods","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial gives an introduction to discontinuous Galerkin (DG) methods with the example of the scalar linear advection equation in 1D. Starting with some theoretical explanations, we first implement a raw version of a discontinuous Galerkin spectral element method (DGSEM). Then, we will show how to use features of Trixi.jl to achieve the same result.","category":"page"},{"location":"tutorials/introduction/#[2-DGSEM-with-flux-differencing](@ref-DGSEM_FluxDiff)","page":"Introduction","title":"2 DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"To improve stability often the flux differencing formulation of the DGSEM (split form) is used. We want to present the idea and formulation on a basic 1D level. Then, we show how this formulation can be implemented in Trixi.jl and analyse entropy conservation for two different flux combinations.","category":"page"},{"location":"tutorials/introduction/#[3-Shock-capturing-with-flux-differencing-and-stage-limiter](@ref-shock_capturing)","page":"Introduction","title":"3 Shock capturing with flux differencing and stage limiter","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Using the flux differencing formulation, a simple procedure to capture shocks is a hybrid blending of a high-order DG method and a low-order subcell finite volume (FV) method. We present the idea on a very basic level and show the implementation in Trixi.jl. Then, a positivity preserving limiter is explained and added to an exemplary simulation of the Sedov blast wave with the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/introduction/#[4-Non-periodic-boundary-conditions](@ref-non_periodic_boundaries)","page":"Introduction","title":"4 Non-periodic boundary conditions","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Thus far, all examples used periodic boundaries. In Trixi.jl, you can also set up a simulation with non-periodic boundaries. This tutorial presents the implementation of the classical Dirichlet boundary condition with a following example. Then, other non-periodic boundaries are mentioned.","category":"page"},{"location":"tutorials/introduction/#[5-DG-schemes-via-DGMulti-solver](@ref-DGMulti_1)","page":"Introduction","title":"5 DG schemes via DGMulti solver","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial is about the more general DG solver DGMulti, introduced here. We are showing some examples for this solver, for instance with discretization nodes by Gauss or triangular elements. Moreover, we present a simple way to include pre-defined triangulate meshes for non-Cartesian domains using the package StartUpDG.jl.","category":"page"},{"location":"tutorials/introduction/#[6-Other-SBP-schemes-(FD,-CGSEM)-via-DGMulti-solver](@ref-DGMulti_2)","page":"Introduction","title":"6 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Supplementary to the previous tutorial about DG schemes via the DGMulti solver we now present the possibility for DGMulti to use other SBP schemes via the package SummationByPartsOperators.jl. For instance, we show how to set up a finite differences (FD) scheme and a continuous Galerkin (CGSEM) method.","category":"page"},{"location":"tutorials/introduction/#[7-Upwind-FD-SBP-schemes](@ref-upwind_fdsbp)","page":"Introduction","title":"7 Upwind FD SBP schemes","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"General SBP schemes can not only be used via the DGMulti solver but also with a general DG solver. In particular, upwind finite difference SBP methods can be used together with the TreeMesh. Similar to general SBP schemes in the DGMulti framework, the interface is based on the package SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/introduction/#[8-Adding-a-new-scalar-conservation-law](@ref-adding_new_scalar_equations)","page":"Introduction","title":"8 Adding a new scalar conservation law","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial explains how to add a new physics model using the example of the cubic conservation law. First, we define the equation using a struct CubicEquation and the physical flux. Then, the corresponding standard setup in Trixi.jl (mesh, solver, semi and ode) is implemented and the ODE problem is solved by OrdinaryDiffEq's solve method.","category":"page"},{"location":"tutorials/introduction/#[9-Adding-a-non-conservative-equation](@ref-adding_nonconservative_equation)","page":"Introduction","title":"9 Adding a non-conservative equation","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"In this part, another physics model is implemented, the nonconservative linear advection equation. We run two different simulations with different levels of refinement and compare the resulting errors.","category":"page"},{"location":"tutorials/introduction/#[10-Parabolic-terms](@ref-parabolic_terms)","page":"Introduction","title":"10 Parabolic terms","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes how parabolic terms are implemented in Trixi.jl, e.g., to solve the advection-diffusion equation.","category":"page"},{"location":"tutorials/introduction/#[11-Adding-new-parabolic-terms](@ref-adding_new_parabolic_terms)","page":"Introduction","title":"11 Adding new parabolic terms","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes how new parabolic terms can be implemented using Trixi.jl.","category":"page"},{"location":"tutorials/introduction/#[12-Adaptive-mesh-refinement](@ref-adaptive_mesh_refinement)","page":"Introduction","title":"12 Adaptive mesh refinement","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Adaptive mesh refinement (AMR) helps to increase the accuracy in sensitive or turbolent regions while not wasting resources for less interesting parts of the domain. This leads to much more efficient simulations. This tutorial presents the implementation strategy of AMR in Trixi.jl, including the use of different indicators and controllers.","category":"page"},{"location":"tutorials/introduction/#[13-Structured-mesh-with-curvilinear-mapping](@ref-structured_mesh_mapping)","page":"Introduction","title":"13 Structured mesh with curvilinear mapping","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"In this tutorial, the use of Trixi.jl's structured curved mesh type StructuredMesh is explained. We present the two basic option to initialize such a mesh. First, the curved domain boundaries of a circular cylinder are set by explicit boundary functions. Then, a fully curved mesh is defined by passing the transformation mapping.","category":"page"},{"location":"tutorials/introduction/#[14-Unstructured-meshes-with-HOHQMesh.jl](@ref-hohqmesh_tutorial)","page":"Introduction","title":"14 Unstructured meshes with HOHQMesh.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"The purpose of this tutorial is to demonstrate how to use the UnstructuredMesh2D functionality of Trixi.jl. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will demonstrate how to conceptualize a problem with curved boundaries, generate a curvilinear mesh using the available HOHQMesh software in the Trixi.jl ecosystem, and then run a simulation using Trixi.jl on said mesh. In the end, the tutorial briefly explains how to simulate an example using AMR via P4estMesh.","category":"page"},{"location":"tutorials/introduction/#[15-Explicit-time-stepping](@ref-time_stepping)","page":"Introduction","title":"15 Explicit time stepping","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial is about time integration using OrdinaryDiffEq.jl. It explains how to use their algorithms and presents two types of time step choices - with error-based and CFL-based adaptive step size control.","category":"page"},{"location":"tutorials/introduction/#[16-Differentiable-programming](@ref-differentiable_programming)","page":"Introduction","title":"16 Differentiable programming","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This part deals with some basic differentiable programming topics. For example, a Jacobian, its eigenvalues and a curve of total energy (through the simulation) are calculated and plotted for a few semidiscretizations. Moreover, we calculate an example for propagating errors with Measurement.jl at the end.","category":"page"},{"location":"tutorials/introduction/#[17-Custom-semidiscretization](@ref-custom_semidiscretization)","page":"Introduction","title":"17 Custom semidiscretization","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes the semidiscretiations of Trixi.jl and explains how to extend them for custom tasks.","category":"page"},{"location":"tutorials/introduction/#Examples-in-Trixi.jl","page":"Introduction","title":"Examples in Trixi.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Trixi.jl already contains several more coding examples, the so-called elixirs. You can find them in the folder examples/. They are structured by the underlying mesh type and the respective number of spatial dimensions. The name of an elixir is composed of the underlying system of conservation equations (for instance advection or euler) and other special characteristics like the initial condition (e.g. gauss, astro_jet, blast_wave) or the included simulation features (e.g. amr, shockcapturing).","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This page was generated using Literate.jl.","category":"page"},{"location":"reference-trixi/#Trixi.jl-API","page":"Trixi.jl","title":"Trixi.jl API","text":"","category":"section"},{"location":"reference-trixi/","page":"Trixi.jl","title":"Trixi.jl","text":"CurrentModule = Trixi","category":"page"},{"location":"reference-trixi/","page":"Trixi.jl","title":"Trixi.jl","text":"Modules = [Trixi]","category":"page"},{"location":"reference-trixi/#Trixi.Trixi","page":"Trixi.jl","title":"Trixi.Trixi","text":"Trixi\n\nTrixi.jl is a numerical simulation framework for hyperbolic conservation laws. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures.\n\nTo get started, run your first simulation with Trixi.jl using\n\ntrixi_include(default_example())\n\nSee also: trixi-framework/Trixi.jl\n\n\n\n\n\n","category":"module"},{"location":"reference-trixi/#Trixi.boundary_condition_do_nothing","page":"Trixi.jl","title":"Trixi.boundary_condition_do_nothing","text":"boundary_condition_do_nothing = Trixi.BoundaryConditionDoNothing()\n\nImposing no boundary condition just evaluates the flux at the inner state.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.boundary_condition_periodic","page":"Trixi.jl","title":"Trixi.boundary_condition_periodic","text":"boundary_condition_periodic = Trixi.BoundaryConditionPeriodic()\n\nA singleton struct indicating periodic boundary conditions.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_hll","page":"Trixi.jl","title":"Trixi.flux_hll","text":"flux_hll\n\nSee FluxHLL.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_hll_chen_noelle","page":"Trixi.jl","title":"Trixi.flux_hll_chen_noelle","text":"flux_hll_chen_noelle = FluxHLL(min_max_speed_chen_noelle)\n\nAn instance of FluxHLL specific to the shallow water equations that uses the wave speed estimates from min_max_speed_chen_noelle. This HLL flux is guaranteed to have zero numerical mass flux out of a \"dry\" element, maintain positivity of the water height, and satisfy an entropy inequality.\n\nFor complete details see Section 2.4 of the following reference\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI: 10.1137/15M1053074\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_hlle","page":"Trixi.jl","title":"Trixi.flux_hlle","text":"flux_hlle\n\nSee min_max_speed_einfeldt. This is a FluxHLL-type two-wave solver with special estimates of the wave speeds.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_lax_friedrichs","page":"Trixi.jl","title":"Trixi.flux_lax_friedrichs","text":"flux_lax_friedrichs\n\nSee FluxLaxFriedrichs.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.AMRCallback","page":"Trixi.jl","title":"Trixi.AMRCallback","text":"AMRCallback(semi, controller [,adaptor=AdaptorAMR(semi)];\n interval,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true,\n dynamic_load_balancing=true)\n\nPerforms adaptive mesh refinement (AMR) every interval time steps for a given semidiscretization semi using the chosen controller.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AbstractEquations","page":"Trixi.jl","title":"Trixi.AbstractEquations","text":"AbstractEquations{NDIMS, NVARS}\n\nAn abstract supertype of specific equations such as the compressible Euler equations. The type parameters encode the number of spatial dimensions (NDIMS) and the number of primary variables (NVARS) of the physics model.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AbstractMesh","page":"Trixi.jl","title":"Trixi.AbstractMesh","text":"AbstractMesh{NDIMS}\n\nAn abstract supertype of specific mesh types such as TreeMesh or StructuredMesh. The type parameters encode the number of spatial dimensions (NDIMS).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AcousticPerturbationEquations2D","page":"Trixi.jl","title":"Trixi.AcousticPerturbationEquations2D","text":"AcousticPerturbationEquations2D(v_mean_global, c_mean_global, rho_mean_global)\n\nAcoustic perturbation equations (APE) in two space dimensions. The equations are given by\n\nbeginaligned\n fracpartialmathbfvpartial t + nabla (barmathbfvcdotmathbfv)\n + nablaleft( fracbarc^2 tildepbarrho right) = 0 \n fracpartial tildeppartial t +\n nablacdot (barrho mathbfv + barmathbfv tildep) = 0\nendaligned\n\nThe bar bar(cdot) indicates time-averaged quantities. The unknowns of the APE are the perturbed velocities mathbfv = (v_1 v_2)^T and the scaled perturbed pressure tildep = fracpbarc^2, where p denotes the perturbed pressure and the perturbed variables are defined by phi = phi - barphi.\n\nIn addition to the unknowns, Trixi.jl currently stores the mean values in the state vector, i.e. the state vector used internally is given by\n\nmathbfu =\n beginpmatrix\n v_1 v_2 tildep barv_1 barv_2 barc barrho\n endpmatrix\n\nThis affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the mean values must be zero.\nThe mean values have to be considered when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes these variables too.\nTrixi.jl's visualization tools will visualize the mean values by default.\n\nThe constructor accepts a 2-tuple v_mean_global and scalars c_mean_global and rho_mean_global which can be used to make the definition of initial conditions for problems with constant mean flow more flexible. These values are ignored if the mean values are defined internally in an initial condition.\n\nThe equations are based on the APE-4 system introduced in the following paper:\n\nRoland Ewert and Wolfgang Schröder (2003) Acoustic perturbation equations based on flow decomposition via source filtering DOI: 10.1016/S0021-9991(03)00168-2\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.Adiabatic","page":"Trixi.jl","title":"Trixi.Adiabatic","text":"struct Adiabatic\n\nUsed to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_normal_flux_function should be a function with signature boundary_value_normal_flux_function(x, t, equations) and return a scalar value for the normal heat flux at point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AliveCallback","page":"Trixi.jl","title":"Trixi.AliveCallback","text":"AliveCallback(analysis_interval=0, alive_interval=analysis_interval÷10)\n\nInexpensive callback showing that a simulation is still running by printing some information such as the current time to the screen every alive_interval time steps. If analysis_interval ≂̸ 0, the output is omitted every analysis_interval time steps.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AnalysisCallback","page":"Trixi.jl","title":"Trixi.AnalysisCallback","text":"AnalysisCallback(semi; interval=0,\n save_analysis=false,\n output_directory=\"out\",\n analysis_filename=\"analysis.dat\",\n extra_analysis_errors=Symbol[],\n extra_analysis_integrals=())\n\nAnalyze a numerical solution every interval time steps and print the results to the screen. If save_analysis, the results are also saved in joinpath(output_directory, analysis_filename).\n\nAdditional errors can be computed, e.g. by passing extra_analysis_errors = (:l2_error_primitive, :linf_error_primitive) or extra_analysis_errors = (:conservation_error,).\n\nIf you want to omit the computation (to safe compute-time) of the default_analysis_errors, specify analysis_errors = Symbol[]. Note: default_analysis_errors are :l2_error and :linf_error for all equations. If you want to compute extra_analysis_errors such as :conservation_error solely, i.e., without :l2_error, :linf_error you need to specify analysis_errors = [:conservation_error] instead of extra_analysis_errors = [:conservation_error].\n\nFurther scalar functions func in extra_analysis_integrals are applied to the numerical solution and integrated over the computational domain. Some examples for this are entropy, energy_kinetic, energy_internal, and energy_total. You can also write your own function with the same signature as the examples listed above and pass it via extra_analysis_integrals. See the developer comments about Trixi.analyze, Trixi.pretty_form_utf, and Trixi.pretty_form_ascii for further information on how to create custom analysis quantities.\n\nIn addition, the analysis callback records and outputs a number of quantities that are useful for evaluating the computational performance, such as the total runtime, the performance index (time/DOF/rhs!), the time spent in garbage collection (GC), or the current memory usage (alloc'd memory).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AnalysisCallbackCoupled","page":"Trixi.jl","title":"Trixi.AnalysisCallbackCoupled","text":"AnalysisCallbackCoupled(semi, callbacks...)\n\nCombine multiple analysis callbacks for coupled simulations with a SemidiscretizationCoupled. For each coupled system, an indididual AnalysisCallback must be created and passed to the AnalysisCallbackCoupled in order, i.e., in the same sequence as the indidvidual semidiscretizations are stored in the SemidiscretizationCoupled.\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AveragingCallback","page":"Trixi.jl","title":"Trixi.AveragingCallback","text":"AveragingCallback(semi::SemidiscretizationHyperbolic, tspan; output_directory=\"out\",\n filename=\"averaging.h5\")\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nA callback that averages the flow field described by semi which must be a semidiscretization of the compressible Euler equations in two dimensions. The callback records the mean velocity, mean speed of sound, mean density, and mean vorticity for each node over the time interval given by tspan and stores the results in an HDF5 file filename in the directory output_directory. Note that this callback does not support adaptive mesh refinement (AMRCallback).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionCoupled","page":"Trixi.jl","title":"Trixi.BoundaryConditionCoupled","text":"BoundaryConditionCoupled(other_semi_index, indices, uEltype)\n\nBoundary condition to glue two meshes together. Solution values at the boundary of another mesh will be used as boundary values. This requires the use of SemidiscretizationCoupled. The other mesh is specified by other_semi_index, which is the index of the mesh in the tuple of semidiscretizations.\n\nNote that the elements and nodes of the two meshes at the coupled boundary must coincide. This is currently only implemented for StructuredMesh.\n\nArguments\n\nother_semi_index: the index in SemidiscretizationCoupled of the semidiscretization from which the values are copied\nindices::Tuple: node/cell indices at the boundary of the mesh in the other semidiscretization. See examples below.\nuEltype::Type: element type of solution\n\nExamples\n\n# Connect the left boundary of mesh 2 to our boundary such that our positive\n# boundary direction will match the positive y direction of the other boundary\nBoundaryConditionCoupled(2, (:begin, :i), Float64)\n\n# Connect the same two boundaries oppositely oriented\nBoundaryConditionCoupled(2, (:begin, :i_backwards), Float64)\n\n# Using this as y_neg boundary will connect `our_cells[i, 1, j]` to `other_cells[j, end-i, end]`\nBoundaryConditionCoupled(2, (:j, :i_backwards, :end), Float64)\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionDirichlet","page":"Trixi.jl","title":"Trixi.BoundaryConditionDirichlet","text":"BoundaryConditionDirichlet(boundary_value_function)\n\nCreate a Dirichlet boundary condition that uses the function boundary_value_function to specify the values at the boundary. This can be used to create a boundary condition that specifies exact boundary values by passing the exact solution of the equation. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as\n\nboundary_value_function(x, t, equations)\n\nwhere x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.\n\nExamples\n\njulia> BoundaryConditionDirichlet(initial_condition_convergence_test)\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionNavierStokesWall","page":"Trixi.jl","title":"Trixi.BoundaryConditionNavierStokesWall","text":"struct BoundaryConditionNavierStokesWall\n\nCreates a wall-type boundary conditions for the compressible Navier-Stokes equations. The fields boundary_condition_velocity and boundary_condition_heat_flux are intended to be boundary condition types such as the NoSlip velocity boundary condition and the Adiabatic or Isothermal heat boundary condition.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionNeumann","page":"Trixi.jl","title":"Trixi.BoundaryConditionNeumann","text":"BoundaryConditionNeumann(boundary_normal_flux_function)\n\nSimilar to BoundaryConditionDirichlet, but creates a Neumann boundary condition for parabolic equations that uses the function boundary_normal_flux_function to specify the values of the normal flux at the boundary. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as\n\nboundary_normal_flux_function(x, t, equations)\n\nwhere x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundsCheckCallback","page":"Trixi.jl","title":"Trixi.BoundsCheckCallback","text":"BoundsCheckCallback(; output_directory=\"out\", save_errors=false, interval=1)\n\nSubcell limiting techniques with SubcellLimiterIDP are constructed to adhere certain local or global bounds. To make sure that these bounds are actually met, this callback calculates the maximum deviation from the bounds. The maximum deviation per applied bound is printed to the screen at the end of the simulation. For more insights, when setting save_errors=true the occurring errors are exported every interval time steps during the simulation. Then, the maximum deviations since the last export are saved in \"output_directory/deviations.txt\". The BoundsCheckCallback has to be applied as a stage callback for the SSPRK time integration scheme.\n\nnote: Note\nFor SubcellLimiterIDP, the solution is corrected in the a posteriori correction stage SubcellLimiterIDPCorrection. So, to check the final solution, this bounds check callback must be called after the correction stage.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CarpenterKennedy2N43","page":"Trixi.jl","title":"Trixi.CarpenterKennedy2N43","text":" CarpenterKennedy2N43()\n\nCarpenter, Kennedy (1994) Third order 2N storage RK schemes with error control\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CarpenterKennedy2N54","page":"Trixi.jl","title":"Trixi.CarpenterKennedy2N54","text":"CarpenterKennedy2N54()\n\nThe following structures and methods provide a minimal implementation of the low-storage explicit Runge-Kutta method of\n\nCarpenter, Kennedy (1994) Fourth order 2N storage RK schemes, Solution 3\n\nusing the same interface as OrdinaryDiffEq.jl.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations1D","text":"CompressibleEulerEquations1D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1 rho v_1^2 + p (rho e +p) v_1\nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in one space dimension. Here, rho is the density, v_1 the velocity, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho v_1^2 right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations2D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations2D","text":"CompressibleEulerEquations2D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + p rho v_1 v_2 (rho e +p) v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + p (rho e +p) v_2\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in two space dimensions. Here, rho is the density, v_1, v_2 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations3D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations3D","text":"CompressibleEulerEquations3D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2 rho v_3 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + p rho v_1 v_2 rho v_1 v_3 ( rho e +p) v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + p rho v_1 v_3 ( rho e +p) v_2\nendpmatrix\n+\nfracpartialpartial z\nbeginpmatrix\nrho v_3 rho v_1 v_3 rho v_2 v_3 rho v_3^2 + p ( rho e +p) v_3\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in three space dimensions. Here, rho is the density, v_1, v_2, v_3 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2+v_3^2) right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquationsQuasi1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquationsQuasi1D","text":"CompressibleEulerEquationsQuasi1D(gamma)\n\nThe quasi-1d compressible Euler equations (see Chan et al. DOI: 10.48550/arXiv.2307.12089 for details)\n\nfracpartialpartial t\nbeginpmatrix\na rho a rho v_1 a e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\na rho v_1 a rho v_1^2 a v_1 (e +p)\nendpmatrix\n+ \na fracpartialpartial x\nbeginpmatrix\n0 p 0 \nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in one space dimension. Here, rho is the density, v_1 the velocity, e the specific total energy rather than specific internal energy, a the (possibly) variable nozzle width, and\n\np = (gamma - 1) left( e - frac12 rho v_1^2 right)\n\nthe pressure.\n\nThe nozzle width function a(x) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the compressible Euler equations one can set the nozzle width variable a to one. \n\nIn addition to the unknowns, Trixi.jl currently stores the nozzle width values at the approximation points despite being fixed in time. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the nozzle width must be zero.\nThe nozzle width values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the nozzle width by default.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerMulticomponentEquations1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerMulticomponentEquations1D","text":"CompressibleEulerMulticomponentEquations1D(; gammas, gas_constants)\n\nMulticomponent version of the compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho v_1 rho e rho_1 rho_2 vdots rho_n\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1^2 + p (rho e +p) v_1 rho_1 v_1 rho_2 v_1 vdots rho_n v_1\nendpmatrix\n\n=\nbeginpmatrix\n0 0 0 0 vdots 0\nendpmatrix\n\nfor calorically perfect gas in one space dimension. Here, rho_i is the density of component i, rho=sum_i=1^nrho_i the sum of the individual rho_i, v_1 the velocity, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho v_1^2 right)\n\nthe pressure,\n\ngamma=fracsum_i=1^nrho_i C_vigamma_isum_i=1^nrho_i C_vi\n\ntotal heat capacity ratio, gamma_i heat capacity ratio of component i,\n\nC_vi=fracRgamma_i-1\n\nspecific heat capacity at constant volume of component i.\n\nIn case of more than one component, the specific heat ratios gammas and the gas constants gas_constants should be passed as tuples, e.g., gammas=(1.4, 1.667).\n\nThe remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerMulticomponentEquations2D","page":"Trixi.jl","title":"Trixi.CompressibleEulerMulticomponentEquations2D","text":"CompressibleEulerMulticomponentEquations2D(; gammas, gas_constants)\n\nMulticomponent version of the compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho v_1 rho v_2 rho e rho_1 rho_2 vdots rho_n\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1^2 + p rho v_1 v_2 ( rho e +p) v_1 rho_1 v_1 rho_2 v_1 vdots rho_n v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_1 v_2 rho v_2^2 + p ( rho e +p) v_2 rho_1 v_2 rho_2 v_2 vdots rho_n v_2\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0 0 vdots 0\nendpmatrix\n\nfor calorically perfect gas in two space dimensions. Here, rho_i is the density of component i, rho=sum_i=1^nrho_i the sum of the individual rho_i, v_1, v_2 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2 + v_2^2) right)\n\nthe pressure,\n\ngamma=fracsum_i=1^nrho_i C_vigamma_isum_i=1^nrho_i C_vi\n\ntotal heat capacity ratio, gamma_i heat capacity ratio of component i,\n\nC_vi=fracRgamma_i-1\n\nspecific heat capacity at constant volume of component i.\n\nIn case of more than one component, the specific heat ratios gammas and the gas constants gas_constants in [kJ/(kg*K)] should be passed as tuples, e.g., gammas=(1.4, 1.667).\n\nThe remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion1D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion1D","text":"CompressibleNavierStokesDiffusion1D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations1D.\n\nequations: instance of the CompressibleEulerEquations1D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v rho v^2 + p (rho e + p) v\nendpmatrix\n=\nfracpartialpartial x\nbeginpmatrix\n0 tau tau v - q\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho v^2 right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations1D. The terms on the right hand side of the system above are built from the viscous stress\n\ntau = mu fracpartialpartial x v\n\nwhere the heat flux is\n\nq = -kappa fracpartialpartial x left(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nq = -kappa fracpartialpartial x left(Tright) = -fracgamma mu(gamma - 1)textrmPr fracpartialpartial x left(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn one spatial dimensions we require gradients for two quantities, e.g., primitive quantities\n\nfracpartialpartial x v fracpartialpartial x T\n\nor the entropy variables\n\nfracpartialpartial x w_2 fracpartialpartial x w_3\n\nwhere\n\nw_2 = fracrho v1p w_3 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion2D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion2D","text":"CompressibleNavierStokesDiffusion2D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations2D.\n\nequations: instance of the CompressibleEulerEquations2D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho mathbfv rho e\nendpmatrix\n+\nnabla cdot\nbeginpmatrix\n rho mathbfv rho mathbfvmathbfv^T + p underlineI (rho e + p) mathbfv\nendpmatrix\n=\nnabla cdot\nbeginpmatrix\n0 underlinetau underlinetaumathbfv - mathbfq\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor\n\nunderlinetau = mu left(nablamathbfv + left(nablamathbfvright)^Tright) - frac23 mu left(nablacdotmathbfvright)underlineI\n\nwhere underlineI is the 2times 2 identity matrix and the heat flux is\n\nmathbfq = -kappanablaleft(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nmathbfq = -kappanablaleft(Tright) = -fracgamma mu(gamma - 1)textrmPrnablaleft(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn two spatial dimensions we require gradients for three quantities, e.g., primitive quantities\n\nnabla v_1 nabla v_2 nabla T\n\nor the entropy variables\n\nnabla w_2 nabla w_3 nabla w_4\n\nwhere\n\nw_2 = fracrho v_1p w_3 = fracrho v_2p w_4 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion3D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion3D","text":"CompressibleNavierStokesDiffusion3D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations3D.\n\nequations: instance of the CompressibleEulerEquations3D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho mathbfv rho e\nendpmatrix\n+\nnabla cdot\nbeginpmatrix\n rho mathbfv rho mathbfvmathbfv^T + p underlineI (rho e + p) mathbfv\nendpmatrix\n=\nnabla cdot\nbeginpmatrix\n0 underlinetau underlinetaumathbfv - mathbfq\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2+v_3^2) right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor\n\nunderlinetau = mu left(nablamathbfv + left(nablamathbfvright)^Tright) - frac23 mu left(nablacdotmathbfvright)underlineI\n\nwhere underlineI is the 3times 3 identity matrix and the heat flux is\n\nmathbfq = -kappanablaleft(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nmathbfq = -kappanablaleft(Tright) = -fracgamma mu(gamma - 1)textrmPrnablaleft(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn two spatial dimensions we require gradients for three quantities, e.g., primitive quantities\n\nnabla v_1 nabla v_2 nabla v_3 nabla T\n\nor the entropy variables\n\nnabla w_2 nabla w_3 nabla w_4 nabla w_5\n\nwhere\n\nw_2 = fracrho v_1p w_3 = fracrho v_2p w_4 = fracrho v_3p w_5 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ControllerThreeLevel","page":"Trixi.jl","title":"Trixi.ControllerThreeLevel","text":"ControllerThreeLevel(semi, indicator; base_level=1,\n med_level=base_level, med_threshold=0.0,\n max_level=base_level, max_threshold=1.0)\n\nAn AMR controller based on three levels (in descending order of precedence):\n\nset the target level to max_level if indicator > max_threshold\nset the target level to med_level if indicator > med_threshold; if med_level < 0, set the target level to the current level\nset the target level to base_level otherwise\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ControllerThreeLevelCombined","page":"Trixi.jl","title":"Trixi.ControllerThreeLevelCombined","text":"ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;\n base_level=1,\n med_level=base_level, med_threshold=0.0,\n max_level=base_level, max_threshold=1.0,\n max_threshold_secondary=1.0)\n\nAn AMR controller based on three levels (in descending order of precedence):\n\nset the target level to max_level if indicator_primary > max_threshold\nset the target level to med_level if indicator_primary > med_threshold; if med_level < 0, set the target level to the current level\nset the target level to base_level otherwise\n\nIf indicator_secondary >= max_threshold_secondary, set the target level to max_level.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DG","page":"Trixi.jl","title":"Trixi.DG","text":"DG(; basis, mortar, surface_integral, volume_integral)\n\nCreate a discontinuous Galerkin method. If basis isa LobattoLegendreBasis, this creates a DGSEM.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DGMulti-Tuple{SummationByPartsOperators.AbstractDerivativeOperator}","page":"Trixi.jl","title":"Trixi.DGMulti","text":"DGMulti(approximation_type::AbstractDerivativeOperator;\n element_type::AbstractElemShape,\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n kwargs...)\n\nCreate a summation by parts (SBP) discretization on the given element_type using a tensor product structure based on the 1D SBP derivative operator passed as approximation_type.\n\nFor more info, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMulti-Tuple{}","page":"Trixi.jl","title":"Trixi.DGMulti","text":"DGMulti(; polydeg::Integer,\n element_type::AbstractElemShape,\n approximation_type=Polynomial(),\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n RefElemData_kwargs...)\n\nCreate a discontinuous Galerkin method which uses\n\napproximations of polynomial degree polydeg\nelement type element_type (Tri(), Quad(), Tet(), and Hex() currently supported)\n\nOptional:\n\napproximation_type (default is Polynomial(); SBP() also supported for Tri(), Quad(), and Hex() element types).\nRefElemData_kwargs are additional keyword arguments for RefElemData, such as quad_rule_vol. For more info, see the StartUpDG.jl docs.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh{NDIMS, ...}\n\nDGMultiMesh describes a mesh type which wraps StartUpDG.MeshData and boundary_faces in a dispatchable type. This is intended to store geometric data and connectivities for any type of mesh (Cartesian, affine, curved, structured/unstructured).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Tuple{DGMulti{2, Tri, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{2, Tri, ApproxType}} where {ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, Dict{Symbol, Int64}}","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{2, Tri}, triangulateIO, boundary_dict::Dict{Symbol, Int})\n\ndg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\ntriangulateIO is a TriangulateIO mesh representation\nboundary_dict is a Dict{Symbol, Int} which associates each integer TriangulateIO boundary tag with a Symbol.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType<:SummationByPartsOperators.AbstractPeriodicDerivativeOperator, SurfaceIntegral, VolumeIntegral, Mortar}}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti)\n\nConstructs a single-element DGMultiMesh for a single periodic element given a DGMulti with approximation_type set to a periodic (finite difference) SBP operator from SummationByPartsOperators.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, AbstractArray}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{NDIMS}, vertex_coordinates, EToV;\n is_on_boundary=nothing,\n periodicity=ntuple(_->false, NDIMS)) where {NDIMS}\n\ndg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\nvertex_coordinates is a tuple of vectors containing x,y,... components of the vertex coordinates\nEToV is a 2D array containing element-to-vertex connectivities for each element\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of booleans specifying if the domain is periodic true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, Any}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{NDIMS}, cells_per_dimension, mapping;\n is_on_boundary=nothing,\n periodicity=ntuple(_ -> false, NDIMS), kwargs...) where {NDIMS}\n\nConstructs a Curved() DGMultiMesh with element type dg.basis.element_type.\n\nmapping is a function which maps from a reference [-1, 1]^NDIMS domain to a mapped domain, e.g., xy = mapping(x, y) in 2D.\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti, cells_per_dimension;\n coordinates_min=(-1.0, -1.0), coordinates_max=(1.0, 1.0),\n is_on_boundary=nothing,\n periodicity=ntuple(_ -> false, NDIMS))\n\nConstructs a Cartesian DGMultiMesh with element type dg.basis.element_type. The domain is the tensor product of the intervals [coordinates_min[i], coordinates_max[i]].\n\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, String}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti, filename::String)\n\ndg::DGMulti contains information associated with the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\nfilename is a path specifying a .mesh file generated by HOHQMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGSEM","page":"Trixi.jl","title":"Trixi.DGSEM","text":"DGSEM(; RealT=Float64, polydeg::Integer,\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n mortar=MortarL2(basis))\n\nCreate a discontinuous Galerkin spectral element method (DGSEM) using a LobattoLegendreBasis with polynomials of degree polydeg.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DissipationGlobalLaxFriedrichs","page":"Trixi.jl","title":"Trixi.DissipationGlobalLaxFriedrichs","text":"DissipationGlobalLaxFriedrichs(λ)\n\nCreate a global Lax-Friedrichs dissipation operator with dissipation coefficient λ.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DissipationLocalLaxFriedrichs","page":"Trixi.jl","title":"Trixi.DissipationLocalLaxFriedrichs","text":"DissipationLocalLaxFriedrichs(max_abs_speed=max_abs_speed_naive)\n\nCreate a local Lax-Friedrichs dissipation operator where the maximum absolute wave speed is estimated as max_abs_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to max_abs_speed_naive.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nA callback that couples the acoustic perturbation equations and compressible Euler equations. Must be used in conjunction with SemidiscretizationEulerAcoustics. This callback manages the flow solver - which is always one time step ahead of the acoustics solver - and calculates the acoustic source term after each time step. The linearized Lamb vector is used as the source term, i.e.\n\nmathbfs = -(mathbfomega times barmathbfv\n + barmathbfomega times mathbfv)\n\nwhere mathbfv denotes the velocity, mathbfomega denotes the vorticity, the bar bar(cdot) indicates time-averaged quantities (see AveragingCallback) and prime (cdot) denotes perturbed quantities defined by phi = phi - barphi. Note that the perturbed quantities here are based entirely on the pure flow solution and should not be confused with the state variables of the acoustic perturbation equations.\n\nIn addition, this callback manages the time step size for both solvers and initializes the mean values of the acoustic perturbation equations using results obtained with the AveragingCallback.\n\nMichael Schlottke-Lakemper (2017) A direct-hybrid method for aeroacoustic analysis DOI: 10.18154/RWTH-2017-04082\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback-Tuple{Any, AbstractString, Any, Real, Real}","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback(ode_euler, averaging_file::AbstractString, alg,\n cfl_acoustics::Real, cfl_euler::Real; kwargs...)\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nCreates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_file (see AveragingCallback).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback-Tuple{Any, DiscreteCallback{<:Any, <:AveragingCallback}, Any, Real, Real}","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback(ode_euler,\n averaging_callback::DiscreteCallback{<:Any, <:AveragingCallback},\n alg, cfl_acoustics::Real, cfl_euler::Real; kwargs...)\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nCreates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_callback (see AveragingCallback).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.FDSBP","page":"Trixi.jl","title":"Trixi.FDSBP","text":"FDSBP(D_SBP; surface_integral, volume_integral)\n\nSpecialization of DG methods that uses general summation by parts (SBP) operators from SummationByPartsOperators.jl. In particular, this includes classical finite difference (FD) SBP methods. These methods have the same structure as classical DG methods - local operations on elements with connectivity through interfaces without imposing any continuity constraints.\n\nD_SBP is an SBP derivative operator from SummationByPartsOperators.jl. The other arguments have the same meaning as in DG or DGSEM.\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxHLL","page":"Trixi.jl","title":"Trixi.FluxHLL","text":"FluxHLL(min_max_speed=min_max_speed_naive)\n\nCreate an HLL (Harten, Lax, van Leer) numerical flux where the minimum and maximum wave speeds are estimated as λ_min, λ_max = min_max_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to min_max_speed_naive. Original paper:\n\nAmiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxHydrostaticReconstruction","page":"Trixi.jl","title":"Trixi.FluxHydrostaticReconstruction","text":"FluxHydrostaticReconstruction(numerical_flux, hydrostatic_reconstruction)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nAllow for some kind of hydrostatic reconstruction of the solution state prior to the surface flux computation. This is a particular strategy to ensure that the method remains well-balanced for the shallow water equations, see ShallowWaterEquations1D or ShallowWaterEquations2D.\n\nFor example, the hydrostatic reconstruction from Audusse et al. is implemented in one and two spatial dimensions, see hydrostatic_reconstruction_audusse_etal or the original paper\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\nOther hydrostatic reconstruction techniques are available, particularly to handle wet / dry fronts. A good overview of the development and application of hydrostatic reconstruction can be found in\n\nGuoxian Chen and Sebastian Noelle A unified surface-gradient and hydrostatic reconstruction scheme for the shallow water equations (2021) RWTH Aachen preprint\nAndreas Buttinger-Kreuzhuber, Zsolt Horváth, Sebastian Noelle, Günter Blöschl and Jürgen Waser (2019) A fast second-order shallow water scheme on two-dimensional structured grids over abrupt topography DOI: 10.1016/j.advwatres.2019.03.010\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxLMARS","page":"Trixi.jl","title":"Trixi.FluxLMARS","text":"FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nLow Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.\n\nReferences:\n\nXi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxLaxFriedrichs","page":"Trixi.jl","title":"Trixi.FluxLaxFriedrichs","text":"FluxLaxFriedrichs(max_abs_speed=max_abs_speed_naive)\n\nLocal Lax-Friedrichs (Rusanov) flux with maximum wave speed estimate provided by max_abs_speed, cf. DissipationLocalLaxFriedrichs and max_abs_speed_naive.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxPlusDissipation","page":"Trixi.jl","title":"Trixi.FluxPlusDissipation","text":"FluxPlusDissipation(numerical_flux, dissipation)\n\nCombine a numerical_flux with a dissipation operator to create a new numerical flux.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxRotated","page":"Trixi.jl","title":"Trixi.FluxRotated","text":"FluxRotated(numerical_flux)\n\nCompute a numerical_flux flux in direction of a normal vector by rotating the solution, computing the numerical flux in x-direction, and rotating the calculated flux back.\n\nRequires a rotationally invariant equation with equation-specific functions rotate_to_x and rotate_from_x.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxUpwind","page":"Trixi.jl","title":"Trixi.FluxUpwind","text":"FluxUpwind(splitting)\n\nA numerical flux f(u_left, u_right) = f⁺(u_left) + f⁻(u_right) based on flux vector splitting.\n\nThe SurfaceIntegralUpwind with a given splitting is equivalent to the SurfaceIntegralStrongForm with FluxUpwind(splitting) as numerical flux (up to floating point differences).\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.GlmSpeedCallback","page":"Trixi.jl","title":"Trixi.GlmSpeedCallback","text":"GlmSpeedCallback(; glm_scale=0.5, cfl)\n\nUpdate the divergence cleaning wave speed c_h according to the time step computed in StepsizeCallback for the ideal GLM-MHD equations. The cfl number should be set to the same value as for the time step size calculation. The glm_scale ensures that the GLM wave speed is lower than the fastest physical waves in the MHD solution and should thus be set to a value within the interval [0,1]. Note that glm_scale = 0 deactivates the divergence cleaning.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.GradientVariablesPrimitive","page":"Trixi.jl","title":"Trixi.GradientVariablesPrimitive","text":"GradientVariablesPrimitive and GradientVariablesEntropy are gradient variable type parameters for CompressibleNavierStokesDiffusion1D. By default, the gradient variables are set to be GradientVariablesPrimitive. Specifying GradientVariablesEntropy instead uses the entropy variable formulation from\n\nHughes, Mallet, Franca (1986) A new finite element formulation for computational fluid dynamics: I. Symmetric forms of the compressible Euler and Navier-Stokes equations and the second law of thermodynamics. https://doi.org/10.1016/0045-7825(86)90127-1\n\nUnder GradientVariablesEntropy, the Navier-Stokes discretization is provably entropy stable.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HypDiffN3Erk3Sstar52","page":"Trixi.jl","title":"Trixi.HypDiffN3Erk3Sstar52","text":"HypDiffN3Erk3Sstar52()\n\nFive stage, second-order accurate explicit Runge-Kutta scheme with stability region optimized for the hyperbolic diffusion equation with LLF flux and polynomials of degree polydeg=3.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations1D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations1D","text":"HyperbolicDiffusionEquations1D\n\nThe linear hyperbolic diffusion equations in one space dimension. A description of this system can be found in Sec. 2.5 of the book\n\nMasatsuka (2013) I Do Like CFD, Too: Vol 1. Freely available at http://www.cfdbooks.com/\n\nFurther analysis can be found in the paper\n\nNishikawa (2007) A first-order system approach for diffusion equation. I: Second-order residual-distribution schemes DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations2D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations2D","text":"HyperbolicDiffusionEquations2D\n\nThe linear hyperbolic diffusion equations in two space dimensions. A description of this system can be found in Sec. 2.5 of the book \"I Do Like CFD, Too: Vol 1\". The book is freely available at http://www.cfdbooks.com/ and further analysis can be found in the paper by Nishikawa DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations3D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations3D","text":"HyperbolicDiffusionEquations3D\n\nThe linear hyperbolic diffusion equations in three space dimensions. A description of this system can be found in Sec. 2.5 of the book \"I Do Like CFD, Too: Vol 1\". The book is freely available at http://www.cfdbooks.com/ and further analysis can be found in the paper by Nishikawa DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations1D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations1D","text":"IdealGlmMhdEquations1D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in one space dimension.\n\nnote: Note\nThere is no divergence cleaning variable psi because the divergence-free constraint is satisfied trivially in one spatial dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations2D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations2D","text":"IdealGlmMhdEquations2D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in two space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations3D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations3D","text":"IdealGlmMhdEquations3D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in three space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdMulticomponentEquations1D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdMulticomponentEquations1D","text":"IdealGlmMhdMulticomponentEquations1D\n\nThe ideal compressible multicomponent GLM-MHD equations in one space dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdMulticomponentEquations2D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdMulticomponentEquations2D","text":"IdealGlmMhdMulticomponentEquations2D\n\nThe ideal compressible multicomponent GLM-MHD equations in two space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorHennemannGassner","page":"Trixi.jl","title":"Trixi.IndicatorHennemannGassner","text":"IndicatorHennemannGassner(equations::AbstractEquations, basis;\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable)\nIndicatorHennemannGassner(semi::AbstractSemidiscretization;\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable)\n\nIndicator used for shock-capturing (when passing the equations and the basis) or adaptive mesh refinement (AMR, when passing the semi).\n\nSee also VolumeIntegralShockCapturingHG.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorHennemannGassnerShallowWater","page":"Trixi.jl","title":"Trixi.IndicatorHennemannGassnerShallowWater","text":"IndicatorHennemannGassnerShallowWater(equations::AbstractEquations, basis;\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable)\n\nModified version of the IndicatorHennemannGassner indicator used for shock-capturing for shallow water equations. After the element-wise values for the blending factors are computed an additional check is made to see if the element is partially wet. In this case, partially wet elements are set to use the pure finite volume scheme that is guaranteed to be well-balanced for this wet/dry transition state of the flow regime.\n\nSee also VolumeIntegralShockCapturingHG.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorLöhner","page":"Trixi.jl","title":"Trixi.IndicatorLöhner","text":"IndicatorLöhner (equivalent to IndicatorLoehner)\n\nIndicatorLöhner(equations::AbstractEquations, basis;\n f_wave=0.2, variable)\nIndicatorLöhner(semi::AbstractSemidiscretization;\n f_wave=0.2, variable)\n\nAMR indicator adapted from a FEM indicator by Löhner (1987), also used in the FLASH code as standard AMR indicator. The indicator estimates a weighted second derivative of a specified variable locally.\n\nWhen constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.\n\nReferences\n\nLöhner (1987) \"An adaptive finite element scheme for transient problems in CFD\" doi: 10.1016/0045-7825(87)90098-3\nhttps://flash.rochester.edu/site/flashcode/usersupport/flash4ug_4p62/node59.html#SECTION05163100000000000000\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorMax","page":"Trixi.jl","title":"Trixi.IndicatorMax","text":"IndicatorMax(equations::AbstractEquations, basis; variable)\nIndicatorMax(semi::AbstractSemidiscretization; variable)\n\nA simple indicator returning the maximum of variable in an element. When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.InviscidBurgersEquation1D","page":"Trixi.jl","title":"Trixi.InviscidBurgersEquation1D","text":"InviscidBurgersEquation1D\n\nThe inviscid Burgers' equation\n\npartial_t u + frac12 partial_1 u^2 = 0\n\nin one space dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.Isothermal","page":"Trixi.jl","title":"Trixi.Isothermal","text":"struct Isothermal\n\nUsed to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and return a scalar value for the temperature at point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion1D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion1D","text":"LaplaceDiffusion1D(diffusivity, equations)\n\nLaplaceDiffusion1D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion2D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion2D","text":"LaplaceDiffusion2D(diffusivity, equations)\n\nLaplaceDiffusion2D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion3D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion3D","text":"LaplaceDiffusion3D(diffusivity, equations)\n\nLaplaceDiffusion3D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LatticeBoltzmannEquations2D","page":"Trixi.jl","title":"Trixi.LatticeBoltzmannEquations2D","text":"LatticeBoltzmannEquations2D(; Ma, Re, collision_op=collision_bgk,\n c=1, L=1, rho0=1, u0=nothing, nu=nothing)\n\nThe Lattice-Boltzmann equations\n\npartial_t u_alpha + v_alpha1 partial_1 u_alpha + v_alpha2 partial_2 u_alpha = 0\n\nin two space dimensions for the D2Q9 scheme.\n\nThe characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).\n\nThe nine discrete velocity directions of the D2Q9 scheme are sorted as follows [4]:\n\n 6 2 5 y\n ┌───┼───┐ │\n │ │ │\n 3 ┼ 9 ┼ 1 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 7 4 8 z\n\nNote that usually the velocities are numbered from 0 to 8, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 9, where 1 through 8 correspond to the velocity directions in [4] and 9 is the zero velocity.\n\nThe corresponding opposite directions are:\n\n1 ←→ 3\n2 ←→ 4\n3 ←→ 1\n4 ←→ 2\n5 ←→ 7\n6 ←→ 8\n7 ←→ 5\n8 ←→ 6\n9 ←→ 9\n\nThe main sources for the base implementation were\n\nMisun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024\nKarsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.\nDieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0\nDieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LatticeBoltzmannEquations3D","page":"Trixi.jl","title":"Trixi.LatticeBoltzmannEquations3D","text":"LatticeBoltzmannEquations3D(; Ma, Re, collision_op=collision_bgk,\n c=1, L=1, rho0=1, u0=nothing, nu=nothing)\n\nThe Lattice-Boltzmann equations\n\npartial_t u_alpha + v_alpha1 partial_1 u_alpha + v_alpha2 partial_2 u_alpha + v_alpha3 partial_3 u_alpha = 0\n\nin three space dimensions for the D3Q27 scheme.\n\nThe characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).\n\nThe twenty-seven discrete velocity directions of the D3Q27 scheme are sorted as follows [4]:\n\nplane at z = -1:\n 24 17 21 y\n ┌───┼───┐ │\n │ │ │\n 10 ┼ 6 ┼ 15 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 20 12 26 z\nplane at z = 0:\n 14 3 7 y\n ┌───┼───┐ │\n │ │ │\n 2 ┼ 27 ┼ 1 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 8 4 13 z\nplane at z = +1:\n 25 11 19 y\n ┌───┼───┐ │\n │ │ │\n 16 ┼ 5 ┼ 9 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 22 18 23 z\n\nNote that usually the velocities are numbered from 0 to 26, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 27, where 1 through 26 correspond to the velocity directions in [4] and 27 is the zero velocity.\n\nThe corresponding opposite directions are:\n\n1 ←→ 2\n2 ←→ 1\n3 ←→ 4\n4 ←→ 3\n5 ←→ 6\n6 ←→ 5\n7 ←→ 8\n8 ←→ 7\n9 ←→ 10\n10 ←→ 9\n11 ←→ 12\n12 ←→ 11\n13 ←→ 14\n14 ←→ 13\n15 ←→ 16\n16 ←→ 15\n17 ←→ 18\n18 ←→ 17\n19 ←→ 20\n20 ←→ 19\n21 ←→ 22\n22 ←→ 21\n23 ←→ 24\n24 ←→ 23\n25 ←→ 26\n26 ←→ 25\n27 ←→ 27\n\nThe main sources for the base implementation were\n\nMisun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024\nKarsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.\nDieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0\nDieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation1D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation1D","text":"LinearScalarAdvectionEquation1D\n\nThe linear scalar advection equation\n\npartial_t u + a partial_1 u = 0\n\nin one space dimension with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation2D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation2D","text":"LinearScalarAdvectionEquation2D\n\nThe linear scalar advection equation\n\npartial_t u + a_1 partial_1 u + a_2 partial_2 u = 0\n\nin two space dimensions with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation3D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation3D","text":"LinearScalarAdvectionEquation3D\n\nThe linear scalar advection equation\n\npartial_t u + a_1 partial_1 u + a_2 partial_2 u + a_3 partial_3 u = 0\n\nin three space dimensions with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearizedEulerEquations2D","page":"Trixi.jl","title":"Trixi.LinearizedEulerEquations2D","text":"LinearizedEulerEquations2D(v_mean_global, c_mean_global, rho_mean_global)\n\nLinearized euler equations in two space dimensions. The equations are given by\n\npartial_t\nbeginpmatrix\n rho v_1 v_2 p\nendpmatrix\n+\npartial_x\nbeginpmatrix\n barrho v_1 + barv_1 rho barv_1 v_1 + fracpbarrho barv_1 v_2 barv_1 p + c^2 barrho v_1\nendpmatrix\n+\npartial_y\nbeginpmatrix\n barrho v_2 + barv_2 rho barv_2 v_1 barv_2 v_2 + fracpbarrho barv_2 p + c^2 barrho v_2\nendpmatrix\n=\nbeginpmatrix\n 0 0 0 0\nendpmatrix\n\nThe bar bar(cdot) indicates uniform mean flow variables and c is the speed of sound. The unknowns are the acoustic velocities v = (v_1 v_2), the pressure p and the density rho.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LobattoLegendreBasis","page":"Trixi.jl","title":"Trixi.LobattoLegendreBasis","text":"LobattoLegendreBasis([RealT=Float64,] polydeg::Integer)\n\nCreate a nodal Lobatto-Legendre basis for polynomials of degree polydeg.\n\nFor the special case polydeg=0 the DG method reduces to a finite volume method. Therefore, this function sets the center point of the cell as single node.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NoSlip","page":"Trixi.jl","title":"Trixi.NoSlip","text":"struct NoSlip\n\nUse to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and should return a SVector{NDIMS} whose entries are the velocity vector at a point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NonConservativeLocal","page":"Trixi.jl","title":"Trixi.NonConservativeLocal","text":"NonConservativeLocal()\n\nStruct used for multiple dispatch on non-conservative flux functions in the format of \"local * symmetric\". When the argument nonconservative_type is of type NonConservativeLocal, the function returns the local part of the non-conservative term.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NonConservativeSymmetric","page":"Trixi.jl","title":"Trixi.NonConservativeSymmetric","text":"NonConservativeSymmetric()\n\nStruct used for multiple dispatch on non-conservative flux functions in the format of \"local * symmetric\". When the argument nonconservative_type is of type NonConservativeSymmetric, the function returns the symmetric part of the non-conservative term.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.P4estMesh","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nAn unstructured curved mesh based on trees that uses the C library p4est to manage trees and mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.P4estMesh-Tuple{Any}","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh(trees_per_dimension; polydeg,\n mapping=nothing, faces=nothing, coordinates_min=nothing, coordinates_max=nothing,\n RealT=Float64, initial_refinement_level=0, periodicity=true, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true)\n\nCreate a structured curved P4estMesh of the specified size.\n\nThere are three ways to map the mesh to the physical domain.\n\nDefine a mapping that maps the hypercube [-1, 1]^n.\nSpecify a Tuple faces of functions that parametrize each face.\nCreate a rectangular mesh by specifying coordinates_min and coordinates_max.\n\nNon-periodic boundaries will be called :x_neg, :x_pos, :y_neg, :y_pos, :z_neg, :z_pos.\n\nArguments\n\ntrees_per_dimension::NTupleE{NDIMS, Int}: the number of trees in each dimension.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\nmapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.\nfaces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.P4estMesh-Union{Tuple{String}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh{NDIMS}(meshfile::String;\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true)\n\nMain mesh constructor for the P4estMesh that imports an unstructured, conforming mesh from an Abaqus mesh file (.inp). Each element of the conforming mesh parsed from the meshfile is created as a p4est tree datatype.\n\nTo create a curved unstructured mesh P4estMesh two strategies are available:\n\np4est_mesh_from_hohqmesh_abaqus: High-order, curved boundary information created by HOHQMesh.jl is available in the meshfile. The mesh polynomial degree polydeg of the boundaries is provided from the meshfile. The computation of the mapped tree coordinates is done with transfinite interpolation with linear blending similar to UnstructuredMesh2D. Boundary name information is also parsed from the meshfile such that different boundary conditions can be set at each named boundary on a given tree.\np4est_mesh_from_standard_abaqus: By default, with mapping=nothing and polydeg=1, this creates a straight-sided from the information parsed from the meshfile. If a mapping function is specified then it computes the mapped tree coordinates via polynomial interpolants with degree polydeg. The mesh created by this function will only have one boundary :all, as distinguishing different physical boundaries is non-trivial.\n\nNote that the mapping and polydeg keyword arguments are only used by the p4est_mesh_from_standard_abaqus function. The p4est_mesh_from_hohqmesh_abaqus function obtains the mesh polydeg directly from the meshfile and constructs the transfinite mapping internally.\n\nThe particular strategy is selected according to the header present in the meshfile where the constructor checks whether or not the meshfile was created with HOHQMesh.jl. If the Abaqus file header is not present in the meshfile then the P4estMesh is created with the function p4est_mesh_from_standard_abaqus.\n\nThe default keyword argument initial_refinement_level=0 corresponds to a forest where the number of trees is the same as the number of elements in the original meshfile. Increasing the initial_refinement_level allows one to uniformly refine the base mesh given in the meshfile to create a forest with more trees before the simulation begins. For example, if a two-dimensional base mesh contains 25 elements then setting initial_refinement_level=1 creates an initial forest of 2^2 * 25 = 100 trees.\n\nArguments\n\nmeshfile::String: an uncurved Abaqus mesh file that can be imported by p4est.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ParametersEulerGravity","page":"Trixi.jl","title":"Trixi.ParametersEulerGravity","text":"ParametersEulerGravity(; background_density=0.0,\n gravitational_constant=1.0,\n cfl=1.0,\n resid_tol=1.0e-4,\n n_iterations_max=10^4,\n timestep_gravity=timestep_gravity_erk52_3Sstar!)\n\nSet up parameters for the gravitational part of a SemidiscretizationEulerGravity.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ParsaniKetchesonDeconinck3Sstar32","page":"Trixi.jl","title":"Trixi.ParsaniKetchesonDeconinck3Sstar32","text":"ParsaniKetchesonDeconinck3Sstar32()\n\nParsani, Ketcheson, Deconinck (2013) Optimized explicit RK schemes for the spectral difference method applied to wave propagation problems DOI: 10.1137/120885899\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ParsaniKetchesonDeconinck3Sstar94","page":"Trixi.jl","title":"Trixi.ParsaniKetchesonDeconinck3Sstar94","text":"ParsaniKetchesonDeconinck3Sstar94()\n\nParsani, Ketcheson, Deconinck (2013) Optimized explicit RK schemes for the spectral difference method applied to wave propagation problems DOI: 10.1137/120885899\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PerformanceCounter","page":"Trixi.jl","title":"Trixi.PerformanceCounter","text":"PerformanceCounter()\n\nA PerformanceCounter can be used to track the runtime performance of some calls. Add a new runtime measurement via put!(counter, runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PerformanceCounterList","page":"Trixi.jl","title":"Trixi.PerformanceCounterList","text":"PerformanceCounterList{N}()\n\nA PerformanceCounterList{N} can be used to track the runtime performance of calls to multiple functions, adding them up. Add a new runtime measurement via put!(counter.counters[i], runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PlotData1D","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D\n\nHolds all relevant data for creating 1D plots of multiple solution variables and to visualize the mesh.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PlotData1D-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D(u, semi [or mesh, equations, solver, cache];\n solution_variables=nothing, nvisnodes=nothing)\n\nCreate a new PlotData1D object that can be used for visualizing 1D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.\n\nnvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.\n\nWhen visualizing data from a two-dimensional simulation, a 1D slice is extracted for plotting. slice specifies the axis along which the slice is extracted and may be :x, or :y. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0). Both of these values are ignored when visualizing 1D data. This applies analogously to three-dimensional simulations, where slice may be :xy, :xz, or :yz.\n\nAnother way to visualize 2D/3D data is by creating a plot along a given curve. This is done with the keyword argument curve. It can be set to a list of 2D/3D points which define the curve. When using curve any other input from slice or point will be ignored.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData1D-Tuple{Union{ODESolution{T, N, uType, uType2, DType, tType, rateType, P} where {T, N, uType, uType2, DType, tType, rateType, P<:(ODEProblem{uType_, tType_, isinplace, P_} where {uType_, tType_, isinplace, P_<:Trixi.AbstractSemidiscretization})}, Trixi.TimeIntegratorSolution}}","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D(sol; kwargs...)\n\nCreate a PlotData1D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2DCartesian","page":"Trixi.jl","title":"Trixi.PlotData2DCartesian","text":"PlotData2D\n\nHolds all relevant data for creating 2D plots of multiple solution variables and to visualize the mesh.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PolytropicEulerEquations2D","page":"Trixi.jl","title":"Trixi.PolytropicEulerEquations2D","text":"PolytropicEulerEquations2D(gamma, kappa)\n\nThe polytropic Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + kapparho^gamma rho v_1 v_2\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + kapparho^gamma\nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in two space dimensions. Here, rho is the density and v_1 andv_2 the velocities and\n\np = kapparho^gamma\n\nthe pressure, which we replaced using this relation.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PositivityPreservingLimiterShallowWater","page":"Trixi.jl","title":"Trixi.PositivityPreservingLimiterShallowWater","text":"PositivityPreservingLimiterShallowWater(; variables)\n\nThe limiter is specifically designed for the shallow water equations. It is applied to all scalar variables in their given order using the defined threshold_limiter from the ShallowWaterEquations1D struct or the ShallowWaterEquations2D struct to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.\n\nAs opposed to the standard version of the PositivityPreservingLimiterZhangShu, nodes with a water height below the threshold_limiter are treated in a special way. To avoid numerical problems caused by velocities close to zero, the velocity is cut off, such that the node can be identified as \"dry\". The special feature of the ShallowWaterEquations used here is that the bottom topography is stored as an additional quantity in the solution vector u. However, the value of the bottom topography should not be changed. That is why, it is not limited.\n\nAfter the limiting process is applied to all degrees of freedom, for safety reasons, the threshold_limiter is applied again on all the DG nodes in order to avoid water height below. In the case where the cell mean value is below the threshold before applying the limiter, there could still be dry nodes afterwards due to the logic of the limiter.\n\nThis fully-discrete positivity-preserving limiter is based on the work of\n\nZhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PositivityPreservingLimiterZhangShu","page":"Trixi.jl","title":"Trixi.PositivityPreservingLimiterZhangShu","text":"PositivityPreservingLimiterZhangShu(; threshold, variables)\n\nThe fully-discrete positivity-preserving limiter of\n\nZhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153\n\nThe limiter is applied to all scalar variables in their given order using the associated thresholds to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SaveRestartCallback","page":"Trixi.jl","title":"Trixi.SaveRestartCallback","text":"SaveRestartCallback(; interval=0,\n save_final_restart=true,\n output_directory=\"out\")\n\nSave the current numerical solution in a restart file every interval time steps.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SaveSolutionCallback","page":"Trixi.jl","title":"Trixi.SaveSolutionCallback","text":"SaveSolutionCallback(; interval::Integer=0,\n dt=nothing,\n save_initial_solution=true,\n save_final_solution=true,\n output_directory=\"out\",\n solution_variables=cons2prim)\n\nSave the current numerical solution in regular intervals. Either pass interval to save every interval time steps or pass dt to save in intervals of dt in terms of integration time by adding additional (shortened) time steps where necessary (note that this may change the solution). solution_variables can be any callable that converts the conservative variables at a single point to a set of solution variables. The first parameter passed to solution_variables will be the set of conservative variables and the second parameter is the equation struct.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationCoupled","page":"Trixi.jl","title":"Trixi.SemidiscretizationCoupled","text":"SemidiscretizationCoupled\n\nA struct used to bundle multiple semidiscretizations. semidiscretize will return an ODEProblem that synchronizes time steps between the semidiscretizations. Each call of rhs! will call rhs! for each semidiscretization individually. The semidiscretizations can be coupled by gluing meshes together using BoundaryConditionCoupled.\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationCoupled-Tuple","page":"Trixi.jl","title":"Trixi.SemidiscretizationCoupled","text":"SemidiscretizationCoupled(semis...)\n\nCreate a coupled semidiscretization that consists of the semidiscretizations passed as arguments.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerAcoustics","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerAcoustics","text":"SemidiscretizationEulerAcoustics(semi_acoustics::SemiAcoustics, semi_euler::SemiEuler;\n source_region=x->true, weights=x->1.0)\n\nwarning: Experimental code\nThis semidiscretization is experimental and may change in any future release.\n\nConstruct a semidiscretization of the acoustic perturbation equations that is coupled with the compressible Euler equations via source terms for the perturbed velocity. Both semidiscretizations have to use the same mesh and solvers with a shared basis. The coupling region is described by a function source_region that maps the coordinates of a single node to true or false depending on whether the point lies within the coupling region or not. A weighting function weights that maps coordinates to weights is applied to the acoustic source terms. Note that this semidiscretization should be used in conjunction with EulerAcousticsCouplingCallback and only works in two dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerGravity","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerGravity","text":"SemidiscretizationEulerGravity\n\nA struct containing everything needed to describe a spatial semidiscretization of a the compressible Euler equations with self-gravity, reformulating the Poisson equation for the gravitational potential as steady-state problem of the hyperblic diffusion equations.\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) \"A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics\" arXiv: 2008.10593\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerGravity-Union{Tuple{SemiGravity}, Tuple{SemiEuler}, Tuple{Mesh}, Tuple{SemiEuler, SemiGravity, Any}} where {Mesh, SemiEuler<:(SemidiscretizationHyperbolic{Mesh, <:Trixi.AbstractCompressibleEulerEquations}), SemiGravity<:(SemidiscretizationHyperbolic{Mesh, <:Trixi.AbstractHyperbolicDiffusionEquations})}","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerGravity","text":"SemidiscretizationEulerGravity(semi_euler::SemiEuler, semi_gravity::SemiGravity, parameters)\n\nConstruct a semidiscretization of the compressible Euler equations with self-gravity. parameters should be given as ParametersEulerGravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolic","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolic","text":"SemidiscretizationHyperbolic\n\nA struct containing everything needed to describe a spatial semidiscretization of a hyperbolic conservation law.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolic-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolic","text":"SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;\n source_terms=nothing,\n boundary_conditions=boundary_condition_periodic,\n RealT=real(solver),\n uEltype=RealT,\n initial_cache=NamedTuple())\n\nConstruct a semidiscretization of a hyperbolic PDE.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolicParabolic","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolicParabolic","text":"SemidiscretizationHyperbolicParabolic\n\nA struct containing everything needed to describe a spatial semidiscretization of a mixed hyperbolic-parabolic conservation law.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolicParabolic-Tuple{Any, Tuple, Any, Any}","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolicParabolic","text":"SemidiscretizationHyperbolicParabolic(mesh, both_equations, initial_condition, solver;\n solver_parabolic=default_parabolic_solver(),\n source_terms=nothing,\n both_boundary_conditions=(boundary_condition_periodic, boundary_condition_periodic),\n RealT=real(solver),\n uEltype=RealT,\n both_initial_caches=(NamedTuple(), NamedTuple()))\n\nConstruct a semidiscretization of a hyperbolic-parabolic PDE.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ShallowWaterEquations1D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquations1D","text":"ShallowWaterEquations1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)\n\nShallow water equations (SWE) in one space dimension. The equations are given by\n\nbeginaligned\n fracpartial hpartial t + fracpartialpartial x(h v) = 0 \n fracpartialpartial t(h v) + fracpartialpartial xleft(h v^2 + fracg2h^2right)\n + g h fracpartial bpartial x = 0\nendaligned\n\nThe unknown quantities of the SWE are the water height h and the velocity v. The gravitational constant is denoted by g and the (possibly) variable bottom topography function b(x). Conservative variable water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nAlso, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is \"wet\" before calculating the numerical flux.\n\nThe bottom topography function b(x) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.\n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography by default.\n\nReferences for the SWE are many but a good introduction is available in Chapter 13 of the book:\n\nRandall J. LeVeque (2002) Finite Volume Methods for Hyperbolic Problems DOI: 10.1017/CBO9780511791253\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterEquations2D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquations2D","text":"ShallowWaterEquations2D(; gravity, H0 = 0, threshold_limiter = nothing, threshold_wet = nothing)\n\nShallow water equations (SWE) in two space dimensions. The equations are given by\n\nbeginaligned\n fracpartial hpartial t + fracpartialpartial x(h v_1)\n + fracpartialpartial y(h v_2) = 0 \n fracpartialpartial t(h v_1) + fracpartialpartial xleft(h v_1^2 + fracg2h^2right)\n + fracpartialpartial y(h v_1 v_2) + g h fracpartial bpartial x = 0 \n fracpartialpartial t(h v_2) + fracpartialpartial x(h v_1 v_2)\n + fracpartialpartial yleft(h v_2^2 + fracg2h^2right) + g h fracpartial bpartial y = 0\nendaligned\n\nThe unknown quantities of the SWE are the water height h and the velocities mathbfv = (v_1 v_2)^T. The gravitational constant is denoted by g and the (possibly) variable bottom topography function b(xy). Conservative variable water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nAlso, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is \"wet\" before calculating the numerical flux.\n\nThe bottom topography function b(xy) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.\n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography by default.\n\nReferences for the SWE are many but a good introduction is available in Chapter 13 of the book:\n\nRandall J. LeVeque (2002) Finite Volume Methods for Hyperbolic Problems DOI: 10.1017/CBO9780511791253\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterEquationsQuasi1D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquationsQuasi1D","text":"ShallowWaterEquationsQuasi1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)\n\nThe quasi-1D shallow water equations (SWE). The equations are given by\n\nbeginaligned\n fracpartialpartial t(a h) + fracpartialpartial x(a h v) = 0 \n fracpartialpartial t(a h v) + fracpartialpartial x(a h v^2)\n + g a h fracpartialpartial x(h + b) = 0\nendaligned\n\nThe unknown quantities of the Quasi-1D SWE are the water height h and the scaled velocity v. The gravitational constant is denoted by g, the (possibly) variable bottom topography function b(x), and (possibly) variable channel width a(x). The water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nAlso, there are two thresholds which prevent numerical problems as well as instabilities. Both of them do not have to be passed, as default values are defined within the struct. The first one, threshold_limiter, is used in PositivityPreservingLimiterShallowWater on the water height, as a (small) shift on the initial condition and cutoff before the next time step. The second one, threshold_wet, is applied on the water height to define when the flow is \"wet\" before calculating the numerical flux.\n\nThe bottom topography function b(x) and channel width a(x) are set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero and a to one. \n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography and channel width values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography and channel width must be zero.\nThe bottom topography and channel width values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography and channel width by default.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterTwoLayerEquations1D","page":"Trixi.jl","title":"Trixi.ShallowWaterTwoLayerEquations1D","text":"ShallowWaterTwoLayerEquations1D(gravity, H0, rho_upper, rho_lower)\n\nTwo-Layer Shallow Water equations (2LSWE) in one space dimension. The equations are given by\n\nbeginalignat*4\nfracpartialpartial th_upper\n+ fracpartialpartial xleft(h_upper v_1upperright)\n= 0 \nfracpartialpartial tleft(h_upperv_1upperright)\n+ fracpartialpartial xleft(h_upperv_1upper^2 + dfracgh_upper^22right)\n= -gh_upperfracpartialpartial xleft(b+h_lowerright)\nfracpartialpartial th_lower\n+ fracpartialpartial xleft(h_lowerv_1lowerright)\n= 0 \nfracpartialpartial tleft(h_lowerv_1lowerright)\n+ fracpartialpartial xleft(h_lowerv_1lower^2 + dfracgh_lower^22right)\n= -gh_lowerfracpartialpartial xleft(b+dfracrho_upperrho_lowerh_upperright)\nendalignat*\n\nThe unknown quantities of the 2LSWE are the water heights of the {lower} layer h_lower and the {upper} layer h_upper with respective velocities v_1upper and v_1lower. The gravitational constant is denoted by g, the layer densitites by rho_upperand rho_lower and the (possibly) variable bottom topography function b(x). The conservative variable water height h_lower is measured from the bottom topography b and h_upper relative to h_lower, therefore one also defines the total water heights as H_upper = h_upper + h_upper + b and H_lower = h_lower + b.\n\nThe densities must be chosen such that rho_upper rho_lower, to make sure that the heavier fluid rho_lower is in the bottom layer and the lighter fluid rho_upper in the {upper} layer.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nThe bottom topography function b(x) is set inside the initial condition routine for a particular problem setup.\n\nIn addition to the unknowns, Trixi currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi's visualization tools will visualize the bottom topography by default.\n\nA good introduction for the 2LSWE is available in Chapter 12 of the book:\n\nBenoit Cushman-Roisin (2011)\nIntroduction to geophyiscal fluid dynamics: physical and numerical aspects\nhttps://www.sciencedirect.com/bookseries/international-geophysics/vol/101/suppl/C\nISBN: 978-0-12-088759-0\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterTwoLayerEquations2D","page":"Trixi.jl","title":"Trixi.ShallowWaterTwoLayerEquations2D","text":"ShallowWaterTwoLayerEquations2D(gravity, H0, rho_upper, rho_lower)\n\nTwo-Layer Shallow water equations (2LSWE) in two space dimension. The equations are given by\n\nbeginalignat*8\nfracpartialpartial th_upper\n+ fracpartialpartial xleft(h_upper v_1upperright)\n+ fracpartialpartial yleft(h_upper v_2upperright) quad\n= quad 0 \nfracpartialpartial tleft(h_upper v_1upperright)\n+ fracpartialpartial xleft(h_upper v_1upper^2 + fracgh_upper^22right)\n+ fracpartialpartial yleft(h_upper v_1upper v_2upperright) quad\n= -gh_upperfracpartialpartial xleft(b+h_lowerright) \nfracpartialpartial tleft(h_upper v_2upperright)\n+ fracpartialpartial xleft(h_upper v_1upper v_2upperright)\n+ fracpartialpartial yleft(h_upper v_2upper^2 + fracgh_upper^22right)\n= -gh_upperfracpartialpartial yleft(b+h_lowerright)\nfracpartialpartial th_lower\n+ fracpartialpartial xleft(h_lower v_1lowerright)\n+ fracpartialpartial yleft(h_lower v_2lowerright)\n= quad 0 \nfracpartialpartial tleft(h_lower v_1lowerright)\n+ fracpartialpartial xleft(h_lower v_1lower^2 + fracgh_lower^22right)\n+ fracpartialpartial yleft(h_lower v_1lower v_2lowerright)\n= -gh_lowerfracpartialpartial xleft(b+fracrho_upperrho_lower h_upperright)\nfracpartialpartial tleft(h_lower v_2lowerright)\n+ fracpartialpartial xleft(h_lower v_1lower v_2lowerright)\n+ fracpartialpartial yleft(h_lower v_2lower^2 + fracgh_lower^22right)\n= -gh_lowerfracpartialpartial yleft(b+fracrho_upperrho_lower h_upperright)\nendalignat*\n\nThe unknown quantities of the 2LSWE are the water heights of the lower layer h_lower and the upper layer h_upper and the respective velocities in x-direction v_1lower and v_1upper and in y-direction v_2lower and v_2upper. The gravitational constant is denoted by g, the layer densitites by rho_upperand rho_lower and the (possibly) variable bottom topography function by b(x). Conservative variable water height h_lower is measured from the bottom topography b and h_upper relative to h_lower, therefore one also defines the total water heights as H_lower = h_lower + b and H_upper = h_upper + h_lower + b.\n\nThe densities must be chosen such that rho_upper rho_lower, to make sure that the heavier fluid rho_lower is in the bottom layer and the lighter fluid rho_upper in the upper layer.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nThe bottom topography function b(x) is set inside the initial condition routine for a particular problem setup.\n\nIn addition to the unknowns, Trixi currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi's visualization tools will visualize the bottom topography by default.\n\nA good introduction for the 2LSWE is available in Chapter 12 of the book:\n\nBenoit Cushman-Roisin (2011)\nIntroduction to geophyiscal fluid dynamics: physical and numerical aspects\nhttps://www.sciencedirect.com/bookseries/international-geophysics/vol/101/suppl/C\nISBN: 978-0-12-088759-0\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SimpleSSPRK33","page":"Trixi.jl","title":"Trixi.SimpleSSPRK33","text":"SimpleSSPRK33(; stage_callbacks=())\n\nThe third-order SSP Runge-Kutta method of Shu and Osher.\n\nReferences\n\nShu, Osher (1988) \"Efficient Implementation of Essentially Non-oscillatory Shock-Capturing Schemes\" (Eq. 2.18) DOI: 10.1016/0021-9991(88)90177-5\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SteadyStateCallback","page":"Trixi.jl","title":"Trixi.SteadyStateCallback","text":"SteadyStateCallback(; abstol=1.0e-8, reltol=1.0e-6)\n\nTerminates the integration when the residual_steady_state(du, equations) falls below the threshold specified by abstol, reltol.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StepsizeCallback","page":"Trixi.jl","title":"Trixi.StepsizeCallback","text":"StepsizeCallback(; cfl=1.0)\n\nSet the time step size according to a CFL condition with CFL number cfl if the time integration method isn't adaptive itself.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StructuredMesh","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nA structured curved mesh.\n\nDifferent numbers of cells per dimension are possible and arbitrary functions can be used as domain faces.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, coordinates_min, coordinates_max; periodicity=true)\n\nCreate a StructuredMesh that represents a uncurved structured mesh with a rectangular domain.\n\nArguments\n\ncells_per_dimension::NTuple{NDIMS, Int}: the number of cells in each dimension.\ncoordinates_min::NTuple{NDIMS, RealT}: coordinate of the corner in the negative direction of each dimension.\ncoordinates_max::NTuple{NDIMS, RealT}: coordinate of the corner in the positive direction of each dimension.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, mapping; RealT=Float64, unsaved_changes=true, mapping_as_string=mapping2string(mapping, length(cells_per_dimension)))\n\nCreate a StructuredMesh of the given size and shape that uses RealT as coordinate type.\n\nArguments\n\ncells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.\nmapping: a function of NDIMS variables to describe the mapping, which transforms the reference mesh to the physical domain. If no mapping_as_string is defined, this function must be defined with the name mapping to allow for restarts. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.\nRealT::Type: the type that should be used for coordinates.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\nmapping_as_string::String: the code that defines the mapping. If CodeTracking can't find the function definition, it can be passed directly here. The code string must define the mapping function with the name mapping. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Tuple}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, faces; RealT=Float64, unsaved_changes=true, faces_as_string=faces2string(faces))\n\nCreate a StructuredMesh of the given size and shape that uses RealT as coordinate type.\n\nArguments\n\ncells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.\nfaces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D).\nRealT::Type: the type that should be used for coordinates.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SubcellLimiterIDP","page":"Trixi.jl","title":"Trixi.SubcellLimiterIDP","text":"SubcellLimiterIDP(equations::AbstractEquations, basis;\n local_minmax_variables_cons = String[],\n positivity_variables_cons = String[],\n positivity_correction_factor = 0.1)\n\nSubcell invariant domain preserving (IDP) limiting used with VolumeIntegralSubcellLimiting including:\n\nLocal maximum/minimum Zalesak-type limiting for conservative variables (local_minmax_variables_cons)\nPositivity limiting for conservative variables (positivity_variables_cons)\n\nConservative variables to be limited are passed as a vector of strings, e.g. local_minmax_variables_cons = [\"rho\"] and positivity_variables_cons = [\"rho\"].\n\nThe bounds are calculated using the low-order FV solution. The positivity limiter uses positivity_correction_factor such that u^new >= positivity_correction_factor * u^FV.\n\nnote: Note\nThis limiter and the correction callback SubcellLimiterIDPCorrection only work together. Without the callback, no correction takes place, leading to a standard low-order FV scheme.\n\nReferences\n\nRueda-Ramírez, Pazner, Gassner (2022) Subcell Limiting Strategies for Discontinuous Galerkin Spectral Element Methods DOI: 10.1016/j.compfluid.2022.105627\nPazner (2020) Sparse invariant domain preserving discontinuous Galerkin methods with subcell convex limiting DOI: 10.1016/j.cma.2021.113876\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SubcellLimiterIDPCorrection","page":"Trixi.jl","title":"Trixi.SubcellLimiterIDPCorrection","text":"SubcellLimiterIDPCorrection()\n\nPerform antidiffusive correction stage for the a posteriori IDP limiter SubcellLimiterIDP called with VolumeIntegralSubcellLimiting.\n\nnote: Note\nThis callback and the actual limiter SubcellLimiterIDP only work together. This is not a replacement but a necessary addition.\n\nReferences\n\nRueda-Ramírez, Pazner, Gassner (2022) Subcell Limiting Strategies for Discontinuous Galerkin Spectral Element Methods DOI: 10.1016/j.compfluid.2022.105627\nPazner (2020) Sparse invariant domain preserving discontinuous Galerkin methods with subcell convex limiting DOI: 10.1016/j.cma.2021.113876\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralStrongForm","page":"Trixi.jl","title":"Trixi.SurfaceIntegralStrongForm","text":"SurfaceIntegralStrongForm(surface_flux=flux_central)\n\nThe classical strong form surface integral type for FD/DG methods.\n\nSee also VolumeIntegralStrongForm.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralUpwind","page":"Trixi.jl","title":"Trixi.SurfaceIntegralUpwind","text":"SurfaceIntegralUpwind(splitting)\n\nCouple elements with upwind simultaneous approximation terms (SATs) that use a particular flux splitting, e.g., splitting_steger_warming.\n\nSee also VolumeIntegralUpwind.\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralWeakForm","page":"Trixi.jl","title":"Trixi.SurfaceIntegralWeakForm","text":"SurfaceIntegralWeakForm(surface_flux=flux_central)\n\nThe classical weak form surface integral type for DG methods as explained in standard textbooks.\n\nSee also VolumeIntegralWeakForm.\n\nReferences\n\nKopriva (2009) Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers doi: 10.1007/978-90-481-2261-5\nHesthaven, Warburton (2007) Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications doi: 10.1007/978-0-387-72067-8\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.T8codeMesh","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nAn unstructured curved mesh based on trees that uses the C library 't8code' to manage trees and mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{Any}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(trees_per_dimension; polydeg, mapping=identity,\n RealT=Float64, initial_refinement_level=0, periodicity=true)\n\nCreate a structured potentially curved 'T8codeMesh' of the specified size.\n\nNon-periodic boundaries will be called ':xneg', ':xpos', ':yneg', ':ypos', ':zneg', ':zpos'.\n\nArguments\n\n'treesperdimension::NTupleE{NDIMS, Int}': the number of trees in each dimension.\n'polydeg::Integer': polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\n'mapping': a function of 'NDIMS' variables to describe the mapping that transforms the reference mesh ('[-1, 1]^n') to the physical domain.\n'RealT::Type': the type that should be used for coordinates.\n'initialrefinementlevel::Integer': refine the mesh uniformly to this level before the simulation starts.\n'periodicity': either a 'Bool' deciding if all of the boundaries are periodic or an 'NTuple{NDIMS, Bool}' deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Union{Tuple{Ptr{P4est.LibP4est.p4est_connectivity}}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh{NDIMS}(conn::Ptr{p4est_connectivity},\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.\n\nArguments\n\nconn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Union{Tuple{Ptr{T8code.Libt8.t8_cmesh}}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh{NDIMS}(cmesh::Ptr{t8_cmesh},\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a t8_cmesh data structure.\n\nArguments\n\ncmesh::Ptr{t8_cmesh}: Pointer to a cmesh object.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Union{Tuple{String}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh{NDIMS}(meshfile::String;\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a Gmsh mesh file (.msh).\n\nArguments\n\nmeshfile::String: path to a Gmsh mesh file.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.TimeSeriesCallback","page":"Trixi.jl","title":"Trixi.TimeSeriesCallback","text":"TimeSeriesCallback(semi, point_coordinates;\n interval=1, solution_variables=cons2cons,\n output_directory=\"out\", filename=\"time_series.h5\",\n RealT=real(solver), uEltype=eltype(cache.elements))\n\nCreate a callback that records point-wise data at points given in point_coordinates every interval time steps. The point coordinates are to be specified either as a vector of coordinate tuples or as a two-dimensional array where the first dimension is the point number and the second dimension is the coordinate dimension. By default, the conservative variables are recorded, but this can be controlled by passing a different conversion function to solution_variables.\n\nAfter the last time step, the results are stored in an HDF5 file filename in directory output_directory.\n\nThe real data type RealT and data type for solution variables uEltype default to the respective types used in the solver and the cache.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.TreeMesh","page":"Trixi.jl","title":"Trixi.TreeMesh","text":"TreeMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nA Cartesian mesh based on trees of hypercubes to support adaptive mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.UnstructuredMesh2D","page":"Trixi.jl","title":"Trixi.UnstructuredMesh2D","text":"UnstructuredMesh2D <: AbstractMesh{2}\n\nAn unstructured (possibly curved) quadrilateral mesh.\n\nUnstructuredMesh2D(filename; RealT=Float64, periodicity=false)\n\nAll mesh information, neighbour coupling, and boundary curve information is read in from a mesh file filename.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.UnstructuredSortedBoundaryTypes","page":"Trixi.jl","title":"Trixi.UnstructuredSortedBoundaryTypes","text":"UnstructuredSortedBoundaryTypes\n\nGeneral container to sort the boundary conditions by type for some unstructured meshes/solvers. It stores a set of global indices for each boundary condition type to expedite computation during the call to calc_boundary_flux!. The original dictionary form of the boundary conditions set by the user in the elixir file is also stored for printing.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ViscousFormulationBassiRebay1","page":"Trixi.jl","title":"Trixi.ViscousFormulationBassiRebay1","text":"ViscousFormulationBassiRebay1()\n\nThe classical BR1 flux from\n\nF. Bassi, S. Rebay (1997) A High-Order Accurate Discontinuous Finite Element Method for the Numerical Solution of the Compressible Navier-Stokes Equations DOI: 10.1006/jcph.1996.5572\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ViscousFormulationLocalDG","page":"Trixi.jl","title":"Trixi.ViscousFormulationLocalDG","text":"ViscousFormulationLocalDG(penalty_parameter)\n\nThe local DG (LDG) flux from \"The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems\" by Cockburn and Shu (1998).\n\nNote that, since this implementation does not involve the parabolic \"upwinding\" vector, the LDG solver is equivalent to ViscousFormulationBassiRebay1 with an LDG-type penalization.\n\nCockburn and Shu (1998). The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems DOI: 10.1137/S0036142997316712\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VisualizationCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.VisualizationCallback","text":"VisualizationCallback(; interval=0,\n solution_variables=cons2prim,\n variable_names=[],\n show_mesh=false,\n plot_data_creator=PlotData2D,\n plot_creator=show_plot,\n plot_arguments...)\n\nCreate a callback that visualizes results during a simulation, also known as in-situ visualization.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in any future releases.\n\nThe interval specifies the number of time step iterations after which a new plot is generated. The available variables to plot are configured with the solution_variables parameter, which acts the same way as for the SaveSolutionCallback. The variables to be actually plotted can be selected by providing a single string or a list of strings to variable_names, and if show_mesh is true, an additional plot with the mesh will be generated.\n\nTo customize the generated figure, plot_data_creator allows to use different plot data types. With plot_creator you can further specify an own function to visualize results, which must support the same interface as the default implementation show_plot. All remaining keyword arguments are collected and passed as additional arguments to the plotting command.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.VolumeIntegralFluxDifferencing","page":"Trixi.jl","title":"Trixi.VolumeIntegralFluxDifferencing","text":"VolumeIntegralFluxDifferencing(volume_flux)\n\nVolume integral type for DG methods based on SBP operators and flux differencing using a symmetric two-point volume_flux. This volume_flux needs to satisfy the interface of numerical fluxes in Trixi.jl.\n\nReferences\n\nLeFloch, Mercier, Rohde (2002) Fully Discrete, Entropy Conservative Schemes of Arbitrary Order doi: 10.1137/S003614290240069X\nFisher, Carpenter (2013) High-order entropy stable finite difference schemes for nonlinear conservation laws: Finite domains doi: 10.1016/j.jcp.2013.06.014\nHendrik Ranocha (2017) Comparison of Some Entropy Conservative Numerical Fluxes for the Euler Equations arXiv: 1701.02264 doi: 10.1007/s10915-017-0618-1\nChen, Shu (2017) Entropy stable high order discontinuous Galerkin methods with suitable quadrature rules for hyperbolic conservation laws doi: 10.1016/j.jcp.2017.05.025\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralPureLGLFiniteVolume","page":"Trixi.jl","title":"Trixi.VolumeIntegralPureLGLFiniteVolume","text":"VolumeIntegralPureLGLFiniteVolume(volume_flux_fv)\n\nA volume integral that only uses the subcell finite volume schemes of the VolumeIntegralShockCapturingHG.\n\nThis gives a formally O(1)-accurate finite volume scheme on an LGL-type subcell mesh (LGL = Legendre-Gauss-Lobatto).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralShockCapturingHG","page":"Trixi.jl","title":"Trixi.VolumeIntegralShockCapturingHG","text":"VolumeIntegralShockCapturingHG(indicator; volume_flux_dg=flux_central,\n volume_flux_fv=flux_lax_friedrichs)\n\nShock-capturing volume integral type for DG methods using a convex blending of the finite volume method with numerical flux volume_flux_fv and the VolumeIntegralFluxDifferencing with volume flux volume_flux_dg. The amount of blending is determined by the indicator, e.g., IndicatorHennemannGassner.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralStrongForm","page":"Trixi.jl","title":"Trixi.VolumeIntegralStrongForm","text":"VolumeIntegralStrongForm()\n\nThe classical strong form volume integral type for FD/DG methods.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralSubcellLimiting","page":"Trixi.jl","title":"Trixi.VolumeIntegralSubcellLimiting","text":"VolumeIntegralSubcellLimiting(limiter;\n volume_flux_dg, volume_flux_fv)\n\nA subcell limiting volume integral type for DG methods based on subcell blending approaches with a low-order FV method. Used with limiter SubcellLimiterIDP.\n\nnote: Note\nSubcell limiting methods are not fully functional on non-conforming meshes. This is mainly because the implementation assumes that low- and high-order schemes have the same surface terms, which is not guaranteed for non-conforming meshes. The low-order scheme with a high-order mortar is not invariant domain preserving.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralUpwind","page":"Trixi.jl","title":"Trixi.VolumeIntegralUpwind","text":"VolumeIntegralUpwind(splitting)\n\nSpecialized volume integral for finite difference summation-by-parts (FDSBP) solvers. Can be used together with the upwind SBP operators of Mattsson (2017) implemented in SummationByPartsOperators.jl. The splitting controls the discretization.\n\nSee also splitting_steger_warming, splitting_lax_friedrichs, splitting_vanleer_haenel.\n\nReferences\n\nMattsson (2017) Diagonal-norm upwind SBP operators doi: 10.1016/j.jcp.2017.01.042\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralWeakForm","page":"Trixi.jl","title":"Trixi.VolumeIntegralWeakForm","text":"VolumeIntegralWeakForm()\n\nThe classical weak form volume integral type for DG methods as explained in standard textbooks.\n\nReferences\n\nKopriva (2009) Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers doi: 10.1007/978-90-481-2261-5\nHesthaven, Warburton (2007) Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications doi: 10.1007/978-0-387-72067-8\n\nVolumeIntegralWeakForm() is only implemented for conserved terms as non-conservative terms should always be discretized in conjunction with a flux-splitting scheme, see VolumeIntegralFluxDifferencing. This treatment is required to achieve, e.g., entropy-stability or well-balancedness.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Base.getindex-Tuple{Trixi.AbstractPlotData, Any}","page":"Trixi.jl","title":"Base.getindex","text":"Base.getindex(pd::AbstractPlotData, variable_name)\n\nExtract a single variable variable_name from pd for plotting with Plots.plot.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Base.resize!-Tuple{Trixi.AbstractContainer, Any}","page":"Trixi.jl","title":"Base.resize!","text":"resize!(c::AbstractContainer, new_length) -> AbstractContainer\n\nResize c to contain new_length elements. If new_length is smaller than the current container length, the first new_length elements will be retained. If new_length is larger, the new elements are invalidated.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.compute_coefficients!-Tuple{Any, Any, Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"PolynomialBases.compute_coefficients!","text":"compute_coefficients!(u_ode, func, t, semi::AbstractSemidiscretization)\n\nSame as compute_coefficients but stores the result in u_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.compute_coefficients-Tuple{Any, Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"PolynomialBases.compute_coefficients","text":"compute_coefficients(func, t, semi::AbstractSemidiscretization)\n\nCompute the discrete coefficients of the continuous function func at time t associated with the semidiscretization semi. For example, the discrete coefficients of func for a discontinuous Galerkin spectral element method (DGSEM) are the values of func at the Lobatto-Legendre nodes. Similarly, a classical finite difference method will use the values of func at the nodes of the grid assoociated with the semidiscretization semi.\n\nFor semidiscretizations semi associated with an initial condition, func can be omitted to use the given initial condition at time t.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.integrate-Tuple{Any, Any, LobattoLegendreBasis}","page":"Trixi.jl","title":"PolynomialBases.integrate","text":"integrate(f, u, basis::LobattoLegendreBasis)\n\nMap the function f to the coefficients u and integrate with respect to the quadrature rule given by basis.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.integrate-Union{Tuple{Func}, Tuple{Func, Any, Trixi.AbstractSemidiscretization}} where Func","page":"Trixi.jl","title":"PolynomialBases.integrate","text":"integrate([func=(u_node,equations)->u_node,] u_ode, semi::AbstractSemidiscretization; normalize=true)\n\nCall func(u_node, equations) for each vector of nodal variables u_node in u_ode and integrate the result using a quadrature associated with the semidiscretization semi.\n\nIf normalize is true, the result is divided by the total volume of the computational domain.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SciMLBase.add_tstop!-Tuple{Trixi.SimpleIntegratorSSP, Any}","page":"Trixi.jl","title":"SciMLBase.add_tstop!","text":"add_tstop!(integrator::SimpleIntegratorSSP, t)\n\nAdd a time stop during the time integration process. This function is called after the periodic SaveSolutionCallback to specify the next stop to save the solution.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{SemidiscretizationHyperbolicParabolic, Any}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::SemidiscretizationHyperbolicParabolic, tspan)\n\nWrap the semidiscretization semi as a split ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The parabolic right-hand side is the first function of the split ODE problem and will be used by default by the implicit part of IMEX methods from the SciML ecosystem.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{Trixi.AbstractSemidiscretization, Any, AbstractString}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::AbstractSemidiscretization, tspan, restart_file::AbstractString)\n\nWrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The initial condition etc. is taken from the restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{Trixi.AbstractSemidiscretization, Any}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::AbstractSemidiscretization, tspan)\n\nWrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiBasis-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.DGMultiBasis","text":"DGMultiBasis(element_type, polydeg; approximation_type = Polynomial(), kwargs...)\n\nConstructs a basis for DGMulti solvers. Returns a \"StartUpDG.RefElemData\" object. The kwargs arguments are additional keyword arguments for RefElemData, such as quad_rule_vol. These are the same as the RefElemData_kwargs used in DGMulti. For more info, see the StartUpDG.jl docs.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.LBMCollisionCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.LBMCollisionCallback","text":"LBMCollisionCallback()\n\nApply the Lattice-Boltzmann method (LBM) collision operator before each time step. See LatticeBoltzmannEquations2D for further details.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.P4estMeshCubedSphere-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.P4estMeshCubedSphere","text":"P4estMeshCubedSphere(trees_per_face_dimension, layers, inner_radius, thickness;\n polydeg, RealT=Float64,\n initial_refinement_level=0, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true)\n\nBuild a \"Cubed Sphere\" mesh as P4estMesh with 6 * trees_per_face_dimension^2 * layers trees.\n\nThe mesh will have two boundaries, :inside and :outside.\n\nArguments\n\ntrees_per_face_dimension::Integer: the number of trees in the first two local dimensions of each face.\nlayers::Integer: the number of trees in the third local dimension of each face, i.e., the number of layers of the sphere.\ninner_radius::Integer: the inner radius of the sphere.\nthickness::Integer: the thickness of the sphere. The outer radius will be inner_radius + thickness.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2D-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.PlotData2D","text":"PlotData2D(u, semi [or mesh, equations, solver, cache];\n solution_variables=nothing,\n grid_lines=true, max_supported_level=11, nvisnodes=nothing,\n slice=:xy, point=(0.0, 0.0, 0.0))\n\nCreate a new PlotData2D object that can be used for visualizing 2D/3D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.\n\nIf grid_lines is true, also extract grid vertices for visualizing the mesh. The output resolution is indirectly set via max_supported_level: all data is interpolated to 2^max_supported_level uniformly distributed points in each spatial direction, also setting the maximum allowed refinement level in the solution. nvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.\n\nWhen visualizing data from a three-dimensional simulation, a 2D slice is extracted for plotting. slice specifies the plane that is being sliced and may be :xy, :xz, or :yz. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0, 0.0). Both of these values are ignored when visualizing 2D data.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nExamples\n\njulia> using Trixi, Plots\n\njulia> trixi_include(default_example())\n[...]\n\njulia> pd = PlotData2D(sol)\nPlotData2D(...)\n\njulia> plot(pd) # To plot all available variables\n\njulia> plot(pd[\"scalar\"]) # To plot only a single variable\n\njulia> plot!(getmesh(pd)) # To add grid lines to the plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2D-Tuple{Union{ODESolution{T, N, uType, uType2, DType, tType, rateType, P} where {T, N, uType, uType2, DType, tType, rateType, P<:(ODEProblem{uType_, tType_, isinplace, P_} where {uType_, tType_, isinplace, P_<:Trixi.AbstractSemidiscretization})}, Trixi.TimeIntegratorSolution}}","page":"Trixi.jl","title":"Trixi.PlotData2D","text":"PlotData2D(sol; kwargs...)\n\nCreate a PlotData2D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ScalarPlotData2D-Tuple{Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.ScalarPlotData2D","text":"ScalarPlotData2D(u, semi::AbstractSemidiscretization; kwargs...)\n\nReturns an PlotData2DTriangulated object which is used to visualize a single scalar field. u should be an array whose entries correspond to values of the scalar field at nodal points.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SummaryCallback","page":"Trixi.jl","title":"Trixi.SummaryCallback","text":"SummaryCallback()\n\nCreate and return a callback that prints a human-readable summary of the simulation setup at the beginning of a simulation and then resets the timer. When the returned callback is executed directly, the current timer values are shown.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.TrivialCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.TrivialCallback","text":"TrivialCallback()\n\nA callback that does nothing. This can be useful to disable some callbacks easily via trixi_include.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.adapt_to_mesh_level!-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.adapt_to_mesh_level!","text":"adapt_to_mesh_level!(u_ode, semi, level)\nadapt_to_mesh_level!(sol::Trixi.TrixiODESolution, level)\n\nLike adapt_to_mesh_level, but modifies the solution and parts of the semidiscretization (mesh and caches) in place.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.adapt_to_mesh_level-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.adapt_to_mesh_level","text":"adapt_to_mesh_level(u_ode, semi, level)\nadapt_to_mesh_level(sol::Trixi.TrixiODESolution, level)\n\nUse the regular adaptive mesh refinement routines to adaptively refine/coarsen the solution u_ode with semidiscretization semi towards a uniformly refined grid with refinement level level. The solution and semidiscretization are copied such that the original objects remain unaltered.\n\nA convenience method accepts an ODE solution object, from which solution and semidiscretization are extracted as needed.\n\nSee also: adapt_to_mesh_level!\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x","text":"boundary_condition_linear_x(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation1D)\n\nBoundary conditions for initial_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x","text":"boundary_condition_linear_x(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x_y-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x_y","text":"boundary_condition_linear_x_y(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_x_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_y-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_y","text":"boundary_condition_linear_y(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_z-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_z","text":"boundary_condition_linear_z(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation1D)\n\nBoundary conditions for boundary_condition_linear_z.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_noslip_wall-Tuple{Any, Any, Any, Any, Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_noslip_wall","text":"boundary_condition_noslip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equations::LatticeBoltzmannEquations2D)\n\nNo-slip wall boundary condition using the bounce-back approach.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_poisson_nonperiodic-Tuple{Any, Any, Any, Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_poisson_nonperiodic","text":"boundary_condition_poisson_nonperiodic(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equations::HyperbolicDiffusionEquations1D)\n\nBoundary conditions used for convergence tests in combination with initial_condition_poisson_nonperiodic and source_terms_poisson_nonperiodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::AcousticPerturbationEquations2D)\n\nUse an orthogonal projection of the perturbed velocities to zero out the normal velocity while retaining the possibility of a tangential velocity in the boundary state. Further details are available in the paper:\n\nMarcus Bauer, Jürgen Dierke and Roland Ewert (2011) Application of a discontinuous Galerkin method to discretize acoustic perturbation equations DOI: 10.2514/1.J050333\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations2D)\n\nShould be used together with StructuredMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations3D)\n\nShould be used together with StructuredMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::CompressibleEulerEquations2D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\n\nDetails about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761\n\nShould be used together with UnstructuredMesh2D.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::CompressibleEulerEquations3D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\n\nDetails about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::ShallowWaterEquations2D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value. For details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::ShallowWaterTwoLayerEquations2D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.\n\nFor details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations1D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations2D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations3D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,\n equations::ShallowWaterEquations1D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.\n\nFor details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::ShallowWaterEquations2D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,\n equations::ShallowWaterTwoLayerEquations1D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.\n\nFor details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_wall-Tuple{Any, Any, Any, Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_wall","text":"boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,\n equations::AcousticPerturbationEquations2D)\n\nBoundary conditions for a solid wall.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_wall-Tuple{Any, Any, Any, Any, Any, Any, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_wall","text":"boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,\n equations::LinearizedEulerEquations2D)\n\nBoundary conditions for a solid wall.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_error_norms-Tuple{Any, Any, Any, Trixi.AbstractSemidiscretization, Any}","page":"Trixi.jl","title":"Trixi.calc_error_norms","text":"calc_error_norms([func=(u_node,equations)->u_node,] u_ode, t, analyzer, semi::AbstractSemidiscretization, cache_analysis)\n\nCalculate discrete L2 and L∞ error norms of func applied to each nodal variable u_node in u_ode. If no exact solution is available, \"errors\" are calculated using some reference state and can be useful for regression tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, direction, equations::IdealGlmMhdEquations1D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations2D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_wavespeed_roe-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.calc_wavespeed_roe","text":"calc_wavespeed_roe(u_ll, u_rr, direction::Integer,\n equations::ShallowWaterEquations1D)\n\nCalculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} See for instance equation (62) in \n\nPaul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044\n\nOr equation (9.17) in this lecture notes.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_wavespeed_roe-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.calc_wavespeed_roe","text":"calc_wavespeed_roe(u_ll, u_rr, direction::Integer,\n equations::ShallowWaterEquations2D)\n\nCalculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} depending on direction. See for instance equation (62) in \n\nPaul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044\n\nOr this slides, slides 8 and 9.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.collision_bgk-Tuple{Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.collision_bgk","text":"collision_bgk(u, dt, equations::LatticeBoltzmannEquations2D)\n\nCollision operator for the Bhatnagar, Gross, and Krook (BGK) model.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.collision_bgk-Tuple{Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.collision_bgk","text":"collision_bgk(u, dt, equations::LatticeBoltzmannEquations3D)\n\nCollision operator for the Bhatnagar, Gross, and Krook (BGK) model.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.cons2cons-Tuple{Any, Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.cons2cons","text":"cons2cons(u, equations)\n\nReturn the conserved variables u. While this function is as trivial as identity, it is also as useful.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.cons2entropy","page":"Trixi.jl","title":"Trixi.cons2entropy","text":"cons2entropy(u, equations)\n\nConvert the conserved variables u to the entropy variables for a given set of equations with chosen standard entropy.\n\nu is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by entropy2cons.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.cons2prim","page":"Trixi.jl","title":"Trixi.cons2prim","text":"cons2prim(u, equations)\n\nConvert the conserved variables u to the primitive variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by prim2cons.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.convergence_test-Tuple{Module, AbstractString, Any}","page":"Trixi.jl","title":"Trixi.convergence_test","text":"convergence_test([mod::Module=Main,] elixir::AbstractString, iterations; kwargs...)\n\nRun iterations Trixi.jl simulations using the setup given in elixir and compute the experimental order of convergence (EOC) in the L^2 and L^infty norm. In each iteration, the resolution of the respective mesh will be doubled. Additional keyword arguments kwargs... and the optional module mod are passed directly to trixi_include.\n\nThis function assumes that the spatial resolution is set via the keywords initial_refinement_level (an integer) or cells_per_dimension (a tuple of integers, one per spatial dimension).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_analysis_errors-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.default_analysis_errors","text":"default_analysis_errors(equations)\n\nDefault analysis errors (:l2_error and :linf_error) used by the AnalysisCallback.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_analysis_integrals-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.default_analysis_integrals","text":"default_analysis_integrals(equations)\n\nDefault analysis integrals used by the AnalysisCallback.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_example-Tuple{}","page":"Trixi.jl","title":"Trixi.default_example","text":"default_example()\n\nReturn the path to an example elixir that can be used to quickly see Trixi.jl in action on a TreeMesh. See also examples_dir and get_examples.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_example_unstructured-Tuple{}","page":"Trixi.jl","title":"Trixi.default_example_unstructured","text":"default_example_unstructured()\n\nReturn the path to an example elixir that can be used to quickly see Trixi.jl in action on an UnstructuredMesh2D. This simulation is run on the example curved, unstructured mesh given in the Trixi.jl documentation regarding unstructured meshes.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.density-Tuple{Real, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.density","text":"density(p::Real, equations::LatticeBoltzmannEquations2D)\ndensity(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic density from the pressure p or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.density-Tuple{Real, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.density","text":"density(p::Real, equations::LatticeBoltzmannEquations3D)\ndensity(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic density from the pressure p or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_dof_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_dof_global","text":"each_dof_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the degrees of freedom (DOF) in dg. In particular, not the DOFs themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_face_node-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_face_node","text":"each_face_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the face nodes in dg. In particular, not the face_nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_face_node_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_face_node_global","text":"each_face_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the face nodes in mesh. In particular, not the face nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_quad_node-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_quad_node","text":"each_quad_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the quadrature nodes in dg. In particular, not the quadrature nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_quad_node_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_quad_node_global","text":"each_quad_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the global quadrature nodes in mesh. In particular, not the quadrature nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachboundary-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachboundary","text":"eachboundary(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the boundaries in cache. In particular, not the boundaries themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachcomponent-Tuple{Trixi.AbstractCompressibleEulerMulticomponentEquations}","page":"Trixi.jl","title":"Trixi.eachcomponent","text":"eachcomponent(equations::AbstractCompressibleEulerMulticomponentEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the components in AbstractCompressibleEulerMulticomponentEquations. In particular, not the components themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachcomponent-Tuple{Trixi.AbstractIdealGlmMhdMulticomponentEquations}","page":"Trixi.jl","title":"Trixi.eachcomponent","text":"eachcomponent(equations::AbstractIdealGlmMhdMulticomponentEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the components in AbstractIdealGlmMhdMulticomponentEquations. In particular, not the components themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachdim-Tuple{Any}","page":"Trixi.jl","title":"Trixi.eachdim","text":"eachdim(mesh)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the dimensions in AbstractTree. In particular, not the dimensions themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachdirection-Tuple{Trixi.AbstractTree}","page":"Trixi.jl","title":"Trixi.eachdirection","text":"eachdirection(tree::AbstractTree)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the directions in AbstractTree. In particular, not the directions themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in cache. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in mesh. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer1D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer1D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer2D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer2D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer3D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer3D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.UnstructuredElementContainer2D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::UnstructuredElementContainer2D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachinterface-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachinterface","text":"eachinterface(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the interfaces in cache. In particular, not the interfaces themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmortar-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmortar","text":"eachmortar(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the mortars in cache. In particular, not the mortars themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmpiinterface-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmpiinterface","text":"eachmpiinterface(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the MPI interfaces in cache. In particular, not the interfaces themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmpimortar-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmpimortar","text":"eachmpimortar(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the MPI mortars in cache. In particular, not the mortars themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{DG}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(dg::DG)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in dg. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{LobattoLegendreBasis}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(basis::LobattoLegendreBasis)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in basis. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{Trixi.LobattoLegendreAnalyzer}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(analyzer::LobattoLegendreAnalyzer)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in analyzer. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachvariable-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.eachvariable","text":"eachvariable(equations::AbstractEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the variables in equations. In particular, not the variables themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.energy_internal","page":"Trixi.jl","title":"Trixi.energy_internal","text":"energy_internal(u, equations)\n\nReturn the internal energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.energy_kinetic","page":"Trixi.jl","title":"Trixi.energy_kinetic","text":"energy_kinetic(u, equations)\n\nReturn the kinetic energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.energy_total","page":"Trixi.jl","title":"Trixi.energy_total","text":"energy_total(u, equations)\n\nReturn the total energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.entropy","page":"Trixi.jl","title":"Trixi.entropy","text":"entropy(u, equations)\n\nReturn the chosen entropy of the conserved variables u for a given set of equations.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.entropy2cons","page":"Trixi.jl","title":"Trixi.entropy2cons","text":"entropy2cons(w, equations)\n\nConvert the entropy variables w based on a standard entropy to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2entropy.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.equilibrium_distribution-Tuple{Any, Any, Any, Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.equilibrium_distribution","text":"equilibrium_distribution(alpha, rho, v1, v2, v3, equations::LatticeBoltzmannEquations3D)\n\nCalculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2, v3.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.equilibrium_distribution-Tuple{Any, Any, Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.equilibrium_distribution","text":"equilibrium_distribution(alpha, rho, v1, v2, equations::LatticeBoltzmannEquations2D)\n\nCalculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.examples_dir-Tuple{}","page":"Trixi.jl","title":"Trixi.examples_dir","text":"examples_dir()\n\nReturn the directory where the example files provided with Trixi.jl are located. If Trixi.jl is installed as a regular package (with ]add Trixi), these files are read-only and should not be modified. To find out which files are available, use, e.g., readdir:\n\nExamples\n\nreaddir(examples_dir())\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux","page":"Trixi.jl","title":"Trixi.flux","text":"flux(u, orientation_or_normal, equations)\n\nGiven the conservative variables u, calculate the (physical) flux in Cartesian direction orientation::Integer or in arbitrary direction normal::AbstractVector for the corresponding set of governing equations. orientation is 1, 2, and 3 for the x-, y-, and z-directions, respectively.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.flux-Tuple{Any, AbstractVector, Trixi.AbstractEquations{1}}","page":"Trixi.jl","title":"Trixi.flux","text":"flux(u, normal_direction::AbstractVector, equations::AbstractEquations{1})\n\nEnables calling flux with a non-integer argument normal_direction for one-dimensional equations. Returns the value of flux(u, 1, equations) scaled by normal_direction[1].\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_central-Tuple{Any, Any, Any, Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.flux_central","text":"flux_central(u_ll, u_rr, orientation_or_normal_direction, equations::AbstractEquations)\n\nThe classical central numerical flux f((u_ll) + f(u_rr)) / 2. When this flux is used as volume flux, the discretization is equivalent to the classical weak form DG method (except floating point errors).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chan_etal-Tuple{Any, Any, Integer, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_chan_etal","text":"@inline function fluxchanetal(ull, urr, orientation::Integer, equations::CompressibleEulerEquationsQuasi1D)\n\nConservative (symmetric) part of the entropy conservative flux for quasi 1D compressible Euler equations split form. This flux is a generalization of flux_ranocha for CompressibleEulerEquations1D. Further details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089 \n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chan_etal-Tuple{Any, Any, Integer, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_chan_etal","text":"flux_chan_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquationsQuasi1D)\n\nTotal energy conservative (mathematical entropy for quasi 1D shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., FluxPlusDissipation(flux_chan_etal, DissipationLocalLaxFriedrichs()).\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations1D)\n\nEntropy conserving two-point flux by\n\nAyoub Gouasmi, Karthik Duraisamy (2020) \"Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations\" arXiv:1904.00972v3 [math.NA] 4 Feb 2020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations2D)\n\nAdaption of the entropy conserving two-point flux by\n\nAyoub Gouasmi, Karthik Duraisamy (2020) \"Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations\" arXiv:1904.00972v3 [math.NA] 4 Feb 2020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations2D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations3D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nEntropy conserving two-point flux adapted by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdMulticomponentEquations2D)\n\nEntropy conserving two-point flux adapted by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_es_ersing_etal-Tuple{Any, Any, Any, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_es_ersing_etal","text":"flux_es_ersing_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterTwoLayerEquations2D)\n\nEntropy stable surface flux for the two-layer shallow water equations. Uses the entropy conservative flux_wintermeyer_etal and adds a Lax-Friedrichs type dissipation dependent on the jump of entropy variables. \n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_es_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_es_ersing_etal","text":"flux_es_ersing_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterTwoLayerEquations1D)\n\nEntropy stable surface flux for the two-layer shallow water equations. Uses the entropy conservative flux_wintermeyer_etal and adds a Lax-Friedrichs type dissipation dependent on the jump of entropy variables. \n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_fjordholm_etal","text":"flux_fjordholm_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquations1D)\n\nTotal energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.\n\nDetails are available in Eq. (4.1) in the paper:\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_fjordholm_etal","text":"flux_fjordholm_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nTotal energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.\n\nDetails are available in Eq. (4.1) in the paper:\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_godunov-Tuple{Any, Any, Integer, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_godunov","text":"flux_godunov(u_ll, u_rr, orientation_or_normal_direction,\n equations::LinearizedEulerEquations2D)\n\nAn upwind flux for the linearized Euler equations based on diagonalization of the physical flux matrix. Given the physical flux Au, A=T Lambda T^-1 with Lambda being a diagonal matrix that holds the eigenvalues of A, decompose Lambda = Lambda^+ + Lambda^- where Lambda^+ and Lambda^- are diagonal matrices holding the positive and negative eigenvalues of A, respectively. Then for left and right states u_L u_R, the numerical flux calculated by this function is given by A^+ u_L + A^- u_R where A^pm = T Lambda^pm T^-1.\n\nThe diagonalization of the flux matrix can be found in\n\nR. F. Warming, Richard M. Beam and B. J. Hyett (1975) Diagonalization and simultaneous symmetrization of the gas-dynamic matrices DOI: 10.1090/S0025-5718-1975-0388967-5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations1D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations2D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations3D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdMulticomponentEquations1D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdMulticomponentEquations2D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations2D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations3D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nLi (2005)\n\nAn HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_audusse_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_audusse_etal","text":"flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.\n\nThis hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations1D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.\n\nFurther details on the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_audusse_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_audusse_etal","text":"flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_audusse_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.\n\nThis hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations2D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.\n\nFurther details for the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chan_etal-Tuple{Any, Any, Integer, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chan_etal","text":"flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,\n equations::CompressibleEulerEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_direction, \n equations::CompressibleEulerEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_ll, normal_rr,\n equations::CompressibleEulerEquationsQuasi1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the pressure CompressibleEulerEquationsQuasi1D and the nozzle width.\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089 \n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chan_etal-Tuple{Any, Any, Integer, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chan_etal","text":"flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_direction::AbstractVector,\n equations::ShallowWaterEquationsQuasi1D) \nflux_nonconservative_chan_etal(u_ll, u_rr, \n normal_ll::AbstractVector, normal_rr::AbstractVector,\n equations::ShallowWaterEquationsQuasi1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquationsQuasi1D and the channel width.\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chen_noelle-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chen_noelle","text":"flux_nonconservative_chen_noelle(u_ll, u_rr,\n orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_chen_noelle on the conservative variables.\n\nShould be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_chen_noelle in the surface flux to ensure consistency.\n\nFurther details on the hydrostatic reconstruction and its motivation can be found in\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chen_noelle-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chen_noelle","text":"flux_nonconservative_chen_noelle(u_ll, u_rr,\n orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_chen_noelle(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average ::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_chen_noelle on the conservative variables.\n\nShould be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_chen_noelle in the surface flux to ensure consistency.\n\nFurther details on the hydrostatic reconstruction and its motivation can be found in\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.\n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_ersing_etal(u_ll, u_rr,\n normal_direction_ll::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.\n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterTwoLayerEquations1D)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterTwoLayerEquations1D and an additional term that couples the momentum of both layers. \n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal. \n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterTwoLayerEquations2D)\nflux_nonconservative_ersing_etal(u_ll, u_rr,\n normal_direction_ll::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterTwoLayerEquations2D)\n\nwarning: Experimental code\n\n\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterTwoLayerEquations2D and an additional term that couples the momentum of both layers. \n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.\n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_fjordholm_etal","text":"flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nThis contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.\n\nFurther details for the original finite volume formulation are available in\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\nand for curvilinear 2D case in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_fjordholm_etal","text":"flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_fjordholm_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nThis contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.\n\nFurther details for the original finite volume formulation are available in\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\nand for curvilinear 2D case in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdEquations2D)\nflux_nonconservative_powell(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::IdealGlmMhdEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdEquations3D)\nflux_nonconservative_powell(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::IdealGlmMhdEquations3D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations3D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdMulticomponentEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdMulticomponentEquations2D.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D, Trixi.NonConservativeSymmetric, Integer}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,\n equations::IdealGlmMhdEquations2D,\n nonconservative_type::NonConservativeSymmetric,\n nonconservative_term::Integer)\n\nSymmetric part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered \"fluxes\" for subcell limiting. See, e.g.,\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\nThis function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, u_rr,\n orientation::Integer,\n equations::IdealGlmMhdEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.\n\nThis implementation uses a non-conservative term that can be written as the product of local and symmetric parts. It is equivalent to the non-conservative flux of Bohm et al. (flux_nonconservative_powell) for conforming meshes but it yields different results on non-conforming meshes(!).\n\nThe two other flux functions with the same name return either the local or symmetric portion of the non-conservative flux based on the type of the nonconservativetype argument, employing multiple dispatch. They are used to compute the subcell fluxes in dg2dsubcelllimiters.jl.\n\nReferences\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Integer, IdealGlmMhdEquations2D, Trixi.NonConservativeLocal, Integer}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,\n equations::IdealGlmMhdEquations2D,\n nonconservative_type::NonConservativeLocal,\n nonconservative_term::Integer)\n\nLocal part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered \"fluxes\" for subcell limiting. See, e.g.,\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\nThis function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_wintermeyer_etal","text":"flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nFurther details are available in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_wintermeyer_etal","text":"flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_wintermeyer_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nFurther details are available in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations1D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerMulticomponentEquations1D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerMulticomponentEquations2D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha_turbo-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.flux_ranocha_turbo","text":"flux_ranocha_turbo(u_ll, u_rr, orientation_or_normal_direction, equations)\n\nEquivalent to flux_ranocha except that it may use specialized methods, e.g., when used with VolumeIntegralFluxDifferencing. These specialized methods may enable better use of SIMD instructions to increase runtime efficiency on modern hardware.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal_turbo-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.flux_shima_etal_turbo","text":"flux_shima_etal_turbo(u_ll, u_rr, orientation_or_normal_direction, equations)\n\nEquivalent to flux_shima_etal except that it may use specialized methods, e.g., when used with VolumeIntegralFluxDifferencing. These specialized methods may enable better use of SIMD instructions to increase runtime efficiency on modern hardware.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquations1D)\n\nTotal energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nTotal energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterTwoLayerEquations1D)\n\nTotal energy conservative (mathematical entropy for two-layer shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used with the nonconservative flux_nonconservative_ersing_etal. To obtain the flux for the two-layer shallow water equations the flux that is described in the paper for the normal shallow water equations is used within each layer.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterTwoLayerEquations2D)\nflux_wintermeyer_etal(u_ll, u_rr,\n normal_direction::AbstractVector,\n equations::ShallowWaterTwoLayerEquations2D)\n\nTotal energy conservative (mathematical entropy for two-layer shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used with the nonconservative flux_nonconservative_ersing_etal. To obtain the flux for the two-layer shallow water equations the flux that is described in the paper for the normal shallow water equations is used within each layer.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_winters_etal-Tuple{Any, Any, AbstractVector, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_winters_etal","text":"flux_winters_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::PolytropicEulerEquations2D)\n\nEntropy conserving two-point flux for isothermal or polytropic gases. Requires a special weighted Stolarsky mean for the evaluation of the density denoted here as stolarsky_mean. Note, for isothermal gases where gamma = 1 this stolarsky_mean becomes the ln_mean.\n\nFor details see Section 3.2 of the following reference\n\nAndrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_boundary_outer_state-Tuple{BoundaryConditionDirichlet, Any, Any, Any, Any, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.get_boundary_outer_state","text":"get_boundary_outer_state(boundary_condition::BoundaryConditionDirichlet,\n cache, t, equations, dg, indices...)\n\nFor subcell limiting, the calculation of local bounds for non-periodic domains require the boundary outer state. This function returns the boundary value at time t and for node with spatial indices indices.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_examples-Tuple{}","page":"Trixi.jl","title":"Trixi.get_examples","text":"get_examples()\n\nReturn a list of all example elixirs that are provided by Trixi.jl. See also examples_dir and default_example.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_name-Tuple{Any}","page":"Trixi.jl","title":"Trixi.get_name","text":"get_name(x)\n\nReturns a name of x ready for pretty printing. By default, return string(y) if x isa Val{y} and return string(x) otherwise.\n\nExamples\n\njulia> Trixi.get_name(\"test\")\n\"test\"\n\njulia> Trixi.get_name(Val(:test))\n\"test\"\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_name-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.get_name","text":"get_name(equations::AbstractEquations)\n\nReturns the canonical, human-readable name for the given system of equations.\n\nExamples\n\njulia> Trixi.get_name(CompressibleEulerEquations1D(1.4))\n\"CompressibleEulerEquations1D\"\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.getmesh-Tuple{Trixi.AbstractPlotData}","page":"Trixi.jl","title":"Trixi.getmesh","text":"getmesh(pd::AbstractPlotData)\n\nExtract grid lines from pd for plotting with Plots.plot.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.global_mean_vars-Tuple{AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.global_mean_vars","text":"global_mean_vars(equations::AcousticPerturbationEquations2D)\n\nReturns the global mean variables stored in equations. This makes it easier to define flexible initial conditions for problems with constant mean flow.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.have_nonconservative_terms-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.have_nonconservative_terms","text":"have_nonconservative_terms(equations)\n\nTrait function determining whether equations represent a conservation law with or without nonconservative terms. Classical conservation laws such as the CompressibleEulerEquations2D do not have nonconservative terms. The ShallowWaterEquations2D with non-constant bottom topography are an example of equations with nonconservative terms. The return value will be True() or False() to allow dispatching on the return type.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_audusse_etal-Tuple{Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_audusse_etal","text":"hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nA particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_audusse_etal-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_audusse_etal","text":"hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nA particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details for the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_chen_noelle-Tuple{Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_chen_noelle","text":"hydrostatic_reconstruction_chen_noelle(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nA particular type of hydrostatic reconstruction of the water height to guarantee well-balancedness for a general bottom topography of the ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. The key idea is a linear reconstruction of the bottom and water height at the interfaces using subcells. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_chen_noelle-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_chen_noelle","text":"hydrostatic_reconstruction_chen_noelle(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\n\nA particular type of hydrostatic reconstruction of the water height to guarantee well-balancedness for a general bottom topography of the ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. The key idea is a linear reconstruction of the bottom and water height at the interfaces using subcells. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_mpi-Tuple{}","page":"Trixi.jl","title":"Trixi.init_mpi","text":"init_mpi()\n\nInitialize MPI by calling MPI.Initialized(). The function will check if MPI is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_p4est-Tuple{}","page":"Trixi.jl","title":"Trixi.init_p4est","text":"init_p4est()\n\nInitialize p4est by calling p4est_init and setting the log level to SC_LP_ERROR. This function will check if p4est is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_t8code-Tuple{}","page":"Trixi.jl","title":"Trixi.init_t8code","text":"init_t8code()\n\nInitialize t8code by calling sc_init, p4est_init, and t8_init while setting the log level to SC_LP_ERROR. This function will check if t8code is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::AcousticPerturbationEquations2D)\n\nA constant initial condition where the state variables are zero and the mean flow is constant. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::InviscidBurgersEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LatticeBoltzmannEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LatticeBoltzmannEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::AcousticPerturbationEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations3D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquationsQuasi1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations1D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations2D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations3D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdMulticomponentEquations1D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdMulticomponentEquations2D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::InviscidBurgersEquation1D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA smooth initial condition used for convergence tests (in combination with BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearizedEulerEquations2D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::PolytropicEulerEquations2D)\n\nManufactured smooth initial condition used for convergence tests in combination with source_terms_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterEquationsQuasi1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterTwoLayerEquations1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterTwoLayerEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test. Constants must be set to rho_upper = 09, rho_lower = 10, g = 100.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_density_wave-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_density_wave","text":"initial_condition_density_wave(x, t, equations::CompressibleEulerEquations1D)\n\nA sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper\n\nGregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026\n\nwith the following parameters\n\ndomain [-1, 1]\nmesh = 4x4\npolydeg = 5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_density_wave-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_density_wave","text":"initial_condition_density_wave(x, t, equations::CompressibleEulerEquations2D)\n\nA sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper\n\nGregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026\n\nwith the following parameters\n\ndomain [-1, 1]\nmesh = 4x4\npolydeg = 5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations1D)\n\nOne dimensional variant of the setup used for convergence tests of the Euler equations with self-gravity from\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nnote: Note\nThere is no additional source term necessary for the manufactured solution in one spatial dimension. Thus, source_terms_eoc_test_coupled_euler_gravity is not present there.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations1D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::AcousticPerturbationEquations2D)\n\nA Gaussian pulse in a constant mean flow. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA Gaussian pulse used together with BoundaryConditionDirichlet(initial_condition_gauss).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equation::LinearScalarAdvectionEquation2D)\n\nA Gaussian pulse used together with BoundaryConditionDirichlet(initial_condition_gauss).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA Gaussian pulse.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x","text":"initial_condition_linear_x(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA linear function of x[1] used together with boundary_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x","text":"initial_condition_linear_x(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] used together with boundary_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x_y-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x_y","text":"initial_condition_linear_x_y(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] + x[2] used together with boundary_condition_linear_x_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_y-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_y","text":"initial_condition_linear_y(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] used together with boundary_condition_linear_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_z-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_z","text":"initial_condition_linear_z(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA linear function of x[3] used together with boundary_condition_linear_z.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_poisson_nonperiodic-Tuple{Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_poisson_nonperiodic","text":"initial_condition_poisson_nonperiodic(x, t, equations::HyperbolicDiffusionEquations1D)\n\nA non-priodic smooth initial condition. Can be used for convergence tests in combination with source_terms_poisson_nonperiodic and boundary_condition_poisson_nonperiodic.\n\nnote: Note\nThe solution is periodic but the initial guess is not.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin","text":"initial_condition_sin(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin","text":"initial_condition_sin(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin_sin-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin_sin","text":"initial_condition_sin_sin(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations1D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations2D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations3D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nA for multicomponent adapted weak blast wave adapted to multicomponent and taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nA for multicomponent adapted weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations1D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations3D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations1D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::PolytropicEulerEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations1D)\n\nA weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations2D)\n\nA weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.integrate_via_indices-Union{Tuple{Func}, Tuple{Func, Any, Trixi.AbstractSemidiscretization, Vararg{Any}}} where Func","page":"Trixi.jl","title":"Trixi.integrate_via_indices","text":"integrate_via_indices(func, u_ode, semi::AbstractSemidiscretization, args...; normalize=true)\n\nCall func(u, i..., element, equations, solver, args...) for all nodal indices i..., element and integrate the result using a quadrature associated with the semidiscretization semi.\n\nIf normalize is true, the result is divided by the total volume of the computational domain.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.inv_ln_mean-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.inv_ln_mean","text":"inv_ln_mean(x, y)\n\nCompute the inverse 1 / ln_mean(x, y) of the logarithmic mean ln_mean.\n\nThis function may be used to increase performance where the inverse of the logarithmic mean is needed, by replacing a (slow) division by a (fast) multiplication.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.jacobian_ad_forward-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.jacobian_ad_forward","text":"jacobian_ad_forward(semi::AbstractSemidiscretization;\n t0=zero(real(semi)),\n u0_ode=compute_coefficients(t0, semi))\n\nUses the right-hand side operator of the semidiscretization semi and forward mode automatic differentiation to compute the Jacobian J of the semidiscretization semi at state u0_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.jacobian_fd-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.jacobian_fd","text":"jacobian_fd(semi::AbstractSemidiscretization;\n t0=zero(real(semi)),\n u0_ode=compute_coefficients(t0, semi))\n\nUses the right-hand side operator of the semidiscretization semi and simple second order finite difference to compute the Jacobian J of the semidiscretization semi at state u0_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.linear_structure-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.linear_structure","text":"linear_structure(semi::AbstractSemidiscretization;\n t0=zero(real(semi)))\n\nWraps the right-hand side operator of the semidiscretization semi at time t0 as an affine-linear operator given by a linear operator A and a vector b.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ln_mean-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.ln_mean","text":"ln_mean(x, y)\n\nCompute the logarithmic mean\n\nln_mean(x, y) = (y - x) / (log(y) - log(x)) = (y - x) / log(y / x)\n\nProblem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Ismail and Roe (2009). Set ξ = y / x. Then, we have\n\n(y - x) / log(y / x) = (x + y) / log(ξ) * (ξ - 1) / (ξ + 1)\n\nSet f = (ξ - 1) / (ξ + 1) = (y - x) / (x + y). Then, we use the expansion\n\nlog(ξ) = 2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7) + O(ξ^9)\n\nInserting the first few terms of this expansion yields\n\n(y - x) / log(ξ) ≈ (x + y) * f / (2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7))\n = (x + y) / (2 + 2/3 * f^2 + 2/5 * f^4 + 2/7 * f^6)\n\nSince divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use\n\nf^2 = (y - x)^2 / (x + y)^2\n = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)\n\nGiven ε = 1.0e-4, we use the following algorithm.\n\nif f^2 < ε\n # use the expansion above\nelse\n # use the direct formula (y - x) / log(y / x)\nend\n\nReferences\n\nIsmail, Roe (2009). Affordable, entropy-consistent Euler flux functions II: Entropy production at shocks. DOI: 10.1016/j.jcp.2009.04.021\nAgner Fog. Lists of instruction latencies, throughputs and micro-operation breakdowns for Intel, AMD, and VIA CPUs. https://www.agner.org/optimize/instruction_tables.pdf\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_adaptive_time_integrator!-Tuple{Any, AbstractString}","page":"Trixi.jl","title":"Trixi.load_adaptive_time_integrator!","text":"load_adaptive_time_integrator!(integrator, restart_file::AbstractString)\n\nLoad the context information for time integrators with error-based step size control saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_dt-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_dt","text":"load_dt(restart_file::AbstractString)\n\nLoad the time step size (dt in OrdinaryDiffEq.jl) saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_mesh-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_mesh","text":"load_mesh(restart_file::AbstractString; n_cells_max)\n\nLoad the mesh from the restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_time-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_time","text":"load_time(restart_file::AbstractString)\n\nLoad the time saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_timestep!-Tuple{Any, AbstractString}","page":"Trixi.jl","title":"Trixi.load_timestep!","text":"load_timestep!(integrator, restart_file::AbstractString)\n\nLoad the time step number saved in a restart_file and assign it to both the time step number and and the number of accepted steps (iter and stats.naccept in OrdinaryDiffEq.jl, respectively) in integrator.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_timestep-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_timestep","text":"load_timestep(restart_file::AbstractString)\n\nLoad the time step number (iter in OrdinaryDiffEq.jl) saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.max-Tuple","page":"Trixi.jl","title":"Trixi.max","text":"max(x, y, ...)\n\nReturn the maximum of the arguments. See also the maximum function to take the maximum element from a collection.\n\nThis version in Trixi.jl is semantically equivalent to Base.max but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).\n\nExamples\n\njulia> max(2, 5, 1)\n5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.max_abs_speed_naive","page":"Trixi.jl","title":"Trixi.max_abs_speed_naive","text":"max_abs_speed_naive(u_ll, u_rr, orientation::Integer, equations)\nmax_abs_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimate of the maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, based only on the local wave speeds associated to u_ll and u_rr.\n\nFor non-integer arguments normal_direction in one dimension, max_abs_speed_naive returns abs(normal_direction[1]) * max_abs_speed_naive(u_ll, u_rr, 1, equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min-Tuple","page":"Trixi.jl","title":"Trixi.min","text":"min(x, y, ...)\n\nReturn the minimum of the arguments. See also the minimum function to take the minimum element from a collection.\n\nThis version in Trixi.jl is semantically equivalent to Base.min but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).\n\nExamples\n\njulia> min(2, 5, 1)\n1\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_chen_noelle-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.min_max_speed_chen_noelle","text":"min_max_speed_chen_noelle(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nThe approximated speeds for the HLL type numerical flux used by Chen and Noelle for their hydrostatic reconstruction. As they state in the paper, these speeds are chosen for the numerical flux to ensure positivity and to satisfy an entropy inequality.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_chen_noelle-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_chen_noelle","text":"min_max_speed_chen_noelle(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nmin_max_speed_chen_noelle(u_ll, u_rr, normal_direction::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nSpecial estimate of the minimal and maximal wave speed of the shallow water equations for the left and right states u_ll, u_rr. These approximate speeds are used for the HLL-type numerical flux flux_hll_chen_noelle. These wave speed estimates together with a particular hydrostatic reconstruction technique guarantee that the numerical flux is positive and satisfies an entropy inequality.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in the reference below. The definition of the wave speeds are given in Equation (2.20).\n\nGuoxian Chen and Sebastian Noelle (2017) A new hydrostatic reconstruction scheme based on subcell reconstructions DOI:10.1137/15M1053074\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_davis","page":"Trixi.jl","title":"Trixi.min_max_speed_davis","text":"min_max_speed_davis(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_davis(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimates of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.\n\nS.F. Davis (1988) Simplified Second-Order Godunov-Type Methods DOI: 10.1137/0909030\n\nSee also FluxHLL, min_max_speed_naive, min_max_speed_einfeldt.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_einfeldt(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nMore advanced mininmal and maximal wave speed computation based on\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\noriginally developed for the compressible Euler equations. A compact representation can be found in this lecture notes, eq. (9.28).\n\nSee also FluxHLL, min_max_speed_naive, min_max_speed_davis.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, AbstractVector, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations2D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, AbstractVector, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations3D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nOriginal publication:\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\n\nCompactly summarized:\n\nSiddhartha Mishra, Ulrik Skre Fjordholm and Rémi Abgrall Numerical methods for conservation laws and related equations. Link\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations::IdealGlmMhdEquations1D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations::IdealGlmMhdEquations2D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_naive","page":"Trixi.jl","title":"Trixi.min_max_speed_naive","text":"min_max_speed_naive(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimate(!) of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.\n\nAmiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002\n\nSee also FluxHLL, min_max_speed_davis, min_max_speed_einfeldt.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.modify_dt_for_tstops!-Tuple{Trixi.SimpleIntegratorSSP}","page":"Trixi.jl","title":"Trixi.modify_dt_for_tstops!","text":"modify_dt_for_tstops!(integrator::SimpleIntegratorSSP)\n\nModify the time-step size to match the time stops specified in integrator.opts.tstops. To avoid adding OrdinaryDiffEq to Trixi's dependencies, this routine is a copy of https://github.com/SciML/OrdinaryDiffEq.jl/blob/d76335281c540ee5a6d1bd8bb634713e004f62ee/src/integrators/integrator_utils.jl#L38-L54\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.multiply_dimensionwise-Tuple{AbstractMatrix, AbstractMatrix}","page":"Trixi.jl","title":"Trixi.multiply_dimensionwise","text":"multiply_dimensionwise(matrix::AbstractMatrix, data_in::AbstractArray{<:Any, NDIMS+1})\n\nMultiply the array data_in by matrix in each coordinate direction, where data_in is assumed to have the first coordinate for the number of variables and the remaining coordinates are multiplied by matrix.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.n_nonconservative_terms","page":"Trixi.jl","title":"Trixi.n_nonconservative_terms","text":"n_nonconservative_terms(equations)\n\nNumber of nonconservative terms in the form local * symmetric for a particular equation. This function needs to be specialized only if equations with nonconservative terms are combined with certain solvers (e.g., subcell limiting).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.ndofs-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.ndofs","text":"ndofs(semi::AbstractSemidiscretization)\n\nReturn the number of degrees of freedom associated with each scalar variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.negative_part-Tuple{Any}","page":"Trixi.jl","title":"Trixi.negative_part","text":"negative_part(x)\n\nReturn x if x is negative, else zero. In other words, return (x - abs(x)) / 2 for real numbers x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_default_options-Tuple{}","page":"Trixi.jl","title":"Trixi.ode_default_options","text":"ode_default_options()\n\nReturn the default options for OrdinaryDiffEq's solve. Pass ode_default_options()... to solve to only return the solution at the final time and enable MPI aware error-based step size control, whenever MPI is used. For example, use solve(ode, alg; ode_default_options()...).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_norm-Tuple{Number, Any}","page":"Trixi.jl","title":"Trixi.ode_norm","text":"ode_norm(u, t)\n\nImplementation of the weighted L2 norm of Hairer and Wanner used for error-based step size control in OrdinaryDiffEq.jl. This function is aware of MPI and uses global MPI communication when running in parallel.\n\nYou must pass this function as a keyword argument internalnorm=ode_norm to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.\n\nSee the \"Advanced Adaptive Stepsize Control\" section of the documentation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_unstable_check-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.ode_unstable_check","text":"ode_unstable_check(dt, u, semi, t)\n\nImplementation of the basic check for instability used in OrdinaryDiffEq.jl. Instead of checking something like any(isnan, u), this function just checks isnan(dt). This helps when using MPI parallelization, since no additional global communication is required and all ranks will return the same result.\n\nYou should pass this function as a keyword argument unstable_check=ode_unstable_check to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.\n\nSee the \"Miscellaneous\" section of the documentation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.partition!-Tuple{TreeMesh{NDIMS, <:Trixi.ParallelTree{NDIMS}} where NDIMS}","page":"Trixi.jl","title":"Trixi.partition!","text":"partition!(mesh::ParallelTreeMesh, allow_coarsening=true)\n\nPartition mesh using a static domain decomposition algorithm based on leaf cell count and tree structure. If allow_coarsening is true, the algorithm will keep leaf cells together on one rank when needed for local coarsening (i.e. when all children of a cell are leaves).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.positive_part-Tuple{Any}","page":"Trixi.jl","title":"Trixi.positive_part","text":"positive_part(x)\n\nReturn x if x is positive, else zero. In other words, return (x + abs(x)) / 2 for real numbers x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.pressure-Tuple{Real, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.pressure","text":"pressure(rho::Real, equations::LatticeBoltzmannEquations2D)\npressure(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic pressure from the density rho or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.pressure-Tuple{Real, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.pressure","text":"pressure(rho::Real, equations::LatticeBoltzmannEquations3D)\npressure(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic pressure from the density rho or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.prim2cons","page":"Trixi.jl","title":"Trixi.prim2cons","text":"prim2cons(u, equations)\n\nConvert the primitive variables u to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2prim.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.residual_steady_state-Tuple{Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.residual_steady_state","text":"residual_steady_state(du, ::AbstractHyperbolicDiffusionEquations)\n\nUsed to determine the termination criterion of a SteadyStateCallback. For hyperbolic diffusion, this checks convergence of the potential phi.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.rotate_from_x","page":"Trixi.jl","title":"Trixi.rotate_from_x","text":"rotate_from_x(u, normal, equations)\n\nApply the rotation that maps the x-axis onto normal to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.\n\nSee also: rotate_to_x\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.rotate_to_x","page":"Trixi.jl","title":"Trixi.rotate_to_x","text":"rotate_to_x(u, normal, equations)\n\nApply the rotation that maps normal onto the x-axis to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.\n\nSee also: rotate_from_x\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.save_plot-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.save_plot","text":"save_plot(plot_data, variable_names;\n show_mesh=true, plot_arguments=Dict{Symbol,Any}(),\n time=nothing, timestep=nothing)\n\nVisualize the plot data object provided in plot_data and save result as a PNG file in the out directory, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.\n\nThe timestep is used in the filename. time is currently unused by this function.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nSee also: VisualizationCallback, show_plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.show_plot-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.show_plot","text":"show_plot(plot_data, variable_names;\n show_mesh=true, plot_arguments=Dict{Symbol,Any}(),\n time=nothing, timestep=nothing)\n\nVisualize the plot data object provided in plot_data and display result, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.\n\nThis function is the default plot_creator argument for the VisualizationCallback. time and timestep are currently unused by this function.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nSee also: VisualizationCallback, save_plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.solve","page":"Trixi.jl","title":"Trixi.solve","text":"solve(ode, alg; dt, callbacks, kwargs...)\n\nThe following structures and methods provide the infrastructure for SSP Runge-Kutta methods of type SimpleAlgorithmSSP.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::AcousticPerturbationEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations3D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquationsQuasi1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the mozzle width 'a(x) = 1.5 - 0.5 * cos(x[1] * pi)' as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::InviscidBurgersEquation1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::PolytropicEulerEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the bottom topography function b(x,y) = 2 + 0.5 * sin(sqrt(2)*pi*x) + 0.5 * sin(sqrt(2)*pi*y) as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterEquationsQuasi1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the bottom topography function b(x) = 0.2 - 0.05 * sin(sqrt(2) * pi *x[1]) and channel width 'a(x)= 1 + 0.1 * cos(sqrt(2) * pi * x[1])' as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterTwoLayerEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterTwoLayerEquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterTwoLayerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterTwoLayerEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_coupled_euler_gravity-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_coupled_euler_gravity","text":"source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_coupled_euler_gravity-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_coupled_euler_gravity","text":"source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_euler-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_euler","text":"source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_euler-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_euler","text":"source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\nnote: Note\nThis method is to be used for testing pure Euler simulations with analytic self-gravity. If you intend to do coupled Euler-gravity simulations, you need to use source_terms_eoc_test_coupled_euler_gravity instead.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations1D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations2D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations3D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_poisson_nonperiodic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_poisson_nonperiodic","text":"source_terms_poisson_nonperiodic(u, x, t,\n equations::HyperbolicDiffusionEquations1D)\n\nSource terms that include the forcing function f(x) and right hand side for the hyperbolic diffusion system that is used with initial_condition_poisson_nonperiodic and boundary_condition_poisson_nonperiodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_coirier_vanleer-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_coirier_vanleer","text":"splitting_coirier_vanleer(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_coirier_vanleer(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux from Coirier and van Leer. The splitting has correction terms in the pressure splitting as well as the mass and energy flux components. The motivation for these corrections are to handle flows at the low Mach number limit.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nWilliam Coirier and Bram van Leer (1991) Numerical flux formulas for the Euler and Navier-Stokes equations. II - Progress in flux-vector splitting DOI: 10.2514/6.1991-1566\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) similar to a flux splitting one would apply, e.g., to Burgers' equation.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::InviscidBurgersEquation1D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::InviscidBurgersEquation1D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ = abs(u).\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::LinearScalarAdvectionEquation1D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::LinearScalarAdvectionEquation1D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ is the absolute value of the advection velocity.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations3D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations3D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_vanleer_haenel-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_vanleer_haenel","text":"splitting_vanleer_haenel(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nBram van Leer (1982) Flux-Vector Splitting for the Euler Equation DOI: 10.1007/978-3-642-60543-7_5\nD. Hänel, R. Schwane and G. Seider (1987) On the accuracy of upwind schemes for the solution of the Navier-Stokes equations DOI: 10.2514/6.1987-1105\nMeng-Sing Liou and Chris J. Steffen, Jr. (1991) High-Order Polynomial Expansions (HOPE) for Flux-Vector Splitting NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_vanleer_haenel-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_vanleer_haenel","text":"splitting_vanleer_haenel(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nSplitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nBram van Leer (1982) Flux-Vector Splitting for the Euler Equation DOI: 10.1007/978-3-642-60543-7_5\nD. Hänel, R. Schwane and G. Seider (1987) On the accuracy of upwind schemes for the solution of the Navier-Stokes equations DOI: 10.2514/6.1987-1105\nMeng-Sing Liou and Chris J. Steffen, Jr. (1991) High-Order Polynomial Expansions (HOPE) for Flux-Vector Splitting NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.stolarsky_mean-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.stolarsky_mean","text":"stolarsky_mean(x, y, gamma)\n\nCompute an instance of a weighted Stolarsky mean of the form\n\nstolarsky_mean(x, y, gamma) = (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))\n\nwhere gamma > 1.\n\nProblem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Winters et al. (2020). Set f = (y - x) / (y + x) and g = gamma (for compact notation). Then, we use the expansions\n\n((1+f)^g - (1-f)^g) / g = 2*f + (g-1)(g-2)/3 * f^3 + (g-1)(g-2)(g-3)(g-4)/60 * f^5 + O(f^7)\n\nand\n\n((1+f)^(g-1) - (1-f)^(g-1)) / (g-1) = 2*f + (g-2)(g-3)/3 * f^3 + (g-2)(g-3)(g-4)(g-5)/60 * f^5 + O(f^7)\n\nInserting the first few terms of these expansions and performing polynomial long division we find that\n\nstolarsky_mean(x, y, gamma) ≈ (y + x) / 2 * (1 + (g-2)/3 * f^2 - (g+1)(g-2)(g-3)/45 * f^4 + (g+1)(g-2)(g-3)(2g(g-2)-9)/945 * f^6)\n\nSince divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use\n\nf^2 = (y - x)^2 / (x + y)^2\n = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)\n\nGiven ε = 1.0e-4, we use the following algorithm.\n\nif f^2 < ε\n # use the expansion above\nelse\n # use the direct formula (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))\nend\n\nReferences\n\nAndrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w\nAgner Fog. Lists of instruction latencies, throughputs and micro-operation breakdowns for Intel, AMD, and VIA CPUs. https://www.agner.org/optimize/instruction_tables.pdf\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.totalgamma-Tuple{Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.totalgamma","text":"totalgamma(u, equations::CompressibleEulerMulticomponentEquations1D)\n\nFunction that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.totalgamma-Tuple{Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.totalgamma","text":"totalgamma(u, equations::CompressibleEulerMulticomponentEquations2D)\n\nFunction that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.trixi_include-Tuple{Module, AbstractString}","page":"Trixi.jl","title":"Trixi.trixi_include","text":"trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)\n\ninclude the file elixir and evaluate its content in the global scope of module mod. You can override specific assignments in elixir by supplying keyword arguments. It's basic purpose is to make it easier to modify some parameters while running Trixi.jl from the REPL. Additionally, this is used in tests to reduce the computational burden for CI while still providing examples with sensible default values for users.\n\nBefore replacing assignments in elixir, the keyword argument maxiters is inserted into calls to solve and Trixi.solve with it's default value used in the SciML ecosystem for ODEs, see the \"Miscellaneous\" section of the documentation.\n\nExamples\n\njulia> redirect_stdout(devnull) do\n trixi_include(@__MODULE__, joinpath(examples_dir(), \"tree_1d_dgsem\", \"elixir_advection_extended.jl\"),\n tspan=(0.0, 0.1))\n sol.t[end]\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n0.1\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.uses_amr-Tuple{Any}","page":"Trixi.jl","title":"Trixi.uses_amr","text":"uses_amr(callback)\n\nChecks whether the provided callback or CallbackSet is an AMRCallback or contains one.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.varnames","page":"Trixi.jl","title":"Trixi.varnames","text":"varnames(conversion_function, equations)\n\nReturn the list of variable names when applying conversion_function to the conserved variables associated to equations. This function is mainly used internally to determine output to screen and for IO, e.g., for the AnalysisCallback and the SaveSolutionCallback. Common choices of the conversion_function are cons2cons and cons2prim.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, Integer, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, orientation, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y) from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, Integer, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, orientation, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y, 3 -> z) from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic velocity vector from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic velocity vector from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.@autoinfiltrate","page":"Trixi.jl","title":"Trixi.@autoinfiltrate","text":"@autoinfiltrate\n@autoinfiltrate condition::Bool\n\nInvoke the @infiltrate macro of the package Infiltrator.jl to create a breakpoint for ad-hoc interactive debugging in the REPL. If the optional argument condition is given, the breakpoint is only enabled if condition evaluates to true.\n\nAs opposed to using Infiltrator.@infiltrate directly, this macro does not require Infiltrator.jl to be added as a dependency to Trixi.jl. As a bonus, the macro will also attempt to load the Infiltrator module if it has not yet been loaded manually.\n\nNote: For this macro to work, the Infiltrator.jl package needs to be installed in your current Julia environment stack.\n\nSee also: Infiltrator.jl\n\nwarning: Internal use only\nPlease note that this macro is intended for internal use only. It is not part of the public API of Trixi.jl, and it thus can altered (or be removed) at any time without it being considered a breaking change.\n\n\n\n\n\n","category":"macro"},{"location":"reference-trixi/#Trixi.@threaded-Tuple{Any}","page":"Trixi.jl","title":"Trixi.@threaded","text":"@threaded for ... end\n\nSemantically the same as Threads.@threads when iterating over a AbstractUnitRange but without guarantee that the underlying implementation uses Threads.@threads or works for more general for loops. In particular, there may be an additional check whether only one thread is used to reduce the overhead of serial execution or the underlying threading capabilities might be provided by other packages such as Polyester.jl.\n\nwarn: Warn\nThis macro does not necessarily work for general for loops. For example, it does not necessarily support general iterables such as eachline(filename).\n\nSome discussion can be found at https://discourse.julialang.org/t/overhead-of-threads-threads/53964 and https://discourse.julialang.org/t/threads-threads-with-one-thread-how-to-remove-the-overhead/58435.\n\n\n\n\n\n","category":"macro"},{"location":"styleguide/#Style-guide","page":"Style guide","title":"Style guide","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Coding style is an inherently personal - and thus hotly contested - issue. Since code is usually \"written once, read often\", it helps regular developers, new users, and reviewers if code is formatted consistently. We therefore believe in the merit of using a common coding style throughout Trixi.jl, even at the expense that not everyone can be happy with every detailed style decision. If you came here because you are furious about our code formatting rules, here is a happy little whale for you to calm you down: 🐳","category":"page"},{"location":"styleguide/#Conventions","page":"Style guide","title":"Conventions","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"The following lists a few coding conventions for Trixi.jl. Note that in addition to these conventions, we apply and enforce automated source code formatting (see below for more details):","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Modules, types, structs with CamelCase.\nFunctions, variables with lowercase snake_case.\nIndentation with 4 spaces (never tabs!)\nMaximum line length (strictly): 92.\nFunctions that mutate their input are named with a trailing !.\nFunctions order their parameters similar to Julia Base.\nThe main modified argument comes first. For example, if the right-hand side du is modified, it should come first. If only the cache is modified, e.g., in prolong2interfaces! and its siblings, put the cache first.\nOtherwise, use the order mesh, equations, solver, cache.\nIf something needs to be specified in more detail for dispatch, put the additional argument before the general one that is specified in more detail. For example, we use have_nonconservative_terms(equations), equations and dg.mortar, dg.\nPrefer for i in ... to for i = ... for better semantic clarity and greater flexibility.\nExecutable code should only use ASCII characters.\nDocstrings and comments can and should use Unicode characters where it helps understanding.\nMultiline expressions should be explicitly grouped by parentheses and not rely on Julia's implicit line continuation syntax.\nWhen naming multiple functions of a single or similar category, prefer to put the general classification first and the specialization second. Example: Use flux_central instead of central_flux. This helps when searching for available functions on the REPL (e.g., when trying to find all flux functions).","category":"page"},{"location":"styleguide/#automated-source-code-formatting","page":"Style guide","title":"Automated source code formatting","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"We use JuliaFormatter.jl to format the source code of Trixi.jl, which will also enforce some of the Conventions listed above (e.g., line length or indentation with 4 spaces are automatically handled, while capitalization of names is not). Our format is mostly based on the SciML-style formatting rules. For more details you can have a look at the current .JuliaFormatter.toml file that holds the configuration options we use for JuliaFormatter.jl.","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Note that we expect all contributions to Trixi.jl to be formatted with JuliaFormatter.jl before being merged to the main branch. We ensure this by running a automated check on all PRs that verify that running JuliaFormatter.jl again will not change the source code.","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"To format your contributions before created a PR (or, at least, before requesting a review of your PR), you need to install JuliaFormatter.jl first by running","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"julia -e 'using Pkg; Pkg.add(\"JuliaFormatter\")'","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"You can then recursively format the core Julia files in the Trixi.jl repo by executing","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"julia -e 'using JuliaFormatter; format([\"benchmark\", \"examples\", \"ext\", \"src\", \"test\", \"utils\"])'","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"from inside the Trixi.jl repository. For convenience, there is also a script you can directly run from your terminal shell, which will automatically install JuliaFormatter in a temporary environment and then run it:","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"utils/trixi-format.jl","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"You can get more information about using the convenience script by running it with the --help/-h flag.","category":"page"},{"location":"styleguide/#Checking-formatting-before-committing","page":"Style guide","title":"Checking formatting before committing","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"It can be convenient to check the formatting of source code automatically before each commit. We use git-hooks for it and provide a pre-commit script in the utils folder. The script uses JuliaFormatter.jl just like formatting script that runs over the whole Trixi.jl directory. You can copy the pre-commit-script into .git/hooks/pre-commit and it will check your formatting before each commit. If errors are found the commit is aborted and you can add the corrections via","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"git add -p","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"EditURL = \"../../literate/src/files/scalar_linear_advection_1d.jl\"","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#scalar_linear_advection_1d","page":"1 Introduction to DG methods","title":"1: Introduction to DG methods","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"This tutorial is about how to set up a simple way to approximate the solution of a hyperbolic partial differential equation. First, we will implement a basic and naive algorithm. Then, we will use predefined features from Trixi.jl to show how you can use Trixi.jl on your own.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We will implement the scalar linear advection equation in 1D with the advection velocity 1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u_t + u_x = 0 textfor tin mathbbR^+ xinOmega=-11","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We define the domain Omega by setting the boundaries.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"coordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We assume periodic boundaries and the following initial condition.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"initial_condition_sine_wave(x) = 1.0 + 0.5 * sin(pi * x)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#The-discontinuous-Galerkin-collocation-spectral-element-method-(DGSEM)","page":"1 Introduction to DG methods","title":"The discontinuous Galerkin collocation spectral element method (DGSEM)","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/#i.-Discretization-of-the-physical-domain","page":"1 Introduction to DG methods","title":"i. Discretization of the physical domain","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To improve precision we want to approximate the solution on small parts of the physical domain. So, we split the domain Omega=-1 1 into elements Q_l of length dx.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"n_elements = 16 # number of elements\n\ndx = (coordinates_max - coordinates_min) / n_elements # length of one element","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To make the calculation more efficient and storing less information, we transform each element Q_l with center point x_l to a reference element E=-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Q_l=Bigx_l-fracdx2 x_l+fracdx2Big undersetx(xi)oversetxi(x)rightleftarrows -1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"So, for every element the transformation from the reference domain to the physical domain is defined by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"x(xi) = x_l + fracdx2 xi xiin-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Therefore,","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"beginalign*\nu = u(x(xi) t) \nu_x = u_xi fracdxidx 3pt\nfracdxidx = (x_xi)^-1 = frac2dx = J^-1 \nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Here, J is the Jacobian determinant of the transformation.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Using this transformation, we can transform our equation for each element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"fracdx2 u_t^Q_l + u_xi^Q_l = 0 text for tinmathbbR^+ xiin-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Here, u_t^Q_l and u_xi^Q_l denote the time and spatial derivatives of the solution on the element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#ii.-Polynomial-approach","page":"1 Introduction to DG methods","title":"ii. Polynomial approach","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Now, we want to approximate the solution in each element Q_l by a polynomial of degree N. Since we transformed the equation, we can use the same polynomial approach for the reference coordinate xiin-1 1 in every physical element Q_l. This saves a lot of resources by reducing the amount of calculations needed and storing less information.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"For DGSEM we choose Lagrange basis functions l_j_j=0^N as our polynomial basis of degree N in -1 1. The solution in element Q_l can be approximated by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u(x(xi) t)big_Q_l approx u^Q_l(xi t) = sum_j=0^N u_j^Q_l(t) l_j(xi)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"with N+1 coefficients u_j^Q_l_j=0^N. By construction the Lagrange basis has some useful advantages. This basis is defined by N+1 nodes, which fulfill a Kronecker property at the exact same nodes. Let xi_i_i=0^N be these nodes.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"l_j(xi_i) = delta_ij =\nbegincases\n1 textif i=j \n0 textelse\nendcases","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Because of this property, the polynomial coefficients are exact the values of u^Q_l at the nodes","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u^Q_l(xi_i t) = sum_j=0^N u_j^Q_l(t) underbracel_j(xi_i)_=delta_ij = u_i^Q_l(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Next, we want to select the nodes xi_i_i=0^N, which we use for the construction of the Lagrange polynomials. We choose the N+1 Gauss-Lobatto nodes, which are used for the Gaussian-Lobatto quadrature. These always contain the boundary points at -1 and +1 and are well suited as interpolation nodes. The corresponding weights will be referred to as w_j_j=0^N. In Trixi.jl the basis with Lagrange polynomials on Gauss-Lobatto nodes is already defined.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using Trixi\npolydeg = 3 #= polynomial degree = N =#\nbasis = LobattoLegendreBasis(polydeg)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"The Gauss-Lobatto nodes are","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"nodes = basis.nodes","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"with the corresponding weights","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"weights = basis.weights","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To illustrate how you can integrate using numerical quadrature with this Legendre-Gauss-Lobatto nodes, we give an example for f(x)=x^3. Since f is of degree 3, a polynomial interpolation with N=3 is exact. Therefore, the integral on -1 1 can be calculated by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"beginalign*\nint_-1^1 f(x) dx = int_-1^1 Big( sum_j=0^3 f(xi_j)l_j(x) Big) dx\n= sum_j=0^3 f(xi_j) int_-1^1 l_j(x)dx \n= sum_j=0^3 f(xi_j) w_j\n= sum_j=0^3 xi_j^3 w_j\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Let's use our nodes and weights for N=3 and plug in","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"integral = sum(nodes.^3 .* weights)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Using this polynomial approach leads to the equation","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"fracdx2 dotu^Q_l(xi t) + u^Q_l(xi t) = 0","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"with dotu=fracpartialpartial tu and u=fracpartialpartial xu. To approximate the solution, we need to get the polynomial coefficients u_j^Q_l_j=0^N for every element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"After defining all nodes, we can implement the spatial coordinate x and its initial value u0 for every node.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"x = Matrix{Float64}(undef, length(nodes), n_elements)\nfor element in 1:n_elements\n x_l = coordinates_min + (element - 1) * dx + dx/2\n for i in 1:length(nodes)\n ξ = nodes[i] # nodes in [-1, 1]\n x[i, element] = x_l + dx/2 * ξ\n end\nend\n\nu0 = initial_condition_sine_wave.(x)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To have a look at the initial sinus curve, we plot it.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using Plots\nplot(vec(x), vec(u0), label=\"initial condition\", legend=:topleft)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#iii.-Variational-formulation","page":"1 Introduction to DG methods","title":"iii. Variational formulation","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"After defining the equation and initial condition, we want to implement an algorithm to approximate the solution.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"From now on, we only write u instead of u^Q_l for simplicity, but consider that all the following calculation only concern one element. Multiplying the new equation with the smooth Lagrange polynomials l_i_i=0^N (test functions) and integrating over the reference element E=-11, we get the variational formulation of our transformed partial differential equation for i=0N:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"beginalign*\nint_-1^1 Big( fracdx2 dotu(xi t) + u(xi t) Big) l_i(xi)dxi\n = underbracefracdx2 int_-1^1 dotu(xi t) l_i(xi)dxi_textTerm I + underbraceint_-1^1 u(xi t) l_i(xi)dxi_textTerm II = 0\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We deal with the two terms separately. We write int_-1 N^1 cdot dxi for the approximation of the integral using numerical quadrature with N+1 basis points. We use the Gauss-Lobatto nodes again. The numerical scalar product langlecdot cdotrangle_N is defined by langle f grangle_N = int_-1 N^1 f(xi) g(xi) dxi.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Term-I:","page":"1 Introduction to DG methods","title":"Term I:","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"In the following calculation we approximate the integral numerically with quadrature on the Gauss-Lobatto nodes xi_i_i=0^N and then use the Kronecker property of the Lagrange polynomials. This approach of using the same nodes for the interpolation and quadrature is called collocation.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"beginalign*\nfracdx2 int_-1^1 dotu(xi t) l_i(xi)dxi\napprox fracdx2 int_-1 N^1 dotu(xi t) l_i(xi)dxi \n= fracdx2 sum_k=0^N underbracedotu(xi_k t)_=dotu_k(t) underbracel_i(xi_k)_=delta_kiw_k \n= fracdx2 dotu_i(t) w_i\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We define the Legendre-Gauss-Lobatto (LGL) mass matrix M and by the Kronecker property follows:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"M_ij = langle l_j l_irangle_N = delta_ij w_j ij=0N","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using LinearAlgebra\nM = diagm(weights)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Now, we can write the integral with this new matrix.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"fracdx2 int_-1 N^1 dotu(xi t) underlinel(xi)dxi = fracdx2 M underlinedotu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"where underlinedotu = (dotu_0 dotu_N)^T and underlinel respectively.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Note: Since the LGL quadrature with N+1 nodes is exact up to functions of degree 2N-1 and dotu(xi t) l_i(xi) is of degree 2N, in general the following holds","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"int_-1^1 dotu(xi t) l_i(xi) dxi neq int_-1 N^1 dotu(xi t) l_i(xi) dxi","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"With an exact integration the mass matrix would be dense. Choosing numerical integrating and quadrature with the exact same nodes (collocation) leads to the sparse and diagonal mass matrix M. This is called mass lumping and has the big advantage of an easy invertation of the matrix.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Term-II:","page":"1 Introduction to DG methods","title":"Term II:","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We use spatial partial integration for the second term:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"int_-1^1 u(xi t) l_i(xi) dxi = u l_i_-1^1 - int_-1^1 u l_idxi","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"The resulting integral can be solved exactly with LGL quadrature since the polynomial is now of degree 2N-1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Again, we split the calculation in two steps.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Surface-term","page":"1 Introduction to DG methods","title":"Surface term","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"As mentioned before, we approximate the solution with a polynomial in every element. Therefore, in general the value of this approximation at the interfaces between two elements is not unique. To solve this problem we introduce the idea of the numerical flux u^*, which will give an exact value at the interfaces. One of many different approaches and definitions for the calculation of the numerical flux we will deal with in 4. Numerical flux.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u l_i_-1^1 = u^*big^1 l_i(+1) - u^*big_-1 l_i(-1)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Since the Gauss-Lobatto nodes contain the element boundaries -1 and +1, we can use the Kronecker property of l_i for the calculation of l_i(-1) and l_i(+1).","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u underlinel_-1^1 = u^*big^1 left(beginarrayc 0 vdots 0 1 endarrayright)\n- u^*big_-1 left(beginarrayc 1 0 vdots 0endarrayright)\n= B underlineu^*(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"with the boundary matrix","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"B = beginpmatrix\n-1 0 cdots 0\n0 0 cdots 0\nvdots vdots 0 0\n0 cdots 0 1\nendpmatrix\nqquadtextandqquad\nunderlineu^*(t) = left(beginarrayc u^*big_-1 0 vdots 0 u^*big^1endarrayright)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"B = diagm([-1; zeros(polydeg - 1); 1])","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Volume-term","page":"1 Introduction to DG methods","title":"Volume term","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"As mentioned before, the new integral can be solved exact since the function inside is of degree 2N-1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"- int_-1^1 u l_idxi = - int_-1 N^1 u l_i dxi\n= - sum_k=0^N u(xi_k t) l_i(xi_k) w_k\n= - sum_k=0^N u_k(t) D_ki w_k","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"where D is the derivative matrix defined by D_ki = l_i(xi_k) for ik=0N.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"D = basis.derivative_matrix","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To show why this matrix is called the derivative matrix, we go back to our example f(x)=x^3. We calculate the derivation of f at the Gauss-Lobatto nodes xi_k_k=0^N with N=8.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"f_x=xi_k = Big( sum_j=0^8 f(xi_j) l_j(x) Big)_x=xi_k = sum_j=0^8 f(xi_j) l_j(xi_k)\n= sum_j=0^8 f(xi_j) D_kj","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"for k=0N and therefore, underlinef = D underlinef.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"basis_N8 = LobattoLegendreBasis(8)\nplot(vec(x), x -> 3 * x^2, label=\"f'\", lw=2)\nscatter!(basis_N8.nodes, basis_N8.derivative_matrix * basis_N8.nodes.^3, label=\"Df\", lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Combining the volume term for every i=0N results in","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"int_-1^1 u underlinel dxi = - D^T M underlineu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Putting all parts together we get the following equation for the element Q_l","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"fracdx2 M underlinedotu(t) = - B underlineu^*(t) + D^T M underlineu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"or equivalent","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"underlinedotu^Q_l(t) = frac2dx Big - M^-1 B underlineu^Q_l^*(t) + M^-1 D^T M underlineu^Q_l(t)Big","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"This is called the weak form of the DGSEM.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Note: For every element Q_l we get a system of N+1 ordinary differential equations to calculate N+1 coefficients. Since the numerical flux u^* is depending on extern values at the interfaces, the equation systems of adjacent elements are weakly linked.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#numerical_flux","page":"1 Introduction to DG methods","title":"iv. Numerical flux","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"As mentioned above, we still have to handle the problem of different values at the same point at the interfaces. This happens with the ideas of the numerical flux f^*(u)=u^*. The role of f^* might seem minor in this simple example, but is important for more complicated problems. There are two values at the same spatial coordinate. Let's say we are looking at the interface between the elements Q_l and Q_l+1, while both elements got N+1 nodes as defined before. We call the first value of the right element u_R=u_0^Q_l+1 and the last one of the left element u_L=u_N^Q_l. So, for the value of the numerical flux on that interface the following holds","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u^* = u^*(u_L u_R)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"These values are interpreted as start values of a so-called Riemann problem. There are many different (approximate) Riemann solvers available and useful for different problems. We will use the local Lax-Friedrichs flux.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"surface_flux = flux_lax_friedrichs","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"The only missing ingredient is the flux calculation at the boundaries -1 and +1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"u^Q_first^*big_-1 = u^Q_first^*big_-1(u^bound(-1) u_R)\nquadtextandquad\nu^Q_last^*big^1 = u^Q_last^*big^1(u_L u^bound(1))","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"The boundaries are periodic, which means that the last value of the last element u^Q_last_N is used as u_L at the first interface and accordingly for the other boundary.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Now, we implement a function, that calculates underlinedotu^Q_l for the given matrices, underlineu and underlineu^*.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"function rhs!(du, u, x, t)\n # Reset du and flux matrix\n du .= zero(eltype(du))\n flux_numerical = copy(du)\n\n # Calculate interface and boundary fluxes, $u^* = (u^*|_{-1}, 0, ..., 0, u^*|^1)^T$\n # Since we use the flux Lax-Friedrichs from Trixi.jl, we have to pass some extra arguments.\n # Trixi.jl needs the equation we are dealing with and an additional `1`, that indicates the\n # first coordinate direction.\n equations = LinearScalarAdvectionEquation1D(1.0)\n for element in 2:n_elements-1\n # left interface\n flux_numerical[1, element] = surface_flux(u[end, element-1], u[1, element], 1, equations)\n flux_numerical[end, element-1] = flux_numerical[1, element]\n # right interface\n flux_numerical[end, element] = surface_flux(u[end, element], u[1, element+1], 1, equations)\n flux_numerical[1, element+1] = flux_numerical[end, element]\n end\n # boundary flux\n flux_numerical[1, 1] = surface_flux(u[end, end], u[1, 1], 1, equations)\n flux_numerical[end, end] = flux_numerical[1, 1]\n\n # Calculate surface integrals, $- M^{-1} * B * u^*$\n for element in 1:n_elements\n du[:, element] -= (M \\ B) * flux_numerical[:, element]\n end\n\n # Calculate volume integral, $+ M^{-1} * D^T * M * u$\n for element in 1:n_elements\n flux = u[:, element]\n du[:, element] += (M \\ transpose(D)) * M * flux\n end\n\n # Apply Jacobian from mapping to reference element\n for element in 1:n_elements\n du[:, element] *= 2 / dx\n end\n\n return nothing\nend","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Combining all definitions and the function that calculates the right-hand side, we define the ODE and solve it until t=2 with OrdinaryDiffEq's solve function and the Runge-Kutta method RDPK3SpFSAL49(), which is optimized for discontinuous Galerkin methods and hyperbolic PDEs. We set some common error tolerances abstol=1.0e-6, reltol=1.0e-6 and pass save_everystep=false to avoid saving intermediate solution vectors in memory.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using OrdinaryDiffEq\ntspan = (0.0, 2.0)\node = ODEProblem(rhs!, u0, tspan, x)\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)\n\nplot(vec(x), vec(sol.u[end]), label=\"solution at t=$(tspan[2])\", legend=:topleft, lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Alternative-Implementation-based-on-Trixi.jl","page":"1 Introduction to DG methods","title":"Alternative Implementation based on Trixi.jl","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Now, we implement the same example. But this time, we directly use the functionality that Trixi.jl provides.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using Trixi, OrdinaryDiffEq, Plots","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"First, define the equation with a advection_velocity of 1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"advection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Then, create a DG solver with polynomial degree = 3 and (local) Lax-Friedrichs/Rusanov flux as surface flux. The implementation of the basis and the numerical flux is now already done.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We will now create a mesh with 16 elements for the physical domain [-1, 1] with periodic boundaries. We use Trixi.jl's standard mesh TreeMesh. Since it's limited to hypercube domains, we choose 2^4=16 elements. The mesh type supports AMR, that' why n_cells_max has to be set, even if we don't need AMR here.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"coordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4, # number of elements = 2^4\n n_cells_max=30_000) # set maximum capacity of tree data structure (only needed for AMR)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"A semidiscretization collects data structures and functions for the spatial discretization. In Trixi.jl, an initial condition has the following parameter structure and is of the type SVector.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"initial_condition_sine_wave(x, t, equations) = SVector(1.0 + 0.5 * sin(pi * sum(x - equations.advection_velocity * t)))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_sine_wave, solver)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"Again, combining all definitions and the function that calculates the right-hand side, we define the ODE and solve it until t=2 with OrdinaryDiffEq's solve function and the Runge-Kutta method RDPK3SpFSAL49().","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"tspan = (0.0, 2.0)\node_trixi = semidiscretize(semi, tspan)\n\nsol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);\nnothing #hide","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"We add a plot of the new approximated solution to the one calculated before.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"plot!(sol_trixi, label=\"solution at t=$(tspan[2]) with Trixi.jl\", legend=:topleft, linestyle=:dash, lw=2)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Summary-of-the-code","page":"1 Introduction to DG methods","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"To sum up, here is the complete code that we used.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Raw-implementation","page":"1 Introduction to DG methods","title":"Raw implementation","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"# basis: Legendre-Gauss-Lobatto\nusing Trixi, LinearAlgebra, OrdinaryDiffEq, Plots\npolydeg = 3 #= polynomial degree =#\nbasis = LobattoLegendreBasis(polydeg)\nnodes = basis.nodes # Gauss-Lobatto nodes in [-1, 1]\nD = basis.derivative_matrix\nM = diagm(basis.weights) # mass matrix\nB = diagm([-1; zeros(polydeg - 1); 1])\n\n# mesh\ncoordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nn_elements = 16 # number of elements\n\ndx = (coordinates_max - coordinates_min) / n_elements # length of one element\n\nx = Matrix{Float64}(undef, length(nodes), n_elements)\nfor element in 1:n_elements\n x_l = -1 + (element - 1) * dx + dx/2\n for i in 1:length(nodes) # basis points in [-1, 1]\n ξ = nodes[i]\n x[i, element] = x_l + dx/2 * ξ\n end\nend\n\n# initial condition\ninitial_condition_sine_wave(x) = 1.0 + 0.5 * sin(pi * x)\nu0 = initial_condition_sine_wave.(x)\n\nplot(vec(x), vec(u0), label=\"initial condition\", legend=:topleft)\n\n# flux Lax-Friedrichs\nsurface_flux = flux_lax_friedrichs\n\n# rhs! method\nfunction rhs!(du, u, x, t)\n # reset du\n du .= zero(eltype(du))\n flux_numerical = copy(du)\n\n # calculate interface and boundary fluxes\n equations = LinearScalarAdvectionEquation1D(1.0)\n for element in 2:n_elements-1\n # left interface\n flux_numerical[1, element] = surface_flux(u[end, element-1], u[1, element], 1, equations)\n flux_numerical[end, element-1] = flux_numerical[1, element]\n # right interface\n flux_numerical[end, element] = surface_flux(u[end, element], u[1, element+1], 1, equations)\n flux_numerical[1, element+1] = flux_numerical[end, element]\n end\n # boundary flux\n flux_numerical[1, 1] = surface_flux(u[end, end], u[1, 1], 1, equations)\n flux_numerical[end, end] = flux_numerical[1, 1]\n\n # calculate surface integrals\n for element in 1:n_elements\n du[:, element] -= (M \\ B) * flux_numerical[:, element]\n end\n\n # calculate volume integral\n for element in 1:n_elements\n flux = u[:, element]\n du[:, element] += (M \\ transpose(D)) * M * flux\n end\n\n # apply Jacobian from mapping to reference element\n for element in 1:n_elements\n du[:, element] *= 2 / dx\n end\n\n return nothing\nend\n\n# create ODE problem\ntspan = (0.0, 2.0)\node = ODEProblem(rhs!, u0, tspan, x)\n\n# solve\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)\n\nplot(vec(x), vec(sol.u[end]), label=\"solution at t=$(tspan[2])\", legend=:topleft, lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Alternative-Implementation-based-on-Trixi.jl-2","page":"1 Introduction to DG methods","title":"Alternative Implementation based on Trixi.jl","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using Trixi, OrdinaryDiffEq, Plots\n\n# equation with a advection_velocity of `1`.\nadvection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)\n\n# create DG solver with flux lax friedrichs and LGL basis\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\n# distretize domain with `TreeMesh`\ncoordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4, # number of elements = 2^4\n n_cells_max=30_000)\n\n# create initial condition and semidiscretization\ninitial_condition_sine_wave(x, t, equations) = SVector(1.0 + 0.5 * sin(pi * sum(x - equations.advection_velocity * t)))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_sine_wave, solver)\n\n# solve\ntspan = (0.0, 2.0)\node_trixi = semidiscretize(semi, tspan)\nsol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);\n\nplot!(sol_trixi, label=\"solution at t=$(tspan[2]) with Trixi.jl\", legend=:topleft, linestyle=:dash, lw=2)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Package-versions","page":"1 Introduction to DG methods","title":"Package versions","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"1 Introduction to DG methods","title":"1 Introduction to DG methods","text":"This page was generated using Literate.jl.","category":"page"},{"location":"reference-trixi2vtk/#Trixi2Vtk.jl-API","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl API","text":"","category":"section"},{"location":"reference-trixi2vtk/","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl","text":"CurrentModule = Trixi2Vtk","category":"page"},{"location":"reference-trixi2vtk/","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl","text":"Modules = [Trixi2Vtk]","category":"page"},{"location":"reference-trixi2vtk/#Trixi2Vtk.trixi2vtk-Tuple{Vararg{AbstractString}}","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.trixi2vtk","text":"trixi2vtk(filename::AbstractString...;\n format=:vtu, verbose=false, hide_progress=false, pvd=nothing,\n output_directory=\".\", nvisnodes=nothing, save_celldata=true,\n reinterpolate=true, data_is_uniform=false)\n\nConvert Trixi-generated output files to VTK files (VTU or VTI).\n\nArguments\n\nfilename: One or more Trixi solution/restart/mesh files to convert to a VTK file. Filenames support file globbing, e.g., \"solution*\" to match all files starting with solution.\nformat: Output format for solution/restart files. Can be 'vtu' or 'vti'.\nverbose: Set to true to enable verbose output.\nhide_progress: Hide progress bar (will be hidden automatically if verbose is true).\npvd: Use this filename to store PVD file (instead of auto-detecting name). Note that only the name will be used (directory and file extension are ignored).\noutput_directory: Output directory where generated files are stored.\nnvisnodes: Number of visualization nodes per element. (default: number of DG nodes for StructuredMesh or UnstructuredMesh2D, twice the number of DG nodes for TreeMesh). A value of 0 (zero) uses the number of nodes in the DG elements.\nsave_celldata: Boolean value to determine if cell-based data should be saved. (default: true)\nreinterpolate: Boolean value to determine if data should be reinterpolated onto uniform points. When false the raw data at the compute nodes is copied into the appropriate format. (default: true)\ndata_is_uniform: Boolean to indicate if the data to be converted is from a finite difference method on a uniform grid of points. (default: false)\n\nExamples\n\njulia> trixi2vtk(\"out/solution_000*.h5\")\n[...]\n\n\n\n\n\n","category":"method"}] +[{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"EditURL = \"../../literate/src/files/adaptive_mesh_refinement.jl\"","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#adaptive_mesh_refinement","page":"14 Adaptive mesh refinement","title":"14: Adaptive mesh refinement","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Adaptive mesh refinement (AMR) is a method of adapting the resolution of the numerical method to the solution features such as turbulent regions or shocks. In those critical regions of the domain, we want the simulation to use elements with smaller mesh sizes compared to other regions. This should be automatically and dynamically adapted during the run of the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Implementation-in-Trixi.jl","page":"14 Adaptive mesh refinement","title":"Implementation in Trixi.jl","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"In Trixi.jl, AMR is possible for the mesh types TreeMesh and P4estMesh. Both meshes are organized in a tree structure and therefore, each element can be refined independently. In Trixi.jl, AMR is restricted to a 2:1 refinement ratio between neighbor elements. This means that the maximum resolution difference of neighboring elements is a factor of two.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"The implementation of AMR is divided into different steps. The basic refinement setting contains an indicator and a controller. These are added to the simulation by using an AMR callback.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Indicators","page":"14 Adaptive mesh refinement","title":"Indicators","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"An indicator estimates the current accuracy of the numerical approximation. It indicates which regions of the domain need finer or coarser resolutions. In Trixi.jl, you can use for instance IndicatorLöhner and IndicatorHennemannGassner.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"IndicatorLöhner (also callable with IndicatorLoehner) is an interpretation and adaptation of a FEM indicator by Löhner (1987) and estimates a weighted second derivative of a specified variable locally.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_indicator = IndicatorLöhner(semi, variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"All indicators have the parameter variable which is used to specify the variable for the indicator calculation. You can use for instance density, pressure or density_pressure for the compressible Euler equations. Moreover, you have the option to use simply the first conservation variable with first for any equations. This might be a good choice for a starting example.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"IndicatorHennemannGassner, also used as a shock-capturing indicator, was developed by Hennemann et al. (2021) and is explained in detail in the tutorial about shock-capturing. It can be constructed as follows.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_indicator = IndicatorHennemannGassner(semi,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Another indicator is the very basic IndicatorMax. It indicates the maximal value of a variable and is therefore mostly used for verification and testing. But it might be useful for the basic understanding of the implementation of indicators and AMR in Trixi.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_indicator = IndicatorMax(semi, variable=variable)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Controllers","page":"14 Adaptive mesh refinement","title":"Controllers","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"The spatial discretization into elements is tree-based for both AMR supporting mesh types TreeMesh and P4estMesh. Thus, the higher the level in the tree the higher the level of refinement. For instance, a mesh element of level 3 has double resolution in each direction compared to another element with level 2.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"To map specific indicator values to a desired level of refinement, Trixi.jl uses controllers. They are build in three levels: There is a base level of refinement base_level, which is the minimum allowed refinement level. Then, there is a medium level med_level, which corresponds to the initial level of refinement, for indicator values above the threshold med_threshold and equally, a maximal level max_level for values above max_threshold. This variant of controller is called ControllerThreeLevel in Trixi.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevel(semi, amr_indicator;\n base_level=4,\n med_level=5, med_threshold=0.1,\n max_level=6, max_threshold=0.6)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"You can also set med_level=0 to use the current level as target, see the docstring of ControllerThreeLevel.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"An extension is ControllerThreeLevelCombined, which uses two different indicators. The primary indicator works the same as the single indicator for ControllerThreeLevel. The second indicator with its own maximum threshold adds the property, that the target level is set to max_level additionally if this indicator's value is greater than max_threshold_secondary. This is for instance used to assure that a shock has always the maximum refinement level.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;\n base_level=2,\n med_level=6, med_threshold=0.0003,\n max_level=8, max_threshold=0.003,\n max_threshold_secondary=0.3)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"This controller is for instance used in elixir_euler_astro_jet_amr.jl.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Callback","page":"14 Adaptive mesh refinement","title":"Callback","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"The AMR indicator and controller are added to the simulation through the callback AMRCallback. It contains a semidiscretization semi, the controller amr_controller and the parameters interval, adapt_initial_condition, and adapt_initial_condition_only_refine.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Adaptive mesh refinement will be performed every interval time steps. adapt_initial_condition indicates whether the initial condition already should be adapted before the first time step. And with adapt_initial_condition_only_refine=true the mesh is only refined at the beginning but not coarsened.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Exemplary-simulation","page":"14 Adaptive mesh refinement","title":"Exemplary simulation","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Here, we want to implement a simple AMR simulation of the 2D linear advection equation for a Gaussian pulse.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"using OrdinaryDiffEq\nusing Trixi\n\nadvection_velocity = (0.2, -0.7)\nequations = LinearScalarAdvectionEquation2D(advection_velocity)\n\ninitial_condition = initial_condition_gauss\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\ncoordinates_min = (-5.0, -5.0)\ncoordinates_max = ( 5.0, 5.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n n_cells_max=30_000)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\n\ntspan = (0.0, 10.0)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"For the best understanding about indicators and controllers, we use the simple AMR indicator IndicatorMax. As described before, it returns the maximal value of the specified variable (here the only conserved variable). Therefore, regions with a high maximum are refined. This is not really useful numerical application, but a nice demonstration example.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_indicator = IndicatorMax(semi, variable=first)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"These values are transferred to a refinement level with the ControllerThreeLevel, such that every element with maximal value greater than 0.1 is refined once and elements with maximum above 0.6 are refined twice.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"amr_controller = ControllerThreeLevel(semi, amr_indicator,\n base_level=4,\n med_level=5, med_threshold=0.1,\n max_level=6, max_threshold=0.6)\n\namr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(amr_callback, stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Running the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"We plot the solution and add the refined mesh at the end of the simulation.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#More-examples","page":"14 Adaptive mesh refinement","title":"More examples","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Trixi.jl provides many elixirs using AMR. We want to give some examples for different mesh types:","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"elixir_euler_blast_wave_amr.jl for TreeMesh and P4estMesh\nelixir_euler_kelvin_helmholtz_instability_amr.jl for TreeMesh\nelixir_euler_double_mach_amr.jl for P4estMesh","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Animations of more interesting and complicated AMR simulations can be found below and on Trixi.jl's youtube channel \"Trixi Framework\".","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"First, we give a purely hyperbolic simulation of a Sedov blast wave with self-gravity. This simulation uses the mesh type TreeMesh as we did and the AMR indicator IndicatorHennemannGassner.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":" \n
","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Source: Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"The next example is a numerical simulation of an ideal MHD rotor on an unstructured AMR mesh. The used mesh type is a P4estMesh.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":" \n
","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"Source: Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"For more information, please have a look at the respective links.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/#Package-versions","page":"14 Adaptive mesh refinement","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"","category":"page"},{"location":"tutorials/adaptive_mesh_refinement/","page":"14 Adaptive mesh refinement","title":"14 Adaptive mesh refinement","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"EditURL = \"../../literate/src/files/structured_mesh_mapping.jl\"","category":"page"},{"location":"tutorials/structured_mesh_mapping/#structured_mesh_mapping","page":"15 Structured mesh with curvilinear mapping","title":"15: Structured mesh with curvilinear mapping","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Here, we want to introduce another mesh type of Trixi.jl. More precisely, this tutorial is about the curved mesh type StructuredMesh supporting curved meshes.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Creating-a-curved-mesh","page":"15 Structured mesh with curvilinear mapping","title":"Creating a curved mesh","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"There are two basic options to define a curved StructuredMesh in Trixi.jl. You can implement curves for the domain boundaries, or alternatively, set up directly the complete transformation mapping. We now present one short example each.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Mesh-defined-by-domain-boundary-curves","page":"15 Structured mesh with curvilinear mapping","title":"Mesh defined by domain boundary curves","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Both examples are based on a semdiscretization of the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"using OrdinaryDiffEq\nusing Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"We start with a pressure perturbation at (xs, 0.0) as initial condition.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"function initial_condition_pressure_perturbation(x, t, equations::CompressibleEulerEquations2D)\n xs = 1.5 # location of the initial disturbance on the x axis\n w = 1/8 # half width\n p = exp(-log(2) * ((x[1]-xs)^2 + x[2]^2)/w^2) + 1.0\n v1 = 0.0\n v2 = 0.0\n rho = 1.0\n\n return prim2cons(SVector(rho, v1, v2, p), equations)\nend\ninitial_condition = initial_condition_pressure_perturbation","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Initialize every boundary as a boundary_condition_slip_wall.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"boundary_conditions = boundary_condition_slip_wall","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"The approximation setup is an entropy-stable split-form DG method with polydeg=4. We are using the two fluxes flux_ranocha and flux_lax_friedrichs.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"solver = DGSEM(polydeg=4, surface_flux=flux_lax_friedrichs,\n volume_integral=VolumeIntegralFluxDifferencing(flux_ranocha))","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"We want to define a circular cylinder as physical domain. It contains an inner semicircle with radius r0 and an outer semicircle of radius r1.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"(Image: )","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"The domain boundary curves with curve parameter in -11 are sorted as shown in the sketch. They always are orientated from negative to positive coordinate, such that the corners have to fit like this f_1(+1) = f_4(-1), f_3(+1) = f_2(-1), etc.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"In our case we can define the domain boundary curves as follows:","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"r0 = 0.5 # inner radius\nr1 = 5.0 # outer radius\nf1(xi) = SVector( r0 + 0.5 * (r1 - r0) * (xi + 1), 0.0) # right line\nf2(xi) = SVector(-r0 - 0.5 * (r1 - r0) * (xi + 1), 0.0) # left line\nf3(eta) = SVector(r0 * cos(0.5 * pi * (eta + 1)), r0 * sin(0.5 * pi * (eta + 1))) # inner circle\nf4(eta) = SVector(r1 * cos(0.5 * pi * (eta + 1)), r1 * sin(0.5 * pi * (eta + 1))) # outer circle","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"We create a curved mesh with 16 x 16 elements. The defined domain boundary curves are passed as a tuple.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"cells_per_dimension = (16, 16)\nmesh = StructuredMesh(cells_per_dimension, (f1, f2, f3, f4), periodicity=false)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Then, we define the simulation with endtime T=3 with semi, ode and callbacks.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_conditions)\n\ntspan = (0.0, 3.0)\node = semidiscretize(semi, tspan)\n\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi, interval=analysis_interval)\n\nalive_callback = AliveCallback(analysis_interval=analysis_interval)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(analysis_callback,\n alive_callback,\n stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Running the simulation","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"pd = PlotData2D(sol)\nplot(pd[\"p\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Mesh-directly-defined-by-the-transformation-mapping","page":"15 Structured mesh with curvilinear mapping","title":"Mesh directly defined by the transformation mapping","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"As mentioned before, you can also define the domain for a StructuredMesh by directly setting up a transformation mapping. Here, we want to present a nice mapping, which is often used to test free-stream preservation. Exact free-stream preservation is a crucial property of any numerical method on curvilinear grids. The mapping is a reduced 2D version of the mapping described in Rueda-Ramírez et al. (2021), p.18.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"using OrdinaryDiffEq\nusing Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"As mentioned, this mapping is used for testing free-stream preservation. So, we use a constant initial condition.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"initial_condition = initial_condition_constant\n\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"We define the transformation mapping with variables in -1 1 as described in Rueda-Ramírez et al. (2021), p.18 (reduced to 2D):","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"function mapping(xi_, eta_)\n # Transform input variables between -1 and 1 onto [0,3]\n xi = 1.5 * xi_ + 1.5\n eta = 1.5 * eta_ + 1.5\n\n y = eta + 3/8 * (cos(1.5 * pi * (2 * xi - 3)/3) *\n cos(0.5 * pi * (2 * eta - 3)/3))\n\n x = xi + 3/8 * (cos(0.5 * pi * (2 * xi - 3)/3) *\n cos(2 * pi * (2 * y - 3)/3))\n\n return SVector(x, y)\nend","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Instead of a tuple of boundary functions, the mesh now has the mapping as its parameter.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"cells_per_dimension = (16, 16)\nmesh = StructuredMesh(cells_per_dimension, mapping)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan)\n\nanalysis_callback = AnalysisCallback(semi, interval=250)\n\nstepsize_callback = StepsizeCallback(cfl=0.8)\n\ncallbacks = CallbackSet(analysis_callback,\n stepsize_callback)\n\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Now, we want to verify the free-stream preservation property and plot the mesh. For the verification, we calculate the absolute difference of the first conservation variable density u[1] and 1.0. To plot this error and the mesh, we are using the visualization feature ScalarPlotData2D, explained in visualization.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"error_density = let u = Trixi.wrap_array(sol.u[end], semi)\n abs.(u[1, :, :, :] .- 1.0) # density, x, y, elements\nend\npd = ScalarPlotData2D(error_density, semi)\n\nusing Plots\nplot(pd, title=\"Error in density\")\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"We observe that the errors in the variable density are at the level of machine accuracy. Moreover, the plot shows the mesh structure resulting from our transformation mapping.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"Of course, you can also use other mappings as for instance shifts by (x y)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"mapping(xi, eta) = SVector(xi + x, eta + y)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"or rotations with a rotation matrix T","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"mapping(xi, eta) = T * SVector(xi, eta).","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"For more curved mesh mappings, please have a look at some elixirs for StructuredMesh. For another curved mesh type, there is a tutorial about Trixi.jl's unstructured mesh type [UnstructuredMesh2D] and its use of the High-Order Hex-Quad Mesh (HOHQMesh) generator, created and developed by David Kopriva.","category":"page"},{"location":"tutorials/structured_mesh_mapping/#Package-versions","page":"15 Structured mesh with curvilinear mapping","title":"Package versions","text":"","category":"section"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"","category":"page"},{"location":"tutorials/structured_mesh_mapping/","page":"15 Structured mesh with curvilinear mapping","title":"15 Structured mesh with curvilinear mapping","text":"This page was generated using Literate.jl.","category":"page"},{"location":"github-git/#GitHub-and-Git","page":"GitHub & Git","title":"GitHub & Git","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This page contains information on how to use GitHub and Git when developing Trixi.jl.","category":"page"},{"location":"github-git/#Development-workflow","page":"GitHub & Git","title":"Development workflow","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"For adding modifications to Trixi.jl, we generally follow these steps:","category":"page"},{"location":"github-git/#Create-an-issue-(optional)","page":"GitHub & Git","title":"Create an issue (optional)","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"In many cases it makes sense to start by creating an issue on GitHub. For example, if the implementation approach for a new feature is not yet clear or if there should be a discussion about the desired outcome, it is good practice to first get a consensus on what is the expected result of this modification. A GitHub issue is the place to lead this discussion, as it preserves it in the project and - together with the actual code changes - allows in the future to revisit the reasons for a particular choice of implementation or feature.","category":"page"},{"location":"github-git/#Create-a-branch-and-*immediately*-create-a-pull-request","page":"GitHub & Git","title":"Create a branch and immediately create a pull request","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"All feature development, bug fixes etc. should be developed in a branch and not directly on main. If you do not have write access to the main repository on GitHub, first create a fork of the Trixi.jl repository and clone the fork to your machine. Then, create a branch locally by executing git checkout -b yourbranch, push it to the repository, and create a pull request (PR).","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you have already cloned Trixi.jl from the main repo to your local machine, you can also work in that clone. You just need to add your fork as additional remote repository and push your new branch there.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"git remote add myfork git@github.com:YOUR_NAME/Trixi.jl.git\n# get latest main from the main repo\ngit checkout main\ngit pull\n# create a new branch for a cool new feature, bug fix, ...\ngit checkout -b YOUR_BRANCH_NAME\n# do some work and push it to your fork\ngit push -u myfork\n# go to https://github.com/trixi-framework/Trixi.jl/pull\n# and create a PR from your new branch","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"info: Why using pull requests?\nImmediately creating a PR for your branch has the benefit that all code discussions can now be held directly next to the corresponding code. Also, the PR allows to easily compare your branch to the upstream branch (usually main) to see what you have changed. Moreover, tests will run automatically.","category":"page"},{"location":"github-git/#Make-changes","page":"GitHub & Git","title":"Make changes","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"With a branch and PR in place, you can now write your code and commit it to your branch. If you request feedback from someone else, make sure to push your branch to the repository such that the others can easily review your changes or dive in and change something themselves.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"warning: Avoid committing unwanted files\nWhen you use git add . or similar catch-all versions, make sure you do not accidentally commit unwanted files (e.g., Trixi.jl output files, images or videos etc.). If it happens anyways, you can undo the last commit (also multiple times) by running git reset HEAD~ (see also Undo last commit). However, this strategy only works if you have not yet pushed your changes. If you did push your changes, please talk to one of the core developers on how to proceed.","category":"page"},{"location":"github-git/#Keep-your-branch-in-sync-with-main","page":"GitHub & Git","title":"Keep your branch in sync with main","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"For larger features with longer-living branches, it may make sense to synchronize your branch with the current main, e.g., if there was a bug fix in main that is relevant for you. In this case, perform the following steps to merge the current main to your branch:","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Commit all your local changes to your branch and push it. This allows you to delete your clone in case you make a mistake and need to abort the merge.\nExecute git fetch to get the latest changes from the repository.\nMake sure you are in the correct branch by checking the output of git status or by running git checkout yourbranch.\nMerge main using git merge main. If there were no conflicts, hooray!, you are done. Otherwise you need to resolve your merge conflicts and commit the changes afterwards. A good guide for resolving merge conflicts can be found here.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"In general, always use git merge and not git rebase to get the latest changes from main. It is less error-prone and does not create problems on branches that are worked on collaboratively.","category":"page"},{"location":"github-git/#Prepare-for-review","page":"GitHub & Git","title":"Prepare for review","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you feel like your branch is ready to be merged to main, prepare it for review. That is, you should","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"merge the current main to your branch\nrun tests if available, but at least ensure that you did not accidentally change the results for one of the existing example elixirs\nproperly comment your code\ndelete old/unused code, especially commented lines (unless they contain helpful code, in which case you should add a comment on why you keep this around)\nremove debug statements\nadd a elixir_xxx.jl that uses your feature (only relevant for new features)\nmake sure your code formatting adheres to the Style guide\ndescribe changes in NEWS.md if appropriate","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"After you are confident that your branch is cleaned up properly, commit all changes and push them to the repository.","category":"page"},{"location":"github-git/#Get-reviewed","page":"GitHub & Git","title":"Get reviewed","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Ask one of the core developers to review your code. Sometimes this will be done directly, either face-to-face or via a video call. Other times a review will be conducted asynchronously, with the reviewer leaving comments and annotations. In some cases it will be necessary to do multiple rounds of reviews, especially if there are larger changes to be added. Just commit and push your changes to your branch, and the corresponding pull request will be updated automatically.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Please note that a review has nothing to do with the lack of experience of the person developing changes: We try to review all code before it gets added to main, even from the most experienced developers. This is good practice and helps to keep the error rate low while ensuring that the code is developed in a consistent fashion. Furthermore, do not take criticism of your code personally - we just try to keep Trixi.jl as accessible and easy to use for everyone.","category":"page"},{"location":"github-git/#Merge-branch","page":"GitHub & Git","title":"Merge branch","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Once your branch is reviewed and declared ready for merging by the reviewer, make sure that all the latest changes have been pushed. Then, one of the developers will merge your PR. If you are one of the developers, you can also go to the pull request page on GitHub and and click on Merge pull request. Voilà, you are done! Your branch will have been merged to main and the source branch will have been deleted in the GitHub repository (if you are not working in your own fork).","category":"page"},{"location":"github-git/#Update-your-working-copy","page":"GitHub & Git","title":"Update your working copy","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Once you have merged your branch by accepting the PR on GitHub, you should clean up your local working copy of the repository by performing the following steps:","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Update your clone by running git fetch.\nCheck out main using git checkout main.\nDelete merged branch locally with git branch -d yourbranch.\nRemove local references to deleted remote branch by executing git remote prune origin.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"You can now proceed with your next changes by starting again at the top.","category":"page"},{"location":"github-git/#Using-Git","page":"GitHub & Git","title":"Using Git","text":"","category":"section"},{"location":"github-git/#Resources-for-learning-Git","page":"GitHub & Git","title":"Resources for learning Git","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Here are a few resources for learning do use Git that at least one of us found helpful in the past (roughly ordered from novice to advanced to expert):","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Git Handbook by GitHub\nLearn Git Branching","category":"page"},{"location":"github-git/#Tips-and-tricks","page":"GitHub & Git","title":"Tips and tricks","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This is an unordered collection of different tips and tricks that can be helpful while working with Git. As usual, your mileage might vary.","category":"page"},{"location":"github-git/#Undo-last-commit","page":"GitHub & Git","title":"Undo last commit","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If you made a mistake in your last commit, e.g., by committing an unwanted file, you can undo the latest commit by running","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"git reset HEAD~","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"This only works if you have not yet pushed your branch to the GitHub repository. In this case, please talk to one of the core developers on how to proceed. Especially when you accidentally committed a large file (image, or video), please let us know as fast as possible, since the effort to fix the repository grows considerably over time.","category":"page"},{"location":"github-git/#Remove-large-file-from-repository","page":"GitHub & Git","title":"Remove large file from repository","text":"","category":"section"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"If a large file was accidentally committed and pushed to the Trixi.jl repository, please talk to one of the core developers as soon as possible so that they can fix it.","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"danger: Large files\nYou should never try to fix this yourself, as it potentially disrupts/destroys the work of others!","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Based on the instructions found here and here, the following steps need to be taken (as documented for GitLab in issue #33):","category":"page"},{"location":"github-git/","page":"GitHub & Git","title":"GitHub & Git","text":"Tell everyone to commit and push their changes to the repository.\nFix the branch in which the file was committed by removing it and committing the removal. This is especially important on main.\nPerform the following steps to clean up the Git repository:\ncd /tmp\n\n# Download bfg-1.13.0.jar from https://rtyley.github.io/bfg-repo-cleaner/\n\n# Get fresh clone of repo (so you can throw it away in case there is a problem)\ngit clone --mirror git@github.com:trixi-framework/Trixi.jl.git\n\n# Clean up repo of all files larger than 10M\njava -jar bfg-1.13.0.jar --strip-blobs-bigger-than 10M Trixi.jl.git\n\n# Enter repo\ncd Trixi.jl.git\n\n# Clean up reflog and force aggressive garbage collection\ngit reflog expire --expire=now --all && git gc --prune=now --aggressive\n\n# Push changes\ngit push\n\n# Delete clone\nrm -rf Trixi.jl.git\nTell everyone to clean up their local working copies by performing the following steps (also do this yourself):\n# Enter repo\ncd Trixi.jl\n\n# Get current changes\ngit fetch\n\n# Check out the fixed branch\ngit checkout branchname\n\n# IMPORTANT: Do a rebase instead of a pull!\ngit rebase\n\n# Clean reflog and force garbage collection\ngit reflog expire --expire=now --all && git gc --prune=now --aggressive\nIMPORTANT: You need to do a git rebase instead of a git pull when updating the fixed branch.","category":"page"},{"location":"testing/#Testing","page":"Testing","title":"Testing","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"During the development of Trixi.jl, we rely on continuous testing to ensure that modifications or new features do not break existing functionality or add other errors. In the main Trixi.jl repository (and the repositories for the visualization tool Trixi2Vtk), this is facilitated by GitHub Actions, which allows to run tests automatically upon certain events. When, how, and what is tested by GitHub Actions is controlled by the workflow file .github/workflows/ci.yml. In Trixi.jl and its related repositories, tests are triggered by","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"each git push to main and\neach git push to any pull request.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"Besides checking functionality, we also analyse the Test coverage to ensure that we do not miss important parts during testing.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Test and coverage requirements\nBefore merging a pull request (PR) to main, we require thatthe code passes all functional tests\ncode coverage does not decrease.","category":"page"},{"location":"testing/#Testing-setup","page":"Testing","title":"Testing setup","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"The entry point for all testing is the file test/runtests.jl, which is run by the automated tests and which can be triggered manually by executing","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"julia> using Pkg; Pkg.test(\"Trixi\")","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"in the REPL. Since there already exist many tests, we have split them up into multiple files in the test directory to allow for faster testing of individual parts of the code. Thus in addition to performing all tests, you can also just include one of the files named test_xxx.jl to run only a specific subset, e.g.,","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"julia> # Run all 2D tests on the P4estMesh\n include(joinpath(\"test\", \"test_p4est_2d.jl\"))\n\njulia> # Run all 1D tests for the Euler equations on the TreeMesh\n include(joinpath(\"test\", \"test_tree_1d_euler.jl\"))","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"For the automated tests with GitHub Actions, we run multiple jobs in parallel to reduce the waiting time until all tests are finished. You can see the different components that are run as jobs by looking at the TRIXI_TEST variable in test/runtests.jl.","category":"page"},{"location":"testing/#Adding-new-tests","page":"Testing","title":"Adding new tests","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"We use Julia's built-in unit testing capabilities to configure tests. In general, newly added code must be covered by at least one test, and all new elixirs added to the examples/ directory must be used at least once during testing. New tests should be added to the corresponding test/test_xxx.jl file, e.g., a test involving the 3D linear advection equation on the TreeMesh would go into test/test_tree_3d_advection.jl. Please study one of the existing tests and stay consistent to the current style when creating new tests.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"Since we want to test as much as possible, we have a lot of tests and frequently create new ones. Naturally, this increases the time to wait for all tests to pass with each novel feature added to Trixi.jl. Therefore, new tests should be as short as reasonably possible, i.e., without being too insensitive to pick up changes or errors in the code.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"When you add new tests, please check whether all CI jobs still take approximately the same time. If the job where you added new tests takes much longer than everything else, please consider moving some tests from one job to another (or report this incident and ask the main developers for help).","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Test duration\nAs a general rule, tests should last no more than 10 seconds when run with a single thread and after compilation (i.e., excluding the first run).","category":"page"},{"location":"testing/#Test-coverage","page":"Testing","title":"Test coverage","text":"","category":"section"},{"location":"testing/","page":"Testing","title":"Testing","text":"In addition to ensuring that the code produces the expected results, the automated tests also record the code coverage. The resulting coverage reports, i.e., which lines of code were executed by at least one test and are thus considered \"covered\" by testing, are automatically uploaded to Coveralls for easy analysis. Typically, you see a number of Coveralls results at the bottom of each pull request: One for each parallel job (see Testing setup), which can usually be ignored since they only cover parts of the code by definition, and a cumulative coverage result named coverage/coveralls. The \"Details\" link takes you to a detailed report on which lines of code are covered by tests, which ones are missed, and especially which new lines the pull requests adds to Trixi.jl's code base that are not yet covered by testing.","category":"page"},{"location":"testing/","page":"Testing","title":"Testing","text":"note: Coverage requirements\nIn general, we require pull requests to not decrease the overall test coverage percentage in main, with a hard lower bound of 97%.","category":"page"},{"location":"authors/#Authors","page":"Authors","title":"Authors","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"Trixi.jl's development is coordinated by a group of principal developers, who are also its main contributors and who can be contacted in case of questions about Trixi.jl. In addition, there are contributors who have provided substantial additions or modifications. Together, these two groups form \"The Trixi.jl Authors\" as mentioned under License.","category":"page"},{"location":"authors/#Principal-Developers","page":"Authors","title":"Principal Developers","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"Michael Schlottke-Lakemper, RWTH Aachen University/High-Performance Computing Center Stuttgart (HLRS), Germany\nGregor Gassner, University of Cologne, Germany\nHendrik Ranocha, Johannes Gutenberg University Mainz, Germany\nAndrew Winters, Linköping University, Sweden\nJesse Chan, Rice University, US","category":"page"},{"location":"authors/#Contributors","page":"Authors","title":"Contributors","text":"","category":"section"},{"location":"authors/","page":"Authors","title":"Authors","text":"The following people contributed major additions or modifications to Trixi.jl and are listed in alphabetical order:","category":"page"},{"location":"authors/","page":"Authors","title":"Authors","text":"Maximilian D. Bertrand\nBenjamin Bolm\nSimon Candelaresi\nJesse Chan\nLars Christmann\nChristof Czernik\nDaniel Doehring\nPatrick Ersing\nErik Faulhaber\nGregor Gassner\nLucas Gemein\nSven Goldberg\nJoshua Lampert\nJulia Odenthal\nSigrun Ortleb\nHendrik Ranocha\nAndrés M. Rueda-Ramírez\nFelipe Santillan\nMichael Schlottke-Lakemper\nToskan Theine\nAndrew Winters","category":"page"},{"location":"license/#License","page":"License","title":"License","text":"","category":"section"},{"location":"license/","page":"License","title":"License","text":"MIT LicenseCopyright (c) 2020-present The Trixi.jl Authors (see Authors)Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the \"Software\"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.","category":"page"},{"location":"restart/#restart","page":"Restart simulation","title":"Restart simulation","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"You can continue running an already finished simulation by first preparing the simulation for the restart and then performing the restart. Here we suppose that in the first run your simulation stops at time 1.0 and then you want it to run further to time 2.0.","category":"page"},{"location":"restart/#restart_preparation","page":"Restart simulation","title":"Prepare the simulation for a restart","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"In you original elixir you need to specify to write out restart files. Those will later be read for the restart of your simulation. This is done almost the same way as writing the snapshots using the SaveSolutionCallback callback. For the restart files it is called SaveRestartCallback:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"save_restart = SaveRestartCallback(interval=100,\n save_final_restart=true)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Make this part of your CallbackSet.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"An example is examples/examples/structured_2d_dgsem/elixir_advection_extended.jl.","category":"page"},{"location":"restart/#restart_perform","page":"Restart simulation","title":"Perform the simulation restart","text":"","category":"section"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Since all of the information about the simulation can be obtained from the last snapshot, the restart can be done with relatively few lines in an extra elixir file. However, some might prefer to keep everything in one elixir and conditionals like if restart with a boolean variable restart that is user defined.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"First we need to define from which file we want to restart, e.g.","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"restart_file = \"restart_000021.h5\"\nrestart_filename = joinpath(\"out\", restart_file)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Then we load the mesh file:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"mesh = load_mesh(restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"This is then needed for the semidiscretization:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"We then define a new time span for the simulation that takes as starting time the one form the snapshot:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"tspan = (load_time(restart_filename), 2.0)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"We now also take the last dt, so that our solver does not need to first find one to fulfill the CFL condition:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"dt = load_dt(restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"The ODE that we will pass to the solver is now:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"ode = semidiscretize(semi, tspan, restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"You should now define a SaveSolutionCallback similar to the original simulation, but with save_initial_solution=false, otherwise our initial snapshot will be overwritten. If you are using one file for the original simulation and the restart you can reuse your SaveSolutionCallback, but need to set","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"save_solution.condition.save_initial_solution = false","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Before we compute the solution using OrdinaryDiffEq.jl we need to set the integrator and its time step number, e.g.:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"integrator = init(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=dt, save_everystep=false, callback=callbacks);\nload_timestep!(integrator, restart_filename)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"Now we can compute the solution:","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"sol = solve!(integrator)","category":"page"},{"location":"restart/","page":"Restart simulation","title":"Restart simulation","text":"An example is in examples/structured_2d_dgsem/elixir_advection_restart.jl.","category":"page"},{"location":"meshes/structured_mesh/#Structured-mesh","page":"Structured mesh","title":"Structured mesh","text":"","category":"section"},{"location":"meshes/structured_mesh/","page":"Structured mesh","title":"Structured mesh","text":"The StructuredMesh is a structured, curvilinear, conforming mesh type available for one-, two-, and three-dimensional simulations.","category":"page"},{"location":"meshes/structured_mesh/","page":"Structured mesh","title":"Structured mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"development/#Development","page":"Development","title":"Development","text":"","category":"section"},{"location":"development/#interactive-use-of-julia","page":"Development","title":"Interactive use of Julia","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"When a Julia program is executed, Julia first loads and parses all code. Then, the just-in-time compiler has to compile all functions at their first use, which incurs an overhead each time a program is run. For proper packages and commands executed in the REPL (= \"return-eval-print loop\", which is what the Julia community calls the interactive command-line prompt that opens when executing julia without any files as arguments), however, the previously compiled functions are cached. Therefore, Trixi.jl should generally always be used interactively from the REPL without closing Julia during development, as it allows much faster turnaround times.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"If you naively run Trixi.jl from the REPL, you will not be able to change your Trixi.jl source files and then run the changed code without restarting the REPL, which destroys any potential benefits from caching. However, restarting Julia can be avoided by using the Revise.jl package, which tracks changed files and re-loads them automatically. Therefore, it is highly recommended to first install Revise with the following command in Julia: To enter the package REPL mode, press ] in the standard Julia REPL mode. Then, execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add Revise","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Now you are able to run Trixi.jl from the REPL, change Trixi.jl code between runs, and enjoy the advantages of the compilation cache! Before you start using Revise regularly, please be aware of some of the Pitfalls when using Revise.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Another recommended package for working from the REPL is OhMyREPL.jl. It can be installed by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add OhMyREPL","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"and adds syntax highlighting, bracket highlighting, and other helpful improvements for using Julia interactively. To automatically use OhMyREPL when starting the REPL, follow the instructions given in the official documentation.","category":"page"},{"location":"development/#Running-Trixi.jl-interactively-in-the-global-environment","page":"Development","title":"Running Trixi.jl interactively in the global environment","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you've installed Trixi.jl and Revise in your default environment, begin by executing:","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"This will start the Julia REPL. Then, run","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Revise; using Trixi","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"You can run a simulation by executing","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> trixi_include(default_example())","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Together, all of these commands can take some time, roughly half a minute on a modern workstation. Most of the time is spent on compilation of Julia code etc. If you execute the last command again in the same REPL, it will finish within a few milliseconds (maybe ~45 on a modern workstation). This demonstrates the second reason for using the REPL: the compilation cache. That is, those parts of the code that do not change between two Trixi.jl runs do not need to be recompiled and thus execute much faster after the first run.","category":"page"},{"location":"development/#Manually-starting-Trixi.jl-in-the-local-environment","page":"Development","title":"Manually starting Trixi.jl in the local environment","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you followed the installation instructions for developers, execute Julia with the project directory set to the run directory of the program/tool you want to use. For example, to run Trixi.jl this way, you need to start the REPL with","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia --project=path/to/Trixi.jl/run","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"and execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Revise; using Trixi","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"to load Revise and Trixi.jl. You can then proceed with the usual commands and run Trixi.jl as in the example above. The --project flag is required such that Julia can properly load Trixi.jl and all dependencies if Trixi.jl is not installed in the global environment. The same procedure also applies should you opt to install the postprocessing tool Trixi2Vtk manually such that you can modify their implementations.","category":"page"},{"location":"development/#Pitfalls-when-using-Revise","page":"Development","title":"Pitfalls when using Revise","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"While Revise is a great help for developing Julia code, there are a few situations to watch out for when using Revise. The following list of potential issues is based on personal experiences of the Trixi.jl developers and probably incomplete. Further information on limitations and possible issues with Revise can be found in the official documentation.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"tip: If in doubt, restart the REPL\nOftentimes, it is possible to recover from issues with Revise by fixing the offending code. Sometimes, however, this is not possible or you might have troubles finding out what exactly caused the problems. Therefore, in these cases, or if in doubt, restart the REPL to get a fresh start.","category":"page"},{"location":"development/#Syntax-errors-are-easy-to-miss","page":"Development","title":"Syntax errors are easy to miss","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Revise does not stop on syntax errors, e.g., when you accidentally write a[i) instead of a[i]. In this case, Revise reports an error but continues to use the old version of your files! This is especially dangerous for syntax errors, as they are detected while Revise reloads changed code, which happens in the beginning of a new execution. Thus, the syntax error message quickly disappears from the terminal once Trixi.jl starts writing output to the screen and you might not even have noticed that an error occurred at all.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Therefore, when you are deep in a coding/debugging session and wonder why your code modifications do not seem to have any effect, scroll up in your terminal to check if you missed earlier syntax errors, or - if in doubt - restart your REPL.","category":"page"},{"location":"development/#Files-are-not-tracked-after-changing-branches","page":"Development","title":"Files are not tracked after changing branches","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Sometimes, Revise stops tracking files when changing the Git branch. That is, modifications to Trixi.jl's source files will not be reloaded by Revise and thus have no effect of a currently running REPL session. This issue is particularly annoying for a developer, since it does not come with any warning! Therefore, it is good practice to always restart the REPL after changing branches.","category":"page"},{"location":"development/#Changes-to-type-definitions-are-not-allowed","page":"Development","title":"Changes to type definitions are not allowed","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Revise cannot handle changes to type definitions, e.g., when modifying the fields in a struct. In this case, Revise reports an error and refuses to run your code unless you undo the modifications. Once you undo the changes, Revise will usually continue to work as expected again. However, if you want to keep your type modifications, you need to restart the REPL.","category":"page"},{"location":"development/#Using-the-Julia-REPL-effectively","page":"Development","title":"Using the Julia REPL effectively","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"The Julia manual is an excellent resource to learn Julia. Here, we list some helpful commands than can increase your productivity in the Julia REPL.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Use the REPL help mode entered by typing ?.\njulia> using Trixi\n\nhelp?> trixi_include\nsearch: trixi_include\n\n trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)\n\n include the file elixir and evaluate its content in the global scope of module mod. You can override specific\n assignments in elixir by supplying keyword arguments. It's basic purpose is to make it easier to modify some\n parameters while running Trixi.jl from the REPL. Additionally, this is used in tests to reduce the computational\n burden for CI while still providing examples with sensible default values for users.\n\n Examples\n ≡≡≡≡≡≡≡≡≡≡\n\n julia> trixi_include(@__MODULE__, default_example(), tspan=(0.0, 0.1))\n [...]\n\n julia> sol.t[end]\n 0.1\nYou can copy and paste REPL history including julia> prompts into the REPL.\nUse tab completion in the REPL, both for names of functions/types/variables and for function arguments.\njulia> flux_ranocha( # and TAB\nflux_ranocha(u_ll, u_rr, orientation::Integer, equations::CompressibleEulerEquations1D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_1d.jl:390\nflux_ranocha(u_ll, u_rr, orientation::Integer, equations::CompressibleEulerEquations2D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_2d.jl:839\n[...]\nUse methodswith to discover methods associated to a given type etc.\njulia> methodswith(CompressibleEulerEquations2D)\n[1] initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations2D) in Trixi at ~/.julia/dev/Trixi/src/equations/compressible_euler_2d.jl:51\n[...]\nUse @which (or @edit) for method calls.\njulia> @which trixi_include(default_example())\ntrixi_include(elixir::AbstractString; kwargs...) in Trixi at ~/.julia/dev/Trixi/src/auxiliary/special_elixirs.jl:36\nUse apropos to search through the documentation and docstrings.\njulia> apropos(\"MHD\")\nTrixi.IdealGlmMhdEquations3D\nTrixi.IdealGlmMhdMulticomponentEquations2D\nTrixi.calc_fast_wavespeed_roe\nTrixi.IdealGlmMhdEquations1D\nTrixi.initial_condition_constant\nTrixi.flux_nonconservative_powell\nTrixi.GlmSpeedCallback\nTrixi.flux_derigs_etal\nTrixi.flux_hindenlang_gassner\nTrixi.initial_condition_convergence_test\nTrixi.min_max_speed_naive\nTrixi.IdealGlmMhdEquations2D\nTrixi.IdealGlmMhdMulticomponentEquations1D\n[...]","category":"page"},{"location":"development/#Text-editors","page":"Development","title":"Text editors","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"When writing code, the choice of text editor can have a significant impact on productivity and developer satisfaction. While using the default text editor of the operating system has its own benefits (specifically the lack of an explicit installation procure), usually it makes sense to switch to a more programming-friendly tool. In the following, a few of the many options are listed and discussed:","category":"page"},{"location":"development/#VS-Code","page":"Development","title":"VS Code","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Visual Studio Code is a modern open source editor with good support for Julia. While Juno had some better support in the past, the developers of Juno and the Julia VS Code plugin are joining forces and concentrating on VS Code since support of Atom has been suspended. Basically, all comments on Juno below also apply to VS Code.","category":"page"},{"location":"development/#Juno","page":"Development","title":"Juno","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"If you are new to programming or do not have a preference for a text editor yet, Juno is a good choice for developing Julia code. It is based on Atom, a sophisticated and widely used editor for software developers, and is enhanced with several Julia-specific features. Furthermore and especially helpful for novice programmers, it has a MATLAB-like appearance with easy and interactive access to the current variables, the help system, and a debugger.","category":"page"},{"location":"development/#Vim-or-Emacs","page":"Development","title":"Vim or Emacs","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Vim and Emacs are both very popular editors that work great with Julia. One of their advantages is that they are text editors without a GUI and as such are available for almost any operating system. They also are preinstalled on virtually all Unix-like systems. However, Vim and Emacs come with their own, steep learning curve if they have never been used before. Therefore, if in doubt, it is probably easier to get started with a classic GUI-based text editor (like Juno). If you decide to use Vim or Emacs, make sure that you install the corresponding Vim plugin julia-vim or Emacs major mode julia-emacs.","category":"page"},{"location":"development/#Debugging","page":"Development","title":"Debugging","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Julia offers several options for debugging. A classical debugger is available with the Debugger.jl package or in the Julia extension for VS Code. However, it can be quite slow and, at the time of writing (January 2023), currently does not work properly with Trixi.jl. The Infiltrator.jl package on the other hand does not offer all features of a full debugger, but is a fast and simple tool that allows users to set breakpoints to open a local REPL session and access the call stack and variables.","category":"page"},{"location":"development/#Infiltrator","page":"Development","title":"Infiltrator","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"The Infiltrator package provides fast, interactive breakpoints using the @infiltrate command, which drops the user into a local REPL session. From there, it is possible to access local variables, see the call stack, and execute statements.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"The package can be installed in the Julia REPL by executing","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> add Infiltrator","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"To load the package in the Julia REPL execute","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia> using Infiltrator","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Breakpoints can be set by adding a line with the @infiltrate macro at the respective position in the code. Use Revise if you want to set and delete breakpoints in your package without having to restart Julia.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"note: Use `@autoinfiltrate` when debugging Trixi.jl\nWhen running Julia inside a package environment, e.g., inside the source code of Trixi.jl itself, the @infiltrate macro only works if Infiltrator has been added to the package dependencies. To avoid this, you can use the (non-exported) @autoinfiltrate macro in Trixi.jl, which only requires Infiltrator.jl to be available in the current environment stack and will auto-load it for you.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Triggering the breakpoint starts a REPL session where it is possible to interact with the current local scope. Possible commands are:","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"@locals: Print the local variables.\n@exfiltrate: Save the local variables to a global storage, which can be accessed with the safehouse variable outside the Infiltrator session.\n@trace: Print the current stack trace.\nExecute other arbitrary statements\n?: Print a help list with all options","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"To finish a debugging session, either use @continue to continue and eventually stop at the next breakpoint or @exit to skip further breakpoints. After the code has finished, local variables saved with @exfiltrate can be accessed in the REPL using the safehouse variable.","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"Limitations of using Infiltrator.jl are that local variables cannot be changed, and that it is not possible to step into further calls or access other function scopes.","category":"page"},{"location":"development/#Releasing-a-new-version-of-Trixi.jl,-Trixi2Vtk","page":"Development","title":"Releasing a new version of Trixi.jl, Trixi2Vtk","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Check whether everything is okay, tests pass etc.\nSet the new version number in Project.toml according to the Julian version of semver. Commit and push.\nComment @JuliaRegistrator register on the commit setting the version number.\nJuliaRegistrator will create a PR with the new version in the General registry. Wait for it to be merged.\nIncrement the version number in Project.toml again with suffix -pre. For example, if you have released version v0.2.0, use v0.2.1-pre as new version number.\nWhen a new version of Trixi.jl was released, check whether the [compat] entries in test/Project.toml in Trixi2Vtk should be updated. When a new version of Trixi2Vtk was released, check whether the [compat] entries in docs/Project.toml in Trixi.jl should be updated.\nThese entries will also be checked regularly by CompatHelper (once a day). Hence, if everything was released correctly, you should only need to do these checks manually if new minor versions with changes in the docs of Trixi2Vtk were released but no new version of Trixi.jl was released afterwards.","category":"page"},{"location":"development/#Preview-of-the-documentation","page":"Development","title":"Preview of the documentation","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"You can build the documentation of Trixi.jl locally by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"julia --project=docs -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd())); Pkg.instantiate()'\njulia --project=docs --color=yes docs/make.jl","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"from the Trixi.jl main directory. Then, you can look at the html files generated in docs/build. For PRs triggered from branches inside the Trixi.jl main repository previews of the new documentation are generated at https://trixi-framework.github.io/Trixi.jl/previews/PRXXX, where XXX is the number of the PR. This does not work for PRs from forks for security reasons (since anyone could otherwise push arbitrary stuff to the Trixi.jl website, including malicious code).","category":"page"},{"location":"development/#trixi2vtk-dev","page":"Development","title":"Developing Trixi2Vtk","text":"","category":"section"},{"location":"development/","page":"Development","title":"Development","text":"Trixi2Vtk has Trixi.jl as dependency and uses Trixi.jl's implementation to, e.g., load mesh files. When developing Trixi2Vtk, one may want to change functions in Trixi.jl to allow them to be reused in Trixi2Vtk. To use a locally modified Trixi.jl clone instead of a Trixi.jl release, one can tell Pkg to use the local source code of Trixi.jl instead of a registered version by running","category":"page"},{"location":"development/","page":"Development","title":"Development","text":"(@v1.9) pkg> develop path/to/Trixi.jl","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"EditURL = \"../../../literate/src/files/first_steps/create_first_setup.jl\"","category":"page"},{"location":"tutorials/first_steps/create_first_setup/#create_first_setup","page":"1.2 Create first setup","title":"1.2: First steps in Trixi.jl: Create first setup","text":"","category":"section"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"In this part of the introductory guide, we will create a first Trixi.jl setup as an extension of elixir_advection_basic.jl. Since Trixi.jl has a common basic structure for the setups, you can create your own by extending and modifying the following example.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Let's consider the linear advection equation for a state u = u(x y t) on the two-dimensional spatial domain -1 1 times -1 1 with a source term","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"fracpartialpartial tu + fracpartialpartial x (02 u) - fracpartialpartial y (07 u) = - 2 e^-t\nsinbigl(2 pi (x - t) bigr) sinbigl(2 pi (y - t) bigr)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"with the initial condition","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"u(x y 0) = sinbigl(pi x bigr) sinbigl(pi y bigr)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"and periodic boundary conditions.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The first step is to create and open a file with the .jl extension. You can do this with your favorite text editor (if you do not have one, we recommend VS Code). In this file you will create your setup.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To be able to use functionalities of Trixi.jl, you always need to load Trixi.jl itself and the OrdinaryDiffEq.jl package.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"using Trixi\nusing OrdinaryDiffEq","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The next thing to do is to choose an equation that is suitable for your problem. To see all the currently implemented equations, take a look at src/equations. If you are interested in adding a new physics model that has not yet been implemented in Trixi.jl, take a look at the tutorials Adding a new scalar conservation law or Adding a non-conservative equation.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The linear scalar advection equation in two spatial dimensions","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"fracpartialpartial tu + fracpartialpartial x (a_1 u) + fracpartialpartial y (a_2 u) = 0","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"is already implemented in Trixi.jl as LinearScalarAdvectionEquation2D, for which we need to define a two-dimensional parameter advection_velocity describing the parameters a_1 and a_2. Appropriate for our problem is (0.2, -0.7).","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"advection_velocity = (0.2, -0.7)\nequations = LinearScalarAdvectionEquation2D(advection_velocity)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To solve our problem numerically using Trixi.jl, we have to discretize the spatial domain, for which we set up a mesh. One of the most used meshes in Trixi.jl is the TreeMesh. The spatial domain used is -1 1 times -1 1. We set an initial number of elements in the mesh using initial_refinement_level, which describes the initial number of hierarchical refinements. In this simple case, the total number of elements is 2^initial_refinement_level throughout the simulation. The variable n_cells_max is used to limit the number of elements in the mesh, which cannot be exceeded when using adaptive mesh refinement.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"All minimum and all maximum coordinates must be combined into Tuples.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"coordinates_min = (-1.0, -1.0)\ncoordinates_max = ( 1.0, 1.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level = 4,\n n_cells_max = 30_000)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To approximate the solution of the defined model, we create a DGSEM solver. The solution in each of the recently defined mesh elements will be approximated by a polynomial of degree polydeg. For more information about discontinuous Galerkin methods, check out the Introduction to DG methods tutorial.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"solver = DGSEM(polydeg=3)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now we need to define an initial condition for our problem. All the already implemented initial conditions for LinearScalarAdvectionEquation2D can be found in src/equations/linear_scalar_advection_2d.jl. If you want to use, for example, a Gaussian pulse, it can be used as follows:","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"initial_conditions = initial_condition_gauss","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"But to show you how an arbitrary initial condition can be implemented in a way suitable for Trixi.jl, we define our own initial conditions.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"u(x y 0) = sinbigl(pi x bigr) sinbigl(pi y bigr)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The initial conditions function must take spatial coordinates, time and equation as arguments and returns an initial condition as a statically sized vector SVector. Following the same structure, you can define your own initial conditions. The time variable t can be unused in the initial condition, but might also be used to describe an analytical solution if known. If you use the initial condition as analytical solution, you can analyze your numerical solution by computing the error, see also the section about analyzing the solution.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"function initial_condition_sinpi(x, t, equations::LinearScalarAdvectionEquation2D)\n scalar = sinpi(x[1]) * sinpi(x[2])\n return SVector(scalar)\nend\ninitial_condition = initial_condition_sinpi","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The next step is to define a function of the source term corresponding to our problem.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"f(u x y t) = - 2 e^-t sinbigl(2 pi (x - t) bigr) sinbigl(2 pi (y - t) bigr)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"This function must take the state variable, the spatial coordinates, the time and the equation itself as arguments and returns the source term as a static vector SVector.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"function source_term_exp_sinpi(u, x, t, equations::LinearScalarAdvectionEquation2D)\n scalar = - 2 * exp(-t) * sinpi(2*(x[1] - t)) * sinpi(2*(x[2] - t))\n return SVector(scalar)\nend","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now we collect all the information that is necessary to define a spatial discretization, which leaves us with an ODE problem in time with a span from 0.0 to 1.0. This approach is commonly referred to as the method of lines.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;\n source_terms = source_term_exp_sinpi)\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"At this point, our problem is defined. We will use the solve function defined in OrdinaryDiffEq.jl to get the solution. OrdinaryDiffEq.jl gives us the ability to customize the solver using callbacks without actually modifying it. Trixi.jl already has some implemented Callbacks. The most widely used callbacks in Trixi.jl are step control callbacks that are activated at the end of each time step to perform some actions, e.g. to print statistics. We will show you how to use some of the common callbacks.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To print a summary of the simulation setup at the beginning and to reset timers we use the SummaryCallback. When the returned callback is executed directly, the current timer values are shown.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"summary_callback = SummaryCallback()","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"We also want to analyze the current state of the solution in regular intervals. The AnalysisCallback outputs some useful statistical information during the solving process every interval time steps.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"analysis_callback = AnalysisCallback(semi, interval = 5)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"It is also possible to control the time step size using the StepsizeCallback if the time integration method isn't adaptive itself. To get more details, look at CFL based step size control.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"stepsize_callback = StepsizeCallback(cfl = 1.6)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To save the current solution in regular intervals we use the SaveSolutionCallback. We would like to save the initial and final solutions as well. The data will be saved as HDF5 files located in the out folder. Afterwards it is possible to visualize a solution from saved files using Trixi2Vtk.jl and ParaView, which is described below in the section Visualize the solution.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"save_solution = SaveSolutionCallback(interval = 5,\n save_initial_solution = true,\n save_final_solution = true)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Alternatively, we have the option to print solution files at fixed time intervals.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"save_solution = SaveSolutionCallback(dt = 0.1,\n save_initial_solution = true,\n save_final_solution = true)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Another useful callback is the SaveRestartCallback. It saves information for restarting in regular intervals. We are interested in saving a restart file for the final solution as well. To perform a restart, you need to configure the restart setup in a special way, which is described in the section Restart simulation.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"save_restart = SaveRestartCallback(interval = 100, save_final_restart = true)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Create a CallbackSet to collect all callbacks so that they can be passed to the solve function.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"callbacks = CallbackSet(summary_callback, analysis_callback, stepsize_callback, save_solution,\n save_restart)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The last step is to choose the time integration method. OrdinaryDiffEq.jl defines a wide range of ODE solvers, e.g. CarpenterKennedy2N54(williamson_condition = false). We will pass the ODE problem, the ODE solver and the callbacks to the solve function. Also, to use StepsizeCallback, we must explicitly specify the initial trial time step dt, the selected value is not important, because it will be overwritten by the StepsizeCallback. And there is no need to save every step of the solution, we are only interested in the final result.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition = false), dt = 1.0,\n save_everystep = false, callback = callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Finally, we print the timer summary.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"summary_callback()","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now you can plot the solution as shown below, analyze it and improve the stability, accuracy or efficiency of your setup.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/#Visualize-the-solution","page":"1.2 Create first setup","title":"Visualize the solution","text":"","category":"section"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"In the previous part of the tutorial, we calculated the final solution of the given problem, now we want to visualize it. A more detailed explanation of visualization methods can be found in the section Visualization.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/#Using-Plots.jl","page":"1.2 Create first setup","title":"Using Plots.jl","text":"","category":"section"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"The first option is to use the Plots.jl package directly after calculations, when the solution is saved in the sol variable. We load the package and use the plot function.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"To show the mesh on the plot, we need to extract the visualization data from the solution as a PlotData2D object. Mesh extraction is possible using the getmesh function. Plots.jl has the plot! function that allows you to modify an already built graph.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"pd = PlotData2D(sol)\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/first_steps/create_first_setup/#Using-Trixi2Vtk.jl","page":"1.2 Create first setup","title":"Using Trixi2Vtk.jl","text":"","category":"section"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Another way to visualize a solution is to extract it from a saved HDF5 file. After we used the solve function with SaveSolutionCallback there is a file with the final solution. It is located in the out folder and is named as follows: solution_index.h5. The index is the final time step of the solution that is padded to 6 digits with zeros from the beginning. With Trixi2Vtk you can convert the HDF5 output file generated by Trixi.jl into a VTK file. This can be used in visualization tools such as ParaView or VisIt to plot the solution. The important thing is that currently Trixi2Vtk.jl supports conversion only for solutions in 2D and 3D spatial domains.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"If you haven't added Trixi2Vtk.jl to your project yet, you can add it as follows.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"import Pkg\nPkg.add([\"Trixi2Vtk\"])","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now we load the Trixi2Vtk.jl package and convert the file out/solution_000018.h5 with the final solution using the trixi2vtk function saving the resulting file in the out folder.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"using Trixi2Vtk\ntrixi2vtk(joinpath(\"out\", \"solution_000018.h5\"), output_directory=\"out\")","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now two files solution_000018.vtu and solution_000018_celldata.vtu have been generated in the out folder. The first one contains all the information for visualizing the solution, the second one contains all the cell-based or discretization-based information.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now let's visualize the solution from the generated files in ParaView. Follow this short instruction to get the visualization.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Download, install and open ParaView.\nPress Ctrl+O and select the generated files solution_000018.vtu and solution_000018_celldata.vtu from the out folder.\nIn the upper-left corner in the Pipeline Browser window, left-click on the eye-icon near solution_000018.vtu.\nIn the lower-left corner in the Properties window, change the Coloring from Solid Color to scalar. This already generates the visualization of the final solution.\nNow let's add the mesh to the visualization. In the upper-left corner in the Pipeline Browser window, left-click on the eye-icon near solution_000018_celldata.vtu.\nIn the lower-left corner in the Properties window, change the Representation from Surface to Wireframe. Then a white grid should appear on the visualization.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Now, if you followed the instructions exactly, you should get a similar image as shown in the section Using Plots.jl:","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"(Image: paraview_trixi2vtk_example)","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"After completing this tutorial you are able to set up your own simulations with Trixi.jl. If you have an interest in contributing to Trixi.jl as a developer, refer to the third part of the introduction titled Changing Trixi.jl itself.","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"Sys.rm(\"out\"; recursive=true, force=true) #hide","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"","category":"page"},{"location":"tutorials/first_steps/create_first_setup/","page":"1.2 Create first setup","title":"1.2 Create first setup","text":"This page was generated using Literate.jl.","category":"page"},{"location":"visualization/#visualization","page":"Visualization","title":"Visualization","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"There are two possible approaches to visualize results from Trixi.jl: either directly from the REPL using Plots.jl or with ParaView/VisIt by postprocessing Trixi.jl's output files with Trixi2Vtk.","category":"page"},{"location":"visualization/#Plots.jl","page":"Visualization","title":"Plots.jl [experimental]","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By far the easiest and most convenient plotting approach is to use the powerful Plots.jl package to directly visualize Trixi.jl's results from the REPL. In the following, you will find more information on a number of topics for how to use Plots.jl together with Trixi.jl:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Getting started\nCustomizing plot results via a PlotData2D object\nPlotting a 3D solution as a 2D plot\nCreating a 1D plot\nPlotting a 2D or 3D solutions as a 1D plot\nVisualizing results during a simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nPlotting via Plots.jl is still considered an experimental feature and might change in any future releases.","category":"page"},{"location":"visualization/#getting-started-plots-jl","page":"Visualization","title":"Getting started","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"After running a simulation with Trixi.jl in the REPL, load the Plots package with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Plots","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize the solution, execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Here we assume that sol holds the return value of the solve(...) method (with type SciMLBase.ODESolution), which is the default variable name when you use one of the example elixirs. This will generate a grid layout with one subplot for each solution variable, convenient for getting an overview of the current solution:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You can save the resulting file as a PNG image file by calling savefig(...) with an output file name that ends in .png, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> savefig(\"solution-overview.png\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In Trixi.jl, two plot types are available: 2D heatmap plots and 1D line plots. If you use plot(sol), Trixi.jl will automatically choose the plot type that fits the dimensions of the sol input: 2D/3D data will be visualized as a heatmap, 1D data as a line plot. For more fine-grained control over what to plot, you can create such an object yourself, which can either be a PlotData2D or a PlotData1D object. For further details on both of these see below:","category":"page"},{"location":"visualization/#Customizing-plot-results-via-a-PlotData2D-object","page":"Visualization","title":"Customizing plot results via a PlotData2D object","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For more fine-grained control over what to plot, first create a PlotData2D object by executing","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol)\njulia> pd = PlotData2D(u, semi)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"where u is an array containing the solution and semi is the semidiscretization. For example, if PlotData2D(sol.u[2], semi) is specified, this will create a PlotData2D instance from the 2nd saved time-step. If PlotData2D(sol(0.5), semi) is specified, it will construct a PlotData2D instance using OrdinaryDiffEq.jl's interpolation to evaluate the solution at time t=0.5.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This takes the results generated by Trixi.jl and stores them in a data format that can be understood by the Plots package, and pd holds all data relevant for plotting sol. You can pass variable names as strings to pd using a dictionary-like syntax, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd[\"rho\"])","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will create a single 2D heatmap plot of the variable rho:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The default plot type and style can be overridden by passing any additional arguments that are understood by the Plots package. For example, to change the color scheme and add names to the axes, modify the previous command to","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd[\"rho\"], seriescolor = :heat, xguide=\"x\", yguide=\"y\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to yield","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-modified)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For more details on the various format options for plot, please consult the Plots documentation.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In addition, you can plot the mesh lines on top of the solution variables by calling the getmesh(...) function on the PlotData2D object","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot!(getmesh(pd)) # here we use `plot!` with an `!` to add to the previous plot","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"which modifies the previous plot to","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-modified-mesh)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default, PlotData2D will convert the conserved variables to primitive variables, but this can be changed by passing an appropriate conversion function in the solution_variables keyword argument, similar to the behavior of the SaveSolutionCallback:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol; solution_variables=cons2cons) # Plot conservative variables","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"There are several other keyword arguments that influence how the solution data is processed for visualization with the Plots package. A detailed explanation can be found in the docstring of the PlotData2D constructor.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Another way to change the appearance of a plot is to convert the solution to a uniformly refined mesh before plotting. This can be helpful, e.g., when trying different settings for a simulation with adaptive mesh refinement, where one would like to ignore the mesh changes when comparing solutions. This is achieved with adapt_to_mesh_level, which uses the mesh adaptation routines to adapt the solution to a uniform grid. For example, the AMR solution from above could be preprocessed with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(adapt_to_mesh_level(sol, 4)...)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When plotted together with the mesh, this will yield the following visualization:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-rho-uniform-mesh)","category":"page"},{"location":"visualization/#Plotting-a-user-defined-scalar-field","page":"Visualization","title":"Plotting a user-defined scalar field","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To plot a scalar quantity, one can call plot(ScalarPlotData2D(u, semi)), where u is an array of nodal values of the scalar field to plot. The layout of u should match the layout of the x and y nodal coordinates of the respective solver. For example, after running trixi_include(joinpath(\"examples\", \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\")), the following can be used to plot the function f(x, y) = x * y:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"x = view(semi.cache.elements.node_coordinates, 1, :, :, :)\ny = view(semi.cache.elements.node_coordinates, 2, :, :, :)\nplot(ScalarPlotData2D(x .* y, semi))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This produces the following plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: scalar-plotting-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This routine can be used to visualize scalar quantities which depend on the solution, such as the norm of a velocity vector or two-dimensional vorticity. For example, we can visualize vorticity for a compressible version of the Brown-Minion vortex problem:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi, Plots\n\njulia> redirect_stdout(devnull) do\n # runs the elixir without any output from callbacks etc.\n trixi_include(@__MODULE__,\n joinpath(examples_dir(), \"dgmulti_2d\", \"elixir_euler_brown_minion_vortex.jl\"))\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n\njulia> function compute_vorticity(velocity, mesh, equations::CompressibleEulerEquations2D,\n dg::DGMulti, cache)\n rd = dg.basis\n md = mesh.md\n @unpack Dr, Ds = rd\n @unpack rxJ, sxJ, ryJ, syJ, J = md\n v1, v2 = velocity\n dv1dy = ryJ .* (Dr * v1) + syJ .* (Ds * v1)\n dv2dx = rxJ .* (Dr * v2) + sxJ .* (Ds * v2)\n return dv2dx - dv1dy\n end;\n\njulia> compute_vorticity(velocity, semi) =\n compute_vorticity(velocity, Trixi.mesh_equations_solver_cache(semi)...);\n\njulia> function get_velocity(sol)\n rho, rhou, rhov, E = StructArrays.components(sol.u[end])\n v1 = rhou ./ rho\n v2 = rhov ./ rho\n return v1, v2\n end;\n\njulia> vorticity = compute_vorticity(get_velocity(sol), semi);\n\njulia> plot(ScalarPlotData2D(vorticity, semi;\n variable_name = \"Vorticity at t = $(sol.prob.tspan[end])\"))\nPlot{Plots.GRBackend() n=1}","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This produces the following plot of vorticity.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: vorticity-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Since the mesh is fairly coarse, we observe numerical artifacts due to the low resolution. These errors vanish under mesh refinement; for example, doubling the mesh resolution by running","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(examples_dir(), \"dgmulti_2d\", \"elixir_euler_BM_vortex.jl\"), cells_per_dimension = 32)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"yields the following plot of vorticity:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: vorticity-example-refined)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nWhen visualizing a scalar field, the plotted solution is reinterpolated using a high order polynomial approximation. Thus, small discrepancies may be observed when the underlying data is highly non-smooth or under-resolved.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"ScalarPlotData2D objects can also be used with Makie through iplot. For example, the following code plots two surfaces:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi, CairoMakie\n\njulia> redirect_stdout(devnull) do\n # runs the elixir without any output from callbacks etc.\n trixi_include(@__MODULE__,\n joinpath(examples_dir(), \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\"))\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n\njulia> x = view(semi.cache.elements.node_coordinates, 1, :, :, :); # extracts the node x coordinates\n\njulia> y = view(semi.cache.elements.node_coordinates, 2, :, :, :); # extracts the node y coordinates\n\njulia> fig_ax_plt = iplot(ScalarPlotData2D((@. 1 - .25*(x^2 + y^2)), semi), plot_mesh=true, colormap=:viridis);\n\njulia> fig_ax_plt2 = iplot!(fig_ax_plt, ScalarPlotData2D((@. .125*(x+y)), semi), plot_mesh=true, colormap=:blues)\nFigureAxisPlot()","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This creates the following plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: ScalarPlotData2D_example)","category":"page"},{"location":"visualization/#Plotting-a-3D-solution-as-a-2D-plot","page":"Visualization","title":"Plotting a 3D solution as a 2D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"It is possible to plot 2D slices from 3D simulation data using the TreeMesh with the same commands as above:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol) # `sol` is from a 3D simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default, plotting sol or creating a PlotData2D object from a 3D simulation will create a 2D slice of the solution in the xy-plane. You can customize this behavior by explicitly creating a PlotData2D object and passing appropriate keyword arguments:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"slice specifies the plane which is being sliced and can be :xy, :xz, or :yz (default: :xy)\npoint specifies a three-dimensional point. The sliced plane is then created such that it lies on the point (default: (0.0, 0.0, 0.0)).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"All other attributes for PlotData2D objects apply here as well.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"For example, to plot the velocity field orthogonal to the yz-plane at different x-axis locations, you can execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(examples_dir(), \"tree_3d_dgsem\", \"elixir_euler_taylor_green_vortex.jl\"), tspan=(0.0, 1.0))\n[...]\n\njulia> plots = []\nAny[]\n\njulia> for x in range(0, stop=pi/2, length=6)\n pd = PlotData2D(sol, slice=:yz, point=(x, 0.0, 0.0))\n push!(plots, plot(pd[\"v1\"], clims=(-1,1), title=\"x = \"*string(round(x, digits=2))))\n end\n\njulia> plot(plots..., layout=(2, 3), size=(750,350))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"which results in a 2x3 grid of slices of the yz-plane:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: plot-v1-0.0-to-0.5pi)","category":"page"},{"location":"visualization/#Creating-a-1D-plot","page":"Visualization","title":"Creating a 1D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When plotting a 1D solution with","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol) # `sol` is from a 1D simulation","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi.jl automatically creates a PlotData1D object and visualizes it as a line plot: (Image: 1d-plot)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To customize your 1D plot, you can create a PlotData1D object manually as follows:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol)\njulia> pd = PlotData1D(u, semi)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The behavior is analogous to the PlotData2D behavior.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In a very similar fashion to PlotData2D, you can customize your plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"plot(pd) creates the same plot as in plot(sol).\nplot(pd[\"rho\", \"p\"]) only plots specific variables. In this case rho and p.\nplot!(getmesh(pd)) adds mesh lines after creating a plot.\nAny attributes from Plots can be used, e.g., plot(pd, yguide=:temperature).\npd = PlotData1D(adapt_to_mesh_level(sol, 4)...) adapts the mesh before plotting (in this example to a mesh with refinement level 4).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You can also customize the PlotData1D object itself by passing attributes to the PlotData1D constructor:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"solution_variables specifies the variables to be plotted.\nnvisnodes sets the amount of nodes per element which the solution then is interpolated on.","category":"page"},{"location":"visualization/#Plotting-a-2D-or-3D-solutions-as-a-1D-plot","page":"Visualization","title":"Plotting a 2D or 3D solutions as a 1D plot","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"It is possible to extract a straight, axis-parallel line from a 2D or 3D solution and visualize it as a 1D plot. This is done by creating a PlotData1D object with a 2D/3D solution sol as input:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot is then created with:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"By default the x-axis is extracted, which can be changed with following attributes:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"slice specifies the axis which is being extracted and can be :x, :y or :z (:z is only for 3D input and default is :x)\npoint specifies a two or three dimensional point. The sliced axis is then created in such a way, that it lies on the point. (default: (0.0, 0.0) or (0.0, 0.0, 0.0))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"All other attributes for PlotData1D objects apply here as well.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In the following, is an example for a 2D simulation of the linear scalar advection equation. First, we have the regular 2D heatmap plot:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 2d-plot-for-slice)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"From this, we can extract a line plot parallel to the y-axis going through the point (1.0, 0.0) with the following commands:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData1D(sol, slice=:y, point=(1.0, 0.0))\njulia> plot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 1d-plot-for-slice)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This convenient method of slicing is limited to axis-parallel slices, but for 2D/3D solutions it is also possible to create a plot along any curve you want. To do so, you first need to create a list of 2D/3D points that define your curve. Then you can create a PlotData1D with the keyword argument curve set to your list.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Let's give an example of this with the basic advection equation from above by creating a plot along the circle marked in green:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: 2d-plot-along-circle)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"We can write a function like this, that outputs a list of points on a circle:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"function circle(radius, center, n_points)\n coordinates = zeros(2, n_points)\n for i in 1:n_points\n coordinates[:,i] = radius*[cospi(2*i/n_points), sinpi(2*i/n_points)] .+ center\n end\n return coordinates\nend","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Then create and plot a PlotData1D object along a circle with radius one, center at (1,1), and 100 points:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"pd = PlotData1D(sol, curve=circle(1.0, (1.0, 1.0), 100))\nplot(pd)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This gives you the following plot: (Image: 1d-plot-along-circle)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Creating a plot like this has its downsides. For one, it is unclear what to put on the abscissa of the plot. By default, the arc length of the given curve is used. Also, with this way of plotting you lose the ability to use a mesh plot from getmesh.","category":"page"},{"location":"visualization/#Visualizing-results-during-a-simulation","page":"Visualization","title":"Visualizing results during a simulation","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize solutions while a simulation is still running (also known as in-situ visualization), you can use the VisualizationCallback. It is created as a regular callback and accepts upon creation a number of keyword arguments that allow, e.g., to control the visualization interval, to specify the variables to plot, or to customize the plotting style.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"During the simulation, the visualization callback creates and displays visualizations of the current solution in regular intervals. This can be useful to, e.g., monitor the validity of a long-running simulation or for illustrative purposes. An example for how to create a VisualizationCallback can be found in examples/tree_2d_dgsem/elixir_advection_amr_visualization.jl:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"[...]\n\n# Enable in-situ visualization with a new plot generated every 20 time steps\n# and additional plotting options passed as keyword arguments\nvisualization = VisualizationCallback(interval=20; clims=(0,1))\n\n[...]","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The resulting output of the referenced elixir can be seen in the embedded video below:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":" \n
","category":"page"},{"location":"visualization/#Trixi2Vtk","page":"Visualization","title":"Trixi2Vtk","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi2Vtk converts Trixi.jl's .h5 output files to VTK files, which can be read by ParaView, VisIt, and other visualization tools. It automatically interpolates solution data from the original quadrature node locations to equidistant visualization nodes at a higher resolution, to make up for the loss of accuracy from going from a high-order polynomial representation to a piecewise constant representation in VTK.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In the Julia REPL, first load the package Trixi2Vtk","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using Trixi2Vtk","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To process an HDF5 file generated by Trixi.jl, execute","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(joinpath(\"out\", \"solution_000000.h5\"), output_directory=\"out\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will create two unstructured VTK files in the out subdirectory that can be opened with ParaView or VisIt: solution_000000.vtu contains the discontinuous Galerkin solution data while solution_000000_celldata.vtu holds any cell-based values such as the current AMR indicator or the cell refinement level.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: \"solution_000000_scalar_mesh\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This allows you to generate VTK files for solution, restart and mesh files. By default, Trixi2Vtk generates .vtu (unstructured VTK) files for both cell/element data (e.g., cell ids, element ids) and node data (e.g., solution variables). This format visualizes each cell with the same number of nodes, independent of its size. Alternatively, you can provide format=:vti as a keyword argument to trixi2vtk, which causes Trixi2Vtk to generate .vti (image data VTK) files for the solution files, while still using .vtu files for cell-/element-based data. In .vti files, a uniform resolution is used throughout the entire domain, resulting in different number of visualization nodes for each element. This can be advantageous to create publication-quality images, but increases the file size.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"If you want to convert multiple solution/restart files at once, you can just supply multiple input files as the positional arguments to trixi2vtk, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/solution_000000.h5\", \"out/solution_000040.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"You may also use file globbing to select a range of files based on filename patterns, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/solution_*.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to convert all solution files in the out/ directory or","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi2vtk(\"out/restart_00[0-9]000.h5\")","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"to convert every one-thousandth restart file (out/restart_000000.h5, out/restart_001000.h5 etc.).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"When multiple solution/restart files are provided, Trixi2Vtk will also generate a .pvd file, which allows ParaView to read all .vtu/.vti files at once and which uses the time attribute in solution/restart files to inform ParaView about the solution time. A comprehensive list of all possible arguments for trixi2vtk can be found in the Trixi2Vtk.jl API.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Further information regarding the development of Trixi2Vtk can be found in the development section.","category":"page"},{"location":"visualization/#Makie.jl","page":"Visualization","title":"Makie.jl [experimental]","text":"","category":"section"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"In addition to Plots.jl support, Trixi.jl includes visualization utilities through Makie.jl. Trixi.jl provides Makie-based visualization options both for heatmap-type plots (similar to the Plots.jl recipes) as well as for interactive surface plots. Support is currently limited to the UnstructuredMesh2D type.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nPlotting via Makie.jl is still considered an experimental feature and might change in any future releases.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"A Makie plot can be created as follows: after running a simulation with Trixi.jl in the REPL, load a Makie backend (for example, GLMakie or CairoMakie).","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> using GLMakie","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"To visualize the solution and mesh with a heatmap-type plot, simply run","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"note: Note\nBoth Makie.jl and Plots.jl export plot, so if you load both libraries, you will have to specify which plot function to call via Plots.plot or Makie.plot.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"As with Plots.jl recipes, one can view individual solution components by creating a PlotData2D object and indexing into it with the desired variable name","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> pd = PlotData2D(sol)\njulia> plot(pd[\"rho\"])","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Unlike the Plots.jl recipe, mesh plotting is controlled using the keyword argument plot_mesh = false, e.g.,","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> plot(sol; plot_mesh=false)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot command also returns figure and axis handles, which can be used to edit plot titles or labels:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> fig, axes = plot(sol)\njulia> axes[1,1].title = \"New title for subplot (1,1)\"","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Trixi.jl also supports interactive surface plots using iplot. After executing","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> trixi_include(joinpath(\"examples\", \"unstructured_2d_dgsem\", \"elixir_euler_wall_bc.jl\"))","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"we can run","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"julia> iplot(sol)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"This will open up an interactive visualization window:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: makie-example)","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"The plot can be rotated (click and hold), zoomed in and out (scroll up and down), and panned (hold right click and drag). Two toggle buttons control whether mesh lines are visible on top of and below the solution.","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"Both plot and iplot use colormap = :inferno by default. A different colormap can be selected by providing an appropriate keyword argument. For example, plot(sol, colormap=:blues) and iplot(sol, colormap=:blues) produce the following figures:","category":"page"},{"location":"visualization/","page":"Visualization","title":"Visualization","text":"(Image: makie-plot-example) (Image: makie-iplot-example)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"EditURL = \"../../literate/src/files/adding_new_parabolic_terms.jl\"","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#adding_new_parabolic_terms","page":"13 Adding new parabolic terms","title":"13: Adding new parabolic terms","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"This demo illustrates the steps involved in adding new parabolic terms for the scalar advection equation. In particular, we will add an anisotropic diffusion. We begin by defining the hyperbolic (advection) part of the advection-diffusion equation.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"using OrdinaryDiffEq\nusing Trixi\n\n\nadvection_velocity = (1.0, 1.0)\nequations_hyperbolic = LinearScalarAdvectionEquation2D(advection_velocity);\nnothing #hide","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Define-a-new-parabolic-equation-type","page":"13 Adding new parabolic terms","title":"Define a new parabolic equation type","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Next, we define a 2D parabolic diffusion term type. This is similar to LaplaceDiffusion2D except that the diffusivity field refers to a spatially constant diffusivity matrix now. Note that ConstantAnisotropicDiffusion2D has a field for equations_hyperbolic. It is useful to have information about the hyperbolic system available to the parabolic part so that we can reuse functions defined for hyperbolic equations (such as varnames).","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"The abstract type Trixi.AbstractEquationsParabolic has three parameters: NDIMS (the spatial dimension, e.g., 1D, 2D, or 3D), NVARS (the number of variables), and GradientVariable, which we set as GradientVariablesConservative. This indicates that the gradient should be taken with respect to the conservative variables (e.g., the same variables used in equations_hyperbolic). Users can also take the gradient with respect to a different set of variables; see, for example, the implementation of CompressibleNavierStokesDiffusion2D, which can utilize either \"primitive\" or \"entropy\" variables.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"struct ConstantAnisotropicDiffusion2D{E, T} <: Trixi.AbstractEquationsParabolic{2, 1, GradientVariablesConservative}\n diffusivity::T\n equations_hyperbolic::E\nend\n\nvarnames(variable_mapping, equations_parabolic::ConstantAnisotropicDiffusion2D) =\n varnames(variable_mapping, equations_parabolic.equations_hyperbolic)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Next, we define the viscous flux function. We assume that the mixed hyperbolic-parabolic system is of the form","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"partial_t u(tx) + partial_x (f_1(u) - g_1(u nabla u))\n + partial_y (f_2(u) - g_2(u nabla u)) = 0","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"where f_1(u), f_2(u) are the hyperbolic fluxes and g_1(u nabla u), g_2(u nabla u) denote the viscous fluxes. For anisotropic diffusion, the viscous fluxes are the first and second components of the matrix-vector product involving diffusivity and the gradient vector.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Here, we specialize the flux to our new parabolic equation type ConstantAnisotropicDiffusion2D.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"function Trixi.flux(u, gradients, orientation::Integer, equations_parabolic::ConstantAnisotropicDiffusion2D)\n @unpack diffusivity = equations_parabolic\n dudx, dudy = gradients\n if orientation == 1\n return SVector(diffusivity[1, 1] * dudx + diffusivity[1, 2] * dudy)\n else # if orientation == 2\n return SVector(diffusivity[2, 1] * dudx + diffusivity[2, 2] * dudy)\n end\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Defining-boundary-conditions","page":"13 Adding new parabolic terms","title":"Defining boundary conditions","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Trixi.jl's implementation of parabolic terms discretizes both the gradient and divergence using weak formulation. In other words, we discretize the system","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"beginaligned\nbmq = nabla u \nbmsigma = beginpmatrix g_1(u bmq) g_2(u bmq) endpmatrix \ntextviscous contribution = nabla cdot bmsigma\nendaligned","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Boundary data must be specified for all spatial derivatives, e.g., for both the gradient equation bmq = nabla u and the divergence of the viscous flux nabla cdot bmsigma. We account for this by introducing internal Gradient and Divergence types which are used to dispatch on each type of boundary condition.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"As an example, let us introduce a Dirichlet boundary condition with constant boundary data.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"struct BoundaryConditionConstantDirichlet{T <: Real}\n boundary_value::T\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"This boundary condition contains only the field boundary_value, which we assume to be some real-valued constant which we will impose as the Dirichlet data on the boundary.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Boundary conditions have generally been defined as \"callable structs\" (also known as \"functors\"). For each boundary condition, we need to specify the appropriate boundary data to return for both the Gradient and Divergence. Since the gradient is operating on the solution u, the boundary data should be the value of u, and we can directly impose Dirichlet data.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"@inline function (boundary_condition::BoundaryConditionConstantDirichlet)(flux_inner, u_inner, normal::AbstractVector,\n x, t, operator_type::Trixi.Gradient,\n equations_parabolic::ConstantAnisotropicDiffusion2D)\n return boundary_condition.boundary_value\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"While the gradient acts on the solution u, the divergence acts on the viscous flux bmsigma. Thus, we have to supply boundary data for the Divergence operator that corresponds to bmsigma. However, we've already imposed boundary data on u for a Dirichlet boundary condition, and imposing boundary data for bmsigma might overconstrain our problem.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Thus, for the Divergence boundary data under a Dirichlet boundary condition, we simply return flux_inner, which is boundary data for bmsigma computed using the \"inner\" or interior solution. This way, we supply boundary data for the divergence operation without imposing any additional conditions.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"@inline function (boundary_condition::BoundaryConditionConstantDirichlet)(flux_inner, u_inner, normal::AbstractVector,\n x, t, operator_type::Trixi.Divergence,\n equations_parabolic::ConstantAnisotropicDiffusion2D)\n return flux_inner\nend","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#A-note-on-the-choice-of-gradient-variables","page":"13 Adding new parabolic terms","title":"A note on the choice of gradient variables","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"It is often simpler to transform the solution variables (and solution gradients) to another set of variables prior to computing the viscous fluxes (see CompressibleNavierStokesDiffusion2D for an example of this). If this is done, then the boundary condition for the Gradient operator should be modified accordingly as well.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Putting-things-together","page":"13 Adding new parabolic terms","title":"Putting things together","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"Finally, we can instantiate our new parabolic equation type, define boundary conditions, and run a simulation. The specific anisotropic diffusion matrix we use produces more dissipation in the direction (1 -1) as an isotropic diffusion.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"For boundary conditions, we impose that u=1 on the left wall, u=2 on the bottom wall, and u = 0 on the outflow walls. The initial condition is taken to be u = 0. Note that we use BoundaryConditionConstantDirichlet only for the parabolic boundary conditions, since we have not defined its behavior for the hyperbolic part.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"using Trixi: SMatrix\ndiffusivity = 5.0e-2 * SMatrix{2, 2}([2 -1; -1 2])\nequations_parabolic = ConstantAnisotropicDiffusion2D(diffusivity, equations_hyperbolic);\n\nboundary_conditions_hyperbolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1.0)),\n y_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(2.0)),\n y_pos = boundary_condition_do_nothing,\n x_pos = boundary_condition_do_nothing)\n\nboundary_conditions_parabolic = (; x_neg = BoundaryConditionConstantDirichlet(1.0),\n y_neg = BoundaryConditionConstantDirichlet(2.0),\n y_pos = BoundaryConditionConstantDirichlet(0.0),\n x_pos = BoundaryConditionConstantDirichlet(0.0));\n\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\ncoordinates_min = (-1.0, -1.0) # minimum coordinates (min(x), min(y))\ncoordinates_max = ( 1.0, 1.0) # maximum coordinates (max(x), max(y))\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n periodicity=false, n_cells_max=30_000) # set maximum capacity of tree data structure\n\ninitial_condition = (x, t, equations) -> SVector(0.0)\n\nsemi = SemidiscretizationHyperbolicParabolic(mesh,\n (equations_hyperbolic, equations_parabolic),\n initial_condition, solver;\n boundary_conditions=(boundary_conditions_hyperbolic,\n boundary_conditions_parabolic))\n\ntspan = (0.0, 2.0)\node = semidiscretize(semi, tspan)\ncallbacks = CallbackSet(SummaryCallback())\ntime_int_tol = 1.0e-6\nsol = solve(ode, RDPK3SpFSAL49(); abstol=time_int_tol, reltol=time_int_tol,\n ode_default_options()..., callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/#Package-versions","page":"13 Adding new parabolic terms","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"","category":"page"},{"location":"tutorials/adding_new_parabolic_terms/","page":"13 Adding new parabolic terms","title":"13 Adding new parabolic terms","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"EditURL = \"../../literate/src/files/p4est_from_gmsh.jl\"","category":"page"},{"location":"tutorials/p4est_from_gmsh/#p4est_from_gmsh","page":"17 P4est mesh from gmsh","title":"17: P4est mesh from gmsh","text":"","category":"section"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"Trixi.jl supports numerical approximations from structured and unstructured quadrilateral meshes with the P4estMesh mesh type.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"The purpose of this tutorial is to demonstrate how to use the P4estMesh functionality of Trixi.jl for existing meshes with straight-sided (bilinear) elements/cells. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will cover how to use existing meshes generated by gmsh or any other meshing software that can export to the Abaqus input .inp format.","category":"page"},{"location":"tutorials/p4est_from_gmsh/#Running-the-simulation-of-a-near-field-flow-around-an-airfoil","page":"17 P4est mesh from gmsh","title":"Running the simulation of a near-field flow around an airfoil","text":"","category":"section"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"Trixi.jl supports solving hyperbolic-parabolic problems on several mesh types. A somewhat complex example that employs the P4estMesh is the near-field simulation of a Mach 2 flow around the NACA6412 airfoil.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"using Trixi\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi_include(joinpath(examples_dir(), \"p4est_2d_dgsem\", \"elixir_euler_NACA6412airfoil_mach2.jl\"), tspan=(0.0, 0.5))\nend #hide","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"Conveniently, we use the Plots package to have a first look at the results:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"using Plots\npd = PlotData2D(sol)\nplot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/p4est_from_gmsh/#Creating-a-mesh-using-gmsh","page":"17 P4est mesh from gmsh","title":"Creating a mesh using gmsh","text":"","category":"section"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"The creation of an unstructured quadrilateral mesh using gmsh is driven by a geometry file. There are plenty of possibilities for the user, see the documentation and tutorials.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"To begin, we provide a complete geometry file for the NACA6412 airfoil bounded by a rectangular box. After this we give a breakdown of the most important parts required for successful mesh generation that can later be used by the p4est library and Trixi.jl. We emphasize that this near-field mesh should only be used for instructive purposes and not for actual production runs.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"The associated NACA6412.geo file is given below:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":" // GMSH geometry script for a NACA 6412 airfoil with 11 degree angle of attack\n // in a box (near-field mesh).\n // see https://github.com/cfsengineering/GMSH-Airfoil-2D\n // for software to generate gmsh `.geo` geometry files for NACA airfoils.\n\n // outer bounding box\n Point(1) = {-1.25, -0.5, 0, 1.0};\n Point(2) = {1.25, -0.5, 0, 1.0};\n Point(3) = {1.25, 0.5, 0, 1.0};\n Point(4) = {-1.25, 0.5, 0, 1.0};\n\n // lines of the bounding box\n Line(1) = {1, 2};\n Line(2) = {2, 3};\n Line(3) = {3, 4};\n Line(4) = {4, 1};\n // outer box\n Line Loop(8) = {1, 2, 3, 4};\n\n // Settings\n // This value gives the global element size factor (lower -> finer mesh)\n Mesh.CharacteristicLengthFactor = 1.0 * 2^(-3);\n // Insist on quads instead of default triangles\n Mesh.RecombineAll = 1;\n // Violet instead of green base color for better visibility\n Mesh.ColorCarousel = 0;\n\n // points of the airfoil contour\n // Format: {x, y, z, DesiredCellSize}. See the documentation: https://gmsh.info/doc/texinfo/gmsh.html#Points\n // These concrete points are generated using the tool from https://github.com/cfsengineering/GMSH-Airfoil-2D\n Point(5) = {-0.4900332889206208, 0.09933466539753061, 0, 0.125};\n Point(6) = {-0.4900274857651495, 0.1021542752054094, 0, 0.125};\n Point(7) = {-0.4894921489729144, 0.1049830248247787, 0, 0.125};\n Point(8) = {-0.4884253336670712, 0.1078191282319664, 0, 0.125};\n Point(9) = {-0.4868257975566199, 0.1106599068424483, 0, 0.125};\n Point(10) = {-0.4846930063965668, 0.1135018003016681, 0, 0.125};\n Point(11) = {-0.4820271400142729, 0.1163403835785654, 0, 0.125};\n Point(12) = {-0.4788290988083472, 0.1191703902233889, 0, 0.125};\n Point(13) = {-0.4751005105908123, 0.1219857416089041, 0, 0.125};\n Point(14) = {-0.4708437376101668, 0.1247795819332056, 0, 0.125};\n Point(15) = {-0.4660618835629463, 0.1275443187232316, 0, 0.125};\n Point(16) = {-0.4607588003749649, 0.1302716685409717, 0, 0.125};\n Point(17) = {-0.4549390945110529, 0.132952707559475, 0, 0.125};\n Point(18) = {-0.448608132554204, 0.1355779266432996, 0, 0.125};\n Point(19) = {-0.4417720457819508, 0.138137290538182, 0, 0.125};\n Point(20) = {-0.4344377334597768, 0.140620300747629, 0, 0.125};\n Point(21) = {-0.4266128645686593, 0.1430160616500159, 0, 0.125};\n Point(22) = {-0.4183058776865576, 0.1453133493887722, 0, 0.125};\n Point(23) = {-0.4095259787518715, 0.147500683050503, 0, 0.125};\n Point(24) = {-0.4002831364505879, 0.1495663976315875, 0, 0.125};\n Point(25) = {-0.3905880749878933, 0.1514987182830453, 0, 0.125};\n Point(26) = {-0.3804522640292948, 0.1532858353164163, 0, 0.125};\n Point(27) = {-0.3698879056254708, 0.1549159794501833, 0, 0.125};\n Point(28) = {-0.3589079179688306, 0.1563774967770029, 0, 0.125};\n Point(29) = {-0.3475259158676376, 0.1576589229368209, 0, 0.125};\n Point(30) = {-0.3357561878650377, 0.158749055989923, 0, 0.125};\n Point(31) = {-0.3236136699747923, 0.1596370274972017, 0, 0.125};\n Point(32) = {-0.3111139160522804, 0.1603123713324616, 0, 0.125};\n Point(33) = {-0.298273064867608, 0.160765089773461, 0, 0.125};\n Point(34) = {-0.2851078039966239, 0.1609857164445887, 0, 0.125};\n Point(35) = {-0.2716353306943914, 0.160965375714529, 0, 0.125};\n Point(36) = {-0.2578733099632437, 0.1606958381868515, 0, 0.125};\n Point(37) = {-0.2438398300730194, 0.1601695719599709, 0, 0.125};\n Point(38) = {-0.2295533558334121, 0.1593797893750759, 0, 0.125};\n Point(39) = {-0.2150326799566391, 0.1583204890160489, 0, 0.125};\n Point(40) = {-0.2002968728818922, 0.1569864927736143, 0, 0.125};\n Point(41) = {-0.18536523146042, 0.1553734778363979, 0, 0.125};\n Point(42) = {-0.1702572269208345, 0.1534780035235666, 0, 0.125};\n Point(43) = {-0.1549924525477129, 0.1512975329264932, 0, 0.125};\n Point(44) = {-0.1395905715122586, 0.1488304493795921, 0, 0.125};\n Point(45) = {-0.1240712652914332, 0.1460760678321895, 0, 0.125};\n Point(46) = {-0.1084541831014299, 0.1430346412430583, 0, 0.125};\n Point(47) = {-0.09275889275279087, 0.1397073621660917, 0, 0.125};\n Point(48) = {-0.07700483330818747, 0.1360963597385416, 0, 0.125};\n Point(49) = {-0.06151286635366404, 0.1323050298149023, 0, 0.125};\n Point(50) = {-0.04602933219022032, 0.1283521764905442, 0, 0.125};\n Point(51) = {-0.03051345534800332, 0.1242331665904082, 0, 0.125};\n Point(52) = {-0.01498163190522334, 0.1199540932779839, 0, 0.125};\n Point(53) = {0.0005498526140696458, 0.1155214539466913, 0, 0.125};\n Point(54) = {0.01606484191716884, 0.1109421303284033, 0, 0.125};\n Point(55) = {0.03154732664394777, 0.106223368423828, 0, 0.125};\n Point(56) = {0.0469814611314705, 0.1013727584299359, 0, 0.125};\n Point(57) = {0.06235157928986135, 0.09639821481480275, 0, 0.125};\n Point(58) = {0.07764220964363855, 0.09130795666388933, 0, 0.125};\n Point(59) = {0.09283808959671735, 0.08611048839446452, 0, 0.125};\n Point(60) = {0.1079241789809607, 0.08081458090718853, 0, 0.125};\n Point(61) = {0.1228856729475325, 0.07542925321638272, 0, 0.125};\n Point(62) = {0.1377080142575372, 0.06996375457378261, 0, 0.125};\n Point(63) = {0.1523769050236616, 0.06442754707512513, 0, 0.125};\n Point(64) = {0.1668783179480157, 0.05883028871526293, 0, 0.125};\n Point(65) = {0.1811985070933818, 0.05318181683604975, 0, 0.125};\n Point(66) = {0.1953240182159306, 0.04749213189240609, 0, 0.125};\n Point(67) = {0.2092416986775084, 0.04177138144606024, 0, 0.125};\n Point(68) = {0.2229387069452062, 0.03602984428372727, 0, 0.125};\n Point(69) = {0.2364025216754475, 0.03027791454712048, 0, 0.125};\n Point(70) = {0.2496209503696738, 0.02452608575629232, 0, 0.125};\n Point(71) = {0.2625821375791982, 0.01878493460541621, 0, 0.125};\n Point(72) = {0.2752745726282818, 0.01306510441121807, 0, 0.125};\n Point(73) = {0.28768709681727, 0.007377288098728577, 0, 0.125};\n Point(74) = {0.2998089100619555, 0.001732210616722449, 0, 0.125};\n Point(75) = {0.3116295769214332, -0.003859389314124759, 0, 0.125};\n Point(76) = {0.3231390319647309, -0.009386778203927332, 0, 0.125};\n Point(77) = {0.3343275844265582, -0.01483924761490708, 0, 0.125};\n Point(78) = {0.3451859221046181, -0.02020613485126957, 0, 0.125};\n Point(79) = {0.3557051144551212, -0.02547684454806881, 0, 0.125};\n Point(80) = {0.3658766148492779, -0.03064087116872238, 0, 0.125};\n Point(81) = {0.3756922619615632, -0.0356878223992288, 0, 0.125};\n Point(82) = {0.3851442802702071, -0.0406074434050937, 0, 0.125};\n Point(83) = {0.394225279661484, -0.04538964189492445, 0, 0.125};\n Point(84) = {0.4029282541416501, -0.05002451391298904, 0, 0.125};\n Point(85) = {0.4112465796735204, -0.05450237026215737, 0, 0.125};\n Point(86) = {0.4191740111683733, -0.05881376343890812, 0, 0.125};\n Point(87) = {0.4267046786777481, -0.06294951494382847, 0, 0.125};\n Point(88) = {0.4338330828434404, -0.06690074281456823, 0, 0.125};\n Point(89) = {0.4405540896772232, -0.07065888921378868, 0, 0.125};\n Point(90) = {0.4468629247542237, -0.07421574789251445, 0, 0.125};\n Point(91) = {0.4527551669150955, -0.0775634913396257, 0, 0.125};\n Point(92) = {0.4582267415819197, -0.08069469742118066, 0, 0.125};\n Point(93) = {0.4632739138007936, -0.08360237530891265, 0, 0.125};\n Point(94) = {0.4678932811302005, -0.08627999049569551, 0, 0.125};\n Point(95) = {0.4720817664982195, -0.08872148869699745, 0, 0.125};\n Point(96) = {0.4758366111533843, -0.09092131844134463, 0, 0.125};\n Point(97) = {0.4791553678333992, -0.09287445215953141, 0, 0.125};\n Point(98) = {0.4820358942729613, -0.09457640559161551, 0, 0.125};\n Point(99) = {0.4844763471666588, -0.09602325534252773, 0, 0.125};\n Point(100) = {0.4864751766953637, -0.09721165443119822, 0, 0.125};\n Point(101) = {0.4880311217148797, -0.09813884569428721, 0, 0.125};\n Point(102) = {0.4891432056939881, -0.09880267292366274, 0, 0.125};\n Point(103) = {0.4898107334756874, -0.09920158963645126, 0, 0.125};\n Point(104) = {0.4900332889206208, -0.09933466539753058, 0, 0.125};\n Point(105) = {0.4897824225031319, -0.09926905587549506, 0, 0.125};\n Point(106) = {0.4890301110661922, -0.09907236506934192, 0, 0.125};\n Point(107) = {0.4877772173496635, -0.09874500608402761, 0, 0.125};\n Point(108) = {0.48602517690576, -0.09828766683852558, 0, 0.125};\n Point(109) = {0.4837759946062035, -0.09770130916007558, 0, 0.125};\n Point(110) = {0.4810322398085871, -0.09698716747297723, 0, 0.125};\n Point(111) = {0.4777970402368822, -0.09614674703990023, 0, 0.125};\n Point(112) = {0.4740740746447117, -0.09518182170326678, 0, 0.125};\n Point(113) = {0.4698675643422793, -0.09409443106501386, 0, 0.125};\n Point(114) = {0.4651822636784212, -0.09288687703518478, 0, 0.125};\n Point(115) = {0.460023449577924, -0.09156171967354482, 0, 0.125};\n Point(116) = {0.4543969102408585, -0.09012177224394632, 0, 0.125};\n Point(117) = {0.4483089331151018, -0.08857009539864649, 0, 0.125};\n Point(118) = {0.4417662922553667, -0.08690999040934186, 0, 0.125};\n Point(119) = {0.4347762351819332, -0.0851449913634191, 0, 0.125};\n Point(120) = {0.4273464693498908, -0.08327885624791403, 0, 0.125};\n Point(121) = {0.419485148335155, -0.08131555684993674, 0, 0.125};\n Point(122) = {0.411200857836944, -0.07925926741086739, 0, 0.125};\n Point(123) = {0.4025026015879757, -0.07711435198240155, 0, 0.125};\n Point(124) = {0.3933997872536054, -0.07488535044544484, 0, 0.125};\n Point(125) = {0.3839022123897198, -0.07257696316779733, 0, 0.125};\n Point(126) = {0.3740200505167618, -0.07019403429336624, 0, 0.125};\n Point(127) = {0.3637638373540689, -0.06774153367408606, 0, 0.125};\n Point(128) = {0.3531444572451353, -0.06522453747557577, 0, 0.125};\n Point(129) = {0.3421731297908021, -0.06264820750853495, 0, 0.125};\n Point(130) = {0.3308613966940724, -0.06001776935966011, 0, 0.125};\n Point(131) = {0.3192211088076166, -0.05733848941811218, 0, 0.125};\n Point(132) = {0.3072644133633567, -0.05461565091590426, 0, 0.125};\n Point(133) = {0.2950037413531683, -0.05185452912263369, 0, 0.125};\n Point(134) = {0.2824517950208982, -0.04906036585632723, 0, 0.125};\n Point(135) = {0.2696215354188702, -0.04623834349241404, 0, 0.125};\n Point(136) = {0.2565261699769623, -0.04339355867155523, 0, 0.125};\n Point(137) = {0.2431791400293651, -0.04053099592384862, 0, 0.125};\n Point(138) = {0.2295941082432855, -0.03765550144139543, 0, 0.125};\n Point(139) = {0.2157849458952252, -0.03477175724299444, 0, 0.125};\n Point(140) = {0.2017657199439165, -0.03188425598348005, 0, 0.125};\n Point(141) = {0.187550679854507, -0.02899727666564914, 0, 0.125};\n Point(142) = {0.1731542441359161, -0.02611486151457043, 0, 0.125};\n Point(143) = {0.1585909865622793, -0.02324079427214604, 0, 0.125};\n Point(144) = {0.1438756220597465, -0.02037858016395433, 0, 0.125};\n Point(145) = {0.129022992251319, -0.0175314277805827, 0, 0.125};\n Point(146) = {0.1140480506645569, -0.01470223310184333, 0, 0.125};\n Point(147) = {0.09896584761949168, -0.01189356587453844, 0, 0.125};\n Point(148) = {0.08379151482656089, -0.009107658532933174, 0, 0.125};\n Point(149) = {0.06854024973648176, -0.006346397826038436, 0, 0.125};\n Point(150) = {0.05322729969528361, -0.003611319287478529, 0, 0.125};\n Point(151) = {0.03786794596792287, -0.00090360465249055, 0, 0.125};\n Point(152) = {0.0224774877026287, 0.00177591770710904, 0, 0.125};\n Point(153) = {0.007071225915134205, 0.004426769294862437, 0, 0.125};\n Point(154) = {-0.00833555242305456, 0.007048814950562587, 0, 0.125};\n Point(155) = {-0.02372759010533726, 0.009642253300220296, 0, 0.125};\n Point(156) = {-0.03908967513210498, 0.01220760427359278, 0, 0.125};\n Point(157) = {-0.05440665578848514, 0.01474569380579989, 0, 0.125};\n Point(158) = {-0.06966345527617318, 0.01725763587663899, 0, 0.125};\n Point(159) = {-0.08484508582421563, 0.01974481207672138, 0, 0.125};\n Point(160) = {-0.09987987792382108, 0.02219618763023203, 0, 0.125};\n Point(161) = {-0.1145078729404739, 0.02450371976411331, 0, 0.125};\n Point(162) = {-0.1290321771824579, 0.0267015185742735, 0, 0.125};\n Point(163) = {-0.143440065923266, 0.02879471001709845, 0, 0.125};\n Point(164) = {-0.1577189448447794, 0.03078883518202784, 0, 0.125};\n Point(165) = {-0.1718563428491159, 0.03268980457290044, 0, 0.125};\n Point(166) = {-0.1858399037768357, 0.03450385196323842, 0, 0.125};\n Point(167) = {-0.1996573773370766, 0.03623748825421298, 0, 0.125};\n Point(168) = {-0.2132966095779342, 0.03789745574015834, 0, 0.125};\n Point(169) = {-0.2267455332406906, 0.0394906831577609, 0, 0.125};\n Point(170) = {-0.2399921583489679, 0.04102424186233269, 0, 0.125};\n Point(171) = {-0.2530245633834605, 0.04250530343879837, 0, 0.125};\n Point(172) = {-0.2658308873846617, 0.04394109901707172, 0, 0.125};\n Point(173) = {-0.2783993233102972, 0.04533888052223981, 0, 0.125};\n Point(174) = {-0.2907181129514687, 0.04670588405019788, 0, 0.125};\n Point(175) = {-0.3027755436824813, 0.0480492955198111, 0, 0.125};\n Point(176) = {-0.3145599472847223, 0.04937621871394801, 0, 0.125};\n Point(177) = {-0.3260597010456697, 0.05069364578437131, 0, 0.125};\n Point(178) = {-0.337263231291058, 0.05200843025992359, 0, 0.125};\n Point(179) = {-0.3481590194623916, 0.05332726256406103, 0, 0.125};\n Point(180) = {-0.3587356108043638, 0.05465664801682354, 0, 0.125};\n Point(181) = {-0.3689816256782782, 0.0560028872679817, 0, 0.125};\n Point(182) = {-0.3788857734692287, 0.05737205908247899, 0, 0.125};\n Point(183) = {-0.3884368690074614, 0.05877000537646382, 0, 0.125};\n Point(184) = {-0.3976238513788748, 0.06020231838219783, 0, 0.125};\n Point(185) = {-0.40643580495675, 0.06167432980291591, 0, 0.125};\n Point(186) = {-0.4148619824472646, 0.06319110180426264, 0, 0.125};\n Point(187) = {-0.4228918297057104, 0.06475741967717524, 0, 0.125};\n Point(188) = {-0.43051501204915, 0.06637778599795482, 0, 0.125};\n Point(189) = {-0.4377214417649294, 0.06805641610468524, 0, 0.125};\n Point(190) = {-0.4445013064933708, 0.06979723470503821, 0, 0.125};\n Point(191) = {-0.4508450981473512, 0.07160387342876083, 0, 0.125};\n Point(192) = {-0.4567436420215075, 0.073479669138689, 0, 0.125};\n Point(193) = {-0.4621881257395756, 0.07542766281688272, 0, 0.125};\n Point(194) = {-0.4671701276898881, 0.07745059884734995, 0, 0.125};\n Point(195) = {-0.471681644606229, 0.07955092452372269, 0, 0.125};\n Point(196) = {-0.4757151179639407, 0.0817307896190848, 0, 0.125};\n Point(197) = {-0.4792634588791559, 0.0839920458658267, 0, 0.125};\n Point(198) = {-0.4823200712220043, 0.08633624620581726, 0, 0.125};\n Point(199) = {-0.4848788726822436, 0.08876464368523246, 0, 0.125};\n Point(200) = {-0.4869343135575803, 0.09127818988394577, 0, 0.125};\n Point(201) = {-0.4884813930704814, 0.09387753278635144, 0, 0.125};\n Point(202) = {-0.4895156730580155, 0.09656301401871749, 0, 0.125};\n\n // splines of the airfoil\n Spline(5) = {5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104};\n Spline(6) = {104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,5};\n\n // airfoil\n Line Loop(9) = {5, 6};\n // complete domain\n Plane Surface(1) = {8, 9};\n\n // labeling of the boundary parts\n Physical Line(1) = {4}; // inflow\n Physical Line(2) = {2}; // outflow\n Physical Line(3) = {1, 3}; // airfoil\n Physical Line(4) = {5, 6}; // upper/lower wall\n Physical Surface(1) = {10};","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"From which we can construct a mesh like this: (Image: mesh_screenshot)","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"The first four points define the bounding box = (near-field) domain:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":" // outer bounding box\nPoint(1) = {-1.25, -0.5, 0, 1.0};\nPoint(2) = {1.25, -0.5, 0, 1.0};\nPoint(3) = {1.25, 0.5, 0, 1.0};\nPoint(4) = {-1.25, 0.5, 0, 1.0};","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"which is constructed from connecting the points in lines:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// outer box\nLine(1) = {1, 2};\nLine(2) = {2, 3};\nLine(3) = {3, 4};\nLine(4) = {4, 1};\n// outer box\nLine Loop(8) = {1, 2, 3, 4};","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"This is followed by a couple (in principle optional) settings where the most important one is","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// Insist on quads instead of default triangles\nMesh.RecombineAll = 1;","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"which forces gmsh to generate quadrilateral elements instead of the default triangles. This is strictly required to be able to use the mesh later with p4est, which supports only straight-sided quads, i.e., C2D4, CPS4, S4 in 2D and C3D in 3D. See for more details the (short) documentation on the interaction of p4est with .inp files. In principle, it should also be possible to use the recombine function of gmsh to convert the triangles to quads, but this is observed to be less robust than enforcing quads from the beginning.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"Then the airfoil is defined by a set of points:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// points of the airfoil contour\n Point(5) = {-0.4900332889206208, 0.09933466539753061, 0, 0.125};\n Point(6) = {-0.4900274857651495, 0.1021542752054094, 0, 0.125};\n ...","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"which are connected by splines for the upper and lower part of the airfoil:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// splines of the airfoil\n Spline(5) = {5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,\n ...\n 96,97,98,99,100,101,102,103,104};\n Spline(6) = {104,105,106,107,108,109,110,111,112,113,114,115,\n ...\n 200,201,202,5};","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"which are then connected to form a single line loop for easy physical group assignment:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// airfoil\n Line Loop(9) = {5, 6};","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"At the end of the file the physical groups are defined:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"// labeling of the boundary parts\n Physical Line(1) = {4}; // Inflow. Label in Abaqus .inp file: PhysicalLine1\n Physical Line(2) = {2}; // Outflow. Label in Abaqus .inp file: PhysicalLine2\n Physical Line(3) = {1, 3}; // Upper and lower wall/farfield/... Label in Abaqus .inp file: PhysicalLine3\n Physical Line(4) = {5, 6}; // Airfoil. Label in Abaqus .inp file: PhysicalLine4","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"which are crucial for the correct assignment of boundary conditions in Trixi.jl. In particular, it is the responsibility of a user to keep track on the physical boundary names between the mesh generation and assignment of boundary condition functions in an elixir.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"After opening this file in gmsh, meshing the geometry and exporting to Abaqus .inp format, we can have a look at the input file:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"*Heading\n \n*NODE\n1, -1.25, -0.5, 0\n2, 1.25, -0.5, 0\n3, 1.25, 0.5, 0\n4, -1.25, 0.5, 0\n...\n******* E L E M E N T S *************\n*ELEMENT, type=T3D2, ELSET=Line1\n1, 1, 7\n...\n*ELEMENT, type=CPS4, ELSET=Surface1\n191, 272, 46, 263, 807\n...\n*NSET,NSET=PhysicalLine1\n1, 4, 52, 53, 54, 55, 56, 57, 58,\n*NSET,NSET=PhysicalLine2\n2, 3, 26, 27, 28, 29, 30, 31, 32,\n*NSET,NSET=PhysicalLine3\n1, 2, 3, 4, 7, 8, 9, 10, 11, 12,\n13, 14, 15, 16, 17, 18, 19, 20, 21, 22,\n23, 24, 25, 33, 34, 35, 36, 37, 38, 39,\n40, 41, 42, 43, 44, 45, 46, 47, 48, 49,\n50, 51,\n*NSET,NSET=PhysicalLine4\n5, 6, 59, 60, 61, 62, 63, 64, 65, 66,\n67, 68, 69, 70, 71, 72, 73, 74, 75, 76,\n77, 78, 79, 80, 81, 82, 83, 84, 85, 86,\n87, 88, 89, 90, 91, 92, 93, 94, 95, 96,\n97, 98, 99, 100, 101, 102, 103, 104, 105, 106,\n107, 108, 109, 110, 111, 112, 113, 114, 115, 116,\n117, 118, 119, 120, 121, 122, 123, 124, 125, 126,\n127, 128, 129, 130, 131, 132, 133, 134, 135, 136,\n137, 138, 139, 140, 141, 142, 143, 144, 145, 146,\n147, 148, 149, 150, 151, 152, 153, 154, 155, 156,\n157, 158, 159, 160, 161, 162, 163, 164, 165, 166,\n167, 168, 169, 170, 171, 172, 173, 174, 175, 176,\n177, 178, 179, 180, 181, 182, 183, 184, 185, 186,\n187, 188, 189, 190,","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"First, the coordinates of the nodes are listed, followed by the elements. Note that gmsh exports also line elements of type T3D2 which are ignored by p4est. The relevant elements in 2D which form the gridcells are of type CPS4 which are defined by their four corner nodes. This is followed by the nodesets encoded via *NSET which are used to assign boundary conditions in Trixi.jl. Trixi.jl parses the .inp file and assigns the edges (in 2D, surfaces in 3D) of elements to the corresponding boundary condition based on the supplied boundary_symbols that have to be supplied to the P4estMesh constructor:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"# boundary symbols\nboundary_symbols = [:PhysicalLine1, :PhysicalLine2, :PhysicalLine3, :PhysicalLine4]\nmesh = P4estMesh{2}(mesh_file, polydeg = polydeg, boundary_symbols = boundary_symbols)","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"The same boundary symbols have then also be supplied to the semidiscretization alongside the corresponding physical boundary conditions:","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"# Supersonic inflow boundary condition.\n# Calculate the boundary flux entirely from the external solution state, i.e., set\n# external solution state values for everything entering the domain.\n@inline function boundary_condition_supersonic_inflow(u_inner,\n normal_direction::AbstractVector,\n x, t, surface_flux_function,\n equations::CompressibleEulerEquations2D)\n u_boundary = initial_condition_mach2_flow(x, t, equations)\n flux = Trixi.flux(u_boundary, normal_direction, equations)\n\n return flux\nend\n\n# Supersonic outflow boundary condition.\n# Calculate the boundary flux entirely from the internal solution state. Analogous to supersonic inflow\n# except all the solution state values are set from the internal solution as everything leaves the domain\n@inline function boundary_condition_supersonic_outflow(u_inner,\n normal_direction::AbstractVector, x,\n t,\n surface_flux_function,\n equations::CompressibleEulerEquations2D)\nflux = Trixi.flux(u_inner, normal_direction, equations)\n\nboundary_conditions = Dict(:PhysicalLine1 => boundary_condition_supersonic_inflow, # Left boundary\n :PhysicalLine2 => boundary_condition_supersonic_outflow, # Right boundary\n :PhysicalLine3 => boundary_condition_supersonic_outflow, # Top and bottom boundary\n :PhysicalLine4 => boundary_condition_slip_wall) # Airfoil\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions = boundary_conditions)","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"Note that you have to supply the boundary_symbols keyword to the P4estMesh constructor to select the boundaries from the available nodesets in the .inp file. If the boundary_symbols keyword is not supplied, all boundaries will be assigned to the default set :all.","category":"page"},{"location":"tutorials/p4est_from_gmsh/#Package-versions","page":"17 P4est mesh from gmsh","title":"Package versions","text":"","category":"section"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\", \"Download\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"","category":"page"},{"location":"tutorials/p4est_from_gmsh/","page":"17 P4est mesh from gmsh","title":"17 P4est mesh from gmsh","text":"This page was generated using Literate.jl.","category":"page"},{"location":"code_of_conduct/","page":"Code of Conduct","title":"Code of Conduct","text":"EditURL = \"https://github.com/trixi-framework/Trixi.jl/blob/main/CODE_OF_CONDUCT.md\"","category":"page"},{"location":"code_of_conduct/#code-of-conduct","page":"Code of Conduct","title":"Code of Conduct","text":"","category":"section"},{"location":"code_of_conduct/","page":"Code of Conduct","title":"Code of Conduct","text":"Contributor Covenant Code of ConductOur PledgeWe as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation.We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community.Our StandardsExamples of behavior that contributes to a positive environment for our community include:Demonstrating empathy and kindness toward other people\nBeing respectful of differing opinions, viewpoints, and experiences\nGiving and gracefully accepting constructive feedback\nAccepting responsibility and apologizing to those affected by our mistakes, and learning from the experience\nFocusing on what is best not just for us as individuals, but for the overall communityExamples of unacceptable behavior include:The use of sexualized language or imagery, and sexual attention or advances of any kind\nTrolling, insulting or derogatory comments, and personal or political attacks\nPublic or private harassment\nPublishing others' private information, such as a physical or email address, without their explicit permission\nOther conduct which could reasonably be considered inappropriate in a professional settingEnforcement ResponsibilitiesCommunity leaders are responsible for clarifying and enforcing our standards of acceptable behavior and will take appropriate and fair corrective action in response to any behavior that they deem inappropriate, threatening, offensive, or harmful.Community leaders have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, and will communicate reasons for moderation decisions when appropriate.ScopeThis Code of Conduct applies within all community spaces, and also applies when an individual is officially representing the community in public spaces. Examples of representing our community include using an official e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event.EnforcementInstances of abusive, harassing, or otherwise unacceptable behavior may be reported to Michael Schlottke-Lakemper, Hendrik Ranocha, or any other of the principal developers responsible for enforcement listed in Authors. All complaints will be reviewed and investigated promptly and fairly.All community leaders are obligated to respect the privacy and security of the reporter of any incident.Enforcement GuidelinesCommunity leaders will follow these Community Impact Guidelines in determining the consequences for any action they deem in violation of this Code of Conduct:1. CorrectionCommunity Impact: Use of inappropriate language or other behavior deemed unprofessional or unwelcome in the community.Consequence: A private, written warning from community leaders, providing clarity around the nature of the violation and an explanation of why the behavior was inappropriate. A public apology may be requested.2. WarningCommunity Impact: A violation through a single incident or series of actions.Consequence: A warning with consequences for continued behavior. No interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, for a specified period of time. This includes avoiding interactions in community spaces as well as external channels like social media. Violating these terms may lead to a temporary or permanent ban.3. Temporary BanCommunity Impact: A serious violation of community standards, including sustained inappropriate behavior.Consequence: A temporary ban from any sort of interaction or public communication with the community for a specified period of time. No public or private interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, is allowed during this period. Violating these terms may lead to a permanent ban.4. Permanent BanCommunity Impact: Demonstrating a pattern of violation of community standards, including sustained inappropriate behavior, harassment of an individual, or aggression toward or disparagement of classes of individuals.Consequence: A permanent ban from any sort of public interaction within the community.AttributionThis Code of Conduct is adapted from the [Contributor Covenant][homepage], version 2.0, available at https://www.contributor-covenant.org/version/2/0/codeofconduct.html.Community Impact Guidelines were inspired by Mozilla's code of conduct enforcement ladder.[homepage]: https://www.contributor-covenant.orgFor answers to common questions about this code of conduct, see the FAQ at https://www.contributor-covenant.org/faq. Translations are available at https://www.contributor-covenant.org/translations.","category":"page"},{"location":"conventions/#conventions","page":"Conventions","title":"Conventions","text":"","category":"section"},{"location":"conventions/#Spatial-dimensions-and-directions","page":"Conventions","title":"Spatial dimensions and directions","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"We use the following numbering schemes on Cartesian or curved structured meshes.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"The orientations are numbered as 1 => x, 2 => y, 3 => z. For example, numerical fluxes such as flux_central(u_ll, u_rr, orientation, equations::AbstractEquations) use the orientation in this way.\nThe directions are numbered as 1 => -x, 2 => +x, 3 => -y, 4 => +y, 5 => -z, 6 => +z. For example, the boundary_conditions are ordered in this way when a Tuple of boundary conditions per direction is passed to the constructor of a SemidiscretizationHyperbolic.\nFor structured and unstructured curved meshes the concept of direction is generalized via the variable normal_direction. This variable points in the normal direction at a given, curved surface. For the computation of boundary fluxes the normal_direction is normalized to be a normal_vector used, for example, in FluxRotated.","category":"page"},{"location":"conventions/#Cells-vs.-elements-vs.-nodes","page":"Conventions","title":"Cells vs. elements vs. nodes","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"To uniquely distinguish between different components of the discretization, we use the following naming conventions:","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"The computational domain is discretized by a mesh, which is made up of individual cells. In general, neither the mesh nor the cells should be aware of any solver-specific knowledge, i.e., they should not store anything that goes beyond the geometrical information and the connectivity.\nThe numerical solvers do not directly store their information inside the mesh, but use own data structures. Specifically, for each cell on which a solver wants to operate, the solver creates an element to store solver-specific data.\nFor discretization schemes such as the discontinuous Galerkin or the finite element method, inside each element multiple nodes may be defined, which hold nodal information. The nodes are again a solver-specific component, just like the elements.\nWe often identify elements, nodes, etc. with their (local or global) integer index. Convenience iterators such as eachelement, eachnode use these indices.","category":"page"},{"location":"conventions/#Keywords-in-elixirs","page":"Conventions","title":"Keywords in elixirs","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Trixi.jl is distributed with several examples in the form of elixirs, small Julia scripts containing everything to set up and run a simulation. Working interactively from the Julia REPL with these scripts can be quite convenient while for exploratory research and development of Trixi.jl. For example, you can use the convenience function trixi_include to include an elixir with some modified arguments. To enable this, it is helpful to use a consistent naming scheme in elixirs, since trixi_include can only perform simple replacements. Some standard variables names are","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"polydeg for the polynomial degree of a solver\nsurface_flux for the numerical flux at surfaces\nvolume_flux for the numerical flux used in flux differencing volume terms","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Moreover, convergence_test requires that the spatial resolution is set via the keywords","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"initial_refinement_level (an integer, e.g. for the TreeMesh and the P4estMesh) or\ncells_per_dimension (a tuple of integers, one per spatial dimension, e.g. for the StructuredMesh and the DGMultiMesh).","category":"page"},{"location":"conventions/#Variable-names","page":"Conventions","title":"Variable names","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Use descriptive names (using snake_case for variables/functions and CamelCase for types)\nUse a suffix _ as in name_ for local variables that would otherwise hide existing symbols.\nUse a prefix _ as in _name to indicate internal methods/data that are \"fragile\" in the sense that there's no guarantee that they might get changed without notice. These are also not documented with a docstring (but maybe with comments using #).","category":"page"},{"location":"conventions/#Array-types-and-wrapping","page":"Conventions","title":"Array types and wrapping","text":"","category":"section"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"To allow adaptive mesh refinement efficiently when using time integrators from OrdinaryDiffEq, Trixi.jl allows to represent numerical solutions in two different ways. Some discussion can be found online and in form of comments describing Trixi.wrap_array and Trixi.wrap_array_native in the source code of Trixi.jl. The flexibility introduced by this possible wrapping enables additional performance optimizations. However, it comes at the cost of some additional abstractions (and needs to be used with caution, as described in the source code of Trixi.jl). Thus, we use the following conventions to distinguish between arrays visible to the time integrator and wrapped arrays mainly used internally.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Arrays visible to the time integrator have a suffix _ode, e.g., du_ode, u_ode.\nWrapped arrays do not have a suffix, e.g., du, u.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"Methods either accept arrays visible to the time integrator or wrapped arrays based on the following rules.","category":"page"},{"location":"conventions/","page":"Conventions","title":"Conventions","text":"When some solution is passed together with a semidiscretization semi, the solution must be a u_ode that needs to be wrapped via wrap_array(u_ode, semi) (or wrap_array_native(u_ode, semi)) for further processing.\nWhen some solution is passed together with the mesh, equations, solver, cache, ..., it is already wrapped via wrap_array (or wrap_array_native).\nExceptions of this rule are possible, e.g. for AMR, but must be documented in the code.\nwrap_array should be used as default option. wrap_array_native should only be used when necessary, e.g., to avoid additional overhead when interfacing with external C libraries such as HDF5, MPI, or visualization.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"EditURL = \"../../literate/src/files/non_periodic_boundaries.jl\"","category":"page"},{"location":"tutorials/non_periodic_boundaries/#non_periodic_boundaries","page":"6 Non-periodic boundaries","title":"6: Non-periodic boundaries","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Dirichlet-boundary-condition","page":"6 Non-periodic boundaries","title":"Dirichlet boundary condition","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"First, let's look at the Dirichlet boundary condition BoundaryConditionDirichlet.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"BoundaryConditionDirichlet(boundary_value_function)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"In Trixi.jl, this creates a Dirichlet boundary condition where the function boundary_value_function is used to set the values at the boundary. It can be used to create a boundary condition that sets exact boundary values by passing the exact solution of the equation.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"It is important to note that standard Dirichlet boundary conditions for hyperbolic PDEs do not make sense in most cases. However, we are using a special weak form of the Dirichlet boundary condition, based on the application of the numerical surface flux. The numerical surface flux takes the solution value from inside the domain and the prescribed value of the outer boundary state as arguments, and solves an approximate Riemann problem to introduce dissipation (and hence stabilization) at the boundary. Hence, the performance of the Dirichlet BC depends on the fidelity of the numerical surface flux. An easy-to read introductory reference on this topic is the paper by Mengaldo et al..","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"The passed boundary value function is called with the same arguments as an initial condition function, i.e.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"boundary_value_function(x, t, equations)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"where x specifies the spatial coordinates, t is the current time, and equations is the corresponding system of equations.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"We want to give a short example for a simulation with such a Dirichlet BC.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Consider the one-dimensional linear advection equation with domain Omega=0 2 and a constant zero initial condition.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"using OrdinaryDiffEq, Trixi\n\nadvection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)\n\ninitial_condition_zero(x, t, equation::LinearScalarAdvectionEquation1D) = SVector(0.0)\ninitial_condition = initial_condition_zero\n\nusing Plots\nplot(x -> sum(initial_condition(x, 0.0, equations)), label=\"initial condition\", ylim=(-1.5, 1.5))","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Using an advection velocity of 1.0 and the (local) Lax-Friedrichs/Rusanov flux FluxLaxFriedrichs as a numerical surface flux, we are able to create an inflow boundary on the left and an outflow boundary on the right, as the Lax-Friedrichs flux is in this case an exact characteristics Riemann solver. We note that for more complex PDEs different strategies for inflow/outflow boundaries are necessary. To define the inflow values, we initialize a boundary_value_function.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"function boundary_condition_sine_sector(x, t, equation::LinearScalarAdvectionEquation1D)\n if 1 <= t <= 3\n scalar = sin(2 * pi * sum(t - 1))\n else\n scalar = zero(t)\n end\n return SVector(scalar)\nend\nboundary_condition = boundary_condition_sine_sector","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"We set the BC in negative and positive x-direction.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"boundary_conditions = (x_neg=BoundaryConditionDirichlet(boundary_condition),\n x_pos=BoundaryConditionDirichlet(boundary_condition))","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\ncoordinates_min = (0.0,)\ncoordinates_max = (2.0,)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"For the mesh type TreeMesh the parameter periodicity must be set to false in the corresponding direction.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"mesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n n_cells_max=10_000,\n periodicity=false)\n\n\nsemi = SemidiscretizationHyperbolic(mesh, equations,\n initial_condition,\n solver,\n boundary_conditions=boundary_conditions)\n\ntspan = (0.0, 6.0)\node = semidiscretize(semi, tspan)\n\nanalysis_callback = AnalysisCallback(semi, interval=100,)\n\nstepsize_callback = StepsizeCallback(cfl=0.9)\n\ncallbacks = CallbackSet(analysis_callback,\n stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"We define some equidistant nodes for the visualization","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"visnodes = range(tspan[1], tspan[2], length=300)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"and run the simulation.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, saveat=visnodes, callback=callbacks);\n\nusing Plots\n@gif for step in 1:length(sol.u)\n plot(sol.u[step], semi, ylim=(-1.5, 1.5), legend=true, label=\"approximation\", title=\"time t=$(round(sol.t[step], digits=5))\")\n scatter!([0.0], [sum(boundary_condition(SVector(0.0), sol.t[step], equations))], label=\"boundary condition\")\nend","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Other-available-example-elixirs-with-non-trivial-BC","page":"6 Non-periodic boundaries","title":"Other available example elixirs with non-trivial BC","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Moreover, there are other boundary conditions in Trixi.jl. For instance, you can use the slip wall boundary condition boundary_condition_slip_wall.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Trixi.jl provides some interesting examples with different combinations of boundary conditions, e.g. using boundary_condition_slip_wall and other self-defined boundary conditions using BoundaryConditionDirichlet.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"For instance, there is a 2D compressible Euler setup for a Mach 3 wind tunnel flow with a forward facing step in the elixir elixir_euler_forward_step_amr.jl discretized with a P4estMesh using adaptive mesh refinement (AMR).","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"A double Mach reflection problem for the 2D compressible Euler equations elixir_euler_double_mach_amr.jl exercises a special boundary conditions along the bottom of the domain that is a mixture of Dirichlet and slip wall.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"A channel flow around a cylinder at Mach 3 elixir_euler_supersonic_cylinder.jl contains supersonic Mach 3 inflow at the left portion of the domain and supersonic outflow at the right portion of the domain. The top and bottom of the channel as well as the cylinder are treated as Euler slip wall boundaries.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":" \n
","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"Source: Video on Trixi.jl's YouTube channel Trixi Framework","category":"page"},{"location":"tutorials/non_periodic_boundaries/#Package-versions","page":"6 Non-periodic boundaries","title":"Package versions","text":"","category":"section"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"","category":"page"},{"location":"tutorials/non_periodic_boundaries/","page":"6 Non-periodic boundaries","title":"6 Non-periodic boundaries","text":"This page was generated using Literate.jl.","category":"page"},{"location":"parallelization/#Parallelization","page":"Parallelization","title":"Parallelization","text":"","category":"section"},{"location":"parallelization/#Shared-memory-parallelization-with-threads","page":"Parallelization","title":"Shared-memory parallelization with threads","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Many compute-intensive loops in Trixi.jl are parallelized using the multi-threading support provided by Julia. You can recognize those loops by the @threaded macro prefixed to them, e.g.,","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"@threaded for element in eachelement(dg, cache)\n ...\nend","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"This will statically assign an equal iteration count to each available thread.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"To use multi-threading, you need to tell Julia at startup how many threads you want to use by either setting the environment variable JULIA_NUM_THREADS or by providing the -t/--threads command line argument. For example, to start Julia with four threads, start Julia with","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia --threads=4","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If both the environment variable and the command line argument are specified at the same time, the latter takes precedence.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If you use time integration methods from OrdinaryDiffEq.jl and want to use multiple threads therein, you need to set the keyword argument thread=OrdinaryDiffEq.True() of the algorithms, as described in the section on time integration methods.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"warning: Warning\nNot everything is parallelized yet and there are likely opportunities to improve scalability. Multi-threading isn't considered part of the public API of Trixi.jl yet.","category":"page"},{"location":"parallelization/#Distributed-computing-with-MPI","page":"Parallelization","title":"Distributed computing with MPI","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"In addition to the shared memory parallelization with multi-threading, Trixi.jl supports distributed parallelism via MPI.jl, which leverages the Message Passing Interface (MPI). MPI.jl comes with its own MPI library binaries such that there is no need to install MPI yourself. However, it is also possible to instead use an existing MPI installation, which is recommended if you are running MPI programs on a cluster or supercomputer (see the MPI.jl docs to find out how to select the employed MPI library). Additional notes on how to use a system-provided MPI installation with Trixi.jl can be found in the following subsection.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"warning: Work in progress\nMPI-based parallelization is work in progress and not finished yet. Nothing related to MPI is part of the official API of Trixi.jl yet.","category":"page"},{"location":"parallelization/#parallel_system_MPI","page":"Parallelization","title":"Using a system-provided MPI installation","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"When using Trixi.jl with a system-provided MPI backend, the underlying p4est, t8code and HDF5 libraries need to be compiled with the same MPI installation. If you want to use p4est (via the P4estMesh) or t8code (via the T8codeMesh) from Trixi.jl, you also need to use system-provided p4est or t8code installations (for notes on how to install p4est and t8code see, e.g., here and here, use the configure option --enable-mpi). Otherwise, there will be warnings that no preference is set for P4est.jl and T8code.jl that can be ignored if you do not use these libraries from Trixi.jl. Note that t8code already comes with a p4est installation, so it suffices to install t8code. In order to use system-provided p4est and t8code installations, P4est.jl and T8code.jl need to be configured to use the custom installations. Follow the steps described here and here for the configuration. The paths that point to libp4est.so (and potentially to libsc.so) need to be the same for P4est.jl and T8code.jl. This could, e.g., be libp4est.so that usually can be found in lib/ or local/lib/ in the installation directory of t8code. Note that the T8codeMesh, however, does not support MPI yet. The preferences for HDF5.jl always need to be set, even if you do not want to use HDF5 from Trixi.jl, see also issue #1079 in HDF5.jl. To set the preferences for HDF5.jl, follow the instructions described here.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"In total, in your active Julia project you should have a LocalPreferences.toml file with sections [MPIPreferences], [T8code] (only needed if T8codeMesh is used), [P4est] (only needed if P4estMesh is used), and [HDF5] as well as an entry MPIPreferences in your Project.toml to use a custom MPI installation. A LocalPreferences.toml file created as described above might look something like the following:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"[HDF5]\nlibhdf5 = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\"\nlibhdf5_hl = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\"\n\n[HDF5_jll]\nlibhdf5_hl_path = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\"\nlibhdf5_path = \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\"\n\n[MPIPreferences]\n__clear__ = [\"preloads_env_switch\"]\n_format = \"1.0\"\nabi = \"OpenMPI\"\nbinary = \"system\"\ncclibs = []\nlibmpi = \"/lib/x86_64-linux-gnu/libmpi.so\"\nmpiexec = \"mpiexec\"\npreloads = []\n\n[P4est]\nlibp4est = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\"\nlibsc = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\"\n\n[T8code]\nlibp4est = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\"\nlibsc = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\"\nlibt8 = \"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libt8.so\"","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"This file is created with the following sequence of commands:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using MPIPreferences\njulia> MPIPreferences.use_system_binary()","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Restart the Julia REPL","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using P4est\njulia> P4est.set_library_p4est!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libp4est.so\")\njulia> P4est.set_library_sc!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/libsc.so\")\njulia> using T8code\njulia> T8code.set_libraries_path!(\"/home/mschlott/hackathon/libtrixi/t8code/install/lib/\")\njulia> using HDF5\njulia> HDF5.API.set_libraries!(\"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so\", \"/usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5_hl.so\")","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"After the preferences are set, restart the Julia REPL again.","category":"page"},{"location":"parallelization/#parallel_usage","page":"Parallelization","title":"Usage","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"To start Trixi.jl in parallel with MPI, there are three options:","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Run from the REPL with mpiexec(): You can start a parallel execution directly from the REPL by executing\njulia> using MPI\n\njulia> mpiexec() do cmd\n run(`$cmd -n 3 $(Base.julia_cmd()) --threads=1 --project=@. -e 'using Trixi; trixi_include(default_example())'`)\n end\nThe parameter -n 3 specifies that Trixi.jl should run with three processes (or ranks in MPI parlance) and should be adapted to your available computing resources and problem size. The $(Base.julia_cmd()) argument ensures that Julia is executed in parallel with the same optimization level etc. as you used for the REPL; if this is unnecessary or undesired, you can also just use julia. Further, if you are not running Trixi.jl from a local clone but have installed it as a package, you need to omit the --project=@..\nRun from the command line with mpiexecjl: Alternatively, you can use the mpiexecjl script provided by MPI.jl, which allows you to start Trixi.jl in parallel directly from the command line. As a preparation, you need to install the script once by running\njulia> using MPI\n\njulia> MPI.install_mpiexecjl(destdir=\"/somewhere/in/your/PATH\")\nThen, to execute Trixi.jl in parallel, execute the following command from your command line:\nmpiexecjl -n 3 julia --threads=1 --project=@. -e 'using Trixi; trixi_include(default_example())'\nRun interactively with tmpi (Linux/MacOS only): If you are on a Linux/macOS system, you have a third option which lets you run Julia in parallel interactively from the REPL. This comes in handy especially during development, as in contrast to the first two options, it allows to reuse the compilation cache and thus facilitates much faster startup times after the first execution. It requires tmux and the OpenMPI library to be installed before, both of which are usually available through a package manager. Once you have installed both tools, you need to configure MPI.jl to use the OpenMPI for your system, which is explained here. Then, you can download and install the tmpi script by executing\ncurl https://raw.githubusercontent.com/Azrael3000/tmpi/master/tmpi -o /somewhere/in/your/PATH/tmpi\nFinally, you can start and control multiple Julia REPLs simultaneously by running\ntmpi 3 julia --threads=1 --project=@.\nThis will start Julia inside tmux three times and multiplexes all commands you enter in one REPL to all other REPLs (try for yourself to understand what it means). If you have no prior experience with tmux, handling the REPL this way feels slightly weird in the beginning. However, there is a lot of documentation for tmux available and once you get the hang of it, developing Trixi.jl in parallel becomes much smoother this way. Some helpful commands are the following. To close a single pane you can press Ctrl+b and then x followed by y to confirm. To quit the whole session you press Ctrl+b followed by :kill-session. Often you would like to scroll up. You can do that by pressing Ctrl+b and then [, which allows you to use the arrow keys to scroll up and down. To leave the scroll mode you press q. Switching between panes can be done by Ctrl+b followed by o. As of March 2022, newer versions of tmpi also support mpich, which is the default backend of MPI.jl (via MPICH_Jll.jl). To use this setup, you need to install mpiexecjl as described in the documentation of MPI.jl and make it available as mpirun, e.g., via a symlink of the form\nln -s ~/.julia/bin/mpiexecjl /somewhere/in/your/path/mpirun\n(assuming default installations).","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"note: Hybrid parallelism\nIt is possible to combine MPI with shared memory parallelism via threads by starting Julia with more than one thread, e.g. by passing the command line argument julia --threads=2 instead of julia --threads=1 used in the examples above. In that case, you should make sure that your system supports the number of processes/threads that you try to start.","category":"page"},{"location":"parallelization/#parallel_performance","page":"Parallelization","title":"Performance","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"For information on how to evaluate the parallel performance of Trixi.jl, please have a look at the Performance metrics of the AnalysisCallback section, specifically at the descriptions of the performance index (PID).","category":"page"},{"location":"parallelization/#Using-error-based-step-size-control-with-MPI","page":"Parallelization","title":"Using error-based step size control with MPI","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If you use error-based step size control (see also the section on error-based adaptive step sizes) together with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.","category":"page"},{"location":"parallelization/#Using-parallel-input-and-output","page":"Parallelization","title":"Using parallel input and output","text":"","category":"section"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Trixi.jl allows parallel I/O using MPI by leveraging parallel HDF5.jl. On most systems, this is enabled by default. Additionally, you can also use a local installation of the HDF5 library (with MPI support). For this, you first need to use a system-provided MPI library, see also here and you need to tell HDF5.jl to use this library. To do so with HDF5.jl v0.17 and newer, set the preferences libhdf5 and libhdf5_hl to the local paths of the libraries libhdf5 and libhdf5_hl, which can be done by","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using Preferences, UUIDs\njulia> set_preferences!(\n UUID(\"f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f\"), # UUID of HDF5.jl\n \"libhdf5\" => \"/path/to/your/libhdf5.so\",\n \"libhdf5_hl\" => \"/path/to/your/libhdf5_hl.so\", force = true)","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"Alternatively, with HDF5.jl v0.17.1 or higher you can use","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"julia> using HDF5\njulia> HDF5.API.set_libraries!(\"/path/to/your/libhdf5.so\", \"/path/to/your/libhdf5_hl.so\")","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"For more information see also the documentation of HDF5.jl. In total, you should have a file called LocalPreferences.toml in the project directory that contains a section [MPIPreferences], a section [HDF5] with entries libhdf5 and libhdf5_hl, a section [P4est] with the entry libp4est as well as a section [T8code] with the entries libt8, libp4est and libsc. If you use HDF5.jl v0.16 or older, instead of setting the preferences for HDF5.jl, you need to set the environment variable JULIA_HDF5_PATH to the path, where the HDF5 binaries are located and then call ]build HDF5 from Julia.","category":"page"},{"location":"parallelization/","page":"Parallelization","title":"Parallelization","text":"If HDF5 is not MPI-enabled, Trixi.jl will fall back on a less efficient I/O mechanism. In that case, all disk I/O is performed only on rank zero and data is distributed to/gathered from the other ranks using regular MPI communication.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"EditURL = \"../../literate/src/files/behind_the_scenes_simulation_setup.jl\"","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/#behind_the_scenes_simulation_setup","page":"2 Behind the scenes of a simulation setup","title":"2: Behind the scenes of a simulation setup","text":"","category":"section"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"This tutorial will guide you through a simple Trixi.jl setup (\"elixir\"), giving an overview of what happens in the background during the initialization of a simulation. While the setup described herein does not cover all details, it involves relatively stable parts of Trixi.jl that are unlikely to undergo significant changes in the near future. The goal is to clarify some of the more fundamental, technical concepts that are applicable to a variety of (also more complex) configurations.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Trixi.jl follows the method of lines concept for solving partial differential equations (PDEs). Firstly, the PDEs are reduced to a (potentially huge) system of ordinary differential equations (ODEs) by discretizing the spatial derivatives. Subsequently, these generated ODEs may be solved with methods available in OrdinaryDiffEq.jl or those specifically implemented in Trixi.jl. The following steps elucidate the process of transitioning from PDEs to ODEs within the framework of Trixi.jl.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/#Basic-setup","page":"2 Behind the scenes of a simulation setup","title":"Basic setup","text":"","category":"section"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Import essential libraries and specify an equation.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"using Trixi, OrdinaryDiffEq\nequations = LinearScalarAdvectionEquation2D((-0.2, 0.7))","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Generate a spatial discretization using a TreeMesh with a pre-coarsened set of cells.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"coordinates_min = (-2.0, -2.0)\ncoordinates_max = (2.0, 2.0)\n\ncoarsening_patches = ((type = \"box\", coordinates_min = [0.0, -2.0],\n coordinates_max = [2.0, 0.0]),)\n\nmesh = TreeMesh(coordinates_min, coordinates_max, initial_refinement_level = 2,\n n_cells_max = 30_000,\n coarsening_patches = coarsening_patches)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"The created TreeMesh looks like the following:","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: TreeMesh_example)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Instantiate a DGSEM solver with a user-specified polynomial degree. The solver will define polydeg + 1 Gauss-Lobatto nodes and their associated weights within the reference interval -1 1 in each spatial direction. These nodes will be subsequently used to approximate solutions on each leaf cell of the TreeMesh.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"solver = DGSEM(polydeg = 3)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Gauss-Lobatto nodes with polydeg = 3:","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: Gauss-Lobatto_nodes_example)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/#Overview-of-the-[SemidiscretizationHyperbolic](@ref)-type","page":"2 Behind the scenes of a simulation setup","title":"Overview of the SemidiscretizationHyperbolic type","text":"","category":"section"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"At this stage, all necessary components for configuring the spatial discretization are in place. The remaining task is to combine these components into a single structure that will be used throughout the entire simulation process. This is where SemidiscretizationHyperbolic comes into play.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test,\n solver)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"The constructor for the SemidiscretizationHyperbolic object calls numerous sub-functions to perform the necessary initialization steps. A brief description of the key sub-functions is provided below.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"init_elements(leaf_cell_ids, mesh, equations, dg.basis, RealT, uEltype)\nThe fundamental elements for approximating the solution are the leaf cells. The solution is constructed as a polynomial of the degree specified in the DGSEM solver in each spatial direction on each leaf cell. This polynomial approximation is evaluated at the Gauss-Lobatto nodes mentioned earlier. The init_elements function extracts these leaf cells from the TreeMesh, assigns them the label \"elements\", records their coordinates, and maps the Gauss-Lobatto nodes from the 1D interval -1 1 onto each coordinate axis of every element.\n(Image: elements_example)\nThe visualization of elements with nodes shown here includes spaces between elements, which do not exist in reality. This spacing is included only for illustrative purposes to underscore the separation of elements and the independent projection of nodes onto each element.\ninit_interfaces(leaf_cell_ids, mesh, elements)\nAt this point, the elements with nodes have been defined; however, they lack the necessary communication functionality. This is crucial because the local solution polynomials on the elements are not independent of each other. Furthermore, nodes on the boundary of adjacent elements share the same spatial location, which requires a method to combine this into a meaningful solution. Here Riemann solvers come into play which can handle the principal ambiguity of a multi-valued solution at the same spatial location.\nAs demonstrated earlier, the elements can have varying sizes. Let us initially consider neighbors with equal size. For these elements, the init_interfaces function generates interfaces that store information about adjacent elements, their relative positions, and allocate containers for sharing solution data between neighbors during the solution process.\nIn our visualization, these interfaces would conceptually resemble tubes connecting the corresponding elements.\n(Image: interfaces_example)\ninit_mortars(leaf_cell_ids, mesh, elements, dg.mortar)\nReturning to the consideration of different sizes among adjacent elements, within the TreeMesh, adjacent leaf cells can vary in side length by a maximum factor of two. This implies that a large element has one neighbor of equal size with a connection through an interface, or two neighbors at half the size, requiring a connection through so called \"mortars\". In 3D, a large element would have four small neighbor elements.\nMortars store information about the connected elements, their relative positions, and allocate containers for storing the solutions along the boundaries between these elements.\nDue to the differing sizes of adjacent elements, it is not feasible to directly map boundary nodes of adjacent elements. Therefore, the concept of mortars employs a mass-conserving interpolation function to map boundary nodes from a larger element to a smaller one.\nIn our visualization, mortars are represented as branched tubes.\n(Image: mortars_example)\ninit_boundaries(leaf_cell_ids, mesh, elements)\nIn order to apply boundary conditions, it is necessary to identify the locations of the boundaries. Therefore, we initialize a \"boundaries\" object, which records the elements that contain boundaries, specifies which side of an element is a boundary, stores the coordinates of boundary nodes, and allocates containers for managing solutions at these boundaries.\nIn our visualization, boundaries and their corresponding nodes are highlighted with green, semi-transparent lines.\n(Image: boundaries_example)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"All the structures mentioned earlier are collected as a cache of type NamedTuple. Subsequently, an object of type SemidiscretizationHyperbolic is initialized using this cache, initial and boundary conditions, equations, mesh and solver.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"In conclusion, the primary purpose of a SemidiscretizationHyperbolic is to collect equations, the geometric representation of the domain, and approximation instructions, creating specialized structures to interconnect these components in a manner that enables their utilization for the numerical solution of partial differential equations (PDEs).","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"As evident from the earlier description of SemidiscretizationHyperbolic, it comprises numerous functions called subsequently. Without delving into details, the structure of the primary calls are illustrated as follows:","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: SemidiscretizationHyperbolic_structure)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/#Overview-of-the-[semidiscretize](@ref)-function","page":"2 Behind the scenes of a simulation setup","title":"Overview of the semidiscretize function","text":"","category":"section"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"At this stage, we have defined the equations and configured the domain's discretization. The final step before solving is to select a suitable time span and apply the corresponding initial conditions, which are already stored in the initialized SemidiscretizationHyperbolic object.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"The purpose of the semidiscretize function is to wrap the semidiscretization as an ODEProblem within the specified time interval. During this procedure the approximate solution is created at the given initial time via the specified initial_condition function from the SemidiscretizationHyperbolic object. This ODEProblem can be subsequently passed to the solve function from the OrdinaryDiffEq.jl package or to Trixi.solve.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"ode = semidiscretize(semi, (0.0, 1.0));\nnothing #hide","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"The semidiscretize function involves a deep tree of subsequent calls, with the primary ones explained below.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"allocate_coefficients(mesh, equations, solver, cache)\nTo apply initial conditions, a data structure (\"container\") needs to be generated to store the initial values of the target variables for each node within each element.\nSince only one-dimensional Arrays are resize!able in Julia, we use Vectors as an internal storage for the target variables and resize! them whenever needed, e.g. to change the number of elements. Then, during the solving process the same memory is reused by unsafe_wrapping multi-dimensional Arrays around the internal storage.\nwrap_array(u_ode, semi)\nAs previously noted, u_ode is constructed as a 1D vector to ensure compatibility with OrdinaryDiffEq.jl. However, for internal use within Trixi.jl, identifying which part of the vector relates to specific variables, elements, or nodes can be challenging.\nThis is why the u_ode vector is wrapped by the wrap_array function using unsafe_wrap to form a multidimensional array u. In this array, the first dimension corresponds to variables, followed by N dimensions corresponding to nodes for each of N space dimensions. The last dimension corresponds to the elements. Consequently, navigation within this multidimensional array becomes noticeably easier.\n\"Wrapping\" in this context involves the creation of a reference to the same storage location but with an alternative structural representation. This approach enables the use of both instances u and u_ode as needed, so that changes are simultaneously reflected in both. This is possible because, from a storage perspective, they share the same stored data, while access to this data is provided in different ways.\ncompute_coefficients!(u, initial_conditions, t, mesh::DG, equations, solver, cache)\nNow the variable u, intended to store solutions, has been allocated and wrapped, it is time to apply the initial conditions. The compute_coefficients! function calculates the initial conditions for each variable at every node within each element and properly stores them in the u array.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"At this stage, the semidiscretize function has all the necessary components to initialize and return an ODEProblem object, which will be used by the solve function to compute the solution.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"In summary, the internal workings of semidiscretize with brief descriptions can be presented as follows.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: semidiscretize_structure)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/#Functions-solve-and-rhs!","page":"2 Behind the scenes of a simulation setup","title":"Functions solve and rhs!","text":"","category":"section"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Once the ODEProblem object is initialized, the solve function and one of the ODE solvers from the OrdinaryDiffEq.jl package can be utilized to compute an approximated solution using the instructions contained in the ODEProblem object.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"sol = solve(ode, CarpenterKennedy2N54(williamson_condition = false), dt = 0.01,\n save_everystep = false);\nnothing #hide","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Since the solve function and the ODE solver have no knowledge of a particular spatial discretization, it is necessary to define a \"right-hand-side function\", rhs!, within Trixi.jl.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Trixi.jl includes a set of rhs! functions designed to compute du, i.e., fracpartial upartial t according to the structure of the setup. These rhs! functions calculate interface, mortars, and boundary fluxes, in addition to surface and volume integrals, in order to construct the du vector. This du vector is then used by the time integration method to obtain the solution at the subsequent time step. The rhs! function is called by time integration methods in each iteration of the solve loop within OrdinaryDiffEq.jl, with arguments du, u, semidiscretization, and the current time.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Trixi.jl uses a two-levels approach for rhs! functions. The first level is limited to a single function for each semidiscretization type, and its role is to redirect data to the target rhs! for specific solver and mesh types. This target rhs! function is responsible for calculating du.","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Path from the solve function call to the appropriate rhs! function call:","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"(Image: rhs_structure)","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"Computed solution:","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"using Plots\nplot(sol)\npd = PlotData2D(sol)\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"","category":"page"},{"location":"tutorials/behind_the_scenes_simulation_setup/","page":"2 Behind the scenes of a simulation setup","title":"2 Behind the scenes of a simulation setup","text":"This page was generated using Literate.jl.","category":"page"},{"location":"overview/#Overview-of-the-structure-of-Trixi.jl","page":"Overview","title":"Overview of the structure of Trixi.jl","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl is designed as a library of components for discretizations of hyperbolic conservation laws. Thus, it is not a monolithic PDE solver that is configured at runtime via parameter files, as it is often found in classical numerical simulation codes. Instead, each simulation is configured by pure Julia code. Many examples of such simulation setups, called elixirs in Trixi.jl, are provided in the examples/ folder.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl uses the method of lines, i.e., the full space-time discretization is separated into two steps; the spatial semidiscretization is performed at first and the resulting ODE system is solved numerically using a suitable time integration method. Thus, the main ingredients of an elixir designed to solve a PDE numerically are the spatial semidiscretization and the time integration scheme.","category":"page"},{"location":"overview/#overview-semidiscretizations","page":"Overview","title":"Semidiscretizations","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Semidiscretizations are high-level descriptions of spatial discretizations specialized for certain PDEs. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Such semidiscretizations are usually named semi in Trixi.jl","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"(Image: semidiscretization_overview)","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"The basic building blocks of a semidiscretization are","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"a mesh describing the geometry of the domain\na set of equations describing the physical model\na solver describing the numerical approach","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"In addition, a semidiscretization bundles initial and boundary conditions, and possible source terms. These different ingredients of a semidiscretization can be configured individually and combined together. When a semidiscretization is constructed, it will create an internal cache, i.e., a collection of setup-specific data structures, that is usually passed to all lower level functions.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Due to Trixi.jl's modular nature using Julia's multiple dispatch features, new ingredients can be created specifically for a certain combination of other ingredients. For example, a new mesh type can be created and implemented at first only for a specific solver. Thus, there is no need to consider all possible combinations of meshes, equations, and solvers when implementing new features. This allows rapid prototyping of new ideas and is one of the main design goals behind Trixi.jl. Below is a brief overview of the availability of different features on different mesh types.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"Feature TreeMesh StructuredMesh UnstructuredMesh2D P4estMesh DGMultiMesh Further reading\nSpatial dimension 1D, 2D, 3D 1D, 2D, 3D 2D 2D, 3D 1D, 2D, 3D \nCoordinates Cartesian curvilinear curvilinear curvilinear curvilinear \nConnectivity h-nonconforming conforming conforming h-nonconforming conforming \nElement type line, square, cube line, quadᵃ, hexᵃ quadᵃ quadᵃ, hexᵃ simplex, quadᵃ, hexᵃ \nAdaptive mesh refinement ✅ ❌ ❌ ✅ ❌ AMRCallback\nSolver type DGSEM DGSEM DGSEM DGSEM DGMulti \nDomain hypercube mapped hypercube arbitrary arbitrary arbitrary \nWeak form ✅ ✅ ✅ ✅ ✅ VolumeIntegralWeakForm\nFlux differencing ✅ ✅ ✅ ✅ ✅ VolumeIntegralFluxDifferencing\nShock capturing ✅ ✅ ✅ ✅ ❌ VolumeIntegralShockCapturingHG\nNonconservative equations ✅ ✅ ✅ ✅ ✅ e.g., GLM MHD or shallow water equations\nParabolic terms ✅ ✅ ❌ ✅ ✅ e.g., CompressibleNavierStokesDiffusion2D","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"ᵃ: quad = quadrilateral, hex = hexahedron","category":"page"},{"location":"overview/#Time-integration-methods","page":"Overview","title":"Time integration methods","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem is a wrapper of Trixi.rhs!(du_ode, u_ode, semi, t), which gets called in ODE solvers. Further information can be found in the section on time integration methods.","category":"page"},{"location":"overview/#Next-steps","page":"Overview","title":"Next steps","text":"","category":"section"},{"location":"overview/","page":"Overview","title":"Overview","text":"We explicitly encourage people interested in Trixi.jl to have a look at the examples/ bundled with Trixi.jl to get an impression of what is possible and the general look and feel of Trixi.jl. Before doing that, it is usually good to get an idea of how to visualize numerical results.","category":"page"},{"location":"overview/","page":"Overview","title":"Overview","text":"If you like learning by doing, looking at the tutorials and trying to mix your own elixirs based thereon is probably a good next step. Otherwise, you can further dig into the documentation by looking at Trixi.jl's basic building blocks.","category":"page"},{"location":"callbacks/#callbacks-id","page":"Callbacks","title":"Callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Many of the advanced features of Trixi.jl, such as adaptive mesh refinement, are implemented as callbacks. A callback is an algorithmic entity that gets passed to the ODE solver and is called at specific points during execution to perform certain tasks. Callbacks in Trixi.jl are either called after each time step (step callbacks) or after each stage of the ODE solver (stage callbacks).","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"(Image: callbacks_illustration)","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The advantage of callbacks over hard-coding all features is that it allows to extend Trixi.jl without modifying the internal source code. Trixi.jl provides callbacks for time step control, adaptive mesh refinement, I/O, and more.","category":"page"},{"location":"callbacks/#Step-callbacks","page":"Callbacks","title":"Step callbacks","text":"","category":"section"},{"location":"callbacks/#CFL-based-time-step-control","page":"Callbacks","title":"CFL-based time step control","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Time step control can be performed with a StepsizeCallback. An example making use of this can be found at examples/tree_2d_dgsem/elixir_advection_basic.jl","category":"page"},{"location":"callbacks/#Adaptive-mesh-refinement","page":"Callbacks","title":"Adaptive mesh refinement","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Trixi.jl uses a hierarchical Cartesian mesh which can be locally refined in a solution-adaptive way. This can be used to speed up simulations with minimal loss in overall accuracy. Adaptive mesh refinement (AMR) can be used by passing an AMRCallback to the ODE solver. The AMRCallback requires a controller such as ControllerThreeLevel or ControllerThreeLevelCombined to tell the AMR algorithm which cells to refine/coarsen.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"An example elixir using AMR can be found at examples/tree_2d_dgsem/elixir_advection_amr.jl.","category":"page"},{"location":"callbacks/#Analyzing-the-numerical-solution","page":"Callbacks","title":"Analyzing the numerical solution","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The AnalysisCallback can be used to analyze the numerical solution, e.g. calculate errors or user-specified integrals, and print the results to the screen. The results can also be saved in a file. An example can be found at examples/tree_2d_dgsem/elixir_euler_vortex.jl. Note that the errors (e.g. L2 error or Linf error) are computed with respect to the initial condition. The percentage of the simulation time refers to the ratio of the current time and the final time, i.e. it does not consider the maximal number of iterations. So the simulation could finish before 100% are reached. Note that, e.g., due to AMR or smaller time step sizes, the simulation can actually take longer than the percentage indicates. In Performance metrics of the AnalysisCallback you can find a detailed description of the different performance metrics the AnalysisCallback computes.","category":"page"},{"location":"callbacks/#I/O","page":"Callbacks","title":"I/O","text":"","category":"section"},{"location":"callbacks/#Solution-and-restart-files","page":"Callbacks","title":"Solution and restart files","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"To save the solution in regular intervals you can use a SaveSolutionCallback. It is also possible to create restart files using the SaveRestartCallback. An example making use of these can be found at examples/tree_2d_dgsem/elixir_advection_extended.jl. An example showing how to restart a simulation from a restart file can be found at examples/tree_2d_dgsem/elixir_advection_restart.jl.","category":"page"},{"location":"callbacks/#Time-series","page":"Callbacks","title":"Time series","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Sometimes it is useful to record the evaluations of state variables over time at a given set of points. This can be achieved by the TimeSeriesCallback, which is used, e.g., in examples/tree_2d_dgsem/elixir_acoustics_gaussian_source.jl. The TimeSeriesCallback constructor expects a semidiscretization and a list of points at which the solution should be recorded in regular time step intervals. After the last time step, the entire record is stored in an HDF5 file.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For the points, two different input formats are supported: You can either provide them as a list of tuples, which is handy if you specify them by hand on the REPL. Alternatively, you can provide them as a two-dimensional array, where the first dimension is the point number and the second dimension is the coordinate dimension. This is especially useful when reading them from a file.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For example, to record the primitive variables at the points (0.0, 0.0) and (-1.0, 0.5) every five timesteps and storing the collected data in the file tseries.h5, you can create the TimeSeriesCallback as","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"time_series = TimeSeriesCallback(semi, [(0.0, 0.0), (-1.0, 0.5)];\n interval=5,\n solution_variables=cons2prim,\n filename=\"tseries.h5\")","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"For a full list of possible arguments, please check the documentation for the TimeSeriesCallback. As an alternative to specifying the point coordinates directly in the elixir or on the REPL, you can read them from a file. For instance, with a text file points.dat with content","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":" 0.0 0.0\n-1.0 0.5","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"you can create a time series callback with","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"using DelimitedFiles: readdlm\ntime_series = TimeSeriesCallback(semi, readdlm(\"points.dat\"))","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"To plot the individual point data series over time, you can create a PlotData1D from the TimeSeriesCallback and a given point ID. For example, executing","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"julia> using Trixi, Plots\n\njulia> trixi_include(joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_acoustics_gaussian_source.jl\"))\n\njulia> pd1 = PlotData1D(time_series, 1)\n\njulia> pd2 = PlotData1D(time_series, 2)\n\njulia> plot(pd1[\"p_prime\"]); plot!(pd2[\"p_prime\"], xguide=\"t\")","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"will yield the following plot:","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"(Image: image)","category":"page"},{"location":"callbacks/#Miscellaneous","page":"Callbacks","title":"Miscellaneous","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The AliveCallback prints some information to the screen to show that a simulation is still running.\nThe SummaryCallback prints a human-readable summary of the simulation setup and controls the automated performance measurements, including an output of the recorded timers after a simulation.\nThe VisualizationCallback can be used for in-situ visualization. See Visualizing results during a simulation.\nThe TrivialCallback does nothing and can be used to easily disable some callbacks via trixi_include.","category":"page"},{"location":"callbacks/#Equation-specific-callbacks","page":"Callbacks","title":"Equation-specific callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Some callbacks provided by Trixi.jl implement specific features for certain equations:","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"The LBMCollisionCallback implements the Lattice-Boltzmann method (LBM) collision operator and should only be used when solving the Lattice-Boltzmann equations. See e.g. examples/tree_2d_dgsem/elixir_lbm_constant.jl\nThe SteadyStateCallback terminates the time integration when the residual steady state falls below a certain threshold. This checks the convergence of the potential phi for hyperbolic diffusion. See e.g. examples/tree_2d_dgsem/elixir_hypdiff_nonperiodic.jl.\nThe GlmSpeedCallback updates the divergence cleaning wave speed c_h for the ideal GLM-MHD equations. See e.g. examples/tree_2d_dgsem/elixir_mhd_alfven_wave.jl.","category":"page"},{"location":"callbacks/#Usage-of-step-callbacks","page":"Callbacks","title":"Usage of step callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Step callbacks are passed to the solve method from the ODE solver via the keyword argument callback. If you want to use a single callback cb, pass it as callback=cb. When using two or more callbacks, you need to turn them into a CallbackSet first by calling callbacks = CallbackSet(cb1, cb2) and passing it as callback=callbacks.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"note: Note\nThere are some restrictions regarding the order of callbacks in a CallbackSet.The callbacks are called after each time step but some callbacks actually belong to the next time step. Therefore, the callbacks should be ordered in the following way:Callbacks that belong to the current time step:\nSummaryCallback controls, among other things, timers and should thus be first\nSteadyStateCallback may mark a time step as the last one\nAnalysisCallback may do some checks that mark a time step as the last one\nAliveCallback should be nearby AnalysisCallback\nSaveSolutionCallback/SaveRestartCallback should save the current solution before it is degraded by AMR\nVisualizationCallback should be called before the mesh is adapted\nCallbacks that belong to the next time step:\nAMRCallback\nStepsizeCallback must be called after AMRCallback to accommodate potential changes to the mesh\nGlmSpeedCallback must be called after StepsizeCallback because the step size affects the value of c_h\nLBMCollisionCallback is already part of the calculations of the next time step and should therefore be called after StepsizeCallback","category":"page"},{"location":"callbacks/#Stage-callbacks","page":"Callbacks","title":"Stage callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"PositivityPreservingLimiterZhangShu is a positivity-preserving limiter, used to enforce physical constraints. An example elixir using this feature can be found at examples/tree_2d_dgsem/elixir_euler_positivity.jl.","category":"page"},{"location":"callbacks/#Implementing-new-callbacks","page":"Callbacks","title":"Implementing new callbacks","text":"","category":"section"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"Since Trixi.jl is compatible with OrdinaryDiffEq.jl, both packages share the same callback interface. A detailed description of it can be found in the OrdinaryDiffEq.jl documentation. Step callbacks are just called callbacks. Stage callbacks are called stage_limiter!.","category":"page"},{"location":"callbacks/","page":"Callbacks","title":"Callbacks","text":"An example elixir showing how to implement a new simple stage callback and a new simple step callback can be found at examples/tree_2d_dgsem/elixir_advection_callbacks.jl.","category":"page"},{"location":"#Trixi.jl","page":"Home","title":"Trixi.jl","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"(Image: Docs-stable) (Image: Docs-dev) (Image: Slack) (Image: Youtube) (Image: Build Status) (Image: Codecov) (Image: Coveralls) (Image: Aqua QA) (Image: License: MIT) (Image: DOI)","category":"page"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is a numerical simulation framework for conservation laws written in Julia. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures. Its features include:","category":"page"},{"location":"","page":"Home","title":"Home","text":"1D, 2D, and 3D simulations on line/quad/hex/simplex meshes\nCartesian and curvilinear meshes\nConforming and non-conforming meshes\nStructured and unstructured meshes\nHierarchical quadtree/octree grid with adaptive mesh refinement\nForests of quadtrees/octrees with p4est via P4est.jl\nHigh-order accuracy in space and time\nDiscontinuous Galerkin methods\nKinetic energy-preserving and entropy-stable methods based on flux differencing\nEntropy-stable shock capturing\nPositivity-preserving limiting\nFinite difference summation by parts (SBP) methods\nCompatible with the SciML ecosystem for ordinary differential equations\nExplicit low-storage Runge-Kutta time integration\nStrong stability preserving methods\nCFL-based and error-based time step control\nNative support for differentiable programming\nForward mode automatic differentiation via ForwardDiff.jl\nPeriodic and weakly-enforced boundary conditions\nMultiple governing equations:\nCompressible Euler equations\nCompressible Navier-Stokes equations\nMagnetohydrodynamics (MHD) equations\nMulti-component compressible Euler and MHD equations\nLinearized Euler and acoustic perturbation equations\nHyperbolic diffusion equations for elliptic problems\nLattice-Boltzmann equations (D2Q9 and D3Q27 schemes)\nShallow water equations\nScalar advection\nMulti-physics simulations\nSelf-gravitating gas dynamics\nShared-memory parallelization via multithreading\nMulti-node parallelization via MPI\nVisualization and postprocessing of the results\nIn-situ and a posteriori visualization with Plots.jl\nInteractive visualization with Makie.jl\nPostprocessing with ParaView/VisIt via Trixi2Vtk","category":"page"},{"location":"#Installation","page":"Home","title":"Installation","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you have not yet installed Julia, please follow the instructions for your operating system. Trixi.jl works with Julia v1.8 and newer. We recommend using the latest stable release of Julia.","category":"page"},{"location":"#For-users","page":"Home","title":"For users","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl and its related tools are registered Julia packages. Hence, you can install Trixi.jl, the visualization tool Trixi2Vtk, OrdinaryDiffEq.jl, and Plots.jl by executing the following commands in the Julia REPL:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg\n\njulia> Pkg.add([\"Trixi\", \"Trixi2Vtk\", \"OrdinaryDiffEq\", \"Plots\"])","category":"page"},{"location":"","page":"Home","title":"Home","text":"You can copy and paste all commands to the REPL including the leading julia> prompts - they will automatically be stripped away by Julia. The package OrdinaryDiffEq.jl provides time integration schemes used by Trixi.jl, while Plots.jl can be used to directly visualize Trixi.jl's results from the REPL.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note on package versions: If some of the examples for how to use Trixi.jl do not work, verify that you are using a recent Trixi.jl release by comparing the installed Trixi.jl version from","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg; Pkg.update(\"Trixi\"); Pkg.status(\"Trixi\")","category":"page"},{"location":"","page":"Home","title":"Home","text":"to the latest release. If the installed version does not match the current release, please check the Troubleshooting section.","category":"page"},{"location":"","page":"Home","title":"Home","text":"The commands above can also be used to update Trixi.jl. A brief list of notable changes to Trixi.jl is available in NEWS.md.","category":"page"},{"location":"#for-developers","page":"Home","title":"For developers","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you plan on editing Trixi.jl itself, you can download Trixi.jl to a local folder and use the code from the cloned directory:","category":"page"},{"location":"","page":"Home","title":"Home","text":"git clone git@github.com:trixi-framework/Trixi.jl.git\ncd Trixi.jl\nmkdir run\ncd run\njulia --project=. -e 'using Pkg; Pkg.develop(PackageSpec(path=\"..\"))'","category":"page"},{"location":"","page":"Home","title":"Home","text":"If you installed Trixi.jl this way, you always have to start Julia with the --project flag set to your run directory, e.g.,","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia --project=.","category":"page"},{"location":"","page":"Home","title":"Home","text":"if already inside the run directory.","category":"page"},{"location":"","page":"Home","title":"Home","text":"The advantage of using a separate run directory is that you can also add other related packages (see below, e.g., for time integration or visualization) to the project in the run folder and always have a reproducible environment at hand to share with others.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Since the postprocessing tool Trixi2Vtk.jl typically does not need to be modified, it is recommended to install it as a normal package. Likewise, you can install OrdinaryDiffEq.jl and Plots.jl as ordinary packages. To achieve this, use the following REPL commands:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Pkg\n\njulia> Pkg.add([\"OrdinaryDiffEq\", \"Trixi2Vtk\", \"Plots\"])","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note that the postprocessing tools Trixi2Vtk.jl and Plots.jl are optional and can be omitted.","category":"page"},{"location":"#Example:-Installing-Trixi.jl-as-a-package","page":"Home","title":"Example: Installing Trixi.jl as a package","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":" ","category":"page"},{"location":"","page":"Home","title":"Home","text":"Please note that the playback speed is set to 3x, thus the entire installation procedure lasts around 45 seconds in real time (depending on the performance of your computer and on how many dependencies had already been installed before).","category":"page"},{"location":"#Usage","page":"Home","title":"Usage","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"In the Julia REPL, first load the package Trixi.jl","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Trixi","category":"page"},{"location":"","page":"Home","title":"Home","text":"Then start a simulation by executing","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> trixi_include(default_example())","category":"page"},{"location":"","page":"Home","title":"Home","text":"Please be patient since Julia will compile the code just before running it. To visualize the results, load the package Plots","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> using Plots","category":"page"},{"location":"","page":"Home","title":"Home","text":"and generate a heatmap plot of the results with","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> plot(sol) # No trailing semicolon, otherwise no plot is shown","category":"page"},{"location":"","page":"Home","title":"Home","text":"This will open a new window with a 2D visualization of the final solution:","category":"page"},{"location":"","page":"Home","title":"Home","text":"(Image: image)","category":"page"},{"location":"","page":"Home","title":"Home","text":"The method trixi_include(...) expects a single string argument with the path to a Trixi.jl elixir, i.e., a text file containing Julia code necessary to set up and run a simulation. To quickly see Trixi.jl in action, default_example() returns the path to an example elixir with a short, two-dimensional problem setup. A list of all example elixirs packaged with Trixi.jl can be obtained by running get_examples(). Alternatively, you can also browse the examples/ subdirectory. If you want to modify one of the elixirs to set up your own simulation, download it to your machine, edit the configuration, and pass the file path to trixi_include(...).","category":"page"},{"location":"#Example:-Running-a-simulation-with-Trixi.jl","page":"Home","title":"Example: Running a simulation with Trixi.jl","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":" ","category":"page"},{"location":"","page":"Home","title":"Home","text":"If this produces weird symbols or question marks in the terminal on your system, you are probably using Mac OS with problematic fonts. In that case, please check the Troubleshooting section.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Note on performance: Julia uses just-in-time compilation to transform its source code to native, optimized machine code at the time of execution and caches the compiled methods for further use. That means that the first execution of a Julia method is typically slow, with subsequent runs being much faster. For instance, in the example above the first execution of trixi_include takes about 20 seconds, while subsequent runs require less than 60 milliseconds.","category":"page"},{"location":"#Performing-a-convergence-analysis","page":"Home","title":"Performing a convergence analysis","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"To automatically determine the experimental order of convergence (EOC) for a given setup, execute","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> convergence_test(default_example(), 4)","category":"page"},{"location":"","page":"Home","title":"Home","text":"This will run a convergence test with the elixir default_example(), using four iterations with different initial refinement levels. The initial iteration will use the elixir unchanged, while for each subsequent iteration the initial_refinement_level parameter is incremented by one. Finally, the measured l^2 and l^infty errors and the determined EOCs will be displayed like this:","category":"page"},{"location":"","page":"Home","title":"Home","text":"[...]\nl2\nscalar\nerror EOC\n9.14e-06 -\n5.69e-07 4.01\n3.55e-08 4.00\n2.22e-09 4.00\n\nmean 4.00\n--------------------------------------------------------------------------------\nlinf\nscalar\nerror EOC\n6.44e-05 -\n4.11e-06 3.97\n2.58e-07 3.99\n1.62e-08 4.00\n\nmean 3.99\n--------------------------------------------------------------------------------","category":"page"},{"location":"","page":"Home","title":"Home","text":"An example with multiple variables looks like this:","category":"page"},{"location":"","page":"Home","title":"Home","text":"julia> convergence_test(joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_euler_source_terms.jl\"), 3)","category":"page"},{"location":"","page":"Home","title":"Home","text":"[...]\nl2\nrho rho_v1 rho_v2 rho_e\nerror EOC error EOC error EOC error EOC\n9.32e-07 - 1.42e-06 - 1.42e-06 - 4.82e-06 -\n7.03e-08 3.73 9.53e-08 3.90 9.53e-08 3.90 3.30e-07 3.87\n4.65e-09 3.92 6.09e-09 3.97 6.09e-09 3.97 2.12e-08 3.96\n\nmean 3.82 mean 3.93 mean 3.93 mean 3.91\n--------------------------------------------------------------------------------\nlinf\nrho rho_v1 rho_v2 rho_e\nerror EOC error EOC error EOC error EOC\n9.58e-06 - 1.17e-05 - 1.17e-05 - 4.89e-05 -\n6.23e-07 3.94 7.48e-07 3.97 7.48e-07 3.97 3.22e-06 3.92\n4.05e-08 3.94 4.97e-08 3.91 4.97e-08 3.91 2.10e-07 3.94\n\nmean 3.94 mean 3.94 mean 3.94 mean 3.93\n--------------------------------------------------------------------------------","category":"page"},{"location":"#Showcase-of-advanced-features","page":"Home","title":"Showcase of advanced features","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"The presentation From Mesh Generation to Adaptive Simulation: A Journey in Julia, originally given as part of JuliaCon 2022, outlines how to use Trixi.jl for an adaptive simulation of the compressible Euler equations in two spatial dimensions on a complex domain. More details as well as code to run the simulation presented can be found at the reproducibility repository for the presentation.","category":"page"},{"location":"#Referencing","page":"Home","title":"Referencing","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"If you use Trixi.jl in your own research or write a paper using results obtained with the help of Trixi.jl, please cite the following articles:","category":"page"},{"location":"","page":"Home","title":"Home","text":"@article{ranocha2022adaptive,\n title={Adaptive numerical simulations with {T}rixi.jl:\n {A} case study of {J}ulia for scientific computing},\n author={Ranocha, Hendrik and Schlottke-Lakemper, Michael and Winters, Andrew Ross\n and Faulhaber, Erik and Chan, Jesse and Gassner, Gregor},\n journal={Proceedings of the JuliaCon Conferences},\n volume={1},\n number={1},\n pages={77},\n year={2022},\n doi={10.21105/jcon.00077},\n eprint={2108.06476},\n eprinttype={arXiv},\n eprintclass={cs.MS}\n}\n\n@article{schlottkelakemper2021purely,\n title={A purely hyperbolic discontinuous {G}alerkin approach for\n self-gravitating gas dynamics},\n author={Schlottke-Lakemper, Michael and Winters, Andrew R and\n Ranocha, Hendrik and Gassner, Gregor J},\n journal={Journal of Computational Physics},\n pages={110467},\n year={2021},\n month={06},\n volume={442},\n publisher={Elsevier},\n doi={10.1016/j.jcp.2021.110467},\n eprint={2008.10593},\n eprinttype={arXiv},\n eprintclass={math.NA}\n}","category":"page"},{"location":"","page":"Home","title":"Home","text":"In addition, you can also refer to Trixi.jl directly as","category":"page"},{"location":"","page":"Home","title":"Home","text":"@misc{schlottkelakemper2020trixi,\n title={{T}rixi.jl: {A}daptive high-order numerical simulations\n of hyperbolic {PDE}s in {J}ulia},\n author={Schlottke-Lakemper, Michael and Gassner, Gregor J and\n Ranocha, Hendrik and Winters, Andrew R and Chan, Jesse},\n year={2021},\n month={09},\n howpublished={\\url{https://github.com/trixi-framework/Trixi.jl}},\n doi={10.5281/zenodo.3996439}\n}","category":"page"},{"location":"#authors-index-md","page":"Home","title":"Authors","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl was initiated by Michael Schlottke-Lakemper (RWTH Aachen University/High-Performance Computing Center Stuttgart (HLRS), Germany) and Gregor Gassner (University of Cologne, Germany). Together with Hendrik Ranocha (Johannes Gutenberg University Mainz, Germany) and Andrew Winters (Linköping University, Sweden), and Jesse Chan (Rice University, US), they are the principal developers of Trixi.jl. The full list of contributors can be found under Authors.","category":"page"},{"location":"#License-and-contributing","page":"Home","title":"License and contributing","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is licensed under the MIT license (see License). Since Trixi.jl is an open-source project, we are very happy to accept contributions from the community. Please refer to Contributing for more details. Note that we strive to be a friendly, inclusive open-source community and ask all members of our community to adhere to our Code of Conduct. To get in touch with the developers, join us on Slack or create an issue.","category":"page"},{"location":"#Acknowledgments","page":"Home","title":"Acknowledgments","text":"","category":"section"},{"location":"","page":"Home","title":"Home","text":"
\n
\n
\n
\n
\n
\n
\n
","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through the following grants:","category":"page"},{"location":"","page":"Home","title":"Home","text":"Excellence Strategy EXC 2044-390685587, Mathematics Münster: Dynamics-Geometry-Structure.\nResearch unit FOR 5409 \"Structure-Preserving Numerical Methods for Bulk- and Interface Coupling of Heterogeneous Models (SNuBIC)\" (project number 463312734).\nIndividual grant no. 528753982.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the European Research Council through the ERC Starting Grant \"An Exascale aware and Un-crashable Space-Time-Adaptive Discontinuous Spectral Element Solver for Non-Linear Conservation Laws\" (Extreme), ERC grant agreement no. 714487.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from Vetenskapsrådet (VR, Swedish Research Council), Sweden through the VR Starting Grant \"Shallow water flows including sediment transport and morphodynamics\", VR grant agreement 2020-03642 VR.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the United States National Science Foundation (NSF) under awards DMS-1719818 and DMS-1943186.","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding from the German Federal Ministry of Education and Research (BMBF) through the project grant \"Adaptive earth system modeling with significantly reduced computation time for exascale supercomputers (ADAPTEX)\" (funding id: 16ME0668K).","category":"page"},{"location":"","page":"Home","title":"Home","text":"This project has benefited from funding by the Daimler und Benz Stiftung (Daimler and Benz Foundation) through grant no. 32-10/22.","category":"page"},{"location":"","page":"Home","title":"Home","text":"Trixi.jl is supported by NumFOCUS as an Affiliated Project.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"EditURL = \"../../literate/src/files/adding_nonconservative_equation.jl\"","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#adding_nonconservative_equation","page":"11 Adding a non-conservative equation","title":"11: Adding a non-conservative equation","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"If you want to use Trixi.jl for your own research, you might be interested in a new physics model that is not present in Trixi.jl. In this tutorial, we will implement the nonconservative linear advection equation in a periodic domain","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"left\nbeginalignedpartial_t u(tx) + a(x) partial_x u(tx) = 0 \nu(0x)=sin(x) \nu(t-pi)=u(tpi)\nendaligned\nright","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"where a(x) = 2 + cos(x). The analytic solution is","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"u(tx)=-sin left(2 tan ^-1left(sqrt3 tan left(fracsqrt3 t2-tan ^-1left(frac1sqrt3tan left(fracx2right)right)right)right)right)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"In Trixi.jl, such a mathematical model is encoded as a subtype of Trixi.AbstractEquations.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Basic-setup","page":"11 Adding a non-conservative equation","title":"Basic setup","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"Since there is no native support for variable coefficients, we need to transform the PDE to the following system:","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"left\nbeginalignedpartial_t beginpmatrixu(tx)a(tx) endpmatrix +beginpmatrix a(tx) partial_x u(tx) 0 endpmatrix = 0 \nu(0x)=sin(x) \na(0x)=2+cos(x) \nu(t-pi)=u(tpi)\nendaligned\nright","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"# Define new physics\nusing Trixi\nusing Trixi: AbstractEquations, get_node_vars\nimport Trixi: varnames, default_analysis_integrals, flux, max_abs_speed_naive,\n have_nonconservative_terms\n\n# Since there is no native support for variable coefficients, we use two\n# variables: one for the basic unknown `u` and another one for the coefficient `a`\nstruct NonconservativeLinearAdvectionEquation <: AbstractEquations{1 #= spatial dimension =#,\n 2 #= two variables (u,a) =#}\nend\n\nvarnames(::typeof(cons2cons), ::NonconservativeLinearAdvectionEquation) = (\"scalar\", \"advection_velocity\")\n\ndefault_analysis_integrals(::NonconservativeLinearAdvectionEquation) = ()\n\n\n# The conservative part of the flux is zero\nflux(u, orientation, equation::NonconservativeLinearAdvectionEquation) = zero(u)\n\n# Calculate maximum wave speed for local Lax-Friedrichs-type dissipation\nfunction max_abs_speed_naive(u_ll, u_rr, orientation::Integer, ::NonconservativeLinearAdvectionEquation)\n _, advection_velocity_ll = u_ll\n _, advection_velocity_rr = u_rr\n\n return max(abs(advection_velocity_ll), abs(advection_velocity_rr))\nend\n\n\n# We use nonconservative terms\nhave_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()\n\n# This \"nonconservative numerical flux\" implements the nonconservative terms.\n# In general, nonconservative terms can be written in the form\n# g(u) ∂ₓ h(u)\n# Thus, a discrete difference approximation of this nonconservative term needs\n# - `u mine`: the value of `u` at the current position (for g(u))\n# - `u_other`: the values of `u` in a neighborhood of the current position (for ∂ₓ h(u))\nfunction flux_nonconservative(u_mine, u_other, orientation,\n equations::NonconservativeLinearAdvectionEquation)\n _, advection_velocity = u_mine\n scalar, _ = u_other\n\n return SVector(advection_velocity * scalar, zero(scalar))\nend","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"The implementation of nonconservative terms uses a single \"nonconservative flux\" function flux_nonconservative. It will basically be applied in a loop of the form","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"du_m(D, u) = sum(D[m, l] * flux_nonconservative(u[m], u[l], 1, equations)) # orientation 1: x","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"where D is the derivative matrix and u contains the nodal solution values.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"Now, we can run a simple simulation using a DGSEM discretization.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"# Create a simulation setup\nusing Trixi\nusing OrdinaryDiffEq\n\nequation = NonconservativeLinearAdvectionEquation()\n\n# You can derive the exact solution for this setup using the method of\n# characteristics\nfunction initial_condition_sine(x, t, equation::NonconservativeLinearAdvectionEquation)\n x0 = -2 * atan(sqrt(3) * tan(sqrt(3) / 2 * t - atan(tan(x[1] / 2) / sqrt(3))))\n scalar = sin(x0)\n advection_velocity = 2 + cos(x[1])\n SVector(scalar, advection_velocity)\nend\n\n# Create a uniform mesh in 1D in the interval [-π, π] with periodic boundaries\nmesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=4, n_cells_max=10^4)\n\n# Create a DGSEM solver with polynomials of degree `polydeg`\n# Remember to pass a tuple of the form `(conservative_flux, nonconservative_flux)`\n# as `surface_flux` and `volume_flux` when working with nonconservative terms\nvolume_flux = (flux_central, flux_nonconservative)\nsurface_flux = (flux_lax_friedrichs, flux_nonconservative)\nsolver = DGSEM(polydeg=3, surface_flux=surface_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# Setup the spatial semidiscretization containing all ingredients\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create an ODE problem with given time span\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan)\n\n# Set up some standard callbacks summarizing the simulation setup and computing\n# errors of the numerical solution\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes\n# the passed callbacks\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false, callback=callbacks)\n\n# Print the timer summary\nsummary_callback()\n\n# Plot the numerical solution at the final time\nusing Plots: plot\nplot(sol)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"You see a plot of the final solution.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"We can check whether everything fits together by refining the grid and comparing the numerical errors. First, we look at the error using the grid resolution above.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"error_1 = analysis_callback(sol).l2 |> first","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"Next, we increase the grid resolution by one refinement level and run the simulation again.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"mesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=5, n_cells_max=10^4)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\n\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback);\n\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false, callback=callbacks);\nsummary_callback()\n\nerror_2 = analysis_callback(sol).l2 |> first","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"error_1 / error_2","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"As expected, the new error is roughly reduced by a factor of 16, corresponding to an experimental order of convergence of 4 (for polynomials of degree 3).","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Summary-of-the-code","page":"11 Adding a non-conservative equation","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"Here is the complete code that we used (without the callbacks since these create a lot of unnecessary output in the doctests of this tutorial). In addition, we create the struct inside the new module NonconservativeLinearAdvection. That ensures that we can re-create structs defined therein without having to restart Julia.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"Define new physics","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"module NonconservativeLinearAdvection\n\nusing Trixi\nusing Trixi: AbstractEquations, get_node_vars\nimport Trixi: varnames, default_analysis_integrals, flux, max_abs_speed_naive,\n have_nonconservative_terms\n\n# Since there is not yet native support for variable coefficients, we use two\n# variables: one for the basic unknown `u` and another one for the coefficient `a`\nstruct NonconservativeLinearAdvectionEquation <: AbstractEquations{1 #= spatial dimension =#,\n 2 #= two variables (u,a) =#}\nend\n\nvarnames(::typeof(cons2cons), ::NonconservativeLinearAdvectionEquation) = (\"scalar\", \"advection_velocity\")\n\ndefault_analysis_integrals(::NonconservativeLinearAdvectionEquation) = ()\n\n\n# The conservative part of the flux is zero\nflux(u, orientation, equation::NonconservativeLinearAdvectionEquation) = zero(u)\n\n# Calculate maximum wave speed for local Lax-Friedrichs-type dissipation\nfunction max_abs_speed_naive(u_ll, u_rr, orientation::Integer, ::NonconservativeLinearAdvectionEquation)\n _, advection_velocity_ll = u_ll\n _, advection_velocity_rr = u_rr\n\n return max(abs(advection_velocity_ll), abs(advection_velocity_rr))\nend\n\n\n# We use nonconservative terms\nhave_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()\n\n# This \"nonconservative numerical flux\" implements the nonconservative terms.\n# In general, nonconservative terms can be written in the form\n# g(u) ∂ₓ h(u)\n# Thus, a discrete difference approximation of this nonconservative term needs\n# - `u mine`: the value of `u` at the current position (for g(u))\n# - `u_other`: the values of `u` in a neighborhood of the current position (for ∂ₓ h(u))\nfunction flux_nonconservative(u_mine, u_other, orientation,\n equations::NonconservativeLinearAdvectionEquation)\n _, advection_velocity = u_mine\n scalar, _ = u_other\n\n return SVector(advection_velocity * scalar, zero(scalar))\nend\n\nend # module\n\n\n\n# Create a simulation setup\nimport .NonconservativeLinearAdvection\nusing Trixi\nusing OrdinaryDiffEq\n\nequation = NonconservativeLinearAdvection.NonconservativeLinearAdvectionEquation()\n\n# You can derive the exact solution for this setup using the method of\n# characteristics\nfunction initial_condition_sine(x, t, equation::NonconservativeLinearAdvection.NonconservativeLinearAdvectionEquation)\n x0 = -2 * atan(sqrt(3) * tan(sqrt(3) / 2 * t - atan(tan(x[1] / 2) / sqrt(3))))\n scalar = sin(x0)\n advection_velocity = 2 + cos(x[1])\n SVector(scalar, advection_velocity)\nend\n\n# Create a uniform mesh in 1D in the interval [-π, π] with periodic boundaries\nmesh = TreeMesh(-Float64(π), Float64(π), # min/max coordinates\n initial_refinement_level=4, n_cells_max=10^4)\n\n# Create a DGSEM solver with polynomials of degree `polydeg`\n# Remember to pass a tuple of the form `(conservative_flux, nonconservative_flux)`\n# as `surface_flux` and `volume_flux` when working with nonconservative terms\nvolume_flux = (flux_central, NonconservativeLinearAdvection.flux_nonconservative)\nsurface_flux = (flux_lax_friedrichs, NonconservativeLinearAdvection.flux_nonconservative)\nsolver = DGSEM(polydeg=3, surface_flux=surface_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# Setup the spatial semidiscretization containing all ingredients\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create an ODE problem with given time span\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\n\n# Set up some standard callbacks summarizing the simulation setup and computing\n# errors of the numerical solution\nsummary_callback = SummaryCallback()\nanalysis_callback = AnalysisCallback(semi, interval=50)\ncallbacks = CallbackSet(summary_callback, analysis_callback);\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes\n# the passed callbacks\nsol = solve(ode, Tsit5(), abstol=1.0e-6, reltol=1.0e-6,\n save_everystep=false);\n\n# Plot the numerical solution at the final time\nusing Plots: plot\nplot(sol);\nnothing #hide","category":"page"},{"location":"tutorials/adding_nonconservative_equation/#Package-versions","page":"11 Adding a non-conservative equation","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"","category":"page"},{"location":"tutorials/adding_nonconservative_equation/","page":"11 Adding a non-conservative equation","title":"11 Adding a non-conservative equation","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"EditURL = \"../../literate/src/files/differentiable_programming.jl\"","category":"page"},{"location":"tutorials/differentiable_programming/#differentiable_programming","page":"19 Differentiable programming","title":"19: Differentiable programming","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Julia and its ecosystem provide some tools for differentiable programming. Trixi.jl is designed to be flexible, extendable, and composable with Julia's growing ecosystem for scientific computing and machine learning. Thus, the ultimate goal is to have fast implementations that allow automatic differentiation (AD) without too much hassle for users. If some parts do not meet these requirements, please feel free to open an issue or propose a fix in a PR.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"In the following, we will walk through some examples demonstrating how to differentiate through Trixi.jl.","category":"page"},{"location":"tutorials/differentiable_programming/#Forward-mode-automatic-differentiation","page":"19 Differentiable programming","title":"Forward mode automatic differentiation","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Trixi.jl integrates well with ForwardDiff.jl for forward mode AD.","category":"page"},{"location":"tutorials/differentiable_programming/#Computing-the-Jacobian","page":"19 Differentiable programming","title":"Computing the Jacobian","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"The high-level interface to compute the Jacobian this way is jacobian_ad_forward. First, we load the required packages and compute the Jacobian of a semidiscretization of the compressible Euler equations, a system of nonlinear conservation laws.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, LinearAlgebra, Plots\n\nequations = CompressibleEulerEquations2D(1.4)\n\nsolver = DGSEM(3, flux_central)\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi);\nsize(J)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Next, we compute the eigenvalues of the Jacobian.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ = eigvals(J)\nscatter(real.(λ), imag.(λ), label=\"central flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"As you can see here, the maximal real part is close to zero.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Interestingly, if we add dissipation by switching to the flux_lax_friedrichs at the interfaces, the maximal real part of the eigenvalues increases.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"solver = DGSEM(3, flux_lax_friedrichs)\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\nλ = eigvals(J)\n\nscatter!(real.(λ), imag.(λ), label=\"Lax-Friedrichs flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Although the maximal real part is still somewhat small, it's larger than for the purely central discretization.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"However, we should be careful when using this analysis, since the eigenvectors are not necessarily well-conditioned.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ, V = eigen(J)\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"In one space dimension, the situation is a bit different.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"equations = CompressibleEulerEquations1D(1.4)\n\nsolver = DGSEM(3, flux_central)\nmesh = TreeMesh((-1.0,), (1.0,), initial_refinement_level=6, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\n\nλ = eigvals(J)\n\nscatter(real.(λ), imag.(λ), label=\"central flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Here, the maximal real part is basically zero to machine accuracy.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Moreover, the eigenvectors are not as ill-conditioned as in 2D.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ, V = eigen(J)\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"If we add dissipation, the maximal real part is still approximately zero.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"solver = DGSEM(3, flux_lax_friedrichs)\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ = jacobian_ad_forward(semi)\nλ = eigvals(J)\n\nscatter!(real.(λ), imag.(λ), label=\"Lax-Friedrichs flux\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"As you can see from the plot generated above, the maximal real part is still basically zero to machine precision.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"relative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Let's check the condition number of the eigenvectors.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ, V = eigen(J)\n\ncondition_number = cond(V)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Note that the condition number of the eigenvector matrix increases but is still smaller than for the example in 2D.","category":"page"},{"location":"tutorials/differentiable_programming/#Computing-other-derivatives","page":"19 Differentiable programming","title":"Computing other derivatives","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"It is also possible to compute derivatives of other dependencies using AD in Trixi.jl. For example, you can compute the gradient of an entropy-dissipative semidiscretization with respect to the ideal gas constant of the compressible Euler equations as described in the following. This example is also available as the elixir examples/special_elixirs/elixir_euler_ad.jl","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"First, we create a semidiscretization of the compressible Euler equations.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, LinearAlgebra, ForwardDiff\n\nequations = CompressibleEulerEquations2D(1.4)\n\n\"\"\"\n initial_condition_isentropic_vortex(x, t, equations::CompressibleEulerEquations2D)\n\nThe classical isentropic vortex test case of\n- Chi-Wang Shu (1997)\n Essentially Non-Oscillatory and Weighted Essentially Non-Oscillatory\n Schemes for Hyperbolic Conservation Laws\n [NASA/CR-97-206253](https://ntrs.nasa.gov/citations/19980007543)\n\"\"\"\nfunction initial_condition_isentropic_vortex(x, t, equations::CompressibleEulerEquations2D)\n inicenter = SVector(0.0, 0.0) # initial center of the vortex\n iniamplitude = 5.0 # size and strength of the vortex\n\n rho = 1.0 # base flow\n v1 = 1.0\n v2 = 1.0\n vel = SVector(v1, v2)\n p = 25.0\n\n rt = p / rho # ideal gas equation\n t_loc = 0.0\n\n cent = inicenter + vel*t_loc # shift advection of center to handle periodic BC, but only for v1 = v2 = 1.0\n cent = x - cent # distance to center point\n cent = SVector(-cent[2], cent[1])\n\n r2 = cent[1]^2 + cent[2]^2\n du = iniamplitude / (2*π) * exp(0.5 * (1 - r2)) # vel. perturbation\n dtemp = -(equations.gamma - 1) / (2 * equations.gamma * rt) * du^2 # isentropic\n\n rho = rho * (1 + dtemp)^(1 / (equations.gamma - 1))\n vel = vel + du * cent\n v1, v2 = vel\n p = p * (1 + dtemp)^(equations.gamma / (equations.gamma - 1))\n\n prim = SVector(rho, v1, v2, p)\n return prim2cons(prim, equations)\nend\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsolver = DGSEM(3, flux_lax_friedrichs, VolumeIntegralFluxDifferencing(flux_ranocha))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_isentropic_vortex, solver)\n\nu0_ode = Trixi.compute_coefficients(0.0, semi)\nsize(u0_ode)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Next, we compute the Jacobian using ForwardDiff.jacobian.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"J = ForwardDiff.jacobian((du_ode, γ) -> begin\n equations_inner = CompressibleEulerEquations2D(first(γ))\n semi_inner = Trixi.remake(semi, equations=equations_inner, uEltype=eltype(γ))\n Trixi.rhs!(du_ode, u0_ode, semi_inner, 0.0)\nend, similar(u0_ode), [1.4]); # γ needs to be an `AbstractArray`\n\nround.(extrema(J), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Note that we create a semidiscretization semi at first to determine the state u0_ode around which we want to perform the linearization. Next, we wrap the RHS evaluation inside a closure and pass that to ForwardDiff.jacobian. There, we need to make sure that the internal caches are able to store dual numbers from ForwardDiff.jl by setting uEltype appropriately. A similar approach is used by jacobian_ad_forward.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Note that the ideal gas constant does not influence the semidiscrete rate of change of the density, as demonstrated by","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"norm(J[1:4:end])","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Here, we used some knowledge about the internal memory layout of Trixi.jl, an array of structs with the conserved variables as fastest-varying index in memory.","category":"page"},{"location":"tutorials/differentiable_programming/#Differentiating-through-a-complete-simulation","page":"19 Differentiable programming","title":"Differentiating through a complete simulation","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"It is also possible to differentiate through a complete simulation. As an example, let's differentiate the total energy of a simulation using the linear scalar advection equation with respect to the wave number (frequency) of the initial data.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, OrdinaryDiffEq, ForwardDiff, Plots\n\nfunction energy_at_final_time(k) # k is the wave number of the initial condition\n equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\n solver = DGSEM(3, flux_lax_friedrichs)\n initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\n end\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k))\n ode = semidiscretize(semi, (0.0, 1.0))\n sol = solve(ode, BS3(), save_everystep=false)\n Trixi.integrate(energy_total, sol.u[end], semi)\nend\n\nk_values = range(0.9, 1.1, length=101)\n\nplot(k_values, energy_at_final_time.(k_values), label=\"Energy\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"You see a plot of a curve that resembles a parabola with local maximum around k = 1.0. Why's that? Well, the domain is fixed but the wave number changes. Thus, if the wave number is not chosen as an integer, the initial condition will not be a smooth periodic function in the given domain. Hence, the dissipative surface flux (flux_lax_friedrichs in this example) will introduce more dissipation. In particular, it will introduce more dissipation for \"less smooth\" initial data, corresponding to wave numbers k further away from integers.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"We can compute the discrete derivative of the energy at the final time with respect to the wave number k as follows.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"round(ForwardDiff.derivative(energy_at_final_time, 1.0), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"This is rather small and we can treat it as zero in comparison to the value of this derivative at other wave numbers k.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"dk_values = ForwardDiff.derivative.((energy_at_final_time,), k_values);\n\nplot(k_values, dk_values, label=\"Derivative\")","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"If you remember basic calculus, a sufficient condition for a local maximum is that the first derivative vanishes and the second derivative is negative. We can also check this discretely.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"second_derivative = round(ForwardDiff.derivative(\n k -> Trixi.ForwardDiff.derivative(energy_at_final_time, k), 1.0),\n sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Having seen this application, let's break down what happens step by step.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"function energy_at_final_time(k) # k is the wave number of the initial condition\n equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\n solver = DGSEM(3, flux_lax_friedrichs)\n initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\n end\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k))\n ode = semidiscretize(semi, (0.0, 1.0))\n sol = solve(ode, BS3(), save_everystep=false)\n Trixi.integrate(energy_total, sol.u[end], semi)\nend\n\nk = 1.0\nround(ForwardDiff.derivative(energy_at_final_time, k), sigdigits=2)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"When calling ForwardDiff.derivative(energy_at_final_time, k) with k=1.0, ForwardDiff.jl will basically use the chain rule and known derivatives of existing basic functions to calculate the derivative of the energy at the final time with respect to the wave number k at k0 = 1.0. To do this, ForwardDiff.jl uses dual numbers, which basically store the result and its derivative w.r.t. a specified parameter at the same time. Thus, we need to make sure that we can treat these ForwardDiff.Dual numbers everywhere during the computation. Fortunately, generic Julia code usually supports these operations. The most basic problem for a developer is to ensure that all types are generic enough, in particular the ones of internal caches.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"The first step in this example creates some basic ingredients of our simulation.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"equations = LinearScalarAdvectionEquation2D(1.0, -0.3)\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=3, n_cells_max=10^4)\nsolver = DGSEM(3, flux_lax_friedrichs);\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"These do not have internal caches storing intermediate values of the numerical solution, so we do not need to adapt them. In fact, we could also define them outside of energy_at_final_time (but would need to take care of globals or wrap everything in another function).","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Next, we define the initial condition","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"initial_condition = (x, t, equation) -> begin\n x_trans = Trixi.x_trans_periodic_2d(x - equation.advection_velocity * t)\n return SVector(sinpi(k * sum(x_trans)))\nend;\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"as a closure capturing the wave number k passed to energy_at_final_time. If you call energy_at_final_time(1.0), k will be a Float64. Thus, the return values of initial_condition will be SVectors of Float64s. When calculating the ForwardDiff.derivative, k will be a ForwardDiff.Dual number. Hence, the initial_condition will return SVectors of ForwardDiff.Dual numbers.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"The semidiscretization semi uses some internal caches to avoid repeated allocations and speed up the computations, e.g. for numerical fluxes at interfaces. Thus, we need to tell Trixi.jl to allow ForwardDiff.Dual numbers in these caches. That's what the keyword argument uEltype=typeof(k) in","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n uEltype=typeof(k));\nnothing #hide","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"does. This is basically the only part where you need to modify your standard Trixi.jl code to enable automatic differentiation. From there on, the remaining steps","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"ode = semidiscretize(semi, (0.0, 1.0))\nsol = solve(ode, BS3(), save_everystep=false)\nround(Trixi.integrate(energy_total, sol.u[end], semi), sigdigits=5)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"do not need any modifications since they are sufficiently generic (and enough effort has been spend to allow general types inside these calls).","category":"page"},{"location":"tutorials/differentiable_programming/#Propagating-errors-using-Measurements.jl","page":"19 Differentiable programming","title":"Propagating errors using Measurements.jl","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"(Image: Error bars by Randall Munroe) \"Error bars\" by Randall Munroe, linked from https://xkcd.com/2110","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Similar to AD, Trixi.jl also allows propagating uncertainties using linear error propagation theory via Measurements.jl. As an example, let's create a system representing the linear advection equation in 1D with an uncertain velocity. Then, we create a semidiscretization using a sine wave as initial condition, solve the ODE, and plot the resulting uncertainties in the primitive variables.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, OrdinaryDiffEq, Measurements, Plots, LaTeXStrings\n\nequations = LinearScalarAdvectionEquation1D(1.0 ± 0.1)\n\nmesh = TreeMesh((-1.0,), (1.0,), n_cells_max=10^5, initial_refinement_level=5)\n\nsolver = DGSEM(3)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test,\n solver, uEltype=Measurement{Float64})\n\node = semidiscretize(semi, (0.0, 1.5))\n\nsol = solve(ode, BS3(), save_everystep=false);\n\nplot(sol)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"You should see a plot where small error bars are shown around the extrema and larger error bars are shown in the remaining parts. This result is in accordance with expectations. Indeed, the uncertain propagation speed will affect the extrema less since the local variation of the solution is relatively small there. In contrast, the local variation of the solution is large around the turning points of the sine wave, so the uncertainties will be relatively large there.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"All this is possible due to allowing generic types and having good abstractions in Julia that allow packages to work together seamlessly.","category":"page"},{"location":"tutorials/differentiable_programming/#Finite-difference-approximations","page":"19 Differentiable programming","title":"Finite difference approximations","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Trixi.jl provides the convenience function jacobian_fd to approximate the Jacobian via central finite differences.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, LinearAlgebra\n\nequations = CompressibleEulerEquations2D(1.4)\n\nsolver = DGSEM(3, flux_central)\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_density_wave, solver)\n\nJ_fd = jacobian_fd(semi)\n\nJ_ad = jacobian_ad_forward(semi)\n\nrelative_difference = norm(J_fd - J_ad) / size(J_fd, 1)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"This discrepancy is of the expected order of magnitude for central finite difference approximations.","category":"page"},{"location":"tutorials/differentiable_programming/#Linear-systems","page":"19 Differentiable programming","title":"Linear systems","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"When a linear PDE is discretized using a linear scheme such as a standard DG method, the resulting semidiscretization yields an affine ODE of the form","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"partial_t u(t) = A u(t) + b","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"where A is a linear operator (\"matrix\") and b is a vector. Trixi.jl allows you to obtain this linear structure in a matrix-free way by using linear_structure. The resulting operator A can be used in multiplication, e.g. mul! from LinearAlgebra, converted to a sparse matrix using sparse from SparseArrays, or converted to a dense matrix using Matrix for detailed eigenvalue analyses. For example,","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using Trixi, LinearAlgebra, Plots\n\nequations = LinearScalarAdvectionEquation2D(1.0, -0.3)\n\nsolver = DGSEM(3, flux_lax_friedrichs)\n\nmesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), initial_refinement_level=2, n_cells_max=10^5)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test, solver)\n\nA, b = linear_structure(semi)\n\nsize(A), size(b)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"Next, we compute the eigenvalues of the linear operator.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ = eigvals(Matrix(A))\n\nscatter(real.(λ), imag.(λ))","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"As you can see here, the maximal real part is close to machine precision.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"λ = eigvals(Matrix(A))\nrelative_maximum = maximum(real, λ) / maximum(abs, λ)","category":"page"},{"location":"tutorials/differentiable_programming/#Package-versions","page":"19 Differentiable programming","title":"Package versions","text":"","category":"section"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\", \"ForwardDiff\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"","category":"page"},{"location":"tutorials/differentiable_programming/","page":"19 Differentiable programming","title":"19 Differentiable programming","text":"This page was generated using Literate.jl.","category":"page"},{"location":"meshes/tree_mesh/#Tree-mesh","page":"Tree mesh","title":"Tree mesh","text":"","category":"section"},{"location":"meshes/tree_mesh/","page":"Tree mesh","title":"Tree mesh","text":"The TreeMesh is a Cartesian, h-non-conforming mesh type used in many parts of Trixi.jl. Often, the support for this mesh type is developed best since it was the first mesh type in Trixi.jl, and it is available in one, two, and three space dimensions.","category":"page"},{"location":"meshes/tree_mesh/","page":"Tree mesh","title":"Tree mesh","text":"It is limited to hypercube domains but supports AMR via the AMRCallback. Due to its Cartesian nature, (numerical) fluxes need to implement methods dispatching on the orientation::Integer as described in the conventions.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Unstructured-quadrilateral-mesh","page":"Unstructured mesh","title":"Unstructured quadrilateral mesh","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The UnstructuredMesh2D is an unstructured, curvilinear, conforming mesh type in two space dimensions.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Next, we describe the conventions taken in the implementation for two-dimensional unstructured quadrilateral meshes. Principally, this relates to how a file with the extension .mesh encodes information about the numbering and orientation of elements in an unstructured quadrilateral mesh with possibly curved boundaries.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We use the following unstructured mesh with three elements for this discussion:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"(Image: example-mesh)","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Further, a simulation using Trixi.jl on this example unstructured mesh is provided in examples/unstructured_2d_dgsem/elixir_euler_basic.jl.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Mesh-file-header","page":"Unstructured mesh","title":"Mesh file header","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The first two lines of the mesh file lists the mesh file type as well as the total number of corners, surfaces, elements, and the polynomial degree that the mesh will use to represent any curved sides. For the example mesh these quantities are","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"ISM-V2\n 7 9 3 8","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"corresponding to seven corners, nine surfaces, and three elements. The mesh polynomial degree of eight is taken only for illustrative purposes. In practice, this mesh polynomial degree depends on the particular application for which the curved, unstructured mesh is required.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-corner-nodes","page":"Unstructured mesh","title":"List of corner nodes","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"After these global counts that prescribe information about the mesh skeleton, the mesh file give a list of the physical (x,y) coordinates of all the corners. The corner nodes are listed in the order prescribed by mesh generator. Thus, for the example mesh this node list would be","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 1.0 -1.0\n 3.0 0.0\n 1.0 1.0\n 2.0 0.0\n 0.0 0.0\n 3.0 1.0\n 3.0 -1.0","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The corner nodes are internally referenced by their position in the list above. For example, here the node at (1.0, -1.0) would have node id 1, node id 2 would be at (3.0, 0.0) etc.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-neighbor-connectivity","page":"Unstructured mesh","title":"List of neighbor connectivity","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"After the corner list comes the neighbor connectivity along each surface in the mesh. This includes local indexing and orientation information necessary to compute the coupling between elements in the mesh. In 2D each surface is defined by connecting two nodes indexed as with the numbering above. We adopt the convention that node id 1 < node id 2.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Each surface will have two neighbors where the element on the left locally as one \"walks\" from node id 1 to node id 2 is taken to be the primary element and the element locally on the right is taken to be the secondary element. If, however, there is no secondary element index, then the surface lies along a physical boundary. In this case the only available element index is considered to be primary and the secondary index is set to zero.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The final two index numbers within the neighbor information list are used to identify the local surface within each element. The first local surface index (on the primary element) will always be positive whereas the second local surface index (on the primary element) can be positive or negative. If the second local surface index is positive, then the local coordinate systems in the primary element and secondary element match, i.e., the indexing on either side runs from 1:polydeg+1. However, if the local surface index of the secondary element is negative in the mesh file, then the coordinate system in the secondary element is flipped with respect to the primary element. Therefore, care must be taken in the implementation to ensure that the primary element indexing runs from 1:polydeg+1 whereas the secondary element indexing must run in reverse from polydeg+1:-1:1. Finally, if the secondary element index is zero, then so will be the local surface index because the surface is on a physical boundary. Also, there is no flipping of coordinate indexing required at the physical boundary because only the primary element's coordinate system exists.","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Three-examples:-One-along-a-physical-boundary-and-two-along-interior-surfaces.","page":"Unstructured mesh","title":"Three examples: One along a physical boundary and two along interior surfaces.","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {8} we connect node (2) to node (7) and are along a physical boundary in element 3 with the local surface index 1 and the neighbor information:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 7 3 0 1 0","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {1} we connect node (2) to node (4) such that the primary element is 3 with local surface index 2 and the secondary element is 2 with local surface index 1. Furthermore, we see that coordinate system in the secondary element 2 is flipped with respect to the primary element's coordinate system such that the sign of the local surface index in the secondary element flips. This gives the following neighbor information:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 4 3 2 2 -1","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Along edge {4} we connect node (1) to node (4) such that the primary element is 1 with local surface index 2 and the secondary element is 3 with local surface index 3. The coordinate systems in both elements match and no sign change is required on the local surface index in the secondary element:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 1 4 1 3 2 3","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We collect the complete neighbor information for the example mesh above:","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 2 4 3 2 2 -1\n 3 5 1 0 4 0\n 1 5 1 0 1 0\n 1 4 1 3 2 3\n 2 6 2 0 2 0\n 1 7 3 0 4 0\n 3 6 2 0 3 0\n 2 7 3 0 1 0\n 3 4 2 1 4 -3","category":"page"},{"location":"meshes/unstructured_quad_mesh/#List-of-elements","page":"Unstructured mesh","title":"List of elements","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. We connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion thus making the element right-handed indicated using the circular arrow in the figure above. In turn, this right-handedness defines the local surface indexing (i.e. the four local sides) and the local (xi eta) coordinate system. For example, the four corners for element 1 would be listed as","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The mesh file also encodes information for curved surfaces either interior to the domain (as surface {9} above) or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (1, 2, 3, or 4) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y) data will be given in the direction of the local coordinate system.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"The last piece of information provided by the mesh file are labels for the different surfaces of an element. These labels are useful to set boundary conditions along physical surfaces. The labels can be short descriptive words. The label --- indicates an internal surface where no boundary condition is required.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"As an example, the complete information for element 1 in the example mesh would be","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3\n 0 0 1 1\n 1.000000000000000 1.000000000000000\n 1.024948365654583 0.934461926834452\n 1.116583018200151 0.777350964621867\n 1.295753434047077 0.606254343587194\n 1.537500000000000 0.462500000000000\n 1.768263070247418 0.329729152118310\n 1.920916981799849 0.185149035378133\n 1.986035130050921 0.054554577460044\n 2.000000000000000 0.0\n 0.0 0.0\n 0.035513826946206 0.105291711848750\n 0.148591270347399 0.317731556850611\n 0.340010713990041 0.452219430075470\n 0.575000000000000 0.462500000000000\n 0.788022294598950 0.483764065630034\n 0.926408729652601 0.644768443149389\n 0.986453164464803 0.883724792445746\n 1.000000000000000 1.000000000000000\n Slant --- --- Bezier","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"where the curved boundary information is encoded \"back to back\". That is, the first nine (x,y) nodes in the list above correspond to the interior boundary curve along local side 3 in element 1 and the next nine (x,y) nodes denote the curved physical boundary named Bezier along local side 4.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We collect the complete set of element information for the example mesh","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" 5 1 4 3\n 0 0 1 1\n 1.000000000000000 1.000000000000000\n 1.024948365654583 0.934461926834452\n 1.116583018200151 0.777350964621867\n 1.295753434047077 0.606254343587194\n 1.537500000000000 0.462500000000000\n 1.768263070247418 0.329729152118310\n 1.920916981799849 0.185149035378133\n 1.986035130050921 0.054554577460044\n 2.000000000000000 0.0\n 0.0 0.0\n 0.035513826946206 0.105291711848750\n 0.148591270347399 0.317731556850611\n 0.340010713990041 0.452219430075470\n 0.575000000000000 0.462500000000000\n 0.788022294598950 0.483764065630034\n 0.926408729652601 0.644768443149389\n 0.986453164464803 0.883724792445746\n 1.000000000000000 1.000000000000000\n Slant --- --- Bezier\n 4 2 6 3\n 0 0 0 1\n 2.000000000000000 0.0\n 1.986035130050921 0.054554577460044\n 1.920916981799849 0.185149035378133\n 1.768263070247418 0.329729152118310\n 1.537500000000000 0.462500000000000\n 1.295753434047077 0.606254343587194\n 1.116583018200151 0.777350964621867\n 1.024948365654583 0.934461926834452\n 1.000000000000000 1.000000000000000\n --- Right Top ---\n 7 2 4 1\n 0 0 0 0\n Right --- --- Bottom","category":"page"},{"location":"meshes/unstructured_quad_mesh/#Trixi.jl-on-an-unstructured-quadrilateral-mesh","page":"Unstructured mesh","title":"Trixi.jl on an unstructured quadrilateral mesh","text":"","category":"section"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"We provide an example simulation on an unstructured quadrilateral mesh by executing","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"julia> trixi_include(default_example_unstructured())","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":"Note this may download a copy of the mesh file described above for the three element unstructured mesh. This elixir provides the solution for the compressible Euler equations in two spatial dimensions for a smooth propagating wave solution. Below we provide the time evolution of the pressure wave for this example created with the Trixi2Vtk tool and visualized using ParaView.","category":"page"},{"location":"meshes/unstructured_quad_mesh/","page":"Unstructured mesh","title":"Unstructured mesh","text":" \n
","category":"page"},{"location":"reference-trixibase/#TrixiBase.jl-API","page":"TrixiBase.jl","title":"TrixiBase.jl API","text":"","category":"section"},{"location":"reference-trixibase/","page":"TrixiBase.jl","title":"TrixiBase.jl","text":"CurrentModule = TrixiBase","category":"page"},{"location":"reference-trixibase/","page":"TrixiBase.jl","title":"TrixiBase.jl","text":"Modules = [TrixiBase]","category":"page"},{"location":"reference-trixibase/#TrixiBase.trixi_include-Tuple{Module, AbstractString}","page":"TrixiBase.jl","title":"TrixiBase.trixi_include","text":"trixi_include([mod::Module=Main,] elixir::AbstractString; kwargs...)\n\ninclude the file elixir and evaluate its content in the global scope of module mod. You can override specific assignments in elixir by supplying keyword arguments. Its basic purpose is to make it easier to modify some parameters while running simulations from the REPL. Additionally, this is used in tests to reduce the computational burden for CI while still providing examples with sensible default values for users.\n\nBefore replacing assignments in elixir, the keyword argument maxiters is inserted into calls to solve with it's default value used in the SciML ecosystem for ODEs, see the \"Miscellaneous\" section of the documentation.\n\nExamples\n\njulia> using TrixiBase, Trixi\n\njulia> redirect_stdout(devnull) do\n trixi_include(@__MODULE__, joinpath(examples_dir(), \"tree_1d_dgsem\", \"elixir_advection_extended.jl\"),\n tspan=(0.0, 0.1))\n sol.t[end]\n end\n[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.\n0.1\n\n\n\n\n\n","category":"method"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"EditURL = \"../../literate/src/files/time_stepping.jl\"","category":"page"},{"location":"tutorials/time_stepping/#time_stepping","page":"18 Explicit time stepping","title":"18: Explicit time stepping","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"For the time integration, Trixi.jl uses the package OrdinaryDiffEq.jl from the SciML ecosystem. The interface to this package is the solve(...) function. It always requires an ODE problem and a time integration algorithm as input parameters.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"solve(ode, alg; kwargs...)","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"In Trixi.jl, the ODE problem is created by semidiscretize(semi, tspan) for a semidiscretization semi and the time span tspan. In particular, semidiscretize returns an ODEProblem used by OrdinaryDiffEq.jl.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"OrdinaryDiffEq.jl provides many integration algorithms, which are summarized in the documentation. Particularly interesting for Trixi.jl are their strong stability preserving (SSP) methods and low-storage methods. There are some differences regarding the choice of the used time step.","category":"page"},{"location":"tutorials/time_stepping/#adaptive_step_sizes","page":"18 Explicit time stepping","title":"Error-based adaptive step sizes","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"First, we treat time integration algorithms with adaptive step sizes, such as SSPRK43. It is used in some elixirs, like elixir_euler_colliding_flow.jl or elixir_euler_astro_jet_amr.jl.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"Other error-based adaptive integration algorithms are for instance RDPK3SpFSAL35, RDPK3Sp35, RDPK3SpFSAL49, RDPK3Sp49, RDPK3SpFSAL510, RDPK3Sp510.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"They already contain an error-based adaptive step size control and heuristics to guess a starting step size. If this heuristic fails in your case, you can specify an appropriately small initial step size as keyword argument dt=... of solve.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"If you run Trixi in parallel with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to enable MPI aware error-based adaptive step size control. These keyword arguments are also included in ode_default_options.","category":"page"},{"location":"tutorials/time_stepping/#CFL-based-step-size-control","page":"18 Explicit time stepping","title":"CFL-based step size control","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"The SciML ecosystem also provides time integration algorithms without adaptive time stepping on their own, such as CarpenterKennedy2N54. Moreover, you also can deactivate the automatic adaptivity of adaptive integration algorithms by passing adaptive=false in the solve function.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"These algorithms require another way of setting the step size. You have to pass dt=... in the solve function. Without other settings, the simulation uses this fixed time step.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"For hyperbolic PDEs, it is natural to use an adaptive CFL-based step size control. Here, the time step is proportional to a ratio of the local measure of mesh spacing Delta x_i for an element i and the maximum (local) wave speed lambda_max related to the largest-magnitude eigenvalue of the flux Jacobian of the hyperbolic system.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"Delta t_n = textCFL * min_i fracDelta x_ilambda_max(u_i^n)","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"We compute Delta x_i by scaling the element size by a factor of 1(N+1), cf. Gassner and Kopriva (2011), Section 5.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"Trixi.jl provides such a CFL-based step size control. It is implemented as the callback StepsizeCallback.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"stepsize_callback = StepsizeCallback(; cfl=1.0)","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"A suitable CFL number depends on many parameters such as the chosen grid, the integration algorithm and the polynomial degree of the spatial DG discretization. So, the optimal number for an example is mostly determined experimentally.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"You can add this CFL-based step size control to your simulation like any other callback.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"callbacks = CallbackSet(stepsize_callback)\nalg = CarpenterKennedy2N54(williamson_condition=false)\nsolve(ode, alg;\n dt=1.0 # solve needs some value here but it will be overwritten by the stepsize_callback\n callback=callbacks)","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"You can find simple examples with a CFL-based step size control for instance in the elixirs elixir_advection_basic.jl or elixir_euler_source_terms.jl.","category":"page"},{"location":"tutorials/time_stepping/#Package-versions","page":"18 Explicit time stepping","title":"Package versions","text":"","category":"section"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"","category":"page"},{"location":"tutorials/time_stepping/","page":"18 Explicit time stepping","title":"18 Explicit time stepping","text":"This page was generated using Literate.jl.","category":"page"},{"location":"contributing/#Contributing","page":"Contributing","title":"Contributing","text":"","category":"section"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Trixi.jl is an open-source project and we are very happy to accept contributions from the community. Please feel free to open issues or submit patches (preferably as merge requests) any time. For planned larger contributions, it is often beneficial to get in contact with one of the principal developers first (see Authors).","category":"page"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Trixi.jl and its contributions are licensed under the MIT license (see License). As a contributor, you certify that all your contributions are in conformance with the Developer Certificate of Origin (Version 1.1), which is reproduced below.","category":"page"},{"location":"contributing/#Developer-Certificate-of-Origin-(Version-1.1)","page":"Contributing","title":"Developer Certificate of Origin (Version 1.1)","text":"","category":"section"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"The following text was taken from https://developercertificate.org:","category":"page"},{"location":"contributing/","page":"Contributing","title":"Contributing","text":"Developer Certificate of Origin\nVersion 1.1\n\nCopyright (C) 2004, 2006 The Linux Foundation and its contributors.\n1 Letterman Drive\nSuite D4700\nSan Francisco, CA, 94129\n\nEveryone is permitted to copy and distribute verbatim copies of this\nlicense document, but changing it is not allowed.\n\n\nDeveloper's Certificate of Origin 1.1\n\nBy making a contribution to this project, I certify that:\n\n(a) The contribution was created in whole or in part by me and I\n have the right to submit it under the open source license\n indicated in the file; or\n\n(b) The contribution is based upon previous work that, to the best\n of my knowledge, is covered under an appropriate open source\n license and I have the right under that license to submit that\n work with modifications, whether created in whole or in part\n by me, under the same open source license (unless I am\n permitted to submit under a different license), as indicated\n in the file; or\n\n(c) The contribution was provided directly to me by some other\n person who certified (a), (b) or (c) and I have not modified\n it.\n\n(d) I understand and agree that this project and the contribution\n are public and that a record of the contribution (including all\n personal information I submit with it, including my sign-off) is\n maintained indefinitely and may be redistributed consistent with\n this project or the open source license(s) involved.","category":"page"},{"location":"performance/#Performance","page":"Performance","title":"Performance","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"Trixi.jl is designed to balance performance and readability. Since Julia provides a lot of zero-cost abstractions, it is often possible to optimize both goals simultaneously.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The usual development workflow in Julia is","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Make it work.\nMake it nice.\nMake it fast.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"To achieve the third step, you should be familiar with (at least) the section on performance tips in the Julia manual. Here, we just list some important aspects you should consider when developing Trixi.jl.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Consider using @views/view(...) when using array slices, except on the left-side of an assignment (further details).\nFunctions are essentially for free, since they are usually automatically inlined where it makes sense (using @inline can be used as an additional hint to the compiler) (further details).\nFunction barriers can improve performance due to type stability (further details).\nLook for type instabilities using @code_warntype. Consider using @descend from Cthulhu.jl to investigate deeper call chains.","category":"page"},{"location":"performance/#Manual-benchmarking","page":"Performance","title":"Manual benchmarking","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"If you modify some internal parts of Trixi.jl, you should check the impact on performance. Hence, you should at least investigate the performance roughly by comparing the reported timings of several elixirs. Deeper investigations and micro-benchmarks should usually use BenchmarkTools.jl. For example, the following steps were used to benchmark the changes introduced in PR #256.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"git checkout e7ebf3846b3fd62ee1d0042e130afb50d7fe8e48 (new version)\nStart julia --threads=1 --check-bounds=no.\nExecute the following code in the REPL to benchmark the rhs! call at the final state.\njulia> using BenchmarkTools, Revise; using Trixi\n\njulia> trixi_include(\"examples/2d/elixir_euler_sedov_blast_wave.jl\")\n\njulia> du_test = copy(sol.u[end]); u_test = copy(sol.u[end]);\n\njulia> @benchmark Trixi.rhs!(\n $(du_test),\n $(u_test),\n $(semi),\n $(sol.t[end]))\nBenchmarkTools.Trial:\n memory estimate: 10.48 KiB\n allocs estimate: 67\n --------------\n minimum time: 4.510 ms (0.00% GC)\n median time: 4.646 ms (0.00% GC)\n mean time: 4.699 ms (0.00% GC)\n maximum time: 7.183 ms (0.00% GC)\n --------------\n samples: 1065\n evals/sample: 1\n\nshell> git checkout 222241ff54f8a4ca9876cc1fc25ae262416a4ea0\n\njulia> trixi_include(\"examples/2d/elixir_euler_sedov_blast_wave.jl\")\n\njulia> @benchmark Trixi.rhs!(\n $(du_test),\n $(u_test),\n $(semi),\n $(sol.t[end]))\nBenchmarkTools.Trial:\n memory estimate: 10.36 KiB\n allocs estimate: 67\n --------------\n minimum time: 4.500 ms (0.00% GC)\n median time: 4.635 ms (0.00% GC)\n mean time: 4.676 ms (0.00% GC)\n maximum time: 5.880 ms (0.00% GC)\n --------------\n samples: 1070\n evals/sample: 1\nRun the @benchmark ... commands multiple times to see whether there are any significant fluctuations.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Follow these steps for both commits you want to compare. The relevant benchmark results you should typically be looking at are the median and mean values of the runtime and the memory/allocs estimate. In this example, the differences of the runtimes are of the order of the fluctuations one gets when running the benchmarks multiple times. Since the memory/allocs are (roughly) the same, there doesn't seem to be a significant performance regression here.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"You can also make it more detailed by benchmarking only, e.g., the calculation of the volume terms, but whether that's necessary depends on the modifications you made and their (potential) impact.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Some more detailed description of manual profiling and benchmarking as well as resulting performance improvements of Trixi.jl are given in the following blog posts.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Improving performance of AMR with p4est, cf. #638\nImproving performance of EC methods, cf. #643","category":"page"},{"location":"performance/#Automated-benchmarking","page":"Performance","title":"Automated benchmarking","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"We use PkgBenchmark.jl to provide a standard set of benchmarks for Trixi.jl. The relevant benchmark script is benchmark/benchmarks.jl. To benchmark the changes made in a PR, please proceed as follows:","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Check out the latest main branch of your Trixi.jl development repository.\nCheck out the latest development branch of your PR.\nChange your working directory to the benchmark directory of Trixi.jl.\nExecute julia run_benchmarks.jl.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"This will take some hours to complete and requires at least 8 GiB of RAM. When everything is finished, some output files will be created in the benchmark directory of Trixi.jl.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"warning: Warning\nPlease note that the benchmark scripts use --check-bounds=no at the moment. Thus, they will not work in any useful way for Julia v1.10 (and newer?), see Julia issue #50985.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"You can also run a standard set of benchmarks manually via","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> using PkgBenchmark, Trixi\n\njulia> results = benchmarkpkg(Trixi, BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`))\n\njulia> export_markdown(pkgdir(Trixi, \"benchmark\", \"single_benchmark.md\"), results)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"This will save a markdown file with a summary of the benchmark results similar to this example. Note that this will take quite some time. Additional options are described in the docs of PkgBenchmark.jl. A particularly useful option is to specify a BenchmarkConfig including Julia command line options affecting the performance such as disabling bounds-checking and setting the number of threads.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"A useful feature when developing Trixi.jl is to compare the performance of Trixi.jl's current state vs. the main branch. This can be achieved by executing","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> using PkgBenchmark, Trixi\n\njulia> results = judge(Trixi,\n BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`), # target\n BenchmarkConfig(juliacmd=`$(Base.julia_cmd()) --check-bounds=no --threads=1`, id=\"main\") # baseline\n )\n\njulia> export_markdown(pkgdir(Trixi, \"benchmark\", \"results.md\"), results)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"By default, the target is the current state of the repository. Remember that you need to be in a clean state (commit or stash your changes) to run this successfully. You can also run this comparison and an additional one using two threads via","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"julia> include(\"benchmark/run_benchmarks.jl\")","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Then, markdown files including the results are saved in benchmark/. This example result was obtained using a GitHub action for the PR #535. Note that GitHub actions run on in the cloud in a virtual machine. Hence, we do not really have control over it and performance results must be taken with a grain of salt. Nevertheless, significant runtime differences and differences of memory allocations should be robust indicators of performance changes.","category":"page"},{"location":"performance/#Runtime-performance-vs.-latency-aka-using-@nospecialize-selectively","page":"Performance","title":"Runtime performance vs. latency aka using @nospecialize selectively","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"Usually, Julia will compile specialized versions of each method, using as much information from the types of function arguments as possible (based on some heuristics). The compiler will generate code that is as efficient as comparable code written in a low-level language such as C or Fortran. However, there are cases where the runtime performance does not really matter but the time needed to compile specializations becomes significant. This is related to latency or the time-to-first-plot problem, well-known in the Julia community. In such a case, it can be useful to remove some burden from the compiler by avoiding specialization on every possible argument types using the macro @nospecialize. A prime example of such a case is pretty printing of structs in the Julia REPL, see the associated PR for further discussions.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"As a rule of thumb:","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Do not use @nospecialize in performance-critical parts, in particular not for methods involved in computing Trixi.rhs!.\nConsider using @nospecialize for methods like custom implementations of Base.show.","category":"page"},{"location":"performance/#performance-metrics","page":"Performance","title":"Performance metrics of the AnalysisCallback","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The AnalysisCallback computes two performance indicators that you can use to evaluate the serial and parallel performance of Trixi.jl. They represent measured run times that are normalized by the number of rhs! evaluations and the number of degrees of freedom of the problem setup. The normalization ensures that we can compare different measurements for each type of indicator independent of the number of time steps or mesh size. All indicators have in common that they are still in units of time, thus lower is better for each of them.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"Here, the term \"degrees of freedom\" (DOFs) refers to the number of independent state vectors that are used to represent the numerical solution. For example, if you use a DGSEM-type scheme in 2D on a mesh with 8 elements and with 5-by-5 Gauss-Lobatto nodes in each element (i.e., a polynomial degree of 4), the total number of DOFs would be","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"n_textDOFsDGSEM = textnumber of elements cdot textnumber of nodes per element = 8 cdot (5 cdot 5) = 200","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"In contrast, for a finite volume-type scheme on a mesh with 8 elements, the total number of DOFs would be (independent of the number of spatial dimensions)","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"n_textDOFsFV = textnumber of elements = 8","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"since for standard finite volume methods you store a single state vector in each element. Note that we specifically count the number of state vectors and not the number of state variables for the DOFs. That is, in the previous example n_textDOFsFV is equal to 8 independent of whether this is a compressible Euler setup with 5 state variables or a linear scalar advection setup with one state variable.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"For each indicator, the measurements are always since the last invocation of the AnalysisCallback. That is, if the analysis callback is called multiple times, the indicators are repeatedly computed and can thus also be used to track the performance over the course of a longer simulation, e.g., to analyze setups with varying performance characteristics. Note that the time spent in the AnalysisCallback itself is always excluded, i.e., the performance measurements are not distorted by potentially expensive solution analysis computations. All other parts of a Trixi.jl simulation are included, however, thus make sure that you disable everything you do not want to be measured (such as I/O callbacks, visualization etc.).","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"note: Performance indicators and adaptive mesh refinement\nCurrently it is not possible to compute meaningful performance indicators for a simulation with arbitrary adaptive mesh refinement, since this would require to explicitly keep track of the number of DOF updates due to the mesh size changing repeatedly. The only way to do this at the moment is by setting the analysis interval to the same value as the AMR interval.","category":"page"},{"location":"performance/#Local,-rhs!-only-indicator","page":"Performance","title":"Local, rhs!-only indicator","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The local, rhs!-only indicator is computed as","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"texttimeDOFrhs = fract_texttextttrhsn_textDOFslocal cdot n_textcallstextttrhs","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"where t_texttextttrhs is the accumulated time spent in rhs!, n_textDOFslocal is the local number of DOFs (i.e., on the current MPI rank; if doing a serial run, you can just think of this as the number of DOFs), and n_textcallstextttrhs is the number of times the rhs! function has been evaluated. Note that for this indicator, we measure only the time spent in rhs!, i.e., by definition all computations outside of rhs! - specifically all other callbacks and the time integration method - are not taken into account.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The local, rhs!-only indicator is usually most useful if you do serial measurements and are interested in the performance of the implementation of your core numerical methods (e.g., when doing performance tuning).","category":"page"},{"location":"performance/#Performance-index-(PID)","page":"Performance","title":"Performance index (PID)","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"The performance index (PID) is computed as","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"textPID = fract_textwall cdot n_textranksMPIn_textDOFsglobal cdot n_textcallstextttrhs","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"where t_textwall is the walltime since the last call to the AnalysisCallback, n_textranksMPI is the number of MPI ranks used, n_textDOFsglobal is the global number of DOFs (i.e., the sum of DOFs over all MPI ranks; if doing a serial run, you can just think of this as the number of DOFs), and n_textcallstextttrhs is the number of times the rhs! function has been evaluated since the last call to the AnalysisCallback. The PID measures everything except the time spent in the AnalysisCallback itself - specifically, all other callbacks and the time integration method itself are included.","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"The PID is usually most useful if you would like to compare the parallel performance of your code to its serial performance. Specifically, it allows you to evaluate the parallelization overhead of your code by giving you a measure of the resources that are necessary to solve a given simulation setup. In a sense, it mimics the \"core hours\" metric that is often used by supercomputer centers to measure how many resources a particular compute job requires. It can thus be seen as a proxy for \"energy used\" and, as an extension, \"monetary cost\".","category":"page"},{"location":"performance/","page":"Performance","title":"Performance","text":"note: Initialization overhead in measurements\nWhen using one of the integration schemes from OrdinaryDiffEq.jl, their implementation will initialize some OrdinaryDiffEq.jl-specific information during the first time step. Among other things, one additional call to rhs! is performed. Therefore, make sure that for performance measurements using the PID either the number of timesteps or the workload per rhs! call is large enough to make the initialization overhead negligible. Note that the extra call to rhs! is properly accounted for in both the number of calls and the measured time, so you do not need to worry about it being expensive. If you want a perfect timing result, you need to set the analysis interval such that the AnalysisCallback is invoked at least once during the course of the simulation and discard the first PID value.","category":"page"},{"location":"performance/#Performance-issues-with-multi-threaded-reductions","page":"Performance","title":"Performance issues with multi-threaded reductions","text":"","category":"section"},{"location":"performance/","page":"Performance","title":"Performance","text":"False sharing is a known performance issue for systems with distributed caches. It also occurred for the implementation of a thread parallel bounds checking routine for the subcell IDP limiting in PR #1736. After some testing and discussion, it turned out that initializing a vector of length n * Threads.nthreads() and only using every n-th entry instead of a vector of length Threads.nthreads() fixes the problem. Since there are no processors with caches over 128B, we use n = 128B / size(uEltype). Now, the bounds checking routine of the IDP limiting scales as hoped.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"EditURL = \"../../literate/src/files/custom_semidiscretization.jl\"","category":"page"},{"location":"tutorials/custom_semidiscretization/#custom_semidiscretization","page":"20 Custom semidiscretizations","title":"20: Custom semidiscretizations","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"As described in the overview section, semidiscretizations are high-level descriptions of spatial discretizations in Trixi.jl. Trixi.jl's main focus is on hyperbolic conservation laws represented in a SemidiscretizationHyperbolic. Hyperbolic-parabolic problems based on the advection-diffusion equation or the compressible Navier-Stokes equations can be represented in a SemidiscretizationHyperbolicParabolic. This is described in the basic tutorial on parabolic terms and its extension to custom parabolic terms. In this tutorial, we will describe how these semidiscretizations work and how they can be used to create custom semidiscretizations involving also other tasks.","category":"page"},{"location":"tutorials/custom_semidiscretization/#Overview-of-the-right-hand-side-evaluation","page":"20 Custom semidiscretizations","title":"Overview of the right-hand side evaluation","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"The semidiscretizations provided by Trixi.jl are set up to create ODEProblems from the SciML ecosystem for ordinary differential equations. In particular, a spatial semidiscretization can be wrapped in an ODE problem using semidiscretize, which returns an ODEProblem. This ODEProblem bundles an initial condition, a right-hand side (RHS) function, the time span, and possible parameters. The ODEProblems created by Trixi.jl use the semidiscretization passed to semidiscretize as a parameter. For a SemidiscretizationHyperbolic, the ODEProblem wraps Trixi.rhs! as ODE RHS. For a SemidiscretizationHyperbolicParabolic, Trixi.jl uses a SplitODEProblem combining Trixi.rhs_parabolic! for the (potentially) stiff part and Trixi.rhs! for the other part.","category":"page"},{"location":"tutorials/custom_semidiscretization/#Standard-Trixi.jl-setup","page":"20 Custom semidiscretizations","title":"Standard Trixi.jl setup","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"In this tutorial, we will consider the linear advection equation with source term","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"partial_t u(tx) + partial_x u(tx) = -exp(-t) sinbigl(pi (x - t) bigr)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"with periodic boundary conditions in the domain [-1, 1] as a model problem. The initial condition is","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"u(0x) = sin(pi x)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"The source term results in some damping and the analytical solution","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"u(tx) = exp(-t) sinbigl(pi (x - t) bigr)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"First, we discretize this equation using the standard functionality of Trixi.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"using Trixi, OrdinaryDiffEq, Plots","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"The linear scalar advection equation is already implemented in Trixi.jl as LinearScalarAdvectionEquation1D. We construct it with an advection velocity 1.0.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"equations = LinearScalarAdvectionEquation1D(1.0)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Next, we use a standard DGSEM solver.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"solver = DGSEM(polydeg = 3)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"We create a simple TreeMesh in 1D.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"coordinates_min = (-1.0,)\ncoordinates_max = (+1.0,)\nmesh = TreeMesh(coordinates_min, coordinates_max;\n initial_refinement_level = 4,\n n_cells_max = 10^4,\n periodicity = true)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"We wrap everything in in a semidiscretization and pass the source terms as a standard Julia function. Please note that Trixi.jl uses SVectors from StaticArrays.jl to store the conserved variables u. Thus, the return value of the source terms must be wrapped in an SVector - even if we consider just a scalar problem.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"function initial_condition(x, t, equations)\n return SVector(exp(-t) * sinpi(x[1] - t))\nend\n\nfunction source_terms_standard(u, x, t, equations)\n return -initial_condition(x, t, equations)\nend\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition,\n solver;\n source_terms = source_terms_standard)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Now, we can create the ODEProblem, solve the resulting ODE using a time integration method from OrdinaryDiffEq.jl, and visualize the numerical solution at the final time using Plots.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"tspan = (0.0, 3.0)\node = semidiscretize(semi, tspan)\n\nsol = solve(ode, RDPK3SpFSAL49(); ode_default_options()...)\n\nplot(sol; label = \"numerical sol.\", legend = :topright)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"We can also plot the analytical solution for comparison. Since Trixi.jl uses SVectors for the variables, we take their first (and only) component to get the scalar value for manual plotting.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"let\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topright)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"We can also add the initial condition to the plot.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"plot!(sol.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"You can of course also use some callbacks provided by Trixi.jl as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi; interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/#Using-a-custom-ODE-right-hand-side-function","page":"20 Custom semidiscretizations","title":"Using a custom ODE right-hand side function","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Next, we will solve the same problem but use our own ODE RHS function. To demonstrate this, we will artificially create a global variable containing the current time of the simulation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"const GLOBAL_TIME = Ref(0.0)\n\nfunction source_terms_custom(u, x, t, equations)\n t = GLOBAL_TIME[]\n return -initial_condition(x, t, equations)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Next, we create our own RHS function to update the global time of the simulation before calling the RHS function from Trixi.jl.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"function rhs_source_custom!(du_ode, u_ode, semi, t)\n GLOBAL_TIME[] = t\n Trixi.rhs!(du_ode, u_ode, semi, t)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Next, we create an ODEProblem manually copying over the data from the one we got from semidiscretize earlier.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"ode_source_custom = ODEProblem(rhs_source_custom!,\n ode.u0,\n ode.tspan,\n ode.p #= semi =#)\nsol_source_custom = solve(ode_source_custom, RDPK3SpFSAL49();\n ode_default_options()...)\n\nplot(sol_source_custom; label = \"numerical sol.\")\nlet\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol_source_custom.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topleft)\nend\nplot!(sol_source_custom.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"This also works with callbacks as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi; interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode_source_custom, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/#Setting-up-a-custom-semidiscretization","page":"20 Custom semidiscretizations","title":"Setting up a custom semidiscretization","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Using a global constant is of course not really nice from a software engineering point of view. Thus, it can often be useful to collect additional data in the parameters of the ODEProblem. Thus, it is time to create our own semidiscretization. Here, we create a small wrapper of a standard semidiscretization of Trixi.jl and the current global time of the simulation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"struct CustomSemidiscretization{Semi, T} <: Trixi.AbstractSemidiscretization\n semi::Semi\n t::T\nend\n\nsemi_custom = CustomSemidiscretization(semi, Ref(0.0))","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"To get pretty printing in the REPL, you can consider specializing","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Base.show(io::IO, parameters::CustomSemidiscretization)\nBase.show(io::IO, ::MIME\"text/plain\", parameters::CustomSemidiscretization)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"for your custom semidiscretiation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Next, we create our own source terms that use the global time stored in the custom semidiscretiation.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"source_terms_custom_semi = let semi_custom = semi_custom\n function source_terms_custom_semi(u, x, t, equations)\n t = semi_custom.t[]\n return -initial_condition(x, t, equations)\n end\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"We also create a custom ODE RHS to update the current global time stored in the custom semidiscretization. We unpack the standard semidiscretization created by Trixi.jl and pass it to Trixi.rhs!.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"function rhs_semi_custom!(du_ode, u_ode, semi_custom, t)\n semi_custom.t[] = t\n Trixi.rhs!(du_ode, u_ode, semi_custom.semi, t)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Finally, we set up an ODEProblem and solve it numerically.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"ode_semi_custom = ODEProblem(rhs_semi_custom!,\n ode.u0,\n ode.tspan,\n semi_custom)\nsol_semi_custom = solve(ode_semi_custom, RDPK3SpFSAL49();\n ode_default_options()...)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"If we want to make use of additional functionality provided by Trixi.jl, e.g., for plotting, we need to implement a few additional specializations. In this case, we forward everything to the standard semidiscretization provided by Trixi.jl wrapped in our custom semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Base.ndims(semi::CustomSemidiscretization) = ndims(semi.semi)\nfunction Trixi.mesh_equations_solver_cache(semi::CustomSemidiscretization)\n Trixi.mesh_equations_solver_cache(semi.semi)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Now, we can plot the numerical solution as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"plot(sol_semi_custom; label = \"numerical sol.\")\nlet\n x = range(-1.0, 1.0; length = 200)\n plot!(x, first.(initial_condition.(x, sol_semi_custom.t[end], equations)),\n label = \"analytical sol.\", linestyle = :dash, legend = :topleft)\nend\nplot!(sol_semi_custom.u[1], semi, label = \"u0\", linestyle = :dot, legend = :topleft)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"This also works with many callbacks as usual. However, the AnalysisCallback requires some special handling since it makes use of a performance counter contained in the standard semidiscretizations of Trixi.jl to report some performance metrics. Here, we forward all accesses to the performance counter to the wrapped semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"function Base.getproperty(semi::CustomSemidiscretization, s::Symbol)\n if s === :performance_counter\n wrapped_semi = getfield(semi, :semi)\n wrapped_semi.performance_counter\n else\n getfield(semi, s)\n end\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Moreover, the AnalysisCallback also performs some error calculations. We also need to forward them to the wrapped semidiscretization.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"function Trixi.calc_error_norms(func, u, t, analyzer,\n semi::CustomSemidiscretization,\n cache_analysis)\n Trixi.calc_error_norms(func, u, t, analyzer,\n semi.semi,\n cache_analysis)\nend","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"Now, we can work with the callbacks used before as usual.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"summary_callback = SummaryCallback()\nanalysis_interval = 100\nanalysis_callback = AnalysisCallback(semi_custom;\n interval = analysis_interval)\nalive_callback = AliveCallback(; analysis_interval)\ncallbacks = CallbackSet(summary_callback,\n analysis_callback,\n alive_callback)\n\nsol = solve(ode_semi_custom, RDPK3SpFSAL49();\n ode_default_options()..., callback = callbacks)\nsummary_callback()","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"For even more advanced usage of custom semidiscretizations, you may look at the source code of the ones contained in Trixi.jl, e.g.,","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"SemidiscretizationHyperbolicParabolic\nSemidiscretizationEulerGravity\nSemidiscretizationEulerAcoustics\nSemidiscretizationCoupled","category":"page"},{"location":"tutorials/custom_semidiscretization/#Package-versions","page":"20 Custom semidiscretizations","title":"Package versions","text":"","category":"section"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"","category":"page"},{"location":"tutorials/custom_semidiscretization/","page":"20 Custom semidiscretizations","title":"20 Custom semidiscretizations","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"EditURL = \"../../literate/src/files/DGSEM_FluxDiff.jl\"","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#DGSEM_FluxDiff","page":"4 DGSEM with flux differencing","title":"4: DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This tutorial starts with a presentation of the weak formulation of the discontinuous Galerkin spectral element method (DGSEM) in order to fix the notation of the used operators. Then, the DGSEM formulation with flux differencing (split form DGSEM) and its implementation in Trixi.jl is shown.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We start with the one-dimensional conservation law","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"u_t + f(u)_x = 0 qquad tin mathbbR^+ xinOmega","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"with the physical flux f.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We split the domain Omega into elements K with center x_K and size Delta x. With the transformation mapping x(xi)=x_K + fracDelta x2 xi we can transform the reference element -11 to every physical element. So, the equation can be restricted to the reference element using the determinant of the Jacobian matrix of the transformation mapping J=fracpartial xpartial xi=fracDelta x2.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"J u_t + f(u)_xi = 0 qquad tin mathbbR^+ xiin -11","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#The-weak-form-of-the-DGSEM","page":"4 DGSEM with flux differencing","title":"The weak form of the DGSEM","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We consider the so-called discontinuous Galerkin spectral element method (DGSEM) with collocation. It results from choosing a nodal DG ansatz using N+1 Gauss-Lobatto nodes xi_i in -11 with matching interpolation weights w_i, which are used for numerical integration and interpolation with the Lagrange polynomial basis l_i of degree N. The Lagrange functions are created with those nodes and hence fulfil a Kronecker property at the GL nodes. The weak formulation of the DGSEM for one element is","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"J underlinedotu(t) = - M^-1 B underlinef^* + M^-1 D^T M underlinef","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"where underlineu=(u_0 u_1 dots u_N)^TinmathbbR^N+1 is the collected pointwise evaluation of u at the discretization nodes and dotu = partial u partial t = u_t is the temporal derivative. The nodal values of the flux function f results with collocation in underlinef, since underlinef_j=f(underlineu_j). Moreover, we got the numerical flux f^*=f^*(u^- u^+).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We will now have a short overview over the operators we used.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The derivative matrix DinmathbbR^(N+1)times (N+1) mimics a spatial derivation on a discrete level with underlinef_x approx D underlinef. It is defined by D_ij = l_j(xi_i).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The diagonal mass matrix M is defined by M_ij=langle l_j l_irangle_N with the numerical scalar product langle cdot cdotrangle_N defined for functions f and g by","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"langle f grangle_N = int_-1 N^1 f(xi) g(xi) dxi = sum_k=0^N f(xi_k) g(xi_k) w_k","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The multiplication by M matches a discrete integration","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":" int_-1^1 f(xi) underlinel(xi) dxi approx M underlinef","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The boundary matrix B=textdiag(-1 0 0 1) represents an evaluation of a function at the boundaries xi_0=-1 and xi_N=1.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"For these operators the following property holds:","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":" M D + (M D)^T = B","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This is called the summation-by-parts (SBP) property since it mimics integration by parts on a discrete level (Gassner (2013)).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The explicit definitions of the operators and the construction of the 1D algorithm can be found for instance in the tutorial introduction to DG methods or in more detail in Kopriva (2009).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This property shows the equivalence between the weak form and the following strong formulation of the DGSEM.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"beginalign*\nJ underlinedotu(t)\n= - M^-1 B underlinef^* + M^-1 D^T M underlinef5pt\n= - M^-1 B underlinef^* + M^-1 (B - MD) underlinef5pt\n= - M^-1 B (underlinef^* - underlinef) - D underlinef\nendalign*","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"More information about the equivalence you can find in Kopriva, Gassner (2010).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#DGSEM-with-flux-differencing","page":"4 DGSEM with flux differencing","title":"DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"When using the diagonal SBP property it is possible to rewrite the application of the derivative operator D in the calculation of the volume integral into a subcell based finite volume type differencing formulation (Fisher, Carpenter (2013)). Generalizing","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"(D underlinef)_i = sum_j D_ij underlinef_j\n= 2sum_j frac12 D_ij (underlinef_j + underlinef_i)\neqqcolon 2sum_j D_ij f_textcentral(u_i u_j)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"we replace D underlinef in the strong form by 2D underlinef_vol(u^- u^+) with the consistent two-point volume flux f_vol and receive the DGSEM formulation with flux differencing (split form DGSEM) (Gassner, Winters, Kopriva (2016)).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"beginalign*\nJ underlinedotu(t) = - M^-1 B (underlinef^* - underlinef) - 2D underlinef_vol(u^- u^+)5pt\n= - M^-1 B (underlinef^* - underlinef_vol(underlineu underlineu)) - 2D underlinef_vol(u^- u^+)5pt\n= - M^-1 B underlinef_surface^* - (2D - M^-1 B) underlinef_vol5pt\n= - M^-1 B underlinef_surface^* - D_split underlinef_vol\nendalign*","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This formulation is in a weak form type formulation and can be implemented by using the derivative split matrix D_split=(2D-M^-1B) and two different fluxes. We divide between the surface flux f=f_surface used for the numerical flux f_surface^* and the already mentioned volume flux f_vol especially for this formulation.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This formulation creates a more stable version of DGSEM, because it fulfils entropy stability. Moreover it allows the construction of entropy conserving discretizations without relying on exact integration. This is achieved when using a two-point entropy conserving flux function as volume flux in the volume flux differencing formulation. Then, the numerical surface flux can be used to control the dissipation of the discretization and to guarantee decreasing entropy, i.e. entropy stability.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#fluxDiffExample","page":"4 DGSEM with flux differencing","title":"Implementation in Trixi.jl","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"Now, we have a look at the implementation of DGSEM with flux differencing with Trixi.jl.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"using OrdinaryDiffEq, Trixi","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We implement a simulation for the compressible Euler equations in 2D","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"partial_t beginpmatrix rho rho v_1 rho v_2 rho e endpmatrix\n+ partial_x beginpmatrix rho v_1 rho v_1^2 + p rho v_1 v_2 (rho e +p) v_1 endpmatrix\n+ partial_y beginpmatrix rho v_2 rho v_1 v_2 rho v_2^2 + p (rho e +p) v_2 endpmatrix\n= beginpmatrix 0 0 0 0 endpmatrix","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"for an ideal gas with ratio of specific heats gamma=14. Here, rho is the density, v_1, v_2 the velocities, e the specific total energy and","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"p = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"the pressure.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"gamma = 1.4\nequations = CompressibleEulerEquations2D(gamma)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"As our initial condition we will use a weak blast wave from Hennemann, Gassner (2020). The primitive variables are defined by","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"beginpmatrix rho v_1 v_2 p endpmatrix\n= beginpmatrix 10 00 00 10 endpmatrix textif x_2 05\ntextand beginpmatrix rho v_1 v_2 p endpmatrix\n= beginpmatrix 11691 01882 * cos(phi) 01882 * sin(phi) 1245 endpmatrix textelse","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"with phi = tan^-1(fracx_2x_1).","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This initial condition is implemented in Trixi.jl under the name initial_condition_weak_blast_wave.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"initial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"In Trixi.jl, flux differencing for the volume integral can be implemented with VolumeIntegralFluxDifferencing using symmetric two-point volume fluxes. First, we set up a simulation with the entropy conserving and kinetic energy preserving flux flux_ranocha by Hendrik Ranocha (2018) as surface and volume flux.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We will confirm the entropy conservation property numerically.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"volume_flux = flux_ranocha # = f_vol\nsolver = DGSEM(polydeg=3, surface_flux=volume_flux,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"Now, we implement Trixi.jl's mesh, semi and ode in a simple framework. For more information please have a look at the documentation, the basic tutorial introduction to DG methods or some basic elixirs.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"coordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=5,\n n_cells_max=10_000,\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_condition_periodic)\n\n# ODE solvers\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"To analyse the entropy conservation of the approximation, we will use the analysis calllback implemented in Trixi. It provides some information about the approximation including the entropy change.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"analysis_callback = AnalysisCallback(semi, interval=100);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We now run the simulation using flux_ranocha for both surface and volume flux.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=analysis_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"A look at the change in entropy sum partial Spartial U cdot U_t in the analysis callback confirms that the flux is entropy conserving since the change is about machine precision.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We can plot the approximated solution at the time t=0.4.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"Now, we can use for instance the dissipative flux flux_lax_friedrichs as surface flux to get an entropy stable method.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"using OrdinaryDiffEq, Trixi\n\ngamma = 1.4\nequations = CompressibleEulerEquations2D(gamma)\n\ninitial_condition = initial_condition_weak_blast_wave\n\nvolume_flux = flux_ranocha # = f_vol\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\ncoordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=5,\n n_cells_max=10_000,\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_condition_periodic)\n\n# ODE solvers\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan);\n\nanalysis_callback = AnalysisCallback(semi, interval=100);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"We now run the simulation using the volume flux flux_ranocha and surface flux flux_lax_friedrichs.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=analysis_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"The change in entropy confirms the expected entropy stability.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"Of course, you can use more than these two fluxes in Trixi. Here, we will give a short list of possible fluxes for the compressible Euler equations. For the volume flux Trixi.jl provides for example flux_ranocha, flux_shima_etal, flux_chandrashekar, flux_kennedy_gruber. As surface flux you can use all volume fluxes and additionally for instance flux_lax_friedrichs, flux_hll, flux_hllc.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/#Package-versions","page":"4 DGSEM with flux differencing","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"","category":"page"},{"location":"tutorials/DGSEM_FluxDiff/","page":"4 DGSEM with flux differencing","title":"4 DGSEM with flux differencing","text":"This page was generated using Literate.jl.","category":"page"},{"location":"multi-physics_coupling/#multi-physics-coupling","page":"Coupling","title":"Multi-physics coupling","text":"","category":"section"},{"location":"multi-physics_coupling/","page":"Coupling","title":"Coupling","text":"A complex simulation can consist of different spatial domains in which different equations are being solved, different numerical methods being used or the grid structure is different. One example would be a fluid in a tank and an extended hot plate attached to it. We would then like to solve the Navier-Stokes equations in the fluid domain and the heat conduction equations in the plate. The coupling would happen at the interface through the exchange of thermal energy.","category":"page"},{"location":"multi-physics_coupling/#Converter-coupling","page":"Coupling","title":"Converter coupling","text":"","category":"section"},{"location":"multi-physics_coupling/","page":"Coupling","title":"Coupling","text":"It may happen that the two systems to be coupled do not share any variables, but share some of the physics. In such a situation, the same physics is just represented in a different form and with a different set of variables. This is the case, for instance assuming two domains, if there is a fluid system in one domain and a Vlasov system in the other domain. In that case we would have variables representing distribution functions of the Vlasov system on one side and variables representing the mechanical quantities, like density, of the fluid system. To translate the fields from one description to the other one needs to use converter functions. These functions need to be hand tailored by the user in the elixir file where each pair of coupled systems requires two coupling functions, one for each direction.","category":"page"},{"location":"multi-physics_coupling/","page":"Coupling","title":"Coupling","text":"In the general case, we have a system A with m variables u_Ai i = 1 dots m and another system B with n variables u_Bj j = 1 dots n. We then define two coupling functions, one that transforms u_A into u_B and one that goes the other way.","category":"page"},{"location":"multi-physics_coupling/","page":"Coupling","title":"Coupling","text":"In their minimal form they take the position vector x, state vector u and the equations of the two coupled systems and return the transformed variables. By passing the equations we can make use of their parameters, if they are required. Examples can be seen in examples/structured_2d_dgsem/elixir_advection_coupled.jl.","category":"page"},{"location":"multi-physics_coupling/#Warning-about-binary-compatibility","page":"Coupling","title":"Warning about binary compatibility","text":"","category":"section"},{"location":"multi-physics_coupling/","page":"Coupling","title":"Coupling","text":"Currently the coordinate values on the nodes can differ by machine precision when simulating the mesh and when splitting the mesh in multiple domains. This is an issue coming from the coordinate interpolation on the nodes. As a result, running a simulation in a single system and in two coupled domains may result in a difference of the order of the machine precision. While this is not an issue for most practical problems, it is best to keep this in mind when comparing test runs.","category":"page"},{"location":"meshes/dgmulti_mesh/#DGMulti","page":"DGMulti mesh","title":"Unstructured meshes and the DGMulti solver","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Trixi.jl includes support for simplicial and tensor product meshes via the DGMulti solver type, which is based on the StartUpDG.jl package. DGMulti solvers also provide support for quadrilateral and hexahedral meshes, though this feature is currently restricted to Cartesian grids. On these line/quad/hex meshes, the DGMulti solver also allows to use all (finite domain) SBP derivative operators provided by SummationByPartsOperators.jl, including several finite difference SBP methods.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"We make a few simplifying assumptions about supported meshes:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"meshes consist of a single type of element\nmeshes are conforming (e.g., each face of an element is shared with at most one other element).\nthe geometric mapping from reference to physical elements is polynomial (currently, only affine mappings are supported).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"StartUpDG.jl includes both simple uniform meshes via uniform_mesh, as well as support for triangular meshes constructed using Triangulate.jl, a wrapper around Jonathan Shewchuk's Triangle package.","category":"page"},{"location":"meshes/dgmulti_mesh/#The-DGMulti-solver-type","page":"DGMulti mesh","title":"The DGMulti solver type","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Trixi.jl solvers on simplicial meshes use the DGMulti solver type, which allows users to specify element_type and approximation_type in addition to polydeg, surface_flux, surface_integral, and volume_integral.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"DGMulti(; polydeg::Integer,\n element_type::AbstractElemShape,\n approximation_type=Polynomial(),\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n RefElemData_kwargs...)","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Here, element_type can be Tri(), Quad(), Tet(), or Hex(), and approximation_type can be","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Polynomial(), which specifies a DG discretization using a polynomial basis using quadrature rules which are exact for degree 2 * polydeg integrands, or\nSBP(), which specifies a DG discretization using multi-dimensional SBP operators. Types of SBP discretizations available include: SBP{Kubatko{LobattoFaceNodes}}() (the default choice), SBP{Kubatko{LegendreFaceNodes}}(), and SBP{Hicken}(). For polydeg = 1, ..., 4, the SBP{Kubatko{LegendreFaceNodes}}() SBP nodes are identical to the SBP nodes of Chen and Shu. More detailed descriptions of each SBP node set can be found in the StartUpDG.jl docs. Trixi.jl will also specialize certain parts of the solver based on the SBP approximation type.\na (periodic or non-periodic) derivative operator from SummationByPartsOperators.jl, usually constructed as D = derivative_operator(...). In this case, you do not need to pass a polydeg. Periodic derivative operators will only work with single-element meshes constructed using DGMultiMesh.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Additional options can also be specified through RefElemData_kwargs:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"quad_rule_vol = quad_nodes(Tri(), Nq) will substitute in a volume quadrature rule of degree Nq instead of the default (which is a quadrature rule of degree polydeg). Here, a degree Nq rule will be exact for at least degree 2*Nq integrands (such that the mass matrix is integrated exactly). Quadrature rules of which exactly integrate degree Nq integrands may also be specified (for example, quad_rule_vol = StartUpDG.quad_nodes_tri(Nq) on triangles).\nquad_rule_face = quad_nodes(Line(), Nq)) will use a face quadrature rule of degree Nq rather than the default. This rule is also exact for at least degree 2*Nq integrands.","category":"page"},{"location":"meshes/dgmulti_mesh/#The-GaussSBP()-approximation-type-on-Quad()-and-Hex()-meshes","page":"DGMulti mesh","title":"The GaussSBP() approximation type on Quad() and Hex() meshes","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"When using VolumeIntegralFluxDifferencing on Quad() and Hex() meshes, one can also specify approximation_type = GaussSBP() to use an entropy stable Gauss collocation scheme. Here, GaussSBP() refers to \"generalized\" summation-by-parts operators (see for example Ranocha 2018 or Fernandez and Zingg 2015).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Unlike traditional SBP operators, generalized SBP operators are constructed from nodes which do not include boundary nodes (i.e., Gauss quadrature nodes as opposed to Gauss-Lobatto quadrature nodes). This makes the computation of interface fluxes slightly more expensive, but also usually results in a more accurate solution. Roughly speaking, an entropy stable Gauss collocation scheme will yield results similar to a modal entropy stable scheme using a Polynomial() approximation type, but will be more efficient at high orders of approximation.","category":"page"},{"location":"meshes/dgmulti_mesh/#Trixi.jl-elixirs-on-simplicial-and-tensor-product-element-meshes","page":"DGMulti mesh","title":"Trixi.jl elixirs on simplicial and tensor product element meshes","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Example elixirs with triangular, quadrilateral, and tetrahedral meshes can be found in the examples/dgmulti_2d/ and examples/dgmulti_3d/ folders. Some key elixirs to look at:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"examples/dgmulti_2d/elixir_euler_weakform.jl: basic weak form DG discretization on a uniform triangular mesh. Changing element_type = Quad() or approximation_type = SBP() will switch to a quadrilateral mesh or an SBP-type discretization. Changing surface_integral = SurfaceIntegralWeakForm(flux_ec) and volume_integral = VolumeIntegralFluxDifferencing(flux_ec) for some entropy conservative flux (e.g., flux_chandrashekar or flux_ranocha) will switch to an entropy conservative formulation.\nexamples/dgmulti_2d/elixir_euler_triangulate_pkg_mesh.jl: uses an unstructured mesh generated by Triangulate.jl.\nexamples/dgmulti_3d/elixir_euler_weakform.jl: ´basic weak form DG discretization on a uniform tetrahedral mesh. Changing element_type = Hex() will switch to a hexahedral mesh. Changing surface_integral = SurfaceIntegralWeakForm(flux_ec) and volume_integral = VolumeIntegralFluxDifferencing(flux_ec) for some entropy conservative flux (e.g., flux_chandrashekar or flux_ranocha) will switch to an entropy conservative formulation.","category":"page"},{"location":"meshes/dgmulti_mesh/#For-developers","page":"DGMulti mesh","title":"For developers","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/#DGMultiMesh-wrapper-type","page":"DGMulti mesh","title":"DGMultiMesh wrapper type","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"DGMulti meshes in Trixi.jl are represented using a DGMultiMesh{NDIMS, ...} type. This mesh type is assumed to have fields md::MeshData, which contains geometric terms derived from the mapping between the reference and physical elements, and boundary_faces, which contains a Dict of boundary segment names (symbols) and list of faces which lie on that boundary segment.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"A DGMultiMesh can be constructed in several ways. For example, DGMultiMesh(dg::DGMulti) will return a Cartesian mesh on -1 1^d with element types specified by dg. DGMulti meshes can also be constructed by specifying a list of vertex coordinates vertex_coordinates_x, vertex_coordinates_y, vertex_coordinates_z and a connectivity matrix EToV where EToV[e,:] gives the vertices which correspond to element e. These quantities are available from most unstructured mesh generators.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Initial support for curved DGMultiMeshes is available for flux differencing solvers using SBP and GaussSBP approximation types on quadrilateral and hexahedral meshes. These can be called by specifying mesh = DGMultiMesh(dg, cells_per_dimension, mapping), where mapping is a function which specifies the warping of the mesh (e.g., mapping(xi, eta) = SVector{2}(xi, eta) is the identity mapping) similar to the mapping argument used by StructuredMesh.","category":"page"},{"location":"meshes/dgmulti_mesh/#Variable-naming-conventions","page":"DGMulti mesh","title":"Variable naming conventions","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"We use the convention that coordinates on the reference element are r in 1D, r s in 2D, or r s t in 3D. Physical coordinates use the standard conventions x (1D), x y (2D), and x y z (3D).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"(Image: \"Ref-to-physical mapping\")","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Derivatives of reference coordinates with respect to physical coordinates are abbreviated, e.g., fracpartial rpartial x = r_x. Additionally, J is used to denote the determinant of the Jacobian of the reference-to-physical mapping.","category":"page"},{"location":"meshes/dgmulti_mesh/#Variable-meanings-and-conventions-in-StartUpDG.jl","page":"DGMulti mesh","title":"Variable meanings and conventions in StartUpDG.jl","text":"","category":"section"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"StartUpDG.jl exports structs RefElemData{NDIMS, ElemShape, ...} (which contains data associated with the reference element, such as interpolation points, quadrature rules, face nodes, normals, and differentiation/interpolation/projection matrices) and MeshData{NDIMS} (which contains geometric data associated with a mesh). These are currently used for evaluating DG formulations in a matrix-free fashion. These structs contain fields similar (but not identical) to those in Globals1D, Globals2D, Globals3D in the Matlab codes from \"Nodal Discontinuous Galerkin Methods\" by Hesthaven and Warburton (2007).","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"In general, we use the following code conventions:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"variables such as r, s,... and x, y,... correspond to values at nodal interpolation points.\nvariables ending in q (e.g., rq, sq,... and xq, yq,...) correspond to values at volume quadrature points.\nvariables ending in f (e.g., rf, sf,... and xf, yf,...) correspond to values at face quadrature points.\nvariables ending in p (e.g., rp, sp,...) correspond to \"plotting\" points, which are usually a fine grid of equispaced points.\nV matrices correspond to interpolation matrices from nodal interpolation points, e.g., Vq interpolates to volume quadrature points, Vf interpolates to face quadrature points.\ngeometric quantities in MeshData are stored as matrices of dimension textnumber of points per element times textnumber of elements.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Quantities in rd::RefElemData:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"rd.Np, rd.Nq, rd.Nf: the number of nodal interpolation points, volume quadrature points, and face quadrature points on the reference element, respectively.\nrd.Vq: interpolation matrices from values at nodal interpolation points to volume quadrature points\nrd.wq: volume quadrature weights on the reference element\nrd.Vf: interpolation matrices from values at nodal interpolation points to face quadrature points\nrd.wf: a vector containing face quadrature weights on the reference element\nrd.M: the quadrature-based mass matrix, computed via rd.Vq' * diagm(rd.wq) * rd.Vq.\nrd.Pq: a quadrature-based L^2 projection matrix rd.Pq = rd.M \\ rd.Vq' * diagm(rd.wq) which maps between values at quadrature points and values at nodal points.\nDr, Ds, Dt matrices are nodal differentiation matrices with respect to the rst coordinates, e.g., Dr*f.(r,s) approximates the derivative of f(rs) at nodal points.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"Quantities in md::MeshData:","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"md.xyz is a tuple of matrices md.x, md.y, md.z, where column e contains coordinates of physical interpolation points.\nmd.xyzq is a tuple of matrices md.xq, md.yq, md.zq, where column e contains coordinates of physical quadrature points.\nmd.rxJ, md.sxJ, ... are matrices where column e contains values of Jfracpartial rpartial x, Jfracpartial spartial x, etc. at nodal interpolation points on the element e.\nmd.J is a matrix where column e contains values of the Jacobian J at nodal interpolation points.\nmd.Jf is a matrix where column e contains values of the face Jacobian (e.g., determinant of the geometric mapping between a physical face and a reference face) at face quadrature points.\nmd.nxJ, md.nyJ, ... are matrices where column e contains values of components of the unit normal scaled by the face Jacobian md.Jf at face quadrature points.","category":"page"},{"location":"meshes/dgmulti_mesh/","page":"DGMulti mesh","title":"DGMulti mesh","text":"For more details, please see the StartUpDG.jl docs.","category":"page"},{"location":"meshes/p4est_mesh/#P4est-based-mesh","page":"P4est-based mesh","title":"P4est-based mesh","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The P4estMesh is an unstructured, curvilinear, nonconforming mesh type for quadrilateral (2D) and hexahedral (3D) cells. It supports quadtree/octree-based adaptive mesh refinement (AMR) via the C library p4est. See AMRCallback for further information.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Due to its curvilinear nature, (numerical) fluxes need to implement methods dispatching on the normal::AbstractVector. Rotationally invariant equations such as the compressible Euler equations can use FluxRotated to wrap numerical fluxes implemented only for Cartesian meshes. This simplifies the re-use of existing functionality for the TreeMesh but is usually less efficient, cf. PR #550.","category":"page"},{"location":"meshes/p4est_mesh/#Construction-of-a-P4estMesh-from-an-Abaqus-file","page":"P4est-based mesh","title":"Construction of a P4estMesh from an Abaqus file","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"One available option to construct a P4estMesh is to read in an Abaqus (.inp) mesh file. We briefly describe the structure of this file, the conventions it uses, and how the mesh file is parsed to create an initial unstructured, curvilinear, and conforming mesh.","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-in-two-spatial-dimensions","page":"P4est-based mesh","title":"Mesh in two spatial dimensions","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For this discussion we use the following two-dimensional unstructured curved mesh with three elements:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"(Image: abaqus-2dmesh-docs)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"We note that the corner and element connectivity information parsed from the Abaqus file creates a straight sided (linear) mesh. From this linear mesh there are two strategies available to make the mesh curvilinear:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Apply a mapping function to describe a transformation of the linear mesh to another physical domain. The mapping is approximated using interpolation polynomial of a user specified polynomial degree. The default value of this polynomial degree is 1 that corresponds to an uncurved geometry.\nHigh-order boundary information is available in the .inp mesh file because it was created with the HOHQMesh mesh generator, which is available via the Julia package HOHQMesh.jl. This information is used to create appropriate transfinite mappings during the mesh construction.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"We divide our discussion into two parts. The first part discusses the standard corner and element information contained in the .inp mesh file. The second part specifically deals with the mesh file parsing of an Abaqus file created by HOHQMesh.jl.","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-file-header","page":"P4est-based mesh","title":"Mesh file header","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"An Abaqus .inp mesh file typically begins with a *Heading. Though optional, the *Heading is helpful to give users some information about the mesh described by the mesh file. In particular, a .inp mesh file created with HOHQMesh will contain the header","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*Heading\n File created by HOHQMesh","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"This heading is used to indicate to the mesh constructor which of the above mapping strategies to apply in order to create a curvilinear mesh. If the Abaqus file header is not present then the P4estMesh is created with the first strategy above.","category":"page"},{"location":"meshes/p4est_mesh/#corner-node-list","page":"P4est-based mesh","title":"List of corner nodes","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next, prefaced with *NODE, comes a list of the physical (x,y,z) coordinates of all the corners. The first integer in the list of the corners provides its id number. Thus, for the two-dimensional example mesh this block of corner information is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*NODE\n1, 1.0, -1.0, 0.0\n2, 3.0, 0.0, 0.0\n3, 1.0, 1.0, 0.0\n4, 2.0, 0.0, 0.0\n5, 0.0, 0.0, 0.0\n6, 3.0, 1.0, 0.0\n7, 3.0, -1.0, 0.0","category":"page"},{"location":"meshes/p4est_mesh/#element-list","page":"P4est-based mesh","title":"List of elements","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The element connectivity is given after the list of corners. The header for this information block is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*ELEMENT, type=CPS4, ELSET=Surface1","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The Abaqus element type CPS4 corresponds to a quadrilateral element. Each quadrilateral element in the unstructured mesh is dictated by four corner points with indexing taken from the numbering given by the corner list above. The elements connect a set of four corner points (starting from the bottom left) in an anti-clockwise fashion; making the element right-handed. This element handedness is indicated using the circular arrow in the figure above. Just as with the corner list, the first integer in the element connectivity list indicates the element id number. Thus, the element connectivity list for the three element example mesh is","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*ELEMENT, type=CPS4, ELSET=Surface1\n1, 5, 1, 4, 3\n2, 4, 2, 6, 3\n3, 7, 2, 4, 1","category":"page"},{"location":"meshes/p4est_mesh/#Element-neighbor-connectivity","page":"P4est-based mesh","title":"Element neighbor connectivity","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The construction of the element neighbor ids and identifying physical boundary surfaces is done using functionality directly from the p4est library. For example, the neighbor connectivity is created in the mesh constructor using the wrapper read_inp_p4est function.","category":"page"},{"location":"meshes/p4est_mesh/#Encoding-of-boundaries","page":"P4est-based mesh","title":"Encoding of boundaries","text":"","category":"section"},{"location":"meshes/p4est_mesh/#HOHQMesh-boundary-information","page":"P4est-based mesh","title":"HOHQMesh boundary information","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"If present, any additional information in the mesh file that was created by HOHQMesh is prefaced with ** to make it an Abaqus comment. This ensures that the read in of the file by a standard Abaqus file parser, as done in the read_inp_p4est function, is done correctly.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The high-order, curved boundary information and labels of the physical boundary created by HOHQMesh is found below the comment line","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** ***** HOHQMesh boundary information ***** **","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next comes the polynomial degree that the mesh will use to represent any curved sides","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** mesh polynomial degree = 8","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The mesh file then, again, provides the element connectivity as well as information for curved surfaces either interior to the domain or along the physical boundaries. A set of check digits are included directly below the four corner indexes to indicate whether the local surface index (-y, +x, +y, or -x) within the element is straight sided, 0, or is curved, 1. If the local surface is straight sided no additional information is necessary during the mesh file read in. But for any curved surfaces the mesh file provides (x,y,z) coordinate values in order to construct an interpolant of this surface with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes. This list of (x,y,z) data will be given in the direction of the local coordinate system. Given below is the element curvature information for the example mesh:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** 5 1 4 3\n** 0 0 1 1\n** 1.000000000000000 1.000000000000000 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 2.000000000000000 0.0 0.0\n** 0.0 0.0 0.0\n** 0.035513826946206 0.105291711848750 0.0\n** 0.148591270347399 0.317731556850611 0.0\n** 0.340010713990041 0.452219430075470 0.0\n** 0.575000000000000 0.462500000000000 0.0\n** 0.788022294598950 0.483764065630034 0.0\n** 0.926408729652601 0.644768443149389 0.0\n** 0.986453164464803 0.883724792445746 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 4 2 6 3\n** 0 0 0 1\n** 2.000000000000000 0.0 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 7 2 4 1\n** 0 0 0 0","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The last piece of information provided by HOHQMesh are labels for the different surfaces of an element. These labels are useful to set boundary conditions along physical surfaces. The labels can be short descriptive words up to 32 characters in length. The label --- indicates an internal surface where no boundary condition is required.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"It is important to note that these labels are given in the following order according to the local surface index -x +x -y +y as required by the p4est library.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"** Bezier --- Slant ---\n** --- Right --- Top\n** Bottom --- Right ---","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For completeness, we provide the entire Abaqus mesh file for the example mesh in the figure above:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*Heading\n File created by HOHQMesh\n*NODE\n1, 1.0, -1.0, 0.0\n2, 3.0, 0.0, 0.0\n3, 1.0, 1.0, 0.0\n4, 2.0, 0.0, 0.0\n5, 0.0, 0.0, 0.0\n6, 3.0, 1.0, 0.0\n7, 3.0, -1.0, 0.0\n*ELEMENT, type=CPS4, ELSET=Surface1\n1, 5, 1, 4, 3\n2, 4, 2, 6, 3\n3, 7, 2, 4, 1\n** ***** HOHQMesh boundary information ***** **\n** mesh polynomial degree = 8\n** 5 1 4 3\n** 0 0 1 1\n** 1.000000000000000 1.000000000000000 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 2.000000000000000 0.0 0.0\n** 0.0 0.0 0.0\n** 0.035513826946206 0.105291711848750 0.0\n** 0.148591270347399 0.317731556850611 0.0\n** 0.340010713990041 0.452219430075470 0.0\n** 0.575000000000000 0.462500000000000 0.0\n** 0.788022294598950 0.483764065630034 0.0\n** 0.926408729652601 0.644768443149389 0.0\n** 0.986453164464803 0.883724792445746 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 4 2 6 3\n** 0 0 0 1\n** 2.000000000000000 0.0 0.0\n** 1.986035130050921 0.054554577460044 0.0\n** 1.920916981799849 0.185149035378133 0.0\n** 1.768263070247418 0.329729152118310 0.0\n** 1.537500000000000 0.462500000000000 0.0\n** 1.295753434047077 0.606254343587194 0.0\n** 1.116583018200151 0.777350964621867 0.0\n** 1.024948365654583 0.934461926834452 0.0\n** 1.000000000000000 1.000000000000000 0.0\n** 7 2 4 1\n** 0 0 0 0\n** Bezier --- Slant ---\n** --- Right --- Top\n** Bottom --- Right ---","category":"page"},{"location":"meshes/p4est_mesh/#Standard-Abaqus-format-boundary-information","page":"P4est-based mesh","title":"Standard Abaqus format boundary information","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"As an alternative to an Abaqus mesh generated by HOHQMesh, .inp files with boundary information encoded as nodesets *NSET,NSET= can be used to construct a p4est mesh. This is especially useful for usage of existing meshes (consisting of bilinear elements) which could stem from the popular gmsh meshing software.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"In addition to the list of nodes and elements given above, there are nodesets of the form ","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*NSET,NSET=PhysicalLine1\n1, 4, 52, 53, 54, 55, 56, 57, 58, ","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"present which are used to associate the edges defined through their corner nodes with a label. In this case it is called PhysicalLine1. By looping over every element and its associated edges, consisting of two nodes, we query the read in NSETs if the current node pair is present.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"To prevent that every nodeset following *NSET,NSET= is treated as a boundary, the user must supply a boundary_symbols keyword to the P4estMesh constructor:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"boundary_symbols = [:PhysicalLine1]\n\nmesh = P4estMesh{2}(mesh_file, polydeg = polydeg, boundary_symbols = boundary_symbols)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"By doing so, only nodesets with a label present in boundary_symbols are treated as physical boundaries. Other nodesets that could be used for diagnostics are not treated as external boundaries. Note that there is a leading colon : compared to the label in the .inp mesh file. This is required to turn the label into a Symbol.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"A 2D example for this mesh, which is read-in for an unstructured mesh file created with gmsh, is presented in examples/p4est_2d_dgsem/elixir_euler_NACA6412airfoil_mach2.jl.","category":"page"},{"location":"meshes/p4est_mesh/#Mesh-in-three-spatial-dimensions","page":"P4est-based mesh","title":"Mesh in three spatial dimensions","text":"","category":"section"},{"location":"meshes/p4est_mesh/#HOHQMesh-Extended-Abaqus-format","page":"P4est-based mesh","title":"HOHQMesh-Extended Abaqus format","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The 3D Abaqus file format with high-order boundary information from HOHQMesh is very similar to the 2D version discussed above. There are only three changes:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"The element connectivity would be given in terms of the eight corners that define a hexahedron. The corners are numbered as shown in the figure below. The header of the element list changes to be\n*ELEMENT, type=C3D8, ELSET=Volume1\nwhere C3D8 corresponds to a Abaqus hexahedral element.\nThere are six check digits included directly below the eight corner indexes to indicate whether the local face within the element is straight sided, 0, or is curved, 1. For curved faces (x,y,z) coordinate values are available in order to construct an face interpolant with the mesh polynomial order at the Chebyshev-Gauss-Lobatto nodes.\nThe boundary labels are given in the following order according to the local surface index -x +x -y +y -z +z as required by the p4est library.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For completeness, we also give a short description and derivation of the three-dimensional transfinite mapping formulas used to compute the physical coordinates mathbfx=(xyz) of a (possibly curved) hexahedral element give the reference coordinates boldsymbolxi = (xi eta zeta) which lie in -11^3. That is, we will create an expression mathbfx= mathbfX(boldsymbolxi).","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Below we provide a sketch of a single hexahedral element with curved faces. This is done to introduce the numbering conventions for corners, edges, and faces of the element.","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"(Image: abaqus-3dmesh-docs)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"When the hexahedron is a straight sided (linear) element we compute the transfinite mapping directly from the element corner points according to","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\nmathbfX_linear(boldsymbolxi) = frac18quad mathbfx_1(1-xi)(1-eta)(1-zeta)\n + mathbfx_2(1+xi)(1-eta)(1-zeta)-015cm\n qquad + mathbfx_3(1+xi)(1+eta)(1-zeta)\n + mathbfx_4(1-xi)(1+eta)(1-zeta) \n qquad + mathbfx_5(1-xi)(1-eta)(1+zeta)\n + mathbfx_6(1+xi)(1-eta)(1+zeta) \n qquad + mathbfx_7(1+xi)(1+eta)(1+zeta)\n + mathbfx_8(1-xi)(1+eta)(1+zeta)quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Next, we create a transfinite mapping function, mathbfX(boldsymbolxi), for a hexahedron that has one or more curved faces. For this we assume that have a set of six interpolating polynomials Gamma_i_i=1^6 that approximate the faces. The interpolating polynomial for any curved faces is provided by the information in a HOHQMesh Abaqus mesh file or is constructed on the fly via a bi-linear interpolation routine for any linear faces. Explicitly, these six face interpolation polynomials depend on the computational coordinates boldsymbolxi as follows","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":" beginaligned\n Gamma_1(xi zeta) quad quad Gamma_3(xi eta) quad quad Gamma_4(eta zeta)01cm\n Gamma_2(xi zeta) quad quad Gamma_5(xi eta) quad quad Gamma_6(eta zeta)\n endaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"To determine the form of the mapping we first create linear interpolations between two opposing faces, e.g., Gamma_3 and Gamma_5 and sum them together to have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\n boldsymbolSigma(boldsymbolxi) = frac12quad(1-xi)Gamma_6(etazeta) + (1+xi)Gamma_4(etazeta) -015cm\n qquad+ (1-eta)Gamma_1(xizeta) + (1+eta)Gamma_2(xizeta) -015cm\n qquad +(1-zeta)Gamma_3(xieta) + (1+zeta)Gamma_5(xieta)quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Unfortunately, the linear interpolations boldsymbolSigma(boldsymbolxi) no longer match at the faces, e.g., evaluating at eta = -1 we have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"boldsymbolSigma(xi-1zeta) = Gamma_1(xizeta) + frac12(1-xi)Gamma_6(-1zeta) + (1+xi)Gamma_4(-1zeta)\n +(1-zeta)Gamma_3(xi-1) + (1+zeta)Gamma_5(xi-1)","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"which is the desired face Gamma_1(xizeta) plus four edge error terms. Analogous edge error terms occur at the other faces evaluating boldsymbolSigma(boldsymbolxi) at eta=1, xi=pm 1, and zeta=pm 1. In order to match the faces, we subtract a linear interpolant in the xi, eta, and zeta directions of the edge error terms, e.g., the terms in braces in the above equation. So, continuing the example above, the correction term to be subtracted for face Gamma_1 to match would be","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"left(frac1-eta2right) bigg frac12 (1-xi)Gamma_6(-1zeta) + (1+xi)Gamma_4(-1zeta)+(1-zeta)Gamma_3(xi-1)\n + (1+zeta)Gamma_5(xi-1) bigg","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"For clarity, and to allow an easier comparison to the implementation, we introduce auxiliary notation for the 12 edge values present in the complete correction term. That is, for given values of xi, eta, and zeta we have","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":" beginaligned\n textttedge_1 = Gamma_1(xi -1) quad quad textttedge_5 = Gamma_2(xi -1) quad quad textttedge_9 = Gamma_6(eta -1)01cm\n textttedge_2 = Gamma_1(1 zeta) quad quadtextttedge_6 = Gamma_2(1 zeta) quad quad textttedge_10 = Gamma_4(eta -1)01cm\n textttedge_3 = Gamma_1(xi 1) quad quad textttedge_7 = Gamma_2(xi 1) quad quad textttedge_11 = Gamma_4(eta 1)01cm\n textttedge_4 = Gamma_1(-1 zeta) quad quad textttedge_8 = Gamma_2(-1 zeta) quad quad textttedge_12 = Gamma_6(eta 1)\n endaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"With this notation for the edge terms (and after some algebraic manipulation) we write the complete edge correction term, mathcalC_textttedge(boldsymbolxi), as","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"beginaligned\nmathcalC_textttedge(boldsymbolxi) = frac14quad (1-eta)(1-zeta)textttedge_1-015cm\n qquad + (1+xi)(1-eta)textttedge_2 \n qquad + (1-eta)(1+zeta)textttedge_3 \n qquad + (1-xi)(1-eta)textttedge_4 \n qquad + (1+eta)(1-zeta)textttedge_5 \n qquad + (1+xi)(1+eta)textttedge_6 \n qquad + (1+eta)(1+zeta)textttedge_7 \n qquad + (1-xi)(1+eta)textttedge_8 \n qquad + (1-xi)(1-zeta)textttedge_9 \n qquad + (1+xi)(1-zeta)textttedge_10 \n qquad + (1+xi)(1+zeta)textttedge_11 \n qquad + (1-xi)(1+zeta)textttedge_12quad\nendaligned","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"However, subtracting the edge correction terms mathcalC_textttedge(boldsymbolxi) from boldsymbolSigma(boldsymbolxi) removes the interior element contributions twice. Thus, to complete the construction of the transfinite mapping mathbfX(boldsymbolxi) we add the transfinite map of the straight sided hexahedral element to find","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"mathbfX(boldsymbolxi) = boldsymbolSigma(boldsymbolxi)\n - mathcalC_textttedge(boldsymbolxi)\n + mathbfX_linear(boldsymbolxi)","category":"page"},{"location":"meshes/p4est_mesh/#Construction-from-standard-Abaqus","page":"P4est-based mesh","title":"Construction from standard Abaqus","text":"","category":"section"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"Also for a mesh in standard Abaqus format there are no qualitative changes when going from 2D to 3D. The most notable difference is that boundaries are formed in 3D by faces defined by four nodes while in 2D boundaries are edges consisting of two elements. A simple mesh file, which is used also in examples/p4est_3d_dgsem/elixir_euler_free_stream_boundaries.jl, is given below:","category":"page"},{"location":"meshes/p4est_mesh/","page":"P4est-based mesh","title":"P4est-based mesh","text":"*Heading\n\n*NODE\n1, -2, 0, 0\n2, -1, 0, 0\n3, -1, 1, 0\n4, -2, 1, 0\n5, -2, 0, 1\n6, -1, 0, 1\n7, -1, 1, 1\n8, -2, 1, 1\n9, -1.75, 1, 0\n10, -1.5, 1, 0\n11, -1.25, 1, 0\n12, -1, 0.75000000000035, 0\n13, -1, 0.50000000000206, 0\n14, -1, 0.25000000000104, 0\n15, -1.25, 0, 0\n16, -1.5, 0, 0\n17, -1.75, 0, 0\n18, -2, 0.24999999999941, 0\n19, -2, 0.49999999999869, 0\n20, -2, 0.74999999999934, 0\n21, -1.75, 0, 1\n22, -1.5, 0, 1\n23, -1.25, 0, 1\n24, -1, 0.24999999999941, 1\n25, -1, 0.49999999999869, 1\n26, -1, 0.74999999999934, 1\n27, -1.25, 1, 1\n28, -1.5, 1, 1\n29, -1.75, 1, 1\n30, -2, 0.75000000000035, 1\n31, -2, 0.50000000000206, 1\n32, -2, 0.25000000000104, 1\n33, -2, 0, 0.24999999999941\n34, -2, 0, 0.49999999999869\n35, -2, 0, 0.74999999999934\n36, -2, 1, 0.24999999999941\n37, -2, 1, 0.49999999999869\n38, -2, 1, 0.74999999999934\n39, -1, 0, 0.24999999999941\n40, -1, 0, 0.49999999999869\n41, -1, 0, 0.74999999999934\n42, -1, 1, 0.24999999999941\n43, -1, 1, 0.49999999999869\n44, -1, 1, 0.74999999999934\n45, -1.25, 0.25000000000063, 0\n46, -1.25, 0.50000000000122, 0\n47, -1.25, 0.7500000000001, 0\n48, -1.5, 0.25000000000023, 0\n49, -1.5, 0.50000000000038, 0\n50, -1.5, 0.74999999999984, 0\n51, -1.75, 0.24999999999982, 0\n52, -1.75, 0.49999999999953, 0\n53, -1.75, 0.74999999999959, 0\n54, -1.75, 0.25000000000063, 1\n55, -1.75, 0.50000000000122, 1\n56, -1.75, 0.7500000000001, 1\n57, -1.5, 0.25000000000023, 1\n58, -1.5, 0.50000000000038, 1\n59, -1.5, 0.74999999999984, 1\n60, -1.25, 0.24999999999982, 1\n61, -1.25, 0.49999999999953, 1\n62, -1.25, 0.74999999999959, 1\n63, -2, 0.24999999999982, 0.24999999999941\n64, -2, 0.49999999999953, 0.24999999999941\n65, -2, 0.74999999999959, 0.24999999999941\n66, -2, 0.25000000000023, 0.49999999999869\n67, -2, 0.50000000000038, 0.49999999999869\n68, -2, 0.74999999999984, 0.49999999999869\n69, -2, 0.25000000000063, 0.74999999999934\n70, -2, 0.50000000000122, 0.74999999999934\n71, -2, 0.7500000000001, 0.74999999999934\n72, -1.25, 1, 0.74999999999934\n73, -1.25, 1, 0.49999999999869\n74, -1.25, 1, 0.24999999999941\n75, -1.5, 1, 0.74999999999934\n76, -1.5, 1, 0.49999999999869\n77, -1.5, 1, 0.24999999999941\n78, -1.75, 1, 0.74999999999934\n79, -1.75, 1, 0.49999999999869\n80, -1.75, 1, 0.24999999999941\n81, -1, 0.25000000000063, 0.24999999999941\n82, -1, 0.50000000000122, 0.24999999999941\n83, -1, 0.7500000000001, 0.24999999999941\n84, -1, 0.25000000000023, 0.49999999999869\n85, -1, 0.50000000000038, 0.49999999999869\n86, -1, 0.74999999999984, 0.49999999999869\n87, -1, 0.24999999999982, 0.74999999999934\n88, -1, 0.49999999999953, 0.74999999999934\n89, -1, 0.74999999999959, 0.74999999999934\n90, -1.75, 0, 0.74999999999934\n91, -1.75, 0, 0.49999999999869\n92, -1.75, 0, 0.24999999999941\n93, -1.5, 0, 0.74999999999934\n94, -1.5, 0, 0.49999999999869\n95, -1.5, 0, 0.24999999999941\n96, -1.25, 0, 0.74999999999934\n97, -1.25, 0, 0.49999999999869\n98, -1.25, 0, 0.24999999999941\n99, -1.75, 0.25000000000043, 0.74999999999934\n100, -1.75, 0.25000000000023, 0.49999999999869\n101, -1.75, 0.25000000000002, 0.24999999999941\n102, -1.75, 0.5000000000008, 0.74999999999934\n103, -1.75, 0.50000000000038, 0.49999999999869\n104, -1.75, 0.49999999999995, 0.24999999999941\n105, -1.75, 0.74999999999997, 0.74999999999934\n106, -1.75, 0.74999999999984, 0.49999999999869\n107, -1.75, 0.74999999999972, 0.24999999999941\n108, -1.5, 0.25000000000023, 0.74999999999934\n109, -1.5, 0.25000000000023, 0.49999999999869\n110, -1.5, 0.25000000000023, 0.24999999999941\n111, -1.5, 0.50000000000038, 0.74999999999934\n112, -1.5, 0.50000000000038, 0.49999999999869\n113, -1.5, 0.50000000000038, 0.24999999999941\n114, -1.5, 0.74999999999984, 0.74999999999934\n115, -1.5, 0.74999999999984, 0.49999999999869\n116, -1.5, 0.74999999999984, 0.24999999999941\n117, -1.25, 0.25000000000002, 0.74999999999934\n118, -1.25, 0.25000000000023, 0.49999999999869\n119, -1.25, 0.25000000000043, 0.24999999999941\n120, -1.25, 0.49999999999995, 0.74999999999934\n121, -1.25, 0.50000000000038, 0.49999999999869\n122, -1.25, 0.5000000000008, 0.24999999999941\n123, -1.25, 0.74999999999972, 0.74999999999934\n124, -1.25, 0.74999999999984, 0.49999999999869\n125, -1.25, 0.74999999999997, 0.24999999999941\n******* E L E M E N T S *************\n*ELEMENT, type=C3D8, ELSET=Volume1\n153, 54, 21, 5, 32, 99, 90, 35, 69\n154, 99, 90, 35, 69, 100, 91, 34, 66\n155, 100, 91, 34, 66, 101, 92, 33, 63\n156, 101, 92, 33, 63, 51, 17, 1, 18\n157, 55, 54, 32, 31, 102, 99, 69, 70\n158, 102, 99, 69, 70, 103, 100, 66, 67\n159, 103, 100, 66, 67, 104, 101, 63, 64\n160, 104, 101, 63, 64, 52, 51, 18, 19\n161, 56, 55, 31, 30, 105, 102, 70, 71\n162, 105, 102, 70, 71, 106, 103, 67, 68\n163, 106, 103, 67, 68, 107, 104, 64, 65\n164, 107, 104, 64, 65, 53, 52, 19, 20\n165, 29, 56, 30, 8, 78, 105, 71, 38\n166, 78, 105, 71, 38, 79, 106, 68, 37\n167, 79, 106, 68, 37, 80, 107, 65, 36\n168, 80, 107, 65, 36, 9, 53, 20, 4\n169, 57, 22, 21, 54, 108, 93, 90, 99\n170, 108, 93, 90, 99, 109, 94, 91, 100\n171, 109, 94, 91, 100, 110, 95, 92, 101\n172, 110, 95, 92, 101, 48, 16, 17, 51\n173, 58, 57, 54, 55, 111, 108, 99, 102\n174, 111, 108, 99, 102, 112, 109, 100, 103\n175, 112, 109, 100, 103, 113, 110, 101, 104\n176, 113, 110, 101, 104, 49, 48, 51, 52\n177, 59, 58, 55, 56, 114, 111, 102, 105\n178, 114, 111, 102, 105, 115, 112, 103, 106\n179, 115, 112, 103, 106, 116, 113, 104, 107\n180, 116, 113, 104, 107, 50, 49, 52, 53\n181, 28, 59, 56, 29, 75, 114, 105, 78\n182, 75, 114, 105, 78, 76, 115, 106, 79\n183, 76, 115, 106, 79, 77, 116, 107, 80\n184, 77, 116, 107, 80, 10, 50, 53, 9\n185, 60, 23, 22, 57, 117, 96, 93, 108\n186, 117, 96, 93, 108, 118, 97, 94, 109\n187, 118, 97, 94, 109, 119, 98, 95, 110\n188, 119, 98, 95, 110, 45, 15, 16, 48\n189, 61, 60, 57, 58, 120, 117, 108, 111\n190, 120, 117, 108, 111, 121, 118, 109, 112\n191, 121, 118, 109, 112, 122, 119, 110, 113\n192, 122, 119, 110, 113, 46, 45, 48, 49\n193, 62, 61, 58, 59, 123, 120, 111, 114\n194, 123, 120, 111, 114, 124, 121, 112, 115\n195, 124, 121, 112, 115, 125, 122, 113, 116\n196, 125, 122, 113, 116, 47, 46, 49, 50\n197, 27, 62, 59, 28, 72, 123, 114, 75\n198, 72, 123, 114, 75, 73, 124, 115, 76\n199, 73, 124, 115, 76, 74, 125, 116, 77\n200, 74, 125, 116, 77, 11, 47, 50, 10\n201, 24, 6, 23, 60, 87, 41, 96, 117\n202, 87, 41, 96, 117, 84, 40, 97, 118\n203, 84, 40, 97, 118, 81, 39, 98, 119\n204, 81, 39, 98, 119, 14, 2, 15, 45\n205, 25, 24, 60, 61, 88, 87, 117, 120\n206, 88, 87, 117, 120, 85, 84, 118, 121\n207, 85, 84, 118, 121, 82, 81, 119, 122\n208, 82, 81, 119, 122, 13, 14, 45, 46\n209, 26, 25, 61, 62, 89, 88, 120, 123\n210, 89, 88, 120, 123, 86, 85, 121, 124\n211, 86, 85, 121, 124, 83, 82, 122, 125\n212, 83, 82, 122, 125, 12, 13, 46, 47\n213, 7, 26, 62, 27, 44, 89, 123, 72\n214, 44, 89, 123, 72, 43, 86, 124, 73\n215, 43, 86, 124, 73, 42, 83, 125, 74\n216, 42, 83, 125, 74, 3, 12, 47, 11\n*NSET,NSET=PhysicalSurface1\n1, 2, 3, 4, 5, 6, 7, 8, 9, 10, \n11, 12, 13, 14, 15, 16, 17, 18, 19, 20, \n21, 22, 23, 24, 25, 26, 27, 28, 29, 30, \n31, 32, 33, 34, 35, 36, 37, 38, 45, 46, \n47, 48, 49, 50, 51, 52, 53, 54, 55, 56, \n57, 58, 59, 60, 61, 62, 63, 64, 65, 66, \n67, 68, 69, 70, 71, \n*NSET,NSET=PhysicalSurface2\n1, 2, 3, 4, 5, 6, 7, 8, 9, 10, \n11, 12, 13, 14, 15, 16, 17, 21, 22, 23, \n24, 25, 26, 27, 28, 29, 33, 34, 35, 36, \n37, 38, 39, 40, 41, 42, 43, 44, 72, 73, \n74, 75, 76, 77, 78, 79, 80, 81, 82, 83, \n84, 85, 86, 87, 88, 89, 90, 91, 92, 93, \n94, 95, 96, 97, 98, ","category":"page"},{"location":"troubleshooting/#Troubleshooting-and-FAQ","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In general, Trixi.jl works best with the newest Julia release and up-to-date dependencies. If something does not work as expected, try updating your installed Julia packages via the package manager, e.g., by running","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> import Pkg; Pkg.update()","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"If you do not use the latest stable release of Julia from the official website, consider updating your Julia installation.","category":"page"},{"location":"troubleshooting/#old-release","page":"Troubleshooting and FAQ","title":"Installing Trixi.jl as a package only provides an older release","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Trixi.jl requires fairly recent versions of several of its dependencies, which sometimes causes issues when other installed packages have conflicting version requirements. In this case, Julia's package manager Pkg will try to handle this gracefully by going back in history until it finds a Trixi.jl release whose version requirements can be met, resulting in an older - and usually outdated - version of Trixi.jl being installed.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The following example illustrates this issue:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The current Trixi.jl release v0.3.6 requires package Foo with a minimum version of v0.2.\nAn older Trixi.jl release v0.2.1 requires package Foo only with a minimum version of v0.1.\nA user has already installed package Bar, which itself requires Foo with a maximum version of v0.1.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In this case, installing Trixi.jl via Pkg will result in version v0.2.1 to be installed instead of the current release v0.3.6. That is, a specific release of Trixi.jl may not be installable if it has a dependency with a higher minimum version that at the same time is restricted to a lower maximum version by another installed package.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"You can check whether an outdated version of Trixi.jl is installed by executing","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> import Pkg; Pkg.update(\"Trixi\"); Pkg.status(\"Trixi\")","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"in the REPL and comparing the reported Trixi.jl version with the version of the latest release. If the versions differ, you can confirm that it is due to a version conflict by forcing Pkg to install the latest Trixi.jl release, where version is the current release:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"julia> Pkg.add(name=\"Trixi\", version=\"0.3.6\")","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"In case of a conflict, the command above will produce an error that informs you about the offending packages, similar to the following:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":" Updating registry at `~/.julia/registries/General`\n Resolving package versions...\nERROR: Unsatisfiable requirements detected for package DataStructures [864edb3b]:\n DataStructures [864edb3b] log:\n ├─possible versions are: [0.9.0, 0.10.0, 0.11.0-0.11.1, 0.12.0, 0.13.0, 0.14.0-0.14.1, 0.15.0, 0.16.1, 0.17.0-0.17.20, 0.18.0-0.18.8] or uninstalled\n ├─restricted by compatibility requirements with DiffEqCallbacks [459566f4] to versions: 0.18.0-0.18.8\n │ └─DiffEqCallbacks [459566f4] log:\n │ ├─possible versions are: [2.0.0, 2.1.0, 2.2.0, 2.3.0, 2.4.0, 2.5.0-2.5.2, 2.6.0, 2.7.0, 2.8.0, 2.9.0, 2.10.0, 2.11.0, 2.12.0-2.12.1, 2.13.0-2.13.5, 2.14.0-2.14.1, 2.15.0] or uninstalled\n │ ├─restricted by compatibility requirements with Trixi [a7f1ee26] to versions: [2.14.0-2.14.1, 2.15.0]\n │ │ └─Trixi [a7f1ee26] log:\n │ │ ├─possible versions are: [0.1.0-0.1.2, 0.2.0-0.2.6, 0.3.0-0.3.6] or uninstalled\n │ │ └─restricted to versions 0.3.6 by an explicit requirement, leaving only versions 0.3.6\n │ └─restricted by compatibility requirements with StaticArrays [90137ffa] to versions: 2.15.0 or uninstalled, leaving only versions: 2.15.0\n │ └─StaticArrays [90137ffa] log:\n │ ├─possible versions are: [0.8.0-0.8.3, 0.9.0-0.9.2, 0.10.0, 0.10.2-0.10.3, 0.11.0-0.11.1, 0.12.0-0.12.5, 1.0.0-1.0.1] or uninstalled\n │ └─restricted by compatibility requirements with Trixi [a7f1ee26] to versions: 1.0.0-1.0.1\n │ └─Trixi [a7f1ee26] log: see above\n └─restricted by compatibility requirements with JLD2 [033835bb] to versions: [0.9.0, 0.10.0, 0.11.0-0.11.1, 0.12.0, 0.13.0, 0.14.0-0.14.1, 0.15.0, 0.16.1, 0.17.0-0.17.20] — no versions left\n └─JLD2 [033835bb] log:\n ├─possible versions are: [0.1.0-0.1.14, 0.2.0-0.2.4, 0.3.0-0.3.1] or uninstalled\n └─restricted by compatibility requirements with BinaryBuilder [12aac903] to versions: 0.1.0-0.1.14\n └─BinaryBuilder [12aac903] log:\n ├─possible versions are: [0.1.0-0.1.2, 0.1.4, 0.2.0-0.2.6] or uninstalled\n └─restricted to versions * by an explicit requirement, leaving only versions [0.1.0-0.1.2, 0.1.4, 0.2.0-0.2.6]","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"From the error message, we can see that ultimately BinaryBuilder is the problem here: It restricts the package DataStructures to version v0.17 (via its dependency JLD2), while Trixi.jl requires at least v0.18 (via its dependency DiffEqCallbacks). Following the official Pkg documentation, there are a number of things you can try to fix such errors:","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Try updating all packages with julia -e 'using Pkg; Pkg.update()'. A newer version of the problematic package may exist that has updated version requirements.\nRemove the offending package. Running\njulia> import Pkg; Pkg.rm(\"BinaryBuilder\"); Pkg.update(); Pkg.status()\nin the REPL will remove BinaryBuilder and (hopefully) update Trixi.jl to the latest version.\nReport the versioning issue to us and/or the development repository of the conflicting package. Maybe it is possible to lift the version restrictions such that both packages can live side by side.\nInstead of installing Trixi.jl and conflicting packages in the same (default) environment, consider creating new environments/projects and install only packages required for the specific tasks, as explained in the official Pkg documentation. For example, if you use Trixi.jl for a research project (Bachelor/Master thesis or a paper), you should create a new Julia project/environment for that research and add Trixi.jl as a dependency. If you track all your code and the Project.toml, Manifest.toml files (generated by Pkg) in a version control system such as git, you can make your research easily reproducible (if you also record the version of Julia you are using and leave some comments for others who do not know what you are trying to do, including your future self 😉).","category":"page"},{"location":"troubleshooting/#font-issues","page":"Troubleshooting and FAQ","title":"There are many questions marks and weird symbols in the output of Trixi.jl","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This probably means that the default font used by your operating system does not support enough Unicode symbols. Try installing a modern font with decent unicode support, e.g. JuliaMono. Detailed installation instructions are available there.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This problems affects users of Mac OS particularly often. At the time of writing, installing JuliaMono is as simple as","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"$ brew tap homebrew/cask-fonts\n$ brew install --cask font-juliamono","category":"page"},{"location":"troubleshooting/#There-are-no-timing-results-of-the-initial-mesh-creation","page":"Troubleshooting and FAQ","title":"There are no timing results of the initial mesh creation","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"By default, the SummaryCallback resets the timer used internally by Trixi.jl when it is initialized (when solve is called). If this step needs to be timed, e.g. to debug performance problems, explicit timings can be used as follows.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"using Trixi\n\nbegin\n Trixi.reset_timer!(Trixi.timer())\n\n equations = LinearScalarAdvectionEquation2D(0.2, -0.7)\n mesh = TreeMesh((-1.0, -1.0), (1.0, 1.0), n_cells_max=10^5, initial_refinement_level=5)\n solver = DGSEM(3)\n semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_convergence_test, solver)\n\n Trixi.print_timer(Trixi.timer())\nend","category":"page"},{"location":"troubleshooting/#MPI-ranks-are-assigned-zero-cells-in-[P4estMesh](@ref)-even-though-there-are-enough-cells","page":"Troubleshooting and FAQ","title":"MPI ranks are assigned zero cells in P4estMesh even though there are enough cells","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"The P4estMesh allows one to coarsen the mesh by default. When Trixi.jl is parallelized with multiple MPI ranks, this has the consequence that sibling cells (i.e., child cells with the same parent cell) are kept on the same MPI rank to be able to coarsen them easily. This might cause an unbalanced distribution of cells on different ranks. For 2D meshes, this also means that initially each rank will at least own 4 cells, and for 3D meshes, initially each rank will at least own 8 cells. See issue #1329.","category":"page"},{"location":"troubleshooting/#Installing-and-updating-everything-takes-a-lot-of-time","page":"Troubleshooting and FAQ","title":"Installing and updating everything takes a lot of time","text":"","category":"section"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Julia compiles code to get good (C/Fortran-like) performance. At the same time, Julia provides a dynamic environment and usually compiles code just before using it. Over time, Julia has improved its caching infrastructure, allowing to store and reuse more results from (pre-)compilation. This often results in an increased time to install/update packages, in particular when updating to Julia v1.8 or v1.9 from older versions.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"Some packages used together with Trixi.jl provide options to configure the amount of precompilation. For example, OrdinaryDiffEq.jl precompiles many ODE solvers for a good runtime experience of average users. Currently, Trixi.jl does not use all of the available solvers. Thus, you can save some time at every update by setting their precompilation options.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"At the time of writing, this could look as follows. First, you need to activate the environment where you have installed OrdinaryDiffEq.jl. Then, you need to execute the following Julia code.","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"using Preferences, UUIDs\nlet uuid = UUID(\"1dea7af3-3e70-54e6-95c3-0bf5283fa5ed\")\n set_preferences!(uuid, \"PrecompileAutoSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileAutoSwitch\" => false)\n set_preferences!(uuid, \"PrecompileDefaultSpecialize\" => true)\n set_preferences!(uuid, \"PrecompileFunctionWrapperSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileLowStorage\" => true)\n set_preferences!(uuid, \"PrecompileNoSpecialize\" => false)\n set_preferences!(uuid, \"PrecompileNonStiff\" => true)\n set_preferences!(uuid, \"PrecompileStiff\" => false)\nend","category":"page"},{"location":"troubleshooting/","page":"Troubleshooting and FAQ","title":"Troubleshooting and FAQ","text":"This disables precompilation of all implicit methods. This should usually not affect the runtime latency with Trixi.jl since most setups use explicit time integration methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"EditURL = \"../../literate/src/files/shock_capturing.jl\"","category":"page"},{"location":"tutorials/shock_capturing/#shock_capturing","page":"5 Shock capturing with flux differencing and stage limiter","title":"5: Shock capturing with flux differencing and stage limiter","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"This tutorial contains a short summary of the idea of shock capturing for DGSEM with flux differencing and its implementation in Trixi.jl. In the second part, an implementation of a positivity preserving limiter is added to the simulation.","category":"page"},{"location":"tutorials/shock_capturing/#Shock-capturing-with-flux-differencing","page":"5 Shock capturing with flux differencing and stage limiter","title":"Shock capturing with flux differencing","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"The following rough explanation is on a very basic level. More information about an entropy stable shock-capturing strategy for DGSEM discretizations of advection dominated problems, such as the compressible Euler equations or the compressible Navier-Stokes equations, can be found in Hennemann et al. (2021). In Rueda-Ramírez et al. (2021) you find the extension to the systems with non-conservative terms, such as the compressible magnetohydrodynamics (MHD) equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"The strategy for a shock-capturing method presented by Hennemann et al. is based on a hybrid blending of a high-order DG method with a low-order variant. The low-order subcell finite volume (FV) method is created directly with the Legendre-Gauss-Lobatto (LGL) nodes already used for the high-order DGSEM. Then, the final method is a convex combination with regulating indicator alpha of these two methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Since the surface integral is equal for both the DG and the subcell FV method, only the volume integral divides between the two methods.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"This strategy for the volume integral is implemented in Trixi.jl under the name of VolumeIntegralShockCapturingHG with the three parameters of the indicator and the volume fluxes for the DG and the subcell FV method.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Note, that the DG method is based on the flux differencing formulation. Hence, you have to use a two-point flux, such as flux_ranocha, flux_shima_etal, flux_chandrashekar or flux_kennedy_gruber, for the DG volume flux. We would recommend to use the entropy conserving flux flux_ranocha by Ranocha (2018) for the compressible Euler equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;\n volume_flux_dg=volume_flux_dg,\n volume_flux_fv=volume_flux_fv)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We now focus on a choice of the shock capturing indicator indicator_sc. A possible indicator is alpha_HG presented by Hennemann et al. (p.10), which depends on the current approximation with modal coefficients m_j_j=0^N of a given variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"The indicator is calculated for every DG element by itself. First, we calculate a smooth alpha by","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"alpha = frac11+exp(-frac-smathbbT(mathbbE-mathbbT))","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"with the total energy mathbbE=maxbig(fracm_N^2sum_j=0^N m_j^2 fracm_N-1^2sum_j=0^N-1 m_j^2big), threshold mathbbT= 05 * 10^-18*(N+1)^14 and parameter s=lnbig(frac1-0000100001big)approx 921024.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"For computational efficiency, alpha_min is introduced and used for","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"tildealpha = begincases\n0 textif alphaalpha_min\nalpha textif alpha_minleq alpha leq 1- alpha_min\n1 textif 1-alpha_minalpha\nendcases","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Moreover, the parameter alpha_max sets a maximal value for alpha by","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"alpha = mintildealpha alpha_max","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"This allows to control the maximal dissipation.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"To remove numerical artifact the final indicator is smoothed with all the neighboring elements' indicators. This is activated with alpha_smooth=true.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"alpha_HG = max_E alpha 05 * alpha_E","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"where E are all elements sharing a face with the current element.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Furthermore, you can specify the variable used for the calculation. For instance you can choose density, pressure or both with density_pressure for the compressible Euler equations. For every equation there is also the option to use the first conservation variable with first.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"This indicator is implemented in Trixi.jl and called IndicatorHennemannGassner with the parameters equations, basis, alpha_max, alpha_min, alpha_smooth and variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"indicator_sc = IndicatorHennemannGassner(equations, basis,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=variable)","category":"page"},{"location":"tutorials/shock_capturing/#Positivity-preserving-limiter","page":"5 Shock capturing with flux differencing and stage limiter","title":"Positivity preserving limiter","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Some numerical solutions are physically meaningless, for instance negative values of pressure or density for the compressible Euler equations. This often results in crashed simulations since the calculation of numerical fluxes or stable time steps uses mathematical operations like roots or logarithms. One option to avoid these cases are a-posteriori positivity preserving limiters. Trixi.jl provides the fully-discrete positivity-preserving limiter of Zhang, Shu (2011).","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"It works the following way. For every passed (scalar) variable and for every DG element we calculate the minimal value value_min. If this value falls below the given threshold varepsilon, the approximation is slightly adapted such that the minimal value of the relevant variable lies now above the threshold.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"underlineu^new = theta * underlineu + (1-theta) * u_mean","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"where underlineu are the collected pointwise evaluation coefficients in element e and u_mean the integral mean of the quantity in e. The new coefficients are a convex combination of these two values with factor","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"theta = fracvalue_mean - varepsilonvalue_mean - value_min","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"where value_mean is the relevant variable evaluated for the mean value u_mean.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"The adapted approximation keeps the exact same mean value, but the relevant variable is now greater or equal the threshold varepsilon at every node in every element.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We specify the variables the way we did before for the shock capturing variables. For the compressible Euler equations density, pressure or the combined variable density_pressure are a reasonable choice.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"You can implement the limiter in Trixi.jl using PositivityPreservingLimiterZhangShu with parameters threshold and variables.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"stage_limiter! = PositivityPreservingLimiterZhangShu(thresholds=thresholds,\n variables=variables)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Then, the limiter is added to the time integration method in the solve function. For instance, like","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"CarpenterKennedy2N54(stage_limiter!, williamson_condition=false)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"or","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"SSPRK43(stage_limiter!).","category":"page"},{"location":"tutorials/shock_capturing/#Simulation-with-shock-capturing-and-positivity-preserving","page":"5 Shock capturing with flux differencing and stage limiter","title":"Simulation with shock capturing and positivity preserving","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Now, we can run a simulation using the described methods of shock capturing and positivity preserving limiters. We want to give an example for the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"using OrdinaryDiffEq, Trixi\n\nequations = CompressibleEulerEquations2D(1.4)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"As our initial condition we use the Sedov blast wave setup.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"function initial_condition_sedov_blast_wave(x, t, equations::CompressibleEulerEquations2D)\n # Set up polar coordinates\n inicenter = SVector(0.0, 0.0)\n x_norm = x[1] - inicenter[1]\n y_norm = x[2] - inicenter[2]\n r = sqrt(x_norm^2 + y_norm^2)\n\n r0 = 0.21875 # = 3.5 * smallest dx (for domain length=4 and max-ref=6)\n # r0 = 0.5 # = more reasonable setup\n E = 1.0\n p0_inner = 3 * (equations.gamma - 1) * E / (3 * pi * r0^2)\n p0_outer = 1.0e-5 # = true Sedov setup\n # p0_outer = 1.0e-3 # = more reasonable setup\n\n # Calculate primitive variables\n rho = 1.0\n v1 = 0.0\n v2 = 0.0\n p = r > r0 ? p0_outer : p0_inner\n\n return prim2cons(SVector(rho, v1, v2, p), equations)\nend\ninitial_condition = initial_condition_sedov_blast_wave","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"basis = LobattoLegendreBasis(3)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We set the numerical fluxes and divide between the surface flux and the two volume fluxes for the DG and FV method. Here, we are using flux_lax_friedrichs and flux_ranocha.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"surface_flux = flux_lax_friedrichs\nvolume_flux = flux_ranocha","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Now, we specify the shock capturing indicator alpha.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We implement the described indicator of Hennemann, Gassner as explained above with parameters equations, basis, alpha_max, alpha_min, alpha_smooth and variable. Since density and pressure are the critical variables in this example, we use density_pressure = density * pressure = rho * p as indicator variable.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"indicator_sc = IndicatorHennemannGassner(equations, basis,\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable=density_pressure)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"Now, we can use the defined fluxes and the indicator to implement the volume integral using shock capturing.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;\n volume_flux_dg=volume_flux,\n volume_flux_fv=surface_flux)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We finalize the discretization by implementing Trixi.jl's solver, mesh, semi and ode, while solver now has the extra parameter volume_integral.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"solver = DGSEM(basis, surface_flux, volume_integral)\n\ncoordinates_min = (-2.0, -2.0)\ncoordinates_max = ( 2.0, 2.0)\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=6,\n n_cells_max=10_000)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)\n\ntspan = (0.0, 1.0)\node = semidiscretize(semi, tspan);\nnothing #hide","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We add some callbacks to get an solution analysis and use a CFL-based time step size calculation.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"analysis_callback = AnalysisCallback(semi, interval=100)\n\nstepsize_callback = StepsizeCallback(cfl=0.8)\n\ncallbacks = CallbackSet(analysis_callback, stepsize_callback);\nnothing #hide","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"We now run the simulation using the positivity preserving limiter of Zhang and Shu for the variables density and pressure.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"stage_limiter! = PositivityPreservingLimiterZhangShu(thresholds=(5.0e-6, 5.0e-6),\n variables=(Trixi.density, pressure))\n\nsol = solve(ode, CarpenterKennedy2N54(stage_limiter!, williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/shock_capturing/#Package-versions","page":"5 Shock capturing with flux differencing and stage limiter","title":"Package versions","text":"","category":"section"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"","category":"page"},{"location":"tutorials/shock_capturing/","page":"5 Shock capturing with flux differencing and stage limiter","title":"5 Shock capturing with flux differencing and stage limiter","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"EditURL = \"../../literate/src/files/DGMulti_1.jl\"","category":"page"},{"location":"tutorials/DGMulti_1/#DGMulti_1","page":"7 DG schemes via DGMulti solver","title":"7: DG schemes via DGMulti solver","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"DGMulti is a DG solver that allows meshes with simplex elements. The basic idea and implementation of this solver is explained in section \"Meshes\". Here, we want to give some examples and a quick overview about the options with DGMulti.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"We start with a simple example we already used in the tutorial about flux differencing. There, we implemented a simulation with initial_condition_weak_blast_wave for the 2D compressible Euler equations CompressibleEulerEquations2D and used the DG formulation with flux differencing using volume flux flux_ranocha and surface flux flux_lax_friedrichs.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"Here, we want to implement the equivalent example, only now using the DGMulti solver instead of DGSEM.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\n\nequations = CompressibleEulerEquations2D(1.4)\n\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"To use the Gauss-Lobatto nodes again, we choose approximation_type=SBP() in the solver. Since we want to start with a Cartesian domain discretized with squares, we use the element type Quad().","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Quad(),\n approximation_type = SBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"Run the simulation with the same time integration algorithm as before.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"sol = solve(ode, RDPK3SpFSAL49(), abstol=1.0e-6, reltol=1.0e-6,\n callback=callbacks, save_everystep=false);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"plot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"This simulation is not as fast as the equivalent with TreeMesh since no special optimizations for quads (for instance tensor product structure) have been implemented. Figure 4 in \"Efficient implementation of modern entropy stable and kinetic energy preserving discontinuous Galerkin methods for conservation laws\" (2021) provides a nice runtime comparison between the different mesh types. On the other hand, the functions are more general and thus we have more option we can choose from.","category":"page"},{"location":"tutorials/DGMulti_1/#Simulation-with-Gauss-nodes","page":"7 DG schemes via DGMulti solver","title":"Simulation with Gauss nodes","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"For instance, we can change the approximation type of our simulation.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\nequations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"We now use Gauss nodes instead of Gauss-Lobatto nodes which can be done for the element types Quad() and Hex(). Therefore, we set approximation_type=GaussSBP(). Alternatively, we can use a modal approach using the approximation type Polynomial().","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Quad(),\n approximation_type = GaussSBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/#Simulation-with-triangular-elements","page":"7 DG schemes via DGMulti solver","title":"Simulation with triangular elements","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"Also, we can set another element type. We want to use triangles now.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq\nequations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_weak_blast_wave","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"Since there is no direct equivalent to Gauss-Lobatto nodes on triangles, the approximation type SBP() now uses Gauss-Lobatto nodes on faces and special nodes in the interior of the triangular. More details can be found in the documentation of StartUpDG.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3,\n element_type = Tri(),\n approximation_type = SBP(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))\n\ncells_per_dimension = (32, 32)\nmesh = DGMultiMesh(dg,\n cells_per_dimension, # initial_refinement_level = 5\n coordinates_min=(-2.0, -2.0),\n coordinates_max=( 2.0, 2.0),\n periodicity=true)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n boundary_conditions=boundary_condition_periodic)\ntspan = (0.0, 0.4)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=10)\nanalysis_callback = AnalysisCallback(semi, interval=100, uEltype=real(dg))\ncallbacks = CallbackSet(analysis_callback, alive_callback);\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"plot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/#Triangular-meshes-on-non-Cartesian-domains","page":"7 DG schemes via DGMulti solver","title":"Triangular meshes on non-Cartesian domains","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"To use triangular meshes on a non-Cartesian domain, Trixi.jl uses the package StartUpDG.jl. The following example is based on elixir_euler_triangulate_pkg_mesh.jl and uses a pre-defined mesh from StartUpDG.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Trixi, OrdinaryDiffEq","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"We want to simulate the smooth initial condition initial_condition_convergence_test with source terms source_terms_convergence_test for the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"equations = CompressibleEulerEquations2D(1.4)\ninitial_condition = initial_condition_convergence_test\nsource_terms = source_terms_convergence_test","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"We create the solver DGMulti with triangular elements (Tri()) as before.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"dg = DGMulti(polydeg = 3, element_type = Tri(),\n approximation_type=Polynomial(),\n surface_flux = flux_lax_friedrichs,\n volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"StartUpDG.jl provides for instance a pre-defined Triangulate geometry for a rectangular domain with hole RectangularDomainWithHole. Other pre-defined Triangulate geometries are e.g., SquareDomain, Scramjet, and CircularDomain.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"meshIO = StartUpDG.triangulate_domain(StartUpDG.RectangularDomainWithHole());\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"The pre-defined Triangulate geometry in StartUpDG has integer boundary tags. With DGMultiMesh we assign boundary faces based on these integer boundary tags and create a mesh compatible with Trixi.jl.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"mesh = DGMultiMesh(dg, meshIO, Dict(:outer_boundary=>1, :inner_boundary=>2))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"boundary_condition_convergence_test = BoundaryConditionDirichlet(initial_condition)\nboundary_conditions = (; :outer_boundary => boundary_condition_convergence_test,\n :inner_boundary => boundary_condition_convergence_test)\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, dg,\n source_terms = source_terms,\n boundary_conditions = boundary_conditions)\n\ntspan = (0.0, 0.2)\node = semidiscretize(semi, tspan)\n\nalive_callback = AliveCallback(alive_interval=20)\nanalysis_callback = AnalysisCallback(semi, interval=200, uEltype=real(dg))\ncallbacks = CallbackSet(alive_callback, analysis_callback);\n\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt = 0.5 * estimate_dt(mesh, dg), save_everystep=false, callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using Plots\npd = PlotData2D(sol)\nplot(pd[\"rho\"])\nplot!(getmesh(pd))","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"For more information, please have a look in the StartUpDG.jl documentation.","category":"page"},{"location":"tutorials/DGMulti_1/#Package-versions","page":"7 DG schemes via DGMulti solver","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"StartUpDG\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"","category":"page"},{"location":"tutorials/DGMulti_1/","page":"7 DG schemes via DGMulti solver","title":"7 DG schemes via DGMulti solver","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"EditURL = \"../../literate/src/files/adding_new_scalar_equations.jl\"","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#adding_new_scalar_equations","page":"10 Adding a new scalar conservation law","title":"10: Adding a new scalar conservation law","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"If you want to use Trixi.jl for your own research, you might be interested in a new physics model that's not already included in Trixi.jl. In this tutorial, we will implement the cubic conservation law","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"partial_t u(tx) + partial_x u(tx)^3 = 0","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"in a periodic domain in one space dimension. In Trixi.jl, such a mathematical model is encoded as a subtype of Trixi.AbstractEquations.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Basic-setup","page":"10 Adding a new scalar conservation law","title":"Basic setup","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"using Trixi\n\nstruct CubicEquation <: Trixi.AbstractEquations{1 #= number of spatial dimensions =#,\n 1 #= number of primary variables, i.e. scalar =#};\nend","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"We create CubicEquation as an empty struct since we do not use any parameters for this equation. Other models could bundle arbitrary parameters, e.g., the ideal gas constant for the compressible Euler equations.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Next, we define the physical flux f(u) = u^3 using the calling structure used in Trixi.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Trixi.flux(u, orientation, equation::CubicEquation) = u.^3\nTrixi.varnames(_, ::CubicEquation) = (\"scalar\",)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"In Trixi.jl, the conserved variables u are usually passed as SVectors of variables at a single physical location. Hence, we must use u.^3 instead of the scalar operation u^3.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"That's already enough to run a simple simulation with a standard DGSEM discretization using the non-dissipative central flux at interfaces.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"using OrdinaryDiffEq\n\n# Create a simulation setup\nequation = CubicEquation()\n\ninitial_condition_sine(x, t, equation::CubicEquation) = SVector(sinpi(x[1]))\n\nmesh = TreeMesh(-1.0, 1.0, # min/max coordinates\n initial_refinement_level=4,\n n_cells_max=10^4)\n\nsolver = DGSEM(3 #= polynomial degree =#, flux_central)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"We wrap the return value of the initial_condition_sine inside an SVector since that's the approach used in Trixi.jl also for systems of equations. We need to index the spatial coordinate x[1], since it is an SVector with one component. In multiple space dimensions, all spatial coordinates are passed together.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Next, we create an ODEProblem from the SciML/DifferentialEquations ecosystem. We can solve this ODE numerically using any time integration method, e.g., SSPRK43 from OrdinaryDiffEq.jl. Before, we set up a callback to summarize the simulation setup.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"# Create ODE problem with given time span\ntspan = (0.0, 0.09)\node = semidiscretize(semi, tspan)\n\nsummary_callback = SummaryCallback()\ncallbacks = CallbackSet(summary_callback)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes the passed callbacks\nsol = solve(ode, SSPRK43();\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"That's it, you ran your first simulation using your new equation with Trixi.jl! Now, we can plot the solution at the final time using Plots.jl.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"You can already see that discontinuities will develop and oscillations start to occur around steep parts of the wave. That's expected from our central discretization. To avoid these issues, we need to use dissipative numerical fluxes (approximate Riemann solvers) at interfaces.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Advanced-setup","page":"10 Adding a new scalar conservation law","title":"Advanced setup","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Thus, we add a Godunov's flux for our cubic equation. That is easy for this equation since the wave speed f'(u) = 3u^2 is always non-negative.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"@inline Trixi.flux_godunov(u_ll, u_rr, orientation, equation::CubicEquation) = flux(u_ll, orientation, equation)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Let's run the example again but with a dissipative numerical flux at interfaces. remake will recreate the semidiscretization we used before and only change selected parameters, in this case the solver.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"# A new setup with dissipation\nsemi = remake(semi, solver=DGSEM(3, flux_godunov))\node = semidiscretize(semi, tspan)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot!(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"You can see that there are fewer oscillations, in particular around steep edges. Now let's increase the final time (and also the spatial resolution).","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"# A larger final time: Nonclassical shocks develop (you can even increase the refinement to 12)\nsemi = remake(semi, mesh=TreeMesh(-1.0, 1.0, initial_refinement_level=8, n_cells_max=10^5))\node = semidiscretize(semi, (0.0, 0.5) #= tspan =#)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"You can observe that nonclassical shocks develop and are stable under grid refinement, e.g. for initial_refinement_level=12. In this case, these nonclassical shocks can be avoided by using an entropy-dissipative semidiscretization. Thus, we need to define an entropy-conservative numerical flux","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"@inline function Trixi.flux_ec(u_ll, u_rr, orientation, equation::CubicEquation)\n return SVector(0.25 * (u_ll[1]^3 + u_ll[1]^2 * u_rr[1] + u_ll[1] * u_rr[1]^2 + u_rr[1]^3))\nend","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"and use a VolumeIntegralFluxDifferencing instead of the standard VolumeIntegralWeakForm in the DGSEM.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"# Let's use a provably entropy-dissipative semidiscretization\nsemi = remake(semi, solver=DGSEM(3, flux_godunov, VolumeIntegralFluxDifferencing(flux_ec)))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...);\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"Possible next steps could be","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"to define Trixi.max_abs_speeds(u, equations::CubicEquation) = 3 * u[1]^2 to use CFL-based time step control via a StepsizeCallback\nto define quantities of interest like Trixi.entropy(u, equations::CubicEquation) = u[1]^2 and integrate them in a simulation using the AnalysisCallback\nto experiment with shock-capturing volume integrals VolumeIntegralShockCapturingHG and adaptive mesh refinement AMRCallback","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"For further reading, Trixi.jl provides another example on adding a scalar equation. In the elixir about the KPP problem, the 2D scalar \"KPP equation\" from Kurganov, Petrova, Popov (2007) is implemented.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Summary-of-the-code","page":"10 Adding a new scalar conservation law","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"To sum up, here is the complete code that we used (without the callbacks since these create a lot of unnecessary output in the doctests of this tutorial). In addition, we create the struct inside the new module CubicConservationLaw. That ensures that we can re-create structs defined therein without having to restart Julia.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"# Define new physics\nmodule CubicConservationLaw\n\nusing Trixi\n\nstruct CubicEquation <: Trixi.AbstractEquations{1 #= number of spatial dimensions =#,\n 1 #= number of primary variables, i.e. scalar =#}\nend\n\n@inline Trixi.flux(u, orientation, equation::CubicEquation) = u.^3\nTrixi.varnames(_, ::CubicEquation) = (\"scalar\",)\n\n@inline Trixi.flux_godunov(u_ll, u_rr, orientation, equation::CubicEquation) = flux(u_ll, orientation, equation)\n@inline function Trixi.flux_ec(u_ll, u_rr, orientation, equation::CubicEquation)\n return SVector(0.25 * (u_ll[1]^3 + u_ll[1]^2 * u_rr[1] + u_ll[1] * u_rr[1]^2 + u_rr[1]^3))\nend\n\nend # module\n\n\n# Create a simulation setup\nimport .CubicConservationLaw\nusing Trixi\nusing OrdinaryDiffEq\nusing Plots\n\nequation = CubicConservationLaw.CubicEquation()\n\ninitial_condition_sine(x, t, equation::CubicConservationLaw.CubicEquation) = SVector(sinpi(x[1]))\n\nmesh = TreeMesh(-1.0, 1.0, # min/max coordinates\n initial_refinement_level=4,\n n_cells_max=10^4)\n\nsolver = DGSEM(3 #= polynomial degree =#, flux_central)\n\nsemi = SemidiscretizationHyperbolic(mesh, equation, initial_condition_sine, solver)\n\n# Create ODE problem with given time span\ntspan = (0.0, 0.1)\node = semidiscretize(semi, tspan)\n\n# OrdinaryDiffEq's `solve` method evolves the solution in time and executes the passed callbacks\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)\n\n\n# A new setup with dissipation\nsemi = remake(semi, solver=DGSEM(3, flux_godunov))\node = semidiscretize(semi, tspan)\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot!(sol)\n\n\n# A larger final time: Nonclassical shocks develop (you can even increase the refinement to 12)\nsemi = remake(semi, mesh=TreeMesh(-1.0, 1.0, initial_refinement_level=8, n_cells_max=10^5))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)\n\n\n# Let's use a provably entropy-dissipative semidiscretization\nsemi = remake(semi, solver=DGSEM(3, flux_godunov, VolumeIntegralFluxDifferencing(flux_ec)))\node = semidiscretize(semi, (0.0, 0.5))\nsol = solve(ode, SSPRK43(); ode_default_options()...)\nplot(sol)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/#Package-versions","page":"10 Adding a new scalar conservation law","title":"Package versions","text":"","category":"section"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"","category":"page"},{"location":"tutorials/adding_new_scalar_equations/","page":"10 Adding a new scalar conservation law","title":"10 Adding a new scalar conservation law","text":"This page was generated using Literate.jl.","category":"page"},{"location":"time_integration/#time-integration","page":"Time integration","title":"Time integration methods","text":"","category":"section"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, explicit Runge-Kutta methods from OrdinaryDiffEq.jl are tested extensively. Interesting classes of time integration schemes are","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Explicit low-storage Runge-Kutta methods\nStrong stability preserving methods","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"Some common options for solve from OrdinaryDiffEq.jl are the following. Further documentation can be found in the SciML docs.","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"If you use a fixed time step method like CarpenterKennedy2N54, you need to pass a time step as dt=.... If you use a StepsizeCallback, the value passed as dt=... is irrelevant since it will be overwritten by the StepsizeCallback. If you want to use an adaptive time step method such as SSPRK43 or RDPK3SpFSAL49 and still want to use CFL-based step size control via the StepsizeCallback, you need to pass the keyword argument adaptive=false to solve.\nYou should usually set save_everystep=false. Otherwise, OrdinaryDiffEq.jl will (try to) save the numerical solution after every time step in RAM (until you run out of memory or start to swap).\nYou can set the maximal number of time steps via maxiters=....\nSSP methods and many low-storage methods from OrdinaryDiffEq.jl support stage_limiter!s and step_limiter!s, e.g., PositivityPreservingLimiterZhangShu from Trixi.jl.\nIf you start Julia with multiple threads and want to use them also in the time integration method from OrdinaryDiffEq.jl, you need to pass the keyword argument thread=OrdinaryDiffEq.True() to the algorithm, e.g., RDPK3SpFSAL49(thread=OrdinaryDiffEq.True()) or CarpenterKennedy2N54(thread=OrdinaryDiffEq.True(), williamson_condition=false). For more information on using thread-based parallelism in Trixi.jl, please refer to Shared-memory parallelization with threads.\nIf you use error-based step size control (see also the section on error-based adaptive step sizes together with MPI, you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.","category":"page"},{"location":"time_integration/","page":"Time integration","title":"Time integration","text":"note: Number of `rhs!` calls\nIf you use explicit Runge-Kutta methods from OrdinaryDiffEq.jl, the total number of rhs! calls can be (slightly) bigger than the number of steps times the number of stages, e.g. to allow for interpolation (dense output), root-finding for continuous callbacks, and error-based time step control. In general, you often should not need to worry about this if you use Trixi.jl.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"EditURL = \"../../../literate/src/files/first_steps/getting_started.jl\"","category":"page"},{"location":"tutorials/first_steps/getting_started/#getting_started","page":"1.1 Getting started","title":"1.1: First steps in Trixi.jl: Getting started","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Trixi.jl is a numerical simulation framework for conservation laws and is written in the Julia programming language. This tutorial is intended for beginners in Julia and Trixi.jl. After reading it, you will know how to install Julia and Trixi.jl on your computer, and you will be able to download setup files from our GitHub repository, modify them, and run simulations.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"The contents of this tutorial:","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Julia installation\nTrixi.jl installation\nRunning a simulation\nGetting an existing setup file\nModifying an existing setup","category":"page"},{"location":"tutorials/first_steps/getting_started/#Julia-installation","page":"1.1 Getting started","title":"Julia installation","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Trixi.jl is compatible with the latest stable release of Julia. Additional details regarding Julia support can be found in the README.md file. The current default Julia installation is managed through juliaup. You may follow our concise installation guidelines for Windows, Linux, and MacOS provided below. In the event of any issues during the installation process, please consult the official Julia installation instruction.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Windows","page":"1.1 Getting started","title":"Windows","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Open a terminal by pressing Win+r and entering cmd in the opened window.\nTo install Julia, execute the following command in the terminal:\nwinget install julia -s msstore\nVerify the successful installation of Julia by executing the following command in the terminal:\njulia\nTo exit Julia, execute exit() or press Ctrl+d.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Linux-and-MacOS","page":"1.1 Getting started","title":"Linux and MacOS","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To install Julia, run the following command in a terminal:\ncurl -fsSL https://install.julialang.org | sh\nFollow the instructions displayed in the terminal during the installation process.\nIf an error occurs during the execution of the previous command, you may need to install curl. On Ubuntu-type systems, you can use the following command:\nsudo apt install curl\nAfter installing curl, repeat the first step once more to proceed with Julia installation.\nVerify the successful installation of Julia by executing the following command in the terminal:\njulia\nTo exit Julia, execute exit() or press Ctrl+d.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Trixi.jl-installation","page":"1.1 Getting started","title":"Trixi.jl installation","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Trixi.jl and its related tools are registered Julia packages, thus their installation happens inside Julia. For a smooth workflow experience with Trixi.jl, you need to install Trixi.jl, OrdinaryDiffEq.jl, and Plots.jl.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Open a terminal and start Julia.\nExecute following commands:\nimport Pkg\nPkg.add([\"OrdinaryDiffEq\", \"Plots\", \"Trixi\"])","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Now you have installed all these packages. OrdinaryDiffEq.jl provides time integration schemes used by Trixi.jl and Plots.jl can be used to directly visualize Trixi.jl results from the Julia REPL.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Usage","page":"1.1 Getting started","title":"Usage","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/#Running-a-simulation","page":"1.1 Getting started","title":"Running a simulation","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To get you started, Trixi.jl has a large set of example setups, that can be taken as a basis for your future investigations. In Trixi.jl, we call these setup files \"elixirs\", since they contain Julia code that takes parts of Trixi.jl and combines them into something new.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Any of the examples can be executed using the trixi_include function. trixi_include(...) expects a single string argument with a path to a file containing Julia code. For convenience, the examples_dir function returns a path to the examples folder, which has been locally downloaded while installing Trixi.jl. joinpath(...) can be used to join path components into a full path.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Let's execute a short two-dimensional problem setup. It approximates the solution of the compressible Euler equations in 2D for an ideal gas (CompressibleEulerEquations2D) with a weak blast wave as the initial condition.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Start Julia in a terminal and execute the following code:","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"using Trixi, OrdinaryDiffEq\ntrixi_include(joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_euler_ec.jl\"))","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"using Trixi, OrdinaryDiffEq #hide\ntrixi_include(@__MODULE__,joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_euler_ec.jl\")) #hide","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To analyze the result of the computation, we can use the Plots.jl package and the function plot(...), which creates a graphical representation of the solution. sol is a variable defined in the executed example and it contains the solution at the final moment of the simulation.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To obtain a list of all Trixi.jl elixirs execute get_examples. It returns the paths to all example setups.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"get_examples()","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Editing an existing elixir is the best way to start your first own investigation using Trixi.jl.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Getting-an-existing-setup-file","page":"1.1 Getting started","title":"Getting an existing setup file","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To edit an existing elixir, you first have to find a suitable one and then copy it to a local folder. Let's have a look at how to download the elixir_euler_ec.jl elixir used in the previous section from the Trixi.jl GitHub repository.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"All examples are located inside the examples folder.\nNavigate to the file elixir_euler_ec.jl.\nRight-click the Raw button on the right side of the webpage and choose Save as... (or Save Link As...).\nChoose a folder and save the file.","category":"page"},{"location":"tutorials/first_steps/getting_started/#Modifying-an-existing-setup","page":"1.1 Getting started","title":"Modifying an existing setup","text":"","category":"section"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"As an example, we will change the initial condition for calculations that occur in elixir_euler_ec.jl. In this example we consider the compressible Euler equations in two spatial dimensions,","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"fracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1 rho v_1^2 + p rho v_1 v_2 (rho e + p) v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + p (rho e + p) v_2\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0\nendpmatrix","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"for an ideal gas with the specific heat ratio gamma. Here, rho is the density, v_1 and v_2 are the velocities, e is the specific total energy, and","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"p = (gamma - 1) left( rho e - frac12 rho (v_1^2 + v_2^2) right)","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"is the pressure. Initial conditions consist of initial values for rho, rho v_1, rho v_2 and rho e. One of the common initial conditions for the compressible Euler equations is a simple density wave. Let's implement it.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Open the downloaded file elixir_euler_ec.jl with a text editor.\nGo to the line with the following code:\ninitial_condition = initial_condition_weak_blast_wave\nHere, initial_condition_weak_blast_wave is used as the initial condition.\nComment out the line using the # symbol:\n# initial_condition = initial_condition_weak_blast_wave\nNow you can create your own initial conditions. Add the following code after the commented line:","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"function initial_condition_density_waves(x, t, equations::CompressibleEulerEquations2D)\n v1 = 0.1 # velocity along x-axis\n v2 = 0.2 # velocity along y-axis\n rho = 1.0 + 0.98 * sinpi(sum(x) - t * (v1 + v2)) # density wave profile\n p = 20 # pressure\n rho_e = p / (equations.gamma - 1) + 1/2 * rho * (v1^2 + v2^2)\n return SVector(rho, rho*v1, rho*v2, rho_e)\nend\ninitial_condition = initial_condition_density_waves","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Execute the following code one more time, but instead of path/to/file paste the path to the elixir_euler_ec.jl file that you just edited.\nusing Trixi\ntrixi_include(path/to/file)\nusing Plots\nplot(sol)","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Then you will obtain a new solution from running the simulation with a different initial condition.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"trixi_include(@__MODULE__,joinpath(examples_dir(), \"tree_2d_dgsem\", \"elixir_euler_ec.jl\"), #hide\n initial_condition=initial_condition) #hide\npd = PlotData2D(sol) #hide\np1 = plot(pd[\"rho\"]) #hide\np2 = plot(pd[\"v1\"], clim=(0.05, 0.15)) #hide\np3 = plot(pd[\"v2\"], clim=(0.15, 0.25)) #hide\np4 = plot(pd[\"p\"], clim=(10, 30)) #hide\nplot(p1, p2, p3, p4) #hide","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"To get exactly the same picture execute the following.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"pd = PlotData2D(sol)\np1 = plot(pd[\"rho\"])\np2 = plot(pd[\"v1\"], clim=(0.05, 0.15))\np3 = plot(pd[\"v2\"], clim=(0.15, 0.25))\np4 = plot(pd[\"p\"], clim=(10, 30))\nplot(p1, p2, p3, p4)","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Feel free to make further changes to the initial condition to observe different solutions.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Now you are able to download, modify and execute simulation setups for Trixi.jl. To explore further details on setting up a new simulation with Trixi.jl, refer to the second part of the introduction titled Create first setup.","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"Sys.rm(\"out\"; recursive=true, force=true) #hide","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"","category":"page"},{"location":"tutorials/first_steps/getting_started/","page":"1.1 Getting started","title":"1.1 Getting started","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"EditURL = \"../../literate/src/files/upwind_fdsbp.jl\"","category":"page"},{"location":"tutorials/upwind_fdsbp/#upwind_fdsbp","page":"9 Upwind FD SBP schemes","title":"9: Upwind FD SBP schemes","text":"","category":"section"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"General tensor product SBP methods are supported via the DGMulti solver in a reasonably complete way, see the previous tutorial. Nevertheless, there is also experimental support for SBP methods with other solver and mesh types.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"The first step is to set up an SBP operator. A classical (central) SBP operator can be created as follows.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"using Trixi\nD_SBP = derivative_operator(SummationByPartsOperators.MattssonNordström2004(),\n derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Instead of prefixing the source of coefficients MattssonNordström2004(), you can also load the package SummationByPartsOperators.jl. Either way, this yields an object representing the operator efficiently. If you want to compare it to coefficients presented in the literature, you can convert it to a matrix.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Matrix(D_SBP)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Upwind SBP operators are a concept introduced in 2017 by Ken Mattsson. You can create them as follows.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"D_upw = upwind_operators(SummationByPartsOperators.Mattsson2017,\n derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Upwind operators are derivative operators biased towards one direction. The \"minus\" variants has a bias towards the left side, i.e., it uses values from more nodes to the left than from the right to compute the discrete derivative approximation at a given node (in the interior of the domain). In matrix form, this means more non-zero entries are left from the diagonal.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Matrix(D_upw.minus)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Analogously, the \"plus\" variant has a bias towards the right side.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Matrix(D_upw.plus)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"For more information on upwind SBP operators, please refer to the documentation of SummationByPartsOperators.jl and references cited there.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"The basic idea of upwind SBP schemes is to apply a flux vector splitting and use appropriate upwind operators for both parts of the flux. In 1D, this means to split the flux","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"f(u) = f^-(u) + f^+(u)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"such that f^-(u) is associated with left-going waves and f^+(u) with right-going waves. Then, we apply upwind SBP operators D^- D^+ with an appropriate upwind bias, resulting in","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"partial_x f(u) approx D^+ f^-(u) + D^- f^+(u)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Note that the established notations of upwind operators D^pm and flux splittings f^pm clash. The right-going waves from f^+ need an operator biased towards their upwind side, i.e., the left side. This upwind bias is provided by the operator D^-.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Many classical flux vector splittings have been developed for finite volume methods and are described in the book \"Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction\" of Eleuterio F. Toro (2009), DOI: 10.1007/b79761. One such a well-known splitting provided by Trixi.jl is splitting_steger_warming.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"Trixi.jl comes with several example setups using upwind SBP methods with flux vector splitting, e.g.,","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"elixir_euler_vortex.jl\nelixir_euler_taylor_green_vortex.jl","category":"page"},{"location":"tutorials/upwind_fdsbp/#Package-versions","page":"9 Upwind FD SBP schemes","title":"Package versions","text":"","category":"section"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"SummationByPartsOperators\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"","category":"page"},{"location":"tutorials/upwind_fdsbp/","page":"9 Upwind FD SBP schemes","title":"9 Upwind FD SBP schemes","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"EditURL = \"../../literate/src/files/hohqmesh_tutorial.jl\"","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#hohqmesh_tutorial","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16: Unstructured meshes with HOHQMesh.jl","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Trixi.jl supports numerical approximations on unstructured quadrilateral meshes with the UnstructuredMesh2D mesh type.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The purpose of this tutorial is to demonstrate how to use the UnstructuredMesh2D functionality of Trixi.jl. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will demonstrate how to conceptualize a problem with curved boundaries, generate a curvilinear mesh using the available software in the Trixi.jl ecosystem, and then run a simulation using Trixi.jl on said mesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Unstructured quadrilateral meshes can be made with the High-Order Hex-Quad Mesh (HOHQMesh) generator created and developed by David Kopriva. HOHQMesh is a mesh generator specifically designed for spectral element methods. It provides high-order boundary curve information (needed to accurately set boundary conditions) and elements can be larger (due to the high accuracy of the spatial approximation) compared to traditional finite element mesh generators. For more information about the design and features of HOHQMesh one can refer to its official documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"HOHQMesh is incorporated into the Trixi.jl framework via the registered Julia package HOHQMesh.jl. This package provides a Julia wrapper for the HOHQMesh generator that allows users to easily create mesh files without the need to build HOHQMesh from source. To install the HOHQMesh package execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"import Pkg; Pkg.add(\"HOHQMesh\")","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Now we are ready to generate an unstructured quadrilateral mesh that can be used by Trixi.jl.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Running-and-visualizing-an-unstructured-simulation","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Running and visualizing an unstructured simulation","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Trixi.jl supports solving hyperbolic problems on several mesh types. There is a default example for this mesh type that can be executed by","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using Trixi\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi_include(default_example_unstructured())\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"This will compute a smooth, manufactured solution test case for the 2D compressible Euler equations on the curved quadrilateral mesh described in the Trixi.jl documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Apart from the usual error and timing output provided by the Trixi.jl run, it is useful to visualize and inspect the solution. One option available in the Trixi.jl framework to visualize the solution on unstructured quadrilateral meshes is post-processing the Trixi.jl output file(s) with the Trixi2Vtk tool and plotting them with ParaView.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"To convert the HDF5-formatted .h5 output file(s) from Trixi.jl into VTK format execute the following","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using Trixi2Vtk\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000180.h5\", output_directory=\"out\")\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Note this step takes about 15-30 seconds as the package Trixi2Vtk must be precompiled and executed for the first time in your REPL session. The trixi2vtk command above will convert the solution file at the final time into a .vtu file which can be read in and visualized with ParaView. Optional arguments for trixi2vtk are: (1) Pointing to the output_directory where the new files will be saved; it defaults to the current directory. (2) Specifying a higher number of visualization nodes. For instance, if we want to use 12 uniformly spaced nodes for visualization we can execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"redirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000180.h5\", output_directory=\"out\", nvisnodes=12)\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"By default trixi2vtk sets nvisnodes to be the same as the number of nodes specified in the elixir file used to run the simulation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Finally, if you want to convert all the solution files to VTK execute","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"redirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\ntrixi2vtk(\"out/solution_000*.h5\", output_directory=\"out\", nvisnodes=12)\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"then it is possible to open the .pvd file with ParaView and create a video of the simulation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Creating-a-mesh-using-HOHQMesh","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Creating a mesh using HOHQMesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The creation of an unstructured quadrilateral mesh using HOHQMesh.jl is driven by a control file. In this file the user dictates the domain to be meshed, prescribes any desired boundary curvature, the polynomial order of said boundaries, etc. In this tutorial we cover several basic features of the possible control inputs. For a complete discussion on this topic see the HOHQMesh control file documentation.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"To begin, we provide a complete control file in this tutorial. After this we give a breakdown of the control file components to explain the chosen parameters.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Suppose we want to create a mesh of a domain with straight sided outer boundaries and a curvilinear \"ice cream cone\" shaped object at its center.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_boundary_cartoon)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The associated ice_cream_straight_sides.control file is created below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"open(\"out/ice_cream_straight_sides.control\", \"w\") do io\n println(io, raw\"\"\"\n\\begin{CONTROL_INPUT}\n \\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_straight_sides.mesh\n plot file name = ice_cream_straight_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n \\end{RUN_PARAMETERS}\n\n \\begin{BACKGROUND_GRID}\n x0 = [-8.0, -8.0, 0.0]\n dx = [1.0, 1.0, 0.0]\n N = [16,16,1]\n \\end{BACKGROUND_GRID}\n\n \\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n \\end{SPRING_SMOOTHER}\n\n\\end{CONTROL_INPUT}\n\n\\begin{MODEL}\n\n \\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n \\end{INNER_BOUNDARIES}\n\n\\end{MODEL}\n\\end{FILE}\n\"\"\")\nend","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The first three blocks of information are wrapped within a CONTROL_INPUT environment block as they define the core components of the quadrilateral mesh that will be generated.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The first block of information in RUN_PARAMETERS is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"\\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_straight_sides.mesh\n plot file name = ice_cream_straight_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n\\end{RUN_PARAMETERS}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The mesh and plot file names will be the files created by HOHQMesh once successfully executed. The stats file name is available if you wish to also save a collection of mesh statistics. For this example it is deactivated. These file names given within RUN_PARAMETERS should match that of the control file, and although this is not required by HOHQMesh, it is a useful style convention. The mesh file format ISM-v2 in the format currently required by Trixi.jl. The polynomial order prescribes the order of an interpolant constructed on the Chebyshev-Gauss-Lobatto nodes that is used to represent any curved boundaries on a particular element. The plot file format of skeleton means that visualizing the plot file will only draw the element boundaries (and no internal nodes). Alternatively, the format can be set to sem to visualize the interior nodes of the approximation as well.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The second block of information in BACKGROUND_GRID is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"\\begin{BACKGROUND_GRID}\n x0 = [-8.0, -8.0, 0.0]\n dx = [1.0, 1.0, 0.0]\n N = [16,16,1]\n\\end{BACKGROUND_GRID}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"This lays a grid of Cartesian elements for the domain beginning at the point x0 as its bottom-left corner. The value of dx, which could differ in each direction if desired, controls the step size taken in each Cartesian direction. The values in N set how many Cartesian box elements are set in each coordinate direction. The above parameters define a 16times 16 element square mesh on -88^2. Further, this sets up four outer boundaries of the domain that are given the default names: Top, Left, Bottom, Right.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The third block of information in SPRING_SMOOTHER is","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"\\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n\\end{SPRING_SMOOTHER}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Once HOHQMesh generates the mesh, a spring-mass-dashpot model is created to smooth the mesh and create \"nicer\" quadrilateral elements. The default parameters of Hooke's law for the spring-mass-dashpot model have been selected after a fair amount of experimentation across many meshes. If you wish to deactivate this feature you can set smoothing = OFF (or remove this block from the control file).","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"After the CONTROL_INPUT environment block comes the MODEL environment block. It is here where the user prescribes curved boundary information with either:","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"An OUTER_BOUNDARY (covered in the next section of this tutorial).\nOne or more INNER_BOUNDARIES.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"There are several options to describe the boundary curve data to HOHQMesh like splines or parametric curves.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"For the example ice_cream_straight_sides.control we define three internal boundaries; two straight-sided and one as a circular arc. Within the HOHQMesh control input each curve must be assigned to a CHAIN as shown below in the complete INNER_BOUNDARIES block.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"\\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n\\end{INNER_BOUNDARIES}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"It is important to note there are two name quantities one for the CHAIN and one for the PARAMETRIC_EQUATION_CURVE. The name for the CHAIN is used internally by HOHQMesh, so if you have multiple CHAINs they must be given a unique name. The name for the PARAMETRIC_EQUATION_CURVE will be printed to the appropriate boundaries within the .mesh file produced by HOHQMesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We create the mesh file ice_cream_straight_sides.mesh and its associated file for plotting ice_cream_straight_sides.tec by using HOHQMesh.jl's function generate_mesh.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using HOHQMesh\ncontrol_file = joinpath(\"out\", \"ice_cream_straight_sides.control\")\noutput = generate_mesh(control_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The mesh file ice_cream_straight_sides.mesh and its associated file for plotting ice_cream_straight_sides.tec are placed in the out folder. The resulting mesh generated by HOHQMesh.jl is given in the following figure.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_straight_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We note that Trixi.jl uses the boundary name information from the control file to assign boundary conditions in an elixir file. Therefore, the name should start with a letter and consist only of alphanumeric characters and underscores. Please note that the name will be treated as case sensitive.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Example-simulation-on-ice_cream_straight_sides.mesh","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Example simulation on ice_cream_straight_sides.mesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"With this newly generated mesh we are ready to run a Trixi.jl simulation on an unstructured quadrilateral mesh. For this we must create a new elixir file.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The elixir file given below creates an initial condition for a uniform background flow state with a free stream Mach number of 0.3. A focus for this part of the tutorial is to specify the boundary conditions and to construct the new mesh from the file that was generated in the previous exercise.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"It is straightforward to set the different boundary condition types in an elixir by assigning a particular function to a boundary name inside a Julia dictionary, Dict, variable. Observe that the names of these boundaries match those provided by HOHQMesh either by default, e.g. Bottom, or user assigned, e.g. IceCream. For this problem setup use","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Freestream boundary conditions on the four box edges.\nFree slip wall boundary condition on the interior curved boundaries.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"To construct the unstructured quadrilateral mesh from the HOHQMesh file we point to the appropriate location with the variable mesh_file and then feed this into the constructor for the UnstructuredMesh2D type in Trixi.jl","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"# create the unstructured mesh from your mesh file\nusing Trixi\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file);","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The complete elixir file for this simulation example is given below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using OrdinaryDiffEq, Trixi\n\nequations = CompressibleEulerEquations2D(1.4) # set gas gamma = 1.4\n\n# freestream flow state with Ma_inf = 0.3\n@inline function uniform_flow_state(x, t, equations::CompressibleEulerEquations2D)\n\n # set the freestream flow parameters\n rho_freestream = 1.0\n u_freestream = 0.3\n p_freestream = inv(equations.gamma)\n\n theta = 0.0 # zero angle of attack\n si, co = sincos(theta)\n v1 = u_freestream * co\n v2 = u_freestream * si\n\n prim = SVector(rho_freestream, v1, v2, p_freestream)\n return prim2cons(prim, equations)\nend\n\n# initial condition\ninitial_condition = uniform_flow_state\n\n# boundary condition types\nboundary_condition_uniform_flow = BoundaryConditionDirichlet(uniform_flow_state)\n\n# boundary condition dictionary\nboundary_conditions = Dict( :Bottom => boundary_condition_uniform_flow,\n :Top => boundary_condition_uniform_flow,\n :Right => boundary_condition_uniform_flow,\n :Left => boundary_condition_uniform_flow,\n :LeftSlant => boundary_condition_slip_wall,\n :RightSlant => boundary_condition_slip_wall,\n :IceCream => boundary_condition_slip_wall );\n\n# DGSEM solver.\n# 1) polydeg must be >= the polynomial order set in the HOHQMesh control file to guarantee\n# freestream preservation. As a extra task try setting polydeg=3\n# 2) VolumeIntegralFluxDifferencing with central volume flux is activated\n# for dealiasing\nvolume_flux = flux_ranocha\nsolver = DGSEM(polydeg=4, surface_flux=flux_hll,\n volume_integral=VolumeIntegralFluxDifferencing(volume_flux))\n\n# create the unstructured mesh from your mesh file\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file)\n\n# Create semidiscretization with all spatial discretization-related components\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,\n boundary_conditions=boundary_conditions)\n\n# Create ODE problem from semidiscretization with time span from 0.0 to 2.0\ntspan = (0.0, 2.0)\node = semidiscretize(semi, tspan)\n\n\n# Create the callbacks to output solution files and adapt the time step\nsummary_callback = SummaryCallback()\nsave_solution = SaveSolutionCallback(interval=10,\n save_initial_solution=true,\n save_final_solution=true)\nstepsize_callback = StepsizeCallback(cfl=1.0)\n\ncallbacks = CallbackSet(summary_callback, save_solution, stepsize_callback)\n\nredirect_stdio(stdout=devnull, stderr=devnull) do # code that prints annoying stuff we don't want to see here #hide\n# Evolve ODE problem in time using `solve` from OrdinaryDiffEq\nsol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),\n dt=1.0, # solve needs some value here but it will be overwritten by the stepsize_callback\n save_everystep=false, callback=callbacks);\n# print the timer summary\nsummary_callback()\nend #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Visualization of the solution is carried out in a similar way as above. That is, one converts the .h5 output files with trixi2vtk and then plot the solution in ParaView. An example plot of the pressure at the final time is shown below.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_straight_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Making-a-mesh-with-a-curved-outer-boundary","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Making a mesh with a curved outer boundary","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Let us modify the mesh from the previous task and place a circular outer boundary instead of straight-sided outer boundaries. Note, the \"ice cream cone\" shape is still placed at the center of the domain.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We create the new control file ice_cream_curved_sides.control file below and will then highlight the major differences compared to ice_cream_straight_sides.control.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"open(\"out/ice_cream_curved_sides.control\", \"w\") do io\n println(io, raw\"\"\"\n\\begin{CONTROL_INPUT}\n \\begin{RUN_PARAMETERS}\n mesh file name = ice_cream_curved_sides.mesh\n plot file name = ice_cream_curved_sides.tec\n stats file name = none\n mesh file format = ISM-v2\n polynomial order = 4\n plot file format = skeleton\n \\end{RUN_PARAMETERS}\n\n \\begin{BACKGROUND_GRID}\n background grid size = [1.0, 1.0, 0.0]\n \\end{BACKGROUND_GRID}\n\n \\begin{SPRING_SMOOTHER}\n smoothing = ON\n smoothing type = LinearAndCrossBarSpring\n number of iterations = 25\n \\end{SPRING_SMOOTHER}\n\n\\end{CONTROL_INPUT}\n\n\\begin{MODEL}\n\n \\begin{OUTER_BOUNDARY}\n \\begin{PARAMETRIC_EQUATION_CURVE}\n name = OuterCircle\n xEqn = x(t) = 8.0*sin(2.0*pi*t)\n yEqn = y(t) = 8.0*cos(2.0*pi*t)\n zEqn = z(t) = 0.0\n \\end{PARAMETRIC_EQUATION_CURVE}\n\n \\end{OUTER_BOUNDARY}\n\n \\begin{INNER_BOUNDARIES}\n\n \\begin{CHAIN}\n name = IceCreamCone\n \\begin{END_POINTS_LINE}\n name = LeftSlant\n xStart = [-2.0, 1.0, 0.0]\n xEnd = [ 0.0, -3.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{END_POINTS_LINE}\n name = RightSlant\n xStart = [ 0.0, -3.0, 0.0]\n xEnd = [ 2.0, 1.0, 0.0]\n \\end{END_POINTS_LINE}\n\n \\begin{CIRCULAR_ARC}\n name = IceCream\n units = degrees\n center = [ 0.0, 1.0, 0.0]\n radius = 2.0\n start angle = 0.0\n end angle = 180.0\n \\end{CIRCULAR_ARC}\n \\end{CHAIN}\n\n \\end{INNER_BOUNDARIES}\n\n\\end{MODEL}\n\\end{FILE}\n\"\"\")\nend","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The first alteration is that we have altered the second block of information BACKGROUND_GRID within the CONTROL_INPUT to be","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"\\begin{BACKGROUND_GRID}\n background grid size = [1.0, 1.0, 0.0]\n\\end{BACKGROUND_GRID}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"This mesh control file has an outer boundary that determines the extent of the domain to be meshed. Therefore, we only need to supply the background grid size to the BACKGROUND_GRID control input.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The second alteration is that the MODEL now contains information for an OUTER_BOUNDARY. In this case it is a circle of radius 8 centered at [0.0, 0.0, 0.0] written as a set of PARAMETRIC_EQUATION_CURVEs.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":" \\begin{OUTER_BOUNDARY}\n\n \\begin{PARAMETRIC_EQUATION_CURVE}\n name = OuterCircle\n xEqn = x(t) = 8.0*sin(2.0*pi*t)\n yEqn = y(t) = 8.0*cos(2.0*pi*t)\n zEqn = z(t) = 0.0\n \\end{PARAMETRIC_EQUATION_CURVE}\n\n \\end{OUTER_BOUNDARY}","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Just as with the inner boundary curves, we must assign a name to the OUTER_BOUNDARY. It will be included in the generated .mesh file and is used within the Trixi.jl elixir file to set boundary conditions.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Again, we create the .mesh and .tec files with HOHQMesh.jl's function generate_mesh","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"control_file = joinpath(\"out\", \"ice_cream_curved_sides.control\")\noutput = generate_mesh(control_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The files are placed in the out folder.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The resulting mesh generated by HOHQMesh.jl is given in the following figure.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: mesh_curved_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Running-Trixi.jl-on-ice_cream_curved_sides.mesh","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Running Trixi.jl on ice_cream_curved_sides.mesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We can reuse much of the elixir file to setup the uniform flow over an ice cream cone from the previous part of this tutorial. The only component of the elixir file that must be changed is the boundary condition dictionary because we now have a boundary named OuterCircle instead of four edges of a bounding box.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"# boundary condition dictionary\nboundary_conditions = Dict( :OuterCircle => boundary_condition_uniform_flow,\n :LeftSlant => boundary_condition_slip_wall,\n :RightSlant => boundary_condition_slip_wall,\n :IceCream => boundary_condition_slip_wall );\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Also, we must update the construction of the mesh from our new mesh file ice_cream_curved_sides.mesh that is located in the out folder.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"# create the unstructured mesh from your mesh file\nmesh_file = joinpath(\"out\", \"ice_cream_curved_sides.mesh\")\nmesh = UnstructuredMesh2D(mesh_file);\nnothing #hide","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We can then post-process the solution file at the final time on the new mesh with Trixi2Vtk and visualize with ParaView.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_curved_sides)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Setting-up-a-simulation-with-AMR-via-P4estMesh","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Setting up a simulation with AMR via P4estMesh","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"The above explained mesh file format of ISM-V2 only works with UnstructuredMesh2D and so does not support AMR. On the other hand, the mesh type P4estMesh allows AMR. The mesh constructor for the P4estMesh imports an unstructured, conforming mesh from an Abaqus mesh file (.inp).","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"As described above, the first block of the HOHQMesh control file contains the parameter mesh file format. If you set mesh file format = ABAQUS instead of ISM-V2, HOHQMesh.jl's function generate_mesh creates an Abaqus mesh file .inp.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using HOHQMesh\ncontrol_file = joinpath(\"out\", \"ice_cream_straight_sides.control\")\noutput = generate_mesh(control_file);","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Now, you can create a P4estMesh from your mesh file. It is described in detail in the P4est-based mesh part of the Trixi.jl docs.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using Trixi\nmesh_file = joinpath(\"out\", \"ice_cream_straight_sides.inp\")\nmesh = P4estMesh{2}(mesh_file)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"Since P4estMesh supports AMR, we just have to extend the setup from the first example by the standard AMR procedure. For more information about AMR in Trixi.jl, see the matching tutorial.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"amr_indicator = IndicatorLöhner(semi, variable=density)\n\namr_controller = ControllerThreeLevel(semi, amr_indicator,\n base_level=0,\n med_level =1, med_threshold=0.05,\n max_level =3, max_threshold=0.1)\n\namr_callback = AMRCallback(semi, amr_controller,\n interval=5,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true)\n\ncallbacks = CallbackSet(..., amr_callback)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"We can then post-process the solution file at the final time on the new mesh with Trixi2Vtk and visualize with ParaView, see the appropriate visualization section for details.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"(Image: simulation_straight_sides_p4est_amr)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/#Package-versions","page":"16 Unstructured meshes with HOHQMesh.jl","title":"Package versions","text":"","category":"section"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\", \"Trixi2Vtk\", \"HOHQMesh\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"","category":"page"},{"location":"tutorials/hohqmesh_tutorial/","page":"16 Unstructured meshes with HOHQMesh.jl","title":"16 Unstructured meshes with HOHQMesh.jl","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"EditURL = \"../../literate/src/files/DGMulti_2.jl\"","category":"page"},{"location":"tutorials/DGMulti_2/#DGMulti_2","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8: Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"section"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"For a tutorial about DG schemes via the DGMulti solver please visit the previous tutorial. The DGMulti solver also supports other methods than DG. The important property a method has to fulfill is the summation-by-parts (SBP) property. The package SummationByPartsOperators.jl provides such methods, like a finite difference SBP (FD SBP) scheme. To do this, you need to create an SBP derivative operator and pass that as approximation_type to the DGMulti constructor. For example, the classical second-order FD SBP operator can be created as","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly\nD = derivative_operator(MattssonNordström2004(), derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"Here, the arguments xmin and xmax do not matter beyond setting the real type used for the operator - they just set a reference element and are rescaled on the physical elements. The parameter N determines the number of finite difference nodes. Then, D can be used as approximation_type like SBP() in a multi-block fashion. In multiple dimensions, such a 1D SBP operator will be used in a tensor product fashion, i.e., in each coordinate direction. In particular, you can use them only on 1D, 2D Quad(), and 3D Hex() elements.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"You can also use fully periodic single-block FD methods by creating a periodic SBP operator. For example, a fully periodic FD operator can be constructed as","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"D = periodic_derivative_operator(derivative_order=1, accuracy_order=2,\n xmin=0.0, xmax=1.0, N=11)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"An example using such an FD method is implemented in elixir_euler_fdsbp_periodic.jl. For all parameters and other calling options, please have a look in the documentation of SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"Another possible method is for instance a continuous Galerkin (CGSEM) method. You can use such a method with polynomial degree of 3 (N=4 Legendre Lobatto nodes on [0, 1]) coupled continuously on a uniform mesh with Nx=10 elements by setting approximation_type to","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly\nD = couple_continuously(legendre_derivative_operator(xmin=0.0, xmax=1.0, N=4),\n UniformPeriodicMesh1D(xmin=-1.0, xmax=1.0, Nx=10))","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"To choose a discontinuous coupling (DGSEM), use couple_discontinuously() instead of couple_continuously().","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"For more information and other SBP operators, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/DGMulti_2/#Package-versions","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"Package versions","text":"","category":"section"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"StartUpDG\", \"SummationByPartsOperators\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"page"},{"location":"tutorials/DGMulti_2/","page":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"EditURL = \"../../literate/src/files/parabolic_terms.jl\"","category":"page"},{"location":"tutorials/parabolic_terms/#parabolic_terms","page":"12 Parabolic terms","title":"12: Parabolic terms","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"Experimental support for parabolic diffusion terms is available in Trixi.jl. This demo illustrates parabolic terms for the advection-diffusion equation.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"using OrdinaryDiffEq\nusing Trixi","category":"page"},{"location":"tutorials/parabolic_terms/#Splitting-a-system-into-hyperbolic-and-parabolic-parts.","page":"12 Parabolic terms","title":"Splitting a system into hyperbolic and parabolic parts.","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"For a mixed hyperbolic-parabolic system, we represent the hyperbolic and parabolic parts of the system separately. We first define the hyperbolic (advection) part of the advection-diffusion equation.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"advection_velocity = (1.5, 1.0)\nequations_hyperbolic = LinearScalarAdvectionEquation2D(advection_velocity);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"Next, we define the parabolic diffusion term. The constructor requires knowledge of equations_hyperbolic to be passed in because the LaplaceDiffusion2D applies diffusion to every variable of the hyperbolic system.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"diffusivity = 5.0e-2\nequations_parabolic = LaplaceDiffusion2D(diffusivity, equations_hyperbolic);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Boundary-conditions","page":"12 Parabolic terms","title":"Boundary conditions","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"As with the equations, we define boundary conditions separately for the hyperbolic and parabolic part of the system. For this example, we impose inflow BCs for the hyperbolic system (no condition is imposed on the outflow), and we impose Dirichlet boundary conditions for the parabolic equations. Both BoundaryConditionDirichlet and BoundaryConditionNeumann are defined for LaplaceDiffusion2D.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"The hyperbolic and parabolic boundary conditions are assumed to be consistent with each other.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"boundary_condition_zero_dirichlet = BoundaryConditionDirichlet((x, t, equations) -> SVector(0.0))\n\nboundary_conditions_hyperbolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1 + 0.5 * x[2])),\n y_neg = boundary_condition_zero_dirichlet,\n y_pos = boundary_condition_do_nothing,\n x_pos = boundary_condition_do_nothing)\n\nboundary_conditions_parabolic = (; x_neg = BoundaryConditionDirichlet((x, t, equations) -> SVector(1 + 0.5 * x[2])),\n y_neg = boundary_condition_zero_dirichlet,\n y_pos = boundary_condition_zero_dirichlet,\n x_pos = boundary_condition_zero_dirichlet);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Defining-the-solver-and-mesh","page":"12 Parabolic terms","title":"Defining the solver and mesh","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"The process of creating the DG solver and mesh is the same as for a purely hyperbolic system of equations.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\ncoordinates_min = (-1.0, -1.0) # minimum coordinates (min(x), min(y))\ncoordinates_max = ( 1.0, 1.0) # maximum coordinates (max(x), max(y))\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4,\n periodicity=false, n_cells_max=30_000) # set maximum capacity of tree data structure\n\ninitial_condition = (x, t, equations) -> SVector(0.0);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/#Semidiscretizing-and-solving","page":"12 Parabolic terms","title":"Semidiscretizing and solving","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"To semidiscretize a hyperbolic-parabolic system, we create a SemidiscretizationHyperbolicParabolic. This differs from a SemidiscretizationHyperbolic in that we pass in a Tuple containing both the hyperbolic and parabolic equation, as well as a Tuple containing the hyperbolic and parabolic boundary conditions.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"semi = SemidiscretizationHyperbolicParabolic(mesh,\n (equations_hyperbolic, equations_parabolic),\n initial_condition, solver;\n boundary_conditions=(boundary_conditions_hyperbolic,\n boundary_conditions_parabolic))","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"The rest of the code is identical to the hyperbolic case. We create a system of ODEs through semidiscretize, defining callbacks, and then passing the system to OrdinaryDiffEq.jl.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"tspan = (0.0, 1.5)\node = semidiscretize(semi, tspan)\ncallbacks = CallbackSet(SummaryCallback())\ntime_int_tol = 1.0e-6\nsol = solve(ode, RDPK3SpFSAL49(); abstol=time_int_tol, reltol=time_int_tol,\n ode_default_options()..., callback=callbacks);\nnothing #hide","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"We can now visualize the solution, which develops a boundary layer at the outflow boundaries.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"using Plots\nplot(sol)","category":"page"},{"location":"tutorials/parabolic_terms/#Package-versions","page":"12 Parabolic terms","title":"Package versions","text":"","category":"section"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"","category":"page"},{"location":"tutorials/parabolic_terms/","page":"12 Parabolic terms","title":"12 Parabolic terms","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"EditURL = \"../../literate/src/files/index.jl\"","category":"page"},{"location":"tutorials/introduction/#Tutorials-for-Trixi.jl","page":"Introduction","title":"Tutorials for Trixi.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"The tutorial section for Trixi.jl also contains interactive step-by-step explanations via Binder.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Right now, you are using the classic documentation. The corresponding interactive notebooks can be opened in Binder and viewed in nbviewer via the icons (Image: ) and (Image: ) in the respective tutorial. You can also open the raw notebook files via (Image: ).","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Note: To improve responsiveness via caching, the notebooks are updated only once a week. They are only available for the latest stable release of Trixi.jl at the time of caching.","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"There are tutorials for the following topics:","category":"page"},{"location":"tutorials/introduction/#[1-First-steps-in-Trixi.jl](@ref-getting_started)","page":"Introduction","title":"1 First steps in Trixi.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial provides guidance for getting started with Trixi.jl, and Julia as well. It outlines the installation procedures for both Julia and Trixi.jl, the execution of Trixi.jl elixirs, the fundamental structure of a Trixi.jl setup, the visualization of results, and the development process for Trixi.jl.","category":"page"},{"location":"tutorials/introduction/#[2-Behind-the-scenes-of-a-simulation-setup](@ref-behind_the_scenes_simulation_setup)","page":"Introduction","title":"2 Behind the scenes of a simulation setup","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial will guide you through a simple Trixi.jl setup (\"elixir\"), giving an overview of what happens in the background during the initialization of a simulation. While the setup described herein does not cover all details, it involves relatively stable parts of Trixi.jl that are unlikely to undergo significant changes in the near future. The goal is to clarify some of the more fundamental, technical concepts that are applicable to a variety of (also more complex) configurations.s","category":"page"},{"location":"tutorials/introduction/#[3-Introduction-to-DG-methods](@ref-scalar_linear_advection_1d)","page":"Introduction","title":"3 Introduction to DG methods","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial gives an introduction to discontinuous Galerkin (DG) methods with the example of the scalar linear advection equation in 1D. Starting with some theoretical explanations, we first implement a raw version of a discontinuous Galerkin spectral element method (DGSEM). Then, we will show how to use features of Trixi.jl to achieve the same result.","category":"page"},{"location":"tutorials/introduction/#[4-DGSEM-with-flux-differencing](@ref-DGSEM_FluxDiff)","page":"Introduction","title":"4 DGSEM with flux differencing","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"To improve stability often the flux differencing formulation of the DGSEM (split form) is used. We want to present the idea and formulation on a basic 1D level. Then, we show how this formulation can be implemented in Trixi.jl and analyse entropy conservation for two different flux combinations.","category":"page"},{"location":"tutorials/introduction/#[5-Shock-capturing-with-flux-differencing-and-stage-limiter](@ref-shock_capturing)","page":"Introduction","title":"5 Shock capturing with flux differencing and stage limiter","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Using the flux differencing formulation, a simple procedure to capture shocks is a hybrid blending of a high-order DG method and a low-order subcell finite volume (FV) method. We present the idea on a very basic level and show the implementation in Trixi.jl. Then, a positivity preserving limiter is explained and added to an exemplary simulation of the Sedov blast wave with the 2D compressible Euler equations.","category":"page"},{"location":"tutorials/introduction/#[6-Non-periodic-boundary-conditions](@ref-non_periodic_boundaries)","page":"Introduction","title":"6 Non-periodic boundary conditions","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Thus far, all examples used periodic boundaries. In Trixi.jl, you can also set up a simulation with non-periodic boundaries. This tutorial presents the implementation of the classical Dirichlet boundary condition with a following example. Then, other non-periodic boundaries are mentioned.","category":"page"},{"location":"tutorials/introduction/#[7-DG-schemes-via-DGMulti-solver](@ref-DGMulti_1)","page":"Introduction","title":"7 DG schemes via DGMulti solver","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial is about the more general DG solver DGMulti, introduced here. We are showing some examples for this solver, for instance with discretization nodes by Gauss or triangular elements. Moreover, we present a simple way to include pre-defined triangulate meshes for non-Cartesian domains using the package StartUpDG.jl.","category":"page"},{"location":"tutorials/introduction/#[8-Other-SBP-schemes-(FD,-CGSEM)-via-DGMulti-solver](@ref-DGMulti_2)","page":"Introduction","title":"8 Other SBP schemes (FD, CGSEM) via DGMulti solver","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Supplementary to the previous tutorial about DG schemes via the DGMulti solver we now present the possibility for DGMulti to use other SBP schemes via the package SummationByPartsOperators.jl. For instance, we show how to set up a finite differences (FD) scheme and a continuous Galerkin (CGSEM) method.","category":"page"},{"location":"tutorials/introduction/#[9-Upwind-FD-SBP-schemes](@ref-upwind_fdsbp)","page":"Introduction","title":"9 Upwind FD SBP schemes","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"General SBP schemes can not only be used via the DGMulti solver but also with a general DG solver. In particular, upwind finite difference SBP methods can be used together with the TreeMesh. Similar to general SBP schemes in the DGMulti framework, the interface is based on the package SummationByPartsOperators.jl.","category":"page"},{"location":"tutorials/introduction/#[10-Adding-a-new-scalar-conservation-law](@ref-adding_new_scalar_equations)","page":"Introduction","title":"10 Adding a new scalar conservation law","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial explains how to add a new physics model using the example of the cubic conservation law. First, we define the equation using a struct CubicEquation and the physical flux. Then, the corresponding standard setup in Trixi.jl (mesh, solver, semi and ode) is implemented and the ODE problem is solved by OrdinaryDiffEq's solve method.","category":"page"},{"location":"tutorials/introduction/#[11-Adding-a-non-conservative-equation](@ref-adding_nonconservative_equation)","page":"Introduction","title":"11 Adding a non-conservative equation","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"In this part, another physics model is implemented, the nonconservative linear advection equation. We run two different simulations with different levels of refinement and compare the resulting errors.","category":"page"},{"location":"tutorials/introduction/#[12-Parabolic-terms](@ref-parabolic_terms)","page":"Introduction","title":"12 Parabolic terms","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes how parabolic terms are implemented in Trixi.jl, e.g., to solve the advection-diffusion equation.","category":"page"},{"location":"tutorials/introduction/#[13-Adding-new-parabolic-terms](@ref-adding_new_parabolic_terms)","page":"Introduction","title":"13 Adding new parabolic terms","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes how new parabolic terms can be implemented using Trixi.jl.","category":"page"},{"location":"tutorials/introduction/#[14-Adaptive-mesh-refinement](@ref-adaptive_mesh_refinement)","page":"Introduction","title":"14 Adaptive mesh refinement","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Adaptive mesh refinement (AMR) helps to increase the accuracy in sensitive or turbolent regions while not wasting resources for less interesting parts of the domain. This leads to much more efficient simulations. This tutorial presents the implementation strategy of AMR in Trixi.jl, including the use of different indicators and controllers.","category":"page"},{"location":"tutorials/introduction/#[15-Structured-mesh-with-curvilinear-mapping](@ref-structured_mesh_mapping)","page":"Introduction","title":"15 Structured mesh with curvilinear mapping","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"In this tutorial, the use of Trixi.jl's structured curved mesh type StructuredMesh is explained. We present the two basic option to initialize such a mesh. First, the curved domain boundaries of a circular cylinder are set by explicit boundary functions. Then, a fully curved mesh is defined by passing the transformation mapping.","category":"page"},{"location":"tutorials/introduction/#[16-Unstructured-meshes-with-HOHQMesh.jl](@ref-hohqmesh_tutorial)","page":"Introduction","title":"16 Unstructured meshes with HOHQMesh.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"The purpose of this tutorial is to demonstrate how to use the UnstructuredMesh2D functionality of Trixi.jl. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will demonstrate how to conceptualize a problem with curved boundaries, generate a curvilinear mesh using the available HOHQMesh software in the Trixi.jl ecosystem, and then run a simulation using Trixi.jl on said mesh. In the end, the tutorial briefly explains how to simulate an example using AMR via P4estMesh.","category":"page"},{"location":"tutorials/introduction/#[17-P4est-mesh-from-gmsh](@ref-p4est_from_gmsh)","page":"Introduction","title":"17 P4est mesh from gmsh","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes how to obtain a P4estMesh from an existing mesh generated by gmsh or any other meshing software that can export to the Abaqus input .inp format. The tutorial demonstrates how edges/faces can be associated with boundary conditions based on the physical nodesets.","category":"page"},{"location":"tutorials/introduction/#[18-Explicit-time-stepping](@ref-time_stepping)","page":"Introduction","title":"18 Explicit time stepping","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial is about time integration using OrdinaryDiffEq.jl. It explains how to use their algorithms and presents two types of time step choices - with error-based and CFL-based adaptive step size control.","category":"page"},{"location":"tutorials/introduction/#[19-Differentiable-programming](@ref-differentiable_programming)","page":"Introduction","title":"19 Differentiable programming","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This part deals with some basic differentiable programming topics. For example, a Jacobian, its eigenvalues and a curve of total energy (through the simulation) are calculated and plotted for a few semidiscretizations. Moreover, we calculate an example for propagating errors with Measurement.jl at the end.","category":"page"},{"location":"tutorials/introduction/#[20-Custom-semidiscretization](@ref-custom_semidiscretization)","page":"Introduction","title":"20 Custom semidiscretization","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This tutorial describes the semidiscretiations of Trixi.jl and explains how to extend them for custom tasks.","category":"page"},{"location":"tutorials/introduction/#Examples-in-Trixi.jl","page":"Introduction","title":"Examples in Trixi.jl","text":"","category":"section"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"Trixi.jl already contains several more coding examples, the so-called elixirs. You can find them in the folder examples/. They are structured by the underlying mesh type and the respective number of spatial dimensions. The name of an elixir is composed of the underlying system of conservation equations (for instance advection or euler) and other special characteristics like the initial condition (e.g. gauss, astro_jet, blast_wave) or the included simulation features (e.g. amr, shockcapturing).","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"","category":"page"},{"location":"tutorials/introduction/","page":"Introduction","title":"Introduction","text":"This page was generated using Literate.jl.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"EditURL = \"../../../literate/src/files/first_steps/changing_trixi.jl\"","category":"page"},{"location":"tutorials/first_steps/changing_trixi/#changing_trixi","page":"1.3 Changing Trixi.jl itself","title":"1.3: First steps in Trixi.jl: Changing Trixi.jl itself","text":"","category":"section"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"If you plan on editing Trixi.jl itself, you can download Trixi.jl locally and run it from the cloned directory.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/#Cloning-Trixi.jl","page":"1.3 Changing Trixi.jl itself","title":"Cloning Trixi.jl","text":"","category":"section"},{"location":"tutorials/first_steps/changing_trixi/#Windows","page":"1.3 Changing Trixi.jl itself","title":"Windows","text":"","category":"section"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"If you are using Windows, you can clone Trixi.jl by using the GitHub Desktop tool:","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"If you do not have a GitHub account yet, create it on the GitHub website.\nDownload and install GitHub Desktop and then log in to your account.\nOpen GitHub Desktop, press Ctrl+Shift+O.\nIn the opened window, paste trixi-framework/Trixi.jl and choose the path to the folder where you want to save Trixi.jl. Then click Clone and Trixi.jl will be cloned to your computer.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"Now you cloned Trixi.jl and only need to tell Julia to use the local clone as the package sources:","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"Open a terminal using Win+r and cmd. Navigate to the folder with the cloned Trixi.jl using cd.\nCreate a new directory run, enter it, and start Julia with the --project=. flag:\nmkdir run\ncd run\njulia --project=.\nNow run the following commands to install all relevant packages:\nusing Pkg; Pkg.develop(PackageSpec(path=\"..\")) # Tell Julia to use the local Trixi.jl clone\nPkg.add([\"OrdinaryDiffEq\", \"Plots\"]) # Install additional packages","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"Now you already installed Trixi.jl from your local clone. Note that if you installed Trixi.jl this way, you always have to start Julia with the --project flag set to your run directory, e.g.,","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"julia --project=.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"if already inside the run directory.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/#Linux","page":"1.3 Changing Trixi.jl itself","title":"Linux","text":"","category":"section"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"You can clone Trixi.jl to your computer by executing the following commands:","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"git clone git@github.com:trixi-framework/Trixi.jl.git\n# If an error occurs, try the following:\n# git clone https://github.com/trixi-framework/Trixi.jl\ncd Trixi.jl\nmkdir run\ncd run\njulia --project=. -e 'using Pkg; Pkg.develop(PackageSpec(path=\"..\"))' # Tell Julia to use the local Trixi.jl clone\njulia --project=. -e 'using Pkg; Pkg.add([\"OrdinaryDiffEq\", \"Plots\"])' # Install additional packages","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"Note that if you installed Trixi.jl this way, you always have to start Julia with the --project flag set to your run directory, e.g.,","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"julia --project=.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"if already inside the run directory.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/#Additional-reading","page":"1.3 Changing Trixi.jl itself","title":"Additional reading","text":"","category":"section"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"To further delve into Trixi.jl, you may have a look at the following introductory tutorials.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"Introduction to DG methods will teach you how to set up a simple way to approximate the solution of a hyperbolic partial differential equation. It will be especially useful to learn about the Discontinuous Galerkin method and the way it is implemented in Trixi.jl.\nAdding a new scalar conservation law and Adding a non-conservative equation describe how to add new physics models that are not yet included in Trixi.jl.\nCallbacks gives an overview of how to regularly execute specific actions during a simulation, e.g., to store the solution or adapt the mesh.","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"","category":"page"},{"location":"tutorials/first_steps/changing_trixi/","page":"1.3 Changing Trixi.jl itself","title":"1.3 Changing Trixi.jl itself","text":"This page was generated using Literate.jl.","category":"page"},{"location":"reference-trixi/#Trixi.jl-API","page":"Trixi.jl","title":"Trixi.jl API","text":"","category":"section"},{"location":"reference-trixi/","page":"Trixi.jl","title":"Trixi.jl","text":"CurrentModule = Trixi","category":"page"},{"location":"reference-trixi/","page":"Trixi.jl","title":"Trixi.jl","text":"Modules = [Trixi]","category":"page"},{"location":"reference-trixi/#Trixi.Trixi","page":"Trixi.jl","title":"Trixi.Trixi","text":"Trixi\n\nTrixi.jl is a numerical simulation framework for hyperbolic conservation laws. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi.jl is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures.\n\nTo get started, run your first simulation with Trixi.jl using\n\ntrixi_include(default_example())\n\nSee also: trixi-framework/Trixi.jl\n\n\n\n\n\n","category":"module"},{"location":"reference-trixi/#Trixi.boundary_condition_do_nothing","page":"Trixi.jl","title":"Trixi.boundary_condition_do_nothing","text":"boundary_condition_do_nothing = Trixi.BoundaryConditionDoNothing()\n\nImposing no boundary condition just evaluates the flux at the inner state.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.boundary_condition_periodic","page":"Trixi.jl","title":"Trixi.boundary_condition_periodic","text":"boundary_condition_periodic = Trixi.BoundaryConditionPeriodic()\n\nA singleton struct indicating periodic boundary conditions.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_hll","page":"Trixi.jl","title":"Trixi.flux_hll","text":"flux_hll\n\nSee FluxHLL.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_hlle","page":"Trixi.jl","title":"Trixi.flux_hlle","text":"flux_hlle\n\nSee min_max_speed_einfeldt. This is a FluxHLL-type two-wave solver with special estimates of the wave speeds.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.flux_lax_friedrichs","page":"Trixi.jl","title":"Trixi.flux_lax_friedrichs","text":"flux_lax_friedrichs\n\nSee FluxLaxFriedrichs.\n\n\n\n\n\n","category":"constant"},{"location":"reference-trixi/#Trixi.AMRCallback","page":"Trixi.jl","title":"Trixi.AMRCallback","text":"AMRCallback(semi, controller [,adaptor=AdaptorAMR(semi)];\n interval,\n adapt_initial_condition=true,\n adapt_initial_condition_only_refine=true,\n dynamic_load_balancing=true)\n\nPerforms adaptive mesh refinement (AMR) every interval time steps for a given semidiscretization semi using the chosen controller.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AbstractEquations","page":"Trixi.jl","title":"Trixi.AbstractEquations","text":"AbstractEquations{NDIMS, NVARS}\n\nAn abstract supertype of specific equations such as the compressible Euler equations. The type parameters encode the number of spatial dimensions (NDIMS) and the number of primary variables (NVARS) of the physics model.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AbstractMesh","page":"Trixi.jl","title":"Trixi.AbstractMesh","text":"AbstractMesh{NDIMS}\n\nAn abstract supertype of specific mesh types such as TreeMesh or StructuredMesh. The type parameters encode the number of spatial dimensions (NDIMS).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AcousticPerturbationEquations2D","page":"Trixi.jl","title":"Trixi.AcousticPerturbationEquations2D","text":"AcousticPerturbationEquations2D(v_mean_global, c_mean_global, rho_mean_global)\n\nAcoustic perturbation equations (APE) in two space dimensions. The equations are given by\n\nbeginaligned\n fracpartialmathbfvpartial t + nabla (barmathbfvcdotmathbfv)\n + nablaleft( fracbarc^2 tildepbarrho right) = 0 \n fracpartial tildeppartial t +\n nablacdot (barrho mathbfv + barmathbfv tildep) = 0\nendaligned\n\nThe bar bar(cdot) indicates time-averaged quantities. The unknowns of the APE are the perturbed velocities mathbfv = (v_1 v_2)^T and the scaled perturbed pressure tildep = fracpbarc^2, where p denotes the perturbed pressure and the perturbed variables are defined by phi = phi - barphi.\n\nIn addition to the unknowns, Trixi.jl currently stores the mean values in the state vector, i.e. the state vector used internally is given by\n\nmathbfu =\n beginpmatrix\n v_1 v_2 tildep barv_1 barv_2 barc barrho\n endpmatrix\n\nThis affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the mean values must be zero.\nThe mean values have to be considered when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes these variables too.\nTrixi.jl's visualization tools will visualize the mean values by default.\n\nThe constructor accepts a 2-tuple v_mean_global and scalars c_mean_global and rho_mean_global which can be used to make the definition of initial conditions for problems with constant mean flow more flexible. These values are ignored if the mean values are defined internally in an initial condition.\n\nThe equations are based on the APE-4 system introduced in the following paper:\n\nRoland Ewert and Wolfgang Schröder (2003) Acoustic perturbation equations based on flow decomposition via source filtering DOI: 10.1016/S0021-9991(03)00168-2\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.Adiabatic","page":"Trixi.jl","title":"Trixi.Adiabatic","text":"struct Adiabatic\n\nUsed to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_normal_flux_function should be a function with signature boundary_value_normal_flux_function(x, t, equations) and return a scalar value for the normal heat flux at point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AliveCallback","page":"Trixi.jl","title":"Trixi.AliveCallback","text":"AliveCallback(analysis_interval=0, alive_interval=analysis_interval÷10)\n\nInexpensive callback showing that a simulation is still running by printing some information such as the current time to the screen every alive_interval time steps. If analysis_interval ≂̸ 0, the output is omitted every analysis_interval time steps.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AnalysisCallback","page":"Trixi.jl","title":"Trixi.AnalysisCallback","text":"AnalysisCallback(semi; interval=0,\n save_analysis=false,\n output_directory=\"out\",\n analysis_filename=\"analysis.dat\",\n extra_analysis_errors=Symbol[],\n extra_analysis_integrals=())\n\nAnalyze a numerical solution every interval time steps and print the results to the screen. If save_analysis, the results are also saved in joinpath(output_directory, analysis_filename).\n\nAdditional errors can be computed, e.g. by passing extra_analysis_errors = (:l2_error_primitive, :linf_error_primitive) or extra_analysis_errors = (:conservation_error,).\n\nIf you want to omit the computation (to safe compute-time) of the default_analysis_errors, specify analysis_errors = Symbol[]. Note: default_analysis_errors are :l2_error and :linf_error for all equations. If you want to compute extra_analysis_errors such as :conservation_error solely, i.e., without :l2_error, :linf_error you need to specify analysis_errors = [:conservation_error] instead of extra_analysis_errors = [:conservation_error].\n\nFurther scalar functions func in extra_analysis_integrals are applied to the numerical solution and integrated over the computational domain. Some examples for this are entropy, energy_kinetic, energy_internal, and energy_total. You can also write your own function with the same signature as the examples listed above and pass it via extra_analysis_integrals. See the developer comments about Trixi.analyze, Trixi.pretty_form_utf, and Trixi.pretty_form_ascii for further information on how to create custom analysis quantities.\n\nIn addition, the analysis callback records and outputs a number of quantities that are useful for evaluating the computational performance, such as the total runtime, the performance index (time/DOF/rhs!), the time spent in garbage collection (GC), or the current memory usage (alloc'd memory).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AnalysisCallbackCoupled","page":"Trixi.jl","title":"Trixi.AnalysisCallbackCoupled","text":"AnalysisCallbackCoupled(semi, callbacks...)\n\nCombine multiple analysis callbacks for coupled simulations with a SemidiscretizationCoupled. For each coupled system, an indididual AnalysisCallback must be created and passed to the AnalysisCallbackCoupled in order, i.e., in the same sequence as the indidvidual semidiscretizations are stored in the SemidiscretizationCoupled.\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.AveragingCallback","page":"Trixi.jl","title":"Trixi.AveragingCallback","text":"AveragingCallback(semi::SemidiscretizationHyperbolic, tspan; output_directory=\"out\",\n filename=\"averaging.h5\")\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nA callback that averages the flow field described by semi which must be a semidiscretization of the compressible Euler equations in two dimensions. The callback records the mean velocity, mean speed of sound, mean density, and mean vorticity for each node over the time interval given by tspan and stores the results in an HDF5 file filename in the directory output_directory. Note that this callback does not support adaptive mesh refinement (AMRCallback).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionCoupled","page":"Trixi.jl","title":"Trixi.BoundaryConditionCoupled","text":"BoundaryConditionCoupled(other_semi_index, indices, uEltype, coupling_converter)\n\nBoundary condition to glue two meshes together. Solution values at the boundary of another mesh will be used as boundary values. This requires the use of SemidiscretizationCoupled. The other mesh is specified by other_semi_index, which is the index of the mesh in the tuple of semidiscretizations.\n\nNote that the elements and nodes of the two meshes at the coupled boundary must coincide. This is currently only implemented for StructuredMesh.\n\nArguments\n\nother_semi_index: the index in SemidiscretizationCoupled of the semidiscretization from which the values are copied\nindices::Tuple: node/cell indices at the boundary of the mesh in the other semidiscretization. See examples below.\nuEltype::Type: element type of solution\ncoupling_converter::CouplingConverter: function to call for converting the solution state of one system to the other system\n\nExamples\n\n# Connect the left boundary of mesh 2 to our boundary such that our positive\n# boundary direction will match the positive y direction of the other boundary\nBoundaryConditionCoupled(2, (:begin, :i), Float64, fun)\n\n# Connect the same two boundaries oppositely oriented\nBoundaryConditionCoupled(2, (:begin, :i_backwards), Float64, fun)\n\n# Using this as y_neg boundary will connect `our_cells[i, 1, j]` to `other_cells[j, end-i, end]`\nBoundaryConditionCoupled(2, (:j, :i_backwards, :end), Float64, fun)\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionDirichlet","page":"Trixi.jl","title":"Trixi.BoundaryConditionDirichlet","text":"BoundaryConditionDirichlet(boundary_value_function)\n\nCreate a Dirichlet boundary condition that uses the function boundary_value_function to specify the values at the boundary. This can be used to create a boundary condition that specifies exact boundary values by passing the exact solution of the equation. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as\n\nboundary_value_function(x, t, equations)\n\nwhere x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.\n\nExamples\n\njulia> BoundaryConditionDirichlet(initial_condition_convergence_test)\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionNavierStokesWall","page":"Trixi.jl","title":"Trixi.BoundaryConditionNavierStokesWall","text":"struct BoundaryConditionNavierStokesWall\n\nCreates a wall-type boundary conditions for the compressible Navier-Stokes equations. The fields boundary_condition_velocity and boundary_condition_heat_flux are intended to be boundary condition types such as the NoSlip velocity boundary condition and the Adiabatic or Isothermal heat boundary condition.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundaryConditionNeumann","page":"Trixi.jl","title":"Trixi.BoundaryConditionNeumann","text":"BoundaryConditionNeumann(boundary_normal_flux_function)\n\nSimilar to BoundaryConditionDirichlet, but creates a Neumann boundary condition for parabolic equations that uses the function boundary_normal_flux_function to specify the values of the normal flux at the boundary. The passed boundary value function will be called with the same arguments as an initial condition function is called, i.e., as\n\nboundary_normal_flux_function(x, t, equations)\n\nwhere x specifies the coordinates, t is the current time, and equation is the corresponding system of equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.BoundsCheckCallback","page":"Trixi.jl","title":"Trixi.BoundsCheckCallback","text":"BoundsCheckCallback(; output_directory=\"out\", save_errors=false, interval=1)\n\nSubcell limiting techniques with SubcellLimiterIDP are constructed to adhere certain local or global bounds. To make sure that these bounds are actually met, this callback calculates the maximum deviation from the bounds. The maximum deviation per applied bound is printed to the screen at the end of the simulation. For more insights, when setting save_errors=true the occurring errors are exported every interval time steps during the simulation. Then, the maximum deviations since the last export are saved in \"output_directory/deviations.txt\". The BoundsCheckCallback has to be applied as a stage callback for the SSPRK time integration scheme.\n\nnote: Note\nFor SubcellLimiterIDP, the solution is corrected in the a posteriori correction stage SubcellLimiterIDPCorrection. So, to check the final solution, this bounds check callback must be called after the correction stage.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CarpenterKennedy2N43","page":"Trixi.jl","title":"Trixi.CarpenterKennedy2N43","text":" CarpenterKennedy2N43()\n\nCarpenter, Kennedy (1994) Third order 2N storage RK schemes with error control\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CarpenterKennedy2N54","page":"Trixi.jl","title":"Trixi.CarpenterKennedy2N54","text":"CarpenterKennedy2N54()\n\nThe following structures and methods provide a minimal implementation of the low-storage explicit Runge-Kutta method of\n\nCarpenter, Kennedy (1994) Fourth order 2N storage RK schemes, Solution 3\n\nusing the same interface as OrdinaryDiffEq.jl.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations1D","text":"CompressibleEulerEquations1D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1 rho v_1^2 + p (rho e +p) v_1\nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in one space dimension. Here, rho is the density, v_1 the velocity, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho v_1^2 right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations2D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations2D","text":"CompressibleEulerEquations2D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + p rho v_1 v_2 (rho e +p) v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + p (rho e +p) v_2\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in two space dimensions. Here, rho is the density, v_1, v_2 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquations3D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquations3D","text":"CompressibleEulerEquations3D(gamma)\n\nThe compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2 rho v_3 rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + p rho v_1 v_2 rho v_1 v_3 ( rho e +p) v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + p rho v_1 v_3 ( rho e +p) v_2\nendpmatrix\n+\nfracpartialpartial z\nbeginpmatrix\nrho v_3 rho v_1 v_3 rho v_2 v_3 rho v_3^2 + p ( rho e +p) v_3\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in three space dimensions. Here, rho is the density, v_1, v_2, v_3 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2+v_3^2) right)\n\nthe pressure.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerEquationsQuasi1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerEquationsQuasi1D","text":"CompressibleEulerEquationsQuasi1D(gamma)\n\nThe quasi-1d compressible Euler equations (see Chan et al. DOI: 10.48550/arXiv.2307.12089 for details)\n\nfracpartialpartial t\nbeginpmatrix\na rho a rho v_1 a e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\na rho v_1 a rho v_1^2 a v_1 (e +p)\nendpmatrix\n+ \na fracpartialpartial x\nbeginpmatrix\n0 p 0 \nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in one space dimension. Here, rho is the density, v_1 the velocity, e the specific total energy rather than specific internal energy, a the (possibly) variable nozzle width, and\n\np = (gamma - 1) left( e - frac12 rho v_1^2 right)\n\nthe pressure.\n\nThe nozzle width function a(x) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the compressible Euler equations one can set the nozzle width variable a to one. \n\nIn addition to the unknowns, Trixi.jl currently stores the nozzle width values at the approximation points despite being fixed in time. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the nozzle width must be zero.\nThe nozzle width values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the nozzle width by default.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerMulticomponentEquations1D","page":"Trixi.jl","title":"Trixi.CompressibleEulerMulticomponentEquations1D","text":"CompressibleEulerMulticomponentEquations1D(; gammas, gas_constants)\n\nMulticomponent version of the compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho v_1 rho e rho_1 rho_2 vdots rho_n\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1^2 + p (rho e +p) v_1 rho_1 v_1 rho_2 v_1 vdots rho_n v_1\nendpmatrix\n\n=\nbeginpmatrix\n0 0 0 0 vdots 0\nendpmatrix\n\nfor calorically perfect gas in one space dimension. Here, rho_i is the density of component i, rho=sum_i=1^nrho_i the sum of the individual rho_i, v_1 the velocity, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho v_1^2 right)\n\nthe pressure,\n\ngamma=fracsum_i=1^nrho_i C_vigamma_isum_i=1^nrho_i C_vi\n\ntotal heat capacity ratio, gamma_i heat capacity ratio of component i,\n\nC_vi=fracRgamma_i-1\n\nspecific heat capacity at constant volume of component i.\n\nIn case of more than one component, the specific heat ratios gammas and the gas constants gas_constants should be passed as tuples, e.g., gammas=(1.4, 1.667).\n\nThe remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleEulerMulticomponentEquations2D","page":"Trixi.jl","title":"Trixi.CompressibleEulerMulticomponentEquations2D","text":"CompressibleEulerMulticomponentEquations2D(; gammas, gas_constants)\n\nMulticomponent version of the compressible Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho v_1 rho v_2 rho e rho_1 rho_2 vdots rho_n\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\nrho v_1^2 + p rho v_1 v_2 ( rho e +p) v_1 rho_1 v_1 rho_2 v_1 vdots rho_n v_1\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_1 v_2 rho v_2^2 + p ( rho e +p) v_2 rho_1 v_2 rho_2 v_2 vdots rho_n v_2\nendpmatrix\n=\nbeginpmatrix\n0 0 0 0 0 vdots 0\nendpmatrix\n\nfor calorically perfect gas in two space dimensions. Here, rho_i is the density of component i, rho=sum_i=1^nrho_i the sum of the individual rho_i, v_1, v_2 the velocities, e the specific total energy rather than specific internal energy, and\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2 + v_2^2) right)\n\nthe pressure,\n\ngamma=fracsum_i=1^nrho_i C_vigamma_isum_i=1^nrho_i C_vi\n\ntotal heat capacity ratio, gamma_i heat capacity ratio of component i,\n\nC_vi=fracRgamma_i-1\n\nspecific heat capacity at constant volume of component i.\n\nIn case of more than one component, the specific heat ratios gammas and the gas constants gas_constants in [kJ/(kg*K)] should be passed as tuples, e.g., gammas=(1.4, 1.667).\n\nThe remaining variables like the specific heats at constant volume cv or the specific heats at constant pressure cp are then calculated considering a calorically perfect gas.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion1D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion1D","text":"CompressibleNavierStokesDiffusion1D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations1D.\n\nequations: instance of the CompressibleEulerEquations1D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v rho e\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v rho v^2 + p (rho e + p) v\nendpmatrix\n=\nfracpartialpartial x\nbeginpmatrix\n0 tau tau v - q\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho v^2 right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations1D. The terms on the right hand side of the system above are built from the viscous stress\n\ntau = mu fracpartialpartial x v\n\nwhere the heat flux is\n\nq = -kappa fracpartialpartial x left(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nq = -kappa fracpartialpartial x left(Tright) = -fracgamma mu(gamma - 1)textrmPr fracpartialpartial x left(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn one spatial dimensions we require gradients for two quantities, e.g., primitive quantities\n\nfracpartialpartial x v fracpartialpartial x T\n\nor the entropy variables\n\nfracpartialpartial x w_2 fracpartialpartial x w_3\n\nwhere\n\nw_2 = fracrho v1p w_3 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion2D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion2D","text":"CompressibleNavierStokesDiffusion2D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations2D.\n\nequations: instance of the CompressibleEulerEquations2D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho mathbfv rho e\nendpmatrix\n+\nnabla cdot\nbeginpmatrix\n rho mathbfv rho mathbfvmathbfv^T + p underlineI (rho e + p) mathbfv\nendpmatrix\n=\nnabla cdot\nbeginpmatrix\n0 underlinetau underlinetaumathbfv - mathbfq\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2) right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor\n\nunderlinetau = mu left(nablamathbfv + left(nablamathbfvright)^Tright) - frac23 mu left(nablacdotmathbfvright)underlineI\n\nwhere underlineI is the 2times 2 identity matrix and the heat flux is\n\nmathbfq = -kappanablaleft(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nmathbfq = -kappanablaleft(Tright) = -fracgamma mu(gamma - 1)textrmPrnablaleft(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn two spatial dimensions we require gradients for three quantities, e.g., primitive quantities\n\nnabla v_1 nabla v_2 nabla T\n\nor the entropy variables\n\nnabla w_2 nabla w_3 nabla w_4\n\nwhere\n\nw_2 = fracrho v_1p w_3 = fracrho v_2p w_4 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.CompressibleNavierStokesDiffusion3D","page":"Trixi.jl","title":"Trixi.CompressibleNavierStokesDiffusion3D","text":"CompressibleNavierStokesDiffusion3D(equations; mu, Pr,\n gradient_variables=GradientVariablesPrimitive())\n\nContains the diffusion (i.e. parabolic) terms applied to mass, momenta, and total energy together with the advective terms from the CompressibleEulerEquations3D.\n\nequations: instance of the CompressibleEulerEquations3D\nmu: dynamic viscosity,\nPr: Prandtl number,\ngradient_variables: which variables the gradients are taken with respect to. Defaults to GradientVariablesPrimitive().\n\nFluid properties such as the dynamic viscosity mu can be provided in any consistent unit system, e.g., [mu] = kg m⁻¹ s⁻¹.\n\nThe particular form of the compressible Navier-Stokes implemented is\n\nfracpartialpartial t\nbeginpmatrix\nrho rho mathbfv rho e\nendpmatrix\n+\nnabla cdot\nbeginpmatrix\n rho mathbfv rho mathbfvmathbfv^T + p underlineI (rho e + p) mathbfv\nendpmatrix\n=\nnabla cdot\nbeginpmatrix\n0 underlinetau underlinetaumathbfv - mathbfq\nendpmatrix\n\nwhere the system is closed with the ideal gas assumption giving\n\np = (gamma - 1) left( rho e - frac12 rho (v_1^2+v_2^2+v_3^2) right)\n\nas the pressure. The value of the adiabatic constant gamma is taken from the CompressibleEulerEquations2D. The terms on the right hand side of the system above are built from the viscous stress tensor\n\nunderlinetau = mu left(nablamathbfv + left(nablamathbfvright)^Tright) - frac23 mu left(nablacdotmathbfvright)underlineI\n\nwhere underlineI is the 3times 3 identity matrix and the heat flux is\n\nmathbfq = -kappanablaleft(Tright)quad T = fracpRrho\n\nwhere T is the temperature and kappa is the thermal conductivity for Fick's law. Under the assumption that the gas has a constant Prandtl number, the thermal conductivity is\n\nkappa = fracgamma mu R(gamma - 1)textrmPr\n\nFrom this combination of temperature T and thermal conductivity kappa we see that the gas constant R cancels and the heat flux becomes\n\nmathbfq = -kappanablaleft(Tright) = -fracgamma mu(gamma - 1)textrmPrnablaleft(fracprhoright)\n\nwhich is the form implemented below in the flux function.\n\nIn two spatial dimensions we require gradients for three quantities, e.g., primitive quantities\n\nnabla v_1 nabla v_2 nabla v_3 nabla T\n\nor the entropy variables\n\nnabla w_2 nabla w_3 nabla w_4 nabla w_5\n\nwhere\n\nw_2 = fracrho v_1p w_3 = fracrho v_2p w_4 = fracrho v_3p w_5 = -fracrhop\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ControllerThreeLevel","page":"Trixi.jl","title":"Trixi.ControllerThreeLevel","text":"ControllerThreeLevel(semi, indicator; base_level=1,\n med_level=base_level, med_threshold=0.0,\n max_level=base_level, max_threshold=1.0)\n\nAn AMR controller based on three levels (in descending order of precedence):\n\nset the target level to max_level if indicator > max_threshold\nset the target level to med_level if indicator > med_threshold; if med_level < 0, set the target level to the current level\nset the target level to base_level otherwise\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ControllerThreeLevelCombined","page":"Trixi.jl","title":"Trixi.ControllerThreeLevelCombined","text":"ControllerThreeLevelCombined(semi, indicator_primary, indicator_secondary;\n base_level=1,\n med_level=base_level, med_threshold=0.0,\n max_level=base_level, max_threshold=1.0,\n max_threshold_secondary=1.0)\n\nAn AMR controller based on three levels (in descending order of precedence):\n\nset the target level to max_level if indicator_primary > max_threshold\nset the target level to med_level if indicator_primary > med_threshold; if med_level < 0, set the target level to the current level\nset the target level to base_level otherwise\n\nIf indicator_secondary >= max_threshold_secondary, set the target level to max_level.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DG","page":"Trixi.jl","title":"Trixi.DG","text":"DG(; basis, mortar, surface_integral, volume_integral)\n\nCreate a discontinuous Galerkin method. If basis isa LobattoLegendreBasis, this creates a DGSEM.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DGMulti-Tuple{SummationByPartsOperators.AbstractDerivativeOperator}","page":"Trixi.jl","title":"Trixi.DGMulti","text":"DGMulti(approximation_type::AbstractDerivativeOperator;\n element_type::AbstractElemShape,\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n kwargs...)\n\nCreate a summation by parts (SBP) discretization on the given element_type using a tensor product structure based on the 1D SBP derivative operator passed as approximation_type.\n\nFor more info, see the documentations of StartUpDG.jl and SummationByPartsOperators.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMulti-Tuple{}","page":"Trixi.jl","title":"Trixi.DGMulti","text":"DGMulti(; polydeg::Integer,\n element_type::AbstractElemShape,\n approximation_type=Polynomial(),\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n RefElemData_kwargs...)\n\nCreate a discontinuous Galerkin method which uses\n\napproximations of polynomial degree polydeg\nelement type element_type (Tri(), Quad(), Tet(), and Hex() currently supported)\n\nOptional:\n\napproximation_type (default is Polynomial(); SBP() also supported for Tri(), Quad(), and Hex() element types).\nRefElemData_kwargs are additional keyword arguments for RefElemData, such as quad_rule_vol. For more info, see the StartUpDG.jl docs.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh{NDIMS, ...}\n\nDGMultiMesh describes a mesh type which wraps StartUpDG.MeshData and boundary_faces in a dispatchable type. This is intended to store geometric data and connectivities for any type of mesh (Cartesian, affine, curved, structured/unstructured).\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Tuple{DGMulti{2, Tri, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{2, Tri, ApproxType}} where {ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, Dict{Symbol, Int64}}","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{2, Tri}, triangulateIO, boundary_dict::Dict{Symbol, Int})\n\ndg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\ntriangulateIO is a TriangulateIO mesh representation\nboundary_dict is a Dict{Symbol, Int} which associates each integer TriangulateIO boundary tag with a Symbol.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType<:SummationByPartsOperators.AbstractPeriodicDerivativeOperator, SurfaceIntegral, VolumeIntegral, Mortar}}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti)\n\nConstructs a single-element DGMultiMesh for a single periodic element given a DGMulti with approximation_type set to a periodic (finite difference) SBP operator from SummationByPartsOperators.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, AbstractArray}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{NDIMS}, vertex_coordinates, EToV;\n is_on_boundary=nothing,\n periodicity=ntuple(_->false, NDIMS)) where {NDIMS}\n\ndg::DGMulti contains information associated with to the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\nvertex_coordinates is a tuple of vectors containing x,y,... components of the vertex coordinates\nEToV is a 2D array containing element-to-vertex connectivities for each element\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of booleans specifying if the domain is periodic true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any, Any}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti{NDIMS}, cells_per_dimension, mapping;\n is_on_boundary=nothing,\n periodicity=ntuple(_ -> false, NDIMS), kwargs...) where {NDIMS}\n\nConstructs a Curved() DGMultiMesh with element type dg.basis.element_type.\n\nmapping is a function which maps from a reference [-1, 1]^NDIMS domain to a mapped domain, e.g., xy = mapping(x, y) in 2D.\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, Any}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti, cells_per_dimension;\n coordinates_min=(-1.0, -1.0), coordinates_max=(1.0, 1.0),\n is_on_boundary=nothing,\n periodicity=ntuple(_ -> false, NDIMS))\n\nConstructs a Cartesian DGMultiMesh with element type dg.basis.element_type. The domain is the tensor product of the intervals [coordinates_min[i], coordinates_max[i]].\n\nis_on_boundary specifies boundary using a Dict{Symbol, <:Function}\nperiodicity is a tuple of Bools specifying periodicity = true/false in the (x,y,z) direction.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiMesh-Union{Tuple{NDIMS}, Tuple{DGMulti{NDIMS, ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar, <:StartUpDG.RefElemData{NDIMS, ElemType, ApproxType}} where {ElemType, ApproxType, SurfaceIntegral, VolumeIntegral, Mortar}, String}} where NDIMS","page":"Trixi.jl","title":"Trixi.DGMultiMesh","text":"DGMultiMesh(dg::DGMulti, filename::String)\n\ndg::DGMulti contains information associated with the reference element (e.g., quadrature, basis evaluation, differentiation, etc).\nfilename is a path specifying a .mesh file generated by HOHQMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGSEM","page":"Trixi.jl","title":"Trixi.DGSEM","text":"DGSEM(; RealT=Float64, polydeg::Integer,\n surface_flux=flux_central,\n surface_integral=SurfaceIntegralWeakForm(surface_flux),\n volume_integral=VolumeIntegralWeakForm(),\n mortar=MortarL2(basis))\n\nCreate a discontinuous Galerkin spectral element method (DGSEM) using a LobattoLegendreBasis with polynomials of degree polydeg.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DissipationGlobalLaxFriedrichs","page":"Trixi.jl","title":"Trixi.DissipationGlobalLaxFriedrichs","text":"DissipationGlobalLaxFriedrichs(λ)\n\nCreate a global Lax-Friedrichs dissipation operator with dissipation coefficient λ.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.DissipationLocalLaxFriedrichs","page":"Trixi.jl","title":"Trixi.DissipationLocalLaxFriedrichs","text":"DissipationLocalLaxFriedrichs(max_abs_speed=max_abs_speed_naive)\n\nCreate a local Lax-Friedrichs dissipation operator where the maximum absolute wave speed is estimated as max_abs_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to max_abs_speed_naive.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nA callback that couples the acoustic perturbation equations and compressible Euler equations. Must be used in conjunction with SemidiscretizationEulerAcoustics. This callback manages the flow solver - which is always one time step ahead of the acoustics solver - and calculates the acoustic source term after each time step. The linearized Lamb vector is used as the source term, i.e.\n\nmathbfs = -(mathbfomega times barmathbfv\n + barmathbfomega times mathbfv)\n\nwhere mathbfv denotes the velocity, mathbfomega denotes the vorticity, the bar bar(cdot) indicates time-averaged quantities (see AveragingCallback) and prime (cdot) denotes perturbed quantities defined by phi = phi - barphi. Note that the perturbed quantities here are based entirely on the pure flow solution and should not be confused with the state variables of the acoustic perturbation equations.\n\nIn addition, this callback manages the time step size for both solvers and initializes the mean values of the acoustic perturbation equations using results obtained with the AveragingCallback.\n\nMichael Schlottke-Lakemper (2017) A direct-hybrid method for aeroacoustic analysis DOI: 10.18154/RWTH-2017-04082\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback-Tuple{Any, AbstractString, Any, Real, Real}","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback(ode_euler, averaging_file::AbstractString, alg,\n cfl_acoustics::Real, cfl_euler::Real; kwargs...)\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nCreates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_file (see AveragingCallback).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.EulerAcousticsCouplingCallback-Tuple{Any, DiscreteCallback{<:Any, <:AveragingCallback}, Any, Real, Real}","page":"Trixi.jl","title":"Trixi.EulerAcousticsCouplingCallback","text":"EulerAcousticsCouplingCallback(ode_euler,\n averaging_callback::DiscreteCallback{<:Any, <:AveragingCallback},\n alg, cfl_acoustics::Real, cfl_euler::Real; kwargs...)\n\nwarning: Experimental code\nThis callback is experimental and may change in any future release.\n\nCreates an EulerAcousticsCouplingCallback based on the pure flow ODEProblem given by ode_euler. Creates an integrator using the time integration method alg and the keyword arguments to solve ode_euler (consult the OrdinaryDiffEq documentation for further information). Manages the step size for both solvers by using the minimum of the maximum step size obtained with CFL numbers cfl_acoustics for the acoustics solver and cfl_euler for and flow solver, respectively. The mean values for the acoustic perturbation equations are read from averaging_callback (see AveragingCallback).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.FDSBP","page":"Trixi.jl","title":"Trixi.FDSBP","text":"FDSBP(D_SBP; surface_integral, volume_integral)\n\nSpecialization of DG methods that uses general summation by parts (SBP) operators from SummationByPartsOperators.jl. In particular, this includes classical finite difference (FD) SBP methods. These methods have the same structure as classical DG methods - local operations on elements with connectivity through interfaces without imposing any continuity constraints.\n\nD_SBP is an SBP derivative operator from SummationByPartsOperators.jl. The other arguments have the same meaning as in DG or DGSEM.\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxHLL","page":"Trixi.jl","title":"Trixi.FluxHLL","text":"FluxHLL(min_max_speed=min_max_speed_davis)\n\nCreate an HLL (Harten, Lax, van Leer) numerical flux where the minimum and maximum wave speeds are estimated as λ_min, λ_max = min_max_speed(u_ll, u_rr, orientation_or_normal_direction, equations), defaulting to min_max_speed_davis. Original paper:\n\nAmiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxHydrostaticReconstruction","page":"Trixi.jl","title":"Trixi.FluxHydrostaticReconstruction","text":"FluxHydrostaticReconstruction(numerical_flux, hydrostatic_reconstruction)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nAllow for some kind of hydrostatic reconstruction of the solution state prior to the surface flux computation. This is a particular strategy to ensure that the method remains well-balanced for the shallow water equations, see ShallowWaterEquations1D or ShallowWaterEquations2D.\n\nFor example, the hydrostatic reconstruction from Audusse et al. is implemented in one and two spatial dimensions, see hydrostatic_reconstruction_audusse_etal or the original paper\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\nOther hydrostatic reconstruction techniques are available, particularly to handle wet / dry fronts. A good overview of the development and application of hydrostatic reconstruction can be found in\n\nGuoxian Chen and Sebastian Noelle A unified surface-gradient and hydrostatic reconstruction scheme for the shallow water equations (2021) RWTH Aachen preprint\nAndreas Buttinger-Kreuzhuber, Zsolt Horváth, Sebastian Noelle, Günter Blöschl and Jürgen Waser (2019) A fast second-order shallow water scheme on two-dimensional structured grids over abrupt topography DOI: 10.1016/j.advwatres.2019.03.010\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxLMARS","page":"Trixi.jl","title":"Trixi.FluxLMARS","text":"FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nLow Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.\n\nReferences:\n\nXi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxLMARS-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.FluxLMARS","text":"FluxLMARS(c)(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nLow Mach number approximate Riemann solver (LMARS) for atmospheric flows using an estimate c of the speed of sound.\n\nReferences:\n\nXi Chen et al. (2013) A Control-Volume Model of the Compressible Euler Equations with a Vertical Lagrangian Coordinate DOI: 10.1175/MWR-D-12-00129.1\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.FluxLaxFriedrichs","page":"Trixi.jl","title":"Trixi.FluxLaxFriedrichs","text":"FluxLaxFriedrichs(max_abs_speed=max_abs_speed_naive)\n\nLocal Lax-Friedrichs (Rusanov) flux with maximum wave speed estimate provided by max_abs_speed, cf. DissipationLocalLaxFriedrichs and max_abs_speed_naive.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxPlusDissipation","page":"Trixi.jl","title":"Trixi.FluxPlusDissipation","text":"FluxPlusDissipation(numerical_flux, dissipation)\n\nCombine a numerical_flux with a dissipation operator to create a new numerical flux.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxRotated","page":"Trixi.jl","title":"Trixi.FluxRotated","text":"FluxRotated(numerical_flux)\n\nCompute a numerical_flux flux in direction of a normal vector by rotating the solution, computing the numerical flux in x-direction, and rotating the calculated flux back.\n\nRequires a rotationally invariant equation with equation-specific functions rotate_to_x and rotate_from_x.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.FluxUpwind","page":"Trixi.jl","title":"Trixi.FluxUpwind","text":"FluxUpwind(splitting)\n\nA numerical flux f(u_left, u_right) = f⁺(u_left) + f⁻(u_right) based on flux vector splitting.\n\nThe SurfaceIntegralUpwind with a given splitting is equivalent to the SurfaceIntegralStrongForm with FluxUpwind(splitting) as numerical flux (up to floating point differences).\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.GlmSpeedCallback","page":"Trixi.jl","title":"Trixi.GlmSpeedCallback","text":"GlmSpeedCallback(; glm_scale=0.5, cfl)\n\nUpdate the divergence cleaning wave speed c_h according to the time step computed in StepsizeCallback for the ideal GLM-MHD equations. The cfl number should be set to the same value as for the time step size calculation. The glm_scale ensures that the GLM wave speed is lower than the fastest physical waves in the MHD solution and should thus be set to a value within the interval [0,1]. Note that glm_scale = 0 deactivates the divergence cleaning.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.GradientVariablesPrimitive","page":"Trixi.jl","title":"Trixi.GradientVariablesPrimitive","text":"GradientVariablesPrimitive and GradientVariablesEntropy are gradient variable type parameters for CompressibleNavierStokesDiffusion1D. By default, the gradient variables are set to be GradientVariablesPrimitive. Specifying GradientVariablesEntropy instead uses the entropy variable formulation from\n\nHughes, Mallet, Franca (1986) A new finite element formulation for computational fluid dynamics: I. Symmetric forms of the compressible Euler and Navier-Stokes equations and the second law of thermodynamics. https://doi.org/10.1016/0045-7825(86)90127-1\n\nUnder GradientVariablesEntropy, the Navier-Stokes discretization is provably entropy stable.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HypDiffN3Erk3Sstar52","page":"Trixi.jl","title":"Trixi.HypDiffN3Erk3Sstar52","text":"HypDiffN3Erk3Sstar52()\n\nFive stage, second-order accurate explicit Runge-Kutta scheme with stability region optimized for the hyperbolic diffusion equation with LLF flux and polynomials of degree polydeg=3.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations1D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations1D","text":"HyperbolicDiffusionEquations1D\n\nThe linear hyperbolic diffusion equations in one space dimension. A description of this system can be found in Sec. 2.5 of the book\n\nMasatsuka (2013) I Do Like CFD, Too: Vol 1. Freely available at http://www.cfdbooks.com/\n\nFurther analysis can be found in the paper\n\nNishikawa (2007) A first-order system approach for diffusion equation. I: Second-order residual-distribution schemes DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations2D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations2D","text":"HyperbolicDiffusionEquations2D\n\nThe linear hyperbolic diffusion equations in two space dimensions. A description of this system can be found in Sec. 2.5 of the book \"I Do Like CFD, Too: Vol 1\". The book is freely available at http://www.cfdbooks.com/ and further analysis can be found in the paper by Nishikawa DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.HyperbolicDiffusionEquations3D","page":"Trixi.jl","title":"Trixi.HyperbolicDiffusionEquations3D","text":"HyperbolicDiffusionEquations3D\n\nThe linear hyperbolic diffusion equations in three space dimensions. A description of this system can be found in Sec. 2.5 of the book \"I Do Like CFD, Too: Vol 1\". The book is freely available at http://www.cfdbooks.com/ and further analysis can be found in the paper by Nishikawa DOI: 10.1016/j.jcp.2007.07.029\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations1D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations1D","text":"IdealGlmMhdEquations1D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in one space dimension.\n\nnote: Note\nThere is no divergence cleaning variable psi because the divergence-free constraint is satisfied trivially in one spatial dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations2D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations2D","text":"IdealGlmMhdEquations2D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in two space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdEquations3D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdEquations3D","text":"IdealGlmMhdEquations3D(gamma)\n\nThe ideal compressible GLM-MHD equations for an ideal gas with ratio of specific heats gamma in three space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdMulticomponentEquations1D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdMulticomponentEquations1D","text":"IdealGlmMhdMulticomponentEquations1D\n\nThe ideal compressible multicomponent GLM-MHD equations in one space dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IdealGlmMhdMulticomponentEquations2D","page":"Trixi.jl","title":"Trixi.IdealGlmMhdMulticomponentEquations2D","text":"IdealGlmMhdMulticomponentEquations2D\n\nThe ideal compressible multicomponent GLM-MHD equations in two space dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorHennemannGassner","page":"Trixi.jl","title":"Trixi.IndicatorHennemannGassner","text":"IndicatorHennemannGassner(equations::AbstractEquations, basis;\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable)\nIndicatorHennemannGassner(semi::AbstractSemidiscretization;\n alpha_max=0.5,\n alpha_min=0.001,\n alpha_smooth=true,\n variable)\n\nIndicator used for shock-capturing (when passing the equations and the basis) or adaptive mesh refinement (AMR, when passing the semi).\n\nSee also VolumeIntegralShockCapturingHG.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorLöhner","page":"Trixi.jl","title":"Trixi.IndicatorLöhner","text":"IndicatorLöhner (equivalent to IndicatorLoehner)\n\nIndicatorLöhner(equations::AbstractEquations, basis;\n f_wave=0.2, variable)\nIndicatorLöhner(semi::AbstractSemidiscretization;\n f_wave=0.2, variable)\n\nAMR indicator adapted from a FEM indicator by Löhner (1987), also used in the FLASH code as standard AMR indicator. The indicator estimates a weighted second derivative of a specified variable locally.\n\nWhen constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.\n\nReferences\n\nLöhner (1987) \"An adaptive finite element scheme for transient problems in CFD\" doi: 10.1016/0045-7825(87)90098-3\nhttps://flash.rochester.edu/site/flashcode/usersupport/flash4ug_4p62/node59.html#SECTION05163100000000000000\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.IndicatorMax","page":"Trixi.jl","title":"Trixi.IndicatorMax","text":"IndicatorMax(equations::AbstractEquations, basis; variable)\nIndicatorMax(semi::AbstractSemidiscretization; variable)\n\nA simple indicator returning the maximum of variable in an element. When constructed to be used for AMR, pass the semi. Pass the equations, and basis if this indicator should be used for shock capturing.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.InviscidBurgersEquation1D","page":"Trixi.jl","title":"Trixi.InviscidBurgersEquation1D","text":"InviscidBurgersEquation1D\n\nThe inviscid Burgers' equation\n\npartial_t u + frac12 partial_1 u^2 = 0\n\nin one space dimension.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.Isothermal","page":"Trixi.jl","title":"Trixi.Isothermal","text":"struct Isothermal\n\nUsed to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and return a scalar value for the temperature at point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion1D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion1D","text":"LaplaceDiffusion1D(diffusivity, equations)\n\nLaplaceDiffusion1D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion2D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion2D","text":"LaplaceDiffusion2D(diffusivity, equations)\n\nLaplaceDiffusion2D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LaplaceDiffusion3D","page":"Trixi.jl","title":"Trixi.LaplaceDiffusion3D","text":"LaplaceDiffusion3D(diffusivity, equations)\n\nLaplaceDiffusion3D represents a scalar diffusion term nabla cdot (kappanabla u)) with diffusivity kappa applied to each solution component defined by equations.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LatticeBoltzmannEquations2D","page":"Trixi.jl","title":"Trixi.LatticeBoltzmannEquations2D","text":"LatticeBoltzmannEquations2D(; Ma, Re, collision_op=collision_bgk,\n c=1, L=1, rho0=1, u0=nothing, nu=nothing)\n\nThe Lattice-Boltzmann equations\n\npartial_t u_alpha + v_alpha1 partial_1 u_alpha + v_alpha2 partial_2 u_alpha = 0\n\nin two space dimensions for the D2Q9 scheme.\n\nThe characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).\n\nThe nine discrete velocity directions of the D2Q9 scheme are sorted as follows [4]:\n\n 6 2 5 y\n ┌───┼───┐ │\n │ │ │\n 3 ┼ 9 ┼ 1 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 7 4 8 z\n\nNote that usually the velocities are numbered from 0 to 8, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 9, where 1 through 8 correspond to the velocity directions in [4] and 9 is the zero velocity.\n\nThe corresponding opposite directions are:\n\n1 ←→ 3\n2 ←→ 4\n3 ←→ 1\n4 ←→ 2\n5 ←→ 7\n6 ←→ 8\n7 ←→ 5\n8 ←→ 6\n9 ←→ 9\n\nThe main sources for the base implementation were\n\nMisun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024\nKarsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.\nDieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0\nDieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LatticeBoltzmannEquations3D","page":"Trixi.jl","title":"Trixi.LatticeBoltzmannEquations3D","text":"LatticeBoltzmannEquations3D(; Ma, Re, collision_op=collision_bgk,\n c=1, L=1, rho0=1, u0=nothing, nu=nothing)\n\nThe Lattice-Boltzmann equations\n\npartial_t u_alpha + v_alpha1 partial_1 u_alpha + v_alpha2 partial_2 u_alpha + v_alpha3 partial_3 u_alpha = 0\n\nin three space dimensions for the D3Q27 scheme.\n\nThe characteristic Mach number and Reynolds numbers are specified as Ma and Re. By the default, the collision operator collision_op is set to the BGK model. c, L, and rho0 specify the mean thermal molecular velocity, the characteristic length, and the reference density, respectively. They can usually be left to the default values. If desired, instead of the Mach number, one can set the macroscopic reference velocity u0 directly (Ma needs to be set to nothing in this case). Likewise, instead of the Reynolds number one can specify the kinematic viscosity nu directly (in this case, Re needs to be set to nothing).\n\nThe twenty-seven discrete velocity directions of the D3Q27 scheme are sorted as follows [4]:\n\nplane at z = -1:\n 24 17 21 y\n ┌───┼───┐ │\n │ │ │\n 10 ┼ 6 ┼ 15 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 20 12 26 z\nplane at z = 0:\n 14 3 7 y\n ┌───┼───┐ │\n │ │ │\n 2 ┼ 27 ┼ 1 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 8 4 13 z\nplane at z = +1:\n 25 11 19 y\n ┌───┼───┐ │\n │ │ │\n 16 ┼ 5 ┼ 9 ──── x\n │ │ ╱\n └───┼───┘ ╱\n 22 18 23 z\n\nNote that usually the velocities are numbered from 0 to 26, where 0 corresponds to the zero velocity. Due to Julia using 1-based indexing, here we use indices from 1 to 27, where 1 through 26 correspond to the velocity directions in [4] and 27 is the zero velocity.\n\nThe corresponding opposite directions are:\n\n1 ←→ 2\n2 ←→ 1\n3 ←→ 4\n4 ←→ 3\n5 ←→ 6\n6 ←→ 5\n7 ←→ 8\n8 ←→ 7\n9 ←→ 10\n10 ←→ 9\n11 ←→ 12\n12 ←→ 11\n13 ←→ 14\n14 ←→ 13\n15 ←→ 16\n16 ←→ 15\n17 ←→ 18\n18 ←→ 17\n19 ←→ 20\n20 ←→ 19\n21 ←→ 22\n22 ←→ 21\n23 ←→ 24\n24 ←→ 23\n25 ←→ 26\n26 ←→ 25\n27 ←→ 27\n\nThe main sources for the base implementation were\n\nMisun Min, Taehun Lee, A spectral-element discontinuous Galerkin lattice Boltzmann method for nearly incompressible flows, J Comput Phys 230(1), 2011 doi:10.1016/j.jcp.2010.09.024\nKarsten Golly, Anwendung der Lattice-Boltzmann Discontinuous Galerkin Spectral Element Method (LB-DGSEM) auf laminare und turbulente nahezu inkompressible Strömungen im dreidimensionalen Raum, Master Thesis, University of Cologne, 2018.\nDieter Hänel, Molekulare Gasdynamik, Springer-Verlag Berlin Heidelberg, 2004 doi:10.1007/3-540-35047-0\nDieter Krüger et al., The Lattice Boltzmann Method, Springer International Publishing, 2017 doi:10.1007/978-3-319-44649-3\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation1D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation1D","text":"LinearScalarAdvectionEquation1D\n\nThe linear scalar advection equation\n\npartial_t u + a partial_1 u = 0\n\nin one space dimension with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation2D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation2D","text":"LinearScalarAdvectionEquation2D\n\nThe linear scalar advection equation\n\npartial_t u + a_1 partial_1 u + a_2 partial_2 u = 0\n\nin two space dimensions with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearScalarAdvectionEquation3D","page":"Trixi.jl","title":"Trixi.LinearScalarAdvectionEquation3D","text":"LinearScalarAdvectionEquation3D\n\nThe linear scalar advection equation\n\npartial_t u + a_1 partial_1 u + a_2 partial_2 u + a_3 partial_3 u = 0\n\nin three space dimensions with constant velocity a.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LinearizedEulerEquations2D","page":"Trixi.jl","title":"Trixi.LinearizedEulerEquations2D","text":"LinearizedEulerEquations2D(v_mean_global, c_mean_global, rho_mean_global)\n\nLinearized euler equations in two space dimensions. The equations are given by\n\npartial_t\nbeginpmatrix\n rho v_1 v_2 p\nendpmatrix\n+\npartial_x\nbeginpmatrix\n barrho v_1 + barv_1 rho barv_1 v_1 + fracpbarrho barv_1 v_2 barv_1 p + c^2 barrho v_1\nendpmatrix\n+\npartial_y\nbeginpmatrix\n barrho v_2 + barv_2 rho barv_2 v_1 barv_2 v_2 + fracpbarrho barv_2 p + c^2 barrho v_2\nendpmatrix\n=\nbeginpmatrix\n 0 0 0 0\nendpmatrix\n\nThe bar bar(cdot) indicates uniform mean flow variables and c is the speed of sound. The unknowns are the acoustic velocities v = (v_1 v_2), the pressure p and the density rho.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.LobattoLegendreBasis","page":"Trixi.jl","title":"Trixi.LobattoLegendreBasis","text":"LobattoLegendreBasis([RealT=Float64,] polydeg::Integer)\n\nCreate a nodal Lobatto-Legendre basis for polynomials of degree polydeg.\n\nFor the special case polydeg=0 the DG method reduces to a finite volume method. Therefore, this function sets the center point of the cell as single node.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NoSlip","page":"Trixi.jl","title":"Trixi.NoSlip","text":"struct NoSlip\n\nUse to create a no-slip boundary condition with BoundaryConditionNavierStokesWall. The field boundary_value_function should be a function with signature boundary_value_function(x, t, equations) and should return a SVector{NDIMS} whose entries are the velocity vector at a point x and time t.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NonConservativeLocal","page":"Trixi.jl","title":"Trixi.NonConservativeLocal","text":"NonConservativeLocal()\n\nStruct used for multiple dispatch on non-conservative flux functions in the format of \"local * symmetric\". When the argument nonconservative_type is of type NonConservativeLocal, the function returns the local part of the non-conservative term.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.NonConservativeSymmetric","page":"Trixi.jl","title":"Trixi.NonConservativeSymmetric","text":"NonConservativeSymmetric()\n\nStruct used for multiple dispatch on non-conservative flux functions in the format of \"local * symmetric\". When the argument nonconservative_type is of type NonConservativeSymmetric, the function returns the symmetric part of the non-conservative term.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.P4estMesh","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nAn unstructured curved mesh based on trees that uses the C library p4est to manage trees and mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.P4estMesh-Tuple{Any}","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh(trees_per_dimension; polydeg,\n mapping=nothing, faces=nothing, coordinates_min=nothing, coordinates_max=nothing,\n RealT=Float64, initial_refinement_level=0, periodicity=true, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true)\n\nCreate a structured curved P4estMesh of the specified size.\n\nThere are three ways to map the mesh to the physical domain.\n\nDefine a mapping that maps the hypercube [-1, 1]^n.\nSpecify a Tuple faces of functions that parametrize each face.\nCreate a rectangular mesh by specifying coordinates_min and coordinates_max.\n\nNon-periodic boundaries will be called :x_neg, :x_pos, :y_neg, :y_pos, :z_neg, :z_pos.\n\nArguments\n\ntrees_per_dimension::NTupleE{NDIMS, Int}: the number of trees in each dimension.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\nmapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.\nfaces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.P4estMesh-Union{Tuple{String}, Tuple{NDIMS}} where NDIMS","page":"Trixi.jl","title":"Trixi.P4estMesh","text":"P4estMesh{NDIMS}(meshfile::String;\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true,\n boundary_symbols = nothing)\n\nMain mesh constructor for the P4estMesh that imports an unstructured, conforming mesh from an Abaqus mesh file (.inp). Each element of the conforming mesh parsed from the meshfile is created as a p4est tree datatype.\n\nTo create a curved unstructured mesh P4estMesh two strategies are available:\n\np4est_mesh_from_hohqmesh_abaqus: High-order, curved boundary information created by HOHQMesh.jl is available in the meshfile. The mesh polynomial degree polydeg of the boundaries is provided from the meshfile. The computation of the mapped tree coordinates is done with transfinite interpolation with linear blending similar to UnstructuredMesh2D. Boundary name information is also parsed from the meshfile such that different boundary conditions can be set at each named boundary on a given tree.\np4est_mesh_from_standard_abaqus: By default, with mapping=nothing and polydeg=1, this creates a straight-sided from the information parsed from the meshfile. If a mapping function is specified then it computes the mapped tree coordinates via polynomial interpolants with degree polydeg. The mesh created by this function will only have one boundary :all if boundary_symbols is not specified. If boundary_symbols is specified the mesh file will be parsed for nodesets defining the boundary nodes from which boundary edges (2D) and faces (3D) will be assigned.\n\nNote that the mapping and polydeg keyword arguments are only used by the p4est_mesh_from_standard_abaqus function. The p4est_mesh_from_hohqmesh_abaqus function obtains the mesh polydeg directly from the meshfile and constructs the transfinite mapping internally.\n\nThe particular strategy is selected according to the header present in the meshfile where the constructor checks whether or not the meshfile was created with HOHQMesh.jl. If the Abaqus file header is not present in the meshfile then the P4estMesh is created with the function p4est_mesh_from_standard_abaqus.\n\nThe default keyword argument initial_refinement_level=0 corresponds to a forest where the number of trees is the same as the number of elements in the original meshfile. Increasing the initial_refinement_level allows one to uniformly refine the base mesh given in the meshfile to create a forest with more trees before the simulation begins. For example, if a two-dimensional base mesh contains 25 elements then setting initial_refinement_level=1 creates an initial forest of 2^2 * 25 = 100 trees.\n\nArguments\n\nmeshfile::String: an uncurved Abaqus mesh file that can be imported by p4est.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\nboundary_symbols::Vector{Symbol}: A vector of symbols that correspond to the boundary names in the meshfile. If nothing is passed then all boundaries are named :all. \n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ParametersEulerGravity","page":"Trixi.jl","title":"Trixi.ParametersEulerGravity","text":"ParametersEulerGravity(; background_density=0.0,\n gravitational_constant=1.0,\n cfl=1.0,\n resid_tol=1.0e-4,\n n_iterations_max=10^4,\n timestep_gravity=timestep_gravity_erk52_3Sstar!)\n\nSet up parameters for the gravitational part of a SemidiscretizationEulerGravity.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ParsaniKetchesonDeconinck3Sstar32","page":"Trixi.jl","title":"Trixi.ParsaniKetchesonDeconinck3Sstar32","text":"ParsaniKetchesonDeconinck3Sstar32()\n\nParsani, Ketcheson, Deconinck (2013) Optimized explicit RK schemes for the spectral difference method applied to wave propagation problems DOI: 10.1137/120885899\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ParsaniKetchesonDeconinck3Sstar94","page":"Trixi.jl","title":"Trixi.ParsaniKetchesonDeconinck3Sstar94","text":"ParsaniKetchesonDeconinck3Sstar94()\n\nParsani, Ketcheson, Deconinck (2013) Optimized explicit RK schemes for the spectral difference method applied to wave propagation problems DOI: 10.1137/120885899\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PerformanceCounter","page":"Trixi.jl","title":"Trixi.PerformanceCounter","text":"PerformanceCounter()\n\nA PerformanceCounter can be used to track the runtime performance of some calls. Add a new runtime measurement via put!(counter, runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PerformanceCounterList","page":"Trixi.jl","title":"Trixi.PerformanceCounterList","text":"PerformanceCounterList{N}()\n\nA PerformanceCounterList{N} can be used to track the runtime performance of calls to multiple functions, adding them up. Add a new runtime measurement via put!(counter.counters[i], runtime) and get the averaged runtime of all measurements added so far via take!(counter), resetting the counter.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PlotData1D","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D\n\nHolds all relevant data for creating 1D plots of multiple solution variables and to visualize the mesh.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PlotData1D-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D(u, semi [or mesh, equations, solver, cache];\n solution_variables=nothing, nvisnodes=nothing)\n\nCreate a new PlotData1D object that can be used for visualizing 1D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.\n\nnvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.\n\nWhen visualizing data from a two-dimensional simulation, a 1D slice is extracted for plotting. slice specifies the axis along which the slice is extracted and may be :x, or :y. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0). Both of these values are ignored when visualizing 1D data. This applies analogously to three-dimensional simulations, where slice may be :xy, :xz, or :yz.\n\nAnother way to visualize 2D/3D data is by creating a plot along a given curve. This is done with the keyword argument curve. It can be set to a list of 2D/3D points which define the curve. When using curve any other input from slice or point will be ignored.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData1D-Tuple{Union{ODESolution{T, N, uType, uType2, DType, tType, rateType, P} where {T, N, uType, uType2, DType, tType, rateType, P<:(ODEProblem{uType_, tType_, isinplace, P_} where {uType_, tType_, isinplace, P_<:Trixi.AbstractSemidiscretization})}, Trixi.TimeIntegratorSolution}}","page":"Trixi.jl","title":"Trixi.PlotData1D","text":"PlotData1D(sol; kwargs...)\n\nCreate a PlotData1D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2DCartesian","page":"Trixi.jl","title":"Trixi.PlotData2DCartesian","text":"PlotData2D\n\nHolds all relevant data for creating 2D plots of multiple solution variables and to visualize the mesh.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PolytropicEulerEquations2D","page":"Trixi.jl","title":"Trixi.PolytropicEulerEquations2D","text":"PolytropicEulerEquations2D(gamma, kappa)\n\nThe polytropic Euler equations\n\nfracpartialpartial t\nbeginpmatrix\nrho rho v_1 rho v_2\nendpmatrix\n+\nfracpartialpartial x\nbeginpmatrix\n rho v_1 rho v_1^2 + kapparho^gamma rho v_1 v_2\nendpmatrix\n+\nfracpartialpartial y\nbeginpmatrix\nrho v_2 rho v_1 v_2 rho v_2^2 + kapparho^gamma\nendpmatrix\n=\nbeginpmatrix\n0 0 0\nendpmatrix\n\nfor an ideal gas with ratio of specific heats gamma in two space dimensions. Here, rho is the density and v_1 andv_2 the velocities and\n\np = kapparho^gamma\n\nthe pressure, which we replaced using this relation.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.PositivityPreservingLimiterZhangShu","page":"Trixi.jl","title":"Trixi.PositivityPreservingLimiterZhangShu","text":"PositivityPreservingLimiterZhangShu(; threshold, variables)\n\nThe fully-discrete positivity-preserving limiter of\n\nZhang, Shu (2011) Maximum-principle-satisfying and positivity-preserving high-order schemes for conservation laws: survey and new developments doi: 10.1098/rspa.2011.0153\n\nThe limiter is applied to all scalar variables in their given order using the associated thresholds to determine the minimal acceptable values. The order of the variables is important and might have a strong influence on the robustness.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SaveRestartCallback","page":"Trixi.jl","title":"Trixi.SaveRestartCallback","text":"SaveRestartCallback(; interval=0,\n save_final_restart=true,\n output_directory=\"out\")\n\nSave the current numerical solution in a restart file every interval time steps.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SaveSolutionCallback","page":"Trixi.jl","title":"Trixi.SaveSolutionCallback","text":"SaveSolutionCallback(; interval::Integer=0,\n dt=nothing,\n save_initial_solution=true,\n save_final_solution=true,\n output_directory=\"out\",\n solution_variables=cons2prim)\n\nSave the current numerical solution in regular intervals. Either pass interval to save every interval time steps or pass dt to save in intervals of dt in terms of integration time by adding additional (shortened) time steps where necessary (note that this may change the solution). solution_variables can be any callable that converts the conservative variables at a single point to a set of solution variables. The first parameter passed to solution_variables will be the set of conservative variables and the second parameter is the equation struct.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationCoupled","page":"Trixi.jl","title":"Trixi.SemidiscretizationCoupled","text":"SemidiscretizationCoupled\n\nA struct used to bundle multiple semidiscretizations. semidiscretize will return an ODEProblem that synchronizes time steps between the semidiscretizations. Each call of rhs! will call rhs! for each semidiscretization individually. The semidiscretizations can be coupled by gluing meshes together using BoundaryConditionCoupled.\n\nwarning: Experimental code\nThis is an experimental feature and can change any time.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationCoupled-Tuple","page":"Trixi.jl","title":"Trixi.SemidiscretizationCoupled","text":"SemidiscretizationCoupled(semis...)\n\nCreate a coupled semidiscretization that consists of the semidiscretizations passed as arguments.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerAcoustics","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerAcoustics","text":"SemidiscretizationEulerAcoustics(semi_acoustics::SemiAcoustics, semi_euler::SemiEuler;\n source_region=x->true, weights=x->1.0)\n\nwarning: Experimental code\nThis semidiscretization is experimental and may change in any future release.\n\nConstruct a semidiscretization of the acoustic perturbation equations that is coupled with the compressible Euler equations via source terms for the perturbed velocity. Both semidiscretizations have to use the same mesh and solvers with a shared basis. The coupling region is described by a function source_region that maps the coordinates of a single node to true or false depending on whether the point lies within the coupling region or not. A weighting function weights that maps coordinates to weights is applied to the acoustic source terms. Note that this semidiscretization should be used in conjunction with EulerAcousticsCouplingCallback and only works in two dimensions.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerGravity","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerGravity","text":"SemidiscretizationEulerGravity\n\nA struct containing everything needed to describe a spatial semidiscretization of a the compressible Euler equations with self-gravity, reformulating the Poisson equation for the gravitational potential as steady-state problem of the hyperblic diffusion equations.\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) \"A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics\" arXiv: 2008.10593\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationEulerGravity-Union{Tuple{SemiGravity}, Tuple{SemiEuler}, Tuple{Mesh}, Tuple{SemiEuler, SemiGravity, Any}} where {Mesh, SemiEuler<:(SemidiscretizationHyperbolic{Mesh, <:Trixi.AbstractCompressibleEulerEquations}), SemiGravity<:(SemidiscretizationHyperbolic{Mesh, <:Trixi.AbstractHyperbolicDiffusionEquations})}","page":"Trixi.jl","title":"Trixi.SemidiscretizationEulerGravity","text":"SemidiscretizationEulerGravity(semi_euler::SemiEuler, semi_gravity::SemiGravity, parameters)\n\nConstruct a semidiscretization of the compressible Euler equations with self-gravity. parameters should be given as ParametersEulerGravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolic","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolic","text":"SemidiscretizationHyperbolic\n\nA struct containing everything needed to describe a spatial semidiscretization of a hyperbolic conservation law.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolic-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolic","text":"SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;\n source_terms=nothing,\n boundary_conditions=boundary_condition_periodic,\n RealT=real(solver),\n uEltype=RealT,\n initial_cache=NamedTuple())\n\nConstruct a semidiscretization of a hyperbolic PDE.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolicParabolic","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolicParabolic","text":"SemidiscretizationHyperbolicParabolic\n\nA struct containing everything needed to describe a spatial semidiscretization of a mixed hyperbolic-parabolic conservation law.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SemidiscretizationHyperbolicParabolic-Tuple{Any, Tuple, Any, Any}","page":"Trixi.jl","title":"Trixi.SemidiscretizationHyperbolicParabolic","text":"SemidiscretizationHyperbolicParabolic(mesh, both_equations, initial_condition, solver;\n solver_parabolic=default_parabolic_solver(),\n source_terms=nothing,\n both_boundary_conditions=(boundary_condition_periodic, boundary_condition_periodic),\n RealT=real(solver),\n uEltype=RealT,\n both_initial_caches=(NamedTuple(), NamedTuple()))\n\nConstruct a semidiscretization of a hyperbolic-parabolic PDE.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ShallowWaterEquations1D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquations1D","text":"ShallowWaterEquations1D(; gravity, H0 = 0)\n\nShallow water equations (SWE) in one space dimension. The equations are given by\n\nbeginaligned\n fracpartial hpartial t + fracpartialpartial x(h v) = 0 \n fracpartialpartial t(h v) + fracpartialpartial xleft(h v^2 + fracg2h^2right)\n + g h fracpartial bpartial x = 0\nendaligned\n\nThe unknown quantities of the SWE are the water height h and the velocity v. The gravitational constant is denoted by g and the (possibly) variable bottom topography function b(x). Conservative variable water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nThe bottom topography function b(x) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.\n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography by default.\n\nReferences for the SWE are many but a good introduction is available in Chapter 13 of the book:\n\nRandall J. LeVeque (2002) Finite Volume Methods for Hyperbolic Problems DOI: 10.1017/CBO9780511791253\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterEquations2D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquations2D","text":"ShallowWaterEquations2D(; gravity, H0 = 0)\n\nShallow water equations (SWE) in two space dimensions. The equations are given by\n\nbeginaligned\n fracpartial hpartial t + fracpartialpartial x(h v_1)\n + fracpartialpartial y(h v_2) = 0 \n fracpartialpartial t(h v_1) + fracpartialpartial xleft(h v_1^2 + fracg2h^2right)\n + fracpartialpartial y(h v_1 v_2) + g h fracpartial bpartial x = 0 \n fracpartialpartial t(h v_2) + fracpartialpartial x(h v_1 v_2)\n + fracpartialpartial yleft(h v_2^2 + fracg2h^2right) + g h fracpartial bpartial y = 0\nendaligned\n\nThe unknown quantities of the SWE are the water height h and the velocities mathbfv = (v_1 v_2)^T. The gravitational constant is denoted by g and the (possibly) variable bottom topography function b(xy). Conservative variable water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nThe bottom topography function b(xy) is set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero.\n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography must be zero.\nThe bottom topography values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography by default.\n\nReferences for the SWE are many but a good introduction is available in Chapter 13 of the book:\n\nRandall J. LeVeque (2002) Finite Volume Methods for Hyperbolic Problems DOI: 10.1017/CBO9780511791253\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ShallowWaterEquationsQuasi1D","page":"Trixi.jl","title":"Trixi.ShallowWaterEquationsQuasi1D","text":"ShallowWaterEquationsQuasi1D(; gravity, H0 = 0, threshold_limiter = nothing threshold_wet = nothing)\n\nThe quasi-1D shallow water equations (SWE). The equations are given by\n\nbeginaligned\n fracpartialpartial t(a h) + fracpartialpartial x(a h v) = 0 \n fracpartialpartial t(a h v) + fracpartialpartial x(a h v^2)\n + g a h fracpartialpartial x(h + b) = 0\nendaligned\n\nThe unknown quantities of the Quasi-1D SWE are the water height h and the scaled velocity v. The gravitational constant is denoted by g, the (possibly) variable bottom topography function b(x), and (possibly) variable channel width a(x). The water height h is measured from the bottom topography b, therefore one also defines the total water height as H = h + b.\n\nThe additional quantity H_0 is also available to store a reference value for the total water height that is useful to set initial conditions or test the \"lake-at-rest\" well-balancedness.\n\nThe bottom topography function b(x) and channel width a(x) are set inside the initial condition routine for a particular problem setup. To test the conservative form of the SWE one can set the bottom topography variable b to zero and a to one. \n\nIn addition to the unknowns, Trixi.jl currently stores the bottom topography and channel width values at the approximation points despite being fixed in time. This is done for convenience of computing the bottom topography gradients on the fly during the approximation as well as computing auxiliary quantities like the total water height H or the entropy variables. This affects the implementation and use of these equations in various ways:\n\nThe flux values corresponding to the bottom topography and channel width must be zero.\nThe bottom topography and channel width values must be included when defining initial conditions, boundary conditions or source terms.\nAnalysisCallback analyzes this variable.\nTrixi.jl's visualization tools will visualize the bottom topography and channel width by default.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SimpleSSPRK33","page":"Trixi.jl","title":"Trixi.SimpleSSPRK33","text":"SimpleSSPRK33(; stage_callbacks=())\n\nThe third-order SSP Runge-Kutta method of Shu and Osher.\n\nReferences\n\nShu, Osher (1988) \"Efficient Implementation of Essentially Non-oscillatory Shock-Capturing Schemes\" (Eq. 2.18) DOI: 10.1016/0021-9991(88)90177-5\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SteadyStateCallback","page":"Trixi.jl","title":"Trixi.SteadyStateCallback","text":"SteadyStateCallback(; abstol=1.0e-8, reltol=1.0e-6)\n\nTerminates the integration when the residual_steady_state(du, equations) falls below the threshold specified by abstol, reltol.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StepsizeCallback","page":"Trixi.jl","title":"Trixi.StepsizeCallback","text":"StepsizeCallback(; cfl=1.0)\n\nSet the time step size according to a CFL condition with CFL number cfl if the time integration method isn't adaptive itself.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StructuredMesh","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nA structured curved mesh.\n\nDifferent numbers of cells per dimension are possible and arbitrary functions can be used as domain faces.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, coordinates_min, coordinates_max; periodicity=true)\n\nCreate a StructuredMesh that represents a uncurved structured mesh with a rectangular domain.\n\nArguments\n\ncells_per_dimension::NTuple{NDIMS, Int}: the number of cells in each dimension.\ncoordinates_min::NTuple{NDIMS, RealT}: coordinate of the corner in the negative direction of each dimension.\ncoordinates_max::NTuple{NDIMS, RealT}: coordinate of the corner in the positive direction of each dimension.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, mapping; RealT=Float64, unsaved_changes=true, mapping_as_string=mapping2string(mapping, length(cells_per_dimension)))\n\nCreate a StructuredMesh of the given size and shape that uses RealT as coordinate type.\n\nArguments\n\ncells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.\nmapping: a function of NDIMS variables to describe the mapping, which transforms the reference mesh to the physical domain. If no mapping_as_string is defined, this function must be defined with the name mapping to allow for restarts. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.\nRealT::Type: the type that should be used for coordinates.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\nmapping_as_string::String: the code that defines the mapping. If CodeTracking can't find the function definition, it can be passed directly here. The code string must define the mapping function with the name mapping. This will be changed in the future, see https://github.com/trixi-framework/Trixi.jl/issues/541.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.StructuredMesh-Tuple{Any, Tuple}","page":"Trixi.jl","title":"Trixi.StructuredMesh","text":"StructuredMesh(cells_per_dimension, faces; RealT=Float64, unsaved_changes=true, faces_as_string=faces2string(faces))\n\nCreate a StructuredMesh of the given size and shape that uses RealT as coordinate type.\n\nArguments\n\ncells_per_dimension::NTupleE{NDIMS, Int}: the number of cells in each dimension.\nfaces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D).\nRealT::Type: the type that should be used for coordinates.\nperiodicity: either a Bool deciding if all of the boundaries are periodic or an NTuple{NDIMS, Bool} deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SubcellLimiterIDP","page":"Trixi.jl","title":"Trixi.SubcellLimiterIDP","text":"SubcellLimiterIDP(equations::AbstractEquations, basis;\n local_minmax_variables_cons = String[],\n positivity_variables_cons = String[],\n positivity_variables_nonlinear = [],\n positivity_correction_factor = 0.1,\n max_iterations_newton = 10,\n newton_tolerances = (1.0e-12, 1.0e-14),\n gamma_constant_newton = 2 * ndims(equations))\n\nSubcell invariant domain preserving (IDP) limiting used with VolumeIntegralSubcellLimiting including:\n\nLocal maximum/minimum Zalesak-type limiting for conservative variables (local_minmax_variables_cons)\nPositivity limiting for conservative variables (positivity_variables_cons) and nonlinear variables\n\n(positivity_variables_nonlinear)\n\nConservative variables to be limited are passed as a vector of strings, e.g. local_minmax_variables_cons = [\"rho\"] and positivity_variables_cons = [\"rho\"]. For nonlinear variables the specific functions are passed in a vector, e.g. positivity_variables_nonlinear = [pressure].\n\nThe bounds are calculated using the low-order FV solution. The positivity limiter uses positivity_correction_factor such that u^new >= positivity_correction_factor * u^FV. The limiting of nonlinear variables uses a Newton-bisection method with a maximum of max_iterations_newton iterations, relative and absolute tolerances of newton_tolerances and a provisional update constant gamma_constant_newton (gamma_constant_newton>=2*d, where d = #dimensions). See equation (20) of Pazner (2020) and equation (30) of Rueda-Ramírez et al. (2022).\n\nnote: Note\nThis limiter and the correction callback SubcellLimiterIDPCorrection only work together. Without the callback, no correction takes place, leading to a standard low-order FV scheme.\n\nReferences\n\nRueda-Ramírez, Pazner, Gassner (2022) Subcell Limiting Strategies for Discontinuous Galerkin Spectral Element Methods DOI: 10.1016/j.compfluid.2022.105627\nPazner (2020) Sparse invariant domain preserving discontinuous Galerkin methods with subcell convex limiting DOI: 10.1016/j.cma.2021.113876\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SubcellLimiterIDPCorrection","page":"Trixi.jl","title":"Trixi.SubcellLimiterIDPCorrection","text":"SubcellLimiterIDPCorrection()\n\nPerform antidiffusive correction stage for the a posteriori IDP limiter SubcellLimiterIDP called with VolumeIntegralSubcellLimiting.\n\nnote: Note\nThis callback and the actual limiter SubcellLimiterIDP only work together. This is not a replacement but a necessary addition.\n\nReferences\n\nRueda-Ramírez, Pazner, Gassner (2022) Subcell Limiting Strategies for Discontinuous Galerkin Spectral Element Methods DOI: 10.1016/j.compfluid.2022.105627\nPazner (2020) Sparse invariant domain preserving discontinuous Galerkin methods with subcell convex limiting DOI: 10.1016/j.cma.2021.113876\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralStrongForm","page":"Trixi.jl","title":"Trixi.SurfaceIntegralStrongForm","text":"SurfaceIntegralStrongForm(surface_flux=flux_central)\n\nThe classical strong form surface integral type for FD/DG methods.\n\nSee also VolumeIntegralStrongForm.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralUpwind","page":"Trixi.jl","title":"Trixi.SurfaceIntegralUpwind","text":"SurfaceIntegralUpwind(splitting)\n\nCouple elements with upwind simultaneous approximation terms (SATs) that use a particular flux splitting, e.g., splitting_steger_warming.\n\nSee also VolumeIntegralUpwind.\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.SurfaceIntegralWeakForm","page":"Trixi.jl","title":"Trixi.SurfaceIntegralWeakForm","text":"SurfaceIntegralWeakForm(surface_flux=flux_central)\n\nThe classical weak form surface integral type for DG methods as explained in standard textbooks.\n\nSee also VolumeIntegralWeakForm.\n\nReferences\n\nKopriva (2009) Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers doi: 10.1007/978-90-481-2261-5\nHesthaven, Warburton (2007) Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications doi: 10.1007/978-0-387-72067-8\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.T8codeMesh","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nAn unstructured curved mesh based on trees that uses the C library 't8code' to manage trees and mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{Any}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(trees_per_dimension; polydeg, mapping=identity,\n RealT=Float64, initial_refinement_level=0, periodicity=true)\n\nCreate a structured potentially curved 'T8codeMesh' of the specified size.\n\nNon-periodic boundaries will be called ':xneg', ':xpos', ':yneg', ':ypos', ':zneg', ':zpos'.\n\nArguments\n\n'treesperdimension::NTupleE{NDIMS, Int}': the number of trees in each dimension.\n'polydeg::Integer': polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\nmapping: a function of NDIMS variables to describe the mapping that transforms the reference mesh ([-1, 1]^n) to the physical domain. Use only one of mapping, faces and coordinates_min/coordinates_max.\nfaces::NTuple{2*NDIMS}: a tuple of 2 * NDIMS functions that describe the faces of the domain. Each function must take NDIMS-1 arguments. faces[1] describes the face onto which the face in negative x-direction of the unit hypercube is mapped. The face in positive x-direction of the unit hypercube will be mapped onto the face described by faces[2]. faces[3:4] describe the faces in positive and negative y-direction respectively (in 2D and 3D). faces[5:6] describe the faces in positive and negative z-direction respectively (in 3D). Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_min: vector or tuple of the coordinates of the corner in the negative direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\ncoordinates_max: vector or tuple of the coordinates of the corner in the positive direction of each dimension to create a rectangular mesh. Use only one of mapping, faces and coordinates_min/coordinates_max.\n'RealT::Type': the type that should be used for coordinates.\n'initialrefinementlevel::Integer': refine the mesh uniformly to this level before the simulation starts.\n'periodicity': either a 'Bool' deciding if all of the boundaries are periodic or an 'NTuple{NDIMS, Bool}' deciding for each dimension if the boundaries in this dimension are periodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{Ptr{P4est.LibP4est.p4est_connectivity}}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(conn::Ptr{p4est_connectivity}; kwargs...)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.\n\nArguments\n\nconn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{Ptr{P4est.LibP4est.p8est_connectivity}}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(conn::Ptr{p8est_connectivity}; kwargs...)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a p4est_connectivity data structure.\n\nArguments\n\nconn::Ptr{p4est_connectivity}: Pointer to a P4est connectivity object.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{Ptr{T8code.Libt8.t8_cmesh}}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(cmesh::Ptr{t8_cmesh},\n mapping=nothing, polydeg=1, RealT=Float64,\n initial_refinement_level=0)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a t8_cmesh data structure.\n\nArguments\n\ncmesh::Ptr{t8_cmesh}: Pointer to a cmesh object.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.T8codeMesh-Tuple{String, Any}","page":"Trixi.jl","title":"Trixi.T8codeMesh","text":"T8codeMesh(meshfile::String, ndims; kwargs...)\n\nMain mesh constructor for the T8codeMesh that imports an unstructured, conforming mesh from a Gmsh mesh file (.msh).\n\nArguments\n\nmeshfile::String: path to a Gmsh mesh file.\nndims: Mesh file dimension: 2 or 3.\nmapping: a function of NDIMS variables to describe the mapping that transforms the imported mesh to the physical domain. Use nothing for the identity map.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree. The default of 1 creates an uncurved geometry. Use a higher value if the mapping will curve the imported uncurved mesh.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.TimeSeriesCallback","page":"Trixi.jl","title":"Trixi.TimeSeriesCallback","text":"TimeSeriesCallback(semi, point_coordinates;\n interval=1, solution_variables=cons2cons,\n output_directory=\"out\", filename=\"time_series.h5\",\n RealT=real(solver), uEltype=eltype(cache.elements))\n\nCreate a callback that records point-wise data at points given in point_coordinates every interval time steps. The point coordinates are to be specified either as a vector of coordinate tuples or as a two-dimensional array where the first dimension is the point number and the second dimension is the coordinate dimension. By default, the conservative variables are recorded, but this can be controlled by passing a different conversion function to solution_variables.\n\nAfter the last time step, the results are stored in an HDF5 file filename in directory output_directory.\n\nThe real data type RealT and data type for solution variables uEltype default to the respective types used in the solver and the cache.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.TrafficFlowLWREquations1D","page":"Trixi.jl","title":"Trixi.TrafficFlowLWREquations1D","text":"TrafficFlowLWREquations1D\n\nThe classic Lighthill-Witham Richards (LWR) model for 1D traffic flow. The car density is denoted by u in 0 1 and the maximum possible speed (e.g. due to speed limits) is v_textmax.\n\npartial_t u + v_textmax partial_1 u (1 - u) = 0\n\nFor more details see e.g. Section 11.1 of \n\nRandall LeVeque (2002)\n\nFinite Volume Methods for Hyperbolic Problems [DOI: 10.1017/CBO9780511791253]https://doi.org/10.1017/CBO9780511791253\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.TreeMesh","page":"Trixi.jl","title":"Trixi.TreeMesh","text":"TreeMesh{NDIMS} <: AbstractMesh{NDIMS}\n\nA Cartesian mesh based on trees of hypercubes to support adaptive mesh refinement.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.UnstructuredMesh2D","page":"Trixi.jl","title":"Trixi.UnstructuredMesh2D","text":"UnstructuredMesh2D <: AbstractMesh{2}\n\nAn unstructured (possibly curved) quadrilateral mesh.\n\nUnstructuredMesh2D(filename; RealT=Float64, periodicity=false)\n\nAll mesh information, neighbour coupling, and boundary curve information is read in from a mesh file filename.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.UnstructuredSortedBoundaryTypes","page":"Trixi.jl","title":"Trixi.UnstructuredSortedBoundaryTypes","text":"UnstructuredSortedBoundaryTypes\n\nGeneral container to sort the boundary conditions by type for some unstructured meshes/solvers. It stores a set of global indices for each boundary condition type to expedite computation during the call to calc_boundary_flux!. The original dictionary form of the boundary conditions set by the user in the elixir file is also stored for printing.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ViscousFormulationBassiRebay1","page":"Trixi.jl","title":"Trixi.ViscousFormulationBassiRebay1","text":"ViscousFormulationBassiRebay1()\n\nThe classical BR1 flux from\n\nF. Bassi, S. Rebay (1997) A High-Order Accurate Discontinuous Finite Element Method for the Numerical Solution of the Compressible Navier-Stokes Equations DOI: 10.1006/jcph.1996.5572\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.ViscousFormulationLocalDG","page":"Trixi.jl","title":"Trixi.ViscousFormulationLocalDG","text":"ViscousFormulationLocalDG(penalty_parameter)\n\nThe local DG (LDG) flux from \"The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems\" by Cockburn and Shu (1998).\n\nNote that, since this implementation does not involve the parabolic \"upwinding\" vector, the LDG solver is equivalent to ViscousFormulationBassiRebay1 with an LDG-type penalization.\n\nCockburn and Shu (1998). The Local Discontinuous Galerkin Method for Time-Dependent Convection-Diffusion Systems DOI: 10.1137/S0036142997316712\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VisualizationCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.VisualizationCallback","text":"VisualizationCallback(; interval=0,\n solution_variables=cons2prim,\n variable_names=[],\n show_mesh=false,\n plot_data_creator=PlotData2D,\n plot_creator=show_plot,\n plot_arguments...)\n\nCreate a callback that visualizes results during a simulation, also known as in-situ visualization.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in any future releases.\n\nThe interval specifies the number of time step iterations after which a new plot is generated. The available variables to plot are configured with the solution_variables parameter, which acts the same way as for the SaveSolutionCallback. The variables to be actually plotted can be selected by providing a single string or a list of strings to variable_names, and if show_mesh is true, an additional plot with the mesh will be generated.\n\nTo customize the generated figure, plot_data_creator allows to use different plot data types. With plot_creator you can further specify an own function to visualize results, which must support the same interface as the default implementation show_plot. All remaining keyword arguments are collected and passed as additional arguments to the plotting command.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.VolumeIntegralFluxDifferencing","page":"Trixi.jl","title":"Trixi.VolumeIntegralFluxDifferencing","text":"VolumeIntegralFluxDifferencing(volume_flux)\n\nVolume integral type for DG methods based on SBP operators and flux differencing using a symmetric two-point volume_flux. This volume_flux needs to satisfy the interface of numerical fluxes in Trixi.jl.\n\nReferences\n\nLeFloch, Mercier, Rohde (2002) Fully Discrete, Entropy Conservative Schemes of Arbitrary Order doi: 10.1137/S003614290240069X\nFisher, Carpenter (2013) High-order entropy stable finite difference schemes for nonlinear conservation laws: Finite domains doi: 10.1016/j.jcp.2013.06.014\nHendrik Ranocha (2017) Comparison of Some Entropy Conservative Numerical Fluxes for the Euler Equations arXiv: 1701.02264 doi: 10.1007/s10915-017-0618-1\nChen, Shu (2017) Entropy stable high order discontinuous Galerkin methods with suitable quadrature rules for hyperbolic conservation laws doi: 10.1016/j.jcp.2017.05.025\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralPureLGLFiniteVolume","page":"Trixi.jl","title":"Trixi.VolumeIntegralPureLGLFiniteVolume","text":"VolumeIntegralPureLGLFiniteVolume(volume_flux_fv)\n\nA volume integral that only uses the subcell finite volume schemes of the VolumeIntegralShockCapturingHG.\n\nThis gives a formally O(1)-accurate finite volume scheme on an LGL-type subcell mesh (LGL = Legendre-Gauss-Lobatto).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralShockCapturingHG","page":"Trixi.jl","title":"Trixi.VolumeIntegralShockCapturingHG","text":"VolumeIntegralShockCapturingHG(indicator; volume_flux_dg=flux_central,\n volume_flux_fv=flux_lax_friedrichs)\n\nShock-capturing volume integral type for DG methods using a convex blending of the finite volume method with numerical flux volume_flux_fv and the VolumeIntegralFluxDifferencing with volume flux volume_flux_dg. The amount of blending is determined by the indicator, e.g., IndicatorHennemannGassner.\n\nReferences\n\nHennemann, Gassner (2020) \"A provably entropy stable subcell shock capturing approach for high order split form DG\" arXiv: 2008.12044\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralStrongForm","page":"Trixi.jl","title":"Trixi.VolumeIntegralStrongForm","text":"VolumeIntegralStrongForm()\n\nThe classical strong form volume integral type for FD/DG methods.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralSubcellLimiting","page":"Trixi.jl","title":"Trixi.VolumeIntegralSubcellLimiting","text":"VolumeIntegralSubcellLimiting(limiter;\n volume_flux_dg, volume_flux_fv)\n\nA subcell limiting volume integral type for DG methods based on subcell blending approaches with a low-order FV method. Used with limiter SubcellLimiterIDP.\n\nnote: Note\nSubcell limiting methods are not fully functional on non-conforming meshes. This is mainly because the implementation assumes that low- and high-order schemes have the same surface terms, which is not guaranteed for non-conforming meshes. The low-order scheme with a high-order mortar is not invariant domain preserving.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralUpwind","page":"Trixi.jl","title":"Trixi.VolumeIntegralUpwind","text":"VolumeIntegralUpwind(splitting)\n\nSpecialized volume integral for finite difference summation-by-parts (FDSBP) solvers. Can be used together with the upwind SBP operators of Mattsson (2017) implemented in SummationByPartsOperators.jl. The splitting controls the discretization.\n\nSee also splitting_steger_warming, splitting_lax_friedrichs, splitting_vanleer_haenel.\n\nReferences\n\nMattsson (2017) Diagonal-norm upwind SBP operators doi: 10.1016/j.jcp.2017.01.042\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Trixi.VolumeIntegralWeakForm","page":"Trixi.jl","title":"Trixi.VolumeIntegralWeakForm","text":"VolumeIntegralWeakForm()\n\nThe classical weak form volume integral type for DG methods as explained in standard textbooks.\n\nReferences\n\nKopriva (2009) Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers doi: 10.1007/978-90-481-2261-5\nHesthaven, Warburton (2007) Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications doi: 10.1007/978-0-387-72067-8\n\nVolumeIntegralWeakForm() is only implemented for conserved terms as non-conservative terms should always be discretized in conjunction with a flux-splitting scheme, see VolumeIntegralFluxDifferencing. This treatment is required to achieve, e.g., entropy-stability or well-balancedness.\n\n\n\n\n\n","category":"type"},{"location":"reference-trixi/#Base.getindex-Tuple{Trixi.AbstractPlotData, Any}","page":"Trixi.jl","title":"Base.getindex","text":"Base.getindex(pd::AbstractPlotData, variable_name)\n\nExtract a single variable variable_name from pd for plotting with Plots.plot.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Base.resize!-Tuple{Trixi.AbstractContainer, Any}","page":"Trixi.jl","title":"Base.resize!","text":"resize!(c::AbstractContainer, new_length) -> AbstractContainer\n\nResize c to contain new_length elements. If new_length is smaller than the current container length, the first new_length elements will be retained. If new_length is larger, the new elements are invalidated.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.compute_coefficients!-Tuple{Any, Any, Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"PolynomialBases.compute_coefficients!","text":"compute_coefficients!(u_ode, func, t, semi::AbstractSemidiscretization)\n\nSame as compute_coefficients but stores the result in u_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.compute_coefficients-Tuple{Any, Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"PolynomialBases.compute_coefficients","text":"compute_coefficients(func, t, semi::AbstractSemidiscretization)\n\nCompute the discrete coefficients of the continuous function func at time t associated with the semidiscretization semi. For example, the discrete coefficients of func for a discontinuous Galerkin spectral element method (DGSEM) are the values of func at the Lobatto-Legendre nodes. Similarly, a classical finite difference method will use the values of func at the nodes of the grid assoociated with the semidiscretization semi.\n\nFor semidiscretizations semi associated with an initial condition, func can be omitted to use the given initial condition at time t.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.integrate-Tuple{Any, Any, LobattoLegendreBasis}","page":"Trixi.jl","title":"PolynomialBases.integrate","text":"integrate(f, u, basis::LobattoLegendreBasis)\n\nMap the function f to the coefficients u and integrate with respect to the quadrature rule given by basis.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#PolynomialBases.integrate-Union{Tuple{Func}, Tuple{Func, Any, Trixi.AbstractSemidiscretization}} where Func","page":"Trixi.jl","title":"PolynomialBases.integrate","text":"integrate([func=(u_node,equations)->u_node,] u_ode, semi::AbstractSemidiscretization; normalize=true)\n\nCall func(u_node, equations) for each vector of nodal variables u_node in u_ode and integrate the result using a quadrature associated with the semidiscretization semi.\n\nIf normalize is true, the result is divided by the total volume of the computational domain.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SciMLBase.add_tstop!-Tuple{Trixi.SimpleIntegratorSSP, Any}","page":"Trixi.jl","title":"SciMLBase.add_tstop!","text":"add_tstop!(integrator::SimpleIntegratorSSP, t)\n\nAdd a time stop during the time integration process. This function is called after the periodic SaveSolutionCallback to specify the next stop to save the solution.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{SemidiscretizationHyperbolicParabolic, Any}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::SemidiscretizationHyperbolicParabolic, tspan)\n\nWrap the semidiscretization semi as a split ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The parabolic right-hand side is the first function of the split ODE problem and will be used by default by the implicit part of IMEX methods from the SciML ecosystem.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{Trixi.AbstractSemidiscretization, Any, AbstractString}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::AbstractSemidiscretization, tspan, restart_file::AbstractString)\n\nWrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem. The initial condition etc. is taken from the restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#SummationByPartsOperators.semidiscretize-Tuple{Trixi.AbstractSemidiscretization, Any}","page":"Trixi.jl","title":"SummationByPartsOperators.semidiscretize","text":"semidiscretize(semi::AbstractSemidiscretization, tspan)\n\nWrap the semidiscretization semi as an ODE problem in the time interval tspan that can be passed to solve from the SciML ecosystem.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.DGMultiBasis-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.DGMultiBasis","text":"DGMultiBasis(element_type, polydeg; approximation_type = Polynomial(), kwargs...)\n\nConstructs a basis for DGMulti solvers. Returns a \"StartUpDG.RefElemData\" object. The kwargs arguments are additional keyword arguments for RefElemData, such as quad_rule_vol. These are the same as the RefElemData_kwargs used in DGMulti. For more info, see the StartUpDG.jl docs.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.LBMCollisionCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.LBMCollisionCallback","text":"LBMCollisionCallback()\n\nApply the Lattice-Boltzmann method (LBM) collision operator before each time step. See LatticeBoltzmannEquations2D for further details.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.P4estMeshCubedSphere-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.P4estMeshCubedSphere","text":"P4estMeshCubedSphere(trees_per_face_dimension, layers, inner_radius, thickness;\n polydeg, RealT=Float64,\n initial_refinement_level=0, unsaved_changes=true,\n p4est_partition_allow_for_coarsening=true)\n\nBuild a \"Cubed Sphere\" mesh as P4estMesh with 6 * trees_per_face_dimension^2 * layers trees.\n\nThe mesh will have two boundaries, :inside and :outside.\n\nArguments\n\ntrees_per_face_dimension::Integer: the number of trees in the first two local dimensions of each face.\nlayers::Integer: the number of trees in the third local dimension of each face, i.e., the number of layers of the sphere.\ninner_radius::Integer: the inner radius of the sphere.\nthickness::Integer: the thickness of the sphere. The outer radius will be inner_radius + thickness.\npolydeg::Integer: polynomial degree used to store the geometry of the mesh. The mapping will be approximated by an interpolation polynomial of the specified degree for each tree.\nRealT::Type: the type that should be used for coordinates.\ninitial_refinement_level::Integer: refine the mesh uniformly to this level before the simulation starts.\nunsaved_changes::Bool: if set to true, the mesh will be saved to a mesh file.\np4est_partition_allow_for_coarsening::Bool: Must be true when using AMR to make mesh adaptivity independent of domain partitioning. Should be false for static meshes to permit more fine-grained partitioning.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2D-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.PlotData2D","text":"PlotData2D(u, semi [or mesh, equations, solver, cache];\n solution_variables=nothing,\n grid_lines=true, max_supported_level=11, nvisnodes=nothing,\n slice=:xy, point=(0.0, 0.0, 0.0))\n\nCreate a new PlotData2D object that can be used for visualizing 2D/3D DGSEM solution data array u with Plots.jl. All relevant geometrical information is extracted from the semidiscretization semi. By default, the primitive variables (if existent) or the conservative variables (otherwise) from the solution are used for plotting. This can be changed by passing an appropriate conversion function to solution_variables.\n\nIf grid_lines is true, also extract grid vertices for visualizing the mesh. The output resolution is indirectly set via max_supported_level: all data is interpolated to 2^max_supported_level uniformly distributed points in each spatial direction, also setting the maximum allowed refinement level in the solution. nvisnodes specifies the number of visualization nodes to be used. If it is nothing, twice the number of solution DG nodes are used for visualization, and if set to 0, exactly the number of nodes in the DG elements are used.\n\nWhen visualizing data from a three-dimensional simulation, a 2D slice is extracted for plotting. slice specifies the plane that is being sliced and may be :xy, :xz, or :yz. The slice position is specified by a point that lies on it, which defaults to (0.0, 0.0, 0.0). Both of these values are ignored when visualizing 2D data.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nExamples\n\njulia> using Trixi, Plots\n\njulia> trixi_include(default_example())\n[...]\n\njulia> pd = PlotData2D(sol)\nPlotData2D(...)\n\njulia> plot(pd) # To plot all available variables\n\njulia> plot(pd[\"scalar\"]) # To plot only a single variable\n\njulia> plot!(getmesh(pd)) # To add grid lines to the plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.PlotData2D-Tuple{Union{ODESolution{T, N, uType, uType2, DType, tType, rateType, P} where {T, N, uType, uType2, DType, tType, rateType, P<:(ODEProblem{uType_, tType_, isinplace, P_} where {uType_, tType_, isinplace, P_<:Trixi.AbstractSemidiscretization})}, Trixi.TimeIntegratorSolution}}","page":"Trixi.jl","title":"Trixi.PlotData2D","text":"PlotData2D(sol; kwargs...)\n\nCreate a PlotData2D object from a solution object created by either OrdinaryDiffEq.solve! (which returns a SciMLBase.ODESolution) or Trixi.jl's own solve! (which returns a TimeIntegratorSolution).\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ScalarPlotData2D-Tuple{Any, Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.ScalarPlotData2D","text":"ScalarPlotData2D(u, semi::AbstractSemidiscretization; kwargs...)\n\nReturns an PlotData2DTriangulated object which is used to visualize a single scalar field. u should be an array whose entries correspond to values of the scalar field at nodal points.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.SummaryCallback","page":"Trixi.jl","title":"Trixi.SummaryCallback","text":"SummaryCallback()\n\nCreate and return a callback that prints a human-readable summary of the simulation setup at the beginning of a simulation and then resets the timer. When the returned callback is executed directly, the current timer values are shown.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.TrivialCallback-Tuple{}","page":"Trixi.jl","title":"Trixi.TrivialCallback","text":"TrivialCallback()\n\nA callback that does nothing. This can be useful to disable some callbacks easily via trixi_include.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.adapt!-Tuple{T8codeMesh, Any}","page":"Trixi.jl","title":"Trixi.adapt!","text":"Trixi.adapt!(mesh::T8codeMesh, adapt_callback; kwargs...)\n\nAdapt a T8codeMesh according to a user-defined adapt_callback.\n\nArguments\n\nmesh::T8codeMesh: Initialized mesh object.\nadapt_callback: A user-defined callback which tells the adaption routines if an element should be refined, coarsened or stay unchanged.\nThe expected callback signature is as follows:\n`adapt_callback(forest, ltreeid, eclass_scheme, lelemntid, elements, is_family, user_data)`\n # Arguments\n - `forest`: Pointer to the analyzed forest.\n - `ltreeid`: Local index of the current tree where the analyzed elements are part of.\n - `eclass_scheme`: Element class of `elements`.\n - `lelemntid`: Local index of the first element in `elements`.\n - `elements`: Array of elements. If consecutive elements form a family\n they are passed together, otherwise `elements` consists of just one element.\n - `is_family`: Boolean signifying if `elements` represents a family or not.\n - `user_data`: Void pointer to some arbitrary user data. Default value is `C_NULL`.\n # Returns\n -1 : Coarsen family of elements.\n 0 : Stay unchanged.\n 1 : Refine element.\nkwargs:\nrecursive = true: Adapt the forest recursively. If true the caller must ensure that the callback returns 0 for every analyzed element at some point to stop the recursion.\nbalance = true: Make sure the adapted forest is 2^(NDIMS-1):1 balanced.\npartition = true: Partition the forest to redistribute elements evenly among MPI ranks.\nghost = true: Create a ghost layer for MPI data exchange.\nuser_data = C_NULL: Pointer to some arbitrary user-defined data.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.adapt_to_mesh_level!-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.adapt_to_mesh_level!","text":"adapt_to_mesh_level!(u_ode, semi, level)\nadapt_to_mesh_level!(sol::Trixi.TrixiODESolution, level)\n\nLike adapt_to_mesh_level, but modifies the solution and parts of the semidiscretization (mesh and caches) in place.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.adapt_to_mesh_level-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.adapt_to_mesh_level","text":"adapt_to_mesh_level(u_ode, semi, level)\nadapt_to_mesh_level(sol::Trixi.TrixiODESolution, level)\n\nUse the regular adaptive mesh refinement routines to adaptively refine/coarsen the solution u_ode with semidiscretization semi towards a uniformly refined grid with refinement level level. The solution and semidiscretization are copied such that the original objects remain unaltered.\n\nA convenience method accepts an ODE solution object, from which solution and semidiscretization are extracted as needed.\n\nSee also: adapt_to_mesh_level!\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.balance!-Tuple{T8codeMesh}","page":"Trixi.jl","title":"Trixi.balance!","text":"Trixi.balance!(mesh::T8codeMesh)\n\nBalance a T8codeMesh to ensure 2^(NDIMS-1):1 face neighbors.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x","text":"boundary_condition_linear_x(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation1D)\n\nBoundary conditions for initial_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x","text":"boundary_condition_linear_x(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_x_y-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_x_y","text":"boundary_condition_linear_x_y(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_x_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_y-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_y","text":"boundary_condition_linear_y(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation2D)\n\nBoundary conditions for initial_condition_linear_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_linear_z-Tuple{Any, Any, Any, Any, Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_linear_z","text":"boundary_condition_linear_z(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equation::LinearScalarAdvectionEquation1D)\n\nBoundary conditions for boundary_condition_linear_z.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_noslip_wall-Tuple{Any, Any, Any, Any, Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_noslip_wall","text":"boundary_condition_noslip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equations::LatticeBoltzmannEquations2D)\n\nNo-slip wall boundary condition using the bounce-back approach.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_poisson_nonperiodic-Tuple{Any, Any, Any, Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_poisson_nonperiodic","text":"boundary_condition_poisson_nonperiodic(u_inner, orientation, direction, x, t,\n surface_flux_function,\n equations::HyperbolicDiffusionEquations1D)\n\nBoundary conditions used for convergence tests in combination with initial_condition_poisson_nonperiodic and source_terms_poisson_nonperiodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::AcousticPerturbationEquations2D)\n\nUse an orthogonal projection of the perturbed velocities to zero out the normal velocity while retaining the possibility of a tangential velocity in the boundary state. Further details are available in the paper:\n\nMarcus Bauer, Jürgen Dierke and Roland Ewert (2011) Application of a discontinuous Galerkin method to discretize acoustic perturbation equations DOI: 10.2514/1.J050333\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations2D)\n\nShould be used together with StructuredMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations3D)\n\nShould be used together with StructuredMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::CompressibleEulerEquations2D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\n\nDetails about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761\n\nShould be used together with UnstructuredMesh2D.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::CompressibleEulerEquations3D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\n\nDetails about the 1D pressure Riemann solution can be found in Section 6.3.3 of the book\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction 3rd edition DOI: 10.1007/b79761\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, AbstractVector, Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, normal_direction, x, t, surface_flux_function,\n equations::ShallowWaterEquations2D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value. For details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations1D)\n\nDetermine the boundary numerical surface flux for a slip wall condition. Imposes a zero normal velocity at the wall. Density is taken from the internal solution state and pressure is computed as an exact solution of a 1D Riemann problem. Further details about this boundary state are available in the paper:\n\nJ. J. W. van der Vegt and H. van der Ven (2002) Slip flow boundary conditions in discontinuous Galerkin discretizations of the Euler equations of gas dynamics PDF\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations2D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::CompressibleEulerEquations3D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation_or_normal, x, t, surface_flux_function,\n equations::ShallowWaterEquations1D)\n\nCreate a boundary state by reflecting the normal velocity component and keep the tangential velocity component unchanged. The boundary water height is taken from the internal value.\n\nFor details see Section 9.2.5 of the book:\n\nEleuterio F. Toro (2001) Shock-Capturing Methods for Free-Surface Shallow Flows 1st edition ISBN 0471987662\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_slip_wall-Tuple{Any, Any, Any, Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_slip_wall","text":"boundary_condition_slip_wall(u_inner, orientation, direction, x, t,\n surface_flux_function, equations::ShallowWaterEquations2D)\n\nShould be used together with TreeMesh.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_wall-Tuple{Any, Any, Any, Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_wall","text":"boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,\n equations::AcousticPerturbationEquations2D)\n\nBoundary conditions for a solid wall.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.boundary_condition_wall-Tuple{Any, Any, Any, Any, Any, Any, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.boundary_condition_wall","text":"boundary_condition_wall(u_inner, orientation, direction, x, t, surface_flux_function,\n equations::LinearizedEulerEquations2D)\n\nBoundary conditions for a solid wall.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_error_norms-Tuple{Any, Any, Any, Trixi.AbstractSemidiscretization, Any}","page":"Trixi.jl","title":"Trixi.calc_error_norms","text":"calc_error_norms([func=(u_node,equations)->u_node,] u_ode, t, analyzer, semi::AbstractSemidiscretization, cache_analysis)\n\nCalculate discrete L2 and L∞ error norms of func applied to each nodal variable u_node in u_ode. If no exact solution is available, \"errors\" are calculated using some reference state and can be useful for regression tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, direction, equations::IdealGlmMhdEquations1D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations2D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_fast_wavespeed_roe-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.calc_fast_wavespeed_roe","text":"calc_fast_wavespeed_roe(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)\n\nCompute the fast magnetoacoustic wave speed using Roe averages as given by\n\nCargo and Gallice (1997) Roe Matrices for Ideal MHD and Systematic Construction of Roe Matrices for Systems of Conservation Laws DOI: 10.1006/jcph.1997.5773\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_wavespeed_roe-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.calc_wavespeed_roe","text":"calc_wavespeed_roe(u_ll, u_rr, direction::Integer,\n equations::ShallowWaterEquations1D)\n\nCalculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} See for instance equation (62) in \n\nPaul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044\n\nOr equation (9.17) in this lecture notes.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.calc_wavespeed_roe-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.calc_wavespeed_roe","text":"calc_wavespeed_roe(u_ll, u_rr, direction::Integer,\n equations::ShallowWaterEquations2D)\n\nCalculate Roe-averaged velocity v_roe and wavespeed c_roe = sqrt{g * h_roe} depending on direction. See for instance equation (62) in \n\nPaul A. Ullrich, Christiane Jablonowski, and Bram van Leer (2010) High-order finite-volume methods for the shallow-water equations on the sphere DOI: 10.1016/j.jcp.2010.04.044\n\nOr this slides, slides 8 and 9.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.collision_bgk-Tuple{Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.collision_bgk","text":"collision_bgk(u, dt, equations::LatticeBoltzmannEquations2D)\n\nCollision operator for the Bhatnagar, Gross, and Krook (BGK) model.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.collision_bgk-Tuple{Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.collision_bgk","text":"collision_bgk(u, dt, equations::LatticeBoltzmannEquations3D)\n\nCollision operator for the Bhatnagar, Gross, and Krook (BGK) model.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.cons2cons-Tuple{Any, Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.cons2cons","text":"cons2cons(u, equations)\n\nReturn the conserved variables u. While this function is as trivial as identity, it is also as useful.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.cons2entropy","page":"Trixi.jl","title":"Trixi.cons2entropy","text":"cons2entropy(u, equations)\n\nConvert the conserved variables u to the entropy variables for a given set of equations with chosen standard entropy.\n\nu is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by entropy2cons.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.cons2prim","page":"Trixi.jl","title":"Trixi.cons2prim","text":"cons2prim(u, equations)\n\nConvert the conserved variables u to the primitive variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by prim2cons.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.convergence_test-Tuple{Module, AbstractString, Any}","page":"Trixi.jl","title":"Trixi.convergence_test","text":"convergence_test([mod::Module=Main,] elixir::AbstractString, iterations; kwargs...)\n\nRun iterations Trixi.jl simulations using the setup given in elixir and compute the experimental order of convergence (EOC) in the L^2 and L^infty norm. In each iteration, the resolution of the respective mesh will be doubled. Additional keyword arguments kwargs... and the optional module mod are passed directly to trixi_include.\n\nThis function assumes that the spatial resolution is set via the keywords initial_refinement_level (an integer) or cells_per_dimension (a tuple of integers, one per spatial dimension).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_analysis_errors-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.default_analysis_errors","text":"default_analysis_errors(equations)\n\nDefault analysis errors (:l2_error and :linf_error) used by the AnalysisCallback.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_analysis_integrals-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.default_analysis_integrals","text":"default_analysis_integrals(equations)\n\nDefault analysis integrals used by the AnalysisCallback.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_example-Tuple{}","page":"Trixi.jl","title":"Trixi.default_example","text":"default_example()\n\nReturn the path to an example elixir that can be used to quickly see Trixi.jl in action on a TreeMesh. See also examples_dir and get_examples.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.default_example_unstructured-Tuple{}","page":"Trixi.jl","title":"Trixi.default_example_unstructured","text":"default_example_unstructured()\n\nReturn the path to an example elixir that can be used to quickly see Trixi.jl in action on an UnstructuredMesh2D. This simulation is run on the example curved, unstructured mesh given in the Trixi.jl documentation regarding unstructured meshes.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.density-Tuple{Real, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.density","text":"density(p::Real, equations::LatticeBoltzmannEquations2D)\ndensity(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic density from the pressure p or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.density-Tuple{Real, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.density","text":"density(p::Real, equations::LatticeBoltzmannEquations3D)\ndensity(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic density from the pressure p or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.download-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.download","text":"Trixi.download(src_url, file_path)\n\nDownload a file from given src_url to given file_path if file_path is not already a file. This function just returns file_path. This is a small wrapper of Downloads.download(src_url, file_path) that avoids race conditions when multiple MPI ranks are used.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_dof_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_dof_global","text":"each_dof_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the degrees of freedom (DOF) in dg. In particular, not the DOFs themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_face_node-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_face_node","text":"each_face_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the face nodes in dg. In particular, not the face_nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_face_node_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_face_node_global","text":"each_face_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the face nodes in mesh. In particular, not the face nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_quad_node-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_quad_node","text":"each_quad_node(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the quadrature nodes in dg. In particular, not the quadrature nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.each_quad_node_global-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.each_quad_node_global","text":"each_quad_node_global(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the global quadrature nodes in mesh. In particular, not the quadrature nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachboundary-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachboundary","text":"eachboundary(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the boundaries in cache. In particular, not the boundaries themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachcomponent-Tuple{Trixi.AbstractCompressibleEulerMulticomponentEquations}","page":"Trixi.jl","title":"Trixi.eachcomponent","text":"eachcomponent(equations::AbstractCompressibleEulerMulticomponentEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the components in AbstractCompressibleEulerMulticomponentEquations. In particular, not the components themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachcomponent-Tuple{Trixi.AbstractIdealGlmMhdMulticomponentEquations}","page":"Trixi.jl","title":"Trixi.eachcomponent","text":"eachcomponent(equations::AbstractIdealGlmMhdMulticomponentEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the components in AbstractIdealGlmMhdMulticomponentEquations. In particular, not the components themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachdim-Tuple{Any}","page":"Trixi.jl","title":"Trixi.eachdim","text":"eachdim(mesh)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the dimensions in AbstractTree. In particular, not the dimensions themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachdirection-Tuple{Trixi.AbstractTree}","page":"Trixi.jl","title":"Trixi.eachdirection","text":"eachdirection(tree::AbstractTree)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the directions in AbstractTree. In particular, not the directions themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in cache. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{DGMultiMesh, DGMulti{NDIMS, ElemType, ApproxType} where {NDIMS, ElemType, ApproxType}, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(mesh::DGMultiMesh, dg::DGMulti, other_args...)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in mesh. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer1D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer1D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer2D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer2D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.ElementContainer3D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::ElementContainer3D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachelement-Tuple{Trixi.UnstructuredElementContainer2D}","page":"Trixi.jl","title":"Trixi.eachelement","text":"eachelement(elements::UnstructuredElementContainer2D)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the elements in elements. In particular, not the elements themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachinterface-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachinterface","text":"eachinterface(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the interfaces in cache. In particular, not the interfaces themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmortar-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmortar","text":"eachmortar(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the mortars in cache. In particular, not the mortars themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmpiinterface-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmpiinterface","text":"eachmpiinterface(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the MPI interfaces in cache. In particular, not the interfaces themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachmpimortar-Tuple{DG, Any}","page":"Trixi.jl","title":"Trixi.eachmpimortar","text":"eachmpimortar(dg::DG, cache)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the MPI mortars in cache. In particular, not the mortars themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{DG}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(dg::DG)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in dg. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{LobattoLegendreBasis}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(basis::LobattoLegendreBasis)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in basis. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachnode-Tuple{Trixi.LobattoLegendreAnalyzer}","page":"Trixi.jl","title":"Trixi.eachnode","text":"eachnode(analyzer::LobattoLegendreAnalyzer)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the nodes in analyzer. In particular, not the nodes themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.eachvariable-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.eachvariable","text":"eachvariable(equations::AbstractEquations)\n\nReturn an iterator over the indices that specify the location in relevant data structures for the variables in equations. In particular, not the variables themselves are returned.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.energy_internal","page":"Trixi.jl","title":"Trixi.energy_internal","text":"energy_internal(u, equations)\n\nReturn the internal energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.energy_kinetic","page":"Trixi.jl","title":"Trixi.energy_kinetic","text":"energy_kinetic(u, equations)\n\nReturn the kinetic energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.energy_total","page":"Trixi.jl","title":"Trixi.energy_total","text":"energy_total(u, equations)\n\nReturn the total energy of the conserved variables u for a given set of equations, e.g., the CompressibleEulerEquations2D.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.entropy","page":"Trixi.jl","title":"Trixi.entropy","text":"entropy(u, equations)\n\nReturn the chosen entropy of the conserved variables u for a given set of equations.\n\nu is a vector of the conserved variables at a single node, i.e., a vector of the correct length nvariables(equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.entropy2cons","page":"Trixi.jl","title":"Trixi.entropy2cons","text":"entropy2cons(w, equations)\n\nConvert the entropy variables w based on a standard entropy to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2entropy.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.equilibrium_distribution-Tuple{Any, Any, Any, Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.equilibrium_distribution","text":"equilibrium_distribution(alpha, rho, v1, v2, v3, equations::LatticeBoltzmannEquations3D)\n\nCalculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2, v3.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.equilibrium_distribution-Tuple{Any, Any, Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.equilibrium_distribution","text":"equilibrium_distribution(alpha, rho, v1, v2, equations::LatticeBoltzmannEquations2D)\n\nCalculate the local equilibrium distribution for the distribution function with index alpha and given the macroscopic state defined by rho, v1, v2.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.examples_dir-Tuple{}","page":"Trixi.jl","title":"Trixi.examples_dir","text":"examples_dir()\n\nReturn the directory where the example files provided with Trixi.jl are located. If Trixi.jl is installed as a regular package (with ]add Trixi), these files are read-only and should not be modified. To find out which files are available, use, e.g., readdir:\n\nExamples\n\nreaddir(examples_dir())\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux","page":"Trixi.jl","title":"Trixi.flux","text":"flux(u, orientation_or_normal, equations)\n\nGiven the conservative variables u, calculate the (physical) flux in Cartesian direction orientation::Integer or in arbitrary direction normal::AbstractVector for the corresponding set of governing equations. orientation is 1, 2, and 3 for the x-, y-, and z-directions, respectively.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.flux-Tuple{Any, AbstractVector, Trixi.AbstractEquations{1}}","page":"Trixi.jl","title":"Trixi.flux","text":"flux(u, normal_direction::AbstractVector, equations::AbstractEquations{1})\n\nEnables calling flux with a non-integer argument normal_direction for one-dimensional equations. Returns the value of flux(u, 1, equations) scaled by normal_direction[1].\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_central-Tuple{Any, Any, Any, Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.flux_central","text":"flux_central(u_ll, u_rr, orientation_or_normal_direction, equations::AbstractEquations)\n\nThe classical central numerical flux f((u_ll) + f(u_rr)) / 2. When this flux is used as volume flux, the discretization is equivalent to the classical weak form DG method (except floating point errors).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chan_etal-Tuple{Any, Any, Integer, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_chan_etal","text":"@inline function fluxchanetal(ull, urr, orientation::Integer, equations::CompressibleEulerEquationsQuasi1D)\n\nConservative (symmetric) part of the entropy conservative flux for quasi 1D compressible Euler equations split form. This flux is a generalization of flux_ranocha for CompressibleEulerEquations1D. Further details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089 \n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chan_etal-Tuple{Any, Any, Integer, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_chan_etal","text":"flux_chan_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquationsQuasi1D)\n\nTotal energy conservative (mathematical entropy for quasi 1D shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., FluxPlusDissipation(flux_chan_etal, DissipationLocalLaxFriedrichs()).\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)\n\nEntropy conserving two-point flux by\n\nChandrashekar (2013) Kinetic Energy Preserving and Entropy Stable Finite Volume Schemes for Compressible Euler and Navier-Stokes Equations DOI: 10.4208/cicp.170712.010313a\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations1D)\n\nEntropy conserving two-point flux by\n\nAyoub Gouasmi, Karthik Duraisamy (2020) \"Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations\" arXiv:1904.00972v3 [math.NA] 4 Feb 2020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_chandrashekar-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_chandrashekar","text":"flux_chandrashekar(u_ll, u_rr, orientation, equations::CompressibleEulerMulticomponentEquations2D)\n\nAdaption of the entropy conserving two-point flux by\n\nAyoub Gouasmi, Karthik Duraisamy (2020) \"Formulation of Entropy-Stable schemes for the multicomponent compressible Euler equations\" arXiv:1904.00972v3 [math.NA] 4 Feb 2020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations2D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations3D)\n\nEntropy conserving two-point flux by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nEntropy conserving two-point flux adapted by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_derigs_etal-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_derigs_etal","text":"flux_derigs_etal(u_ll, u_rr, orientation, equations::IdealGlmMhdMulticomponentEquations2D)\n\nEntropy conserving two-point flux adapted by\n\nDerigs et al. (2018) Ideal GLM-MHD: About the entropy consistent nine-wave magnetic field divergence diminishing ideal magnetohydrodynamics equations for multicomponent DOI: 10.1016/j.jcp.2018.03.002\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_fjordholm_etal","text":"flux_fjordholm_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquations1D)\n\nTotal energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.\n\nDetails are available in Eq. (4.1) in the paper:\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_fjordholm_etal","text":"flux_fjordholm_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nTotal energy conservative (mathematical entropy for shallow water equations). When the bottom topography is nonzero this should only be used as a surface flux otherwise the scheme will not be well-balanced. For well-balancedness in the volume flux use flux_wintermeyer_etal.\n\nDetails are available in Eq. (4.1) in the paper:\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_godunov-Tuple{Any, Any, Integer, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_godunov","text":"flux_godunov(u_ll, u_rr, orientation_or_normal_direction,\n equations::LinearizedEulerEquations2D)\n\nAn upwind flux for the linearized Euler equations based on diagonalization of the physical flux matrix. Given the physical flux Au, A=T Lambda T^-1 with Lambda being a diagonal matrix that holds the eigenvalues of A, decompose Lambda = Lambda^+ + Lambda^- where Lambda^+ and Lambda^- are diagonal matrices holding the positive and negative eigenvalues of A, respectively. Then for left and right states u_L u_R, the numerical flux calculated by this function is given by A^+ u_L + A^- u_R where A^pm = T Lambda^pm T^-1.\n\nThe diagonalization of the flux matrix can be found in\n\nR. F. Warming, Richard M. Beam and B. J. Hyett (1975) Diagonalization and simultaneous symmetrization of the gas-dynamic matrices DOI: 10.1090/S0025-5718-1975-0388967-5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations1D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations2D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdEquations3D)\n\nEntropy conserving and kinetic energy preserving two-point flux of Hindenlang and Gassner (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdMulticomponentEquations1D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hindenlang_gassner-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hindenlang_gassner","text":"flux_hindenlang_gassner(u_ll, u_rr, orientation_or_normal_direction,\n equations::IdealGlmMhdMulticomponentEquations2D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux of Hindenlang (2019), extending flux_ranocha to the MHD equations.\n\nReferences\n\nFlorian Hindenlang, Gregor Gassner (2019) A new entropy conservative two-point flux for ideal MHD equations derived from first principles. Presented at HONOM 2019: European workshop on high order numerical methods for evolutionary PDEs, theory and applications\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations2D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations3D)\n\nComputes the HLLC flux (HLL with Contact) for compressible Euler equations developed by E.F. Toro Lecture slides Signal speeds: DOI: 10.1137/S1064827593260140\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_hllc-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.flux_hllc","text":"flux_hllc(u_ll, u_rr, orientation, equations::IdealGlmMhdEquations1D)\n\nLi (2005)\n\nAn HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_kennedy_gruber-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_kennedy_gruber","text":"flux_kennedy_gruber(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nKinetic energy preserving two-point flux by\n\nKennedy and Gruber (2008) Reduced aliasing formulations of the convective terms within the Navier-Stokes equations for a compressible fluid DOI: 10.1016/j.jcp.2007.09.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_audusse_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_audusse_etal","text":"flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.\n\nThis hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations1D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.\n\nFurther details on the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_audusse_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_audusse_etal","text":"flux_nonconservative_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_audusse_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point surface flux that discretizes the nonconservative (source) term. The discretization uses the hydrostatic_reconstruction_audusse_etal on the conservative variables.\n\nThis hydrostatic reconstruction ensures that the finite volume numerical fluxes remain well-balanced for discontinuous bottom topographies ShallowWaterEquations2D. Should be used together with FluxHydrostaticReconstruction and hydrostatic_reconstruction_audusse_etal in the surface flux to ensure consistency.\n\nFurther details for the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chan_etal-Tuple{Any, Any, Integer, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chan_etal","text":"flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,\n equations::CompressibleEulerEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_direction, \n equations::CompressibleEulerEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_ll, normal_rr,\n equations::CompressibleEulerEquationsQuasi1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the pressure CompressibleEulerEquationsQuasi1D and the nozzle width.\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089 \n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_chan_etal-Tuple{Any, Any, Integer, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_chan_etal","text":"flux_nonconservative_chan_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquationsQuasi1D)\nflux_nonconservative_chan_etal(u_ll, u_rr, normal_direction::AbstractVector,\n equations::ShallowWaterEquationsQuasi1D) \nflux_nonconservative_chan_etal(u_ll, u_rr, \n normal_ll::AbstractVector, normal_rr::AbstractVector,\n equations::ShallowWaterEquationsQuasi1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquationsQuasi1D and the channel width.\n\nFurther details are available in the paper:\n\nJesse Chan, Khemraj Shukla, Xinhui Wu, Ruofeng Liu, Prani Nalluri (2023) High order entropy stable schemes for the quasi-one-dimensional shallow water and compressible Euler equations DOI: 10.48550/arXiv.2307.12089\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.\n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_ersing_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_ersing_etal","text":"flux_nonconservative_ersing_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_ersing_etal(u_ll, u_rr,\n normal_direction_ll::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nwarning: Experimental code\nThis numerical flux is experimental and may change in any future release.\n\nNon-symmetric path-conservative two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nThis is a modified version of flux_nonconservative_wintermeyer_etal that gives entropy conservation and well-balancedness in both the volume and surface when combined with flux_wintermeyer_etal.\n\nFor further details see:\n\nPatrick Ersing, Andrew R. Winters (2023) An entropy stable discontinuous Galerkin method for the two-layer shallow water equations on curvilinear meshes DOI: 10.48550/arXiv.2306.12699\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_fjordholm_etal","text":"flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nThis contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.\n\nFurther details for the original finite volume formulation are available in\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\nand for curvilinear 2D case in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_fjordholm_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_fjordholm_etal","text":"flux_nonconservative_fjordholm_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_fjordholm_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point surface flux discretizing the nonconservative (source) term of that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nThis contains additional terms compared to flux_nonconservative_wintermeyer_etal that account for possible discontinuities in the bottom topography function. Thus, this flux should be used in general at interfaces. For flux differencing volume terms, flux_nonconservative_wintermeyer_etal is analytically equivalent but slightly cheaper.\n\nFurther details for the original finite volume formulation are available in\n\nUlrik S. Fjordholm, Siddhartha Mishr and Eitan Tadmor (2011) Well-balanced and energy stable schemes for the shallow water equations with discontinuous topography DOI: 10.1016/j.jcp.2011.03.042\n\nand for curvilinear 2D case in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdEquations2D)\nflux_nonconservative_powell(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::IdealGlmMhdEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdEquations3D)\nflux_nonconservative_powell(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::IdealGlmMhdEquations3D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations3D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell-Tuple{Any, Any, Integer, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell","text":"flux_nonconservative_powell(u_ll, u_rr, orientation::Integer,\n equations::IdealGlmMhdMulticomponentEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdMulticomponentEquations2D.\n\nReferences\n\nMarvin Bohm, Andrew R.Winters, Gregor J. Gassner, Dominik Derigs, Florian Hindenlang, Joachim Saur An entropy stable nodal discontinuous Galerkin method for the resistive MHD equations. Part I: Theory and numerical verification DOI: 10.1016/j.jcp.2018.06.027\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D, Trixi.NonConservativeSymmetric, Integer}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,\n equations::IdealGlmMhdEquations2D,\n nonconservative_type::NonConservativeSymmetric,\n nonconservative_term::Integer)\n\nSymmetric part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered \"fluxes\" for subcell limiting. See, e.g.,\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\nThis function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, u_rr,\n orientation::Integer,\n equations::IdealGlmMhdEquations2D)\n\nNon-symmetric two-point flux discretizing the nonconservative (source) term of Powell and the Galilean nonconservative term associated with the GLM multiplier of the IdealGlmMhdEquations2D.\n\nThis implementation uses a non-conservative term that can be written as the product of local and symmetric parts. It is equivalent to the non-conservative flux of Bohm et al. (flux_nonconservative_powell) for conforming meshes but it yields different results on non-conforming meshes(!).\n\nThe two other flux functions with the same name return either the local or symmetric portion of the non-conservative flux based on the type of the nonconservativetype argument, employing multiple dispatch. They are used to compute the subcell fluxes in dg2dsubcelllimiters.jl.\n\nReferences\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_powell_local_symmetric-Tuple{Any, Integer, IdealGlmMhdEquations2D, Trixi.NonConservativeLocal, Integer}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_powell_local_symmetric","text":"flux_nonconservative_powell_local_symmetric(u_ll, orientation::Integer,\n equations::IdealGlmMhdEquations2D,\n nonconservative_type::NonConservativeLocal,\n nonconservative_term::Integer)\n\nLocal part of the Powell and GLM non-conservative terms. Needed for the calculation of the non-conservative staggered \"fluxes\" for subcell limiting. See, e.g.,\n\nRueda-Ramírez, Gassner (2023). A Flux-Differencing Formula for Split-Form Summation By Parts Discretizations of Non-Conservative Systems. https://arxiv.org/pdf/2211.14009.pdf.\n\nThis function is used to compute the subcell fluxes in dg2dsubcell_limiters.jl.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_wintermeyer_etal","text":"flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations1D.\n\nFurther details are available in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_nonconservative_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_nonconservative_wintermeyer_etal","text":"flux_nonconservative_wintermeyer_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations2D)\nflux_nonconservative_wintermeyer_etal(u_ll, u_rr,\n normal_direction_ll ::AbstractVector,\n normal_direction_average::AbstractVector,\n equations::ShallowWaterEquations2D)\n\nNon-symmetric two-point volume flux discretizing the nonconservative (source) term that contains the gradient of the bottom topography ShallowWaterEquations2D.\n\nOn curvilinear meshes, this nonconservative flux depends on both the contravariant vector (normal direction) at the current node and the averaged one. This is different from numerical fluxes used to discretize conservative terms.\n\nFurther details are available in the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction, equations::CompressibleEulerEquations1D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nEntropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerMulticomponentEquations1D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha-Tuple{Any, Any, Integer, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.flux_ranocha","text":"flux_ranocha(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerMulticomponentEquations2D)\n\nAdaption of the entropy conserving and kinetic energy preserving two-point flux by\n\nHendrik Ranocha (2018) Generalised Summation-by-Parts Operators and Entropy Stability of Numerical Methods for Hyperbolic Balance Laws PhD thesis, TU Braunschweig\n\nSee also\n\nHendrik Ranocha (2020) Entropy Conserving and Kinetic Energy Preserving Numerical Methods for the Euler Equations Using Summation-by-Parts Operators Proceedings of ICOSAHOM 2018\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_ranocha_turbo-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.flux_ranocha_turbo","text":"flux_ranocha_turbo(u_ll, u_rr, orientation_or_normal_direction, equations)\n\nEquivalent to flux_ranocha except that it may use specialized methods, e.g., when used with VolumeIntegralFluxDifferencing. These specialized methods may enable better use of SIMD instructions to increase runtime efficiency on modern hardware.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations2D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.flux_shima_etal","text":"flux_shima_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::CompressibleEulerEquations3D)\n\nThis flux is is a modification of the original kinetic energy preserving two-point flux by\n\nYuichi Kuya, Kosuke Totani and Soshi Kawai (2018) Kinetic energy and entropy preserving schemes for compressible flows by split convective forms DOI: 10.1016/j.jcp.2018.08.058\n\nThe modification is in the energy flux to guarantee pressure equilibrium and was developed by\n\nNao Shima, Yuichi Kuya, Yoshiharu Tamaki, Soshi Kawai (JCP 2020) Preventing spurious pressure oscillations in split convective form discretizations for compressible flows DOI: 10.1016/j.jcp.2020.110060\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_shima_etal_turbo-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.flux_shima_etal_turbo","text":"flux_shima_etal_turbo(u_ll, u_rr, orientation_or_normal_direction, equations)\n\nEquivalent to flux_shima_etal except that it may use specialized methods, e.g., when used with VolumeIntegralFluxDifferencing. These specialized methods may enable better use of SIMD instructions to increase runtime efficiency on modern hardware.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation,\n equations::ShallowWaterEquations1D)\n\nTotal energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_wintermeyer_etal-Tuple{Any, Any, Integer, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.flux_wintermeyer_etal","text":"flux_wintermeyer_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nTotal energy conservative (mathematical entropy for shallow water equations) split form. When the bottom topography is nonzero this scheme will be well-balanced when used as a volume_flux. The surface_flux should still use, e.g., flux_fjordholm_etal.\n\nFurther details are available in Theorem 1 of the paper:\n\nNiklas Wintermeyer, Andrew R. Winters, Gregor J. Gassner and David A. Kopriva (2017) An entropy stable nodal discontinuous Galerkin method for the two dimensional shallow water equations on unstructured curvilinear meshes with discontinuous bathymetry DOI: 10.1016/j.jcp.2017.03.036\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.flux_winters_etal-Tuple{Any, Any, AbstractVector, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.flux_winters_etal","text":"flux_winters_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::PolytropicEulerEquations2D)\n\nEntropy conserving two-point flux for isothermal or polytropic gases. Requires a special weighted Stolarsky mean for the evaluation of the density denoted here as stolarsky_mean. Note, for isothermal gases where gamma = 1 this stolarsky_mean becomes the ln_mean.\n\nFor details see Section 3.2 of the following reference\n\nAndrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_boundary_outer_state-Tuple{BoundaryConditionDirichlet, Any, Any, Any, Any, Vararg{Any}}","page":"Trixi.jl","title":"Trixi.get_boundary_outer_state","text":"get_boundary_outer_state(boundary_condition::BoundaryConditionDirichlet,\n cache, t, equations, dg, indices...)\n\nFor subcell limiting, the calculation of local bounds for non-periodic domains require the boundary outer state. This function returns the boundary value at time t and for node with spatial indices indices.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_examples-Tuple{}","page":"Trixi.jl","title":"Trixi.get_examples","text":"get_examples()\n\nReturn a list of all example elixirs that are provided by Trixi.jl. See also examples_dir and default_example.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_name-Tuple{Any}","page":"Trixi.jl","title":"Trixi.get_name","text":"get_name(x)\n\nReturns a name of x ready for pretty printing. By default, return string(y) if x isa Val{y} and return string(x) otherwise.\n\nExamples\n\njulia> Trixi.get_name(\"test\")\n\"test\"\n\njulia> Trixi.get_name(Val(:test))\n\"test\"\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.get_name-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.get_name","text":"get_name(equations::AbstractEquations)\n\nReturns the canonical, human-readable name for the given system of equations.\n\nExamples\n\njulia> Trixi.get_name(CompressibleEulerEquations1D(1.4))\n\"CompressibleEulerEquations1D\"\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.getmesh-Tuple{Trixi.AbstractPlotData}","page":"Trixi.jl","title":"Trixi.getmesh","text":"getmesh(pd::AbstractPlotData)\n\nExtract grid lines from pd for plotting with Plots.plot.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.global_mean_vars-Tuple{AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.global_mean_vars","text":"global_mean_vars(equations::AcousticPerturbationEquations2D)\n\nReturns the global mean variables stored in equations. This makes it easier to define flexible initial conditions for problems with constant mean flow.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.have_nonconservative_terms-Tuple{Trixi.AbstractEquations}","page":"Trixi.jl","title":"Trixi.have_nonconservative_terms","text":"have_nonconservative_terms(equations)\n\nTrait function determining whether equations represent a conservation law with or without nonconservative terms. Classical conservation laws such as the CompressibleEulerEquations2D do not have nonconservative terms. The ShallowWaterEquations2D with non-constant bottom topography are an example of equations with nonconservative terms. The return value will be True() or False() to allow dispatching on the return type.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_audusse_etal-Tuple{Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_audusse_etal","text":"hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation::Integer,\n equations::ShallowWaterEquations1D)\n\nA particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations1D. The reconstructed solution states u_ll_star and u_rr_star variables are then used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details on this hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.hydrostatic_reconstruction_audusse_etal-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.hydrostatic_reconstruction_audusse_etal","text":"hydrostatic_reconstruction_audusse_etal(u_ll, u_rr, orientation_or_normal_direction,\n equations::ShallowWaterEquations2D)\n\nA particular type of hydrostatic reconstruction on the water height to guarantee well-balancedness for a general bottom topography ShallowWaterEquations2D. The reconstructed solution states u_ll_star and u_rr_star variables are used to evaluate the surface numerical flux at the interface. Use in combination with the generic numerical flux routine FluxHydrostaticReconstruction.\n\nFurther details for the hydrostatic reconstruction and its motivation can be found in\n\nEmmanuel Audusse, François Bouchut, Marie-Odile Bristeau, Rupert Klein, and Benoit Perthame (2004) A fast and stable well-balanced scheme with hydrostatic reconstruction for shallow water flows DOI: 10.1137/S1064827503431090\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_mpi-Tuple{}","page":"Trixi.jl","title":"Trixi.init_mpi","text":"init_mpi()\n\nInitialize MPI by calling MPI.Initialized(). The function will check if MPI is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_p4est-Tuple{}","page":"Trixi.jl","title":"Trixi.init_p4est","text":"init_p4est()\n\nInitialize p4est by calling p4est_init and setting the log level to SC_LP_ERROR. This function will check if p4est is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.init_t8code-Tuple{}","page":"Trixi.jl","title":"Trixi.init_t8code","text":"init_t8code()\n\nInitialize t8code by calling sc_init, p4est_init, and t8_init while setting the log level to SC_LP_ERROR. This function will check if t8code is already initialized and if yes, do nothing, thus it is safe to call it multiple times.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::AcousticPerturbationEquations2D)\n\nA constant initial condition where the state variables are zero and the mean flow is constant. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::CompressibleEulerEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::IdealGlmMhdEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::InviscidBurgersEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LatticeBoltzmannEquations2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LatticeBoltzmannEquations3D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_constant-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_constant","text":"initial_condition_constant(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA constant initial condition to test free-stream preservation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::AcousticPerturbationEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquations3D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerEquationsQuasi1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations1D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations2D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdEquations3D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdMulticomponentEquations1D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::IdealGlmMhdMulticomponentEquations2D)\n\nAn Alfvén wave as smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::InviscidBurgersEquation1D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA smooth initial condition used for convergence tests (in combination with BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, LinearizedEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::LinearizedEulerEquations2D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::PolytropicEulerEquations2D)\n\nManufactured smooth initial condition used for convergence tests in combination with source_terms_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterEquations2D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::ShallowWaterEquationsQuasi1D)\n\nA smooth initial condition used for convergence tests in combination with source_terms_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_convergence_test-Tuple{Any, Any, TrafficFlowLWREquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_convergence_test","text":"initial_condition_convergence_test(x, t, equations::TrafficFlowLWREquations1D)\n\nA smooth initial condition used for convergence tests.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_density_wave-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_density_wave","text":"initial_condition_density_wave(x, t, equations::CompressibleEulerEquations1D)\n\nA sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper\n\nGregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026\n\nwith the following parameters\n\ndomain [-1, 1]\nmesh = 4x4\npolydeg = 5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_density_wave-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_density_wave","text":"initial_condition_density_wave(x, t, equations::CompressibleEulerEquations2D)\n\nA sine wave in the density with constant velocity and pressure; reduces the compressible Euler equations to the linear advection equations. This setup is the test case for stability of EC fluxes from paper\n\nGregor J. Gassner, Magnus Svärd, Florian J. Hindenlang (2020) Stability issues of entropy-stable and/or split-form high-order schemes arXiv: 2007.09026\n\nwith the following parameters\n\ndomain [-1, 1]\nmesh = 4x4\npolydeg = 5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations1D)\n\nOne dimensional variant of the setup used for convergence tests of the Euler equations with self-gravity from\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nnote: Note\nThere is no additional source term necessary for the manufactured solution in one spatial dimension. Thus, source_terms_eoc_test_coupled_euler_gravity is not present there.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_eoc_test_coupled_euler_gravity or source_terms_eoc_test_euler.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations1D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_eoc_test_coupled_euler_gravity-Tuple{Any, Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_eoc_test_coupled_euler_gravity","text":"initial_condition_eoc_test_coupled_euler_gravity(x, t, equations::HyperbolicDiffusionEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with source_terms_harmonic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::AcousticPerturbationEquations2D)\n\nA Gaussian pulse in a constant mean flow. Uses the global mean values from equations.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA Gaussian pulse used together with BoundaryConditionDirichlet(initial_condition_gauss).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equation::LinearScalarAdvectionEquation2D)\n\nA Gaussian pulse used together with BoundaryConditionDirichlet(initial_condition_gauss).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_gauss-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_gauss","text":"initial_condition_gauss(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA Gaussian pulse.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x","text":"initial_condition_linear_x(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA linear function of x[1] used together with boundary_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x","text":"initial_condition_linear_x(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] used together with boundary_condition_linear_x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_x_y-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_x_y","text":"initial_condition_linear_x_y(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] + x[2] used together with boundary_condition_linear_x_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_y-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_y","text":"initial_condition_linear_y(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA linear function of x[1] used together with boundary_condition_linear_y.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_linear_z-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_linear_z","text":"initial_condition_linear_z(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA linear function of x[3] used together with boundary_condition_linear_z.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_poisson_nonperiodic-Tuple{Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_poisson_nonperiodic","text":"initial_condition_poisson_nonperiodic(x, t, equations::HyperbolicDiffusionEquations1D)\n\nA non-priodic smooth initial condition. Can be used for convergence tests in combination with source_terms_poisson_nonperiodic and boundary_condition_poisson_nonperiodic.\n\nnote: Note\nThe solution is periodic but the initial guess is not.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin-Tuple{Any, Any, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin","text":"initial_condition_sin(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin-Tuple{Any, Any, LinearScalarAdvectionEquation3D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin","text":"initial_condition_sin(x, t, equations::LinearScalarAdvectionEquation1D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_sin_sin-Tuple{Any, Any, LinearScalarAdvectionEquation2D}","page":"Trixi.jl","title":"Trixi.initial_condition_sin_sin","text":"initial_condition_sin_sin(x, t, equations::LinearScalarAdvectionEquation2D)\n\nA sine wave in the conserved variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations1D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations2D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerEquations3D)\n\nA weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nA for multicomponent adapted weak blast wave adapted to multicomponent and taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nA for multicomponent adapted weak blast wave taken from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations1D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdEquations3D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations1D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, IdealGlmMhdMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::IdealGlmMhdMulticomponentEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::PolytropicEulerEquations2D)\n\nA weak blast wave adapted from\n\nSebastian Hennemann, Gregor J. Gassner (2020) A provably entropy stable subcell shock capturing approach for high order split form DG arXiv: 2008.12044\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, ShallowWaterEquations1D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations1D)\n\nA weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.initial_condition_weak_blast_wave-Tuple{Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.initial_condition_weak_blast_wave","text":"initial_condition_weak_blast_wave(x, t, equations::ShallowWaterEquations2D)\n\nA weak blast wave discontinuity useful for testing, e.g., total energy conservation. Note for the shallow water equations to the total energy acts as a mathematical entropy function.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.integrate_via_indices-Union{Tuple{Func}, Tuple{Func, Any, Trixi.AbstractSemidiscretization, Vararg{Any}}} where Func","page":"Trixi.jl","title":"Trixi.integrate_via_indices","text":"integrate_via_indices(func, u_ode, semi::AbstractSemidiscretization, args...; normalize=true)\n\nCall func(u, i..., element, equations, solver, args...) for all nodal indices i..., element and integrate the result using a quadrature associated with the semidiscretization semi.\n\nIf normalize is true, the result is divided by the total volume of the computational domain.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.inv_ln_mean-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.inv_ln_mean","text":"inv_ln_mean(x, y)\n\nCompute the inverse 1 / ln_mean(x, y) of the logarithmic mean ln_mean.\n\nThis function may be used to increase performance where the inverse of the logarithmic mean is needed, by replacing a (slow) division by a (fast) multiplication.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.jacobian_ad_forward-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.jacobian_ad_forward","text":"jacobian_ad_forward(semi::AbstractSemidiscretization;\n t0=zero(real(semi)),\n u0_ode=compute_coefficients(t0, semi))\n\nUses the right-hand side operator of the semidiscretization semi and forward mode automatic differentiation to compute the Jacobian J of the semidiscretization semi at state u0_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.jacobian_fd-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.jacobian_fd","text":"jacobian_fd(semi::AbstractSemidiscretization;\n t0=zero(real(semi)),\n u0_ode=compute_coefficients(t0, semi))\n\nUses the right-hand side operator of the semidiscretization semi and simple second order finite difference to compute the Jacobian J of the semidiscretization semi at state u0_ode.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.linear_structure-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.linear_structure","text":"linear_structure(semi::AbstractSemidiscretization;\n t0=zero(real(semi)))\n\nWraps the right-hand side operator of the semidiscretization semi at time t0 as an affine-linear operator given by a linear operator A and a vector b.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ln_mean-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.ln_mean","text":"ln_mean(x, y)\n\nCompute the logarithmic mean\n\nln_mean(x, y) = (y - x) / (log(y) - log(x)) = (y - x) / log(y / x)\n\nProblem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Ismail and Roe (2009). Set ξ = y / x. Then, we have\n\n(y - x) / log(y / x) = (x + y) / log(ξ) * (ξ - 1) / (ξ + 1)\n\nSet f = (ξ - 1) / (ξ + 1) = (y - x) / (x + y). Then, we use the expansion\n\nlog(ξ) = 2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7) + O(ξ^9)\n\nInserting the first few terms of this expansion yields\n\n(y - x) / log(ξ) ≈ (x + y) * f / (2 * f * (1 + f^2 / 3 + f^4 / 5 + f^6 / 7))\n = (x + y) / (2 + 2/3 * f^2 + 2/5 * f^4 + 2/7 * f^6)\n\nSince divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use\n\nf^2 = (y - x)^2 / (x + y)^2\n = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)\n\nGiven ε = 1.0e-4, we use the following algorithm.\n\nif f^2 < ε\n # use the expansion above\nelse\n # use the direct formula (y - x) / log(y / x)\nend\n\nReferences\n\nIsmail, Roe (2009). Affordable, entropy-consistent Euler flux functions II: Entropy production at shocks. DOI: 10.1016/j.jcp.2009.04.021\nAgner Fog. Lists of instruction latencies, throughputs and micro-operation breakdowns for Intel, AMD, and VIA CPUs. https://www.agner.org/optimize/instruction_tables.pdf\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_adaptive_time_integrator!-Tuple{Any, AbstractString}","page":"Trixi.jl","title":"Trixi.load_adaptive_time_integrator!","text":"load_adaptive_time_integrator!(integrator, restart_file::AbstractString)\n\nLoad the context information for time integrators with error-based step size control saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_dt-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_dt","text":"load_dt(restart_file::AbstractString)\n\nLoad the time step size (dt in OrdinaryDiffEq.jl) saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_mesh-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_mesh","text":"load_mesh(restart_file::AbstractString; n_cells_max)\n\nLoad the mesh from the restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_time-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_time","text":"load_time(restart_file::AbstractString)\n\nLoad the time saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_timestep!-Tuple{Any, AbstractString}","page":"Trixi.jl","title":"Trixi.load_timestep!","text":"load_timestep!(integrator, restart_file::AbstractString)\n\nLoad the time step number saved in a restart_file and assign it to both the time step number and and the number of accepted steps (iter and stats.naccept in OrdinaryDiffEq.jl, respectively) in integrator.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.load_timestep-Tuple{AbstractString}","page":"Trixi.jl","title":"Trixi.load_timestep","text":"load_timestep(restart_file::AbstractString)\n\nLoad the time step number (iter in OrdinaryDiffEq.jl) saved in a restart_file.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.max-Tuple","page":"Trixi.jl","title":"Trixi.max","text":"max(x, y, ...)\n\nReturn the maximum of the arguments. See also the maximum function to take the maximum element from a collection.\n\nThis version in Trixi.jl is semantically equivalent to Base.max but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).\n\nExamples\n\njulia> max(2, 5, 1)\n5\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.max_abs_speed_naive","page":"Trixi.jl","title":"Trixi.max_abs_speed_naive","text":"max_abs_speed_naive(u_ll, u_rr, orientation::Integer, equations)\nmax_abs_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimate of the maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, based only on the local wave speeds associated to u_ll and u_rr.\n\nFor non-integer arguments normal_direction in one dimension, max_abs_speed_naive returns abs(normal_direction[1]) * max_abs_speed_naive(u_ll, u_rr, 1, equations).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min-Tuple","page":"Trixi.jl","title":"Trixi.min","text":"min(x, y, ...)\n\nReturn the minimum of the arguments. See also the minimum function to take the minimum element from a collection.\n\nThis version in Trixi.jl is semantically equivalent to Base.min but may be implemented differently. In particular, it may avoid potentially expensive checks necessary in the presence of NaNs (or signed zeros).\n\nExamples\n\njulia> min(2, 5, 1)\n1\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_davis","page":"Trixi.jl","title":"Trixi.min_max_speed_davis","text":"min_max_speed_davis(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_davis(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimates of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr.\n\nS.F. Davis (1988) Simplified Second-Order Godunov-Type Methods DOI: 10.1137/0909030\n\nSee eq. (10.38) from\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction DOI: 10.1007/b79761\n\nSee also FluxHLL, min_max_speed_naive, min_max_speed_einfeldt.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_einfeldt(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nMore advanced mininmal and maximal wave speed computation based on\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\noriginally developed for the compressible Euler equations. A compact representation can be found in this lecture notes, eq. (9.28).\n\nSee also FluxHLL, min_max_speed_naive, min_max_speed_davis.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, AbstractVector, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations2D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, AbstractVector, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, normal_direction, equations::CompressibleEulerEquations3D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations1D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nOriginal publication:\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\n\nCompactly summarized:\n\nSiddhartha Mishra, Ulrik Skre Fjordholm and Rémi Abgrall Numerical methods for conservation laws and related equations. Link\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations2D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation, equations::CompressibleEulerEquations3D)\n\nComputes the HLLE (Harten-Lax-van Leer-Einfeldt) flux for the compressible Euler equations. Special estimates of the signal velocites and linearization of the Riemann problem developed by Einfeldt to ensure that the internal energy and density remain positive during the computation of the numerical flux.\n\nBernd Einfeldt (1988) On Godunov-type methods for gas dynamics. DOI: 10.1137/0725021\nBernd Einfeldt, Claus-Dieter Munz, Philip L. Roe and Björn Sjögreen (1991) On Godunov-type methods near low densities. DOI: 10.1016/0021-9991(91)90211-3\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations1D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations::IdealGlmMhdEquations1D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations2D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation::Integer, equations::IdealGlmMhdEquations2D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_einfeldt-Tuple{Any, Any, Integer, IdealGlmMhdEquations3D}","page":"Trixi.jl","title":"Trixi.min_max_speed_einfeldt","text":"min_max_speed_einfeldt(u_ll, u_rr, orientation_or_normal_direction, equations::IdealGlmMhdEquations3D)\n\nCalculate minimum and maximum wave speeds for HLL-type fluxes as in\n\nLi (2005) An HLLC Riemann solver for magneto-hydrodynamics DOI: 10.1016/j.jcp.2004.08.020\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.min_max_speed_naive","page":"Trixi.jl","title":"Trixi.min_max_speed_naive","text":"min_max_speed_naive(u_ll, u_rr, orientation::Integer, equations)\nmin_max_speed_naive(u_ll, u_rr, normal_direction::AbstractVector, equations)\n\nSimple and fast estimate(!) of the minimal and maximal wave speed of the Riemann problem with left and right states u_ll, u_rr, usually based only on the local wave speeds associated to u_ll and u_rr. Slightly more diffusive than min_max_speed_davis.\n\nAmiram Harten, Peter D. Lax, Bram van Leer (1983) On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation Laws DOI: 10.1137/1025002\n\nSee eq. (10.37) from\n\nEleuterio F. Toro (2009) Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction DOI: 10.1007/b79761\n\nSee also FluxHLL, min_max_speed_davis, min_max_speed_einfeldt.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.modify_dt_for_tstops!-Tuple{Trixi.SimpleIntegratorSSP}","page":"Trixi.jl","title":"Trixi.modify_dt_for_tstops!","text":"modify_dt_for_tstops!(integrator::SimpleIntegratorSSP)\n\nModify the time-step size to match the time stops specified in integrator.opts.tstops. To avoid adding OrdinaryDiffEq to Trixi's dependencies, this routine is a copy of https://github.com/SciML/OrdinaryDiffEq.jl/blob/d76335281c540ee5a6d1bd8bb634713e004f62ee/src/integrators/integrator_utils.jl#L38-L54\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.multiply_dimensionwise-Tuple{AbstractMatrix, AbstractMatrix}","page":"Trixi.jl","title":"Trixi.multiply_dimensionwise","text":"multiply_dimensionwise(matrix::AbstractMatrix, data_in::AbstractArray{<:Any, NDIMS+1})\n\nMultiply the array data_in by matrix in each coordinate direction, where data_in is assumed to have the first coordinate for the number of variables and the remaining coordinates are multiplied by matrix.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.n_nonconservative_terms","page":"Trixi.jl","title":"Trixi.n_nonconservative_terms","text":"n_nonconservative_terms(equations)\n\nNumber of nonconservative terms in the form local * symmetric for a particular equation. This function needs to be specialized only if equations with nonconservative terms are combined with certain solvers (e.g., subcell limiting).\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.ndofs-Tuple{Trixi.AbstractSemidiscretization}","page":"Trixi.jl","title":"Trixi.ndofs","text":"ndofs(semi::AbstractSemidiscretization)\n\nReturn the number of degrees of freedom associated with each scalar variable.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.negative_part-Tuple{Any}","page":"Trixi.jl","title":"Trixi.negative_part","text":"negative_part(x)\n\nReturn x if x is negative, else zero. In other words, return (x - abs(x)) / 2 for real numbers x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_default_options-Tuple{}","page":"Trixi.jl","title":"Trixi.ode_default_options","text":"ode_default_options()\n\nReturn the default options for OrdinaryDiffEq's solve. Pass ode_default_options()... to solve to only return the solution at the final time and enable MPI aware error-based step size control, whenever MPI is used. For example, use solve(ode, alg; ode_default_options()...).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_norm-Tuple{Number, Any}","page":"Trixi.jl","title":"Trixi.ode_norm","text":"ode_norm(u, t)\n\nImplementation of the weighted L2 norm of Hairer and Wanner used for error-based step size control in OrdinaryDiffEq.jl. This function is aware of MPI and uses global MPI communication when running in parallel.\n\nYou must pass this function as a keyword argument internalnorm=ode_norm to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.\n\nSee the \"Advanced Adaptive Stepsize Control\" section of the documentation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.ode_unstable_check-NTuple{4, Any}","page":"Trixi.jl","title":"Trixi.ode_unstable_check","text":"ode_unstable_check(dt, u, semi, t)\n\nImplementation of the basic check for instability used in OrdinaryDiffEq.jl. Instead of checking something like any(isnan, u), this function just checks isnan(dt). This helps when using MPI parallelization, since no additional global communication is required and all ranks will return the same result.\n\nYou should pass this function as a keyword argument unstable_check=ode_unstable_check to OrdinaryDiffEq.jl's solve when using error-based step size control with MPI parallel execution of Trixi.jl.\n\nSee the \"Miscellaneous\" section of the documentation.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.partition!-Tuple{T8codeMesh}","page":"Trixi.jl","title":"Trixi.partition!","text":"Trixi.partition!(mesh::T8codeMesh)\n\nPartition a T8codeMesh in order to redistribute elements evenly among MPI ranks.\n\nArguments\n\nmesh::T8codeMesh: Initialized mesh object.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.partition!-Tuple{TreeMesh{NDIMS, <:Trixi.ParallelTree{NDIMS}} where NDIMS}","page":"Trixi.jl","title":"Trixi.partition!","text":"partition!(mesh::ParallelTreeMesh, allow_coarsening=true)\n\nPartition mesh using a static domain decomposition algorithm based on leaf cell count and tree structure. If allow_coarsening is true, the algorithm will keep leaf cells together on one rank when needed for local coarsening (i.e. when all children of a cell are leaves).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.positive_part-Tuple{Any}","page":"Trixi.jl","title":"Trixi.positive_part","text":"positive_part(x)\n\nReturn x if x is positive, else zero. In other words, return (x + abs(x)) / 2 for real numbers x.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.pressure-Tuple{Real, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.pressure","text":"pressure(rho::Real, equations::LatticeBoltzmannEquations2D)\npressure(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic pressure from the density rho or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.pressure-Tuple{Real, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.pressure","text":"pressure(rho::Real, equations::LatticeBoltzmannEquations3D)\npressure(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic pressure from the density rho or the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.prim2cons","page":"Trixi.jl","title":"Trixi.prim2cons","text":"prim2cons(u, equations)\n\nConvert the primitive variables u to the conserved variables for a given set of equations. u is a vector type of the correct length nvariables(equations). Notice the function doesn't include any error checks for the purpose of efficiency, so please make sure your input is correct. The inverse conversion is performed by cons2prim.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.residual_steady_state-Tuple{Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.residual_steady_state","text":"residual_steady_state(du, ::AbstractHyperbolicDiffusionEquations)\n\nUsed to determine the termination criterion of a SteadyStateCallback. For hyperbolic diffusion, this checks convergence of the potential phi.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.rotate_from_x","page":"Trixi.jl","title":"Trixi.rotate_from_x","text":"rotate_from_x(u, normal, equations)\n\nApply the rotation that maps the x-axis onto normal to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.\n\nSee also: rotate_to_x\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.rotate_to_x","page":"Trixi.jl","title":"Trixi.rotate_to_x","text":"rotate_to_x(u, normal, equations)\n\nApply the rotation that maps normal onto the x-axis to the convservative variables u. This is used by FluxRotated to calculate the numerical flux of rotationally invariant equations in arbitrary normal directions.\n\nSee also: rotate_from_x\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.save_plot-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.save_plot","text":"save_plot(plot_data, variable_names;\n show_mesh=true, plot_arguments=Dict{Symbol,Any}(),\n time=nothing, timestep=nothing)\n\nVisualize the plot data object provided in plot_data and save result as a PNG file in the out directory, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.\n\nThe timestep is used in the filename. time is currently unused by this function.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nSee also: VisualizationCallback, show_plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.set_log_type-Tuple{Any}","page":"Trixi.jl","title":"Trixi.set_log_type","text":"Trixi.set_log_type(type; force = true)\n\nSet the type of the (natural) log function to be used in Trixi.jl. The default is \"sqrt_Trixi_NaN\" which returns NaN for negative arguments instead of throwing an error. Alternatively, you can set type to \"sqrt_Base\" to use the Julia built-in sqrt function which provides a stack-trace of the error which might come in handy when debugging code.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.set_sqrt_type-Tuple{Any}","page":"Trixi.jl","title":"Trixi.set_sqrt_type","text":"Trixi.set_sqrt_type(type; force = true)\n\nSet the type of the square root function to be used in Trixi.jl. The default is \"sqrt_Trixi_NaN\" which returns NaN for negative arguments instead of throwing an error. Alternatively, you can set type to \"sqrt_Base\" to use the Julia built-in sqrt function which provides a stack-trace of the error which might come in handy when debugging code.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.show_plot-Tuple{Any, Any}","page":"Trixi.jl","title":"Trixi.show_plot","text":"show_plot(plot_data, variable_names;\n show_mesh=true, plot_arguments=Dict{Symbol,Any}(),\n time=nothing, timestep=nothing)\n\nVisualize the plot data object provided in plot_data and display result, plotting only the variables in variable_names and, optionally, the mesh (if show_mesh is true). Additionally, plot_arguments will be unpacked and passed as keyword arguments to the Plots.plot command.\n\nThis function is the default plot_creator argument for the VisualizationCallback. time and timestep are currently unused by this function.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\nSee also: VisualizationCallback, save_plot\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.solve","page":"Trixi.jl","title":"Trixi.solve","text":"solve(ode, alg; dt, callbacks, kwargs...)\n\nThe following structures and methods provide the infrastructure for SSP Runge-Kutta methods of type SimpleAlgorithmSSP.\n\nwarning: Experimental implementation\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, AcousticPerturbationEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::AcousticPerturbationEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquations3D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerEquationsQuasi1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the mozzle width 'a(x) = 1.5 - 0.5 * cos(x[1] * pi)' as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerMulticomponentEquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::CompressibleEulerMulticomponentEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::InviscidBurgersEquation1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, PolytropicEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::PolytropicEulerEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterEquations2D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the bottom topography function b(x,y) = 2 + 0.5 * sin(sqrt(2)*pi*x) + 0.5 * sin(sqrt(2)*pi*y) as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, ShallowWaterEquationsQuasi1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::ShallowWaterEquationsQuasi1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test (and BoundaryConditionDirichlet(initial_condition_convergence_test) in non-periodic domains).\n\nThis manufactured solution source term is specifically designed for the bottom topography function b(x) = 0.2 - 0.05 * sin(sqrt(2) * pi *x[1]) and channel width 'a(x)= 1 + 0.1 * cos(sqrt(2) * pi * x[1])' as defined in initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_convergence_test-Tuple{Any, Any, Any, TrafficFlowLWREquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_convergence_test","text":"source_terms_convergence_test(u, x, t, equations::TrafficFlowLWREquations1D)\n\nSource terms used for convergence tests in combination with initial_condition_convergence_test.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_coupled_euler_gravity-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_coupled_euler_gravity","text":"source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_coupled_euler_gravity-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_coupled_euler_gravity","text":"source_terms_eoc_test_coupled_euler_gravity(u, x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_euler-Tuple{Any, Any, Any, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_euler","text":"source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations2D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_eoc_test_euler-Tuple{Any, Any, Any, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_eoc_test_euler","text":"source_terms_eoc_test_euler(u, x, t, equations::CompressibleEulerEquations3D)\n\nSetup used for convergence tests of the Euler equations with self-gravity used in\n\nMichael Schlottke-Lakemper, Andrew R. Winters, Hendrik Ranocha, Gregor J. Gassner (2020) A purely hyperbolic discontinuous Galerkin approach for self-gravitating gas dynamics arXiv: 2008.10593\n\nin combination with initial_condition_eoc_test_coupled_euler_gravity.\n\nnote: Note\nThis method is to be used for testing pure Euler simulations with analytic self-gravity. If you intend to do coupled Euler-gravity simulations, you need to use source_terms_eoc_test_coupled_euler_gravity instead.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations1D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations2D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations2D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_harmonic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations3D}","page":"Trixi.jl","title":"Trixi.source_terms_harmonic","text":"source_terms_harmonic(u, x, t, equations::HyperbolicDiffusionEquations3D)\n\nSource term that only includes the forcing from the hyperbolic diffusion system.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.source_terms_poisson_nonperiodic-Tuple{Any, Any, Any, HyperbolicDiffusionEquations1D}","page":"Trixi.jl","title":"Trixi.source_terms_poisson_nonperiodic","text":"source_terms_poisson_nonperiodic(u, x, t,\n equations::HyperbolicDiffusionEquations1D)\n\nSource terms that include the forcing function f(x) and right hand side for the hyperbolic diffusion system that is used with initial_condition_poisson_nonperiodic and boundary_condition_poisson_nonperiodic.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_coirier_vanleer-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_coirier_vanleer","text":"splitting_coirier_vanleer(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_coirier_vanleer(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux from Coirier and van Leer. The splitting has correction terms in the pressure splitting as well as the mass and energy flux components. The motivation for these corrections are to handle flows at the low Mach number limit.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nWilliam Coirier and Bram van Leer (1991) Numerical flux formulas for the Euler and Navier-Stokes equations. II - Progress in flux-vector splitting DOI: 10.2514/6.1991-1566\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) similar to a flux splitting one would apply, e.g., to Burgers' equation.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, InviscidBurgersEquation1D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::InviscidBurgersEquation1D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::InviscidBurgersEquation1D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ = abs(u).\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_lax_friedrichs-Tuple{Any, Integer, LinearScalarAdvectionEquation1D}","page":"Trixi.jl","title":"Trixi.splitting_lax_friedrichs","text":"splitting_lax_friedrichs(u, orientation::Integer,\n equations::LinearScalarAdvectionEquation1D)\nsplitting_lax_friedrichs(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::LinearScalarAdvectionEquation1D)\n\nNaive local Lax-Friedrichs style flux splitting of the form f⁺ = 0.5 (f + λ u) and f⁻ = 0.5 (f - λ u) where λ is the absolute value of the advection velocity.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_steger_warming-Tuple{Any, Integer, CompressibleEulerEquations3D}","page":"Trixi.jl","title":"Trixi.splitting_steger_warming","text":"splitting_steger_warming(u, orientation::Integer,\n equations::CompressibleEulerEquations3D)\nsplitting_steger_warming(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations3D)\n\nSplitting of the compressible Euler flux of Steger and Warming.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nJoseph L. Steger and R. F. Warming (1979) Flux Vector Splitting of the Inviscid Gasdynamic Equations With Application to Finite Difference Methods NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_vanleer_haenel-Tuple{Any, Integer, CompressibleEulerEquations1D}","page":"Trixi.jl","title":"Trixi.splitting_vanleer_haenel","text":"splitting_vanleer_haenel(u, orientation::Integer,\n equations::CompressibleEulerEquations1D)\nsplitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations1D)\n\nSplitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nBram van Leer (1982) Flux-Vector Splitting for the Euler Equation DOI: 10.1007/978-3-642-60543-7_5\nD. Hänel, R. Schwane and G. Seider (1987) On the accuracy of upwind schemes for the solution of the Navier-Stokes equations DOI: 10.2514/6.1987-1105\nMeng-Sing Liou and Chris J. Steffen, Jr. (1991) High-Order Polynomial Expansions (HOPE) for Flux-Vector Splitting NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.splitting_vanleer_haenel-Tuple{Any, Integer, CompressibleEulerEquations2D}","page":"Trixi.jl","title":"Trixi.splitting_vanleer_haenel","text":"splitting_vanleer_haenel(u, orientation::Integer,\n equations::CompressibleEulerEquations2D)\nsplitting_vanleer_haenel(u, which::Union{Val{:minus}, Val{:plus}}\n orientation::Integer,\n equations::CompressibleEulerEquations2D)\n\nSplitting of the compressible Euler flux from van Leer. This splitting further contains a reformulation due to Hänel et al. where the energy flux uses the enthalpy. The pressure splitting is independent from the splitting of the convective terms. As such there are many pressure splittings suggested across the literature. We implement the 'p4' variant suggested by Liou and Steffen as it proved the most robust in practice.\n\nReturns a tuple of the fluxes \"minus\" (associated with waves going into the negative axis direction) and \"plus\" (associated with waves going into the positive axis direction). If only one of the fluxes is required, use the function signature with argument which set to Val{:minus}() or Val{:plus}().\n\nwarning: Experimental implementation (upwind SBP)\nThis is an experimental feature and may change in future releases.\n\nReferences\n\nBram van Leer (1982) Flux-Vector Splitting for the Euler Equation DOI: 10.1007/978-3-642-60543-7_5\nD. Hänel, R. Schwane and G. Seider (1987) On the accuracy of upwind schemes for the solution of the Navier-Stokes equations DOI: 10.2514/6.1987-1105\nMeng-Sing Liou and Chris J. Steffen, Jr. (1991) High-Order Polynomial Expansions (HOPE) for Flux-Vector Splitting NASA Technical Memorandum\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.stolarsky_mean-Tuple{Any, Any, Any}","page":"Trixi.jl","title":"Trixi.stolarsky_mean","text":"stolarsky_mean(x, y, gamma)\n\nCompute an instance of a weighted Stolarsky mean of the form\n\nstolarsky_mean(x, y, gamma) = (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))\n\nwhere gamma > 1.\n\nProblem: The formula above has a removable singularity at x == y. Thus, some care must be taken to implement it correctly without problems or loss of accuracy when x ≈ y. Here, we use the approach proposed by Winters et al. (2020). Set f = (y - x) / (y + x) and g = gamma (for compact notation). Then, we use the expansions\n\n((1+f)^g - (1-f)^g) / g = 2*f + (g-1)(g-2)/3 * f^3 + (g-1)(g-2)(g-3)(g-4)/60 * f^5 + O(f^7)\n\nand\n\n((1+f)^(g-1) - (1-f)^(g-1)) / (g-1) = 2*f + (g-2)(g-3)/3 * f^3 + (g-2)(g-3)(g-4)(g-5)/60 * f^5 + O(f^7)\n\nInserting the first few terms of these expansions and performing polynomial long division we find that\n\nstolarsky_mean(x, y, gamma) ≈ (y + x) / 2 * (1 + (g-2)/3 * f^2 - (g+1)(g-2)(g-3)/45 * f^4 + (g+1)(g-2)(g-3)(2g(g-2)-9)/945 * f^6)\n\nSince divisions are usually more expensive on modern hardware than multiplications (Agner Fog), we try to avoid computing two divisions. Thus, we use\n\nf^2 = (y - x)^2 / (x + y)^2\n = (x * (x - 2 * y) + y * y) / (x * (x + 2 * y) + y * y)\n\nGiven ε = 1.0e-4, we use the following algorithm.\n\nif f^2 < ε\n # use the expansion above\nelse\n # use the direct formula (gamma - 1)/gamma * (y^gamma - x^gamma) / (y^(gamma-1) - x^(gamma-1))\nend\n\nReferences\n\nAndrew R. Winters, Christof Czernik, Moritz B. Schily & Gregor J. Gassner (2020) Entropy stable numerical approximations for the isothermal and polytropic Euler equations DOI: 10.1007/s10543-019-00789-w\nAgner Fog. Lists of instruction latencies, throughputs and micro-operation breakdowns for Intel, AMD, and VIA CPUs. https://www.agner.org/optimize/instruction_tables.pdf\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.totalgamma-Tuple{Any, CompressibleEulerMulticomponentEquations1D}","page":"Trixi.jl","title":"Trixi.totalgamma","text":"totalgamma(u, equations::CompressibleEulerMulticomponentEquations1D)\n\nFunction that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.totalgamma-Tuple{Any, CompressibleEulerMulticomponentEquations2D}","page":"Trixi.jl","title":"Trixi.totalgamma","text":"totalgamma(u, equations::CompressibleEulerMulticomponentEquations2D)\n\nFunction that calculates the total gamma out of all partial gammas using the partial density fractions as well as the partial specific heats at constant volume.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.uses_amr-Tuple{Any}","page":"Trixi.jl","title":"Trixi.uses_amr","text":"uses_amr(callback)\n\nChecks whether the provided callback or CallbackSet is an AMRCallback or contains one.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.varnames","page":"Trixi.jl","title":"Trixi.varnames","text":"varnames(conversion_function, equations)\n\nReturn the list of variable names when applying conversion_function to the conserved variables associated to equations. This function is mainly used internally to determine output to screen and for IO, e.g., for the AnalysisCallback and the SaveSolutionCallback. Common choices of the conversion_function are cons2cons and cons2prim.\n\n\n\n\n\n","category":"function"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, Integer, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, orientation, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y) from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, Integer, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, orientation, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic velocity for the given orientation (1 -> x, 2 -> y, 3 -> z) from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, LatticeBoltzmannEquations2D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, equations::LatticeBoltzmannEquations2D)\n\nCalculate the macroscopic velocity vector from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.velocity-Tuple{Any, LatticeBoltzmannEquations3D}","page":"Trixi.jl","title":"Trixi.velocity","text":"velocity(u, equations::LatticeBoltzmannEquations3D)\n\nCalculate the macroscopic velocity vector from the particle distribution functions u.\n\n\n\n\n\n","category":"method"},{"location":"reference-trixi/#Trixi.@autoinfiltrate","page":"Trixi.jl","title":"Trixi.@autoinfiltrate","text":"@autoinfiltrate\n@autoinfiltrate condition::Bool\n\nInvoke the @infiltrate macro of the package Infiltrator.jl to create a breakpoint for ad-hoc interactive debugging in the REPL. If the optional argument condition is given, the breakpoint is only enabled if condition evaluates to true.\n\nAs opposed to using Infiltrator.@infiltrate directly, this macro does not require Infiltrator.jl to be added as a dependency to Trixi.jl. As a bonus, the macro will also attempt to load the Infiltrator module if it has not yet been loaded manually.\n\nNote: For this macro to work, the Infiltrator.jl package needs to be installed in your current Julia environment stack.\n\nSee also: Infiltrator.jl\n\nwarning: Internal use only\nPlease note that this macro is intended for internal use only. It is not part of the public API of Trixi.jl, and it thus can altered (or be removed) at any time without it being considered a breaking change.\n\n\n\n\n\n","category":"macro"},{"location":"reference-trixi/#Trixi.@threaded-Tuple{Any}","page":"Trixi.jl","title":"Trixi.@threaded","text":"@threaded for ... end\n\nSemantically the same as Threads.@threads when iterating over a AbstractUnitRange but without guarantee that the underlying implementation uses Threads.@threads or works for more general for loops. In particular, there may be an additional check whether only one thread is used to reduce the overhead of serial execution or the underlying threading capabilities might be provided by other packages such as Polyester.jl.\n\nwarn: Warn\nThis macro does not necessarily work for general for loops. For example, it does not necessarily support general iterables such as eachline(filename).\n\nSome discussion can be found at https://discourse.julialang.org/t/overhead-of-threads-threads/53964 and https://discourse.julialang.org/t/threads-threads-with-one-thread-how-to-remove-the-overhead/58435.\n\n\n\n\n\n","category":"macro"},{"location":"styleguide/#Style-guide","page":"Style guide","title":"Style guide","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Coding style is an inherently personal - and thus hotly contested - issue. Since code is usually \"written once, read often\", it helps regular developers, new users, and reviewers if code is formatted consistently. We therefore believe in the merit of using a common coding style throughout Trixi.jl, even at the expense that not everyone can be happy with every detailed style decision. If you came here because you are furious about our code formatting rules, here is a happy little whale for you to calm you down: 🐳","category":"page"},{"location":"styleguide/#Conventions","page":"Style guide","title":"Conventions","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"The following lists a few coding conventions for Trixi.jl. Note that in addition to these conventions, we apply and enforce automated source code formatting (see below for more details):","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Modules, types, structs with CamelCase.\nFunctions, variables with lowercase snake_case.\nIndentation with 4 spaces (never tabs!)\nMaximum line length (strictly): 92.\nFunctions that mutate their input are named with a trailing !.\nFunctions order their parameters similar to Julia Base.\nThe main modified argument comes first. For example, if the right-hand side du is modified, it should come first. If only the cache is modified, e.g., in prolong2interfaces! and its siblings, put the cache first.\nOtherwise, use the order mesh, equations, solver, cache.\nIf something needs to be specified in more detail for dispatch, put the additional argument before the general one that is specified in more detail. For example, we use have_nonconservative_terms(equations), equations and dg.mortar, dg.\nPrefer for i in ... to for i = ... for better semantic clarity and greater flexibility.\nExecutable code should only use ASCII characters.\nDocstrings and comments can and should use Unicode characters where it helps understanding.\nMultiline expressions should be explicitly grouped by parentheses and not rely on Julia's implicit line continuation syntax.\nWhen naming multiple functions of a single or similar category, prefer to put the general classification first and the specialization second. Example: Use flux_central instead of central_flux. This helps when searching for available functions on the REPL (e.g., when trying to find all flux functions).","category":"page"},{"location":"styleguide/#automated-source-code-formatting","page":"Style guide","title":"Automated source code formatting","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"We use JuliaFormatter.jl to format the source code of Trixi.jl, which will also enforce some of the Conventions listed above (e.g., line length or indentation with 4 spaces are automatically handled, while capitalization of names is not). Our format is mostly based on the SciML-style formatting rules. For more details you can have a look at the current .JuliaFormatter.toml file that holds the configuration options we use for JuliaFormatter.jl.","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"Note that we expect all contributions to Trixi.jl to be formatted with JuliaFormatter.jl before being merged to the main branch. We ensure this by running a automated check on all PRs that verify that running JuliaFormatter.jl again will not change the source code.","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"To format your contributions before created a PR (or, at least, before requesting a review of your PR), you need to install JuliaFormatter.jl first by running","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"julia -e 'using Pkg; Pkg.add(\"JuliaFormatter\")'","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"You can then recursively format the core Julia files in the Trixi.jl repo by executing","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"julia -e 'using JuliaFormatter; format([\"benchmark\", \"examples\", \"ext\", \"src\", \"test\", \"utils\"])'","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"from inside the Trixi.jl repository. For convenience, there is also a script you can directly run from your terminal shell, which will automatically install JuliaFormatter in a temporary environment and then run it:","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"utils/trixi-format.jl","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"You can get more information about using the convenience script by running it with the --help/-h flag.","category":"page"},{"location":"styleguide/#Checking-formatting-before-committing","page":"Style guide","title":"Checking formatting before committing","text":"","category":"section"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"It can be convenient to check the formatting of source code automatically before each commit. We use git-hooks for it and provide a pre-commit script in the utils folder. The script uses JuliaFormatter.jl just like formatting script that runs over the whole Trixi.jl directory. You can copy the pre-commit-script into .git/hooks/pre-commit and it will check your formatting before each commit. If errors are found the commit is aborted and you can add the corrections via","category":"page"},{"location":"styleguide/","page":"Style guide","title":"Style guide","text":"git add -p","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"EditURL = \"../../literate/src/files/scalar_linear_advection_1d.jl\"","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#scalar_linear_advection_1d","page":"3 Introduction to DG methods","title":"3: Introduction to DG methods","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"(Image: ) (Image: ) (Image: )","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"This tutorial is about how to set up a simple way to approximate the solution of a hyperbolic partial differential equation. First, we will implement a basic and naive algorithm. Then, we will use predefined features from Trixi.jl to show how you can use Trixi.jl on your own.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We will implement the scalar linear advection equation in 1D with the advection velocity 1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u_t + u_x = 0 textfor tin mathbbR^+ xinOmega=-11","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We define the domain Omega by setting the boundaries.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"coordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We assume periodic boundaries and the following initial condition.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"initial_condition_sine_wave(x) = 1.0 + 0.5 * sin(pi * x)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#The-discontinuous-Galerkin-collocation-spectral-element-method-(DGSEM)","page":"3 Introduction to DG methods","title":"The discontinuous Galerkin collocation spectral element method (DGSEM)","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/#i.-Discretization-of-the-physical-domain","page":"3 Introduction to DG methods","title":"i. Discretization of the physical domain","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To improve precision we want to approximate the solution on small parts of the physical domain. So, we split the domain Omega=-1 1 into elements Q_l of length dx.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"n_elements = 16 # number of elements\n\ndx = (coordinates_max - coordinates_min) / n_elements # length of one element","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To make the calculation more efficient and storing less information, we transform each element Q_l with center point x_l to a reference element E=-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Q_l=Bigx_l-fracdx2 x_l+fracdx2Big undersetx(xi)oversetxi(x)rightleftarrows -1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"So, for every element the transformation from the reference domain to the physical domain is defined by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"x(xi) = x_l + fracdx2 xi xiin-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Therefore,","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"beginalign*\nu = u(x(xi) t) \nu_x = u_xi fracdxidx 3pt\nfracdxidx = (x_xi)^-1 = frac2dx = J^-1 \nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Here, J is the Jacobian determinant of the transformation.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Using this transformation, we can transform our equation for each element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"fracdx2 u_t^Q_l + u_xi^Q_l = 0 text for tinmathbbR^+ xiin-1 1","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Here, u_t^Q_l and u_xi^Q_l denote the time and spatial derivatives of the solution on the element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#ii.-Polynomial-approach","page":"3 Introduction to DG methods","title":"ii. Polynomial approach","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Now, we want to approximate the solution in each element Q_l by a polynomial of degree N. Since we transformed the equation, we can use the same polynomial approach for the reference coordinate xiin-1 1 in every physical element Q_l. This saves a lot of resources by reducing the amount of calculations needed and storing less information.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"For DGSEM we choose Lagrange basis functions l_j_j=0^N as our polynomial basis of degree N in -1 1. The solution in element Q_l can be approximated by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u(x(xi) t)big_Q_l approx u^Q_l(xi t) = sum_j=0^N u_j^Q_l(t) l_j(xi)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"with N+1 coefficients u_j^Q_l_j=0^N. By construction the Lagrange basis has some useful advantages. This basis is defined by N+1 nodes, which fulfill a Kronecker property at the exact same nodes. Let xi_i_i=0^N be these nodes.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"l_j(xi_i) = delta_ij =\nbegincases\n1 textif i=j \n0 textelse\nendcases","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Because of this property, the polynomial coefficients are exact the values of u^Q_l at the nodes","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u^Q_l(xi_i t) = sum_j=0^N u_j^Q_l(t) underbracel_j(xi_i)_=delta_ij = u_i^Q_l(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Next, we want to select the nodes xi_i_i=0^N, which we use for the construction of the Lagrange polynomials. We choose the N+1 Gauss-Lobatto nodes, which are used for the Gaussian-Lobatto quadrature. These always contain the boundary points at -1 and +1 and are well suited as interpolation nodes. The corresponding weights will be referred to as w_j_j=0^N. In Trixi.jl the basis with Lagrange polynomials on Gauss-Lobatto nodes is already defined.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using Trixi\npolydeg = 3 #= polynomial degree = N =#\nbasis = LobattoLegendreBasis(polydeg)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"The Gauss-Lobatto nodes are","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"nodes = basis.nodes","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"with the corresponding weights","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"weights = basis.weights","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To illustrate how you can integrate using numerical quadrature with this Legendre-Gauss-Lobatto nodes, we give an example for f(x)=x^3. Since f is of degree 3, a polynomial interpolation with N=3 is exact. Therefore, the integral on -1 1 can be calculated by","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"beginalign*\nint_-1^1 f(x) dx = int_-1^1 Big( sum_j=0^3 f(xi_j)l_j(x) Big) dx\n= sum_j=0^3 f(xi_j) int_-1^1 l_j(x)dx \n= sum_j=0^3 f(xi_j) w_j\n= sum_j=0^3 xi_j^3 w_j\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Let's use our nodes and weights for N=3 and plug in","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"integral = sum(nodes.^3 .* weights)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Using this polynomial approach leads to the equation","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"fracdx2 dotu^Q_l(xi t) + u^Q_l(xi t) = 0","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"with dotu=fracpartialpartial tu and u=fracpartialpartial xu. To approximate the solution, we need to get the polynomial coefficients u_j^Q_l_j=0^N for every element Q_l.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"After defining all nodes, we can implement the spatial coordinate x and its initial value u0 for every node.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"x = Matrix{Float64}(undef, length(nodes), n_elements)\nfor element in 1:n_elements\n x_l = coordinates_min + (element - 1) * dx + dx/2\n for i in 1:length(nodes)\n ξ = nodes[i] # nodes in [-1, 1]\n x[i, element] = x_l + dx/2 * ξ\n end\nend\n\nu0 = initial_condition_sine_wave.(x)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To have a look at the initial sinus curve, we plot it.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using Plots\nplot(vec(x), vec(u0), label=\"initial condition\", legend=:topleft)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#iii.-Variational-formulation","page":"3 Introduction to DG methods","title":"iii. Variational formulation","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"After defining the equation and initial condition, we want to implement an algorithm to approximate the solution.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"From now on, we only write u instead of u^Q_l for simplicity, but consider that all the following calculation only concern one element. Multiplying the new equation with the smooth Lagrange polynomials l_i_i=0^N (test functions) and integrating over the reference element E=-11, we get the variational formulation of our transformed partial differential equation for i=0N:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"beginalign*\nint_-1^1 Big( fracdx2 dotu(xi t) + u(xi t) Big) l_i(xi)dxi\n = underbracefracdx2 int_-1^1 dotu(xi t) l_i(xi)dxi_textTerm I + underbraceint_-1^1 u(xi t) l_i(xi)dxi_textTerm II = 0\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We deal with the two terms separately. We write int_-1 N^1 cdot dxi for the approximation of the integral using numerical quadrature with N+1 basis points. We use the Gauss-Lobatto nodes again. The numerical scalar product langlecdot cdotrangle_N is defined by langle f grangle_N = int_-1 N^1 f(xi) g(xi) dxi.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Term-I:","page":"3 Introduction to DG methods","title":"Term I:","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"In the following calculation we approximate the integral numerically with quadrature on the Gauss-Lobatto nodes xi_i_i=0^N and then use the Kronecker property of the Lagrange polynomials. This approach of using the same nodes for the interpolation and quadrature is called collocation.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"beginalign*\nfracdx2 int_-1^1 dotu(xi t) l_i(xi)dxi\napprox fracdx2 int_-1 N^1 dotu(xi t) l_i(xi)dxi \n= fracdx2 sum_k=0^N underbracedotu(xi_k t)_=dotu_k(t) underbracel_i(xi_k)_=delta_kiw_k \n= fracdx2 dotu_i(t) w_i\nendalign*","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We define the Legendre-Gauss-Lobatto (LGL) mass matrix M and by the Kronecker property follows:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"M_ij = langle l_j l_irangle_N = delta_ij w_j ij=0N","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using LinearAlgebra\nM = diagm(weights)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Now, we can write the integral with this new matrix.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"fracdx2 int_-1 N^1 dotu(xi t) underlinel(xi)dxi = fracdx2 M underlinedotu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"where underlinedotu = (dotu_0 dotu_N)^T and underlinel respectively.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Note: Since the LGL quadrature with N+1 nodes is exact up to functions of degree 2N-1 and dotu(xi t) l_i(xi) is of degree 2N, in general the following holds","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"int_-1^1 dotu(xi t) l_i(xi) dxi neq int_-1 N^1 dotu(xi t) l_i(xi) dxi","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"With an exact integration the mass matrix would be dense. Choosing numerical integrating and quadrature with the exact same nodes (collocation) leads to the sparse and diagonal mass matrix M. This is called mass lumping and has the big advantage of an easy invertation of the matrix.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Term-II:","page":"3 Introduction to DG methods","title":"Term II:","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We use spatial partial integration for the second term:","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"int_-1^1 u(xi t) l_i(xi) dxi = u l_i_-1^1 - int_-1^1 u l_idxi","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"The resulting integral can be solved exactly with LGL quadrature since the polynomial is now of degree 2N-1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Again, we split the calculation in two steps.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Surface-term","page":"3 Introduction to DG methods","title":"Surface term","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"As mentioned before, we approximate the solution with a polynomial in every element. Therefore, in general the value of this approximation at the interfaces between two elements is not unique. To solve this problem we introduce the idea of the numerical flux u^*, which will give an exact value at the interfaces. One of many different approaches and definitions for the calculation of the numerical flux we will deal with in 4. Numerical flux.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u l_i_-1^1 = u^*big^1 l_i(+1) - u^*big_-1 l_i(-1)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Since the Gauss-Lobatto nodes contain the element boundaries -1 and +1, we can use the Kronecker property of l_i for the calculation of l_i(-1) and l_i(+1).","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u underlinel_-1^1 = u^*big^1 left(beginarrayc 0 vdots 0 1 endarrayright)\n- u^*big_-1 left(beginarrayc 1 0 vdots 0endarrayright)\n= B underlineu^*(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"with the boundary matrix","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"B = beginpmatrix\n-1 0 cdots 0\n0 0 cdots 0\nvdots vdots 0 0\n0 cdots 0 1\nendpmatrix\nqquadtextandqquad\nunderlineu^*(t) = left(beginarrayc u^*big_-1 0 vdots 0 u^*big^1endarrayright)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"B = diagm([-1; zeros(polydeg - 1); 1])","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Volume-term","page":"3 Introduction to DG methods","title":"Volume term","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"As mentioned before, the new integral can be solved exact since the function inside is of degree 2N-1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"- int_-1^1 u l_idxi = - int_-1 N^1 u l_i dxi\n= - sum_k=0^N u(xi_k t) l_i(xi_k) w_k\n= - sum_k=0^N u_k(t) D_ki w_k","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"where D is the derivative matrix defined by D_ki = l_i(xi_k) for ik=0N.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"D = basis.derivative_matrix","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To show why this matrix is called the derivative matrix, we go back to our example f(x)=x^3. We calculate the derivation of f at the Gauss-Lobatto nodes xi_k_k=0^N with N=8.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"f_x=xi_k = Big( sum_j=0^8 f(xi_j) l_j(x) Big)_x=xi_k = sum_j=0^8 f(xi_j) l_j(xi_k)\n= sum_j=0^8 f(xi_j) D_kj","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"for k=0N and therefore, underlinef = D underlinef.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"basis_N8 = LobattoLegendreBasis(8)\nplot(vec(x), x -> 3 * x^2, label=\"f'\", lw=2)\nscatter!(basis_N8.nodes, basis_N8.derivative_matrix * basis_N8.nodes.^3, label=\"Df\", lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Combining the volume term for every i=0N results in","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"int_-1^1 u underlinel dxi = - D^T M underlineu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Putting all parts together we get the following equation for the element Q_l","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"fracdx2 M underlinedotu(t) = - B underlineu^*(t) + D^T M underlineu(t)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"or equivalent","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"underlinedotu^Q_l(t) = frac2dx Big - M^-1 B underlineu^Q_l^*(t) + M^-1 D^T M underlineu^Q_l(t)Big","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"This is called the weak form of the DGSEM.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Note: For every element Q_l we get a system of N+1 ordinary differential equations to calculate N+1 coefficients. Since the numerical flux u^* is depending on extern values at the interfaces, the equation systems of adjacent elements are weakly linked.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#numerical_flux","page":"3 Introduction to DG methods","title":"iv. Numerical flux","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"As mentioned above, we still have to handle the problem of different values at the same point at the interfaces. This happens with the ideas of the numerical flux f^*(u)=u^*. The role of f^* might seem minor in this simple example, but is important for more complicated problems. There are two values at the same spatial coordinate. Let's say we are looking at the interface between the elements Q_l and Q_l+1, while both elements got N+1 nodes as defined before. We call the first value of the right element u_R=u_0^Q_l+1 and the last one of the left element u_L=u_N^Q_l. So, for the value of the numerical flux on that interface the following holds","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u^* = u^*(u_L u_R)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"These values are interpreted as start values of a so-called Riemann problem. There are many different (approximate) Riemann solvers available and useful for different problems. We will use the local Lax-Friedrichs flux.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"surface_flux = flux_lax_friedrichs","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"The only missing ingredient is the flux calculation at the boundaries -1 and +1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"u^Q_first^*big_-1 = u^Q_first^*big_-1(u^bound(-1) u_R)\nquadtextandquad\nu^Q_last^*big^1 = u^Q_last^*big^1(u_L u^bound(1))","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"The boundaries are periodic, which means that the last value of the last element u^Q_last_N is used as u_L at the first interface and accordingly for the other boundary.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Now, we implement a function, that calculates underlinedotu^Q_l for the given matrices, underlineu and underlineu^*.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"function rhs!(du, u, x, t)\n # Reset du and flux matrix\n du .= zero(eltype(du))\n flux_numerical = copy(du)\n\n # Calculate interface and boundary fluxes, $u^* = (u^*|_{-1}, 0, ..., 0, u^*|^1)^T$\n # Since we use the flux Lax-Friedrichs from Trixi.jl, we have to pass some extra arguments.\n # Trixi.jl needs the equation we are dealing with and an additional `1`, that indicates the\n # first coordinate direction.\n equations = LinearScalarAdvectionEquation1D(1.0)\n for element in 2:n_elements-1\n # left interface\n flux_numerical[1, element] = surface_flux(u[end, element-1], u[1, element], 1, equations)\n flux_numerical[end, element-1] = flux_numerical[1, element]\n # right interface\n flux_numerical[end, element] = surface_flux(u[end, element], u[1, element+1], 1, equations)\n flux_numerical[1, element+1] = flux_numerical[end, element]\n end\n # boundary flux\n flux_numerical[1, 1] = surface_flux(u[end, end], u[1, 1], 1, equations)\n flux_numerical[end, end] = flux_numerical[1, 1]\n\n # Calculate surface integrals, $- M^{-1} * B * u^*$\n for element in 1:n_elements\n du[:, element] -= (M \\ B) * flux_numerical[:, element]\n end\n\n # Calculate volume integral, $+ M^{-1} * D^T * M * u$\n for element in 1:n_elements\n flux = u[:, element]\n du[:, element] += (M \\ transpose(D)) * M * flux\n end\n\n # Apply Jacobian from mapping to reference element\n for element in 1:n_elements\n du[:, element] *= 2 / dx\n end\n\n return nothing\nend","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Combining all definitions and the function that calculates the right-hand side, we define the ODE and solve it until t=2 with OrdinaryDiffEq's solve function and the Runge-Kutta method RDPK3SpFSAL49(), which is optimized for discontinuous Galerkin methods and hyperbolic PDEs. We set some common error tolerances abstol=1.0e-6, reltol=1.0e-6 and pass save_everystep=false to avoid saving intermediate solution vectors in memory.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using OrdinaryDiffEq\ntspan = (0.0, 2.0)\node = ODEProblem(rhs!, u0, tspan, x)\n\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)\n\nplot(vec(x), vec(sol.u[end]), label=\"solution at t=$(tspan[2])\", legend=:topleft, lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Alternative-Implementation-based-on-Trixi.jl","page":"3 Introduction to DG methods","title":"Alternative Implementation based on Trixi.jl","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Now, we implement the same example. But this time, we directly use the functionality that Trixi.jl provides.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using Trixi, OrdinaryDiffEq, Plots","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"First, define the equation with a advection_velocity of 1.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"advection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Then, create a DG solver with polynomial degree = 3 and (local) Lax-Friedrichs/Rusanov flux as surface flux. The implementation of the basis and the numerical flux is now already done.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"solver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We will now create a mesh with 16 elements for the physical domain [-1, 1] with periodic boundaries. We use Trixi.jl's standard mesh TreeMesh. Since it's limited to hypercube domains, we choose 2^4=16 elements. The mesh type supports AMR, that' why n_cells_max has to be set, even if we don't need AMR here.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"coordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4, # number of elements = 2^4\n n_cells_max=30_000) # set maximum capacity of tree data structure (only needed for AMR)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"A semidiscretization collects data structures and functions for the spatial discretization. In Trixi.jl, an initial condition has the following parameter structure and is of the type SVector.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"initial_condition_sine_wave(x, t, equations) = SVector(1.0 + 0.5 * sin(pi * sum(x - equations.advection_velocity * t)))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_sine_wave, solver)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"Again, combining all definitions and the function that calculates the right-hand side, we define the ODE and solve it until t=2 with OrdinaryDiffEq's solve function and the Runge-Kutta method RDPK3SpFSAL49().","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"tspan = (0.0, 2.0)\node_trixi = semidiscretize(semi, tspan)\n\nsol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);\nnothing #hide","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"We add a plot of the new approximated solution to the one calculated before.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"plot!(sol_trixi, label=\"solution at t=$(tspan[2]) with Trixi.jl\", legend=:topleft, linestyle=:dash, lw=2)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Summary-of-the-code","page":"3 Introduction to DG methods","title":"Summary of the code","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"To sum up, here is the complete code that we used.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Raw-implementation","page":"3 Introduction to DG methods","title":"Raw implementation","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"# basis: Legendre-Gauss-Lobatto\nusing Trixi, LinearAlgebra, OrdinaryDiffEq, Plots\npolydeg = 3 #= polynomial degree =#\nbasis = LobattoLegendreBasis(polydeg)\nnodes = basis.nodes # Gauss-Lobatto nodes in [-1, 1]\nD = basis.derivative_matrix\nM = diagm(basis.weights) # mass matrix\nB = diagm([-1; zeros(polydeg - 1); 1])\n\n# mesh\ncoordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nn_elements = 16 # number of elements\n\ndx = (coordinates_max - coordinates_min) / n_elements # length of one element\n\nx = Matrix{Float64}(undef, length(nodes), n_elements)\nfor element in 1:n_elements\n x_l = -1 + (element - 1) * dx + dx/2\n for i in 1:length(nodes) # basis points in [-1, 1]\n ξ = nodes[i]\n x[i, element] = x_l + dx/2 * ξ\n end\nend\n\n# initial condition\ninitial_condition_sine_wave(x) = 1.0 + 0.5 * sin(pi * x)\nu0 = initial_condition_sine_wave.(x)\n\nplot(vec(x), vec(u0), label=\"initial condition\", legend=:topleft)\n\n# flux Lax-Friedrichs\nsurface_flux = flux_lax_friedrichs\n\n# rhs! method\nfunction rhs!(du, u, x, t)\n # reset du\n du .= zero(eltype(du))\n flux_numerical = copy(du)\n\n # calculate interface and boundary fluxes\n equations = LinearScalarAdvectionEquation1D(1.0)\n for element in 2:n_elements-1\n # left interface\n flux_numerical[1, element] = surface_flux(u[end, element-1], u[1, element], 1, equations)\n flux_numerical[end, element-1] = flux_numerical[1, element]\n # right interface\n flux_numerical[end, element] = surface_flux(u[end, element], u[1, element+1], 1, equations)\n flux_numerical[1, element+1] = flux_numerical[end, element]\n end\n # boundary flux\n flux_numerical[1, 1] = surface_flux(u[end, end], u[1, 1], 1, equations)\n flux_numerical[end, end] = flux_numerical[1, 1]\n\n # calculate surface integrals\n for element in 1:n_elements\n du[:, element] -= (M \\ B) * flux_numerical[:, element]\n end\n\n # calculate volume integral\n for element in 1:n_elements\n flux = u[:, element]\n du[:, element] += (M \\ transpose(D)) * M * flux\n end\n\n # apply Jacobian from mapping to reference element\n for element in 1:n_elements\n du[:, element] *= 2 / dx\n end\n\n return nothing\nend\n\n# create ODE problem\ntspan = (0.0, 2.0)\node = ODEProblem(rhs!, u0, tspan, x)\n\n# solve\nsol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)\n\nplot(vec(x), vec(sol.u[end]), label=\"solution at t=$(tspan[2])\", legend=:topleft, lw=3)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Alternative-Implementation-based-on-Trixi.jl-2","page":"3 Introduction to DG methods","title":"Alternative Implementation based on Trixi.jl","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using Trixi, OrdinaryDiffEq, Plots\n\n# equation with a advection_velocity of `1`.\nadvection_velocity = 1.0\nequations = LinearScalarAdvectionEquation1D(advection_velocity)\n\n# create DG solver with flux lax friedrichs and LGL basis\nsolver = DGSEM(polydeg=3, surface_flux=flux_lax_friedrichs)\n\n# distretize domain with `TreeMesh`\ncoordinates_min = -1.0 # minimum coordinate\ncoordinates_max = 1.0 # maximum coordinate\nmesh = TreeMesh(coordinates_min, coordinates_max,\n initial_refinement_level=4, # number of elements = 2^4\n n_cells_max=30_000)\n\n# create initial condition and semidiscretization\ninitial_condition_sine_wave(x, t, equations) = SVector(1.0 + 0.5 * sin(pi * sum(x - equations.advection_velocity * t)))\n\nsemi = SemidiscretizationHyperbolic(mesh, equations, initial_condition_sine_wave, solver)\n\n# solve\ntspan = (0.0, 2.0)\node_trixi = semidiscretize(semi, tspan)\nsol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);\n\nplot!(sol_trixi, label=\"solution at t=$(tspan[2]) with Trixi.jl\", legend=:topleft, linestyle=:dash, lw=2)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/#Package-versions","page":"3 Introduction to DG methods","title":"Package versions","text":"","category":"section"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"These results were obtained using the following versions.","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"using InteractiveUtils\nversioninfo()\n\nusing Pkg\nPkg.status([\"Trixi\", \"OrdinaryDiffEq\", \"Plots\"],\n mode=PKGMODE_MANIFEST)","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"","category":"page"},{"location":"tutorials/scalar_linear_advection_1d/","page":"3 Introduction to DG methods","title":"3 Introduction to DG methods","text":"This page was generated using Literate.jl.","category":"page"},{"location":"reference-trixi2vtk/#Trixi2Vtk.jl-API","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl API","text":"","category":"section"},{"location":"reference-trixi2vtk/","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl","text":"CurrentModule = Trixi2Vtk","category":"page"},{"location":"reference-trixi2vtk/","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.jl","text":"Modules = [Trixi2Vtk]","category":"page"},{"location":"reference-trixi2vtk/#Trixi2Vtk.trixi2vtk-Tuple{Vararg{AbstractString}}","page":"Trixi2Vtk.jl","title":"Trixi2Vtk.trixi2vtk","text":"trixi2vtk(filename::AbstractString...;\n format=:vtu, verbose=false, hide_progress=false, pvd=nothing,\n output_directory=\".\", nvisnodes=nothing, save_celldata=true,\n reinterpolate=true, data_is_uniform=false)\n\nConvert Trixi-generated output files to VTK files (VTU or VTI).\n\nArguments\n\nfilename: One or more Trixi solution/restart/mesh files to convert to a VTK file. Filenames support file globbing, e.g., \"solution*\" to match all files starting with solution.\nformat: Output format for solution/restart files. Can be 'vtu' or 'vti'.\nverbose: Set to true to enable verbose output.\nhide_progress: Hide progress bar (will be hidden automatically if verbose is true).\npvd: Use this filename to store PVD file (instead of auto-detecting name). Note that only the name will be used (directory and file extension are ignored).\noutput_directory: Output directory where generated files are stored.\nnvisnodes: Number of visualization nodes per element. (default: number of DG nodes for StructuredMesh or UnstructuredMesh2D, twice the number of DG nodes for TreeMesh). A value of 0 (zero) uses the number of nodes in the DG elements.\nsave_celldata: Boolean value to determine if cell-based data should be saved. (default: true)\nreinterpolate: Boolean value to determine if data should be reinterpolated onto uniform points. When false the raw data at the compute nodes is copied into the appropriate format. (default: true)\ndata_is_uniform: Boolean to indicate if the data to be converted is from a finite difference method on a uniform grid of points. (default: false)\n\nExamples\n\njulia> trixi2vtk(\"out/solution_000*.h5\")\n[...]\n\n\n\n\n\n","category":"method"}] } diff --git a/previews/PR1624/styleguide/index.html b/previews/PR1624/styleguide/index.html index 3b404bd4eac..7e268d252cf 100644 --- a/previews/PR1624/styleguide/index.html +++ b/previews/PR1624/styleguide/index.html @@ -1,2 +1,2 @@ -Style guide · Trixi.jl
GitHub

Style guide

Coding style is an inherently personal - and thus hotly contested - issue. Since code is usually "written once, read often", it helps regular developers, new users, and reviewers if code is formatted consistently. We therefore believe in the merit of using a common coding style throughout Trixi.jl, even at the expense that not everyone can be happy with every detailed style decision. If you came here because you are furious about our code formatting rules, here is a happy little whale for you to calm you down: 🐳

Conventions

The following lists a few coding conventions for Trixi.jl. Note that in addition to these conventions, we apply and enforce automated source code formatting (see below for more details):

  • Modules, types, structs with CamelCase.
  • Functions, variables with lowercase snake_case.
  • Indentation with 4 spaces (never tabs!)
  • Maximum line length (strictly): 92.
  • Functions that mutate their input are named with a trailing !.
  • Functions order their parameters similar to Julia Base.
    • The main modified argument comes first. For example, if the right-hand side du is modified, it should come first. If only the cache is modified, e.g., in prolong2interfaces! and its siblings, put the cache first.
    • Otherwise, use the order mesh, equations, solver, cache.
    • If something needs to be specified in more detail for dispatch, put the additional argument before the general one that is specified in more detail. For example, we use have_nonconservative_terms(equations), equations and dg.mortar, dg.
  • Prefer for i in ... to for i = ... for better semantic clarity and greater flexibility.
  • Executable code should only use ASCII characters.
  • Docstrings and comments can and should use Unicode characters where it helps understanding.
  • Multiline expressions should be explicitly grouped by parentheses and not rely on Julia's implicit line continuation syntax.
  • When naming multiple functions of a single or similar category, prefer to put the general classification first and the specialization second. Example: Use flux_central instead of central_flux. This helps when searching for available functions on the REPL (e.g., when trying to find all flux functions).

Automated source code formatting

We use JuliaFormatter.jl to format the source code of Trixi.jl, which will also enforce some of the Conventions listed above (e.g., line length or indentation with 4 spaces are automatically handled, while capitalization of names is not). Our format is mostly based on the SciML-style formatting rules. For more details you can have a look at the current .JuliaFormatter.toml file that holds the configuration options we use for JuliaFormatter.jl.

Note that we expect all contributions to Trixi.jl to be formatted with JuliaFormatter.jl before being merged to the main branch. We ensure this by running a automated check on all PRs that verify that running JuliaFormatter.jl again will not change the source code.

To format your contributions before created a PR (or, at least, before requesting a review of your PR), you need to install JuliaFormatter.jl first by running

julia -e 'using Pkg; Pkg.add("JuliaFormatter")'

You can then recursively format the core Julia files in the Trixi.jl repo by executing

julia -e 'using JuliaFormatter; format(["benchmark", "examples", "ext", "src", "test", "utils"])'

from inside the Trixi.jl repository. For convenience, there is also a script you can directly run from your terminal shell, which will automatically install JuliaFormatter in a temporary environment and then run it:

utils/trixi-format.jl

You can get more information about using the convenience script by running it with the --help/-h flag.

Checking formatting before committing

It can be convenient to check the formatting of source code automatically before each commit. We use git-hooks for it and provide a pre-commit script in the utils folder. The script uses JuliaFormatter.jl just like formatting script that runs over the whole Trixi.jl directory. You can copy the pre-commit-script into .git/hooks/pre-commit and it will check your formatting before each commit. If errors are found the commit is aborted and you can add the corrections via

git add -p
+Style guide · Trixi.jl
GitHub

Style guide

Coding style is an inherently personal - and thus hotly contested - issue. Since code is usually "written once, read often", it helps regular developers, new users, and reviewers if code is formatted consistently. We therefore believe in the merit of using a common coding style throughout Trixi.jl, even at the expense that not everyone can be happy with every detailed style decision. If you came here because you are furious about our code formatting rules, here is a happy little whale for you to calm you down: 🐳

Conventions

The following lists a few coding conventions for Trixi.jl. Note that in addition to these conventions, we apply and enforce automated source code formatting (see below for more details):

  • Modules, types, structs with CamelCase.
  • Functions, variables with lowercase snake_case.
  • Indentation with 4 spaces (never tabs!)
  • Maximum line length (strictly): 92.
  • Functions that mutate their input are named with a trailing !.
  • Functions order their parameters similar to Julia Base.
    • The main modified argument comes first. For example, if the right-hand side du is modified, it should come first. If only the cache is modified, e.g., in prolong2interfaces! and its siblings, put the cache first.
    • Otherwise, use the order mesh, equations, solver, cache.
    • If something needs to be specified in more detail for dispatch, put the additional argument before the general one that is specified in more detail. For example, we use have_nonconservative_terms(equations), equations and dg.mortar, dg.
  • Prefer for i in ... to for i = ... for better semantic clarity and greater flexibility.
  • Executable code should only use ASCII characters.
  • Docstrings and comments can and should use Unicode characters where it helps understanding.
  • Multiline expressions should be explicitly grouped by parentheses and not rely on Julia's implicit line continuation syntax.
  • When naming multiple functions of a single or similar category, prefer to put the general classification first and the specialization second. Example: Use flux_central instead of central_flux. This helps when searching for available functions on the REPL (e.g., when trying to find all flux functions).

Automated source code formatting

We use JuliaFormatter.jl to format the source code of Trixi.jl, which will also enforce some of the Conventions listed above (e.g., line length or indentation with 4 spaces are automatically handled, while capitalization of names is not). Our format is mostly based on the SciML-style formatting rules. For more details you can have a look at the current .JuliaFormatter.toml file that holds the configuration options we use for JuliaFormatter.jl.

Note that we expect all contributions to Trixi.jl to be formatted with JuliaFormatter.jl before being merged to the main branch. We ensure this by running a automated check on all PRs that verify that running JuliaFormatter.jl again will not change the source code.

To format your contributions before created a PR (or, at least, before requesting a review of your PR), you need to install JuliaFormatter.jl first by running

julia -e 'using Pkg; Pkg.add("JuliaFormatter")'

You can then recursively format the core Julia files in the Trixi.jl repo by executing

julia -e 'using JuliaFormatter; format(["benchmark", "examples", "ext", "src", "test", "utils"])'

from inside the Trixi.jl repository. For convenience, there is also a script you can directly run from your terminal shell, which will automatically install JuliaFormatter in a temporary environment and then run it:

utils/trixi-format.jl

You can get more information about using the convenience script by running it with the --help/-h flag.

Checking formatting before committing

It can be convenient to check the formatting of source code automatically before each commit. We use git-hooks for it and provide a pre-commit script in the utils folder. The script uses JuliaFormatter.jl just like formatting script that runs over the whole Trixi.jl directory. You can copy the pre-commit-script into .git/hooks/pre-commit and it will check your formatting before each commit. If errors are found the commit is aborted and you can add the corrections via

git add -p
diff --git a/previews/PR1624/testing/index.html b/previews/PR1624/testing/index.html index 86771912753..76542b77d89 100644 --- a/previews/PR1624/testing/index.html +++ b/previews/PR1624/testing/index.html @@ -1,6 +1,6 @@ -Testing · Trixi.jl
GitHub

Testing

During the development of Trixi.jl, we rely on continuous testing to ensure that modifications or new features do not break existing functionality or add other errors. In the main Trixi.jl repository (and the repositories for the visualization tool Trixi2Vtk), this is facilitated by GitHub Actions, which allows to run tests automatically upon certain events. When, how, and what is tested by GitHub Actions is controlled by the workflow file .github/workflows/ci.yml. In Trixi.jl and its related repositories, tests are triggered by

  • each git push to main and
  • each git push to any pull request.

Besides checking functionality, we also analyse the Test coverage to ensure that we do not miss important parts during testing.

Test and coverage requirements

Before merging a pull request (PR) to main, we require that

  • the code passes all functional tests
  • code coverage does not decrease.

Testing setup

The entry point for all testing is the file test/runtests.jl, which is run by the automated tests and which can be triggered manually by executing

julia> using Pkg; Pkg.test("Trixi")

in the REPL. Since there already exist many tests, we have split them up into multiple files in the test directory to allow for faster testing of individual parts of the code. Thus in addition to performing all tests, you can also just include one of the files named test_xxx.jl to run only a specific subset, e.g.,

julia> # Run all 2D tests on the P4estMesh
+Testing · Trixi.jl
GitHub
Testing
During the development of Trixi.jl, we rely on continuous testing to ensure that modifications or new features do not break existing functionality or add other errors. In the main Trixi.jl repository (and the repositories for the visualization tool Trixi2Vtk), this is facilitated by GitHub Actions, which allows to run tests automatically upon certain events. When, how, and what is tested by GitHub Actions is controlled by the workflow file .github/workflows/ci.yml. In Trixi.jl and its related repositories, tests are triggered by
  • each git push to main and
  • each git push to any pull request.
Besides checking functionality, we also analyse the Test coverage to ensure that we do not miss important parts during testing.
Test and coverage requirements
Before merging a pull request (PR) to main, we require that
  • the code passes all functional tests
  • code coverage does not decrease.
Testing setup
The entry point for all testing is the file test/runtests.jl, which is run by the automated tests and which can be triggered manually by executing
julia> using Pkg; Pkg.test("Trixi")
in the REPL. Since there already exist many tests, we have split them up into multiple files in the test directory to allow for faster testing of individual parts of the code. Thus in addition to performing all tests, you can also just include one of the files named test_xxx.jl to run only a specific subset, e.g.,
julia> # Run all 2D tests on the P4estMesh
        include(joinpath("test", "test_p4est_2d.jl"))
 
 julia> # Run all 1D tests for the Euler equations on the TreeMesh
-       include(joinpath("test", "test_tree_1d_euler.jl"))
For the automated tests with GitHub Actions, we run multiple jobs in parallel to reduce the waiting time until all tests are finished. You can see the different components that are run as jobs by looking at the TRIXI_TEST variable in test/runtests.jl.
Adding new tests
We use Julia's built-in unit testing capabilities to configure tests. In general, newly added code must be covered by at least one test, and all new elixirs added to the examples/ directory must be used at least once during testing. New tests should be added to the corresponding test/test_xxx.jl file, e.g., a test involving the 3D linear advection equation on the TreeMesh would go into test/test_tree_3d_advection.jl. Please study one of the existing tests and stay consistent to the current style when creating new tests.
Since we want to test as much as possible, we have a lot of tests and frequently create new ones. Naturally, this increases the time to wait for all tests to pass with each novel feature added to Trixi.jl. Therefore, new tests should be as short as reasonably possible, i.e., without being too insensitive to pick up changes or errors in the code.
When you add new tests, please check whether all CI jobs still take approximately the same time. If the job where you added new tests takes much longer than everything else, please consider moving some tests from one job to another (or report this incident and ask the main developers for help).
Test duration
As a general rule, tests should last no more than 10 seconds when run with a single thread and after compilation (i.e., excluding the first run).
Test coverage
In addition to ensuring that the code produces the expected results, the automated tests also record the code coverage. The resulting coverage reports, i.e., which lines of code were executed by at least one test and are thus considered "covered" by testing, are automatically uploaded to Coveralls for easy analysis. Typically, you see a number of Coveralls results at the bottom of each pull request: One for each parallel job (see Testing setup), which can usually be ignored since they only cover parts of the code by definition, and a cumulative coverage result named coverage/coveralls. The "Details" link takes you to a detailed report on which lines of code are covered by tests, which ones are missed, and especially which new lines the pull requests adds to Trixi.jl's code base that are not yet covered by testing.
Coverage requirements
In general, we require pull requests to not decrease the overall test coverage percentage in main, with a hard lower bound of 97%.

+       include(joinpath("test", "test_tree_1d_euler.jl"))

For the automated tests with GitHub Actions, we run multiple jobs in parallel to reduce the waiting time until all tests are finished. You can see the different components that are run as jobs by looking at the TRIXI_TEST variable in test/runtests.jl.

Adding new tests

We use Julia's built-in unit testing capabilities to configure tests. In general, newly added code must be covered by at least one test, and all new elixirs added to the examples/ directory must be used at least once during testing. New tests should be added to the corresponding test/test_xxx.jl file, e.g., a test involving the 3D linear advection equation on the TreeMesh would go into test/test_tree_3d_advection.jl. Please study one of the existing tests and stay consistent to the current style when creating new tests.

Since we want to test as much as possible, we have a lot of tests and frequently create new ones. Naturally, this increases the time to wait for all tests to pass with each novel feature added to Trixi.jl. Therefore, new tests should be as short as reasonably possible, i.e., without being too insensitive to pick up changes or errors in the code.

When you add new tests, please check whether all CI jobs still take approximately the same time. If the job where you added new tests takes much longer than everything else, please consider moving some tests from one job to another (or report this incident and ask the main developers for help).

Test duration

As a general rule, tests should last no more than 10 seconds when run with a single thread and after compilation (i.e., excluding the first run).

Test coverage

In addition to ensuring that the code produces the expected results, the automated tests also record the code coverage. The resulting coverage reports, i.e., which lines of code were executed by at least one test and are thus considered "covered" by testing, are automatically uploaded to Coveralls for easy analysis. Typically, you see a number of Coveralls results at the bottom of each pull request: One for each parallel job (see Testing setup), which can usually be ignored since they only cover parts of the code by definition, and a cumulative coverage result named coverage/coveralls. The "Details" link takes you to a detailed report on which lines of code are covered by tests, which ones are missed, and especially which new lines the pull requests adds to Trixi.jl's code base that are not yet covered by testing.

Coverage requirements

In general, we require pull requests to not decrease the overall test coverage percentage in main, with a hard lower bound of 97%.

diff --git a/previews/PR1624/time_integration/index.html b/previews/PR1624/time_integration/index.html index 3410eeb0cf4..f221057ab63 100644 --- a/previews/PR1624/time_integration/index.html +++ b/previews/PR1624/time_integration/index.html @@ -1,2 +1,2 @@ -Time integration · Trixi.jl
GitHub

Time integration methods

Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, explicit Runge-Kutta methods from OrdinaryDiffEq.jl are tested extensively. Interesting classes of time integration schemes are

Some common options for solve from OrdinaryDiffEq.jl are the following. Further documentation can be found in the SciML docs.

  • If you use a fixed time step method like CarpenterKennedy2N54, you need to pass a time step as dt=.... If you use a StepsizeCallback, the value passed as dt=... is irrelevant since it will be overwritten by the StepsizeCallback. If you want to use an adaptive time step method such as SSPRK43 or RDPK3SpFSAL49 and still want to use CFL-based step size control via the StepsizeCallback, you need to pass the keyword argument adaptive=false to solve.
  • You should usually set save_everystep=false. Otherwise, OrdinaryDiffEq.jl will (try to) save the numerical solution after every time step in RAM (until you run out of memory or start to swap).
  • You can set the maximal number of time steps via maxiters=....
  • SSP methods and many low-storage methods from OrdinaryDiffEq.jl support stage_limiter!s and step_limiter!s, e.g., PositivityPreservingLimiterZhangShu from Trixi.jl.
  • If you start Julia with multiple threads and want to use them also in the time integration method from OrdinaryDiffEq.jl, you need to pass the keyword argument thread=OrdinaryDiffEq.True() to the algorithm, e.g., RDPK3SpFSAL49(thread=OrdinaryDiffEq.True()) or CarpenterKennedy2N54(thread=OrdinaryDiffEq.True(), williamson_condition=false). For more information on using thread-based parallelism in Trixi.jl, please refer to Shared-memory parallelization with threads.
  • If you use error-based step size control (see also the section on error-based adaptive step sizes together with MPI, you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.
Number of `rhs!` calls

If you use explicit Runge-Kutta methods from OrdinaryDiffEq.jl, the total number of rhs! calls can be (slightly) bigger than the number of steps times the number of stages, e.g. to allow for interpolation (dense output), root-finding for continuous callbacks, and error-based time step control. In general, you often should not need to worry about this if you use Trixi.jl.

+Time integration · Trixi.jl
GitHub

Time integration methods

Trixi.jl is compatible with the SciML ecosystem for ordinary differential equations. In particular, explicit Runge-Kutta methods from OrdinaryDiffEq.jl are tested extensively. Interesting classes of time integration schemes are

Some common options for solve from OrdinaryDiffEq.jl are the following. Further documentation can be found in the SciML docs.

  • If you use a fixed time step method like CarpenterKennedy2N54, you need to pass a time step as dt=.... If you use a StepsizeCallback, the value passed as dt=... is irrelevant since it will be overwritten by the StepsizeCallback. If you want to use an adaptive time step method such as SSPRK43 or RDPK3SpFSAL49 and still want to use CFL-based step size control via the StepsizeCallback, you need to pass the keyword argument adaptive=false to solve.
  • You should usually set save_everystep=false. Otherwise, OrdinaryDiffEq.jl will (try to) save the numerical solution after every time step in RAM (until you run out of memory or start to swap).
  • You can set the maximal number of time steps via maxiters=....
  • SSP methods and many low-storage methods from OrdinaryDiffEq.jl support stage_limiter!s and step_limiter!s, e.g., PositivityPreservingLimiterZhangShu from Trixi.jl.
  • If you start Julia with multiple threads and want to use them also in the time integration method from OrdinaryDiffEq.jl, you need to pass the keyword argument thread=OrdinaryDiffEq.True() to the algorithm, e.g., RDPK3SpFSAL49(thread=OrdinaryDiffEq.True()) or CarpenterKennedy2N54(thread=OrdinaryDiffEq.True(), williamson_condition=false). For more information on using thread-based parallelism in Trixi.jl, please refer to Shared-memory parallelization with threads.
  • If you use error-based step size control (see also the section on error-based adaptive step sizes together with MPI, you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to OrdinaryDiffEq's solve, which are both included in ode_default_options.
Number of `rhs!` calls

If you use explicit Runge-Kutta methods from OrdinaryDiffEq.jl, the total number of rhs! calls can be (slightly) bigger than the number of steps times the number of stages, e.g. to allow for interpolation (dense output), root-finding for continuous callbacks, and error-based time step control. In general, you often should not need to worry about this if you use Trixi.jl.

diff --git a/previews/PR1624/troubleshooting/index.html b/previews/PR1624/troubleshooting/index.html index d7bc16966a0..03fdc04ffcb 100644 --- a/previews/PR1624/troubleshooting/index.html +++ b/previews/PR1624/troubleshooting/index.html @@ -1,5 +1,5 @@ -Troubleshooting and FAQ · Trixi.jl
GitHub

Troubleshooting and FAQ

In general, Trixi.jl works best with the newest Julia release and up-to-date dependencies. If something does not work as expected, try updating your installed Julia packages via the package manager, e.g., by running

julia> import Pkg; Pkg.update()

If you do not use the latest stable release of Julia from the official website, consider updating your Julia installation.

Installing Trixi.jl as a package only provides an older release

Trixi.jl requires fairly recent versions of several of its dependencies, which sometimes causes issues when other installed packages have conflicting version requirements. In this case, Julia's package manager Pkg will try to handle this gracefully by going back in history until it finds a Trixi.jl release whose version requirements can be met, resulting in an older - and usually outdated - version of Trixi.jl being installed.

The following example illustrates this issue:

  • The current Trixi.jl release v0.3.6 requires package Foo with a minimum version of v0.2.
  • An older Trixi.jl release v0.2.1 requires package Foo only with a minimum version of v0.1.
  • A user has already installed package Bar, which itself requires Foo with a maximum version of v0.1.

In this case, installing Trixi.jl via Pkg will result in version v0.2.1 to be installed instead of the current release v0.3.6. That is, a specific release of Trixi.jl may not be installable if it has a dependency with a higher minimum version that at the same time is restricted to a lower maximum version by another installed package.

You can check whether an outdated version of Trixi.jl is installed by executing

julia> import Pkg; Pkg.update("Trixi"); Pkg.status("Trixi")

in the REPL and comparing the reported Trixi.jl version with the version of the latest release. If the versions differ, you can confirm that it is due to a version conflict by forcing Pkg to install the latest Trixi.jl release, where version is the current release:

julia> Pkg.add(name="Trixi", version="0.3.6")

In case of a conflict, the command above will produce an error that informs you about the offending packages, similar to the following:

   Updating registry at `~/.julia/registries/General`
+Troubleshooting and FAQ · Trixi.jl
GitHub
Troubleshooting and FAQ
In general, Trixi.jl works best with the newest Julia release and up-to-date dependencies. If something does not work as expected, try updating your installed Julia packages via the package manager, e.g., by running
julia> import Pkg; Pkg.update()
If you do not use the latest stable release of Julia from the official website, consider updating your Julia installation.
Installing Trixi.jl as a package only provides an older release
Trixi.jl requires fairly recent versions of several of its dependencies, which sometimes causes issues when other installed packages have conflicting version requirements.  In this case, Julia's package manager Pkg will try to handle this gracefully by going back in history until it finds a Trixi.jl release whose version requirements can be met, resulting in an older - and usually outdated - version of Trixi.jl being installed.
The following example illustrates this issue:
  • The current Trixi.jl release v0.3.6 requires package Foo with a minimum version of v0.2.
  • An older Trixi.jl release v0.2.1 requires package Foo only with a minimum version of v0.1.
  • A user has already installed package Bar, which itself requires Foo with a maximum version of v0.1.
In this case, installing Trixi.jl via Pkg will result in version v0.2.1 to be installed instead of the current release v0.3.6. That is, a specific release of Trixi.jl may not be installable if it has a dependency with a higher minimum version that at the same time is restricted to a lower maximum version by another installed package.
You can check whether an outdated version of Trixi.jl is installed by executing
julia> import Pkg; Pkg.update("Trixi"); Pkg.status("Trixi")
in the REPL and comparing the reported Trixi.jl version with the version of the latest release. If the versions differ, you can confirm that it is due to a version conflict by forcing Pkg to install the latest Trixi.jl release, where version is the current release:
julia> Pkg.add(name="Trixi", version="0.3.6")
In case of a conflict, the command above will produce an error that informs you about the offending packages, similar to the following:
   Updating registry at `~/.julia/registries/General`
   Resolving package versions...
 ERROR: Unsatisfiable requirements detected for package DataStructures [864edb3b]:
  DataStructures [864edb3b] log:
@@ -44,4 +44,4 @@
   set_preferences!(uuid, "PrecompileNoSpecialize" => false)
   set_preferences!(uuid, "PrecompileNonStiff" => true)
   set_preferences!(uuid, "PrecompileStiff" => false)
-end
This disables precompilation of all implicit methods. This should usually not affect the runtime latency with Trixi.jl since most setups use explicit time integration methods.

+end

This disables precompilation of all implicit methods. This should usually not affect the runtime latency with Trixi.jl since most setups use explicit time integration methods.

diff --git a/previews/PR1624/tutorials/DGMulti_1/fbd159b8.svg b/previews/PR1624/tutorials/DGMulti_1/0064e13f.svg similarity index 58% rename from previews/PR1624/tutorials/DGMulti_1/fbd159b8.svg rename to previews/PR1624/tutorials/DGMulti_1/0064e13f.svg index 64f7ea02a5b..60e1d919426 100644 --- a/previews/PR1624/tutorials/DGMulti_1/fbd159b8.svg +++ b/previews/PR1624/tutorials/DGMulti_1/0064e13f.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/fadf51f7.svg b/previews/PR1624/tutorials/DGMulti_1/10abe6bf.svg similarity index 90% rename from previews/PR1624/tutorials/DGMulti_1/fadf51f7.svg rename to previews/PR1624/tutorials/DGMulti_1/10abe6bf.svg index 6194e618c3a..787f50792f0 100644 --- a/previews/PR1624/tutorials/DGMulti_1/fadf51f7.svg +++ b/previews/PR1624/tutorials/DGMulti_1/10abe6bf.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/61e49aa9.svg b/previews/PR1624/tutorials/DGMulti_1/208d42b7.svg similarity index 77% rename from previews/PR1624/tutorials/DGMulti_1/61e49aa9.svg rename to previews/PR1624/tutorials/DGMulti_1/208d42b7.svg index 017ede98d17..a6914442f66 100644 --- a/previews/PR1624/tutorials/DGMulti_1/61e49aa9.svg +++ b/previews/PR1624/tutorials/DGMulti_1/208d42b7.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/9bd05ce4.svg b/previews/PR1624/tutorials/DGMulti_1/5d9b08f6.svg similarity index 76% rename from previews/PR1624/tutorials/DGMulti_1/9bd05ce4.svg rename to previews/PR1624/tutorials/DGMulti_1/5d9b08f6.svg index 4cfb73825b6..05f6a241b71 100644 --- a/previews/PR1624/tutorials/DGMulti_1/9bd05ce4.svg +++ b/previews/PR1624/tutorials/DGMulti_1/5d9b08f6.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/f8f84043.svg b/previews/PR1624/tutorials/DGMulti_1/95b64f7c.svg similarity index 90% rename from previews/PR1624/tutorials/DGMulti_1/f8f84043.svg rename to previews/PR1624/tutorials/DGMulti_1/95b64f7c.svg index 8a8eacb9227..fb608895746 100644 --- a/previews/PR1624/tutorials/DGMulti_1/f8f84043.svg +++ b/previews/PR1624/tutorials/DGMulti_1/95b64f7c.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/165f977b.svg b/previews/PR1624/tutorials/DGMulti_1/fba4076c.svg similarity index 89% rename from previews/PR1624/tutorials/DGMulti_1/165f977b.svg rename to previews/PR1624/tutorials/DGMulti_1/fba4076c.svg index 10aa8411e73..ca7dedac91d 100644 --- a/previews/PR1624/tutorials/DGMulti_1/165f977b.svg +++ b/previews/PR1624/tutorials/DGMulti_1/fba4076c.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGMulti_1/index.html b/previews/PR1624/tutorials/DGMulti_1/index.html index 41712c0179f..df4d1a4984a 100644 --- a/previews/PR1624/tutorials/DGMulti_1/index.html +++ b/previews/PR1624/tutorials/DGMulti_1/index.html @@ -1,5 +1,5 @@ -5 DG schemes via DGMulti solver · Trixi.jl
GitHub

5: DG schemes via DGMulti solver

DGMulti is a DG solver that allows meshes with simplex elements. The basic idea and implementation of this solver is explained in section "Meshes". Here, we want to give some examples and a quick overview about the options with DGMulti.

We start with a simple example we already used in the tutorial about flux differencing. There, we implemented a simulation with initial_condition_weak_blast_wave for the 2D compressible Euler equations CompressibleEulerEquations2D and used the DG formulation with flux differencing using volume flux flux_ranocha and surface flux flux_lax_friedrichs.

Here, we want to implement the equivalent example, only now using the DGMulti solver instead of DGSEM.

using Trixi, OrdinaryDiffEq
+7 DG schemes via DGMulti solver · Trixi.jl
GitHub
7: DG schemes via DGMulti solver
  
DGMulti is a DG solver that allows meshes with simplex elements. The basic idea and implementation of this solver is explained in section "Meshes". Here, we want to give some examples and a quick overview about the options with DGMulti.
We start with a simple example we already used in the tutorial about flux differencing. There, we implemented a simulation with initial_condition_weak_blast_wave for the 2D compressible Euler equations CompressibleEulerEquations2D and used the DG formulation with flux differencing using volume flux flux_ranocha and surface flux flux_lax_friedrichs.
Here, we want to implement the equivalent example, only now using the DGMulti solver instead of DGSEM.
using Trixi, OrdinaryDiffEq
 
 equations = CompressibleEulerEquations2D(1.4)
 
@@ -28,11 +28,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       1.06200000e-06 s
+ #timesteps:                  0                run time:       1.24200000e-06 s
  Δt:             0.00000000e+00                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:         16384                alloc'd memory:       2386.347 MiB
+ #DOFs per field:         16384                alloc'd memory:       2585.584 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -41,18 +41,18 @@
  ∑∂S/∂U ⋅ Uₜ :  -4.07691781e-17
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 
-#timesteps:     10 │ Δt: 8.7782e-03 │ sim. time: 6.2446e-02 (15.612%)  │ run time: 1.5205e-01 s
-#timesteps:     20 │ Δt: 1.2426e-02 │ sim. time: 1.7307e-01 (43.267%)  │ run time: 2.9937e-01 s
-#timesteps:     30 │ Δt: 1.3710e-02 │ sim. time: 3.0622e-01 (76.554%)  │ run time: 4.5032e-01 s
+#timesteps:     10 │ Δt: 8.7782e-03 │ sim. time: 6.2446e-02 (15.612%)  │ run time: 1.4813e-01 s
+#timesteps:     20 │ Δt: 1.2426e-02 │ sim. time: 1.7307e-01 (43.267%)  │ run time: 2.9160e-01 s
+#timesteps:     30 │ Δt: 1.3710e-02 │ sim. time: 3.0622e-01 (76.554%)  │ run time: 4.3570e-01 s
 
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                 37                run time:       5.57844587e-01 s
+ #timesteps:                 37                run time:       5.40248358e-01 s
  Δt:             9.57329122e-03                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  9.40224390e-08 s
-                                               PID:            1.00798758e-07 s
- #DOFs per field:         16384                alloc'd memory:       2391.208 MiB
+ sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  9.08240690e-08 s
+                                               PID:            9.76392068e-08 s
+ #DOFs per field:         16384                alloc'd memory:       2590.445 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -65,8 +65,8 @@
 Trixi.jl simulation finished.  Final time: 0.4  Time steps: 37 (accepted), 37 (total)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
using Plots
 pd = PlotData2D(sol)
-plot(pd)
Example block output
plot(pd["rho"])
-plot!(getmesh(pd))
Example block output
This simulation is not as fast as the equivalent with TreeMesh since no special optimizations for quads (for instance tensor product structure) have been implemented. Figure 4 in "Efficient implementation of modern entropy stable and kinetic energy preserving discontinuous Galerkin methods for conservation laws" (2021) provides a nice runtime comparison between the different mesh types. On the other hand, the functions are more general and thus we have more option we can choose from.
Simulation with Gauss nodes
For instance, we can change the approximation type of our simulation.
using Trixi, OrdinaryDiffEq
+plot(pd)
Example block output
plot(pd["rho"])
+plot!(getmesh(pd))
Example block output
This simulation is not as fast as the equivalent with TreeMesh since no special optimizations for quads (for instance tensor product structure) have been implemented. Figure 4 in "Efficient implementation of modern entropy stable and kinetic energy preserving discontinuous Galerkin methods for conservation laws" (2021) provides a nice runtime comparison between the different mesh types. On the other hand, the functions are more general and thus we have more option we can choose from.
Simulation with Gauss nodes
For instance, we can change the approximation type of our simulation.
using Trixi, OrdinaryDiffEq
 equations = CompressibleEulerEquations2D(1.4)
 initial_condition = initial_condition_weak_blast_wave
initial_condition_weak_blast_wave (generic function with 13 methods)
We now use Gauss nodes instead of Gauss-Lobatto nodes which can be done for the element types Quad() and Hex(). Therefore, we set approximation_type=GaussSBP(). Alternatively, we can use a modal approach using the approximation type Polynomial().
dg = DGMulti(polydeg = 3,
              element_type = Quad(),
@@ -95,11 +95,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       1.52300000e-06 s
+ #timesteps:                  0                run time:       1.43300000e-06 s
  Δt:             0.00000000e+00                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:         16384                alloc'd memory:       2422.014 MiB
+ #DOFs per field:         16384                alloc'd memory:       2618.868 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -108,19 +108,19 @@
  ∑∂S/∂U ⋅ Uₜ :  -1.01164379e-01
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 
-#timesteps:     10 │ Δt: 5.0548e-03 │ sim. time: 3.3436e-02 (8.359%)   │ run time: 4.7752e-01 s
-#timesteps:     20 │ Δt: 8.8313e-03 │ sim. time: 1.0601e-01 (26.503%)  │ run time: 9.4270e-01 s
-#timesteps:     30 │ Δt: 1.0521e-02 │ sim. time: 2.0438e-01 (51.094%)  │ run time: 1.4082e+00 s
-#timesteps:     40 │ Δt: 1.1006e-02 │ sim. time: 3.1245e-01 (78.113%)  │ run time: 1.8753e+00 s
+#timesteps:     10 │ Δt: 5.0548e-03 │ sim. time: 3.3436e-02 (8.359%)   │ run time: 4.8293e-01 s
+#timesteps:     20 │ Δt: 8.8313e-03 │ sim. time: 1.0601e-01 (26.503%)  │ run time: 9.5003e-01 s
+#timesteps:     30 │ Δt: 1.0521e-02 │ sim. time: 2.0438e-01 (51.094%)  │ run time: 1.4173e+00 s
+#timesteps:     40 │ Δt: 1.1006e-02 │ sim. time: 3.1245e-01 (78.113%)  │ run time: 1.8849e+00 s
 
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                 48                run time:       2.27908016e+00 s
+ #timesteps:                 48                run time:       2.26586407e+00 s
  Δt:             1.00192333e-02                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  3.04877188e-07 s
-                                               PID:            3.18857922e-07 s
- #DOFs per field:         16384                alloc'd memory:       2424.384 MiB
+ sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  3.00558227e-07 s
+                                               PID:            3.17029669e-07 s
+ #DOFs per field:         16384                alloc'd memory:       2621.238 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -133,7 +133,7 @@
 Trixi.jl simulation finished.  Final time: 0.4  Time steps: 48 (accepted), 48 (total)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
using Plots
 pd = PlotData2D(sol)
-plot(pd)
Example block output
Simulation with triangular elements
Also, we can set another element type. We want to use triangles now.
using Trixi, OrdinaryDiffEq
+plot(pd)
Example block output
Simulation with triangular elements
Also, we can set another element type. We want to use triangles now.
using Trixi, OrdinaryDiffEq
 equations = CompressibleEulerEquations2D(1.4)
 initial_condition = initial_condition_weak_blast_wave
initial_condition_weak_blast_wave (generic function with 13 methods)
Since there is no direct equivalent to Gauss-Lobatto nodes on triangles, the approximation type SBP() now uses Gauss-Lobatto nodes on faces and special nodes in the interior of the triangular. More details can be found in the documentation of StartUpDG.jl.
dg = DGMulti(polydeg = 3,
              element_type = Tri(),
@@ -162,11 +162,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       1.08200000e-06 s
+ #timesteps:                  0                run time:       1.17300000e-06 s
  Δt:             0.00000000e+00                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:         30720                alloc'd memory:       2464.805 MiB
+ #DOFs per field:         30720                alloc'd memory:       2652.463 MiB
  #elements:                2048
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -175,19 +175,19 @@
  ∑∂S/∂U ⋅ Uₜ :  -6.74736856e-03
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 
-#timesteps:     10 │ Δt: 6.6947e-03 │ sim. time: 4.5960e-02 (11.490%)  │ run time: 8.2017e-01 s
-#timesteps:     20 │ Δt: 1.0524e-02 │ sim. time: 1.3699e-01 (34.249%)  │ run time: 1.6243e+00 s
-#timesteps:     30 │ Δt: 1.2096e-02 │ sim. time: 2.5197e-01 (62.993%)  │ run time: 2.4411e+00 s
-#timesteps:     40 │ Δt: 1.2622e-02 │ sim. time: 3.7597e-01 (93.993%)  │ run time: 3.2655e+00 s
+#timesteps:     10 │ Δt: 6.6947e-03 │ sim. time: 4.5960e-02 (11.490%)  │ run time: 8.1025e-01 s
+#timesteps:     20 │ Δt: 1.0524e-02 │ sim. time: 1.3699e-01 (34.249%)  │ run time: 1.6059e+00 s
+#timesteps:     30 │ Δt: 1.2096e-02 │ sim. time: 2.5197e-01 (62.993%)  │ run time: 2.4140e+00 s
+#timesteps:     40 │ Δt: 1.2622e-02 │ sim. time: 3.7597e-01 (93.993%)  │ run time: 3.2298e+00 s
 
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                 42                run time:       3.44302981e+00 s
+ #timesteps:                 42                run time:       3.40467542e+00 s
  Δt:             1.13592115e-02                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  2.85553711e-07 s
-                                               PID:            2.93127595e-07 s
- #DOFs per field:         30720                alloc'd memory:       2473.716 MiB
+ sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  2.83179097e-07 s
+                                               PID:            2.89989257e-07 s
+ #DOFs per field:         30720                alloc'd memory:       2661.374 MiB
  #elements:                2048
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -200,10 +200,10 @@
 Trixi.jl simulation finished.  Final time: 0.4  Time steps: 42 (accepted), 42 (total)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
using Plots
 pd = PlotData2D(sol)
-plot(pd)
Example block output
plot(pd["rho"])
-plot!(getmesh(pd))
Example block output
Triangular meshes on non-Cartesian domains
To use triangular meshes on a non-Cartesian domain, Trixi.jl uses the package StartUpDG.jl. The following example is based on elixir_euler_triangulate_pkg_mesh.jl and uses a pre-defined mesh from StartUpDG.jl.
using Trixi, OrdinaryDiffEq
We want to simulate the smooth initial condition initial_condition_convergence_test with source terms source_terms_convergence_test for the 2D compressible Euler equations.
equations = CompressibleEulerEquations2D(1.4)
+plot(pd)
Example block output
plot(pd["rho"])
+plot!(getmesh(pd))
Example block output
Triangular meshes on non-Cartesian domains
To use triangular meshes on a non-Cartesian domain, Trixi.jl uses the package StartUpDG.jl. The following example is based on elixir_euler_triangulate_pkg_mesh.jl and uses a pre-defined mesh from StartUpDG.jl.
using Trixi, OrdinaryDiffEq
We want to simulate the smooth initial condition initial_condition_convergence_test with source terms source_terms_convergence_test for the 2D compressible Euler equations.
equations = CompressibleEulerEquations2D(1.4)
 initial_condition = initial_condition_convergence_test
-source_terms = source_terms_convergence_test
source_terms_convergence_test (generic function with 14 methods)
We create the solver DGMulti with triangular elements (Tri()) as before.
dg = DGMulti(polydeg = 3, element_type = Tri(),
+source_terms = source_terms_convergence_test
source_terms_convergence_test (generic function with 13 methods)
We create the solver DGMulti with triangular elements (Tri()) as before.
dg = DGMulti(polydeg = 3, element_type = Tri(),
              approximation_type=Polynomial(),
              surface_flux = flux_lax_friedrichs,
              volume_integral = VolumeIntegralFluxDifferencing(flux_ranocha))
┌──────────────────────────────────────────────────────────────────────────────────────────────────┐
@@ -242,11 +242,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       1.02200000e-06 s
+ #timesteps:                  0                run time:       1.42200000e-06 s
  Δt:             1.49245207e-03                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:          5980                alloc'd memory:       2507.877 MiB
+ #DOFs per field:          5980                alloc'd memory:       2703.909 MiB
  #elements:                 598
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -255,12 +255,12 @@
  ∑∂S/∂U ⋅ Uₜ :  -1.23444983e-02
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 
-#timesteps:     20 │ Δt: 1.4925e-03 │ sim. time: 2.9849e-02 (14.925%)  │ run time: 5.4895e-01 s
-#timesteps:     40 │ Δt: 1.4925e-03 │ sim. time: 5.9698e-02 (29.849%)  │ run time: 1.0717e+00 s
-#timesteps:     60 │ Δt: 1.4925e-03 │ sim. time: 8.9547e-02 (44.774%)  │ run time: 1.6095e+00 s
-#timesteps:     80 │ Δt: 1.4925e-03 │ sim. time: 1.1940e-01 (59.698%)  │ run time: 2.1313e+00 s
-#timesteps:    100 │ Δt: 1.4925e-03 │ sim. time: 1.4925e-01 (74.623%)  │ run time: 2.6591e+00 s
-#timesteps:    120 │ Δt: 1.4925e-03 │ sim. time: 1.7909e-01 (89.547%)  │ run time: 3.1829e+00 s
+#timesteps:     20 │ Δt: 1.4925e-03 │ sim. time: 2.9849e-02 (14.925%)  │ run time: 5.3836e-01 s
+#timesteps:     40 │ Δt: 1.4925e-03 │ sim. time: 5.9698e-02 (29.849%)  │ run time: 1.0654e+00 s
+#timesteps:     60 │ Δt: 1.4925e-03 │ sim. time: 8.9547e-02 (44.774%)  │ run time: 1.5912e+00 s
+#timesteps:     80 │ Δt: 1.4925e-03 │ sim. time: 1.1940e-01 (59.698%)  │ run time: 2.1177e+00 s
+#timesteps:    100 │ Δt: 1.4925e-03 │ sim. time: 1.4925e-01 (74.623%)  │ run time: 2.6445e+00 s
+#timesteps:    120 │ Δt: 1.4925e-03 │ sim. time: 1.7909e-01 (89.547%)  │ run time: 3.1722e+00 s
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 Trixi.jl simulation finished.  Final time: 0.2  Time steps: 135 (accepted), 135 (total)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
@@ -269,11 +269,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGMulti(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                135                run time:       3.57582854e+00 s
+ #timesteps:                135                run time:       3.57029004e+00 s
  Δt:             1.14223386e-05                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      2.00000000e-01 (100.000%)     time/DOF/rhs!:  8.76292946e-07 s
-                                               PID:            8.83090824e-07 s
- #DOFs per field:          5980                alloc'd memory:       2508.350 MiB
+ sim. time:      2.00000000e-01 (100.000%)     time/DOF/rhs!:  8.74988798e-07 s
+                                               PID:            8.81678523e-07 s
+ #DOFs per field:          5980                alloc'd memory:       2704.381 MiB
  #elements:                 598
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -283,7 +283,7 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
using Plots
 pd = PlotData2D(sol)
 plot(pd["rho"])
-plot!(getmesh(pd))
Example block output
For more information, please have a look in the StartUpDG.jl documentation.
Package versions
These results were obtained using the following versions.
using InteractiveUtils
+plot!(getmesh(pd))
Example block output
For more information, please have a look in the StartUpDG.jl documentation.
Package versions
These results were obtained using the following versions.
using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -303,7 +303,7 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
+⌅ [91a5bcdd] Plots v1.39.0
   [472ebc20] StartUpDG v0.17.7
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.

+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`
+Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/DGMulti_2/index.html b/previews/PR1624/tutorials/DGMulti_2/index.html index 289f8a84398..3e765fe53ae 100644 --- a/previews/PR1624/tutorials/DGMulti_2/index.html +++ b/previews/PR1624/tutorials/DGMulti_2/index.html @@ -1,5 +1,5 @@ -6 Other SBP schemes (FD, CGSEM) via DGMulti solver · Trixi.jl
GitHub

6: Other SBP schemes (FD, CGSEM) via DGMulti solver

For a tutorial about DG schemes via the DGMulti solver please visit the previous tutorial. The DGMulti solver also supports other methods than DG. The important property a method has to fulfill is the summation-by-parts (SBP) property. The package SummationByPartsOperators.jl provides such methods, like a finite difference SBP (FD SBP) scheme. To do this, you need to create an SBP derivative operator and pass that as approximation_type to the DGMulti constructor. For example, the classical second-order FD SBP operator can be created as

using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly
+8 Other SBP schemes (FD, CGSEM) via DGMulti solver · Trixi.jl
GitHub
8: Other SBP schemes (FD, CGSEM) via DGMulti solver
  
For a tutorial about DG schemes via the DGMulti solver please visit the previous tutorial. The DGMulti solver also supports other methods than DG. The important property a method has to fulfill is the summation-by-parts (SBP) property. The package SummationByPartsOperators.jl provides such methods, like a finite difference SBP (FD SBP) scheme. To do this, you need to create an SBP derivative operator and pass that as approximation_type to the DGMulti constructor. For example, the classical second-order FD SBP operator can be created as
using Trixi.SummationByPartsOperators # or add SummationByPartsOperators to your project and use it directly
 D = derivative_operator(MattssonNordström2004(), derivative_order=1, accuracy_order=2,
                         xmin=0.0, xmax=1.0, N=11)
SBP first-derivative operator of order 2 on a grid in [0.0, 1.0] using 11 nodes 
 and coefficients of Mattsson, Nordström (2004) 
@@ -32,5 +32,5 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
   [472ebc20] StartUpDG v0.17.7
-  [9f78cca6] SummationByPartsOperators v0.5.52
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
This page was generated using Literate.jl.

+  [9f78cca6] SummationByPartsOperators v0.5.53
+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/DGSEM_FluxDiff/464dc32e.svg b/previews/PR1624/tutorials/DGSEM_FluxDiff/081aafce.svg similarity index 94% rename from previews/PR1624/tutorials/DGSEM_FluxDiff/464dc32e.svg rename to previews/PR1624/tutorials/DGSEM_FluxDiff/081aafce.svg index 557d75bb58a..37f6af2e22b 100644 --- a/previews/PR1624/tutorials/DGSEM_FluxDiff/464dc32e.svg +++ b/previews/PR1624/tutorials/DGSEM_FluxDiff/081aafce.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGSEM_FluxDiff/7294c625.svg b/previews/PR1624/tutorials/DGSEM_FluxDiff/b2d0b36c.svg similarity index 95% rename from previews/PR1624/tutorials/DGSEM_FluxDiff/7294c625.svg rename to previews/PR1624/tutorials/DGSEM_FluxDiff/b2d0b36c.svg index c63f1a637af..4fda4762dd3 100644 --- a/previews/PR1624/tutorials/DGSEM_FluxDiff/7294c625.svg +++ b/previews/PR1624/tutorials/DGSEM_FluxDiff/b2d0b36c.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/DGSEM_FluxDiff/index.html b/previews/PR1624/tutorials/DGSEM_FluxDiff/index.html index 478e4c00226..92ee5435899 100644 --- a/previews/PR1624/tutorials/DGSEM_FluxDiff/index.html +++ b/previews/PR1624/tutorials/DGSEM_FluxDiff/index.html @@ -1,5 +1,5 @@ -2 DGSEM with flux differencing · Trixi.jl
GitHub

2: DGSEM with flux differencing

This tutorial starts with a presentation of the weak formulation of the discontinuous Galerkin spectral element method (DGSEM) in order to fix the notation of the used operators. Then, the DGSEM formulation with flux differencing (split form DGSEM) and its implementation in Trixi.jl is shown.

We start with the one-dimensional conservation law

\[u_t + f(u)_x = 0, \qquad t\in \mathbb{R}^+, x\in\Omega\]

with the physical flux $f$.

We split the domain $\Omega$ into elements $K$ with center $x_K$ and size $\Delta x$. With the transformation mapping $x(\xi)=x_K + \frac{\Delta x}{2} \xi$ we can transform the reference element $[-1,1]$ to every physical element. So, the equation can be restricted to the reference element using the determinant of the Jacobian matrix of the transformation mapping $J=\frac{\partial x}{\partial \xi}=\frac{\Delta x}{2}$.

\[J u_t + f(u)_{\xi} = 0, \qquad t\in \mathbb{R}^+, \xi\in [-1,1]\]

The weak form of the DGSEM

We consider the so-called discontinuous Galerkin spectral element method (DGSEM) with collocation. It results from choosing a nodal DG ansatz using $N+1$ Gauss-Lobatto nodes $\xi_i$ in $[-1,1]$ with matching interpolation weights $w_i$, which are used for numerical integration and interpolation with the Lagrange polynomial basis $l_i$ of degree $N$. The Lagrange functions are created with those nodes and hence fulfil a Kronecker property at the GL nodes. The weak formulation of the DGSEM for one element is

\[J \underline{\dot{u}}(t) = - M^{-1} B \underline{f}^* + M^{-1} D^T M \underline{f}\]

where $\underline{u}=(u_0, u_1, \dots, u_N)^T\in\mathbb{R}^{N+1}$ is the collected pointwise evaluation of $u$ at the discretization nodes and $\dot{u} = \partial u / \partial t = u_t$ is the temporal derivative. The nodal values of the flux function $f$ results with collocation in $\underline{f}$, since $\underline{f}_j=f(\underline{u}_j)$. Moreover, we got the numerical flux $f^*=f^*(u^-, u^+)$.

We will now have a short overview over the operators we used.

The derivative matrix $D\in\mathbb{R}^{(N+1)\times (N+1)}$ mimics a spatial derivation on a discrete level with $\underline{f}_x \approx D \underline{f}$. It is defined by $D_{ij} = l_j'(\xi_i)$.

The diagonal mass matrix $M$ is defined by $M_{ij}=\langle l_j, l_i\rangle_N$ with the numerical scalar product $\langle \cdot, \cdot\rangle_N$ defined for functions $f$ and $g$ by

\[\langle f, g\rangle_N := \int_{-1, N}^1 f(\xi) g(\xi) d\xi := \sum_{k=0}^N f(\xi_k) g(\xi_k) w_k.\]

The multiplication by $M$ matches a discrete integration

\[ \int_{-1}^1 f(\xi) \underline{l}(\xi) d\xi \approx M \underline{f},\]

The boundary matrix $B=\text{diag}([-1, 0,..., 0, 1])$ represents an evaluation of a function at the boundaries $\xi_0=-1$ and $\xi_N=1$.

For these operators the following property holds:

\[ M D + (M D)^T = B.\]

This is called the summation-by-parts (SBP) property since it mimics integration by parts on a discrete level (Gassner (2013)).

The explicit definitions of the operators and the construction of the 1D algorithm can be found for instance in the tutorial introduction to DG methods or in more detail in Kopriva (2009).

This property shows the equivalence between the weak form and the following strong formulation of the DGSEM.

\[\begin{align*} +4 DGSEM with flux differencing · Trixi.jl

GitHub

4: DGSEM with flux differencing

This tutorial starts with a presentation of the weak formulation of the discontinuous Galerkin spectral element method (DGSEM) in order to fix the notation of the used operators. Then, the DGSEM formulation with flux differencing (split form DGSEM) and its implementation in Trixi.jl is shown.

We start with the one-dimensional conservation law

\[u_t + f(u)_x = 0, \qquad t\in \mathbb{R}^+, x\in\Omega\]

with the physical flux $f$.

We split the domain $\Omega$ into elements $K$ with center $x_K$ and size $\Delta x$. With the transformation mapping $x(\xi)=x_K + \frac{\Delta x}{2} \xi$ we can transform the reference element $[-1,1]$ to every physical element. So, the equation can be restricted to the reference element using the determinant of the Jacobian matrix of the transformation mapping $J=\frac{\partial x}{\partial \xi}=\frac{\Delta x}{2}$.

\[J u_t + f(u)_{\xi} = 0, \qquad t\in \mathbb{R}^+, \xi\in [-1,1]\]

The weak form of the DGSEM

We consider the so-called discontinuous Galerkin spectral element method (DGSEM) with collocation. It results from choosing a nodal DG ansatz using $N+1$ Gauss-Lobatto nodes $\xi_i$ in $[-1,1]$ with matching interpolation weights $w_i$, which are used for numerical integration and interpolation with the Lagrange polynomial basis $l_i$ of degree $N$. The Lagrange functions are created with those nodes and hence fulfil a Kronecker property at the GL nodes. The weak formulation of the DGSEM for one element is

\[J \underline{\dot{u}}(t) = - M^{-1} B \underline{f}^* + M^{-1} D^T M \underline{f}\]

where $\underline{u}=(u_0, u_1, \dots, u_N)^T\in\mathbb{R}^{N+1}$ is the collected pointwise evaluation of $u$ at the discretization nodes and $\dot{u} = \partial u / \partial t = u_t$ is the temporal derivative. The nodal values of the flux function $f$ results with collocation in $\underline{f}$, since $\underline{f}_j=f(\underline{u}_j)$. Moreover, we got the numerical flux $f^*=f^*(u^-, u^+)$.

We will now have a short overview over the operators we used.

The derivative matrix $D\in\mathbb{R}^{(N+1)\times (N+1)}$ mimics a spatial derivation on a discrete level with $\underline{f}_x \approx D \underline{f}$. It is defined by $D_{ij} = l_j'(\xi_i)$.

The diagonal mass matrix $M$ is defined by $M_{ij}=\langle l_j, l_i\rangle_N$ with the numerical scalar product $\langle \cdot, \cdot\rangle_N$ defined for functions $f$ and $g$ by

\[\langle f, g\rangle_N := \int_{-1, N}^1 f(\xi) g(\xi) d\xi := \sum_{k=0}^N f(\xi_k) g(\xi_k) w_k.\]

The multiplication by $M$ matches a discrete integration

\[ \int_{-1}^1 f(\xi) \underline{l}(\xi) d\xi \approx M \underline{f},\]

The boundary matrix $B=\text{diag}([-1, 0,..., 0, 1])$ represents an evaluation of a function at the boundaries $\xi_0=-1$ and $\xi_N=1$.

For these operators the following property holds:

\[ M D + (M D)^T = B.\]

This is called the summation-by-parts (SBP) property since it mimics integration by parts on a discrete level (Gassner (2013)).

The explicit definitions of the operators and the construction of the 1D algorithm can be found for instance in the tutorial introduction to DG methods or in more detail in Kopriva (2009).

This property shows the equivalence between the weak form and the following strong formulation of the DGSEM.

\[\begin{align*} J \underline{\dot{u}}(t) &= - M^{-1} B \underline{f}^* + M^{-1} D^T M \underline{f}\\[5pt] &= - M^{-1} B \underline{f}^* + M^{-1} (B - MD) \underline{f}\\[5pt] @@ -54,11 +54,11 @@ ──────────────────────────────────────────────────────────────────────────────────────────────────── Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3) ──────────────────────────────────────────────────────────────────────────────────────────────────── - #timesteps: 0 run time: 7.81000000e-07 s + #timesteps: 0 run time: 8.42000000e-07 s Δt: 0.00000000e+00 └── GC time: 0.00000000e+00 s (0.000%) sim. time: 0.00000000e+00 (0.000%) time/DOF/rhs!: NaN s PID: Inf s - #DOFs per field: 16384 alloc'd memory: 2516.749 MiB + #DOFs per field: 16384 alloc'd memory: 2700.882 MiB #elements: 1024 Variable: rho rho_v1 rho_v2 rho_e @@ -71,19 +71,19 @@ ──────────────────────────────────────────────────────────────────────────────────────────────────── Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3) ──────────────────────────────────────────────────────────────────────────────────────────────────── - #timesteps: 61 run time: 8.85897993e-01 s + #timesteps: 61 run time: 9.11395485e-01 s Δt: 2.65388338e-06 └── GC time: 0.00000000e+00 s (0.000%) - sim. time: 4.00000000e-01 (100.000%) time/DOF/rhs!: 9.09829036e-08 s - PID: 9.74935153e-08 s - #DOFs per field: 16384 alloc'd memory: 2518.476 MiB + sim. time: 4.00000000e-01 (100.000%) time/DOF/rhs!: 9.30133547e-08 s + PID: 1.00329540e-07 s + #DOFs per field: 16384 alloc'd memory: 2702.610 MiB #elements: 1024 Variable: rho rho_v1 rho_v2 rho_e L2 error: 6.17814257e-02 5.02178088e-02 5.02253900e-02 2.25981851e-01 Linf error: 2.91149630e-01 3.21787795e-01 3.22040740e-01 1.04645370e+00 - ∑∂S/∂U ⋅ Uₜ : -2.34481695e-18 + ∑∂S/∂U ⋅ Uₜ : -4.30663196e-18 ────────────────────────────────────────────────────────────────────────────────────────────────────

A look at the change in entropy $\sum \partial S/\partial U \cdot U_t$ in the analysis callback confirms that the flux is entropy conserving since the change is about machine precision.

We can plot the approximated solution at the time t=0.4.

using Plots
-plot(sol)
Example block output

Now, we can use for instance the dissipative flux flux_lax_friedrichs as surface flux to get an entropy stable method.

using OrdinaryDiffEq, Trixi
+plot(sol)
Example block output

Now, we can use for instance the dissipative flux flux_lax_friedrichs as surface flux to get an entropy stable method.

using OrdinaryDiffEq, Trixi
 
 gamma = 1.4
 equations = CompressibleEulerEquations2D(gamma)
@@ -113,11 +113,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       7.72000000e-07 s
+ #timesteps:                  0                run time:       8.51000000e-07 s
  Δt:             0.00000000e+00                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:         16384                alloc'd memory:       2603.176 MiB
+ #DOFs per field:         16384                alloc'd memory:       2699.726 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -130,11 +130,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                 37                run time:       5.19967342e-01 s
+ #timesteps:                 37                run time:       5.44143161e-01 s
  Δt:             9.70561500e-03                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  8.58841763e-08 s
-                                               PID:            9.36932658e-08 s
- #DOFs per field:         16384                alloc'd memory:       2604.902 MiB
+ sim. time:      4.00000000e-01 (100.000%)     time/DOF/rhs!:  9.01071495e-08 s
+                                               PID:            9.81093957e-08 s
+ #DOFs per field:         16384                alloc'd memory:       2701.452 MiB
  #elements:                1024
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -142,7 +142,7 @@
  Linf error:     2.61815838e-01   2.48816692e-01   2.48316760e-01   9.30972696e-01
  ∑∂S/∂U ⋅ Uₜ :  -1.40306972e-04
 ────────────────────────────────────────────────────────────────────────────────────────────────────

The change in entropy confirms the expected entropy stability.

using Plots
-plot(sol)
Example block output

Of course, you can use more than these two fluxes in Trixi. Here, we will give a short list of possible fluxes for the compressible Euler equations. For the volume flux Trixi.jl provides for example flux_ranocha, flux_shima_etal, flux_chandrashekar, flux_kennedy_gruber. As surface flux you can use all volume fluxes and additionally for instance flux_lax_friedrichs, flux_hll, flux_hllc.

Package versions

These results were obtained using the following versions.

using InteractiveUtils
+plot(sol)
Example block output

Of course, you can use more than these two fluxes in Trixi. Here, we will give a short list of possible fluxes for the compressible Euler equations. For the volume flux Trixi.jl provides for example flux_ranocha, flux_shima_etal, flux_chandrashekar, flux_kennedy_gruber. As surface flux you can use all volume fluxes and additionally for instance flux_lax_friedrichs, flux_hll, flux_hllc.

Package versions

These results were obtained using the following versions.

using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -162,6 +162,6 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

+⌅ [91a5bcdd] Plots v1.39.0 + [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl` +Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/adaptive_mesh_refinement/1bd6887d.svg b/previews/PR1624/tutorials/adaptive_mesh_refinement/52aac2a9.svg similarity index 85% rename from previews/PR1624/tutorials/adaptive_mesh_refinement/1bd6887d.svg rename to previews/PR1624/tutorials/adaptive_mesh_refinement/52aac2a9.svg index 9c53eee82c9..637a08626b3 100644 --- a/previews/PR1624/tutorials/adaptive_mesh_refinement/1bd6887d.svg +++ b/previews/PR1624/tutorials/adaptive_mesh_refinement/52aac2a9.svg @@ -1,35 +1,35 @@ - + - + - + - + - + - - - - - - - - - - - - - + + + + + + + + + + + + + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/adaptive_mesh_refinement/index.html b/previews/PR1624/tutorials/adaptive_mesh_refinement/index.html index 4cdab2ec780..b0c8f8e8814 100644 --- a/previews/PR1624/tutorials/adaptive_mesh_refinement/index.html +++ b/previews/PR1624/tutorials/adaptive_mesh_refinement/index.html @@ -1,5 +1,5 @@ -12 Adaptive mesh refinement · Trixi.jl
GitHub

12: Adaptive mesh refinement

Adaptive mesh refinement (AMR) is a method of adapting the resolution of the numerical method to the solution features such as turbulent regions or shocks. In those critical regions of the domain, we want the simulation to use elements with smaller mesh sizes compared to other regions. This should be automatically and dynamically adapted during the run of the simulation.

Implementation in Trixi.jl

In Trixi.jl, AMR is possible for the mesh types TreeMesh and P4estMesh. Both meshes are organized in a tree structure and therefore, each element can be refined independently. In Trixi.jl, AMR is restricted to a 2:1 refinement ratio between neighbor elements. This means that the maximum resolution difference of neighboring elements is a factor of two.

The implementation of AMR is divided into different steps. The basic refinement setting contains an indicator and a controller. These are added to the simulation by using an AMR callback.

Indicators

An indicator estimates the current accuracy of the numerical approximation. It indicates which regions of the domain need finer or coarser resolutions. In Trixi.jl, you can use for instance IndicatorLöhner and IndicatorHennemannGassner.

IndicatorLöhner (also callable with IndicatorLoehner) is an interpretation and adaptation of a FEM indicator by Löhner (1987) and estimates a weighted second derivative of a specified variable locally.

amr_indicator = IndicatorLöhner(semi, variable=variable)

All indicators have the parameter variable which is used to specify the variable for the indicator calculation. You can use for instance density, pressure or density_pressure for the compressible Euler equations. Moreover, you have the option to use simply the first conservation variable with first for any equations. This might be a good choice for a starting example.

IndicatorHennemannGassner, also used as a shock-capturing indicator, was developed by Hennemann et al. (2021) and is explained in detail in the tutorial about shock-capturing. It can be constructed as follows.

amr_indicator = IndicatorHennemannGassner(semi,
+14 Adaptive mesh refinement · Trixi.jl
GitHub
14: Adaptive mesh refinement
  
Adaptive mesh refinement (AMR) is a method of adapting the resolution of the numerical method to the solution features such as turbulent regions or shocks. In those critical regions of the domain, we want the simulation to use elements with smaller mesh sizes compared to other regions. This should be automatically and dynamically adapted during the run of the simulation.
Implementation in Trixi.jl
In Trixi.jl, AMR is possible for the mesh types TreeMesh and P4estMesh. Both meshes are organized in a tree structure and therefore, each element can be refined independently. In Trixi.jl, AMR is restricted to a 2:1 refinement ratio between neighbor elements. This means that the maximum resolution difference of neighboring elements is a factor of two.
The implementation of AMR is divided into different steps. The basic refinement setting contains an indicator and a controller. These are added to the simulation by using an AMR callback.
Indicators
An indicator estimates the current accuracy of the numerical approximation. It indicates which regions of the domain need finer or coarser resolutions. In Trixi.jl, you can use for instance IndicatorLöhner and IndicatorHennemannGassner.
IndicatorLöhner (also callable with IndicatorLoehner) is an interpretation and adaptation of a FEM indicator by Löhner (1987) and estimates a weighted second derivative of a specified variable locally.
amr_indicator = IndicatorLöhner(semi, variable=variable)
All indicators have the parameter variable which is used to specify the variable for the indicator calculation. You can use for instance density, pressure or density_pressure for the compressible Euler equations. Moreover, you have the option to use simply the first conservation variable with first for any equations. This might be a good choice for a starting example.
IndicatorHennemannGassner, also used as a shock-capturing indicator, was developed by Hennemann et al. (2021) and is explained in detail in the tutorial about shock-capturing. It can be constructed as follows.
amr_indicator = IndicatorHennemannGassner(semi,
                                           alpha_max=0.5,
                                           alpha_min=0.001,
                                           alpha_smooth=true,
@@ -53,7 +53,7 @@
             save_everystep=false, callback=callbacks);
We plot the solution and add the refined mesh at the end of the simulation.
using Plots
 pd = PlotData2D(sol)
 plot(pd)
-plot!(getmesh(pd))
Example block output
More examples
Trixi.jl provides many elixirs using AMR. We want to give some examples for different mesh types:
Animations of more interesting and complicated AMR simulations can be found below and on Trixi.jl's youtube channel "Trixi Framework".
First, we give a purely hyperbolic simulation of a Sedov blast wave with self-gravity. This simulation uses the mesh type TreeMesh as we did and the AMR indicator IndicatorHennemannGassner.
  U
z$ggcjNgq$VOVwhhoAZOryG$Ox;KgYdypPdW9<+HPzN=iaAwUQ3F*p@_=5M9%9AA5_
zGgFZ>J0+LjDra%NKHGGO4Br!WuiEf)J?hE7O7D~n%5P}lZrlp>xH7Dw=VknEU*c<@
zl5NPZH0E6TrG7Y1wA1PP1^1_)>>D4qJqGVMoK9}}X)5l?e~mCdlZboP?UM#X?v%3)
zS2hfKuw2f*<{7R36oPl>-NND=#>m+SuP&bQxpHC86VyAle<9Ca6n!mik=JYe}-fv7E
zTi9_`^ofch_O+p3)_~?EO#S-0aP{C4HuKZRyRwC0ta*$}S=I1!Y;pd^2lKi{v8KK3
z-c89%Vhd;Z8ZDic%HA~u#SfRvy25!SN;1F?|-P>y|9eh`#bhmn;%ThezHAwM@Gt7i&Xav%VQ-h{Pwmmy*?#Ovsugm{Y7`#
z*mh%$_kS*8{^=D~>yr!F*5caHeH>ERq0QeMe5+E}UajaM8oc{?eaEK7I62#`uDk8DnVjWc*6`LIDrdvnbzjl$
z9{k!xi;K?Sz4@~2xlM*ol`*3im#<8WE@TV&eulo`u_?!@*Tads$_o>`n{40r#f7V&+!Lk%KK+kdftNavp(C)h3LG=HiGS#uCFz(u0
zpZh@ftvWVIo?Uud#wv&U?o7n*Vd`z!to7YV#-8=jo+&?ub=Ilao!fL!|6Q)M)NX`w
z4o$SF`h;^3$Lh~N7Ur=02>Hywm|HCI;Ika1fn_9;x-$oKV`CaEmR
zz;FKX`-!YkZU={r_Ltblcc;d>9*AK3+;p6?PRLl$!5x`(co#2G=i1^befa-oXJInqa-PCIt=J~>ot}8Z9$9Jtchr(U)y>@KB5qJIYJ$8!!k+`3D
z-{(ibn3yflQ}#Z6-Xrw8>Bs7yTUlVfYHio2_b@qr&)Mmwe-AmkYq#fte}`f=HmQC2
zAl*V1TK@h{?d1aI-RbuyBZnfEJ=!PhwX~R-<>g#@u_uBx{+@Vemg7;DFmdOc^}GGp
z&l=?@8?BQpZa|2t_pC5B>Y|a8+R^>&LZ-^Y+q(kT05#jCD?*R5pN%5Ac4jBpvvuPm
zhfWM-qehPCVwfjmCp>>0IganZXZpIeD=Nl){@pF3;veDuOyH#WIDB6p67gg7wtjN9
zxv8UHFHJcMNuJ+3U=QwFeKM-;PLhYeNy_Z709LEAO=
z553KN+V`rS_&tl=pTGEv+R;qb=8&sRY-ku;TJb$RVE7rPzW(Z|&DE!vM)2j&yZ=V9
z`ddBk-YrdF${!m>U+xyl7N2dbA@zu1RwIoIuBoK5Nm(~j3jNYpd4x&0N82=3+dln7
zW7tQXW{>294`JP$G%e+?x}1&m``c+j0qk|=3;nh0@%`EEjM2G{#jK0+VdHJtcUkR&
zs6|7U$=Cs-LB4YrVZM&ousu8lc4_T0?Q4OVY*v6~)`RS9Rvwz-Dm`$U4Sk~Q-!B&J
z^QGA1$-`pyrkSJqqEiX1B-Qsx!~10X{>6+3uQp2A$1l-61C6s;{emp@q3$=>?LjMc
zx;0H_^E;)l%ht?fr!RZ0+HR1;D$9NhxL%RVHjl7z>5h3s^M$V;R$WvSHpiSuwM`qpxZ|r}BWUF<);!lbQe1G>?6`cW<)f
z#vSI^*xGqsNHHrlx#@8pza!FZL+`<^2_?++s7}~9yjyU_`Ig1eFr2gQoxiJOExyxw
zyQ5te{DQxoC*2%~@6~#&-aDnIk&MmojNjGlo{Z(VF8rkHg#C?*S>{%F=PESHVDxzS
z1H)V%Zt}6Z%QlUiJ9wyC2|KjqZ!SC
zOR9>;pPh{THMN%Ia`-pPE$21JAa_X3DMTOkvS)s^)Dd_fzdJYTM(LI6it5%k#k4&If
z_3e!l$720D$tYm2E6zvDe$MF(JqH^FKJE+uZg8{Z4My6yR~hBCEgt^##pHrjyn;TYY-}APDzCf>K<1c@N{-Dn-(X(QITK74xF5-H)w>Mt(O}(c){Oz60&V}lp6ZSz@PCR
z^DvK2Ui0UDQVCoA&bqLw4*ZkzC+S|nIo@omg=a86R8QN-eTAPqN8eH@Xq1fIvFX!L
zpn>+;n)Id5e4O(wpQCgB4!$D@8^0^CtDJ>DId*Uf`s-2uhy6cnE5>*HgMAm96tmc7
z!Ai4XFLbk-O<%d;F6;GrOP?WmciD=v@cml%ikQ!H&ysa}3fT5K)mua6-e!JxXE}B5
ze3NzWJom0=Z4Mie_)cdQ#_{zH^T&N^S;GG6Jo^5O6|-L^A5@nwzQa~)J!qVI`!*YI
zuxv=W72fC5@>&_NR>q1(`sgK~-;PlB@{2x%{%L;m{_!rQ?1Oeq=qdPx^Og+Q-a13h
z&i-ArYBt{So3t@J&vTEQoj!Q)V%7_s8`fR!>G=3AEA&mvo3pBz8SR*RB3GWzzPPw)
z?DV_I%yOoD}?OUF*}Y)OJFy{c0*
z^SbqVsb-Tr)-JlUo3F+#cIDyaBd_{m{nhTEneOuItljDCiB}CW*pAcJZhJqMvy8`o
zl%@va9{UDAwOM5{Hni3-=3<77#k!1YNJRO@?VA>qFc5Zoz|_^XnR3?k^sncKx5=53
z<^FFAQ!#FWzAxB@`FQf|pI2S9vYF>?@A)Gh=CX_J=5PJAq=>mD#?0SVRKm1JPX9YU
z5U=YFIiZyCESfzVwm7MWB#fO6PoGs|dzQ8FY4K=(#d)@^bL&jA%otXfTYY{{gOo*7
z`5yVu`5H4#8o8oTU>bYT!Ku|f>u7emD*1DZuoLX#=tps5SHqs2oAg3if_G?2H?7ve
z??uIwv{BmM8^5P;>C58oSbvz`TI3M675DJXnpv95;ICW=9%>3dt8T@(zRAxqUi!MW
zOGo%5&&2*JM`TR(JW6a_z~rtzBdupg?bFB9eMG0z>_e#Hy#LaNW}
z6R$o6zrT6*_93fPGPY8=I=e?J%v-Ike6C_%c)vQUu9;c_yAu7%G&wtg85&1j?--KA
zT>cCSNcKo&@|#J@KG9cM$gn&;J@ZBEhqmjcC)3Zcpvlt~wwe>l*7#e$xL16Xtx1Wy
zJgZXx>tp5LIv`)hx=i%HN~UbCB}EOKAT2Osz`nd!A#Kfzy?o*UV-cT@@6
z^LVGdWKAKnb&DCQE2aVc=J-gl$<1%H2LegVzpTIICG?a<;^fwQ5(G?
zGvpj=p*z9VB={ifplm*W{mC%abXL*lpRP&F&U?h2CJUu3_>q6-ye{diRp8&Zw<58x
zf8>7QQS_@vH>POp@GfC%N4$A4FBR*A$Sw^Ik4jkkzZ24LgqAYpwU(H205?Tc8fesK#1zf5Cw
za`=xm`2DgCL!JgKILEfwEbuRE6V5!ZOxtnQ=OXLS<9glGqFAQmS9`B>R2&Qbv{GGV
zNjeJ;=sPg8G>t{x>po`LfK*of$1UTlV$AnMCu#2L+xHmCnbVm#7x`=412XD!?Cl!(U;DZZ3CDX<
zl^eSsSoB85vNR^FAGj5ML>rGlzZO{6A2fWGQzK(vzZx`TZNWLxqGz@N8Srz`=ed|~
ziDRF41+H4)mcTw{S2s60mdqZmTsC{+t5p16>Z?@kqiL+qCOhMq9`USQqT0t9pQD-1
zo9tITUY%nPJ`PG-y73TmF8DU?Lyui7!eXWW4D{QFZJyTcR>%FfgIh+#7GfRvAYrMm
z3w|#*(r-Uo%JrF?&w)LYur2cfy0ARR-xIc
zb9p}YY1F=Ma9o1*;jdj6x~rl*!z3#6dKI(d-Kz)Yc;>TTjUB3D*QPP&b&amQNlRh3
zni$A$uZd&tI$5aMDMhgGLqCjPUJPZ;HwNy?$=S`4{l2Djn{t}T2822|FHT^|FDLC;
z{v@3>*Rx&lpidebd3ucXIQ0LCGYV%9+g-xme%7lC@hD-T1VM9(k<+D@QPbu$-
zFJM(_l3%5fcUfTPA1PCs%UD!E-#N(`_fLDS@!p&q%7(3fC#}%A$ObK)q5bo6G`oDJ
zc(X@T3e$1y@IbTk4Q4YsC#cuiG&ZbYK-Do>D(ikX@A?8W8}$%C#-94bRTU$)GWOnl()9pjTJ*`irF*LPH=WXb-ke@ZpCNj5|e3=L|MBwcR4
zzw&+8Na>+m%|qTDW2F(7u53_K3zyzcui0|{@p8#$^B-XiUJ+8QQP+)+ciAW1Xn54j
zX{x(qaO)X2UF!x*&t%w~-DEXN>RoUx*W^;FB)VDG4_@A>5~Z?%~$6$awNen>!Y`{y&-v>_<2+L&+C%vhXdaJH$3iqH;Gq)%0qnGjI+pvyz4WJR|Nc2jrcNJ;~rjjZ^dFDdBS?%?TT
z*^*QJ#(1uFO_A)&^55QV&sNEYdwYz!8Xb}}Qv5w%{5rJ7*t3#m<(V({t6Y;f)fK6%
z(kzfzSldn>+AdEr#9`l8o7cA_`Z4ay3tQw$jv1EQwCWTnEf3T!S~TdiR7b;Q5;%?Y1$dha}flYR>E@50HkgtiD-uT`Dzb;!%6D=T&KC*|9^{l_I4J
z+*h=@GWxP~&cy<|zi(3{<92%g>a+5;L~Bpv@%-UAlF|3}9o*P0PolB5@`_$nqQu-d
z%DU`FoW!qo@WKbr&PYrw!}gu}<0oDI?p4ef^+l2>m42UpDQ%OK%Yx4IeSAZ5cH6o5
zGusO!7we~`_I#QzX`9ltt>3CUk}Vyl$*nHml=z%{ddaeni!`ELOP8rPQl+~}SH|pf
zJ}>o}RWd(1D^_YfcS_&!#}7%@Jh`41(kWaTSkpLY;rsJaljfg?-dKB1y20?(w)?YU
zr573pFX*-=R(f&#{V7oa7bOWYqirQOb0u0$TQ~0BHB7S3u~VPTKNBQAt~yIKgA%22
zDl=a+Y(FmP`77yn>WOqok7ky2Q`e+PcF*5&Z0tuYw9iI
zC~4G=U}LwVFD1%*j^1lGJ3(UK$y>=~OR&VI{feUSrWd7WdxT7@Ir&`jI$OL-DsBwVX+SPG*BH6g`)ee#%seH%XChzPrh6
z?~!ZLXB`6K#(r*M`c^w~X2;6mrd@VvDdT=s%iWs-DkY~`qsC<}F;aicI~FBd
zbTekII>{=0r)Knh;pU#_)-GetaD(EMK|L~Nb^FwNtMgz}tG4dmZU)X7dB5$geS(8B
z;#k(>)`Fvwb^{n6Ij8?n5>AZP6*|g-!
zu7>V=%}q!5jUU#lheyVWQwNfdmL1P9?3$>;TCXu3|5Zcsbnn^>XV;l|J%-QAuy|*;
zx7Kin>6rLQC3)3S(~$Su9Mm3qnC|>830To~OGbHIlY*H#!5L3|5BOT7<(kfmeRW)8
zaGhzzlIk;yW5P1ro-h5FRvc#fNmbos^_AF+)G11B-i|6Y)jJ^ZoRdg|cTY3gKnq-(|EI~jjuo!n|BJTOh$
zTz_Y*Qw;L`Ra&5xG
z;VCb!7?_P|W}CWcd3DBB&*-ohld>}6rZ1k9WU@YE(ZNHFcC_`+u#ud#KhgcX>4Hab
z*{Y!l8T%Y3IrWdsGrfBA*V2Q7kDJ~cWq#wr+_f1CG}~{iO*oh_ZrY1mI+1cyceOQ2
zmmEqnPV8(tw`ubyE{&gXNk^SiF!_qxt~-LMAt
zP0<`dhj?Mo;Pttt1YZ+f0vm-%*DkOZkKPN
z9)u!V6+Kd@4f?{|^NuU@?c+XYtC=0N$>`PipRyTB)bqw6eo+HTedfK%l(>bEZ*`7+
zCW%BVdrSK)=H1Y!@e?yMc3Y@}(DDkq@+UglMWH0zdhM|`!I0YirwC}w2#pM8YpDF*4wagFR1)O2Koa8Q$S@a*3frggYJ0~+
z+;1;F-MJQu{tHuY{v#_2k!*HWUVFs~UD!ILjiez)&##K>)5#nH{t)U6I{8JUzO!8J
zdk%y$pD;Z<;vq-D@Uc8!5g|mnkSAqrw~N$~XY&hd?0`w1;7h%kD$w(@!aMnAI>^p5
z+(BxZ0;Q>Tjfna}03~yr;ugtIkWT+kL0?jx(Iy7Xa~HBXP#Lj*qY5X2kX60JVfOE<
z=yU0ZUjL18K|^9DzJ`h^=oQw;l(NAm==0=W=<^dJ$P1JH*QBM1Qhc#)Rd&7)u}C)s
z5DET5*f!6X5OZ*;?b6)=v={O1~
z%B^duLOOvKd0yt!tpV`E@^SS&nRbwBq>`@YLV!B|zFqfka{y^L7+RLPTZ|k+(v${p
zE23nQQNik#g-E@XcC|XZ3W`S
zF>$QUtf~!+CD*_E{HPw>Ff>dU+iwQ;G3S~}s5?N(O%6PKu>&aab6MW`)dD({FU!$a
zwtxwnrsJnqVrCU;qJ{GV1E{!i%4k(Zz66Ob=S25L9=16gvB%cMOXe}aHth{O1mxJ>_Pxt*K%A!
zd?CW4`bB{AM*%`8V#8nJSA;}Qe&+koorTmQ4faEilaYnMcVZK*Dd6i^8BqM5f{amC
zY0H=vAuEK@E~5Nd$fJ%0g8_P9ge5@%YrokJ>ecjB(3d5Em%sde#bd*Cj!JVSoMFiYz7OWbn$r~T7hsKTr4Kl2JF>mQm9Z2
zC=Gpolx81`oIVr^P_aos)-2`18Ix*|bmIkH^51!g)2BEaImdG3p9njdXyYq{HsW_%
z5cdZp@Z4HX@27I)ryrHue-;G@jc^)8H(NRa&phnR(DVlP#@~*`T0RHwXU|AGd=0oq
z5~_IWm;lVv^BV@-vOtp1Skn!PGaW{W-h7+c0p6*l+dktS1PaO<9}_nRfwc63l$&8I
zFwJS5zkt<)Q`h-O$;C6>dtv4^JwXe&Bm6md+qxMnkuwx~lVE^!`BG?;SS3PNBKetb
zvI;yMD*4##tbieB8?0spY~SkK&Sn+OHBz4AbH=&h`_2EM6%^ik<)d6QCs?FZtn+x
zN2K;0q2e;IiJ(7k8@7N(AcdN1y!L=t^F|i$h
zwwG7`y?ry-J!y^(U#SHQ70q7EYLm#{=g!v$el`QP0}oHVt_IMaXUAy!sSdqH@$X3nkR
zAw9INa>_TGBLkgizn~{sEDm)Y@HVXeR7M%~oT<#>b)n=Cw^qEKH>zwdtv7|wMzv2B
zoN@7FkQxaIc1kV_1*DVi-gQtx8`YnXl)ZF;Dwo%8N)cs3D~>+9rN4Plmsl11cQB6!MfN3kr@8zO@RMpwHqzxE3-FLdT>;zG04pFo{hr
zy-{!q>O7E4zj=}mWhan%Qr<{LnLjh}t`B~OTFMG+FL>^t?%m3%+)5$no21q2V~TcA
z`SVwo=@l}d#%jj5McUJ-;CuW773ojt0WEmdTTB9+ODP!GF>|J1xBcI{(9@VKfXrlV#(eWLyi+UHa~P@vdF
zFL?d9a+pQ{9}DquYTuW@KJ?Q}FY4>VpV(bqSkMq-o|;{MqK`@8o{;Fe-=F9)W4Rj!
zok_=s6l3jV`PKIfFJfEmYifAkx7gk4^qR0!?lU_vdGNT9WyC
z?~h%G#j=-%x1I{)s@8BMAGw1~(14`@^Bb7KT+~}-&lA*|m(W2>rWo>#Uv5@WKZU#(
zFJ!#kR)Sv$OI4oK%EAU6E-}3rC47g|k&7H!M3aOpTk>{Duzr)B2^A?Tn0@=nRjupR
zSohR$l=3TSOqr_HtD2b`{_13|`qi2WHmdS{95Ydh4LLo%#n1f;lZheJzwkr``&hDj
zzDVLC_B-Be?rX?@@K@WT*f2w440>>bnl$_O-&H
z#epypLD6}LrnDlhNOK0osm9be_4PveYHt|EL{rdM>e)i~TLI9oOjpRGqza8~bG#WA
z^c6CxWKNbPoTRf+On
z=pk9<0NGdqdMnd^EM(!$*?ne{r2b|P*|j*=3^$vhg;eG=ZX2r5c<9mJn6+H!@r&zn
zTmJ>^R~7hTo5
zMps!iiTH04!FNnL$^LE6K{})CUChpSv^Y|AJmFgD_V?n|^g)dY<}n^;DU$emU|`a7h)DT)xH`h^<(jv5EGHjO2fpaXQ!<=qnO9R?b_KPY;0
zIzYU~P-r(bB^siby+ORT@(2wpzIj&w-zNb`l%@ZDdjh{7c$zcB(4D4e|K
zpssu#tkR5$)L)-QQZ`!MzYU;(Z%HkF0vSVClr;~MjZ`6?WS6mhVJ0Y0gE%ZvE
z&{Au0X$O>jno?yTTt!&6!zQDw&+34ihJ?)dG2pU$tAw>&F4)y)Y1r+v0g2>6L+eyc
zfXVis)%bcF@UU7icRthS^mwtw+fFT@R+Y%+Ie!~~ng}+lwVMI!W=rqy-h9M6`1FJ`
zrv@=!YFsRutwy@1dotQ$n~>*1RjTdAO~?ZvP#^+zA}KG4UeCw5f}2CJ=?Cb4$ooi%
zm(CVyh+)Au?HRE&@L8|VyT5W7IZ=0w77ywM3UNiThiCOrC!=cr4c7sXx%kbp_g)7G
znWq*lhEULvPh%Kg*A48_Ok1KhD#6-zj)UEq9^>d9B6#aP2tucHyW<1!;QHUJs{yyV
z!S(8#;Rd2^aP`zC{RV9ppzK0Akh~7?<>~JCOr~0-f9^U7>h&HGC*Yc6(zZlCgvNU!
zkw&2LzH(B`QW8QR`?FE`Y4TaUuHd3sV->=l_*9X^u>_&)YYV)Vkcn`1+gn`a0tph)^eE2-Ma)7ac|A?PzE_jW2^K5yQgQq;XL5~Eo!1c15rKY^ypj16m2Y%WL
z(EA(p2EnZ$lwSIm@C*tVCazk%KB)rM8kspvB<-NzIQMX}sS1=2JLxaz76P9Au6drC
zLO>8RQQah$1qy9bJY)I%kiMBoBZrAJq<)9qNt8GP@sD=nk7o4&t0y{y(YY^?{*W@?
z?DJ_z%{#5}AD*>H0TJJo6qb5~l(YC_fkriA$5dhX8O}yDq>T25d#k`3^GoG@UmAfp
zG4=987Y*y_6CU-Soi*B_7O}|^To_VbxKlJI7
zN37<^`5K1p9_uOuJstD)cdkMdR_GZ|MT?N?%Ib&h#U;p`Z+MuGq#59_nrtzWN02%*
z$7qIc<%pct7kg>?JcQeJBg}&HJtCrzn?h|C21pr>3U^+qmLJWH&
zxUzM@=L&HXAlTdM?XN2W9*ft_^gWt^zaGnO^}lK$=+7=hOtv{h))ef3J**pl7=me>1MsnLK9e{WF(ylI4inx2<3SC^PK{6CC
zzTly*L#|4;*xa~Pg?KJH>$z1lA$s9^h=ZFm@_d%9=R?FRpyl0Py{=>cAioDTL0N7$3Zszoit#zG(ubW*BdylvFg*Tk2iUh4-6hlhafriJc~ZY
z-C&sBx-YD10PN{@^5T1*n>>=aeX>7-%f^tsC)#SU4Gf^e-e!k;j-lHkNU3>h2
zx@ZA$yK$%8G=l`{T;6R>d;A}yt#!1=aB~9mKC)G!a;ijU
zq$co1z8tETRdA89VH`<)=awVO-vchz1f-k_FCzt6bV=!sM9{*tSP0#?3M7q}6qpSf
z1taADo+o>6BhqcUw>?vjz@tw(1{yE9Q0GSsK4d30=yr#-)+8+#)Y~E}NTqodjg7l@
z_X4{Rgz50YO8?zQ2|gfV5%1rj#AKiR~gjKr0r+p^;-z40RGv_~XUn3^1KKMU}7UZIV60
zpfM6LJn>gzIFi1GzrYni6TT0Z+&@L2LYvt~nvXltCe3qX7Z@*Gmb_
zBP)m0|C2v3nyp4V#Pj&$8k(VJF*TJ3;Y;YVzKo+`#4^jnNwkY;Tfv<1sZ8e9tH6Z#Nq~_SHX<$VQJDrkbgI$tWp-5t(x30>Vca9cyv`
z$$4g2K8QKxJl9nWE2@`PuvtLKI|QUxoRqLYIovwr=nVhe?X?M^B*8|z&llg&ql0=^etglevQ{qbZ{
z1N(c<?+5RkV20{qkLG4KxJzJ8U&o^b!=G+Ah;_(g>sJ)M$KO)J@^eMMKl^P4k4q>|
zXeOs&MWKyjTw#^4!=~P!m3$d2;VONu^tcs#UNeBD@R=(%b0hZ?n|U!jy_S7%V80f7
z{O2r2H-^JgF??w>h3~Nw(rh+zmuxsj(U4;9$s{(Kh|Db~Rbh?BP3I_5D&Pp?^P-y@
zmDsvB(&a*iz&ZMv82z;!EO6gXH+7y2xApqBAkVoD%sPAntB4JUCEwd;8?hK+%-;g4
zHge>!X{nSFTgS(+LmG9unPffuL(ORG)qXfkU}9^g;rj$;vq+N{cqNBb%@a*cJ%50i
zMKg@HEMeHCU;oKy{7u2`H+gi`4!dDoMUPddW%b~na+`=flLdT}b&iDSRt9!GT2ja7
z^8oz%;H!_EYBtQ%+dnN75RLg$y7N_Bh2aFMgjb?rOgJaI7Yx)RtFZai@nX|I!SL<6
zvDfu!n8HJ(;f7HtpeANC{@9S)fr<);
zBgSf%L1O!g0^T#K8Cw8{P@%zY(@NebDs<@KzwF!&4P#%X-`%J{YlFWlob)ncCyX2x
zOcp!n*Q*l4WhCR!0)JO9*<=?=mI$SYE8roqRr9ub0xme)IIsUZ>oJs{A?1If^95Ct
z+`Rh~8-t1|T?eO=`_SL#77ZNB?_%6uF(v%&7cf3u&z|I1Vp!eh4wJ~-nPNmM8g^@b
zLq^55t{x+nSd-hkIK7_#FiX7|mz`IaVXyZbGY-tuSf7Z~b~%0zQf4HT{eH6o6@JF3
zR8H|8z5Um%P5g&5)NApQCT`CX%@+MR*vq5=aSoZ^`9X|_1V2o^>S%ZeMUqZy|5|WC
z6XcAJ41J$MmK&u7w{0!ZkA94gb|v~y>kqiTLdq0WMmZFkRk>=e3rCF#5RxUF%lLXyrW~wA6OF}i+8Sao7Z9@v?
z5mk3M79s6gsoL@OdX#a5P5OprAr#cRKjM`dj4ocFr70~pg5KNL)zMH(K#pOP<~@CM
zXwS!@NKtxHC^*5y<}Lqkq>=mh&21VBRGofmso;hn%2h9u+`2~xdCxj8&yN$M1_^X2
z?Bi!XwF-fvc!vYz<>7dlS6dRgByc@Q9CsP5`^9Q3xK0NNa~fCmIntpwavsvAmByjC
z|H}EJP)&47gqch`6+q=k;;BTX1bSkbd%k^63R?QZ9V*0~hThHV(L%LCr?IXoI$=(8fyspz}I@9|BR&7G&#?63d_qvnC8U=c+
zsQBsc-%i9x^X1O(gkoSxuvSv}ED;gZp>x=r%m%#W6)&B6MWA<+2A0yfqrl>}4#|&`
zL4^U-9H}5@^srAXU_~1o>A^
zRl>vGg9izeeMX8EATahzuPLYinYazvl{X0-mRcY!#kD@$pNl+ov`J8Zn1XyxqmcU*kc1?(E*L=r8OUye!kR#6
zB(kdWOsSgiA>ts#oE#cZjc7-cNPhIsM2HL+sIa^w#BI~=b&-DvBFt^2_s}96sD09v
zs3C6wTn&kFf`%Qymsc(>?t3GkZNyJR9_0e-=fPqHdFdckSA)R#=Od&=e5Tl>p&793
zeJI7wzI1LiOyd4+TxM0gk|)r_QfhL2pU)fI>7L
z)JGSz3Ha$CB|PIC)MxqCy`i@j?uI3RMMkV4iaQn9%i6T)`$Qs11vK4BEH9CSII+hc
zIO7nG31u;X_ADgvN}dPbQZ&Nh{PPh(Q#RtgXJp1d>JMJhU%TkD@d^Yi=3Bmd7>P9f
z0(){TZ$R6VnFkhgZ-MdS3N_pLdLXitZRjS~3_3eo_~uIrf!b9vZ%ykmAYeF3s9jf%
z?5}D*?GbAQ?+~+
zt2DzOfkJ$erTz64pjh&zz$VBJ5uOfy+PP~D-pM!1guQzWtncTp8M8No54BTa!568Z
z3zrSo1mp{m4&0Zi2D&jo$c{&@rnMqIKY}UaD7OIKpe)+FS^^qOCw9`V=>Vw#el9K@
zIlzjp|7p`RD2eT$sR_y%z4^Ntf5phwes67Q=PnxR7H
z)!z3JP(zAF7=z7M8k8-I#+0x55=3E;J8ncPh4#G5SXGY?Mn(BAnm*5jAddk<^j<7L
z`&b-&$`bBCgtCm<$rr6rKdR^E`_yTu!jdp2UAP3g#xC5aFeC-ljk$)9JeEff$akXu
zleK_;mON`yewYb`25?hWpV`G%qh9YiS<9oZK7^7*{J96oSFX-!*V~~hf5e`Jhn1o&
z1grD4kvb@ScSWBnof#BSxb5fLZG*ny4h}EfkAj|blpH
zzlBH=Ra}CD-=pcm>ER^NRgmUVovqUuC$ph^dz`H?AB{QP@9$M8fXc2%Z%*1(q2GlX
zVhQWUpwxA-ZdOwwcy~ZN;caL-R54CjUrdsVs&d<_-jAw)=m#(M#a51?aQl_}7YS&w
zz4;Xd>A+m{#k$q{k!%HIyZN}-p}h%REojy$Ir$77lJS3OF}MuF>8mY@U5ya`lIFhn
z<}ez0KukOOY7Y7{aof5^j0jsr6mzNA<9Tr=7@M02ld{6~D17`|A5
zgKP@8u&F(pL4z!FtcLAMD0}yHxa{M-LVjLOEIu;nPsrAFIE;kPjG9OuvmlFknk=J<
z?T)NV8<}xogi~P^_)2zI6)|N0yLcVT)|nL^OaBjM?Fv~i6tjjmK#6lxIzRk;F7s6h
z1s^6&mKe@2A`Wv!=wF`}*22y#yCQYE=a@$Pjcy7^5xcMSTjTjtW%%vaAlA(xZEW!B
zXPqcO~KuA{k4G!Ppyl?{}peuO;Z+c}@}
zcm^XPpw!F@YJneHJ^ju9V+vy>z8tf+HivQcoonby-p2l`5tbhAtA|A*CFQCcmM~UF
z`{+8eX;`zYBFpLa7wnR`9;z)m4Bx_T@mExG;4EJr`b+F|!%SI`ycQgJScn|?$HBL5
z@W%(L;-5MRaE<3yb#@r(@fVJwAKa_x!tPQ}>{dABVDjI%@^+8g;iESzdvmxt
zYI}be*7|X|?Z~+c8({&7u@g`ctKmxwo-VpY3JWX4+o
z*wx&vNZNnln5%I4ulcHKY^kil#Hl|So1c0*w|*K0!)-dq{~N@9j{SGOzU3
zDXtSXn=zC>sLO?~tN*C>=t;$n+U_hD*m+?SDz(7|mG7|W<(r&EQv~?+v^X6(Fo5kS
z6k4qg_rSz=VhoD#A25>GK`OJ*8rWnuD3QDVGnO4HHLb}lb*=)MN0H8yL%=)H_-$PtC+$GyT@M`!v!m6%{t
zVrJC34mg*8b#j=roXw41K^5#^{?t}Hg#&J%b=ZqGp~Zr7WK9dJrcwM{1Flgm7CPqfB2Ks(
zfmU6t+A_-1ho<6L?$Q2)(dp?HjzV93v^viz?$o9ZdhAq6o2LC5`Z)a0P9oJ7ji_X#
znmO~2FT8bq%2uL@)+x7!F${O1q-?j!ax>qfj31}&@i%%v0k#-jhLAn_lQZFK0F@P_
zb?K)Jx7#1cm$6Lm&h2qdB&DOWXJn+f0q(y@Lk
z-|rCLT_sY_(-u@gP9kPQxf)8v**XzAr=qXGGw-Xdau9FS^ZD*3L1=wiit2(VF)Bdj
zSX9Kd2Tp~orXwRWkT&Zo|vrlpRRMo5YIfj4m2=2%1vaL_GuU)Lu_Cy!sA
zdzsUWbn1|Zj&NooWZ4BjR=jJGQ{``}jczf>mA8XRs;bwacR10k<-SOSS;kIgn6C}l
z@?NG}axVvu2>yLA67E2f8Vz2>-qL`|o`>8%iM;^PhAHu9TJ0g0dFt8ppEd#4bFBn2
zv7ZRRvDbr+`{Cd_xJ`|Bz=MxsO+$hg5MU60fk^F)$2L5^)fXUK3K}x1I%?vZLA9Hw
zoXzJJaLVpeyUo}N%E;s{Dmk=*EN`;yyS^3Rv&Vrr8BYdM(q;ZmXR-(>qhe<&*e^h6
zU0bv`T}qJHS;p4fONmJPsQm{K!Ahi)`FGum=6J-j$;7uXT?ahRmwr)l)eV`?Ik6K+
zd~wF}&FknrSVF2OgJZ#T41!Y$>Kt=*1*7hY!_>sth_iau14*P2
zX^2T*)DS;x>W4P$i8v#-#X{0d|(_h=*-M>
zFwzNbxJu-HAH)N`CbQ=XfBHc&<*3SblhLpxd^}uZj9yP{QXl{-vg(P}n}8T7D>sR>|6?Q?N)w1AQEWIsXh%J9P&onf_E%
zuaD(AkJA?
zm;}n}m}*L<0zChT1nX
zI16kCK-;TEv5uYgeT!P|60P`m6K9(BrlUni_}{<_{Gl^_tIwg5vwm
z73pQ6J1pi}mKr6{0gM{ihy?lK8O&}92TCW@=YS;~JMT`GAaUMdOP6pby(x;)!dP>z}hGX=NP~(TYUc2Np_frI#{zYquG5(8oJ70iR-c39SLO^iuICh+FDOYoDh
z(2tw+7qE8uRi!T^obXQqZ1=L99F~BE)&o-$OtWjQ=`JB17V@5}@6UI7xGZZ!Kdk-=
z=Jq0WcdGz`P4$TQ2zwpj@k^Q+){id1mk*5-M*m#FoW_gjehl-#Outn~`e&3evjkOG5O_!O<#~Ce1^!UJ;JW+P7i$cF&e`b@l^&9rb~_A&cGzYikEIa8^>yi3W;|M
zKEog^!eRcx2}Zct+wT
zqKZ;(oP11`-k9x^X3!aZRWnNVV@^p
z2_qw?KKxHi@O=v1=ISEWuNSgE;d=@1uJW~v^F=S*E#0#;LUSG`TBRP$#JF>Ij>d*x
zTF>DQN`jt#-N)gs7;x&&klw*bs~&=n)LDML5Cbpv-5`O{Wm%|
zDlg-jx8NbJuE&w1gz7fF!nb1kfSUj>k5dn_l<9<{?+zs&(09Yy-3JG7SUc=6^pCxZ
z{xlMSRGDdiTH!TD;gJ
z%x%S&5I6e1Feq2H6>iG)>AH3C7yLQ&{B!Ze=QzvRH`RmjGC1eg`mkGyB%b#kO#DKK
z3Ab`X&6@MkWqh9=F!NgR$DMmP#HBN)iM#$SvSMcW7T#1vMZ&a+9BTSaa$&-qs!2y$44C&5wct;Ue(;ZPpG@b19>GhE3QapXG8q06
z!&1nQFr3Vi7-`OU5i9oeHS4Z^gH1e$teb(KVmDi~y&cw+;Fa%HKXPnkFiE<9_s>7%
zhnGKlEGqc4z&)0)#80Q9;Co`LM)&rfV_A0?C)^jFz$&BeJ0V{zut>X%p+M$wSU|xl
zERL=c##qf`HTK_N^#?+-SXuy_A)oN-F7V-|Y
zK>k#C;}kLGV|ilQ#X`1g;Fm6gDWIDx_(q%L;x==%SnX%tA}i-d-@X3XQNB!
zXs>Elq(JpYKS@=$GSD{+<#prjm*Gw}h6e&E-=Iu>F>OPkF0}bD?MGB03MHi3A)H1{
zXhZb;aM|_8kk(B86CpoQDD?djiQE<)N|VvGt4Dqg>Md|`Is8C^ez1}`7k6TXR^Jq1
zzBk5=YIc;$y<9jC?bIv@XC9m)_p~0c#rTjwZ`Ab9eLZl3G8@wpvWKQ
z=#+q3MC>)x{?wq8AtX=2FS?=Xfn%I>Ud9j;MYgbeyB2zeI!QFP>q5B%DMqFg1gK-}
zO!Ukw9?{wFJ8ixB8u^7f88{hKA}moF6>8su!0`wy;MSV#tw*34@D5phUuUWS
zHiB<%GSIhxG0)P6=U42+~kFf$N9eiu1#N2>Le-b@m
zbn-+J_FUv6?{GrRz5BU!inB=WO`MJf;TbnHVtR1IR}SjWEUoP;waEA2h=eT}6^Qj=
zkal0y9`F~w-qqkRi%j<`hZ1-%1JKUV5aa#@A;1~@m3p6u5dGK$Y`gtHFhX{4qN5g^
zF&}0Dk0$`>F(MQ#$_!miIEYnwlmqsY4NUw-vVc1Jo$L~GGpJiA0ylYc5vhOP(h|&6
zsHjEiFt1H2!XKBI;dc-YY)NwL9Uhb;o4xT`so&c{{@*}qm!?Zlgp`0{zE&MLf*7RnKl3jLL8j^Vc-Fgy^{F<>s{m&1=V@e^=*N4l!`#pn@
z9mjbKHmCRKaL)7z#f=1L>-vXJIvXERk$+ubU(#Bk8>#Z`CxZkq$PS5ISBQYx1k0Kd
zb|cYAXBg=|CLfHj~l;kv9Yv&XlA~z6OOP0eZ+T;JL%RUDm{=7-*
z)lH5?Hdr01Suw&-9Sf?Qh;Cx%wvSd;x-Bt9<%Vp
z$ynjr7l5m;RI6bIX0VZzU@zqy8SF*N_U~5+1t!BWILjWw1k-CK=O>Te#M<89p0YBr
z!-mbrSeOsFFnT8Sl5d=6`v)4O
z^09Z|b%Sc^mT#KaUD9l
z<{eMPev~4*ImUT#*ug)M?dCpgbKy`j60S79~oBm4d>8vOPuTj?Y7
zbu6UwZ?(lrQg2>hCpS3s0$9*d(C==hBh;iP>j22LB;adLOB=W6^|g6XEl
zx(Q5{FrA`iLTTJ4oSS6!mH!?c-bHXjIQ8>o{8%QR&hE3{FjIrli!s$j*kFNdtl4oN
z8+@u{_Mbc-j_4Bg0p-9$oVLn}sgTeC<}uXorFl$)lf>6vaunslM=5T{SrjYde#OkX
z4;~oc{!1}rP0BcgmAh+Pw6Y{{-MlCq>;mj>xeuD-I7Mpr{tm0)
z$`fWB`Fectr4^(%;mI&OAsyj@8KWl7qv}7cF4PC-cF=IH>#hgRc(#2B{F1^GDiLd)
z4I#suyE)diwC%$It-A7|helxCWD5yll+Py8&nON5)f##Fdpzk!9=
zr&&IYUxY2)BL|Nohp|WP4{iqWDdL~mzBlV&BEl=yY?552J%AU)hPQiDHZd3L6i
z;vX|f;4aZgr@WJp#5?owRxQa3;BHmx$5k0$!`}^j`0aE399%stlE~WM1aHIW>3nZ8
zri=6UGBSvQO-bE{R9?Dbp83~`)&u6SAk!&jTETKmAly$Ixf=sZ{F;)>bPmBjCA|vf
zmVN=3u3s@!=vje{{Y{ytI~w7r!qv4alWEuoDaGm=tTFJ_sD?%^yAaIb
zvTil*=3A`3T_Ids27y!e9`^ltQic80Mi2jz)W8uItJibHRDrJ#adM*uQr@!5E}gEu5TZg8jYj<);w?!3Tn}>joWufzoF{zxvBgIgR7!gJ|gzQx8xdJEtzOL{F5D%f`x4!UX8?J
z^QHwa#7lbUS-?JVh8J+L
z53xgAwl91?b>RPen}_R-4KPP0FXHDCy)dZ!^53mzF>w9YHGBUhFRVn2+l{`^32wV<
zDRXb#9(%v8KkDHj0RIYAb$X?O^zrU=6X)zthxtG^+AF<#mW
z9y2IQT3sKb-!SyKi8NTjeh8J25Kik-)WV822kxl}@1g}Nl3a0NyAb?e@kPzzbrip_
zCGeH+H)M6q6|Y)qi~W`5xcZn!5EDpp*M~!y;P$qZl=xIi?1ACX5|;!G9PsGN_CjAb
z8aBUS7RV8ame6~q4kIorgPCZy3kPULho#
z?7gy`=U7QJh^#1kW{Zq|S!G4C%1C6dtYp3CIhk46BgsgLB4tQE5#qyT40l
z7RV&D2&}6!U2&L^Q$HYBWxUJ1nC*w92q;zX_Mph4U_VzeiGKu{<&IaLUKIpMo4t><
zs;7~}6qjZ$?q^tmo8|o!Ad7Tf8`%)H5<&WUsETR@1u;G&-p-DCC8WGzY~ljLm^$(N
zcZEI6ae~9=wU%AN7j;Sj>Q4q+L|DG7$i*cxL+qwdZsZm*re1YP%-#B09HEelQ7@xy
zLVY0OauxH^GT}cBkH`EiYS`{&Ufv|(Ty<`Xj{o8ZZmVZ~A7WDfOlwEos^ftn>SGN0P?Trx`
zb|(4ZIcsd~AUk^X?0v+UcleJ7(uG|6Zt=EEOa)VonU@Y62FUHO?wZ$r7qAZ=syDP=
zx*X5Tn|ov)HOTevlZJ1U_XtnKseeq={82X`lm3(Vbc5iv5X~>t|DKR>ds&G*Js2C&
zVhfOsRm9A-JU2~%7V_km&?rm5CCqW(f+^>P7sBG0OkpM1fE39Whec$$VhmPyoG+QW
zAQ!qlIci3Hu*jOE#aXT-L^Ce$^{UMnV#mQwlX|Wld4784rZjI4mZwopw_QJq7`+Ik
zr!3sYZV4vr8Ph31iox0rP2y6dmVD3b>c1N7rgPd%mv9Rrewu;hBK(9oQmI-NgmNCw
zg(DfQ5n0$N)f?h}DN+%)D_4aU`|~i+Z4v00+<`2(ykZ!zBL-7FArrKpDiAYq(K{wt
zZ!nn&N9Qb>R^*+Yde*Yw0+wvlu5xJ02V-PIShcP%A(ZTQldsh-V74CG0)(C)h|}M%
z_NhVS(DkC(s?<#>7)AA{&(>TB&geWb|JI`b;#8mh-WP=0GTvSbdpclo-lXUST>yA<=X08HJrOve6A)HI
zK@GLrNA~2{*+IIl`@UnI4s0dciEjN61-Vk3?~w>FKvY_G_N*B%cr1!~tZxWIAIp#f
zEr$yrr1fQrkys)Kd{y4UIj0TJ2yB=W`Za;$r#p&OQ+n`zj^2O4XYE1OgjaiSRSl4%
ztr|!?p9H3&_Hx=vav_zv*{}VM3J~#ixT0Wp6cRspxHfkCD+~_pll&C%5WG4J(dqyB
z5RRUGvqI*W0!(R#-v8=sguRu7T!FPOAm3oT=3#skoG960Ni>TCewn7+Ns%cq&U&Bk
zwt6)v6YTlV?)n~(VK@UbpOym!p803OARj)-D4$$4t^l7;jN^H?-@(_)4p*1S4nZ_w
zHbp6T0)$R`-w}P&f1Hbjb8IeS;C>-ulWDyI3i_3DMNC=oR(L&3tN_jVJ?Dd=`TPb5VRebv%?
zugIXrYL++7e>!fszbP)Olwd*A!yGviwis~P&XKgj`IG2(zx-cU7#UEjK29~xXmi|-
zR0}12$c3tw$QwjEvf%ID{{8LjDS^^-j*UcRO#-PMNB+c5II3_d|Fv%4tZfe({9>}#6_p(mL|%Fer9KogVDvuW1yxNW%+
z^ym;oUt3wfGL_`ThmTSmyWPA|c1o?h6;*vSj#afA8c5=_zhw5R_n$64GYs165u(-I4oK0IDgQ;208*;#
z^>VnGasKz3XN0mA-~-_!zgvIbf@p14NrsR*s9C3L=6<}VcfO^y@P%U^DoU1^d{cS{
z<1SfCZFz@+``&&F6^k~|mYpO=-6{(B^7ua))ThDu6;vV-@gdmq2XQW$@PO=2F$n1Gyzz2F8qYT9vt~#?fzvd|JhU$S1kk*=@4xR4!KT}UuGdXCSVv8r
zK6~My2e*IBl==wdnFXHRsrG>qI@NJ^d7MDJ3)2_hOdFU#_;$}ftOSrBr3`*}7z~to
z=-y`rTf&GZhukbmh9D^B=lZy_CKN{l*~%_Nz>k>m+e#fhNZecK)G?(B*!GRKPp->A
z+JV%%JZn`@ef>5w@6TK?TiUPXt?Ud)q(s>2vJIfm*G6lOAr0VF_~O#&J2Ci4w%#?<
z0SC?1yC&Gh%R}T@WU`q8)
zYKk7*X3lonoFIjVU-A!8K~f-kQt@8z@e}s;FXBGwSvE*r8PC-Aa}PNmuvjhKvW}fl
zw^(9yTS4MO$wvR#NC6)s!}QoE>lovc%$D_c^+^9_tJww3Tr92jeY&nt4|eWM*tF3L
zWk?rX8mi863U-Pq<9(ktkYHZ9Z?vRKm>qM2CIev{8T|hK`yC$(xZ3^lX-Nx!J8jFA
zjFYDUSsKw6O9~ZK_+mPr|7ablSGPSNu_OYHjp=!$Zp}z{PHwMbdI3fy@e}R$%tl&!
z^?4pV&BII^;=?nnR*{jcOXjn#HAvQQyXy;^G;H31?8I-cM+j>kmqP!^MC=XMT?bG1
zN<_kc{!w5B2X%yVxCE*#u4TKvXM2ly!-0F_A@yIllXgNZlQ40ZqFaBxK0{
z`K`|>n9Zx49remsWd7v`al`!w7*CMqaOWX0_$L!Vr+MiE!m1_Qk|R@(srF5Vnmj5+
zkjV$Oe6Mpc>`|P|(zjGhWJuaiAVmgy>N_I0d{!D!&C3mMn-<30Po;R1EuBOBe58-*
zD-W^D`yxB&8|<(kx;aZc>hGYuwd0RcpBx&2guf-|-Z
zd8>9OL-7n2xS=^xxvPxaVyxe0!o!h7xeQuq2_~d~xzbQig$-jbzF6v{!;k3Xm+w6b
z&_3puby4m4Ji_qeGCPT1GFUgI(fOuO6+~$IigI&}Hnxyq;FuZjfV|%tRP}mXjpXXb
zrMy04gc*Ffr#eAyiio6IT#Q|`!J>R|>+34v2q`C>npl4~!kx7l^-SvtM*T*fYcJyo
zB4F(P1yjnxY*cC}zQ_=e=?6tF4pe^;OV<~p4uTJn#5Q3;i}YyBhnki}R3jO=oZTKj
zzFUiN=6XEgUE0KW`)Qn<=$>E~zh`s33`s`(zR$<*rDtQ0t~_xY9Kev<<)?a{#2q2U
z;9JFM*J^~}jgeGsW*ycgF}F0_4-wPb9L=2#^Vm7hg5qimewg~Ofu`^7d*mWtkMn2t
zAuQNAif){K3gK(}xZipIANJ9|nImXP6$X*qYko?00{pxXvHypf1b!-he+b^vfr61M
z#zjA+;l88R?P*ynSRE{snws_xQ)9y%GF!VMk7>v@dd
z3j)oHJXT5xvT$Upn;`0B0)__ej3A@QptrW`q_rjjfBLpn{6#6rsw3D{dfWZ*f-hV`boO2Uo2EpMqxKuLGF`|8-(?E-WZY!@S9WJI~$5%c6ljxn%M9;E^u;sqcMEwj-xltTn*RRX$|577G
zP|onIr`?|C(Cd!2)Qx4A@qOLfyxwcx_@9OTlNRQJs0>}j(A;fBe6_m@ZkuW0e>?~T
zdhrYR+p-E5u~1vQGxS0A`aT0r86)r{P*efM2bHN=_RZ0q%=3_1c?OJLO?vn5zBld=
z!X_rZVvTO?7PaVTBdGK_C$f1bJKUlpp5N+EIIe^>amtb2!NbG0JWeEQ;UVeS9d6HJ
zaXuTC{}imB;tH>Xi{03g@eSRM9x3}MG}v0#W6I4K?K>|)Nk3tNw%mK-Vl%0ZQcx7N
zF-NnZPhM^02$6{5qU~Qwirc0!k(qPQY
zJg7?uW|l>bLopZ6ZzL#@qzdWB(N$=EF3D`CWFDBre`argIsos^Of(ZiZTxY8ecuTe
zDm=zNP-Vqv7vu=!%}yAtLQ9#zp|IsIz>k+YbR@(81^!JpYDuN>tv8GjdV7a(scF}F
zKWh*8nO83f9g*N8K8%q)CLyS@PFt>Ajw*_KNEb$L$>0ynSw>!*6GYdH?_*|-ytw_M
z22oYZFnF_`w&-9|1Jd({gPQV^ph{NdR3d2vXlNhzU88Y<^+a7aeT6^4TlO`8?f9Om
zSeaAO8+`)iZI#3GG#`VQ~;g{rv@+X?~Z&y5o{BoW%qdTThp3~6G>9z_GdtR?&ryvba+G~F@mKO#Io7ShZ
z`CftCtEbi2jm|aP1ugbYD
z#I=mCFNV;Du@C(WTi54?sSSj*Pd#6-%=C)gnit)O(h&eB)^KcKm-9#?5Jw{HAD$kQ
z69Hc@?fqK4vX0P2doBAejA2!G4i|zpdXZ7N9EyYoJ(z$&oP>&LaNw(26WY?764t$r>D$Tsyj9a%{Nw#
zCCNsF&sU?j@2Vf-wpk=Geqe?9F=_uP+%QI%wP$6&P?}+O;zzIjYdSC=U&Uupz2Vqn
zq@tDQi7QeZ?`J_CV2izeFBR7P${yhvb`GK;`+}G!Jf;e4i^W7p7O&D}1YiyEZ>K$J
z{gM3jk#A<653uu=I$XzY93X%7?E~8UTBKDw_}_NlOAJ?zjK4OYiNx_%s*BC%VH72w
z$das3>|mpYvENr8iQw-~JmqSFUCsTr?h|nbu~Jv!RTziJ=hH9Zgi#?ZB-#g={wI&<
zp4$Aaunw@Q6AIc6&ukD%|CjgXGK!ErJMtT{F()uzlfMtzB96H$RL1m{WNes4+5T&n
zWL1QZ?awSy;enL@lqg)bqQ#~uj8kWlxe+5#VSR385iCUM{#gSPeT4ItusHHF9?^TI
zq_*F)MG*AKJ&GVciPCzhEq1qD)-!=S|-^bjczG+njcIG
z=C31t*O{cfb&oL3A5?=8@X$KKv&F;3V!Q_CKq(&`y05^8=e*HagCOWM*L{2P
zhdF3!4Sw*e!w#C9l}z6Aivr~RdA0!x8^Fi(LM-iv06@#SI?dPQ4Xah0b9NMiL4?47
zevNk)4EU>2B;>XLLNk}I3u-(Al`0?k8%@&SG_}(2IJ;cnuW!qd@Zv4JQ1tz5Z^;-;
zzW3r~E7b~+pu3XzjrS9nb!qN7$lipEuhv}nmZi{WkN>KRV@>gdTJEM-vt;OEZ=v_t
zuO0Zto$7nX`X2cFA7@LF84DVE|GNb(#W~#IOHtu=3=NtRH~0C59tqA*o0^faNrPG{
zR@~vH-iL?PEQ2Xd)RFKNYIiJBOODoe^7sn^^}`2IVw5oFPxP?jNeea*K9h#
zi7!vRogdo!3R17OpW7F20M1S&jDAEtV8&|vZ;C7_YQpi&G~)+?a+F6{Ho4tFU#Iok
zyz9_Ky@bRvUgk>R!YWe07|M3{-w`0S;%BFQ
z@eN+pZ7(pJ)tI-OXz=TgD?6KQHb9I*=&JDJA()7Mv~(ZTM92J#
z%Y3|9(I*pcJ~0)Zz!kPt74Q7r1}#nK<-<2Wp;ZNBHKu39PisXy^5|KDo<;w@t?c%K
zhT9S>g;_Xs(a5Ww*Qo=|EG@(VKPhlh_WI^>;Taf6*O90i-UQ%}9nGDl3ivSU=JNPN
z7VxZ;llpA^8xlvC8fj#-LuS`|GWAXgARyuCa}WQ!(3X2-mG-*>q(8~0IYPb)bqPT=
zq7JX&ZibJ9@RI~UMpLQ5x)chJs)ZXYQ=I{DeUig0zYC>eH=dE2w7_ewyYGE|m4bhY
z_LCyQxsdc|R;mBWa}e5g_JM6$1IVLOxw&&O2FzVbS;@TQ4qvv6f8VaM0r4LX7#hBt
z!Mq^WmlqFTg79mK$-NT6Ad$sd<$z!gOIxA`!?iDi@UX6FkpT@Tuy9)Pdb1r2G@PO~
z`+NgX_z}A#Wbs2m(l*N3P70v)lY@A?N)TA+iOfH4(*p&My4MK(#^Bn|?=j^ABCrpA
z)_C(9GnhA{V(v;|1CL8Xb_GT?5~4Gr5$+bm8o~z7iV^
z7RWD>LtB!&gCO}I%6fxVu_fltcB+;oq#Iq0lsfknJCHG%Xem6z+#^qs$37ZEMt;7O
zN#*LoSV~va+K<&#L>1KqugYm4r(o~r(`x(Jkdq}NNAL$EZmIGGKVv^8p(Tkb1=J!e
zb;LO$A}X+JPM+k#?+xrPVacATmr;(_@Ctk^kyU1&X@Zr(y
zdPIIiI(nod2W#7!Ta{l;MI^ShBq(1z#d_`Y6<5l?A|}x8*H!XTKKA6SLAsH$kH{{6o`ykIfU+ge3G2HE;Cg#$x?BBzxi}2Ox+&ZE(!onU(
z^}G%6K_bfBKdyS!BX;{&QoO2|u#Z#k5UyvuNbp^uL8V|(Z1!DTpKFH!vhd|jCt)=M
z$qljFYxqo#ut#6Hd*&7$M#fUo^XG^eX&fouFU~%P-M|M&-OKDTU3=+w636+s-BY*a
zZZ)Z4E)x15-#$E#+&5~XxWwp;?G*fW=$?6nDKftx-Ayw<+G`#d8~nSD8TWeUzdEu(
z%BBlGe;W+PO6&zC&YteZ_{@YS$C;ZE(^czdj3We$Q(sV?Ij;+Gab=73ytRz!TSne=
z4(EeyBs-UX9ONNP+_y7+Z02D>`CZ40*9wHP(ska+>n-Q8+RV
zid{-;9I#Hs&lw*By^w$(E1*sw9DBxGw%TxZ9jQ|4;jz~!zy#NP)`Lz}Be%Hxdp}G!
zVMa+WKPI1^MXbFl(jT+(s@>C_el8hXPWz*XA8gDS{xU>x=Os5x|gZ>VqnByl|j1Q|Xi?h3D=Rcav*V
zg9`PcD2F;uc%`>dZG+Jl@H$WFsQ=w3TolFAA7MR^6(^Q6+e;Pj@{i+S;VD){+l~6Ja|M_XHzB`An*BigP}cdT_SQ
zYpNB-Lao%|P&@b^)fL;gmHWVr#UqYSGXWw=!_`q|>cQ5>f9p&K^AIH}(7p0J3Hmu@`BH{=%D_H5dO~r@~3PqnxYPt&V+i6&?waC-4w)
zw?qT`O%U#WQ7)HZAF`81$@%m$;`VfqpQg$T{m~&F+cVVxT9|U!bI+{91;cv&!4pJi
zSxae)Yrh;`ySShsrs;*AA*Rxxp*fEdWmaBxSCmAZrAFCQdg<}9j
zQY80HWn5CzO%^tu7QEVz>=EG31S-ysU-Z{@+IR$bx~
zQ%<7GB8pxtHKQ=bjbcT&c@oZAao6nJAplVCOM(M&Zpno*A
zrMlk+6O-iHyIFF9*;$&b9@rO|St7g(+IvHwEUCK83bOwkN&PZQ*+zj_>XV2r4m7pkCxlVCD
z1J-Yh`!?pXpdVtn9H{R72K%pH%TGNUf=_F%eEAjD26B1qOcC2xkmUB&r_z(DASfV#
z!Mi6EtOULj4WO}r<#U$2NB<1K#IHN!dZ*4oY?k$d(wjse;mGr)__rhYGIk;xYF>bC
zG`aMLaq7TfyUaKwRS_Ec{m6;_V*@SMs23ugRp9OlIs4Ms7sNh**kL4X6iF)d1m@jL
zP}J|P?47!Nm^^Uh-8I2`uwc{Aiz?>^$T?;7enG|%9$+jz8dVw~^O9Auc0x0pUv!u6wC`OinO3e=IOy&;m4`k2akE=OL4*WaLF(cBuOGw|`3UKP1fc
zF`uf*Z)_c^7ar{`A(m<gFa9
zyC%(-gU`3Hb0?E%M-ILr?%%tjGu$xDky*|7+_frX5dW7-eXO9Wf+ZJ&YV^S6>5*v4
zbT+VV`vSd8yN9j3v5tDV{2MXtGcsJL`+>cAd#3CdIAuJI!V
zWBaDu?U6Q!$Xg)>sZL>t{
zzj!qEZ-GvA{XYl9`3yb&OT+>rU!iqhkupF6Ecvf1^8LX4Xi3ur`KvLNPyD-dsVRsK
z_g0U=Oe|K=ckIu+AB)_TkR9AZh~ddE#M6^u?O3}Y2I?qEk=Lhb)LYAPv1hJ)){Q6g
z5T6OMNcXFz$g??new}@5>?6PD;6*xTMBx44EXa$-?olsY%=Q?<41I-02p<#>>Qh%f
z-4N2mXzn)poiEZyELb0V&TP12>py#U^Ip|sXIIVLea6pV<=?Wm0w;M9-Lvn-_6efc
zb=o?!Ukdt2j_$WEMXh9{pCX>PLx2Ia93r2cok{M(-$VmeyY+m
zIPBZ*I0hFW4zvH&+CmzMKz#g7Jc>rMU`abuoD2?e$|Wjsh@EDlq)yU;#7I
z!aME>r9q}!GB5LSa(MQhO{M5875KDKW!$UA3}dMMK9BrR0AD+lHy)6?fZgZkUme^z
zL|#cveEgU~1mEh?^{`G-flCt9{xCMsAAhvoKsB-55nu8=8cg9H2E5j
z!X6_`(>gGD(ku^#V_%Zwr``Z4^Xbb`9VIY4F1~K#Q#H7HlWSO}panWmdGF&5^T6pZ
zNxv)6U+~mSh(5d^4HExk(YvdZ4c_>Ok9U75f))RM$^AXZXbF^5pq?KQQLBh|@V{!tLMZ
zy;7DS9Mz_HxhzJCW^dQ9)*3P3YKn_52Yzv(Ov%_V(Yg-4HZLRfSHuE!>V5kA{^UAb
zoVJB%_dKko;*&~r-v(`pvr{>LgmHJXqYpwGCir4!72{Q^ukZ|M>gQi?7eJA3>xjF~
z8Z0`Q5{OqapcfvrU2qOt0sR)Y7?mGR0PDYk#xF_6;k$_rEiavUuy4Qp@Y~!8+$VkI
zN&n6vy!J*rAk2Rj3N?oc*ob@tR|TIX){iejo)=FWmvCy-e?^F@If({mQT}!9ZuV_d
z=hD0JvGT|0W$)fkjtR-=rP7XFuA%$5%z4)t>E$fETe?=Wyw(Dzu=n9O;c*4^Ila4c
zrOyw&V0B%eNci@MAGlC3^;L(Z|uYkb+oNEd|Ho630<6B7-FGNMJ=aY<1DR1&}aGwLke9i
z_=I!!%UEIo-1ykQ5%~TT`e*N3E$ash^rc4ga<*FpIz0V`(p~v3zCToOL(V$}HSDE)
zYC-%6%|!=oDL$M-2PdEFFSB~%k6Q;`>rdOEl`hwQwAI{4lQr7TB<)5oYlpvYy|(S-cdvuXKF`ZP%)tni-(nxs3PuHxBXe3)CSc%LGgQn-U!cL
zdB&(UpoXsv|4H94RlzUj`I}L46QPk_f$lfsg7N3Z^&(lK1~}hA(|;S53aFpnl!rlu
z2>$BJ>*S#bPE?|6fWp4c3Lgp;EsI&FML8~MQq`@WLUX@QC56yZJWinr2xm4E^-^2GHNBvd9?paZ^+3m1U#rhOFk(jbYpxuCa{4@to
zz2-qln4=d7YXMY@XUlp#%#UKLZz)oWn>BiF+3bqf6X&i242
z=a_RtsDER@6$Bjxq(YX9-ovf$2?NeNjnGzeMed(l3TO}}pC3w(ghMnxs_j&DL8ZFh
z^gv=a%rjvO&S&=@yd=~E_^J070pw)%F_xJ|#
zgiZC`AIBZchvjH3^gs+c$$g;3DgR=B^GJpBn5Gebl5*02`$L#{rg&Wx=ts(@ZV@?q
z@PZv_YU4Sb6(r&8xN2$G0CsSuVDsGhcI53D#Wx|}-(rVSY8Nc7@E`Z0+0_HIcd=tp
z)YN5@Nu;fgg?crsA6u*9^(isyL}~`rluMZ?p5H=kbyCJ7JT)Yk*R?22w#b$*+v+h=_Wh#9
z(?31PaP1A3g~!YG4+@SU^>M+9l+q2w$uW%6X-uwM?LH={
zI7)2G7EiqTOLVbIq8%UPE+~j}0orq(Q!IEiUA|gB=cx^Mt2Xmz_dixo7
zLh`cfy2sfZu}7iO3H$@wg!9XD>n~dm)Lp7X8B1o!Fp8FLGI=)vL}lckQahs$LXlj1
zy6a~XLHLss+5OBJ^%wCKE0UvQgapwdF{c4yM3sWkX{6c^`NHA#CLU&~`|ePts>bCI
zY^NpPE}FHf&oCr=ovxWCNcL4G3ZTdQ#qz6j4BW+peDgxDt-KP#kO;4J!dwTToh_2H
z#=B4biYoVXtUE7q;BfWG`qv?|s8Nt_RB
zjGP;kqrZzN-yjP49TO
zi-wr6>vpAM|8XDjb54Bw*L&D{v8C|gZs#$lfKgQJuL!op`%6?VQwre#kBV9bl(28@
z8mx_y7RWO;*AEq=xd`83p@u{$57KnOm~UK~|ClErYu)A}i99d68bGyk1q*#5;iWR0
zf`Oq#g^@HtOs6u4u*5Bl#8|%jWMCnSG0St@FIBa5LPjyu
zt?TzU|Jz35I_IKp^2vh`>+Q@*xS)KAvZ7IWD9YA_09Q)XH5Q7|5G1WA9kSlDCRjL+)Pz+d?-|T3C(XijZkYz`6wx
zTDbky+&mawNWZl!;_H0SFQ0r{VA~X@%nb(A_uGt$<}E*#w;u)>_q$8c>#ZP4Kw&g3cgx<
zoT;)Shg%$^x%MX+K?QY?4sVq#oNJgewMwvomHf%&-32Eh!dipZG*AFLKjIUoRT$x4
zB!?f>lLp#t(682P9Vd)PD`3&5+ta;F1@Cek<#;8-<
zK7_Mh>dSwoCjqYoo3!wia=0E}uyoko1>V2Ui<4cBfFcE`yJ>F-D9&)o%#@CT88$wWLhU%n4Un&H#F?4p5q|Y
z4qRQ5;KCCwld)`fE8tSXWH*~!l<>c>P4e8P94dWrU^?_dI9dyyOJmAM5V`rcGD2e=
z4qGYw$d6*fb%%th9aP2e;W$f?!37QU%=vTJTVHZyhEFadHphl6y
zuH$DM&<%Yv)8E~9aIPn!4g#(S{z1Ew^qZYFuJg9me@y#49;qqxvW?ajy__UsqIbd@
zJ0LmMuN-rW-1#3PhZg8HKUcq}d6t}5zKQKgJWgzTN?Gj~RVugu2PrO4q%>!pF1
z%povu;aaUQ(LfJAZl;fqUqB^VwW~zg98hz|P5(1$H_*oT6CXZ!#G;#fK2L4LlF+UB
z`u%&oA?T`mu8ZW9Dn8gemCv>uf_jX`aycV+(Et^Ok5wL`==#z51y6Ac+|ZO>0p?oa
zBKq-_%p=}t2?>kwdp{%m@PxZsl(iAQ852K8UCD`Z=Df0>We`O>>O1QeURvYCOy0D*
z2L^caxfN=I1Z}j0-}(I>2^*SA(Q>neuL!T0wZ1z*y6Sx-=#U_S@099&gXcN`0&|@mEyIB
z0_cox#P{DMcKAO^{i)BLinwGz+j4IoFS?f2zQ258177%T6R)OAhsN`|obd5iLhZL@
zO!<}s(1S9%-p4b)A){TQO*s1lm?hbj(PuUWDLYM+6c>bWug6a1Mu)7pRjU4k*{NwD
zt@MIe)G8Fv*-zAjNOpkXOKbcXt20{yPd{5T9#dTdg3Gi;{3~
zZ2(++=#}oyC5u+Bec?0&)6?P&Mo7+5ABAaOnwdj;==;K
zP!s=K`AjN&q+U6Jw_JlBM%*oohEsrkuE%xdVLND;;at&cN{9C}Tx)LMdjPUdl%*<%
z-UKu+_xh!N1c3WXM-P^w+u%LYGSPo$qd}I0QZC~!2bf~Re70u$A*>+R2!D6j1KYGE
z(g$>EV11-c+V9dB;A&>RI(jo0vVW)0r|`A_zoQ8(B*$ma!m
z?okb-GulI|HW=B1mS@m8~~q
zfTf)i#to9bu;=OMO#GS+l&;)?e*zUjWi|gO3ym=B7h`&4e{Jq~J|bD0e&GzS{y52<
z#%BYBjbu`GKZ!yPrFV3t@{AzE;Ij2=+T$hsX!69tmKU(A4l_UX?J{`sBU4_uO$3V2
z#&&Vkasp)r<-Y6nbWn8S4|l29&10SfpU&YW088nx0YmX$tW34U3PwI}Nxw`d{
z>B<4ZlqmKff*wMzsT9)lOvo9O(&Mrx^=l)eyI;
ze~qYK5^HS%~X|jNqi=2y$7r
zL5hp*B~teHKS9qs;n=@=ZS~>8JBWkjk9eb9Z>;HVc=6QbcI>VFX0td?I=21h*L1nA
zKN6o9I#yWkg5BNx@apiPEy6n9S}sf4g#6U|d3e$&6Un@~ezVQXA7iWwdB*1Nj11nO
z*m3eU!E6S++?vrk#G>YoZG%b>a-RMu^h(4HEI@Q`%=fnuA|lQ4>>Ix!CbdeFQFrV~
zaPt~uZ$ONVN*b&%o#H9e2r;59
zl|=FQ!6D&A=eoO5lP&hgGbT%tRt{^jJBH5fF(70=NiRM;=6h2bJ-QjKb)XKvC^CG1
z{Z#$V)A{UT21xix|GDv%-CKg^jmSiXfFS~{60iTelT?`G$Gpkc&r8(1a&LU(sOeSb
zn&Z=wYV1{?*1H`|l=e6}N|dbrPeZdJ<8_0&#MR+@MM#c%Gdq<+A6bWbhZ`|9JV%YSws~p>
z5@-?f+~n%{Mixx17jYdwHoZArNJGw*h9kXaq^pj&I|(EgN@Or>S$(i9`Cu}0nK1n9
zkJwJnDI`?r^sn}m+KzIXDZ
z2p0N&4!xcDnvj0$Uk!W;f>7N{H;HAdlx+QV-aV5MWc4L&zd|82=7+s<)fy
z?59aX!e$4;JWtW9o(#U=tzK`XfjlyKGP#^Hx3ydGAa``g1WJV}9qspCy-y
zeAJ^YcMq$g|i;Rgge*~!_E=%|VOeYs0}gUbM0xi%yDw&^xvKi9m+wNZv_rMx1vJU@rT
z9!PHf_e~M=rJbT*tkp!~S_()7?d>qtPC9W#v0Uu_*4E51?g^86emBIe5kd4rB3#71
zb+Id)G&M!^PKaw*tkk=S41`Yiqd8OaCf0PRIMo5$MC7QIDvF-&V}8j+shMPKfb-_t
zyy#DsfK>3B)K`T@jP+V}hH1wDLTmR?u8Zs=7OH31-}aISbQV-8ndYhh+I@Nx7w>#z
zg2%~{a{U$d;6LvB>ugPk&&>-zzsb*F?7^zBa@5R_lN`}F#hHZ>aeDi|$MO*Y-D&(@
zV-;r0VQ7>@Jc1Ag>VMsorT{Zh?gT3JLZr5zZYDar2rDwJa+yr2K^7h>S)36cIrf0R
zGq)Yx#I{S{F`wEC21Jog5vYqVbkz=1Ehh>BW}`nsO7KMJ(jb3Ewx|dM@TvUFjQ}9X
z?3|YIZ!H)$WF@EzuYeDzSRt(OhD|aRly%8Zpm2+6*T08CQ2nevnT4SMcv~uIIpQY;
zGY$iVZD|aEYPQdJ^*9?aGT-V+@q`EHJdYrf4rYgzVGn(
zs{zeFCU>xxnn_?};U61z?YJVB}V-Di|cYeMV)$1bZ_OOY{2xLp*n?S5X!7svYg(&2guZ`
z>ogW(A)5M^KKuChOrP*Z-Y9SYZed!j*Dc16dybUXR01S%W7DYlLpdq@rl-?_^krI9
zK#SmUnfM%f0sP=wd1{6JhvMtOttpb1q~9%r5gu?cayBOj=79>~?kbgu^{i?g52
zC!dEy+hWZ0GDEO?<2so{D-Qpp`Zt>gk)aK_y^H}m!yt8hB}_DS1pK-1t@@T$2TbFn
zH~$JhLzT`QY)hR9f77%~`|zwisxwUc>HMlE+8sDeV>#+bTkZ*qQ;Pfd}JPxVUCZ)wIi65j6
zEAiuP{mpXMIr(rkN4^{B-mb0caCUUs;n8V)Vz#SO-Vym}!aSgCm^2Gg6U`3Cp
z_B}oQ&!Xo_NB0XK+Lbs+^>68!n1;}AdUoy-wVer5;gv~l!zP>jK!NoRgG$20sdS4&1(Ew(TuA4i?5(w+QyBaDuWiyZ4n4`h{1NXRi48|JMBYXfbF2
zQi~rYR}7cJ5UsguF9my`mfb^&%y}AA^!+a9Rpbl!r7dS-^Y9(Kd`6bDEiwa;RIam}
zSgV4KJQSyrgq9)4-fugvnm+iSE8T|P$vp7If}=(+z!0j`gj}@z6alV|BW4Ut(}2fI
zHXyvP6-1hHJZhy&hP$_xiP%TUH!5wf#z
zN`8BV{IMS}+x3q7)MEu_{haSoFO44edQGG2&$d8Cavrn$rU_tRrGa~r>K@FQY4?s3
zVS&#~Z$2NhuZOLy8EX8wvG9-OUDMzPXu;hAq_G+~<=sYv6nooHdJ{(DfP5X_LGH-0o%kR|uVtk_D
zln)5@>U|fSWMFJ%{4xdZS3d2!=ZOo>aU}WLnIqP)mWRQD9hWgaaqV5unn@j=>3g5v
zIUyQXw))i>bln-Mng%JV~(Hk2c$+!?Cbqv4Ce_r5Vos
z_D|)f{D&}kfuUga>kIIj_v(D1M-g1I^VlDArWbHJ&A`#|0Bw9x{qbi)9dAa0`
z*G;_V+qF3u=Y-u)-eloodTgboh_>{b@oIRPg*Foo?>S!F
zX8c-GMMMpDms6o(^?eU?{8AQFsrV9#wK!a@{;~_|d=Mu5&TKF2Pj>Xfz0G}+l?AV&
zliYQf29a@3o-qqs5j6>U-(7~~MXP7L5Fo=Tvp-mVpKitmUh@P*s-|J*Cy{osT?N=B
zMTeH0-;R*&@z>m6^~hir&MCpfyFCzZ=b4qIOSRD9<~B|pPXY>Tlb~y!4uTTI><)i?
zWry7BSRe8{!0~Xq+E_y-T9YOcWtO=-j
zswmT`z7Bd{mmU+hxzF8^Nxk*bgEdf$Meo;>-fq~&uIQ=zw@a`iht?}}nToMp{J*YN
zxRqj;dj3>+*ydqe-zt~}9PdM1*P2mBJQzDSemwFJcM|q%;9_2He;Q`>uxpnQUm&Kc
zALkgCmkIICGCAxQX@TIqelH)cv_gGFKmYLBwLtIucMWQ!G(eLNchSAlbAu>7Rqm-X
zWkQN|cLhY|iXbCFHNGh62F%Lrq4>!INAXixY||>gs-UzI=;FQMN+{?<+wD-<3P}5j
z3>VIk4J9e@M;;Dc#J(Qz(=Q87!Hk@4*u40->4P
zA2RZY9hPB?!o2LtTgP!
z2c15QH$Zbd7gBWR2QV@tE}gw%E0E|CP_|mzJYS_K@cJ?7e(3F={nNj6nlUwL@f9|O
zx6qVe{8zSzJ8*Wp>$C9GMZ84H;n(7X8~pAtlVi7NAbyta#7!~gD7dTd)P%Wx6@GX5
zfM{bBSW*p4KcMYAB=X!Y__d@MG?^sI17wUxam8Ndu)w4ANJ7}$7
zTG|Ku&m72tA??J_u^05jl3Cd2A2J6JDaGoe)x3-ZhqfKlQAR-|nvUDMz3R^9O4lKC
zt~D1o9;(wVqwR$mKJ~ZJKAy&Rdr#U>f<4H`)3fKNzltHS4QsmCfm~Rvc=_1~MFpO0
zbGh^GYCYWhwCZZr*I7Jg->jqa=67Ue<>gOZk#s!%$L_BaZUu1AhSHJ1sVclKrTG3&
zn{jxb2`7FKP#{(}{GC5Nsemh{0)Bpxsld;hzMe0Lt%G6yeeT>Zrg64HmqJC=Iox}P
zpx}y=6GG23EUD?|N+?)w+-y%bB{V-8dC6T3Bj}_q+-yGEfVjO=ruMmW1=$<7!*0k$
ziqO^Y|B?%{A|4g7-4eGf2ss|APM_VIkIHka54#-&32miYk?ckSh_Cz8npf++_|){z
z@czu3$kAGvNAcd8gr}0*MJ~1D$j2OXsqN-*!t_u3vCk@ApU%pXu8gKA{*;KMBAwp5q%$M&@r$RnEu(}ociV#f@y}xoS;ZU8lt-{t-Z=Y
z_}>QBIYLzkm(__3`du}M^mikRvy*S}!bKl+h;a|)N
z#7)odVz`SM*Fkt(7n{Wfj$yWPl{S$Vx8#7~)iXB2ZSQr@-7
z`sh;?Qgo>28JlbxA}KM}Y-R0>r1;gdD-iF9gd909FgJz7g>P;rVY~-zsG%*lZDK&b
zR_HuEpLP(v5ttJowOUFd(tME|?rX~W1mi{>OrEj
znixq`GYC^5h?ex16ML<~kr;-Qob`2VjJ7Al7kT$3
z9n~Q!<(HFs3zLXY-PzX8RW0N~gQ}7Ys|>Oey)}|^&jR7G
z+G&GX^%Bi0bmEIYDv04H^bt`7-iT9}a-r7zZG>fyO}}uZ2_pDWaJ7d%ix?FRyI%Oq
zfk>^}ttER$7h!kRJB0*G5pkZ%0Vn0^ITs*x>vn3CY$RqApaeg~hwG69(BhVUitI{`-<
z8esk3?1
zIdzY
zii<7%9yzUS00#!;=mlR9#~WPF8kp1~=b|D?6aw{6jGI
z<5@kcE?p{gyiGN0)nYen#*5~qDUCto3d8#K+BH!Az)c+*?=GnR?6@3qBn^6^z2hY^
z^9^DZW~b`3N`(ekOlY!>MPPv%&3fk+BCyiNMV9o}~&zl=`wS1LV`RD4s
zsaP-U<&e*-tO*w=p5=Z2kxOov<%b|zDG^WT;fs$K!*)eu;atOxs$(Wlw2MsAqG~&I
z`lzgqO6hAzK2P;i!N+Fk70c$=3$sm-g>fFgduARqgde=PFQx+PR^a{ANl}aGPTW=t
z*;j*QI}|p4Uk}BSujR$|+-`!)%!;b8FITp_xYNuR)Dn(8MZ3i1(K+jIMz-w#4Hk
zUi+rw`sFhiR5irIAb+R|67voEIS^e6)wrv0DqE&(zDLe~_HzG$L8759ht!g>=w0Vx
z>9t>C_1Q-YSUJ5hIHn)#JravOk+c
zOh7Me3jcf`{Ea!J`3>wflxKciZ=_OmgecrNg{|
zYwjp4?p|zmf3+5B90tpG-n`Zs1wAJ-hK&@iV$wpDyEK-VD2=bAn(H
zXF$`-9*Zv~$iLt_mOS
z2cD$6m49D362EiRfJXIFFYZfM!P8qJ0gGyBb?@E~hd-zv%o)&-#~<7Q)(=!{;VYDL
z?DF4>;c4T*&VHdN-iFb;0vla8mY1cYs1js?ZX8RA*67Lx6K%!{44F&lNRCyW(2_JYMc
zesToA$KlY(NF{-E2dyU(XXOa~=}!;tfli#0^Yzu0%at&h)qy^NcfB}M_ZVfG{~jdl
z0aXm2$q}TH{Y`n|Hv-P4r)AH2SdPmamRYE}S_}^smNN5xo5$yz_OG(%zQbKD=owzx
zm*EfhTQ1(EDT4oyGwc3j&A_Gnri`p_O~J#d5~Dg#o8jCn-(Iw_7WON6njFJii$i1!
z61Rm?VB_?M;d_i{@H4GeKMhO1!yRSUVs@}gA>@bon34`WBW$O|>7$2jh{`|o
z8bmI-S4ezw9r%M|+Bi^?-0KykMHrDns0ezXy#Z
zo*@!{vn9pI_#pB3(qk`Qa3_v+1$tP9<`8;%Vd}_OIq?#K)$tqU#JZ8-y14C4MDERk
z!^%Y|f{X_^>0}ub^UQUkH((n=O}w0=c-4$}*MHRKikuiB!!NECTquN8^x{`Omfk^*
z&K1Ap8P-HfQwL)^4?jXoOas_^)lVbD>SR;891qduH&UD`s6e=AC_0_`ewT>rP^WYd
zS0!%jjyyc$AWYCUy`4?N1rb`W8;7*=6%qFWMpG*Lhe%FX8u@I|LnPQ;%5(f&C7~Zk
z-D`F~m}vTL0TNqmkS_8LLzZT3MA0VBrR9qy5|XjwbJ-0G;_F?TrSr1#$QdxRTcS!1
zsoc@#x%5Z~$!=M651mDksv_4PBPE*1=C7Cb_V@UZocKm(%4-6|?@q&>?mKh@B=PC|
zxG^VjGuQuth37qjjq82J(=Vq8Go`iTlA~Nm#i}L~PXq;G+v?CplgEUd_Kds3XQYYP
z$Xa>?F`|fAjKl*!PcEWj4{Kun^d3TRpm$K=I5R<}U%9qEZ;jAXH7Hh-DI=pMDLfqO
zXNb72ea1oYo9F(0h?;H&w1h(MbWw@hMZ}_Y-0_^nZiLPIwrb)02ELYNc+6RR7Isk7
z4b?gM7Qf6DxA^VkaU$^uKjVw)6`Y0_gz5jDg5g79@7rG2>sm>PG!Z~l)A4!C;
zB7}yBp_;%qm@Sp^S|L1uH_pi-rOXxZ5Nam(v#b>N4GZso03RWKe3tLxF8qyi&lE*)
zGY`WmbaR>+{Dt_#aEOxiRRO$@$K%7pkwe7pV8bbw?@M_27mn{_6P@t6Ug#L*N-fU*
zR-3=^O&&by0pEN7Vh-mYPzk$|oQh*UVIxCgF>u^#?Q?UBLHMMZ*xT+4f$&S&>Dx9~
zGkon<;RfsFNVxI0&$Yv5&+(*;*kau*3%Egad1AcM0MCC=8@bnvz@_(pY;<}bgb&Cy
zs^W}Ra0->!JL)s~xQ@)hggwsc@QYQ?Pbbm=T;Rgj*!Ut2c6(Ckaa;W<&g+LZHt#lq
zpO$ABH;5|Y`>fN)4hfgTRF+rn32g4md+k&vj}@gc&a3F#}uP?7sZ9(L{_GJ&+`m;-*J(+
z6;Vo9M6h36MV1^tVGvRAkwJ6w`-A0^6U-9$``z8;%omyAvkEF#qmJ#zUC^qOP|h8&
z>pS6N)E`vve9@OC2i8vDRW&iC>Pfp{wld1w-||;5y`MEceYvZU#O*rm1ua3`RUlMW
zxT^=-Ji$Tx%K9)QJ^K9px_K%V+}rZN|H~jG;$lVJ8c7Y8l=vi`(rJS-i>a>%X2e
z@j%iPk;P#1cLY;Lz8(r>5HcIi_p?rjz`AACZ8E7Upr176{GJqLSiQO{*~pI{(27*j
z)C<-gC^bBxSXQwEDyM^DFAsM?zZ(vWE1ESzN)Ew4RGxN1a;=|zN7XB#4t)cU@Z>Ux
zRd+CXBE1}PY9^QX+Ef4yDU_%tp|X8#?XqC)IafKh0k
zXgRj_O7rF2tY9ojx~eh!xFdAuht=gO`)EkAw>{RuCjfdfWojB1o&-t$%IKD>jK(C%
zYU!@sQ}q$ye#S3ZUeJ#|sBDYM`8Ds{5ShORy|z0S!m0
ze9ZG$nEYe)`_Qn-Va|_dy)nt$q|a-M#!0weDNgF3@IK8
zu`7I22L0|#>OsAWF{#G5Qwbh-p(Am1&U2eMke|**u{R8&32MLXd(Z?jT-oVMJLd~o
zo*Z^9KUaiNzg0OP5>$cpsSRqJd{u%m$`2$-DkWkDY-jhGekg>@IPPFBPXW{`80*PQ
z*8sh6;SqD%-3Y;M_*dt`D#$1AQSF#T8WgB+bn;?&8+5RQjn(d88$_*V9;eUL2KDZD
zq5OHO1-iizci=X8z~*{+=Si}$X2?2b_sKNQR_J_oiZ}z(0+m^ZH}B(Xg`^bt_`Chi
zV?yLZ2hZnOVS6r?-r~O$53N$WrOZ(}K*RfA($j0EU>ki0!Xq8MFxM~>yD`}sj7kP~
zlnlTy^DFl_{pJ9sla_EwXm<_95L*0ZFh>i*d-onK5%huBW*8GwykBCH0$#1xT!OKZ
zqgn>kcOx-n+q=T9Lv7H)I@f-8o(AZ+t`>gvNh6eeOt8s$tOk;o)9%ENZ=T0_@on(U
z@m47B>`CqO(rr+K*615fwkF8&I^*x;&G&*PVMW^)pan`oqOOe6=5GG~ex||){!)y~
za`t+PQs4}it6~~+&^6i!iufz^^iJjre+B`FUqo_G{A`|O974kElycqjot$p<1
z`7$i=V#@oGUHRCIk9x^X|0mcdvmQ<6_zH|{`(OE0)uy$!5NuBRqj))40ml7A{6>3a
zCDi|z-&4i50lIyynfulS45C_Jec=4Q22ve4Umri-4$W)q62cdnp<|y~Xj%JPp-1_v
zVXplx(Di{{&N=61XxEWimHLEMXy4R2+n%LjEF_>UP9v`hbF)~FPR-23=4jbGsBCgE
z&)oxTfuZ?W^sQURu`@ZCM!e|=vaU?^`=tuUMH|Swhhl?
zI7=|gjHgwH-OI7{ba`6t_;SoTirem_PYKo&#OzqgSB06<4!Gus*h6V^H!kl%iZBh%
zdG^EaN+CVh^0}_B?=aJezI*3NPU4O<>4}BGq%yXZqhnm>pZG;9&
zq1N^!i5B*^7-OQUDyw1|_Ij;0!_OfZ+B}{x@mFRRbaQ5Re(tecC_m1v`S|s2NN6Oz
zx!_|pmV5N)&cKFx?4c$(Qp;M0**_I1c|~1^VF%;5R3v&a?d77M>lD@4JDGtjo`coc
znoPtSN&ix8{j;Fj^3ej!b6v_ombnd@rC!++vA_=-MqmA`F1QoE>{=+6bd(eC>0itV
zUQ>p7u4YNb86?0dwtB+so_pY_8~s%g9@Cghht%wj=6)!#de5tTa!&Y3zD-a~`XhMB
zKe1+}@-5b^Z8CV|Pa1S#jz2i!Mj=KfpDTIYU?=|Zw1e*N*2{RiITdA3crO&xr13gw
z^LxF#3(=`f)G3(an_pLbrHiial`HNzk7?8fCZe|}{7
zI)-0+?({{>@IKz~sNi$VbA5O;$8%v$p^B{i!0#P&Xx=$XgN#+)-KFV6i}V@$l?(yL&3UANqc?d`IKv
z-&=iOx@h+bUJdZqw$F@#y_XTk-Diq%ZI;IZ-2w%;2rBS)Pj4Hpkuu6i)>sF-wHC>(
z#7^T2rf<7_<8Bbi_tt{cnQss>BKtxVGbM=l61wF5*G~|YRMT%?x=|rH-Y4h!xRD)kT
z&fWa62Zwiu2R)c`UxM#)cI&K5euvFg&lbk^Pr+BGm6ivKT4DXg({t
z9Eg&w!AzNl#gi6#E8CzRiYa1
zW%V`U(wrWjO4b(e-_jd*h68`%=MS2(Xj1&d%^sfb=~B6k_|l;U!4zCb>9riWwNEUB
zIgO@HdodYtxsr8qAd8w1{#YUdU<@94nHKeHeHs58z^whadldgI=Ax+%72@ZfT^D^e
z*@`#2D
zuGVt>qX7C~u`>I~b2jji!(r;O!GVn-7flkPN71s-!eid)ihPy2K(lrg#
z!cO(Qfd}kkV2T*ICl7dw;Pj=zQIJDRlt<>jlp}wKmt=^ZGl*@4X9LVl9z^EjVLCQ&L`Vv!T6Uo}nW?X>qgA!sIDzEOp;9CaoA&
zvADH=sM-W()wp|UtlJC^oSd3C`QjmbQ8QnzK~EKz2!F}D%PJc`-1GV~zn>Xyq$k?fT0}A23o!nJ!-Og~XQD|Gyj5bcKp%O$ta~Hl!=PLGK;5L5O_}(=R
z6&v`}eBK$~3Q_z!Uya4ti4(ZHMNC3J8z;;gs}5L}PT)0ahE;Yo_i)$x4p@nC4}Ng$
za!mIk70kIaF88!OIbJjE3G;m21&^fmFxc1Lf^TP7GhP_n2Zu@Gf(}CDcvzOxrFx?^
z=-QXcH`!EcSh%p~Y264_93L>utJ6P)|9qNulY@sArj+sfl=)#7J|(H)v#7TQ#l(l+
zJ^V}w7cYKlkV(#t-|-K4jhF0z3Eg)0Ta+tU*8I)|b=PI+8oj?n&j(Jts(M91NcJ^!
z^XxhQXI};&nZiaKW+fx;Nfa1lG_P0F8%7dux+FbF(f0ZAPWF>yv}F=f+W$JqUeSc0tMjpGuv*l&@3@?YNHAv&$SS2WAZ
zA>p!WT90$#m?QhE{QJxa(EEdlvvj?FSh9_ZmIB@lxvLD_AyOMRk71O`Ww})c!5W;S
zT0J$8&i>O*#twB5F7R~e7SxEPH&$^)d6!`6n)<>}a5XkgADVvjX9Z@|ns9_7xe!zQ
zU~+MK$scMA`*b2cAQ5{#+Og0Skb>n;_3%aWx?y?Z9JQyOW?}|PoL{f$S7JNj=(HAE
z;-DQUQ+e*O1gLF#XgTGz5MZeFD8=6WYbi8jm03t}I3`4axy4dM`$
zsJ;cbNJJq{T4gFw`ww^4jgI>P6IOrIN
zL#2Act$-{UJAO=HoZ@i=w0N{F;M~bHsAzwaVgFnPez(H26xM6bV`V~m4t
zHT@Y*2ug(D9YR@g{_fZTx0KiQn^;&VvX{Zr9^z)1xvpVVhN+KkZfkE@jPX+FU!E>0
z#a`z1j7Vx^V1=jWB;}K$FslImvEAIwm^jOFOPwE%$?w;^{-PDffPD1f;#eHp`0d5Y
z{-_kQJ4$m##91HZ{`~R6kl+=RrSFaJimxzu^h{WQI}w;r9;?GYf%Q(NYGbtdpqC*q``coL(rjP;#g&+7h)ZalLf?
zj|?FAKc%jzT|iyA7)qUsguz?2jDU02O&~2mv5|Ks7kq!VIO2i0qIGXRX*G>o1JC;$
z&!1{qqFKIXI>&kXz!(q9!pi;%V3gSFWx!>Jjve>lZEkuDSZa=rRYho_fA*Whwh=L4
zt-_&=y|ZZZSI!ehcByNDanoUkDYb`aHEp>~%P%W%Z0g9O^7&TuTdivIVsaY#Eq>2T
z_Ou;XoEj?)giO$YDmA|DXAi(xHe;1>u257tacFMe6=$@`a;MFYBPxLVn{Z{)nmpRm
z#dAU`_a@+%&7=(0bOAcPS8hH=ErINxMXpk1S@gyiv62{6614oPI^^Jc2~`M~KmIMm
z1HDakWN+?MOZ57Q@3+s=-vBGZU!T=aTtbT}$*vuEBnN&oYShp7kX!y^)~d@9Q1KZj(L^204w;x-5;TrAZa1^{rTpn;B{vBI(YCLjD)=?KGkmp
zKq-e}C!~uCM!DQZ)O^tU987cy{gLQrtDJp3cT!LAwg458lA$73
zZ$tHVe(Wx&tVe@d;%75%0rV#Mued9WwdiZI1^dSN@@)#7gctcvBwuzJkvaNKFTH7yMEIBGOJ)PKGw9
zZuK5H=@@(T_4#Ept1Fh9*QbB_%Av=gZJe3cZps}5e>G=jpmGDLPV2pcOV7aC2$g`D
zPXXE>?|2=L&q5i7c#|7Hyy5z0Rd;^g%_uRD*Ya^+yk~
zwT02V3q>WV-i}xD#G*PX5|X^RbNDAfV`5&G3Dwo0IXtCd-vet<`5plWmQH{^I3oQev;s$G%}!ZTQsU^%;d=+X
z1`+hiZD=FnJp%GS=~kxJ>j6EoqC%=eo8$lVKL4`QE}*75ec1wY1Q(tZd^~Mp58m^{
z(t9pg0@a=L-(4z2fH0Ny>()=UsDj@2z|U-ns3Fx7b3{NgY7Fcj7^Y^T%|UDIU1bF*
zNjZb>T4%hjMNJr{efjn<&-)_3N*57nY8lK0KYC~DD#)(bri
z_i^+Elgn~S
zz~P1;wR>Mb1K*c(KlVEmp@*3|EI@HCnwH8f#zvQnE^9G0MD3112e5P5-xmGQdU~1}
zCf-_4PJv^?ZI%gC2x!uf&YiqGyc{CdigOMPu>Mf)a7b1MT>gFQ$9tz+U@kVp
z>D>u`bh0>6Ea0S&(9p|zA;stn#e?ejKe^~g#`8j+i?ygjh@zT^CV>|D2Lu+Fm!XfnZCj4r
zEJevY+7&leN>QPtr=DLf{zNEct&|buD!Sihko*F637x*7Rd!J43J@Rf7nQQt2TIc?
zzU#Z(1>#neV+r2J(ZP6r`mPBrbkYp@{ax1*J^CX%>&6i$^c_RJNgJ0F>KJy=I9o;;
z*ellQAM*7APB*yTmCXf#-#3qqs#S%8-Y$te$E$g0l*i7FcC~PH>X@P8^~5xgs~-5d
zBC!}8YGog<$Sei!r(g-Q$`bUula%7^8<}Wo!Kw?bVFbtvv11ppECDC(s8$^qE(ZhB
zZTd=VT0kJ$JZ0Z0J#^mw>V392A5>HG=*#<`Jy5@Xy)PTIdDg6U
zH702Zsu3
zZ;G;4fI$Vno~6Djz|_Ji!{14OuFI-&3+^?56@Nu!gSG^
zBOCTorq9vmz4fPFFW8{T>z7N(aZ7Yho}v`@%S6DN9etURArjCUh28l5%^uuzNR}Dy
zw?bXNTcm0u&gf2Bx}C;F7XVZAKxoMC)DTjEUXF~dKUQ6kG2Uc4xNQjz9ztQ
z+2!Sq7mc7vu;yjKNIkGEY^P3Wr~}k@>ZHE(H-Jnk#l;g*4ZyS9xqM8a9+1iIw0qNA
z3#4;QcHjC`11@)kc+;Djpp@%8krDZJXzro+3C~~aqLjler@II50_EQ`QwNQ8fM}Ea
zSL@G_V4T}WOKZp$&}kgLm)mQF&S-0ieLifBuA7avJ>O-4eq4G~_q3x5e55L}Q+ZYb
zILY$p;ghK#OAZq;j1C5G4y1p8MLjm3>yb3K((1sB8}GOTTOAN`@O+`VL4dk_EiR+s
z<>32SDz{&S#Xy#WvfqHW7BrNb+8;6|z~m@D_G6_A1jk7dIA0}@<4!npAh;a-;k$X|
zje|B&J0q0VoS+Jt>s~t1n8<-k2OyOY?lS=29o{-KC5%QrRtqo}Hb=i@-g-n|{umuP
zRV*1FtciMB$t*6=3!xtDjcqY?GC*m+#D2y`chFaMMdW~~8!$LXCAO3NDTv~I^ZZl-
z48-@lcW*dppwD`i(EE%AmSEk;>GYA}*72nVyQvkA6(jV7-MA5~SybLlac~nP>
z^YfvX_tEs|U3U$%^wEsd&1@nU%~3&tI~>KTvVdabjju_i8VFsWes)CC6!6m>EWG;5
z0vNQM`j!917<4bF>**J3pdqp5s>%)KXd6XTjaIBPdiIpxt`h}r=t+LZ3&(dlqpX56
zo!O8*%Jbk?V#U4)R5&()GJ!e^wO2krAt+porvAyAQK2kDW$7FPN}O+kOAQLKQTP2(
zyPmt!*KeewPhKpF2lEu8Crv$`y3v%O*AI!YS{6M(;VE(|8TN4&jm0iG7B)2xC0w|lalLso5!UL
zDW?VXJE2#(wT=NkU(}UqSE$}bB8q5Vzj9Ql7(F1Fm%CfE3YB<2eMozM8T!yU=$$QF
z6>3{f=@OPxgZ2sZCfr@GL&t`)4bSt|qrk^bc~!6sO}`kDlwn_nvPs?ev39*2jf{vt
zpG1bC!6O?NqTT>hcu7Kxsy76f&2bBky$J%mxT{ABqcdoowA2undV-!kAL5vy`~o$-
zVkwt=?hz_K@p^=5{uxMFn8{U}4gsDqr#qs!LIM8|9-5|~zJOchnw%=AM4k7k87EUz
zpbrn`(O-d!QO1Rnp5)G%=-yekiVHf4XvV1%u*m&-^yb6-#!2r6G&>CvjxuaSb&A*iH)%6gdKj}|#(y7#-Un(y%-E^r%>(4E{;FYdJJEyzd6?WF4
zDKh7jSZ2LY@T=Zc{fjkP_i*uo4A4VQKd=v$HrhPPS8eTVv|2QnmCn*OJmL@5hw`la
zb$kF%feSn5l|ax@#^DsH5eWR}jRhlO-9YcS0&PWOBDw>vCbxK%imKOH^^H3gpzN%f
z+^o!H=YRmgwEcBQV0Co^+B^5@JwrevdRCd?nnBTK{*2rU2ZS0>Kl0-`GLvNsX}<
z4`9KwpfkMgg))@S2JIaWK<#I=yPFqZpvhHFs=nzrE-lcW2PjdS
z}I4c1RYX|s_%abpw{ww_V;;Cp??x|CGprgS@+9UYy
zfbsid>vxrJ0y7sm8>V<6Fk;RX*4xa4+8N&rzh|NX>XnCUmLk=GT0PVdR-=xB5C-oc
zFI7ObOgrlzqy;WL80C=yY5qxw-pv|;af{V8pNS-41a1N>hd&~zP(5KXZ
zZ_@BUd!#w|vCk3qe(nNHs?H?`eAs+0107px!VUZ)b76XD6^vf23v!Tnbrqz@GGB!6
zTBC<39G3m2g3vFIlUA)4qfw``M6URWP!y13x6$
z6g4=?N;tpSyNVFJUO3$nR*!{6DAl|UitZp9*}89Ksvef;-F
zSCAz(!@_&W1eiXIyR+z}0`}}9bC_SU0_-s@+!9SuXuSJvzaEJobWHkO_k6MwDioqx
zM5AqiDjk2VMSs#9^}ZFOn%tEOX7&dyKQT=M9V(@hU;Cp$9mO%(umpcldg|z4X01Db
zKa5gUR@H#hH)*E=#%n+)n@nEd;~F6HaSG!10l;tiP1(D80(`QunP(fS1Tzx{kBPce
zfFFF>{$+g?VC_SDLb6vCP<9PwkICYFc3;OevQU4=QvO8=poR9%b4WFHH
zty^ep=%3@y>=B?J8gCP2ZiH&_kcZ{kYoK);`ZsB{XhGpjnQoS@4ru#+FXG6wF?beS
zrTC-O4Jh80DBPK63z}FDjD7s50M2>%eK;WV7!5KN(QNwi1f}>aWFIs^08Q$K@Q&X&
zknGi((4)h^hwk3gOsy)A(Cf3VBvc8G^HTjLdsYp&4_sCGQ-XoP=gnffxof~J;j55)
zNG(`7;LD$pR0ooMbI$(sH3Gp{`vlD&7qBZRi6e$O2=vO#=wkCeU_3Y8*`UB5O&mF+
z>?WiR$fIwP8;Y5t)RrOCb7?*(O(cClf_nt|iAQnDdT$77lzzW>T=y~RO6|z@RQ&<4
zMdaNRm;wP)#PhV<E)e_^$8k{cL39~?NxAAuIS&lQdi`l8P_
zFmPej5*2p77hPj}5FHwsRo$7rIsOKq_V_cIfGmgIFCixp)CnXn>huQ!)}P+&dyFi=
zk%V-Pv06uzGqRdWI==?6ACT;u@TdXtSA|(QJ`=#NDsichxdtdcdbsm_X+2=3Zptd&
zg8@0~IIn}hE5LXjt2WuAYLMjqAZ4Ju4(PFGM^e)CgL0G6GLz9+wB&za@wR@`>8~Hj
z|6eTqKlCHT`7gf0fA~mwklK^tk@Q8|{v`QdB()zeuW&gd~5GOhQsSQXT&lmy|at&&uui
zB>lhmUwkB<{B559nJS2?Y=J~fjN%r7v
zd#B;neM_pBbR9_bd2jngZ~KA&k>BvPUeRoej}#x@)|2Y;+?L-PZ23``Zeh%}{?#_B
zZ|k+}w(Qi~yt&)@OGaCIO_wdauQj&W6|%YjYJpLD*Mu-36c*d6eXJ5}>xsmJ5U!**_#Ya?Z@i>@m
z`OhDo{cE?+pZbeza$D_lthVedt`mRtEMl#HvB0+DFaBxk|BK$(7C-l;&cE`Z>{FXV
zXBK^Fq4XDRrsDqU^RFEKE5jnze`WW|+nYmY7M)W6zo(2@bcfwm9zv_rf9*0=TXkp4
zZ(+pEEu@a$%10`As~ZS)rJ+vY7j__O6d>%D~^mQ?@d
zWtEcl*WdDB?_U(#j$@Rx6;G61ar4%jMUUqvZQlB`s5?vTUq5aBtvKGx{C|1wR&V8T
z;mHD+8-HxxirN6X24qk4LAC2wt`CMl!Z?eEGzn{um7`i>)zi#^l
zx@}z#|K=^;W0PBc8XQ~bvpsHo^0(v#$*nqTFHiohL>75DMBd|>R*|JOWQK(c}3Fy
zi!R%Frf&OV+emtTlID+coh|;5ZGW2WI-hU*5B}4R^!)!{_-`Kn%B1@KpZMGS|N8x(
z%A`Cdws}bWBs~d9{r>NKO`4xbNb)D~{I5Lz-{w!^|6hD0J!w4r%TLOOWGC^F^dx_h
zOtO=B|Nll(Ka<8ANhak_x*nuF|Fx6+NOh6&C)G!i|G!33{{Jg3Nl!x3`1uz}^^)=>
z`IBT4lJq1b>HkHNozy=h`CmVhorEMmQosL;q&!J}BqX&b)kV_(8}DEFzx+tgITHWB
z?f;cYe*Yq=4$?Rx)$y-P(vz+$iT}Uy{4YDH4w4@!KT=+#ek18g`dtOi9&ZBob&f#qy!JS*{*Hhc`Y4*1N%WZ#Ur7gaouUqR|
zp>4hDc3egME&G+QE&N@+#V2`tOBUPi&*o=aa%Jn*ddVnwYrXyL#g?9pcT4Y+veo{T
z-PZazCdqTND`wF{FlYZlw<2S~u!=GyUaR_`i62
z^QauZ=zTas6rxZGAtY0hIXvc<Kt0gpeXd5<&#
z>9>b-|M6?B&-#AX`@ZY_$7P?h&p!J;_c>=@XL#=VPJyA6FB_p=MSANlD^q&wuU|=C
zzV9pX_OH+EpeyCes-0(1`h4Rtq^J4PisX0GI*?xeGA9`7`Lfcz7VKxyO3J?!?L~5t
zoj!E2eA%A9*GNAux(DS?t@ERNW#44z^7%3=jov`H@@3l_89sBavAn|t#GS`m1NF+64ZZCZfmr)0ISJ8f{xEIC;9GT`j)1POeV}XR
z%PKuAsJ>nqwdBW!Q}+q|OH|1prLSA49+kDaRKIBy>rE(U9^Lfq9?}mF3WXt;FLQkM
z3#Pz)nabh)l%8b9>xman7m&Rz4jn=~h
zK2_s&Yruv#{2Vo`mnr5UMKQm<6$qV^^rZRK&XwgnF)z*)`4s}CdG($R>!pc#ZK2Sg
zCiIkrT%OB%r^I?lU%dAwqWvO7dEMdMF5ADdzTBL#v1rE)OIViHJCeUr9FgKU3@&;e
zPk~b0m*O&X3d)C-#0ez540(WvM)LcCsVM^-=OeikDJ6{VzSqPpRC0q2!NbN0KF!
z%1QEn=}GNTUtcOGp`<6FWKU`@38nt?KiQMYOZMx__0#_=U(%ECf2;Rj?OZ=ydS2;y
z>Z8=Z{yzsJe~Xg8|J5$iIFtO8^dywZ
zODOpx@m^`%N$HZFR9-?UUqY$ee?7OwrV7
z_rad|&eVl0Cog977JmH|ygh0y$NMh5x&HZT9QXHJ=eT&FJM(14VGiUs{8EL`nqELV|hPXamgmiPpD)&bDD6z@f+p^vQXyVFAL_ttIe1Fh+x0gw_$sS7ch^^8k<6P
zK74IX*x?rM&!2k5`|UNg%h{m-ff7yzPiv
ztL}rJ{@cF0*@0}Bc7DuCz#hT&{ZP+8wY$+eK=5%XoklM|?a|4^Mlc7obxQn_NS
zpVWU&C4cRXdb8d;opIqHKK5)z2RNLA`}e
z2g(o+�jOQ+DeI>Ce8vek2v}_|&Q7c_QLDkLPDuy5z6Qk8;wt>%!}(#Ev}w)Xn4Z
z-OFY&<=^iaO_*_+*VDTiu>R2@Y`0BkI~u2Tr+vuoc+~*XuL^8PvZE*a`PnU_
zIv)KtZaSQvbE@wZ(#u=V<2E~({j>~ZKgNoA+Dq)O)QELQZ#h5T(=wj7e0^B|r0{>D
zSVx!`^8BPdf&H}`#CmTJFy_n^c=i_gtRz2PFirLg0=j_a`!^Q>xXp7#?(d0UbHNbD~)J;M5H@AEurDDF4N
z1@HfGkn^?gupLKV-cKLk$Mw5R;q=D?xxC6JzTZ&SocFyuiTfX``mmmf8S5$TVcsrN
zJwomK_af)7FJikpe{=jSwP)Vm$bU7w+I*Rp1?Su7a@=ff&T&zGg?W4U9_HO8b6D@Z
z0q>j6b9h4aOqel@>@@518eTo{DBYu!K3z7B(w7eCPjZ98*2H__HfzF_r+isTX)8)E
zPj)B0H;rqdO9G#aXa`*k_-E#3;@8oArc-|59^OZ`)oVojDmpq7x>&xfpj8W4qY=}*6F~K!-$HLnj0bxZgZ0O_uH9Y
zrlQ>S)5mX!yRUs9K9_fH8eEacm)-93avP$oSKC~)o8$0%KI-XA?@s&v+jq?!hO+0v
zS^Y5IQvFy4>Sy*K!d@B}#`>a+6^K=l%ht8T!D6HhhXxdPo%F36OZ5%c@EF13Qe#wn3n0)8tkYV+Gd^udC7x!r|`LtUL0kF&vP-nru|~TA#GK4w}ZXC
z#=oAbfLt`lAk`jF=lG~r_c2JFklrrk${h5=
zZ`Qo`+JGSwBVFD9<;$!uhg{$P3EE-H#*n4C(C&v0{(TPRTj}304FJ6}T5fjh0UvEL
zQ9!?9TcwW1exOm+6t>D?9d-Hsag?i@zK_HAUvs3*Axw|E({eMW_xjsy24btAzO-)D
zRhdoe8=r6M>+rq#bbL7#^MLi5p+Ar{YiObA$Byo>w5uYQ*)Q{K+apwd^?K3CHPo>rzr}B<+
z+=^_$_ASD=eCw*^q*vJGDxo}&&-JQOd3~Ah$a+d5?nerg?wh#@zVJ|>kGQvLB=|>j
zk=|AC2y-E42|t4bPms>9ss+FC5c*%&bN%t+T**i5*X0V{5G{B?RttWAHj1pTZOZN7
zC;TxN`+4c2et0?byb>>hPEE3e(t1ylC4EpLJxLxZ{De+G>Cg!Xq4N;JKZI`uzd0vl
z>HO-lkP`*!i~cb{v_l7B-%;dO2$cMm>iaL0{QRF#@>9b9(uYAz&+}jTQaS{4(gSiq
zSl_NBOMYJ$di7D#Sm5
zQo6*$rE-!UOm3tv`6u-cNtWLKC-J@|-Y(@!{YB#Go?;vZ2)j=O`USa
z5aTsm?5A(c=J6UX)&;%AIrS7D_Gfz`+eua9donQ-S$}#kr%yc3@~?$FUo}{Kn%dbp
zmeYM>SkKUbxwl;61)Pf9lCNzs)<|
zr)wU<`)W7ujUju@zw!EE**LyGzT@H}N`L0tpX?-0(j(rkcwil+-+9lxecQ9aBqv-C
zgCUbIJHK@+#oNfscPV{DEXOm$1+PhV4GkvDTA&3}2;76O;CRziYdfWHH&{>gRlR%$
zT?qJ({&3g#D9jI^8XAqftz&>{C8n>K+=%=L
z?sh%{WwSTWUL($#>pLHLqU(Ye#QzVz)gn9pY#-WVe)u_`zsM7=`?Y_HcH16bQ;6yA
zmqL~i8c!L8`S)~>(0+MFHTmbDZ>@cwf9CBpQ)_Z`33RyRX0bevftv+Zs$icl-%+RjOAMct#
zz}M}+v{we~n6HqBy!53}ZoqVqJxz-`V7+5+UOqGt(4gjl^?bmJv5`;pfDb>Z&3biTt#-XU?b%58%0dTv}Vd
zhV|T`X6YY+kDnbqcb69AKij{2Ev_F@;%=-B35E4R=;P(T??U@%4103_Bdjyj{30~L
z{AXD@Uxt_fqhByd3fVkk8a#2Wzge!FAbO)4|2qE?ytz
z(Y}AQ>MY{>6~7x$KTkd!{|o)JEFX3spWjByG$)2+bE-`m1Ri
zml|nNdGq{{)tGOkFW{$WkZ0BTnr1up1nQjtlL6lrQ<8F7!?a{x(p^>cU^?y();z!@>4YZX&z0PHyoBe`LZjjqiQg&%-K7Pcw*rB2^7CH*91{qQnuLE2lZwi
zzPJbZSNqM(Gkx=OVMqW^h_WMInjyb(66GD*PqHzN|9+xG&vSg^ahSqjzx4BYvj4R0
z3KNuHXIjG1L%vMA+hMq33URA+Yyjf*9t$cEJH7nW1N~a}V=LXy@R47gjp<7h)M0t^?XO?`n?eJ{NmVL_a3v?plZYtTwMpdz}INbFNCE3B>tJ
z!wL$KXRJT`{tiA@a7rBQ?;cIvvJrV@mpRWeoq)F)?Y6JM_h`U*YyIPa@e$C
z_U300*X{paocAbS*6*|5+Bo2k-hQ6<(xLqH;r~8bfPA*Sa=!+^A2ntN8b25*m;LI!
z`tCQ7hnN~X>j(R+YGt2}bOn3UkB&6&1<%>sPq}RB^SC+vp!~3wV+%4N|K#Qaae-(uXf@3nk-~K~CyKBj12RBVg-yQ_dttsoQ2LAO>nKAu)KJe}7Z3-`e{&+vV
z_y$pO*+Y}Y&gzhFYImyhR^YX3ep;{p{3Ku2>`Uj)H=n{fsOnm&Yq(ry{P^zf#>?e0
zYnN*wrO+;Zn;L!9_XWSIR<;+<_4^Xqr`Mf`%%L~n^N8!0`O)j~#L3<|}
zpG3aC!F$NY^?>R(^2?ClUXXt@)PjDo^ULCZg|OdVFh%cS1JJit7_D0Z<>lFbVtRtC
zdrRraSU0e*b5ln<9q!3YKAe^fyuIO=xoOtW9`8zD=2n7UN@BM_w6EtK?S
z$JOd%J!e+I`&y+i`&kyqvQ9O<%>c(WjH*T6J~?k#Jf3gdvpr7dM%6~^Y2P^Y+isc%
zJ*rpGyqGg=E#b1#6_oC|vlg~`pq=vf17(5p1WiSnceRTLk>1dXy+9pc{V;1^7Ut_{
z=+Zp*sKI{X?KVa&Y5uO({Ie3r`5QO>58Vg0o+TQv11dy2$ZIh2v>P3CHWBmW<8C{?gz@EUOf;
zpA`-Gxst_x^iFTicM|31_bH_MTZ-==sJG($;4sG5eHeo(;z`eDKihkMfzyWyUO%sn
zR&va+xeW}>4{?B?~}Oir77gW
zqMXDhy9oJ*SZ6Fh#P+1$yV!~Pu8Df2d%TGetQR5DcL+V{_wt9GIsZ&7;|8G@E%XKn
z`GL4UaPTSXN#6-j5&m2h?bAR!$LKQFvw6sPuN&jW_l((Rn8(>%Ormy4uHp36Kba@?
z>dXDX*^1YbV_)4My)&&i{lk0K>phj_Gf%i5w9evjxV|T+pGsmq!woF&7Wd;;EamaC
zX(6YV`mo-;=Pcho%SI^)4rMrulL5;FUQjpOF_I$N2Nac+GC)f$?un4LkujY1A&c{uJrRk#*Z;XHqT&9MC#
z&n(^IGjN}BQBKO=Ac)sGj%{4<-a>Dkx$m%kqd=$qh&#&;sUHTsnm)RB8t}cOYx*u|
zr~5NEKdg70V!K<&cdK(64ZQ>yGxL&eW6(>K88|KnyJn_)+V+NY&*X7E_duNOpET~8
za~hoE$2PWX_7&bw)6@;_pr5gG)4u&^r?9zaEZ~5h7P0xDSL5{B-Li#Trnsz0!7Shl
zcP~WQp`AJRmsvgp`!Q}F*T({{a9!K~Y4rfP?0u6PsinYk_GTwWJ%@O`ajZ@r^59*=
zJ)AEixc(<+7e%
z6Nk(90S`JDe&;>xm#WTgSbP@vi&pNOF5%#h`ADr~V^zi7L$L3e^rv>R2Jp#OqlPrP4|)mr
z=1fe5_K;`&Eoc&-FKg=5T~p^S@btSa7fb?sO~(Ja`vm-38*=(XGteKm<;16Vz@I(S
z_Nk7%1?SX%%O3oO=RTKmtot&+FD=4adxM_-`^KLSfSmUEMOGU4p}Rg{n+2fu!q#7r
zPrbKww#5FUSkS_0F6cGVU+B^jo~yueWy{HsKGSp4GvqmA4-U$}ejD`d`|t6PUzJ&F
z*A;M9ME8kike@~7a9w5AxZ6o<(3@M^bE_rDk3tmoATNv_=yPHZpqFWi65b1OsT}Ye$Ma8z
z5hJiaKKW2+kA4-t>rUqYO0muIa6kA#f*Y+PHDY$JK|QbV=;^3;;gw<*qROj;cKFB?AIfd(i0St;rtCrV
zmnXExba#&fHOM1J+AUs&@{OW=+K2A8_AdFERIN~k`o874bdKU!@P_!`lQSJ@o;1Gw
zfOx^+BcmE%d5<}#Y2Io#{wmEAQw$PF?~{c;>23S%4wN8Y*5dI{y3g%jIE(aj)eA|_
zCX?;WcTg@ty@SbpXx>ctHh}a}<`$7&Vk|%JmW9f&H3;{Y`o=sZOpKs?QN5wsH18Hi
z@;-ra{7=AYiJ3d!>@V0p7fqZ@%?Pe$$TC*+=K5UWDMkezh?T=
zNw54G$FZ2L%mY^pVm|5el=bp@j3>QW(>X3Z68oKf#rLWVDp=2@Ip4!L=*_=3`SziN
z;ztkhyYgi
z%klzK#sG27DHDA1q9NCBA-*@&X)Nna7d&dO8Ou3weEt|bmgD%Ift}`8#tK)(<6X(#sK
zI*9y5qQ3Ux-lpAR=0)x?T<;A7M*nAw`@e90b?X%JOLZvQb;x>4aL(*?NZj$9>jX-O3bf@Jd2_7S1)tApN#dqC-Xj>?rQE=i+}DRz2$oB$BS0nAG-JA
z`f8h%QhsW3Um7QoeR+Q`sAnCe$NfA{xHbAE@qyUtHpH7s=N_f>$V+@LY1?)$l0S@^
zMs@-PR|^F!JnxI#H0W47piM^X8e6c_;&Y>%6R^MK`e#zXU`RJ~xfKxsaWl@;VsH(_!)sTs
z$g+SBS=)MFl>?rx&aslizEk;+vi1(#(S+;3Ong4S*H{UfYYSIVDKlZD?YB98L3+)N6)^b^~)8B=W
z33Az`(wi&C!#>=k!jdcLljO3;dx!eAf&DC(-i3F)edMy5$+zMRU_bF&gyzycz`GV0
zk9w*P{Keo~w=)s&{JLwledz^g@$_=_eBdt!R0A*kYzRDX(e1Ad?*P9#x_rhp*mv6C
zVKb{W=tXQv+GzvNzaqIrmbz9hEBn@Npxs8f%cL;
zGQCw8E0>wtoQUtLAeU{~80C007~*&6jI}>uKlJjK2FEm^or?=vB~|Z}%SvO4zWv<|
z{<{Xee{l!S0rk$#JgGeijFlw_(4@x|}%<$B)JY
z^9ct5gH{cjx)9oVXVUPC$YZRPc0R}Z10magt)2;J-*fqgBY^Q?7xM2zzffH?JM}im
z&6b1&;&`}d*g75enYzB+_T>=ZgnmvRieVqF>4`=w0^q&B)wl2R8`9mMt<|Une<~Nb
zy;4Sh6xOvkwgvc>{(|u-a9_8yaP#G+z+=DGPBwZC?OPJwF7`g?t*==y0`HwZSv2G}
z_U9h4PJeLxtt&i0`*2tE^o>x~^;)|Gd0XEv)o`Q(<7|M>DNL`@RG@VY{I!wpD3^bm
zkc&KU@#GG#&|dM>((X8p_x_Ei{d+B)^SWf;Xw{usNPpGUX)WeYw5#lk`iDa|!x0jU
zE7Pm+2~FVJ1E$ME;5pZ|s?kgV*(T3y0rJGe+$c-rv5r1H*W-F5LsgsRxs(3(G#^g5
z>Q4O1qty?Z7YBv~5k5HS6-a)@PJ4jpGlqFZ>U#ceOkve&())DbIN9mhYru2VoBuqB
z=E2J&>S%r*{mqi}M)mpm1mAc2D~)I!YTWo9mHXA6{po$-T?&@}J$CO8QCGhw~f;&NE%TS^4y^ZJAM1JVjwL-+@jP#S
zv!0C}>m6Fadgq$oBE1KB%wLQj^ZWePkN0n9mL^j_Hpv=5@!i{q?}uG@z~_hs*}hcX
z^9=K#eVIwrPZDCe-|3y=?;Gv$_a?pVoA}&v$y+Ph*Nqs)_IwVr-gafyi@dss`pdv@
z7qU0?IomVOQ6qnBPj4f8@rs*CuSvQu*?XB^O}PJ0Jn;$Fn?=v?Jvm$|n}pbMMFCLQ
zeAzUuQDir?)ioNI<4U`wV}5FJNH4@Qx2|tNz2dXpv`=Gxr~~m5E7f+S*LuVYnx~=%
zENqGM-3R}4;#+G{r-Wj8({vT$O?r!4^+DN4dBAJbi!{|D{&RL_ck(YWstfH)$&7#M
zpq{h$B`4IIH2?fW#I&)KXCZ$mAGv8X@{_f9`i{eMnF=S1iU)v0I?0~kJmETX!&aOZ
zBUT;M#C2wQX2Nfb8~rNB^qB!T=*qMz%rOK-p(L+a26OZlSSK(2$7cfdaF6JEQ&3_+%p8QgL@&fBmbd6qw^IU#G
zt_#|$Uf^AY>-=e3Ub*4EjB~q#2`wP~yry!-GKedW_XdW%hq#l}V6t-6-kM=q-y
z=Pg?a_ImVLnxPGPu3=vjzk=Sf)bt8>=-R2jyM2PalMFKGO7-vCDbj84-gzo`UncewMus89`iKzTUWDN6^b!KHnRk
zZth{c)ehA=2d;^5nh)1I&xzdAu*Gway25
z?LoiS(*c)5ltsNdQ#0D9tb85mHWk~gX8XiSNH<+qWsJPG)yL1zkvCNC8v7FaQ=0zD
zc;q*FfBV!H16Ch8^luJeM3awI$iG8QC%nYp9qF};>4^Q*ICpoOAAp00ZLGrmA^X1?
zbno&}`|UqaFMf4`3-T#>->DsOoa8OM(i8jPl0buX*l#NjDSbiAn}1XR`H{)dYE$gT
zhfGQ*pq`fQWP8l-pWpBm;?A5;e~>4PD;x(`mq0(|IQ<@{Un@t_JDPNv=8I0+$J${2
z;O%pMB1UaVABy(QnW?yeU&?B!qYLUi-1MH#
z<)SM-6`*W7IGpY=EdHoZ@p0X?Vw&fSXB3dUqc)26hgOGep>^bp+pCDbM1@+?{O1$=
zn)JNRzaTxOJ11x!G#j8ndbOW6^1x8W%=jzOJ
zUOi(zl-P*%+RivndMa!AzRaIvtmn9e^)3!zy_e!V=ZyHyigpig(=rcRvCWF|5{m{`mH1c`H;@?4I;AI!L1?*!I6@R843yac}q68DF88nIrX5!ZXU
zF}K6zCmi>6bQm|t8Be+~kBEJqNbNABWjLX(ImgSmo!d!1c!=NIIR9bPKTQ44QhL$R
zj?~{`9M+O-@RWbQrTnFo-utbxQF4K*~qpim1AKHSwKq>vKuZI$k*w^)@db(E_o&p+;}i`}491wd3dZ
z&uQEnF8@w(f6)4e!%(lQiAMs?J1KL^C|-7{nRpZDi?-T758}Mws(x~TAK-)Nx%$^(
zo<8gJsxi)om-_hAe4vrP;s%y~+a>2H>V0q6<=zy~Ymu#!vkv5die)sfc&j|kKzkOk
z>!)Jex9(g0$qw`ewfOM`<8)<@ftncigOwAX;JoBubTJn7zA9&Zi~+q1W4rYuzy5}=
z#rgNvH2Vd}Bf6ZOHXqlOb54~knhzLYJtb%<#1Sp04lldIykGP)JO4G5U-IP2<#LFZ
z$LkW5!gb{`qc$~_ju3ytD-}ZDfS&s4<*ARMoOM|Er8tPcX494)b($iFzsvfd!2r+;
zeckUw4(KgZ`!p&G^g>G?ML2-o^>2+;TEISncINeqbdL0B;3EsrQ(f9N{|MCEq{kqe
z>2Qwox3f=5SGZ?1$FhH1A(Wrj?&yv%@b_8McJR|McyD%-FU){*kpVgv#_m^^%kCci
zn=mmMc-G$ee{SOaB4=3!?q7IY-f1-s+oRvD94oMMzj=qP?qL7@xl8vFVgDob%C8x=
zpf_eyJ`E&umE5rOu$YXYkIRa?lIu
z&~)R(1c=X*+pZrC@%-(&r2$!>r*}U!EDq|s(7o&Y#-Lx6I5N9sFr0hroZWdK-2WNz
zYRroS&@*jgT8;ZmL&9g!J+FIts|R9yQ*v9d;5(p+%ECh!R}TeER&fEe+qPR1<8b1Z
z?M2FfX*U(d)&U-gTat@$RbzGXcIJ(PMCWFF>&L}4D7#Yy&F$Kz41!D=AgY>Z}u<4
z{C7v4i1(|G>D3E)_4HF$FQQ&Iw-3`%uUvB?#r=XreVUiF?(R**cD`P`;1Jed_jvda
z)VucX`5LsR`Qy$jM2kHJ6j#;TUZ8odReyDgmp(_=(!6%iV<6cZ-Cn;4_4k_HrFnBj
zKp367WZw@ZJ=>=}NpH<)!=I?9cDhM2;x893(tBxQ;q-EiE>WC+Tvq3}Z7uGxNcU@o
z2>HG!*UN{?eHP~|ZN<3@Y?6^ZccFLHmSx*Qeh+bz_`A$=1CNm28{gLS-pX%mB7XWn
zk^MATBKTG#=1W#T|B(LD$5%){&zb#vw}^R3Crv*8sBOXRyno*b(r>Vv>pwY(@n;w#
zEKVr@U=IJDAUm7WHHI>-68E;f53t^lH6^5Pta^&#Wh29bH2$Wq;@^SW$GRF|zcuNY
zY=q-0-KP!t>CyW+^-o**bm9SPcQ2;+o?etE$9ZF2QZU)IUc>h;CTmR~+3y?Q2kEzS
z9mP=>c_pQv>&X4XQGCC+Lp<+0#26`1J-W%vORk+LBz>csj|kTVmJwFFjvyZLEAhq-
z^jCeyIQT>f%>Qmv+v7aqKC7nxF~Hd~!%Hx}UhvYhLA{fwZqfJ7+^;1OZ|7F0_M+P56J)bdSiO0-!)THy3+0AZ`!E~cOrL!>JXKeg@4&{pB@9BMiaBM>U
z7@ygEa8kMjQd2X^>f5_MWrEy~#pzb}D&T)V-zCD|_
zg?RMw^U0laK|i2p<#Ak}otdlA7Uy68L96c(PpNv6wg}|-UEc>zhPZxAbG<+QzAj_I
zBRQrodl%drskR~LFCZ+SU#-(76q1MBZwV;nNw&{^LA&}O6g
zx8q>X=ieXealkvwn>zR0zfvyCGk)ID4$eDP-QIS;?GE5~-u^!(pZF>>)Ho7R2mV&2
zC@1wUfPDeG`4?9-m&*!O|GhsqR4&^%sA7tjJ@`3MEh7}-b@U}|<&}U-%=Wt$K)hBN
zdHJD56S*v8NYizYK{j;x<+6El
znbutUn@xa6JXh-YY%A2SIe2vT8?c+AGp=xv9NupWgLS*df&Z66w|+kZe4&k%veRO?
zr`&Du_L5a{S!Lj>xm)1Aht>UkBi;ZnIIezf{J)NJnZahgHLcC%vK4o)9$NXD7j4E=Zx!1u=8DSYzf{UE-R^7Iu!C%d!N6Z3;ZrJCwQIz^OQC+snGy*@ZWrAC}41A)`oe|?z=9h%r}K|te|s&PcOj!hN7B|{0Tg_)o+>G
zOK7L5O}~D+0{QxLMj6}!d3?W`R7>t{Gf%F!yZa>@)_@&E+w+8{!
zx*T|kyru2JS24H`5FMnK>kVk=RqI{}ekQ)rZvPL;_f=oyei+h+Dve9F1ibeAc*SnO
zvrD&ci-!CghsSKj-wWD%d^N>!q217&`q^m5fa5zr_6e{rnFISJn+vagDg@c@@SX4G
zfRh#e%tgM`qrEHrF4WB{H%bHS#3;41#{Chk@}hURkMh&ys|t=sjY`vA$S*dYbCqMh
z+3zEpah}_dHG<|RD+PF@2T(SNM};u}fKchwSqrB@Hnh%of9Ok3mUd6auxc{=d+lo1;_v?0`3EJ`Q
zb9D*k2YYvThv@3uz5>?=KVE!Tjq+8e)nsSsYfs`|{yoP(#`J2vb^8&24PKOjXg%r@
z&7Z5!UZr^_%$~pZ`LAW@Hq7sI*M#Qto=y24;3xAPG@tg7^Y2uQg2MKpp5^%`v_1Mc!
z=juNG{bTR*`$>M$U6Ziig>a#M
zvo(bI=$Or1Zl(ARM~1i$uJVW9$671qvmeFp&#c8gm{x*U2JhtZ_8Dy7L-0H$(Qebl
z@7wYgv)oa7;{Z)yCs2uP-V
zm8#>+^QJ2EzFNgOKIi)Tp$C=QV!e&{md9@HFHvK-{|4l6f6hG0`+UP^S(1Gt@%Iue
zH}Z2OJZJq$?YUpaeB$r#oP5gp4G!^hO$cS#csIB6&qd7ldmk^Nc|fk{MLe-oqe}*k
z`??30v=M(d{|l5E;?rE7Cytin(RwGM{S2B<4M#3Lih6?zqShc!)Uo6H*J@?<^n9uj
z{QKskQ~7&Lt2gzh`OmyL^VEQ{Kzgn%4^2paS#1c-GqrweXnr01FoyESW|R{K>|nX?
zyb+{-Wlm89)~}{h=8rr`SK(wP#sPCb2W8x6ZfX%-brkTbiEl4&h?k2x+Koni!&&p_
zd}eHgQvt>c>#qJ97-s^m&ehrnn3K?&{JifO>4y5r-#6%BJNWFK{~G7@(M#ovF)p6m
z*KRN7|1#|S5z*_*4p&6;am|jSzLQOE1==65!7&fh4>x*E^ZU>Ze_@D%KaE^c5ew&h
zwm`o1_)h4Cbr82(xYe%s3;t%_w0|E4eCF2;7u$)zORgVW9T7HJE?akc_r5lu-|61o
zkb!`*jiuf3J6eP1y-K?Sf#sbGt}821cT
zL~YasUN^DPv}3rBuxgar`&ob;t~Hr44{&OPvJ1AO^@}x2??Bwux-e%0K9|?J%P!bn
zF>6+J!8jXr@?L}%;I>EGR^h&h$E>*)dS>9K&2oj@{&Ja)w_of>;4#mWHaxc+0DSat
zYqv~QcyHU2Hr;{y$d8uy`xjOL=L1t5Z~d)<_J`0%;YEV*3Mdu)Zuqvr88S^T?2lWC|Bor!t<{-dfyYj7qMeY
zQG6rhAq@hjTn1iwztgaXanMewdBdiS0KaYKES~rl_}B5QtAlz2e>#+TB0i-Y&Xo#3
z`u+f3-fdlcw1%l%7CXdIr8Dr}Nsh}^ly&4X!<4icFX6p3?>%H{#bCM2YOi^_=HN&F
z%kA1)L46dYaR>tN7fR?(S=HUJ3kxO43XbaER%R5!|614N}WYxOQ@%gel
z)t&#`1wIy|VKwOvq_65Rb6**}*V4^pC)R-ic!<`uxtI|I@K#{u{vGB^!n;
zhV-#r*X(fz%u~9x1jk>qz14s$U|HwRlFEZ
z{Om!^8Fl8u||EIhTz`Ow&OF>RF;SnX2Z*O`kM^tX%d>qku?&C*@-?KiJAXa@D
zmyM`z^NfD4_RFb+{PG#k@#d*>!AZ>j@MeD);E4CXKdzomc4Y6INmks*`(&=V(@EY?$@?imiyo7#y_NY_@NF9^w^Fki;g)#b
zA5kB`{%f`3xHnHbl=7F~c}V43oagwRE&g7B(p`>+8p_f1{HqEtlHRs8a>D-OIesS=
z^FEK2*>+0L>&yE+c0L@p+ce_(?lx!H+kp4oth=&YuE6zrL@~Y*_lO6ye?#_;WpH^@
zKhCe2&8Q^48#SvX=O22(>C**oaTWK54~c!X18MyHzO}4pDfW{*1s^K4=JaZDFULdd
zzwH)0>ZagD@Np6HD|ZCrCm~DUqpB7>YM0=DeLPstRPZKPL{a%lf$jn?i+w!lzI3@r
zheZkL0~sQe_~R8JLuaP+Qv%^*q$ERUBZPkl;bs3{7$9Vc=l+)-baE;On{9E6a6=QDeIMs{vYAY_A5kxgdf>bxer3WSoGt;W7(gx
z!rmF7zfs7FVjW>2?t2+1@_FsdRK6$ue4z!|ZK$=A=B1NM7Sev-(^FiZ(E*-sYGRm&
z|1{|juNLkxyX=Rif^#u(&+CE<^YDj>%-apt@qCpP#^(Y3#Q4s+%H`IDNZbXphx!sOMm6kcX99OXVqbR
zjWFiFBXX_aI1B;!u7ddnx@XnvY`+628~^P`e@7vs#DMG?+mw+XyLNV>`7rYde`o2Y
zo-&=cdJW?5`1H@~Mf#zuOGrQC5Pw%VHCh9@8tju;DpCH~DbAEHx6vSZOQ*>c7hX2k
zJBNA?Lx%Oh_#V7MuPyFVKF+$>0CCc>MP=xJ--wlVZonTh7e9Q1>x=%c!`9+HWpeWa
z=P=IZwDlT?=Ue&~su6fUT-9MZtyikMfA56$k9{p#j_8=6@eq0N@oz0dkZ;(%U$Pm~
z`;83JKs$yjJW5ekzqAzo^gR52zkJ?A+*eYK^nZux>1(geM?5-ZknVeS{xo$&
z?P>A2e>SWn{0RR3t!!V;hBbh?FNU;1KGQMxPvCCAiqDHbBHy0bx~mVK8!CRwt7!?S
zG3fjWdZcLCMfb{Rv@9T4P)`_Z-C1C)?g{OYm!
z&~(6{L|xS{P_M!)Pt6g)yG{O7&yJ~r@7Z|{Kj#VjTD8!!sul3`hhvTy_yJEH+RAee
ztslay^Sgq4@7c?uZh#BSHD1O8?}{nExVs_XgbPXv8DLj+Nz0!D058tF?Ys!^j7sX>
z3c&O(or}jn{h79oU)}>cKYo^q_suGY_W#=s@UY^nlek}Ly{yz7+u?0agbaD&n%58B
zUj_WQ=fKz>fQ8ymW*I}iQiIF8updmZw3?j+{=9i?G};{AhxfGkwXVRo-Bc%xHV1w?
z`Q}ZFPjG%5mwGS@>xqe1DIW&q&I~G>hxgPrH=2E?8}Q$Y@2AJd00y4WR2>QVBg>0c
z?g#vEWd5~oP_FZ{^v1^k$A#x2Ec&Wh|K%bWukk+#tR_o+Vw22RKCT6j)q
za{V@7#>jQC$djK&oIZ^2)2hAw`v|}m!*^TaKCDTL#O8S3`};wkdqH>Mz5LEG_5gnW
zL34wLH#M*q=gweD9CnUmG^orD+hzIrEGZv7cIvY@LTZJIhXq&W}u;vijmZ
zcP*W>Z~FiS>%COM@ik-m#1+VcO$^T+z;XTb!Vt4hfbG{BX>0?G@{DPO{$1P?o{aTR
zi<_#9emb1H+7s8QWv7~GVSf6Bope7k*)f!OZ`;G~X*~P-*1F^T&{F%zXXNcK9gOtR
ze-kx(+BZns)h!TZjkdLP&VRB?BOmm~xz`_mOdoQ)lz6pGgJ7T{P_Fs$k2t45ob7yJtzhbNeTizcOGx$xmAGK5k&gXC$|K%)B%pkiUEFUCHZ_
z`TIEErHJ>h?2m9<{5NnJ+3y|8@!D@8$JM;Q92XtOGH;)(Kbq`YMV==#PU3iKoX-4w
z$R3WLHWv?5euggd`4#@mFZ!#pJ+~{Ym$ib+S88#7Y%}HoisC$W`8vqN2#2YF8L{?OYJH>
z_c8ID-r_w;djD0gZ@qT;FMVl!@?ZIqeQ7->`6c-+m6P6!B*Tv(;lB`XL_hr{`nltF
z@xI0S1STocdnQowTlzktT>K7L@dHhelzpE|-y4|~c>Oi<7@HQu$xn@jgJ>Pq
z`F;-7V>Q)~)={_j8lS{+W2**JJM`5KqUT>B){h%+4W)QLV~Zi}yVW(|@4RUHULbuN
zc|1K|O7M5+a{s2lH+BNlH|Xn{CmdE0E{El2(a(R%Xd$EaFN-!x(EZ^YP-ty^LK
zF~!GW$k!~EKcMrah5ZbjFuycAf_PF@=o>0;_1LBc{Rs#>rq%*5N5AC3E#RNa#&>y`
z4LGiM@>v5&PxxIGg}l0pZR?@99~`YRY33rpTYDa_#eLj$$)WJ431FR+_>;an^JMFi
zo@jUY`1zyoJ>J!-yNAy=r101mOmCT(30qjOzdEd$2m0B&W9mQTb#X&C5HAgoshVN>
znRf#RqTjEt-tU3++Kf=l$M$MxF;*S1O_NdTXiw>NNH?^%c8Pq$Y{1i@m1(%1b;wF>
z{{%3r$l-5O;J=FT)pmHV({ttZYUGKLBhtoW`RkVoJ|eHq(D-SE#Q4O@89@j-7pDD6XTA1spWM{Nrwpolt++<(M+GyW&@!Bl`Jz
z;d3j*lLwbTh==#&vcnSfN6*)$?~?g@Ef|V+kA6F*fVkvF`AJ>yb7t_g!T^x>?;d{(
z+b8JcIS9E>f8&#lseavLCFK2z1}7F{d-b<;O2YN$ww{R{NY8(`0j5tHmmG%rTjE#J
z_ip+c-5@`AwI7#*dU={lnxkLt6RM-oPYV-w+D|WW-LV<-o3%Z57V-E4wH=7%ngc`8
zAGKUJ_yiT4BPfK?xOuWC6{d7pUq1-VMAR8w*$3~Hj&0D?5T9%B=VLvvp18+lq?di$
zhsO85l4u3Avw8avH!L@5`Q|jlMk%^9F0HfO@1m?R@J$CSU--`xLJjzvwV(2@zdcCf
z)4fGJjqCI6I&H-KFLi#wh|ByHXnxeGJxJqSPnW+BU#G423G)+=^-M+7_2S=)exGGc
zdJzk+5GFY_`i%MOj(#Hk9zJVx$%FE)r9C??y9@|`wPbnH)ey8~E#z}LQTQ%Ty!zwS1Bh}(}snU0%_lf<&&K0ab
zehk}-v10j**x#HZ_92a%aQieBe)Jm4deP!{v!-I7e$sfh(^2qtL%}cXL>$f%{90`i
zm!BeJxrp1+emh*QruJATc+Y;pKNJKHkp8}lwUA3iJ`nDp3B5l8D+M2bphNkvI3x@gDA|+#9*dL@gMsv*^AJV~TqqEN4W+|Bgc5&+K}GT_
zfiOr(meM6XiBDFDe5oEOUqa~YR382zl=SMOlrQnq`uX)`$seg){d`H5?7^V?U)3+!
zlTd0$={Y2n@}+ttlHSLe
z0LdkNFM&B?JtDmiX&oiC;~Pm|%!Bp)hhRze`~^N1{>~QjcFTS|k0+ewxL_&%j+oP5
zzGvub!R2>}d9AxRPt|zB<(ll`IMzVS=h0%_(^%AF?9b)RTzMT8Ana(0b&{s=J5zjb
zY1%H1Z}8(b^82l5=X@~__Y?a<^SZIW+NPSMH~CZ~yi&NQ((@t7cZR+uo?y3KllCc#
z*YJK;!wB}HZZ)qLhZv5ca^bi6TxIB6&VMZaZnTm&*VFBFBH48d=lgYYG|p$~Z9?)8mI()&|V1YH2mlax1XLoA+c;*D5!
z;_DK`z=VNJq@?+{J=;=zZTFFR@&i
z_EYG}us^?I2Jy1MPVZ@5qwdf5;U-%yCjNJ*qZ6U;*czA$fxpgHqjo#dN`H@_WDe
zsOFIGuYGebaWAc#u5$9XoZNW^vcK_noN{`8!;3m4#B=9*?t1^b{MyK#IEM!Z-zDCA
zismq3r$;Ik5d2eQREYQZk5xX=ZvT~|q?}3<$$26~+M~Otllk?VO^OJ4d)65d_`_Oc
zUX`W&V$|C3;X6+>ATZl9n*CsW1k@H*556y|h`6P4t
z9-{xvuhj@XS)~yKee<0?gq(wEIz;^yk0hf0Lk{Wx7#jYND3{7)63-FUBJan~NRs~4
znfFLLd1Vy2Zd;172|h0Oe-Z8Gydb~756X~w!gM^TufCa0AnAVf^6ZlffANv-S7gzK-$**L9)FZp4*#sR>
z`FZy(BAqwIFiVY2-Z>c0LBwJ*^Bb(W;|gD@+S?is~Do!Tc=sC^5A
zeu;jXQ*z`}`wWerMB(>Q^aguKdiuVOzK{2&@bwgAQTOfElpJOhA8QK#je_(&FDf5G
zjyej`b{K8XV3H^TPum|j$%ryeA_C&1B_J&)ZO_qonjVz}LHCY=IH`#;T~E`|cEI2m
zIwU6OX}+lRh%(Iw7w`YW({y9`)AfIXW1lmYj>gmd{u7?=$JpnOEz|ty{*UEHm;YBf
zn*LAn(Rdn;m4|LOww|V^<)F(n94r5y=;->fkmf@}8c&!1S4g*`AuSi(ud(H^bacPz
z@%y*u(tK$?f1;l>QtP>j(w}47jV+_`g^xymn?L9Qn*QJLG#!1uqsw%^==VTFT3#i}uD?R*OASi@
zPCG%$^`Mg6S0pHW3MJ<=3OZBu`qX}NiK?g1PxSNglAP$LGj)za=xy2$#
zy37(1O3GCc^5uK&z#&Bc4QCRbvn1oyVzXBh_yx|Sy~FAu+pVVlKQXmyh`{?p_0UGNbz67ew~!`*RY;9VGo7{7#Ygqiib9Yo;E_=ddyf
zL!BtQ$%mw$U#UmPk+XI>VK>iouqEKUf5LFd;CV@d{2nCHZ$|XfvVc5)iJ6oB&t-Px
zxv;#H9KR{uq?~@jJc4g!KWRsvlOeyCMTnF3SHK}s&iN)!2)Z6N`Q43cNAhjAB<-&?
zhNRt8=tTOH)W0DA4`RB9v_GB4k$jwX`{NYFd;Z{$gq+_qci`0^epi?@nXvO;OSu#0
zj?g4hPVFY@9J=5O@qB&*$^V!vX%Z$$+vFab%LK8m*n3ti{!VN(m%7Qe9L*376jdBCwULx
zAV%Joznx0*X{O>8UmPdDj|Ncw@r`fD=eNu!VGVVke&il0XGJ(!AHAN0dXyiy8)ZK~
zr|kc93lbkh=|>(F|9F$~e|Z!`^2=x<>1wI_c3D5NjKN)m{0ajkolpZ=&Zq2dOUka6
zqwHq)G!n1vM#6=ZKBE2C*HUHooh06k(nGYJ?Man|C_SA+$$`$IgdAQJo=ZVIEEDxc
zw7#SGk5cuK6#PH^Pkqp`hM&uRZrW;NJP-l
z?dbjD^qBTEo~Eba>oNRjI(k2$%e0(yJ9;0bFD+}q~)XSf*w
za6xi$sj?o%;Z+E_MC2nr>%~L)H%M4nA{1-cChm~PmBY0Z4Vt$EN3!3!ol-l5C
zefDg@Mh;jXy_J+ZjQ4j^vP~CI-Wr)`7cR=DYI7o0A(aDqI#RDD<)dHk%Ic42Q2y;7
zd!#Pl{&dxQY0VO(^ODUg*TZb7!Syftf`#(75x@;6m>Ju&)E5$;T9^CI5IKYM7v*m>y5#PGmd_mM~
z%p)GVPB5he?|Zd7pU;iqK#fga<~CLbNZf4<@|Em>^3e;{_QLJ3>`>P6A(u9|Z2D!m
zuC)ycUAFN*pJBmOrk_!=Hv{SyE3I9=h5_b6R~{uRFu-Ybx@gj|_mJ%&>YV1wf`_#a
zKk^lsz>VK?T2+w+&vW}DE$6bpA@dw&Sl~k7&vk6UbzCU7FJ3!y9OgY6IsL8_$M<-^
z46S+i9*wsZ6|boN9oIWg957!FLiGt@QDKgPpt5kH~Q$*J9YDU`HF=
z@iyBKw51&mtn)h+@uCA3MwSF>R&ijvQ(%vS62`ZV%nK--%>ce>z2&1R3@C9+A24Sy
zpn89W{>N!7kSP9J%g})Z7y1T!3s1I9pzroXbB)FfU7>n3tI&mWwO*CZfEL#ZH_vz;WPg=l|1&^R;g}
zxN5oM{IqM-%8luOU~WmqRILsOIxL!FwYMG8MjniA32K8SIcJ-+1lu6Zy6e`+2L`aM
zwzk%1Gr%hGUTWD21}vPgT{#%#)YOR5Sv}keg@0Ykvk+p!-^MRZ^DLNP;Ba}m&o~yC
z)$Okhw`RfY$!MP+TXCAMKw;aiI8~uK7^`)K}M*ykbsfz+q1>iS47UFr>cm&d(wShy^v+
zr}r^oq2lIH6P!mov;2-ETd}~U*8kZm8y0L$@f5l1z=UIkzQT;URtVm+^x)K)tx%Qx
ztGE#7)rFDBz8KUSJ8F&BSfL&|-*Nnq2%eV+L^F5#6?Jw1!Lpb>xZory5Uk#bV~B<^w`WWATsw?!O~P|A095aodA
zlBMeVG&o?c(R@r>hXp6IY&yI~nPBc)WYE#ege`%B*FNEUK2-Quo0JC&j>(&dDC2x;
zt>zg&)L?*`+l&N#CKEia<>=3!&jKlSTH$kL7W^kIu2UNG4P6x*EXh{q!ic-P(HG1w
z{4DuO(&QPauZ9F=-ebS+xTdQ<#klPWGtZyYH~by1E}5=o#f9Ez6Q(%!$7E~HXe+K8
z)7MtJBvH;8E4~IYoM&>?@#emL9pJls6+7FA1CQUl@7Wv4g3r3%W-ih!
za9mq|I;WWdj-&E<&BF{xwO`U8y`72w-yATySjPm3rjYCR_OL)icCKAiITQ3xXLmz9
z0|Yz<#qM8Wz)kHox0x7!K5w8_az5@~v#LG?&qjR|WcF>c7WOOb%fqEks83@!zibcT
z`*uD1GwtU})VoE@6}A{>Zf+CvEB_}4E)P20u5Ian70(!=mB4`v<>uaLX&un_tMc9+
z^>&D!H|?f$1{)S$1I(9mS
z!`GN#neb|2W(5N(lpL+>9x!2g0b`|O9Si*QP7b{lV#CAl@#n5^SWuYsWLOmQDLvh@
zE&M9xVKS-G3G2f3epXvH$RFoRU9*1BQ=C`HF1vRm;(lN`<2mz5COPnn<=zO%z<0SnSRXDDyVX2CDRh^k3nSRm`FT^@(fYmxUuT?01U
z56fugM6khG$%5fl%!Z>L+6$bRY-lS>l>Q6lylrEp{aB3)lXRDh_2K;BMP>OupUH*L
zW$Vo*jH2IN)!0QK+Zp
z>Gx5IGg%5wrC?Vxsc-ApHxPem}nPd(aW-myWXz-z!@X&a1xez+#+bsHGU`8Ad5wF4jiJ{I8C
z4nMq-+{J!i-pi-L?ivcXpNYqvT!8DlRO68FI^0kAj`2%WG4DXnou2^?-%-w?@nr=`
z=qH>0tfNZ^^IVmDu$hW5eY>AZC(h4V%fyV21hm6_@4(F!&;e_C3*7#5lmj1ZOv5Ly
zLchG)9f1{LcposEJ-ze}j#JB=gs*LF;O)3^-1-CU;3>Y3ZMmca&XkU3p2_F{ec{5r
zhaB4>z<9MmLzefh@``TT^c{=yA
zxav3#Tvrf?R;c9w?|9~>xD4>(}EP>wtZ|qMM}KI^fX4U5@uNIgs0s
zcUQ{+>C`P-@-YrgFh}e8Gqlg(>^z?vXqVYc+iNR<|96QyGiCG@?(22EA`{XOZVjC?
zIdnj
zyAi1WCLVE~qNIcNo92V+V>n;MQ@ZDCqrGsazHMD(2p8VThzFSL?u2a-)tUv%yWjxC
zzKkvDf?Y4dES}b&{axc2^GJh>dCHb627l**(<%GR`S&`(c*EYv5veZNGW_keD%!z3
z*S{1`eMEoGO(`cE(9Sl@jlEtKhdCAMcLjQJ46pRRdhfc_9`A0Bq_z_Sx}i91L4Me+>9-4e
z;pyr9;-Lk7keRG~a#wIaICU9lpMTR25$j!z&xCcu_{q;WS3A1l?YHBGEhl>*WmQmy
znqn`=wChxy4DE%HtXk6)9AEj!X44;-r{u(kvsH#C&>z}zPD+s|j?c_9&+FqjV0)Tz
z|CI~oS$kGu%+)|Y-Nrp5HbR)UhOPd7$6pvPyJ3FXT*Nzc&6vT`?}WG6*K~~fJ0ZWi
z?2Cd=H`F90350I%1vlr*9eb_%z^r0Rly*x$m_<*N`ml)yaz8zyme1hB$f27MyO|H0
z{OY%=9pFP0``7Vzr+UGDxk}3~b}vlaHRGZBnLar9UMHkVwjU-psh3Q3?S~~I9>+2p
zIPgWnD=1kX{eaJ@Ok?7?L{qka?Tq{3^2@)zyP{nybRi^4cq8VAIUMcs{VNC7>c6@(
z7yDVN9+IAb>&@q#>yXZET>tJx8HG-LAhOC}NIts<{^4|fGgR$@$wlo|^#zz$`GtFC
z9)?A1Ey-=EFn-!>^jp_|TYs*wM+457YMzm{}O`;rY(SM^Y~RU%$*R;g-3Qy>!N+}|59a^S#oo^kO5kQlNRwIgUR;^EmS9-Oi3wr&6(
zWIU-AS~Q&xJsS?F^}2JR*c)``YQEeYk%t(*02M->_HDrvg
z`1JX8njd`|LVaaB?IbX}?wTd3f#+Q{5b&gyS=-sMBO1kBfOB
z4}UAW!G*%Co4H9I9Jp}HA^vC%2c(Yqe}6Q@0mBUqu6nBI$5VK=BG3xsr{{(-tR|q|
zTzM|$)|76zB(vgeVsj^W9+_U<<_y5;>1sYEN&~R$SGa28
zFdw2FCQfG*@v|HvYvUSmp_MdQh
z#VRJ+)ir^q#Bg5C5^2yH7s&@F7Nllw=fe*%u^*?qc_3k|X)mC}gJ~8XpLaL(gLf&%
z*uRH|aoUgXD5>y4w$M;k;UXV4rj0M|#BpoL%l(_LGXUu#ahCh`aH05mg_AJOtMozb
zC7Qu_&rq-4>w)JCxAL^GhNZav3vCL`(9Q}AnE6B*<$qn`erc^O?(0(%lHZJQKyaa)
z$lr*O?jgv9ZF+jf!O3_o$(0WN_Vw+-UaOiFqpC4dl;>{s?Ujr`JTecZ1^kqdMno
zy5Pv%Sm`R<&*Mv(EB$a>yN}_2_g3P)-%G=A_-g=0eXB0E5Aop`Yvz;9wR{*XUYa`T
z8xJBst-88@S|0@L;|g=Uc+g;9eaLA!52VINitqf&gQSN)QsWc)!P-V@%9GF@_%A-J
BB6$D+

literal 205696
zcmeFacU;f!_dlK$DzaxOrO1|*`9Lxfg_ML6Q6!}zdy}0#Ba)Ss$S4_QHH?hxjLh(S
zin96L&$+J0=k@ya`FOuR>-PQr@%z30cpc|l_v>8en&({O@oZs_uHDM$)zH(S|4Nn8
zDXvqF|Ed1(xBPFWc`+8LyqfMQ|4~HJFiCe875^xrQ%#4`s!4e{sozwn)A5lw=YDo}
z1X1;?RA&4C$^XY&pod*&Cu)L$vK>lEdP$BS$?{mEe=}GW{tu`+Uf1giC8}Nnou(`(
zrlYH4pi@exi2n%Ru|YcJV3#nh)4!SW|FY0Pr<2U52O#o!2Zjs__8sRRq*Jk+j*bx$
z)v07VZfwAK@1P)GFF$X)5I^tnb}H1mE6P;MFD1$k80R->g!icb(1PMzB>(TPd?lsb
z9utO*|Bns-TeiQIr?2Mo_VXU=?H~Lf=lvmzG|;i(e-+b_^&RUSdqi)bP)nV^mHYSnBYpiu{*r&oPnsM1wB+KDPc65EP9UHy-0MEOckok5EhEy@Z?NqxnN
z$Oh1PoI*JDzY0=J;5QW_y$y8gaZ)kXMB{rH+K+}PE!@H;9i8z$<5UGYJ^t-q{~P~M
z=Uvk)LR+UoN@+BxR#1|Pf1ECWodo%}|NozB-^Vvaq5ssrrVRV{?MusrTC)kw{`Xcg
zVO8P(3qPw-e(^v4J^e1_BA-^yj(<;8J^x#{pM~sE^{RHN<*D}lD1=vWg{hiP)%Pc=
zmj9=8Rj%q+`NC8!@0+wwV;)lQYF
z_NwWBqAKp6%KxYI!gl@>UQJi+`cwOe8K_lMI}4Yi^1``m6;-b)FHF_`FDx(2tNIGd
z3-hnIK#lT6(!zG$m-3q2F&afJU#+)Vznc>GgcWF%!sV;&W5@<*6(`Q9Ro=*as$FWn
zt5V)n>Jwf>tGtwQ5gcii{!-sSj;pjvQ>k~7l&_L@_hc8TRrXG%1g)ZxdQ2p}A4K37lhrTUiO^m&<7L9No~rAVJ&PNeUfEA;((_@4ExoFdBg
znnxo+qr6)$^cQ^l$o!E4OV-oOwFA@eC+AtdaJvcfItiPZ-*vVv6l)b9^BL$NjS|#G
z*wLb)(C5z`u2sH2037L+p9(tawy?*-{wvE5wioir6!AizZZAQvul>a71EvVQ9b<(&
zVyVC%t|jP=XXJ{sO56
z+I2Emt0YkMLBGiHP=U+<{sW02e@o2@{zQnNX*>Zmih~>n_vQH6L&?DHohtGLmlxEM
zC#*)9B*#0oD(O95M)b!XSCLP{VXIadeOT}lsT#n!*AcXMf~aSSccOmx775&uA%fb;
zdGmx=*kM4~kY8Ods8xx7D)rWse7eM|^E_Abv#GnFCsy)2tdak+$
z+iSk$8%z9gJAq#&=h0i9&>Ce5Z#J|FjVdaSn-%4#>s8_POO^951D@4`a&gF4nDQ`#
zTwSLMr>lIrw3}uP*`eZTRua$6MtWp_XGwfRR-jdA)(}o@7j8cATrE(B|Bb40wfsL(
zR*QUpN-r!|?Jg`=dA0p$14iRPm8Tm(
zoGMp&)jm~U;dB+JrmOt_7X4Fsh4rgAwSQE*)%vMa#i_iCSE-s_SpWZ&|Ih0E&)T83
zw@TIes=PY>{>kqDtX(QjrGL_^rmIxNsk}-H+o{S`yh{HhSJVHy^gpXtwX-l@m8;|E
zPgLy}wO{|c_WoD%srIT=EnnqTs@kX4M;$k6df{^aX?;}j>iVtnf1;{iRm(4oQ|-9T
z|IqxKU!y4G?|vmy@;TfQf8o^nsQ%}^#4YCr`S*T7wWBN>qE(7>g|!MrdsH41P_-R0
zSfEkvNPjPmoBrC@sQz!5)OS#y{-ITl%6u(lJh4~W(})8Pt@0>Q;I4BAX_QYAXT=rL
zD7&PdbEyL7&4z1~9}=&X_!BRvAdM0{hX5KSMap|gx<~4Zttj*aSx|yj8BF0G@DZ{-
z^JM%rXNZt5F&FXFbZN(}B&ol;$X~pQh(}BG-N$`n5c5R5Z%)&&QZvx+>z~#IM4h
z?*>$1wBIgc)?@7($8=3e>T#{|SzDFqu=}FCGTrB)LRzJ^-#GNJMmafYI_E$0q#5Ii
zCroBOJkFGHiGJ~ja5T!;`D~4+&DdB
zqy`n!D61-3BBItRu?9=fqZ;MBK8+NuQto4M=ASJOzz{?{wWww`rHAYedP_92aetOO
zF8huSf}YhW=sB%&=UcH6#Ge}#_k{S5PB$=xG>Xa2*<9XR9STIXO6NAu*`6uuYA~NZ
zr~~WEyko@WuSpWP@87`!AJg&x<9%BgVgzfHwg+hkt5NQMJ;?HL9|i8v`Ob_V)vPBv
zNTWP0H5fZZ*mY_+>y3WYiSZ6cA8`Fz4y|6c2-1^wcQ+wiRDB0-k9${-vOS+YTftC`
zQpH#3KX^-w2a7R+zdJ?HP&qG0&Zmi`Rg53fMn$WH^1#<973Kcd*9>CF7zZtKh5iX!U?$2Pc<&kG!-FiDTJnI^C@v*Qp;mE|`#Iy=%EjH|xt{Gm
zXK;Ie93sZW>FQ+|Uqp`kph{qozd8+5)O96r2~*vCG2AoMmIB=AL63Vg>}fL@om>ofm2&ofVHc|fUZwwp=Vt#G`c!*Vf2raN>o07tnyyk+kIMg(UH?^_
zs{c=V)pWHTRsDsjI(}6ApY*A6wLMh+ze`m+{@i>OXay{K-x=zgnMv>IbzS)qH=#|I<3A>bb>={yP6r?Nsf4DARXv
zhQDxpjEMC{#XXidb$wLpLtcp1M^!I}vEZXP<6rBn+Ry6xc~R<9<6m|g$j9yxG)tyW
zk^0BUxZGIU$;YI?PI4~&#{4HzQi-Af36ziX%{
zfAbEZKhRv{pLtBg$?M~mv!2P)UzTqoaM^T>jdCI#$pDRVAbbnc8_R_LT4_S>>9JHH
zjqk7s>lXyq3NlVI$Jp
z1~p+ln9>$hXu6@*M0rFHjsTnS;H
z;FZJiacpE2j9jhqFsS20N{@UOIhFW&Q_2nh1-@kL6GWU^<%#!uOjV7dHD72#e1m5*
z8&P^?^TT?i&(f~fNJ^iy@AGI%pD`zP1Esf(d3dV?c>OiMQ9-S8Z$!CqWT)93t7$)x
z-qWiorixY>Y1H&;S0+#1zLkPh5>iOsni_H=o<1
zb+wvXNY7@czOIyRSS%I;5&QP6v%Jp=*)a?oH^lSr?IK8DVE1T*lv<_o_l>;oi9UOb
z`_;wE_A2E|e=(v3(KTVcCw>9_T~W`F_yOUUvGLO?7j(`C5cMqiZBIYQTaGR$SpZsN
zV$6mgpfhS~{Rm&h)8YcvbNAP=JRW|GEyi|r%eCfl5I3Lz1044V@5K0RSnW8EpC^MX
z829sH6yp+FjAz=QoR-Ihr#)>vv2RTI!sBJY(NNaY%lRAUYxbQcG{(We;BlmX^2R|3
z6*bE34fna-f&yF_cO`*t0&vf>F!lh^arPU{2zTN54O8N)JPE`G825)8Kk)cz(Lm^}
zY$N8G&qSyJzUBp}0nTr*n6G`a&6!W=K?|Z*(K|v5phj8l3Uk5NlyPaqC4ZK8+)OK}
zMlm&|l~=2DT`ln4yugBw@V4dok`y|S`8Hvg4#2mLpp{&!SQ`m@KDs|+JMU&J=k^Ji
z>cRD#eaeS%9v|*;zSpgVor7r`kNn-pEYS1pAT&ySb^uz%X0oVX`W$3Kd0N{31NZd3
z;O*o-V$~B-ug_yd`bu^j8f7hCRA`lka(@z9{~_xMi=cwA|2-`5@6hVRXZ8`ieIFB3~LtAK7t_#GpB3CUCS%BwoFbQtz2=$ohPg>C*@fd_Bn@l6*K{
zBx8NWD5Ug>GJT9px0U>Nx&KR|;e+xnQ+EM>ShnL>nZJAk(Jnn`*#&L{RUf!Y^8VLF
z`X6>vkfW0s#{_tla`ljYin4=pH-W0@g=q^GVBMEg&9CBAp4|`hsBr@(2mXWqk>zl+
zf&WI+4mfpxDGJ8Cmblt7A6FYVRliDAxw_BdX8K$GIQ!pt70>GbMLpGg|D;dFsrx+@
zuiB@ktL3Tvq4I^(Rryc;$DjK_%~!Y_RbH6>6HYBx#TDjNeX9JQcvX*z`@cmC+pF54
z%GGu*OjY|;%G2;)j29KB>QT!nOx5vGSYDV{&mUB|YOl)kw1GX>xZuCybk$C^990jU
zGSWO#$G^HBsPkRD_jTZmf895$>FRt%ki&W+IYFcFF*Wq6_wVX?yn3JhT&_nZvixUU
z0j<)K8%m?Rk?UuN!IYp?7D&B&r5%+dzlIkOtzzLUaA#z?3s*>^JmLyzmF;q!T^dCN
zVttqSZ_4y)IF%*&F5akUmG*oeL+k4KB7VeQBG(dfj2c1aryieLf%uJ*FEP7
zX_c`7v;yH=ZN^ciQTqhW^0csHq(<aCy2C)AvJ2ecw
zSLngCUZ6-fju+{@`GP^KgxTWQ8F*(G5qGDuW6>&)J;;F~uGA56&B%9+nLjmYDp~?@
zL%>PSH>(t_=!la&oteK@aWh(0t5kbvfvp(gaf8+@kBV_*oVkHH^CKG7WZbwB2hhSA
z<)ty5fMH+pvm2+MIK7Jb_&2{WB;kK2H01MfJC})^KDYC7PM=lV3Js)Dsy^t!amAPX
z?&ukN3``q3|bj-2>u2+e!IkSo1=hX(0n?`B8mtb;C#(!B4qD!J%twl2Q)=YSCF(iOU-+!9LzI*>0Ws8GO|)Sqr*=pV_C*s|n!U;&MNc
zT?y+GX59q8<+Mu)EAS2SrzYI}j{B`UN%dT!|^bK^(=mpMH
zzs<*gOf3!kry?DSv;hBhOue4tifEPFeHPlbg4}Ln!A!Dar@d8;1EBfuA|7w4fPMCh
z9#e}I!#!mCKE3mSEBD2v4DFvvKY65U2tFpii)%5^n#I~%l3h)A`!1w@E_fG@l}f7|
zir4UdvR{kPpENEOX?wH%UE9t)Pq@XQH6n->aXHEJ@@}!?h=?^xUHv^h2!E_px;^PR
zbJ(*p(QWtl@V+al#pF#jz@OWBy)w1iwag|pslKCQ;t@e>l-=Kg5vt<8ab_0;yc%W0
z_vT9o-|<$UCAF`Gtx^Avpow!F+Yzn&V;Ikij$iuoJaj#QX#l*=t}Q$d^=eXJjQhER
zw8KEZjs48@-p=1Fzw^8_<81Fz0FV0?3rAd3q5f4gJdccTT41HdxI0n6c++?~QP(IP
zQ!e;WJ>wSy?%rt-F#OhDK>Mn_iG%9DWh?eAj{^Q;g47C=T*x
zeZ;yQ+n@={$6VaT)ZURC8s0Z95dOfTk-)bx+<*le^E&G%r@w3(%y_enR?L6RC(o%-
zN@Ts@`b?hIoAIAow&V4$Z+Q$0hR7IYq8
z5M#elU!>pQ1AXiVFzN`u53NtQHy0=)&gqT`zEZSE-%HH^`7PQe0$-)>X{NKb34eCE
z5;Pv
z;V0vSPE;Jyon^ZCEJmxWlDyd+k?*zi>wMe?d{>!1g0h2;og-)oS^sMsc;miNw%ZpD
zl<=!5n+^3-sha*Ls`lTX(p7s@{e^L=
zoVx+#s=Qjh%G0Vr{V`ZlRj*n;i;=GOuNqIQ>1w&EK6N~-=7hiEb~T-s3$&ver>DyC
z#LEN9;bD&S`_djhCISCR#={&&1GiY}@5+ibia#sXDu?C0c}*#&)tAPZo8+VAe48fc
zTWzi|&da48DA@AoI}Vw?MCPAC$7Il7N}k_OmHUFAcOripPgKMS
zQv_cwTKJjSej?x6HbUNequ{H2B*hxV`29JiM|KK%|DFP8K11+^wq!8gpDZ$jK&$v}
zp$cN&M~U5-(YyBu|6+5;rR48}
zri5bYM?BfL2OKs0`K+bzTv}yzYuoRXKBU*?w-mS4|J2iz*J9jHCMmF&DR
z{_q=$10QEzbNK*%)UrK^^@#u0`qe4KzoxUU29^Q8`B3n1D%XC9MRHZ}duK-FZUSHM
zaBnSpP`8}%5zj!AntVy7aznej$KC;7zVx_!4d}66IaggFe^xjAXf@EvZ@kt~+?OBH
zsC6yyVJAI&%s`VWIZP@6`KR*->YhaTO~We0*+Slbw$}zz$cqn(XmjER;=^U5lS$rm
z(c{bE->{GG88fOJ_|mW1FZl5V=k;@4BO(#E7g;hs)~Fuh&qTe~o{0BL7PBsO4|>`S
zeQ?YO_@T2`#61H4+Uk30vgh`}tlKjnKeW1CyLj-;?vO|Kg?_L0&%zA!MU%#V*lmxYf2l$
z?@g;&{)j>PLznqgDGrX#Z5{IV4))WIN85CI27f=I(wP-SwaSXhy}k}dx{=fPfwzHg
z5csGe;VrFJx1#>ggjS0unqH$ekDq&Et6)ilKTp^6rhQTAayvKDe&gWoG5#d~9-UN&
z`X{@scu)55)m#U%^M2z)C8_?A&W|=xdif$Xh7n#_w2J4!?|lwM9U(WYv1>bxpMC{x
zN9H4a_Y*bR%vtZCC|^3j=P!Pd-4Vr6vU4MM{!(zd+iWjFRs;I!8ol(+@rZ=
zH#4LBQN5T5gl}6Z3?Uls1v7Lb2-kep8QwoX>W^s#xh9qlXi)!nasXHt
zC!ko!GkwK6o%+O^`Oxm-{b2cF=F+e3O%?SiE&bhS4ty}4WZYf`R#E;=4~3mm
z_Y3~xcA;k#RUhe=mj&;)aU<(r#1G3f$_@^swMrKk;g|Yq1b$;h65yV|;4I@~Hc$kP
z@pMS|$<|XuJF4+x3}pwNw+X;KA{M+FM^2Y<@_kZ|^w~1qO7e6&O!(wmf;Oh+0Us>W
zu^2M%A^H6r$l&~ksttWwY9`>*Xc&OsDtWySVdwZhLf(e9A;4K<)Dq4}wj+oC;7Lky
zt`2a7CcMh06yjBUVcMJpf5lmx4SH0Z%BRVC!a3rj?qtc6yuEWcT!Hh
zLvjnrGx=LRRGxV;o>wQ
z(0+r&du@l{MCQL-5a+Tr5FQY&-fV;IW#8mJm&
z|0tpFc@4Nr@G;9$7}sQJi{g~O{v}_QZ;y6id4-$858Z1PhMvT;ZT*E@*)*M@mRFvD?PT00|z(sL^NUZMCa
z#Qq2Czq8jE4Wv=x%ww5fo6zYw@zH*#7m|N{rG1A$8*!fbbEY=>Jy^fT-8N`o#2Yb1
zBgxNud-}r>VcqVzlHWxsKFm_WeLC2Ctw@ltS7hiru4U?oFhqY
zSLen~ig&U{)M#T4`g2N8JV`>nX0IH`jw{Q5?x%Cs+ygtWQXKcaU%grOpzm@Tz!4#?
zyjswZ;?2m~{q@9FI*lfJ)>Af7C~bkMoth#z|_s%{g|2=6^Yd%Yb9-aRnS
z;XUHCpe)bzsn8dAE~slgl=~pebAJy<+{fp&nmGsgb4EJl*&*)JZF}AK1Io#3Z(iv%
za8+K6sd{NX-W7b*sbQ(XKJrMNArwFP`y}XI292>`U~d1cAxN_r+YoS--stKo(pu1
zfxhM8S3A}%ZT3DwQ>N8L8e)F`^*gshGK3RZt
z%lOv?zcUcGwms2jx)%HKuGw++9}rhPE~jtqqg6)TU%qfqH?2}3Xve4R@9<7)*)`={
z9QJuRl{S2jMY|_uZ*YR1{8pZKLP4F2E{Q4rl8ukaDcXH$pir2qCa&)6{ADv?r
zP4R7_$z6}Vu=8tfai>Y>w}#DE>(e-IxF^MX2Jo*|X3QP}{@@D7I=?q#Tt1F)I$41B
z+dXW^5sZgHHd%Ua(S8mON_R~GZq9X+8Kl39du9fW^EMTHv2fwtxNCo`?AY(=t~)@q
z+j#}QX{1$l44qYx#(m4Tb9mjj_@*qcdwVMEn@PAf`+Jum+Wq;0wKR`TUTm>|^iA(x
zbrH$E-zIM$+UN7-Y`TZ2ceV+yEB<>^@iYT|WP{}`YCntGSxaeNRq=WGwJYk`S;dOP}Rcl_C
zvl`YoLhbf^h&z|>{Y@8+1NR%odw5^)bkqipZ=YBVUP|lDjt3*PvlfT=F3V@vfV+VF?lPDQ-0dSGU&Xjn+-~X91z%^;6;A(}E6y{v
zO+U;0-r=iFDgUIt$6*-OmE@bO-{j~4)>}T|GPkdpqwuGdIzbHFgB)Z>c@5?H(hPY%
zRVPW?&2{NBUwy>*Sw7_}ExP2lY#G;>0i+rwOltH6pFBiDw9B5-b
zy(H4F_7L(u-*gzad7tpZ@f^UwPo;=kCe^=ca0ksal%(^DI9=bwYa5mwVZj`GxV*
z8TW44Q9h3tvRjLZiF38(ftUf<*LGfrtqgvLWLAUe#P6GYxeu*(D<1b}+}GJwxg=kf
zy@&5Jmwq3QtvvFj#NloM>)?BHu6No9Pd>+R&8^Kii22o~UULy&Tg+&}
zhQ+F$#;d9rlDdRC^u`XT-(x>#ypwev5Ifr-H
zI$<3@@S}kIPx~po&(S%{{Pd6kRlw^^J{#E>^y!Z+iIo4``;%FO+vnA7HsM-y4j5Du
zeD3O((=x$d|2BLd#ijm{53w*J{?UDPi2PZb<>N2VIprCri@}Y-U#eB2E5(`T#~v!?Xvqm;;1)I^1{CY|FIya4%M%I
zh{s!F&}LIRl-UJb>vm>OrdP%LAG2Sv*AOpMNqshA4bp3PTy9724BVTPovZ&D?~e2L
z#nh^=RfaqEYB(Hm&PRhl-L~-8%|2aN-4FTv<|%U~LB2kw|DI1BP|lCZS>3k0YnB=`+Nfe5#7Xtfk4{FM_2BlFA=7bQTH>hpgLQ~sjO%O)
zQDC=m;Q#Ox)H|+mioq|Grz{y;yejmj?`@i3
z2>Bi_(_c*?9~Zan##892yr)Ygil;29)W^**&NEHyo__*A%dDD0ab=mdbw}NYd|~@O
zy1d_LV}cz&&Wr57SdIn_*D)GCGL){LFd4ko&%le{JY+m
zg}M|^Mx_`Az5ze}?ZbE)KQqTXI;IC2G<@y`;=dRfrUiqp4RyGm16);Yzu^=YZg+3?
znaTt*lDpv%$zcUC5r!W2?r}xSg}!l6RelO_2Ha#?mBpOjNdvoyxuI^Qm-HJ3p}2>
z6YhMe+Hi%~cX~J+C;2gJ_e`Qgg0fqZd`yM9-DsZe-*(iM__d+N*jQtq)i?tibBx1!
z<#5u6_dgY8@wyNiPB&+m=M}c`Iu|)W+{28j)yRTy9a{|(>s#JUI0W?5z>kb;mUrwI
z@diO@Sa7wk^SZVq2Jg>geub$OkV<7|iY
z;`&AfuH<%H74?qe$lbm-5RhR1m~Fzig}uZ%!34Lej87jb;<$#z$f2VDO9;PplwT;L
zo`LwR0r?x*-DUiNu?X{ki!($x3;8_sGVvu+LKxSjn~3{HEkv~;PrZn3J8**!3f%3p
zcwPcNCP(YvxP!w5-*%(G*}6Vp+~lo7{(ydV3;LhE6a3WQVqED@!uAsK=o7-u6@A5c@JYn5
zhrIq3F>bQ!>_JBOu@fS0^Up?D0DjVpi;Qz_AkM#{=56BmVPM)xmhZL1whFilpLa2D
zwe|$_QID7Kyt1Dp>eIiZh{Kjn68T#8O<>$lQ#%Y4tzxigE9XzJ?ZEPumaUnexcL#!
z%ZeVjMgq>waT=#@9!Cd{8pU#XXHL)QyBs4K=Sto74{hnGnop6&U8dj$Gq32qEj+eIOx8?Fn
zjkLr_!23ssqr88gv1;QRidRb3XvyV$?>GPz*C?l-E#mcfP>(ldh(9%>$uPp{)w;Tb
z_|Zeld6AxgpG)It-Oro%(t-TK#oP{WOhD%r*^y6n?td}Ql>Er*Nxv7>AKBB+I1F(@z?7G%&wHW;W!@*>Y2H_U^+
zukGqPKS#cbV@~8|fxeqNEU6x)51Mo@4|*0PRYSG(ZcDXQV-4HbZU?K`ceFZkyM#XGyg&-?fW1T;eY
z((Y`2=4_OoXxF9H4a6C3Un=)|LSOrpD}OI7u2m|w8en{F7}}@V$gq2eBLc1#R6dXN
zjJ(nxSG>|FU4Lb?SzJ=9yg&P{=K`F=1gJ+V)|A`PRr4I{`IsLI
z;t_YCH2!S9;qUkKZBAId^p1n<|wM_m0Wza`G|G-3qc6
zZb3c2T&vxj&g+T|C=-_pJ*~^<&Ae@o_1R}*oWmoW2NxOk(gf`io;}*y5O(&f;~$ua
z`07SakL5dH=STOAdqeRabfL!LSEY2Et3_`9x(MfwzR#yD*iY@;xy&mgH?6YH!LUv;
z?2ZY0WTEI{Jo~RLr#*>&a`p6SeFX1j?3)j14Lio2wSF)c_76FK_CAf9UaKq5rg$M%
zXIeU)n7UriwHWe`x5lZXBzq=&GEqLN@$OIlNz?^3V8+npM#bImzF+1
zybt&Yy?~;9q7WzE%Sjl4bGYuS0!lfc|4vV=*U%T^@4(j|S9SxpWJKTvirZplBo-yR
zTiTv6rFmDo{JC#557$g=JC*3#ope-W=ssm3FiuGHVe_Xu5m=!YZmzNkRN6?H=C1O)5IjbTh{2j68qX~r?|;5csz
zy2R_Y`%`)dgm){+?HK2gU#T*!L+!uT=l!5z#Sfg%>@EG&5$;3Q%w*olFRMNEf67HG
zeh)k^%$)aC4O(tFN;sokK^%WQ%~g1P5AU1Ev>*RqqgHtx{e^j#Whfry_g~ka`HLAM
zZhARP#8;>HiukIxe4o_rb}RM=aYhl0@6?q763ovnbU=f9y;mY0>EL6|?J;qVh`;tK
zH#ptdQ24vJ8^Yh6U7o}7EV
z#yPI91V5TD0#UynfCK-Q!bRZJ-*2iU>}_9jp{Q$&{6o|ciV)$7BfT~F_?d_3ik;zemR*e@-J>I@|T^9
zYC>OW3lRsm!7hO8IKM;SZg&&>MgAZi_jek)pwuW++Y7nxT7hfaPw*Z3i8$ol$-
zKl$7V#_c;H^u4MkaP^kPFz!*CWXAOkwP)PTo(|}HqXz=8(XfV`sD;q-%j)KzzR#oJMMqR^D<`nCO#jkTu?liaIX&!
zJWDu_kW`Md+%0Gz;C`g@-D^&_Wa;$Yu6lU{b<{otLT
z5aQoVh^b8UgmYO8B%E`k4`KVNH>%9~7Ekm=WQ*^gpSC+h`h3QmnnkorqTWqf?~Vpc
zn?~`;j$?K1&^mv#__G5PFKk}56oDOnr!`=kAK_w}o$5vR74wE2?nHh=_tUL)Pr&av
zFnT4Wmn&NLgY=b~xu#hN_zQ{OTq#~UIPahp;hrb1eoO68F?4x4
zeOI3NFtatC57~EVfS+i^Inv018wht~X6Z0mk857JQIYgbE@k-04*Z_v!?Arp=Re=P
zCmM8L`!@UOUdrHm>X7;HJK6X5S4&3Rk#y`p?J4kYUBVB=(>c_XZS7KsRyOE4aS;6V
z@vfiuuEn`e@%AbAHo!k^t(Ccr&S}z9PZsG2y)*8f-5rnfo6%*;7wKhz{qeR|ZUMlh
zS?jJ}0bHQV;sIx%zuc70U!4&@TsEGtJEX2wX%Q0eI|sO^sFq`g0{7hE-QD5P*C2IE
zPdn%ENyK`6FrrN)?B06THohYKWmbNZbLWtMRuP-ubCBQjShK*0
zi&*dLzN}%3`@7wq^}IcS)AzpTdKkFo6DmcJ-!DFSN8Lkr@V#&8Qo$C8FIGf$33dT)
z-?NHdwP0tKZp)lopuH8LXhYTHP
zGUE;6r<0A}m&AF~3xl#(z61AraQME?==j
z;*RgwTMU6?#Qk>EiYmmfKKZau0r;=BUTbI^7hf~0JIPbFh8`kZu|7-Tac~dRW@~ra
zA3SfD)QE6%XWg_Xd1TLRjJxGc4ixeJRJ-9c&WDUkY)7~q6+U(%d4hX89CYLRo8!%S
zod}GMXim6Qizd|}+`HJ{Sc!2iRy!(*aPO9VhGW2ebEppc_0#)b^185jDSeQFaatox
ztZVrOyw1&U=)>njp;NE(I$V0t3XWgGH;Db<&Mx0L-^JMz80R)Jo9)ei3NxY4K31%g
zE`EBfZ`<)bSitc;l$kZJmz5g!V4UCg3taBN^uCO1S9%0IB))I^I-YSI9-w)kZ|*fL
zBakn6*pG3B71}fXeh<}0dfGHp2k8$F3qP~?iHJ+iN7FA=Ym|+Rg}<9#4b=f|&S`{`
zkXPR%^3}L0?&~5d3cu7TU*Mje6!A+geO`ln+0W?35awy)+nm2zj{NVXzicA?<>dlVkEu2fIsXMGfqNYQ7W%fx
z`xzS-f!n`G*b~@S;5ta0VSwNl;%ha^*I(YJ=E(R)%NJtsQ(gk+a6#Z2t-itzXIA?gmf3QTW)ZG-r<72
znMW{R+VlzYj{-)qzGVwJn~yXSU9=7l>0G7jWB3vK3Ts1$L*8o9ksJLXAKyLt
zz$y64Y`^dxo6YdOXzuHMFW_fKl??w(ek34r^;J9A6S`@1%o4~41XV6^5a$3xy-eQi
z0lzVAJOYw%#6it2!~fnciJKDUy!$DCdy5E-{&aB4;^oO9$g9Leso>!
zT?6MCb4$5f88(6cdU~n#IOP9YITbgK_?~R|$@SEp&NZ#_i({W{Wq;@Q@G5w3ZDVIT
zCKumzB>J!Sg1eT{R6&OYz+
zfji~q6j%}EKX_LntO;G6P<(Nu3`Z`2X;mnV9c
z+PV?)F{XE;XpDJ)PQ0iNpI%Blv+nX-8L3m
zL*L())!VU{^ld7C2n!3|??!%{L-V6;vxl>&K7D)5WB+UPVO3`ukC`invtO;W`NUhA
zPct31InjLRI631q;Xioj_9UED#3~$+!oM%~gJVHICA~JIalX6U(p7}3RJ$h66F0qc
z2PwU3mwCKSRK2@#62&_?c1<;eo49X6Z_+nq(4<3z)4AJ$*QG_L=9Hszgt(%~_o@84
z2K71-ZeH1+#|XFlRW8R7=if)OU!9gRoYy(Suk=GXc-J&9mDj%xHQZQVVW?#_v`dZfpC^2hMcb>z!!rzBR&zMg=iy*W48_-PL1|Rynn4t%#ThN-x(KME9|uAhtSv`;!w$0pMJvJTwL`1jB9czkxNoxpLx&#MdA
zA17}8*p%?&Ym0IBYDGiFcMioi8uA`hXE1(K(gpM^+OeUi|EFZ?V8r=@sUvV-H&C2A
zl#Qbo<~VnB5OG@zdcK16vmHBw7`)w?rOSJGd#gK%boioas%8$s)S*s6yrM&tdqw{s$wlBVog?#$AIE!!a
zXM+Nc=@c)6_Z2#C9M2*BLgxC;be^!g#Q1rZz`5PjAG{m>=v-)3uT0>B6DK&)`tLEw
zAdKRN#}!v0Qpf&kWQoh@f4uKdnO^
ze}Ny=^|XCV_C9E^t32uZ;BuVbKX3S&x{dBde%l!Ss)snCh5z&9TZlXCy5t-6!u!~s
zCYSS$ZI|$a^LJPW2*mjy~~=Y0zw(+dE&4|1qo@epkb2Q6*i(
z4^Hh4mv*0rcp~qixgqMcLg#C@Dxm8o^qOsp{dX^;KCgFI)+*g&AJ0488vDp6O}lP|
zzMTCP51zt(ihIPGhtptJ-8wl-#^8S8;r6@PWfZixU&7U8h>vu__cy5ue6^wr8hJu*
zUE{D{#bC$CkQ(Z<+V_EZ#3?b^kczA?_>obUqN2HWcS5VXeQ@
z{Yy$_#~w*IC)ktJEI9?|E|dLSCPpBhsbjGC?9^sxkFXsxr{jDh;p;Oq`X0^G^nJld
zq-%Z{HuQ)6#kw!Pu?+U^%V_cQDB`BYdXwvhqaKw~F5KG-YI6IiT`Fiq+4-sZdvJbH
zZRf97@A2NFak1|kD9)KZFW&>{nZZ@sHG}>R?{}puxaV>&o!iO+ywCH{u<@XJ-Qqfo
zLOXV^@}_?~@T=y$&7|`uorA^SQylDe;_DiUFWRQx&8KsK#T_SAqk9ua3!}>mU}xa6
z@2$ci_n7l@6unn+pQ9~F@rliw`P&J%JACDruHY5lxvum*VzIR+9cY{xEu2}5#@Un?
zZ^zR+>4d8X2jzpVbN49L5p-aez)ti|=4bf9RFT>sR-)g-^=sb7zHo8mF^i1)hJGnwYwigPpS(YjJz(~j52(k`jI9(=B`
z7zd@e2RpQ)1hv!PovD2OdBVpHJ5Iz+_aE-3bn6eBDiGhnphE|;bHZ3%j%#LIxx{+(
zH|*f}%VNxt(u8Z=Z7CK?{0>V(7cA^p?;g+>=eS2-Gn&`U_Ol-_wY57BkAQQcV)4uy
zyO}XH$=uEB>Q(uj%8lYzu_46wv28+my*gLx6sOPkf5)^*t;wA4OhF(#5c;#$Ox~aE
zvZ5b2#qR{!2>o}b;IxwGzO
z;ivAp7O*^C-utKj#=yo`YiHpHr=qo}+&1IIe*4}galTPLS>P6r5V+B$
z1a8O@a=bX_aTf8%*fXd$^c!1>_$7QYh6VW0{7lw6yP3#W(+y(Cw@eoN9C_|AODpo{
zg^P6S<05@z6~XiIKFZTe7yLbQLC=wUgM1K$S2kY`ICgOK^=GJc}gqI6aRJk^SLT*~1;?4!`z
zM7elSPW~I7s{wwN%%|$%Y=6VU8M2;rlA87rK4n8X|Lh3x?i6-_&!;c~{GG#sJ|hC&Nw%9kHvrkkurPZu%}=X-6$ycmB_q&w6B9Qcb7f?rZYwDa%D)PwkbuY-68*X-f}juW>(HRg3F
z(B#Dk%Gc)pN(6jZFCQebo*DB?!9c8I{StXS9d>_VHt|lD=kOp7=d|y|xmdTZ-B|xg
zo06O_))O>u>QkoY&X6LmqJYL3$p8bZcY5
zUmHUcMyoh%EQ>%I{?ETRz5vF%tk33a$q&>DKHHi0ku8F6o}%*?y@}0cRY5%P+j`9?{5@z1oxhy1#m3g+LtYDt_NA(_;mu
zDB{f9S9o1t@Zm-V#ff3AtFpf%&a(D+wQ2?I9Xof0Rdd8~2GtLj%7RB_evnS~O^1I9&;rIKD>>ODX_?u1U)FpfU`gEC1etpFo_w(eZ
zOL(>WL4Gef_`F*j^uOxr!oRl=*ZNcA*T4ro$wvohl%*~;uTwi-x$U!jC)w9!&Tz83
z+18p5sl6_AZg-OW@SxTXxiP?5_Z+^N-qlRD+GnC?gSb9%s$m_x4}R&|t9KK`F@v2N
z?QcIoqfBq?95JC8;;W`351HaSt#`d{_KU{vm>u(csO^Y%A<lvS254*}V8RJOtk8ewdZLOid|A%F}Lh)Yt
zRHa8poDh$NO>@iX4f%+z-Ye!{99>D%%_)KUt=N@Rb3Ab0t;T=ai1*K#r9JM_`<+oY
zFQz$w7E4^&t0CxJm#lU#5Vt=3QnK|Uo)7xN$Rd{JU-q0EhiKeXN-ASb@lv+ijd^1}X
z0P19t8%6u1Gi7)2dDu`3EBpo)-nnJCZ}tZ*KCA1-a-f?}_qU?BZ06EoC28J1yV#!N
zvm>YH52gFEpKtX|>0YYlfdF`RjPqXxyiT0UtF(;fw|<$1Z)u!|M}?mxJ6hc^oka2h
z2MSCHet@HIh(FY~MdeQmX_92N*#P{(E<$EIk{cE2RzbAaA@r
zi|b!?D#B>+J~|XA;dd-=i*p(y`JQ_JV}$ug4>=?JdPZ4P2RQGwqP#g*h2A_*k>BB-
z;9IXkHss5xE%;g&1@FC2l>gjY@b)sk=$tF~-J}NkQ|b5*`NuXB{Q6HKeHe8Y(%(t?
zLH?bAw-`+%-+WKta%4QznwkN6YR?eqBd9rnyDZ~OOfu$YNWRWhfp1J^BHs#12EH4y
z;C&^3QpRg(lE)^B(`gq#_zRNf>pkS7Q%cIOrc=~MJXZsFP6kh{PjWR5RpVRy7%9uy
zETnTW$fr_DrF@tqOev}&UQI7d3zu6sUFB6fR9>a5{%`tK`x{C-?=leQCTs}y{S4)Xri
ze>DwgtoQq^SpP3=F!Md`IU^_bGl90K0oqxr$=@c)BF
z{oa%mb{ND6yZe`*jSJr6+f-vZUe53JTLr%4dunmSh5FQB81LP?Fg+3~(r4B{GSVOI
z5`4o@su1G5jpBZIeuEdh?$+ECeV^p}w(a8ee_rkHB+A(jN70-OJL4DDK*vD-B=}&RP
zw~uq0(79ye#x*R68kHy=P4*kWXI{j&9v%H35BZ2QphdLP%
zZ8>bf6rv~2EDfi4;7IpJybs7*cM}Ug;;UN2V~BQKHG|tbb6pNzL}HwNpBWvAc(VMS
zpADx`To!*Ut}^7O>XwaF5Z8H^T2i~N9?nTUC-OV<=!L`AM^pJ*rhTRJvkfWpv%Ns`
zHM{YR2;zgCdIw4G73id_Lmq%m!oDUc)(N`y#8tR-it6l8fA?{llRR$0G)Tf3U20a4#E8ac&jwjPVwr?I&&VqIR%*KWGWXiPPUyE=S)Xu8kTveKx3B
z{}&M%pxWRiSE(J!PdvIS2KKbdcroM&`o(H-((=-Xx4T}cROun&+h3de^bE#5bI$i`
z;Z%-M*YoYtk?-#E{*ga`zueU9LpG;j&lBd$d2?-`Q!JJLvDc
zo0e-p`+xmrvkLUgp)F%*pH=6YmkGVQ8aKVmkyW63n>laj4eHtV#*bd0<--k&(eKaI
zo_yDz`qQf9`&Fi(d;NO#p>xB#E?fLnQHxVtUvOzyO+f&=(9ql?_3?;dN$9xqDgtDj%rI%y5MvwdQdt)TPA
zg;u9~lM8#!C(!xnu~RnfXvFn(EIJzV4~yRZrR$ggxl(Q
zvoej(z^sf@eL&AIY`uqm581r8J^nxuy@x+nnd0oCo06IkzWJr{wlpq|zcgFv0cyQD
z*^%a#iR&1=TtYi^y`D?@M}Jv`KbeYi|E@m{(sv*)0y2A(eYgBxFQ9b$$)ozyJbvo!
z9{39M!W|p_y_Q93SxYGWVOBcFy|fUy
zar$qQY_>nH&pP%;WvsXIx|w2ckw|*HT5RWav;C9{XL1}K68B=C6X1|=PC8Aj=a;Tx8HIdY>j|75
zzLI{p7~fag`?9{iu5{q2Ri^d9GKzfbI#p#m)?S#!{POD*X&`PrDE#K|18|q%eK(C`
zx*)}Z>GE26jQ2}|TSEGHoK}#&8lHB{cXhbN{JcrRuNGel7U?f;!rg%GJb+^=(Egn%
zpvLcuG(h#C$7KXXJlJP01q=RI>l4g7)1R}2{B{MQZ`L+ZUXJw3X1|2LH@+?M71<0p
z@RssEcEN4I@2V-vpNC%|r}SSKB~;F@Mys(SykdQ`k>hsvvz!)3G!H!~=AA1HS->i2|!cxNp8;ZNnN`tC~n
zpKz+2hcoo6{GX`WPpVyoaca40{W?fHL*+OX6@q`}iw?x=v~2yYTo!}BFYONI0`T`8
z(jHZxS|6ItWDng=5xxdj1nsk&NF0FF^Uu(eem-LaqZ}2MfNPDUNSQPAsVUcaj&%d_Uwqsvv|6
zM7&4EfYM*zCVUMx5a*1q1@GCE48%Gr?W`i}mB0fDailzlNI4?(=q7Do{1>N52h00O>!rob0*6L{+{V%@hj=*am#$n|}Me6PDn-Y?#BrUt^hQ)A(`
zN?fEr%z$&^$puGAe!E4hheYc%7}JHQrkXCl6D?&Rey3{6OE+{e&Wo&Payfkx2Jk*=
z-Ho0(B)_xU9W97{$#}-;Ic0@E35jite;{thSEQ8Q=q$n&@Dm$~_tyiaAUovImK9ll
z!(%T|A;kX|Vp-qT@)sCymhHeaq39rPXV)1~T<(^WmGFyCc(+jM+z^U)mOXX*MDdqt
z#~%2@>iByfj_Gl9-h26M6OM~qdiIDV{i(A?(DN%=?uw)NxL+
z*=Z%w@M`w!iFUnd8a_3{gy_4hh=xSJ-6)F-Sop`+
zFR>v;KL++cNwmW1)ayjIb^F2j53D$7M(3&}KDYAR3i*nh#kv$Hw_P@`v>)jGHP2qr
zJCvVpKfmSvKkU7ASeDDzHcS|Rg$g2K5EhDntq2Z?7^tX7*oX-RVk?MlyRf@J#TFYy
zF%Vm^;1&~6K}i8AMFjoMHFM3ke+Td5yZ8Hi@z=v~?0vdsty!~XO|6-0uImPFo}O<{
z@)Hd!gcg7{&$o)Ab6k_jZ=O@U_O@PVbczP2bxX?6EGG_HFyGzdUORe75D}{(eCBgN;4t-e=31
zue+0-uPT?IfiYkCsq?t{`{p1Q%IDiVI+NN{(`C&h;t%XM*PHBk*>&kEvU5`%t1h&E
zwz!e9h%o2%g}KD}7Ig1Ke5;(JyGcIua##Lc
zvTEiF`TfvUTGx7${_C4<^XWSyJwogaX+QhkPg|YR_eSoSlnxFc?B~To{u_pd`y_TYY4w*{WLfPpP)j&nUc=OO-}e6
z%vg9|W?dSlYMjq+d_U_O<-cbgK=vQL_@N`Ee-2vB<;G}<^PIM)5_^-Lx6UK^bH7xd
z0hoCa=X&~N5dCPScuo*du_Mp#nG?w6%yTeYkUy)%
zRxVepxrE2>zHv08V;?hfGcWgJHMkFYc5BD-`NpsTMSj>fg^T$*brl7&*q;x1%k^C`
zsmserGvCE&%{#Pm;yJQrA*XM57W43cdK5_GJp8r^$MLpe{c%&i8*E$^YGB98O*nP|
zj_E7vR~c^;QNCLzX(2(}-d4y}Z8e?K*G7qR!z2qaf44aZH$sm8oDA}LZhgY(%lIlg
zrT2OPbx6NdAoR?iDCh%?#5@}pDQKPJ5Qn^vp`btVfCu;aHe-54`c-|wZ>;chk_A1P
z2cCG3)luZT=PU5{XTdKbGg1CPW&IrdNUU$o6gy@pad!}U4asdGH_|6J*4ZI%MEW{16a4MO2ERLzp!+NIC~798Pg3&BxV^o?4^`wlDtbI=nM8gW
zw_j8EjEf#jq&-C)@YGTkikUeA187Zt2hRNKqs8
zGbu0od7_doznlF}_PkQ$IY8f^%jX}gl;2sIH^Pq?q4%`Hr`4Y9vQqS2QO<=1y%p)!
znYLWMp&l(Dh{G?Fr*ZE)y%agIPp`rFIi-mPITH5)LWLb#YBXcw{TlDqJm0DIwD95l
zRk=6T!G;Wl9aV+=sR?8t-fwQ7!0{2D4iJ5(hTu2jAELs2keM{2<6cxUEmU}JWI~R?
zd5I}G9_s+2CZhG+#JOFxv3NdTZI>=;jPrQ?-Pl6oIYuWR9NgjD)40+p!nX^W&7gdm
zalLu|w)OR30L~jrR>Cv5&;IHF6ypAY=54IBc+dOc172S7jP;^CJG#Jz=NHq1xL*qM
zj5+PeSKQNs-0_qpDg46S~=Wz@>e}H4w-D7Yt
z`h&3tp1|RKb=}o-DE@5dT;GN8eW>>aO7Es?$>UY4Ei=$iIESCy7()f;ciX?CE92hv
zP+K0CS-PuZ2xA{LMYLzTp(}X&S6Z-@#OwWA=LffQcT@TT8up87Z7@aQ9@01szVGe6NgV?e
z`|p?wTwgQOd@euoVGg(B=62yfFCQ(g@A;4ccwr0o^X82nP5E>~#`FF9;471PxrYKB
z`MzP>?IJ
z>+v*h_oWIid4HUfJC4(>KW^gkf%euMlZFoAcw8r%+c{`+0ryMG;EG(&=99}f?m6C%
zqsOT+22_6UElJ#OKM$_t^?%x-8Q0Ul`x!1*ZC7`0XSiNVp59yiI!`aEPXQ&yWj#7T
z#=W;Qceq_MZ$9PpV^^2~evT_mG}veBig8=GZyZk#r_Zf|zkF;F=kIrixuAo$zzoPA
zNVMR1qP7@!Tl8Qi_#sV@41UZlkf5D*cjR=Q4z^!N|5i*Z75eK*F+UC7D&*>@QaRs*
z+y}WJ8g7tt8v_#bF9$(C%=Y90EUC6g*yvYCNz2HwL_aHr6SwF_#5prEdiu5xtgj}MsuEi^H``{jeKboqE
z{Ed|Sm=rkeMec)raTqm(G)f5Nd80p(pu3Pb^7mHkC{yUE3crOypH}EqG`%C=8bzPX
ze_7#kHPB~H*PXRtb|GOyl
zNc@u?8g*1}X(#U{pk;k!eo6mHk4%^H5@q_I@So%)U)m?_km=>=Kb0fXWj;y&3IDB}
zEdNh>WcvT9K3U&C<(GVkQtnSE+x1U;nf`C}{o8t%*Y|JzAno}-#eb{+-`X$r$@xOo
zhfkvz=ZRbZ-|^y%zvq)b*)8X>Kk;SyKg}0XPL5w`r%ac0`FeIx=6}Ad!Tydu3CB9h
zxz;BI7f%xDDg1?QJQpJp@~gPRa4xCXJy}oaEm$M)1_8?9YJlx|L*bl|9)lqNiD3fG
zmHbr{{#(WFW4yw+f6f;;Ja;}$3oy=`7ua*X-{%W#TXccb`krJcp1=K|hNAOxB%|Cy
z{z5vQPhTU2ct0nfo>=4eS9}G(eK=JF=i6EW&3e#|8}AMLCWl~sR_ecLm5|pzCvYa7
zGLRmPb+q!~x#Luk9&k^TA4}mc@<;F=PR2R-VjjR-R8i3-?rF5%xm9mc580JrhKHH{qEm
zw`8fXv)v1E&pLXhc<$EgtI*prg&dE25Wdut@SgJg+8lef5ojAo5ee?y^%it
ztVHC0=7Mbx^vC!oaCD4};dZ`Rx`pQ(ajrg3|5AZ&G-IFob|uH;P~p#mYR@=r;Vt}r
zvV-V1>sE?>v$p}AoML}JMA$bKPus|zZTw&b?-w@`>96_Y;!6=jmLJ%Y1PT(Cw*cCv&_7LHcVx5^^(Y(SXBu
z^pv=GB|lKcJ(KIA+~tNs&eL3!vxwXZ{fX}dKbi-YxEH9z&lXDjpY=iTQFNFC3s9gj{1QA@@UxyZiEh81GjpdWR_ILvyKms80a7
z3rM>-?r(*@zz?o*|Bcw-^WzQJb4j6_D)QeIJGgn^pI7+U)Ny@J6?!c-1N6xI>5UaW
zHm5wDyNmK;cS&-*8G$nHmhr#T`%tlyq)DDe1KH1~LD1w4qD>W~-b6GP2cMJwLDQ~{
z^w20Gl>G9@w^zs~^_AyKx(@%x-{ncWxY^Lp-T%KSN9vLC|74ex=jQ(ly;A<4P}(W&
zD6jwD+9m7%Z_AVI_b2&(Dp%TB-u^%7lj$;_r2mA{AAjP@^na`G-^#trGU|8}zZYVE
z{kQvPZM=V)I^E_Z-uu(7zacFg_j=XN#1Ej~8@ZDhyJOP}jPK0g+avLQ+%5mLZ)4DJ
zoB5wk>5cobX1|i7DM}yuJIgJz`7ne*sgsn?m1VvYMP1fUaz*xQ0d}5#mkQI)+V|*54H(uba4mn6%O89
zX8acCk_Ywt817BKytiS8AMO=Le~t2Xs)zIQciGjd+Ga3A8~vZRPGa40s9N&U5c5>i
zrgzNnUCPs5^PU8lB2F}PHF@zWll?f+xOJPn95y&{WrjQMU!Uoa9WoW)rL4ZEMJ;34
z=d$#;Zqw&EtWn$Kyo`9fPxn!6MOE1U({$pVhN$O
zfbU3n&T^W(3jD)s+z0GLy{q5<(cKSzN$k`->IP!dkIqn(XZ?4@k
z9eUgP_KCj-dSJceF^Ayq^@kJO7`_Lt+qcHmMeR}FLp%HT?t^orCkC(1;d|Y)ihrzL
zXn}Y)HS*yeXS^qR>8xwt-Z-atuIF3R1kb$#yiR&1;GWSvHy6AUCbJE%V
znZ`>xjK%w%HkyalK>K!T*Jdr9_l{X&_oFrLUB5f^rDiIA
z=kUekEkzMs5kGZo5;xuwzjt`j{aP{J^Q^J>Oouk3vsgQ8Q?p>-FRbUoAICc6XS04A
zbMr^1>-2A0Uxb}rneV<8wS?R|_wDsj?|U^W
z?CNcc=gL=3+#QYYrkFqW+T*!1xg8otrZVGHJmJvBUs
z_PJbW8Pg_%9o}GaW?t0{mT>H*ebC8twyaIk<>Ef+?CAV{3r?7&v&4{U>xUjnW!Go-
zXgl{-D(gCF_^oWyGBST-@9+l$&
zzb0y?>)?AVlXNOPOYD#P;)}x`nIkz#aP`S^KX*#0{UE3ICc#TIDJ`KG3s#kAKqJDJfeosBcN
z`*47EIy0-iyy~eZY0P?8l>O!Vc&C;>-R1I=ase{`7w?Ciin;6rh7U&
z99CMTbWb`PIPsUh@9A`A>vYq79e(d}?tY_2EzwVkyA=&;iE-|?F~4?*7QRzbv3uBS
zyiaiP@ZKbs4LE;yZuRmL#={TG_8rf3!|!qSdG{j%{VaLsv4vBR9(`oojg>z$*{%0?
z^SY>iVWBQX{!PQO+0WV&GKQVYVc*w}S#_uh;+5T}tseQM8|&IwOr&)(VSY_6Hk
z7WaO4;M$FJw!XgSaG&ZK%0*B6&(yQQb0-R)u*S4?9qG%8he
z3{7PZo4fR(-yg!5%ud0)(Q$o2trwVwzG`eAZ&nvrSbFyf=Fw4W>I6JPyFENtq_YJ5
z%6Yx<7gam@{;#)BTa1rsW48x7p?xmCsJ(Gz3!K||$6o&MF^3Jc+p+TUkR0~qKUwhQMJ^Y57OD--1yG5qSM*cW~HH*nq;s&gIgHwWf}M`
z^9-H1>#1z!=oigGTR)xYjB&71pOVhjm0VhvKQx`KIRAO^A=vxi`+(w9
ze1G`!y8Bu6n&4i*P(#=H4RDTL;q%mN)a&zrv6?3_U;Q$8exy6T%TxTqz&X4#-b+60
zq;>`Mc^9-aaWweGk7qQl13OFhX8Xn`WwS=Qecv@alg-BNbC{i9J%=rM7;+_eYz}j1
zZ*c82=BX_+dimRbOlR>fna99rHE-~~b=tKno#F4j=Z=+D0^d30pzC%c{2s=T^QU$-zTTmY0hz2u)i1R#XJ)dkP3&$p4$Wd=iLWzG>3(6=qBVL4_y59D&zv2vhd5?i
zhg#bMG2eve6(;AQUKT@A4`^UL8Gmxsw%hnFL9IS*w*6&ik;1vp*9dpJ%JQ|F-_
zKHb$>WDd+}G-q5#^v@7I$C?$^ePIn%k!N*>X0z@g<82Ow=CFRrbKZOT;`dY@4Y|=Y
z9QSiR^*Wn-Ba4lg^19;j%q;e3zwi5YOTMr-9zLG;2V}FFKGs=J(J#{+j5~jH{KDFg
zo*STX<_o*o>S?Z-Nj4kWbFk{LS2lYPJv-v&#%z}GJtZ^=^M6DG^LPXN4vAK~q>oyN
zSGKrzn|uIq&UTl{W=}Dn1*>1PvcPz^vFTiV1h}wt-NFutE4NoH`tSn$7N>KkwnV-u
zl_%t!!}qi@CZ4`s;Us;hVD|Zpr}$mC;7!d>VE$hFCL?NIWvruDzNrph&t_(QmVDDN
z&SC0pTD&NHS(K@vB
z9=-(YRp71Z+wdLJ^Bn`{pM@Qpm!1p@MO-#^XV#;qD5qYmO_x^H@SfJe%}sLbL}U4zBJ#Y6IdzRp$45v(T?5?G8NH8S&(B(~?>P@!ivDrh!Es$p5`-i;8_}qMZ!h
z=g)79b!N);DVOm*j+1N3-p6D99^U)dMNin-X;Q!Wo!n4P(zqpcZ{mJv`lOV{7I{ql
zsB!ML@p&vJ{9Q%_^zQmNxoigdePHN`MtBF2m5p!O{@rIhZ^|_s(x_S};q^=NO3Xk!SXaGARHYLb
zAGLH-^mgMsVoA-iub3xpe;M%A5_Znq{%dCt;+2kG<~sSIxEIlAw|<*(d92UJ^R4vL
z^O${~zmg8a&RH$=TW>j>$9jCz(9UX@&-_izerHX~XIr$aRfTKvS(nn>EA@2qnB|rM
z;h!4jvDq(J>TH`lc5XxbEU)Q#?ESfo7MH{G*aE}YAGW>V$HA&gZwBIdYsATg-B;jz
zbA9tR8u*S{Q9Hd~>iCYw)jKYiORB&g9reJ`4-rQn7^H7G0rwrEZq6$hg?lP5SD$OU
z1@HNquMOJzDUap+_2lJ~$9e3|l@AN`=H)Tput)Q+mtdcfGHm3|V>oANTQPRR`g~SA
ztJ8w`_@Y2
zO2osS74q05n<|ZsqVw3$ms(K^4(GEuv8$?HHY{Ko%PxLw7gxa2qqj8ksanYDIAuFp
z`4zH7>F)--I#$Rs@5csrx>d*q^}Ts5znizUg+
z1F#+r7~AbgeLUALO&e~~3j3WA=BM$_XQ-VwKxyE}9#VAZP~O*^!qfTgVUtJ3~+0bBL>
zO`T=kvF~#FX?A5K&Tp@-tNLXW<{`87gL)3a`Zj0#P+KRgCzW=eio$n^E6*8y$OGeH
zjc@&kn`IXK{4sd_-1rsr2?kZ~jUO>e|;
z{g&34Yl?m~^2`If;6hgALLDo&bA>FeNrGd|2Zd~Ahtf|c91GcmFI5(_cw4}{w6m@1
zd_ueJ4xO>GsDQmHu70<9SOGIVI%#A7H3dxVSZt7SUIFtS|8#f~>^$E2m&PNM`~38y
zk716JO}J(){|YbRXBy>Wl1;2$x#pFU~puV!y>KmYsG
z52cvLZXNs8XFA6DrhQ%aA45A9ZJK3x2lJHsZuKjR5I>)>ni5tE->;c)uY%@Fls|lL
z(bU1P@2Xep-F^)U@%wH`J{JuN+0DiqcZ{uA$egF1xV`>j0b3fjBYfWV0=CO&)s={A
z1+4#iXYD6Cg)HUzI>+E)h3sQR|9a!Fk2R?Ktj&v|_)b*YWlsIE@7Uh<(6g3U-(GI$
zUaEulhAVA3=x1Ao?@~I2r`ltEXswEhG#QP3oqpPSvkJJE^|C>ywpi~rG_L3I1@rL;
z!`2u4u>Wc^DaG3`2=U~|`-5wB$zx6zPN@nt@>z^etm9&hLiXZi;g{Z>3fYBty$-cs
z6|#*JAJ3ZVS;R_{Q^NO*EMlWt~TF9)2RX*>5_FLYqZEjL(0c)M|b&koL0%lO}
z!F8wL0@m*Q{eHKx?nHEN^=%*av%`y2YLgr>&o?a@`)Ul%7arKRsWBeEBk-t}-$shD8`-etw=%Qoldet+@qhPV?{`siKxUmR8DVl_v+r4qTVVtWTfv
z8~Y9SF=ocpsk#GxvbWgW-VgVdwP%bE_bS3ZXU-Ng*jr}PC1iwc5nF6l+M?6XLe{;q
z!Tb3Mg{(nW%U1S(6|i}38cAxK3-G%DthZH20UO)7$@aaE3YeK?z`EZ#1V5EDz2mR(3+ChZyqd=c+8{nWn-$=m
zaRv8@N1r+x{sQZuW6QwgFX(rnN$b8(#6Bw_tH&IShohCBww;Q3J8Y-pV0w>xfotz5
zLwv`x;j8Fetv7|tT$gDrI$6lv_Qk)=98k!Lt~!}qEG%FX=8kRW^Q3@%-*m>v0OL5o
z^ktxPwL-S?T-2z#frTu{;%>ipp9|TQ0bBRHE-7S96Z>n8!oIGk>yQyM(XS>ixgGLy
z0N!W*+3x34>~mKRG+7xjnIjh99$gd9)9cOf{>$md
zzD2uoSlR66ulHcQ%&2hU@Iu6wvoEY!7B~X$S+8&5R`nOw`D2%kZ-~fapS$^{8e)Ao
z_Hy;L3;Km@xS4DJl&gho)1@T~l6n`h&Eca)E%GR0HNJj){y49Y)oqgYvBpKrYwr44
z&oQ1hu~Rya0t=XL&9ugw))ugwsbg)u5kJ&lqjuKc2K%W;rK1<4Ki7Iuv($AK&cSle
zb}zX#c0Uhc(wG1oNgku&bqxAz`ofc=2+k#p`R1{N{Xi^tU;*c7o7ep=u2
zsur=+tJ_&D^1yfqty^z^RslP2`rRh|NCB&5ob;;x?gG}h`O&A%G7H!}qt5R0hZHc6
zV?TmZHsHC)fb-(OETsM!nnzyDuJjXy9@Q`c3mGTVE)ulMN5<7RlZNf*ys)$oy|
z%J>g|;=mT5$g7KlJ%Ut|s0luCY8o>&{%8Uv;?YyP-~2g(np#Dq5?@1uf9M31P@t)P
zW2eyg;WiZ~|7s~(h?iMw-?*x3#+;?eH9?o3_mpvJwNXhru@@TD)MmF)QyakL{@?ze
zhGMUpnmEz@`~RY(|Jt8zjUOw9{Qu|AeIvy)jQ_4b$(Pcf{`mL*H~##u{neU(4W3-~
zU-R4fKOyG-{QZvy{>KCVWVX+%#FRc}QGQZR(v6aG?<^8E1QeXMS#OzM9%Z=_%k%$4
z|4Fa(n?z|}c|Fozea=v^-U{j|xKF`tT!9MbXY{9H&I;C3aJ7Qhxj`z{Pod2fl=AO+
zfhv~C8(PIQ75;DqYbY3@;44MnONGu>^6yvZor<1reBi3s4hLeWSXW-Kip49mVHssq
zF}+CwxAB3iVsCi|SFu^#Q7TqYQ{;dCNXY3n7W7L#FjXw4xuAa!5mGA;;t+wDS+)h#3=y?PJoNh%iXOg&J=(r1vPRLpb)0Juu2|F&1ugH=v@&c%A5S@7YtFc&51=EZ6aI++RxK!tNF!gx
zwha~XX-kCsxC(|meORxi99?sS-7Ag=`D?$0{EUqkIRB#WY>sQU*W$QmL3fUcottnx
zV4A?~T>fJ^3{kPFcHMbBYHA$ee9b0FoHn`(68wuw`_7@_L4V*Aii!n7is+o#^bhnq
z-l0`2wWmmrQu@z1MQ-W@QE&4@fZVu#l^?{O3R_v7i;>`s
z<;tU!m-Uc(%A>4r`Sy`=GF_G{QSv3q^8Ts5(x1|PiPC;qp42P#Njb@v<;nJ!w3L_m
zCCd1^eEE_uQOe1FEz>1RJ9)T=eqY|M^7cu&W?aGlI&Vn5Qcl)Grk6)K5ByvE{?zZK
zpZY6y)lsmog8B;XmNd`scbx66q|0@Qhb!>M3`K5~!Z%ZJu!2Xq16AxduYihuQ)pj>
z|5KqWa7U_Gl%lVhLLXN61`6GfH>iqjRN^Bw3b$Z~A$PcnIXDU0_bX+@`ahFAtYS0y
z!i#l=!ZOfVO5F51UC23K5OOOT3I1Ir9`I7=0f8dD(HJsB#jXw%^xB33M=G@K9ieaI
ze4(%RWkC;LEaXye2)ZZVpr}|EKM{}js7XB=amZ0}fQo&bL_Jf*ju!OecwJ)?M~xVA
zq>4qSP)}5`3s!>mW+LviUriN5{LxJ4{oS7&reaq@gx=F@j5&RDR|Agk@~OwbZ_$E3
zsiP=&S;zpMzHhDvM@RGF9KFlPAu4A4OYj{BUPOgdY|hDT9ACCQ!TEO4$2q-W*iTNE
ze)@t6s@U>v5t#B&uMmoKRBXHNHqLjv6wBqUR%>(Y;h2jGsF+F9v2eJGJzGTqsft~y
zMG=FFZFuAj2Otjnp@IWcY*9Um5LL`Ka3AOE?TP2|0WYU;e$a-As368~j4P%z6|2*=
zDd%_1c*5nEHu3C5>G3B$c={?2(_o@U_BvUM@U3l@Ct>P{`P&I=vLgM~z_~UrG36qz
zzZ=i>7)Cnt{6}ve<@A_0UpRfnz9)vfiY={BkK5TG@-(kkg28mopS12emv3X#4;v;G
zThqQWw>RwQ2{fRJH4U22pq@c$y!?JsOgP>R6#d6MTlAk$H_`7e
zg-zr9X&*$t^f=Oz(<(Q(0eYHG0066;6YbfeZzWDgzM;@l#TMk!%!mH69@PMztmMJl
zF`=vxuixRGG;meya$bMVFIWUOfbKb}9`85Cy6t$qGIv(xayyR%a(?}Ht2hqYC+sYn
z4HEfUDe-+R9!RU$G-aQ({}P%3{D!xM9XEyw`3dy})+2Ml@4i8dv;41uF3c73i9-Zk
zO_?7S?WPF`?ZL!&oWE1p;Wk#tzgG5P&&FhNyM49MY>*#bL@TU{%}){eB035A!=w)B
zpU6z$K>{GHULs@BANq`R}c6?v4gY{
zUm{NiO}&xQV-<8!kXPqF_|&^dt`aXu#rQDC7{z;Cz5ppuFlJ%5&B+7Jjdi?!ej`W9=lSsXvS)P=W^(k+!d6JfTB`r~=%X&&$@+C@rk}py6
zW&6walj*Ykq<(34`EeupGM|)_dS!X#6kN_5K*a_tG+!3McUS1E+&~rEqpTMc9Z-I61tp!Y&~0o4zYDL3img`iw^8T|
z+`*WS4TM}ZzOhlU2xXnHMD#-X)zt-Vb0p8G*aRh>pW0NU*F>~J=_N{hZ+M4VM8*0v
z5a|KRIv=L22cxP~TI~e*EB6#j>NwP!-$1lPZ9H%WlEnxI@r|
zUP8~#HPrGdcBV2pNX71NBZuN#d$x$5pVp@pO2w8nP2m{)TcodCLn9ON4Mf~@Wb6P=
zmsNeuu_&1wsbc%S3HdG?X{NRGZVmAgg=9o8f636Y&
z^x%0FJAZH@r*+b3=cZy##?v{5J$2xE&TVc62jYCu;{#83D6Y%XL*q43QTWU77Fr1V
zht0L>5^l&J&h?%fuf^rmX5Qv>=H3dJl2xo;yL2?1isf~R;P~+`A%9(a4W~C-EkXs+
z&Zf&boi-qXmov$0G^afW#&bHxdMY9yw5#?kosTKX$Xd~x8YmnT^RG{(pEhEwravqqK#5^S0eo0sj*i(VC$xnN?HLmt?%!l^iK7|Y7<8Gv~nX1
zdE&f*FwLRDLTli`gf`g0so3Pp1JP4d>~7aK)d|~AvzbQez3*t^5toV`OLs1!^ts=H
zODKJC)ZNKMzqaVMhRU6DvE;H9P~EAmHR;`4)TWsDjZ)hNk$la&GZI<=?*-P5?E)M&
z$H%)NaO1DztB7uXu-0I6px;uxH&ub%E~&!x+Ns#{arU7U-%oM<)sytSUvg*`VYKU=
zYJ`y`gZmTixp?jn*?CR<+Hs=KZEAg%%B_2GGne#brjW^RhT=~OSi#u72%4G9{otJR^&F0wiB27ptb~6Co7WnQq
zt+*;SW|NR#8(y31TbCl9f2r^8#rZDXg?x%i=zltBF(wwY@1|Jpuda_y@O=IT3%UID
z$2B=lo*Rw=u`hBT%g15H<#*h#NyaBRzn?qJ;IQYsuyZ;OWU&8M=5K#+i9ZLq2~Rw2
z#Zhw>iUU1kqDXI7OQ4N1Uuzu|^VrBxQlMf%`~eQ^oi5}q=m>NUp%WpT6BW{m1V3a8
zJ8rLpnV?-~i~Q4uZRB#UG2|9TG){Nc)Z6<@X>PhO;a0P3Dt+l63iUC7%x$=;L7y
zkT*M!tN$-h)?51LPja$dWclUgWcku>(mvh{C|Ba2$#IOTI+uH`y+-|H%AOPPPNh-qaudRIaSA%vZi#nQy3~N7h^R
zE1BP2;n(09u+A$q4~J1cMR{b$1O?CWf>cb7Lq5HMkHw$M(QN{fk5|TR4I)9iDOg>>
zXa&1-M__$a=tw1hoKHjDD&A>L&0B6%N3Qcq}%d}V!hx4
z8gUyp0OP`rT1>_6DD&Gs?ih?eGok0apTKV+(H4*6-`U%{?FDR(kJP&GF
z6$_x_XkgQ|f?pag=u#7*$K(yoc!)0yMEM?FMf&?$#H8t*jx5&T*x*ti$4L)FoKv}ND-?t{YQ-0xK4a5JuK$hpYMy_v
zg&F5xm_JIN15YN|d;iz-OKXy6>k%o%(Xmcl$XxHRkFDVZF
zVmGuO=`+E1K8Rm;P-#6(9ygz-zUW!k9MBVvH^o7?maKj3;_D(yZ
z^>pSZQ9PSI)!EtwsP8t{jpE;jqbfgW34E8@el6LxfAi$^y1>*|s=(QZ+ea_hRM`ad
z*b(V_$quimhHpr3(C&8|D**S-FMd)Faw(oGGh9KJ4m#sZ`-7xSGqfmfu3{6^W;t-A
zzh#_N8~A%)FaJ_t>lN0u9g*I@B&CiSu=HHX{{2X=Kj7-vJJ7f0{{2^Nfjj5mS8|YU
z8^u0Mhg{2`*c7diREdeZ=KQSX9DDI*Q>OpC-Qk!NOinxk8||a
zSGHa+j%UXP+!#_B>F=x?ziJBkGg}|5tPku{+ON9<@IrCLkVe3V4ht%2LC>B6D|^-g
zy{_f_
zyDY9I`_}5LG5)i;0kWm;d$b?f~!}N
zwi#V(xgF5%YL!T8=Nz?EJjuX*ZhG@+#CKlmUy)EF*}=^eSgH!>N&AMIQTL`&KltVP
z^cu;BTYHWnxdTq0{B{5z>7EH04D>x!V+!r_P8nAK=&?A35|hA&sj9>3M?5M%Jq643gqK_|8feE(BHO==Cp~~V@}7_NyCN$
z=QO7rvE-ujD<$Tk!T?^Mub=%n-MC21r@bOX
zzUl=?hFsdob9|nfx8ee?*NTLJJl)aNlw#WC4n1U1lDoDGzz|MU$`G=6ei@?tP%XW1%i(JNIMo4+jNi?
zSd?ch{I)BS48VS2mdJmj@i#7)7Afvu&Mgz`)+!6J?tM8R^i5OtD;pKN*HUrN!?!~~
z5!hnh@f7JN$el=+`=ZxGf?iDOfZcM0URtGy)+2L(1&aP)c#r7qW%LikA8vw9X(lk1
zniuJ1<^n5M67@6I6Y?FXI^Z`PB=`-n>JUAg07$hYhp>
zO(O+}UzGft6@4QVyJM8`Uu&9x7z
zaId*c=yBp59rNBtL7(yya`~?XZLnDANw`HB5$A89imBL;gCf0OK2-?u@(F4o6+7)u
z9>DnBBlwj+3q3pLAQ|cR97Ou#HPbmAdsWPXH)DjJnsn^~atRI0IEKx}at8V@TK0fO
z)>L7{9rML}-D8_8r$astgJQ(Due>;IUP+hZ{#`yCdo3Ho^-R+~2v6d^&5lDnea>QX
zD4zcxYr*rkk6DAEp<>OnL_Chq=5e`8-Z~tEY-nU*o(*~o2I7{ff5)Cd
zSPPsmD~GSk`@>H2bzCFs=?*U6V^Bqke~v%W=jB^`G4-bOJ~J(nDgVwO%O4c)1Qd-*
zrZ{GCZp<^{ci8j@D-_C!&0IwEu7ihG5#4B9#uCD9-aoY{-Kom5YJ^4hhyEh8{Mn%s
z#S2CwE_qWtW$muR>pAG_kNw1Vc>iJ&rz%d-ywBdV
zu9o}k81T-V4_9xs#=iY@g?a&?=Y}Orr?`L7ki7BUK--5EZ*)SrNp02@y+pYtExir@
zg1ml~d9)egK&Cojstf)eli?Ll+F-xyVX{7`G4ux9oAwU!E)g1+RzZ)!sBkt7;
z-gPp;8S(0`SrrBb#xs*cS`B6*|I5KA3T=R;iF3A6ydBeXX;c;9w)?*~I{|BDE!of;
z<=VL>Bo#nzTK=bo{r}s~gFMjT|S>x}<
zvuXEZ?&y`pv+)+{Ee3SOxai*evN!m1PM4K!w?erqFD{{WiVQdCLhXF@+*S-U?9Yy9
zUZH-mW3@#v(cM2h+RT7$7Ckkj{np&$hM|;xW5!DiP_$dyraaDdyx;!35on`sI}I#=
zss2GvNPoeIRf|YI^3{;zgbP!jg?#{awF(If6Qr6%=aBtbw?qhfj!1_;DC0!zn_oO
zrO}r;{n~xd53aXHH$Lvxn6KpN_M5l!@oqGJ1fNf`tasj_{1-#&3?qz44a5wL{!#ph
zkM}!v81A6ml9Saa9#5OQ{w9sDuAAPjqx44i<{^SYe^f8TQi%QCj&>!~zhnAmeI}eY
zzde=)Jm-5>84nupJgv)a%;50Phk1NlemR4%9`wzp6riDBG@(cU_0^*Q4&!ah9WHmR
z$1+Ykn9zwb>TN?SJkEQ1QG@`$&g{YC>>AJDe$dgi#rzsl8PgHy8=jb+fCVF`4f&5x2=wW0I@CE^JsY1W~LjS=0K+OZBRwOi0_`4OlV-Hc!
zU3~<PU$pl_9%$fs^#`H<2rZkC%i#|ma8Hd$&^$Ykyg@WC7Rb#3
z%I7>(I`O$VKwf>IOn*ToE(-Fn2{iRSN{_E8usc_Pc}YpX
zruZdJp{3oO6`BL`%XYZ;2RUhvj6*9Ze&%5q@{62o_kIMUb2aE-6#Tx{1WFtY(4-nYwt{5MuX}Ul@tp_6Ap^u>5
zvjzWQGpYg3J4Xushw%cxDC4lFiVVfLx0}S#yn@JY(v2K~@wA32g!RFi3`HChJcpyr
zDRKbfD&;=xl0j4<@OO&wU+;7^D8Tx>fhvIg@qBUw&JXX2`1aAXQWS*#zPJco1?!jX
zO3tsh@I6m&R`n96yI5s%erXRU>=?0c_(1_Po@;g%^MzVPp+ELzL!MsL#L9&Dev7co
z0c!&cJ;xZu
zy3!*35ycfTqc#pDzV`fDa2(<+J2h_KlGX8;DG~3C+OU^!(asY?2yNZPd3r?B%H~AB
z8?@p*;jd2WeQ2HPp=qy9^o_*kTM3;{j$A|K4(w7GMR7p7*1a2qs#5n3gmwF>VZlW_
zWwIB6G4ijdQ;p71m*3PLeGBV$=f(A&klclDQ;yO(T}JH2mfe95Mt6&S3jW=k(R(rx
z$2{*-xu7=EbBu~7+5>xqm&B5ub%)19QM~vuBfTgUaT;rJG_nrprdu5=(Rp6dGph+E
zz}45EV+z4|UOutydfw{^-}QCk%A-LVUa7`=?Uy!#AH*U4_2un`d~|gFO~o9sONQEuTYmi+&(1y@lxy%Vo@s++7pv(37h
zZEKgrGrL1whS$=-Ie2Dt)wZxVcG47=+weon#bq_(Q2$3Q-wv9d8_!PpF8X;mIi5ZK
zb=v1(H~8DrsK^ie6MJsP+B8+ML&XnGN{*pl6c-yjdl=6~hm|aLLp$hfoKn22JKFc&
zleSvsX!pI3x^=k!oT-oXKQ+J(@&2@&$#EI+Z0GaUHdhd@*NJvWaW}+0Z)fAS%@;gp
z`>IDL?<~PSt#0g@J~h!U&KF&_U5RI1mg=fy9F1pJ#!e2O2|sMT=TUqC*uEt0BDJHX
zsk#>Rvk&S#JRw`Kxe{By|j1O!_G&&HYd?I{{A^^
zCXM4G*4@%*{yCpvV@mTzUhc%UpHa^~=dX7B2Kupo@})x5v#O?pp9koZ`e*n!opSfN
zC)p7>b_XVS#D6IiIAY&clf-#xAQe0VB$
zVO^pR)-LrRoZE4Ox*_mk&?G0?|JvRT=JRpmgwuTAJ=d?^L6RGD>>1Y^oS)9ug@gXx
zxIJt8i}-ft8nLb%o~FU&HrBt&UeOQ;9
z#Qk^R)I;v~UNyve98?j@7vze5BRg>HAOJ8&U*LtB6#2m~GlZNOEw7+k`U?I|T84>!
z*Oy~Z7>WUHxKN~DJ1XSDm3yO!xfJlLSllb2FGu0O;4c!OA3KTk4p}|8|E8^|#^t|V
z7j{%h5`I3LD9*9wU{xo5n;}hjf&loDiU)2a0Di45&{&cCprljl5dWb(o-h8K!yRW2o
zP~>`2*a!YM3KxKhbA%k8KVDst`u<}J0jA{
z)(g4HdV)5oE&9n{E4K3dr#(cvbFz>dzMd+Cc+ZD+23Y@H4|DuAr*g{lRG$z4JUyR5xXHNbW+8$ZxKJV-e752D&f+
z^W}60F86tSsXo#9U!!^cpq*Q}+?QR{vk(_o-p|MJfhOm;T-1c)T+VqNIUMWrrS4p=
zi94?Cf-bvlq(`{XFs%_`@{Q;Lg!T@9@i^~F@-{1)_s48s0gvO{>0okw;=5?=8%DV4
zj5eM;A-=H>#t^{zsN?mv6L7-9Yyc%zMVgA*vc!K1H)K6YZ>&1&R
zbb$l=`Lc&Y3DY*zz?V3%u3UZ}Lvle;tuR#4-y5H|C+ym2)Ih?I{<_~uE^%bDSS}a$
zq8`zfajUn`I{Ru{gG()dhRf$Iqd3rOTKJY1SZ_9+QhjR+EZaQDnAXYRRsqlGT*~V~
zTrREm2?P5l-vfXD*ADA^K{q!>qzi2EvAGAui9Tb06m8ecN>(>)*m@
zoktlX{k`|eOH{sNquTw*E{(Yrs$@bgruxOysaU65SgiP15$U6RJeE=XV7NuSnJI9u
zuKg!kSM*YxE1tzVxx)76tpw0Fz5G6fL4V`PIx8ITT>jGjRT204V0;H|>UXs{@^?yz
zErH&-M-oD)oR@u`bsh)INILZPD&ooF#0ejRp2jn^m){R)Azr=yxXMyGe{5qBvhgtV
zzN%7qD#A&{tj-6Ff43r@Ehvl%9)hXBVX~^4YWzO02D~=sFFmL$f)A6i>v43JD
z;xi+ygSy+HC)U4R)nddiZH7nJ%tBmSqu&)wg`bS)H%f5*CI3Z|ekt^~w=oPhcR6#s*JEqrp
zGwNYJut~mT(^anx1U10tjn{TpA9>%z3UPG9exh0cxjazXxn;r{C>!l
zCnd*CA=jj#xtBlkCmmWFlL5IOdd=5)Lhi!BshGGhzTCCDm4M$V>-g_|z`wMiq_U#!GlSFcU9o#+~+Y4EB7>05dI=UCF$=|sJwlzw~Q^yh@lE-wh6@%6g?
z0d3M}aIjKOlG}eZ^C9t18{Y~dYw6Ykzt
z1xq&8sTS)Ic&J!~B}o>9zY^83GNJvp(?xRZe*=19LRGP;p1Q7tx8uyXTuukNP>lQg
zwSRHB5j#3@|0F&RcBk}LFHmjZym8&RT*Ey2fDinZ;6I!2*+&yTPhOoJ#pQIZT5-8`
zU0(6|_($0gE?;6=gZpWX1>-owffi~N8{lp${C9LTFaL0F@!s~ScfEM}aSu85f}PI(E<$r6(HOL_Ft>Zx*R#*l45BcW<+ZUz$%uGSUxHvjewx7ieZEu#QU9hn`*$
z-&MI+Q05`%K8U_4-8xpRvsSgST!Hph?l-7=3w~cEp6t3y)O(MYkn>E0nUHTjS;!gT
z=@`-9HAHA`3{q8}@C
zN^?Q?2oZ9v$t~d1>P^@iqn|KE@sI0kQh@y*h5Nu>RBhk^YA#@)zrYuxI&eB>ztH>P
zq)2z4CFmyhf}i&W-3|G6V6+grRRtZAE!wH-zy+NDT2rJyrLX|Pd&I?K;=uQIL=ePeB%6jyJM4$LKZTqWH
z+*D_53C<={2b@8WiVseP3dLW1Tu=
z_mZ%*#IrT+uj@2f!q<_VU-x-Y`d>eMvE#xz<)@D)1-KW}>)a%YZ!}t@AauvN-Nu%m
zbE%!nJ4tbu)|-cK8UvqxJb5h^>(1un=Ql}i*OBSIH1E#r^3s&f)jT^)yz~+LGewp1
zDgu8mYd`B3FfQ^#8Ld
zv`%bco&KWor>*zNY>IfK<5DYITBquUpFKQfKRfgW=r_BYJ=lQq+$TIaKXV}J8?h-h2>Ct?@&0rhawA9GFEm2j
zba?(I=LEz#$zN|Od5l4N(uPe#)|AAt
z=-j1_C5Q*c^a?4xg!88@8#?+n1V6yx+R$#GL#|e~e|R*GJ<*TZb{fx9`-PtUHXrFt
zZ>+KD?vDMW`ja!;E1|qowQon(k7F9GBK^nVJS(h112c7;CqtkbzbG|BXz8{8f3SI*u+Q!JJsA}EE!}rCrx;H{gKTX0s?lOgH!q%@m_EO;GwjWG)$fst`WJ=Gn-GC=KCEK?mjdy;d;5h!Nnhic{j$?5
zXTfi8N?rtXL;ThL+TlCD5uc5^I5hlRzvrw#zu$KUobTy79j|%97;#H0hYBO$zquD9
zlNjQmrY2!;H@|((j_RBq_5gY>9Im(bL}omj*zmv*Kgf>@D*4M1c>U!q^Qt|cvxFft
zW4>2=&VJnpG7Ck#TA^3&+qKd0tb$9A>$YcMk5-*&ebJuL73O%ph26g2qHfT0;|67~
zH&WbPuUFrI$-p{0Gc&^B-v
zjU5mA(81V!6i=m>O~Drlu}{vANSOib7cew>Jp8}oLgR`fC_aDge%~JDFYr=}r2DMT
zy421c2-SDiD56?OlYG#E-q(kLPgXf9lKMROXcP-##q(nyjtr+Lm$G?@zxRK`n15*cqa
zDnuy_D$V2Xv-Y{?dEck!`##UlKObxFwbx#IT5Iin_Br=7kB7`w{D+6pc%EL7E5uiw
z8p(0PUzv2}erH^sAD3qz;LOtxex(2h=eZ$5oZ<7n-6*0TDlO&d*LqN30>3`POW-?y
zo6}z23a2T(^V3gooY?otqIv&?YOUt|d@r>d*Q>`_!LQ~WM>m0Q$|Dp9YLj^22R!{D
z+!JES@#HdrZ&ZztuDD3x^J^wxmL7%|{0`Gl4e*&fURiKN$ZsVR%DK6>Y|kjpEyY{TP`1zE>=daJHkTwc%ESDd!}
z{f5u47qe$^dDbKTa6L7TE}cm7pDk}_ML64eW_$BTg8x
zkG%GE5}~i_ucm|xX2*^vKI{EjwS;#{ym|c7z2ZBU=lCY|C&^13^}v<#Z`^;qk>VDW
zNyjpYc76JH64|*fwyIK!JKUJVzB<^qubG{jj0aYh@3tm?Vy(Bik?iTM{`*y}fW>{L
zYSkcrz}Mr-GeP$m7!gWxVurTtL;Nk&x__m3r04U-6gp?=9Tn~&fxW$QdhBD0$1WWF
zjDtAXr(1`|Q9l%i*f@U!U+(_;s=MIRy!~`F@ePs;)gZo*+VQT`uWwF%{?-7#7PGuJ
z1c0yNMO{fV;LWN&k&}UG9c#i$5uX&P{MPI~2K%68I+JVho`$hWy7xl(n~)Ai=iVm0
zwJfesoR*O{{@^A&U#se2@$nAMeat3X>{$cb0Bv^3+)7N#4B%$
zadYr0;43rV
zHByiGtfmd?2)B)tsNWU|id+(=p6j$?R4)JwS9uPo$27_Mm{f)9qyT5%v2Ts@%
zMSP`aj}Iljp$}$XCO+xB>zX9*gZlK9wB9O=eEf;8_54x<1n93PSqa3~L9KsV;wx~u
z0YeKveAbQoOPhl-ZqG7qP3QiyM=4!R!u?pqqp^f3&sXp`=3z8l9Kk)f3;I0Xm_6Nq
z^F`RXoTK!d^hZfN$1Sl!xjTON90c+Q&Z8S!SO+b~@O*#y`;o1Sh4?{n-viE9
zA%E}Y#C8l5_>aFA;+)ty0$&TgJkEDxdpHO{y9C4JBi{T)_=frnjnNB9R^-z1&$<=K7abWsC+;Ro|%qB;Ug+$&KQ+T(eP
z0y?}m7MjTU!u{#OKK5A&D%}XX*)1_7jBjpcN4R(3Gwy$X#9Ga#^|affyF2;I@6kJY
zQ~r~Y$EOiijCSPDQ);ih+fRJY^M-sNG-%d@%NvOE=uug1kM4hc1KYD*4d?a
zV2u6j_>G+JemkS-#5do50hjl?@zoaMo7d#?R`UDrdqwl}nlmf4xZhuLPQN+vuWx!9
zktE{w2Wl_Ke?H#-ryu#t>W6Z~phWKm)BiC`t0e784QHU=SGHqL*DnQ1eZwH~rhIR8RQL!_A_4
z^ja;$@4YUa>5KGVhf`aTzwmw5%-RY1uinvLy*=(3Z8Rxxu&j|Wr-{isSKu6DgNfH=
zbNJPMQ$`0*g1;`qPw9^Rrj3%pFz`2?)#%g?d|Mx8RV_gK
z*Tz@tOvUq@>*i}l6?TLi)1?F6E5Z-7(CktIzQszJO`{+$Jn*fX0{rxrv-@8}eUY(E
zZ>A_2q28+QHU09yCz)+o*MR!8@2LG$1pnEQH)P}Q$k>B;%Qd}p&~NKYlHY-E#3|q8
zBFM3;#ZSSbefiH~Ey%BJ{HxQ^6yvyO#G-2Qmn%DKJ5l=$m#-X5{xV>{hqpia;da7;
zAr9d4K2_RX3F(O=S}74D*w{*Xb@4*F}Z6Y*YfDq*;+3
z?aviD>wcPq{6h*(#-#tLWhH0HUk;2dT@Nzll#-6
z+egk5|KOM1c^tmDyju*dt2qtbc|2_yJ9Z4Mx5=#>9EneRPdguyclg1IBgEJJ(Qplt
z*Cix>Dy_FGt-h*Ky(iY!EG51(s%D%oeUh->I91(yD8&)Vl1O~o0QcwIths&x$E4V4
z!ftLF;YIRH-JA{)cI%eG{pF!05!_!6c5&wZa?Yk_FznbLHLK=+|7=8)sYLglTnI;r
z@wg+rSGs=X9nRNP(N^&L4!b$u6O(P6FF{2K1|Rcp2{|4-{|Pw5`F_=h+Y&v?u_v$B
zSuvT{yY=c^&KESjHk|Tb&kl#WH=K_xuiuxUAh^J$DJ;&ymlj^Hor?F4^0
z&{y!6(nEs3jBm4w^Hp6a;CB4z8#EhoyN=f9xWQ6*p7FHWmh)vN&;?pN7uScHzy-qPD;4TJ*-WT+p!`0JN3L*BW=~u*E@z_Q2#zVHa04jc$zQ^8
znEZKyZ#hV$uNaEzfob%33s|*l5yyTwjUoCgdWBG_qk!gz1uRe;#?$o(kZ%reM(9B9
z8Msf*Um?G@p#$$o>D*ld@z9Uc_tD#wK39Ie!H+2@pOS%`4cLj)1cqTY6UGU+H5<@$b~#`*QC0LWkIfnf)h
zBvON5C&mirJ&_L+Ic+Kv>Th!dB=Vo05a^^^LOG{h!aSH>CFIv(7fpPmb
zUm|A{YQ5RFhWJ;P?C0{GXYE48Sl5x$(1oxI<6QatKQnd@CK&bwUc!7Z3PU%6b{y}(
zG1PkrFSp`4y@`l@&HYeLS890j{O#wfIem9RGN;Q9)pG1wcWo`L`@1`|lDXZwGW8+@
ztsJ^incfF`>EE#L8qncf!)J;oyk8!{#Dyk8*pV
z#lL7s=OSUsq))S5c#o&l~Rd&?YQ#IqpPg-18w0WN>c1XooJv
z1>r_*wo8UoWi5
z0=f@Uck)S#3Bcr|Tdr#WwSSvfZUc@h-kI?V{GM|bD^gtY_O|^yII23%@8vE4R1*u*wLCaDv>yy8@(JXdJYv9iN=Z;(g_UgMgejNBdBrdd8#{Ji<
zlNY8>jFd5tmd}4>Wa0Z0rY7G_%CK)J_n+g~qC4BCa~1NnEiyywp5dHZGwJE)
zPKbZ|uUr@*ld*m(Q!I2-VW(V&Zm@cT^M#4N^Lmg3L-f`s}e@(`iI>>r`8=Fx(oGc+;j_XkN(?M*X|_Eub=)3J{^Jgp0w9;
z1m;a_Ui<;$yExSJav;j<*Y#Sw0qJ9FXWXOs#H;5;!!@8U>n5&z1)LY_d3_P&H%se(
zlg<-#5`b=iwu(p>H#OKPLZs
zH@ta+xU;7k4yFW(NC`)!I5MBhGfBkVBFJZaQmyiGM&dA
z_nN=re6RPXz*E9*FBJUIiuKpIT|E`N1&#!Es&W!9_o}})0s;8DPxOOPurIf+aeMnW
zbS2LpHDt&+O5Z36u_3g16~W_|vS&ZJoa0OI*?puZNK3h1^{(Z-oaIh`UZ2il6i0r<
z8uGlbH#x02esl}r_&Cpw=Qr`|&aor^avq+`DMK9i?RzS6tg;f$|Ns0ThmQL8p*ZCm
z1aUcr@BebFQ55{nRy@6?{MTCmz*%L+Tu+ByO*tLiFoI*3*j^m1O@(ui^S_^QIdl7h
z9pz*JXby0aiBR9w`vU!QmQa556m$dfpV2lQxR8z)fZyu`dLM<2pmiS$=|!pnJ{m9J
z^X>w+mVcKr>aIW^Z7bjp9;hI$fmbCtC#X8AhtvZ$$EXwj=_%m%FrnTl{}q?U!W<<
zAwD=)jxl&TN%Tb?fnh%Li!fLx6A*qNJu}6dqv~scPqr8&(mT^NS>W#3!Z_5_Ba8rT
zLIA9%;RH^h;Q{W$=>=iU1tI=w!WBZiL&JjfuZZpm*Gv%b_EbUcV7Z=|@^&4R_q!{i
zBH~x~5#;usHj~r8zf~duaYmPHPRC_SP$AAmulCg=H0p`40jSW}j$_Ht`5X((H}P^d
zedh7_DOpXJw}0f{)%jD}hx2v2H@!Wo8H7M%OK&-7OTMvOYvU=1uDsy>MJkpT}bULFPfb|urm1pcLo%DP6<^YMOqvl#U~
z&&Mw@!f*6`<<$W=Kr%a$&WRcWY&VnKiFtoq37N#yQ%cdBm?x=w%@wmLpz6bS>dx1ezGR0LW|C`k{9UYP6XfB<>Lo716nAJ&OZqJ
zZ590L58^6a$0-BI&xTkBSdt#|&))TzPDZ~N*S&?lyw
z=N1ppuNU8R-Hhi!!yv%-FmRMnss30kJYO4eU3nPl
zZP6{yLVFv=V?*M`r_qq-AGms%8}3&=4STk$UoY@CpPjI@)ImtyhaDHupTica158o(CFUnpi^fdBk3A
z`~(&D4Ob!uQ9L;#e{!M;@*CQ@OmPFfA$MIA-Lt=cvd1zyZ!N2zw2Id6vC@+Ev~TP-
z>(h8T$7*_dLLS+Ni}TEaXQcHP~H
zujh#w?tMTntC}A~aiaBXWiLt(%<<|%m^op;0-@8nfhJ@ZJa^YFr~TnDzbD@*eW=TQ
zY=oh2<4~LkAg&tZu3Eg6T
z!XsjT9C4fX#qoTLOkG}XX_r^rJ{~$dg72%1wPWGPaK6$enCG83fo(BpB^yI-7qcbD
zxIVhcwcLK|t!}~Vv5>6ba{U`3d3@V<*9=Zu=At<$-(huoj>is>gU7k3z6LKhwfQnm
ztMeZo!}|~wP#5_I>mVL@n9MiOs4X2FA%0m00Ja>0;y^7I3IwojRMvBSj?|U&dW}B|
z=P1?{f?w@0T8Kl9nxE$RC6N?(VO?}UbAXFN;qHL@MhpJ+?bWfIo`tVvQo4l~iUFv2VNBPJY*rX
zuU3BldFLD<|6~gRXFnJ6d)o-KBZV6zXPAH|dkZ)VvysxTQ8xjNF{*@>)V;({Vt|d~
z1Uw_pr(FQ0hZ6vKm<;5v*8#2M<&Vkp8|D0P`kbG7h00mW5tEtI_HtB|*Tb6+J_IlL
zgB_A1&j!6nPSfa6KFuCN4*!KbyaO%r@$USue374K7nSGg1I2p8e6fB6b)1iG0TaJ2
z9s2>7aRL9i=l&ki6{WwXY#^P26IK!cuXBYkZ?y$lL!QpJ5y;n+(--CFC6`aLm&%Ln
z5@~wMOX>VGy+9ZNPK)go`+c4~-IAs+_#|%BV(f2JsKq!p#cdeEw+h2KT91t2SWb^Y
zksh%P034n!(8u2j^KHY{kvx4rR|x(`U*PkU?~7s|3;e2=sKel&X_*17X?q9U%N54?
z*J6QQ=ql*Le`XK)KD+^lhvn{G!%=p1C@*(@5svkc-fR{rhWl0MGC@RwkHK)r!NF^
z`62S}?o7_2nSpucCahCq-FBQdHg)839rzD};dy7SU^hD9*Bprd)r>J5N8PI6n331*
z45ionh~;tk>SOUdE^q5=ltSsdUw^|F_+bxU3iqlq7YFft=BCZtXMZmI0Pzhr%k?FH
zrxGz19SQ$3DG>vU{bj5nx6{p=Y}F(BXHF!xl<5DO(Ol0HF?)HtcKIozV-b&8Z$Ss(
z{P|4o6vD*$!ak(j0?aQfbNjyJZwHJN?5CE?3W_hYHvOGP{##4Q(4`&l
zc-6rL8-amN(_b*eAD=q5Ur+V#DSYcr*vme%8R@;K#`-Flh>5rQlI%$FaSw75dH}K81
z{huig&sA=B{W|0xKGslv4tTC#@yA)fL5dE%e-Hc_VbLA*Hq*rm>3n{Qrd|iyfBo$D
zzM96>Jjif~Ik3~A^c6He23&dn&`AB?JS|e2=8^h~RU=jdvr0DSCIRP1zS12BH1M};
zpy#+lMqXM+=k9xsuUbOojeBPs(R+P8sy)7Hg7%g~%_n^w8c*Ca0Pfs!zV2xY
z5AedLr2a#(9~ymb%CaED_iIz9bRU9!fXcdiLq0&SIj%cP923x=?vqudzh6n`oWJ9@Cq4rfq-dX|xO&t?rGeCMV;vSvp?PrBELoeL
z7x?vyFQ?}!&wiP!l3sNMB)#Oz4T=p@V2u-
z*(l(rE1IXNKE*|TI7q<0Zt98C6gRIt*m
z^SiVzPnF_*-D7H>sr=8DL-~HYo!SjA%3pUVy$_*D=Dn%3&R%&aETnP$Gu#V~3gg*J
z@K=)^9dw9}?sYnc>MJvj7*6~vZgjmx=}(WQ|0a9TX8%+iB+`AtKbuKT!_>hiiPn(x
za3#JT_I=>VVF#w%gX4ssiurS#@cfs~(+MvY@8bHzKJ3fwZ=WTR+OwDiVgEcl
zkkgMlIr98flb|lrhcpWQZspJIT<)2QeLVe1N0?jWYeidfyhfkd13d$?k+55~Do52S
z6a&2=4ghS3L^lC1EZ@fK8+T54kN)uqGz;kmw+Q(+EQNB;orHVy!wq`!{6Ejo9Pnk}
zEA)iw+u;6y?-PabUa?D{)s8_t(p`=KfahKcwCiX=z5`v$L;BF2LVkP`0WV=RDg9`f
zfVGbVOs8Q&`cw)RfYV4_pi>Y9itry(1$}*SN=CHOOd)>^W)0EX?+N(>rU>*a9^m1=
zsXX3UA&=WvH3;P|$>Vl?Y7WZt$L7Eg5(B*JAYgyYCZgfgI6fx;?I))n$?Mri-Gy`n
z865esENCY=orqOJ`37>nU-I+|ayhATI$a*8(d;08OhS%^a->n9^d@pW$I5An3Mrkd
z3#3sXnx1kH(r6I!VnCk!U+EOpQ9jKsLQx(kL36eLi~nCaA|LN&@QJj@C(0-W!WDWki-v1?!qF>_6ldqQuKH!Kec!OZKDLhBMzWlzY3GV>RZ~1)a
zET@~v=cAo`em0|GC^tur3UaKK<6|@W2hWe>v^X!P$Z4`Bl>b~lkGsj!`F0xVo#gZu
zIZlwTubCcH0QPn;CE$6#h7jLhTseW${>uwF`r`HgC5utcnuycFu
zL4q#sEa3cU0{Y4KU3>j#hQsgig$Da9w>P8Ybnr%jpM?o>XUfk5hLM>>xleNX+9#p@
z&+^}S3=Cez^CutiKnBjW1`BpQd}Ar50Q{OmXb(b(rWfZM`I&kg`_Y#)^PrEWDKEF=
z>S%Nb;y?OXLEzU&!4H(4qm3c#$^{RO7dP8sBoWux(9A)ckFReKZb+wzf%K1R9JTsw
z<#Ma4g?|0>SLo-i?*x18HOrRs&HZwnqh>(ZT`Hg98m~y`oNc$6&~dO0kN0&Nr`;ky
zKj;1lbR^=KS*>BHaW7;W9!CQWCq_CG{&r0`Pbi&f&g&nOipK?@-(0)R>oGpl$oco*
zUx2BE_|iw0x2JaI44xkotHkkhwi}iL?tcudfa0((^ZQ{bBQ6+nl;@{S#bad9Uk9(|
zwC#i6FdeYd5AzpLyc||I2M1VqZaB9T9gcg2J=4tTy!hPY#%C*mnfHgBp!CRqn|8#P
zX;IKf^b#G7E)=I-Qa6dAeL}M8B}YmxlWw&md?cBP0fK)x^gM>}b>8%JLf5BynuING
z_0S`9>-drD@6g@;FUkMBzbV(he%T$IY{4HY>+^P9yYvJDhv#A^v(yROFE;N<{P+49
z4G94n#4pxg0KC@rNCU;6(Pwfd(fRW6bq@9ve@D7E&L@7&Y6BIvQhlfWvsV#L?!0V0;UBMXTjF97X%t3sAm^$$5L6WC?PlN9tSEst4CH|P%Zz2f?_O0RgA}Pg<;`#+Hw`P*Q
zZ%%Z3(h2y;_S`3mb8i;*_g@TrzVhyKDi^0SI*V}8N_*ZvH#V3Rl0F$fo=l>+^-$tf
zL>O5A2X}Iu=fC?R(dRYW^dMBu@5Jvr$IT1EgEGX&KL+smJ2my%U7}~kQ{V!Bnqgi*
zxV|Veg7E8^mmG^TTr-Hia{2BoLTP!=C@N>Lz5Z}G&|#4K8b{zr^F7hD-!*qWe3jav
ze$~>6`opj=UPg4&MQ6uQ+}nQPvR9P;8Ze2sl~>xs7Y+p?DO
zkB$xC^FXKi@?@ejo-Kb!?QeIyjJG$WGEJG%)r!o=5Y{%I#pm0!)@$6Uo|#vct|NJ_
z)0&N^{N9@8ykB0XjE6zNz1Nmw6)0}+ci`DAn&--rTL-wj)(w3A{)|cPPIP$d+j=C&
zxh`Lk=;Mzr<75-(I7VN45_X#G2+sijv&{tt5dHVV*P8H{!7pydf@k)IA;rEZL9jz9
zeDUcwZM1HL+~5<&jq?VRVcsF5823nAipo!Jy7rq(@li@m6&~$AA{LN
zSVYYSE>RNd3A`uJR`R$(jR%g13*ZHaFEbX)6*zT3lF*&h1b(MtKn+R;&Xm*2Qw2H))|S$FbAb1W1d8#`Sl+qc(L$nPSrhi>ms`5XeEvOK?)oW^40=`%MptQ
zfB%WIskey6W)0ON*2_uIH2RdzhZ`u$`(J!wd*IY~IZ-c8qWr%o_QSvFA}y8|^$^oVBNr>0wc5Ce%>JgkTazQx9=M8`z
z;uRw9<$`b@9Nv=nzRL3vEONT39AC)soZL=2$!T5;2TusgTI%f
zXt&Ptfx>%&Qo)Y5xI+oJUu!M+4JEZnoX%(y%kjzqssQoRaR9IdZd(!t@fD7ED2*%%
z&LL+C`58I_s_+Kkxid}^Dc@xnH5~eN6!K5Q`x3pXlYoso1iZ`-T5w;7Hvsz@eIeac
zPOq*J+2ImE-Os
z=O7UAv9ccL-#MHt9`u#cjDWr`g?>^>vgP#m6#*P4e53)!_ze~G_+upSYX_6U@P`|J
za=B|B34GesNJe_qNI`$~G@5ycuh$83=hXi~k0G8om5dCWGll)d$m4ydw!g4s5Z}G3
z!9gLO6D#jPM`GW$JP}J3_f3ir4g&k@`@_^?e0m$tCY;qq(Ce12aBk;5ZY{5`sfQtN
z=Wa&>9&cUe2Vpqp^;*gE$E>B53coQH;XLqu#vxu$ng2!Jo^8k4@%+HEx*WS~Xu{D>
zLMI2;Qd93~^?@Fn_ih58>@wTi2Om>b7ayDSxZyi_L^z7f(4#m^#>i
z>wR$Jyz|sfwS7|FkKx~z<~iq|zZu&xzblWs?@tx-|H$pO*CT<($6$H69C9-#zns(tp2DgjRMQmjh+ICMhEcjRsS8w>
z)9q;(K<~yV6E2bGUm+4SzorXZZY0oSsrx`*AOO-T;^hcHJ9H5cA2Z@~TX{XS+oOD1
zjf7UzeLzhDAa_T=9&%cYOG@Ol7}xV+;G^E6aw1(Nm(QDlbP>6_pn3NI-R0%17rZ_|K;QAgYRGdfBD4rigH90`9wM9
zJmEk6B(`7V7x~2g5fuNAo}xSv#d<|KV*39=v46z+{^b+piR~5Tiu#BsmKVoc#DC@e
z%O}eFmrvA7lq1&vzxe)5|Cbiq@o)P7qQ!be{ls>P;~};KP6U7dc~0`LJc@QGU({cm
zr=lE@Uz8`dTdY^i{})C5MEqBt$S>xLv{=u-{34&Ir^qj&m@k$WQOp-noS*;77xTq*
zu^y2Y^Z!M$UJah{pM8{;oDPyB4@1$<|E{Nhdo4vmc2A_dv^NrXh~!f^e^&7sP#MVn_NVK9ONh<$mUi
zJb$4)zlj`Q$?*%HNbtvUT33!Ufg_h7JOuW=O^zJ;_M#bt{aRXUjsph?ap6>)DpPu<
zyU;#G3L`wo^Ta8B!x`Ha)iw$yQWJ{oJ!(W~Voj9a0+XIQ7?q_0m(xA_$*$K
z6SP7|uT2opwq8KP5g5H}VX~E!o7s~{6&k``Pg@F1ih4Hbo5b8grH+lQI(6$9~
zt3w4;?f??>%D+PUTpDpr4fG+aRG$7xtAw{Bah}kQz>0P}J$I>q315YH{k1#}FLDv&
z?pq@CZ*-`j=aFv0yvxcH^f`7`Xs7hAQ2)4(Lb>l_g>sg?g#0P{g#61_1Ujvp0!5ss
zj1>C8Mp2M=^n;K;=a|6%*pn6@-V<>W&KG+05Y8LkdkN)gw1sj?kA-$KIo&T>D4)|H
z$m_RSpj{^m{14>zYpJ~76ym!uwL(M_`gXMAh=UxKcO1}OhO9a~3S->Ik
z^6B#RPG6Ox@|*Cs24Tn^flks8u#G&fc9O@w{LHVE{gs*F&R9
zbP)j%hZP);5`eap>nYNBsK(P*5FnqMeIP!{%IVYcd~Pn0-iDHaW96tXw|{(lg!H@e
z^gHr$`{i`9Tu<(1z$f~%-tv4=-g-GLIKKa!XVWNAy<99%K~8f3{ck>RCep{q`9!}c
zmKXW?aDwmOdTI9jcYVCu!ADk?(nb9^30g!ko)YDXw3sggI
z5#@++tk^GN|BER0?>H{tKjR|G6H(+7$H7|8FV^=jpV%G|MLrQldH?penE$W5O&@$u
z+c%anUzehEEk)diUU0vBAKq76vggQnKio^b+v0`&Mtsk~?#yTJ(V4I-eU{XEAb+CS
zJ=0CduPa;s^cHYnyT0Al;l8Yio9)3?i)E}+dP1lh-kUq|Ibx0q?x8xheWXtJ%so`y
z?`Girz>ij*9)V47vY&77&%Etd%4Q8+8mqb#{-AX$$G@6*?!W1!avS`R&b>8TAiTFNbi~6tMNt!w#!wTilX
zQ*pP9Ma+#~62A}UqO0_>q+M`7_`#U#78M2T(}O3IH(xAcvkQAyEXTdy_U*cUbHw*C
z`v?E9qVHh#+%s7%_gV?N-l5KJ-Fz9VGwRrB(>UD!FjQA-hwm|r%yAfT0PWbi`P5dm
zR%I-|cC1eOGq`8;GVRDcyifQ$&wKo4^pni1{^Mx0zwM@9gGZnr?`M{qrQ>(|HtxQu
zvkLXiF?X?!>5J#qbKflehWpdHm2NE$UckLS)8YHa*#Xby#NN7$ew8}1^Ju??S(Rha
z@-jBz{OH8QP~02ta`=>uO&QB8?YgCCZW$X8`EsQ!0PE=HsqE-#8SB#I=rTw5Qa0g)
z%YxWZ#cWYacU6t6cUb6lqm}`Mx$Jz^2xo`hIV}Cy$FJ{gZnKQ_PB$GwlUYTqt7=~F
zXuJoIZEocn#b!TjeCswToQ+c$9<6qvfCYEXvn|uU%I2BRvm4*(GQRVvJGo236{cxu
zxwg4%1vG21cOZmZrP{?+)j`ZwMm9axn)s_+KJ&v?M2!5i_N^sYy4m;I5k
z)m>+*AHR=&@NB+mu1*=7w_&z<#4nsH+VmbZ>l3~sH_YJCm*+Q`%ikPmYT(;GZfF=Y+Um8{wo@E?mgYWqLf#do
zFgrANiCqdinBI_g)#ob9xS_v8?^ZHHwCxs25`
zJ27;QWf{9UYyaubtc>l;FJHCK9N%GCqju5ySSbs7*Jr0e?^4!N$6ac;xP%#>`84Wa
zV-YiY{w-wV)k3x?49!yd5|d-dp9@rhk{I@wcB_tHKf8_Z+13}#O~CJAYWH9M;^QnCORmf5eD@gIt@Ac@W7jhFJ@8FT=O$&W!};pj
z)gN)*7x!xXuSMBxnAwE>f#-9XZQkZ3Eei^nU1;sSCpjf7>-cZ=H%m)dpN6XYVK(7R
zsdC&J!wz9A{Fw284|mTprtmh-w=R)YJ&wKYXqm*^cH5gN&PZaz_J3KMsF1>*SjUVE
zyLXu#f4n<3YDO%(5^~3J=Yc48X^FezwGbIQJv_4O4}2GI*(U?j&xTm%*HbrdoQ(5n
zwX930@ScI6mg(jCI^4V6sj8BphWRsW;)8FxWo-P@(H6e=u8-S>&F7ulm$C53X|HR&
zO4-9T>iV}omaw7EZLg1gQp`5anc`5HP{cmD7`|8NP|SX4ES`LM+
z;Jzg6??pD{f#%7t3g?(!W%xI3hbX2VRG)I74Cp2?RS!7Frv83X6Q^FyoU
z?a=c@V-x8;%sb5{tut*#l(E4(G~XKzE@RS(m(Bh4%2=F1c;mMA#jHTt{BXbSh3r>_
zMN;F%0ybe+hr_)cirDNt&BGt1#q3J_+={(>qu9V_(wzq#g4yb%@TQJ?{8_^>rF8dB
zC)mB;n?1VBiePDxzdP*?*w6a!DqOaA_hHuk#`;#r!h_kM$Y)oa*a;S2xvfEGTsYe@
z)zscJPsRdwHaU0!?}0~q&KOfvjC*~aofjrN#yvQrl92P|GN$2P>9(~m#-Vud5M8Y@
z)^%-Om%V%NK7(Nk#S;9^gjXvY%hI03?3vLF$-^k%o+|_Q>?>s9`$Ec(4!X@cv^>Tp
ze7nI?x?1no3%MAotWhW0PsU1clx)~n?ENm->V{qg=GSu9||sV@DN++>RvubZQ+n9gop
z?cJ+d>l%9$YZw#OGn+k_e$O#p9EZYM!fK&aDFc@>{Gi3r;Iy-
z`+ZS&T=VeW|H?f9ag`JBU8IstZr{@J-H6sB=Z@Hb=eTFyz3qce!%qDyBd$x`EqVnSPKl-XPJ+p0`hDCfey8RKN8nyawMzV5Z#+jWN=?Z|Tww7!
z+SU~Bs~wrqu2J6@-*0Hucd6oCd}qpDW03AF+$Xa2P#=%)T&*%_^I-H;%){LolYGaQ
zu+hKVS`BVn%B)B4)x2*gW04CNty9jCv9o?p3gcj>A0F%bs{ai56&LSlw;v^}*WzIB
zf|mFm!;R4T(oz}ozTNoE7je&b#l(l-OG??&LhIEZT=3nnWomoEP_K5g4|DA8%h9
zSB-3(@Vt!8wN#m&y}F!f*M2J-)BYaoTi5a8{CW4-qId<}5ex6J6_bAT`2{^!H|yaT
z551Z!u(I$$+&V5H{J15)4-x7Uwf6iZ#&UsVo!gu%pg813acXOPk7M`72>+M(y*=%*aQ?f{x3fR=^hkI9^)L|MCrYZy>zR%F1@F!`6tBf|a3;ALhTo+z8!Pqr
zUMph_D$x_|LS^iH@v5iZUrL!$%z>#&_ez-hW|?W$*VkRr
z9l?r@W8PgUWvea(-zfip=Zl*a-x(*PpL=Fxox%Fpq4m0<4!V(Fkb`N9_@Eqi`jF#_#$_+V!U6t
zeXaxi#kOJVZm(Eh!cNXzaKR+6gncdcZPE2^5!=^#b`Re@1#Di+Q%b|;-ez0PTI+Pt
z$zdP7t{z!gpUrl7D(BKVp6oxQ^+)AWw$Sih_Cr?8KF6k|+xu>^`4Zc49+f~fl)Tp`LZX7COy>308aI9lFvk#~Vibq`dZC2Ecz8Pig
zYlr(ox8OT|TAvpd?%q?zs?(=Sch_QH@NRC^#3v=}*JNvhiJrww{q?he6J`1A?D@9q
zSNi8LgPS`Q)i!6b(2M>n&Sz({78=HfocwRFyEVNUb}vt5#Rs~6I?*8k@pHfUst)n&
z?8?uXznkW=2r`1>>px3r=Xv64!`TztWo{ACE8c*v2x$je(>j0Dwe;whB(n{+x3vGW$fIx
zO5>F&Xm@Ls<^Hh81p`Z?Y&&PMQI$3GO&;80a}90n89t^-I{pztmV*qLnIOG`K#bAf2W*b%N5O6oUDjub_0fZKgm4L)MvzIuK6itH>w+V
z+|3OJwKYp@;~%fkC_mUeT?3yo8cksm{F$Tc25~I|2g++SInP@o%^rP`&P=n
zRN5%*GLo^H{B->t8)Phh=-t{2Td>ddwkvR!;T$&R$J$khv!sjG%c^Sdym`9kp}WX8
zTRZ-w!ciH^v0nVfx|fVy&lr6EVp1uKY}T(*(p<)DB>DxR$uf5DM?^vA`18!P)7)uq
zkHoNg_eVG1jXKSio23MvVlm8X%!>@xI+3-v`+R5A!w}{l{j%|j=^-|Bf#0?2^M_c;
ztdJZwU=36HWZ8Q5gl*TnOD4NGuEOt54X?hR<);XgbVp5FIR6Ze3oU+6vu--Y}(
z-D8a6WEq=3_=3F+`blZz$Riz9=>6OcUMa8({qzd5Qq&Uh`{g^J>q4<=)m4R<4Re@
z(k2G|*Ay~ii=nft*WPAJcc+><-%VxiTT8A@%)Y{0k0$qi()I!yZEt4#EGL2)zKqYi
z6B@>=KNxwjVZjyl&hz`x8doVxJd&#J)iIq}Xe=|SJd1jt
z9w^%=k+A~Ptd;H7ma>`UhC5S|OWDc8w?f?>!JnENq}~WGXFa^^RxPS4XNrF=>@=KT
z&i1wK;n3k!88g1O7rz@(#-0`g)sH-Sn@xC9JV3THpN%_j8*z3<5$o;0Gq*qDvMG@t
zSJ$m6WjpTWPbzQ?W73S@4?3$wvIcMCPfF{~vFVronsq9^z3Gde|x|Hdj
z&i*#ZUb?M3jUDnx?)b$P{^;P+osM?U!88uKW7
z+@aIX_+606X6DhC5eIKsyQee!NxQL?Nl8X!OnpGm&&F4HZ(3`jWfRzEm1YGe&f|Mh
z(}o63_joO19Ve^i=x!-xZx_6_^i{&X{_(`O(wDGC`u%ELx
zUt*eT^r!ePOl0v@RVBlYU1e{d$M>?WOJQTzwK|~_oW@3fZL9sqErAU(?Y8y7$9R^Q
z`|wQFi!k?v}w^Y=6W8T_Tj(NLT1NU~yRI1Mu%HS6!dN0Sh
zP17vR7uPn+So{6Vwh`wprwVf;Mk3z0F>`w2_axi{cNpVx?hN)*{g;RM-7jZ*BI-gO
z=i|P|({JmhFNL3v)rjqe{^@U3RAAPlm^J);Go{IjeAdM0&(!nYX)IJr*Zy5v5}Q7C
z*6TZKE-}Z-c1@xbqu9v$(9V;?!&%VReeb4b?`Ha6UA~)4KE<9NeD+@AmdFfG=5D@P
zna)=H_V_F9lg8TiHtoF`^S}OAL5||?Qug!jj=Tc5QWpKZE~r6P%A6Y?C~CHcf0<^%
zRz50co@(CnuDmQ~m+C^*Y_hShePr^W$2Xk&&ge9%9oB!U|Dh?JXXUeAVL5Ab6AM`L
zaDy!Ovn5RCXdQ5&g^bx7nmFlT-H(w*&L4LvoE;xS{8bGqbOeY=u;^(@I$y$2BnyD}sp)%%NY&Pb6Q&tG?d)!dP+weRdZL+6NO{C2ha
z!4Gdp2E7Amh|arlO#PmYQmCf
zIkD13q1C5<4!R(n*t{fYT5o~*(&ukiC+A`+Pn^{B^G`K
znO;3^OWFkaHM!XGy2LE`;iWIjvLur%N~0hBxGw40d3E06`0El~-JR!iW3nW}=J!4`
zX2ng(%^^0EYWCcagzhzd@^S7B$>Fd`eb1|2m&CDSzM-wpOVuV0T=L@9WodHvC-Y7&
zN{}k#Z2d6b?2>e3Sg>FBF(;*O#=W_3@Z3itxso)tQ*fkII&NjywD@qTLYI4w@5J~>
z%J(N$-933k`fKaEt^=d4N^Xy?T^g8^CK*zF^t*c8HHrPpiP@ia7DyZhXb;(5cwN$_
zMN+hxMTTVbl&>|P>wF~3pKhF#XK+XowW~#kEuVKuYLuH!GW~l-65Tk~dG3L05|1H2
zD_73ABN=w*XO*H?p2XAUT2Y^qxstuwRHel>NzINw`qz7hN~X*`=5@j(Pck#)
z=E)WIg_35yop-6K-jUqRy#2M+kOIk!sBZ6P4$75uJ-sF&d~me%l#`0RhexusZt2Jm
z`{zVUrN!pY&fmBo-RJLKTwQlsDjQp8VRhx4RK0R^ahOY#^lsC#9y#7&(pgVuT)#i_
zg4C&9drh}*XQh+hEnNM(AVuOdt=XP~X4#UZpYyz$Zx5C%PpMt8uzjMWMW-)QlP(5J
zwZ5lS1cslH{8q|X>by2h;-sa#$$m|m(GhlH`rg?p{BpUXe80=&{$RYoeq=_Cea9R3fQZZDm}nze~Dk%7!oFaw8;)o@csj
z4?8Cb7`SZzwL6z3E?dS9+qdPWblwB4ZJ|M7QoUOj+ZtO+rH)T-N8We6BAw{rT)2&n2Ha?d*4N@-4~UOwHG}i(({sWs84EvaU#5WcW2xOsSO&FX^m)RAyrN
zPTfAzZ*Qfl!6(1s3GuctM$EW8=>*+X4ox%zL+_?c!U(nEtL
zXIA{qmW22A=zh-bw#3$9RZm;}uhQWomUge%&@ZF2hla+{#kCT%?9d3T*^c-M0!aC>g!X|SvY+7|tw!Ijb
zQS_*8ZR-Q38ON%x9qJR+!*bxFz7w4WT3Bwg%UWZiv^-uni&hqO~?-yLr0
zS@?W;uMQ(JW~AJj;;c3#<52hlp9{lWEM@MCUR*9bV7WXwzv=udwizQ%_p|(aWs>FV
z9=msD77w?a5f&abVUT;q7loQheTN>;c%3n%mHsO)OHFmdQ3w3IGosgiS6V%5UdE8O
zhU%Wi+b!2Vt=D_IL~40p|Sj--!aiRqvbuztqluXZPduIJUjA(=78V%8QPyyXLeAy
zZyBe&qA2yFLgqE22d#z|x5?bA6ZGm^d2_3umnPV+*Hg^A^kjwY*RgG^R4R@)eKmbx
zW|NHUxOOcJtt0_vg|qW(GoIht{>|I(dd6zMNzO52He}@L7WeTVvm+zOuCrHN*J#U&
zPtU&&T$+^Oct|bNOFhprSvOkx;OTM8;Z;5<5k+e=5}s_$w*G$%o%cUiZy3gnjOc5G
zNFvD=$qLUoA3I8!W$(Q=$)1_n%9g##o}F_(_Rh#G5wcT>GAm#Iz6DkH4+^apcR-OzYSBasn(Hw;)bIXZ4(_j
zwxLUEHwmJ*htZKYr;_@QKB1MnPqu{96QIiq$JtSNo@kF(AMKMiQ)o!%Pn@u}1e(4O
zyXLSXj@}WqJi=BzA)BJ$x52%R5Y78!HTp-UX#EE|Y7b&{=onHgF568x$
zG7(zMmAu|)jc{fCzJxurrEt*LW@?CPiTiPHO8?#g26jN3-X!;n8PXt8!WYz7
zay*(=X{tBi6Aon7!jQbL050_dt1eRBff_Iw6_w22mQKDC%$75}KBk22pIi
zsg1TahkgX~f=;qrFZXzoplto<`KKK0)tU+BNv>z9dMa&C}7=%zl
zK8-uU!Q|-oKED#5CPJvR-lvMWVh@pdKJmv#aTjO~Pz;3ksX$jqU!|=G(n8X@{Q@Kx
zDbbRqP2aoy1Q4}Z#z9TtALPrATNlpNx}l#R-D!&EyoTZizcty>3qa&Ve}w1vucLp}
zR?lfIvqK-B`FM&eXQ8_~-_KSn+M;!L^@i^F>OpeB#$hqq>S(L=%X&7C`_RoDHSO5o
z9VAII^)C5~YJ~N?-=u0@GcbC+6W9+b0G-E6+A^nJ@KW+Y=*zy}V6Bh~A&4Fa(&w&T
ze>U0!g1UXmA0)g35t1neLCc+>VESHZU@HNdswK$dC^3v=5xHtJyH+B;!D6Z~ec7eK=
z(@te)b(SV{{N8p|4HB#KKC_dz7!e!yTx^aiK?oK5eKXpMk-YPC%;B|Z$dlT%wkeZ-
zT7S7i;$TdZPOyr5@foPGMAJo2|O3^Fbb;3MmlWhf|OJ%
zkvQ{rTGeEE2+Pw8kvB8#L0HqyUFGj6==pc8KAhVg7=QIi+#JmTm6D}X&g+R_wVrM$
zZUP1JnsqFZDtM4t6a`tRwtxxVur!&_WZ>J+RGM350iLRB?t%0az+ss(({rH>2*vFj
z#jE##fWQpKR|1{jYE=+!!@3>(5`4<(x7q-j46ojU*Be}WL-d9
ze|x)}SRbiAJSp#-%|OftNvTvfgMpyz#D(p
z<_dI1l8|5a7nknT=ORs%HY?^1Vz)0(XGD)fS66J
zCcF@IzlA}M^Ba)y`+V5;^;)oNAwQg8)dkG63ybfI^#EZ*W==Ed31IWL%eUlT7dYnV
z`mNvH1PHFP*&2QA0#C&)Gu&l5fsml4u3IJ^JdzN2aZ#=f5K+oWNXoY&W|V8J2frH;
z*12?}xtwNXyo*^a8g4*D367W4kYWU__uCvxi$xyi$o=>?REG#G(UcGVSBJ1(j0utd
zRDgts-%k_1{R+WT`yd)P3gu*6o^Uee}hckf+h`0!>063P?kPL2JQ3$-^?2=
zTfbWYTM*&zClg)Zh!?&sI$8+;r6Y?u9R?m6Kk1DB@*4P-c)$_`t)K?CF;=`Z1R6H}
z6W3Di14V|v_rH`60j_I4PKPHwfSg`3skEpBaa{LFE*k4ZZf$=$32;D=@|B)8((Z0V
z=n&tK!21?istfd}xnGJXJLhZXirIqd2ZGN1b2iACZ=0jK=Z!SepFZPXGzU6!>4S3>
zJz()5(K=CN0+2=7UnrRw0o++XT-P<<1M~TcHSvOB;C1b@_rp^Uped&-mqXAGZjHD8
zNRlWA(=h(DtK_VXklnvrV)Gs$5C0RDe9{2~n8tTa$lJh{@xaihm{#!Nl}FK|M;*Z6
zmBjDwQ!{|}vA$k1c@sk7GA=OMS`2P(QGb1!k_*Pv3VddlNptUdG-OiE|bQ;
zA+G^H_rnr#))l}_?s2~P0Y7B&^;v3fcP032Oytf)ISvFHe!lrBzl6wzTU{MrCxzOn
z+4b-*??Q8@WX)cTWk8E2H0|(DF=8$c{3|pvo-fXYs2~$XiaRVakvUy3K+tcyrx@LNWzYgeX=K9u{Rio!V7oV^}4+
zv2hdWXKAjdmdz088l*)s7)c+n5!4Z1$7|)KVnqXLk?)7Df?$aEDMqk
zZEPl5>xZsiS*ZRMn1j}_k#LLuE`ZkNE~oyu`Wme!@}nan9D%-2T{Jz2B!n-2IE0CG
zvQcBEgpkX8PLp;L(Dse6V3*+=rYkHUo_tb^cS*w_>(6Wx+}gHH=do2{=jJ}UbCn`m3eCyf6FdI
z4?bp0{fhVjZDyC=i&yJ*R|
zwR$}o0+_kp^*HLG2-ZP(V_5#;V_2(=RJlc*2n+h8bPw-F0xSNEdLrk<1@aAlB*e({})*mLoz6Q{YWFeOc>^Ts!2j8fqTj`k)Dr~SVA
zmqkDr<1e`(R68ODKPh?ivD-!t%U8Xsan$1i>$VwP)eW_TttiZb6ei8FczshioYEZj
zloV@!{GS8n>Uy&+y(ksdZBLu8h%UmSUu1UGOB%o_9!;!8l%()V7gtf0LKj5NAXCCL
zw+CtX?+I>yxqu~Rn5wScRK)I$eVbwlx`{E)%+Y(%{6#(4EEhf{)Is&1^`2+CokEtW
zlRi~Dig2zQuc{NNG@O0$FBT(n0e-}wez`$>8I6cNWK$y}!S;Mf*T1n?!Ibh>q%X`^
zVruwu_JT=CZ0pYK#XTzynD>f__=f2PSVI0dpW5*a_QzM_VoO96_AKb%`(k!!>|y&x
zzCt1c_T=d^+c7
z!xt!CJf|U?avFWxvY+E#Pz%u(7E?%>r=nMj+&Q?YIWT;>ZW2{EF}7op@i=b&7+UH{
z=AUd`MVoR9!>YLlA*)+aIaxC|Vc-AOgtT{WU=};%kCqkK;TeA>XikM379He>i6i(6
zO{JZ(XU4riDKvsaX{YH+V!cKcV1-OA9F-iS((2Rrkg=V(+@7ioa&$|OxeFy|7$>N^lrAW#l@nwsra{FPrRY*
zTVLVxZjR`4v6P-~HI|To@pdbvjnqD9fcy$@A^r0i;>Z5^7Z#|+4
z-lRsozC`%E-;763)(L%v|EZ$eZ%H{OeWfAEwSnc!)wfaaoB84K^CFPu)jL+*lVvEB
zmt8TM^%P~c;6IPjn?S#gi??Y^wNP<+1G92@HK)cN5vdOTYWial
z0lxcJ7TYsILCTuH_F8Zo=ui8`FkM0j4HXZ{4vtTP8{sD|b?19QHRt_DU!})^5y?Xd
zy=OfjQHg)MP?{2*=yTaKq@d>dgM$?$UhLi#IjjdLW%`Du%k_a)eJW*}nj!GXq^BlH
zWf)vJ`o{{^>Os=%&F?qU2LV4z~CqolM$~H;|X)$T4c&Src2wQ5#c|kHtv(DLBxxHWvyu!ATOV|tg0^;fZxyjZC)*Q
z0DR+OveZcv$euj75r<%)O3(Pd4QVYn_|lt&l9vE~dcP>?pqC(0mrGvQsT@297gYLW
zUjgD?H^E@634{mpNM^=ofU-x9i}$#Cz)0#WwNrZ!NXs&yw|LYEG}mbd*fKf+YnIiN
zwZ>EAs}-SXs$d;5)weM_`?4Mxw4?ppsc^=fmg_CqBw_8
zYX|sQ*E-|J`5t&1f4`D__w
z%<H%!4tJqEz;x{v^3TY)Uy7yt2~uLOCZ5+9t>6AT_<@=|x+?g1xaMDf3v
zG|=R|(>=Q1e#UL86U2U1piX6g(@OU}ur)k2`0vd%X!8rB`StY%gqx)-^YXKgprA@v
z?7r3_60JjWQ>yVCbb;X$38%dRgm3rLFOr)Cx4ydUKYdz*Eah7`OR6t`4u+io^&baF
zkH^MahEf?+^lL~odEq2dOP^lJq1^{w+O;-h%B&#&X`cp<*%Lv;KZ*_|r&S=f&=ppS
z`x7AN{u1RByn}?Sz3K2zIR+^;ly~$3*wOOuuRFi~u|sW!)h1c@uR**5oG%9NUPo^T
z9BC-92|zt@1GZuh?xP!9{6X(?lhDkZ6!!S{vXDtmlLVZlj5@!+`RWOwHdHM$&GmEB
z8@=VfF_Dy9gZ4Szbze*oLf46N_lH$wp!}uB=1TJS(Z8b>6{)AL5H)(i
zN){vvwK+8|PyMHit^|h0m^o-e1F9MF`Fpe(ApA;
zqTF|GXK+W~`?dCk(Pcm^#xwHq=VzfCEMmSD`_br5{pq^(TrxDaG2t;uor8Y(gA37e
z!JzQJbE2Qth~XND3TxIyP^I$9+~yqwqSKy=dSuj#2GglJDvPZ_sX3kGEmj1u*yCqM
zJtNi76&ZhE+~0sM?2%LaeBBPU2FrZN30+3*q=FS47|F0)T_LNN8jWZQNzsl%DF*qG
z@&{Y(_oE?c6-CR<+tBdq<;}5%i}2N`&DXsm-B7f&WKxLS2lRG``S$z$RY>|*9ETsx
zf0)Qn^0}3C5v)p_qCAk89_G__qMl)<#oT6u%!)Oc;I>Ft@j)>WY$f;Qx`nbaW_-T-
z@){u%d?)0NMB=sU*s{Zr)`86eDwGk~p5Ug0#Yb+=nb^6(%$DBo<2272z9#cLQ49V;$L}EMBLGh-f{tm317k0G8F{7oIyNs$T~iEP;Eml&Kc0}f!!5$v
zYgsNv5s)(cPqPlDBMB+E?Rg8kVD;v&3B4uk!C6hjdF+ARRy!tBva5nCFp>xn3Nz&s@gg{$qfeRFej2NDNYZ}J+K35H_v^ED*1*kSbN>cI
z>#*qT+-J?nZ{b*^*f>aZ7n{Ct>c81bhP%ZZSJ6e%gY8$UD_{N?1(%#R)_!HIi>ZGO
zOf4yr#XMuPQ}_n0VW-6{foG@9a8AsQ0fBTBtn9_ybQoj<`%$rZCYi`!9qx+lVcQ1S
zxT`2DZ5W2l*cStk*ciN9LK6_y{)A?&m8qCA
zN89Y8d?j)(%84c+?sp#+apQgHS!6i&9v>e1KrHVl^!pm@`NPxWFZT)b`1sX@ms*SH
z$Vd>vgiJucbwdodqleJoi|^J-s+2M5clRflpD|#Le-nJZrW3oZV;$DGnao@pq@j}W9APJA#1)Ky8a^!S}x#Jt;8omc!)knYS9BF
z`&SKr2(X3H|Cq7vTbQGj%%Yvjfu7(vcyW+rpbFzO%yKe2&;z;S
zXWCp#&{wac464Bv)K>QwMX6Q=v`@WC*8MabW&i!*Yj=?z#9MKfy5fQe)Rk_B4_=@_
z{fqV!#=tq~Ad(gvCHReafBU<~Ol6K55#+#>?t&0)c=f54H(x7I
z*AHwsV@Bfb386if8L_k^MW2(MQ+8fQ}?F@Bj*X$dH%jA0yKisSM<)iA&wn+M>y3qkbmvaG2~(j@V>u&
z^fJ8}_%sq34x;
z4inQ%*=|6dn%1#4-w5`V-da_}RU!R}F5@cI*=M|*?KfDSh2&=?Q{70&L3V_64*4Qu
z5!1}w>C8)(;FVeKJd4z8mpE^DypKfjbA*j}tkv0d+;W=lEYew@;^P;9!-uRX;d&?FlABb1(S(Aw8PXaZpHg7P6+c=e+X(XdKGg1=`41hd
zxz?@pU4TPn?3dA}4#1=kpjw7QfGB^x$HU1MAe*o6C0sQKg#UiH%gS7VREnPOr)n=p
zj_H$_tg=dxM_ubB(rYEi+{2$_jE{N|VVBxitDx6Nz>DdE>6Lhpx%JsatSlAzvbe)`
z_j(2xynXAthR+(1B}=eaIxYsU1~2!srs*RN{tx|fpggenb=J{Hy%oTTc8Y%;&Q_w_
zR^hrx)qu!NVa%(2xKp#AS|!k13Vo_BNIt#
z0F>%Uh`COJ0u`PAt|l)b4r7({8-2G?)1Y?a&sXq>!N&&atvNhk6D?kT^miO_zu4bq
zCHfOUAw*VVS}+9L+wt8KoCX9jYgHH0ml3TTf~Ym6@1V3Wy=@$$MNi!u6;`TE(Yg+<
z$QsT6AdN)T-L^R@v_zNwQpRgWsAlme4H<(ts(XtV8F(3vwpzK1JgSEv@w3hA)`bAg
zlwfWTs#1h5Bza24@>-#uyKe~o-N{9NtB_0tr--1mg4VxcK8iz4EzvUruCnNiHc`T|
zf;p7%?OvdZT>*rbOv=)!1W?q%8*^$GS#(>03gi>rhX`HBwGO)-(Ot1oUMTYonp3md
zAefov~WLRTkBper>nqkxabW!EYT<}z@?z$m-4Xb;TFn^AY8!m_0O=UP`F-f%lOLn6=CT(ch
z-DAy;X_u`2*Ve@fs}q(>|NX&*xwpSu*Qil|pTT04SFEjI_XC%ytRh}G=?a_+)AL~I
zQfjcP#BF#~;p6k@I~o|7S*N$*9d|6rqJSbA$YWpI_1EQ{?!uPjcEM?rnwamYUxE~o
z7yLVOtb0&96?UVnFZ4?F!;(!`lU}%Z!ZN`)`RkkhSVVJY>W{8en1+EkxvV4#7Pz0<
zc||r3`<(aTJ9%OP>|j#!uT)11CU&Vzdm35@vvtII6
zV&jD4_M{#*XYu)are-O|p!Vm_twjtTH5+%5zZ!~}CPp$a-HL+cyQZxk(pbYou+E40
zpLK9yl=mACdpxE@ly-e;KLndnS=4)uGsWtL^1BB;Vljpsu7P_WCb5vn|CnD!&0;*i
z%p^l!b--i7Rf|>!GuV?ykBby1K4AfmPmk8>cCf~nYqd4wZScBH^Y_W+Wvp#{e?T#4
z7T#$d=Ngh<#9aD(L~t^ru(@ZX^{*2)+{w?D%gMt7Fp0D!>-L{T_a+Vga{VGz&ntD)SU(?>qY?JmdiOuar7xu}X@!F!|?tLq-O7C0+PVUxql&vW--jo0=TYBVPOb
zqYWkQ8SgD2f+KpoI`n6P=!q`Q-ObuIRzn8A#`k#gLkJhnUum70FKidf*LU)}O*01f
zSm}RZAne1e_4m#P?@q!`?qMzmvMl)Pi%#RwE<4yCDkUQM<4KG(Mbr5Ed@t2?cOAE7CSOikL5#D?h^>gdyKkQ!{!${5-7kriDX;8Jj4n8rqBK4%1
z4u2U`KaVUXJnM(e{A-tg1+P-hTUz-%7$e8h6^N@+V5_3A-a1+n%)^{wLxwOE6XZ~m
z?p%n%ro-=sEZ=OvW@~$1wC`qOP!#7PFPR8+o9wjP7h
z-aD>DOLwC<2{}$`Gaale{!zK$LpF?0D}}3QmK2WK6eY27`ir*0&$iTbk0I=9P4sgI
z8H~j+rRoa53U=Tq*!ee&mkFu3P-dX
z`Sm}Bisnbo`^(|bzW@5}o!CD{$?p*?2Y56?y1+uL?gic(5A3YZAN;aU3*s
zp{E^a+`9ZTK~&C?4-6_i%=OEXQOKg^lNI#Sx`
zJP4evyP^ttURd=2@NYzat>Vq|`c5QDwKeIQTqsD76xG{(P=|b)xk>%OC=vP6wWc*$
ze*_vtK6}ookARlKQCAnnZm>^4$eOs+3yg1xM^@XngB$0_H?v%cQRm^^;&V-%i0!w9
zrB?DXkc0pGG9
zSq*~@kRzcM_T_vh$cXhaO=K|bwsNLp(o0<8&#s@D#0K!L^XsvhlY5dBHgCn)UJNMJ
z9elox?*Y6bPb$k6Ly(SZmxFdPMzWXt#A3+foSL;v5Cb66jsu5-U1cf3cSw@!#Lt?O8)&a{yxfJi
zEy($MjOjXzwV;1GIcV%gH=g5V
zh|flG+M8+^Qoi8vqm<7N{U-RAeysE`W}e
z3(jR|&ceq>+4CikYaU@Nl$3Q3hj-{&4Y8$BB01JO%RnPrU4>Q&U;16D{sx*<`;z>at{Ej%+EmK*
znt`-M%p~0+=;0>~NuQ_<`k|aHIr;$4QM9hmk9p$NEOa0>Ff~Shf-YrIc#d&PV-)F9
zFOqKTLxr!F#0+=$&?uU}{?DKPg&y5FezIM41>0lZO11c|GR?S;HfZko#KS)8^?`n=I>%rf6wxo>iX96v-y6l5XbVL!b=-XGCr$`43sh06SbPvQRFHeTYIzNSdXT0d--h^XY3@S$i
zCxx(iUF!W^FMsT{!-rqEcOmdCZ#zZDpMrJpqP%~A6xK2gDV*MvV%Fy{TK)%D1$AwWnyu@=kfP=UKY7eaNr(X
ztum`fp20@@jSLtgR$!+TMN9qX<1l6>i%1k+z%lNZ+X`wD;&WDLq8c=&;glCDpDoun
zG2bv}i~5=+%&c!R-z%LFpDkxZ(nBx{<4bmHj~LG5;_`hcK3>|znA(^%Iq;-7f-XxH
z`x6{afGqmyD1jnQrdy}sea#JArmsg?%-@?h)Re0_utx?*nOx8HUy>GX?S*L=VrYny
zVA1UA`YC}Yj(2;S`}-8*+wmraBYWTt3!yKgUj49v-wCIdT{nD(ptZU4w*>A{=r4lb
z{|IqZ#<7#XjDGTEFD+*Q3{(AO6Pzhh@Vt-vp1yG^~ePcjiqlEtL`5@D2{Y{^r|P*p0QV;^Roz}b)19yWvICzIe=f#$9)lrj=T$>tJGuj&_n?
z&#^4~W_q!l5csbDu7f?jD<)%?2rrsnf}Nx`UwtAZfb*;w)y~1Q=)dc4bAw|Bq5Mw6
zQvSzfs1-ML_enA*Hhhpe+0k=?{w6oOcKhi^h$>AuH-2ywjS$R^2<`2Hh8w4pQ!_>2
z2XXY?4Z|dGcTQ)i(|>DdA#Xaf)zL?&bFyLg>(G0YCjRdiJii6@ZfgK|GVov{(<$HS
zPpRSanq^cIhvzVkp!r-==X0>Xu11%P=o75+^7ma4H8reKt|s7Z{Y_XM_n6y^>^k&=_{z*)IGX;9j_$#W*)C*13cN(q{cYvaQUZ^Txv_bDh>AXe+rqFB~f!|SNxhQ-i
z=DjIp2(;~$K$U;$g<>}he+p%LLe4x+`8-twaI!Hi(G!AJNS1U-&5EK34GZrqSF6i{
zu2aGPW`5+O2aj*?{OV$czvXrkif64t1m=Z#f&yn;kf`s>UV%b$DZ(LtmLfZHg&tFS{m^XZhpL(*ZF|V*(DI+TgryDV
zp(gqFvn%n`DEUIbHYa@=s`{K_S?A$RRFgjJ&HXSz=x>;Rw456^8b#FRH8m^(DRvIk
zlN~goKi-AYooDnyUl%^M_xIO_@>+Xt1$Aqn$EpuVCHu7@7CiSwy?duf)3t!#sksdZ
zZ=n1=w`kmDN0ZA#{u2l&w2Els)19=Ls0m;0g!Ipa?@>IB$nowE=;_v=-L!NOT
zJl^~)w)}D)F@+t)x~W&eLc9RSznVqlw@0^a>_{x=cRZLc5F7-TJY(ur>QLZYoxJiT
zWEy<^QcEu@zy#e%@GJf_mIc1}z0Zd(R{&=lEjqa#1d!%Xt6aTWg%H&a-Vo!sfV%v>
zWSX{?h4_Cv)a>Vu0ilh=vcI@$ko)q{v?SlV!B(Ts3$J!Yh$Gi9Oh~TyDnt|EV(4yvLTz9q+S72WC6syqG*L4pkjy}}3K(4Z~l+#3~E=-eN^nL|-F
zRPZ~#-l>fnay7krBkrm)TK1^N>bM{Yjdu^>f64_=6N?vh%imy#Q(T`U=6f5BXNmsXOM7tfZUm&U>i{+fMj$F_
zn+7G9bo6q%kg5627tk|D4TqBLS18U&V_z}16KazAcp@=E0J~b=j;N7~fye{LKfT(E
zJ-dI5)S@%fpgYW*9&R2@Xkp8zF2A9DwEgx=x9_8==-0t-JTZ0Y(3Fpo=VI_Hv~XRP
z)K{t#a$D`J%F!o=7rW0z>D$*s60-5V@&V20##{5|aNAC(SbmEHPq>1nxy;~{rMNLF
zf;NKdVdD_V`}`lGp%WDTB4C?nQJLlEO86ERimK
zn8#s;
z<<)&Y_^$D|YqgFkOho8xBld|5jzIpMeUSo7o=uvE&-Cqum;0h^@5QiicdFWBBS-AO
zGk){I5j!@YA*)Hk&Ih-sPAnRK6~!2XtvB8%=)ym*ND;SvOM@R~hpj(7ki|r`4828D
z6ycs$x{pn(>R212Wz@Y?ci61*FE%(~)s}u&2iQGff=R;deOxd$#bzJgZdeFY*!K$kN*RYsU+v>7pmg}zw0-mT
zL^dXnlGYWZUkbkvJ`JC5AHv2PIIv-sAK0CZaEENh5X+Pb2Lqpg{5ku9WfobtPgE6ZdHSnp(aE2i4};%k5C%*_U!kJpOKD
z7OIZtIZoeU|F-VTtKArf|Lu%8>tCA3IHMJ7DBFo}xTE)vwR%`_#1GnSj5d~G#9Ht@
zx6v|Y!AGMYhWiftxZV|}xkrQVhHo1E`_7Cv{ckif(*76xwAsV&uG$hTG(p~L?Y56?
z2Wk4z%kkjOaO-i_x4~36<^|^6E
zH^XoKp%B6Im5^SyMPYp4I$t>~B!?FbHL@YCQpHusfAvtgzXn^+-0S|e$&PEzzgL^8
ztBsq#hChrT4LeaG{)t=ib$k@1hj;n8O6EB
zx&#|O|8RRRW(>OxtebA}$m93;snp)k5#ezczV>HO9l~!c>tZyrzGH7%lTPX*SKxY&
z;;&N8cX11UitCqt+{XWbt~~0JCdMy6)e#x6Bg9ekcaYC@li{a)Jsl{{2jC8EH?soh
zL~v6Aa=$f2MDbFTw4xQVd^q8^&~d%~4SX(F8c$5?CwRku4Xqz~3r|`yUTYrA#CU(O
zSQzNU!*01xy`lp=u|3(^f`=iWuoOWz=UKiQ%$VD?D@ZvRMisZ;o47?_?H~B(>hJi$
zjLsTbYy)fX#em`5n>}r?>06!gwy9i9+1hP_j3pWV7|Uv}{U0(;gC!L^bDKZEX88-R6u}_9=OWpmLme20f!8Ggna#vVzK~7nB
z+X*|H3H0Hn4b^xW}AK$TsKBDKK&#~&6V~|IKG(RN#4wd~GSn@)d4IAG2
zz{yXtj#hwbieG18pbY=^zVBHbXyjyWPnK&tlwefDOPzHS)_AY_x^eRtL{3MgnPb??74f=B=NoMT4J>=7wfeV;ygIHQLB5DhtKv!m5S#xF0
z(Dn;g&rjOuKpOX&4OdMJ(1gd|o*Y>Zpd9;3gbWE0=*X3=f@iP`M4{aI&&=2gZS^O~
zBOkPbCN)p(e?+ZAX{mLbZG+Fy+3v;tfUN-Z_r$sHiuu0KV$l_MqUdO}`|1A-o%cUi
zUmVA?vsYv#dqwtk&$UV+8QEKAl94?Mq0EfPC>aea+t>Qsb3;_55K_pdMA@6Z{(*b%
z59e{tx$oEO^?ViMVo79RNhI+0qR4-U>h&x+&EGFEy=N&&CASKY3#roG+}~ber1z@6
z@_RTSD$_!-k5n0vONSwJ`S$w+KJC~2XZND@~zW>o!MRqwJi!g?j*dY*>}!Z3bQE>=wt8;ew>*@}5zBfgcX!^)2)jpnH?Gmq1fgx$
z*w&x4J#Ou$ek00%7csju9v+NzA<1Gds=o^=W77BOtld5WU_tr2Tn|is$AdB7+#Lz`
z6IzHJ@WnK09`yE{Nk!ams!9(z3?e*q5!)~2IuNSZ1TjgT57=XEnU%GMFG%2ON-jZEim*HFez~+)gEi8MWw#2qBD7x*oaEpbR&QUxITy|aRl9Tp
z&qwEB{$V%E^C%u6M4p$-%0CujjUVX7xYOSvFT&fHnjDEhEWZ3DZ>$3GyUCe!H}5};
zrl)njnYs;0mc6b-Bs7hs%n#7cIPpT$vjVH^rVEH@uDGId?KEa(EwzyIVG;2QZ#x|v
zLJrZBv1+>Z=b^UNYSb+YA^45+tmNVcdGKX7$*S?#?|LbluA5oi1(?;-B)?D21xw@s
zhJN>$VDhbVX+4Q&z`GEYXb>p~%U$Mv(;aAog5j}=zjT2hS&}Y_v;G7~zE*owO-=><
z%>Q=2Waj|?5#K=N3$gKx+hPKy(>Z)qM$)}e@
z19ehCX`c)8&C&u`Qt&}_v$FyeR#_MyZGV9YSnz{E|2Zf{X87GdG7kLZyyJLcIS#I+
z+jgZoXMjdEt+z9sO|U-jR=)An3{Xwv<4jJAfn)2x&dFOP0J`isKEc=w$UMuTW{gyW
z6_9nSMSmZ7Ge&t;Wt1PEd;K*=YQ=C^qP*(9Q3W_;kN3Le`3?%#o*SRvJOrQapAD`J
z83x9x5o&P_A0bdw{hRJK2m(rcDQ#_jAERonSM7|mpfByOSxdH$fj+HAO;_E1gEIr~
zmG2nP;8XT0Pw$)pXoAccCB~==c+Cr?D%LGZ^p@^VokNjc4eF8i9^Lpseu9cZ3TTKSEEZotpB-#QuG4Hs`
z9(D%3P<}<`vMmel6(+#`!c`n)wV}Ib6gUP7)M8oY*IjVO1U}6o7a^2oMQ-`uwkWEj
zTMFuAbx;AuUkZgyQuy_h1eWgeUib~~t4VvNA!tY5#{@YK9hAg&t?{0%EM6zOn`Hb}
z5Ty`dcHouZ!PQr8%c*(#pss&~7r0b)(Le1Dzex2Za7vTmaxsJhO{Y5ltUvt}zW!PL
zXROp5{7$UO7){)0N)A
z5Qo)CSO0KuWK#W);im&!@YvINZW9AWclZ9)>rvy1GeHw8VttTO&x5^_whqvHdTEQ|ie`gCtLvWW@U^ByA10FS;BY&WC?Dx4%OWmV+3Fa_=Wi|Y!0zR(C7k^oi
zg1cl<6S%Dkh}E1=ySiKex|LiKek!_w@A${#e6$`kP|YvX>r)3^|A{1-eh`B-)f@w;
z3l18h!`Y$(3xPo$CsAm7AS@g2m9(CA17|#pi=USOD3^QgBiiZ+t@4^J*t|rJ=brbA
zS4Zdp%UqN4&@d@H8DbY=AV>=CKla!%VZ99FI1j_25puQzJI~Zk34D+vdV51GlxuUn0r9F(>n<%-sqy^pijZPPa*OC
zUayf6vF|7TYkYE?lF@5ZA`ix1Ccm+NmhOcxKAd^cN9>4^rmnGj+?_=nc+NKTIMyJN
zQFgm}QyIs3U2FFI*YU`clXG(72eDYm%>sFze?)+0!CAOXZv+YDmZbHOYQXZC#fdeN
zN)gGBY;L40#v#O82?n{aN<`hdQwx&St|2N1t%NvXJ+&+BpYFiV_B=Fp<=@Ug<5`D7l
zDtR%Q^lSyY_xwnl_4B*{rebWxl%YyX2s>u9EmObYyT(tXFAadZ*tfy-ajQ_=Biau*5fck#I-7>A^Herur0x{pb)2kWz{rz1o~
zO*)<1wbRzKP#k`aT!#^8u0DiSChIky=`!O9KN$NPKqi*UmS!pXc=
zy@Ft|Y0X(RSsIcsIXR2EUISr8oBN7i(m_vEEa9340?%O5Xk4xakS^0=sQ9J_huGPF
z9?staR`Ii`Lb5MFr`YOilMU-*eT4ar+L83|3@5u(2uo2XUiA
z@xp5p;GTnCIla{=v^ps&{+@OoaHonZ(O#g!Rr{SwGqq3RWSHLxE|CHFAn`JFKVbkg
z#X&o#@NtMn1~#>>5usvbn$>LXoG72?1%%&t7T}c8#oF4_P|t$5=a6g}=-v5#eb|Z)
z*H=L;DK
z7b0U+{U?p^Cx5j04T
znortK4pk73rV4m@1s5LNF0otj!RIc%Y64dUQIneOq|_ShX}Btl#seIPBtl5}8!_Ft$_zOO$4knqzL@d-XS@mQOF
z-fvqp;9AR3sy2d-yGIleyE)-i;YF4;yHWU6A!ZM^BY#|p^j+cy2TlC9zoD&uc03-C
z?Xh5E_ZXiwx_rRKk&d&WdkXY!F=*g_{F+X_Ca6^9+vJ+B*U(SDZ*htbtD(Hb38TFC
z*wL}quUA4yMe)xwV_sp#5-5O#7w_dY(f`~}Jv_2HhrgELvMYZhi~2@oKc{t;KnY`(
zPxO1eai*^R>wB8!=pd~RyY3(_>PemZXo&O@PNBp0EYB6-3J*m^M_?E}nJHxHD(rxN
zY`ZM`nZ*Z{x)A==A@BUnYzpSt!P@gteIOcdy7<_jP8tvUAKN2Q2y9`|FY2%T-^I`!TRS6sP_A+P7|hwe$zdZU_vH|x5nPCxfPj~#8
zYjXF&+EPG&=Mf2B(blatdo2vTlYiJ=c&xL_HpEl8ZAjtub=+fjiDG)S(vRV7L74x>o)#)X8(ph}7Ew9vaM
zD95Y%ZE_$DI9@RGkynp}Yyt-|LG$6@xpujyJM|n8c>0G!UZW8RZz1h-139qlU4u5A
zYYIru9rR*siiWM9imD_+$nfW}_9`SBcfr&)_kRp)4Zwt`Xuq_+61qC|+wob~02=4L
z`8Q1bc=73!*P3g~u%<}+qOZU(xZL$F2?X@Oo~p?Ojf76{U6ud$+2~;Sv-?U;l8*_T
z^m)9>OCt+VbDygUPD0QkeB3~tg9F@=CBgJgMuNj{CJnw<^})#vc@cX~DcE(+zqxRo
zAFwY_ru-J?fe&ZvL-;G=AWX*%=(fz@w>t>|U7pI|a}3e-202OiGGG1=pR6#bPQTte
z%Tozz3_jm2ziJM!L+KToUNiWWCUVxE!Vm~0KFU9IFo6@!-^1uRxniu=2PcV&VM^7LrSr*q;9$`6B*4c47&~GL$hPpFkX009PCw
zWkf*o^IT~ugH_}p(Wtg_Y7lb@dHt$vwHwLSm=HJ>`vEI|A2N19MIWAzxxuKrGlwy5
z1lD*_{YA3U$iEOs7O?eye-8wnt{_vStv!wEu27sKZm3{U2x?ce&!|x`0v6Vxpu}Es
z*cLZ)rg-8ch-MWo#`4}^aRlz6(7Jn=MbVvE-`nm;ck5|4KUoK?Ky9w?T7UzheHf`n
zDba)^nA`Yl8Qew^0&;u0_HD4aGX>gFt0qWKk)883N^>mrGU45i>bF?){j+{{AEGeM
z*KM5-A9^5OgwV~8{*Ku9%PNm7Dz77~mz&D@NN13|ZS9Ps&Uoww=|B@9D-dJdQjR>J
z@<*NiB#bUh*v$<3SM1ZA_{9>oc-`K}(innBvIcXXHzI`X-t?F{yC;hf
z2Oq@i`~}#za}2XoISvSY!`0%0oR`ScOGq(U0x>p6xi-}7#efvge130C#E!ka-0QNI
zq=JZkzLMwq+6!q!Z}v5N(qQ+w1f3J}e{BWBL17UUd7%Q9(r
z8`jMkdEs+K53(~7Klh>j52jl|D*EBF1YDR3=heQIk4*jUp&TJ^jx!?D--#&Or3xGLklJ&-%zQnzJc*4Q0ldx%|&J&+!2Mv>^Pl_Wlz`V7{%=7nTa9`zjz6}Q}Fl_PcJ>|^@x57>wP^#&GfYveVpbLKB
z*ES_hef2cfl(;>-)&C9osk3#qxT-Mj52)+R$YD%vL{wu{I_Zx3J
zzgk6DoCi*)>;J=Ua*szzD+mFvgr0(c!Q;99JgLO*NAsBUGrkkQ?P!5v8XtLIH9Nd5
z`Ht=(p9Vm=m>U{C#{<`qg^%K=6G7Aq=~HCA*--q$^P=F{=is6+Hn@4M2No4SmCHzN
zf?OJnewW1E!Nbe%E>Q=B!Oalcdsy)Uz@0*R#(KC0KE0GN`|@xW@YNpJs|^Z(2!)Dy
z?llXrmZM_v_k$CxRb;KA35WsA>)dJi5^JC>?fk!q_CO%R3T%CIjO1FO8XhS
z>2p{~TK^nKD7M?I-D?CVjZ$MC*}n&O(9xzYz38v@^%pjR2Ifc`$9Y
z^CDV;jb~mc6373BO}uWQ=Rr+t($M+2(>Ub`0hSX?Zs?_EYGO%IZPa$u&iuhoK78??
zT^`*dCiG`_1!L?yDNaTGC*WkIF}iCv^{8hEpqq?#pZ8i=a7&x@#=OISAo&DclxER9
zWZmmNX*;EYHvSIESl*OJGe`d3+;yeKABrlucRH>CzVnt&Gx>c`R3+MEwO0cTw)`*Q
zruAvmP=Ky@HIEoSJz_s?xwZ)+0{$&{wlBj*T88o_hSPYd-^rnho&}gdM_K=1uN%Dl
z7~<2IkHht(>FO63>%er@$lc4!6gbD#a7hQDakyDWkwYKZ46^3J6tJcW$oI(oBfZT2!G~Pna^d=PU{L1(akb?$TKw?+Xwc-f#f0xce^k@(veq2|?
zCESHwkF5zQeH?{6GgHsWPo;(%0x!<8N*`bkDaMcgoJWw`IXkEo*+(qQpl`?2wHC1s
zbmQ+jr3{xDhD`V28ur13@gb0#LI_;hL#^8?{3M0<4yuBb>Ca*R@odKKC?UfaSoy({f)*zIS@N1S1Lu=zkvj}p2Zc<*kLgX
zGRV~{?+|^_oevhr+?BG7-znH`2o}?#9~9!?gD`hra=aJhhB5d4eU-2=fDEBi3?zR8
z5!Mw}i}HG3%w0=|Itaai^qr6M>2mVJwAHxOpGIq7?`^av^SCY|92v_EB(z4@*_*h@
z=>T8kWj`_Zd+!FMrP}pw%q!;O`B@ml=}aC($ir43I7AfF5`COa{9X?c_gVByU&=w)
zf5=d5kCP)t!!%?-h8E-M^G%fIWI-Hq(Gr*?htc)(PR^8H$0(UI{Qp!j?L%`h~it;HFhR9ER2&qcYnMROMD$8(~_x&$i0_&LUdq?`G2lu2|wY8@MTa6
z-~Akg5uKE@Im7xM(~FmL8#~p4*r4w(u?-Ng7cZh)NejD>Edfd~Gy9*|PO@~ZNHi~O
zTbOU(J}gA?6H7V2t`}n2Hx#zQH7gJ*wSW!#zr9$q)IIVxN)pI*v-Vr`Q516fw}nsl
zYiF$AiY3Iy&l@p}*LW|;8-+bW%E{*hSCK2m{T*M`pJR_T)bpVCJ_X#38ksj5RRFD}x)+Saqrra`Wx+Lo0X!%-JYtd~g)*n^yimSK1s=6@
zP4qQ!!8?IUCJyv|z>a}pcHfTy{CV^BVr|K>?p`>arhh>c_&N7j(&h
z=z~-SZgO#u>+N{eSwjgDL@evs=C#19s8WSDs5?B&7A(_$^BBIo-c~y!6a|D2sd!v!
zlAtsVovl)0Hh7!8#8D^M1}mh;oStp}1=PU>Z(l_xgZV*O72}CSsJl=3{pZ(A(8{64
z!amakyNU8HlZQ-$Umj;7UrR&*TkXMoFZoc22u*Pq3*Q5`PmNFhb1#Nfxu#7P^($l2PKGSHc=MTc%+x%l}l&j4SVN7yo#K{
zmqJF=-r&7K#I_nrQt^_@DtiO$r=$+fEDpg;?%sO(kWpy2cUWPlM}U4`j-Gm0nxl%<
zP5WPD{y?FQSshEsMer?5PP%`32VRxvwOoHei3@AsGlhR`Q0r3>6=M9vxOti5#ATy(
zz-qMW&AqY@-+%c_>AS*+-zg|$dRuCa#?;#)d=#DFyzm#UQl3>ft4aOrE!7E>=u9T-
zTAvI)43~$qA^8vI}UFSosO4EP+zj5+P6C(c}9?P8Qt3JK&)kGuyX{X^`gr
zQR0gACA3a_QjV%p06*82nW8E50U&&GJpIjsKrzvzDpt52t}=~sE_(Cg$=L{ht-TWN
zm}E77LzDr%7)q!6^Trk&e6gh=xbhX0EhgM*+_%It3UU{$$+ht!$AHc$dmfa==1b7&
zE;`(e{cJSV>?Y790s2ynw{X(wE`{hdJv`p7uQ=uWd9+NU(#5LnEbdn}-)fdlhl+p1
z3Njvkfs4ch?MUn^)KpukBX_C+1y>g`&VvF-Lq$cndnX3&o0+YMP4)xM2$4HgCpw@Q
zVN))Hy#P??qczLqsW6G|9W~<(XDHMrM{$vT9bC^&iVR<^24gnopTmLp
z>E$&udP|tGK9Lc6qz630qyGCuBL`C{LnZIGrT~&Zr@Id}T|n~IEc|*=2V$A=8mp;l
zz$H>v&G)ea+5G#!f@Tn8)_pT5QSj(RKMX+)Y5of0)Fj=f%
zh|!?@(!T}7+ON;Qzh41{L}~GK?d@XQ6uMl-{*%boLLpOi*$}26_sLLU?knOP>6PEY4xU+T
zyo;jV$5_uMtxEh{LH<2G|0TX|5$kiDB&={df{>RRY)_s`0%2loh`xpqf`iJnFkN!M
z$}!{c;|Vde{h_pa?>JXGRKt~?8C!$U65B6R**wDT#SSwrB_tprKAIFaS07+sXR3?}
zo_$6E&4bvxti}Qui^N
z82M1d}e2qggG7wj1$tQwi
z@t8LPBR=1WM^5Yz5`zH|G#tIX%^B5!WvDlWpE+5I+45%
z=^`D4?WMOtbkE1JW@N_C?m21@p4f-TO<;~(
z`O2g847S)Kyk?%ti^Oc3$yZv6VRvi4>GId>A}Noz$aKV0k)#9nTfQ{|m|LoEi8JXC
z;<1p;G3+*pk?X#r>Wd}sC?$s0
zGu1Zwyop%(6F8}PGa2dj%4i~d&cgmM|L|3xX+*}NZb-E3Y$4HkZr_DUGZ8026L+W9
zEG(jr&y(p@A@YmVIUpgj9Xsch)R?Gngz<*XQs0o&05tyeiR76Vpx0LO)d%86peJlx
zI@RVj+<5O}s5|)*Qp+`d`G)a;JDeoWZfat%>ec`epQ$`>pc+AUoGsztuX@Wep-f0@
zzUuoNCx@T&%SI)CPy$t(8`7UtPr;*sh`!-Pc@WLfYAZ+K4uokc{AL1<5VH3htDPAq
z;51){$IdV%*eqM&W2-s`n`@^s&L!Bw=K7tymTFSSC~e1AGe{1Ykw;YO`;1U)Fuy9h
zMHZCb+P`X)Wer-~4O3xf3EaGVmc?Vd8W<3SM!!C9g+*DvjmeuP0m(D!WqPt*aKng9
z_4e~j$ar$cAVu*BnCBzj9iMp#rO)5kVb|#aUlpYm5P=1-of%c^)93-TcY^trXMEx2
z^w!7?&IoYXz?^eAqY&zx4iym+{|Jd1^O>R%M<@`-N0K(;3bI+qIGO$Z;P1dYk2=q1
z0U|8Vdoa5V&}Ltm`>bOHuJe*t%E`Gw2|teKvj5!%1dBJ(qb|A7H<(7dYXd@k;py!r
zK~hvY>ht1~Ap@S+ru_c)pED>SNSZ8tMH^RRfl;Qatg(4C8a;B-v7du1g9
z+G@9#oEEnPTCMbTo+k_gqKe$ukK{v;tbL-q>Ea~F%S{(2ny_+bL{DW^a*{Rrvrb(Z*_s`kOb8wRMS+&5SAkGIi?tHhY<
zdIV}|%rD3q9FF@dUii;jg#*WWD5-u^Yv4&oRi>q_7I=>9WS}(3b(Ds;l3Aa~553>Y
zltlZL0aqO0$WgaeLshmDb-d^lQP=+Lx*vziC_U-;{<9ll$31}V(kufkIQ6FOeFh2v
zynE`+Kc5aJbj+rfg>lppl{va*Xn89drS<(dEU6fb8yihX1bII|e>iv2wUZ>FQ~$76
zDYzU;;MVvt!{&o$7-VHQe|18&?LTHbskw{Z<(?$s*u0F#-l^O7YjnZw+%GFw)?47t
z+L=@j=weYK&6$Q5pN-M$pEzt>-3izFv-#1Iy8nT+N8g#Guhl`Zf#DTjt2r1(6(INL)Eg++kyFnjkq;Ea@!XE7
zRM;)>OMX%QE_fl3ob_~K6rAPZOB_8}4}NISiV4uzKK_R$M_nOXOsk_?!#Rl5i@)ScqKro4x}IyQwHi66@$lYT3cis`ejZ1!{*rqgij(m-={7(?d)ctcw
zrjLZ#5*@aAOvdmqr9=Kkn9OnBbN0lRybyF(kMRxs#SNrB{(3QG9t`*;g!XPbS%GYe
zrCT426ro1Tz1|21G2nQkpRpx}6)ue3mmYQU1j1@-VJ3>Q;4Vp#xPdquyknX~*JMi%
z;`1(N%NoS)gF@KvW2l#CPx_@
zib2!OHOB*rT}+ZGlbxD*6tU6is(yLUhY7k)r8$9*NPA48fSeZ((B&`cn$h}=kVY*%
z7`^)mE7dz0{pOhGe_7D1^UAL`nE!)hmoo;(egNWIJxiL~n9h0nogR~~$hX=_AN#zI
z80NGV?PlJIu;ZueVV|mu*?-Tmw!2%=vsE3RZrr1jKa3W`L;2_8cyC$}vwn!^Ek!!nSK6OVd)k61&ecZ9ASpVm7WEgfoVCHd+@?FdmZ7d(-c0TrpvTOwQ
zTVa=;f7Mzsv_^!Kg?%~wI+3=})N!T26r`q0_On=sFZMToEr=1iA~h1ZJJf70*h;9M
z{w3xug7!1ZnzKI-)ymwj3)#h!VHbX!_P^#RfP5o623#`wB3RMq?d<4g!uRpnl4qM^
zYD;=92M(%(gz&L2k%&Pe#GH7)zP-v2`Pw~1F-V`QHl|$Jf2r*mp*p0}Fvy@yEjbGB
zLtl*&e0u!`gr&}5|5<kFV!(HJ+p)Kk+w11Y7o|s_(qKSWu*9=wJF^q>ad-
zmMA0$Q$EQy{5B~H86aPm`d)MF2m0kz{Hb*YQ_qk2*wUkeVDHk}(&r2@clMfSKZW~}PuRL)K`jLU}H_(xxH3A2<~2Io7c-1u=sU(LXq2ZOs>v(8Det`d*U+0|(d8LJGKQWOntFl>oiX|(r5e%D*fs#-X^p?e9NqZ$O6cI!*aee}779QkRq#qdl@UVTmI^w3NszZq
z{P(N;QPtDjpqz*
z=Oal=*CUS)ds?Tndnn>=!>+;9KV;As<+iWBT#Q29I&^OqN&JJGHC`Wrbyi_4i?cvr
z6g&Rwtc5K>TMXCf+s!T;QAg{@5-l^M2BF9!gA*8Q~dQyKHu5Xt+9EM7w+$D+wip6!R3
zm++sy?wP`OGVs5x3ZI{S%fv-~jkV7l>n)PP)|LSxF6jP|Fy|__FUmS^8Q8?`jPA7m
zo4ec{fL|Y&K7Ypt!OOUTc8IeUo};o~`~1QM{0~p%Q9Z3ATJ|{XTOO$o+7;A()-6R8
zZ58-2=oq7f%1>Pr3%f&u8$Q&}5^Ge#pI2nm(%8D8@eP7EzCV+}>q{Ff|6@4Q_=K_yJk7PVQAK0
zcC?FD6}LY(bNg^144r&!7Q2K5qdO0Ja~>y%qI;=kXF4P;@$VMD2(3jnxHezhNWsAI
zcM^j25^fve7rp<+blVx>N0Y?=r8aV*m#GX1cBe$qC6d9v1%GAKY*v(#@S4`s(-@tGP9yE>-DVM_k*3N#&I^3tm9fz8v-N*e9jOdFjgo0W?o8UAvjg
z3GXAZ>oxCEz*A;INVPxmpjt+EH8x1r;M1h8DuNCzdL^9;SqV}^D_^~8_gfG^uHu?`n=+DJKwd(HmEbBp-a}Y~GwIGDkowQ+{Iy#LLjI&=XvWx;1CjGf>=5Sz^
z^+-5y%+w1E^WqqXaLpTPS!zBj*|
zkw#@-zq#}dXF|8b1lrqPEkaHbl6@}CQIO0Vsu|8z3#b2U@Eg3wf*KD9)!%#Z6A&1(
zoEa{Cg3JF+?8PLvfqRuF`#Y}I!a?TV5`E|6yq;eyovTwH%oZ3Al)C>M&=&Fa*Rwr>
z=_#f3r>c!%1G9Q%v(Yy|DHuaJ5Z3`%OxmutHXn1tAMM{}zY_~2+TNYn@jvFPu^(oh
zx)2C9==5eOAGE`hVY}N;1nvX=MVSMGb!WKsC&Z{?BMwfiNw|mq`*6%{uRUHAH4yuB
zarA821Mn+tQC8GG6b^1yCo}n3f}@m6v`3%?QlSwXOf8A9rIX5uj@KJZ%gHwg9H6drmd}k~1rR3{Uiy|=7%J#lsgQY3Aat=DF#3rbe5ClSyqnhn#4R^4
zJo+IDU!74K{UFZ>?h1^JbW-z!)D32-kxg%4lNWvN=iC)=3{M+>hf_E9&9*#R>8AqxL0A%kAn{{8U_$mEM(G)KE-Xd@
zQf>_Gu?(Mt5^p2#cMs1ZbF$gY+DaqXsk#+63A6WzH~ZE1pL)1}MMxqkiNPT<9#eJp
zDa}{xyJ<+sRkm*A*0=rbhrRXK^CY*~p8J2W(Xd2(KII4_YO|3XQ90(0rkuYAf3{(T
zQs!sg1QcQWJ5}9|w@$$}iiVs&s`Hrr{?x_=l}<$I#!O?Gb_=$yM`ZQ9p&mhh8g0LP
zvX3QjzhCb*X~C{>xiHii6(JUHZ_!^W%f>eF0TzeqJj8Y_>%=+5f#aObJ
zW5iSTkEtIxB6K8DRAlcL*d
zu14#SX|^8%`N|>4f8iIqd1I}yiieYoe7}s4BYo!6K*$hVFZ4Sw|CEj})qKgB{tghH
zA7Za%^`x-QU#I%h_(hPA3OXZ6OKGf8&W`BmbssE}`dw_}wk%fOc*(VGngx;Qp)DU$
zqQyqJR!9{ns1akEy;yV12{AI+_EI1bL#)Qze!tnJ!WP8O3NgPXLS_`+BnG7%5xiA~
zfoY>7)_K9ykdj;m8{)iILng+EY;4q4pUxw}nziVj&YU<@o1H949o$M(>-ZBtyeESZ
zct5XuI+(p7{F`}b^76tb0#U@PJsW09?8MoQ(xZ+7HL+gJ=6T;9wR;xr!>{7H)ij8{
zPu?wlr{>Re>J@xBP9W~(?95_ICDh-Hy;k<7j_^*9jlQ20CtUVxOqjBV1OYpeJvrG_
zHMc7!D^UqS1dM`=j?SurP#t;u1hY~NVfOZ0EzxtegoiAsH%M&A~_y^nd>UZLNvdmB(;U1rs{OkU9-6aW3JiRz8o!$WK4C!Fns+}DAAoImH))FxaFytQNycd>^Q8gm_L=6)!cdX5o3x<)gG
z?R2Zr`_z~jKmS8eQ!V}x=P!#rUDn!WkxNA`{#J5bnoY#6(tY^)TILa=ep=<6XJid_
z&)Xp&DC#H1nldLYtaHo>ks9&yzG;t*2AXx;xaNU`n#E!tt|Vh!P5o>t#}|zcPjoat
zPa;^|)6#3otDEj1Wi
zku-~QFK>U%L5c*ik*bz;OgVz1;}5oubZ%RY+{`|}xGyjXYm&2ri&KBvtH-T?-|LRc
zKjoXS@J@?&p>IDSxh#=Y3uMQInW;pYBSj}b{Q2hb9|g+b#|>Jt0H0#y-{h52#RE3fLP8%;H
z8bA_#O`N2pD8T#sOx-+cFOb<5xzxNTFR@ZSVxfhM8pLL{bpt&&fR#LhGV4R@Sj&ZP
zq+WZW;7<4#z0>Y~kbbH2dccVg;4XgAv;t3o!*{M2K2CcH{Ofcz$D;tq+G8uyUD1S(
z#1PMm&;WeCYfnB_;{)l_LaUx-roq)<*U8p6A$a%lza&>f0dTV!?OVBh9uA~#pRA_V
z1BK7mgg>M@06*-{?%UL}!1N8JmTo8ql#V<3=W7T*fb4#1G}qMOq&m&3=oM`!&~QEe
z=#&stpHD7`0Q!3`<3CGMAl
zKyd5_*Pr`~Fyi)FtHO;`U}WCthuc4b*|}fSkULTE<6rZi-_df|b3mB;y*~tKtj-Kl
z+>C*)i{zfMUJ2lVv%XWrwP1isub*#8djdEQaK0t_&#>jp`-fGV5y0S#T}JD4JoH}V
zg7U}D(;6j}Zgg=5riGyoD9ndIbjuXYZjc0?#_-=}h~jyiLb1x?t`-eyAS_qZcR~(b
znrW^t$*@8Bm3*U_)wbb=bADpDF*!c-rEm27XXN!?s1G}OSEgEot&s&asO%eh9Y&0r
zH^#j+^#28dy;NSGHD$&>JyN>NuzMQk`0^`^wvGWmFEgOAxv~v^zZvS
zInEYVDSWK^W;PzDjJHn)U44H|2G2jZ@vAIo3-KGHi1u-A@m7}lzh)u)tE1z>{m;B;swhc_`;-j6cWsNKy^aRO*gS+3ws$~3+@opu
z_!9^%!7zgt6sSn#h^uKIKU$?0D6@Qe3#{`g4xcs!u_ANi%PuU%h*@oql}8)|LfBCa`vJ)sA@tJH-6n<-W;E#z$&~7yPmpPJ@!!_x
zQW$nkpsiH+1MC-Z68SYrjoLX2)9
z@>-R{bLJg0BZ*js*{;(T^C*CWqCLJZ+5kCN
z?+Bzzhd0mvDRA0WgJNcV$6Zleac7CKw=f3t!+_I~%E|e65h`+I5pnvt*#RXs8@d6#->$x1%RkPQeWwuIbh5
z4KPnxg|6^fJnT%=m;2p68s)o{^oDY}z3Tt)E*w
z|LJyn-QJ(yw}0-p$K!Fo9^*RKIoI{nIoEYKhCy<#0+GO~a>bzZ4DwlnNZ(=3Ld2&u
zwKjNdKF4a0S0*u;5I5|qMfUAEhZH(K)UMibl2G06%6V7kC*0SRs9#oMjC3xUCh=tzc)#OlqqGm$R
z{ed(qV!AK)mvqbw?v(%F=H6hP&2x5s8eL}LL!@&u%T~ws5LfMy$klieUxRaB#&*NQJe-@aNR{G)luAv}A1B}#IiI*OS2E&drfTo#=6=#4
zht(a0SZeT#Y#LpfJ8N-Y-O<)@DOv)&>F#ZaTEu%V@RJAgb-;?p1N>qb>v7pjfto^i
z23#vW7(4WQ9S2439Gi92O5=SW?8Gqg
z`&e=YEW(&wBkSmmTV(Kyb8|SsH@l83_VQTa65bjjJvL2vW%avh$h9o2*{>gQFENA_i(fN5
ztUU+cy3(_#7f=S{$z6rC%z?0~EFxDBZj49Wel;1Midrti&x^1JcVB!Eh|y0xPzMrSt;j!kir8?V~ZZG
zv%(Um?(`TvV8G+d_sEF9riJG$6}bL*-GY@{&JLwD_!R}+pK|_2WA)S8Ky*6hA@R#fN8cL5hZtkbs
z#Q5v^AXH(B
zNlua3u-z=fLTvm+1=PrJ(R$f_eV2ADs4Ppw>P0G+^75!oKt(Z@cxg;0n!^nmzQETp
zBToi@FNStJP#lE31|OyPKCOjhvjlJOoNs}G$;h0PoIfK=JG6l%7fTj7a2ck6+u53>;zg`eW8>%&AjqR08yG=4tyur0%cwqiEG`l
zd7k)#k7xEuHA76&&&;xm+_1OOytYwN6_}T&PTuX>N-RfaF@oB+3e$RS{<~m*2}U-_
zeX?2D3nLtYpkj+qOh_faaK_pWqO`OTjo#cxIqRlr@>!WY?8Jd(tz*tPkcwJ_{pFXv
zkn`AY?OWOdP{(CmDMV-xT3dbp(!8+~Vi}b~q>P?I2WDvP-})CoqNQ9O$paP8u~Dgh
z^Sg~$yga415<4e;GLI(W9U!2DI-v)8FKZ$B9XNiEsRlCWTnf>w%!hiVv@Pgn7BNBP
zHKm=Znb>TNoKw$q24oZ-c=AhaIYiTYd`aY86*PJN4U=W;Ff{aOYCk@djostR3ygpx
zF{OKZ|EM?`LO;}xu3k+_#=0GyLkjlvVF>T%QEmPj=vgh3#YOI9O#MeqIa_Qc*7`yD
z#L@Rr7`lJozO;kAkWcN~3ypDHxGwoc&Wbp0xSH%@_0swY{EzA@@}CEl;ZK|asyBne
zV9sAEvNJxdShu{R)rQ+LbVF>t=ZL~8rYa?vACYwg{<9EVe6!39-pD=CSH;*48LIcI
zz7%c|Vs
zdh`DYSyk2`uUh^p7cvB{b=N9dcs~GH&>OPo=yYPzf0D~2*XE#b^$Lln5IKH=*KgzV
z%mv)K$EmP!!WWKK3OX1l`3TqF6Zb`#C4O`NI;ymY2X(mi(J}MqiDZPtbW34-v=poy
zP_p1(ZiHvAX0jbjzXRJ$E$gST#^SK6$~jA92zP^wih5*3@FnB8oA7Bl7^mhtyz7)Y
zZhJj)r*E(`%o@@4{%Bn}94YiRrN){M9v*dOGrj@g_qeY$Z0kF@K?nT_v!D#i&6X&k#u=*!U?J|%fS6a
zds*Pw)Qfb;aqYSvJ_1Zc$4aLU1(Og$PqE4)OlTp=-rmih+Y0d`=^FyU`vzeFC4rW8
zn<;!(QO)LV61$Mf62Swy%cl{O<l08_4NBocQ}uUV>AzGV&WPfJDAYbtZZ_|Z_}>MwZ*=cPb`N^_w)rYeefGlhug6sIF%3U_|exA7t&#vf0=H1tL&
z$}ZbV8y>eRPP0RFO&H(am#U)Y&HoBkhUD*U0Hw
zuUZhAu3wrf&zB(wd$L#V6j1_t{&&GW9n9cKW3Y~L;eG%r3zvH~eiJRQvSf|&96^%-
zeR%t+0kQQcY;KHyL0o1LX3?}BBdCQ{{9o?xB;3uVy)H3S5;ig)VRPp+q;voDXv3~H
zWVu%TS_x+h^5$JEmuFfjVid6`l%$Y=I7a=>xOFg<;L3fpV)g7!G!{9mcA?PMV=X~#jHoAhVv%Up0hFeJB1w9^lE_FugQ7h-
z2v?lV(^-=tBQB%Lju{M4=}qK|K6<92^NVfi{1
z@cwEeuB}DKylH494p*mMc2tcZe)F}@&Rz{h@=vEb@B31URFJpT##ZCVlj}Sg@Dhd)
z_Y=L|ggrzeX*dp<$p#b5RuOI=^GgV_r~}i}&+Ca@h0V)PwFzP=EU9eGff=0`QF0ZZ
zT1Jiz(CR)7tRSxI*nE+jNFtu_`*>Ar^%1OT13&ZKULhya)g9Av(h0N5$`?yc`9z5N
zb+&$!7$V0k0m5{$iJ*lW^F&iS;h6XCT-!zh;T1l?6x|$%T==zc;rc~;q+3@uvSY{(
z>AZ5k?K%z+v4@56nQ76;#q95fMTrjx(PO9b7`RLb^ZSwmV<|2KO^GPu;-?0}2JdTg
zElnpVH=_1JKW`x-{&&Y0xGy4JmjY^7_3V+l;M;-w90v)?;g03j)f!^<_mZl1X)t1K
zax>-qygKq_HZDTB&KfDRd>$IRKacP(m+Q&ca3k6uuG5KXnIh|RYU*Z@(!_JwQgJ;M
zB?Q(w?L1K(goM@o7TTR;iWF;K-oY^?OFY{RQJ)lIM(po?VxFftN(lM&oKYF}AoM(Y
zB*!ZCiJ-yv^lw7WA`?v(E@~T;1b23mwUIFsqI^Q2^>^JpB(BN+TYkL}V!4CRlplL8ae
zZe#`Tk4WF;DSQPv$hdr=eJ2-^`ceNpQ#TbcIZy8JGiw8;I(Lx#^r2-uXkp4^a7B=K
zWH;oo^CdN**llb}56{6%R)V*A8AfsbqKnjwtu63IZBjS~BR669kYv%AuD}hNlXl(u!m|1O(}XT%G!lKfhSz$na}vK(~jQadhp
zNlS<(>yp0su+|MKTszcf4dm`*{G&>)SLXUEP
zbwiG2m<@KSs-tfd`o2PO^0XU0oD<%mbUdpYI^})HgHxjkt3J}VRJG)ZMWJL1Ud|2J
z%OQ`qC_5z%Tb=jt-2D4@>otCE<{fT+znAKnMTx}>4wylg*(LjdLXLEF+*V`
z9Fl*{6yTytVEu`sbo~7>n2S)wzSgQjEQblAAa
z8W!=y6o2z+J*tVtB5J7Ryn-AdGphaXVhq$US^E5Ic|DcTqwv#8q|sX&=#iLtc+X;z#Y<^K*)&?m`DV#*_wLv_eW-MqJ5;2O?
zdryBGsltTWf0#dIY{aI{>EOdMwOE(om6&W```e$Ab0^>XVH&vA-Mnllokxhj>uR9KFUk0>R&d@jRw(!{^Rrz){;
z@ssQ$8rfKGOdNhnwiN<gvmsPUtYZ@rC-jA5pGk&43cqG95aZ0YVWOxW)267Vp^8+^`o|Kd=>Gg&3pj`EnmyYOSSMw*-ZW}eFGRFE0O
z@9(xge56_m4p-@qxA}Pn)_fM$F6~1qE5sFi
zxS#*(b1`T5{bZ3q)tNl_;@8Z5O#z3AQ4tsW`6yPTlJO8@&(s)hc{{Ri53Lk(qQN@<
z%tZWc~4`4soY$Y+n7Q84;Fa^QkbDoQPC?>~n>W8xf+2
zVlh`biY)1Sy9Vw!gD{TD2AcAc5!6@N->e&bgC&jb$4%ZkMCc56b((M=A~d5DoUSSe
z5w(<8Y31&7Bk4CDeW`r<85fj+smB|*5HCTkN|6CJ#0eIATR3q7IsE1!i^CUrBFKkA
z)4L!R@k&^q++kgfe7dSi`}=1FGIt{N>7GAnNKAgV+KOK=lDyA%VA0_=k>ixF&q{V1
zaj_(OtyE?S5brL6jrmnM?!;%$V{%OGouBByM71(B3f
z)$T_t^2G8kh9GN4KlaO?
z^CPJ2f
6%c|66R$*GR}qCz*Nktca3l
zGbffHXTs{rwYO61cL$H1DjWg1rC3Kz+JS2`RZ8_;vcG2vSRt
zZoV>yBJ`pr_2CspNUEx_`+J=eh~paz+M~*X#PhG!2IE=`M1~G^eS#Gi!F{e}l_}sF
z@%~Aoru}DrqEuo^?%FGEWcqM9zkKXYq|LrYvbm5M`7T4tvDeZFS^4rUjE)IK{L4Md
zrUJN$@oe7<@boSsU{cgviid@mOZdY6>a7#P>cwj)Po{|^=^2L{{V7T~TcG0tsbq+?
zz^f;_8}<><8=VL>-wOz5k;I-0cc_q0?dAuW-u}jUJLH(v&dkG=lnS2me6u*3BjJ_E
z$3ukcHk3Hkt>G1{bYnulCSjWF(d#YGn(06r%kZ^s
zT|--V1b=nYU$TXz2JSD7vSg^P#P3gjx+4N}5TS1jhCE7s;bUjXbq}+=gzbg}&5sC{
z;w-Z@m1&(Nu*T}fLFci9#9?WH?;Fc2c=9P)<(Bd1@FnTS>EGWQaXY#;HO<)~*!QVk
zd_>>^KD?a2S0ODEFD$M({Zld#E@;@r@cB<^Og!50{`}*wcRovzs
zl|rCcHN2{Rp(=;{5xg-FaD$)P3cnDx`WenMgf(cX*1{7t@G80)I`U(W;1j|3AE{kb
zhV{z61<>nD;SgW5ca69hJmkXkvgxQe?%^X}AG2bOYw6P2lg~)vC??dA2-soEe0Itn
zT1LDgymi`&d>>p>hnA~!8NmZtB9>;b2yD^E=i4mIh*Q;n^}@ud;eo-%_nVt|;g&R_
zf{6!=@SX7iwQU?y_*M04#p@g_@MD_Qv7mi4c!;FiyzvJz_{O(P`PBt&yt{(dU-!c?
z{IS&a`LQl4xXw3V!@Fh;`$T)Dk;-iyPY#Og{Mu&>9W4IG&nl02bW<_53pu~hKh3ExBL+NAo&OWHGN9-M4SwCd|
zK0Rwez8@m*-Nk&yzXkI8eZxRf?->+ztu;PGy&C#FUE_CzHXAZ7RyNKdtAJ=5lsai0
zN+Gvyfs-j^RanH2G0qL1RP6rD&zHOw=~(#lk9YZ!N-@i4Hp7n?%dtHH2}z4xbdZ@%
zx!>ou5J>Z$iwCxm1T`$Gg;7w2LBA!qdX$}0p@$ACdqbW(L5%?%*ZlQmv3Q+B>4|MQ
z7)&d1<=mc7NW$(?w1rVEwQJPAxwKj-TakR7hdNt%0$)Eh#s1$=V
znQc=1bs?LFn!lq*xv(mik9MvjQIJx{gSTzh+#tGxZ96%SL_>YaxXYSe1vd9wny1L2
z8he^1;cXjGj`@31wcO25#W+Hn4Geq$u22nI2+*SJE4IffzJ6T!
zX_yL8>+u{|iS&mIeweg)D5heM^mhmEo)5yV#!s$)uWZ0Plvs5NNnwv>oc`YJSczko$&c2>}~Ac(g^wvsO3`D
znvO#Y6kpH2Aa2zPd6~wZ2gNN=YtX>?8-EBWe%RDVq^uocVpg+I`Oyje(d$eS`_l@w
zKF+Q@bfN>wyv7wRsn-tmnm>wMV#&b5__TNW{wl|)cdd-%^j2f{xsL)+6R#rnf_)i^R^7V;J<7zD(`m
zW<47NSp!ce+92cE;(Zb`tq>OQAg*h<3xY3lax7_gLQBn$%7iz6N15{bCC(mghq%Fl
zNv&`h#?AcT?a;nD%t^llHfk=wmSQ5gOpVGgl{b39#~HJ*5ATcP{6FSkT%VY4>x5Kd
z;;*X2i`jFra#L&Jnf_AD=MUAfU>L_9OZZE^G_SyBMr`{GH$Po9m5bnX=BUA%=)!1)
z_EckUj7-;~1vZa~vl6xLJZuM1(W%Z<)MjBEAmZJ|!xD&PdeQpk+d1s+==D451RtJw
zMK@lY;0V)hbBkz-5EKeY>%`|tgmYm7V5;3
zq|niXU3-3lH(7lEqZ%=xZAhrc>~|M;C|s(?vPIsqZ!}b4o(zNZnZ9LMg}5SbSN1b#
zo&0*{+ohAR6QXq0pK1pjneZXYZyy(arfw%~*19H4lOed<`*td9VZM7%#%>pEey3!J
zqG1Y4t(*H}5IGF_z8#MU{&@t}m8@u?$hir}x%ZmK&dy?atZ&JFq-Q{_LnZRr!ll@i
z7152cD?4y~OEcw#ZaKV9X*r{(Y1mSLycYYPZ_p__QR_>_
zFr3@cxEyQdx#r$HIILGq
zS=B2IKbfmsc+%<#Tzpk|$H5(&-<|w1e@^x7HGZHhV_wTj4}WB^S4rH{9A?AFKG_6;bc0S$gE!Z}ST#
zA^Q-|ew$*Yh@(hkW0f8C=6TQsyIyOb$ScQhtkCZkpz?v`-d@ppTGIv_dB)+ZJ|Ey1
zE%wFr$27?ML)y6=Tq$sM(ZM*+giJi*2C9=8Q>B6M^Rm>T&EN4iFSyf)m2zhi1zsv8@^xoTbHGU5mEku^c;
z+xQ#&f=?=(Dvy}Ja+j^K$v<_6IwZq;P-_<*#
za9ldU$=2)1EIuFU<;TXogoAumxq0r7_(uh*&37pp@!!{^)9;_ChMkQr1!tR2z!YIU
zEs*GYxa#U>A?~bs`0nHN$H|h<;4_^457UQQ@z{kwcEKAT@D+{8>xqxPHmoVxKtRy&mPM;c)^iQOp{u8HtirC87ge;I7gDXn$8U>`B<
z^O)u`#T=fP>K{Cj*ah$L=zT34SBxi8$4#^76u=03!<*N8yYccj5v*CXv9RVT`Tpw{
zg5c}R(m%i6_P{;Gir=?c-GfE-E|7i8EQj?xC9Q5&S;NqGuTOmackp+&Y`mib4B@4l
zt(HvYI(Wr{J>ZZ-J}z~=%5f(64jy{bWMzWa40e2TPvfhJCeBlTO?=(upr~al`A_>0>4oadA5Mw-MFV6KeeU
zYWOoXDYAXAkTYL>k|PEFN`pq~$H@&ycwdmkyZajW=UAIPINM>Iqm}7E%kdrXm}OpR
zXYO|_DfI67o0Q+6XFmlj)P8f}tlb;
zx6trX{My|eaK|mHs9>`D*v4#yIF@wy
z*LoDhgqy!W?+-vN{*OOZP_#nNIi{YmQZ++g8e-|H&Ne~M%zsU_J_b+|&>Rf;)r7T5
zQ6k2k?o!f|
zOzeBL=@(SV1ta5Lp({U?jeTrE>iZ64VlPe9Ko73+)Y9wJCX!1NoIQem))53gtY|%t^Fw
zfRgoc0#BZ4hWf}tfXIOyu$`SZvtL_=M_%HtT+|(o;8L-qPzjQ
z5`0xS$VA|wHW_eo=b2Sj@f1Daa{g`VJR27^dB6@u&$&_?c=x1(RDhHW3o@=P(LlH
z^^rkQu+v-6wSD0z8lu@EFkZ(8TxssyJ4xXVZd4SV-;gi?+DnF>b=44Rq9NXr*C7n3
z3t8j^PVu43r{gm$ID^m@nIH+S#||jBzza>&wu>ORr?f6Qpp3Wx`aKBNfdfQP3ie}Uo70YjdA~{6IF5VE`
zDePOUrRoUin34vm&UB$6H4afxPqNUp{D+6ny>J0@zw|jDov}uR(hH9tFT4Rdy$@Wt
z%pHa9%aE>DmGeS*l%;RIV$cQ$tYXzgXH?KhvHP0w32NYupw+}xeQ!XS7FR2PIDmfi
zcF?%2A{uyCG~QbDJn$VJWBYSo20amw6t^?ZA3eUKciGOt0X->KxQ?x>f|G0#jNZL6
zXhNjeYnp3H;8w%Y$;UtJP{H{6O3w`oly=bk!H-yOl=#j)Z8vQPejbev@Tl?z;VB>b
zPeeF?hPK}-#x3`Nyhy${?G1O(crxGn=I~vhpF__6M9LIZeo(`uqY;8SUCQ7*Jrsw2
zV1CwLua$wuv88@9urEYU%L#pCm@h$f`NcPXuy&)uYx_)->Y7m)>%L%p1)u|MUtDyV
z8qtxXXEaRK8cedT_LFQqSBGul8tD5M_|>1ga$P5;m(6paRi!r=Lx=%S&yC*&t;xER)>mZ
zZluLC)uYjKlr;$z_2~0sS<3o>IQsdfZKcA_2Gsdp^O(V#1~k1Z&5qp74Rzc@8S-1;
z4$Ay!=1bxieN@&mw63*72K`jJr0NyLh>qdyr!dGJ#Iaw%*wgQWQK@@3-sRl~JPnO!
zyeVD5F!SBI1r9fG^?vd!U26!sdYEsJZYd0{GR4cE?RLgj&Yd{QW0(6&fEl1<)Mkn@-vL@fW4lkFsR8wK^}O=+
zQ3IO2d%=(CIY5WYo}CG=!O`>m_tt*z+Pof*FIuH!B5177;B)0eD1Zw2V+0$`z(s|=
z#Jfx`;G%f$mrKIlz`yC(15L~WxIJlexM=kN#5}q1W#+vDaQI-i+T0`oYSO0ULO$I`
zO@vu|m)X)#x+K6H6#f*Q_IFFXnVE}*PlvSkR+XT!=G0!|Z_3bbL5uolI{^9_?-Bl}
z)PS~!DtuOZ)qsXyy%0BWuo2Dc!oL2YYD9PJ?UQNL!_j+O5-P3P82ZS~G9b;P9yK%N
zOrf%>L!Z8nKgw-fhk6TK8Vabq50WcI3;p9Bf}z_pU!^PUfyUQ8lv1;H=vwVwL
zsB1L4Qj)y~+Q|UxA9;2K)TN!WeZ}PmYCXi&UjOk09{c#-%y+wjFSl0=d)>;=+TMF#
zE6NMez2tTNdl;Ug&(iDjA5q1koUJC;I~E_JDegS9viS+kfiIy&s&(ie33k(gsm=Tg&J2VhII8b{AaUtN16tal?LYgv
z5&e0vahG`YJv6EH@hqc_F)CpH<|)0944PT@+VT^=6Q#cY$n?t-L3GHzpe4!E4b1Dk
z5Zd>}2~?iq)S(Ht2WLlkqb-;mZb
z{7p580-9?4P4-Kt40`sBS<8ZkJRp0!p@uv#2hk#GugJW0L5TW%U!w;Pst(cew!gWB
zjbD=@NwXRt`ej#G>JBh1#
zSX~sTqa|cNC={U*@cEcBy%==tr*bfRS{6tZH>T`LD+jLvHJw3jC0MbgH273ofnGbB
zo#ApR7d^-3LDDDVKKe|E~zl<+`R-*!;7n^&O`#i6XMCd
zd}*MX>pW{q(qquiU^UU^mI8z$9=&!M$_KLYnN^*wRp9lg^Ip#k448IJ&&yU;g0k;9
z7Iaj#fb+%69MgILg4(Srasrybo%K)=Q~hQTC}M^<9IgRNlJ#s~hw8v-NpmHt=bNwN
z*A0tL_%(oE)r@g>78?LI^Q~E?T!iwQ0gjtaz^)_pWbL`
zaX@8U?=l=uP6HWJ(cdTb#{v3dr4ynn4*-*VI;FzUUDWGf#OCQkUZ@PWK$&eB2B47V
zADkNML0IsMsG~B~V8Ge^yM{|CurLqRB?~A3nvth98l;-R;;Lg29MJ+iIMsae^x6Pn
zYz4-G+dw|WZmI(mZ9u@+xm{kR5!|_QHRF4FBe<$;<|FyM34Hh-dv5q(3rLqUw_xaN
z0Tt1w>>M(j(YMU+h7o;#v@vYlOr7Q-x?|_bASE7eROuX>3Dj(ejwPq=I0|{9y{^I?
zhlcG@vD9|3+UfwR0_%)rN+LnrF|2UmVm!E218eS3DgdWc$_=80NcDMzcTdsJ!H@*chr`9u{y#AX2wy(u5jmK02t7mg_KT6BNdrA*JJAp>Q(>
z5f;HK3TDLJ?Yk$%P^wmE7xV_c3=pdi*dOhtJQlx!ID3#Y<>R;0^}+RFg_U^B(|r=c8Bn&;&D
z<(Gi?+{<=xp-(};Ejc&T+zZ@Xt@R3(e1MK9JM@}|R&UM+;b;4*%E9ByE#=1#<$&k*
zeFgpj5n%3PDZ44zWssF|`(bj3B|-~_@wqaaNc3GJN>*3uuz=N&ir8yJUo?YyB4j0P*Ag>eYpX88Z=oM
zw%DP1oZW->lDtrt3NNk;CBA4HDEb@23Z>eyB<-lmbGp_JZIoA^Ky-&2E
zyk!R>o_FlmE%pZ8CIO~rEBt`>y_0ha&r#qGv@RA5d!n`)MZAhiugYHmp3!13(AUVRK73Jit^y24gMUBosk-qSv1*JDW$~tKX&}S!d
zoq4qz&||kNFBo_?qU>aosi
z@&(^Klto^*rlFyCM?9xHGtu-}gzu9_3EGeU*m0Dl3T3oBf+v5dM(g%?hUV6{ph8dS
zS1H0<(VFG=?s{dd=!eCy4Mvd`)bi8W{*x1%d9E18IB7JYZ5h6G?!k>{#v!}-2Pq9`
zNRDgB>Ael8euMp?TYUiSeiA-ueCY{rL-${2=<)||yH%B=eg>hBj?F2XPKKl7yqr0y
zyMjc3>tfe!=})6bS!^E86jF~;
z&NPa1!^;4{$*&xkD@0LO8|{KEU9Jk8$-p<9bRGW!O>~;9Idaf@y*|JBQzzU
zb*NqzRUSL+iKdmK8pl`F(R9o6kfy{9bjnyd+Tgt*y3S0-su`z`vW`=pc9}N?HxF9T
zM+Tk+mA~r6e~oE@(aEry50nPL@%D#-#U_2AqUhI?d)XfS#=-sY(-T89&!$W1eY+Xz
zSmMq-a7OYXV$L@$8uDp$=JN_NfeqNf3}xf@hQkFRuDT+UkV~9##JN51yNB^
z5BsmFeCW(VU`n7EJz#Jl5_CQ)pcdk#tU{_JYQ!zjP~B7|yHgYWAa?Lv
z0IMV6h~Y7rr3CbY_d}!2
z`yFkTlz%&WZ}S}bh?hQWa}MB6gZ0=8w`d?`TlQnjKMTa*rExvR1%TnZu)l$SEl5hf
zkn+&54g@I-SLn23;M%^+VfE7haDQOi!_ov@m*EhcG^*m)`DgxMWE(g1JKsPC&z|r0KMHj{YX$9Pzd2Q5v{BT
zX9^ZR)p}Y1I`tYK-Nlk{=CbF}GowKIH@Sh9#AW(cCh~JFh3|PJlp2>MGcx!71%_%3mB&S?fF^_zeM4go9*Te&ImQhu88J0|peVzEfu3sspv`&uKM9YC+$j
z#RzWKdZ5E|V-H6q2Kb+;((UDK0Pn0H@&-L_1fTD;z0OZ>0*4A!^~nn@!C;@Zj0Twx
zQ0i_gDWQu1vV#?B`tL%(>2yUi!|E{fX2(56%Om4^s{F|lkzDL8k49~I+X)h
z3R2GNuhf7`a~&6#zSRP?2Ql5_W*A766_WUvALLPLiHvsp_vOI;-CQ$r)*HF(~
z6FALgQWrzh2xx@Qo#wcK0gRw-)~2ilZxUnHlcP*Ppx+t4Bx)Zpx^a&4RK-OQv)n@E
zc1|9od?DAPnLM)j_odE!3ei{;HNG3>Jrs(zc6%PVM}8MQpU9k~B(WdmN$-fh*nJmN
zcV(6ym&gUybyk0(^3#CHRFKxni$|bo`B32JYxW>0Tk`3PNe|RM%F%hDya9aXc-lYc
z+W_kCyYBh8L;&|VcXHW-4ZvDiCn&Cbb6xM273)rm0XB_e^Gn}rKyYGyv9wt|SWWn@
za=oGnNZlUi<=Q_Cs;ysDS-+Y`E4H!xAGvH>-X#6yk&x7mgrs)=Whe1SasC&N6!+hD
zq&Orb@kmJ0|BIx!BqaI$7i+icB>DSad6xc1zmVFIuw!7S@wPuw{iMA9uQG{8%8wN1zmUWuU7!Erk@Ea6J4sK<
zpVZGJJqby1{>x5^Q~w|RO7cU(|MEk!laS<(%RT>
z`jF}i+UCV>^T0p#kjBl+ZN285Eq_n8Wqcb+Jil#u{`MAcr_B~7Z0kF>QGZ)+;<9Bw
zxa~J{Td!=nrC0RcLh4(uH@jpW&E9T5WVNLa-p>1-=axKsY>T(F9e>nrOMh+K|4xG~
zSrH%m>o4r?mMm+!g|*x5SbVl*VZSYZ&tcHSYp+u7`ipY!w&JpW9sVnaJ(>QCIp)v*BC{U<=Fpi(53+0h
zMTN_0fAt+@tbgSSzI%V=KZ|OcLuVdkjkvaXOU|Q7lXHLdoNZfqJSMj4E?3&Z?#o-~
z!8-Am_quwkA8zGs)m4zSWpAd~s?UeA?=R2Tc&k3)zAe4>k1f7s=oaco=>5%WMmqa1
zKX!Sm-#E**>h6r)!lwB(XcJM
zZr>N5w`Kore6fvb+bChS71wH8rr*9ly|?Ap+juZyix<0XKe*jrP22L!wtQ@Ti#ND^
zAEn*clJlIlP<;EoZ`t-snkPulxqqKyk=s1d^M{0+7xiyHk?`xbOnP3C^d#IQ|KtRuQop!w^G3FjlrQP||6fSTi?@BNIcSfF|nQJ
zzj{)hr2Z$#{~~E#Ch7nG=1GeCU--Y)C8TzwdPp)U4-%5>q&!J}Nl4<6>LL00zapvq
zfAL9i{>x5kPwH>dcqipe(vy%>$A8)XSG@nqK
z|GLg5)lJIhUq2+7G_R84lk_Ac;PcNO=rz^T76epT8}W?AW&KvyBh8=lg|inRGo!^WXjL`Fy|M*8FF^%{#KaPBGZd
zXP^I;-C<#C9Tc!VucvJ1F|i$oa@)@timmnQ;qCdfY@7dhn{T7BmLZuRl>>;v*
zBxUy|OZFm^tq_tB6=f@1l%hpMC<-aEg(8FyLTFP_i1!;&5{0DaEOYWo_jf(l
z_59=I%$zwh_uMmcUUO#NcVb^Kq9OC;v0od&tCGvoWL#h7NVcmN$(ZQ#i0rglw1@3!
zDF|0MF>k%Jj(MZ%^%i7DZ@2{v4Y{m%LpA9gx6+{WwVk=Xr#0`1x4+3B4?|imTcwyu
zdZs2a(!0`a2FcH|+mhbOf)E&ba#??^NtFM^U^(TdsB9*AabF`CLUP%xA-hRm*0Cq$
z>v->?{H@s!Vd%(Zg<)NRa>->&*EzuykjsYojwb!P^IlW=wY#ptRFunZS6A&t?AyXB
z22n3#Go}Bm$%UybmzD2Z2$WDRYwBFw8gb7Ai*>Gmu{X|~Mmbx@p6c7OUjL&%2(O6!OXNPbpDEhiUd+=Bv7b6=#qD@Cl=alb{$1wCvXhwK;cIvvCM9#a
z=Oo7MLjSwq;jOkjq54C{vt5l&Jddu`vD`0Sly^QudUH3V5sqxi`_>*_FG#NV$$Bru
zI;0}jw?cu(1xoApLQmFzE7nCjkzXZHT33^1v)&`It}PJylZ2j{kd2C1FIMb_dWdza
zf#|=pqI~%P?w7>(tp8r@v+Ij-uz4Y;!=R_8xX*1-Uh+Ruc~h~^t`z#cgndV$UoKE;r_{cGQFy|FxW|7lBIkJrFMK2dQyK&`BFKleG*FVLqf@pB>!7pk|mV%C0VLRk|mVV
z|J9fBC6xL@(vxHfrFKc4llogiY1~WCBiWPm
zC6x3f`Tw?l$)1E#x&N;y*^y9c_rEB$Q%e6Y`G0GN^uGT$Jt<#mmn8oeO6~bCeQBOa
z{Vv&&+9Q>hP)e7|{X6d@JO5RmWMAUH|D`9@7x2&gmga#}?_QBVQ>eTm;o?cF2v
zP6$2eeMxaC$?hwmS0mQ_t3sB>ue6?S6zOvW|CQPUqy!#GF54wAL9DC$1TTLk{7Q+i
ze^QjMFL>}4!9#}&KlnoU?d76eN5P{j3)s&}=K)K_bKMdCTg8^!;S=tBK)O8CpW
zhqxUcC9F5&x`;>rT}yg9|FE4(u^;g)WM1HXl==5G?OE_@<+7u(+^*p*+1|>T%n$cl
zrIVe2bDD$)GhY)9O6Pd{_9o27`b--{c0SEn3a?f!i#xNP^tv})L+PF8zJ?(qmtEOB
z5nicWcB#DUJ;c=-p`8xlc0Zee6o;+HDTl1cCTk8r}$szAaYi#%aS
z1AngPLwrupvmcdDuGAy^mEE5(Eny$=Q`OHUu(g%TTzbENDJ7Tv8TOaznO5md_6|0R
zJBaDB^~2l{Ph}*-keAD@UyeM0sGn^YjM(+>)pyGPE0zwaTLpOFyh9GA@68!}0?}~L
zReHYG+VOCN1pHrfY|E8^1(~VDkM+)~Wny~Nn^s|nZsXgXTLG9=R@w^fsyvN+jp=?_
z4+0TyT>A0@@%r>R?_2<%-jqE;9`1VCupOq)zV|Q?%g_7~XpQBs90*^q7Eq&2VFBq4
zD-Ux6`Jk_MKP<1M+V>5%r$RkqdH~=Zzb$KC6v$;Ydk*);=b3DK6hgpqSwpWT;b=cW
zUjIJgl(_lSA7_HArlFm#sXZHG`kj@5^nO!E9R_M6mmR+xnT7bw!Kni4F|O+(M|(w%
zO=;iLpkn+D%!pCH^Bx+YUQlG$k?w#FPUC1iBor3Gk|dW|j2c1XV0qdzlAl}h
ze&dbr9vVMk_v=yn&(wCJ@nDpPR|({@_Zu^4T=*!5(zr3GdPL(}%U~MWsk@a!<@6R-
zlAd?El_8evr)CaRQ7)TzG?ev6}zmGVl^%4F}U-+HQqI`ZD`++fPtlwq=x1(?~V-v9-mHe{gU*RF>IV6<)@_LaD
zgPrmvegZFmEN#zZMKZQk>((ekC_EA##M0?Jc_LnBY{(PZdB~ao;
zQafNWlU*rY+Gk1afK1AlWCOTpk
zcKmBklBM$ho1Rp^RPMh}vM1?F?T};%rTYF0rE>pO&;Q@)|F-kLWa)YT3#D?>Jd&RG
z-}e7o{@;8_md2B0M?xuIlEJCKf8{bsmf8VMi_)e0Bs;|VB&AF3l;*EA|N09(Y5Ym;
zkmi4$uxlmOYbjr{mnQPJ3A7RGxzx@ak-k}Aw3x@6#C?=(u^wy`>!d=gV>`wAX==#p
z*kEB-L#%fX#QIn4&CjjWg7pT7{KmqLyF2IGiFMYZDX;r?MR^C2|5(JAl=iZ|@l9Sw
zjl}(i&w}^og@}9+|IQ2IczRke*RLevbv|ZXzDH3Oy%*2V!GyKb
z?Wwc(lHTAG+|I6_SYM@@{b&3l=Is-fFNdK4{LGs3vkcg8PM^$v(c(Pw_J((gVTj0O
zljgFXdoPZghOb~=vt{f+vJ*5v8(ytk_Ar<0&)PDQ(xdJhQ+r%ubzljT%e2)y!<8q9
z|E_ID=}qshBfXThf2p2=`%Php$z{sv{;;IVWoP58D8H9WC(^h6qD=hihiw7zv(c5R
z;4v2cMcq+FICyNmSlE}bCz%l!k1N7l_>xfAop
z?L3r)d}@}_loFH!HUz+xFu81<@r=KSo;}P$FyA7sH5~QGWhKs^wjwHxwhKe-dqmX)
zF(PCATg2j{fddfX2a%r^A8SeP5ij$)11x7OGjm#E=Ifae9qgW-eA+`^jz@M
zs(~n9z0|xn=3h0MoQwMXA9>bCUOZrn*(XfjUB7S)^1(ezexl
zI+-p4+*y1#{TpDLrYeWuLA}Mk?V>$FzUXtmzXjmW!Eh-W(Db;kpJkq07C*Sl?C5-G
z@4@jQGC-}5<(+oFg#Niub4CI3WyAK0qQoNuYu1u;@XRIn*hzt)N_x5-s%elGba|o{$DY1eP_8k|IDqmtHJgk0{*T*uQu{q(XK%pthw>jbh|gPD
z)4sLckkCOWPno%`4&PgnpZi3#lm7V7Kt#EfFYUW7T`Bz%4f0guO>c4kcg9NNIX>?l
zlTZlt%Vk@~T0>|GF^B0QIy+EvG)-5rI`X@BUFQb6;3kIj3U$Mr1v-d?ZkYx7ZW
z$iXWQ5g(6G0v8GW`uaZUKQ`d=%K)Qjee9PRKdk0rzVptyXNXSStHH%VKbL=?^~kQ+
zEq@8fZN`o2iv9BBau=YYFb{%m)4U$K<2bh?zJ}UqvgZc1+g--%;h#EQ&x`7q=XZB&
zMtV;6O$lS7Iqvdu7@w;gyvFOS(}6~0N8fBCp_S%WLcbsE59-o+9X1Hmg3uPM8*THb
z9Z`{1q*r&eH`zb$7*F!iZ}Ie=LsqQ;3Jl})XbbAMR-uiz|HNK#~F8yW_%JBLGA3X
z&3?)%oc)%o7TaGR&*h!EIFnvTi*tmR-}64I$$j2mX1TJSuJHTPxwv%S%uDbDX&KSWHr1Ps^g5PWu`ps8!{b}M{$xp=ViUe=CAb5dmV}5_W
z;vQ!cv5(p-+A~4yr?W);@N(#RrG3P|xJ2Yjyh+lBLR1b02O%upgwj4p(v$Kfxlr(%
zBcfcIKxrQ($(_acFcJOGM%Z5|^nVDH+69A->X%UB>k>-!|8FR@7mSg;|I&v+|9{Mv
z%DoqQ|C?Q@KB@f@O6~ubk4ySeJrYXklD;HM`4UQeSgPk=JtG^bzwhT*!|E1_`tlFTJp)b7KY5AslUDJ1ZFhv%!5^!;z|fJ)x(D<688wXaKfdHR
z-PU6qSJMs~LH0I$=KVv@k$itV`}hM&|GB#l*$Fl2O1!@4#*Rq?7U0--{E!-vQw^l(h2iFwKJ?io}f5ml8JVorY`)5>2W59
z7bB`naU6#EBa>bYMP89I;Ds*gX%t8N!hFY@mzyJx)?X4*AM;l{wA=0tcw%Jh*&%>q
zYuw-`C=k~*9<^{apjC9lXa~T*J%V@M06c!^rRyNzi{?H0ZSVv6Ohrylw3`*P%-}F!
zRBCO7I$($2Z!OM(-uh{+y{3S?MDJxctoN$&kL_asCqCD=nhR((wn3aB@Zns|UsaJH
zza8>m3G$+@ZztA^2VDC;^5+GVXQbb80(%kXdTMtEbor3H=sToe-t}o}5XfzA`cxad
zz;lmY+QY$~&-#VAeRAPkf(DK8}2>HWMhJ#0*`Ze;Nfekkh!MIE@y#W
zyH`11EkUp2rA@730Vi13cb^Ejt*iQFn-6dwrQVc*=*${OW`=`sUyjtYp
z@PWC9_}&U!vKt!#wmP@7OH0sSbv-%{?T*&o-fkM;Gv9Be$WL4Ciw<9X3;KDzrth>I
zxy-w}!=N|7$1`Uo*|&oH*)QKt!2QF)PntIAadKI=$c2jP>vEaqq>YbKU&FqrP5-g2
zz<$ZupS8$){R#{Q)^0
zQ3Kl1sxGZI7}8&=$G1g3uA{MYIqtt24Y56e?J@6`nKK@qPv3V&{kfp;@#gTP(U3k!
zdxs|O3qxWCT)_7Kz2Q&$(3uXGXy7(Rr6*-&
z*xu_in$i4!@lkU>>L)a@Nk=qYDVvDrJV7-o{ZKz}TK7%ZFLpJ_Hkd#2{L>gjh3+6~
zXN*3tg9-y)Z(Gmi_0m6&*RR!mW2v2Sf0`3Mc*5tYW!ISB?nq>QpXG8ETp;ZK$1+dt
zCaa=#DNV#ff;RAZ$b}m{Gf~enG?V&coV+3FB`M~SeX9?A4x6xC`#0)&ng>%{z+UqY
zxNN!XaG^St^Z2PlIwdOY6{!ws>|~v{T|CMofwVYGJf36`6@SV({rbaIP|0joFAjg^<-PIUh)Y}
ze<1encguL5_c3R^0Cip`ekNQdd(*XezWROP@`>pzA9LpW+4Yo{lb(}4=T95Y>*8jy
z-?nNe&W~5v(K>o!-C5GNYu6Blj$AhFV>rnM?>R1;J@6UDA6lsM{R+jaMx>{Dj^o1q
z;~$e;n(syHzv|5URNk`-+p8ZRLh|Cy3#mSz
zFncrcsM*dD{>Z;vrZym+*>!ax40YfeTbziOI<-4|5ambDZswyO_w1ud&tuwjKMZBy
zTW-h5{>CfLmMB}bU#pJsvNak}bbq^D??K_1eqf8scf`uarwy@P@87h3i+s@FM;a_~
z;D=YLQhfNf?LJ$~f3Rng3i6#RmsX9x2>9gH&&il>duiktY{&gc_U;>{CW*Q54|S
z0jF9v1V6t#EwdDP#>P>7uHkdd^I1vpV^6EO>ycNwy`D3|4S3s+ww6`s|5w```hxzo
z^U*H8$fu?`E{s4w+o1E9K|3Js>e9GtG4ieJ^63w-z2SXk$6>zx#oPkqiTWGv=jTKG
zwr<~&;1M8~?@RYV{-v)pRO=4lpxwF&9l?&_F%R$Hbcm-;Q*Dm%PyOe^T~YxT`KhP%
z0G{Gx_Wr{9#}L=P_~hq^2hcyG-d0@&{^;N^?(uUdA9?afrZvc4yBfy-0sdAxG64SS
zqeAw=W)X?$u&uLrAxryw7C#UWH7#S^D+mNeWqalX%xILliRi
zsNRWWbE_8xV&+=9p6kc9lOYw5)~af2GsyVhW%|YM&#YF9VNP40-+Pyj-^V
zh}M~sw{YL8d1{wQ&_1UlDYKKZfd6I;xaiwTAv1h*+dOuJu)7Hdu^
zWOGwKWUDPv$jXx+XqH01k?GYEovbgQc4hwg%S?eAOl^`YIXJ=HDd0sh`{@d559
z6cz^lINk;wFdV)O(01bC9v9$!sI)&cG#TW$S?Vb`&W8IM$Ik)ubV-fIxOcxcc2@h~
zxj#R<w#iN4~2Dy`z8j
zMxuSs7f-h0{0JHP={?TB%l#S>9}Aj&ViC%Qqgp=R2SH0Sfk#Y&Cx5HLj79rP7@CfnD%fJ
z?z@@{XgVL)i!sp;N8!90Wl?MU3*J|`V@gII$d^)E_nZXy{QBiu^LOpWX_TpW{Yd5x2p>=na+>G>=
z#RjWmyU))L%tO6de>>4SXW3#GwfC!5V_KJu)-`a!e*5j$pXxts{E^}X2TyOKeZgm)
z#w2G{Egp{hnRN%6#NxW~?9ZZcXy5tuE(f&ll+phyqU*u^WOr#H$AKqbWnQ*^8T0U$
zN7>(nk9$SWF}y3^zZsOrymsAh=Bp~x_`c!XOW$ez4vldjyCu=T2-o~!{}Y(^f#f4M
zXHdVLS<14;PmXJBAIxHb2KQ6|TA^!jo?+m7LOsfl|(!GbUS5j<>%
zz+#D)iF;PJg`UJay+wWl!Mj$E;`YP{KI?89$>R{|{b}5R^uCrYBYD@-Y|>NOwTAQ#
zRd7AK1G)6$E}U_R$^0
zona$SAYSNOL+6d&PmB7Y{QE$=M!4S13ph&a?Z<#`M=^ba-narpt7i3JD=(Mz%IsB#
z>*&lSV~O`#*ma>eM#9%&`KVXn>H$Yma@oh0cOVoFai+Eds9oznlpV)>qoZwe5iJh)
z|B3moYUMDc;JmYN7Ohhr$?GCf9v4~m2JzI?v|Ff`@yYKz$`6-ME<{;jQg;CTkI@cm
zKjeXNUt5?(027a=FD$#90n+I)njs
zGT4@x34HJU(C)CMhx-)gjo_!Sz<*DDOGLhFmFha~1Yo#qP;mp$%V=JcwFKoe8Dm~_J|+YxzuXQmV02|s=)i~P~a8ePD1n<}AzNxR|CS;U+A*-AtO(+zT`V5HI|BU0>FxQBr=UGsy_)8nh4LZK
zH`MDD0zApT@pvELX@ksPKe`3H_rRh97ra1E^Hkvkd@ly=UjA(u3Gr|vy?LP!_Z)nF
z;$ryI-5@{NvNksp
z+Oc|N{AO!FM_JqV$fw5X^_+_1=SINLwX;C){wMR*jp4bJxA$p04$@s-o_~xy=kc&N
z&vBeS-PcroG~_!a-B{QOP|NV-=t+>?@zucb>jB3a{eF%6mY{+hb>z_*eN?rPPX?X6
z(A*BvPo<9NgX400v(PqM0e`(-T#5TCv()#8Y(dXv=x^_-AnT8M>W{o|k^Aif{(vgB
zr|3RRTF=2xa6W594yOAum9-z8uw7Gob)|EFhkw%FVmvs_dOhtU#~<=vf#bHrxeHTJ
z?^DZ$d59_-oLk^?M>jLp$9Z3Wy7n^E3%@sSCgzvrbffj=Mw6O)D2MMI-wgG_3kTbv
z-mAMhu9)7VV3I##VPd1!m|om?`yb?yeruf;qrBR(gyPUUqHa+;D^!#!Q2*!cSG2#q
z`Sk_ymum;x(K;EC#OE*@H_Mc;yz+oZTDNwOIY&Io-uO1@ofx-=_=A<>MxX?+|Gqqc
z?sMN%%pg6D79UBk%Okc|JXjO1w8Fed=}GHmg`zL%-EjCsdV|jx((`)W(E5aWwu6%%
zAzr_5hIshY(QRnmEjYk&f;G3lQT+~&n7`-8X)ucP)Tgpvx*+16{Tp$-)ABp(
zIVp>K7=8iNj_TqsggR575`HOQz2ExGU%HEP#?FEl1_-}jDELdzFy>=c*V(Td*)X0F
z_Yn$bu)Yn$r+jLsMyl8C_*Pm$282XrT_dCv4YsU8;n+#yPOY@k&luc#%
z&qp3tuE~7=aZVYRPbz0UeKXd3pvwH}X+{#g=akc&9w5GN5HybCa1EF9xH?Wc1?r-i~f{l>Yu(G-2k2XT%4-{?sMjlw`Y)(lbu#
zP`ZQ9R+8t9pGDi{+<50e(R~cO5t?UQ@YZO
zWdDBqX+QYhx+Tr)7UOa$zw4D*G*8B++S9x(@3=G`=kxu?k(!7-9^2EpV(b1T7Uh^@
zV=@uVCYo3v#;2Yu$Mvr8`4K7~6S1Azdv$*c;zvnyX2Vth>?JJuitG2qS38L>1?cpk
z`mcvMP<(K|^~Z&%mr@)CpWJ}BQ05%kmre~=rh4m}_kDr+18gS=z$7?&W7Uknav
zHVV^s_P)Lj*TJslx0ho6%t?_}=ogd{4dW6}-dA`9{lTYpaph=V=hcGAxUPRsby^+{
zXz)p=e+8f-zj7Crv-Wm$$GCv%IcGcMJ?o#Iyn=qNru|ce67Yqms-MC(gX~#VrGn>p
zS~Y`IHvqP{dtc8UP@`IH#dN)4e+`gV9XOmg3hRG4aLiYXujm9X%WVMm
z{Pte(wn4p~`g8ES&&xx;Hx)2Te$vYU?9B0raXSF&M^oGOTY`We&5F=E33l(!s=n?C{r7zCJs%u5
z`q@oJQvU$+Y+82JE#`~=ag%cm+{ej;n^+Ae-YIK21Tnp^FwAidhM
zuqDPZFHGK>-3ZzJM-1s=()Nd+}Tm$|qX&`gjtab4AFP7U>|Hoei@`o|w|)@Kofn9xtmX{%)?@k=D5}
zKb>ejjM~1Q_|?;%Uuj($xif}v_RVdFu>PmdCeyjeiXqG^lDvW`Kkpm=u2soDlI*mb
z()S7K?@5fIbuiDmj@D;`CsRq^Yh7&yzHe=1RoaJkSbCkx?Ool5+T-QF
zoFRS7z*@q<5ix{Y^G=rG^9;yWA|C5IPo3<~_E4gJDP8)Eow*T5$Cxbg`e5~d#+&~+eYx06((FR;~`_J`2Nz*JPwV!db|3Jdd>KGUZ*mDv2#u#y@sDTe@QL#oSL?*cQcUnjQy^X
zUK8;=`Bkhpw=3%f+p}Ij&1BLmEMfkl|B&COo-xO7Bg*d4INp`rm;C!uQ@$T|_7E
zK8=^v_I!VO(<8RG@g?8qjy%No*jhH)zNuM6$c(l1vF%~iS
zS)q_;k{2jYT~P2Q+i;OS3snkzHg5HG1I)+4>mM;_2R
zb)wT1EnM&7RC0)Kz1AH^=bA6=8xwE(Q{1p8+IyVsl8t)GR&9v?I20LB`&2#{P+ZF6
z@kf2syHxOZCF*(o^&Ep3_SkU-@`qeao8ib$Mqf-CiRUs~TlH#u8*rcQ#)r60yfys2
z3D?C3XCCR{KJ#E`_%HMuvbc_}(*V1T^QgxB=I`oc7$-_HoALzL&n~}o>!IH;Qy+Bd
z1NeuVH%xuqK;Ayvts2{tyfgVa_CsE87v%uJ7X2F^JsC2rb6Zz)NoEK&@)OKcYh$*$qteC
z$9l5rN4ow5Kib_Vet}yzg)DSd%N7;j$DVGl&fW%k(aJ@ahJoIb2V*k!0RMTbUOFtj
zr$Y9=?2y}1uvaScTiOQn#v7!#%0X{&kL}@JV6WTdtvg&gD`Xd|Y&%>Ehu>LT7_?g%
z_=)O`7R~kp-`Uh}$d%quzE_7+rs(J29Ibio77hM7EBo~!IKQiXR2yvyepR_HuDUJg
z)#xk;#OHI)Z=|&$5$bJ!XpuMYi#Ve?tq!2K#cp7(ALw0qba3DpXvf><4g)%YKOf%k
zne|uDTXJC9hrUqn>aX*%rbGS*gRQ=EK>zf?#vNzfh5MGli{{ROcvD(am)S0$*Y#k*
zGgHvBU96uu6ZAq_%{Lwc{_>;l(qQD9MGAKV!48cL_2$C>C-ylxb{E^&aZ%hl?ABh=J%cAQU?z&
zUyXV{j>wN87F~S(0nxRBe=lP7wLAGoydugSl+izcV>SX+JXJdGx7WW
zqnh{WdDRV85@g&$wE_ctCu2J@rW%^}{O7Q-oI5>@RD!ZXx;eetvJC%?Hu=NDn?j
z>Ff8mqw(hAyprUB*$?RbhJF4@@7=fjFr^;~)hE5O70XFp@R!eVYW6G9`HZS9$32o?
zwIn^oQ7_V~^g2xTzMVTy_F|N{f6}_~cP9Vn9UwiG#CoKsIkqp2^G-QG$gjpOT|@qI
z*2ajZI39KK!#g2XUM}B_^MA$Sd*oNcjXu%2%Y@y{b1>g_dpmW+x>@G0Q7=g=fq1Fs
z{eCq6x+$mRVZOn{xqmUgFm~=j#Azn3#CP21z9c<0d(B+b`w?KSh4`-h#vZ7r=&efq
z*tg^{&HLCZW#so=X5JZydcSO~ZsWSsWtA`aOP616Nw{8Q2Xzj{bwPDS-lAYY^;wfP
zUxanK|GXMyTo0X;^NDw)8h<^H<@-5=9YehbUnV(C0KLlgwR!F!XU>04_u92qwtI^9
z>}Echg?@k2M74y4pl9sSy9E98am`Jw(C@!I{`di|OZ66ZxQ%)m83}WfLC@T23w+uR
z{QtBzt8o3*GrKt#d4%HBO$X!$<(rL9%mthg=k{_D_>seB>YTg4y1(+{g}_`W?_;mJ
z<16^fM!7c2BMcNW)u%@0E5ZL-Dz5&{1HCT|JicW>xmlz8@4W>6ciQn;I`hXXWKm76
z&-Mj9zvn^63P5j-Yfjb+(92Ph87~Aq;|`Z}OJ^u#MvXF@8$$ZSMH!mbp!ZPiM3Zo+
zSGQN_tch@rQ&ZeCSsCsb9X0P3{0a0jXTJX&2kkAae>WEIMLpkgf5a3x7cuiXsW-oo
zLbfPSWAaeA@1r?K>-c5xmm@lDZ9NYBHU#r-7YrR$*g3z<%bDIStYv
z{vneO-Q)mzvv(xKZ3MkB6>Wl|p#2+)R9D);d64?T3lF=2UTo(hztcf)N>%&q9YF6_
zclFT<&>J|v#kpa(;U3w|pm#Fx=j#5qujGN=jXt_zDNtXZQm0i7K;Q62@|mWwa@l;B
z9-jT+{*Ok&zO>t*S2=S_-8|6CHRzd%<0AWO%6jx~t?ul#F9V!+pe`8wYC!)@nyUeq
zmUy>7Kdce1tE2&FI(+lUI>3yt`7hD0ZhhZ%D~`8um5af}Ya|fcm%6-`n(2johA!ay_dZ=Za;P};f^0x%hrCGZ{I3AN!dwb$M
ziby&cgLv@dxO({9c~ge3LA|}lROr5ncGa9!n6HsFCjv3|YWZ^|1Su|_k|
zUY&j5V$9#;>_NPLa%4_-ITIR7}W!+v{)
zxW^)WH?_ZzZ;5if0=S&?cPFI#)DR>id(!taE;z6}^&`KBYhzW(ug*NQm-Nzun$mmA
zxwDS==`$5>=ir6>{7&_mFE#!4ll0^7C6Io#8@DraKJyZ@mVEwkN{jos)Ak6`SNG%k
z!^Se!#4~D);c`VU_+xFhgdNY@Q;r1>&O#ZU-uV9*gwbyH(;kfpU
zT4Riu?zj*B#0K~yKV52P|2|J>d?pl6As$eCY61E8!I32jTsPWZjwQR3*YS7aN*azP
zS+##YpKoY+(*5AqMb(s^W5DC1?O=|R&3wyo2aUbTRL_Sr<|VF&_+Hn}#0OMvbZiBo
zO~6p%A&dM6Ot8x*z3vn|kj9RN2yNHtN|9xp1Iy
z5w^RL@5)Aa&Tuuz#`HAcH?Khz3AG_YPQppNvc!
zU+^~pZM&Bmf^1WLU3D6)*UB^AjKuFa+~}2d^h*`^`4l&|K9HZSl7D$JVB2!1b}4}4
z=eTJ%2Y=K(b7jyB&=2e3aRm2g#=gUw;rbh4Vs;zb9j<%fKE^jY`VVL}4*Yt`^V(qi
zecjAWUp`{`l0}oc`2ZfCxLDT$P&=^UKv!s2!t-7`k+-C*czX@|*K^7iJM=^2>zunA
z18%A?4T}JKiQ)FC?rA&`zIk+4XYaJY{Oa)T>I^t|iKYf5Q2~#Ao*B
zTFzC-_O%|Os|GwG<@ee@K2U!br?p@6!ERKy^gi|qIQNJ)yO}UjA=`2OxPDC(o)>ND
z>$X548$4FY`{y!+%sSj7c`e-caLiQB$pK!V_V+=(s&)#QepzyNB+IpAOPSF2;
zDPu49OoiXIjbEcR4$dnqLfQ|vJfM($saok0RRR9Bw{b#0;60~yeE;<09q_U1UaIkM
zZgRF*aYt*ILiWn_;H+n0=W}GeS9pK;uT8Uhc*EE#bLTPTD>g0QlGL*U;(E@7q?^Et>%6SVwEVXB~t1hNtu2R~eu`
zyP@~)XV6bugWR$bAb-Y?C1%MWzcMhsHx=rQKXsrg8~XL*s1k>FkY0QuDr^^EBj3=h
z5Wu?a4YH88^mWd?jB$YG8`l=?01UCYvbY-B`NUkQ{a+|Q<4~{lVUWJm#rob
zcRxTCd(VA|kbgD2yEpzGG4K0<@i;H~tx%=k^$qd2IpxZC*~IOkH|9SFByB-KNT(DIEa}e
zpR@Fj;&NsB?6++{uzW+@3%xa&<0$)sIRB$)&z1;om!Y^1lHGv)yo>nV;(BpEc)9pq
z<9Bhd_?@u(`U#h}6nep)EJusqS0##jA_v925i21}_dQG>u>N7eYcd4x6$pcn`nA3I
zJs3DWl7YMuLJ)xPpum-af5BlJrNhhpANqHMoFeQ(Cgn@|(CH+DGbEfO@RvIG+d6T-
zB+ZWH%xxT}IUCD-bod4?w^V$GBS+i^*A{>0Y^4qJ*<#T@w&EU46TvG__;Pt?@!d*q
z!3R}Eze(R^d$)k~M&|H)mcGZ_MCeQ3**_+J-*!8Q^`mN;7kZ88Oyho?+7f!sk}drG
zo_0m)6j!M}z&y{Y5y#cakMcQ}&TB&|*T-Qq@vY5%JYEt;@c4}=Gk8I83`qCTz{?36%oL?`LpUWbS`nHM|Ytk`)QMxPA>^!0W``_a!i7V4rU|jn-2y^+o$o&KPrI1@f&S^ZEYu
z%t|MEzJPfCeRIX^UZlT1x-YGN;~OzgP5gFzk@<=A&P1F}e9p%$KKkc#Y^XbDJmPb)|*}P=Czu6+h6v>B-e4m~Q!szi+nd_#c>}
z5dV+bcpq`irp{B5Z~0h18s-6hyTOt-H-ABU7ucI^I|Y1Z+oIlgh668&+%omurEz$F
z;-#l9=wG!6ywDGD`M!(oo`9cqymD*nH7M_O^k;z=#4+BKEc=M<)oQZs;!eP2=3B>O
zJnFpniBR-=H9^PI41m`yS>7cK;|Px{Hocw!xO&MEt2uzretdDmevCgmtJgK~yGNh-
zZ@}kDw0*S-`>V+OYifSIA}^-0YhP@8wiN&9!gl
z3Yp!KyVJCw9hM=f52itV+4G&u6X4#?+x@3hn!))}*10u`wz_anCNk6c2BdEull$!(
z#IrgMn-Ogc^=HStzE=Y6(H@x_?Ew66)^+DOy8(X`oxY6spRWWC3~3I}XQ}(X{5bS;
z`>j>?-vST)mG!IgI`FX_*48-~-wN2=VnhYJSI4gxFSApkA|j`Xvn^F4oDU
zckh-^kNY8e^{bGtbM%gH1YrLj{WseJZ?%}8*>xiHPraY(LLNeWeYd@z+X3_^j4Ye?
z0?@eIsND-7J@J|T&h>!vC(L!k`RBO5d?}8biWNzExIfty*4YB@uiriMBEAajpL%z6
zE{@0CjyLDyJT1`dv<%06*682X`1{B7%nj~%-Wrf_CKku(CRLA@G_Oy1U%_~(`O4)J
zUqOA2s{>0X1HOsxHywFm=d0ILke`iI`f7r4wO27+zl?|cnu+~OcZ0mX|IpXS%gWUD
z-Nk*9`-hIi&k74W=-}_NR>Kc}ahw@?xiiPf4tAbG^498~gaPv(5^mYV{HyXd|J_%U
zmg-dAA(i7JCVjd6McVB5S~iQLe8-f#RDM=8`|nus_X7G|XMb3tlt|Bi{^N1d3tgcg
z>@bS`_xDnc^DLdbh0?q9=6KJH0QTGM>vMfqG+B-^=D3@6Czik0<9auqXUr4#h}*S(
zLG})2arp_soL_6ls4Bi2<4e`jD&1&5TkzC7ul=lk#D*6227g
zN$NKsmy~{1AOrzPt`_b2cOM|xKPL3TL_a5OWd5k;z~eBbIgjhic|1S9_KYICut{e7
zGZ{m6Gnzl*au0K#Qof;R=Myo0PcLG
zy@Y)gv5&A3_r3I#_`G)NL;jBBgSpn!o+e#so?me~-
zykhV>+jl`%!MZN)c}1;e9-f)Tyj{bc*DH?`d>+tU%*a|etbfqq&EzM}mUJ&e*(3Hj{4^B!nd3;mXi8H$
zZ%ydQ-|;Da--Gm{mVP1q!NJjFe`#$q7;34*jCB#jSwc%5DaBk)89^wW3L2v0d_)}~U$NCtu5_$U&4L|tg
z4a9AFb{&Q1Tl)fLB;)j+3U5C47P6U4Q&do+oBSq?mNuU2+%NPijKbdnBW
z$kg;ijGqPVbPjq7xUPOphZTUo-|FfjpLu^(E7A|}@p``!5#`Sn9F9<3YXqK>@6X{|@d
z6u`{v5xV$Y;>fgJjfVm6?m9_(UGguuf7Z3sa})4upT1W*HU*xpuI@2975
z-vQ?~{`_eK*l~F~_hk{_x9_IU20=Oh&<*u4&S!GB-JxVixBRP1-+QY&7c~ZXqQ)Mh
zjB|iV$=}Vt0@fruOh+EyTW$Gv90wX%NA2%GdlpU$l3Bs~c;t2dnJ4h=yVF8$PX>P5
zdby$f8#q69T-EJG3#g}LiF)NgDCge5Y7X90J9APcr8Dr~emT}@Nq`;}>5Yd%zGvf?
zOLqaj8Xj?>GnA|RX4Y^&pnk;99$|o&{`}d6`>ZB4<8x;MR-PZ`f^q7kv3;iC{W;??
zAs4RzX1X@JiahyU@wPC0pLYzDi_ZZ1mj&Bk9Mmlm
zbKm+f^<_`M${%YR;{39iya|?e*k51ny&LDXYqIX-H-JBWJ!`QU(8TC$eQe+H-3j-w
z{=(RijnLjGv+><{HUDTyPIgIMeK*cDBDj~CtiMeuPeoObtapnV)})Q
z{pQ&2{`++)UvKwS=IxCdQCz6?k#)3gu6SI}9rb@kZ=`kehvNXs4?Eb4)@g@J97k32
zs{={^=SEMiP`o%QyNdFw8}RR&cKDXjdR%wB4dn-&nn1i=V^KK8U6P$XQ~oTi!Gx+W
z)9<3ar=j)9KN=*DqV~*gYfky5*Z5wO`
zP8IJ*#s+eJ&?k;xb>7Q!B>{nBMvtK-D!Mxq8?QpW+=xj9MraSCU
zHJ&p+-{#N$bNTTw%2(^me8pi8^NW6BpRg{0^{Sn@e0eL*zoO1OKuMg(4jsj^tv~x?
znG3h){BcfCp2#>s@U;?s&cD!x?f((HK2`AUY!O$CIl%ji1aS|}YBbwP6TCg!p5=58
zwv#Qsqmk#t>1Rd!vr@!A^G0yGvEV^p1&=Hjd~UUnt%dzPf;UOu)iV_NV+9YKC+t`W
zJ}Z6CZk@=7OBmF@i^O+h^Mqb^VP}!ZuM$`(_*|@zVUSUIiC;^yOq7@GOZx&C9HcMt
zbq65>`Ji+u|6lzAkq<#n(v$WDl0GE5=!=^dZqM#QF>3(K1OQ4WcOcL
z8ZW2BIIR-n+;t1TuN1LAk?MIQ+8ZSBh!{r-@jGBiPwFS>JOhF=&;;-k!oJjRu-H<%
z)b3m%Ulr|<>`D2Oez?$+zULgO%=7n*jMs&!%XpnJTgL16+k5M&9vd;Q*Nfjj8;SFe
zGY`4kmtFk3Ay;94iJ1Q~QSL$rw>M@cubXgrn(WHN`z{muv&BA2FOt_?4Y5AF5&J_|
zF;B0HbE6@7+ir?v2H{f7t^Vpx*%mjy)=G)<7V7%7w&QVq?J9#Yx-}u
zCA<5@`rcC9*ZnAdR~&Mi*Z<6&?6;zJPyUSU{bclf8RGRZO~Vk?E?Ls|Mj~7@vysOH
z_zt9Y&YW&W`<&uug;dXp?k2R4>d?Q-K`eK$MStpt&g%U4wEY8k|LFB}0Qvg~kIX6V
z*5K=RDsLKbjPx6POr_`Z3@d{n2fU~tp6>4i_Tzh6rNK_rpXxnVQ#oz-Ao9m{<=#|p
zpH4Rq;eM~=LXaEs9f#^EQ5erZbiLtzOn><4z>*e#jq(STqkR|MWpqz9abCO+@~*df
zl>x|)nyNTjpnkk{#c9<0*`Vh*#8ZJ?tPmBZKQ$4%?05^IAlO%Ril+Tx?BDaXm_E{R
zO%>u^L%k-L|EkBQc;suAbJOU2X+~l1RhWOd;R)hNM&(&l-sIY#KiH1Uu8w+IfIjz>
zZsL8}Gj%Dq@&UIbv^-)A>3_#7&Lgk3(z!VRM4t2_M#Cx4*zg&cn5ZK+EISR`k{!m2U?}!@u!UoH!51t)}
zap%X4zU@N0@hLVfu^pXnI}FEiX*U{u!{^;@pE>~d(K=&8j4?lOi)(w-?_Mvd0`1QC
z`s<4AOf-66gE(^fqOO?Vch0t{sIP3&n!Zc+_S~!iX!m^3{(6YZI-Ndf0PS=>*6m|B
z$f*fGldyk$KXd|@3(w{0(g5qe_4e*1f6N?X)(^&D!9=~M1!
z#iPFGUe}k@uklIL&Ze@_g{U{>tD`2iYi+~t=dqo>&R!HxA9rmF?bq)e*mnf=b|*I5
zhM3mTjK06*$IS6i4ESO-VcvA#NQ{&mVm2-?=H^Z`yEI8jj8_XO{-`hP!g+0>xN2e
zUax&O^X~<`(l3*}6+x9$J|>*=Q>wpGy7{V1LfIQ$hi66ecc}uZ+LQgZ@$7#V*)ea=
zU(4(EuzI{dXnK{OFHM`*zYT48oj!Dw^>QO7k$v|k9MAGhXWsp>Df8--W9+ZARrvRa
zs~T|n=+?aMyNdgiXT^6p{+{M~Y+tc|y06FOcZlz9L|C!>yB_ZwR&HfKQX_ts8e`4*
zyG6Wk;&+xujbMApHY~@8_$GYBoZ7#rq3~0p9X%{K{havS?06BUw;aWG+6dlmD)@z?
z@WVNRU+awJ@)LwC7k(Q)-b(e^3E5rnp56cO_D&+-R`7x^q8>Hj_u-HHkR4Bv4~s9!
zMY9+segK!WDBVWn{}fm)_y9N^%9nWhDIrU|dZ;Kb@hX@Mqz{9KP~!RXgbdDx(%~OM
ziI>5oA~{=Niomn~=t(@XQlv}uNa>P231L!@T}ki1P}2Xed?{V3M{2*6{x3>;|EB*d
zOLqPXVbW6jrGAz4{(UYW3IDHp|JEm!lkEJPFQxyBlAe?O)Wp
zs)9I@K&idb^GNAZdzuQpu43I-Al8}5A`Wmu{2g~`zJU`V`-cVY5%c<**hfkBrTV1x
zWUJ6G5c?6S-O_lJ+Ln-J93(>yhjiY%1Ar4gbma
z3`4}azC)~koyB>op7{Imn%mirsfhJFQS5u_i+YCc;pd<1$@{2qVW*{7k6Vg%=ZWtv
zP2SG_ZIQ??6#Xn0>u@g-7n<9d+gobelKQu7?^);~;0K0xN&Y>If9Jl|yEVlrU$}BS
zYefRLBXv3N7fXzXQMnfRe6CWL&-tHY+288!;Cftg(#WoN4BxNwQE>i3Q~VH-gx2{x0PHcyNp3b|z=5N$-*RXBYy&SGTO$jOcT6W~g8u7BGEsJPh6JN;p;Xb4`Y?GbO%75qQdx*UL4euSSB=T3CCjF2G0wsxl#=dJL=J)4N19AR+qn$|f
zTO_Z9zz-i;fUE}N5L_h3Z5i#f56_@1Ur#!V>~CH2uvc{0BK
zj^I3kUcC*AiFKr;OU6~X$tUl{Lo^bI`5J$49ntUgr=JsgebaXqfp4p{iCAyW@^c87
z{w9O_n{K=0bHupamGLC{?^-{HsAp7O9D(;P-IJhm+*v*1yU6ww;`^nmr2XU7Bd>_G
zfpR+0&U9V!etgv=(w;i|DXAwfPayYgW^Oi7kHoWIL^;b^^8b6EJjo}Pj3oJ$n~gY;
zzbPtvvCli&F#Nhn9%PsM#D`Hu*PEbn%T%$K0<(>1C4DGf8S
zU&4iCeJ59vc&U`$okhv%=zbDUOOwR&rSA2EDLEKR>Frww$nvzEOdTZ`X!~JlN*)|_
zB=P>H>Zk7;XHjxXo06Nh6dk#goT2mSc+q|o{vuDZK1-^847xy!QxKJpi8Qc6r|-JG9RKGZY~1S`U5fnB8`^<1pJf!fs2;N_opB-BO;BPiGa8$2}tAN
z<|NX%s0m2RcU+YJE8Rz>X?aiAGaQf3r}YRrpN4e1bo$@m@OFmdBasB1|5h)pAJTYq
zKk4*vNY^{Oz2WKqYrNrf(PEV#Z9}fR3-tcn&H9fq3ntr-p|CUeV
z)9d(8?b3L3eSgBE^J(}eyy5iy36IvZ>H26o=<;;>PvzKh-F`K8NC$b1D8SeURiA
zFN;aJf|uafG~UBgbktDLhx%?p$(t+GcbX5={XpjeqP!1vjzXo7NYnkH;XCR(lRA&k
z_c3%mG~cApaoZ_5Hu?=or|tq$u70{srmb8^Jz@KZ3Zgwxcd}g;MelyTuY1x2?Cb^Z!lP2z1SC+lsZ%2!eSN~GwPh$HcG
z?o;^kB?O&X{_C;JXfNVJLM0c{U#;@RG9q8ahdQ5iQ1rBsb_sHA{Y3uc)s95@BJSj(vX*GH25YWbUlX<=l&508@fqb-!5myRJ>&p{0m26&Tt
zQ)>{3FZ#)lprdBbWKwV5=R&|j<5=7>XxBB5{2!z)uqNnq%O}rY(kn^(=SX+*Tv(A$
z&R=O8NvEBdfT*{ji_{~1<;nkLM<hOB?{6jHxDx`L#0Nij*s<8o3D?pG)v=gonf|MTi(ev$)8=iV6|
zME;759s(Z!)J;Gw50cJu>Ry@ULza&zyh-GTGwu-3w85RA*Ed`YSq%OkQ|%}EbtO2R
z$TwTzNx;3gS_tT}jnv24RphvPW|8v@z2v%ah$|%UMAb<8lKM$Kooh|5o833baqt;K
zj>8sPa-BA~lIQjP)c@bDZDhU4b8Zp!-Qtn;znMnXw}#?BQ>b{$XlO;?jcOclFou7RsOpH6&gcb>FTKN~V*k`xB*ZvfUAH$?vx*z1xA(s}(7|
zd21S(k4BJ0f96yC=pTyj+EHo$O=Lc;&(d};ffQd7qwud%bjVZbZB#ywg0U2|r1%y6
zUd#~nJ)VNq6!f9~>*G%`f7nuegh%Jo<3X3B@#%Kxe$f4=^Xc+5L}ray_{>Y9a$I;JDPPbfJTZs0dFZO7EMZdrALwdD5wClYX)>Cx{
z)9KI8%RbA-|KD$)rX`@iU&Y}(`|)URJyP=Kj107oJ@Ko_@dSQ9{jKIux;zhhON61O
z=!d5*lgqn-`S+HtTKW6^yCjQ`iY)Xev}#HjwJ5I(Ow3CxHA7mVFC+c&n?khfP)b{G
zjP}o!eZ{>hKQOL=^_pJ)4ehY3vP)IJpcO*;1H~iwXlI;J$+jh?0k%W#T}@S
z5L>~9pUXvWjd5v(HFZZ1pOxT2ziaWjk_Tuvyz$8)Rg5oDWF2;)U$q&MB2!}>s<<$;
zuOV;O0R9hV`)|65^vcbNVR4bZ^TnG_t-ip8(rmk^ueZ?eb94H()4%&G=quyY8
zd(lF~t4%D8nuh+06WSk`Kf*W?C+C!l?85)}g8uSkDYT~;!+yKj68#A+uGbxwEycJa
zJ4232_u##wVqteC+8K+wZr?v2$4$=Ia*qX$bC1A6;swT!SU02JFAU>>&z>sp_}PjF
zFO||9*9PHxEs`d?=O*yL>hYZX-o-o^8{GR?5Q6c|O>EX|z
zrK-XLPNB?(H8WW7VVO#8WG4D&ihGV8ZN~#2e=S>W5gte!{V}=_=l84fa*sJUj?2Fo
zk1>0O_t;_|CCo8ih?`HJ{u)<=vxAKt7x3VAK=_vkMT{3BYCHPw`ex95R_AlzFE-rR
zGX}NXuoY$Dp{b$!1!bH%3ew`z-Uc$>2d}GayRvfw@9;q
zRez!kJXla^eBko;r7UoN`*2CzA{NB;sID$jV}j~v&yyzb|^U)GQZp*`)DS?HI`
z4{*2F!FX?3Bh9rl)?|{!wzZue(YqRH#I}cvc+_>7021Xm7J<*^(P4
zIgodH#%*bB4rm;EVz}0e4LWtVm$^hXL(HQ?6ZJ)#A*EBT>SqH3T6>nMmuE8|^o6VK
zJ#z*O-hO>-KhilxE=GLmKqE|eRk`0$j0qte>v4DOnedUba#ffJ3!J>I*l}xE7?x6Sy6}XRr9J87xnav`jqA
z0oImrN|gy*IKI>N!u}B$cje*!=0*tyT;vXYT{qMSQOo7$4CXOFZ*@pbS{D<7F8>(0
z7}wFG%-Joejx0!Q-S%X;6ARY;@F*zpV8XU6waqkH8IC0tJ_u1kk%=Q3dK
z&;FZBnM`2bIMgVv#R8RO>32s@$NiHSy+#A$4Mmk{%_z|1fzk*YuPHO|Ju9!ViHfqg
zuAVc*?{{-&Y^JWM5OcZ
zYZHrA#T-ynKik!d>#RaENk6!Y1B3Nt&es=nG0v{}ZZG6h?7#=L-f}F^y&Qcy=N$vy
z$0lB>9$>%^*6Zs^ZcKQ+U8#vx#sq~_k;~PdESQ}1Huz8h6D~U|C)=E10H>xsuP}!J
z1=THbD(HXir&rm$80FV$mDceykRLf^Uhdb#eodWOJU1Bmsp~tbsxTby3#UYKzc0mo
zv{&og8uT;2SNQ(e%^zGSmcN%!`kn)?R3yF?1McgWTXs%N+PlH?}-d>z7zhjZZ;F#p_>z$VB*4k_YNIzf_2xFw6Y#FgTm?^{QEs@XqL1XnG?qWp{U2a4Vg@^
zm~P8DUC9J)U;h%&Mf$4uyOhGOdtJLzI6@{q>XmU@o{||csKZ?N{I*a=ZoweKZJ38
z4qh1Xay}2@?&@AKo`Ub-<|r3&QIAj#HI*E@5dGqtHV&l-xKJCK_Q)WC1COkVcHJ@H
zK)`tG%G^6_&|P18H)j_c%#VJk5uIcu7HZUXBDgA~h!&EOK{;Su+#85Yl;eXhWe4RZTq*U
z3S&|K(7)w;|4B2PQ&wScPQ$YTG^78hOP3>rBTNCQNAp`k9{>;IdA{g&%
z)z`zyFS#&j`Xk5F=TV=`w*PTPiU)c9Ph+_IIgl1-x?irD1AS}m20l;cf@@dqQ*95l
z!wj*jyZze_P5Fqh8uhb5?uM9L)XTE=cDE
z>WOhPexhE;{`;Ux>l6I{{KYRa8{?c`)t@U9it!z0`uWLCMLHww&n8O#!tt0CArp`B
z)t6pLj8WCYb!xdL_yDfg>tMc62la*QV0E+Ohj?%!&1LP9Ev=yHZq<;xs10U}@%Br+
z-v+nsUb|MJo~U@cKkM~e9=sT(e?00t4-8NW&wbho+f(H|e@$tF@TJ%H&O<%eDfu)f
zz6Im5^JMF)P|rqTR#9{s|94B5wks&1zOs~KW%d&9hmK3@zq^BeQ<~p1p5nN>Mwoi6
z55@nzL4uH}D1UAQj4v`r{Y6T4M7F&X4_X6$dE_g%g1oV$NdCPx5Qg81X)bMtka@Ga
zt|6UU^gc_=Eog_XWxF5S7<9nX4-Kg`%^jfo$m-A?#ZD;tI=D(OwGH}Jk1f7r(FW5N
z&2|Vs(gs%3v)#5#Y=_<`Qz?byc965&exiCNuIs;q_kDEGPT{-ch$EgjABSv9KVZE1
z*!7MPskpE6XBE`1MgK^{T?Pl=XQ3Zt#EvdaOXPDiBMYld(a-UTmxN;=+Mm`{)w>q9
zLo4Hq$IJM37?@X=Q`^u6xhI&Dg>|i99y#vmCWBTOz30&Rx$2$Z7vvoH@NOqar0Z2g
z-|2##3UB4yqxjHZdPyyzjt@4frms)Lwu8Bn*T!3%c2LXYIn~8D^wdS#9vy5DF{r;5jAO<3%3s-vCv2
zlhwXLxDa~N>ef9WWc6vN&Z`!}@yXH)Jwt^c?Z^+ZGZn(GG$vo(mk&qdG+#^#=EL2S
znpy@Y_z+~YqSw$-0Qaq(WZK3G!7cKl_V}$BH}SFGo#mH#z)qvK=m<9rPnXsh2v+
zANHOIsV#Wk{Cv_)_5=^!E&X{qc_+S$uIKz>8tOf-J9X-%jD&D(ce>rmjY5!|zxDI3
z-9pgr4j+5#jR0P!zhS)M@*(4B$hTy1AzVA}_WG=<5Ez}=&RGcpm^ilYBOe5i_$qR*
z=42r(4@h+|59ERBwY9%`vE99O%sf4R9*p)#ntKlW7yDO%zOxIq`}MQ-qcr5dWjj_|
zpF+Q5^QHQ2si^PYbxyaX=l3F5?*k%0J@JoPsmsaAfCTc
zr~0}8>?I^c^-={exW9C-N*d;0{ANCGcUOK7MQR
z1eFLbwAQu+Czhc9#h5bZlQ`~?w(I=vf5tf22R1y2@@Ruh*=(`1BRcT?CfE2^=Gc$K
zVzcr20(fbkl(isE2-lUH&ju-UL#R%^p{80l7#e=DT6IwfTd)3z{P9o-!p{0=E*^QS`kFb}Mh;JRDe-Wr*uq
zVlbc~A&?8)mYD9XSGf>;H~7kpelEnS?-fp)h4%}L9|k_Se;)e}D7uSxz^K^G)9^`3
z2ylI&bmLtsh##GK@skYhduhLd^ddZ`T+AA$uigzJ$B#TSSM3Is2Dj^p143YbES~bu
zT_Nt{ppf=hAUZQ+?Tp$Se@&M)9xX&{?gL$w&Y;=bS>d~rwH`JFi@qRsrlo8%isHfU
zA$6@Cc+PnE$YY>sA?mvgU7a?lXLTIAU8RQfkI-8E$;t)&LXPhp{P2?tJ2wmK?83Mp
zCAcPxDL^_!qt7pM=DDU;7)&10@LAsuO8wKeXW~v>*HM2)#7_X_
z(?xw{0)^ny5wXH`TsQO`IJ(eCz8gNhD7bxPRyQOm+IFjRg~0dGi4Pbr1kaxz%m$?ST`)o6TR=Nl6g=-+@6?9z^1qJ~
zS7`=2-zzKXohA^|w^jPBa>2p~)SM!okrK2#0qJUJfO0sjL(%9k4e

diff --git a/previews/PR1624/tutorials/out/solution_000010.vtu b/previews/PR1624/tutorials/out/solution_000010.vtu
index 431cdad652b6ab649cbfcc86f87d14cf88c95415..af6fc48038bc15581ac963e9fd5d1b0a50b35374 100644
GIT binary patch
delta 12052
zcmV+vFYC~flmVHP0kEb60XUNl1T2%&133XUlMMu80yHqQas=2Ge{T!`000000C=1g
zR(CwsT@=5v5+xCtNy#XL6vDkSD=Vv%*|0|l$tfXObM+On&FaiAUdPICI2cGQ`M8x_{pXCcKMAC}nf3sAZ5!vYB$v0VpNGZn}
z*O*&~+}e;+7hZ~p*p{V39Yu&VwmmqKnUBa%)8xGcxrpQiw?qeAN953EYxjvOh~!Qm
zuqsYPUOhvkzpvdZ
z2I8-0fud#*f7g%3s&Ybn`;u>W*;7Ouwt57vhv-zFWY++B;{W+MCDnmQ@E74U0k}7x
z-JE+#9^6M{w~k;HA_6Xf#?KxgVxjPPonHeYam>55%@BXm@0!2YBeFAeWt>|bBGsew
ziF|PGs}5_lD@0$tQ2xts?j2ugiA5{)%i&LwBGfBre~&Bv=8eesbJnP%F^F_5*NgB@
zLgag=e0)|4BG=?PHN_GU`RM8uxdYy9BsSbz7lBC7etqdLL5Osjyll?(M}#>%yp8)L
zB9eou78+5AX!r3d`lcfyb!K?9NClDN*w3s_OrSsY;sU$}5b=9y=;mRA2)q4hhw8nE
z2=2U7f2|5QGnGt@(g4p2E+4gIz|p){#$aLvA|Dd0MGBY0yW`pG$M_-d>m#L|N{GCW
za*1WWr*ZxGn+x~g+@%ergKOY>slTLr&A)Sg
zpTy14lf@KPb`bql_lUytwMW8!^-y^5f9g3Mqh<;RFLCK~q*3_J8`U&E(6RAa+4^0d
zDO|=bcay!9!rqFN3mf1)7TmicN4`+F=SJ4Pe+6iKLl|%Aw4ia+u{|s8;%QuWFj9N4
zn#SRCN|PQ9fXB>m=F@ZVXKB;sWVZi0+rQgt6Ix+KR1ID2-=e>8^a
zAyWBS4NnFma@ONmUo+@-$*`~NT_Yl2@L5mA`64_4BF|9v)<9|%AsoI^}Dl7Wa`iMro3=;w}2
zfmsgZ^%PSZ2!*@?Uo)gTA#Z-Zf5sOk<6(2;dutg8?Y)541f*%B0g*a1?w@9(ab_Dz-=Kb>xf6AZ_Cr8rNB0(pXPmD`;gD(Z2Ko@8l&u=k0
z)E`P?|0KzOJK|{kXr{$)1t0i>Q_z$)`1Om+Jzo!Hfd6I8EUScmmxTY?(E`3!pAT5I
zAkuo@FJ37N5%u#KHC9gWe@^o~agT(+Z`bN@3ue){MR$9s#u^&iYNnZ`03SFU&F>BV
zrg261xaH-2h$vp^Fna25DS#K=b82EgCBp7_mHV1U(64iV3czF*Eb_1*;|+AFmowfA|SLQ_$1aZW@P3
z)9l?U9>}jUzb0cJh^{N1c5$04`q%oR?)a>
zoX2B)n!+(khTiNE6#h1KAicte!pDbg@8p1<3;oX-{0)aWf1aq-+oym?t#h#U&SXS5
z6#X9u0be$K*gCwf4UrvL56^W%-87pa;S@(+w}Q);a+;WWiQWC*!KFJy@9?I&hkz>Vbe=t>|wXf0QL{g
zYa;aBa?d2KRv-GFy`<(^y1p?(`nLgxHEL{iskr2p?f
zO{o_rO~EJn8UzejpkC3zvA}^(6lT1#wt8wvITtJnbrz-NTt)zo&wsw#
z-wL2`weH(HB_$NzX~TJlmxIQs(*2K)z&vAJRg|0z`VapYjWrkgD|@ve|HccU`(R(lN;u0Q4s|%%UoN%7
zdmbgYbNtWeHLtOV_Gua`-29rr_(@^b^e8L4R0_*aMV}4@yzJH;3mOWe@V5E0Bc`3u
zf4|_U&^rb&=S!6J-9CaJ%-HDhgP&JO)w5iL_pEar=rDnMDsz9(&Z$QAH$yt2Nbe;<5Ny;19+2;ks$VyDYvw13iRjd4gFQXgVMI;fr|PL@Po47o;~M4uhDbb@1DavZw+4g@dV6|0i&#qg76;M-(kxKf1z*YD|LNY4CoM=YLWLu_QH0zMZH!4_s=FSbEdF}%d#X7Wq
z^vcv{PvSi>PoH42pJ?8kvYOW2NsBH&OJfDLq=NS42?M2fA=)uS9jC**zY|ZJRg~Kh
zRYo0d3lj}|8qOIJ?Uoy?Vu|Z|%fi%X|4P@VvnF(&Uw@F1T>HX=f&REp+j^Rxl#^|(
zBaRe!j`$GGyfqWCPl7jlJv;5&Cs}`2h`5bsr}`@5`t<0$4a77Xg{!8-Y|(w;Khk=I
z?ufBc(Q!+%QM0HPD+83yIPTY>h4NI{PnZ!ZoBgsW%A4u&q#(A11
zZFRee&t`X=*+fi#A(6{PlxEiMgx^Cm|K7%URNzcXChfcU@J>RlF>Zaco(s`w;eY2Ke6P0dNEL(jz3sQj!2PE09UnYS
zY&>dG;6ZE+-dz?)6giQef_qGx_ULROj=g(c7D4p?DmIGq#Jg)f;r9ye^!;MQPvHvX
z&v0H%$FKgxeI2ZisI_TI!6J2apLNt*DTy?tT*hlDDi<|
z8h?L1F|x&CAcgoV=djROP~qlj+_$ZF!$Dcv_t7+1-k9jJFcoS-R5sL%upoBqUc-&N
zcv*>fV*L}b{91mZxexF50Aiue1@~OytWbDlBGHD^XeNnRG-=qDm?_2tfuibPKFgX}!S#7$M^yMJiCK**#?2=T>YHeUnT?SsBtLEPBpYi6Yo
zzb3Wl&BTBKrVa9`ulcT*Me9pf`wI)eZ#IFAh|6;6wFg#Y&nZ(Wfxa0riaWOu<@=eV
z&`au#nAVfD?#|pyIO?WPklp+w@v4&Z_$wmk@R{mTqO0^nFAL(+$ZjX-Pu@*SSAUDx
zE^eWAm?*v850KhpW@nq-xcZ$3*_9{k-ZpdKU*oiF4o
z>>_F%-pJG^o-{iZBaeG6ojamLbd0Mm+DqJ`ZhHvzu|q?s=^U}SuEHS;_1C>YjD>xt
z`Y)ss7nh{$(HBl0s9dh3dK2w`8-JWiPn-Kn
zBYuwzwckU(t_)ELu#-!)wvqT)$xW=_jj%Vd50K#`QV$
z!~B3soS!0=bQkOFzE%1ykbgbB%_4)3#y%P)rO>6r%fjgA&_`j5!5Fc}bhWn}*(L4z
z&Z`o~O*syW5<922zL=$XWqm{431Un^h#BG$vF=zC$kpW`fj%2J(R^D3x{mtHgM7QE
zX0)O2%BppnAP#yyMFa3R^Q9i!TGaimZ&kwRPu`j}xYrv72OInpEq`peh4aSi%d!?}
z{O2)isU_mZOCApE$=)a)#($L<+!N5JOJsQTruWkPFprrG>iOC3u>CE>v<~SC`2D5Z
zMj!dz61`{y9vr%wvw|4PTl)Ab%|}IX)gbTp5@vnz-S@*ce%FIOBThu2zob|mb%TF$
zUvCV3!nyg3l7EOy|9^RzZp816m=f|h7O(LM{a~)OE|#09)|6QAne1C%x#=QKkM>rK
znbG*2l(j43d+M%E6m-gaO>#BPk4?VZiv0hz{m*G$BL7Ekcl%6SnjybjQBLaSYjJ7^||au#J$Jy
z_G#1umw(eGC2vGCESHkdL8>xVGUwF(UB}#1#udet)6FKeA`%(H~TYrvlL*_CJuFL7p~5
zI@Tf|d%7)OP1AhEy|n}5sEa}F>xfq>Kf_}Z@7S|!P4qkQ`i^gR&=+?cA4T7M7!o>$
zdS3i6rHy`(uaHxS`bl$(mqh)f*0nRikfqP@=(oo&-F=3Bcf08`YWT?&<*itK4E0rK-UEyE+NlzT^<*qv#bL-SE0+1l%s-(!@VbyDu_?9
zk+CWK%;q^}h5nep7i)+AU=_QW37l`DxZh8kSegHIU?Z`h>!-$-_^(i$6~uqDDu=cU
z{82mG(1$u0iNA0N_2j;*$QAkg%}R>x9e>SVJ9Fa=`o(&+8rFRp|CZzzgLrQ9^KwVu
z<1w>rLEUywyO$rqJ9AyV3-6@}yK^<>FM2P@E7N*=;UsC)Z4O6!-v{U@n9m9ItYxeD
z32_LLiJigyvRWfT*U|W#e4zgcycgT2RncEcBCY!nMmbs{llIEpr=wEJpJ;I3h%cxC<@P85Vi?9azV3m;{JM?*Iv2qo3K5=#==Qwnm
z6<~rm1ZG(`&QTv?K9c{cg#U`JL^*fpY45B!6aKUaHn*t|WBOZm;eW%nIX4;lJpZ8a
zNi_O{uAx74Ew?UOU;zCqZ@S?I&Ho2sMegyFF+CW6myclFTOZ*D`jr5PxS%M)wdcHPSb6lisV(tNnP1UWpTJmLz|!qGD@A
z>j^`KyD0f_T}*f)gyd5~R!5*GaXe?4HE~hFw9G@|g`<2Oa)`sO2zOE9ql)g|o{-#s
zDl2Ip?h!LT9Y0IV-WsTi`rB`TblaOWp34(s2*N9@%Ys#PVb{h&JIN&AC;J&w<&
z(0xx;j5Mo==dUhK+f4F~1M264i2KWv6at9f{go?Xi9W$!L^#Cu>RVT#|B!6XM);Lh
zQq_RCn2j~xT21j6`)bIPq9~)zFLM#?J6oy$YCBQzVN$C(ae-fK
z68>wS3^C}WeFNV{y|FL!ahDK(>QZ!smE=h4ojN~h9wxRL^wlgO9)42b5=*q)68Zr3
z&TYQK&5f8|%=O|Du}(h0?*ehOyqiyt*rgjj-3z-!njd!&H5D}W6{5bYuWyE54!3-S
z6NqL@3-YfMhXTWEkRNXSGlyT(`lEEl4dg-3zKIz{EH!kw7EfIF^u5)8AH-unwJmKY
zrk1*SqF(K@ytb_%D$Pjfg_Hb5ips@1I6q;hriOR%M*D0%t=SfGvTS0dQm!iU)vCtl
zIZkVxj_p_Kh(AYPzK4A){CK>+5I>ogMMK@0fB6#H`l`kJ<~JrLh}OJA+T+G8l~?hJi@&$PRLorU^-=A39oG!5>~
z`xSXBSB@Pb7U;Oyq5j%GH(A2pyNn8F3B2dvLUZWj*VnH8M*E7MY?*>y9#&5oy>b_>WDfclNmMePR#C{rpp+g`Lzf)X8>1robezkat`jad=P@Ka75M
z##itU=xy+pbHY7;g8$rq#{GXDR1;B8hvT1nC_w*-nXrw-v3h~O(Z}o}+lU)0Ck|A8of{-?qa>O85EWet0GC>^pv
z94=1>uR+}1oXg8@WB=W2E;=OFQ1NL;|8koftHSq9jss4}@7OIx2}Ppcgj}#Z`nmUF
zSG?ae8S%Ce+9$d5X~HOxVeF8JcPcsHlY{lKqTQDHM3=|+#e7Ka|K?xTX5w*uFK^Vx
zKbAVl$V2FV<@hn^IlUxQKoWf*FTM}+D}1?=FYJ)(9?bqs^ykU?8Smw~pK|a=T65j?
z7f~iE)bY-pL|+b#UULNfPO(Y}@zi`2#zr4fe7kb|FJeXawn@Yz|1)M
zIa&l(6n!3q-&2ijNz4=Z5{OGhP>Vh6=jH#W1p5*ubcH5J|E^RX9bb}v
z%Xe+XJGe1$TO0OfUU2Qg|H9Yx>;~kUr_4qhd2}>Cd>;N!?+ed|-kiZyALLDFXf(5r
z@ReO1hzhWBBz=lLbXvmp8OEBMD$O$~&+
z!j>J0dAz)6=WfjD#IBotn1d%Qc86a?zp54%!#j!-oXdc}`F9$YOA&>e*O$RAGjm5F
z)J^&^DOc26mW5$C^t^sK6ohwOdNr&baTAsZ2tl3P-=VI3gUDNZV-j=8=AF^}|NjGj
zjE->VGsxYYfp|0TElq-*oxhX|!7o3j?e564Q?>W}1m<1F|wjTOVaBncH8vQS*y9)Vr>1VX!yuVa!KhDpUZQKic
zYm^!5V1KiQOC6JxDMvfN=c7R=vSYo%J>ln;bc}l>VI>2
zN9A|?Ur6R>AnsEpzHH=aiH~zS?2gJlQG`62`>k!lJo1+<(fR-TBzLwWtvv>l91MwS
z-!jAsFn4n6Zeu=K>4iK&y`J9{>y7$ojampozDqb|dj_#DaU>4;>ee*vL_N)arS;fA
ze|G0=9s0nNM4d(Z(NEOOWa5bKY36reZ$qhRJn}p06j2U4?ldXb&5`_MyPQArV7>SG
zHsqn8|AW3g!-&VK
znb{@i&nCO1gJI`0VIv*fD=H&@J&U>u{o?ik`|KLOyhDGljEotvCeAuORK~t==_7g2
ztC;uVJLJrh4d)OKT|=v_@c*Lp$#}fa^2Whh%o*E?SwHk4Ymc%h_+{`=X%ph!y2L0J
zeizSXuKShNx6h>jzOg7;E=)*zpUk(sON-Oq*#rh0U=+`pAo@=g$s-
zkXLa=_|V69d!}ctL_A!QN^vgS00000004NL6<2vYR%;h6i4-DnRj5>wDWUGC
z(njWMxGHml3?6!;c`O}GK36y&w1Z7P*KMC_V?HMo%cNN
zShVeOP{q82m^
z4LG#Bt)oHxOq)RdR|+;+8P<-!qd+%ddbLI=1$iR9#gca@Xe}+RDv6>XWBBlNzZV4u
zG^BsUAEclwz5KtVItBHdHGdMz_$i1;GwWROmIv1BOB%#1SjedqFzb|IL4AXf&cZAM
zcHfeie})-2FwT)3HagjDZo&rFoL~8K|f^7M&BvfUty@x8F$y
zeymbYKcLP)_R)JPzMpAuIbLQjewl^(xlr@aWEO-yzBRm0VL`CbE`MP!-p>@@)$vPW
z;oP?Q_l=1x$koWk=p?XEEvw!3*EJTx2c7p#U18yAVc5m4a276qC>4nGV&U(`(mKYN
zgk9t-m`rBj-FSdg}p+B>)d
z-xIYxXkLkhu7Np|N`Dm=ETWsT2l08G;2X<
z!0vJ9j2~j4_t=k#qGJqvw^GiVwP4`2te8To5(A#A>Gs$K8d`-F)_m%w;U!_VO+o
zQUt^Oxr=EStr1PQi02$^t)ceGo`P2eV+O4P6dc-mHYBx$2id*v&%4+2;8W|m>y2|f
zSaaS$nm!NYVjHWv8hMb?e&4uuf(Jj-THCwYdGI6AN9;`q58@)$yVVHs
zpiM1#a6W+x?ti>L7JriGLT8>rpm7%oc1jEMY1C(T;#W(H0TQHj{PLBNp{+qWOQ
zO@O0oO)cq6K;2*Q3SdUSO{u0`T8;#a>DHEh$R}XKiGSeR^6ms=<(?I5c|m~No(|uR
zUIN5#s>fSYIi&Y7jn((yao8iFcgzb0gzHp%HQ0hg&5e~Fh$Y`D7
zao|j2omm5i1NVCG1l}e&Ku_)nd^E~|$C?karSX1Z+vl~D6bGDTl+yccYko212-e8T3)CSpm5T{!Vz(;Lq07%{EPt2-647Uodk3wg$Zbs5b&@6jb|xX
zPnMfk)mS+R?y;|g3V#xyFW|?1ufTPe&h_2Gdg|Rp5_S1Vh;k{N$~;Sg99g6%IZpy#
zj9c%j7%muBIc}*SxnLxre|Gy67xD}YQ`6J9V1M27Z+K@V30tCVsthnMdUp5xUHF6x
zHSR^FE~o>8+>rar`zg?iD&)2F)6ll^lK<*b2BvrcI&MZRR6Vh7u0?+_{*J{bQRnaU
z%C$-K|4`Rgx(s#s@{|3chHEq=zir&=jd(#93I-gCc@Q`nKO^5j!Km|n^YCUG3ZK7-
zynmd>z!NFwRvTRwyi5wGE#p}DEkjG-Jo;5T?yGobJx#@#BB25H^i
zm6eecoY%2(9iHUEC6B;aH+c#+Fh48G4RF7Q%dTuN#9W!o`CPS(g%RC+Z{Z*o#vgYs
zwBx=%rM-M=i~eQ~6!aBkYZa8D-p?_8iK87U@_2}_H!8J-iy_p1yi`O;k(O&N}uTQp>KU@rf>-|2K&2=muO82!?uAo>*7t|goc6)T07)`*eNutQph+)RLh*U|3WMI?*~@$NlJ;eW!5+gWoG
z-W1$)-8VV7g@K_HFQ0pvU>{hj(mQpPh30&lp0_;=M3wGLA3lNo^{>Y{@a
zjU)uT&iCAV5&h*?IAnx=ZS&1vYg|jf2M;OXd^;}K1odXmSW&QLOQhk}ZKz|#^2N=z
zEZlV%)M7(e7?$j~8<~Q2g?}8(v>Kw|S4Op##W4D+r{w#aO(cZ3{5&rqMSzK#?en2=
z4wToHpHe_SSBpF5?mx~25jiCn_8bM5TAU8mIx=8-msV{;y^Omsf~`kca5IeV>QH3B
z!6*1*;qMgeO!=JlT>;NC<6^*{HwaKHwE&Ln?aQ*UTA8Rk^6C*NJfQMYhv7aXAAdx=-YkpUX&k8P`3!2GHEIJoc-
z^T%8EUFZY!vGOg~U`xbr%@)s@!u;7~)t}mn`Q!Cti1!im=YLCCOH38wEw{bCDi8a{
zwygQ2*&z}tB5l%$UtFlvs*C=jM!|fpQr0ay8kPnNb{9A>u))h#IdBaNSCumi`y+7f
zXrrfg#jv31d`6Rh8_ofKr!!t64rFDEe8=-J+E6X5JzrpG{L3DNCl_)4Q1j>QL7X@1
zZOdk{o|l<%sS?OvTwhA7s0abuYYZiCVP9F~_n`abP6D1vlfrU82HOMsO!Tak2
zBLRrByU*9d8S5D?aIEu1-by#O8VBL|WDULMWR)03bfi0@~b+bzd5Lsa+e!q->&Ce=L0ukq1TS*#)^%TtJj5;E3qkjB*?&JAL
zSgx}cMI9Uvt<_aQ-RY_)%ZJWmj`TT6y!w*|QJ>2vns82T+B&%UeH;O2sCdDSuN;Wg
z^Z0g}AN9zeZ1u-$#4$`PZb06oKdPz;;`vO;oPS^2jXK!iTKb0)>TSuly_z2A`>4n7
z`?8)gFzO(}>KM~7`Ox*W@@1TRehnG#YA2yDz0cF(8UcLOJFked98lPs@P1C5fQHy8
zn;g`gg{s{2DC!_H*L2;{?G&tVj!XQAJ{o=ZPEFsGg}JHyW9HcR{7m)z<#uMh
z3g$55VEaS{{UiQbCMI%-2GfanpH(=whI$tHPh+2oUaOj0*2#duOjxRw5e-2cqou*t
zJlLo|PF@HlA!F|sxCTOu}vK3!;@5jViA
z@EAcyMG-y1OhjU4wHM@d~yk*6?
zyg&ap=Kc|N`R~1`n?++^ri*dY!@A9X}t6)^*=RS_Az<;mfwf=I%#
zXHNerAY$s)yHZOQ5u+es{7D*-6XtzScZwil{ql|Zzr2Vff97vWV>KbN-up}NiHc7S=$Ns5a?7gf%vD&X#Ml2hzz`8
zdqhB-dg?1-e+p4%B$CAg@kO_w&h00VU%}m18lpp8oNYbiiM{w^g3*SE-$$_|VYs)T
z?TlMtHX^y=I$9#-hzL3R8a=H+WUu1;)yL}*iCMT;-30MB{krK}9U?mdmqokQB64r!
zZ>%7k`=rGg?gH_IZlKT=IQL4hsBmu!BF6T=k1>|wc}w8;C9b-!LWpd5aj2+W8Iewg
zbL7GRjq65VMpeSOD{G7TSHkzgHG_E}{D{bf&T0<+rg5sOm%2tPja4`tw!Ds`@xj6^
zMys7^oYj+AEUHc8lQS1Yxm9VLz|3AZq6gpCe+ZsX4WjW@Tr#^#oyKDRzB7BgX`I6L
z5X}y!aiVt9;?ZL?uDI*q%L(W0oZVQGE8sip?3zPwDQs_?zSFjv!U5wVg0C|uoHF_3
zu6qiFYa$k9^rcgn3h31qYNqg+=R!>t@cs9!c40#sg
zfAjeizT@B*oe@RhJp-M`ncfuEutgzSt`zo{$~U%zSfgt&G3ZX=r~;>~a`-(^eQHsf
z8--W&Nbb%)O5r%akYam?XY11C<2@+M8j_mkh3_t*mUHHCzVgh~?Q)PeUub#*xki_E*q2
za7KCDy`IKbr-o9WB-2=?SM8qoRvN!zB+H5cF8eS4vJA?gv7*LhIbOi|c)M&ve~>Ql
z{Jkn3_eUhs{m7dp&@E-~=Jx9b;C*&LOe5&Z?PPP61n_*)e;>~h(ElPH)g0z!MC8*a
zB6M63;f%ZV_7K#29jRryo(mC$X*+*qE%>|lCY$pN-Cb$ora?7_!~Jd!vX{sHp0skKaIK)qnLXoGk6ZYmU+Qu(U>+A@uM1dDX4sYY^e*(-{8hfXG?Ve-y3`xKF{t
zihwNe13!xZCkpcBDRo&7gTI6ibgun<8GnT?qG%e=$r~Q%
z4Fp``WInY;)0jQge0-@O+{@8#LKFP@`IWBE2U4I;%G9D#=yyTz&uz`%Yc)A=r)ET&
z?jOIToQ8<%`IIUPM?{p(e|AMb6h%aQl@^~!8jb6vLisagX0uTl2962Z@8bC2Ghib6!hZa85HL1fHn
zR?)d1^tfNWyHbP3n{o{~9yL(7ML1PbWEG7$uHKEZXr%E!<->|Uf52yQx?0&L(TFrm
z-z(>b{9FI7OxXkZt3y<;Zv;G~vMSSZ!Osp&#EX50xl-2F#U8szV_DaNFK50{_)<5|
z>97b2yXKneP6NLY(&~MaNaOx%436tCmjazkzRE`;;+oZ;9|-w}yc9&;p#L=iVTvyb
z!51@)pQQm#r|fN!f66?KoxLB)D8Rk@)4l#NDWdS8dZC0A=+oDCp>#9&K=U<@qWn`d
zZe5$JZ?pjs#|^VyA&~#iz^NSJ2jKs{#%H8}kGR;2lXV@Dy4?XjexMJR1F55Pf6p9AXU*?jfwdq?3L&n+#U7{FfCcue3K@O58@(6SQH^&{;78MA|kn1n8^
zGzMHIe|&^)xkA22-U~7BAb+8L{GXQ{h&VZi8STA~h|&2(a7T!O~CCKfmApSz*Y|=if2ucV*oXrT?uQ1RmhFOlPk!D$;scE%6B_SL
zzs;rf6Q_U
z{|XN~q7CzZTF)+QWUjdta+MVnXG0aPH(wd=hw-v*99<)T4
z%QQz*cv-)*-$x$cnar*+Ht>4Uf2s^?Z_rQJ)@2t%q0XnQ*|7|mQ>CUd8IIrwcM8Jy
zMnfHLo~1ckY+&E*>tOzP4|P29UeAfa{}V-gOqT@ML$2M_TMj%Z4Gm5{2)=pKPKA3D
z;9s7*Wydz~gIm9jcAW#ghRX;=k#XB7p0zz%uNy%qd{
zv#R%sFVu1G3rl|nejrgI*DJ_h(MWKjNkqODZPQ;63%@Q)oQV9v!
zw~W2_T>bU^-0Sb$XSvQ#e>&xuEXTNBwnu!@$ikRjRkT=tI??O&>r~I>F+H`Z_>Mhn
z<9a>g++I(Jf4K=^vag9-rahNL*vIvXVoHOJi5gtOs^Y{6yH#nvv@UcYBzGH~AG$Iz
z$|A;zu{;B4-mQLf{A$|w@NszISsE*{#bmzzHm0ZC9jF;W+;QwM&t2ka+
z`@%$R&+2oAMC)(W=8?p;9R=4_=sdf_i&-Y!=jU%IfAe~AOiydXvvEDGkIFC_Dv86H
ztN~A=v4?u}_j5cc>)GktZ|kaJg^1eRJ5+xl$|r`WZy+XE%3U=gW{K_#-c9?Z+d_ud
z(*5_uF11@Njp=nvoxdkT@5wN>;@72hafZ{wa^ec@V3pN$&bs%P2YST!fsSf+#G)nj
zehcD6f4QBy8ZoHqFyJSK1^K8em<`sYE6KQSnpFT0qilXE{;oz}Zc
zO+EI|JN5-j@?EE~)35GxEaIK@^*SerhZcYHf5-O@o+MV-
zGc#R?4FUQEcZdS6iSc;HtWo>6EyTg@nt~9b&y3g*?vrk-V8L$ARf+q>h+V;Qg)b40
z^5#E$h>hb~)rd>~fNVi3v0GW%o{jbmUyv79CBCo~@DV49=e(c7y9PV1kBSmA4IcAX
ze-Xo8nfAmJPjU7Ooh3TT-8zl;)^}_;$e?orMgg+Bi1v#UK}@2ofqIB3v00yQ726#(TH1lpZD9A^d+o0d{|t7-nI|JN(<8{0zptR(WH
zP9=OZ@vj~?OT<%4y}M(I_7|-75f;FA*aQ^e7xrh@vdw8ewLmEk`Ry1Xzhetgw##h@
z`H~nK(eNj}J2xGSy7?f;Zt^GbvVz^nJ0fTQnX-JMgVaCnro{Z4ZO4#*GEQ5yf436f
zh?}Y$CdzS@?I?qP9nP+JPuwt=5?DaIamOTNH{CB*$JmH?|7bRq4tq5ORQ;uh|69zE
z+e!TCu%eqT(bf1=ge=~*eD0_M(dJHB&R*z-s?{OXho+iP-8o`oWwA{%>aT5s*c9~P
zRM*7>;?lCDHFUx4K*_ffsyE)Qe~$qP#QF9w!8$bW)OjcP6EW(~#Th}`ui|`{!AabG
z&z|!$&HEj28<~I~({(H1-z(aDmytK2rg8;dMEinUCz^>O-3IZne@AQF9q7r-lfc6%n^Q#N(Qtp$P|(-)$}pdCrk8(~Nvmf4mfKxQ@s&
zIWGnOWf^FGLjKx0ib{YDT%r}V#3H4S8efS*kG6K8emUwZ^I*T|#Vd=`&^eXfmH!c2
z*?+HuE~yGe*dZT6EG*}cA8)2v(y%9AB1s+RSvwsyVV4od?}kG(|L9lqd;P@!*wt>#
z5vMYIOK|@av6xuwJ2GG5f4zu$YBYVqOXD+YdHKl86Rla{H8fwG^-6D;_^;7w4;h-5
zIMR7R8To3&aaa`fJgLz%3*FW=&>ka3WCj|;AE9fH*MVGbT_m8hSI6q_ix9VmjdCO2
zE(uQ>p}SHlZA|zg9OXJ4X&Sf7mIU_3Ae68>%Wu
zUZQdCaSO?1^an2&oAoqbFMgf>D$%dqw^N(w;L?%ULF=Enjiph~wffihze4>qOBKWJ
zuWgpPh_6cck|AjM`D*G4VgOHm(G0EM4C5+C+#lSVJ&*79_J`ipL4Jm~hCyEv%5NS&(icR(}vvKpaP+)CQpka}AY|tDyUJ(V3$(ulCMK8-7aN
zTRdzG{gbqCfPdX%w}l~3`LC^8jr$|xE;k_lMOsfMd5CMicf*M
z8xD*(K);1F{{7=E%?n;6hPeApDdaz+`5zyCjKg~*oEb09Q(r9N
zJLO&se@u65O`vsyDffhQ^hISOJ=B+iR#7nYqHCq|e#E0`%~C}*k;xQJLf%a>jt3x)
z@-2lrh!dMtf9X^&v28M_>Im6y9DHH{aR~j*xHXFAC2S>ZP~Q`e^y{HBwQ@F>`OpU&
zJOr9(zDBactO5PY=z=i(cswR{1@emdsXYvFKC^eLE8?UrxQ^9DtUJc0%1`t6tX1~I
z&rCZhYsBs6Q%l=${x9a9@AbaLZ@A`rKCTqKh5X>!f7J8^`KVyg&H?*YKKhsgc9t26
zW4rxd?F`{Zh;gKQ)QmEJvQHzi&SV
zKlhriG0UZKbXejR{0}5ICrlt7?FTjGk?*nxs+Ljjx1LSp!rwWfw+ryD!k$mIuwVM{
zj@!^R-p7$sxIZR&mpSgw^>%b4O2Q-e|esT!9nG+Z^
zjC!8$o6v+_WXq*yp?=bxqSm2)5-VT3f&R;*QP5i_uh^H+8*lw>-0*jSsZwJpjU^kq
ze-e==e0>XX&{x+E11I?3?YU!$C2=~l&kJ#$d+Txud6s@4^(yjmAl_aOeu#TMt_1%)
zHQZ$cJJY$3n?oP(^F|)Qf3U)l=`q~DM}EJzCb1-Arbm&O`S#b^R$6bCpB2P^vl@ry
z3D~1_wz?B_Fc5X|5bDWwXO08n>1!^@f7VUwL1#igKra|7b7Olx$r3Z&kfbL=$B&bc4b)i_E?tPNc*jYW28{GnH;Y>`*2SH?=jS~hL!pt
z{NO7cIgR%vHiQJNrSW-LKOa{bZ`PVrhQ8$8wCIE%4Rx}<_nr3Y%6<5MiqiF+e@7gm
ztaOxH(0`JwV)LLUeUByJzq=!TBk)H{OI6)=_$%{NHTn-D=j>keVTLs;2ysnHPc%p0
z+Sp;1hPueBU%m$Ya^h_lhQC|GHXXwMMd0tkYS6(FLvMEEr{z+~56JWJvjeFoiDk*Y
zO!y%%*`juibcpp#c18jD`Zn4bIQg_^R@@Et)C$%&DnYNiUhTyH#^L(3IOOMF*|pEZ
z@osGcALOQGw$KVp#Uq?5rX7=M-u00000004NL9hP}KmRS_XktN9%
zLkMLkOJm7aCcTp-q-2RGDod6^T4>Z;vL<1WrHmz+vd4_b+d`>`WQ|IVEu*OzuPsYr
z&ac0|pL?GBoO`x=&Sfg=M@TJ>GTb(5G;7U_GE~Jb1oF*~GVDt$jPk#YG8{d473PR*
zH-F?8y;~S%tgo-(LR
zd^{5r^Aj=ftVos;G2Ny#v6arbIGtiWK7TjL80Z*fG3mU%=v0rBB)6~FTq;DIHF?_k
zhUDUXhl@05oyZ|2kwfIIx7^l7OmUWoeoHjh%+%#1y~)pN19*sjNmCsbB!8l!Vr@w4
zF#}6a5%S}*=<=x$l8*}PI0QZMlh=iJ5*G_*WbP3o4)b-&BM!SFJVl7C``woxk$>F!
zTCyX~5imQQFi*_h8l;BvTwaWX{7P|AkN&c|oaDYsGN*$Op9+QTAw=CnpX%+2Lp(~7
zi0>xhg7%GEF)Da&F6vBU+AG-L7@f>n}
zu=~~Y1up?&jUc72*}eer7eX_Ra?F2;#}+Dn%=z
z4&$}aW#Y0imj814PfGB&@-;+-6H}f`VOLn6Dj%^|bEIC4sQIJXoHy-fdVd|APo?u3
zR1CFhh|y`>8JkJoxliMKFwudPtPqI2A5*>`OY~z73$lrAwRvgKzhCb9dia%5UekoQ
z7*Dk3t)lpg{%7DULHqVRWX~KU2CmYO&m@KnI%Y4SeP^pTzA_~)yO-QVHybY%tk@*5*1tYq=BKHucu=65Ssay2c{TTSD)m-nq=S
zd3eB{GLGk$h>fcg11=E9S>1g4#CLk(Gw)!RVC%!zL@fnP$0DL#?bXfjYfqlv@xUx_sVWysi5sVE)YVCTq(}Q~Bdv$b<>iKm)k?W)$XA;>pZ6s8
zb(;Rx0KL)ix3Nz$fXio?IBd*{MxM(y4XENinZ|9Bvn2lfATo0uQPuxor9bU6^V5H;
zL3w2)PDsQNKaB`D-+#sZ*{QZ*Mf(I-_qMqaTbiT{^oSGjE`z8O@y5u)8QQnOi{AwK
zRj{FQA|6$99yB33imd%XmUzkdb&3Yj*SaMXb+tw$v|5xnZ#msNin{&!Rt|CVn&;f#
zM|>f--yD9qh~8R+zV@v{UWjjn(5sc-^^auyIY-~$JKL08Lw|ifbxATIS}?owPZ6D1
z%CRHFd>s!P)L-l877O@WYI)y9oM^}_GK0RQ8?|d^Y2U3!Tc)9xd+==~Y@h@iO$CJ@_m{
z3VrKJSvt;>_)StA{X8z*yB6_CXiv|_|2)0Udp{6mZht9^pw82(2Y13=Q>BAeh(p2*
zQxZ
zkP+(`qkSv4KTaGcZfESy#yyqj_q&eufnr;W0^-4kRib_*Ki>2A;AY~njXu7pk6H_z
z6y)J-e1F0O^aSyS@=GAz`3b$4Um+`;{9%WD_s5*i#1mZC4&c6A4NzwOOY6z9V}i;=
zr3RjFCvgu#qh$}F-znB8A)Zb(OHo$jm_ib(4EU
z(hc>NVQ#>Jo{kqI!MN9~w6I3RZK-%*2d*KsFPPvqIY)^}lw7#hgVh5a{I?P8*yJR@WcRuEk)*sadaTcT
zNEBot?$f{c4}GsoKbK6{9i4Nc7Lj4=A44AO_B=B}o(8H{
zT|$4qIF!%Q#r@tfl@v8R>b{Re$)f#|P}QZ61D){{A2`W^5(Xi@3M)8pgu!Tl3j#
zPr=_a>E@UR1N*)-;eGVae&6H!Fo!Dc?*(sdv=o7U?bA~x@Y_B$xPKXQ=t;T11m4Zo
zU9q{@xJQ3`w<5o-iQQe$uQHOg9q)JAS0~8Pl!j(9fJc?+$|e5qpde
zecaYN^V&+p!!@}A|AqhZunqRMd<*kVq3?G-SPJ4@YHhs3(t_k_Eg}i%&s7cfTM*~5
z#`X7?K%e9ue`n%l^D{^l^o_-CzB>ght|dHOBfTjo;NqJ)-W=2u_`C=2A#wsNBfc>j4d!K!1M4@<%kz8lR2d
zSw`bmRc+0p$287eD=SSR>-JB20r90YI)hCTHj(|eM%lKido((hCoPtBVDYKT?cKIn
zEDZL(70>y~K*Pzhtp7Cw(}ZCq!vY3ZW!mzTFEeN;C@9H~WpL%A=WvG~gLQ^8e#Wn3
z(3pDvZ(5(hbHPkSNq;d0M^fyXrd5mJG$+4G&Vfg6vAA868V|#{R;J@495(cw)P`dG5c;hLv;J^`p^5xsXGb(UffE>m0I6w;ann#bL6dpMTH}4nqq1sekHo
zNZou*C$N*nuC0X|<&W}c7>%@#JkLXJXJ1ua5)av$4GAmBwSUCjtENF`c^q0iR#%hA
zL!(qZ&NP8XsrvG_f1TzL-0ih$=s1tO+^9p%hk3-lDG-nGj-aHsEnQcFDmpEgtW>
zMs139csLw;m4DGqoovOl8lXHD7;8h?bH&}0TEc=|3`vhEUvjXtGiv=@aZf5mIU
zj=}|%SIdmuDSR<2D|nMbVS#V>C5`PAu4L_#tABq=!P}%Uu(6GT;syP9hes4LrbLex
zA0YR+Ns)QF6ov+NZWb8icZmepQ&VnNB@p0R+vSi;
zv;f?oLCBpi0^}Oq$e8i}_2N#|K}G;iHSKE;$iB-l;=ipYzi;)JwQAk407=m$^-pyu
zXn*Z+aBwHMZV?}*cs``CY(+%2MH7X(vr*!P`4nmnoO_r=&XcyyucY@rja?_7OXdDc
z!9qNU|5iky>B#=}i{w02+hr2XCeb+RT`+WIAB|b`Epz2D8WZETwJF32VO8S3pokV?
zrJ}{YB|}1_uCYu`O%cMg_2uEFVj4!rTz^W|ki2-kqV-AcJt2y>-zxAXI&jR2xIVRm
z!P?kdQGExC_UT6sC>0QYih@nIS@9^i=UiJx{6!yd&)Y$C{=)qJa$4H@za>2*&Lpzcs00~@qgHF
zlRNBqiib*?v3Mx)>+hoy!NYTTTpJSG^6>x%gEw*8of28dnzcN_M@Cq*yco6w6HsMsXn)B=|D4ha
zN0M{lzFD40&V=V(ZQf&v9IhPLI%K_@g>PB&)?cL=jO7`vl)NE?p?UVy$-8N&8Ma@H
z{Ym5W?r9Dl;Ue4%i;ejx!NR3-Q$>>($(v@*K8xhwTHAuy3FQA#B(0;qghzzQNx3{@
z4t^igmacDL@Fps;r`22pO@E(Qas65v)`!j*RuX?rN#5>J8%Seh^)El)j|ssj>18_?
zF<7eQpWL^K!;;HEDIMhf9UZFD8%ZwrU2pO@D#fGPV}J0LMiwi3epvl9B6Zv)+)#g5
zhzHZ9rj*LjcxEuel%7vvtKa69tO+z2Dbck%NkY_IN*`79XYjYrs(-=m1sp#6K6~tE
z!{hrDowlJ9JZf@WTB}<*L=`Me{pibL%U8#~sxu-iN}jN8poT{1KRLTs9->egl-pxP
z{Ms3qqiS77;q6Wp>6{HhxE^fF_~FE0@q!r3g^M}-v*`ZBT2~&~Zr#Rw1P_679AHhQj)d-??^EW5u;llZwp-aTve
zR+4A4wY>TL3=Y?OtS@utV11R%)
z?0I>i5axNiCtW2tIu0K1B)UjWS9r`WpT+k$)88>yT@;n{MoDiI`RhbvHnG$
za7TjQk|CctMDoYLsUx|Kr`zt@MiXmUACTFCP=YBUNkSih~qx3uS_p*3vjKTvC|4M~GaniYo{2
ziXbmFdVi;TkU?mS)HnCp94<5jv^R}$i1=ipdG8R9mwE?8CIqK+j%(ovIZyqSQ^|^i
zztinW4YD$%&ZU;h7fD|k-+!a!*-}!k8MO55?-W{l@5Bwu3L$Z9!{ry=BE0u!a*QfT
z|2CEMDIoI#m$7m0qHzv&TUQVElk=nu&-(05cz;W8iPHU@^xkJZ+wC+3@kdLK%ku)vUzt!hDo^3*$yk?6qPx|)vxmPB
z9i(L0&f2_$fuz@|#P`HUL$6=!S=jOz9a`OMPx@Y%twqr677nkXk35Xn%)-_9&wtRR
z1m|jWz?;it&L4}<*P9hcVe_X~+dh-IV$m&i9emX3Y
z#^sd?QhDYSV)`6nD+tc3jYk=oaDNJ7Yf5Cl6aUn%nk+X&d^%S9Lw*fe-IoQ6(JxtC
zxV=c`IKlDI(>twtl84`ld*&&{9H>{=EzA%_=YYyy4Ukm$|kj$4N^G#pFAv|6qTlh|DFS&iphX;q8=nT}9{WalQB2`?^0nY^`jXsOhLF
z&_7jG1!aZi>`&=Gf5d+qyC^Y{#7pV3#6KE{bh}8;EAl@YD6~0
zQE#GywKb_oQ8FNNl?i<@ToFH`IlDUI-SU%K4Y%gb%f5?5>QC9cb@CDlGws8gz6mDj)2bIIy?t_%KK3zS)v
z>_7j`*GRN`=E~X2|9iv#OSb<~pSqOKb&0FHtH+A}KJQPlN)3g9>@Ot+vAyoDUJGZg
zSiRJhwO12~QhS{|yjCptp1Z<(x$C^K&_XK}{!6|8HUE6KB|Lwl|B|0dHd)x8Rfl|1
zy($XY)Xyv2++7t~bQ1nGJpI%a`u)K!Bvn7?;L_oLM!plI{8Ila@%-~v{4aBlxU2a-
zSqK&W-M@`_KT*A2D_ocUw*#ZiD`Sml#4Qocp?#D7HE6&EA)Hl^sFJv^{&y{+d@9st
zP(?*WR)DHlo>Bv`0~Cr=Nl*HxDT+z?&%R<;4TbJ3Nr{z2^LsY>kESSX+}tJwh2;x9
zr2-Ue|Ml1Z6MsnSuGbS@n&n@G(rl2dpd?AZ_%TJS()&I{qb-5Q{=ykRMO-5<>D&&HPnj)2xzJdYmy2=^l;y%5k=JD2T`p9K
zyi|^)KS<0!fi*-XaGxp{Bzszj>6T*tK(StOYzhi;K}tU@rk@tm&x`4sMcPA54-@&T
zA~m>1EtUx%L_6(R!)3xUhQH;)^EZ@HCKP*c`h|6nOc)^MS0XXe$BFz1)*!hsP?TR5
z<#)t<9;^f8LN|Y&&y+ko+Ht>#^GPi^uW8HqZyKDB?8^BYdpO_8gxkHxi42tsy#vVs
z%7tV$0CK@Qj;DWi=d^e(^{iYN8l%JL`+mH>1PBvFh02WEYs6^`-0IaCurnl
z!Y8MGEZro@h4Fi1a~SU)(+-Bpg$ADoGv3kBlcjfG;LZ4RIbT>g=KfEx6lKEeN-vf^
z+kQVwpYQpOr6(MufsqTbG1g6phK}_6Li&b}-LM4Y!lRqJi-}(p-4#AbF4T@WVoCJ3
z^Kq7c+;n@Ezs_d}qfvlYlzy&NKCADHz6!H5!BC&c*KG}8@&jQs)8)eX0-6YNVS%$U
zmZVJhWu(i})%AJ53>=ipc-!AJVP!)1x2>6-!#pQp0OZ2j&=;)#8=nkel0my`7wP+UfR`loNp4bj7c;7!X
zq2vyQ@V0*(dq~^I7D_i~5s9eznNk_n|-J!f9#-^@EtsL7iL(
zq+x`-u{IxfZrH!~Q5DQe%C8{yiz>0;S$Bf|?Y}W8S$XS4O1(hkOZpPeia~x4u{<`M
z&|_wTvhG6qH!=TLalCu}rO(WQep69?@*S`58>>hr92Vsr#d4&6lKiAp-{191a_SY<
z{(qoSd!&9~YOwpCC_i?~|J6=r7VMX(RNhjtKFM#^i1th6O8p|WL$Y7WFY!{kB$x94
z6J^sA_WV=&|5PrO^HsE)O*51$*&*>#KTCE>^XG55RGvitr*`}&J;|Pb(v!wTvP&foJwYKN5XZz|a-wM(MXeEVDfpX!zBk^14E^rZFuPvb1fZ?lSJ|JvUq
zyQFf{#Pk(RAQvP%C0s7m^IFs^WroTAg~O$NL)ssW#Bv-(eF=vbihOG^eX_`(68Xj=
z?gbd&4pWK+d5VfZrgbaKH#hvOLo7Iu>h
zZ(Vq~2?1Q+>pnGDChXgm&S)GZgCAzj(}Vx;@*}TsyP|5kvT|QECQl#};(Bu&6Rt!%
zlT5fWyctVBu#OHMGU3G~j>9@QpJIH~uF)(#qvd@V3V**{2QCEXrdc#oMQ`h!lOa^Yo=fGH~%Mmp-DL*>HK&b^r(
zh0e!u)RYOC1|}m&9ymGKpXedK@vPkaCMqygE`)_wGko9c>ksrO@U7Y&L!!>3q8a|*
z(4liFl`~=-Mrb*ma2
zN#(-wz*uygTo}FQVl<@(oV7&YP%gY`smt2+`I4pL56G9QJSii2^@X^~pWv_QB|cLG
zpJ*|?UnTfn19m+u1AR5100So%BAYIKM&+IEd-^lc&9lp}v}8iH`fJudW2fJ=BHrMj
z5wGm5hF*{AC+pzHuHRbcO;1Y%l`Y#(eGMcJhi3uzd9yn}b{Po|9*!-FH
zdKjZ73(y_V-(Rf7>c7}9gtc!&Vi1$7&!QbtE?g<=!Sr))9mftK6HE>pvH7^d&YI1u
zOP$a?&};6W#q2sV?)U*}=h&WhtY0pkal*oq3HLX*$s+o>*)gVX(%yp605>{F$b`oS
zP(5fWnT7It1@d(|okbYr!sK?elj3}o12yQSbm#MYE*)FJZ!D$=giN?OCWg_re!QHs
z5qv+mHx%X}|0ch)%ueTt`xrm_EqOAT@USnMg>;p^bO6A5NnXPA#?P3|=*DaU{Bpr}
z5I_I6yui=z`3Je3x9K>G^y*w>NB$!5+>^loj!bZA%k?%rAoWWL*zFxM=KND
zC>iPHV){}s{hP?U&!9i#!XT_h(!=CoG>L{6@?Olb!=63d?s=A6-tIQn8z}aZCJU6w
zglUwFe3*QUN;sLZkh9@HJr!a-n4GLW@-CE5!Y{0P@CBlrk^hok(r3q9ly^z&x6z_L
z%}QzyJ2rtbGm(GH6E0sPj@NflKU%cMRphZLv;0hq{EP*a%9VJ@u79GEo@D2L(!(Z>
zfB!y5{jD$IzHm`bI=4#rNjhIj{r*3-SL*M->-)dUOYQxi=9BuF4F~#HYM(?U`z8Kw
zDy2*L|EKj!XYIh(mIvullnu#=~6w?{FU^ibHG)x
zzC9w9@=5DN!k3>|M!8@JZ%XzCh&=Bg;1O1_Ob8a!2U55e?J*Pgxm-37GNA)&uw2Ly
z<3c;hy`ns0QO`mwM^og>ME_FRi|cu@4GefoEGJ3i+XB5&Iab2ltbqh=V)P2*`yFO~`S`
z1#9tqF?s^mx7f|~*Bes@$b^UwQ~}~|4S4>hqW)awIAlVnLau*uJT+J@C{Cmfl?#Ih
z&|7?#(fP{sQu5VUedj+f?@H-ETpqHx
z^ep@53~&Fc{KfEiRlnZ|Wy*zLyOrQl;OC+g8BSAs{+6}ty1`a-h+I&5xDc*fEdoJjaS;c82Z{V$0O!$65#V~8CN9XdC%jMveiVuzX
z0@`}jje%rm-ONUnq(A7m`|R)FUwfW%Zv@`(=*XTeKsBFVzenl!hb-(~kM!sI@#ZSv
zeJ9NKl!Fg_67e7z^uAZw8^Y~1Tf&YsM0%IB=ry0fm!yXs&H?>n5?1I0d9G)TPy~L;
z9kVWG;8!d)H*E{P^+UhW+6{nTt~m@11wB2%uU9qXi^6u@P=cOa_N|3v$D(NSRBC@)
z&5(1{KczGM<~IXhT;3KTR`gr3{yA#bW{cNs{1bX*13=4#W(om~9zx#aXZh(m&}(mBuG|>%yOTSL;Xew^uvo
zki224dIHh*n^m+(Uhv&One?WXG`FJucr~u*7Rl}QE3*0OK3%gZ>EAhhQbPlB1G5Fu
zRp37@I+jZP(!MIwp88$kyxvjD=d|AhJ1qQ{e4!A>g^oTuuubT4}+rhjXjBjeTc1!iZHH!5u1tr~Tfr%&`}>Gmz@<`eT*$#DqL
zGGSAzwYbtie5OQ^_2V^zENpm)
z=X5^I>KV5P$6D~ZJ+O?C@52D{h%%wKH{TD>)eth33u@D09_WhqJl*vxc}%(R+KlgK
zYm`1RJDPSS4=WRl#CxfMzc5_j>$1o*;(k9s*ZG#x~UDo6L1MGUlXRqXTsZufE
zwT*fDC-*{;H0b
zH!b1=v;X=)Iw)ZuXw3WJ*cfilUh#a={sGVDzyJxxo4Oz6+9NmRvm+M#FVQY+MvQ+c
z`u}udk&Z>d^p|57rhG$0p1EbDhl%>j#Qb50c)r?PuFuqwk7hTO*OLKq+}kwc03k|?;M%_Wj1kvTSU2=sOK%tr(iab@jd|FjmAfs
zpHlk(zF0Y}#QIjyGJ_q9#r!s6xj7_;T#pSj=D$c6iFUxbv-B2XIr^+Ya^a>JcX=t!
z_qntU!H)SrTU1^Ta!2586)B@k2#ewTIt!T;3L$w(%ovjIeW&5Y}9a^rf}thiq0Djqk>y~@*%Jm>W&Wpn=Z-0!R$c|FJ1
zt%7JoWkT9m>R7~i=kR)NuHo_h=~c5?x_%w^3-2e<2+IT~IrncVn_nKt
zSYf1P!o}IkS^3ZPk7H|+3C(9(G5j<-#Pkd4sWuBjBn;<*{1b4ba-sC@jyJ?F_)t55
z_@jnKI6%mS{uWnQeHCq;So@>8S)pU_Zeov+MEYAoCd3g=s_pO-CrIGMXRDZ9-uwKR
zU7eJFV=I*l&#tXBBRpMPzQ~^P#gwM8dL|8znoqozYl;csrx%|lGu*Ci+p>VUnZP-eB*ux!sW~UwS`S9)M2%wEiyPUA2AlZjJ)Z?Xgw8rWW$~
zed~*EgWm8xHuVhXVVjnBVvyea-SDaI2L?f@5_@b
z!JjbGzn==aW`$ok;p;sHb0_acdf^Ad>bc^
zbCu!aK^ecW@2)=T@C5maSFYQBrb;F(?s4ZuX;ZwzD!c5p>nrY~VmEy$IE(YzNvFE?
zDDR!}vq>c=XGV0FyL#Y{Yfl(&0lrbodpo;>e-LB6;o?u3&@C-0tR3X{wNhtL`^Rta
zT~D|=;78EK0`Pfl&Hl_$lnZ?peRyq%{yK1{&w+zTSNL39*#Y!intT}5Q#EO4D4jn}
z7zqlQSoxA(!|TBO(b>%}`{zqvNa+s*mv&g`jNh=cK&jr3kH=xq4Ti3-9kOG^O>c=b2>l8P1t{g%758+
zK^*A~Ocz-Bh4Hk}<9(j`9A^KNB?Qdz&aHzXqZgXbXK{$q^|zV+xdDVI5vQ_V!>Hwu
z^-M1-_&7rMazW!Q0c-S&>nJ9#`^5L-iavhuWayu#p-iu=a3hN|Xj^!(@=b5G0Cd6m
z-_!$+1^9JVhfL}h%T?d{6W@1tYj$3AyR1@3sbN#F!0{%FsY{B8#GP`*wUPv2-xh!FRS
z_1KmnU#9qh(T)StnEp3Q-d}k^6kwAHZ7Z)b+AHB894GECvlsJx8aNIkU4Pgk#vfS0
z-<9@{=KeU(3u5R`6Yq;$(Obknre+|Yu{pQ%BW4B3oy2p~=chdVduJ|RU(EgO#ysAB
z3$Z*G2Ba`A+_`>4G$DALXXo>Cmz5mN0soSQ59Pnf=k(wRGiJ9gyRgAJ_vHHDzw&;`
zaUSGb*-y?A*FG)$0pCHR7J;uAPOQLG2foXH8r
zkRBtT)Jw#xh?IJp@=N%e!9AoivmuWb)9G~s<&*RzDwThLrUCTdiuxx-{Wc;mm3Q$z
zFNankmA6++?b((US
z%Qsn(A&BRgQbpJg4pIfcomo8H4!akn+hI2*`DGDrAN)ZL2d)<5`juk)y9IE)F{{X6
z;N5{df6Wmtx7y3;h_zh)+=uUPs&_bbvZIkh`{wcTFP`W7;HlWIth|1cQy5jxgGWq#p*zO&yoUJ`;ODKIt@Pem#-u;OfIBmsa_qoUIM#b^Hjr*uU-4*+`#pZ`12rD6dPO4};D%M;ttA
zUc?CC;(=E3A!|`iBbWJ+4?(|tF__-270Ruy?{MQg@Y&@_o#Irn&u2S)tbl%D^KFGz
z4S~z{t8|Ok#Jec3thlhmO5s(g+{mID@4DPeE9u;!a4F};IF$E3zU0{(v{yUkQ{4f`
zcXzq!v=sO_^Lof9b;v*Z-rNuYy)MUgj_L#Yd-*UR}MfnOtb)nE56M_5S{@
zP_`ZUX7pNOJsEn3pWZsO4)j;t_D0d);Ab~y)!x>T3nA0bZZSl;yN)EqJ%{}{`xkxi
zL;mz!g>AlLaXxaNpmEp$_{w$5m1XL3;m@U^>XvngZ%oz=NLP~!D3Lz4{a3
z?6JZYEt)@>5ozyf-qhEwxlQvZ?n7Job*pZ(3FkMT@a^m%oa3fOE{h2T|Nh;gEh|7*
z<@|^jpdazJyq@~C>&+cAf}7x;bz8yB1jy?bf9vds@%0$IEj9&wxJ!$fH13TQY6IwA
zX!@S>XUX35cIj|UxaYAL`;7R2#R#d&pw;V`0`=MG)G8S?7xy+_eF?mf}1
z74>JqT6t%(=hua-%cQrhYvVq{!Mj!2D#d{Ik2rygLc|f`FKnap_wDOTlxh8Snbt##
z^m7-~C=q?pZZzAEJreG){Y}3-lI@SnMrRa~JsaJ;nch)$k&Sm?Yh79TkR(DFh!6kz
z&hB*_wohUEnC-)#Ouok&+aJ@2Q=Cl3eGKeBk*CcK%7;>Vyp)=iTvc
zEI#qm^YIzt8@N6&B>k@sui!!k=XCdtY(LcUXy1|2r|stV6^HbanLS~79It-+O-Kam
zH-!5+$M67VM_1i^CO@7=jvaozj;BY3aQta=pg;2mLxY|dK)JAvHGS+
z=Yxw#MmfqSkc@JNig@w?c7Nhye{w&Zil_d}BZqAkUw$ggC8ST0Nw^~Q_x0?LMZEJb>i0Y%&ciu!HH9He7%
z;2-uYk-n+s_El4u3+WeeN~ZJ*f)(H=iTP{Z^7gb|&d2|DSFShoHP?FvuT1*VIl@^i
z{}ri2ew><%atDhvUzBeb^=6B_8g&QsMu_JVIt^0(c{JT1-%sX(4;1YR7{TRB=~#jE
zMKlaZ?}1%__zplD#2?M!)Nh1+wF^`Xq-E@QiJ8fyuzJ~c6iCx>7;TxXvu{e_8bSAxeaFg*@XTm0PEo>HAF5<@!;dU
zwH2Rtw-0a}SRXeF33!h&o9itq;prDlzp#3*y9{9YAAR_PkpQk4!}ke2X9DWDf4_f^
z>7|&aGk%fwKvv%P=ls0A?QkkO4Et#hWoD1>I)AoL^$x!hOLC#_OqTCO!w@taanm>S
zLpX?I9B#;Pn}wd?MUo%NHe&hw%93HAT==X%nDL$S7Qt}5%UIDhp6CiAIm3tg)uBwU
zysH)4XD)2Swuf}17RNC45qGt_&G7211M~|gcqdyE$l8~7S1Fb1OO3r}PWxx|1$XA3
z_O4lOM0&A$HExuyCSTE>(pA2Ay3jj`I=_yX3Ba$DOUDs?_+nce`LBDEf5lS#M|tj#
zorL4m+AQfs`TX#qF{Nj;eYcVLCx#j~$p7x#b|fEHO
zR8dxm6XC@AMmOG5IllX{TARUd_qvtP^9}fWe-h7BfU3Vd-lQG$nup}5Z32$Da=lgD
z3sBoVYCeSPZ2DSf4n_K$uX{8{gLj^@FLEGw>q^0VA@Jn7O&$gokdHWClIINhSJlXj
zTI8SaQs0^0cc?ly`_mtB*;8w(ikzW0M7Q^eZqPgMW?ega&k|SrM{^MDx_fVp-%u5~
zV6xw7EyYPE?uc8^7x{ave15+zaM%csZ38bMJvZ}o+GRtzkT4-DvL1R5UUt-70K64?
z?#0Xbz=OLUehf+1L%n@(&RuK-ygNW~_g>&h)g`G~WPjn=m=eM(%V$Mg(a{3FKW4ew
zEWc7%`^zsq1m#~F>N)J$Tf`OAl*;d;9m7i#l$6>7r>!|Nb>iSkVUfbN=5n%6mbPsc
z@b9qgm7^LAz&XQhPJ5jU;H6IS&i$HH2oE2Riz@AebJ*yFmU?aE!h{ozH#f@yZcFcA
zwPj7E;O=#AOvju`;i}E9j{Pg~USZOdb1U0nTom0yis~wa-?i}?VahUL&(?;a>RrDG
z7L~Qi^;>XWIbNhb5_mFkL|My2z{L^X%L-Czfw%e(NIeQXsn_XYcV(1+F7srE>!`op
zW4&uw3%s}b6RI!3zOdm-_q9d8Hoo*{${y_sK_j(8`s{f!!F)mdm`Rp6r_5RBT~Lnm
zM$-@4*PhCRc3WDX3r;8#^v0!(H=9}^7&j7d-GL_`uNdWNgmK&A
zwQX7-jK6ii{DGIyuZ78K?F>QHt$uFp0leJ#TIAV6wDXJApZuNBvpLnLkrmSW1^wuF
z8Fb5s<~DbqVO)!2{o5iwU$hZ1OSCW5BRJazSKBTk`(_WhU}>|(#sxUt
zEql0M3-nu^QJrlW?8u)qD}->NscN$m!E+G@OH^L=R3Gnyt~5;y{Ro^LrSSL_^s0{D
z>wN&YS*>cV%5~`HeQSC>0#vh~!x-8p3T`-sQC#&{^N;TcpUW0DsneD%mg12@fxz44<<%Exq
zxed5VeDZxc!+EZ+=w$-V=gkfDN&aYh&my7&UQJ{BUiD0`F_ivk)%Z>%FG@VX_AkZA
zj?G>{t~_B;D&cjP=w;@V{-b59E3_YW`GxB!@cZo4HKu_dcX<<@;LC(ZK>($4(A6|2g>An1U6F>p`RsD4q&)E3=9P=Zqo^n5#Ke-R{CoA&(n~~e^wHLV
zW5s>OSW_0ia4bq-eyEYbts1g(u<>?#s{h@qyep&^Qdt7Tf_HZI^kF{U%lETkb`-oL
zAOn9E%kjc{L@y}a&{mnrH9K?v{=Ag?J@ril(12eVz`}WX9mf}I3uiO^b%vZK9nEKY
zT^4dbf0h9&oI7IHv$)JJUygTI<@5gQsT;=hZ+i}7|FQ^yF#E^T6IO-32o%nwEdWvYOJ;yD1?173*xBxFq`EF`C&hsz1f2f;E$tR?>QYuV8`aqd_GM1!OJ^G!v(pU
zSe_A30P%|hxZZI?-e09|)bV)t$$roPe!4Xc0L~M-e4IC$s1B=Jc<;FFd6xeXzB(iQD-L|V8SW@#dLx_7
z=klF^U$EolH(pNSOy2(&r}KNTs2|gqUQ?$Ls1WK~eqFeBuH7Yr?+{7u%T`x3uZ^#j8>(<-k?Z>;Q4MJf0V+I0^pz1S<2?JE~<&t?5#P_NOM^t8NptbGK%Q4hb^
z(SCLM=PVr2;NOjA?WKI(o_q)*xw&_iR}|zE*0;(c-eT}z1f=o(;1^p@;)mWphYeCD
zBzydsL-^2rLUmdpXyTw<+bDiD^wUZurTU8{FB%2Kc@+QKIqoOCl0j1>Urxa
zzS-#h{OuH%3Jq_2J`{Qv>rIOPz;C}UJYcmGdg-o8!zk`}Hrq5X2fXX}<4K)CPqeuC
z@B#d9%%Gd&wvt}a>j8u(&uy!4J_kOiVZq7=p!==ULdF4~#)j{@N`Bqirr}{@=v{Ow
z$oB{Tb$2HxYFDQ-X**g1HyxY2x@!*VUuKnZJP@>3+VPiazy}T9Jzev%gIpMHq5rHN
z@j<&%|2i4`>Zf}rwr)jvdoFf&d5CzP>fNmYo*jY9F1~4&avgrse(4BB)L%O-zkCSr
zMl+|iLpDNR(>TMnKo{}V^1b`ry5l>_qBEl#0iWswOuIM>?-EiT7#hU_FQ1Joz3r}n
z-x)F6otfcZDdfFszGoohn|A8CJVk#U(b2td0(iB{{loU_<-l(@oL+4nP$|rr;j#Sj
z2;kxct+Vty%Z2$9)eNrz&wV$2_~--L^|Jr8c@F5mc6(bN8UdW#Y~dl}nr^@yL)#Bj
z`GWJk{^W-5P=2GM`HQIk3~FU>ye`83Zkc8?3V2#O%fPfX`YqYw<(sz9U)*TK(9f{H
zhkXANv$RSf=6J(#-oU$R3X^k{@ZMs~5ZkS*-UGL;b2v2)_HMb;q%gZ(r4U(q*j@?t
z6eXBu`nCj~9{u+60+jdXeg7&S)T8Ll-_+WZp^rO?@
zOU0|eS6xu-w*vZxW8S~6gxyzLb$zQ1`CR9>)eE3!b|F)jaE{`zcQ0drC%>3oe%}W5
zuD-Lk$N~C0$+2%9=4yJ9|cUcFPJIq=>GkCAOOp%<67Y{o9=&7W4i
zljiN5r#ib|0mrU%p6K5ZdK!-d1{=f9aQCd?*O0!(^2Tnt6UWDP^{Aj@uT7N&&
zjqAFCKcTH{`vf#J&(=BTOXq<1Ie$f+-W}y>RY#HD<%Ds+
zsXX)1Ri=bfZ#2u%@&+BVDETA3^9erL`yIta8yn4ZBHS7KF*BXYJ+2WPM);-Oq^1Eax1Sn*})8@Y8X}_8Gv0701$BVK91LXqd5-QZpZ!SG4Xf0lBOSLdW*{_0E+ih=TR)`fAWsw
z$R09|bL`HYV(CRc0pp=}MUVU6^HC~{x99|SiG0UZeHj17$&}HbpDA!E6Z-gZy?&h`
zMt+5+fX$FsDZ+h%_gCO}CG!Q`EBFmIynYLycH;X!ggW%k_2K-1t7scZZ!WJn&N&y#?R&U^*Q?)__xtDSREA3w_VRJ9UB$<9RCkWogk65DyelCJ2rSBk
zFKJW}ezze1J)1|}OFlAwvmQMx#k)AqRduHGw?L!ZYH!{7^@5V8`Vetj%DBPpv
zk7w~Zhd_s$q~CexdQ2poSCdAv{LX1xnck|vMQlAZNnV48QrLe-bq0XMe&wZ(jsVUX
z+1rD}-BT}d(SY~2ZyZK2y}RS^iy1OuN!qOLw4azXO?RVxDm)+w
zKN*63%s7F)pIo>$;s^2lOPf{E{x@~f%t7-(s~aY>{ioxePBv7|k)+W>Nv~CH>nPG|
zWUXyQ?-u8T|9V07O_H01liqPfho{6xYFbUA`odc6SQZGm>ZxnVB%kXN)R*wknw2|9~>i`);jNVljI
zzVvscv#_riz$9LDdi5EkT2-LMb$w9!AkRn&0LJWD4c#eX&Sm7;|)HG3-s
zyPP%2Dr+#_jn*A(`55D4ap6P%no6Nz(^wO$@(N*X(=M9vhy!i-5;oIPu}XNI6kuDd
zRVgT>xqUhW9F<@^I{!Z6Yh!xv)hsRpez<(_NDbmuF)C>)o}(&+?;|{9<_UOL+3JvA
zQ9HzyLLSE4$i#br$uBM#Azqnv__30MX1QS7GOwu;;&zW3J}f>n2)NZQd02al*D|B+
z4tdbO@v6&a1KDT6H*nPfhaKg@lEDQndo0DdY4-1QElaH9^m+Bq5r;f+@b>XX8lQ!p
z{>KOU1(gYf2a}aDsw;#ZLO;9dXRto|@9Dd!J8<@9OAUA6mjt^5R_B4Ulx%l5P{X{;
zYdWm;8H`J`%FDIWK=1heRx?KW@iv}|I-yQ6yKu4VZs
z-^2XLS$sGZxY;UfT6#3b>!-|7{RH&)HhI5wJoM^61ZR8qL4RvlDW;-6C5IzPyP@B|
zY)V)-=E1F*wJ%ICA8RxQPm8k0yULksEVg$A4h!iuQMCfNZmMDCaOhob@@cKYQ9%BLsB0OYu;`IbrMG*%H37ZqY2A-d~r#Q`2&Xe744|_w;>W(ekJ#w2p6@
zma=!=W1bB#AidWGQE%r#ufv^lN)$hTv}l<#t>=*R;xR-O-F}~?eIY9GZp$#(?{VVy
z@M~zF`F2y!E=a$Ba!NRrvq$ghO0uhQ-YunHu($OZ=gjBO?^kkEi_SCmh8d^Pd1Qg%
z{?~MlI_n!Tmh3%MCC4X0`2Cse)okA$GuVsyn-wEdSlp~+$*mJ~U){0Mq>W@xXeM1~
zBK}oD2YTGQcF*rYe2ig(E_A;6aIp0Q;y+(lkH`ehmxd?U{@hpN6Fe=}r{<6hdcV@u
zXk;gftCc2x(IdJ2%~3z8e`0RCs*ycCTP=C&1AeJzqd2lthl6qrVZ3%}w+!
zCB29Z)d)<Vjc_zT(ec
zKn}pGM~xYtD(N?R1z|A86^k0{m-tIoujJNp2al~)-@Dlq=
zHDDj~26s2EqVoRW1V#MsLmX#?1Vpp^E3OY<^5iFVuSw76++7yG=({D0$zQdlfE&);
z1rr&s-J098ncOnU>-&P^H5XMA#s|lKV)Vfx0!~;D@35gX_&s<~Yx6jV()`u6>HpKM}}|{K{8&z1x$}t>7DY^Y)K^I*IAs8O-&X1oHCY
zB6)e+x{qb)q3@S4>ekYMQ5&PlZ2m0k?TwDbI@kZs#_O$zJc-I5xj>!OGd^b<(|a1Z
zpXqt4Mif(eqUIMiFUH>)!_w2XyR!aTf3PXjdm2{G^q%h%bg2AJed$X<3SV&y(H
zp)VKlu4BPnhMR5;T91xI9A<78rsok!Kcof!=X9OvWvYK=<(Yqo$3VjWPMOE%{g3=w
z+#~}x#xH10`kS-5=hM8jG;F(q*3s}4r`W#ZJ7ZE?(px+I+E~(a^|fh9>rVTe&1yOL
z;3>HWDcxh^sXR(=nmgbj(O2_s9j5X$P8J+50$=KTA%^sVBE64O`djxXP2zL+wqfVb
zo#B7H>AMrljEDqQpN|g$zbL2B0Du+iWuyHxDlgisK^pB}w{lHp)BWAfxgkGkzsoom
z5X$gG|NCoX*bg`V8f`=S!s>xzM$>(STdrJ*_Lr~?lgtUnY3)Az{1@~Z>8qVM4?khs
z+onY<^se1)ewFZ2zwL3sgo|`N<4fuLm1)^WryIZz`KOF5x&}V@LQ--I=pD;dMhDA;
zmNrq}wt?<+-`gz_dduEAgOvFV_icf2PSL(ez+i>g2B
zAO3umhc&3rOCe%6;yGEpGZQ8tj&aFwOHCy7x4xNE*b(~OAE$c01fS99-aHlH4~OWn
zUD6Q0n7gCf{qiHgxsMl2Q3F2O^saqH1Efd0d(Xat`g;u0`F;%V;Oetgt~9qs{BCIW
zjIq$0shrwoVLfn;rTfRPsDH<WD)*co~>fAug0ZHCksA^lRT_Shj$_
zJoNorur2VE|Gv{ZH1xp#h##GW??_fX$(!vB|Ma`n@*qw4d9&ilzx>DG{J+>arziSh
zzRWu`9)8U6(!MD=h*K#mR6Xhre3D~#A>{C@O5w`r$GQtpZpnAYr-6`v9;ZM46uwtU
zGB~#97worvTK(KiRw;~K+P2d?#9*Pb@J!8jRIbnpuFT$&3`yU&%05FA%~%-_^?gt66ozH
z2+S;m-mjfM0x3Qj(&X}N@|WB1>*~_F?XT8o9=%J;nWT_F>-)L(WNB8k(
z&rI({@zDUQV{<6}a(7PQdeZBd@Zh8k`0=V+&b$WITzTdyt)q)t-mZiPytP`Kr*-=C
z=xpbO;HOOswj_J)x?W#PdYvu~ok-$RHn
zGS_*)1IUCj7bUiTE$*Ji;yx|SY@$eS!=B%4Khj^VdX)4o$2!4*0%tUT!u;9Fx;?ej
z?$^(fZB=W;|GL;VKAq&n7CSQU1XvCFJjrgu4yxj
z=PTNjPxTCMVZhGE%86r{-pG;k125RO9w6L>e6<(XF+Y*fVmksfn4f*|tQhi0tD)GP!kgj%#ei`?G{3RVF|By^!g(7~{hDAy0WcY#3d)
zAb&t-t~ag`FYm@--fufQtY!Lx=OOF>z50{f@4w2T54W%$&Y7|Fse`UCy5X=c8;9ty
z7A)OLm&eT}Rs_8v{m^yqnZMk2^C1ExSQl**nckXG{#{jybt2Pq2})tOeDBb&OfRZ`
zKc;7*7SH-`Mgye+%CGpg)hOx@<&C>py-N?{XICJPSu10D`~AI(h`(`tY){hr;@5_i
zV?6&i!wD%S^b17TpSp#!IOpvY{EP_n(>5n1lDzn3BJ-=EVdofLy>j)$Xqx8*qvwyL
zb=d!5=tt6fmFqH<^0|J5ko0p2MSEQG_F$mbSY~{&Hwk#VPWaZQ^p9lfQgjTyuuv$ePyizS;19`#h^v
zPJurqpZkU0nP&u8UZi^^qXp%9bZ?||zUXTy{LI>Q*9;DUS2|&v
z75=hyPqQNZ30Ma;!&a?<|1i1TY2jx&uUMDgPOM>0V
z&mA7q?8q?aeKFOyQGh*}$pJsFLhp*%HHRUQmBOc-sT=#DUiq)w()Xx$^6tT|CsD8b
zpiPM*?m|dm93IdPr27O^M0_y`tHHWcJP-Ymv}Dxiu$5|`CJR|sT3R^X}S~s8Y@g|
z;|Bkl=HF<+2E-$58}wDa1HHl_4PvcaDuow%<19C!e)mf4fn%Vz_o0D)$!E+n2a}`b
zDF1+aH}m0YcrP}5{qOO3Z~rRibcZ0c_f@#AYdYRL?f0LVrvSacVoV2w|^Yd
z7I9hqr)K9d4tinljOOCLYez_0us!-We6_i23;0WgW~~R&@950Ao_6FI+WSj3U|yek({#@z
zkiYygbU+{Smp*x&-;lpNtbfjw?wbtqo)uA?Ei!KhenAuWlUL@bk>1Ypi;L)7{w7f8
ztR3iu%7h|XKjwKCa00~d;NR@{tYTQcMe>3^aSdtqf{U;HGYdgu`PmM>VE>#(%M+oM>kK$%mBh
z{d0vAv>%LB-p2f82eX%M1BClt(+$
z+5Wv>hsQ;xH@eQweJ!_7VS07H=$9oipJ#?Ly;%o2KJY4O#q@0Y&;uIars|DM@5JLY
zrl&RaQ4!@+>~((&v?!0YIqOQZzkvYXEoyfJN7wGKM52?=}HvdK)zdFIK6&^*F#>FDon68j7vql&R{JXBa-@=z%WxNnr!^Wvlr*1Uq8MF>y
z=l}2KBOFM7aZ~$pnwRQL(vOnAT%6IA;R2U=)AUGxKtSAD(o^@DGlbSnv}fNvwEp5f
zo})*Bqs#g4#ND*lVe_@^R|}nh#wtr(S_`IP=3_nBxnoox0N|K56+HH>`&+YD_4?d2I3wf
z>b9(UWNY~K3HQIXJcM)dEQ_>Kq`!PxKPL{r6}14oJIi(&y-Vm~
ze#aW=`U_2K=-lTweV&;k_>8;_)_(y#^1p*KY1SuZ;O-WW1=>#xqh
zcR34!9;~Vqy6X(w^9}LruB(p)Rn=4p_Df>NZVIXvoD;XdP{%#Rz(=l%>)+wt;K2Ez
zt*R;oo8tP~dA+^}Ig2k$%Kus|giI-Yu?=wwCD(h`?xI}*KEC7ZQLm%Tq82mO;ymrO
zw9UB-UxXojy@yP~d0nN)Zp)*1$J%ePdDI#B^?udPa}Qmi@AnJdwD17VYIy3?O!Tws
zfVH#kL;v8h`M$T%&(Ci<7JmMK`EhYu=MzPh!uw^qg#(gGg~}?2_iDXA3!3eY^fjt3
z6MPGOYkXQ(3Nj__6nT5x<8Ih+&u83c;j>cuK%=zJ!l6eRU*C?a5H_0H@8}x`|F$IA
zuf7N3_=9%E4OvzuY;oSSv{7QY;AH>y*joH9N3!McrbW1?h>hCS_%-@(ZNFcodt%Ck
zneRKl$aq~ToL=>LQ)ApOEr@cf-+=eUbNnxCSpEj@c{g23Xoh>Acb$sc4m<{2yL|W3DJ5%mKcxdhW0kVah*D~`ZWsuWA^4kTW|DNp>9N9!k1%=QY+?RJS}IQF;B)j
zmWhL7#?U+K8E<;jBw<`?n)^R5$2v-jeb%QV#^H(m*g;vy-@fILQS|#G&qs_je~$E$
zw%&d>!7uGsrd0x5a$|3FSt#uF(Y{wrcrjvv-7uPupZq?SQ@&wolbw#i&M>9o5{e(#
z<~Q@z1KpVKolWoY&+6Hg?EwEVu7w8S^|3JpNr5P*YM9^VCg3;TxwV1bdo(y4e(*l@
zer=EZPI%{1{?4mSA-{L)BE#8}qMIjB9K+Od$~Icprr-7lUPgMwZA13%S$A$?9qo%r
zYQ{||e&M@uRw=Fb3tM!8ZNPsEJhz|DEjR8~)zW^je3Z{CTJIjCR{GKV=kBM66gYx<
zo_p7c?nS@kInJhXTz=?!5^lOQq}xE!AKuwHneJ~Q9;`h`{FEMx@jwE2Ixe}2_K_!j
zdg1>pgnP^M=+$Irh+X^bRNuzaO`Z`Sl+mB<^Npt*?MCuHpOR|G&U9tv8I*q8_z9DT
zo=n0^1MClmqZp2v*=XoJ!e#HPjh9ot-(5y(5!Ezb|CIES@7EX+KdMJ@AnmtHmIV6}
zuU;D%N$Cr_J!H74anCDoRJcF6d8;q8-^mn?2>aU)eHO~94SR>9v%a3mKSh=@|GND?+y=^PGWHqcD<`%=
zU;_7BefYge^1{1JZtTI|iD;_PhfO%Iev1WyzUV%wr~Gz6heUUa=_5paI}$@aLFB!t87ODF8OJqC<1l@ZzUB&V
zZ>T?p9qFm1wXg)|bvxcJL6x^FOqD9ceWMXI9Ouih#f)BF$?N$<*GfqDoDDJ3^%Z#k
zEfnv!x;5kSc0k{hUMJq`tZzjP!aM5&eEfXGex5~UL0(iC~-Ci^Z{gE@1=^OUj!`@Fd)U{^uzvB~<
znf*~(fNRiy;t|96!b3M0?{z7Gt)ucdj`t3oA;Yka?oz{0f7lMjH>zt0%W`5Ky(9
zGg+KxdEOe9Un?>xgX+K8y}KLfXI&eaLg%v};L-Y@d00
zu?53f>&MTBf!P0xyD)q>UPBc#7je>%dqC#!Uwd-#Nha`AD}mw84JT_cgAmW|{G9be
z?(y92#7{b4h`=89ebb0z^xn|9I%{(`@G0+P4cu5URiF0P
zCO-^*lHC_a7N5%i)wxqJQ5k-6WA&EQYe#G1p8eaKqkDntY9rp(
zXr+GMz3RM}FNxfDg7jP!I}W@2PU(qF*}RxpY`R_$1zgw3%wNGu=fl>{|2Ew;}8d
zk9pr@CUBo^rvu+k;$Car1=~xNdAJAMSMs#f6>-43Z<@~oP8xh^?hH%doN2S_9&o_s3ItY_kd>rY!ErSG>DZg!4ogwyArG3u7SXPpFQsO?FhMF
z^22d?z+L6j4F<2o_jZ4j+JF3vb_}sHDX&92)5Fy>+W_}RP5fhe1h~4#rzV~ShM2e|K^09DEpr-;3BBIM5I2x5kIhBwS=~A3SFi=okM;8;U!Y_1v=YJn-b)
zwR0=lK(3^-Zd)Akt+Yr+-~sW>J+irvuw)yi1>-go_Xw!a5Yu?5fpZoNN&z4s|L>hgs6WS2O$f3H3_lHtRVO4dLU
zn9qjyHq-tWm~jOUCU8&if3WxFVL5i+-*Bm9mJ*psDH2LbQadUYMHw
zGv{$#*o)?Tz0)Ze&-qIVwfXw=@alC=@BH+ZuR9|*Tk-n&#)|J?Bxj3t^^zCDT9gx*
zyq%v=m>-xgY#k><&o&KB)oSngAF<$wf(a6V^)k06cH6vdHze+fB&Chw9y@(v+4ALuK|
z^}^FVN)P6OIJd`WBt7TwmK3G`-b?{E-n*3e_jrL=@4124Kl4BuzY}OJ+EFFxy&>Tx
zUNG*ZP%}agjT+MXO{$M8k)SU~@+)}-agJ#qM(e|LE$FyFTt*qXwS`pj(RTW3Y{Fz-)$$-vnPr4c5y;)eQhz02GD2w
zk-x)@XdX`*|1hjh^amRs1jz7L12+_i^R;m~JRaV)p?3JfFv2dccfx0CiE%
z>^2i|u#5A^>SCX`^71auUvkrt%ZDwB;dDGZTuAxe4~pe<>XtT~pI^0;)2D1hIDIRC
zF6Lt#wlv{%a&{``5B6~6bhP0vbO1cJc;<#KjC(FI2bYn&{`8Ike0^!_=|S^;`x+gN
z7J;8{cCh_g;M9F<`8<5*%ua6a-7VYjKmq50`ByPxp#N^zXhi27$K9KAKLod<7Z^XY
z>O023u{d`aWQq+ko_CGD_=@yTEE>h_RJ}6f`9cj~ex!3p|Jrr~S|feTl@peRz%v==RF@G?
z2E0ge7>{-Q@!-4syk;@`$mm>U=AC&*>A6v6yPiGh`PG{2^$-67*4F!;Mfyix&&ejV
z^WFA@${p42E4~aw_r(gnPXX?!G4v(b`7?9?CUBIqA<>4g_ML)tbYAps=~j0mU~Y}5
z1yt|;f4S|U^>Ai~Wjvqhs(QtAPV6wyl&@1wf6gBI19G3e-O61Muh+iwz-kemTP7c{
zyAp;tDOJzz+y}&So7c0js$%92Z-EFg^o$tBbpH
zk0S0bbMzX24fYu~-=Y5=_u+PLd|-)qu^>Mp)D(Ij+zs?h0zGSFBZseuM{kVXII1J=
zVcjhowsarjaK8xu0l%Tw<5ukR4v5#D?(eoY2lX_}_++{s>6=z25Ap`b
zTQnX3dmGsOR$8GQO?&BYqkY!H@^wxWpB>y3H+2l;Uq?r3p8{@tReymd?9VN;wK@d8
z`Jy(VuYgV;lJx??*Dkp@DFe7OuD{!QpkbR{3Bh>pV5RH3waxIIk2w1R{~I`;)ks=j
zX%0U+cKJ1I4W17xY&xa~qo1jjoS#qUs;?b7ZqtMRZk9A&WdJ#^PGutkQSO(on|cMK
zzP2`3UL677*K2@z0OSYWP4uJZvz~g|Z{ML_wn3vajUZn?S$#t$_-{V0)=CH8eMzP+
zjf>->SETp=+vWHi-3!b#^LCvHe$BPjT2R01lW_Yh^{bY<->;!KdtMiX)j{lEO8PZ!
zt^xmc@f;Vk6y@0WS9PVhcV5)bn&gKbA8O_VpxmRj-xB+PHqyV-Fdi5bJxQJJa};&f
z^~wc3)>65iz6ZF)IAS)9>%N`a2NA#ZCDRAw_mzWxXNV&
z=$pBBUmDP4?3z-t|J}j3dlXMMtw^6h=L;s^mF5)xFFG~&4(aI?>(HO{GV
zxhwg(LZ_2kadR5q^|@{qM)nRKVx&Xm%srHZh2k?ib;#6+&rkdJ;eZtD?563>sT}>P
z)_h)$v}wlY;UM=ZJf1#t)AIo3ufF})ZOS+G9{n&E;&R>ZeEzMrIhdcX)*m$)7j5t!
z!fneH#Gm+PGa?Rr&&qK)_sg|!TD<(+?{9ISiuWJ`l6w$Ny_|^?NIVzWifV(dHL8C+{!e>(*{>g!|x6iLmACsd7*`Usrk=iSwMI)~z`|@D)Cz2fgM?
z=|B$m%-7Xz%CVuiCvOmB!Pmb9i$8LC?amZ9z>kj~KmpF_E(h`Ta`ry@r4`(Zx$v6H
zwSR`~6zFyS;(fH&y7#ypB@g>>e*Q+djr1@PS1bFk<@xU)fP10WOCioDJywahXOkh^
zg!IKB?>IWF#N&J9e;}Pt+R@J-fqy2N0y&%`mo4Ju=e(py73bxiB5qoIU+8;KCG^K>
zHsW$4Z-{)|mWX&|Xch%*IEOkW^afrO@=dE_nMOG?JHTA<@5X=xZF*SX68_;a#QE+}
z3x4Mt0H7U(7m)ioOwjim3p?!bbb!+P7z?a@N$|_e1wAH7>>Ku+7j(xJf?r`M=vPua
z-uk$(bGwwUcZ!hnmG%XCoWPd51)4~4@ttpi*2Hd*-vH&oIF67M-l{n+lI;9r#b
zMQT17hJDg4A+M75K?af^&PucfZQr5iD--!&({KZ=$t#3=1rmK1w}46CY)8TOClYim
z30)+=nc%dV(iiR&80AF+4DXW$UFPZ9v#A33Zc~Z4N7BRgG@s9}OW()2_gmOKUiv=e
zs`Hq>A%DSq2hVSKmOO@gT?>UBqosYmjk_rC)0;M2zSkweztvf!)7RxlkJfpSpDlX8
z^)TuCHL(eTAC9kPkp5Xa#JF}kE6U%Tk%lJ4eeJwbFUJ$SPf9y@PsbT%~
zw4QDHDeUaoc0Jd#)WeCF`!M+v&%e6*>W(B|=F^&|Te^sO_>ql0ryVx5h68Ya!!9I;
zFm&iMu16V;5kEB=$M13P
zNl1M_c3xR_eGIM3W5?#KCH@cVp{aDAFQM~X%SXTyZUN{ac%J(+3I`{+e|b;Qg)k=a
zC_YV(_a~<2o~QFdx7vo>4_E7*;`Vs;KE~I>v1k8^A^A3oK5wJ)YQJ;h>!pb^K6izD
z$GeF2)JNIRf#~>-1(><Ae@>Xaf7V>|XqP
zeWza@J7*4B<78fq(ZHz-yPxcgy`!UxB
zc22o}WljOY)Odv-USA?s9H4v5NNR9X*q6$j+w*-!n+x{sA}eQv-O8
zSu5}j-M^gO_DVI1%bSlk=KdeA{)YFL!FA2%(Z2WK_`s&remnlEjHmZJYlm;<`^BmS
z%Mz))s(C}wC~njp|HOsL4bVwDL+9Da4}MvZowGXz^L9HhS7#lyzw_cgb;vK){$a(0
zH$FD$kqX@Hw=(4@;_b3>NRQ
zpT602%@6ztkL5qofeZHPYg2m--rwai@%_(y#0r7;u-{qY3kCRH)K6x+DF1Q0Vt#-0
z(J=q^kW&;
zb_4v}+2;IwO0lnkuR9ZRHL$|rxnpGjpI39EYV-MbxbFyrD40K{;xl)^`)z!Ay2}k5
z^Pt?huM_!vwvhdKj8Fm3_1>s)e#^)uoPR~FE~f+SPw;v)xU&@t4BGA5MUIP}
zedBT(?OSpEz8+mUIz8LV`EgA|T<&9s>L6c}UKEkx`v+m?xt#3>v7VmtfEw7zc2b0a
z=Wb_3+-)DTl&6Q>d(P8GO6T4^rS&#@rHF?cC12(G&hDcN7ihQRBHlh?K!F0jbMTP@
z0X!eL@{rr*tw9H`=;ta?PC`7~33|2@1umG+1{835^lL#c;(-{xvw++rx1$PTkPBW&
zffk-)(Q7qG@3l|xow3?b`hiJ;ZZcNbae%fXNFRk&lyJrw!Jpef(49}BxsdKF#XXb4
z1mDY4@W;It^cU&c^C(IqKGfcl}h=XCH-=IOkO8B
z?hfR`;Nf(dg#2*__|!|t4w)~<$-LS~=j?wVHyiYyaXwcIegB44T+!e5|C`)@OaCV6
zk?oM}u?6{%S-*iR-#Yyj(|9Wk%3qTrG79|vV(`?NcWeGT
zl*SvMK9FuF*|~x@6xJC#A0qtqlnTPVm&9(oejm2f<+!-MIUIqwaioSJ
zVZXM0xZGVk+9+c`+&75h$6KvnAmY}&34GrZxNSbSXG^&_&#Kiz(06*%FCF0?$%@)M
zj=4RDdNS;^7{=v$go=GvQ0GBh-&@<|@EpG1;&1{BGv4EVyRa6;IXPX1=TW?AqLa<<
zTV1^Ru8Q90aqHu+Lvm*C-eYCKdbh)72j#CdcvCK+g?Z!%isLrjq7xeEjS+V1bXd&a
zn_0E*1&?p8oXCx!{69udLc=27&cJ6vAb)E1`nQC4XAM&jnr;sXBy1l?4^Xl1>9VR8
z$v2Mc&f}uAyC3=f$#3i(6H2ey=*#(H$X(mmWIx6Ed#7%~Pch@U+qS1>#6NXz?r6f%
z-+b@^2ei|vj$u^Kh|zbZ6LuRArbp@Lx>mzZ8sq1n52iSr7uM>tjd01P;Ms(l0rzYP
zua+9{^7kJhi`S%q4Gs_F{i5gYxmYm~hn?#2gwj(-3`E1>
zJnL)NVZtwU8$2S6zdzfXjo$T^cgfydgbf=RU{~b=<;Oy;W=d}&5
z_`dbz=(iu$oITB@)V)w^tN*dKTfGXl+djD
z6mIAHYcz21T<38Z@7GCD*j9nQGf)dBEZE2Pp33dKnSO+~^KYxw-2WYthx8}D?R;xb
z!e0kgtt7u5yIC=Z_64~=di*B)GQP*(B7WY$x5tV0TUiuCb|$P2;OGALTU%KXe@A5v
zOtmrquX+AU-iNaih1zhYyKS6cW4}bi3HE@
z<{W%R@>{!hKTiBj`{#co?EOB9>upuM3I`wf&h~|~m_e{!z8S>lF^6&Ict1ZZy_c*g
zO~4F=^NJp?xZF0Q9h@JlyO+05zvDYO7Dc|SL;9-ky~yW@K6l&maWJb5Vyk((eeNXYu~ETdepJPX_tvUi
zr3nvl?Y_x8Uq-qO*VAY}MOxTreSn*Q59ntUP=4iSYzKf}if9GHb0i+9p`Z7r1s1;}
z^&0L5hGmHS7qdma6Ol+pdiW2RgM6)5iv2*-I5Y>+lU(5z;0+CmF!0@|rXpX{ls#Cf
zurDa3NDb!_d*OcI?aLIg;eCpcqFkNRNJe@O{(>3aHJFdL9gB}_{$`IsY1~4
z{e|5FFGA89>ppC;%@eFR^ZjvZ0XUUCcMF-dUjCauF?G3yZDm5K-dCIOIF
z8|l+9s)>HxLZE!!9!}K;{~#rUFUP~t*F?G<52*0GNy_ga$;U|YshGVe|0~IFqa-;U
zNe-KAuJHn)VHxKs7
z`s8xCJCTmT%I%j?wpY$4qwGf>9wMJC_iy&d`u>gozncE1KDm53U8d!FVKT$t-}@!G
z-n={hBVDf7f3ru{|KH?fd;Xi;e=Aq^yWC&$rTSn}#~FL
zO7gOQWd5Hh>-jTXre#0K{*=@I4dwBqEzLWNxdPlz;2ChPOVcpy9Nvu##Cfb_r?tel
zkmyJ$e+n%J|EQ-tu4vY$bUjJF1ve1us6@--pHFuv$5E2kkmd_oMW}qaUh;V1(=_zD
zN&L1hq8xdCxwBm4>oJrJ#CsOm-8o*C%8y1=N_wA5>%n`zal`(%Lg;D4I}G0IljdD{
z|1bd2GU;1bpBzT_$OX+m6AOB@T9p?Gf%Q31&#IO6$x5(#?S3xOJRtOnZ9
zPV6h5JP`Y+1BNsO;XS;8OJF(fU!5V~exoKza{X
z$k*lvMEL#IQrh9-J==A4IcBtIiGuJxfS$OIu-f_tk9$tn4dij_9)nkoB(MKtKi5~S
z&o&r{_{(?%&mU{vO$DbC5;AfY-TRL-}23{h}$!*BpCq}p4-&;
zXhJw}&lFS?@r%7R*K>Ifj`LBj;(RE#({i9U-%nL*SO*4U-BoYN<=@vBff0r0`RgmW
zodefvVamhzoAw)U+TL1=m$&-VF9n?cPq}n3`n5_1}*Z@0aTbZ07cSoK}hk!+A-6FWx?v3fy_R
zuBI2yS9LHGLmBH-BR^cMKt1bq;`dF8U!Uf7DmROM*waG1#}`)ef|ozDV_S>_+?z_L
zfC>Aw1ESxAZ=(^5ec2)UB{%E~=VQAARP_zya!=;c$i{m+`>zTEK0xEp}
z>Zo`>@xf~GeZDRi#C!8)P83-n-lyv}z#jd@_#6C1yq}n7|CZbFPOS#lWBiRSa^N|n
zf#}E0FNymszwJYL`nkU2xE>1)(VuT?i~eY7I)K|{b*1-~v2yy*QL6o>j4tq|>Yl>gv2_Q{zdo=?MVQsNKXhiu6Asf`#v
zzi*0hpOqx|9%gV88_AuC5Loy^LDq;+y9RTq3MzL4WSs8Y~pC4MB4;Bzy8*`yYF
zC*kyi&|1oePR!+awU93V-r)p#?16lJ2!MMj8~8dBJyfELq;vU)EvCaX%;5r?*9*b
z?go@|MUs>A$?=4&SEjib^vL;STCN9gZs?QaW10Ua%K7Db{mGYU-ptU4$s2!v$J^YU
zpk=)%LQc?b6@
zjbDEZ!hPN<+izbia36eB>mxBn_P5r4;(fQ4TQj{Krs3S+MC$Hn+;5FZNptRp
zd#0N;tu)MV&sA%dPP2}dqcm$JO>oi1L&ec*jgTbhOA{_;;NTZ@x;A8y$~
zldjo;WlVRUNzrxVGL|*?Sp%bYXz#M+&ck-$J)6cu*WH8MvqiI)Ej?YxPWL+U%i~Hh
zekbn8x~uVc-=*G=kBw3O@{L>Ond4q=WdDHv>uo-;$Z_=&x=ld*R+P7}`V%}ijtY-E
zaueqx`Az1wX-mI%oO|<3s*+8+YyQlqqL`hi_I}Gk+?O3WxN}Yj>bo@M&6TIX)w;U%
ze6JL;C&hh!y5RjnGovWK*DLV;*wEwK_uzXQubo~iN2lW+bL!J&j^m4&PuzFAscd3R9KYVjF
z{8#Numb=RboOf$jhgM&Ud$tdkH&F{JVKo=;oVIIBF>6$+`@6ZakW~hF<>+M;uo|D+
zx4k~Fm>q4I@=K9WgmZ@H&1ZivU@M28?Y%cn#k~6V9_-d!#g3ZZ9hI@9h@Gutd~4R!
z0+yPx{lck{O6Glf=72?|`K+{4(|N@;m8__L(c!PlRm|
z*7bH)vIkH4n=SHEvJFGGq<@(8nl1FT$g(k0u`QkT?&_XYvdkKTBifD0XC3cs^f0WQ
z&&tz`BJ=JQFz>tzJ9SwROSMi(-?*y;?~SHrxG6Ay6{Ln;#ru}eemjRvY^`GP_cmzM
zE>^M2;bSk|u2aNHrW-wI@w1o>3{ceQaHE7Zh-ztW3On~&Sa&?2t7Lawk}l2nRWj#0
zgH!cvRqSBYzHe)9s95*tOAg}(D4A2u`QcG>lx+3&wtDe>lq@va-1ODGeD?MC{p*vq
z=CeWeThz6$=CjO~J9g>J&SxKNGxBrK=d)C`v20nal9~2j)z$j9l7+aw&C!CL=NreX
zba5Yi&)IH=B0>;vKhS8||02G(*1zm<+wLVS>V5Q4lWX{W#){0#Z~K(6hSj#84)rT$
z>CxdnS!0S=mz(FT?qw9Rm`d%_`gc?;wqk&5?+q%J@!NEIk6J2LEAO!9*+oj$Wz~mP
zj^0Yvz_#^5i@1Civex`u*zSD%E~DWBr^oqh(pj7K!}OJG;_)E!t{=RfgZM~8ObQ!&u)rX$i9i2RK^6~rM6JM{jEnyQ!eA|(Iq?m2{)HU&V
zWDy(FT0La%w*vNH-TsON-3k~x)l+BL)k4;4ON-4N%?nvGC;Pi!^a|J+?R%f>nyK);
zzH@W3Qk3|8lBO+eVCUk1#(5PMuw&4PIWM)8tl`9F`_eR(Ec>Kq##KEfb4?j?-l~(5
z%~K=?+Bz!PBC}4ryH8ZIfwB2D9?U@dbbY_H+7u=Iu6WOkynJ@;r~iYDbNP(5pIN>+
zFrPUcd0yw{$b2^aQ)L_TpnNvxr&ZE*8+_NkW9JSt;tSbIovv#SbSz=spF&nnID+%t
zU0=2?!F-_KGAe7tMI|%L9d7tB5`H%5`|R8LLS}W&Qp;?2G4qQ3>{y2H=Ql4HUfH*y
zl3DteHmidD{<}8mMmZ?)dw4qninLYi*q4aHzME96?Q6eg7lM>5Xi4SrQ4#3x$){Ho
zyD3?pPGvXpe&jRV!a)|H9h7Y2*0^>7GxOOd=QAB@jLv5k8ui;>+ndi)vJY<7dZJ{X
zHY}O)sGo}2&RCpMhIzqat%k398qTw}*rY}1VLrLM|CjCQ61HRYj#*bSO4y?PHMW0Q
zP{MljsyDi2X))9N<#8$5pqTkLwODmCtB76Cb!@mir-&Jk?p^7Rao=rOwTRxcRBUFO
zu-vXIRcv+Q+>bM_s#uP`Zq~U+Dt1l%m-|yACCl6T@ZkoGkC}saOc}1BV&T41m8U#Z
zOnI!L{K6;|+chPl-(q7GJ26Oibf_slqUjsa(i#2Cv31iU4H~N0ftrTKyPK$3jf|?Z
zm#>$wy+`bu2K$$=cRgR+PH`?_%Qb3voP1WoW=>ZWwll~1uVdEO@^?k7bDqwj6TwAH
zD>mSGVC^CnaiRYGiyMmA^9R%Jo9p6z?#b$d3T~;`Gw(ys?jKh%k5cC)%X+EUh0-dI
zkr8jY2g1JZHhpCwFlnAZwThYD+XVMyHsMurX(AC*n#S~q&C%%Q1
zOG5q1^Ac{WHHF{LEbmg>O~vlE*x2fJnUXa*8m9IkT*(>@Uv)!ajCgU`lGGI=O4!ge
z{nCDoikOeh{5i?^Zjr+=ukL3uRqUK^*TbLcVcu1{*vq9}5i=OrDxcXEuxryY(w9tC
zv95-(dxor2v9G(m_Q&X}*z}ZN`a9n#S^ce(R+;OmSYq}i-5l^|mR2pebymeZL-K6h
zTB_KDjNew2JCtFAszE?PK$x5U0-3`rLXJfjd`$1p>bx@_61C(*bx#nyO8NyW%^{+DQ1>0
zb9cntFJ=kn>owZ4NX43lF3~)4P{ls#B%d-KRls`tZ)(+MW+9vZxxGe0vm$mSX;J>m
z%PLmw$=+RKb5(4OUyITHS_MpP;i_&s@x3PRVR>DjEiYicPE9(li&wE`8#dkDqfjxm
z1p}Kkb;oy{%<~(hoG)Nn{<}Lr{8GTm4_)ync&%g=z0AEkqg~2QHh-69p<)qT4Z?rn
zW32(PjyIDYt60*mz?Fv{7PIDEPkyf)TFhpj%)fJbdJ$VV_TD9Bry};Ven5qtZxPE}
zetcKQRz>V$a;AFrQ>@ojHaCqhFU)F{Hv3XRA$zvBciFFQMa=bPmq}e#6tMOEZWO)4
zx)f#+oESL0km=@AEWcr8=P4Bh?9#~x^PEi!+0SoUx7Ah@vH_-j
z`quwh$Pz~%3*6SLh>dPJyCO7O#e8;;*4bcPz`ozOFnI{p%{}G)<{aE!$aaRjjQpS~
zWX0QbjK2-R`l}I@v>e|nsG3~zq}_4+ZijBulUwmUhTn~g&pYA!0g7*qdn;{=*uafV
zSIt{p#F}LcS@H6EF-w^;YI$`vJSTHI*q|73SI)%Gx?M&Uv9yld69h)3%vEF?Yy-`Y}(hgs_=Hj>_GClxrrN!S?-Ef
zC;QwgW?Mg-Y`9!p$c+0xFx4F{52dZ~ywa^XkS(V2!&uO|PtnbCOV+*F^`)2#*?c8dC=lBgBnwB9RYM|>^
zd$0rU!w(7QRC%qKt+!a$cbyU9fS=8+gYrw*jY&u5wZnYhD(`-^OuUb^VQ1gu0mqA(
zOpDDER_cfauU1#&NIvRj*%b)SD2_wA}BZ><4;>ARI=*!kc&
z{;lG1qqgJxZ&0Vtsw});JI>dd;XT2&DHZ|0@%z)`8*X$D9far4v2V`x$tcBp;J2Tg
zUsc95cl7_EhToA{HaWAy$LXJ0TAHoe%HUE~GXLcDVPnd0Pv7Olq!nfCeR|uUdGkwI
z!IklPo@Yy$f!*iet+mS7P_|`TYP&LKblTG4caJhw&8N=YwBDsGEUQiU#bKpb*BjP0
zT3N~}m2Zw&zXsj*@Yv-w%hPuanPo}WB`J&O-o~vh&e(tKoHsiBMr;||
z-=NT@`^7R=c&&8M)1)${)@o|~n75^DQQOx}g#$~OS;~WyX5GtJlE$FScOhkLn){nC
z-smo`|!^b-dmS3*XYg>;WNru`^iNiUgyeK
z%*-#B23M9b@1!@=KfNzyH@3zM(Ckpgb_`HXE{!TH@#(GCeEft|Ww|D%d|F#KlsayBq??A#%*%2~3`>eKFnN|`b-b=tm$W$aAp
zQ}?fv%UK8io_Y;hRIm+h9h1)5SFq?FXO-&{%9)eB$!4RA<;-A!x$f9o13lSXS<@F)OGz)&P)t86x8Th!Iq^=sxij5f+g5BA9i{P>U}1z>SI|s
zTdGxFv)Y|$?1YJmilp0lY|3hY-X54EBzk0=dpZesXyjhJDb_-&n1*FtF3k3
zm*te;d!F&S_6;%qKeu{S`#k2SiPsJCy5Jtf%UG}9!|}d)?3B>O*r$Ejb@R3}-n0E!
zV_mmAl(S}K;;V=E%bC%s#PByS%h{0z_KSC;UyVIE@_C(J
zuVAGnmuo9w|9t;-TQ8%&=_3q_KY3Iz?Hrf=?#Ig59fQ6zL++NbdlR(B;A0}pp}3`f
z(33K@)UtuixWqE1qudp;y)K=ns@FEHfqVS3o|hF>l(2Ornw9zWaqsn=@x5Fdd|&XS
zZ_{;{mwQf+cl5@-xmM+fNo}yMys$ZE*MAS{X|!Va?O?p0J^s^(jSl7P0E^lb?}_$)
zb8yf)mvYvBM271}wQ}a^uTksk(=uj$IPU#K?+VuJhyJweVHGUn+V<9K)>W|33z3I<
z+^Jv|x!a92U!Wa_E*s%{2>tKYs|#rvW!QgA${&dSXqywRmRDTHQkuk?Ev#S8LSIaQyo9xr99CtvzK07h;rBNm8tv)|d?RHYz+8(>Q@etZ`gt^O-i4`pDO2>trS68sD^>00S^Rtwt_o$QC
z=n{TsA>~Dtn-%UC>)iF$I$y#R%euZ-+g-x;Sd17nvql9Q`0-u#zD4D1Mg6IB)ti^$
zyW{hdSO3PjZBUbu&g=0UA*WsJ@2-{X`SIYnh1QkqPTkJkwd^ZdSC6*Y>7^B{i`9(I
zw@y~DJwL}599UVwTpN{yS^@j*xc^jbX9a7#TQAb`9_q2OiF1;&f)!r5Gemn~1q;9C
z<+6Tz1yl4cH@n-bf}PTN_-ezua;E85USf8ooVky8-q2ue37hluo==PQCG5b-y9YM!
zEn!1Wq*M%iLF41A*B2eEuO;FB#j*I#-Iq^MS-tQ*8S8+M+ZXZt-_>eZ^{KdrXF0~a
zpD)hu^-a#0pQ>Or-?~0a99zNs($kKmry)z#x7@;eVqUFvIP+EhhYEZb
zDQx%z%}O@>dG6U3tdi+>G@BMXt&;h+n%=}Wypj#Rl)LUjRyq3_RaLML^OS#GyX;k?gI((MN`RX%q?y`GOPPTa&+q%l&Tb``cEs<`4X>E={jwPu@0cQgdu2^SSc-
zc-Z#}_A_f!ggWN8JsovY=fMAsCS0xmeN8!A=eOH#0>0OpRmV3w>ts3W)2!y$6?PS@
z&!p>xUIWlCFSb2sV_nKxn2pG?%fkDt>&~n`IIM&v)F{*Z6js6}zqzr~`#}j)55E&%
ze>v6*oi1YzLtO(*K0ekm(zH2bGR51fxX)o<_%=N0ZLfzI|el`NZmaeQxC$*exz
zUv?V)e!0Epvk$+^*`h;Qw@S*&Sp9xuhBidMZuTm(?)n22%w*rPBfa-ku)DWPops^g
zI%?^C6Z)00U6JpVr_t}a`6oJg1yr#9qv5uD*=npOSE@nDv{6VT(
z)B|zi=qTX-{1U~=r!GUK5-hfW@_Rth@+ZSZ8fXE!~wIK7E*~{t(v$|Ttm{Qp?;Yj#NqGFH4f50x{?;+
z{k8s0KpnN#oTI@tLDy9ODFJGFs3e^z3k_;&vn`(Hg?9b`=|NkC;{%3z#EED%W)YSen-)j1O$)EoN|KoxG@xcFh
z;D0>upFPl{EBz^$wS*i%%h*Lqm*s5#MXtA$ZYiPtzsSku$Y}M4oJ`B*$$Dfom*izR
zi$Cb@65aU^y>hy&*G1yX_Hq*SZZD<5{a`7)YGa%V7gNJ00~=4*ib@i?nnjmlW0>3y(Ns1P}Z|eqQ6SzY>?=-QhCeT
zQi6gVHWYZ0H;jVWw-fYX#Pd?Y-SmZzzP_Wo6
zp?`ZMHK>9O@S~YT!G?5*zA}d-cVLV|+3>UcuJQ+J=G@%)I4bE?+ewmZv`|^5t~qwMZDQ!0%49hXD#UH(kN`
zTK41^1yimNlLm%pIUg&#+1H)oSFgQ
z$PdT4oX-{u^Z*5WyTz;061cK-7F`^yA#|+`L-x|?Q
zdaK;IJ=6aZepEgea+e$UayiX1VSm@d*WnNaE7aBD=|4t`cGtFg$;(ltc*8LY_Hlm`
zuJ1(LZI0WWsDmlk)iJ$Lpn?T;iG%@YFQ3LU2?Itt@O-Mh3%Pus3-7tUk*}yjE7+w<
z(QhW^3VTm@3j0PWArAi8QNuVsydwIO#~aaZ<~xOb=bDQ2jN~6&?`{LpKQ=cP`3+|Z
z{b>fG-%YD`itBs4sD#VwB{*=}dmhbv7`K&;I5vOy8W|MqjuFfUYCk=}b7W&^v^?zGu6Swo$XKgNjm@l{r)?rvF6VypZSB0m|iYvp~z`8%lN#
z;tCXuW^XEoMiHSLe{mA@zu70J%lTx#$>si;Urv|XOXkZc>-)2waz0s3u7~VrSzh+D
zjC{Dje%a5moLr79C!?G$%gN=*DD!2M>n-zTl=*+^`LiB!|Ks5a?3CM0*7K*l>^B+Z
z{Ia}^vYtQVKbbH02U@-V+3qr5Mp;g7SGhmR{*vY7_TbYe>LJ_5hY4u8o#cFS`k(%h
z%ahY(KguYV_b1BrknNNCvj1d%$a>^*WWJ1YzCZ1g?U(b(^_AxtxqMkapUzNkc|Mct
z{il6$dGdIZ^UL*@`Eq`_pUAXqhisqR&-t{9@@0GFejwYko-0=1IS&2j@A*>JE63Z3
zk{n-dAoqR&!zGmM8p#btd?A&eBk_YI{jn15!7Hd>
z&A34dX38C{V11?5U<0KYW9tV7`1{DwvM6Uf$swZp>qEglp(XE^m31f)wmUxsVH(MHN!8e=t?oJfYDA>EzEjh;Q
z5^+?r-z!c}=n%{0jg0Pc+BknCr#n2&K?N1e>2VA^u3+=8w&m%TcIUX9_96sTaW0#2(W3@A0=G4^o=W^7@jJo^HEx|WCLFoBG51%{hhTIF1xs7G6g^$R;+EXRMhgAl
zd^B&@FFnlAArU7Z_vLa2ENbv}=|84n6s13RTF>>g^DN?6wbIjx(wDz;OS1w7^>ycd
zQ@)syO?u}PcW+Jnb=LxK5Umlbd%+ACFkrwg6X3Yz!s}-jhGfb&!-Mm$6`thnb!<+C0j1A4z8nKY!JfKb=l%0h>(4x&^>0_c@91GN
zPlN0YkLcHzjNlzCb8g
zI$t0Z?DAM7BYoULF+cSiEa(M%VZuC%(MS9l=q-fm1t<=vNA3aE&;tN_j}e&lU8Fk)
z)#2$!y@kKxXG#1p!PhDi^`A7UA(!(Xh3*DD^L&N=tj5AGkyS!Inx3n@apw@H+8_OSHBm=Rn(3q^~44a3D?pgp&MYiM9`*j0$$Hq2QZHcD$qJ
zLV6pFQo<{9(CvXUX9)S?^Fn{41%kh}hQI>^kfYv0D986)9cZdGr8kxA`GZYcRh
zwv&_K%XZ4hvm+gYm*?lpFlbKx1LbXYSUz5k6b%gc7j
z<;vyA{+0WW>^C{?mCN}Pd2{|h)I%I9xgW@U8Rhnn^~ipd(`Dr0
z9?FyZx9lIe99f@So=nSfGA*O*4>|vzap2|Ctf>rT?(9b3QWnNJQ`!0>^j@(cM%Lx(reI&UU34coZ0{H+^uzoQ@ZUc8X
z&MElBfb%ZiVX==PHtgNCh!iMTl7XNzxnmWq8lRxxFFwGrAN8gJus$6XazWYx4aNyH
zl;Zh5%c+Md*vm0wIMy|9p?{*4$Um{Ukazu}#rYv`1mD_Mr0eSo47*AWfPKyKA%M89
zjgaeGDfn&;1$}3#(0isItr!ZnQ(E8WO6z;|Pl9hC)&HvgDXwp^p3w81mP6zlzedQ#
zy{8scu-fMbAVId^wBV7Io6G6<
zv7O)u#2H&$xm?`UP_Fl4W7@f)-6JeH9llQ=6~?(^K|Wd<>({HRa3s#-eBwC2u)!?O
ze>lVvQ-y-f3hCXSu=l#2@VtU8>#-k06>)0w>pbqz$~?y7&D?S0aU`i=V;g5-X~O=<
zz@io5rL0+AgdMIo;c?_&*L2mpfbQ+=k0FKmG@(xe!bNsrlL(*oG8sd9(&p(kqd2?A
zW?g^p)t*P(bb=Jyp-Yv7L)pD+X!Z2B!%^cV#@zj65%LZ2f2
zZ=~l^VR+L{z(oJkeE(qI;L#!CADc3vfcR&(>ZMXVd#HE2G8f>CZ=pCkMI8L#L1WVY
zq+xJLN1$i-JtGGK^(Hwx&IC618fn)9IQ&X<1nFzO#(NXhZ-l|v`4rb%KQ``X2lO>C
z!h;G0D@tk_M40@&-XvS#@Q61LO@UuqIULjlzPOi)tvup7wXq|Jzpt64C&k-d>K#oT
zfUVMpu~6WZ)MZzFfJb|=wGDw^eLDOe4U9P#p5qB@wC|&*y`_TP&bs7x5ExZw+#Dsa
z`0MXzbt}YAzYR1uTc90&nAl(GqF@W^MW?nyzS;d99}@nW5@b*H*73IwCjXRbx5*;^
zoEmkkiu~^M+a{_fFr%9K4zl;tp5S+s?(k{!c=Chp>dL>!{uM`$?&|-YP8C;YKvSd$#sm0h;mfa)JSM{&~$m>4_Rko*>}d^X(`sL=e-_$hFJoyHjh
z*xz?k_)e13k6c|935>RHwEi^oTo3-v``e(Bt-Z;P)EqxPP6iGP#SDmgx`$LF9QpF$
zH^LLy$+yV<@T_^5kr02)S>ev*`j_y2{WZkwE7`4P`voWe*vCfy(k1>GKi&F-lQfKn
z5>E4Mh)@jsx&C?B`72m`{fv1ezs)5DGY+0hG%T(nJx%JyUeE&GoKc7GCv#htb|kr)
z@>CPAJrjgv`l|{UspY8KG|Jc@992Ykn!EwNtpVi49EU29!$?Jwq
z=lNeRRO9o)UmCG|-1{F?ZY}a!F$-AW5CM<1~8#66v!Wp&5WrrT9KJM5Kq28Awm6fo1@n_CPjZ*vw
zc{I8wXzP!HzP&*3^D_i(yF$>l)M!HmzqX*r3FlM$1V2_z14AVF
z;CMPf*$`v)YJ8ZkUzU!=r5=4
z0r>;eY{0XUozavG8jZ-2Hv?$u4U}##mD5I|-6WLbHu4h5jhFn;N21aBc{;Tc@%1IN
zrRD{lC6(V)O6TDa(pO9LiUMJ0Js!|v-IVm_OMXj{{3yqLoQ?ckEwF>6N9N1*qR~U;
zbNCM^*NZpXKlJ@6&)p6_HxEeGl3w0yKweGY8>yZ{B|h&qNSF0Dk<#UTkF1v`Lr&H!
z*GH!1ddhM#^6v7Fe6n9;zKn7_&6^kUvYf16-e;`g8UFT{vqa1F<6$N8x0329`%kWq
z+;5gh`F2U|#=~9Yll|9E;$P+lV4aiNS?-sECBC+Vc@i$=4T^I?iQX;cm-p#6rEzqp
zF$rMam-;iE#uAN1hU0AB;T5c{)X#oN<1R-Um#*9)h|3xY`44%d0OLdAkCnz@sziN65oLj49xR<6~ftWFhbLTBN_-NQNrd!ECB9;`~p7
zUyH(C(6K%ey-vuv=1|YVJiA!r*KZ-xYaJyAAkK3kgP})P@MlMo!x3ja0|4h;6X~P&
zkOQ$#*yPR8Y>vpcyra;gZc7cSU@1oAIIJUY5pDt}8#d$gm8-(uk=MS7aNyxjqpkml}HGV#Tu){|e
zu3xb+g6D6LDel4ORC|G`4*ScdwiUE)XF6NM0eJp!LW!Y@bs?uZS{Uo?DfWZ-_45*Q
zC|=t+FB4rJL){S2ahf{ju)6trQr(dST
z5Ne(9m}&#OU%l-MqGyyEV1-0GWWL%#^xh29jYMBiZ{LRE?#3~{aYYSs?=6>8oD}q7
z+BJ%&Zk2k&0oa$X@zr(*x*e#lOL4x-t?wzM-?m%ZLzF-CXHa@SVB&=yz19QGo?dmL
zIM+UALp;S>YlEk~rSt9fv+lI$58U&@VPgVt{`ymGDeit5lXr>Yx{MzOorqo%k~*RR
zFwxNC#a6_fOK(RsGX_1;);W^m)^B?XzqSJ2|6*@75x8}WPgF0&t^PjY-$#MYY`osH
z5pZ+l@TyQ?Bqk$JgW^%o87v4ZREdnOUv;ua|PS2
z*p)H99ri=V%l&haZ|~ulZVKS5>8&;wK)(3ttGAC3-xs>n7&I3+B7TkI3-DimsMEDN
z^o-vc=l>k(<3}Gpe-84WHyqrvxx0cb>~{0kZtz1lH*UTF{CNWg8@=oXKb{!fqAAiV
zhB#iIZi;j9Rm1LG?125zOovG$k+0*1!MiuY?$9R>6w^&`o}lq$*vH=3kL;^ZHaEh4
zrPsHKNdw`(MWKT-zCL1I+PJ6rLa$!E1LG!NLcACIHvB#87@+LZ`U0?Vq0a6=pl7Q0
zhwZ?rRs|EUz@EG9rZ?#ZJE|JZv|EDoGS`JZ>8MwzbKcpj!0J8cM&3gD&fBiZtzeI(
zM-R>I*xy8+>~S>^`OaJIasCBKfKS6$J?H1I}G%qu?{=-ozo?G`Ta
zM}hBC+b;Nsa?_8@QZvCLoBGSl%sfLCOw(!PjC+mYSC^0O+WOcZopY{p*8_2EQ1X>!
z$oIO&Q5V`rdag^jN&BD&{olQ#eNc;Ej-UDh=O6#Db{DXF`*qH}fmQk!^JpKK*Ls&Z
z_1ijzvw}!IJ$#-0M(DezHGa--U|RCIQPj`X^JMm^?Q*2a9+R22YgkR&@I$&TMhdX+Se81^V7?{&*1+jo1=ARLoT;^=a%n)?^oV9
zHWhO3Lwq`t{DhEjEJ-+*O-sN;hjnz>rAQiYQ^p>^15pKQeQH-6jmz`YY`?cZqS$TOzh5aG+u2>zD%ZkO`9ljaDn|rsOwPT$Aul{=M1qHcRM(P)>3w!
zNPM5kLw?eD&)jg^mT04rzBoyOzwX|fMdKo(>zX-zfEFib)T41cHcxvy?ZXy&m#-lI
zgjr4J^Vj4~NBR8tZF4e5i!3qEulD%J>53XVv7{^5UzL$qIpLok2l)I}I+qp_jJtr{
ze4dM1=;unbb@P~Zgxa1}JpVF}5YFGOX@>;|^UC#HKEHa|8^;oTqCpTs0mvs_!vci<
z)l;t*p?~GAP4qnQh=n&+aIBBRvamzIJZI6lkm~v6n$~QhO@C5kiTU+V6OP_*C<0Tk
z$A>#{)Z7CAeK`l)NMKQTYffuP&!yL0qX-i5=};^mBxiaL3nuzsGSmZq+=4m4et0G=DZ?~DZ!^MKA1Uax`;CwThn2EvXXuMTtnBrUg$Ah~|C7V-LRYu|y(yLM~D
z*A@Se)4V*FePX?}r{xUtsUMIXSUMH}eA!gsoGepL?+y_3#U%oDX!(SkcB-K3HFoE;
z#%Zyh|Hu&hcBcftmvo+Y{-Yt6>mP&ei2OtIU>@*i8)3(hK4Sgr;q`GGX%iOoXZK2gX$
zp<_evTTA5(rex4OyaV#$fNiDpMkEdzgOwwBi)d~hP;5y4-Y@WGLOS(QlH=|HQdC8B
z9(oBOcRO&6q<=9bgQi|excH^8^8z&&Xbf(S!z8~vlW49EeCov{$Hjpj63x3CXcvh_
zP=>$1pQB;0ar%_hPO_atC>e5k5}uGS%ZC1g{hH*z^-_Lki7u1uonAx8@$21?Zz0)z
znL7}E<_=acgD0Y%a=k}N`YuZPtPpikc_-=$l%L0?N_v(_{csMiFy@7uLM~IX^HN7a
z$McOO;wLRZPiQX6{rXOz(L9k}|B9gJKPLlljCYNckHNg#
zOw2Q%rEzROL&#M<6mo7`guWr=R3WTuj|D%FKajvW@l^2dt`hv*;nX8B50;SQa9-4y
z8XWVZ0W}oPyH(WUn5V+rI3CQP9UkJ$@D?09Z1UmwD{3bUz`J!_p>SyC
z5a+>NCl4ch^?u(J!rb*AJPFI27h>e&yf)ul5zD_dN
zZcK6&TN?BxOgLbHA%i&bVfGWkUZ(@b)4JxReUPu$52h&}QGTy=SM6y%e`ZzriQ=1%
zy3gj)`MGVQhj{jlxTf}p8g#DMWa6P)l>ez)|0k6G>q1ar7GYIl0qGCCSl5;GTq>Mw
zNp_zdI^>KS@NLkyqho+60}UC)Rkd6k;wi45yZCgop1`!LZ7#P3?)O)slsDrjmeLBz#
z^H9bB=b6yAZpsVuS&*xHtB3QH{6}n;&b7Uv&}-hYz|LL|&smJTqiZ8B9(;Ua&zddK
zfArKJu6yx_En6MhR?F-OYj3|Mw^@J0>&I=C^GGt>4a(#7}lQlUxqG*1TbTvq>0_SErfBVE>@AJ>zH}
z*l&Dy)q#4DA7*B7dK>C#I?hi$-yP%k?!2hH6$*CpehY2dAJi^zJa+>7jaOqfPE2>g
zdBD-_FT2doVh?T)-!vo@?OWh*e47v6H`=`BU|fBaw{z0no1Taxy@wwsJKi^#QRo4`
zGz~GHPyNi!@&aBpLOVt{9;bd-W75q*)L&ilHHOo3nUij}9NPm8Rv+`DaiRWFH=OvV
zOKiti0@F7fzfW
z9<W3TS!
z<6ToXA1e~#*tT!1X#bqLY2EPB5=BT#3q^|-qOv8WMV2-tt)z%Z
zA`zuUDM@KpD)f7uxu56yey;a*{rczCnK}2&neCj{%stO7oYvomJMX5`I`7R|bA$S4
z<;4b9Y3PO9jEz15{C!UdS9Wj?9G~tt23YEoDo6e5Gjkl6AKo)|q!%rWH384
zKhS|1-b2~|fD#kAc+)#tg~{2SQ7nEL3rR8k?{yrn-Lr>_Poa69OiyZOycj*-UiFQ|
z1>Y4CtRE6zCo+4AB(E~Pzj0*uIK6eWoj|%NH5V9lh(o7j3>V}TWeAJFk~%JZy(EVN
zS90Yk_V;4#6dt#V>CGEKgbDlbJH+5%5C0zCP%R-Q`#9ZZm~xWqujDOU|4J?3_{&r3
zUOe|3(Ty;I=mhWtdWG-U4g~$WpGzN_NI%WS0%
z!#=5G7Q^;&;tbD-TC;WXaHj>A|3QXfQ@Rhs%G-Pc?hUfxd#AC4
zsvYYC344WQUlMM8x6_>Psr)f?49;cd(~cD5JP_9)fN}BHJ`J6dL!)lH%gcvC1n*-{bkK8}+~{TUl(45ci~J?;=0mye4cJog4WN
zDmG>OaqqKfs))me*Pqo=xpRei&ndoNJ!?`FweK&2_{&pmg*RB18H#la>@PVDRyt!{*
z$r`%dNoJjy!d6{Lrb(*&G^KL7W8-5{TJ$Ad+za!_w-vU
zKD?H9^``9XX6ZWK>Z%*LH)U`>YdET)=4CC#k+IjO%shRZidM1viF$>+`0T
zx3+ZKTgCo$y!@_xMjl!4ySrXRmgpDrj`R%}6JcG>b6&Fj@E)%k-p)hKozh*kyuI5t
z51U$#_*O5W_0t1hL(i<=l&lk&H@Zp*-#no|JvCUb{B13-W>4#gAn^Mt)vn1ZQ?Va-wR>6#
z#;I@FaarQ~1%pmzI-CSQRll4agL&1Pq3&P=zJB&-s_Ad+Gt4B0SkXEd()XM~(9SyE
zANPto!_LCbY_9Yh-yiQ$j+k9-GR8jeWAWMR;P+3w{6DXQ{((`4T+RUFJqBydf!zgK
zuU65zG_$g|QU|&Q4@t-a?&>g^cLI6|qVnf?kdrenKOw(U@P2XiaNyNbjjy%PuGMAw
zCxjqZukIRe1B|qhYbKum!T0PLdav{BXl<(^_Iq;bmQ!MZqvoD-yn*y3?LBLV$BuW~
z?MmnGQ+j^A6^FeS;LBFk>DU$XQ=}+467XzUu*Lgi~DCDPy{&tYR9C&F;Ir%FO
zv6=VCzon0Rz<6z%%@C@0X2;!stC91{yt9S
z)$bepjrOaBGDH3m&sRUAoJjdrs=j^6zosV{9H)JnMAI)lQ^>Wcf7ep_)FYEq==qHi
z;zQ_p>)f0cap=8oNLow%VE*8>0_p4a&R$7;{X(5CZf>C6tAt&t{I?@RM-UpQ^&${s
zetB{6(75k!7Oy`3wP0~!iMs-b8uo=RwOIUk^V^)osk^_IvpAP4m%!Q|o%61j+BwNK
z63hf~sH>Tk^Dy}Y7z_48Z~91*|L%~FktBK5-ac}KmxgNF6TU8X9Y|PZeALnyxGB(d
z1-0|2O3x80cW&jE{gfYk(x!{#O3x%_$4bK^Ebe;xB3vSU%GJpXk5qGf;Y!|aCd;-L
zFceGR;=7b6$2U_BBi!P-7n9^!Kb&+r%&<$B+Yfwp<>G%pnkCaSiS3dl{R`uXpyK|X
zi!{^ktme*#jD}8Na?^x#Fe32g%8!iSzjvS?reWM46tI44T3EsKwzQxbNcW@bTfj#<
z0l@gd9B#lafn>v791lvGipL?pYva=0D!J#C(2JDbW~j5=<$=rC1h~xUNg?}IELmj7gQk&D;zJlY4c4AzAJ7jX{fgPM)vM?v_ay4Lf
ztP0`uZlB@OFA8(*RC~hpV~Ptmt~;L0Vfxb4JhbD}T@Gvc=c=D~a(d$$Fl~|FyBOOo
zpwPad3|r!`?S`B&o4W^acvLpyQx`mOtcG;M-x3V5bm7kIB*l@wyCpD|c|?
z^>?i;OLy0O%k-Y>C^FtT#O4vx3+w8F0a!16Y}h<>O+T`V^yLuZWn?b5bTXugn`DsXSNdC=LgE=Fh8IgAIbRVv*n|iUR>JZMKs?JH|C;a
z(2r41nLT5-4`c0%d^c(>rJwoy!+`wUklUtCB!?Vc!s<(#m4^k1^R~W*_Jq^U6=V`0
zeDJ~_H}$~NB=4~IlCu;(DH1>2Raw57{CL4s^*O{VwPS;@f?((7)mXUTF=HN_CI6B5
zT)2UF;kbhA5AMKCosTC|xwCKCPpABgq3$n9|53X42lA6$tHrM!0_p`0b|ij!p|s0`
z>P!CqK%L6(H`VYw@lnsY)I(gZ+qP(bCzZ1~%Z~2nR>w|_35GmS^u3rQ$~h++x^dYEocq~$
zIys9%_Dtx>uQIEQj;9_=$a}e!ar>@bARi91R<5p!X$1
z#Wo7hRra|T5|8sZ@AFb4cj2lTd6!Ub?jGmao4_B|r{$_%M}B-@ex?oh&{6B@m8)d2&zWrBkq-WvW2X1#
zFUoEIa;<%KMJ@000>!ZEYjwOJ_sx4+WDxhP^lxs#d0@@O4Ud+>?uSL!@811a%L`5a
zwL_5{-
zGEaEe%s7g;@pyi^-axeLgP-V=8F-(XWz_Kx{N=gPb$sI&zXMS|lUG)PeL%knv2m!k
zT_|d74S2=6w8$SFb8vq`uj|Zr@a?FR%HyOTgMX`JdGOGWd+f(F%79njSX(dM-v#s3
zw)>CKry8E)Fx$0i2XN0NXUnv;;BQKE=ZSqrJI=>n)N!^%yz8-z4)(6*l{M)(WT9Uy
zN@H_IZAF}z+txbwPc3ii=GNqR@c6?wg4;jA-hh(RW)bMOx*uiTpD;rn}qc7wqowo_usY?3@4i=A$x<^B89{%Yo2)YF!#X(-U#o&qv!7
z{BW~{_Rh!Pi?;D0KK>ZTpSG)e!k||_Z*7mvV$?G=T{Q^fy8SYKKn%UGamE(&p(j!s
z@z@{rD-271sJ008&HVIoHS9F0pA=5_V_Ke$Qz(R$_hi^
zn-5)AseMbnBz9CoFL8Io92cb94iHTm2^_AUYf0;~SkB|B81f$}oR_9^{0;uO(@3v$
z{-6Z%8z(k+Z6|wNr#vnWME;ES_NP06(v$jlx*^@;h1nAF8#60%zEQdUE7v#90O~s`
z+f(@;ZQ2pZa4wm>H-`M$t~wWX-(l{Ydc3j2`Z^+dh5Sp$wY|6qj``H>lS_J22R1UE
zWVqx*4Dn#EpHeT0PtKS2okV(9s_tGPy+^h4nE!qxmEM>3y|PAihsX~tJv^C5dK2zm
zO(VV2dN(!_?#?{#MSd)8fy_eUjR_~8hmzj&PLBh5;dK5;7-|CheDV)D}GNLd+lK3A-s~aQ&kqji|6`@EN1BSC4`Mfk5E509$ozN^S+H-J=#;y
zP0&yG#q;H>@sgnV)uqb`%86}|eg2lM<0Q(l*@+4fOT0W5c%>GeT2gzS0uT{EE{w>+z
zX9w{a?e>micc%H^s#D~w9UnY>=>uk>R&s)dnWCxw;1e~l?PhZcfObh
z?7y~;_z(EiPNiS(iJxd{n;sm4`147ru5k(S;|E-tH5XVS{%8%o2XuSapuGq2LuGPV
zyAkl-jM}&iJlEI%)iELDcddw$dJb8^;e2m4;=r;oiZ{neWBi_#)~`f4ZE+JUT%i~4
zaNRBidWoU+CMJkWx#h3BPf1{$CU*MiB7I`9*Pk`e8@Ft5k3Qt00W;bf@%-11lD{Yp
zJ#snXrH1oDXM5Sz<#-S38~a){195IrWQp);oQvt1MvotJ6+FflEO9P!Cy8l@ZBvhin#M9ZCD9-
z*SlS{{@FQ?dEb3U-0U|5;}m%PcguuM2;=_X({?|@;>!)Z>-Rs!
zDlSC3(wAI%fP0i{J-^A0PpswbDAS&lx3G?2uvpuC&nbJSzzT9x|pX)Ui|
z+QemX-w=nt8}3qHU(buVam`Fa_94$&$#lTVgV;BSj+b1rpbqa5KlNl@tmmCx5PLIj
zOD(VT;0~{?{VI5xn|)tvx>WOc&-R5a$9rP!hg%ccqws#o_m8}d5aL&>zS~;xE2*#X
zPCI)ldGG9cb9^t?@^ba#H~Ht_z4Y)6H@k4Ic53&|@#^4_rrttB0=mn1%lD>j3SC#j
zi;6b5pYinh4
zw&5IZwy?O@_%iU2T{$y;q!sasWrnCXb(iu+nk;dsGOy+Z-ZXf)0-
z)aTzYosalABUr-^=S0UJ?VaQao^@uuu+;+0JF|}}&wVj3wg!gp?Lm7CUyU4Ug!kU-
z`BBXBSj>lwujb96^S!syss-NQWh=Wp
zRaQfx$
zZ(6;b-gl2QS{b_;?Rm0qwS6)4R4p~Wl7AU~e7Pp^i`Iw^czE!wo0I(PVQ1dDok9+f
zZ)LueB|j5Ay3&{Y$jFTIZp4SSuIdQRKtH5xZq(=td*`<}2WKNaw7t4o6==~C{g%$(
zt;&aIlD!Xg16{^LKPjWo;49L97Cu=}54~c;hu>yGZcxeFLw<0;itrYy$8ptzlT_~j
z?8uUtfqk0E|_
zZrS}CR9^K4Cl+_~#MMNo{h2AB+=PJnZ?5zbZ`-9A+=tR{9!tS56|jyyca0=_tP*V_
z>35xt)6DD1j}Gs*_#N@L;sk?aI+vThRrLY&YI|T@>)a17Sbx_O6vz(|*^gnN&ilP1jBf0*$hzwyZ|
zo}JyLqD|?ka=mWUPOrL|Lr4xUFjgcCklA5PC=&LSOZPmuW;*1+t!Ap!&Z%p+U#EVN
zNf{JLdTr{9z%+2*QUu3yz>kxfm_0sglo)=AKo~^2hYQD_a>rj_diF_KEPeRYkqle-
z-z&*2L70YqV1OjkTc$pdwR2n+cW$z*p5qn|+5Mlq2+
zFXYgVJ^!^*=nL7cyHOFu3~^wQ2kYnc`MQiB>{z^?>4mvq`XJxqB)S7AvS}g1vfJUT
zyg|O&te+PZaCXeREX(pk@>VdMwv*$Xmism{y?%?h_UEqP^xltL$@F%{4`JoWpE|_&
z$Y-2#Qhm`pd4})0mN86RMh8^5H#X)uvs2p$We2n(vM_OvB#@P
zPwYh|(~HfqWc7`Lmylk{
zgkx<~&WGGODZ)u2x5SC>49>W5H43Lr-zjzh+#hP1ro+U_MHE+He?SFlZ
zrx(~m4m?mg=>p0L9vFO3X$;;!W@z)s&^RVzY~i6%%OMv+|1>lp=b1V+YT-857%X*PItgR8=pwj;GympyCGVy*V|kBmi;b=G1!yCsCe@
z|FUBJ@3p)QmI0e`qw9D{zXLK)>ecbShg_9ZPs4cU_zbp^LBEuqewqP0h99+FM)7iO
zq0r%)H5ivBktX{K@av0T8W@elIdJUM^Aw*P7V2HMx5j+fVemft?L*$)6Qha(6Tp`X
zR=4S+{;l^uU7xRn_;q#v&tdj>zu?^G^0}uqyodUp(;~npUN2c5{~75L!;_B=L;b~{
zH%bSCHx%U!Kj{3ehSz*|adlb09{xY@!;ikD@%-#S}h_wqzutrX-_
z7SU%dfT5?hOH!Oyy0XZa_8X6i{lsrVFWJm)D7`nU+f(|>4(TJGj`~ac9eP%piYa
zEBSLL>7{&^W9QD%`=iw;Uhez4T$22cS>gkBFKu$+%5d@v+dLFS2pz_|$SpGLJ8j0w
z&^~RAkbxe>^)8R+*l^=MYpPlt=^YFiZ%ydzm;RCZ(IV{~0wB%_9CQLmPdMmNJmm-3
zPIDl1ZP3Jl1pIk*I*XV8Zck=@KkTUaM9S}L63P5~=>s=*KGf&XZ*@w4Qk}@+NQTI<
z2_%2?ox$`xHLBiGd9(KK8A0iPkNdKCBDTei=}l9*#Nt}hxvAZxr!rGlh4f-8RhXXb
zl#5KyI`kcbVS7
zNK6;xcZPBGu9cHz`BglBhCdp*SpMT&S%xaVxpRuXuH1Ra`ivz^Z_{PEF^2uXjBdsw
ze)=V`c-atZih{vocKNb+*{(1=kmTY!>w>92WQ)(RbA^RBinwyNjbi-quh}%Fm%2W_
zo96SE=pHI8CvhqGX5ecL6rmhJaK;?Z4<<4P_>#t(<9!vYg=EH`CkJpf%
zzL!!2>8aUfu>Ahs@@)JAF4nPmxn<5w5F_x!u=i~KEpVFRPw~=e-IW$fzy9ro5*L5uR@_;rxK?Gf8;
z3BNXFxY5obxGym%{6Nodq+c_UUws#PDlrx=#~>eFeJ<)5Fj3lPV-4ck68ZhhB=MYo
z_87<8(0irPbL9l`_k~Q~q=57{z7MICNZb$bh_zRe_l`n5lWQ0{
zdJN)-taqcRDEza;k;x}Q5P!|1Uv#Y4R>$*9I3U~)^>0yAGvi?&zq&VXmpk}{hwiM@
zH+}IQV1LTq5aW8@qvOSn$wP6UAnAe0d%O=iQE3%c3wyF{cCS0xA8~8q=-J21>Uo(;
zBKXs`xFacw0{lnh3fEu35fHD&d(VBeG%@Zh+bOy9PfuVxo5cEgr5Fh=bC**
z)ja)Eey#zKCl<~1ulB$^IC*fa74&vAy&9;6@_psj)zFsA3JY%dPhBNL$1+
z)1Y&{h?lb(2W96VUZ!+j&G*50NQ`GXqpH;!_HzhNLTm4$o-f`XBI$R(X`6+FoRQvC
z?WsobyuYUxHc~i09(K|35%ffr+s@E=LiBe_b*gvYcdI>gzi7?q(?YbK``SjY>V@9#
zjeEWggnYQ~MlW;VI|CO>S}*#KpLNqde!9WKR}?P~k6k#A;-#|R9_C;4v&Y1ff6)1-
z>Oh#TVe3bFqAzD(C%y88YnPJ0`n#`rKjpV5kA!2#_?Zt~MEu~S(eAgT=O5)^M|$1L
zZO16T^oRLe!qE7~_et->b2VksGk0|MBfTmm{X>*5!+Xo(!Hmb(E|Q*^rs*$gm%C*5
zA<|oL=GAV}%d_p2Cye;Gn#IdGRZ7YvKL|Bv@ltv~<^+;odfaF6=G25RA(GXq_9?RG
znvO8NdHeSleQAmsUrgujwK8r^GH
zJbxd(fR*?53U}Y3=%^WsQ!6I)XL=2rC$Mw$gB{W=jNAPTQ|!;}yOmOg
zvFD~HE@fym-kQ}XH|PnAmobf;UdQdREMNZ@_g=zW7TY+KGd*((i@yUKkq!9=g<4*<101>hR@#0ey$h!>gnD;4jC(YS$=82zaxc*o0g$@#K)d1*wmA3b;MA(jhpp$|S&1~mBS#rRiadS4dLf8>~0Q2NUl
z_cX%jl|=j3
z;4!)1DPEosS-OYvhtC_ci{`_b0p1HpZ`F@4_O!k)Jrj9B?`?IWuh^5`&CB#fTkssq
zy{sQ3+?8NB{8!9^DvI~}SH5EXaYgn5<0C_tkHvuFUhBTgY@REH>ayR{^<=kVf+0?=
z{=)caoLdg-x2BCreJFjKmrpGD4Q2CyZaSY8!oeT;ov8K?dli9oD*9y#@RL)LdUM0!
zuO)*b@##RssY6?m?196@&OYsipK;pf&xB40^=p-I+ak2`&
zoC5zcZ^}Z_+aXdihU^g9;jKFh=-p?nb}ex8UG3-uRow6EcstP?emrmP@5Mu>;yk7#
zYe@#~eY&q`8s1n0f4jpt*bVu|tO`G+QXEPseJhRk73rGS4XkAG`*7)X(7@b$}djv>A-`$j*MYg%)9c#bpnxnAc=+>sy9@L|Vq;5Q@r{6^@7
zgk`N+f%1C`pO%#&zONWpsoj8dolS%O`9e?mj)g=3`fX{9LDy*b<5f-9+b4o&G_}^H
z&Gf6|^>Z&v9Y-e(D&EW73+-rUh8>(U3|9A(E
zy_Z{XmnT$qWL^54YF_i`)Zrsz9`XYEFJ3)2yO`(dT{(2}-V)xr)_-zxP4{_k3r)7v
z;(bu^Y3=kk=P-|+i!La?aF>@>YT=%jSIPSrIIr&Y_FCSIMN@v!Q}D|;-O*bBmJSEjr&f^&R(v?jOyMEWGG}@m>E%>t7+}=epNE#Zj0q
zFAavJ*&_eZ1&ar-QJ$D*>-h}K!|~oL&v&Dq>9$9Py#mkKUp{TV6ZAL6E!biQl#ltl
zWgP6(`rd2w5cO<5Jm&6kl=E<_@|rN@$BkS4$`0v%QeK_pxBkvq7ghxx)Gf58bUx(u
z9?#~w1OGU;+Y(=iF}EorzG5AGN7@fKwP-^bt@p;4D=Nq@HukMoB>b+wTq+fIs(Pqq
zQM?|RP!LZ3#bEtz6Eom-{9Wsjz}|v?<2M6y{okcEpnTz=r}o4vqLa^;(7OL*>TgSW
z#wQNl{R+FK?e3`bhur+9HHpqa_RCwHq;rr>&K?S6$Lbq_JL*u*0%=2C;xjr9N@;}Q
zpCio3@8n0g8It@b?1~NfryZ_0kC9(?&Fd>m{!ytSsUKm-mCQ;)--gL<5tGWP_1?jcLC$y5U7%@^`DwK45-*@c7Z}9CeUHWd-SJewG%c
z^No35`VJ#KmFlnisT?`6?pVq{m~s^X82;^MQv})hEXyQ>x@oIakA&a+dH_F&PcSv+a4COzvX=l75Abb|%1AUL)VY0Pd
z4vV`IDWM2>h$}gNS$w|w=`+)tTl9qKRisBVzM(bsHzGd#iInGP(szAe%Hq0mZxbs=
z+f|b3Ki-$Z#j&cxEd7sI97~ruJCnuNqk*zaE`M>Km8(%o1OwyqLz`hoe}p^8clh7^
zxMEn%>^UoJ&eF#n#pB3VU4!C*%Rgb857c&B!SrU0TFm6mVI031JR04B{DD#2^QD&{
zLLN-Xzz0NsfHRh(`+#y0V;J824RayC_XPm&JmcgITe(55(NybNG7&@Qea1;U3hx7I6Rx-)A$8op~td!&cTOwdsPu7l1
z_754%aQ|gFhVt1j7|$^=<@zUd6X*f-pP#g0SU!<{aEExhE`{}5O&wS6`Tbn`9QeOi
zD?C4w>A(KN^>;`ZL_Ft#(>%gcE}Z_TU~c@Dk0nEKPR8etk&&Ff#A#0d4IQ%}-&Tji
zlLJ;bkSu*C1H=*MPV3KuxPhN2U1xmj$7Bav;sphtMezXccSqesh47>C57_*?3T
zztbN)i_%jg(^-3twUn`X&e$ffc2?hh$WT7taW?5$ow>^Lw+$S`aQ+%+h9#@8ys7+_
zRjeEt{@+Cr*}?VSjCY`G$iMcohLz{OMV{F+&irK@`J2)Xk<;YI6l`;ah`-dEmZZ@7
zA9BFtXC3gucPGAx52t+L_BO^N
zYeaWrs^MP3#5-6h;7MOg*O4FZY*Y7&K%DTgy-$tK&KoIPGcX
zPukBuuWP(y0-U2fL*5Pg{r9NKQ@fvDQ29*#m^#pKHT9$N#W@4YGW@aoWdUQS73WeMW@$OF2|YY@NWB^M2iKwK?+8=P<#=Rb-Lep9>GAbn^-
ztO5AsK7A$kcj$+)!s~^+q5nSimeVBYw>wxUNP`C)vh_Lg8ToDnf8s-6$H<%(`B$*}
zRI8HI6tpw7Zn}yS&MEw(E8MBS9d-{|LGi8fg^f0~%l)mCsS}=8UzJ^;q=|m3FfvWQ
zTgQt?n{YY%E`EQtsyec-`~#j%UhniP4)=NLM}9mzXY+t}=|#z`!<*m-i*j3^oCWV#
zF?;azzqPy@Rx8Bn6KZ%PWXleZn(%-(?A7|mDX{l^><^nxMew;(i_TvJk2$|n|ILp?
z^jpg5iHld@o{ms+VI$71>i)gr-NE^b%L3&?y0SRO3#t7*8Rs%Oe|8m2kgenW8X23j
z4}3iQpnpFR9?rj1T^7n<+ymY@Wmw=jjeR1Mc2vO6+AeBx)2re6jS{NQ#D
zR(ln%)*$=X>J<-o3-8BEuN+*1`v`XiX-Pihi7fe;ur92Ir~Ehfv{Yve@8^f9(St4V
z`@9>xqZLiHyjfx9eSVIt<()@q+#eDcXZm->IoL^-p5B+`z<9{dUg^`3E{QvR$mH+I)pJe)M
zoQ3h)?ej=J8{@1eC4B1?c;bX5wZF{fRr6jCtDov~y_R>I2Rk6y*NPu^G)_<
zURjL=%9ZvrS&92q=Q=a3v)s`?9aqA=P~N}~Pm~;RF1c9m)C_fFqz|Z7Hl%&Y<;>%%
z829-30|pqN|5mM)-trmxNo&{cF2wvXwd(q}9`j*B%K*iJuxHBd8SVbaZ<&{}CKdUC
zeeB=$fgHW_-Q{_(bFb2f_oCpN5nGG96_CF)Qmbbh=GVI6wrW2i@Ac99I01HPoh#gM
zA9j3iy%qBqe7B~LRWsGA(tqKVF+e-jXMQ~x*MIkfmC4`ORtAoki03^Q+3o6qtWy7I
z(gVm>qwCZt-F($A>2l!ZOS88vNB+e0Axp_WWJo4>6P9>83w;L~RHeic{|su|xrO-N
zvEP$F(t9nL-gIMH*D_-}%Ri!=;mb3}4uC8q{VJFE(c!+9bI8ve&H0c{_Jk~Yc9PD0
zXI#&me;xXJwqI#5g>1R=b`Gtt2;cRoRB!uz;T_wMFS5)xUKa9uf3a)y{%zcyN_+tb
zsY*45y#_L(lgSz2!l2adG4mvSVz+#E)dh
z$r-LgDLuSkt2E((`BIEej+tw3L%dfiRCYPpvDZ(wV%{4_uRn8hpWd$~++vRtJE>7_O#u{bc(+?nNPnWI@q|H3=Xp39w-Zpu|&Y&->>~<^%l+C#p0}0l`qquF=7IHejL6IPkP1LpsPU5C;klO4c@SP
zo6)Ei>7`#dy+Kzn9FSw|Ml!S;%Kbjk;U;(9Gc5t(68X=bYcT8>fo_2O*zF9n<6a|%
z3-Ueo!7Q9Z#=;!P)3VJNIvL8cdawI0VeWYkUWo%Klxbb!nISafotco{hVHn0+((quEL(Pr(wnO
zd;GZkH%*+p^9-lgNS`%8{v`q+el*W8kQsva8uYn6q{p>$XpLDwG7hg8$}j`5?{nhh
z;pQC13H0bO=vNa^`8pit@yF3Pn|{cL^W+jD((qG(SiZoLj#?%^Y`f2RyWR`XKcw%m
zpbkYG;_oA_Z28CZi@(~l^y^*J!`R2RP=~_L9KbRH_8&(Exx16@-*Iy>`j`b0UlyGZa2Pth`2K~AeZ9cG8JO0QE>AN23%Tfn~(|hMe&ADk?p?_=C&)^rp
z=w=!GMRxGq^qDeSkw3oQHN$;KzuTM|)B`y%VpbvXlPjaPPP`1>^~3v0
zca3jXMQtU0@9A3;4nvkcsM%i{*&~Jzyw|AhhgYXKgSR|O8?`xWhk{X;>8>tn(v<_j6@272AjrL9EP2rUknbC8G-vd
zT84+q@IHO>_BOS4@KMtrOCR0?Km6(cE6*AAb$0{}-huMMjQd}w_md(Sr-f&OhkBn4
zy)_l>n*I6g6FKNjZ=Z5M82B(JCDaPIDr0Yf0_sbZsqK))b7#a4>MjNj+GMqb&MS*{
z9`RoZyx$u&h|cLxxb--tL4VJ&M?y`&4KJL^D}iyQjW1(>mTUcd+<{TCZ3k$*X`bBr
zN!0~@Nv(fSH_pZT2Te8AK!46{nq=z$-fOjU&c$cg@5Cm~=%q-}IHCv7jnPW08;X7`n;Mio6Y0hqq9@RKruLpp!>;eBW+&{4g5(gd;1R
zY+-M%iJGqy%Hf?+@uc4e{*a3Dpz^AY@A8)bR@~~p-WuuNty4ykKTw-ebYLEEyHnR`
zdQW5;GWjO?o76$lT{It7^sznd2pnkkQ;6n^#gTj7#7BP#Uz4Qutn8JMMeD|)d2NRn
zN{T=EkRQghaPhTgK~{FmQ>9p`*D7$f~rGmZm*%Gc@-)A{Pt
z4Ecag2?~BGTTMF!B7-AVS+k1*tT@>?Be?lYFN=JBn~~gv4vAU#vhah2+nLDJnF6Mim?1
zkbKwYuqa`i#A7fB%%@o=S=?Xw`Jf%iuNL*n5aPZL0y^f!W&;*K`v+KofxsV3Yn3AG
zNjwGSf^mPlgZb6LQwvzU^xBw+fQS9XfMu%*+a6tG<@B4veXnA$&|xN@s4-#sO)JV#
z0Q~p}Q5M(B?i+wGpx>@2v2xTr>{#3ma{tTX{rO|lz--|E9w{-qj`%)haem5lBBI!z
z`OIYcgA9wAU7)OVa#yb67IVb(%HgHe(;#nOZ~{z`>Pz?jpt?N+cONcLHLEd
z8^3}lEW#)e>MY}MpOy`iJ?MHK(tqRhj!-!l01T4gzOOI)r481H`*1)&4{RN(*Z4XeN-|y^o4&=a4
zF1;d^77E@MTnJ-&>!eDV{Agbk48wW*VoTJBxTeLmZ@?0&80UZIjTzcM;m+?r7jyDE
zzWqlobLGz+#M$#kjus4f=U&e5N2puAthX+w{7>T_=%;ly8wG7Oc5!!e@$5}F$+ek`Bug+E~heiG@ADT?=>lRjYJ
zz;kU+vvPi?Mq+7UKilNZ;*H4Uy)3`==QEc7wjqGkvuRm1lUpEK4^#YBQ#;U?JHBOZNG)S)=J
zB3I!X$vn^GO=N#eO1rTwu+`&mWiYU8rIQ`S)i<7b@Sxy{Cq!*XkC(MfnVwgy^Cz$DC6?AJ`~;OTJQnInUlia2urH9R!VvF?47A{=ToV0#MiFR~x|k8*VKNy)#49rN@dY~|{=Rl1?VnC$
zMNA~#()L;ifgJrc)-{CcN&WN(o)+uz_wrz8pv8`LA+$ew`e&RC^-tZszKN87!uq`%
z$&cmzHqw6Tgl#ouCU{KFvp^~@?~O?b$r~dg?-TD?T>9WH?WdkaT>bJC`-Bi1vl3<4
zedfrdblUgH-M&~p06g@=J%b=IVA}}OMEZ`#0!ghYKY+XRq?^c&lk4s^&jCKFF;k-P
zQ|hrlNp{)|b$vwT^JZk#QM=B0%-KUcu^|7(=`p}{hNp^008I=IU!eWsn(4}Oi66d5
z(>!7Xl+Vu*BCKz%#hXd&_dfXDqIQ^$knYg|E)|Jo@3oHkz57aX+uq#hfv_`G=B>_O
z$VC@lv{eBeYW@~)1p2!BgbxNT4NpwB0>10j^t1yC9WXoY0z7@bnSDR8rfzfw@zrsA
z>~_%oM={aNQo_mGPJCc?FEQUm?P)zdPM`WwrXmU_c(7Aip&wzFevdBW)q0uL)ISsC
zez5PDO^VCiNb~TL!h!)b{&D5UaiI-7O2Qw1;0Wg$5s3~oJ|RxO=86C#zm=c*0={nd
zpuwDe$KrT<$0Xwa_WH(-vOr~z(MI=?zVgnzAR3SCZN2@eJ@5AKF{1Kpk6-_*2HZ9K
z4hSjEBO=Q8QoASrivnT7x%`3=rZiuQ)f`=^9fKy94yArn%qh4^?R+S-jGb3s47Tv6
z^>%jRU3MQl=}F{IZOAt5of~o
ziW8srFHbflJ=;4ScGO=NZ@l$q?QzMsB#aU-drsxfS$EQc^3%O{!vTV4n}j1m;XL~A
z2~|S#fq{}__uo4Oi>bVkqs-+fJ^q0!7!mfney3&;`oD9*4;1iT<3NiU>Fs`@IgZk|
zot`dAs9_-yNqB5voG9flO**!m05$0dj&wgiqIXjbnfC>KoIgK4^x1+(Qfoe&xA2Lqip=yiY|$0I<%pwlll_yeDRbdx0+zrhs|T
zaPPpMn=Kd$t%ZAsyjqGY=k)@Z3t1ekBWzfTW&`CVxO~e(F8}Zu6pM7{Tb!QE3{F11
zn6ra@rUUuMbGdx^N-lp^0ViwH=MRy;xRL9RDf_v6jRY>=T!_QL@?5>U`1^=GZk*oJ
zGA@5)HYewIb8->Co=!wBDt|Wt*}(`L;ue1OP4as_9}=8*zT~%82dsedb>le{+%M^a
z-lX)r$sFF|>wDo=0HxzcqYTIM%e}?de@M+keq%AGmx5J6@+Q9C6u#be>Sm&x?HEE1MJlaGpgRHETRxs3J!1?A9eqI4`ahTVKSL-{hC083}x
z2tAr5q|X4zXy*UO>~YA9jslr^K!H7i=LAUXj>==r1#$)dU-<$(LAe4w)-BK%C#Z)FAM`+K@eh8DkHi8=X4^xcpj=vo|7))`UuN4<=&@-76!fRS
zZh<{)dLUmg4*$&;*oUCQ>Jiw_rU{-Cv_pWj%E+E2Y#=e8*nogfk=sCeFMoc`Zs%ki
z-mr8>0_4x(BiBIqxlvqtDnDINzQ7*xZuDF`GaP)8U!Gt;C76GL`5M8`7tB||e52J&
z<=FC(y@dy)GPn#jPKbyL&l(b@svx
zdaOfnuZX`tddJ^yJ)&b3=*JG=p4TK1@_7Dwa^mYB^5)X_%5gY=pI^i#1n#Zy`?r|f
zDW1P4@JHs<)0qF8KCpC)dOarJ^*W3az-t!Iz|;ZHGp7zmysoDLF@N%DrQzJpkGof8
zV?GoO!a2e$&Y!>G_rvCTuAQGY(oDno*jRGJh`S%iaPTT2YB+c={(1_bZ3m}6V@WcT
z`xS7{y$|Q=^{=6ZVP9cIBZ&JJ*Kct4n(f3?1wVYiegEi|;UpM>^VfUBa)>XK|Lxj9
z@xJ?M8>TAkKGuT*v7gF3h@QcF=UvrUVu+iyeL6_rMC$|$0-xG&kku2ey@=KGOXmv=
zhCgxjML@#53T({H|f-6_n8cb9q}d}Gt%DY8`<+<**VM<;4JGv;(Pffewea2$9z1B)$=p*e`ZAiex8&Mcwy!!lZ0i!z|9L-2k@UhUmoPhC
z9KX!URgN9U+WEKOS|R1@PH$!T`X@>mU()r_Vf|A1?k0Qw>0AABDzC`XxR>^Ejz9Yi
zB7RqXXxd=np#g^5qv^af%2|kg@7!>oSuVA6rs~?EghE<&c*6_+b)!+6*;T&6gz!h1
z4g0;3sAW4gtT=ab%U??NjJK&^`zY;w0nsEEl)PcQ@TzM?D9J|0on#2pGzVkpV*bu8
z_(#|#K8f|y=OrCj>bOr9dI?J$e0g*TTn+r}#)0;P`+F)Zh!+pg`PD@@pH~n_?TD$|
zTu6M>&FYXR>0NPem_y}i9}71lT-7;9op5fk>=8oK>cyAS(SHHus!MtWD$!SYAvw1wiD%FPM)Ayz@p8V!6
zT{OV{;PjiBgjJX1*!*hqpCn52KQOxYHtl0vM{Jxyd_3Rc^H{R4-(8#Qr0+7~`#2g0
zY1;v6gy|pZm_1A9CL_e+{f(A#FyWFd!C&dQ_-|2I(BO&nHu98iGejDm3-i)8n$3@`
z%?FuXPQjx$QF^&UGn-GcQVMLnws&&hv)kso-I4Mq2T3R~JJL<^)1c?nWgG1XUC*`dBaDqZfzXX}(T(TfiIE;)$L8neQI5{uHst=ba%>70YJ+qi|I?
zE64LoCBspR=V(#>M&AfHIK-_@oh*H%Njk&rDxCil-DF`-`F;1iW`0LZ7Q+qMN{vnc
zasTj3C|iH^xAR#&C1r-pj;(XgF+H)Mehk$Q4q_+~iD`l7I-2LHMl05sif
z!q8j8pP|GAELWuWQAKwE#pk1&fj0g`_^@wJE@JhGoj`LT*V1)wU}qk>6WFc8Jr^jI
z%j~>08Ey{g*IO{$z{hdKz_8A(xq8d|x#u^j&SL4o
z1%Hoy16DKXsqxRb>pjuEkZXT(>1&p7sDWEmls;-dCvVW^%Imno$*p|8{cIyAC*iF;
z>9=QacCAk5;!kN1mu`8P(_3+ylY_5u^1D}@?8WEbf^%N$QC$8?y4FGVQuV+jY93JV
zcm0~FJ0Ld%bLd6tkbhB_11j?Mz4`KUcHj&nLAPzMwpT9vdFy3+ff5vuS{I)_kC#K4Jm?Ti^e$
zet{mbA<`G1pgfk0bb+2g{{NxCZUO$coc}&2=yyTAg8mTbfn4C<|LwEjESbyz=>ily
zFW7$w_A`R~|DvEhg6G&W`JeIx`hxKkwCDe)7r-9C`&K>*<|TPevVRpVOgFS}l1oK@mu7dgn^F*NcUljB^c}=Qc(69e3NAR3re$d-j(m%@zK-}Ql{Tk7e
z(zAYWSVRE%7(d^Zua8SeEM06chk|t|=m#N6#&fUu`giz#4#AtHkL9B~A3=6ldM2V7
zJ;#a%8j}bV^!HxA+{`aGvz`hDfA~ia;(X1GMiBh9o;(fC*%ou}b%bOgLN;H+;aspx
zl4l;Ip2Genp9;i2g&BtPRet|yF$2IS`M=+>rq}4uGhqXZ_X&_lzR3m(JZ~N+_fF!<
zPg3P%_U99zw~6n6SGsWN`*CVX&u2Hi#RE9cOQ9aYd08ZAEKuxRF$}~#wN2evLf}s?
zMOl8;E}Buq6Sz2e(u0d<0ja`R}i;Z>9!eKdq><9r2s(
z7p~ivu=LqCTHxyOyvVPM#J6UCtj3bV`TO$aEI+R60z)6+42F8S=ivHr-g(rV+3Eh0
zW;n({LWW`1pi2w`dOguFobRjLV6vkR%|ygA7h0L%*@4{n`%GVfhG9RKN|6Y>)6|Kf
zWj~Hztov}1$vwA|SpT&?c#fSjzKe6NY5?h{Ze8d=IO+}gC20!6Kjj%3hD$Si>&eY`nWepK{%<{MBu>xW*OD7f7&TXw*>`zs3oB1Ftenjoxp|6g
zJg#2j&beb%ezWp?E6LO0K2F^pX4kic672cu)m;A6tK9qRGaDV5{=~quOh2P5nYB~z
zZYGocrR-V0YBf0wyobFQ!`1(tTW2~h8(I3WIovwjpbN90SL@r(>=Rl$69b1hKA?@+
zF|27Z%YSRXnUxbc_8F^h$)nXQ-SlV-Ly`Q!3=jO{#(mt)-As1E+a@ZH7fTKtaXc8q
z2z)b%vuAi8Y+E6}e8$G_+
z=gUT9F64*&bFXB6e$pu}y%ejH@^_PYK*>cMRuAOt)NXs;hxEO(xqd4q
zGpU}(T>ZjZICFJT34n>zO~7G{xMIEX>p6+pLg_fvVEBvxvSSA&;Jl4ro;PMS$uIfOrSj{g
z)j;W&`Q=EF2$@EUFqDs;{QMz&nH7h8)+``vE^sctJl4&S1^k8$8)WjXRBj_5o%tw;
zn{YDt_kZWTefV}d^V3=QM81F@3FNnW4M86hRsVl#qt+zVBMUwchppHS63v_s*V~v-jD1pS_=l
zF#ey__y2zXpPc{ePW+$i{qJdn@&7OX|4v`R=l@;z|3B})=kee3`tQ2`9sm2k|9=19
z_`m1&-|@fy&3@>VV-P$H<|WpMf~E=ireun%t(pD6E^~7+={dpQ)wTH;#7^)-p3we=
zqWS?Jy}BeZ)gX}jA!sBSLZ}yho~`ug6rn!&S(;U}Iw7C5uSC(-+5pO|t$I>OyFg>D
zO2P>O&xuF@$E~gZ!9&m^vk3gx4fb>Wrnv16ZlahJX#ZJOHts888wz?nq@)hrGRS@bHSj73b
zu_;ibF46zD=m)rQrYb3#bP$LXTg1L5$X7Ji;&ao6;9nKOi>Kj!Kxv=05xDyeDAS$s
zyx}znBqGy1{tfqoyXw>Y1D^?TGGF`T8?+U;Bw1V2y%l#$D7bd4=z~w=x->{2gYfI|`4zD6`KEXcVvAV0b
z+}a7aLYYih)0zOMzsTT1Q9YrKc(#{L&A~R$%f7Kk82C1;?nMy&<8`fLD%=
zy|I^&Uy_l#d+uvFP?ovtem$1
zz`Od?&rW9okIn_5Wxh0^+RtU1(wYVMWq&~!V%I5wrF1Ezo=;iY<21nOzoP%q|i;
z^vX3rsu09qkhyw3I3Fxfy_?=zPX%v|$VXze(?DjD$m*5xOi<$e&5dFt7i6A4y!J0K
zA6WSgiQmwz0#WNQ7nfBffT<&V$mGg^d+c8S`sHE}|9XKdezXvfY@B?qP}~E^#MB#P
z?{@*0_mWoUiCck`fwEg)UOli(ZMxlZx)xk@*%}>r-VCy%eOmS88o^-MM4!gf8Uml8
zHmB;%a^PfgBke^<33y9AB1;mQ56T>$ufEyJ0dnNFAC_Ndfd}k1l_EM>V6uy&s$e-2
zFj2w@)zA?G%8BnX;VSlSN>#Kgry2
zxd3qfktDkEFc(}A)jF=O&jRmBTXNbiW`V2P&jkx-Yrt{T*tyu|CPLp?ffjjE4-k$t
zQHR_6LB^N7lj`vVeY>^#Xt%lqNTjy>Ff6SCuRW^8wS=2NChPo7ErPyAaCaX}yy_*y
z54RSyShIlEn|Omm?Mwo05trMWgdCtB*znj?xr9)^v9PE8vJ!;wb~34T7J%O*>LgL3
zg+Pgc-jYGT0GNCme=q-3GUo6Q161Nbk
z1FjNK|3=&A=jAfcC@~)LzLKz>Plc9~lM>cv4aVO#?tK7SSlic4?FBdPJwiH)djavG
zPvks7&UY`Qc+bRigO`>G)8-rPz-myScl>WF*c3QQ`!>@G*cW9JeGDo=J%!-uMB%^a+;l)3F6~dUkOC6Bl+}AP2=i=iRPQQP
z8tA4U##Y&rfM0-pnT2W@;NWBN4@fEj?%757`8x#wxwHFjnvC$CPbk|TJWT|z#x%=7
zT`wpNbgxfp?g2TCjK0cOdVu2ko#nkhg!}M1St-lf4_;Ru=KD^x0@suK8~lyUKtVEU
zAfIHWC75V#Iim55`p1-08zAh0+3!o&Wl*Zftv@}QATtrKy1m#orgXPuzm@Qw&x88
zr#r$nsT*H_-wZdS7an;6EP<_3cQ6*vsI{4q>%;&d(?sW1Y66b(>t>I2P67|9u<2#~
z;ub(HE-j+^uoi5erKHffUjq~_t9~_^X$R#hel15on?Zw5wLqF>4d^s=Xo}OS0=3l4
z8ttSQu$wT80>K5~$#duT2AO$)!&t7}J-iSsq;H2^Z><1p9!o!G4ywU*-@J9JqfGFJ
zsOQJ5KmlM@nfq3pT>{n=URO2Ll!L0GT^H7b8Wezgv;SbPvN|JVW?K35-p{L%$lZcNk7
zTK9q{#BuZ65h&2QE^*6Us1`V`rATZrHi9Z+={vuVTS0lK3PW;E2RNZGElzb41x6yH
zoLWUS;LK@}H(nd{p!AA*Tkug62pXIv4~}aA($lIDEd+ZuO5edqv4Ml~!twOXW>7DGgx?8=VLzSIpth58n(wk^q
zb?pEGSC3@Ly*ddv=ckTcU-p2Z7WOtCZD4*{BgMI<3lOW<
z_mD02f&i}I9-H<)aQjLu35!cR@Y*A9m~ZI>2Akg$>u>i0)xFPG9}wnoI;rXt%QJ+&
zx1l_bLsYQ^Jy0F*lmqb4}t>O4Bp1>E}_?<=l6`?;V=JVzIP8x)I
zpi3)v`Pc~f90_LMx@!pWzqF2$f7T#a{4`-S@pK4;7MS|z)(?SN$B-quSwh|hW+-S)
z83aMw1zP6ELtyY4MHqg11PE0n+6HKhfY0i4$xVd*Go$V}`$j%Oz25zens0;6V_g#5$7qau8^oHf+rx;HApcTX|II
z4}~BDj^5nQ
z{QZ6uM3M@Xnz)aGX+=eUnBZ4>jh?qNAoQn6a^F#XM7R%0VLfp>L&7?LZrff%U=Z|v
znBUbS$Qcy92agc?+x81L-u@3>2(h25rx$^z=E=hSlg0$T=5Yej
zfT|=}axa|_M};T`CARj1-6k5j&jdagMRWg%muE)6m3IA1>%(CX-E*V#g%jb3r`d3c
z4*@#nwttd02LaIjD8lMPkaJ(_dhgRwumE0V-)R{Eqn~4i9DfXh>{2r%D|{GyxF~V^
z#jQawCQ@q|w)8*wN|hU@`1?U|*V2$sUpL@CJPo_ID@vDa`r=zMN+XUEMc`2Z2FdxI2?IA^*h(U2-Deg82CKtBhv!fiDMkF+<72
zAZ+_*!^6A7;E|5edS34!fWY>bCrka{DcyoVuS`EUz((>of(Ul0XFJ8fzY~-RIrU+O
z-N3Kr`}soxo_^=2;xAh@gP@0P&3#I;3;g81;8Oao1E?_AHYlpK1D;d^U0H&iTH1PD
zAn4r%u3U5=VHNEKC63lw`-FA=999j>`L}^7w#f0w-X@@qQ1(!XHiDfq&a_Y&M
zR~XPtcyv8upbMZz%M!Wdk*ZjuG@TZRC=@X*n>O*(sWN
zQ4VbS{Hi_Pw1A0D=dvOK@Ay?Cm1u;}hnR6bU`8gc0wj5FYE!Y6fF!;XypMlpf=ju#
zm4~#8zzWmJP;>Pvp!PmEg+{&t2n4Of?3b1UUW@Afjjm##FVG;hm=_NY{P+~a6+M9G
zrGaAd{1ULW8)%`vk_r01tW2``#{nXBvBseKbWl*v!@ZNA1>_6G-;DDOftyq=HgVVx
zAURy#zM?)1R2UuZx|R}1R2T{;1Z0LlL`Pu+a*MEUy%#v-Yt|1e=TAOVOX>s9*&V0~
z?(~Bo+>84U!A_9aoYTlXTSdV2RET&tQwCIb%tpJSih)GYNSJU#Iau{AliVIh0f*9L
z_^(+{pyT^zEo@4Qz(=!=e?EQ*^ekmVuC_S?-ix*4x?kJ?Wpl2ho@*8WJ;&A&{;A;m
zH1!e#O9Y^mD#~3iiv;!LA5I+w6a4;$QsQNTeGu|(plDGY0)o5#}IixqWqL>>%%yrv8hZ1f;w4yS(hWj&UW|e%U
z`ZO3+tq=c*cYY6gRCyM!ef$Kpeb22iSLc9QJ@+QQ%p%}NX%>H>vIH#p@OiInlmRjk
z>Xbh6GT?bP<4k=Sp`PSCV~f5wp7Hxeb5Hvm`dQm6~P)CLS)6Fji7hGaR
zl~RfT2e+UZM_DP5&Pwf@t}F%At5=72atc92ajW0$s8rw^6dl_umJa;-C%zP~wE?dI
zzc8w*9+3KI_hB?)esxuxJR74p0=8JX@~n4DVTA&%HIP+rvc`wA|;zF>ID
zs|c{w$VRf9F9xv{Cy6c+^mO(O0iV>c5HPA7n|4uW173ak$+nF&AWs(YX_O-kJR_Hq
zQp!jNyBh~Tc8RhK!Z#&6J1pg;L$jXu6k3?zH8uPah_MM7PnM-vL?7v4*a23xG
zSiW?qy!fgJG(FTjw5BWrIRhV(X>?KWQh8^#&8-f+l`UlsxZeQ2GY)n|vZjNQBaSCB
zWl4a1V^-v5ZwxR9JC5#{iv-TL6{lW22?uKp)t5!Hi$MEGQncW52?*CMt9`gq0r(kb
zbSA@55M8FWXpn*eW2^j;GD5$+<>6~d+Gs*wOI>5=>t{p2x`_VecgI0sDo!qG|En9E
z5ehDPb!`Zo`&$!4?mGzb(@2aI%lp8)O~dn4;Vs}fr36acQw`e93yi(45%MU0>;>$2
z4WMutp<&h{*deb^6(-kH!KLWpl+Nl5aHkn6uUiQR4fjj7lzO8<`Ck!6VdHcFwN}k5
zbQb}&SPuW!{)NDNwYVt!HVR&SAM?=vMuA!kaAM>77ghLFq)e%E4(=j0tMXMmPnl)>B0sA
zZ?V%ISCuYMBP9FAd8rS0B9ZM`<^AB43=89C72*B+pF5^WOasRy#JEar9vHdkMtkad
zHMo26Vdv^n1DLILeSXff5aj(*n{t*fBG^yWsiS*2;MQ3rp$4r2WM<743UDi^92Io2
zvC0Hhd5h=fJ2OC$O&M?TP9b+U$7AC{pf(E|otRzd3*=P)zD5J%dZb((nSm>u~9VBBMBRj2I3@tgH8Jt{b
zfJT)9(1n3mh$%t*e)n7p^v6zX^+2T!D(en}99<#m`{vH%;MsI2B~WBC>~Atkp*_Sfc;tan
zx=)Rv8i`Q2;_v9v>;foCUQsyM<1_TODfmQ+rXTujb$e?xuML_v4{ZI%+YUv)P$h|e
z)d!vZZt~A^vI8nnd;f@M#UCyE_w4)w796@LXZfR$w-Tb@etyzKHy&cXnX>mxIuS}`
zIsW!=r5RdAO1wY+>V`gC%TQeE#v!XWoviL8y^yGqzvdIaKB$s|cC0+M2(qP7E0N8Q
zKxcZ7#&3vap=&JbfwfvOP-IZet#R$b+_`0D#^m{+SwKcs0A{O2-
z+|{yx>SQcIsc$Cgv+rRo+S&-ws17r5iS$B`zev$#puNz;S?>n1ns(@x4N>%A9g6C<
ziSb}9g%F)_jP^ByG^i-oJTvic9whvU3?{SfgRw!5~J
zz0l07r-mF@3#2ml@s}>^Afz@GXxmL$k1F+C5t=;T2;JRhxa|(tqbsj|s8%Z#pk8jT
z{hdNRUR
zDp7;C6Xm77laNQrm6x`+YEeO_eJ|I`>KOa%o#^E^?WiF~Qei>s0Hlu&F$)DHpiB)7
z$qOqRXlMhC7ORdN{N2~<^2H|$(64uOX6MYiQL0$-w>`{*ki2##xos3F{L1a|;iTvd
zj6?Eo%JbXwn2XGhiQzX#Q23o!k%guIP_#elnZw5mSht|n>GwzcaP+;X8MDfNXhcij
z#K*f-m~F=H9DO?{d}{4nhr^|N*l5YR6+He0Zdky@=f7)UOiv4(nCLX&PxC@`c~A7Q
zq!>|^h=;ClUd4Nxs+Ks|$iif>^MW`$&qQQ2u_1vy`73LoZMvDuZC;pS&cgm^~(l
z+#n@BdK;H9KY!&;x13HGL@!wTc+(j~P8@R{0s`^rA!SjDaK
zM~3Zt*sod1txltC_zd0i`ymzu7-#YA?dJi(pfXS!N~
z<+Ja4zr0@sKiybc@T6>nmmKImrFZpUOaGdM$puf~2P{0nBW&I9$FBUotb2pli5;@w
zE9d7il{fju4c_gr1-mJ2$_5kiCGNm1vgQv=e{-(ARp11Embu{TS4Jv)=cscVkGz0f
z5nGc~6dtqTVs~P-8U-&Sb?gDze{YiF
zEfuDR)}>GJd|BD#7CKYpkLC>oJ2t_2f2jZMSAT(spNQ2|>-5DXI5S56A;I{M9-@G6
z)|ZffH52hL%!&h(F1HwUYFrifjzO}2VRY|pEX}_i!+Gz7s&lW&;WSbOMQ=q<;_r*A
z&R<}lLWB)*$urD9qkTS)9p5H4tIuzNQ>~qi$@G82pKOfFl9g`b;SM%*N_4D<
zn?XFafE+bmo~tX9#(V}Bpn{##UlAczJU3>Ia2b13H;XGw$0$U!uecPaCjAbk7
z=oA@rVqS6=#1f6Nu|mfaTi5B$;GKVbDOF}Kum;VavG^rxczo6R>~nz?II5ug;cQSd
zOy^VH`d#leM*XFY_=3|r__ZkW_-DpIED0FH0(&ILUCJx!YD{bJAQU*t)gB4YB-a1c
z4@`!6%lkcBL*p<*ca`F)eG^!#FYTaT?J{glej!;dp98y<8qRp`9W(qs--Bn2l@4PN
z;+UE}HpbXFMO64$DX~H`R*L{da`=ZfxtP**D$I|^IJ~x)0Y1#J_n|O+0l$fNWagy!
z12s&ADLuPIjV+|4Qb*l916zoP-U&@Ohl%Hzwoq7qz*w#*x%&OR4G-lu7f%kbV86Vk
ze4nWCU>pJXBT)qymfW^7IyOkb6e>3f#aYJKSr$o|mNFeUVrts=bg~|Hmyw3wty>2s
z-Md@WYt4*dZJbob_lD5>)*~f6or92Bw5_CoWID?7&T^undjKldS*91?W`K8zl$TFc
zCP5$CCOyxnJ3!)BZ(Njp(gQJlo)cNOaY6USX($a0IkB^vew-l*gAmUS+qJ8l_0R*?
z`QXo4b!dFBAMNumosboU*t+w!IIPyiDaFr0ft7Bm^707xLYiB*zQ`GOLz|~$CP!R5&w*-i{9e|fv+c}^QN?&tooIj;rU
zyO5
z`M;460xg<|II+TssNvz!LnO2WO1`(?
z^6C0ZRFqyxh51x6^x>9S^`n6j^oQc_HFu*oP|dv^uMq!1$h(0%0zWYZU5u^$<@k0G
zV)*@5BI(&6^l(MWM(ub4I*&Z`dG##?!uc6)ov9jdCV^xrCnf%BaX(F2Y
zUSR%wXd@c;VD|8$Q4^Xb*dfL+RgY>?H@R@?6`@}HpVDr1XF#?}A{H?MpCG|Gk{$k=
z$NpH7e8Mt4|X84LW^xt1(ur9LoC4!lKlg33Js=ipdq!Rf4`f$Ab0V5()4;QP@n%Dz3BKbbjFbTx>Bzf
zs&=03Vp)F++QpK$jl&Bxow~h5{p*)ooNNhkU)e2wQJS4;;
zGOtBlqS8@O6WU>Z^ISAT+UX$|Uph+5{^lQIkd0om;I3FX4nZj%)`rvGtc3D=DEqoD
zwL#fse*Mu?y-@avF9B4L*1`X6*MaWDRe^Q&wBcO`WvCh;~Cb(34-2n
zPsFueu7$)>%pyn{Q_+ot0MVq}W^{I;uECcaL(zeM+~l1tDCLik$G@MnpvEJo3WHr6
zAz~V*XB;ff=mYlcWc}cG&)|_%a2;Lj
zaYp=YT?ZeyTKT5Ko6Cxr!V$ZVhLt?5^137V^uJqJSFRTCoQxSY0J9g1C#~iN?&aYg43L9sIIfL*w
z@V;`*n4(w-_JaKH#L>PN%xm3cm%?L@6_G|u9!$8wReu~sK6AXqJnYCK3VC<2GnP8+
zc3HJBA7$8)DS}}VOBs^XkxlT2hS8~-xu2L}_tEV@l7HCC?0}shvOG-aN?JUrRt}s@
zQncXzvJ|uFrlV*(Jpe1k9W>bVY`~cteQBTW=EAhEyJD9t^Dt4i31;G_6>ycX{I%Jg
zAuN$+`ir#A5hhGGpBdKD4*#UkVY^CMjdknLOz^-JFmCB{5KPmF#k}T?Hx=bV6!o14
zw-!lpei8?cQneF^T75+m3m+AZJI3ECf-WGv%zhzd{HndtTmH-hAv;p%exBFOk`the8t1B~Nr!YQhFQba~wA%gEM4K+^e{K3Nd@C9eU210iPk0U4m!$@ObjNn1WUa;WGI6>e&AwBDM2$
zo`jYiH}@Hsf#-w}qC;7uHVYL*N|Qx&?!-O(+>7;LjZ9lS$W=i4ofrk4#s8%s`rtmo
zn8*HGjqx$^sAx>;spv<%HbVJ^-LnjwGbq*xM_zV*KQ>%
zhX>=YNE}}_KC;K}dfK0RM{kI~FPPS3mo-B)o#uvvy6y1a334Jf;vW%jCKF4cf-vL)
z%q2WAY=XS@y|ynLYK3chBze4JcSC}C=UCVb0`cmx`i`FPWSrz}&kwmp1$^gcIoFSW
zZU}YE5>6#q)+E;%q!w(iLTfFt4zeJv*s^ym|HNNiDZA
zUJv*(SWiVDRe}ixR?PPhS|3$4b|N#pk23VliQ#%U%GIzcyif-T=KDLRw~~UFXvche
z;^lxp7>m3}UHu$+Yx>*w6_FhNm~-d6@Q=s1&=S!@II0#A3Edvqz*3QK`Z5~|udm@%
zrzq?mlR6_dUE%z9KUpGf7Cre%eT&H7wPW{|h*UgN>s6@UKn1@4Zd6rGPz^V-Os%N{
zX^2nr{x}1tKE5vA?QcA^4&Q*3cP{Ym!0)#2CXky8;^+TD{1%1_SkgcDHSKjRgu&`Z
z!^NHRcuDBrYt}FwVo&@*hQ^WuF^e4XTlb^Gzf3$(RBF=29s9pU(PT&>MKuPWbIu_k~lZA&CC_A?|&bmxi);p=A#}?C4A{c{l8lXQ6vG7d;Sl$=p+4$
z%Ss9NsTAH<985xDcn)I!RXgHNwUT-gicgT9g21`XA$z2FL|fdLN)TZcqI_0V*aUy=
znf&G&I11~*m1P?gMeqUNhe+G32$<1%jM7i>Ew;P;{{HI`23$kZ#AlJqklDk*?zs7V$I!+ojVmz|HN?yrX~`kaIq&s
z{;kmb_=#S}KenOg3$HTbrK&-(>ltJ4!6%ERPdq=c8oMTu-N)J3e#!;9OYGKIS^a|R
zXyY@OP1W-9s;&w4bLz8Qn4b{#umrE+5ovLcrlL#T;7`%XVnCu!D0#%h-KB
zkHMWcXfx6Gr5LF%Lu2Vq5R7d<@Vq~&0P8dxU2|Uy#_}`tq*kRD;GczMNqPBS-~guc
zr0cWASkqq}MX%6Oc$K?6=1*4^c2`(bZqel_Y=5uBcl|U3J8xDB-7yfr7?ZV=6eT&}
zznlRR+A&M$?WCr)RZdGxi^G&V5&ebkCJ>n4gVoz#WR@B$Xat|6a&0earm^<
z*%6lgZCAeCD}X8F2|JA)aAK)^#oE?BXJNtkNoe&ZH}*KZtib8bCk&%nk`)79YXxPY-DV)
z_@_S<^OK*#dkuf$X-qk=32xa$F5NXWo7hywfq4jOD&W`mK3#|IAHo4q`>l}P>xYSf
zRGe@%L)qucPX-`W_LvjYI<@GH-V1jxZ}dXp2E`Y=YLZc2{jsrJlJgkt6=6Sb-Y<|q
zbotr|#V#m9+^24GzZ;!Y6wfZ-ZG9D`YKQmIoo1rfF@|oYp
z1iwV~k9ssbhq_X&s5yDs!3mU{`V5#D?D8x=GCY_M`;eHwv3G(KJ{)$Zlapq_%C|MY;?QJSs(t_zD6Gbb2Ab
zl(78g?|UHVWbNXs&|YYs^EY=Fo)p0q11
z*Fv=0SC?mcvLKb4r`c|l1)*#K)Z?eBgHiJcQL7=ze01rh;+r!!($IIWtDR)ELeQz)
zJ&W~*M92Z?>b_7;K^wjcxJWvbpl+vPuH;vjqfQ!7a`f{g^m86;axbV2`hIjJfkAB$
zDt`2y`+4*LOC{HY^RT_mNyq)JJSL+~YY@Y(+Zh@Sa9@24W
z6hgvFnmYk*$&k{I+&Pk`R%i}slg*-kC$wZA9om-E4n)rl%9c*U52|eNy%8N6Shy?B)J8QDr%LC+@(t&9}Aa
zoX(C|K30ixFe;yK1JZ}h!jsEBQ74fG3?F)VCXLybrOe)g_-LkvS~Bn=@UXdCM~
zxAROL^s`+v#;B|Z+Q>0kT{P^1>P&l#?1*|G1$@5XZgMZ=tsd`ko~IP0-hdOFZq%cK
zAHp6B3ZbYq13%9Yc>|hoV{kNqrV0JANah{D6$y1MOEJHeDTb!4bt4nJ{h{UhB=+os
ze&~8}`Lj1kb5Q?~@}2|>A@1o*YH{Ehgxa30sBK7=qd`NwjyebHQ2L~;xIXnd#4uoF
zCO+4VG84-=-TgiT-3|@yem^#kp31YvkN=Uwz2nMP9!ZE|w52wW#&5MjO5UgCcg^e|
z#qZaX9<9zor`2X>R%S>sht#ErL#svTjMBXqYLf+oI4sUtakc|0r0dsgvztakzlKV9
zvoXL{S{2>w4s`JG&F1z2+70ySqubu6f@aaSrX+iv{38gWtd8iLx{CdMrRm1qsEL{H
z*sJSXlEI6)A|lOr
zVI-ny;o@~JY`0U6!QVy&3(ooIUijb(TwlpPlrt`jH9N_X{F9J???Bqc8|rBA9FKDGe6=m-D*Ct6{rZSS7nSk2P$C~NNyeMxoN;hldtzV+#Ino>h)(3Q8qjp
zX^1>Dw#Te!DkXk@wS~`Fy`6rj<%SKgwObmb#=_Ug%O)k3S}`h{`!6HDy~TV@jVqq<
zy2Cv8|MfJccwqt`1Kd;Gli@wrn}Naa7O~q`1S0d=Kf!8pF@>^A0oXN)fXWm&0$z({
zYcKmk)mj>9ua!Ky<DVn*n5Lo;EAwzXUFr{BUJN0`%oFg4!fI%X&SXB?ZIzr*!3
ztPQ3Mpj#`5XeI1FuxlId$v?xoQ
zTcSmR7_zmuI9TzS_^0vM83|-k*~`*(+Z53>N)|XKx`^NUGjP&`j}rm%xBS&lbK?aA
z^rke`l1MX4+jj1}4=#0%n(Se*E)opcm@}v8;NcnvQdLBz$h(sb)K(wf;)W9|a`loa
z_?6Pt@cqPxh-QQEbuTG^S2;2nzI&sL@D=H|b7@B5zm|9=L)Jbb)IVJk^d05#Yl;+x
z!h7mC&DY891rmFl#V&G9?GLj`&c($s&(S={QZA664jw0_5%Yo#P1d
zW~7hvH)p3!G#*BjMGsw<56f57A@m#XJkwlr5DRmAd0FBrT0L~IXToX2o4bw<(9tZw{I-2y5@HiIbGZkeBuSx2_-^$Ky3oUZB32%!{?B?nSJ
z107vop0250u6*#1h#fe*mFYZzUar}<5Fa3Jm#xWf42f3F!w@5%wkDzAW;~Vsil)D2
zEl%s0wHsy^g!_vfJ{te>3n$N;Dh*68!LN4CHKg&BA#2c5xWjQMZp%f|
z_sqEv?7KcRe0KlsWlukVq`quE+>0&3wQW=SfBWSk)ecfSS5nH5H%AZPv})=a
z#<&YLugeecWrt1CVV4O!deK&R_~a;VR9%USpU=jJ%P8E$*Pxp1ySk%OG=K2@v4*fd
zdLm%I(|)VnU>sM@_mY6aw{Z6H5zEh8S%`-SON;~l5jPew6jqylf;iYr6?Q~BAip=6
zVjf7^A{$IiE)++}c+8zv8B+5gKT$Ib1%u?7IKkQ`^(M*;{{A`byX8lqS
z1-8nqFZr=}ds^?)CeaVboUzW(cNbfPTukyH_U==hF2MePLZAUzTjXOq8G#|+zMh%2
zTmOo49Sf~qy5@`Y^e|5EWQQVe!q`ck;>&oR#}x5gWgnuJ-u}q-YbD01FJeEei1PeAi_;!7_dTP!
zhIE&Ddi-<|MLr2UeW<%5gg`ys*YZy@;OGU42Vz{g~y4LVbJhnRiXz2MM4og#*BGk%T*e9C)?
z1rL%aT<@vC$Y|>CywG7sY=k{ZKTB~VpAA;=5oRL%^yb5yXF0R5$thB+pL^p3l0TS8
zL`oBP4Vq$CKQrMMH(O`(ljsoz`zh^3JQliad%6}ETVZR0bn*@7B(b5Fk#+i{C_Ldtx$m{%gnj5CV{DppE4fU*
zh%+95$BJ*@zqdQEGq&dnIGet}CeM>EDa%@7M!$|8E=h)CM1MI2FMDz#9EY{Z@v$Q?
zrKsTzx6}dbPpOlWmt+@QId@fg^Gho>b;Na$*7O3VlX~FN@LdU(vTipVxgd-=H*{?M
z@a2W=kl*;yJQbFu{OLOL#g`b}$%4qNK`u-~cAch`gdc_-3O1{c&SF{bPCavqI}4LM
z;dgd3g8qy+BTxi&+0eB$?3JAk0fEr?bd9S3~CnOLCTu-KOBdOmtZ{Z=vg`QRGtg<>o$!{KC@1nfh+jziq#X!l+
zrp?CWm6_B5E#EYp6H}{!K85ft2feF>S`-YToANNIQ@C0rbH@b@?j~#MT`ENjhvSR#
zRXd1_p#$DPpM-HXtKD4~uCvuZIU?0+bE(&036J&@HmNdw}T5LA*cFhfE(2vt6#
z`S9H1EVTCYNzM;nBGJpYwWab!I#F}4L&xRg3W%#;Jxmo(K=%gKDnjUjP??_VnSBes
zXx}$)eqz9ZSz$C)%Kyd!u#-V<`}SdV#PQ
zQj601Q|=hPNr9Aqh=e30+<*-KTG^<}H9=!HUNXOt2tqec3WlT_)I)67J+Alnbwm1B
zwhH35J0N?z_^w)k9!QXeQ_OEC6m=zQu?>v;3eB+&%CxFfLDiWxpNdMqLMjYZcK=vP
zQ9YznUdqJ_dY_xxWqvmt`u&vg(MWbYnjy#kaXhI8>KmO{woF`uT91MX9*jgmb#9~O
z-d)|$gdT5qt7{R|UdMK!@yaap?U|$hJRdQ9vZIDPHX4T__9t2cTKb`^p~G?88VsdG
zQk{}%1|U;N)RR8=IkdB|Wv7AufUZo^{(F#fh|bOLz9(9t
zhc!MsXK;kIptax2yZL8Fpg+c`q!M;BDBrIO35Ow+a9yO-7LhL#rYm(=_LgJ?dY=OF
zIv;JJ4hI+C&pjo9!|mq9#pEtvNpr}I`E^;0>B6bW&JbpJo7>Ccr5+0w{0b$52YKQ1
zf2Y2#M=D_vle8A>(~n`dm%c*2f6u~Y*?bRk^bgRmrbqnpx+md#u4>-(C9+uG$qzRy
z-X+3;OP?v8bgw{^4qCKRbS!WgiO&g(podVVLNn7d*muTtUq#4@WWMYnxiP;C4Hsp>}IS
zY@GMD@*&MmgF6zMwYdpw(JW%-no|)i!eY(K}J4tu3=KA$em{#(d
z`ycx{%y~est^-blAE=-4So@fO)vCsj?p@A+k8}SzxF^+NF~{R!YRm_ie2w0@U+!LT
z4!_e{e})UD!74x1!S4e%rZU{Z>I*QBZvq<~?~Y-U`_Qb$pCs&l{PvLw7onc$L%75g
zO)93{)cTJ7par(|O!{>}oErb@R+)Y8UOTM9_q<#@qXk>kd37Y**AJT(4yyZBZ(>>2
zy8>3%`4A_}rRnF@8F)#h$hJ9P0K0SZY9&eN5WH9YNcfl3UyNz{Tw1m5EhMTUz9W(N
z7yR+<+X$D>D=@c^x_8U9V~pki%9i^ffCw^LiN+W{MiefFHWZ#Xf?I`&i{?}Iu)zFD
zs-(0Nh+Wd}-G_n9cvD@_m$QfW@u?-L;1E$pe59Xm*)oR;`N{QOB9)2(_fdH?d|=0g
zP&88KmQ~(IBG#Bl{>g|Vn-xLkUu$`B7Msl{X?r|Kk>I~M+DHlH*PDw7iDVBxMm1KdJl1&K
z*NZBT`JE7#b7T2q1EIK0l8cP|aT3ALFx;a(tB%{gF6>9f9wN=ll)Bv?O>xy5PY1-}
zy%9362d~<17bCZLa#DN7weaFAajn{t*Ky{4r;lXe_mOYGhP-FadE#kwCv|-o>hb67
z)q4jzVMtcLvB_+}M|}Cr)DFX95F-7GLZiPn1Mh9Dm0GqLM=H-Zl4`>4$m8gRoWHNa
zkc%{Q$4@fT@a{;1Pb1>>BMZ<2T7jc&()29X$PTB<`GNcz}3Vw7TbWS`}TfS>J8cp4QqIeUJ
z;XqGP=r!Vh4*iaoDfty*HHao_QU)8yBhh_?|^o~mW{B4RN4ZS>m2pCM$or}Xy(
zYds?UZa~d7{t#*038B?27n^P8gg7&M|{8!oIN8
z$OxSL>i68BrpQ}l!9t!NNVyNRoO`cX!yBSjQ%|Y`_Et{M*E&3@@g%Pk?(e%%^G~vk
z*k{@okQtk*nompAtZmfio$qG_F2e2dZI;~Nhb>p_l+KwNiW13tUkr(BJm;E70-6E1
zxO3-|`8Ff?oLM|~lI+u`Fk)u()Txd+-c_=gobPoVE*X!5!*8r$cj+*=O<8002@7Y6F
z+=p^C5wQQ+;@Ycw=mkO(jmUx#nEx*ok)!AEti#ID<b@CNdY;&?_zSDhP;^@^#`_~e@fLZP&1=1T*Ax;+w<
zOkPy$UKb7>GLEEvwPnNOLi3Kdjc(u|M7^1Mg9h?je`HG_7Xhhd{-LL8vj8J=u=s;J
zihw+KxZ|pA5S+M6ZCZg_fr$IcMnW;7FzwS*0cYq02|2EB1kZ$k4YMx#quy9>%T)&2
zUUrAeyaCn2_k_T*4>dugWCg64x6dY?tbt|{^&f0Riy@tGl!{qh8gOHsqDuT82q)(U
z2iq?108wMf+wHs!(59EO^x@wQs8ekkwjWUl21r_&3msEHtNMf(-&;pWrK&mRo+bd7
zHmGh&l?sFF^wgJpYx&{Z=3ec^#LFPPyI*y{*9esPBo7*kNJAYCB6A(evykzAyp!V-
zMsO`oGMrkM6)s6l`8^=H3kjr|60U6P02?;-)U72-=tOsHDg2TSs%7^IJ1JiWiocJf
zng0ue0|E(lqZA&n0$Y4>5jO?Rq7|B2%BlB6`kHYZilu+JXwRd1Sg-rhIXcY8h
z{;_2oECH*YLIJ`BD{yY_sbn9$D@=!x=N^0Vfsvf^r(NTZkdAZ4^A^)TvHD=j#Raw*
zME8l4-Wcr1`Z>j&KZJ6@ZcBpz6XOcZKWkzkTXzI;Ve9fn+XpaX$FnM@JO>auG6RCq
zmp6d#Q<=$5zIBW`8`t=2KaAzp=52g_J%ZR>?z?q3K8>0BncF)4yaF$nkMD6CaDf2Y
z59RdZJJ@cU)k{v1Z%E?n$47&Gn+W@V`E{Xok08tB!pAZP4v_edwVs=j4ZPvMFKV{S
z0zVowF-}FY1K5e)vlX7e?shS3JMq59Lj9OLlFE~iC<&rnuj)ul$B$ZmI3ftSOe=qF
z_WO5)JhJ+lHr+EMLf>1I?rJ#ZYyM#0@J%3MY58D?&^`z=;dC?%u$aLj*=w?$ZT*mj
z8-{!|_n#v(FFM4QY{D?!pkv*@|DupzYHMu`H)fHzf)zaPLJ!7lvY=WQmx?&lg`1t5
zx5iq(-Dc2#oPkAsXz_^@VFyOj#l%IT!3Y}xqT^xPjlDS}PiT15j&#z~MhbZKU<`w7
z61rpIh`I!uYC`HAgovXcs)S4$%Qzc(%RWE|`A9-?fqdu&7Skp`#vL7mwV0+Xd^YC6
zlHBjKoMz`kVpg|xf4$+tmSYJ@qA&3xMghk^;!GnD>$qVLqaq%R<%IPJ;N!<4YXz-L
zy?Kza3hEC&s=`>aSgijq+UJ;VdJk=mt~>I|W%7tmKptB^gFl#GWE;-R_{mdVQ+{oQ8afs`cO6eTI56`_`WHs1@
zDyOqM4@r|J-5>!y%?;C3BC&EM{21na5?>NLFRgN+-eYj6LE0!-!9@7v3!XhG=OFq$
zD1G@4-3>&S7WpSLBY$9W38OM8R3{qSknTLZ<;6=m(Ew2PU#0s1*1js~?;K%MG46fR4wHveUaG4NE@>(-
zG=UVO+v2te`w=fKU490fj%<}qJAtEr*nVqx8uOvQWBWjQa_@YW;txh7@YKtDj7FcN
zHCW=J_x%2U!?gqWvfT9JZAjh7Ws20T||Wa
z)Tx~i3#_VADS62%7-@{3P@ygTfQVm;{BphC6{V~TJj
zUeA|Z3dS}#!iXciab#Q2?ZX%x#nxN=?_Z;+M8re?_4ditSC5gD~NvT0kcGP(d>${Noa9VMo=``S>x~UW(afpd>DUJ+$CkA0ROhkj{
zufi*PM;#Ut4}jOl7fo%`O^E!N`A;q(bf96kt}8*78-AO46%5+6LFqMtI*V{uK&^Z$
z+AcyEo~uv{QSQ72NGr*^(AqufbD1UbZu!&VYKUJinN#khdL&@F;Sa`zuJdgM
z27)wEVN;kM24$!FZ?Fv}0BzHPy%>%md6#a(e6@mGy4o6`;qzIrQJ%%E5z>Y(^mtqWm2Mn+D2Hd!y)?46bGlj
zNIvs;_z&=98YC^MRKn{qqOzT0serArYc|Fu5%M3IE>_=f2PFPyhYI%3;MKQ^7TFI5
z0FA=$EXsve7@qplm|CI}#J!HyE-_eug6ZGrp)Uivlcjf1T~`S-EbXYbn)|?M%lo_`
z)=Mz2HnpfTiUw`-q%lF?)1g}nb?ARFJJ5$tilT!;$9^c>Ps1XSi?&*v7bQ!Ywg2j#Aw#$EqY2?Dm$xK%
z@8qp<(u7mERYm%Z@FoV_k((koq>CSQr+zbgN=_ebiTjhF%}k3vjCvGf-F65B?op|8
zYu!XWm`fezHHGkwl2W0IHa57+L#-{H3*4w!i?fPdr#Pd}=LlUZ^nuUsGb{cGnyJE|
z{m4TA*B{t(*GpLj682BF`|g~@lXl`0DV%-qUmnMcN_y(Jo7W-l@+U=fjMwYvasdyX
z5+!+S|0x-2RI#PRlV*={hKp7SEm-1`VWCLt0uLI~ef<}0G9Oy7F(8fo<$ZDJRYvH8X@afe}b3AVdtk#!MqJo3{1Mj=%K!~lu
z&HmbN@Zp=qaa-1M5IjA5wnM8Pyp>B6-%cUMdG03o%Q-ZG3#IKMdod$Gcaoj%^PN^G
zL4mx_5Q~NUUuf+<-ls&%wMTM?^7{ba&r&W6+hrJW7i-F={|+YlYXeV_76H5O^p&We
z2%dR1$C`YO1HJF$LoIfF73>x^=Cc(Zz@1?tpPo2-$xGoFTXB?>(f*5i4^=mUQ-1l~U9%><+ylfe%iFHrcE*Mn;F
z4a||-LR`*&fJZa@FB}TufO3+o0AaHe^fBk&xE31<`*!9{mnKi4OLbq$3KSZ_S=pTt
z+g=QYdR;A#60SU{1A03}{yeylV`sSfff98HmgNhh*aR5$ZEh~w?{M;M=D4I%8>k_2
zj$1>X!-i`CPUL6(;iXdUil?i5fOecG=G13?Agf!yAgaX+Z!$5)a}eJI@8bPAEl3Oeck|e`!#9Z)I)l(CindsiNFom9jVfN%G?_|m)E)o6WM@X)4z@X
z4PS#R8S%knVPtS1NiN)ZO#&b+BT~=&N+G&lbDqjF7^-%Nx}W5vR;RC6{gy-wZs(%2t(6SbQjUUY;Gii@g2?J6*1Jcae~X92*s
zsPx57sR8*F?rlWJIgZ_F7J5@i`5pN?q{_+WGmgD{w7j|3!VT#cQvbHdZy^YWFXq(q
z5#g+(O7}i&#iHaoul+3SM{M&y-MRfHO;;n9e
zn_FoP^G+%>5{WSavtJJrGM5EG2KSvNHGdlTI&mcO*YO!}1{s$l$=$(jKfIJqCU^%-
zMcaE>tEhr)YQKUzq8xDdjQ+aU5N(!|d|dR_O*PkRse9#G!`?dv1MsjG%!Fk>!^iPrt5+F{Js(U!L&9a{Gg7);zWw
zaIs$|qr(&kX^+e;gW^lrZkfuw=nz0^rWg5-Z0wO1XDgeYu0v@#q0YhXR8}neZulW*
z=yfE{^YLNpuqGCv0`KCQUKlUQ#azh^K5Qz|?npOB99cX1>>2z^7V}dudG+scU?VKPPky^F3jLNYFS=l8T(VHa6G@UkC0Mc#``H+kwcB7446s$3qv
zn9tKoQ-CH2d8DY&2a?js<_%W>u3Dzi^gt0Ej4PFn66(0|b?okKc
z&J~Q@z1n*#lCA&?4k<|#POZRf%i-QfH&@Kg^1J!+oglIMykLscMOc(e;B^YND&!ypUDLMshS?M4@L6_Ugqdu4P4gT$
zLVnF=I?6v4Mc<&~1$t~VpN3a3oh|-a}kN2L;QuibKE2pu#S+2PwOHyE@
zWMiZ>e*qZUtmGb|=K8=N>Vrff?n2IjvKU6#t)
z$F4k*o;PB?0M-k?({%eVz@t}}&W}701E&nURChWbgVxEZa^HUpuvzqmv5FQqIDJkO
z>-#MPuWr0#N{`V4bT|4pMCts2zr@vp7g2oh$!{-01`aVGJ)W$*y{!N*HVL)Qy?O{J
zWjd)#Ly}=hNpFtI3t8}CiDjZ#QwDloV<;ZyRRcc%T5RsRy1+WXQGF{q4v3C*c@f&m
z1M0$Gql-Jo$oKdG$eAhuNo8L5N$&gyI9a5eY91#7?X9=Jo)x%(!DthLUC3G
zO+N(aGPz~Wy~=?H&T&eoi6I+<+DnEMci99fnod5LQP_rH88XDkEA
zOK3HS5acDqZ0f8DoslW3$w
zuN%F*GNYFHb@z7zV0eA#fjNEzadFb%cN8(r6U&x?MbAdC$
zNjhjuNT0X-~Y|0MW3kG)FsKJugWn#-b%w_u{$T{cuLo
z4;`m-T=6$u;iTJl|3i;$mdj0kyhP)yQ!iz`utZ-6g?0u!@x^(A>lw|B642adA2MPU
z-=a^7ZY-$>CZWCtANy0titssu>CFC%@#vdh3s$GDJi<@o+sh<#wB-7$FFkdtIM1)ym7)Di
zRBCBTP*pP+E#B|0ci8p9#TQM!e`;gHU&~OZB(GVZzGlq)|c;2Coe(YYek>_?4ORNkVLTO>q+C!
z&D$r{`y=r|dp#~185^9(ys14~k{%^K2vU8ua0PX8HyKJgxP*rWf4W^XH$}t{)NvxJ$|Li}!sptl!b*BZ&w=zJQaHxR4xxMTrrg00O7uB-y
zouWab709Qww0?on`iAdZ{oi3st5w4OK_?jOv1O`NtAH|5JAGG3sBo&mn}desouG(S
zHteEJC;08Ro25P22){q~P`*u84`v$oGq|2mpp@fjX5OZM-~*DMMZG%T;7iHlLXUv2
zfHaVL8%^(k$)aBj-vWO0!PU~puWOf3W|i5Dfa`x@Y_1W~UOxqRi6eMQA5DUymcUyN
zQaSL2XJOgbym(MvQ`SD_E3=SgtVk^9@(NURNUDl*JO(cfU@4LPNz68*m3RbU|W?v
zn3yzvufA~`+PuxsT-EyuiCC10=ee?AV`qO-2@(s&sQ(i_?;iv;$rX}i^&LP<)Z6RzR4csma@A;S;webHWi*umf`G~Ori#9c=@3t={+M&}&iBF>&gg4Y!Y={I
zMYSI)VesEYka4RV(0n-Vxhz)>&4;DRss7mkp-~gI2UMEi=Kj>UhXp@ekMj%Cwz~{`
z5BHDT{_SG%vv04{o0>qy_X$GAT{7?>tlh0zpAAg*8YTTZxCrmRG>szbsm6vBTEzBz
zynyWo9bzRlL$Gm|EdL#~6l5PHI5>{y11gvQgEfpKkf)R5)r-Vbm_S<4eY4pLRLfOf
zSGPt1R~GV;EnW_mFiW$%9nk<>YFGf-ok}=b(BAy(lq=lj7ro};Xb-G@WIZ^?YX!%8
z-t}EcHv<~)(na^|sG(sgTVi?bA8h_WEyeciATmWjIJFl08Ow?D`wyEhL1eCHUF2pw
zKvJtL?!2>EM!M8k_rEH(VtSi8&67-3$W(P4rL1!}QYqR*vut|>4p!LuUf?2vtr5Zw
z94C9@KQXSA7NG-}ZHCb7e-Y(~kIX;2_#{<8l35-b^K1|6>o@1=ZQ8`lg8nnBH{U>V
zt{lB=2fLWB8|OD(M@z7vLF1{$!Oq>~algCuQi>vl9=^>lp
zli1F`%zIpn<)g?8D0npknO`b+-K
zhlnVHy3F4vxyUCHG23J@V&v(r>JM9XX5{#S4_i$iC)Uq>FDa{829esnmQ}R<95H-<
zCV*J+CZfxqarbJXJTmtl>576nSe&w$b@`ePVlM?YKM%(ux1YIPr0>wfY;7XtO{Xl7
za{}t!UIdQVw^oCH1^>k&#_l&Pm}?6#A6|0rf#hmLNyTR9qcDzrCFXg0aJmCgsJ{+U
zxt6dG8AZ>?-7mrNeDx&X;g?ANm-!xF;X;h3;}TB>a~1MHfT#ERneSLp-2wLH^&ys*
z{^+D%dWJMn~@4?*iDq(h!zKS8u^@BxAj0lH1rJKrv$^NDNf%n
zB!)v5tqQj3$UsOP)x7%CEAXLGM6;m#U2ryZ;3CE2^T2MgPM6GX23wnQt9sbAj8T}V
z>+-IM0jri_y6H52$XfBnp)H>iDjpkTmxWLRx?IoVy#MInoo&;%`9UX_?6`7bVxyy>
z?Ww>rhMmVis0mSN(+z}*286`25z(M{E|Q%0YB_8ZDSUKT@f5@uatUQ$dI&g<>&@>+
zy23>wHPjqE@*d~w0IAoSrOdViz<;cGZ0GOCa6yP%No&;$SY9+8
zX;+Vc&lJj)O}>qQf8FTix7$8ovGU^XKYkBrQM$cEEgKB1idVPvPrZZ5Sw=5yy@$cy
zt@^#M9$|1>=JJ(PnlP|=>`wNbBNo0`eonY%mJUJ&QrU=h`T^C;r7_!vPT-u`AZJkC
z1o-OZ(xlACpd8wBXJL{U4Z2u6cGjHH=5YLqdj(t%{x45lmH>}=NMK(u`WMh8
z+;@qi8vr#<{Z0s3Gn^}FR}VPf0xk;3Y?LUi0z*Udb-}2Apih}p{*ulgm>1eXFe)nDBcB%+EH%$FTkKG1OQxxXZ3%5X>
zRB7|%{xPAmPdmG2Ls`%>GcSIx2wg%2Oh`peDOusCLWCa@c?6+YN5;5yGZWg(E&MIE
zp9^<5z84mtu7;;rX!J&LwP}McEDB`4G#C;O+8V3fBp^
z@tM*q!fIJAxMucahsr}2Jd32SkbXuHmA?5U&b5>cf3HP(K{QlAwTvR1Eej4I@
z7l)!1P1Mkyb=JL7Z!tW0)MMa1$1C)Oo|y*CuspgH%Nl8{Y=%q4yKVfT)5ELIY+iP+
z*2h<<##@8CX>n%l?k9uq*ieF9>Zoj!@6dcbZ74Kg7Df#aA=s;JK=k#<`6MbAuZ$j3
zxqR^g+WBvti17XfDCo}_AYWXAg9dZj4rHsKrGMpPql+Z&@;Iep){qttWN;bHk=_RN
ze1Tjt?`Gkk0v;6~`dg;|FkU_Q?R
zL|*4ZH?tmpeby3qt8GkhOuYzvvbhpbN%j`r1XmT_=xhMlqvy&gpQ}LHGtHd*@ORKJ
zT1z-uEC=Wk`V|KiCBce_E(PZ=UjX@OM$z+jbs&Pei_PnK68xv2WyGTz4tBKjEOSn{
z-pc{Yz3fklRRHn_!_v4aX#dad=HE3SFKe@oM7+uJ{K-K8zBDi
z?eyZoJ?QveXwORD5bU)pgsYBb!6>QOq9LM3P{+t-JVwD7ELFVM)zHy~x`s1x-x%aT
zYk#8MV1o~=pabD0QuiUe86Farr38*dqV{*9Bp{cPq~=QHS>VbRyY%X`5A4XI=Hbyk
z(Pbt7Da{$(1p!U*A5?baVGmE)9v=@1+taq|3|Ub~pWCVPtvV6TTMqp?YkD6v`%b3F
zHpsyoYQguy%cgMZxonC^kqID-QVZ|Yy#khW?#c=(Qo^GuIpxNTGoa*>y@68<5zI_A
zw-c&U1HVL?_nXq0!SFcXoc1AtTF~)Fo6;UK>S*Q8ExV2_zYClcF%kj+o}Z=U2QMLZ})P83L`#s9gW!Q
z2|movvJeq*8GKD|k%e*Y>Jye9#v_{C8zQLA4=i9{O3mll2dt1_Mz5VX9np~8J$!E-
zhrQCF^XMe;LkP*jrb6|4u@9rS{C{orVmrdpnVA+v$m_p~0|#8G*p3+qSJ!YX*5>23
z%uoLpA){)}PS={kG^nKL*|gsv8CljWQa?-(t_vf!MQ;0scBDS
zh*R6;^xa_W3LvX1jmt$EZ65xh9j`TdD&;PK>Gku>KDgsaTvV!5i
zvvV;8m%#6%9iOnH+>uS8gVu)Cz4^dN$ES#P1wE10vq+3K$zVXcssf4E6mwRY|Awp*
z@+?JTF^KYvZg}iY9!41Vq2$i5GQ{4hqWAfI76#6^No;l$AxtbA!|ZQ9
zVL-R$QH}Y@{?@(Ath!Cdm&6s_pRJSuN$t1e9grA64@3EC?#zr{_IC~9V2v$W@oVe)rp1F8>HZ8+I8P*GZs+u
zHh<%^{T$+8oT(VMzl3r4aHM;@+eVzj=@ZMZF~c(lYG0FCgu(kJ6}}sf_c0TWY*5p1
z7O<_HInViu7GlI3dA|yUL4ki-N_DL9)y|?b8&;j+IH>U#9Ppx=1Ee}sxfFgN7lDH9tSt=
zP1WshqJaA0gZPSbGLUGJUYX~b8jxf(%=Pndf#u{!99nod=t|4*^W%{LAqAOU315l9
z;2(L~52WId{*W^>P{17|rASB3%S8i9%DUJ}{fA(2wa)j!A2+xnbzxW_F#y=&5?J2l
zEU0kpDqFEFgnwN`e%m_u0V^AwO`_MK&@yA|cGk;yaBU~moVlfqLGpu5t>cv~Z1xDK`Z}LHS^~}H{_t|Z5pOo`eit1(5DfieB
zf0Ya_e-cY-u^$9dH-pdRjxNAy(z1Xzn&jwV=bv}7QyeJWtetH1jdeJi^SS0>LML!|
zLqV>6_Xkwi_4I1i+yTE9AL^NUB6tYZysW(OMU*kkJ?nxcDIRsDiB{PPM>#=3
z981aUHQSO$$J%P69Jn?icCPRjs~a&Y)z;)57et49Fe_3n|3gu4>?UET&5VxW6vb7l_N3SyM-M>s^jx#oFamtX`pn^`0+A-b6
z=-G?r5G+(FCeOk{HjVC|WqtG)#IfcEx`8T=Zfvnwznk;cBLCu1K
zh0V2-zGi07sNELd^})7}eSGm3j^w6D>Qj{TQ=O3jNi3RkMv(vc;S>C3=P5LRD|l(1X!;{J-}ZP~vY?f-9*)_>bs_8x6cVxa;YayLp3FcxbO9
zK{vx2be&VZz%k+h`n`*OMDUpruEnZxhJo%rD*o3*sx*QFSN){JqB>B74)M5!*9s(|
zH`>3fJ&yLm!(BS;=(P0lvWUvri<);)8!s)LGe`yM7$k!5yBDEO-+sheg!rKu|DH5@
zz6iwMWnVfPYWBftY6xB!(SJa7G|$N|DVL+|0i3B`Z9$;3TQpQle@_D6siIW+?8J;-
z6IlH@EEt2kFF3suFZRF@<~H6DQDan^+?R8FLk8cM6dskR5yLOhS_bZ8B@pjQ^
zhTS^4){Naj;5KDrO!aL~5dG;<=VK%pf?MRqBu59pPqhC!J4+dy|LPtWYhMPdCglr>
z=!$`sv7|vqd@+ondL<$`rvlRUOIdsM0gySIjMV60fP6I?1Eg$cz^rtxpVrzBf89tL4p_f;@7Nj?yJl?ZB#{(51%6AxFiC6o7lS%7q+_UNcaQBXV_?^u~5
z4YwXdsrWD`gV47t*X}Zw!ll8c?ph0b7o`arBp3gM`-Vum+Fl_XUvj}2^dOny7I^DGWeFB)afNi0q?IkYC&NVaBJM(fT8LLOZ#>uQr00B`Rx6m(axv}
zleM{2Dl7R3;UPD#WTT(Pq`a@RwJKA<%h3gQbjNLwiNeSaG}Wp<8xJIe~&OWz#1K<0`~MmD(j-hPTmra#aP-g<}0Cp_J{k+q5t
zl1oMXAhJMIvs1Qr=Rz^6+_DDE-Z;eH^lS4+?@DYchjAxqxC2R@nKP+<7KNC}5sSZk
zkb`~Cc(*|sU5r?B%IH^r@5Dgx6@}&pyO>>=|9@@|j<9;p2<0-28iaZ)3{I(Df#u;$
z0#|qCfqMQMP0w5hVBKi)}k(d#34dh+sUS~dKQ+9
z>}iOF-vG08?}N1b%|QN~%bl_h&ftgJe*~TOUhry1ZD65ED0u6z|70q?5bCtwxVTsO
z8fdw!u)m$Q1$SR(M*pj{hb1H=Wu1rr0W~waDQezVaBb?vWpVvxz|Uz^7Kh&l8(p=4
zrPdHyv*MH~CKkW|x#%q=?FU6uWjvY!xRCwHW1YeIB
z+j-{0q1rT!+qlyw@N$xmxzF7JkZG)Y3ZHrii)kLN(O>ogyWY+g{MTPX$3#9REzUO}
z@5e4#NN^kQO?MV;>Z=6YxJ&$}$Zt>}$bNCk;Rw`I9+sz9kf1N!q1r-jKS(~kQr*GV
z2-~dE8!DzSa8^*SuR?JU%w1+K%5H9ke0Ar{i_VUMK7Qlp0X2P4c)6wd@vjjuSK1i9
zS-b%=a-!akYafA1YZiaHuE0P^
zJR)VT9I%;?lGGGsgUOM1*Xp-0n8cTzO>y-n2oj-optSo3#7k%2DSW+xc6_JgGjZm@
z{Y^srwFM>6fQ6c$&k4=(*Tzp*B)9IM9cpslgvow`i}Bq$WU-|9u$f(L(X}h6)~y+(
zfhm1F{0iZ&IO>41%Uk7B3DTiw$%rpLC)AZ8-u#mQOy$Ur@u(iq%(0{PaK%
z&-~*e>}Nwq2ntu$>}YTgsX0M*G6s|?aG%0HQUNa*c%qRU@(88BYdE!G`4ccDJmW4X
zrN+}q51fO;+0o3W=ln7lfdn1uVB@`gm}Yx9~5o3W0u3~`zc*)d;hhH*wP^_ppc(jW(_S3~k7Q-i>qv*8v$p562bn`4
zuR^5eX2d3F)1rUpHJJ>38q=y57V_bOK{?~W**Bo;UX>ajXAXS3u}VW|xC+QN?sYUf
zz6YG~KLIOOK6L64%Tx%=1Tpe|6XxD0!2ol&m{q}kz$?j39SLdyo+f2RA@K?-6N_rs
z#XJWa9c4^&LmXhs)Tve3t{MzDB5_Y6i-6pt=)bN%pTj0~?b3JbUf^k1!salI1)v#q
z4(baY2NPnPXZh7LLF<@rnt@+3WUH`ScH@fy^0zkb_!x)5?hF_D%P(``B?+YU##tNK
zwEESt_KOL4Q*a@{+f^41SU+kUK7AWpT^zVPL6{CJJ{3wT1lYl-V*j~QZ}os=BY)9)
zi5jG4GfthZk_Cp(Uz}rl=>_+j-c^Sj*gywmZUW#e2e=*kO&;{#gn=~g?yu7BVAECV
z?v}*XFlIx7Z#VERjMRKYn(8VFGExI|Jw`L`G7ynV#^Tj
zHWW9}B`SiGhW^6f2~RVqB^}q{eG5#FSl3R7-T@x`RGMcV(1Vtr;qTRQNuaLWjgNs9
zM@SCCqR)8qNiYlwyw`s!fk)I|WuH`2gQiR=VP*9rtUEn&|6IZrl7Cq*TgPA?`?oGV
z{onW%F!M^nDrn?9@Z|U#SlG6NiL?+z=2DL!WYrJ9u8-6p$0hg4MEz;N%qU}d^ez!t
z?cBRa{c#4%8Pw3~+3iI*)V0cN?tMfUK32TB^F$nUZ5ht-vl9c;^GC0HE^T0u6I3i-
zbo0p9hk5^Omxi!M#Acerw*nAS_OaFnT{+k>eI}!*eJSEFiUX*1kw8=8vWDv
z9+s#2!!zUS9YpwtzFpGu=g3c|Y3YY{pRt!|smVHiR>)1fK*GXGO5dgqeVGUZ7zd2uy=YD&W^-*|{rM^9#G(%D
zaIP48KW72*w@TRNb0Z+8Ol;-2(E&8$Zq;m`Qv&lF#0rH%x{z>Vm@fvk1e0A+#>2P6
zA>j$H+q9hrc%OYA?PHgMBZM;lcxcZ6a-mbtn!l65n7)Da*ZCaqM(h&`PU$Mx`tDz0
z)08&!qhX1fQnvxCEMsmv1#a-&Kk&PQ
zp#u*Jfi4Q_M=or|V6>Jlv6}J|RM;!{YA!wq8rli3ZQ-&ae
zu!S1|(?xt)hyKaSv|{jub~=Cfn-}cydB3?^nF#X?Zq&S58V7r~fzD_ZCC;2>YvK^q
z0EY<95o{6H0)dOndj=9s@FQsWVO>5Aly2M{u{D|of4S;EIU0U~q#HB8hw?%|#^V~k
zb^UViAn5njcGW7RFyt=s)ucl`*7tLUJ6fS{Hgj_VMVV@t1kD&Z9Az)#e6ge?S3)m;b)7lA-OpslPlL5q#pw*7*xw)~Lzl-(4sm
z#gCKtMym)00GpC(dbtQCnlKs|;9SXzpMPhOvLR)OKm1}Lw$MR}#`r-jfshgpucW%t
zAIF0JXD_nI*UX2W|NMZvUfmPF^fJ$onB@wZ;TiWV^8p#I(lRM6AF+2b0Li-cRjH$+
zjC#vg9Nlo|*tKs)EoX5UgFKdZWkV(VGYcl(oahJj66+to-$S|Bg58A2{PE?BgEI2Z
z9MItMOePv}eUwD<=3BE;+`)l3aqI5~Z!p
zBaIe#Yue#kQmzLmXUvCle=US?=Acj`S*9zv|K>{;BmzZ8I}4X?ta0Iw)s*d$#82Zf
zoH-q(O20rw>xAmqy&3pkv?*&fz5%q<77;k|GT~K^J*tZbr(xJ#w$L5LA7JhKd#IT^
z3KtqnP%oZApu9Mj)BT4FeH15|L^H7sn&p~qIDZ_4+o!3&>RjvqmFu!q@mirkpw6TC
zGM^;sQ+}22c~37OJd@j`^zb+2uc|HTX*BW{-AKMh?(J(>q<=$V?PEMB?wezx3*3X;8ZM8bi9SI~@tw$Ck!k?i1L7W}6hkvh
zgYYx&vq4ELi;(HzQz-my2CJ&Dhrf~wau!bf*Z0^S`CFAqK(c~>qvNs}EF<59thg=|
z#?~}$C-Q5<*9>Z$UOv*`h&3{JmUpL@`hg5LySO>p^v$*r~$|=aPRu_
z_a^+?Lv)4^G<1|su$7rjI?sBnL~!PY~GYy~>xU-$iRqRoja{e@A_gUyv8Qwsq|FB;*~
z55vCi37p|#N(f=k@IOSt`IGWLV+#2Eb6~-XE6kux6F@&`H;g37!Q{4wrWzM<CTF
zy=*sy6uNFZ*3zGW7lzx`CLS#!>N<;5Db>ChbxWpo%->ezeTmq450@cqIgWjyhM5p-
zuI>tj#R$Sv6EYzs=iK3zQBlvsJHcRd>VwVCyg0~e0P_QdYC!Di&Z3FD66oIeuihch
z4hAwYfxEiyV2Px7m!&Nbp5wROa-ewwIt!)6T`22;#Jk4^Vqaz9L{t7o*IPEAEJq+e
zxk>}RI;78Q#=1b;vX`&Eet!&h`R{zqgwk*#EYc`OOcJC@^N~6JRe*gNZqehB4&aAs
z@h;;|G9+fXb7NDw3o&OId>x{)iapg@uPX?r0>&A~Dx_q0AYbpJL0hjkVA{9ZkB71n
z(#7)qzV@*feqZt0GN70T*GTTc>w^?{&Rt}@@#8u44`W=y8?7qvFvh9iVQ4q>wkQzx
za2f}UP3+^A)Ti*j){d{{eFq7a4(*eakg^BkybfCP8fO
z3bA+b4zx9MJmYSi1O4gfO*2YjLBVvu-O`dM$aSZ5Q;WD7u4>V~5-=}^uE_Q99P1oV
z+7}w%;`tgb(yA!ZP3D1`^~7^MKWd?G!Tg+~%N|_U(}{U?a1{@5&mFrOE`rh(*r*{;
z3ICs=^YG{54dA$Jij2sXy|Yz(?>T#9B%+e+oxMs*5<;>`_Q;4L*?jLg5<+ISY!Rg&
zWrWnvUvT%jd)+R&7oV^DCXk^u>$m@Lr&8c$top5myQk21MSkFELSt7dJL7++S;FEr^aI~PTcxzB8+Yc@m`8Ne;E&ar66s}
z*#>WTh7xHTe}U=Awct_pfAGO85?_srMVIz`@~0RgGs
zZr{+vFD?&kcpKT_6MNFLFVb94@^V@Fh($;IM~r-0?bnBRt3y1c>iNs4)PP$u+A5B^
zEWdl+tL%c8zDRuG{lg3|bK3pj$DWMaB`o>2UEr2gR9FE~z9E_RkBx(TnWoGMP=F2;i!Z4=@noN;L5obD;r(Z-NR)J{a7==ere)XFiPO}-JQ#ic-
zm~b5@FV)I9P3L&z!_BBab-^2Vo}^l{<_<(wR(jVh|d``nl~3vn9%JMe_c9vkp$x#YNmX%B34#7ff*Y%8!OAU$JJZG)H|O=w7AM
z(Z}a{PsD{sSHQ3ohKUF{F7(jR+P64Q2oH$99m?ixi;stqa`oSoL|pANl)!GC=|#yV%#;A;yW4noU*kl7EGL
zMBiIC?G4a#aogwd>-6|r%Y91&7Bajlpy$P@>SL%830N^zJ%;B?$5pJCk)dB^JbG^i
za^NK*eC<-|D`#_)=zpuj>3FXt9X8
z8l*4|S3kUb6JyW=-Z_tkmN)UBg0YdC_xZnrNpG*9&#Y7MucSd{Z&x3%DwMRks}7+O
zBMns(1tl(LN;@a?cpHwF@^Am883NzWE>`VfIAj!l6B4wS3AcU-j6J0y$ImcHC1vY&
zz#lxv;!YBAV9*ipH#4&qUb*Dd&hPjdl)bTjY|y#_9mMW(gzGl|vJHQ`$y>?5zbK2U
z{oXTZmtaW=M{|K-cgN8eGzcClpC9|Lz6|(Jx7~a0l?bn^yST*KJOtZM#57eGufQrF
z4zA+5_wb~4(D(sXUSdS;L!h5fv?lClt~0A+~hR)aFNAm5Gb9h+DlpuWRreeD?y
z1igty!XN%&UmsLUmDE|o3y(UK#s5mf$x*W|({rc6w=}QGT0=T`UMJAR`t||BFRCJG
zd&&r2l<8(x?EQz`EA;xB^X(9kYWR%`2JK+QanBQ3$hQz%i|_J}JxP?`HY(d6(=PhHCk3p|{^2SSY-!U@N9^rHMI}rLmQ7rCu
zwO9`|`Nz(c46J!(ElsQWDIm+dpKu~qc^oKq!IsciP_-~#w2q1Tc^om|c8Dlv+DG?4cj}Qax
zBFUF`SP}IXWr2f%r!bBX$!~?zr;*b+CS1*8{D?XG-OxSVRLs)p%Yo7BVr=5Sa|xnv
z>X1y2I|n@W@39aU){|B2
z1V4{8#7j3q(jE4ujiN9Ldy+rr^v>G{Dci5(A-Q)0JLirE$C_7R;acHOECkk&
z@BwY@b43A2u8-t+3h@>;S)6@;!^j;WTGu>ot4%{nJIAu%&MNkufUm}1io~M-yZnTf
zDITfi(B*V9Ovg$fD3@Oy1#UxhOHxZsg?xsIicd?f_g4EA1aRL%b>bZ$SSGaVjPt!L$iwvmdnP>+d
z?SqYNnuUyaVqOn#*8Wx*!{l!D77wpk07Jl@zKf
zx4``E@UXx=UpS96$vu+}1Fe*=hq)~CK{4l@X8+2o;QHORoCE9{Ox}2UciGzxDDAEO
zwM)4Te_ON!V^>~7a#fOpU$T4<=~kt>t9T7uk}m7{qu>O~M*d5BCiNIZ8|7E_nKppA
zw)b~sg}wlDbtZvj)d{d0mtMBpvJN9T%DoaF979cb9#*Cv^{TY;`Lk{cjRD!OOs5hh
z$KW6TxOAt(MUZ5IUe+n6!=DsiINnsmkJ2(^*2q8p041W)W7xrKa3kczv?6&TeCe+x
z<(J`d#C2BjC9E*xVsvl5$B2Xi`R0^`Y)}juyJpCi#v34ejlad*lTYAlgw!0_J~?Wn
z3jfQxUkhZZw0U3AwZb!DU-d@zKLTmv@h=suOOVhW2ulm7@lXF!VJFMS&#LvkXIft+W0E4O_c46}NYRozR7!TjXST_|&JjlKcvI$$M`|w{iv7D;A#4h@0bb6hB4k*;S?(>UWbN_
z>&cPuwW1-BX59?eaa3Tgi+_AK5YO9>Qpyeq!>w+vkaEiSp-zh1cNMO@LElB^GNvpx
zprctvf8{3Z(87TGVyyqn(7g{A<{dij;-q88)XGX8;i;1>k#|_rP%2etvjZk0TzRvp
ztSz1befFI7gsw&a+Q09gp`z=8*XM+nKH`hT%N02cZH&*OSss^_S2XN!qtS$=z`5%<
z>HBYptcWu@JTsIS#Tbq|_vC0CI6?f))yN-n+8F+pd3bu3iimf-%b+_JQHzITHjiy|
zEASh|0bev(>+z^_Wj7Zy>+pJ2uDM^YU*RRcqp!La9D(>Z5@tqcZ{xGmksb~Eg}60G
zUD5gd=lGYvMb+dc88jfXUMV1_p_=G>N{Q4`C-NdZ;ajAHhH3r8#|LL
z7_r-;>fy(k+Gi_qiv>E)80%Yjy3#G4FM)RGWMtK^sx1S2$!|^efY%X?RTixXe-MaY
zdChIAPVayoX{`cNxMk4$WFLe#mRp8tbm)w>yKIOOQJu%1>9HMAj9)f
z^oEt}NmMbY;nJw06H262%OHJVgr@%)KFpxu!#6=hx2Vnubb67NeeLA}yvMv(;XdGk
zzIZz8ryYc%R|`@K+y^Cb+Geio2xCrE(E9K+mGK}{E4?f%Af<_x-{5>yGtYq2j`G*#
zY*OI;BzltCA4$+G?~&ODCfo3d-1sNXaUuL8`Zzgnj1*ta@^*i&_Y0)+4^A+YPs13L
zy~-s19W+x4x%mC4=a-(@s%i4)5BM(li_>9a0V;P6O<(@h0p9Qa=Xq?Y6{bLshi
zNbW5x$5Wabx=DNod|c
zgCK;XqBjY+2Qm3OJ34|F7NX_7gAi~D(za=?C&HmGl*JVvZNZODJez*@2{=`2VVrQ3
z%WfMbY6MEYhCfbG8I1FJ1KKH@#^DQgU`sY?dyFy+vh6XhX8zX)mpUmGO7
zcq@0c(iO;jn>x)W@dB{DNWyfOBS5_L0`1v4Cz#6KIrI7IRq&tK>os0eF<7k^lw`sg
z2GX8&$1JirLe=3qH@a31koM*0JI-TgK>7Lb^oSGy*_Pk=^qdZX;p%Fl|tfg#MZ3~;*54OsFsRR?uN;3s)c=>lGthr
zq+-tQw6}aTa&2Bdc%}0()~68}|C}!jiCh)(5lwuAb@;KKt>R_H23_Co4EUWwSo+jY
zZrGf`&hKe?22`pcHJ333rb#y}mHTVOa>){Lyiw5dyYwM}DNf|w8C!D9@FUlcjVv+b
zfvjCluaO4wWm}p~L|*QLbl;-wF*_3H?VrOq_bo~6Xp
z-`>e{NB-{BueOg4bbb<3QqnDYzlb0ls9+Acw+3?FmNc1>{iv2M98|EBtAd>N%(zKv
z5scmUr=GmuI)q&Lw3SXNEQzEK%Su7K&yu?OV*YN^#F=c(55z>l?RI7A3-1&<{
z%}Q3q-Zp_R=NDRaZ7zd4r-Igg08u_`>sIKUqJYEij3Py^C`4<*+Qq@hrJRtSboVp}8}
zRqyO78Iap=OMnrkP;B_K#-x=4(3~@}`~C0hv0g*Nvi7yCikf3iR-McBknQF14uPaK
zNXK+Bp^v`|+?_Au3e{6VeFY+)zeX>WR7dlLKff?`C#Y$TZmC-E5Y@26^E1=Gq(R#pq7f!qw84Mlgz$>X4b_17x
zgXbFALe*#G@hyv=b5_OUK;{ZCDpFxavtM@p3L-gk6
zWbOCxiFVpHr@e4oZSbDZpafA5@nd_Jl}vHpiJae?d75~5%DZ_@Rwb0-d3_mcx-PnYej|@hSqH6^{60d+
zJ&&eYZ4-jJ#n9TTI-9jIGPn)1#)^y*0k6#LS2yxg!Y?zBP1jEe;xz+WJ=0svxUv?D
z;u@tm`eg2Epf19SI{u@z-Z5gw2Nm43hO#;F>zDk@c`TW52bEi%>2GZC5`*Eq{+Flm
z4uj;^#`c!z2zI`G_r@*szppCqGM#Rtj1wW(|9h*6|0D78UE~$Pt0VuTh}|+o-&bqh
zWQnyzcQe;lX;!RIU32lqUmwI!XHIPJw#`29w%D*VCzrv!bDu>Y@&MeugfXmXR~^qN
z+lrZ@IE6b)2l{aTd<&QkPL0u2v!P4UwYMKRh@c<;4SoEYC4*kPe*C_UIwPKBiB&qP
z{()uDwBFE$18se!b7PxY1Z@^}vQZZoM8n?~u{N08MgMzO^v$*38=V-T&HpiPhkI}}
zWOXa*qd_c6p^Sg*Q2MqCxzpvg=*7IrI*(~aRJoR7N3>5C@9E`YZ)*nVK^$$j|GF%y
z*Y=@5wpHL?vUB1+_HTJyvl}4BgYZ6W(OYg_rWnaGj$%1j9Ox
z@vW02sI%uIGm$MC+-+3vZQe8vs9-%bBx
z>~r}M_s#u=;!-2vtxLNbOj-&VQfqB~vV;R64WWxYPMu(;i*^9!RD(d3L8DapbeNQ~
zy|nf-0L=32CF=bQ247rX-Bxg^hT_0aT0g5CE_WiQW=}={K3OHh$t*3{ajV$1sG=8rVKAV1iCC_d#R6XVE>2v
zZj0i&AY4XO_&CD_U_;KNun^!6TI8nJHFs^`2_ZSZ^NPp8^DXI`ysEHvmX;N+dC+NXH1WkXPTbz7)fIK%Zx3nNiz$c|Ep4RwZq=x@@`>5(HCRtl*MN0D%lQ{od
z<8V&_k~s)h=-BNeLq`NO}0hLMJHDsvqm2tUw{KX6!ijXkhRR#_w+L@FdJ8icrdu-eJ#`U}hhh-8Ek
zTde6c_9cKtSjT-3N$HERG5nr`{gpiNUC;6kVyqFysB$Y4p<0H)eff*XB3(lpyH5gk
z_V3ehlcH$sj*bMShVWygRY!4dN-P1}mmSnw7@oqab|!{*JRf1^tU7PXVHC2Vp#GKo
za~Kx+RF+0AI2^gb+s)c-JA-^}JtX(%sYFuPIiKDfO~rPCk0&fjKS8FlJdZ~Xg-PMOFu3t}!U^h=AA@|Q}U<9tW@ef6xV1J_(PfQc#k>zKXWPEQ*B8#?C
zIhoSpN0vsX^HvM;h+HY$sJM#u)JhK!=)PT1_<9kso
z=6R9v#6%<}U;RmsHq8_xOR%H!^i@IJ)R{Y$f@HC|TJnsv3qnW*-RYZC-i?H$6TkB5
ziYkZ^-YmRhjl+c9Ejbb8nGNFE;v2RPWC931`VpT?=pPVRzF|Xh6GTGrL{n6*`v9Tf
zp$r9cZ9Ne-?x5@>i$pIYE(WSUrbOTVc{8??EkvSWm`KMoPF!z%9HsfXh{)b=^@eEl
zj1cd8JT%VKj9B&5^Q^)|1o1!V>LS(NP(q6$l}KJ*I+1sc_UWv24RL<+%Oysa8iHR}
zbu16}5MjJ1blAVApYYT7)$59Tp9r!%%Z5tc?+F(yMQI{)xUe!F-;*f}oJhH*?)8Wu
z0W9la%#gEO135UJ`X@Eh8N0<*7I|`XneaGy*70NCPeOyPY;)5%DRysf`R3hvS>!L%
zee$QQ*N#*vDt^|s4+y>kWuw`hy@xrt(km(
zN1V}q3qN1@MELz^HTcBN7;z%Y%0abg1;^ANvR?KP3BA`IxA!R
zRp=S#a~g=^RCj+(w>_qI?V7XbQXYaVZINP3vWW0nkHj@G8O-3cuKRcV9MUczK3QF_
zkG=1Sw+x#q!gw>;Z?wT3gj?@v{QbBs?CCg*rQLlJ(9p6jwW@mtej*=g`MzKX?{Z$m
zSiE|Ws^#L(5&Um3wXds2`Id3W67$(X^1bhfvtZ(4VC^YLMoh}OR8WrjO$v6+NaZ8`
zBzgtCxtTK;^1jWr?Rp{^i%P%|5FykazADyHTIsU^ABB^?b)5lE_Y0EQnJBl5;${
z$-?L0trH2pOP@_ZU3zu1VXz%c*;3Lt(2N38`~oa*d8z^DEAy8#ce%l@YOLNn>=X>r
zotF!;WB`Sm+0#bsJTNVDAeV_0ER08{vox8>Xdh?*K2j?r7*a(0sD6!t2Q?tSFYXyD
zi3_OSI$8Ds8bcd8-Kc=uCZM81`;p~TIvj5`e=bcZf&)|c{WVwoAfl3OZ1%zjm^;xP
zW*vu?lvyBbwG9k%`4!BrIf(B2HyA4D=
z2VN(B-(_
z7~cs>%++X+x_3ZENlbR%{R_C(aQZQuWdzK%Z|>rAX*u$a4}J6Reg|#ku^M^26|Ua7
zXym3d4ks#wKCJJ_q4U@K<2tqF(Sha4g~bwnl&#~1+0r2y>LX_1^X={tGb{ZB6OY9o
z&~TWc9Zqu^4gKlLH9^USe(@}3j!2_HV`%-@#+CQMrR1nCN^}F>i8wu$Z?*@Mt|eEE
z+}MKY>shX6Jr?0$9HZC0%putHq+#eWTmrhUwrJ!`z5>2Yd)w-3GoajiVE+Vv7f}8r
z{NQIs9bm1YQWLy94q1^bz83FRp#52NVMu-+SSOxUjtChA8$Kfg*Wx<>Lz=AHO8yFX
z>(JzMy>=1WxX13E-e`j+0@rlu%2%NDJN$9$)6@7Vr)#`~zCFUb6&GmT<)skn$c^@`HHVaR!Hro#9P#H&&F!>46+R=v?hjjom
z<-Ogynqr`SNCrIxMxgejp7MC)67*^p%m_NM1ch&XSdAM!jjsup*>R)_;88qThTl0;
z!PHj#lSkRxuq|G3UFI1h?jsUh67=;n9_$hk=d4VI%gP@`7bFtEd%sRU%9(jc!T;Q~
zKyd*As$bihF&iNBH$}E$*&28eKi{SA(gN&KBn^j}8bS5PG|TqST|g*W;#6105X_!s
zQ<)N)gg(*o91(quP}n;3)caqrVEFEx508{uV2zaMNng?x7&+o2Zro4~ee52yvTl}v
zNmmioZ>SW!fn-Z+l+=J(fBw^<@0oEQ&t!JtgFleF*M|G_t08cX^Wp<(m0GaXxq_-6
zF`EZ-682|*EkfgVvQL&B&p>xajpmSEBgktJTGtFqfGAypkq>t;;HAv67v?kAu-
zx3*wO+Y*>hn+!{j=t5e`XqIrf1`tga607g`7EtPkGiiKOhhN5`O5@FfVCbvY^V~K!
zk9Z%q$EDbcfc6UGXZ>#_fNSsFhontEsAl5nn?7p@C0R0k>&yHAQ~cz~sO=blels5G
z&vORjaU^E>b9Rs?9wlxj>VfixiHfiu89+xh*;TjFfqi{cQbJp)8(I+%=a|`mU*T90efYVO2tJV
zMa*{yTi!Cr8tUD_G*9`3Pu7aUeZc|o#k1t#_2nm3AL-hVU`hAkyr~h4`QX!yez8R)
zG%WVf3obz*=41~}3U6XjF$?D}O%GtCGA{IFLZ1*ovO|_NGmf>qu%veUECtE^N6_Zc
zWo+@pX#Lfg6{P&o%H~Le_!Ge_R&!#ckxv?q7s_cxOROSk;(tD=v7sUSmwS86L3z&
z;<>W(=xClGMWr02$j?yh%^0;83HdOhR7{y*y8Q&95A>JyDSU)AT^FNLkjb2ziCg0BLTvP*&S!kBDpD5q#!~T
z8(1NlJ)@F9ZuM!AMYu^}{bv4mNd87*;uKfZepz3^{1EWLxeY-O*6hx~R
zGW@}uNORz&x6_V;M8|+qQ^S#!o$rG
zVplrFpEC!&g#Fz2T%|sriE@E712A6#3)=-tMY8XS=A@hEA34SdXUowuk{|ks$oD&=
zH=_p#cLaNd?g+{tGRmso>{ovu`QlaQUklF=yNsoy&S9SjVPn^|X;VKD%}r00ycBsw
zSfG!1E92fppbIk@TRZojz_$CnUfZ{e&^25rUSd~4IDU!XIJ9@0Xn3vgKYF!Kgr8%bLd3}A!>^GtX;`7Vn;aPW&PPU;<1D=wW!4k;!84WNfyp}
z;-kpJ{i9Mk(X@pt2HnpjJ}f**H(6am_$(w9mUJbVNMl}T<=9hBs10t(YEG>ofB=O!
zLx+5Vl2*{?7hHXWw|%{LlMDL@o-_&l_jo%90Tsg1a(hSq_=o-!QaMG0O&Zpy$cYZ(
zu!HzJEoF@0*PY6fxW7f*QybZ1&t}C^XQRWjeh>*7x9HYJ)jA1Gf41y|G&=|?CJ9dx
zYu^z2x^hl!vXWq~?T0#3+qVh)LviX4*~*C{jAy>sXf+V^W}aI;dl*eP`0#lza{3eD
zzU15gDk*9SmV+Ngnqmrxx-U9Tm=(qnZ1t~|79TCOI4UkCw@|ecEkq3(NJ)kW&Z-|0
zd}+D}>mEOpTJFCm(B1WG6KWkLOurs^z^{T42@J3EBJQeT{TH4|Q>~pxN;j+5H)B;X
zPniS$S0UFCUUG}~rL{>IiN|}h_dM(v9THeTBFTzWCfKPI_MO2Po7%)VS5=W7U$4KL
z=PzTgc!b(6>g*6?i)xK6Ykv{hdmi-LQXdkw1wNacpW?yoM-D5!RHDsSxl`pIb+A55C;<47b80l7*hdJy*j@3E3~F>FkiH#eSBVXGdIm!8ZSDk4nU=H+v{4
zUmLliQfZ6^i(}ocDppS|n2=Gj^uH{!ZixG%v^$nYg;<>!T>ErO6}vC11l*{kkUHwk
zM>3WY*oPQB5`QrhOrfpg{LHN^gjz=Bn9Q&avdX<{pSGic@yV|ZwY@Y%o_==)cOx9J
zMz@Exg3KKD^G>W|
zA06+mwE`?mj-)4(H3FYK8a*BlhJeDQ+JQ;(BmVg_9iMaOej%C%VuBBY*}zyEAM-PQ
zO~@G2Zm@L58YZJ?qbCPFtd5TxD6?EfM1P?6Exz>d*`IVZ`+Wp>ECfh0M4JG8+oOuY
z0S_45I8U;Hu)T=lK@;dv>44Yi^m)Mq*|U;Ys8
zvlr%zQP;vx#s@Yxh4tafA9tQPp3;PKLCYT;UjzW03se93=st)Gp{Bj7oC9VbSTJce
zXdLzM^{uFmE5V5~x}Kt0_6f6s!1REGd>9!E;Q{bS2sY*Y|$|T(pcb@
z!ga;s;&Y%v=DIm|q8euWCu6qQ+78cf-okxZI^e4}iQKaO^-%MoCw$dd3~#Q!=XB%C
z20F9j_K18T_@-djC-Eg4e5FAD1YYVHt#%Ir;a*0>5$z-xY(29+dprrs_n67Y
z&_qDhvW`9Z79U7EzWLjM#RC3EX1FvYakRdUHd!1$6$q6`25q-y5};$LP(~j~1Z-eQ
zunDURgsZDRHGWcOf(Il9sqOp8K*?$24RnrxHK%g`X9}
zg-Z$7KJs|M;gUz&evMTiy@2tnTSg0DNtJ;SM|_5>Y^>G4m4d*StIZGn1tQ|$^!>wy
zQ;#50Vl7fgOn_m+m(B${rb53&eO9?=*P*l8CfY+K3%0LEDo{7t0O>A9ad{nQ(CMGM
z!@J`LUVqyVZF?FBCBKa`C(0N@a>r=7*y%^$mgC&%UeXjGcXG47cRd9>j(F^5@$4<+
zGJli0z1+g@R+VgXb@8MS^wTl;tyxDKO}^gvl$%qaI2YLVKm=9!wXucNchS2mP)4
zrM93q{QEoa)|Xl@@O3Wy?T%kKNaGcYLAQN@OZL@|T@}}Xi{Qz~KvgI3qu=gCX7n>q
zlolk#mFf*hK8I~kZM#BM3j1N6uV!#+zS_FP)&XWDq}7wYz=5cqN?fI7DQKrQ{-aVB
z20s}&SkcVeLk@(|=73)XoOQLJ{O1@0ZrZAwta|Go^->BSwhBAIsKKN!)dV#l@opf;
z;C3<+_G;JXjh-Kb>#^VRCXC@2-E#2r08P;8T*#&&MglE~MwNL#$sxh2IGx1L6*TO5
z)4ZND0=6S*_2OjbU~pUM*$?6WVZZmY<$Iz@fGgAamN%(^9p>R&|=Keo|FzK>=C^)X7gUok9xGhzpsc=a*I4(UP)`YR+
zbE@e_r=XVI6N&b#h1iT@j12|3AF{@B%vwA?2N6A8Q1^GP4|$teN}p5l68o4rH-10a
z3KQZ#ePp-1gXBA2`rQ!n7}FE!wU16a^2oRPJ}0A0M-~d>F6M=MBR6IFUqlrJV6mJ@
zDdyWD$RG(Do!IXoB*IYm{^~>+!u9T7OG)VyY~vpV;2L^_nAnCvFnb?UwK}jj+89KF
zw|$Qr+uT7C16Jll8pE*9uMFaFxqFBX@yaQNGxso7p0Uep8p(*`EA6l8vbV9T&tAp_
zkt-pB!w!=g4+N0BU+rFNU-dEZ1-_!9M-do>xro&tX$|D7q;+YAhXgX>82YckP6CT<
zJgLw9QXcV4IjB8d9fI_~nDV>g$BVH)nWduP5ySZJ$y`&;6GEPKS_;1N6~o5e0(ABD
zA7OOHA9!wEy@>UR^BQ867m&MI^-c#%G8j$Y8|Cs45#+`TQPZ$@_b@hJ)zwr+SuE|4z&~i4b8B>
z?r$W-Cp>Ntvj0dlHk(OgOg2)=d53Po{ji>@AIF9W
z`s$R32*&`yw%}&`!cqK)j;V;#)i0l*${HknwYitz(yFBcpZ6201FNR84&D;B=0kO5
zBnJq7`T9#A+9QbP%-u#U28Ri}2ZbUzCr1eP2-OCyUc-bs8~)kX@khSOT;{|-=c0(B
z>>>JsP(ZMe3$or~DxVKe?^Q
zcM!iw%nqh(brAVpX3dv9?jll8h3jVw%MjjqKjKT1PbC`We^=DKTts|N8e_m}piU5P
z`}WZ>v6HBF&du1@XoPUm)vG7$iuPKZmk$?tNU|y^%7|MnA^=|x(GouRl?WED~XoO^c-tP@mpV1
z?=F{J6Vb}wtXuYNJCXC!k9ezhtwhtsl9j5w4@An;J6tZD?}*~&>(5#J>WR5;MCNHU
zI*7RONSJeGD=|*#L(C|9FOkBt`y2`yiG>eK#Tl_AV!h6_j{TlwVn`kd_bY>1B3D5x
z$n;+!xbP&`zTR0RZXdT|tfl=ynEjN`{k4t?bJP+M=hqiQsJF4R!Pz~8iNKo0z@z-y
z^3^;=>fliVZ+>b+acv*bTJ!0m!LK&rzutf(Xw*uq4P3}N|GS!~6=yRyeXg1qL(>p_
zX{nHqf03h{eewfQ_JgP9wr2~GcI;>=mC{aJO%Uz+=R_oaB^f2$dEHCQzWcp8=UNnT
zP^`D`tjF^^U
z!v^+C4bd6C&+W=ja>V10VYqy3j=1q(e9(jqMx?moG1Hv*g^*JJzTNmXC&su#S%OST
zB3W(k^Eafr39HwqS**BPi4jfUSwr3@;u&(~+F;=m$eT&>iI@_xqeDDEmg2HOq_i3*
zn41j{R?U6=teR$tx{>_hfyqCKl^drB9e!#^$i>wG+c$+s!9wi5!|6mU{&NAECj0=o
z{^3o&gK0A&ah6c{?JXG~Jj#2q^1vKX-67eLaB;z+kKV$@kPm`>bl;t>c!oXCzX`fL
z5|G*Dc1%ju7^_jeL^of45vdo!ubj&^!rp1K+4DO)AR`(Yd23(GF@^Oy#N_=&gaJ?&
zKdw^4a@-Y;Uk+7ACV~!zV>ETJ*v`S5+mR*6%CVUa28}D2*K5{(`R^)-I%S^;jgc&t
z_3FW|x^f##Pe@^vT%ZxFOuG^lU3wgx$FmrXk37a%Bfmajzo~(P%&x;C@f@6cedUDb
z6FulY>-zN$7{MHCg_D?e=aCz$C$&xG$C1m*Psn_STd=O5f3Cl=Is>4fr^@*E;SZFN3E-|jqC;i4I
z)jcOPf5}0&;*J|$IV^x(HCc~Q`wUP|dv^=*v;sy}hf1c!{~;qwmtS$qlYz@Occm5r
z7@@7qc?0hVemK5=_gZqgCX|HkN?ILjn8@!6@5b6?tkw3P`^dH9ATInn$4DPHXu8~&
z*h#4jJSd&Qz4Z1Fjr-Pp%Z}vG*8a+K6MhzG~X5=t0V8C&{vd61?f<
zMUr^=7+9VZu}aL419!-e(qaN4;QRfg!=h0Ra5yEN?Klo#nwYc&Jb4Z#_#044A90Hu
zH`_{O*ImH4$e8qofd!DC!yTG#Tn4(y=C1#2Sb+1rFgtZCOL*t$w@)t~y1{}MFBHpn
zJYcciaoNc>L6|VCg#lYMw0~GSlnI3mV0L4#-7%mAlz(jb#;p$5V=pwnt
zY;H7REpgc^k>ji|=*1o!WZ;K6WBF=zh%mgUs%0P9&H=hUWnxm9o5-Qp%PVXpB5>%;
z3GoHy3y@EbB#EL<1(tEE{fhbI0OY=y`^Gp2z)bi2nbKF@aGGK9&0ii1m|f{vtea*B
zA80hyx2||XE49kjPqGfs{^w&v@hSmsQu6gbEIR>*0tXVDyZPYN=7=(xx2mA&&n1JO
z>T1C5M}h5Fpa}3I+0-pYj)AwgsyQrMPJ-`61>#nk+cT-s(WuWWv3xmIcd7;zW
z>~*HeKbY0)Z`YI(D37?4*Vgl0QD|)`|KvJy0?s@tFX2e#27xDu1`W29U{2wuCdfMh
zerN|}n5{V+@y_!bW~~e$8s78x;E4=)X34TLCcKAue_nc*b@MmYgrDp%*c5_~dZR0;
zqK||3(Soei9@HSwOI=-ZZXHt_dYaa$R)@77EEKEZOu&pc^W;#@Jkn?Kl4bVF8pglD
zdoPo(1<|m6o|t0!7C9O0wXG~pd*so%K-*cej72&Pb$&R?slPzoJ}_M@!bWz~<&QO#
zV{eP6O6^WC06o9#6=&9e$X9dkQF_f0tg^htgxeh_#o@g8LI&<}>xPBwDW}0cJm@|n!`FTBu#ut!eTH0OFs-Tq%c&+D9BzLmu~oDYk%o;+m#HEm4zFpv^&JG?3ZWm
z{q;EdtxY69y|#D@LoD3-4SyITIkEhIE*w%|58F?kSF2;do{3dRmN$wbX%A`xxGS6y
zwfLKLc(fXpmE)DdSGz@&Bw?*F`p1COr%v{|DDq)9r^Ymj_-|vMI()b0OWBa?mY187
zUBwWX)9e#OFO2;XU~u=!5<=9nGlNP0C+y4Pxq80;rA485(;}59QHivYrLk2cMJhXm
zv{1H6CE2o+M5JtGPob=3cdgkfA|WY-*DE9yQY61w?)~G}B{aQoNjlVCCBbN#7fD?~(!^)|s!^!aJ579}i+6QV
z`E>KIT2!G@ReE>b`*ca=(FZDi_gb5k;yYu`mmFy?a$+BK7&%m>q_%6Bf9~wIqUeHK
z;))ug%8Pj{O!stFD!p=A_{=x_Nl|GIU()f_FO&{f7=?_jk5SSRU1Zq8`A(@qE98h#
z;2otza~ZS1%rYgRBO3(+c2p}(JNZ~D;AewU{+g
zZfjd4G%sY$eU2(6&b8N0&p(%^c_AM)Ikp6ua^ij#X@wD0qLP|zA*RF%IfPCYTN=poP768F)Pq806O
zFTPGoC>nk`U$jrTrfAjj*T!Z%WkstRFU-+8S5ai;m*@QDX<^Znw30ukNoLXCA4dkG
z?_C6ZPpuk5RGlf8QNb|_KUHdGwYvDs-v%JiHz+r)=QOn_`uI@X79Zf;7;XGE_8Q=R
zmoI%zS`UnzQ;c$wGNyRWZ)xs{JPN*iOsw(rDxtVPf9-<5E2Mgx1I%y!)S#OF+a5Z1
zuRb{DEj#nVQcLRcJfVOqRSi_iVXESTS`Ec5xq+um^F9@C5^oUm*B&H4lx~e)=L3Xi
zHSDYv?V!{rOxJB`sGtQ#4dEin>-q-f56I4YF
z=ixg$4pZ-2PmYS8bpZEvj)e*F*8;inJB7k;i$FwNmY$llD-b%dWnb(5!ytI^9x6i^WU
zAHl!7_t2s->WE>$K0C-aH0_m3-!P!I@z1!GX@C
z;%cxXa`V2fEBPQ#uJb^#Mmca(4Iel$o(;-)r0!$H*Q`c;lBeab-a
zpmmn6H}zKTrbrto08_VTHI@nFf|g%b?q^k|gGM)xbGnb>z&Jc3TrCO%;PvvQ)+5^P0e2~|gIyrxBU=nQA{#I1yvzqzM#um7s^x+0nU|8Arb9iCQ=)Ku2@r3TRGm>n
zfpLLJrK8JB!NO@4C+oJCg4KaXdfvf)4{82s-R)0-lS12^cKDb=mW00obLqip71S0Eo_$$#AJC0Kd!{W|W;S2G8=8YwSLifQCXdT^ZAG
zAbGg$_MDrs;9Y;hA}*B_aG|lAqb)lND7(#8dsp=o+}gcu)#R7Q;O&yx>(3@cyLR_C@ltSU_8t{+pHiT(wBx7Gj#BWm
zolEQ+XDL|Vl{<&`atScH4gA)AF9zJ}uNZ&#CnoWBLQf5b
z?D!iCHYJ%Vy|F(5xb(}MB1he+AUTD`(9(M!}q<0m&b`%J;^XM0Qy{jPwPK;;w8Qyu`g
z=hdvuGXs)KN{*eG4**sMRa$#Q7cE^R6xCDa+5$670>m#s(nEf_0C%A
zAfHVov?S{EP_&Cy&%NqneSkaOc1Rpl=k%KPmZb=&xpy`w}mCEtH7r!wpr
zCGw5WIaRNQN?m;U;$^OSYQLMCm`6!Hl~S=K&rznHlAmjJ@9pYT%DZd&_7#O$RR7xB
z!oM^NsQY)MGso{2Q==0LZ@wukriA2wc2_6lQ`cW{tK0JzQ|rEJe4Tl>l-jlR-Hb1l
zrPQ5aAI?bQQmTAb^^z33O6uWG=MU;vDk;eU*DbFGDyX>k?N<`x%P9`GJ9}OfmQho~
zQI8hQucjuSt&d)}pqlDbN;ZpkdPW`U*)ps*y^2cv@-kuhhYIT2emCm<)_Q7h>3X@)
zygG`Z7mgA!md=%tfnH`O5`x@g~cvy|Gjc%j$K8%31RfP?nVqfe;0C5nYNq%x^9E;ctUdMc>O
z9Y({`c9m1d<(kB7t4pZLqEoL@))r7xx0f7xT9Qi*6#|L+4^`CZFB;#&mRC_z@7^7{
z4Ckd>>bV1##NmIZx~i!_<6b@$XRLPk+fg$hvScv%`@@}->%$p^eP`j{{eE*{`|7M!
zAfw00WlHBDHT#dJ`LYEeaHZQ=)@;5>U3cyr2~+W+xX(^|R3C2yYIkhR+_lOD{H`AR
zWxhU>(pd3*isx7wr5pGBm`C7U%2u)BqRoGrAWXA6Gi)#j%v(KZXqlBuQE%Ezv%IsZ
zff4uz;0*3l#+FCJR|#C9q*B+Lx-akqQr|k3x(4J^%Ti}PPE~tCT{oEF;8m7FEw>rc
z>GunzUU;-M9<(>2LO2iYvs!+LS{@XjBXsHK8U+j0+G
zqy9?|NO-H|0^+mYU43Tc10H&(MlJ5Y2_A`y^Eb=h1d5A~Df!Ox0sZoUt0SfdgM+4z
z{)qL)fR3?sFRna#1e!%cc|S_!02kR4gSX?Kf<)UDW`)fjp#Ro14z1`2a3IjqSLYqxqit&^j^20^x16qch|Ll
zu7>&S@sQtWkeCnE%eSRgdD54<@VjTG(Vbi1S={Y~>~ncQswCBCy-q3UsK0&YtF#Yr
zE83vywk90Zlv)hw2W5kS#q$>Jeo+eYAB7y-n^*?CKDIXmx;_O?!g}qdo`qn)Z)oZh
zsub)oKKQ{pxD4zoJ-Q)WyBrK8^#wWlL%xyp>6P8ALcsc6kf0y@MSes4lVe5Rp!D0EuP3^KDWTs7Pwkvl4&KHs?@65~1FDZW
z)P}Pl{}9Pkv(SACz`3(tYh*G&K`PhOQ~7dG^3qanxpW1nQMTg`kE#UheZNI{2P?tx
zl|2ikw^xEwC;d;9i&TK<>O;c@-ztF8WX;k|;V_=q$~VVuRf2}{5v7e+Dj{!6lPU|V
z1iQNnezQ8OfD4#EreInP)KVmk3f9#Co{I?*{Hz*qD@aU1T0WFIfBK^~clj}@Nhy8S
z_xF#%?h`EI%Kb&)__9S&8mEhZ(H|wdtbh#gu}vUSU*!q7)juhv0{O$-Ee2uA*UCV?
zp{T;v1jwVNK;^C#mEc>u@E6Tj%z@46Le8s_fie2DVTbl=h
z*Vh7_9YryNr>elnXbd%PXC*ML`}n%rxDYgZIg8a==K+2m4LMuo4d8(6@6LVP2~_Re
zvvEUB(bV1|w3vx`N)kn6J
z0aLy?f#(IvKw@|!x4^DK@Jm1?*i}0V>`3S{|9deV{64(*_W9}M;AX;$+h#M$KtZFQ
zbv92auwBydUEoU*NUM$0H9nsQeh;?siq}_y9vv6U_Q^7!>+8`aEm;DDx9Gm!8vYc>
zjfq(w9>@fI9Vb6&9Sx$avo8+((lVkJ22kv4S{JFO{WF&4Y&%D(#K=pZ
z_1i}J+akfvuxon$!U2?iu3p^H&~R$Og2k2js`+6152vw{HW}dXIsT*fzTN;JM!R_R
zU=%ePwEAI}R|@59ZaCo3l?k>c2kTYaCxhGRL7oa1u7ci&y-kN&VyMs0UvKbQluZRa
zzkBZGniIg*=l7xQ_2)s5`;M-Rvh$$GLT4~{*&YzvH~Y;#O9Lug?LeBu#161d@*b$Z
zZbmtY--&BKB|%vXNt_w?^#GgQPphqr3;?@%4k;-2M^L(gYk#V6B~Xh$^Y2bS|Ckbq
zu}TW#$)axdD{y{K%%tY=h(s!lJ)|5o#Nf3~1|?`PRCy>fhl>8e#q&w5fKqa^za71%
zm^z`O?ey{u+{cT%BNudKQg5t02X{+kQ=&IC*=q(eC}U;8ys?l+)R@iI1D})Z!7`o)
zLAv+Dz^|sBjlldq5Zxh^EL7`Facxvl)A!A!{&Gsn@8`b`vP9nWUVC6osd&x*Ivf^H
zvH!GRIWYeTm2~mKyphNG)Y?I
zYRe+&Va?Di>c5#GmVd$vDAR6Wzt5eb4ld=2;gm0>;@57JJ@YG*x-=x5LWPL+(@o<>13wC8(84eQV?takfaI%EU|C7f9cx
zHg{(Rep+{(df3EQGxuI8wX)vZF156jn(W#-Ec&UGx;*`B|BES#va4EGK#7%5)kBYy
z6=oGv`=u1sY`u!8#ec0jSBw==y+L(ntzcen6d21-t)-~79yVKf(%v3P)hAtvbB$MT?wW8r()1|aS2sb*S^c&AOPGvE25L6a~*7M8T1+M@B+u*{+X-g
zzL5gg?(VtYb#&r`m;wUg519e`S(X|9~|b&%$k;J-jF43zsu
z?0l_MLYZ>kdissKh}x=OvvfCi9#zgh`(*j1C)BV>_IVkO9O_vb&+&k4is}^^vh0QY
zMYgEOe1&Zp6>AuE^$EPcQMGlx`vCG^(WsaEF9eiOmeCmxb8Z(=BRW&uT^;!pZ(j*pe1fG^&XId6Pgj*uK4nF@KdZ_pC9Wk2eLu@6Jw*w(rLZ4?
z%QSZjCznt!a$dU?>@TLKy|s2ZFe{hJw@kR-I+jblKghFCSD=WpO%3bVC{zkF?C0=u
z#Zu~2a{bX2p%18g4F}uK#QK73yH{^aQ4R+4rDxj&ciMtGro)C3qE~@;j?M1Km3dV0
zfc<5|Z<*8*i3eqJAd+e*I&=8P)+N+|kZYSlhR=X$?RC?%;Jo@h@2e;~wv@_$UL(#*
zq9{F4%`%ZsMbtEj*f)zb3aRiF8%1VlmQZ7D>|M9Yi>OEr_O-RYDXPoH@m5D(4HfgJ
zi;orCNL}3I$yeM`Ms3UL-PyUlnCf)+WAp2M85Okr$;A&3>M7^vUo9Hq?3nV$bU7n4@iOf)@jRmK7(G+Yw8Q8bnb9-xGUZHrO#J^BGI22R
zQ~#qMO#4iH|MrK8gK0mF_QS{wGI9Q^XUZ95`p2~Qzxu)G8D#oVOt)Y3Pk$IYrr(U;
zMB0w&7ZV4Aj6YY}uH2r?_ZnJ`r?H;Kvov-%61xYqTuEabjS{Dcp2wCTGyV^0eLIc6
z?TG%h1Ho##yp2W^8mH6kanbRW)BZc@`lGZgMC;qd;NQJ~>)YFz_~kr7FyaisYx`*V
z9kENL{j8w<I_I-5ua=M+fw0!G4(RYs!fB%IL`JxWNjS&PJo)CS4ACWCW
zh%9mS9gep-svYA?p-&ivZ}7rL$pVVr#9nF@v5)B>b{DqvV7pB(`LSJWJkh&=PFz0V
z@*d+kpEnpKa&BV$RYlrUKhuWg=?96vA-)UCr}yh%f8LK;vD~qw7h{6scZ}y&lK5s#
z*}$pA0$thZxO~6w5?rr7$CT7xUJSQ77MP_X0hbC3gsWQPa*Z@=Y_A{Yf#rRBy|7*N
zzHE%kc-^r+V6g_)v)9RC`ORD3Jme?rr2yQrSRhAL59{ZY?Z)~!O3$!tq5cB<869~D
zr#cHHneM~-i}N^e{Ux0g)~}D4h3&-~Y+fO|Lr+ZMlx6|GS?RS1eODj9gYffn@x`Yh
z?wd7F9$`+^d0bzA^(9=oz_Gabx!llask`=44UU~Q=F7M3f$M|>k
zC`M8E#~6`adc+tW&*!=B*naqaCe}~d7Gs>Vkq5`AxW^wZT^3kBFB{v}rq0Lx5-`cb
zc3;{)!zBp&F`0<{ehVl{LiQ&TWU&42r*&{ivcRHSBu=x9WPa%$CG%=+0(tIESIK-z
zvzvzFk^Int{cl-H=2fo~nP>Yhka^_6m5lYAA6Xa+BgnjJI!pWtX@11z6VD{Dzv2^7
zSoU5#f^lBYYm6)WiZHhAQ^j$b#M)w+D|!L0C$;!KMz%r>ENhJFMJ5<8xd^fEWzzZ%_jZamrlyn
z7Z66ysHw$Bj6HW%+-d9HG{qG8mQ8cPvBz{xr{b>(f
zzMhuX({jKeV#lKW>C<{1T7RFGwVQ~Y(Q$%IUNfTixgGTWSsg~=+>%D@{ONtPlHSh>
z4M}+g?Vs8AnY_s4CpanbI52U-hlypzp0Q)fZ_;rxGBp1G>>2+l^z-JY?V0?YK-)Lc
zXtJNwPo?wIC)&P=9;f}Z%!~sQFXNBN;|#*ZhvQ-5`xoJ6f%T02zxsb=W*nGy;iAR$
z89z+E52NEOrS&N^GWMyo%=lr(m8s9jOk7O8f8$|f7zB=ksn7HqZpOHrv17^^JEokG
znRXbN;ROsIW8(dH{2BRQJI0>z$F#@9%gFx=|1Z0L$DOhN7n%4Nd!{``X6%_b8D!=a
z(+_4oGxZoh%(yZ#Gmea3ru=^)(;j2b_-D%hMP`05@iEBE7v{M!b`1V&$F%pq;$q5~
zco>=Cy^PG9TNt~4k%^yapBYaEnQ|sh#-8a9^Zc0cXXY7W_kZ<+c^*u>{|gzu%*4gC
z&&0>@YG-;~G2_g{$E-IuT7Q?`2begS`piBOO4}vT`x_(!uq!O^jK&xm=hA+-X}N>O
zJ2dvvIG@HT+P;kTYedV1w7iMt?SscidoHw|=?C-te$sX-^&~GR(fLM`J{R1i_v!m|
zx%VMrr%C&ZpxfI;=c}P3q&+*DS1+OM=F$3%w7*t5Z#{cY_zBCD)LTLOTSe#Dl?9|d
zNe#k#cxhfVM~vk0t8_eTZV-LHiVS@8ED$k`)XV)u>|UslcDCsf-fj|4_zc%e(r$Py
zv6l!W_OhD^U)eQsA3rY#dfYd$i2a^Z#9p|Wv^&p)@XV_&udw~fSA@ryJ2qokZ1^|E
zef?8#sE*Ta$GWi4hVe?+am=Udn(d1=0+`3p`F`q{j!Wf}!T+Wg|hTBUJoQ?J08h2xV
za=~HukX`38(y#6*IUH~K{wBCoSYV;`G`Iv8|eFD4*9YAH%Eg(w#JS^@b5y_f79Al`UZ!i(Ytf5)?IAsU{R
z$8j!ReH_n&vFBudI@b4N+1Ak$qt}I6xK*;iF2OD=+t}^I^Jo6Q|2tZr$p6TKouc(srPv6J2V0nhid`2pA;%bx59hC*aNP#ySy%L^S=<2X+&{EPi9(Iw{z
z-IhM0kFCJC;ztf%FQ#knVJw`litW?8W@3E4Faz5KO(*-w`d$*(0}X=b>3!1IhRA|`
zBu+UwBAW*i)TYk^H)xqZn#8#&lIR2I{;J+3@{tV$H)z)4_Gj>rari)=Lv~IldI2BO
z|F0+GupP&Dat_rNCHEDR+sU|$=#csr^m#6T_G3uz``Prq&PC^gophYKwBC`{Q?x#d
zmX)@WI9Aa6Zy;?iMe7r2IdmofLnkz<+R1t;in+ef$0@QO&<
zu8T$vB~nkC&dZ6k|D$v~UUYqizc4(9;a^OAj0`6ujsyOKk%^lrXX3a|>zQ(x^O{e2#?jITc#Q0;zor#yJ$N2dd86Nm={ER=wo{|4A{I|XTWzUQ=
z!&{l>%ZxkYhmo0aXZpq1GsyJg|B9CxZzisPc{|)Bp$isZ^#3B`pJ|VYld)rD1{pi%
zxiZMa{jbdU`CrKRW#VMw`PZJgFK60k>M{Li#@&&)75b
zCzM`COnqj)m(caQX=I)!vp$&T0FxXZe_I+gX>6ylhDIkEAqm0uJoLSGJKa8WzvN7>
z?|xc8#82$SX}*4y=GQ|sulJ+x%c{L-zqH*j!^7$JV(I)6)vkQQf4T_Z+gH
zYjY5}X%XS=KmC5=@!921^qdsYZ}26uTqMcEM-JS@cHO5*zHIO#dLjC}@n;RmSAThx
zaJ$6|$i3+|`u;WLD2e;SDRMswW^!PE3FoD-Jy#CNhi_kP!1A9`!W%40Z^A8%1&p1^
zdjW@V!pGg0pTKs8)tZw-BzKlgA$?|c
zB1}0fuu|j-=G*nV+V3Dap08~!;)%isKR!VGaZX(q+{#(NU$q4H>uh0f1ClkT(&k_u
ze6~lIk7X6TM&&`v4TliUEES%G;#ntf!5ZV-Q?v*_cQEpI8
z!W^P$Wb9>Gh?`2Dha=v8Hh+2lMTps^H#QuC2(De(rw(zh;pmyg?^wWd$+7W0P^Rjh
zOmsk;EPB=N`xd^p(W_Ct-3j^bkMuo{-?KoS(qZ;tsIPV(mOKk_<01op)Zf^52LS5-
zJa-p!G(MLNd%dqiOt#@;qj9~yXPP-0|J<*k97yh~^ZbhNFuzC1R-|8CYit8?Vwuff
zRtEV~(Elca`t?+7G60p=ZD2bh
z9Cv2DL+IPdS%}6#(&EcFdd?f?MQ%j>WruX(b7^7yU?uXC+hfw;198)PX&W@&7II5e
z(DN>QaKZ2l#6{)e1E`(l9Nh3#Ei8B4jMv>eZ8lzawi#)7eKya%jMuZ!(q)h$z;$Q-
z5vE9ZPFg|EO;RPYSTE=5h3&hwEzSP+ctJ~$upMmPaJ;&qsJEe)npxDJAxaXVA3
zk1&3v1~A4ZlKTu{L$V*vHF3mo)=83k2@(A!Z0CKb8)H}!+0QmPmg6`|ZW6s<7rnop
zY`}I8|0Cz5^=DgPs)h6R%U)a_pu@(PpF4=LaRJ%K0}aIS`F6!SW$d>?_8`_Dmh;8&
zTuZ)$!P1;{EW5;I!=oh3Pop;=|7eE*6aNpNC6zZVK2ROoABo=yOyB&7a`q6E0`?_a|CDu$GjM(B~-TJgHB|
z!<*GlAOx!zdx4>
zB<22e`C|IM#`_v6?=C0%#FRbZ(LjWhmtP{-O4mDZkl6V~|G@1`(EI3kDY3JvCi-kg
zk~b8Nzr=QaGl*WYlb~WNu@f01`=b|&+?!dnkbOu`jpz#niT?I0l3xO>{)3O41sXV-
zv0v3Wu_E+4Uxq4#(PK}
zbk|*nWp&4sIPWZ8?+A~2kY8^#$9{m}GORb*MDld`kSmtuI5yyXme4yPM|*
z&R9-;wFB2@@9BbD3kz&L;18D~%!41lVSV87`B+~sUk|4^c{Kw~*
z2Sxq0fmK;mY-GQ?v|M7LV&LL4#TT>cdD?z6=X
zg;r1=pZit{y&pK)>d~|r;w`T#Iai2Ynu40sAkR(x*8T~_yRumx1frmr-2X{fcH8S5X2|>NZ&_)SqsHeWR
zw#T(Yd3NfF#*Zv8d$yyEKeUr|U9&SC>e-*gTallVSG(p_e1-hG^Xd{MXz$(alqwG~
z_M4t^!Atlaq4-_ozF%-WR=O`QZGwFNMCgH#i!h#(yv`?JdwyYiMfITHuuBV-vLVV?
zIm}6gevS6NdR_wMW*>Lw#ciqPFLzL}XnX{>j
z1&a5y1XjRtSowhOQ#kam{HG=?73$so%>0F(OVeeJhcjXM$)))U-f!SIEiP6#3diA!
zqM%+l^dJACe48n(pHiyuMmv!OZqB?Oed#&0UvOviEVO&Ydrjc(EBIcf$HspW)^Apf
zE=K;7J%qU;As*iU-WQ!GL*KjZK+m^=m#-Af$3g`Y9`rt?^zyWqXg(dd29MbXg;_OXPCP{^iC2jKL_z%^T=#8Uti7NuKx9vozxWx>pI5uTo!5uxzioJWH0-D3Q}@s15NCe)
zF`o{pw0humo!w-<3eEc^ORd}yb~v7gSIO`^
zCnkjV9k8$mpYMnE&K5=eQl1RdmmsZO;6vQy_U
z!Tzs(r0+d0kbT>B=VrVgeoI-8vE_;?jx+v*AwF*zBv#}7XjL{j56JYiV7qO>3)_%i
zi+z%~{~GeAad{9Z!&nh(itl0A8q2U^ARqCkoI=h~T$bd!M+c|mJQPQBQUbQW
zGepKKWEy$D^Ts3yx6@HI{R^@S9LU1{rmlRz_8&gmygx?-7^N{Hd6}M}0)&
zg=P}x!3I(8zaMm`RQ@E
zLYFi5*>DnIyGJy3(f#qF?G5Pik2Es=E$H*=X1ctAj<=DHhxz^kI>-K*c$xT<>2juD
z3^Kf&v1_I6KGXi0^8q9y*l#V3t~BnZdAk)YkJGq|MoGF~!!$o;zSo4C8TR*{uFpsF
zc8vs5zLl;&L8C2=OuHhM#BL?+zm@h=O^@d$ny;78{BZcg@zqx5$%-dx0PF0B`{ChNAjm+Wst;^h5tZvoj4xwaDC
zzI+k64=HUE$NN&2u_fM*H1$dQ*&m6W-Lz-;xjdXk?#q0Rll?B{$_Oq`axBAmWfnK~
z;~(}J?-Q1ThPd3bV
z;{uynah_N(FAbNgteuA2`O$qI=M6=HWALg2-Vd)-i$?hE&UbunFc_SN{pn;2U_VFq
zpNvF$cG{LTC_grZ&w)!1@`5aFd_SkpdWrApRTimZzCLx}0{oPi1#D&IULw2oOHJ^3
zK+Cf08IqH3?ubRWV^yLF%CBPTu3ku<;c4;;;n%FNI|$3vn(+B#SYSWqFXnCQxRKu#
z51hUs{jCW=yvGjaY>sJ)(;dVHR^
za$b@H*?N=_2czNE6&N{^3CH8aA^GCesPT&W`OO+!u
z{Lwjq5-+O{g!r;UXO;#FL%A~{2BUp>zXTxz9-6KZ8=d3_04|_WPU@QSD^5@
zaK#^Zj#BX1vl8;+j=JY1JE5OHTh?4#4SAz-xy*%~Tx>9=70>Mp`MOTKf4s#IoDW9n
z6NNCIYux^yW&5GM*T#%n6YyTYT|2C$yceYPeH7$wfbBZ3TCyItyZ(L2*lg%G!%3}2
z7shk9-ZWwwEFYR9fAA{g$HOPM&F!IowUb}oRzkbU?`j;cpgww`xD6VQ4hNH5!3lUi
zoY;Ro5yp91&oX5g)&tVJMMa@~!22=hqY!sA8!eNE?ZzkNPJhDB27)@b9PdNF=l7pF
zsL%rAwUE8*HJ%rguLAYE@&Q6X=n5ILiEz)Q9Z
zn|^Vv9^dh##A)*I4S5iTk1#_PYU@d19Xl@~NK6|J}b8r0t*
z{mkhP@IF(o!tH}(+rOUNXr0BSNnc0(8+UM0K;;Q_&6NmMRMu=m>#=fM>4NVt?r0&A
zDRf>DcDLz9d|a<0sST~Cf}|vPWed+015cVCL%ZQ^e0P>XS+cI{7Uqk7KZd=r=Jxg+$BQtNBuVy
ztQYN$#rFL&^Fn@D*q{himM
zMv||_r_H{L+7}tMhgYHSULx=a?yp+UB(DE@xhQVW<;i3r(kD2I!z)U-KUo)JJL~;=
zINq!lQ*3{fpPUb}D@ooh<{|lb;{$@1Xx?CdlH}p-2E=YR%{PK*-V#H<$B*G6{>CH-
zZ(zS5e%{dES@CfYxsa2*hcP)y-p@Dok@L%D5Ar^vsFT$1ZzsGXks|kOqvhlr^P`81
z-_mR3{Ye4MOTN%?xzao;gyxf{XnyvQ`!)BI}4o!D2?GIPJr
z+yjRmA-v>1eV@sHhuA0T6Ff|}cSo20uHXZ~^|?eptVPg|eoxX$w{K0y8%oRHoQT~f
z8kv4o({>AKy9Q@sx061f+@N`+3(X^S==T=wbX@EaVtI&@dkY_fIW2zzsuhVTyd$7K`
z^9bIjG~XngLhJfvX#%!;t?r2LWfBUc@P4`PZtZQPe>^h-9%bNt#L^A8UZ~nG%vWFE
z+J@zWd$lT%e#_SeS+wt&3=d;|D7bP4wu{tK!~4vLUoMW5a!AGB0q_St$9(l+&mL^&
zD;$pdr)^_-5&66J@{K&&KZEv$;QaJ9H3WbEtslsbd7|ssf8SAiy#DzY(03Abf-Byl
zdP9XidT4)EH_mWE`Rm82J|UEE{93O_
z*(J@#332(}*Dc8Zk5|I0k)PXFPrpENr?SIeWS8l6_2&f4R|W;yr%-wOyy_lgSMmG7
zCe(hg$o0h7xVooz6WzWIC~uN8O~K2p4TD9sK#pkhWf2zB1R8k-j?9i
z57&j|+iOR)j3EZhIINNc@t4lhzXOoRT=Kk8Zwu{hPl=18dR-697xY5CyPWjuQYeRN
z@|4v;Sv-D5B!GOlz`K3~XE*p>4M
z_^=nW9#ZK6mG9X)I|?Ab9hcLaGGhZ$-?FK_n>_--8
z_X0l|$;X_PaGYnKf5UNr4Gy?i2600^sg#)dvk8u43-@mQyYTrNoO<0<59iNV@u|^L
zIPL}3fv0LAKQ3Iqz*8FbbMMc-XTo`|AH?%6yt5@Cs(SgQBcAs|t7QXP
z*Um44XQOjqHn+Xg1BfMu-vLxltjt;m-A7A|U6n`s0asA`I>g)kLq6P)f%EuaeBjq!
zh#JRFzo1|_hjqy1MK;?rxh4=h7=E8o^)
zp7%K38SlS4ha{#UJN8xGvq<0dBQ6*1dnFh4a3X&$>8@i43+-p%@0nWuL{w=)S?5|N
zK<^1y+7$dG6uwvD{eB$nb9e6T-;L^*tqduvg>e=MRKQR0;doRi383?mj-udoG%jCS
z1va4bbJ4=oHspVJx;DOVV4v8H^NwDIG{zu@DZKxW#U|qShpV`auw8+GB+gr_KAp$;
z*A^8sMN3(7&#!$C`SE#@5XD`L$VSQy3Uj4!K
zHcJ71@&V63d+y*o)PBM?5#?(BXfpdL_RqDQ
z^)*COoPC0Krowgty$8N$h4-k@EA31Q*l%)`lYIjV}ub
zf0}7d_|^U)QjhsvId`WbZvT4?Dc3wecv|*063>=IGM+;=1K2KR+c@rD!#3jo*aPyu
zf;~le*$kRbu2dlL9H9ta*g*U@?;t$w75&|k))J!6_(9NS4{3kf3&OWTXuFsEL~lNe
z^dtKniRfobcG7#Rb^kO37qw_dt=%$9g~F$BCbe|DIv;9xUM7
zEo_&)XEo-Jf0D`XxV-1qVwqd%HeQ!J!kZ(|ybRfJ=o~`dJ)U?zRz;L!yG^@{G4H%~
zGU*1=b7m4go=|fX^Vzg8EC)w>-iRc-;i|I9^Yeh
z^+!aZeQ#)qtt0v#bIqAnK9nc4U3JQkU*}nCJCVQ4$5KC$oh;RBhegzOfjU
z|G1I#bty!nXsHvZ{SB9Y9Y;KNhw`=rq`zUPE|2QDd^;h9?p?xnR&Cr1<+$au*I7Vp
zR+4Rrf~aOWTvfU^g{7zX#r%-G0>cei`Hy
z(^D-g(;%Nv@H(fsb1@qfZ~c{cDuM;R_ph5H4cm8-W$|r;`Y2z|p1;r@43#Z@FUSUQ
z_r)rN=fUp`ihE5@eSzb2_q3rB@tv;|BSxvwo$u4jdmV;eJbA
z7{7bPn~xH{kavFiRMxWr@>%0+y8@y9!{PLuKG1&NY3C00LHPaf&Qz7RiM?Rw__x-D
zylmk8?9pPcG01yX|Kebm!g#s)v$-LEKXQ3Ff5Ai#sJMB*Lmcun=eJvOJ!Y}NA?vyF
zt
z6*X^kA5J~uy5t4%fkE~oWjNkL@621E5Z2?lh;>s1CAG8he(k-w96XnaK8fWYMxU~m!aKEy~~o)sSoDhQ9*ZzhwUeGkiR)&Wh``Gdx{-Z
zkM4cvp45yMfO7YTNO!a@OE&QHB7U_`HO341jaFYij(8S_*}y*Je`G)gf9EhS_&ff7
z!oD{!8tpq76C8)o{uS#ZKZWdG&hynmW&L_uiPRIV_ri)^J
z7&aw=`C;Je71&Pjg+9qUIcsrVv@IFN_uG3l=V3lZJ>$pi8lK*R_37SKIG@B8J;#2p
z1pmV2(#uGGubxHcPx|k4#Yrv1{VUcW-&+_hBzez9j=YC_XG-|NixN?6mvM~TW10n%
z{L5WS_~Aex$@k8lwYc4k#R?b`49WXT2?t`gD1-2jeV0pcT&>Q^*lz7Vu9U=C4I)q0(rT^Z6{U)N9{_hKpJ7eVw9QVXM!iNf*$^Du8Bze#3vXb;G
zD~s>|W5Z!wZ$&=wV^8zbfQ96{6;^`^o^N-|3C{@~An|o9CH7qcu(>U`uufE@O@pb>JUEPItKbAVE&KWyh>(T4X6{a{sPv|cwH
zPzXVGp1t+>_dMQ3s#}rVu;M%XDKNYr99tKF?z{G>U3`tox7_4WMC;il+b16J5$PuF
zzi2;ke7uDhp?mAe-DuyakMhCq!P*196e8JI@U<7(|I#K~Io`m1&sZ`!2KmpB?)`>%
z&gG+n@RNLqN3^ygJDE?qj(?zC;nY?)WS6S{-(J+-JExfANIraBLk8`OdQTqk*YM4*137^=Ecw18a=QXAfNB1cPM?(bF
zTn9J8_I6&p>yP&3O~q9#bWT^i@H}!J%qv$@=O6cm@-aTAycQ@Mf6gnp4DU(4+AXs#
zf#q6*k2TQy%DR@QHAkRaA(T?}2lAO;V7hDxysta!_qtRSwv#f4>vbSR>zx`Pjlz!i!6ubaciV=onXFPTvOMFzB_c46#Qij^LUhxRiz=kXR9sR
z+VUE<_vLEZ$9R|rhvWi`<>7aU_uNOXaL#7~o)&ZW3$xikb}H|lE#xohz)0yByzkl-
zcl^Z}=$BHy5t0Gpb2`eUrnR;g>{1cG`W*Vn$ZV}I%!B;yxJTP6$S;&%RX6>D@tkQ&
z4o=GN0bMOo{WoE}>Z>@r>|x%ma+A%650mu=L`(yJT+_Y1=ZqpSEAZ`
zfIx}dtKYJ0AfU1Pak$U`kb4<-q<6Rn9Q(V{W)I|1wgJa8U9
z_Kzs!h5qyxWGxO*VT0jAXY4mXKY1Q&1STNQzB{ZtEg#N10Uz-k^j&Cm$?lq8m)M~3
z>&*fmaW=5wxhi-G@>rg0((LbWe#C9g)#QWYufSEGqXFZv6w4?@>&w)9Y%}6R8?C~o
z5WhKm{=MfBh(Y;6pLW6aQ{2u(%0YRxU7ZhF$JHA{B~U+%b$-4>@81>mjn2tIyVoZj
z3Im`#$-B}9?K6reZTBK%eSKqw){Cv;5GNmuU&U(gUojXje^0`#99VwBF4q*bcg}ur
z9f~VTVnx<*IDV5S?*3AP_U3BcMQX6zV4_W@8e+{^>98SK|FG@d8R#DFZ`!l%X#e@`
zrH}V*gZuk%UJ(mDg3qB*!SM&t`|9m8f^$(kjW?X}KAj>ng89W9{~-L`%Twc#cI0o#
zvH3gDel)hvz#8qFTIE8RU)ZimU5e^kiL^$_QEapR@
z#p-xmM5SKhd{zXn7f_roD-H2E`*NrU&QFEbt>vhm%k0^hPaQh%9fO{?!MMOy)c*~M
z?RcCU`Aa?^JHcR0JdVS8+p+)4e&qh}`bWaA_^Zf0(l0T>pHd^Euw7{w`Ta(-Kj91d
z0))2&T_^90U!MxZb^^^uaX&i`R$;jUka2l*fZUJN-{!(R^4%aAhyPdCn}_B6eE;K!
zN(il%N+~2HTC~teQ7J8=(qbuuR!c~cY;BaZqAVdoi%?XS78OOZwb$Kpm-ZAFT~Bd|4^D5WecAaK$vEJfM)aBDBWff)b02*V>SNvnQr@Q%@r3Ty(mJxg@Lsln
zjIX}rS)`rUramL%Z&=|F(VJGc1&QHt^gCLZ(1lBXAoZ(`SP}ic>PRdpuU75pOicHB
zd}uA^!=P@nJPylkx|WCMnTB1>4y?C)K3*z~=TS;37oKbZe;i2zVkbIbX(xd5iJq}8
z!*pm_$Rw;^tl{Hr$8;IPZ-a1ptbxxtjB8#+eZ%8iWv2gH6F|>oO=&zo-Zkr*;`Ftr
zXOex}we@3TaedY%?uz<=R>>RGF;A&wX2#-u02-*#F
z%fHWG)dzB-xX5|ce272qm3|55D?A?zPw#-9u#hW{&li2ObWZfaI$Ex-v|=8_hb9+B
zg#uo6-^(!rT-71^r5osGe48LM7wj`ye=_gk@~n&Yr=ABCnZq%%0;~<4(w7Hx^4-$2
zjse~4&rSgaQ2)mxf_%jw$EH0u?@m{I@-|=&sANY;Q%eT%h0J+v8?=(B;C+xGst@eA*f_<-&)`MV>
zyN?Hs{eBB{Ige(&Gyr~A{Np-o8-(`ic=-)0f!?mvIDuzfEBgDdbJQ2?P@*rQt9F4s
z<;aoJSMEUn%-Z!*7_`^v!jHrru$PQ`S>xpZ_8&$6TNwr*f5rc%AFadoiwW~2BcVRu
z$)jJJ=Yw7sS)y|Z`oa6e@m=lE4;w^`UC)4>DoubbOPR!82*^5Gdkm%8TfR0MSA0`KMN0^M#jYvp^OJHFqb5A8=#ZlsjVSLP=itV
z^_A-wHE;cxi`&=pRWA|KaSaW2;P?fSP2_y*u_$S(P)P0
zLM#2rdOM@(9+78uwlfoO{}lJ{QNeVt4h&jhI!AZ`4kO>sMMSQzH4RV2bZaCmO)y>8
zBu5LJ9zCh31>-Y$;Wrrj2c*b4ocHQGDPOF{p2(@IH7#Vl7703*f$3W0bIJKt6dYT7V)79K|%9bXW;iS(`jQs34&9;9D%)a?k}iabkr
zIspC$u~tGiYy4P37n8Jul-JjKI~2@~xkb1B5c$0d^94sMpF`kl9n66_{
zn+T>`B0=A^bgueD^n_K;GDJVvK4Vubrn@75m9*D2?yx2vcLDE^367ueudD*&Z@+~q
zc-~w(@OaHyz+;oE!!TX3OvOZ;uBUyGy#GG?YwUJBey#Q`jmG;DzXKJ-e_i*wvk0cU
zywPno<|mrW26+#E<24rO!KZ
zKlJdL{?E9)7JeONjIxs+=3=_C!JD(OoSj;^x(Lh5=!~Up=K*D$2QOLz#yjm8E`oL9
zYut0s?I1s7@zKvqqb~veG5l+S<@p(J#m{(MEXx1)
zqXg>hcwr1rFu=S|QYysz6wwzB^YJ;7^S`ynqJeI@fTQsYke{W&Q-1h?To8D%O6NVu
z%faz&&3dp--_yV4wFu;8XqB96y(+A4=Y>iZg8Z2gzudA82=Zr3_ZF8a3i^N2d=F?neo
zW2XR}v%R^JIne3Fs!O#%`B}YnUiO-{8e>nYMdmL`>eZx1ZV6PBCC&1iTr?K{w)7`3_t^72->Ko^#m
z0Vq*>BD9xhSo$)Szb8^;#7!XI>`B=ds=+?=v{Iq}AJBc3i8}X
zeP@a+)T`+`@B_=~!Or(X(m+>w#o%Q;@VUb-I17&_<$x7ti9jc?{9`D-Pmq!DsREB<
zg-hK!Um*T+@e5-e2>%`NSH$vCc6HNtEH5`V{E5c$vehxM8J{<0)hR}3Lf9nRTng*E
zF}723FkO@6+9+&i%(xc3B@^@Wd3FSz7jv&hTsscvEm4?;?F;tCR*ty5!fjfC*zc~U
zI^yLz2s_J6)57%@-%TL*#s$wtkatiko>*w(dXK8Q{j`Skk?2bsj{wfw(km??kQ&$?c@_Qe~+QrkmrT{?%1ASvi_HU@FsL0wo?5-P_vKh
zzju^>CUo!ZsULc+)@?$!P2vub69wnoi9FZpbR~3?o>6~;fs1rMzdrqDG_G&a-gU$d
zU+PESExX&8N9rw7p?(CmW2m1=)jRqwPpBi^C+oOa5c!hQ?n~rl7ax_s%~I5!;LPmT
zL(6^>zWnzL68a_5RGx2JPqC?p;zE0xUn`i(@6a_gzM45#sbua~i*KXx8}8G3G8jK}
zcRw0GvYDa+<5%CNPs_RxGFp!B+uKE_Wb246MLS?D@`)K%wJusCiQ0{czOlqqml*DOYK}A
z61u!gN65G?eq=}NS8*x1pmqJ)S
z`EJ@h!~2n&m0ND%^)PgrSUXPuYX^zt)ySAIbzIx!j^;B)}s}*=U
zPbuXH=tDApO^jikH_>#=#OESH+iHG2#&ljT2D4#bYI;|^)?`X6vdFfbjQv3mSPr-d
zKzX+6yfxTwIj>mwEA}7d__^;Hg#E3^PQx8@A^nHLC;J+}k>HsDPoWX3
zo%FkK-m)M5Za#W+hmJM)B?@nq4}<-^^Mk#vy!d?P
z())9fz+aBX+{y7dAnzq!#2e*tQBSt1BC6q_0p0!m3AzoaTVBlG>3btm{Iu*q?5}22
zY1r^AeUOXZ`;SLK+Z)gX?R&Y)J~kjB_pfgS)tXUFNx=5omtZ{9T~oQRv=vEguK8Br
z-iV${tz43MuL*tj8xfE^1^P(hze(Gcz`4v@;nNGD|Kim2Cp?I2LQ7(TE7USUKbdQI
zF96O{qU~R79moee)Ix>tmD!*#=Bb*k*x!sqxxZA$9f5tJU{B$fe_(fSO!~3|^qlgM
zPG1)oPyCCvt*USUJ^jE5%L4F|mRaO2E(v->{lk6p&w&2on_;Y940?{%#~B@C!Oyw%
z*#^Nee_&jeX|H|;!O+B7wm`Lsq$swS2&;8AZr=2>^+2aRW@Ni69o%}
z`%7S4{M^wShwWnJ{EwV2!2i(8dn$h#@Zq?#UvCfmZwU|Fe(6B^0{-x#MhLI)uaZ)Q
z`W-sgu1akL`TJNl;VST#^4AL0L3wkO-7n#HOZu-adzt|HY3!piQ9FpA)~meWB9v2;
zF18!{C#~#Xd<*+YeG1kn#rnGQ&2XQ8P)=Hr+6#Oxd0F!9laqiwTH3}cfKwZmJ>v(Q
zn>9fKpMShw1b^~3a|
zVbsrLmY0kV*2@yb)vw}w)ywx1JFd!4YhrKxQBy?xOM=EHS77~Rk>=ENOc&(pOxDd1
zV>hBV37cIY`bB@Zl@#WG&n!){FXP)Ymgp(V78bn4^*Z~TCSVK=OCfq$zSvPBH(uRy
zByxVLJoU4TK*2=5WuGl2_FN;Yk3?V4ka|P(jH`d>x&MyrBogLpruPGjqu-N!D=K11
z{HOu-6A3+9Oyt^xe_`Rc{;50Ych4S}|03nS?YKkeJ1(pte8|tWA?Gr^rPoOOaf1*N
z@7Z8a;0>X647)+
zjh{2`Bq`6&jM_IY2vEP0qn~KGkGq;lIV%`{HlcG3tzmI)?L&JkfXY?ik6B^cI
z?!|!uEpKS@V^
zF!gKZtJ@J#wn%Km2BuB(!?o){(C&+lAm`48pj`IZRdx82Iz8_b+R>91d*=U{(%
zXniV7KObc7|2?pxbbKzf+`$*rA6;Idj@aMQtz3xyqhIil@IBk1hn&+$t`|Ft=daZ|
z`mTVe8GWZLO6@Aq6Eh9>Q2+W$mmEAkgA3zGKaIaQm9#ISDVT(F&KZ;Os#+gI=Fu|l
z9}<6I3O#o-92g?u)Zv{3g*w{_KW`uOl5n>neP_mU{V)lK{8S*=a*~gfdsSFU1<&8{
zJf|<>bw{*Eq!NcE>&JHCecld3%LJ@n1h=ITyXnM9`XqmoWa=5rw{}SndFQt9tl9~D
zemXbJ*cZp!TbO>u_~F`zix?ewEy+4_G;11pzo~9rQZbHyKJhPE5AN{wlJi{7iU?wd
zZQBtHS2DrwGG-RhJ59$H6TQ<r9jVV156tnR`2<6ZDzA6>IT2+&5|dZM)?|&Ib*rAO{W4yC{QR=UF!`-)>hMI`{R(grWL&G}_It*9-c=mtA=kykHNx>m7E}
z;4b(Jy-qlM+ZWE4x+fVoMYp2u771cepih=>x%~PG@H@W6>cdw!*ZREWijK?^u-B(f
zHZ#`)y)7al1)meCpTE915zd$9Ci$9S{{^|8@~$I8&>zNUnvT1Jo!ie#I9{y@eXp3X
z_Tr)@bbNcuqLp)+P)TKP+?lUU$lLswwm@l(UvKah*!(`N(?{N^I%SvHc*VAl$Fe*WM$oKHz?5&b#_
z>{&j`X2}(uf&XuD$Nm7&`-E>4S|o$L=)RoS8iPF0pLiZ8bKin}K|;yR3jUuBla5?=
zf%hlAojYE59_(97bB}$P2>Rcsx2}!?*fI6zj4#UG(u_LJ&lf0v$y
zu6qFO_gg7F-w675p472WgL}}QCqKXc3H`uVn7`@{{J%4Ou0Az~dJcG`d{P8G?x+7+
zizu+?wWn5ZUk~xh8BPA!4<_xy_?6x8|Cf$gC5ruat*b_VY=QE{3|lHpA--nQ-Feu)
z<*i)piS?pr>jS=cA5&KGidzAE_H5?PdES?C?G*?X|BsKL3qApPjW0aM)+6FJ32993GY}fbwJK
zJ)U_9htqfqv7Wdz+3gCR$CXJR$-M<%xAtXt-zRKba9|Z+kl<@AJl`XI#q2TNf?oxH
z@H*+n@oZcV=U%??2jIRcM
zAH(y#Zp7##PEVUa-?h=aeJ~h@m#=XDh1bdbm7``jEN+k`g!63}{6g0EL?M^=IKJYh
z3i0ia8?_)WfP_PdJX13p|Ka;ntj36YEC8&45AyE2!b=yi3oMYAwn?HcModfmE-
z=u=hwyGg$OHP`EKI`^{PD~u{*=M%kfNNxj3k9_G*ONt`oO%q(89cP5hg+Pxq=L;PyL-q=_+F>`Z4b!
zfYF7t0}f#c4m0{{IipANG3CPMhNQzFBdE=Uxs2YL!|21Hbdhv0MiKnRAQ*#57=$)K
z7_#IiydCZ_AmQ!I_x}d1X+OAi)A6uEgX-#
zSte3CXRsLH`B}rrAR-F-|I!fUG*-W--
zkLH-(=+nAEESFv?d6M!fhKx#a{L-c2p%~}8L=ic(CFPYemIp6qsy&$o_*h89AQ>>=
z9^aBQK&MebyB@Gl&Xkz>RSV?!$A~7F+TfSEZ`=U3tNhX3wj1wHR%9QO$9^F`cXs)?
zg8nI@7;J;vWppIR4WnZ5%JVp1@%m$5^Z|QZZzfhlzufy^8;14!_~YBj_acfnteubB
z8LzduA_(y6&3cn`zzC0wr?_9k_i4eWpyB*fVJ*oQsI&I=CJ3J!x)_DaPZde-^nm)c
zeX_ih!S0wOIm7WE*y(~5`Wjb*o+Kf)sebM`@Kd>BGx!1Yh}k`7Jud@&^kCKMCeTwi
z=@e`g1ifddz-*i<#Lt+}kg^QQot?Tz9`}>Tku|ppAYW+7(+RJD|H?gao=>2E^!H8|
z#`?Cj%%OaIuWXY0k@eVa%QuzxD{jXn`-dmKgMKvW+UTy+5FXk;S9KDU<68E)=o^Ip
zrF~J30+dL+@ih$UbzUP;5eIh7N6J?j;4}T<&mC=`zlx|WEBXq0<#ygNDQ#f4Ovn*i
zy|)wYcLpt9E&=tQ0dSL9rjCvhNFVxcs{0Vs3E#BxN
zVDsm|G>7r6sGKu*T`KI0l4=Vw)j{tM)iocB?-yA7oT+pH&YRV0ZcJGZ<4mjUYA&{`
z+MV^^HG
z1n`A3?>W5w+`9U$0?(V_L|aKE;P3qYYwuzq9LoFn8(!~jzquZS=WS5Sglwz_JNf_B
zmIUm$t{1of&}GAe$GDtixB5Oj|G_Z+8Pln*^w|6h>ODQ@q`f?Z9nbSz@dA9Q?{*re
zpI&}L66?)MKA)Rc0-{(YKU|LOEo%#0&X(%(##-DNipskW*T~@z4K-OX1bOU0K`4b|)4b$^R
zd?V-Lij#hS!Qs~9U	FfAvWd4hL!)p2y>J&_<^iuX_h)OeA#2vQqj_bJ|bsbbKm=_h3pJ04-=pHHe
zS!@BJ*O}r?_!Lp3b}0*yr-ZJC8$;+u_4o+g=F;n={_vA`30|2>zx#E0Tq2RPUP)B1
zzLUL=<$6&&s(cK!
zI~#dXy=OF!Vo^vpsc$Hq>hTTARBw{cX7r~5sxO6{rR`dLo5rtwMlokG#T#!a9!RJ9
z+`nd8e(i6X|0eSeL^z|54>NprFz5BZWoUjEW?xXKM8n4xP%Iau$a8|?Ddyhe3+BB3
z5Q$Oc_+{R45f2p-Z5ENM#BZG=y&>b
znfDQYGw+-)dP>vXneU=p{71*{9cCZY>`Bu-m~a+zUh*J`_RGP8G(CMU#ox@iz-c=g
zPU)sNo#8*9X+K4#U
z;_pKFn8UnJS=>y|Ni44(A^AVgqw{;Z2W^))v(N8Nr}-`~okPlB(@E2h#?koS*JwCT
zp5`CuxlhU)7*Eq{lWF=g1)6>S+CUIwE06g1vCfxdM!~AMRbjxT5ynN-QUy
z$8;0<+rRfHDR)7b3qe0#2f~l3L@~+#*zMO;OrJDcg2dOS(RKJh?Pd~A9UUg?V)aAg
z2wcAGn>12RfIa<=qwlOkBz&dvC0XwsWSvq}qZ@NTulkVPydIzbw*1mri0zc#1G5jE
z1562WpN;)mVoz&lqyjD)wVRIjPw_$1PvG)*Z};QD`S=r4G_gMO_d>RTA)o;Nr|~%b
z%B+1O?`ShBiSbzXgYg&())`o0
zeh$@oKCA(K^Jn+3T`vG7E1#@84(-r23fP
eBFk_Ep*I@^ag4cP-Oy#v3$Pjaf{1;0+vSkQ>j?s67(uz5uXqH9OC3^^dhCt
z=jXTc^>1TVP>|D(suhs>G<{J-lH)GN>vpg9b=eYc}{C+a!=cn`K&HQorK8DjH7EgArApp|b41`{MeSDvJ2!FZ8s}-DAReRyGrjHO@|Zkk+-?|#_#0=28o?jOaqT?{dmS1!54a*l^f!Dc^
zKjdmLlgppnUO;m86%*D^XXlV(z`irSlI0t^WU=R4H3BhHaUSY%gylPJyTJ0d8SP_o
zQ&b`*5_m`ITx{)e&rq~ei{fPOwBH>h?~hAl_J_yI@N6Jo`19`NdSF&jhYiKmjc4<(
zkzQ>}ixstVql4~P7Wch&WX*+~X|#>S^|kI-#*$vY3{N-03;D<4G2yq?H7QcR)L3gX
zJB3cW&!YU)ZyuQX*l&K9#?BS}o6&WE;%-?44>J|>JJ$*`3Gs8T3}!0gl*@Z~Xza6^
z^Ti1_89ca4e0=cjP{`7hOwwrY;`&kYDRf)~*0=yfLKL$N}lM8Z{A65YO+u
z`$zW7OpDjI2QJIZNT&VM{(bv&iMIs#e7;I{=vT(F_tulbF2d8|9(=09dS~E7b%P*%
zpj4EqHTBOF%@47Z-+e`FA;}IxHCyF@J03n6PIi9$q_mF8+xpk!G08s8K}E!S9tCwG
z!h!!CPCESx`-Dlw9d%=2_mkz#X|(SX+d25zQ1H;eE5p_e1inmmKG}x+kOuWB^j&hf
zg3v3nX5Zbbib1s@>7U#x1PNz-HS=LjDJY&I1xI!R|nX5-YuN)
z?#)f@ge{lX9Ucxlf4T?i{e`@>M7pgQSTRJgCebol+^QOU$t$KB|^kKTXWxBNmLp-<7g7P4E-;Ac3s=h2G|Q>Y(r8a%)W
z9^$CY*ujKhQB!pp{;n^j{<+%tqL0SqjBMV<@)m0y
zXnb@M+%1KHgN&cYe+OT`G^o*n-Z$DE^;Lfd(&x?GU@HwA)}^6eg!IKC`GGVZ8Exx^
zQhT;23eTeQvPS;sRspuSy>g*;X@B5FQoHjHdcredJ?@!dLUxUo33s7(XslM0p?>u3
z^1e;&Jo;t}JFk9lf0jS3xAhOtr;@&M&Bv~Zz**r<_~B3NTa`a#(ztw8FX*BC4beA_
zQ~A6BlV6s@uC55B>L$#&NE!Vm0M|5_?fY{gJI7bx#4PA-=DM>KT;uDS_;f
zpPj>g=Q+e@B^)>I8$%xsB0j%&el#2sz0do}mio*7zc2ojU*(uNk1*XU=`EGJPXD|G
z2yXZ52
z(v-Q8m4C2yu_~pX9CHH^4EGia?16HF?${s!F@uudyndXGq#|4Bwui1
z7!f&#+0nXYFq7+#P~b=YA}+r6xnMXThwPvqG{e5@FfkaMi@h;tcvOKaPwpWR0Qk)-
z-2X!`_}0br72Yji>5>s}@5o=UgP09?YcSk9@TWLe&Q(TW;XjGKApOn?G#fa3D3>pj
zgXTcKkd9)377qbH8ADDU>CV~FMcWgk$6w=6u8_+&$>wCq5+Xo&{w3ERv!c0t^;j-{
zo)Cw1qqut4@b?i%Tsgh!d@g_NB~Cuw!^wB}_54KiqVo3>08cA%_>mpRV_ot2Pz5y)
z`HlScYJ(M!tb3Hh-P1S}#wb(zO}?DV*Z06%0hEr9Mj2}I%YDMvucqc9|8*{>mxxtC
zGG0z%$bOa&@?O3j*8Jyo@aqwre;boJ^svYos`K?n^JOhQ3iN1{Nq;O27mxuEi`FNH+2tUY|8J$Hw
zqt8H)H2lFI@ljB&09i8f1qdgGKddW0GJq_wPtYD(m82)IPtab0o}j&KTL3+8W;o^r
zs~G%(W#Al&J&1G6X0E&e{Pry1mv75Q1O9w_N!x9dm%vYF(;2d0KHcM|H}U6*U|wQV
zfIs+EKDP4f0a<0~oB8SE`H0|yKln91((Z`l?R8Wz&Q-=z0^(36C(m=E79(!;qlJcb
z!zL#DF>46;PA8`y`;XJVZACK^ae#FwcpB?)oWrq+2!Bb(D#$-I6adsE5va{yPtJV(
zNG~q^*hmih^YicX`zwMi0KDhM78-aC-yijl!Z8`j`|RAp${jUy3X|6+oj?iTx#wnC
z5l(fb4u>Btp#rh~a>!vJ&N_4NRUNr75)Fc%H>80DpI*lGgLoy^PQ~3cz~CGF|7Ui(
zjSL4*WeXYanNQ{PFMZ+k>n&+y(cZhDOI8
z2ZO*v<6>Anw+1a^^^_05F)Y%b*ssRa$N6JXJ*R3PI
zXLr*-l+s0ny*9=p$;
z?<}17jLNGWa{3?b;|>;27A1a{IB@(3;-Sxq4a2p8!UJ;I@10-En|zJhIblKsdq39o
zxdU!^@!nTjLZyp4D*Aq)KUL3QG3Ff6_FOi
zi!HTUe-qAkIuJcp46g6Z#;Tmn}B*LJ1YvM@n
z!OxOigm1((b`mNZk72U!z5Q6hI4`L?szf+@LqUJSC);=f3IC}-W_WjRsSL?`_wGQb
zqw}Ma)`VR{^x>+opSkkHgfQ@CqY+C_9LeU_JgLhfH2)6_cgv-H%!p(^L*nC_`9IXi
zzRdWwcS(QorU9BX9+&KgsSx&lu3+}u@;aeF`Bn-_>j~E!^v4MZc>3C1U{>G@-Zrw7
z{(1EXcrKjBo9}1yBf2Y|*_DwzVH>3{?yY0!B
zRkM1nV@;X;z6z(<`sw_l=TCMG;-#&o`5iOhQybOa`=nZ(o;x95i%aad6#U<=I_#-Z&`WUUvGv-#60@>n%UFvxq+>Z_XjYHNUsdQ@&K+`QNjAN
zTuqGSH+@TG{-bB}4pz>)uZ0Zf#oKC;{%7Y=3%uklyZE%L6n;a2Q+~1U#n-eyS
zf6x4m!ZZvwWEClI=0{gFZ({4O;!q~5Cne8-Js076mFcZsJD6bw|Nng1@PEhmp{_4W
zKNnAd7VRSQfF;`~GNQkf{2AWW!E%LsTN&K}Ojv|&25$Bz!iRgSlMh&Zch94_kWJ}z
zZ=gQ^oNMqT?zs->Ys}7G1GqWl7re)C15MZqz$o8}t9N1`_xxq$xh&sV1CK#37_*-+
zXbM;VNH4D3ZhbC)7iJyhZ^UXQ9LGQBmiOf1$NMfWJtau)ghdG5V!KA|0IQP*BLJU6tCmb&9XQ>e6-B;R@~v_&kdaH$%q@yE9o(mV?B<;
zKzgkMvKmznq}vuk`me`?_0%1}s`VVwEF<|ZKVP1&Pj5?5dM#~>fS)nD2CefTo#F6htU%M2ig^Xpy3PiHg=>4JKt`RQ=Eb-^aOSb@`+WDJ%Z;1`2tyhf^KsMY+
z7xV+0mXO(W2f|t6@Bi2#C{OUbpuOh*4?RKq1ojDJfjxrq1oaEj1?384ft~;Z`1Psr
zQ7|vbYm&Wms`9^flUE>F&@O@f$N1@jei!HoWI_D``3f5_Y}^gwpl^X%W;LuSPTxrvB#FaC1}`S_Y&?&V4<7<@&DdJ5~sg+?6lzm`@I
z{Cz+kO2#?A6hxrQ1`Z9uGD+SXLp_drk^iVb>}%M-qCfckGnqLS+@J8j-^tYH^ro^2
z`hV^V|6>D%^M{38`EJv=^6~O8D>sqfe*8EGlie0@XiL{9s5d#Pk=c3V$|5F5t^|!m
z`tr0~7>M@=_w>SbfiJ${_6cnfv{LaN#wrzt`U|;uw%J$@1)}|nX{5pPytz1d2EV#a
z@}U{jvp8S1d&kN*)2D_aezBQ@{WW{x6Tb(sgX>q%@--}9XFJ!Pfzc45@AaLNgSS(I
zaL%hUmf=wS7zAv*mu~xq?HByoa#{NKtPe7j{`=vN8^pJ66wpQt`=g`e$#FmP^*X~q
z>TidW)AiPqEZaa2j61oS?_^{;haR}9+O+wC^DQhA2b%}!!~f^
z-(j&5BZPg=Q8|__Z|cM_WeChg`mL|$Ijo9j{TFfUHFnPQJ8jR0QhtixQU}5d)s1e1
z$7+`>CEWTpd_3Xcwa%DuxW`gAW8>Ada}qib@mPWWBN_O5DEI%OCJyLg`Ku3d?Ye&r
z|4tnFQ%tz;&lZcuGdr8@Q&@U>RvELaVb~&OS5&h-!_C1jS^7mcZv2`xt5`psT}wB?
zxW7oKXLv8R12YnFe7`Mb9QF^>s?l)t*RC}dg!17Tm=JhRBYG>dv;6c)hV57Bh86Mh
z$$xB|)?dLj667JT_cC7ykN}&Hr6?GY8Rgp*|x)Iq?3(aAjtn
zO$DwmA^$h`8JjnwJ>=MUJbJ{Pb4yG8Vdd52TQa>VpAIs+0+$bC&yOnR^84=62@USG
zcQ`Wr|Gcg+`R~sJ*3RY6E;2cTws+7|C?DxT#V+4o$oKP&vS;UuTkhzfQYd6CC{QMIaxb$aOos^CrDPTB!8Hc&UIXhJy
zI9Y8hhYxKyl)~(x=f+*)`YnUZgj`d@)$=Qqlc&~kvaTp6_owL!JtMw;CBNPse81k0
zU(XUjd87{g{d{~*Wx`-d#?ArUjKZqGjnE~Ip4F~
z=bX>yzW-^p|D+B7cc0~d>3<(5u7m%j|NUGdt@FS4{nH!%dG0@)^G_@IC;d+=_{UEC
z&))f;{ioObzxn_DTp}OY|H^sifA@*OmH7L=`2N3q|BTxIT>rP42}=xC|EB*nMEc+A
zCG5Y&BkX@T?cZu4(*NcA@A^Nz<)8Qc|L^%PpThs{|M&cV_TOqEp8L0&3Hx6?{nI=D
z-}n7@|KFd7$oJoO{4-1alkY$L_ivI&|7ZVu-+%U>-uX}d|Nj4f`Tk4)w+{Mew)`jG
zf8YO~{qH^hJ@;Sv|GWR6{qOn&qrMbI_$OnuuKI_G^Gos3w^8cEII*Is0NNXY
z9UYgW4n;SRlJS!`s2>2XH1aohh&aMz1;LuIErQpw{D-~Ar&jQ6#saYC_5vLP(rNNb
z1E4)%%eI~Pzqjr7gUJYgbgBU}Kibg=P?bNgDvwSB?%uXUMcEzZ;xMl#%`fYTl
zR}y~0qNS_kozvjswFqg8gek!Nl*=ai&=7EVE-G=2ifG5L?l+#ZB=Bmwq4^mNLjUb*
z?=|h+Nl?OPXTR+^1Y*}^79SA2+2DN5!mYOi{<&t{{Jy#kR3L3{-CO!V`$uc5%_GEq
z7)yuwZ^FN}@9#c!Gr`NSZENp@2=oBO_@z?j(tdDvxip2JoanDVfHZrLQt>PC!LcZKKpGx`CYixwM)
za6i!4b-K>g*#**q-#wE}Zw7P1tvSBM4d7$OkzRVgZZNg^n)1d-8&JAa9^FjUNazIR
z`S!-w0F~&e*Q4YWKr*K*XZuD67^)gOm|SWAsC4^L*rX7Q6uSQ0V$B3K;#%2$uX6ya
zT$9HW>TV!j`2B`VV;zv{Se&WrF9A*q%aaRCMPRPc%tWX?7xc>B_VDAb18yguZC|sF
zBlKT-d=Ya#wcwNu$%x!?Xa69gTdN0s?VCjF0;0s;z
zQ&oExlmk9cFqx1X&IZpu+uTQ*^TD%1ldE&z%faEDR2>%u6!cV+deSrJfn&z7v1CjU
z=;yAv@min|Ky7`n3SAaBnz}KfeKH-uQbjRutFl3)rU8HX-E2@&l^)^2TL6ZmJq%A(
zm4le}h_D+yRUqls8?iWor!|j+OH`Nmy(|=TP4}Sz5O_Y!Uu&cfINZObf@bsr)l7CoECe*t(Yp-V_{ujZSsk*b!{6^HB8$udC>w)R%M=@xQGIZhU=E=7S%wM
z(x)h;p$vrR<<-2GDgtP-n4d&w0q|fS`Qx>o2H?X_&IR602U}_{lIg~?KxA-U#QI1c
zpplpE{Fzh$4&AqLBDq%$I!-%ns6DF!tt^8MK(-ukX|Z)noGk%#^X2htWJ#sl0d9;7&yziYf8a;DJ{y|W!2#8xM!%ESPS^IM8T}d-vdNX
zPCotU(g#=_a}rCBWdoPZ3kD3@SpcDZ3WX--f-s+lRi=ui;N-o_BhPKCfWLCwHb++>
zc%#4A5iC&z+6`D`*!2p5>F<~AG%s_3cH2e>>8lKIch0cx>qIt~taR+2k;(+^pBC0P
zOLIZ`_Ch=lUnzJIdP{grv>a5go9?+*5%|*h@GVLT!av8Q60&sd2edE69CXwAK#)IY
zUR6mS*y*LB{7RJb+D#sA);OZvy`R23Uhe>jD`BoyX{YgNuR`^VqdLjVV=*ESg#e>VSx*6JxsX%I8-`f0W
zHkhJs@YXyT0Z2o|2g|xGz_yKY|%Ah|tY4-PE^omn#_`C0kkRNUzAnusExOn2ZS*H#H8RMy7%_G`e0
z2cN2|53+!_Y4a8)R0xodk6*veDFtlZrc}wb6(D4VE~e728q5xV8YX8<1itTi*@oxi
zz$OLV9{1@o|Gj+gbWgJK{Hvu|pmzL_kZGhCq=ulK{C(yKV
zS%sBQkUV*Ilewf8kRS3Z-CJ(}2wA~a%t15Q9{HwG9N!AQd;1)JPK;-Av_039)^GsJ
z8yCwiGy>h?+mYf*ZNM-~+1taZ0|b;e{OWvL0qh@cnD0*3009ow%ju*IUlCKAyG$TdaD1qpqNXJ+P=UEIAOIh{tcj=l$MbiLpoRp|xkF3B8+RQf>p@gT1G
zvQAJc7;N~G8Tg1aY5ZU(rugNfzq
zR?)THU`MuE<CXt={G&1JfDZo)BPcIE#`tR_k%BlUruH*-_o5v
zu;Bj5IAeyG|Jp98&xjFtsHVZpkd5%0`*|r+OOZIQ8akx(jp+A3vveM7p6LhVW0gkc
z3;jU!Z*5==p{K@_Uqv56;1l`fGXI-IIV)D3XO}MyfIF8)wNeSa;N$Dvwk||`66>V=
zzSC{uoS}(Z;3NluqfSNdbk!0#N|Ia2KYIuar|mqQFd=v?-u2epZ5Rf6q&YN9-w1y+
z`xNt@)FIH*;U;PEcNols80Xc|j}rdNyg$QkkAg^-dva|={Mp;7Hibq3Vn1#s+IWJ9
z^YWkY^7}@l_dnlF%Od9W$WGmcY4?=34xayH5QI>6MR%rx}6o>M7&<8-)!bp
z!oSf=Hl$pT;8Qp!u2?!$Pn???^q9Gs4}nD3U1v6Q2!x;6)he(hthTLA2EmtV5*5Qy
zr8ffT4!=E6CLIN?If7!2XGg)YiPEd*m_~u)ukRv)H%GzubgE#x^ieRD{NaeP${3g{
zal5wcGX}`qs6!vSj)8=d?{QZIiFKfby0`%mpZ1D+m{EiHjx^^JJr)lLzgZfWFi2&?~zSS=bX|4%3B+
zyaZ3$fEtO(C8Z&7@q^XiG4C<3DB*K)z+?=FO$Hsg**XgJB_-$Vr$)d}C0s2dVgwYA
zoYgeEJ_MeZ7YQl;Aok5x_B3A!5IQmaqwewjJ)l-NNaE)2Zt%O>*6nZr!Hc2tUZj+9
z02lY4uQL41{g1at95&t9i0hYkTZ`Bl*=;v0ax@I
z%|OZs5ZQNRxPNm5oF*}?eBU<&b`B`cJo_;KoGsjvS1%LwlW|8;EVvs;KL{Qi3hn~l
z>N4g(4|>3$#v$Qj1doTA+`4=eJ_Poz=Wg7T>;_|})nw1T?gW0{cFriNbpUPlg7DzH
zPH<{#>(Sj;-QZ^NvYeDeAINIUopU8{`7N%yW;$=%0kvjY(0*?-IL{tE#4XVT)-AsI
zx)C^g;(RehN<$x5##7)fvK}C`Y+`+QwH>f+7T&TRZv~$zUlHhf08klG1({CtfwPr#
zvpbSqz;P!P5+G{><)6w5a&lULbz>I3MYP)&_mbzw3BHlz^;NRHE*;=&kmCb_(HL+B
zNm%o}#=zw@NO)kV8_cs=t+YMp0EdqrTtB);;N*m^&6%8PpfCLM&kM91v{9P;$)aik
zYTt98#a(UyF6*lsH+gD-vBPVbkYhFAP@*`c%lAq^Em?4BbEOrSoMo-Nv)>Fx1EG*%
z!3MBg`ZBcX5DMJ1PyYPLRSEWvPh@btYXQk3Hl!!7HGw?Gc`}z<#5f)M#fMVA90-WZ
z$v3c-5d0H+myiFh0i|DK>N}#EL1^RknY6wE;PABW?Y)!3V5#6xe)-2C@Mlz;h1#S7
zP<|4Ww6&}NFYjpkM|!k^`x}${wS?~YwjbZK0QOFo
z&w%T0Udu=C8*Li6+R^do%Kx#t!_}P)!aVb^v4C4M67;mWzgO@vzN_GU#
zG*f#?wM)zp;JxkgYY`{n0$Nhfkr8?jNQAw|!uSBV(tr1K2r=$9-5A;PCiI>*G_{J&
z$H6MiVCI5O_PBK&iL3P&5nhru@U@Td@X;
zdRa1$H9rya&<%Ud>gI=pKE6V#YEaBU^8DAs|(6O&80_cvP3yagKX{%
zPoV&ACYchvTL{QbB7g0sQ-Cssd2H$L3=nkn&|2|M0EqT_nYdgZ1sa$ddC)W2K%AkS
z>gb0OU?uhKVSIWq*qy)f{8V`vuuB}`{Zmy2xYTw9RC0?z_`2rPlM!ivEyBB{KqLcH
zo}}SAJ>L#o59pqrP3Q%mm0kxXXAgso-bB_xxlynhyNy2G9RYO~_FO-eD**KC`-ET^
z4kRSRulpQn1zbU;us3NB_`z8^SeMjC%&#GAd!y-KUW&0aaX%IOO`~q1~NaUd2G?d3zy?a0w9qEFE-2s08f!{^8_%SOj{k-p7v8=K$UJJ?bVK>EIwU
zCOzqBI?#6u_CFS#0p3_sF9z;qgBKJ^ukQF1f+Jn4NBbFwb-nDjsDY6Y5b~|wI>lxL
zj3_kAt~?$FNsr^+S8@&kSXb2|jcE4|$SVcnlly>Al;!eE0*@4say^o0CH!YdJJ^&v
zhrkgPsyHEGdE6DLvr*vHq({E3Ndy=RO0UStE(T=3cJv~C5dD39
z?#a+nB`|(0o9`8Yf-uQycG*-EaK*Meyd>h~OKs-7PsR}EO~32vZag9I%YC~xNdiBJ
z&Mu05`rZRR40vpMOAmwhp}wzI2tId%;8~K}75#vw&E{)KL@N+`d>Z*aPy=*%V+%c{
z8-Zrtr&>O|7F_>)h~bw;2@$8U`)*G<4bZP~4aXE`50ixDZ(|
zG4Ut^G#V>9yY&`>fhVb5kpV?Oeu_MRRTTwJ>i65B2~G=ks?lO!4UB4C%L+4)4S80bwFa;s?O0$MYE)`XgBz|CQY
zoFMS~$VG#qCc`X{raf3W+nfnXb5gAuwu^vvmG$MUpLKvooOIy8gE)^nZ4|wrfkRzr
z0I$@X?6Evb#Yhra%Pk&^OUCnU#VAq&j9
zA?|S(%A~X=XrYDO=m~i@)UX)*dt1T>ZTKy+lf6`hYNlz@eI+eIA488%F(DPGLSkzc
zEngCPL-VkC<@0oiW=QcK#+Zzzh!`22eQp5dhrjaAjW2{KWio%rNkpSXhPOkH3Ztw8uilekb(TEbUUFsXhAsHltOaXfGt%)vqX<(g@wkeRKTW
z0S*Z^N(xUIG(rQ551|DW*)MPbJ1dlb1Y~4KL@T
zDp&xoV(&930O|u5o*ALMWoE9VDY?+cQm%}ouc?r?)XAXP#c;?TJC{LD*#{kJzQcVq
zwI8BleqHJ3(+}A)Ug&eX-2o}3*{=v5%7yOTGf1HE&4NT9C*Jw6*#KQK`02ZPCk5(h
zmXahBDuJFyo@o93rxl{E7m8>k?}6<0pXi-$ZGbxY4Q880dZFudZ)G1R^gyx&UKd=<
zbD&x2pYlH*B%=mVtjA!i7+u=2l4i
za$O@6;xK5bpN&X>;#(Ff-wR}+Z&hx>JF6AYQ0;hfcwj$t=w`k7
z1oc9^sxN&*&@QNTY5Df8k`^fb{MYV_AN!#V?rIw`nI<$Rq&K?$5e`+`s6}xkRiS@P
z$WNW+Pesp&(TGV*VJN>>Kv4+)4CE5L^fsQO7&XE>M?9@7AbU>o+b&+wklNK@tC(|x
z&|s2+|0jD=%=?15$Xr%Gdf0E6XZ_3!WSpOSkKTV04Q4je8uKQHomFd}9*CU4qCZxx
z$bBDyOzu}ay&{v2EJc#J&$%GmN*Ov(*)1?o{d+4TFpW!?j+NbD`{e`_BPL
z*x!}A>1r%mxWjq|y1jw!urs8@T$6>R;A
zkpj->y6tHrD2Lr>=%ji(!HCs<}YFd;t+d@swe>5>o5ytv>yLE0Xo`M5e
z-oQF-W*V3tkT*Malr~SEDCQDm{TTAwb~!SR+Iu6lN_~S+0$7sG@C=<
zYIoYhuJbRjcVtIY9Ft68WA5*+;U&+odC?&I7a#p$!&Apkd2454vt^3!bf%MFR{GU?
z_BZdbl#=sd%3>MtTh{kMKl_?658Kdy^Br?ot%#?A`HeIzFT?!dwqyo8J9N)7l%xnA
zVBjHn7c`C)W*nZI_Wg~qQGTVx=yPD|IJv+S^+N0iZ{aBEb_HhR%`kYPpabR)Ts?N@
z^bU->6+ijjRf<9Q+8nXV2YY=DvgSV41dE>4KI}EwizT~%RQq&5ic@77tiQDDfrGCo
zpm}CP7-Y?AZNNT@#Zf;v_R^#S7P8Nu;TT{=hM=5N53l@$2ad0->+O=_MX?93R8(nk
zT^~lJ+o{3`t$y%rV55N`a};jJdKmD&laXPg35-bMWlyW8e4MyY`i%>IR#J#as8ZuX
zr5JMIu4b2H*&zg$PEhMOLWTbwGy3>$gBcNhQrRS|cn0U#tj_9CM)3RA`v@sQjbo)l
zSM1L+B2&KY=d9Ma@FudfFY7hu5alFGnb6PI5&2C0fspNaSn-N?uj2&{+~uaM$0)NX
za}(l8x%pb(;7pL3$Wu)abA1+~I3Z(_(n}6Y@z&Q~+HCg;P
zH&QG-OM%ySt8h>-(jp4N;ssHh16a3{bh+gICZ?q@ntJ%BDiY}7@W)(|7m@z4Q6%ob
zjO)%!3&j6kfx}MIybr&%4=?h~U)hc}#S@N7(JNiKf#*;O+~I%3g`^CjbEK-r@pL-*
ztM6nf@k^G{%h#ZA-0g}E$?sTM#4O~D5NiJzX$1PyqqCO{>2=S|P?uo7SDpq}{EEdsE2Ms#
zEYydUgqQcSj~QXNB^aj2P4(bl>Ktnot2#JZLHE(BcM_bjmbOL9r;8=%(CW)Y8Nw$#
zJ{vzUc!VvOD5`tAc3~GzG&Nc~iz!`6`4gNE(*k1TiuITbNupCde4~VJOMv3uQa9B%M+NL4gbpC<7TC*q|qNYebGQ)-OM*-u$CN?;H
z(@2+&i4n_wqhQz1Vu+1j|2($(juuO97m$@XI1Hy|$FC_D&|>2%s!7xNOt7g`7Fm|B
z1$^$w>}T}qCiLOd^pl=EI!uUuamqN41tz~7_nNHdI5uN)Si~mH7n{FFmay-w3RAqX
zDVA4Y!=A;Rd-E#$1a_fn^6!O5FiZ~?=zqGUVm7M>Bc-8+*hBFCTOyM#%!OD`I8WTi
z;_aWdA0E+xvnqM+`l_>GFC9f?j%$vfTi5+gefu~JeGQv?scDmmA|KNU*OUP$de2{p
ze}M_kIVdJo?$3kTv!unZNBKi>b2oU;KI(;B8a2KveG7w*lV|TPr*LDlcS*b~-xKp$
z7f!x8S`V?2nHZkmBv$gNZx8aRyCF3spWHoSL1;hMS?lT7l{Pq#o#tR$8!MX)#)5UpK-=Rfyi_j#-gc
zD?@o{A`=-Ra!}SUw=<!9SCX<^5T2(+!mzVThCA1ZJzgO!888J2o(
zH(FDfu=4|sDdqwd=!g*#uXJ_*a*~#BZ$4889p|x=V$$k^RBk@acOPhh>`m0RvlGgo
zFV1mWw6p-4d9yW5`?48&eXV+SL%$o6b{tP`DVicuqHyzaqh!w_9V1kU-w92Ng>KyZ6{!s_ZU4c{<_B{s#P*BPa=@bkst^=yPvO+sxvFLguXs*^V^Q8qy$jzc^*Slb~px1|1P7=?o5Yte3|sAOH6dUO|)IXAV_h`#qZ+$OSKjJ8)#wdGKiKpmzb{NB$J
zP>7A^%?Mi#$+}6i0lV45#-*%G)5OF8+Sr&ovOP++8Kl%FrSmc6KZeMTPCW
zn3pV}_)^MyH<(hO^(sAnjgmB!?d^khHBAg6p*8InlNo?!a)M*;-|K=d%CKvGq3ec7
z#;%Geu_d7o80UP_XGDSGcgQx~5=6xyT85uun=0R8le
zzs%s?2MLngavEvqhiVI{Q{TjQK#LY)k<40kkaj9HUm;~Zw8TU~%|u7w=FdCNcP=$S
zk4a?6g#R``Ij<`Q%KuhFx>^>=eCe%_hRoYfixlM$>DHapT7@p?JYb}3qHl#>cx!Zo
z(Dp)!rIj@&XX~N6v2QqI0@6^bA^X6(rDpWaR6ai*Rfpc-&^UUJv;_^mIvm(~qYZUr
zu3q{Qgh9~K)f4CSJkd&yOzW0RFZ9jCPgmqYA0%!r*n0fR2ITO@Q0kU#Jv1ry>Rl7#
z6!g%*aKlcBQK(F@~3+ldQqt`CZxlYCnLJr;q
z(dy?WyZ!v?^G5!i9P@FGg6DUv^Ar
zjuwkD-?lL0Wry8#Q!K>CWiauVzg>G%OyN1HRZjja2{_t$#%tu91Xja0de0k`hTG^X
z*!u44Vo|TYrEP|I!@I-T7koP8F=OvC4eNocu!cUZw7iuH<}YRb;kmUD9Q&eTqJz2w
zo=n`$i))aF`#(t|WwX*4vHg2qwH}6>LR&vacTiy`K5dv7-^s#+%{bdHduU)fziu#7
zoK%1tIP5#0#VKK7ZvFJGKLBRHzxe#z@CP{JOxIXPt^+KOE)^ADd4f^SM5*A1OyDOz
za*_uX?yweRRh>p?DaM>7tRoft3MT(7D`D4Tja4_z;)=gZ^1*J*dI~)^k?k5
z7+s`Wa)eVIET#po9?)PII?H2W{oEt)2y*4)CVpVKe8ROex7#c}0Z
zXLDf>nRFH-gEEYP%14s>$RPY6{I9ml&vn>6|HSkJPaf>_v(A0(em>UA@m96#e|y4$McrgUoy7
zIFg+BGtv^Hz^leuUKC1_A@}&&`|7A@@g@-_)>dg@M7MspcJYY{GVdlJ7nuDEW}Ws}
zVom#n4ww3+mZl|-o1avrVB!hmz}Ra
zHPS{3Z`CjO3~=HPBlQQ9SwxYKs-|0850nv?vk6ZE_ip0GEn7-2qpfhUi`AOf7s+t*
z{H`dsK21b7Th_5%+yL49ldj%((GQ1ss|F9>$i#!i4n5{-cz|3}IsUFT|1mCiPvN>Q
zc!6|ujCAUxgyQxBH>gI3t#RfO56-U*`gk5`{HQv=5mH?&zTngJ0@q~S`l@`z4>6o}
z?hNjGhvZ1>HOwR%B7=pJx~bOB@KAN`da5JNh%)c*amX+jKOx-A{Uk31uavCR{^WcW
zZ{4a_4*24XNGn$9oT(1P9pXQB{9e>YlB|aAH(1spkCrv3i2bu`fGg}UJKavHL#h}(7VKBk5H$W?
z)0Jj`gPk0P`bB#Lts~>$BzcBBq<>v_e`y}6iSiJbK~nJQdr9?{DU~?4&bbpaXEpJk
z@^d%n?xrImANXSBF6iOu&L#a$*-NnPbx~PMWEZZdu+#IkIfdU}{rXrxe-6tJxTITl
zQWLS-byT%a7R2>`dZ?ZqW5l=z*Oy(HvP>X2RF>Pjx;E-@vNh
zb>{mBC1CqZ+q2sTA28`HaoC;Ed^Ht^cC%RckwZo)6dEVVkaG;n_hhf9!{a8W$?V3f
zVU3zltxS;yj52rKM^i%{fBf{OeDm=WxNMk**O}@8>;cv0ttGl%Y;r-#;UMNKOb$+c
zmYUy(UF-+=z`6`BbA9wj%3)@l;;=e!j2VO7>i_!PFrUKS&{M0*E97AKv0k52hUb{J
z3Y-7oPo}U+R)nj=BSY*`N%RJ6CW@^I04h?$GWc;>>5Twq18lTFjP^yR7G@WshEyYu
z;J;1AxP_V_X5io-S~gjaB~x?3yaulD>Ua`^0@Y2ph`LTMj@lG!Ezrw(ekHhRWN<)>JPcD=lK931&B1rVed|`46XDX5
zN7Rp5OELZISz5-dGPr5pu)P(}#==`(#=fyIfsOesi}Wu+aP+O>o>5gHtYqtTobE0s
zjK+tCjw>&sepgO%IEk5KrZl^6MjU^lbG2sZuT2iPuwgz`MHD1&kQD*#EARO
ztw1|?-a?|(Us?!bBs)Da>C1zu(v4!KF+A`?wFnI@h!1O9`XIN51z=mu#hw~Lrtr(p
zk*;*HBA8VnBL|%}2lj;el#q!6A(z#Y90#TR>fj`l;LQB-*t@UDVB2
zMJ{r~bC;K9_;~uD`S2DxrM5;?(%eM2>uVoWERR+P%%r1nhfF8JNdz&auh$}XbCRHR
z)r{@PH(k*05=J%_qz^S-9sh9OsS%R6Xs@XJUIzAjaQ|#D#E50-p57sU&<3$M24}l9gxb1Te>gN~6b&-0)3`;uRWxc5F1Q
z+@a{+1k~kjb3UcC9jdzi<58@8J=7zpSMcBgfzy3EJQA+uzmT=taNan%4F{
zXxH>?s94JY&%Yr^X
zy?63UB^Smb^U7&Lq#NzH{QYVo-vG2-&z*9AAqVnxInj4Ss23W}5>Z4Cbwd5(4+_eI
zyP$kApJb$?30mx|=?_`!fhJ=k=8X93AXUzibLOI5kg@Hpk)LJtkmq-rNJZO5NSKNt
z{bF4s_`XLDwbBulJU+#@6mWCMI%ISmH^=|@h
z@*(In!L+v{JOC+nJmuE58i1Zb(?t|<)zE>VmfJW-CYnI&ezv!~1#;b3YqC)&fdpTb
z&3XD%Losmm#lu4=#Nkl%>3(ep>NQa^_3?Wr6leJ`Li%VQlnifKo|9^a&fPqeS#8(^
zB^LTZ4_6;UDi5b#_?c9rVHvQ@cd8mRlo@{cGNBUPEL|#4I+TD$U3~b4S@Ruaq*1Wy
z-rRs%uAN#+9IQhRoYZMVC@WF=J4S6z^Drc{EpQ^=I0u5pTu%C2uR+fQ7yEn7HbIk7
zN4VcnVGvb7E@=g076f%SS!qsGK)R&S;iC1$5YrFJlFiIQsBgJkokKPPotX@&$0{PB
zOA_Sv>}qvT4eQ6QKj#LZ4d;w;w4)Qca7Tn)LAV!UGmP@rmRXeDn
zEWceVVgM@Q*k!NTnS~^~Uq2WtjzsI|Kz^g(5JYj3=CO}xD!M8#m9aeV3mO{7-p!2s
zglvwdE=vFCLs?Z5!(ETeLUr!BaT_E*&{k5TG4g#XSi1NSg%ytkCg%Ku)`zPZN)UL~
zc(l|XDooka*UJ13eIS2+KR}Wa+fK~(n4$Oq?QuT8gdHx1R>O0I_M2Ow9go9?92zqy
z#hS^J{PRrkx|#Q5r8)+#M_oSrNGlI?xuo1^H{)nt9YxLK={;h-_xd}Ay{Ze(+yVI
zD45X*QX4cc!Cq9rys>1A_#ykxg5C!yn3I}^EQ>`XETP1RZrY6%$Xg>
znrH;yn}@W+`XAuYw0CV-sT;#v+2nDI(qv<>KZhPE+51ejRI`ei9=^%x%kcxQ&el|B
z?O(@|T)Uyud5lP3k7fD0VO1o#9$^yv%a2dj%T-!Cb0ZY*)w2#;3E&1h3hWB0Fe2ZR
zpu)tYiR6f8==uvWA)Kbp?fYT$_@A2zlv#SLhCHR{Nm0ZHRmwe)GN6#}k*I
zQQ|XKPsO>Ga=xGZsEy!5%7izzEPhH|QK>sh3Augqw@BoZXnf^7R_W8`kL1`n&d^-H
zio4GEO4WZ+$2U!Wl{+(8<8Fa=9+gAykYBgV-E#8YBbG7_`kY7Qk@BktP2vFt$d}_#
z&7%}gME&Q=X=$1SJZ&mj|7uY-K9$C=UGmTxziM@Pl3~;iM_>P5_u$LGAF41g$dMEw
zAcFyUx6zEGovfD>c@~4$AG=aorjv!Q^b1jz&E(_$jom8eT)!YQT?*q~4Li4M=KtiYG7UAb#m$2336U
zG_v0~L3YA>8?j(W=r>TuaoUSDf5y^!@RMz~sArFUMN(@%>pT@%Lk<`=J)@ZDYwR9&Ww
zco>djd`mhL?~3r_e9yupN_G$t*&6eYj!P5f_P!na6)9Aqk3oa%vsA3kC&NXcnojEsKH
z=f75LgY3P1^+y4=LJs;Kz1dJo#RFwbH=~gd#HqB%|Guyjz8|A%TJXXWuSkA6U^96Y
z!Cu!tU#3Vyo=haproDWR%c(n#(UABe^QT9SzbUjrdTY7Z2W?I8x0XB3tmO?z)2?*s
zuqTFy=YO?r==y+@?LMh{Kje)#Km8_gtu-8Rpf~yzF|>eloP+6RLi&-_q8D?HYWEm@A_VW)tpN`iS3=cE&dLYGpT1y`CR)|3NANae_HT)9WyTipF
zFCh{UYW3A7v^bTSg<37$EcQ<2%V$pB64=$=^Lf%_JSMXLBIx%r4rg5GoS<6nhuzkh
z&xt&ohy9;kGG^tg#{PQP-ryB>#w>SvXn(4{hp)%%$HU{F;l&-c?1j)RY{WJ{Ga#S~
z{_*(HCA0OXnC;i6f!B5`G4*SG*FrTpkU(<5lYV@Bh;D?YqTBXgj8{ItIl$~2JXy!^
z*+^jm{uZe`B5|B>9J%{fNtcxZsY>$KQN7KA_uFWMSDr?BRC;s}Iv!ry5PGdtR_o#Z{
z5bTN9kvB(6Uci0Nkho+nNo?ZR{yP;mDs1E1$!HQSIk@dox$}xg4^|3zcT-scFy*3O
z`EWxZOcI_f_N-H^Rm9Ldv(
zl1QoVX`6IG_NB@mxjg(ZPL}Km6Xy{zPri8cJg7p=a$a3a;2nT`jVco|O3Too3?!cF
zhX^*;!Pt_v(+W-fS+siD&;yz3sGOe*?n7~#Hf|fIc1Y{?asmT>39js!JaIIK4Lk03
z^2tEbG*q#wLR#@E77}#3wd*JL9d)j4IEP6(z$+Yr|X}R#-#?xNiF>J&nt20
z()3@p-&i5qJ)z~nOIe7z?6$jv9STG5gl$^G18wL-wUFb>4doES$3{>6PvIzu;S~Mz
z%qLJNS#fClY1yY@+
zojg_&3#~NueGrZ$$2|E!hc0AA9(vPa_Wf;phJ8E>9^G`LQ!SM7|Qjev}&1
zr{qIb8FvpqRjEQh1ikrkw89ov%ayLY_KwmV;SqdF?4u5(>;ya|d(JH`sjTAP$_b^!JP$`r{`{bit
z=?HY#EI8CBr4qeIZ5bc8xCt?nI>q;hpT$&nB<^S~_Cfj=l+%rsUZFR|C|S<%OhbRE
zc}xUy_fedQ=tP5U>PdLmhau+|$AbrjzoAkdVb|$mXkZ&Q^`3jy
ztk`WzKhvl~D^QgcXR_$x2HNqw;d+=XIsEq;hw|HN!dRn|D<%S6!FZ&ZKTi6y!77)>
z+c<35F_W4H^UCe~u!lsONNR`zHt|x+e`?YIo+xd4ba;~oK3(Q(oIkpUCg)wHq250P
zZ;*zvg)Lvfiuw=d9QR0qZ4yxP6|rS#npE1DPlz3!wDAbwO*oD5JU%cpRMdsjUq^q5
z%YThU>d^*yU0XTJ*5
zy>75C(Yy|SJfALm{`mt;pI&aq^<-j}>CtpZq_zxc`OcF_=+NjVdR4w#aWsn#O76*I#G^@_R&
zV2ckI#11MqFoxPc64sXl5dQe_U>BjUFxP{Ie1+G8*l()XO~`i`7Cp+N>u~WG=0K;g
zXJdXH>9TrG6FR>D-{l=vW_hy=XI|-ldhG09thH2=Sn&uU!TT*k9rq28xi%Jv?aw~U
zmGg9A_5E)wpxZ2@J&_C{J>5@p-kS|SnM`-|6q6QyaMa{Oju10G+O*H(oj{9Z#5Xh!
zk}=^_-_LOD>+vEU@0nYamo*WVp(I+`ix-fz40GqPDt_GSR@~&{wG#-GuW$4u`XXXg
zZNr{)Ruw^inVFTYsUwvslcza8+`!8WsDjFZw2+ne%pW1d3a4-^Ofu>V!FLo-ZVIbE
zM>3umD5xE?#J|}_vg~l%BcykYp!UIVT*-AlF@!z^NBYi{X2`4KQB!Y3GB>pm3sRBW
zx$ht2ET&No$)&FlXCI-}E7wX8iHHe-txeAa7el6<@zskFO%e2zN=xd}^unHlRue1l?_qJ36z
zO&?Hd8?%Ay{Mi%GC2PV@+!uLaJs69x7AZy=1b)JEIo7Ekk!&L}_Cm{`fD
ze|ngU94zps98ym>UwbpVjwcgVsl6Q(O*<{OhDgROo)x`wq_%Ol5oL0-kb1#y)))xHiy}m6B)Pk72uWWMxCI01Ssjn4k4`A>1-C-dh=FRVU`|=@h
zN%Zf~wQ`PHu_gB~&PV#SH;s-rb$%BBNd4Jr;Gt0~L}6N?K%oQ5!=7_$*c#V9NWZ_`
za!an(dWn75aQ8K^zI#^3GWTLFUw`?6eBd!4JAC^~!%sd?h8XnfA7-iDkhQy|$W2;H
zb<5Drp$LGaY>VcrS}cHuJ#2iqh@|%36xl;n*Q0ofT5!uR&}@6xrLI^y&1-6cV&af5&P>5G8r
zt(Ei2gG)f%#QhX|{fSzaa};+V6L;V~MoIs1_7NzRjlv%!Y60QIk8bi5vOwfk*9%``
zp4#@ocX7QsL-<;>*TO5U9b|U4UvRjmA5Yy=J|pBI
z|A(RT{s-!d!??X=@5&z8d*5?~D5I=U6d{`wWn^#49-%Ulijqy?o@>uiGLy*q%E+ed
zuRq}aa9`*3I*&TpzxBjqQhJ%P}r=>X|OK?nUS=E>Fp%w
zy=;S);w=I+%6uXv#%c-v9uXg3BF_YJ7b9#J|71h#7R>qPAdi3QdFd=U$c7iv>Z|%&
zQls*pTon(d*Wk}&PFc^@4k*(5+WYf?Bp#Y4UFIZo=QoMWfq1=l>JGFS-={nzzt)jtm~
zOC>h9Hfw+YDoMwG*TUgR|D?5Jl^r0_35XaIQ-aa(k;Sm6BUE{~lh$`N3f#PzCmY|E
z3?3IMjm(qzKsn7=62?%6Y;?nl$cjz>&^N&TZT-Um^(R0uU@wK`mK!AD
zWu|F~mVyqWOi>-rppoqzc8_c0Z1-34hfVN^^rIbn?Ru;{L$;nc}
zY_*`}*FhX0H}lQ;$H-aen;|t(Aao7hKI~90n=%FlmlR@C-k<2K&$-e3Rjjb$?yhN)
z<~d;R7gTL!9t{nguM55>djQ8@-ah@f$^wLujXvbp*MLY}B~4$uJZz=pmoe9V3sU!M
zp5?VBK*5KZI}YTkuuAH|(b%#bc=0AiW{24eG6@m*D_s`{#KnV8KO`O@XX_RhMh6$M
z)1>)h^=wlJrAM<_)T;r^FI`#JHB1od8ONCmD_3D<$>k}P*S;eQTLF$S-Gi9r48?)9
z*C1kH!%#7lD+xp;SR*|5{$Lw6lgrsbqgYrh_b;Q8??}Qq5evHUUl{4@mohJy`JmmF
zk&2^{AXwd~h>Pvp!IEUf-TauoB4+AW?~#tKBi3$apYA!|g#@NW5$|>Q0l%Y4?K^fJ
z(DdJ%R*K^@P{cOkPr)NT5UKFo%9rmamOPSG9w}amJt`G-eDE#<(dd(QHZOgIMSY@I
zeis;yEC!lyY!8nh?$R4H8)~sgUL$AhdGQ#GpRwTeQc)OkVQGEU-!U9}@+N}8!+r|0
zz-Wt77Xpzl$6jw{yyK9!f0-BVxkqDq0aRyeA0!}h?c_zq$PDtH<3B{3q7PfU#JzYn
zA`4k>uYReTWsjNX6?F0jJjd2@vVUzT@d4HwMPJ$_qmjQHp}TutpD|Y5j_)phT}Z``
zn!xgaK5S(qrR>8%4DvmZsoCYZ335)D@2@nQI>t8r;;D9!EK=K98?jg-iH&`B;MGkE
z$4)gAE=Z+`Vow*@co*cvk<#bvE#QS97G<)!^_E-=aj8n-9JY%`qQZ!CPT7cJxh~lC
z8EFZO@GP`*ekh7~=|_SmpYX{h4k_De>gI6#|-mSrqr(aATG@>Csl-{
zG3UiNMQ~}Q(Y>uk%D!+~eMsOC>A2h1_@C=;GG;ZR?$hEb|0GoiQDD@xA*}qz(U&I$g3l%l*UF#_Lu)Gd%SYSeFie;q3D{b+I!!p2COPrkoC29&R-BtEU+f;t<&2z^__Skw_4_(joeD6zZ~_Z8c7D1
zF9-Q~B2>pI`U%!&G3aqC{lh*P_8?Vz#c!SzN#V#Qai?d&XtfgEs++3RM=U3ND?A6)
z8*7)#p3l}auG6^g34U2u4|IB;*62-xIdj}HB~-~(hh=^~^1PMmE1dQ-KSjQ(|H%?}
zimzI36nAYrEfY_Hb*|ui)syM!6>;fNmrimRLyj3XnWkSG$tra|9;xiBn|}oI&qp|r
z#9!rRc1V-DbD#DgSKEB!VSWT+%)g;du2OaQ2V=x8!Tjq2H%+hz2M0Uhfq&|iab@&h
z-ce#C9nquXJj@6zQ}g^pp@0##Oewq1J7FAI&K9gs2{5-ByJscIxAxWlxp`_BVFvKPvJEh*ItEzcFvhvAVFP$JY{nB2-R6r%9J05zhcp
zY&P%)@?vMw?r16kd+jO_H2(lcl>2l$%ASv56J*LmOpH|sQk5Yjz|w>j3g^zQH1#6#
z^`(8ECfBe8ispwYv1E|AChl)(We>8x(D)-e@f$|B7)-a4If2Y;`ZytNZF>#gDWX4nshbqcTm5H3u6=+VECytJoFoB#
zd_;C!%fe9I+k5*^%^6IIcF9g$XhrVC9!4rhu!8e<_cvNJgy8b7*0bt%Be1r4HB2eN
z3n=6DTC$JiVX$PMQ1~Y~u;XO++U*;FZoFh9zYgp`3N=+*)Os>p=n1v|<)jX-D`}-}
z9UDWg{WRmV!Z*NME3v(y+uo4mKNo=?o|VvagzA9r`!5qjmzILszu!f%B
zxCo^Rt$?VB@(c|F-R9>#^q)sV&x*=R14C(mz9)P4qCh<)>Pz@j8r}(p66`6yyi0~*Ux@~e
zR09E}PSYQHNiQfK8DMWqRRzy}Q=8dgnF6eg2Xr3uRj|f?zfrNh795^=+h|wT49%n*
z^s)(Y;7+W)?wEE63f=h@JJqXT%C+mb^VuxmrL2(WWswfYtY4J8yxj%l_lpj%Xd51sViLp`kt=nCxmR$u
zN-@jgJTm+Ob?Y6k90L6FbggDZ8#^x5(1}+JNuYo48S-9`zltti&OcX5#e!<({dFT-
zJ^)8YcZDqsWzfCOM`whu%HlVxFd4O*4tQz6w5kl75c)kx+aLR+j5m;2dh6=3;0ivm
zf0m}4aks^nmj{P#;hta}10fo?{=4fcw+F0np^dM#=5Lj8=NUGP;)ylh`|xl5(qI&N
zu8OV6#Plj&6*8l1B`c1T<(Q1~H7$aK*8bRtZ;U7Vj=T5*Iatxf4>6U-T6X9V+s8QF
z00e#W`0N=kPbr+X=JL!_l?A{Nbag%Q3MKwJcU_LyDF8nuADM8^QXj8#s~9(Kx`6Jq
zz3`a=mVQQKL`RC{n&YbU_&_$ir7=Zs8K0Klc9CilY3>s|Tia;^-{}RTk4l
zCH$&vO7c=^8d}G{ovN1Ng}W{`2+SB7;VM2eeIDN~;|jf1)i?gx;7tWuk1MJf(8J4g
zo!a;;uqhUQmQnQ;y6nz}kuq0+#D(YeK}ZA0la+ln6;FcSunOj2tZV^uL1rTfQQtx1
z>ayn_-FDc_ro_lDl?;115HNe40R>DKB+MRv1EDwG`a9Sxz;?`5`4s&KxZOqk&YPnI
zc;>T2y^mJFeL8;ZJf7l5U(P*#QZBy?9+IvQ5WU@lJ96BNAsYvv!B5b#O~M+-%y|1!
zB*f7hr9sc_B>B+bpt_iAyqq|>&pj=yfdXw#&{>~$`v8vjthGFxp27>k{+j0su0wi_
zGNN+Md!R>uR!WjP1WKozNr+4~fW%kF-((KG1o11WYLFxV+GVpoh3V{ogmo24nE8R4
zBi4iH#zJ^y{3)CwuY>kAx!2qCQ$Po~!=`wFJ9HKGsVhm3f)$=;d+tpVpl^IqiLwv9hR>GV?+UVh2ljJHE(i)bqr`$31_-7-8#@q5c|2nGy>PDiJT_2
zJAl`Er?D0i2md_feR+p27`|azskoXW4i-c-^s2`tfF5tyH~s5k@Xk-4zpkV*z=v(k
zhmpb?1Y8ph|2%dDzU5}Ci|8H3g0EcD6Afnr_6qTIdUBF5zaTbi(!(EaJnB;jKeh+A
zQ@rAaT122Q@9ha+NDb*U*?Mg@R6q{Rm&ZauWpLFy>>G()1pFHOLRFE&9T1*LPfzx=
zg?`K!71^8tn8fER0-~zm`!fXFqAdkbluL_qd6+@yJKd?C{jAO{&S;jM`M-Y9x3fIVxag5-$nKj90A$aT7pW}Zjo5py
zR?Gx&DqmV!*E!+k!Z*FW;Yzy&MTq)kP8^w4YjRdNNU&Foz
zU-vJU4ntgQhqGG~0x?Q4HRSnZ1hOV}ohu^YD{}cxOU3!uIT%NuvXyV5G4?mLCp?qi
z5#f3>(3>S6f>}z9e-xG*!v2v7&*@yQL<9}GqqeAhFb=ol*k?a4Vey?>
z)n`dYCv+pnT+VE`%12DX`V8?>t}mAQW|8@!Y7nwOsrpbnfDa2@^;f$w%Y}tuqQ29%
z9EjT~Y3aEmL2U7Z`zIO~FRWX&`Gtt8Hd0Ez=saReh$)v=BnrHmZ*2HaTHCOi2l=7h
z&(w6)3+a45<6hWli7~34rmg4{M#5TtPIP{t#;!KrUs^C1!9I9e2=^)%Bc?{BgON_Y
z$VQ!`4xjc7OvmR-$!vxR!sln8Frlo2E%NMY&fC31$~iT=PSd0z4mJS~`SQFln$k7G
z2mE%3Mdx8R6(=iJaj2Q~eTW|Umo?qBJIaL}l6)@w*(ivN9X(B%mArsaYpy5ejrwBc
z)uDMp7Tb-UMik_ayXld|E`D~>c1}#ijr;1k9)OsC=4hC_>4KDU%aj;a9H?_#d?>wN
zaTeoC_eu~+I*0sNvD;`GFvL{OF$<3x_+jGvj=0K>I97fkrnDhn8L|3}yfj`tkIg%A
z^)eipA$;e%qei}zAhMpUxk@^E80$v=Mrxxa@=v6n(5TZ6E8fZuu-FSh_Bi6MMVK}t
z-Ylb(3wIx4iX_Jiq>SOn-5$}2+8YTNZw_6rm`fRQ`j#&8HMbWbGCFlBDelgR9{Ygn
z`9HvjB2A@#@kb!Z-~VNpGrz{>7gth;Gb*ug!>K&&dtO-jO%1YglW-(^{o_tkTrx&#
z)7e&fSc6daUV7q+4aXYD81V+H2;rwo_l~P)=XI;*G%~6qUGej`162utcUK@NgncGw6}m>Ob>`=PP38SxdU0p$ix!D
zQoy9sv(im_Ex^coef8`a0+5o!b5S-=1@5m}M|}8uAGlr{rY$`2X&wLYMufid2F$ve
zt37jmaKnFXLdz@?=(2G5uw=i0XtGr%1rvn#tFfv}43U7h)HOJ!{t-;LWxZM$^#nW|
zo0llN_Yu}mzaf#5t%6R6Wlt*l(&3|xQw}0rzCdJ$;;p`~7tl#?^E6GvAt7n1)cvSG
z;6U@XYk@AL`1k2KSsHX>XoX|O
zhz#GmAy$0Q$bdHRvC6nTKIZXmE2FPeHE+z{Y5_B14qUpWM}SQGdEkZ+D_+*!
znsoJ%2KsFWXY_h*jP|QeXLiRhqfgw)U8J%E@r2Of!DHwD`c*
zUt9d)q4w<63zq0Ve%7e}7WD9sDw`3x=3yws6`%0$Uvy9cnT|C+8gtYt^z7rkc1QGw
z&f51z>3g^iNWGJ}{182$*DK4bxq|A;@cDcVu*IVu7}s!5`Qpm1Qz0@PHmDx_Y~ov<
zjPl-p#A+WFghxKD$RMil!atE-`uOzveH2%gVj!Q-MoVqCn&xzEPyvnio%`+&ac9qc
z(OKIxG(q%&M!0@4O3O9#bmw6P>PR(=#Ils&bzAcOH_1~`oqQCD;l7LWrOUqN*VRJ5
zP|qK2#hT#i51Q|}?>gYa>`_X;wvy4H$n^ij$~|$P!`&8IQcaXAG_>S)rVaXEa?jsn
z{1R?9@mFc_AQ!zWDs#HTFan*$=amKj`r-3*P&ajm2k$PVh-3U?i}rllcqwE46-Z5bMT2>o3#N|}E&r_j!w}^uzP5*&`_r&9oFBoyj2bqdN8@JK$
zcV?*%w~Ww%(Cqqa2J-mTNfM`+Q*`K?qG<6ocQ&+(I3+kU;358Gyl53js-WKABu1P`
zbkHiRJJb%Y8P6FDPJ
zm~%!>zp4i)i8d5{vhM-Ljyaz$4K+gspXY@)X&OLDz+zo#AU%33%qgPm<`!JJ)K0x=
z^cB)Q6Yv-d`2t#gkTUZ=?}iq{$66X15-8UUvu45yCrULebj4eAAJ)sTJ)(h=AfzTj
zZp~`~NWYAj|B}OxJ4UyASo({iwF;+Xas*~zhF7=0!PzA!cR+{QyB`C1?;H#AvI$=E
zXE|5&z5v$$O0=b#SJ%5{hf$xc!Wf%GL;DrR1uQ!ln;LG)Lt3N0lDz-FF
zo;G&{OyT|0KMV!{@Z+w%{Hq;`v#`FXo(%!&JB(We>fu0E&k(qudIse$PCXBHI|l6&
zqr&kR#q_Vr{o2SUJXcdSvLCCRUSb1Olv@B?-Vo(9ecVv
zUxU5YuR7Wc@B?&Xx?S}
zF`je9MeQI6-M=pN>?UwX4`Io6K53b!H-+BiJ}=`t>X#MbTNm`DGxAF?`5yUpv^
zh0viEV6%pl%GzB{27fUjk5?Bc-EM(S%Yu(~Yyv>;2a%KO7BhUARmNX2P6O8C-`&3X
zg96&_#tdhrHX)u;F-PTBFN3bAiZ
z47Tl*DOmI}ujpl359I#YQ%eUgZeUyIHyQ?Z)v)!GM$2WYUL>&z>(4(^j1(7j%?t~~
zVLcp_0dAIoi0#F9vI~b-u$R~sR*ZWSYi;V3Jo7dO8Dg&%6kEKl>QKYr(NGYMfP&&CNBJ2jDvi8JuQ
zLqp8_+4GX-e$QMjlC4l`TX(4WE(Lg>%
zj{m7Vj6?2BdP;<8%OIz|u*^M6(?X{Ze2GmSpUh{w!&K|=X!%)A0|ntxJ*R7hL>Zk5L|?uD~X
zgQVTal(w4SM}gni1Bt+xiu;_G5<$Yf4Umkt^MUxM>c|;
z2BG02?+!38cP>xs;GYkO(G!1B$vKQ|R#u%&ULyvYS;CJ5&oV-UP(L*yMH$fQFZhXAnt|k82cn=Y
z1|V3aqspC60>`=o>%yq1L6!*TJqll5INe5GaP`48;AS32{LPaF1Ql;YH(i;+-Vgj*
zxU{~2X;VxX=&dP%-$|PPihoK#>GBJP`NaaT>F~yR$s{HqX>OTF9K#CRNM2K@Jk|hu
zVd|C^F^O>SAs2!C{yiX7>7Z72H4J(@OkhlYoCqRWKAMLKS3nB=x{%ngVBp2bdbgFq
z14y$L*o?({!5tE*8?g+5K(C_NWeLrP|01X*t%SOPZukBND@&?~SUwQ=vH=K>SR
zUGf7@K3vaM)Qg92e|Q)!^?V23nO6je{|f*GPM)p?LJwe&kH-2-^$5UC=e0?4Tmt3p
zubJfpj)L|#**Rg4qoFNv|HTU&(SWFnK`c}t877TYi4^EQ0~xOGD{5v2z)PH#xuUcO
zT#{{@`iivx6U@L?$oU66-@7Y0I7NaMhjhyF`E#L!+L_cLawFi_=H>Beqc)hF`j_9a
zrwMp}@Y^QPn*~>BN~^v2h|#2Kl}4`6B|vyS~wupbeIwSN1{w+Ic$xuXe=-m
z+UlZPKlKTldhJpB7}kNL2gzvsH$Kkp;7ELXBK*uD=8hi&=~u7iOwmg2^Stl06!3J-
zR$)3qAv`thjD2$U9lZUUAYb;u9lYcIPp|2z3#dhC=oDu;4}Rmd-vzekDtMRnNSOJ)
z5BkDn;q=mqJHDcMG_?QV2v$=oTxc>sj~f0fp)cFkzRx^{ohbFY*K1bPL?Pjw$`dM_p=m$Rv2q>!Ih^+`
zcHIF~xiiN(FJD3G509SQI`bYnOWq~9rC$PQKH5C1qArG|E4jn|CTrk{|MPf~&T5dy
z+fE&R*a~~Ss!PMXgx4qZd;QG}qhn|ynUc&xD
z(uVFn2o)yjX1Mep$Y#6OvMSOA2PSCQrl~3cg972houYg|X`#@&8D9$N4O>%MNZjFw
ze(j}xM+d;B=0%aVXAPU!)SlX2wE!VPWe0gb@}SroCFxY+yKsJBAf-?HI#~CM?MgE?
zf`^8>!J8bKptYLPUZyz!mXr1qnW#I$<@*s<0r@&$Y_z7vI7tN-KO-TX#wkJlQ03iQ
z)B#ZCAj-GZC>RpFo>4n>{TeW=Uk};-s|5?2V_t>`qbK=m%U`v|8Zad6%iyKjbXY)r
z3z@v>2ttPa61kf-A$)l_WIlfbZYo@erz)`mb%#@`dZxTUU#b7hKWzr+kT?%wpOXWJ
zJwr{WBx2}bc7<-NUJp!RdoM3OI}KjXk_FBO5X0SxJ5%dA|BxGEI*(*DSFrjlkF}s{
zvS4^&;G+322Czc2mPAv&iKY0cjULL+Bk?4Z_DwGR*mTC#@Q=hC;OgyPKO`2jH`t)2fv+QMcf2M7&eq)FbP!J-N{nj5@)v_}99K18{2TQ^`i
zdEaR?cB#O-BXX9`TvAxe)z@-wr4t)S3>kE7tVL)tpynEj7S&aDo)VCuc
z-omB`?@GvwxzO!q9C6il^)BK;P55W@1UH!b~IKNk9WiGh8qAM>IP|R&n^rnW0@Q$KwX-$K{38RijK?voEYVTmu1uCk(Urn6PZ#Cx@84+w7|>&fT$Cw%T>
zF)N6uVp$k6@e9$qnD-Ln1JClFb~j<~K1wmJ*j6;^rs{e9=Ny7CFSGNFMLoif!X}@s
zS5_i+=luA5=e{BnclCHn8j_HXjQg95n?=~WHcAuTKjlc0(C4@Qzk4wr0;82v>y)tb
zf$)59X9@B+$8OKNpb2}vB_gq5-GkV$+lxz;Ent7&_di-V#{~z>Y^^Pyv?E+!T>k1*
z^GCFyksB
z%Mfu2uvk-HNak?{P{@~S&pXW`O8L8ER{wru6REZI5hdG5TRl;*gYap%*>+$~$0iRx
zGDGdxg#npSu8N%KU00%B6+}GW)o*p>Iu9vt^#fa7P#B>d9mD_
z56mTs2k?tX!K2!_on|s)a3xk@;c2ihU<_>aZ2QFr$wqx6AE^p~6;8!CqL*@zEj#|O
zcjr1NI%jm@Ut0?FQEB}`CWQjcOk^y8N&~VO_NMF!>w(X^lG_FM@4)z2s}aYX7@)->
zc;??Z4PcX45}Ptd0(`W54XkOEp^!^Sx{|aHXfo|#*L6w+joptX6ihro{Li4d#0_uQ
z-M{lYDK!){XRW+S<;;T$)R!3z10XC|zrOv!JqV0ZGC8c|M?vkY>&&bO$f
zYM8*SCNHQ|3*mfjts4mJ9tQd=G`M6V6bLuJwmm>sR0FdA5Rf?E}`@fkhA*8D`y;R0HgooDyxWR6$AN
zn=(T#IJm&6d}aD@1`hC_9&eal2X2&{)KdvzP|oIODy2?47!o`vc4r@gyKg893g@?h
zWJIzywG21PdTo@EI3ok3c-~Zs@EZcUlYcnu=l+8Rs$7r%n9-rX#)*brPx7OUEuGP4
zdj3Fp6L}5^`5w?rO{lVHH4PI&r1FO1cR-iXY{NMe!IgGfkDjQXLfb9N=)+7Y@E<>J
zZ_TYRpnJ-m^k*BTaqBcZA>@}9`h2DD=jPBFZ2h^pTxLsxwphLp^Y&-OllH3VI`&T1
z0e$FZl$+K?%e*3mD5~ht&W5I|=TGtDnoV1=x~bBrI`Us!%ko8B@i67Da=st#oF)(w
zjwquF0=_TvYEXRiEe-n^gCRO`(SI;@&mG^GZ28iia&KHXmRcRtnv;o
zyn>-o*HbwKy-osX123VGZebe5KsFju
zv?|i+Rf0B&>>S>mU7D)F}L-=Ds`dWvdF-UtA~9OwYnK{%2qi{B3?%n=bc
z#@mka1?)!ba49aDMx3J%tv$AA;YfBy`BK}mKg3(&v<_ieJ*p%(m-&F-ALg@_DFkmNGG|>-d;AR(%zH${;_>n(XWpE8ON%J2MKq}E^Nx|d$-X&v}=OsvhGWtoX>rbQeccm9bJG@W1LELY5Txmq?9^tsu*5L3y;11I32Rd
zeRPWSxCbT@zWR?myan?;1BJ$7I3VWnwR0$_4FLb5yRA1;KqA)!c~e+0%s2GA!<}&#
zNOCIg%q-f$ub5IayY46W=ce!DNv9`JtNKv3GvN{V<*_f(oE`#|Op2qw$hiROQ$%7P
zY`VaGz+=O1(hA=RwY+us8V)!)48lI?`+`8}<7zS_6XMqkKC2(?0m+|9|Gx2*!xd)E
zk$vZK$nL~-{RzuE&~m1f*f{MSobn%xU!1!LH1i&(oizvG;pQD?y>52c^s1tJlA9d3
z1$z%P{Fz3gi``53nVg}jc$w5w+4Hb|qAJ{+OaK^s>RVK3WP#Kir*ru(5dht63$=#=
z_d(V0Z|^caQ_$ofhckUtgrSwq?N2@n1BR*j7w<)gVcz=d@2|E~VW9mbyMr}b5MCQk
z!;UM0rS$4{P@oRaJV_Zl&8`bNWETPrI8Ns5ay#bAH(lTuOCc{4w*y#B2gUM_t>O5!
zPn>-smcUcBQg$3{U|ItCht>{rSX*1LMh9mLBHWZncS@iV!wIcg4V=?3HzsINHAn@v
zJ|HZB>o+1l7OuFY+l74-3^C4^u1Cr^IGz+WhGD;p1{TNcPQ$axjp|>oDp8q}r8J0$}YC8KEf2fcXiNuCL>9{
zeImma)!4g$>f#oaCZs|26dgb7FO1=A*t3m`^e}DbPpt&OZKT}A*Jm*Q1*UPCfSpqN
zJ+gZw{ccwED`pz*Dg54N8|ysfyt|xdht00Ajk>dVVR=kJ-dndqkaFsmMql>dV!L39
zefZHba%xdj;^HqGL~K2gMrk$*3%IV4%h#WRn3)*;BYs$gJ-_GC7}DE~6p;TC~N0-Om*NlSKS
z>G8sWw0hh59W5aHywgqojVmDB`TcX3^%s(HZMldvYX@m6oGeZ2AcGBRN<`Hy(qN;N
zn?E=2>PfHA^Oq+-3_1@v6U+Bc-s^{Q53K&}BK7av1Pu#G;BUcCt>Q}@U`#H@_T~G(
zh_XeBAUE+e<|+H|m+jUsMCO$}fx(wqeqW$Ko&R8=KaUNu&cuc*3C%=$nBq|
z``N3Ma8qwfUN%M&yqc`EV~D#6>~?6*gjIP0CVLiIJr_S1+az#a!72)H$&ro>J$(b+
z>pOou56cIn2WtKGGq=I79zW$D^)B!MkIiFKqLW@Ev*4e?(mAk$xI0C`s1+>zGouu%
zaRk;QzldteE#S4h@$mz5Td*!qtkR8=lm7`(vgaMdX#u_9v|0;
z6E8|4kqCF!ui_N5TS2oruILzt-{y5mRy3+OM5hzCA9QiTp;q9j!
zZON7&pkbA-ww=0pP!;{9?&Ryt_K_|vzm3ZG)`qRk`?q&ET=
zK#>m9>RyZp1JeuvlUL!nB(9!fIjWQl!zwr4jR7CIomOE*vQ^41v#R~`=
zLQ%6MW=`pPM)YOXxs><)O1Qnc^4S-b(x}bNCghr*1a2ffdcB;(6cv6_xuRWSfVUaN
zX?%zO0GHZG{xJg++PKVIGd?7Wr_8JDywu>wM@5ELF&1`oqpGg9W;YM+y{H;P52xUx
zbY53a)^$*1?2{ckFN=p3Yx@;w?1KqKZW`651@PEVqc2L}H=LAP9hIq_2c0htrq}yq
z@DjG=w+ruy@SpZKo(D>-fv9NBVp+>6_(QUC#9q4v47xt!Mx^-hbBCnEmlU@kM^v-5
z{@xf!Z!nlQjOc|C!@wPM{?D->(S$2y9MC!>bXECaZG(BAQRfYu(*;hmIPwH5Wi5cjfNs*=ln;rU&4&z
ze*0D`2l)3WOkcLo3h4jA@}2LRLJQiVgRf*)z`wk(_)OwwkgFi*33-SUj7Z!VEqP-G
z8jK$dx4hSb()^c13u?{-C1*A-lV^U=q};au%f16#{HRS_bw?A>lxlsV9gu-M_RSn7
z)H~Re+E*86T6-90zahFFdJXc&_V7Hts|W;H_$cn{3&N>|7wLr4q|oO)YcR#YWB8|o
zI6P#*0t)G>^EM`50D**eOxxwup#kG3wTfUlFnlHD$15cha3L%D;-m{35PenTU0g&4
z?~-kafa)W}Mxu$TyloRpw6N7VT+{(8?zCt0tC@h1q40w%dPf-HME~8ClVujy2Sz5=
z%;zx2wGqYQ30}}|t0JiNg9f~pJso*qgYR{#YNAPQ1_3FJY_Ks|Ic+nAjMXxURX(6g;!
ztM#l+I!tqjtf~2>|>xUsc+2KO06`5F_uw*sIU;*;s{ig$@s~IZ|OkC9wo5o7ivp5VU0uk%l
zuG^ai)|iWh|D^h36C`x|Kr8S?98zR%*TcZqiLw3i4O-Q)LvA!uHB^VZLKZ~J+pD50
zu@yr9@jsUb5u0}_3RmtOAGF>^)ZGXaw#2L(M#qpy
z2iO0iXl9Vyf)3r^p<9@P5}_QF<^hG73wc`mn&7s#0)azzC)V?aHmwUZB3}=c4E0(X
zF=)ZYzEnPdS%?_kP{>sU6YA2R&hLLl4&R!R-f1L)M~}H0*xoPzhoOrj^?6G0qo+t!
z*xeI8RYH1Ase$0+9By@_Ww67onTeEeSs^gME^y~@hdyNVpWLSVbsbtx?~4mr8v%KO
zpRZr88$$`yjgMo+7O3i8G+`)@hrjiH7b`4w1E0REd&H-8K;_DX1)KAx@aIjYoQL|i
z0M0vUL3%v~c18ziQT7#q^YpXrh@cu={+?F5)kqH3`)*N{6VO5xk3GGy3w$S>kbke9
zS~V=wxrL~18A0aJrR5J64nXi}P03-oH(b&htaqh+3jPJEjUi9#p^LA+Ena^EW`V_z
zO{(|7=2)rTVt6EE@XyOV;C=^$%v337$eLhA!JKG>+z@yc;^tXf)(a<{t^6|-`T@r8
z;Qis;!efS%y@#|u>q
z_*q5|RPjk8Au-5}YHK?l!@w%Zx?h
zx1)gZ?}U}_iWNYYnu%)d%`#kTxy^dW#EQC{e(TKqwjEv`D}9=-Uj>t|U)ESw_y)*@
zS-bBL5aXF|SBDSovZGjQq7=6?4a$Z_rITNqhM&vpQceSEbm1YfYkVhy@8wZh^#<9a
z0^{_zC@)jssnlTbS?2%{rZJ?9QfEMCkIcV+E)&B?W-d4?8r;GUcm*EN*HEBmpP6Te
z@6zMHA5rXig`Yvg3pjtXEQ+JAj(A@`)%V3Iek=?u_VS?db}!T>PyByn{c0%bz5^_U
zdfpHZ>!XA`V~_9L@y0FcwPqSWP~ru=M5IP;JZQ43^R2a$Tx}qyFJDlbE%lTFgR0WI4(!x9*0qHo{15oJRcNT-wmpw
z?pw%c-6vT*=3{Iu{eldNjT_Fl$A{t5V{ytoqn79|c8*TTtD-3TNlcYKLJnt*K0QVC
z9zkCiQphtk+2ZFV$pvZzoly<;KXKmnviOzN2-2%;yg1pY8mE8UiJzqT3X
z5A!OM;;`^te!uoANVDpi7PXp!=k?M*6-G6J{*o)p{-Rtsnsia@&i7yNt^%uDs_rzn
zU7PmsL(v#CRwX`OI5z~yepODU>hf|)?vrg`cKgPayw*+F$2wI$Z8r#3k_@tn$r}J+vu*mBU(exE
z?dajOOCq@PNqV(UdJ&6~DXfVl`T;guHN6ix3P9hjYBFJ!RQPOau;`RcEO^@aq`c~0
zIEFhSzVfpVn_>}{V*5Iy0q`3+`+n%}BG>+0}|YcA8+
z%;}z;U-eNiZ7lMbLfQy+M;34K1geAiJ-dUKwX%?8cRRf1{4=d_0yhz1Loh?9-2o5fbd+0RYud>FvEL+&H4Ft@NByHX%d?u
z99S-YzICX64RR9b>5rzDA1{V@R3k
z!z=5CJ(%-?c7y)q3T%jI^XH<*MfkaAe|ATl6Sg?CZ;>4n0{;%)MXIPB?B6$D)^x=g
zgh;VVAg$&qRLwJVXciQKH<=X64(8~=C3cUs6&g~geaVT7^7|TMq|N;Fkz@&W^k2=3
z_UCPgl2Oold+t{(6of`-DXb$>Y;kV$qFmrJq$6~xevEyyPS5@pS&Ha2_(fRh*I@+r
z!;&2fh7kP==Pa%l(}D?W^iA}d6efAp;5^F}iiwHkF+BSC7&nd4YHW^vo8D=^
z9(&vCb1|9B4*NbrT^t?r;6$&Qi9c@)$Iggqx6@_5LXz}0;$ILIBjcaqZ(8g*BGF%<
zR>-XYEG6d8hbJ>p$g^c@uiVdXFdK=nDaqxN{TG*~3;i-#K_>^l{&N#v6OYxOcGIjoEjRHtJ750OuslJl($!Q_H(`qr^`Aou!7zH3Ac
zW44R&xqE^{;6}UMW~je3%;(c_wq^B!f*vcYnU)b?pqY6gq$CB7U(~!duTTrhqDd|Z
zbG(Orxqf6%BAg%st?))OAMlTd{myh>S`-sncKquY8IuzCFDI1v2A$NpIc*i$4zkO|_`QBBK)BP7PHM-f-(skoW?W
zVqs$0OmhW|H#Z}*$1>qPhC8ypV*`j2=~3cd-zv8Mp(V3hnF;(G)3?x6G=X;mA4=f2
z3&FSTf%ixKm5{mZZAMFPKO`<^qFUsh1BRqs`8MP9_}zi-=U6!v`r!^?cZFXy2%cQt
zEzSH4&tx>+k@O!29<&#@4%rCs6h>M_En^})F+o(!IqN<2SV@{vx>^h_F1Pe**H=T&
zDl#BWf5IJ3y-qWc>X8
zKiB!^T-S5n&;7pd*BvCSwO9_DNb+i_wO#@R+w}f8vmB`08m|AVtl(%)sPH8VZh)LG
zzFb&v+=lF>ErNr&LU_1{yN|Qq!IX?7_8~3Koo;x8RGt)_FsK}4WTC%$qPOHs$|$;!54t5aeG<3oxyQ6Sgqm`*@&Bwe&xNsQHX!H{H!(>
zb`KACd(q1%=Z=P)dV84X(})M%31Zs
ze8aD{SK-kK`k%TSUf>;uPgX8_yP}b#(?Oa8R=Ai;Cu5fJP5g3?eR@McDL%n#_{#A=
zKRmHIH^YxbA6^2|SqN
zL%k0u@w##o^kv-BGCwFXN2K1V(yNfeii
z@_n%=#)TTJX{w%N`wbb^z5O-r>Y>ZajDjuZ3^+@GP$`j>9M3%A7FAS5itaJlQMsv+
z;vT2I(`3FSN3Wz8Ss|XMaGT@FCwbIo!Si*W<_G2t;I@tBNLbk%q-7DU6YCbl@p`>d
zqTT?Ua1g2`C`|)huba^F!Z5rM;%2pWr3-YZHuG^{JgAuH>-2v9d7z}^#lyz&4Vv#)
z%f@!~0M;)#R(~}ibbF)2Gf7K@Up&!H@gQOY`gA397}5;@5rI=ORPS&|)>*+ew~`J|
z2ixMdG!(d>rNeVStu|Oaqa-0u#6kHxb@#>e8mRt(nf0@6DUekxtBQO#3#Szhc$M_)
zfuk9jVwQapII$%3Si|o*j9DV>x%@R7#FW6FCe-~vAkpAuU)>w<`1^kF_`?L4eCyb!
zla_&i#W?vOw1KR~^&j9|;BfyPt{kYeP9tPV7X+#~&3_%waDhwqv9_20UIWI@
zT6?A$vcPk%hI_l**&uc0XssIN0qG16W2NG)K+SvU(_E-An9AduYGSSer$xt-ws?X-
zDs5;0pOq673FD|ctz-=z|zGANxqM>0Ke1E!@FjbEcw02V(>Mmq)spwzSA+1Dru44tl`
z{yT3CO#b^pFzCMowATIk^W+5}%Mj%{F`XH(J4e`yZ-2wSy%ImJ5hn*FHd^2RNY{i3
zWfmGLF+5<2>VA=1DlHtY=WAPUJ;a=JR1q?H(bm*CKmWFPdWl
zK^9fh<`vkM5^}}*^c?bQMV2DD$Q}8+>uIuF<$xt*dZzOe91&^;3Y(@6DF|nRdD+_X
zAM9|MUy}K180L7KF5~3!IK-U&J?V|hsaQ9na{Q!K1@hCRGA-(8?z`)K_@_Qj6zGbw
zN8EeO0VfJPUGsc+z*YWtO6SWiL(b>H{1T!$AUfYC1b;yS6mzd8^PgpcnL}2(L3C$9
zK$AtmgpCe_k1u%seq{k(3>^zMi6n!%nMhRh@(S|m;pPb4?JdmL)J3uG1{csLnjSYf
zu!FGZ!zJ>W31rJqz_)`Y7x7E6x@TY5j%94<87#xEm|c3w&Ea`7Ko)qqO`CojQ7>=l
zhxCV-c4N(nwE-H?)`{NYk`{oOJZ4sJ|J;Px36X9R8(&bk|XumIg_%>NNLJYZ4*Wmu_UG0^TjI8l#(
zfY-RL3m*IT2s$x44HgYK00(u=iypf!@Y(aLcLK$NK{|+O9WHt9);e|=Z=@qy3rVqO6gQQVZzIw
zrP%;C{+HjrC|-+({tM^K;C+X7((zAM-oVi@v$@5`s~&jD(2VSESP(wY+A5!L*%kFR
z{doo$m!ob?{w=#x_2^4;ODu5I3OzoYdh;Ze3HsYf+bHjo4_;jz`oT9p7}wODdh(Gg
z1r7gp_Ok-}6`aM6m6k@95^c&o{^{N&chsmLbl{iZpYdZ{U|$7o{o`@1e)?$>Mj}L-6lEvbYj#A>LOPcdPyqhEH#%jz6U&
z;u-&rouCh^!AIZdUt-XHi+g@Heh|o3hfizM=YLMG#jjkrZzO
z{zkr-?PsAgdQ0G0TiXRYRIy;sdFu9E+=(GdI5XH1A5ivBhOF9XUYs&{r`jVlGljLL
zX8bcS^oMFz^GvT3
znmnvC8C}17v_FtM;N6$U-)MiS;n3tn)k7%QfA(|Z%42ND4(;yZ?Pk4Pxj*D^Jj_Ju
z{=OJ0%Y?L1YSXaCVl$|F6m?dpkc64b|>n!E77qdAVgq<>-L93CR+!M`|6hI@Qc
z+=k1b8<%r(Jw<7QQ&-UITTzz-7{-`>;+`=NXyB0$pKZ{ZdcJ=
zC#cq>D{UKk3EYqSX{z?t4(MpG+9!(T1Coo>$6hjrf-sq?uC-cQIP=a(;@@>sP;~RQ
z&9u1)^d?W5-R2AeRZ+GYxY>(kB
z)3maO9!Gw
z^6<)1<1d>#GT_`gCuIPe9MmSqY;!qHL43dtCF3~_c>cm#*zkW0Ah)>;*^B>zl^!S6
ze#^=XyVw>VR8i?dgOot)cw;G;rW>|Jv%>)VyB_@aRhR@C22gtIF?>ccbaJ%E)8!y(
z!JV`f4qCum$v60%A_t+9_}yZ+){Ctsn{Da(%pz122CrqrCBbB0eoXr1KUj+2ZtU~s
ze(d26_Z53EfKa8llJMO8j{Rsoe)4LjEaVA(rC8}Pg%GYSAYLLO*2@&NSZLlXRin)_WP&TJf0efWXs@Ju;uHTm
z(*9$8NpNoH2JqGEN+j-uDh=QvEe;o!EpL6c56yK@{Z>i5>gm3Mr1C>xK!v)
zYa8nzw*6Aq-zq;q9{mU@7}tu$PCMx&J0*X?QivB<3B7U1laUQRiG9~2?gd+6@4g$R
zXVzzb=5-LFHMp<7)iH*Ap)H%5mxx3XaX#vIZQ+=_o7VcDGeO83;A*TMABS5DYmc2I`AqlfhzDG!`Tc|;=RUzG
ziuWb`B?|0r0p_+{DuUo~v46Al)Di2eZ)?(2J`$pZ^b_|x$BA39=~I0UlZ3D7HG8){
z&|=?F&D0zFqzK7p3R7>LHsY+%vwTCmlfbLN_X1NMBBsqW1SS64BQ!OhJIFpmh0*&3
zrtTEf5GS*@0a2m||rtdT}5#DeO}34;4Vh|z(sbA+8b@_stW&O;5w
zR(m8#y4zI{aibZtbq`;xKyAiks$l?Od72mzb3zP}8@OVuFK&#bJUd_Q{ni@EWKH)V
zJB-1^MIyXlQyHdxkUe%daRy6Rgad*1B$02w`@Jqq6EMq@YUvZ4wg@fqe1oRf6$|r`
z*mfD=L+&jN`UbZPU^pX-taPz7GIxzs@%X$c7SY|acq$?QAyY9Mr3{k6(7X&0id$TW
z;pKCy-?(`(`a{%kfm9ap|D_n0MV5(as9d<3vM_{AnB2WV;LAtodC5Yxt*epx_ef12
z<207_pX2XaQ@YTLLo*dc2C$Wa7cWkv^&pDbKb_N*zG8bcY2Pf0wvZSup$6MeXW+;s
z1zn!?A|!2U)BZ1QE;i~jBUpc?2szkEoZr>xz~+Vv6zI08f$pN(`KW>|Jus5iahH2JXM>@L<^1maVGkbNEe7?nqNcI=ip&Hc2cYrgLl~*g{!7k
z0aMxAzb0a0XnK?4uq-hY_{;}I=3K4^UKx~T^U+1HqMiLD_rF!RB)z<2rg{>$tlIw4
z5>*BHcTI|k-+7QAw@L6RmEkDY#;1@A!)nW
zc%)m)7RvvKuLVOi2EDbD0;vB|t4}q;j1QJG_HOAZqn9NtR6ibNew>|5gcz=DLDKG=
z>f-Sup4H$l`S`VM5S1(X(NIDb=UmThrF}gNqW^aP=u}}wn+?Xq#pvnKF-uN)+rWJ|
zC4AY6=V%V~s_|$wYW2fqkDy#K|Bmu^ayI$YK9BJZk;3&NaVK2NqnDq&%L0F*c)Q*K`kCkv7L&N!aavaIE*h4@F?%M-kNUm_(kDJ-BaHM
zaKjFN@|ksJe3@y+O^xb2sx>xnc~gcBy}+R+G;)O#Hy#V$&`9ILi&GX38*Vb;6Lrzw
z4$CcYsG+_={e~SsWA&}M(ds7p(a=3x#@qqLWhV_(Y&}uErtP7jN)7yaO|wD>uMnQk
zP()?2p^p}HtY2V@xrxqkbIMN4-bU{ndhT#^ilBouF@r_7cL6c!QCtp%6t0>h#WTzc
zaInWgf^1*JImZ3yCMj9)O$*UW{A5+&gl2GFU=;_tQF$U}(?$r@j`#4sl_`ZP`P4RU
zU1Y>-80U{GUEG83tQs_KS)M{;INW&3j|rh_;vMg{LYiGXMl%-Q6rC&4z(1{1WZL)0;6ax+Pf^CQq1SJczHldD!OtX_A3SF|i#HgT
zM<&KgqZ8>rHm|3hM=7aEDk|$>*bCeICw%aVt
zA3WMean4H|HJIQj;eqR;y0WN|-3ar)=ZyHLhq78rW|X+LekbE`qaR0e_5xR~{U2Cg
zsl;u6pC5PaT0PA}vjMN1Fu7kBvjk-BWM^}~u7dJi7Xk_UL~tfy!UytlqQS`HX@A>o
zV8%YcyKepiI<$*#e7_$D-;0U*FPFA%?>7VI!LlW%IW3T
zb$4;lQ|qM|NtXj{SlFEJbDzX}G$z6ve(XW!zW=_qUKj!_GiOvxv2r+ioLc3t$dM1E
zd?JFAX&H{~bNxO)UJTW}=c^ZfHh>w+Q#x+sMUeWX7W);B5D@y(srYGpI%$XyHF-rjJ|ypj&JZ`nx@j{NKnMzQG)436L!39o8>vKOT2O)6IZ?gz|9
zf3z{vHp21wO;(nzEWm&K>DOj^1Hj+*Mf={-n?j=XmBrIM!6+nzvQ9q(+!hKns=ork
zU~&3&?GOvtBO3hZwb*5#B9=>yFerf_nbQ+}Zf+p^AY+7L+X7y@W3;)YMhCJ>)?e+k
z@Ez&Mr+M0>6rkcg>#5y6H~2QKDdyNGJ#Zpn2;T`9CCI-Q5dG8i~q`RKxiO-9~Iuv9Eg3pB`2W5!Rj0Iu-GqL+Bz9C3R8=*(jY8?63dhO;t?lMH$)4e}i$ckB;o)3w?p4P=XA>Xrqa5VqNvPpd7ej%|dDD%%Q0W0`t}Z_j;{
zMcy2A?pnBuAtGsRJ!0}loaeEhHG<=^Nb0#A!3&YW$af1(;Uy7agthTwLUEKB_U{)(
z%vD1{q`l6!xvy0ioBbR4rff6}vwYZfo+JBiNcf;qG!HG5k{>~=9?zGBR>3?AEFo9
zM4Sp>ejm`7NPLn!K)wj#2?`4*_T{0{(bibD{sG*ukdeBNxyHU$~JNq;y
znbKwx6wY=1@U8NhPDPGO4SjcVu@>yxWhPTiW2uRfc-B8NoXk?-HNAks8Kg{Xs
zZvm#Z$Hr#Br*01tBm=4Wg@DAT&q{KG9yOul)
zyZc*l#R+RhPW(}b^7l2w2IUm4)}-np0~br_FK`%RR2AQj1z&bU%G#{v&1;J=_k-;3
zK9Xw4LGO&6$U|k!ph;-`g1kENP|x$WZkIKtsW9mhH<^R5e^ISAVV6c`rJY>vi%VgA
z!jEd+_b4E)uWT7>b9FK2kSiI;NCCF>uYvJ^xQUz!by2w;y^ft%%b9ibIm&NL#6@!@eDj0Olv-OW3Gbe@wX5bevp837?}uG(C0D@cNU+Pir%0=zPGS`*@_0*Qm(
z=d<;>L7Pdg{v$^gi2qJcMYK+WFjF!~QVAZYj5E_O#khm^x);5Mo_{cIt5@#QdaIbJ
zwCt(<|7d~V`4#bfJvkU=Gg)wR=ss*@3*st$FopOj)>rx0EF%Setm4j=6maq^VX*6-
z6tJ9E`0Czh1x!r7rH=W1MpVCvZX+#JuwV0MUEV7e&{mf!BW6p00tFnJn5_*wQM%k}
zxevkexsM#)qZLp{H@eqoKNJ4v-)`UQGY3aUxRXycEW)3TYEnkVp0LmFqBtNVf|Yuc
zxGd8^pi^D0-FT-IGF?xpRP6c!8GQ6Bf-zeNPy9-Lx2ghfvZTt>%38u)b?wW)v~oeK
z!{4!jz-n-x2|4z9`yS}C|D?TAX#f>}3qQu4jlh0I6IMK$3LPBNkSk~boHTg4PXF5#
z{yO0&N0s3o=fRT)%;VWrCv`2daXdzgI}AAy_Hp%W?{v%vg5^S`R7qq%|d
zV+r<>1kGKZcwZCp0xWg-BdgEM;r9an^BWocaJ#4eSWAB^^qw}Le=4~GnVZgC{ck7&
z4ogJdsa?ti8{v{Q%t`HF*ht~gLd|6w9~M;X7EHeJp5NdlXx7$r5Bxo#
zvrAb=gH3ke*y;`b4@VpoGoM#;$q%NX*3p?qmN5XgDs?Tt!JlC7VYkoy|9%7E1L@xr
zX5#?3oD1&Y>jb3nmok(yYQdiAyxc3pVVKhIyp;Oj9f-R;c*{(75{#>S3f_MF1z=)^
zlVovipljHNsxp5T_$v1U-I{5b`ts1)cew>-DRFCPl+MDrS9t5Ccy@e8r)ejlX9xT|
zZa;cOwG$+B4Z4HV^AkAJ?AqbHbwT$5(g6mtxl~Un9;r%<6eHQdT;V|8NWaF6nYvpgP~jV|VS++4BzBQMH6fdYJUjX`
zB-O=mTKVCW=h7R%WLWok{)89|wF?vJd|eG*&^OE+wld>tOs`)5+uMUTqkiiLlnwy)
zJG^5d$~9oW+w0@c<`h72v$-H|{fMJt6C8kid=8e0C(1tQGysESyVB3W@z7jix#`Ur
zUm)M`^X_CoH#j|{`*Q&ngUov)Ep|s5`y{%>Z`Zny)oGXi9d$>+^NZUmbYbsdJj$o%
z8e9Z5AI@o5eu@U+5;nGd7jFWs`_=gt1D7HBD~`p`3-#b{PvU@?TO~kr9_kwZx(GvW
zu*^wZ^M*skzYDx9Zh`Q%klfpx1t3HJtrYG2E6~)+ZXCJh3i&Qi4vG!y!&~C-Yrnj4
z1r%$qI?cDDK<}Yi7Fq5+@RDc!GTo#V^q{I%VNB9Ks_2ep>-;DMY)x*Oa%dXD3$^W%
z*LTezuKR|A(@_PqDTy=}?vcQ^oDMJ5|BAt;KgnOr7`L$MeZC@oS0fPfjCi)-KoI^{
zmdyB?>i~13a+6Nk5&&y2#ARunZUe90A%9xm{~)uB+LRc{8Tek-X6m}UAh;Ir>W^@K%T`j59>$;;B)I1QQ=qopk}E^gnmsB`lhe-FlaSnn?=6H
zX9Gl`Z0=f;t_TTefwo0AyauqdgIw~W-J6(}>?|2YjW8TCw-G%lMFAc(Mb|!}X+gef
z-0e^v9m1@;buHhCOd~0d!wr{C3jpnz^yg(lE7-i?7x{{BeOO$M1(~wIAR@2*@mJmW
zFveOl%=2wb9Lg*{7(OvLgVCS75;z(;i&P=U;(sz7N~#oKi6jst((akx`gE>(2|BV-biIiT-N5
zbv#xkOy-bxe*~Ml+IJFMlECihXdUKlg(LC*_UUTQ2V;M_#ZNSpgdp)-998^g-w~<-
z;B~&d2q_@S$IsX%V-8&U*dR?bVj~;PRI(C)X}F&uX(a!Q{GBac__q>`jAoaZ6uu0`
zT#OU?5lug&Hm&ZvHn$(8Su`N)w{gdEQcmwrSP3GnWv880sl^cDnuF)l
z#48y8wEy>|Yq6LPy--izO#y@ovG^f3A%{4Ly%wu^E{&aJJ=a#lt$@&EFbcMhg&8I7dQbHX_Omfz
zTg!_F;Y!fDLZzvOHMAdC32Jg72d0}}J3bB)uh+jLvwYu7XuBeH_FmXJ!I#hZPH3|%
zlED@_q!r#rjG_I--IzE?oH4Lm6AkPkn4d-vuZ>!U56`K*DBu=>3-0-
zWqE%kJpbZNTllq?c3rISUZ`z~j*
zc`BdaZ0nNH!rwyZnXWGtaZe(=-OJLS5gi~rKbfz@F40G5KdB&clItVk8OKPD=0-n3
zv|MY3@zHC7iSbj{slV;SbF>Kp3%sp_W-0}VgxC#Y`2+7OkxgvanB8uK6>mL3{~!4t
zXGJ@~y)aPhUrh^vC*`S@T6r-sB*#S+D>x+fn&<5Who%!{?mAyyeNj&2EjK@9KaxY7
zThKU{ml{Aw@{mI2y88(#Qd>Ehk3JB3dnC;I{(DIrS$ps~Vlaf*pqWQ(`Bg%kn_gaN
z+ioGcj(VR8RsKxSw|6B7JgXywZa*KiPh&aR
zx$9BkYDjmfiSa?>9Pz7qi`+=>AHp}@UH=W1d7?#NTiy+3M#M8P;-c5K5|({wC`t@Y
z5CX&wUD9}GqNUH4eX}N
z=YG<3ePm{I?1fa|U2O5QbL-sj8;s$T%m~ZK43_w@INnqF2liD`_g*ITF7m~C_miTp
z2t3sJy8pq@0t&^P4+yDl#w;Ee&C*J~M}({S&l6}{G2c4oG95UL&%Hlspj0^9P!7BvopKT{WRr!>&n$kn*uxiHpZ7(j%#LzFy%j6|f&z)?vFUb2X#1T1XYvg>nDN=Hbcz28V6vaM9XNQzA0OjA_GghEteVLA=3HAt
z{EodoYv|4a)U!VQBR``76>ghzr19T@=Qml2LdO|kpzC1Jl>7`5!!d5?=fD7&l~!0Q
zw-B(aQE#vmWdxGY%xhfRJV%%g`TQ1THaMbX*8VM#6u5U(Hzp`*K&LI@S&vl>Xd0_{
z!i*yxd^Y85Pn&oR75)pJTZ(Ii)K+ON_Cjrt?d42(g2p5)3(dTK#rYl}qcM<7%LxOc
zd*YH?8WB)0t>w0dk+3Q4~X)ByU*%S&S!~m#mMoW
z-Tp;b+R0?Yqht%dTpJ&;KK=;&wQsbjXZHc5X8D!lR^h;Pd5D&RW(qpxe-gU!Z4`$3
zg*p_}*Tb_TN^uEMPngUl%;nx;L8L|KYS^9L2X)qbx^`CL0
z1>nB?$E{rVNILy~HC1)T6-3;lG%}CzfT5d#vua1VEW5t8ddu&@KtR~jGmY#SIDei?
zUUT6Yh_b1h$&9On!tJfBfmzT2Q&mi*p0~E7OffD{^C3)3;
zM>>STvSOSSJk{6xHw1TvO~c%#^w?1!lOOH30=E!Q?s@;w!wnaJOQzn9uDAg_4u$J&
z)og)!>s-)Y#B*SG;vxPa$r)r5UfQs3*ulw%GlFmDuECtRR;RlZj=z5Z-Enpi39$=lb=e>2IgKWQ_ppXrW
z|FxR}5jF7S*F8kKI&;XrQQ&b5OBv9_RhSPt`
z6!3NL?0IT8J8(J?AE5Nge~z>uee9Ea%d
zn88Nb=A-vNv5KsuEt40!h^j=V_>-SMu%%&HIq|t1EMZM%0|T{-MQx-a(SuU^bSlme-g+k`%1-p6wZe}|~xOz#whVb~^eQ9ea=(5(^C
zH>!Rb&hQaa`;b(bnb?ciBJt(5!8|bgg@!1(;3Pu)WlnM>vKM*m(fU#JsNe7(-F2lZ
z{eDDl=(9NU=|*h$ANz@MydSy7HBH()mxuLbaLNnNxgr5}w2d^eSx9=a<=XX$9;AWH
znX|p@H5T#II>kElHuhVwr_f%d(LgCb%=FGm&9jTUt
zybZLxvnTC|ooJdkRY5I}+?cMMy6(@9{Eo92+?>+I?pjUBJ_`=TE=cZ9$e&k7W^@{%
zm8B^1)l~Xvq`N4V!0N?l^x{Zw=XkJTTJ;z)Yzk*S^_UmamIdOg3L=O^H=s;KNHF|-~S-_`jJQ$8+GyN{3s!X@9m%2
z=MqP{ZPoP`B3p#_DXEUXMZXiR;!6IFFHaB%&l07pSqce7D{$jM93L|DSkt_nx`p6>
zd?BDTrJcx0_BSK=buHo7=tahlzTHIaLx#JX_!QxXAWKN3Yy?rUC!J67sdmr<#eWZM#EHcGfYsX+s3iv`r9o{7h)lbQz@$
z9U}CvF~yEs3K5jk-ZI8g=Mbqbg^<(_<`QyNG{o;7&mcr})qY=A&mlgfx>8~O>>fc!
z;DW%j*^k7hEW&R%?z9s}E^9?a54983P9^_WdGQnRU{L*EIH@Wjl+n&$tvrEfwV~gh
zG?7cRE@7RqQF%^0o$-=IYBZ1d%+W1I-nW~e9;{8Hc{h=*ESnDG+UiGG{|Ijx8suu#r9ROu~p5$8Fc&5xK<3lp6c0
zx37rgDaX_Gl>d@ono2Hh&496`uFJ*qkP=`HOOJoL(f{NsztbvPF$Gl=u6nS%@G&
zWH?q?_#>es@;uv8t~JS~
zRVC5Vu@irDyoTthz-N-998E}M`nj-L{E@hNn`1QYQ#rAV%ai!2wUwCm%Fd1487GQ`
zS@`{A=p_!0eF!_u^dSB;B>6+&Arc=dT1vYY))8x%nrj~?*AnGE%h#NqeMxj5Bzc&_
z_LG>C9
zBu}XLknr^Z&jQC)Mr_V!96GP6A$=n+d*87Y8EKp=#6I#`|H_Abgf)sEckMbRh#6*%tux0~iCJ8$_jbM25Vx5eBdh#8
zL?dHV$?8}Fw!Gi|`Rn{+Bwd^0Nv?4dVjmGKF8`JspfT$`g;z|Faw6F>cKtr4bL-Nj
z?aoI??;D%egp%i2xI}sVhDbc35~x$+c)=FAC&mOA{2ReUIjFaqq(~K23j`((pen;E`X>0a<$-%>EhN&Bl
z*~r-<(EvW#=LnM=A>!8MT7;vex6A(8FeY+%p1SMw4tDI9;0kw1GbWhW>&PimiI87m
zrs~$O$8u}P>Ay(LAhazf3WU%9Ly`y=rS1Qpqb}`Ss(+#eiYTnJy*}H6EgI5rj$C)p
z<<ht{1b2_+mL~4rd-?>sjMJvu{HqJI0J@aV@c{C+y?K*n+MOm
zT0+LpFR833kb{8VoQw_bjL_~Z$3vP>K4_$ONq;<51OE4D_C;0O9QJy#f;Qp(4EFOy
zyS&K_IuMOtvGD2P2KDCR+;-H;;JW2$&WFG;GbLPJ4{UTJp8<$K_foYvhgOGzMn)BJ4*nKFEopuR1i
zW)8kwZ(^(Eat7e(%P7@NXCP&^Lf@@!1!kDviM2d51X_P4mk+BK7G&@VeVfLO8uZuP&z
zLAQ@`N&ekYLP7Qru6b+Jp$%l
z&hWO0s(}hsEzWh<<6tB=>%oC>BUX2YU;V!kHn=!(efSy^9~8~GRUC*4!Wl9dwO=1j
zffx7ON%6E5q{fEVR7!xpBEt75JUYaF{&}}{elty+0-K+Vb
z+2gC^yPu>W%i+p_jvp_46Ie~}G`5Gi8vOFJil>4j1??VD_fdHK3NMSL5tzUP>03ALE$>epFGoat!a
zRr;taeG*~${d7m++8l<;$?^W+Yeq&m1SQU!S0bOs8s@CdAA=3i{|PY^&R{ftf$R69
zCJ<)R#C%-80ONMM6sA{OjP2`Db(}nY5=`uUE*51wMB;EtQxDA{j8C=edV_isHu~&k
zOUmDBL>gQAb=Pqb>v=gq$*%qvX}f8IYk0iJ{=1RHWIvvZIl0=Orx$KPMtd%A-8N`I
zo^rVC1Y2ez|Ha4J@17$x6Z!=!-b4
zIYW`Wq<?tC4R9xqx;EqMP96xFJWrwJ(
zl$qu>?SMR${VNJKpEx4jV&teMQ;|G|BLl_g*7rJQiU$^_`sUMw0QRx#fiyGW_9i#3|}8@`;x
zrf>X@u`iG6sf+rj2$ez_h(>87V@2hlD54^wk_;s>6rm_eljc%1s5BCqG*92U&7%fW
zMM;xyD3VZQ_?_Y2KVEA+zxQ41SR8HcAG|Ue$~D9U&T_=Pkk#B-`NZjH&vOk~sX4BV
zJ33eaT5F5mJ5^~!x;%N5dEq5%W`U5pjhBSx53ND&6OQQ`hnF??#jbwD@=06caL49}
z#^>GwomVIFG*WcyB<0;-YYgANP+q^{0!w~w!*~-LCgWD{5WVqN_WB&J#S1l8(S#k%apZH>z
z$(pfx!K=eODXfK$15AaYN?21KocgSCCy6znW~nmj{g}1>=IgU{=W1Dp`CT90nO)7&
z=k+jO&dz2>hiHUHJN3>*A>!sdMV;7KnO+z}V6Fs6=1TbT~!ftne-%xI@`VK+cp(=wF>K6k-MVEDP6gIhRr`
z{$I9m89k!nI7EWKjcQXCFMr;Dyw?bDB=mT#5;#H??LK*Cu&ISodi1w-<)%7n^Yf|S
zHMZTOrtF%++Q4B4_*b*t=fd_190Nn^#5$?J0mtSW7FSYVW!8@$c@j;@nq<6wddHU%
z`N-C-Gk^1z&fs&ulqGfn_>RL!KyEQB_*U~qWGl8zEn}wf29#}HG-roI35m^04<6XrX7MN^Wa?@@H1xAv;
zf=^4zz@J~bx%{4GK-2oRn%|Z(uvB-G)(?&{FpsyVne%EXIISG9OMbKjcp0DTnSHGU
z1aDrvZv0jWh*|q(x8&SXus)^wcuGAlxKgYO~1(rA2
zNLW8B076gnw$0AX1gn;97IWxM1k-;je;GP_8$24C`&xf#8MtpD_{HcZ1-9NUu`~`W
z0Xp3EeZCKifKkcXVbopbZdfy}X%iKhj(f#=a<;wjT&fr@(BcZ;$(
z(BovPub>?U*1Jnzj-BcY)=2(5zM{MU1TUTl$k|>1tS`TR@bzsD;2E}B+jJ%cc$IR@
zs^60U!k+F=-t1EhLW(kMkN*+76m`fq$X#Y?CMh}ruv-^$s{Ynd7@x1EWb1nR8KcL
zt~C-w@!mGsH|yjhN}EG$gK|bT^}Z)v{q@81RC8{Ro$ulZDo2jPJl8Ojs?TWGKJkb}
zZFUg7y!mVyb%{E{D|;-7I#OP-ph7Q|`m{fzeYSZ%rRJ2o_EHW-eOT~?t;bP8c}rFQ
zPPERYj7nCV&`m0!`fm779xN}V%H^aMcg?AwJYVpykNI6eb@W^iivL(dS4z4s%F8LISbMiTi%+ejkw8$NeZQ_8AX&)oo2k%O3TWqAEWOIQlWoi%h0qxCRvn+=lw3r^AD--
zHs||VOUfyiOh2^F^b`jnE$LaM_xqC{1qlB#!!S~|bI
zlB(wUGj?@eHPy1z%Oq8>nmSN-XHB{c#j~i=u)luIACo9^y8mY%Exec328>w(M55L>;
zMoLSY*C$gpg>rH9RrAZsqBbmhD1TY2h|-UX;cL2ALJ2fIbIdO;p&qLo;E-ZkM!iwuRlIYmib@|ZiCyDSMV&c#lXYOI
zk~;SKtDa_Xv(+314#wbZNn8ZQmOTI$>3&*L2SHPjD-)soJf)znBr
z-LTx7N-E%{L{NuHBenA5`o6OT4OG~cXTzg5_0-$(`fYtWwbX#XfiCWmYRXtj%v_dT
zLao%frL7f^PYta6&N^3+NJU4hZMq8_C|#|~Rx6~@1xMs_u}9?&aE2N`o3wNLMJL#D&IebDAu0!5*JE*@%Si>JB8E*YkRRLmgeB4)EbMnw5^mH
z>#<{@vkH~9{+V2Gh78yxdft55&b^epeMOswUFVF$fr#
zIh9&xzC!`c~Aj+dxY!-uYu5Q#TPI56;g5XIoF~#
z=Th%WYdCDnA5sf^ZW;Gp52a+~%zARqm{69%-(?Qn^DiJw@SbIbbz{I
zS(EbclQXsI{rm00`+O<&W=pG=+Z=$|u0!hgrER_M9H+gK1f^U|(cX=$LPKue6R2=`w{57pZVd7yrqqCv(pS73U2
z=ATW`kw8a>d;1jIbP#H9SCn!qAKX_R*fiI=2n4F_uP#zBp$s+~{}??O47U7G<$rxW
z8GL?`qp|KnHUL8MZp&ak3%ez0v`NB~iW-;fl^gYs_T6Vc7Xy7+{^+YZLe6bwa_dGBgY#bgwwVi?8Orz(^(FJd*JBQ&<&95@5|abh8S$
zEG%EhfSLE?G;Po@$_QWuftvgSBA6&dsymDAj>R@wG*l*>wO^H>eEF
z52Z523#x_Z9;I|APJ3VfkPhaX>={3;$pVXhnOCm1XMtA*JJ?PEkAS6Q(p^I}$OpLO
zbxgmN0-KJ7Tb%>Sfz6+hzGsPLV7pnm%pB@RZq^Gt_EVQdKT}rs(@G2^lNc@i^18!n1!vD1wiIm
zSj-rY28eD;Fg(YTNUhvFW#@;MXlfwey5q+XU!A2Fd;`gSVk=w7Bo1T3Ci
z+;WkUTmM@0?>R%@exlfI*vJBeobFEG96L_=1lCv0sJ8bIzAC^s
zRpcO-NjeCTipvvkOb15=1a64FybW&M*X0Tl^`joEb|&r)jez-0D9B<{A*iuTdNywL
z2w3YJx%*)_5WEOjcsp)5ifTG0{wUn-A+^~r^0QQLCRiT%bhY-`WYBvte4zHS4_IO}
zzp(B_EOmO>g!Vn*Z0dsTyF((PCxDKdk;aP_M<649U+Pk|BUq*D^@MxV4)B<(pZ`Jp
zKI-Uhd)DB>t)OaM>ME{V<`hrB%rgrvEv0t;n&1=kaRrMsbLtu+{lL)4M(g!Mk(5Ap
z_L$_1M9Ss%mh<-<)2TN9sNmN;S=4>?Upb5JW>N~V#z9-gA5dF4?N6o2KB7{+FV`Dn
z=1>t57qyN`7EwPA^BPLYmrw`Zt%(_W0qt*V^TqUJQbQ?wZs;z}rap&`h35=CqE0=F
zX#0FKjk*}Ip(T-1nNCUsHeVT9~F
zf1q*ga=J~tC1q}`wp}?ifhvi<+7~I1OSuln6be5oq&UsD+jS3QQW`vke;dkjC}G3=
z5staVRPLyHN8X__O0@fu=abPgN?t&D@s?&5HI=<<>XHoPLu3tll*iC3>h%^wf72^;=$>a)Ei>b+o9TWS+W9Sby
z*d&Swb-4k_AmPj{dsk6btFn6^ue?eP@HnXNmGh${zH7Z1Fm$9M8d%#7thWPQMb7r8
zH~53OhnEc4s@wvRhT&Wzx`YN)E+Z
zbj#vGHbse9J?DM_`OCr~5oMl}<db7Ib@C|u+)el#IWx!l_T0l<~odiE?n8Qt1utt56fA@-D=0^KLs2l6JQ6rN@?0
zI?rEw+>R@wB)%FfKYf>?+TTr)d_TaVxEjSiP6Nf%9pLEtQn!?vy3tx{W;u)UGTgjy
z;0Hx1vbWf#7t~Q+>)aDG?zU2!Eg#M5Y%QmHjF0bi9X^<&oZd*i
z%WqAN-t(HuJKYKYV}mRjAN?bz(Q-D8`7{>&BQyHkf65tq1~X|r(=LNdy&PJfP9qZ+
zQ}18QpzTs)rT@_n#xG;f_+?~fUZm3X8G9x!CVs|_kr_RMOg+Yq884>2f05BM
z<5oQ`3re$7QuPglu
z4hP7u6yo?yBG!R8KPmrDB)qy3cAGN5IukJvNuiqh>k(sqfCL_au6{H?h`
z`ROs^ZGCj4e-K8$p*TYh<&9ZvH$dp*g35m
zz;;R&bFiIsGSN$GyvF6foVqc-acRS-u{i`|Mg?il(4hm%({~a5vXmYyUvt}m{oM|0
z$Fjf7TZ}cwM=^5GCh>&|pMYBc8`R}L!sWvK3voT2bLJQ|ez4$FW&@d&HE`;&!IU*e
zarrtMD{Oz^jSH6l{B^~4XS7l=zFgvp^()P_us(A@70W{sUT{ldgZ?=pa7nO1JIes;
z@0ICc{ameTEFbvRjQuoiNrX#~4W?b+jrDAmKUkmVQ;zl7>A!Hf>Ra0<$d1?Z8l2K>
zFwt{T(vIWQbN+P1|e*Jl=~@#HHXVZHSWHpW*eWL`zt
z5Wjr-y}11GR0Zr$e@PUU8$OI-4Cf~P&5n`&3LM#oEa^iwvIl=I9d*!hN_?dQ|jE>MU(&!{tG{gn=qaf?+Y?G)4dl{3Ase5dz6
zPa30Ww7g9GX3_i8PP$x~mRHfTzA+sK?QcJ==ce_^w0!CrvAb}bAd@#N=zVT8y?o_(v!mCw%DG4*u_d
z*fV)8k$&EDXnVLwv0WmKtuz|#rpxL4)KA+#qsPgJmKncH-e&SEBQyPB@-q|PzsU55
z;RFBb8TsFFVA^5I8D#u~({cWrUzvV0`IfO~{4wz`{bBTs%#1G+*T2ZvGtYXBn9pN5(%>{x34^{+B(|4r9m23^MbE8Fwa5=J_yo|Mr8i
zXZ$kjggI~g8z&PVqi676JH{VV&d5xh3^H;48`r-wW6z8`Gv63w;$p^`iHmu@|Hi?{
z|Am$GKE#X<6DK1x>(-vOW7Y?wXWDnB^>^rffazBv9
zX9ivVn#OP%2WXs6<8(SdFnrmRu3t>Ivzq4Z-PUBky+rGoelYXqJ8f6gK=N`joo_bL
z=Ynv0pN^r+RrV9RowUC&x;q4D!Kx{%IN*Z~d%*O9k>(%QJAwu>qf5HZFhj{22BpcVi~j
z=Um&4{WS<=-a&T$mi^c+{Q6qB#MvP3)pNL1*`VdoRLuWx+s~Vgcy88iUd$7-jU;g#
zg5w9^6la5QXDv;nH=dYTjc~;+E$p|ozx6TF&#>$ZM>yf9x)SMEpYGvBI3Od~fXaLH
zh2W8v4X*q-*@=+HT-O-klh^x)PeJ^+Oe_Y;ueYD-K=^Tp%LSqORgo3Q&-Ms030sK$
zjU8|)vB4Xzs$&SnM>gspz53K#+^_iuee#Y&d7bD|Ut}*(x9z?S#D?rEruGokT8_Oj
zgP0t@)xrd#b@gi0iQQi5%bYvhFpimJ!YzXhIB#`gxg%T`+bQcT!pNhy6qg6@
zT87(SrtOUTH}m{oTwXlA2IKMP)3CkC?~aGa-?D9M@qFaBR>bp)E%FZQeUG~1xaJ8|
z>7ssK*=OE{?5*mq!xX^=F6~dQBV1BB3(wn|M#k8`S>|J0|0D-l=hiH;PWiqOx$}@N
zw)1~A3*%E&qTg9Xcy;q(V)sO*8_$cA)_&NZlo#0#OlFh)!1lu%T<-3#jN^2lAm`dU
zBSc@!PtKR~yeqMtZec!NFPqjTW7P5kxc&B%To~Kj$oc)oZ_@q;dVk5#Ch;Ak_sQa8
zMCS4&cI>r8P6!~lj6M%I({fJ&iPI#S==rq>y2lVavXY>)U_FkLdm0&sRrLP*YAVs!
z)Au7hx8kwgl5%oRw3a3J6)qa2-{N|tzB7HE^PAdGp`;(>hcWL>BB6&`N^f^eA)^DKm+khqUQ|V3Ob{HY@14V*rw+XhH
z611V?ct*F&Va@|go`nw?kK1`VPB;m$%EWJ
zA4Ov?jg#ucj{=>S@6rB`(eb#^^%?%cw8tP5FC)W6i{oPAWq1xFGx~e~*fITK?3nRk
zcoxIsp!5IXpJ|tqj+?pnWX73kkBOg&m*FjpzyCreo_~?C|F`_VWTrhv|6lbP-pKG+
z=J_+o`1u!^eld28%#7oIwa3K6*fH||6&b$_GI25E&XhAUqi2wblR+ka2AT5zEuMev
zn0EiI$E*{keTHu{_RKtUqvsi8$E;JPz7wstqmj`waWFD--V34E5mTR;?~L3>`(f%c
z^Ospa%zeyh+D?bYmo(PVc#g&(dOggb@3mjk?KAgF4)ppSWb|{$x{{^&dH}tDzoYlP
zoAiBIjytiNLEDYdb{%wkv2=cMX(4uN>2rg_IU+mJ@(}IkWd`BDjdXs|()g;ViPB7fm6usQ}^aoFYHqBWHti*NNVd
zBKk-=Pf0|Rdp)VZ7;Lxt9Lblf=seC(pEo|pkbHHqK^;ytHn9Ii?%7K+NIo;zN8%3i
z_=^2R$PnH>ZHykXtZV7O(m!PYE3T>h-V7uUae!5Yi6FYkm)
z3i89=D7Xc(f#7A5Px*s+u)cE}IS)*+b;bOKtBZ|!Rk$AE?OLI^*k2pB8n$1wQUY!P
zY!C|KFz@GsXsLPmdAH3BYj^z8F
zekdTGs5f4cg7~9uXf~X}Y#=hL4fjiFXJ-qN>nFu#A-g1D9}Q&Za^n`}XUqDAYmxq0
zl57(~p7}XiSZ*`ZLcBA@*rN=|9EY5u5q6qAb3wSLzv&3_XWb*qaSoL~+Vc#R3p`u!
z8R2B)dU)hzgS<=rR)`n!oQd=>huG&DISc74FBzyK{o0NTixA&Uyyu_32jbTAI!4+M
zQ-1htI}6b|
zF%8BUd-&IZ1XwO^D0mh1x06brK>gCudOC^5;rjQERxuc7+RsON?odv$xOG|);+CY7
zEv^vb7teZL2yw?kO?V{&-}|(U8=!cmoa=jw(8%Gj4Z=sfR(O7#eV+Ut#rf{K7NkII
z@bl#fydDhYGCm?XF}5C`CnmU><&nRA;?D3@B78p;X629cB4@MCB0TxwYZt;NRri|F
zI7IV`{Y1~1y05wtjgQZ?K0IGDj%=+$eq^h6^LasZSoA6YjrZ?{J8RJM{@nWEzIk%mz)gWNml
zJtFrRJ^Exnv=uvxpAz}&
z6oS+dvOiNM@jl9Wy&R`yi=dlMlpYYP>J1+X%
za+S{8GwA%ko8|=>*>pQ}zGv?1cF=i!1#Lg`Cy8T-&il!_
zj^SNebe!;^<8p@g!A$|n%sFZ&eU7T2$0?T9Gd!)ImUR?Jy%G8x#hfR1)A2C(tT4G^
z|4e=6{K6o^lNsK~#P?sw^cyB;?DxOy7#_)t3pB>|%=j_w^3&tYw9lNI7&~U17@6V6
zOnrt=GyVIoc$o74s~y~A{=fZW+I67qn7H6&;{T`zH!m#1WPy>fW9A1lj_}eD>xF5A
z$pOm@UuWVTqss^B`2jaeY{!gq3oY-aF_&H^_Ov`nBcPE*kN^BD#1F&g{b>J+^n4ni
z-+OZy5xXW@???MtPRlLy`&4I|=Zn+zf75t~ZtwLW;%9=!AbNegqy5y;e#SkCU8O%k
zU%FhFzOV7{CFR2vWS?-fBRqPm7%4AtCOAmf6W>ehvLe6WaZ9H6QJr#P$F3oIGkcOZ
z+?<-R-LdIJ@6ttZHT@mX$Qaolb34erS>rRZ4_(qCegx(beQO)ZFHUxI;Z$S;B~Eg`
zo<4*A9_MKjZfA};x!>pa(8Ti2y(B-uzk-Y9$qgjGa43*{+p4Gm=jE}La@g-2Pm%|V
z+Q|1KO>@ZiHr6p0;gp5cAmL~_)L^>8s;UM2PW&0KJ$xb1j08rf|npZAT@GkA;KSzSK#wO
z3GZ6m|H_g>n7TJzago)6Dli<3;={C+XrT#8bD<-8UEU&6DcJLR%pYRawJNDj<*S=p04%
zo8)dABYZebPYCf&PN^@WEfBTSZ1j-bQ})x@$e+E)#UjLGHR^M04I!TGnyHirdG~|g
z)<2Iz`B|aPOmy!)t4;d%8i=VaYF`oGO>O`urb3=8%rkoK2$Vz4|6ZgFQD@W7^L!A?
zGrJzXg!-S0??2HUfbU64wbw3y<>q%DI%h!CGrH6*4be$a@bz}c@2Bq5l>P(xb&Hiy
zBI4IkwimwrfN{J^3F}`A%k#4oY}rsQx#Rs&vz-lWGw$Ejfp&3w3~NQ8-frg93rbKQ
z+PhRW@gG0T
zle@v3WVG%-yPj=9evQuytwZ|6pM7!&O*Y<%K;v2s;wRBKS(nuuM(5VG+Aceh|M1np
zFr~t{`*&miW}C@(hS8B4c%SLZUW50;+I(e^t^vYf=iS
z1ALD+UjwF0cs}KHmqPt~-!coI7X_jRZz8=~qb5FIz8Utwach=P7#q)$ymS2kxmVJC
zPwu1aRml4>ZV7S@n9L`6ZI1^zzi9m>=dQgH={U~2ZshyB-3r9tmD}VT<$humx0C%Y
z9q;4(OOvp@Lk1Zy_*cKM{ZlRhETkNnz?u!%)lFL|KW!%t}ndN8d5}fAJ9zl
z_ColF%8~qL1I?1=6bVxq8ru|hH5WQ|E
zDc|*l#2d7ZoV$1|$avkILgENnLU_PP+B_U*;JG>+-?#y>k3CHC`}K|F94SM8Z+4VE
z4~WyeiFuztmF7u;$s`Vjr*+c(WA1-dX}OJ;*tuLFcz~{-NbBQh{$+5L=qm&XE};2S
zIem|MpGME;B+l>_VwWmGkn27{TX}*<89qnzM24rr#f$sTUYZ9%DWHA4d36
z8RI;996V^5;m2@NU_HaDyXgM7)AsvmeJ_m+Z{D|t)LToJ!_5HuX`$m`&L_2tb82+{vqLQq7I
ztg~n48+T%Y+_6d15sdsY*!SP9=*Rdq)*7Pme-(HB4`|j)_vL9Bj
zCA>XVnCydGFP7kasqW|zydOO>Ankh&5IddU)%f{p@sazo*bijCJ0JBGmv6T&$9PnL
z8~agocn7Bl8(gqAA$p4uj0UraALWGKSPpO6j1m1!P-H*G;e+L%kJ8u=b>|e8IsT^L
zxZWIniSq=fMk+2J5u1wJnfJ~S=MB*t*6^wW@q;zd2$ygFg3k@}VrOH2A6}5}#&3u){ba9N}&6Dttb9^lT31FBWqp
zxRKw}ZdX1d{rw_GNrd9_lQd9%KX0>(h0Xz8E4G;+UZfxAqk;S%`r&d8#oaMu`xwH7
zR&6TD5DRBROi1Ss>V&0#5>#`DR??LBX
z%rAt$bZ{bjwfzrdoWkkBkZh*Wx@L@C8N9bR1KL1zr9>6V+wfiyDE6gr#
z{sMVkNS1~5reBaZeV(Vj6!PLx?rpYPpda^WUp;xmk2UkR>&zGg?%NCxdP2S)qv0-i
za0H%DvOO=9!gy5I#ri3IV1o?XZ^;iQ;dw-2qfT$dTcG#pu&P80eDAWtcb^h$_Z3Ip
zI&SFKkj=5v5XR#d%xT2|%OC8`JmL-cb!?gH-80ZX|A#N23fiq(aP0gOs4w5f>P0+P
z=Rlakf=T#3f;wz+AI2Ga*6ZaMtXKJX?J03+Z<;#w^D&4bRvt%%V7qqNT0IZu41((H
z)hkn>U%Awp)v7PxcT2lB_B+7(mwJ{T80Cj?buY5kks1WjS)S1c#^AaMc%&}-49-`l
z>Rp>hVLPirz9Ig2x$*8T#IwKZEqsgSU8T5>GeV2b0w*+&3U#O}`Vd*a&Z(kv-1LYB
zGfRjG8V)IF{#JV5+l}l~7PmLoLW~rY8bJK+jhwi-3&gJP0|AJyuN>Pih{6h2>p
z)pMG^L%F5>+5+@mDyHY)4)k8Af;-O~#T)!&D}FDtVd^#?BzGOG#p_=xcfKZybN?Ut
z6tvz_R;pm0DK?gg_ZfrCDW1rlvsGmpT4%39m;0mstrm+}kIKV?(y9>NzhaJmKQr}&
zr|2k*TmFICBs$j!cMkQV@pSbI=s@e~>*g5?QG1pV9&gZjQGEPs_+lvE=+w8q2C?j|
z>^iir4js-dMdKrB%l8`j4-u_})B>KnZI@y``A6*l-Y547lYR7fQT#2W?^sd;DG6MU
zW+XoxEZ&IsPt!H=*gp33_$Q=yp1ua}W5?G#$K`X%$@@X>J
zea6PI$`RG;n&bTvVeM-jd_N&?z7Kw40_B1SuSj2P6uhq^7KIPe}L<1p^8+4JAHey^YeZcpp=6r7i4++Xql
z**)7@g6;Mm--Z21mK?zLfwRf^psS>8BFtr+!G`ZTka!vTWG#pvN=25{kKRZM7zx%Ev4j!7{Gv7gn(md=dZEsEU
zt1%a1UrWo(?;4qVV5P%^m)xZ9GX=wmeJp)ndxUQ9vmrT`G`%6HoKNgVwh{EC-;*%s
z42Hjl((!ybN9@+q$n>k0?w1H{*KAMhbnX%Kqj{tQ%_Dcx?=4=^aZP?DcH#8Bv7{A|
zXVTwOcN-EpS&KXmTbe&M(%(bAt04NyrR2Hp{7#-vf)gqCGobY}56qPycG_RbbL;dc
zdP@a@0W@#zG$-Tq;W*)^3F$;%Ttw<8q>*{3`GxSrV$ViAkDTJj^Jw2r=7;V@GH;Kb
z|Ah5YYNL1_S?wX~)L4_ex7^oQhV|S+EQw$+@5gHd%VwW*BQq3`p3UvEIJyGF?H`*ygo%2FTwuS
z_+;aHN1w*xb!zjcAM0nPAHn;SVR+(cw62vdCt#{>eXqWF9P>j#Te5$o{?Nhu%q6Qt9H(@M
zB>oP-y{HlM)#rhF*xyL?ZQMVl8@n$fe|(*v)X@Ia_mP
zs6BJvvrgzc2}PGSybr%!QN9c9&)x#2=TQFok>-BL~c{
zLQ1x%+-udw0c4jSFS7=0O9Z{R2~mAdm5HT2yr$+VA@-5r*oS
zb$@$00QJ)^9d|5)vd7c7Zp1ILmlm4qKtB2Ak;>2`SiWkhnmhXL^rLjDXT##?sW+q!KfV^z(
z6utA;Bp~0S?uQwx4uZz5RY7C0{WcG$=X+rPuERek7YOUEIz~wc{(|?YU#nu?ymSk6rkt!8pbi9+LbzZxB?9WcWQ_`WEo;pT8*xdFdn5y>r)$0ifRdnAI#k2zK5IQVp3k2oBV(HEBjlbW5i8h(A@BIT{`tEswq95J@&fX@2?j7&J
z>kB&l_PX#p4`q$8X30TNIK1;u)O*<9ShL%Y;C*&NDesZa&yY{8xy)s<5B4X&bcXz*
zL0}30s%Y~cIDX4ZCsXgk`PZO*?H=s^w#^TPbAR;%DG!Cg4*>F|=B3Ad4h#a})+Wup
zb!@OZbFH9K3OwfwC@tQR(+93;6!0F{H~>ye+j(b05gT+bp8NKX`5;&scwb(0?I5u2
zXgVB9^@7vOty|~my#;4_&fok~3CC-Px8HB0L2x0*T51;LleuAWv!20m
zjIIhcje*ZUaBEjXBb-0ppUVE0!EsktQ~g;B`LVmes*THGKksZ>ymtcHHGUlE{08-I
z!o!PNp}o(yE$cPX;X3oyU+WF;!9`DX$_&H)iP(P+&4Kgn`4T6KDbUVdAV!=+8J;uD
zR}I%chv&$uVmlW&ZYRuHVcPJ$LP2}QK{4o8t>LBZg2X{ECg12OEISC+=y37~!T9Cg
zbU%=Wc262!IfujdA5NT{xsuTS^CYD!0?~!z+a9z}Y+sUXgn4(YSS;e_&!*=;dJnOh
zBb!4OBK()IG7!(Jk{QC^>4dDfIs=^p|706G-G}J7Y5GP~Z~V0SPIMo=xIIk`?HdPe
zwUiKV-yl+Ju^i6hvM1O3-$LBv_^FkGv4TI^g(}^|!E4ygd(ICZc`j#2&R^v_DE+YQg(&&`|euw68Q9
zFH1!Gl5_X@#mFwgtOoPESljb>{}mWffLD+3Jqvr?1*A{j^)MgpdmBc#aUg%yYFEY)
zF4dZjzh|;Isa~rK<%VlK8_|1$YNsUlNhlmwi76-0K4)2HupQO6SK_DYVVo!RoBUDx
zS*t|nU4&@1Y=J)-m(?;GR-*H>zeHLG@=s0I!}kqIrrUAei7Hu!vCr%;-v3?CCc&#j
z_?_M9{n)NWUKZ!A@pea?KYk6k;e1w|7z(fK;5c;#dC1P!E(zraE6xx5P(PP#%sGMdvJr{z5SpcS
zL?AqoD2d~URU`K*Z>0&ZaiiZWAO1u5Yi%L9=NZ=$#r@klLf%IwyOI1(9V7f+Pd66p
z&sN7^)SHoq{nu_Fd48Qfxv!mVX@cLo7CBFl_QmwDy|lkGmZNtZ#PU1+RsJZ>l3Tv`
zx$C`ngU8R({1)K>X5>4uf_QS@=5I;(P!P?dGU)Gr2KNxYq+LpQOXzfBH?x|QJJ7!e
zSV;5Rv{G{3U<(rdH1iXJw2K?9_u9EzsA;)@`HN_PfJ=y;#rqO#?#>#`Tnd`
z?I-Tv=XJ#Y{`=&8#mm2hmrbYnr0jYU*8z%P_)6kmY%Aeu&*<-#wu=(|!!HC~c98a)
z9}~V6McdWSA^JmHq#wmyB(96v#Ln;`(TC9YkG>0tUEw}rCvMk-P*M?8R`|}cZhFpOvWO+PQ&*ZXur5U
zHxz&0(Ak=d?=g-C=|rJ@Z!$yY9QqzJv~1;KlqWP7ag`&#!uA(mqxur!x!;go#H)*@
zh<7>&TE9m3j=Sya_9DM6@{>o9ol@{}yzdmPU(}2C9ivBz_t!xjo9}fH?W>h`@4S(p
zou19z$iL^4Pi3gQ*z(8cMGylm*V~}>P1Xz?$M>D=xrs<`(mY!o)w?csS`OX2L>0tp
z>O)z$$#5~cS6r6eIUEJCB`n8q9@Hm)@Scb6TkLk$=*mGXdp|21?caMYDqceOFYkOV
zD$Ia(Y5RONG9Vx5itCX$<}B7I>PpKc2r&82YKt}Eol*DkRO~Y(S7SDG6-%iQcb-+br7sR
ztG{&+-kb4+@h=jCyu!3{foclm6N(!@eAZhq2zr2J{5q)D
z&w26V7qpM`6V7@kFbJND%lr}GgS@&%V8gj#_`A{6Q(@~Ne|_xeS`t12?=@8fIU~Nn
z?`lKd6;#^21r}?k?%{;xHzPOB41{@KU_`_19^|VtUgwRs)x&sz---FsZ$Y?mcU|a_
zK`_`HH!pATASkbLZj*<6&bhn4Tn+Z`L5ODIe8^vnOr3fEiVXsribwexijeP|-B;rb
z`RoDH_r^Qn_^7|0dy^N&|L35(shAg>Uvi$?*((RZhX$Qb0Z^YC$-RgiZOuv<$w-N)f~>9>=;*kzFKe1o6dL;kM&^H6lwM5Np;R
z3PInGSgrR-HG%d?(nrpjLpkB@`-^T6g?WOqHQ{*g`!Mzx@tl)^SF@kP@eGd%ii(5n
zUQx5LDT975DkK~FK)a<8J%i|b4souFD>dPK`22})S`ZwsteSxPF3>JW($d!h+GYF@
ze{mJs+0@m}Rf7HdbHrmxI<#xo4zG)ZJnOjGRkJo|XPC{|#DaEdVOP(+g6%thjFUzD
zMNW3T9a^^&nu#Jv)_UH9`QSB6p0Y@2cfFMV*=|^V>|Bs3;(yPSL$)E_XZ`5i2ekgO
z13_E}w3{>g^-pyur)9m$Me*!^pK%89#=qZMHlg(_DYPr1723T%{GwzTlzH5q3ZQ%7
z?PY;E$lv7TM>e{z6?C{#g6@5P)+t5KgR-)3tSefVkuT5lAburlQRasHMoY;3L_DkT
z*MI@?f2l&*4(&5HO-AwijT;gH(Maa?RWwHXS4HBQzsRn=+;bZ$w|{o@8^Y>82l4x#
z`~xRmp?yjG@HzMsNXS=%Jynr@d4l2(grUt(@&0Nm$!Ci8p_?KNUy$9F55nt_K7W(l
zM}#`PMx6*bB~Iae|LHdKlW3eC*FW2h%7wg~@q48w8teH`Jnz>?-9Y7ipI&1AA|s!J
z+nMaj#{7k6_YJ(iKl^>k8`%}l+=ln_la47k?@_NbvE7T)$Fbeu`oF%&u6a)nzQ^qR
zycF;Iu`0sY&NrSD+x7lkg!y6aLmB4oEtjRs*_F$5KrRxblObH~N)X;O#v=Dw@W&|FUnqUA=g>^#
z=n!I;B1z(7epk%?L-?8Q2EvE##gl$6K0xeW>?AzuoutbehP^vk{fJs!81U1PZ1x}5k~wTrZSPM>_YQdY7V&o|z^gy-1M-`(r*
z6ML|Q{QZPhy)?{MZqj7&*Cb=*UGuOaa?r+?=c_oJ8~DxmEz?03DSqj``qQX
zHo#AVVV-S1=76yG+yiW`yvZG
z|BvSt;qQhbUB~df)%az1d=JOVZGm~!#W%0hQT@eFdGURnpnNMn-@eauPekP@OAK)R
zRDN%4m%s2S-Y*)T`D&u|Drmnp1li3Qs-J?^X;Frb3X-?&&Q3z>K5?CuKf3R--7WDP
zm4Ex7ua4~eqL;v*EW`7+=BB@BKl%JES`4A;f}9;_-w1!2^$wkHYkoDckSx4zgFD*)
zo@zx3wZVPQUSBR2`M+&&dKB>--!4A*Nj}`)Rc5V6cEZW}`+h+?k3)a2BD;@PQusSB
z*2~5dNVcn)uY~r+&avoUQy@CLErg#k!|%b`+MgpnH2E+>#T&_Y7ud~$Xu9>Vf*i~z
ze`2@+;%(Lki$xAVoO53wH5%fthf3d9!uD2cgdFUKNWF~>MEPT`y_Xsv%qyIF`@7ts
zysiHR>p7HLrf~=azoAmey^V_h1^G<&x>*G!@V>5Y
zP%(QGY{%{m*T(>ext&kaQ9pCxzqacQ<8nXe6rBq1(_ExG^QvIE+v|_l&cb{-qhS2J
z1H4~(vnl$aG0fxAYc44tf%j}ixl=`YV0(woRPBp{c~I(danw5aU1FsEWYCQHgP`+U
z+uG!rgTVMrb?8NS4|nw8w#Pp3zRSpc^@I!bYkl%$M?Q@2;)`hg0}5|JwB!<{Cg`VM
z=Uw*w0yv&US+5sEele{puX84BXPtCdKuN&>Fjx3JH5SGjvhHc=d6;*#O8bA^
zPr5hDE)Cw#)h0Q~rN4X&+SCt>yy+MKQWmrh?*~d&kWu}zSjcgsZzZl^jGc`>wrG+
z`RX@5$>#%L@s3c@TUX)V&6qkb&k^#yjIX8I2JjwEVpOP(>Vy2Xb;BJ8IF2>Ki9d7U
zIG+i6TCo%&Uv
z{MqNzv|${sCYB{=eYNI~tVMiiuNl{0#BUCT^jtd((L8$lhz@K&ME7@?DwG>j=6j)i
z!|h7!Qq+%Z*Y#cK{rhNZ^fgszxA@zUQa>n9F*<66_8Gm~&-D>%El4&;>m@m(w}TJH
zZ`HD3OcKUBB>pKp2bSwU?lnd2T^SZ%hT=-Q=plasj-SDF-tNuNKA$>Wx*3)aPKX%P
zLJZ%b9WVmxcbkb$NBh*Vzr7pK{
z8=~;F@^QpVi$2}^h3tw|S7oE|i5k7zi0Yq=ue|ma#yR11;WYC5**WAg;(1L=c~TLN
zbT=ypTavx
zew6G`#=OXHwPXwO*L}z^38CtdIKsnnwD7nnC^qAK_IEh-G>WsYe=mN|zH0e3oS&lk
z+bdAL^9yHUKK0_KZ!CJ=?B%>FsQ*XfUg2@hE-!h5?Eb#c!Q&|0wgLNB@g?_%UOj|g
zQPt!g$yAc?C(*>H|5w+QhvoQu{gP}UR7ypXy-nH>4WiPbNZLfCB2tv1kR+*S-;1)O
zQbedIS|!p#h_vtfDoLgH%*_4#@%vua`_J>aX6DSynR#aBoH=u5ZbIi5M(=N&zE1Up
zfCW@JywaC3J%64+ve;X{u)(pP+W6J<62ss=`S^C$W76q03xN^|M2T
zwBFmBss1uw+>+4APxT{M|C;J)E|+L|CepOL&2DE&dfeqKf{~)v30^%aMaB>6|48gK
z$6a#C|FxpL)d~N9@4gO`_8%%UBy^jq<`6yS@?*0woG#Q}nvAi1+Xj;U_5*zn>eHqf
zq`bX*A_<-Fh!5Fcl)J4a<7-T95$O+;t*^-V8{K%4=uQ1^G#BD=6gJ39=&VyelKMX=
z7!&S{14Z+NH8B22fgZCwqne<&iF_^;a(D(2&PR&>Af2-aIT&8^nsdE|H~=p?2u
zvZME4vJ|Rk;(Uu|4u8gUw^Ho{v3_C6RYdL+e^rNc*g$970g4X~_!kdb_D`5@rf9Pf
zF7J-hSTSDLkYt3DKHxFA)#Uzmv)eb>Un{M_{rrIk9Wc!!0WzzSDl%RZqYT6VfiYeep7}$@qNn#0AI8
zgRqXOS93_NfcV18@BRG%f6Y14rwjNj?YeU>&`FoR*H{MjndThn+qgV)+unp5fLyMl
zrUrm}uSk0516_lcnUXWmC2u+zSq}BT5mdH#4{}U!l1p6?>@^L`EO+{GwV{Q9{bi3q
zF8wT0S6dBoQr)!U%Rk6p!255s9Q2`cZ`endf&Jp$9d66`3t%_-`Z|ac^rK?_lc}PR
z-kI1Mq6p=73(M8K7=!<{`q35b)v%5&J1-v$bX~jV<+_32Sn9QsfC7+f2ab=}2K)kh
zhVUH^J3){qojX?ef!tkY)L#A`=yv-_W~c$bnyJoTEJm@sT5|2*I?zvty2eI#wxPwV
zqK~?P9ZHmMTA(G^Q-sIPTH6A>vQvFPAhdUt&Qq;+u$NrxynDeM&L`E{b4}Di{_vgdzB!x3VMoc#xMbtmZgvh5zk%OZkahz*~-WR@gph
zAE~8z9q11~X|ha!@XlkGyYM}dFPVGq%>?-^yv;MVAHvIwc)GnnfAskqxhVzWZ6z<1
zZGiaiPn;@tKv?s=#j?4${Ho@vbwJmZ`}xps_`fdKidqT)-9WOMTI1o2#a{N(=0yA|Snc;5QS
zmk46Lq(^FoB%XKQ6}i~n0v-2j4OJ5e*Kbe|I0qOMG=CAczbs8@O2ztse7Md|e4aVv
zHe{dyVRmsg5o{k4t1S@1`oQa&qto%c@VxI@9Vdy`BNO+r
z3wS-yY}Km9sI@XO2BYzjMPj&p>}op_G2I*+wZk~xuC9rkZ&`h~L+m~y5jFEM-8Z&I
z0nG2y%k{dLPQ}QJthXxm4~RWp>Ck}(xPM%E?By_Bo%C-5Om|*&8?l?4$F3!E{dlI}
zQ%n~lWwsmBjWw9+VLGmw;#Q3A8@S(Me6m@Dti#uP+e!J4KbsOcCECzR*6aH|&gqzL
zw_Pqd|4d(yMD&u04|BNzdgt
zzUdP=abmMMkvo+gI)qNqg1&no`<|YMJUSyx=yt4mLe3qnT2B$WoNIohJkdi5ME=Zq
zPUR(sB9)iC52(C6VMgy^ek^!G+Tqkd?GGb1lV877di;tq}M;vUuxGaNO>IpLW6L5LP`!=m`*MJ5~(-kfHR>pt|%pR
zYESt>G2J!05ME4I?$1W*OU?U4^aLpmF`^%2JD5ddI{R=P(%!Hax|{I0>+A^FjpIKk
zsD8(|{>?cBJa6iM3Ty5JG-=5Qz;tSB(JY+)qVlIAuE)lrz#Na?!`@oA@V?}3<6Yvv
z?)|5i7t@W6_>g<8VfQy_;PN=`HO0#UzSEisS3N+^UD7{->9~Zyk$ypmylI%uK-!NF
z%jp$zrU!66!M_iE!E|q>!q;Pr9btQm>13~jNMJcDn=tqq%S)BJ1r1nUDzSgMyB|=j
zH!rvd)``u9JrB)5ez;0+bK!ybhhYNiVgNrJaZSK>=%4cYn&$&%j#S@|fb*FX9-D7Z
zhj8Vro!jpLh9~P6V?AIbCV>z
z<8_hag1q%dsP~yr?4KF+ZQ8zj-+!m>`y4rJ@!*j=LPxM?5vvX0dhgv
zYIA;(-jv%M^|Q{#N-`6sqJ=Ee`T$_8ji_wIKgOV)Hvxpq`Sr
z!HGq1&JvYBek!xJ1vR;>t?>tW7Vcg2Lk8$hj5v=jfPLr6;aB!MfX?bN$GBuj3(8FE
z{T2Z2xp3vj&1R@qCS5Ll3-I%1mgvD$*k>OM<8VrCYC#UA5pJnaPwcW^8b%=Bf(ztc
zB?JF2()}*D{e$)hlxIIV3FF1st%<9o1-+EHnw|#d5jm$$e5nC`pRFig{}SY7FS38P
z588Rn^kf~^{ucB#GU~)W=nwnN+vC*-f#3c;XSxf)&rGFPpbPkVtGPsb>!ubYw`$
zR|q|M^U)6IcVLi$&mn@bNWSJrmkn>d%pYYabg;~F2x^BBva+hS1fs%@$hSWf3U?;KhRbcXV_*X{wIy$44M@OZM#T&uWmt&=S+OfQx5oJc+75a8iy$YWD$8dS+nT_O~+4k%i3b>G&6E
zoX2$MW}hef29fFJRIWZgRD$!R9G*e$(meiL^fLB
z^=8nfRtlH5V{HK0U!1%-wG*eGP+mdS*N0216fxbNqWs%<-I{RBIDym8h`PWPJ(y2V
zZV~y&v!jy8%UHJ(629SSLDs)k8S019w=kZp-}jF9k^Q0R4u4_~un6)cbni@j$od~s
z??UKQZVQFs@~s62$^N^#qk_-{=up1^1D*SX?yS_kP)s-5#g@qPpCdMeF6FTsDevKJ
zx}WFy7)#1KbX}F$;qP6}Cvs1!f%@fN+Cu$Q!Y8P_T>Y_wl;3xl?vq7r^oV>hxJ%zn
zdpDh)S3F%p?FpqFbie+hbcE!qvK}S$YnD=ZZlXc4yNKfW0h+(to67Gy+iCne=3Hgi
zoQ5A9q^J;2>v_%ip_{wWc<%ias~Larmpf^BPVs9AA78!C6L}dH_?rBm19oR4aDN?f
z+(`N%GLY&i#_Cj`vTIvN`Y&h#5k0Ega<&VmmuR0LrtM12uR(
zku3Z~>|on;HOT#!65j=6y!y`SAbN+RNop0I54(&^NqekiWr;oS`^wE^e8q=*{>0@c
z>#API^CAB9KBAZ2zMe$rA_vcracwAMKnZa
zlKZh7PY-RY#_?HxHsn6tXw(Z2%;!Osfk|wKJy;<1a1NkQ#&L-oupczun53iwVcXE<
z!@97q{vdmOY!~>|R%^X*eBuOr^1v6pVV!4xZ61ivMcC7g>|X#K|6=2QKG>JuGOW|r
zp3{bI**KoW{-E2W)of-#d2fQWvasKBq`j!GH_$7s=Gig^`&;WAk!n#$FBqPAs~Rxc
z%2f3Q^g~<6^VKC+}z#oymc?wtG7F?pVfvuQk&mI
zyoUY~HT7~VZ$WG4&0lG++=0?|g#47)+J`JpRP`PQJ;TyyYOWmkTX0mj6lnoJ>+dX!
znYIu6=UHXyn|7jn_Ql5@!Fh|Vf}O;^Jm|+q)@`of$C4=S`~IpA=!Z9oJ5%++p8QtZ
zG!XXpiosIJi-A7{)oUhP~ilr8`hXsNk
z`s;J&J-{Eg+ryFVhixM|z98&*X+je+<#USSzXI#)f~2eFl3<^h{bkxZ=)a@d9%659
zHK7C6+ojW~ilh@KwaSx(BF1jK6wj%(y3t|gqDCFVRzH>usi54exh5p
zeFQxxpzXXnJNP+gJ_zn){{!=?(V;9I#s|l~bPGf1udhC7tuNt!&L60@tbp^0ezxk~
zbtSMKZ2fJC{Y-ND+3!ojxbQ!+E!-H!Mc0C5gFf&-l$M@umk8iV%0c~89&SGUq!IQbqpwn;fIpMlQxs~UJk`q6kB$SKxP|+}c+gK3
z*SH8DfcTJ2_KW?Y98pQbQ;Q+}sef8H_LJg{{Pi8{>!pdSPyK^(e(q7q#^;j5iE7U;
z0~TD;(vSlT$oTej24GmvttI&U!^UfyFSgt9*%qG0cK@MJw=H-*3cD2Di~X-l(r#*E
z{bua^o|J=tnxDNZu-*~>DVF#@pXdMk_;08~^LjF?Ub9kSmu~nfFhh3ZQcHwp8
zvfkZ9tXItq)~Lf_Hld1jm`}~dd2j^@*1;+5#Ghk-)B+-hysYQI)7>z?WY#=73dp&s
zg%{Jee>qR$hu9+AaMdf8f?-BCE7&AFq*j=NLAT1#--y|1n6CMuwK-1jl(!{%le(G@(J$b`l1niESqh33I6bz9o#-h`2H(EN;cN|^
z2N-AhJtlhDS+#RSZft*OPUPR2<#b=pq2@#6+xXniO_)Afqnzj&WeeUCeWYv(J@?mt
z_=tqX?Pm>Rx`*r~Bwx~aG>Na1qkbZu7mA5odpYhOgz4m66N$Xh+A~bb&FP3E^b?`m
z2_KSMjmf#p&xYG1zG{ariC1beC35xX9BmR-*r-F&Yj4StaQb)ZU(1n4^^%w$R6c*r
zS0MRn|OE`!i3+Af_-v_Etlsb8~oJ`bVG?@uLozL~Zwa51g--a6XPDK@nH!-DjDYZ2qO
z-O1b=Oc$W^Ns;s%jCU5TPl@U0+M|`!%%v@w&62Q|1Q_mv0d1#rv}4C#?}!zc{OMjo3|PZI97%
zm$+WVe5c*KNc1Mxnoc)-ej2sLz!k?I&r$n|@k_)He~hxq^t`D4^3iNuzE{(OVjSO=
z%trLX<8~bdIQ*jWA+f`9c?t((y=%f}8PPklGYg2`+31!|`l0sQsTp`(%AMr~p9BDX
zq;$z{e2&v4FYlQNIB%0)atEx#vJ*1)Zm_<0Pbb^@-{!n;`?p2gzz^19{h
zLJvTx^#b#-UL>LyFm?pwox{^Dkd}vm^%S%fIRo
zeO2P2)-jx(T+;(j(ttc$`q>QY&CfsUr@jXKb9;9N9uKY;A5F*oY!su%hS%5fm`gA9
z0B02Q9>eyhz;~6x_KNZ_PgyoTmZ=5yP%BRbMraC*9LO-yRZ3O#cv=lte>vNdgAD3
zwzN9HqwQSVgn`fevR%K&{ecguPeS`iw@gDG7sn1kh6L_N&nWl6_
zny&3e$Cfyqdz8?H)>h?gQV4-_pXGbYjn9G|W7NFjBItEC9css{Ejmz9+Ew$Cx=tit
zXIL-*`oS086N)pz9}fGoRjJEr`~skbHRrvzw?H-p{`KgE1{rIezxg;mkRt2
zE8Jbv0OwkMzk?lxo`Za}+dn3?1NPz7c5(Hfx8A+PmHGhAmxQ{q}URz`d}zw@Zz`HlZa_ro5SM%_u{`JVbCt
zGfFtoIuQS-32owG`+c~q8S#A9%)hn}!pa)p*I8{0h?|hU>Oei=%zv2~C*qH#ghjeT?u&AKxFq=Z21jS=peE`fa&djrBYM
z?(=I_0BR&xyvFlq({AoY;wfeTm6
zEW+`7w$*_shV`$ZlH9u;*6w!3^ZoWxopPLR>qOtRk^X(g2Zwd{4gSLG+_3*&T^v?F
zmAwGxZ;Jjz)_03vwk0?|;koNnyiP8>vPBQKf7R?8Wc{uxTuJo9SAUL@?=l%$`@O;I
ze&d^MM{zm3^*Y^g{>f(!pE15_HYD=&a_Af4H>t_(7KP&v@k9|h<-h$ok&9}_t`Irv
z&^?prb!T1pGBKU4IQ1VbJlaI$?9U-fk`MLXt;Ojb_jhDtywS0O=!KFSwMqKld!9s|
z9U0`V#_97G()UImS5SXv@fY--x|j=nzk49t?jEMg7uF&C9$h_;l*6GUL&_;Dq51e8
z9whbnj#K}ScLCJSHE2oouO<&Fk45*(ko*%>RKLmeqjW)8RG({lLgV+F43T!N3A|0}
z(Sj8l^C9w=`o&B#-#6aFIhVwz&X`X6Yo-O&YrZU}deZdURG)lpZ%ER8H~u2@W<}I~
z*Qs7d!hy{9nZA!xe`u|4>Mx_o+#5rTz8G0e%jFuUd{{B>Ap`|cy{UwGccCqThDU^{
zK9`?P^{I(PG`=dD;w9!?1TeY~K9U&yag5Pde=vIFLZ)08jD&6~L#NDy+Zeqym(hoz
zv4jpb0R$@<+{55Tran;0NIED{1YuJ^u!BL?cTid7!6uBP!zP3v40?j#^g{6OZi)g7
z|9#9m#A3T?zlnwXCUn`%yBD5}-O`!44=2mK!%%sT?k|FwcELv#NIoYf-;HZD?6{ZW
zBc?r*41%$m&?81~&%8~;l?)eH-^YBn(4&`zGd57Yy{wPE6FkvI
z>n|yw{MIw)yhC@4J|e=kqBNKI^xHq@TTXo`ik9UXw6e{|=)6Iz3xM!WT4xNO}(&
zEiWsN+M{pYSVQ#Xr1|1ScpW`G-9{M8i`TkucH?#5TzJPZW57qI!k@9fcU4ayS%<^c
z&Z)wD4B9ze!MJ&5<~6)thkc2uGX-4qqK~X=Ywo}OfYXm8D+gmVsh_hL*JJEbY>Vy2
z0hJM?-%{Epi5@=PdMR1IcaAs1R04TuWHgH9lEU7Lq`Z&OJIZjpyJ@r^M%f|{B8T=H
ztdzt3{4+!P(>%cO>dO+30hgQ0qEtYw1uymbz&`2LlDkD2qE#463PQel14_Oe*^>sC+%8Vep%snYjIiAH&t(t5`FQqt+}j8E
zY=>z$F8^48Cm;4Nuj+m`Jq7HJCzNv+jewm_-P>A89rUCyft6W;*FZn7J1SEKdIVQX
zb!`OD%lKG2H-esO65ermF6cczgCddg5U*Ic@x2U``#az>d7m{Y>FanQ#WU*eGdJ@Bg(^m73fYZjj6`>$`XI7&%t)vcU7*#xE;?-n$+7tKdM{&-Q69+Kka
6yDb5b710Vo(?Xc1@12U-!NB#`WwEo{Ryze(+ZP&^PdI
ziRKoKyP(hYi_Lry3gvn&Yu-2??1bkl)=nJ#0D9ea#~qQdUr`T#ev}v5^(pyih&Aj_
zf=ot_6~Mmc<$Kw@HGm^LJEZ)9e=eOHvzI}=>-QAZKZ5d>ZRLm@1bgj;Hus`52%E|!
z7jr{<_#zVYo&su%J&An_^f6((_G13~Pa91;5AhRH#_KWxFN$?q2txeukuQrifX|cH
z&aHX?@i`%*7nEUNRNIBh_m=M?Fhdohz-Oj~M`MMbeAs2w&QY8I-7Vzs(kr8(n%IPW_RZoZd
zUiB>*#pA`ABP~!0`g>{3?2i?I@taeM9HE?R0y9SVAw5OAMoI(l-m&2zscp!`I*8{f?29a8tpqoK-Y*rG$oUcab)|}ZlrNk&^V~Y{N(08(
zk|?V&Y*)3~Kfc2a`tRAl=y@?fr_5&PwjKKW;lE_AD%c<8{XTGA5coe(@_Nw*sOMcQ
z|2%v?%OQC(P#4f%CDSq;_-q}Q=$Zi-vEDol>&x$^AFq1>82|U+3L(gsv+TCUOTa02
z7EabfKD%#k#=IaM`G-Fa1H9xOwF~Dn9m^Ccgni9q%;?m!@c$R=^p_om`hWK?5y5(|
zDcotE&LlZ~XTRm-9om4W5Jo{|5HM>%G
zGFQ9fattg37h}EoTEoVs&45*${BF3MuB{GwxEw3}1T(x|mtFfy>c4nh>DwSRiIM-MsU==_?mSnn!1=em7bWZR)0kgm9oFkny^84+
zGuIx%n5xoB&c#(141dAlLzljg_4kuVMiUNyVJ%Dc2!^
zlq+W^mj
zlQS@C|cXe#O}VQtyxgy|
z5l;0&tMgRv>CC72>t-*h|9cwM|GQ+V{v?(~F_ZZYM3)C`7rIa5l`|+7NKlM@PtiS%
z>T|=*wEUV8s$a!0??41F`uGoq&j#kaeoU0+w`TSQ@7B?<!WATFqad>LyW!}#^|eW8GAJUbD9p1qm%p>8HB-2!c&;=8|HqdDpSuSqwlIQ>EL8W
z=;4PT999s7K}hf{gK%g~!Z3&k!Vf{#z5zC=Bz)#SbS(P+B21$HDTlS6VC^qh`xX{I
zFsKMUi;jh?dRg?t4BdXFT{cWRrJ4Tp?4kc7ILnao;E;x(9)mlX{`F%1A6ES^SP5M*
zgWDKn^*@UbR==^rLk!(72Ekc?lmkjFK~_6pa*^;s=KpbVreUY66mM*zc&(h`zy>Zd
zpOxcic#L`HeDw<&c4xkeBL0t#-*{#p1dnWzdhD5SI&)r<^@#S1y**7&KTdIkITyHc
zfQIk)QsiOyg^!ex^5F6l!6t@ImWzhVX43DA71u|RbZs4F!gs}^AhJ#f+Xj+zffVLm
zaV7ITbAIN1%E?Z8PBI$dMDl-HN$0n(Ep1mZv(KMo{A&YNiIDPByJ-5`TeMw%w`q8C
z4bAU5@{p9Lz`S4c;xSD>xt1wEj;=qa%V_-SWd9L
z)+F+mTb}Y6w#bH*ukLx6@T0Hzf#gr&`oWFqx9=Aw@#e8~9sbavPr}!ajgxgTSq|8tt{1^%a5qYTl=`DQn3iP8p_1##`Yuz*Xgv)y~Ax-vk
z?zu0CJZ&A2B=4f!yp+s_%N<-G>5u8;>$d;E7`C^L_{#)+&T=IE_O1y`6R^K}!cD%f
zn71iZ4GVsQ~*e20XXbz~{e8;&m&qopPt8lOwj*R|#P&v$QE!upKx;4`&dfP-(wrsDJyD$##1
zKU+QTT*UiC?^XM@V14Iudb1DSpM?7h!qN!i!~FbQ
z?3G>(`sSC~kt?qOqu0n;U4(XU>~BBt3evTwwiM!hjrQ9E&MyJ={#cZWL%Oum7C)@#
z^#=|QRzSS%g1tkrfSe0zD=?qY!`**30amFH7;JXMYWe~=J?X6+f34Hi}8#rxjjrr{270JZyOFOer;u7G$dZ|&Y$Lg%o
zCHI3;DwPbpAUw@U#?k;Va{iY<-2SG4jpUr{Sy@E#1n7|_oEa-IpI0U0m2tm4PE|}>
z3V63=S>O*yubPZ`mJOJ3sgU&l%Cu}UUfOCdK6M7V=SA%majcgvR=9f-w^#V<$ECPkF-J2-$^Z8Hiz@#A*?(5%83B&$_rH((
zNAIp>&vHP6_F!{i;6r>CN1X|TcRzkzxd%|#Wa=?|pR%?iIxq?F;8v#^L5P=L{ZJ6M
zM@rziGqzhVmoa^V_w7~(EFx|KuJY49hx=!zjjtG!x&%^yYQ>gAE=IhZcyG3}s
z^^|gK#Ph0`&!8OZzp(*=t$3d9T=3cwj~n+FMQaQJZ%lW%kK?Bg@6I9ZS$BY(S30gf
zdJ>m=dhQ8Ze4eFrA@H#TgyR&d-#P(4ySZEu*Wdm)|2Cd)yH0Jd#ryh7$wNc9oJU<1
z+j070Kk0g$E@QfZ+)Fj)&F;c_yKv4X?>2-B(WF9BY??8Vbje%z>{-0KE{QU{@$q^Vs{9ALTR*-aqeW{)}o!h&T=)s*i
zK7BZR&)u2a6Pfq&a5xT+=ns%}X1PNs@mCE^yG_qU^XtjFac0&OlKwvZZ#Jf{&@UwA
zi9MTk7l%y>%ZS{OIM77qX{_otvaWvm-AdM9Z;gQ0m@dcXDB-KiPLar~*E^^mV)JS-
z(q5mZJ4m^6ny(Q%nTAC*p))gWC+l^@G5YR5``a!3xV(GG4y61XjR=CFyd|U@DbK$o
zJa7A4l0Ryb8wndepnj95fJ*pI!U^+
z;&YO(hOv9+O;Eo~-%+X$e-)yBAYT@~A#_S*RG+rFM)mfxO;m4}lu0IhJ#e7+SGcxP
zyJ*xLs>lEM)kgB)%Ax1=*=wmCl9zd(x012n{y0MInZ363{>q3XmD3!&RBuo!rFiHY
zjTb1VdD0E+^kXY&n57?qvk_@$2%|^+9isW)F#aVxj6USb*umE@@hS`-@YpTMAIPNd
zWAwfhCR`~^)4du0UU-a_(AP2PK}`PYg*3j9@%NQx=-L=MO{QEGrr+Ul4_Ysyzro}n
z;bsQcF!{llN#bGf5Co?Ug0T4}_=_nYCNl}M%3+-cu=IN|roJ|Y4wMi=4~JL;S$aDR
zW)gc
z(ZQlZ^0V5>;>&`Gk752dRyjcYKYX$F1-t(f&*GQWK3FtJJ~-?o2#XIv)_#J8toFd7
zN77k*viN4Dv+8A~!{efa9uA=i!pEoyik+u;gxN17GWvfg(=K>inb2Kg_6_hc4iX+<
z>c7Dt9D0)YB8L7K6TZ%*moezYpe&OPPBJ7P8087VBL6>OUncC&)O$mg&PxY1I*-TJ
z(s0wET9WV2eNnPLnw!ybybn`6;6styoYJ5FO7{hIOuJb9=ECqJe1p;th*4}~_D}8%
z-y;m2@==;TC!ZpGM1=4m!}RYw2O9Rcst>63=U4Msi5(&Ku&sM^>-ouA5l)zsp
z$Mg=yVkO$nC^>33Nlc~ZobDUx`6$1gEevIlhs=5r&a89pJ7|55HgsLKjdUgD_wAwa
zHB5gdb<*$y<~`5WY%9XgPo`ao%=<|_X0*SG-qH1+CzIZb2)%P63fKS6dXns;UhLB)
z`;7P%d@fi|Q26_DHr6kaoQz5RhHK}M_E|M|kbO+b)b&IzXGb*s#pPHZup)FT#Rdr<
zoaIkRcu&I^EE%BJEO8lhAc@}
z*+utVQGQY|Wx)UAojKt{ByKHv*F*YrVJ_Cga#CHQu->y+vv?!cJ2%F9Kf!vXj`p2O
zoX=?IGvZH>r^BO%!!MUQgknq|I5;2YpE7y^rXJkaQTk2z<_KtRz+tuGo4pt>B-}~H
z{iC%#8I}OBFY`|!{W9icP4ppNF*#z#bZl1@#p%!YFCghEp9a8W0eaZhJwz|k$Veo1
zMv0@	x;GW8`V9-`=>DcL3}8_sTvQV18m#b%F=LdOTsr-Im9Hzf#akdNNp>sL$7{kAuTZ^%)V=mdhxhh;-f%!7z
zNneT4T;tMVjIn$dgD~z9lqBQgnn3MzoKMtkF)SUR*WQ+v$Ee7+_Xr73bi_hCNcB~sy&%CPU7ml28c$%jnz;_!H}`AD=@H!Yymj9^d21Yg&od&8apl+gIT(u_)CiyZYpRJJx7w)a9ge^8
zySfx3n|60C#;9c)u#|zHv&wbyKgaGEChs$+%<#~|?TZoHLGG`#mK&&Ge$Tyu^t^Jq$+~TsgwLQi0e^uHzZ-bgj^mR
z?kV2{_YLDvr|3*vj*@xVPaKwYSWtjbNw)JN#LG;7@inAHy9MslZKLPWp$gX6G`Sxa;V};WdIS<^zc>G&6
zEaBisT{|p;%UjkN03r362*y3T
zOUnB({S+z3#_AaP-^yF)Iu^C*GNHTN8%);I@Uc=-&#Q$yNj}XUdT)AYK^h4^)T8=v
zT^?P>m38?@{%WK5B>lh?y6&0}wUF@lWkUpA*3ta=E2zHy){?I8+Q(_X@J~cjx)?T6
z-sy4b&ysbT`W5uurS);FzC`$B)1&1Fq_>lN#ZuJX-XuotU4m7#eD!1+K9f$@@du2Z
zd`UJXx+x=1&fcQ(!H=C{R|Mt5UXkXDXXNC^?KFMb6N;y~X#4CNX?W*M
zitm{8q-`|(c^2h+`wUv1>~AU`H#7SGcgD_s>Mx})kfg}L*fn9Yk#=`6dUX=x7p=v}
z-5rdaV`t=bB%=@8CDME>d#^Z?KAoW-V)S9Op3)64_V)8k{yru>i>cp=@dFBB`t3AB
zSIE%8p+4dBETc!j$Ad_irI$m1gjsrZ50ieIiHE^T(sLPP@L=
z`AQl3kIZ_(nt!0wlJY<}6J+h9G8sNt4LObLkJ>jzT^
z?m4cbeyKjsseNkZHY#@(CeVGO)z$)-;-J4I&~~LT@6(xQ1
z@jQyt)o3~F4b-j`|09m@Tf}_dEN%hqClx))cehR-N&gZ^{U?nrN69{cXGagwgRXu(
zPV^oXr<69rKYt`#O@ecPgYG2Xr*miFFNS@UH$4{$yT6n0CJs(gYK1Iu8Df`54UAClkHCWN4h(^`V(
zLavkY9vpc`_~EZqUWfS<7*8Slk=RXHFxBCniy-|@)~6|9pMo@GJZmTFx(HOKAdy3mJ>cXUe&`?1%3C!5Ny?9f9zRL$`Kc!?*~X)&n0~R(l|n}
z;17?Eh>5hv)4W%bkk=~C3JnOI!Qgo
z!qZ{u!T#XptIrri#vRKrvYXlxJ~^CRNIsQ62?;n}l6_M&#>#&)A~3qB1dx6T;-KR&
z@5d0S&&JM=@H^Bm3QIER>;2SUdlf(Ri}=aDobY*L@&(ENtIUz0_`eiVpY74VBt1#!
z2#Noq{*&@6I!tivl4&(A*F@6Y2jg_#Hvt%J{-zK;Irs$|*-u|OygVGoUs2sa?CGM;
z*GYcczaFG~<+Zm+`H`>w5PlB~Zy@~Yk7|;3XuB>Z{QRCt_xW>F9uRt=*eueIpHEP~
zcDch;5AX4~LdrdgqHIrhpvMbO%(MQKWo_{8n!t^akUCX
zTV`E*(n`Z+TPecF+z3DS8T;7fT3WvBAG!`IG5YLPH%b?#&(L#F{uVBvVc(4uzr|4W
zk)xQuo8ma*@9>)Oe>Y_5&rJD;_EWm$j9-ToBbU?}Jjw9sdWF*UGWl3~0UVZ*cCfy8
zD6C7vV~pHjeUI@S6SiXLAFQKvo=kigQ;#eYUde>v5g1Z_E0Zsb;S2t3690w4Dkl9V
z6K3gIEPW9c2SU&KemN|HBn*>^AWL4e^ca?XxtK|3=_@R~ho#5JGkmh@W9bJlIZ3%J
zdoimXmc5h3KdT&;zRBW)MbE15zw|KKNcr$Xkd@9tRz0lpVX_lCSQH4d!m!AbFzf&P
zZ#pYKt3Fnk)jn2z|3wxZt6%=x4i-JDepdMZLl(clEaCV6!yRFrC0;W?{_HE*)dlw+
zS1;>0c>(;3xlI%_rocV1%HRRxo8Skl|9#WG^|4@Y83`P}uoLdXxCh^?13%pS?c>UZ
z;Ah&}qBfxpex}+Hfxi;6deG-!UA}zLP9!qBWv>Ig$8cX)#`q8TFK_cV6-^ogyU;{L
zz|%urXodd!ifN)<=wD2@FkeDD(y&~h7vlu>rN8=Rv)J0ueC|Ix7j11p6_;N5*o`)$
zhUV9rVO;IV>d%!YzoFh$>v&7>J+^y8eqwI%b*MwgzaaZBd
z-25!~GuXw%;!_Vmf2?+PsFUl2^Yf{biJxbI{gwOE(^lm+q%z~$DPi!_{Sx@)!W!^{
z79GofzcmN$@$V?uK4A!YpcelvTlE(7eA_`i3)NU9Hv{3LevGcZ_aE$
ze3Eqsk8(DlXZrj7T^pK_?VS6nm*QJccR_je>X2F#d{(KYJFyn&&FFsrQL+w=2>A_u
zFs%pwsNu)4we_gUtSt33@VnIZ(KfA}>d?+d%Q}B;s6`wLXS#6K)gXsd
zO>U0PS|s{U|I_Ih=m!DSo`Ust=#w6MiTlBNWTU$`N1>n|zMo>M^AX0Y>d>ByKDcLK
zcg3)8t`^+qNSb95h{5^q5o!Iy63X*8n^QG7Q|4#Mlqk43E^$gWxGWCf4;DeSar|Qs+Xv?wLe0Aul
z^7jXlH8tqGK
zdKXSKppaQhh78mj(7v7(VyQWeXnw>~srPb?NF|PEx~+5rk{l9ISqS54bGFFpE46h<
zAX;=2C-5omVKg5eRzq(#@g*O=S&KH0)P+b4?@wKby`P4b^
zfBYCxoTFEV{!M5bI324)i+%`M1s|wGhaa1j7GJ1CnVM#Qlf>$fRLPC&A-=U}ejb5Kgc8M4etqoPJu57A+0#WizfrJ5ntajo^Qp+VuKNEVTd9%~Go&
z;7j7cyu!SOTI3+(Jl!C&7EOItT$uxWS4i|8v7D?$pNsq!6i3w~UN62Q-TV#cU)I-y
ziLlO0YVJ?`7X<6#Gt=8k`&y7$nc|vTfh|ZrSr7@|f%pAa?+eN}(t_UoTU7R}vKa}P
z8G1TSYeoYON^w7Pn^5_{DyMBFP3YK>n^%@Thk2;=((7qZJ<^vH=#hH>P9{CaG(BR
z+pl^wC&}gFrGNFv**S`HQ)4~6OFyqtuBskMu5+&3)CBWGVc2)@ZVT#Ke@)2()@k3T
z@mG14!~25L9hZln!S@pEjQTf%d`UmG-OKB96YAzGEgwDHga(ZEY?=b!F_4Fc<)2?`
zLU)F5#YwU?Bhj9PIj3PBmrPdd|N0KbapR?fvyapxuGD~|&MSZ)jrHil>!2lq
z*XvREmX)FFVV<6S+4s(^y&gSG=FGb9T94{FR~}t%SdSw5hcCO8!#Z}-u>R|CJvvdT
z`omDH0m-mj+!K}K4&g6?W{KP>5dSh`MqC;6r?jSs`tEvkQEt)gerV@JaF^lN
zNXQ@c{NeAkdh}1;Txe%WJ^InQ&L-Wf9^E^mu;DV4Gb70Ch6?mY;fC`0}^l)C@?Pr>Kdeya~S&V?0muM_s~D>FIPB?
zHz3Y=V$Qx{KzD&NCo99cQ&AfcUNf7
z2xvg(>}|L0muy6}UxFXahjrde_U7y9Gn&xdJax8LP;N5+!*TjzCrThQtq
zHs=3=VgG$ogDV)mfAM@)K-NyFCUkvyzf-`)CUmX4OhxQrGxCX&T;x{Mf)e?T9IMd;
zd1<&iv{(E*?8ns4mxpxa~{t
zPjhQRC+&)tJ)GB!?hn11y5?LnnlYz!$S(=P7BUhuxth>(<-GKr8=8=vzpi4-?It92
z^-he{#%6T8Wy&I}P|%as8nLT`U*F4}U*!+Ny{4!Qw^y%%`&cF)YZGUJzw5~j-)`oC
zz9SMVEv2LlezB5+ANBXZd8+#i`DM_~=Cj9(+2LMT&ZYRLv*4apJMV>!4e(tQXT9~p
zTXf+&$+Nx{ESL*^ezBtebm6_zv~pj5N5eKW!+JRESVA-6?=x1r#R>0R_!(|^SI~l1
z+#7nC=L!0iWl!Twc(38eXsP3w=w?*MR^$GZ
zs}Krk_{)s_hp5btBE6DB38?OSm7dd5#uTC=fC%BE~<-UP)@5Ix7XMxXkLwgs=
zNWgcFFaFd~It$+~4!hTW$f6DDeCFTT5AUc6ay=Vz0X|ozN3)+QY(q=ViN2c%YDY!q
z^fG4Jb|8TxYzcC`9Y{Czoxw`b_wxK)@1KD0?WNtC>+G_n9p&AgwtXE(2RdTd)Yo#P
z16>;08RYH_?~p#+n{o&4o6WD#(y5MaLoNU2_jPS-N2bdnBVIgiM^VZZ50~e{`Lc|z
z`6hU$fSc$#&hrG*aDHqPlnT?bwLC)B!7K&!b6
z=jkpqX0R~nQ1DzpQgD42U&uK11)TN_jKI43+0Nh2wBt8g-UzBsrvKxAdY%X&+~GRPKV^i=AEgDj?Z=>5&(+>Ji9IsMN4-HpsR=RUl>vqay3#PYmryHQX~;j$XkgDU+roi#GSFMC`2sgrl%JKpR~jn_AIBANceb!7=X=(w<3
z-m2Ms=*rF{HNmSLNPYS7;R?PkwCK>`(2kBCWM+?6Ua;>&5qTv=N;!S#dc-$FZ--uV
z^)cT^S0li54_~bU>o?c|rp7+mWOAqo0(`_<_esFL0XsI^qL+$sTT5})oLq+=E
zepi0#L&8Oi6CWSxL!tR6mbVXfA(vBfxkF~~y#j;Cgq+oFHVbu{pm&Dw&vXLnc0W*#IN3nmg+;gkNmuqUI9Pr
zW|{Ar>_g)sk9T{U_n`%!&yFsV>qqlG=B_rI)(`iSJ%08n_9BnoP~P^q9<-rb_iXn{
z@SDE<*~1O=lX%$_g?~XU=x5BugPHJ7vd4(<;S=oO|FJ7SXz@+(=jjZ}7
z^6>rnyMC@Eh45}-eS7?^yCQJ!)$z-&Z69HO%2%*62=3iJIUV172*Ptb4$2QW!n&rS
z*C)Rn);;|f*6Z*0AkEP6!REXkw6Z8cf6LPzbkOq6>%_%9D9Iwl#VNlV-LuR*bEg^j
z6%BdbXiIE0mrV
z*ma^zHKl#q`QhGgTK$cMdGHRWq(s-kqwVls-)9YW?{1{~@l;xsM<2Sh$MTH+&u%n&
zr>5SXtrtDwwt4@-xF2l`Gr89=YXGHI&lzHG?MJ(2%rF=8>_?@BCQ4Ef{U|bIm6P|?
zess(2i>m4atDN+4To2Lx9iyCEPj6^ZJoWc*0ihC$K(=rfyJH>_PL6SN0CN
zcOwDQkG64P{m6Z>DGGW8{4Ba2w{aZWA*9oiwSEA3Ew{SdnOsW=n#hJ;&rn*I-4koV4Aywl!*KEeNd
zwcRr~mw9PDw>PUDJu494t`mUwOY=0&?SlF4xIOttMJC9_Eq`u}Xm%mxWbyQ6r+d-v
zN9^;DM8No8d$x3UQV;6YUS%~!x*rKv{g#mC96%i!np2Hi`cXsC^_Pd`2GDEagjWvB
z2hf4nhvmM&|K+w~md!Rr7+>dN!!y+SQBw1RESdALzOGa~DdOLUMsvch4OhTA|5IBc
z!@VELmd;eT5ZRBk!oFuC=MFTl|J`;a&`%b=oVd|y2Jfd11xWuq*@C7_^Az_QZ$bVY
zv+Xaf??p5B^&e_G+l@G1XyvC*wxj**s=lA#doJJI`K)cez&jN#=gV%vcpC1S-Mea}
zA0_h%_%y<}5a9FFp1z|GnIE^|YgpZbel@w(XaT=&-Q(J|`u&JK<5+UU9titxJM?~M
zFFF_4)~E>bzvFaub4W9cuRZ%81;M=6=G4wG;^;?jm!6kNcj`i04vMsLakrxX1CEi*
AY5)KL

diff --git a/previews/PR1624/tutorials/out/solution_000020.vtu b/previews/PR1624/tutorials/out/solution_000020.vtu
index d90babd33cf417103641662c4331547b685755e3..01aafe1c226d38f3d25f599e79adda35bc8fbafc 100644
GIT binary patch
delta 12070
zcmY+K^;gsl!0aDHMd|L4rKO}Bq`Ny;8l*cG_KV~K!b+o*l$3ygbV-B6N;gY4EX~sQ
zJ?}a9+;itom|x~IXQne1qc#=e*AGk#u@t=LTq%>7xfpzP-p}XA0Z-ij{#Sw7WX3_F
z)0PxcYDtW7MJGm9&dD4Ty<+@|ar{3(cn;HK9Hbyc4!No?1z$YV@+T@kQ>m90C8LWz
zounu4WqVxDLZN|UxieNSK}!gT_!f5GZT@5Nd9|Je)O_TGr}w{H47@0n@`X@kMl<0(
z#ghqk>90@ng#fNi@;B{X4|H?uqKq=QNKf+Nbg-+4H14h^)B~GE|9)+}wwdYV7*)kd
z^YL-O^WrSh0zn;(WqIYf7*DB%?}BRh>yTN0TUnuPj~>tiPY#Jj*R$N35wg&9;r@TF-JHYhK;r0@XCH{!S1m+NCM@7K}40jL;*N
zeMYE~=8g(3ZSY~)UX#PQ4iSUTVi=C1(_tgzLjzG+L}k^1rR;X$gF_L>J&thL6WsSS
zU;*}7v+0n}Kxwn5?Z?8&JZw{DdGopwo2bVO(^^&{P3KXx_9g#0$ci6b%
z6@yU%L#C|0c1*-A0lH~w@;IZUylQJ=;9PNlOWq`G3+wX
zm;ytS5jXrKN(x^p*AK`d3=(P5N0EmhaHc*lE{3V3%=FG25e|W%wQm*jC%;j5?i20(
zGP25lhGg_IY7Hc0(muXI-vuB%kTS_8@q@tM4Ss0ATSBVJ2cBP*uadBHi#NKx6=-Ah
zY$x&=81a_lI>(n7JH^Y0lco)O+*wZ`r!Fh~?r_3+?k>1@P}tyAEBE9_6eIrx@rdJz
zWmW}S&Dm>EhIgBkk$mo2IA_9j0ns=W!5iV()ts_^Mo9is_tWjw)#Fj;&IPwS^#!n&
zFd_7w|ns(;2vrpO-s{N*utJmIr2vFys@bh
zPk;=QCrwbqd-LOih(lWZjMjDMY|B#?Fz`Wf^)BI{>+VOnm)k3Y|vn?KJ3NvHXyfnZW?l~K`rkkaD?Dw
zF9=L94X|_-UE1@BALT!}){U<;vX}lI_-Rlww(y3uqO3{!E4{W?pmuCOnlG*{lGIFd
ziJd+xe{59hea-9yoN1sPV4rcN4AS|NKX*t6fT%iVDGBCOA$m)bQk_#Y$F~K0^RMw_
z;zG+2SEWJvA>dEnoN0?hojmgQae%mlUj9lv9vurtXjt7gexIY$#C@9xDW_0{*}=n)
zz<}X_D5NsS42_InYhlj%c_`Cy?H`)XACIRlVb^@%0|!ltJ{$<}au?F=S=_TCg2bkT
zR-kfNc%xL3EBE}2n-&{V`sHTW2yarZw-=#*GX6>?{QJYNAbu{qwfARviB{e3vRbw%
zq-%bh*RzDICW|Lgw&NAe(Y}>ibURt{?F)B)*YW(mjd-S7#s2g|Ec$n?)2>n&ffg5>
z@`A@NS5-4fbYGzWv$w@6X+5p}V8Lh2Csnt3Vc?XLDyNj5hP^NZNxJj6o-L`m41?BN
z-{??Id&z~XMC*1_zkw_RLfv)hqq)sEmC^0lQH)rffWd>ZM^9zO@;s)=wZuG@*A_TG
z`ht^G@9~j7Gin
zrX5LV@|K@PrLau_-SZ#%$yfR6or$7}052N`J;&#eUt_w>L@#dF=&hf$OzE~oe|yM}
za{ps$@8DH`K{e*H?k|ej;TWvdN_(Ha?v~G**u{ve4)L9x{|~GnWoqp8Y#$z)2}(l~
z5_Px%{@-7>0m|#p$af>fjI%mdXbgpIi3Apps1~L=4HS-tLtxkTt>}SVT>X-j7*>~*
zAC@6O@3fw4mn8$&`tYyPM9nk8w!D1*rs_Aeftp8P1m+*a!)#koob_?mh&faFhddW>
z+E6)Vc;Zx_qz9&@WqOG*)nmL3Gv`Im-t#_y1liZ>yO#UZWTBemS|mz3RO*U@nfp3u
z{J0bY?Jq{I7w|esT$Z@`IeN)(f8wk07iHJx()(_G*vRd=g!oIzeUL{A=vRt0JQQc?
zHq4lK8=r^AD<7R(V17@(8ntq)ixUHlQx~Q@>bwZEPWfe=b4vb>U~<>6wvn*x_D~N%
z;`t;I=~qm1{%4t}(?MU06KE`i3bvy`huhfa2g3qHq@)9A|{=H1#A((7jNakqRF9wNC7pQPnY+k?De~}IbvGrgn?1|
z+8yI58|oS?iFrnK=i6U}^~f>&b8s6YcjN`i2{Ej|Y+RtWF(}j#G&x%b-*b3Di$@^rs@HVzEstHIM%n2g2B8vb;^hp<+
zP;k#P>SooZBC21F~FFK%bNZx+Zq
zLt#a{Kc$z46q(Ij+&^J#)*u1Un|y+xhKu;czdiS~!&_!Sp`;}T742CsZhfL`t7Zs%
z?iCJseDZl8<%b?J;yE52h)77A34PBDv^{;vd%V+*+NI9EKP~m>ihZA)eNbB;$VGh?;pgcf^%Tyu_>+K1yMzyL!*5>%_Z*uN
z;MnVPq0G?^I^V3T*-rO~{_J2zSzmAE=NVT7tw?4rewk0EDgI6{uB3X*`e!oP*d;oh
z8uL*!-<~Ru%4+`m<;*h7TAyKPO~#&7!X5V@?&=eSdGZJJMJShlh9nr0vaVKL(%SMp
z7x;YFhm^Jp+$(rwmCyuE*wn=wEikaWY}5wHu5>u=HJqu?4A)B)g5)73hVQeS_DL1m
z$iHfTHsd=amHY_Jk6ZV0$tbyFMAU4DW>)}y!m`UdVuz@1&?iWb3ze|G*sa1jNsp~h
z;bQ849ui6Kd%;Ln3n*xh@T5RhuER(gJt8|hkNvRfQ{T**^8~#51m<$tZd&M;G_grC
zdC}(v-z-KjybLvZ7f=)NU{XXdLkCeh#jXkKr3(`=JvQ1h`cuE2td+F*FUH49!$S0A
z7VpJqGr=3gYVLF*N#iFh;!c&P=OI%X3l_=5k&uLrCP&VZBN-lTf1bk%!6_|mm
zq4U)|bHm8cnXaWa5QDQPB<#tX1ygT}5dpJL#6jl~(4OY14C)u8SIXeuA??MADxL+r
z<(AY}qO&=7JwRLiJ6AV#qj7r!|9w%$vHYck&zT)b1^5bEy^};zhV0JPs4EYA{xf?P
z6@$6fv9&@${`XN}q*dEKeAC|GsiHQsYaKT6h5)6RKd6|q7r^PblN5?$X
zy_uWFE|HfZL7Y7T|4|KVlgN+jSf&G7530-=CBn6sb0Ff9yGN?uCH{qv3
zT?;C1BDA+iHB_4A1-1TE(VZtUE*D4~fqn5y`uf$hHn}>UI0|_;4Mq&yJWwog}Km94@Z&H%Ho?CA{(XaHtYXts@?{9Z~~K`TTUJ8
z5|VrMt4{C7ey^0h4^#Gx%0CV^|77YEo6iI6%F5n=!(V4sau!Evxj4>I3+r-uV>p6x<~6#`3?@IUcmrqMc3Fa7u-mF~Ks`naoA
zfs%5NE#|nAtqquIeA!g^1CSl4=C__pxIA1XxuwB5#Hp8v^(@xnF=TuI=GS%gP^;iG
zoza!!n^?0Mg0i={;vD|xqi~8VY$~{^-r0N?D#-Y^JhJA7FYyl1mMER;rw
zHhyIvJ~Eb8z72?$XsG+rCa`n*p8%gAw@CZ9i&mO~4b#Tgp#?&xb+=>dR&!hXQyq)e5`@Od{ZM|%KCZi!MiN<;M1$+R=_bF-!IWK#YE-hV^Di@rAf7+4f
zL|<|lbpvMxi&e1!5utcAB)|JqktdFe?K!=C)xZ5HT~v
z($6PahW2KiCX#^Wep?z&$f(VcC5pETm7et`{e&#Jn<&WD4Cu>|o^|3RuNVgEHn
z_P~Gs@*yDEsp5Lz=?wOQ9?vq!7wNdevpv}@Tmysp1SY>A^Sv(p+7Ez`m2ax
za6Aq&9Qz`f>UOpR{Ba1s&^g)ZQ44(!h7ChZ$vC(-+hpg`1IFUR0Tu48y)x&T_;t|9
zD#7(3!xHTHo#2!fFDSN2$616EB+gvAL&wyCOKmKzVHcW|b!EHsONk!!>H-yr)q2Fi
z-gL{PMFvn=mG=(8&hhsI5pjIEnOe2G_Bn}2SewhNs1wP7>2NW?B7KcZuv-$u+fTs_
zCb4&Pj8FYyOnq>sO&D+@=030D>zpn+5{s^{=Zgk=^A2+b{0nhENAkW?$u$c4Lh#g<
zWPKzX@>0UnO^21f$-ts94xRh7L-}v9UCUYK3@BCC(~3fu?i~LKU2tR(SN?BD;-`f%
zRPid{7)hLfCR}(>qVSS(T44)J-vn(PR%zJmE$esp^p5LOo55j{lHV|mdT(R@wrUMB
zjkcZh!p4U}TxR~=Hh0QI^*kxV9OFhOhbgOazS=_=??&)5|==0>!o07))s^F^2Y
zN8*H!C$^~!8+d2!=y(UZvol!ngVfT1tmu*l#dyc>dOR<`V|lk_V`_No?-lG*Z^?cS
zw!<;qG{a9JcRuU+SGao-@>W9D555{Z0C9gc!>IE*)3=r(Nu=E8pVaX8OHU&=p09&C
ziQ6n-O_=O3-fE(djHK%kL#
zE`&+APKx~f@|7R0o2vXh)v%%aA1NURzS-uRCOW*^)pKVYw-{f2
z6g{+F@8(k2$iYNZWae3Q2Kk+UF#u~7KII|Akj8Q?B-^I33
z61NRvN{n)Ny<2;H_IdX0prcpPdD0+U`|P1%P(9Wyul9>EF~xrPsz(A-3h-$TQEvK%
z2+FCqYZ@ggSdGoCm~(UzaITzv#ba_Dc33s};nNL(JZ+zV(NsabV%7x$``VZr
z*<+O;F3@~(fP?9SH$~;R?aQ)jOTaI5Js}yyOD;Qg^j9WigfPZ={jS%uBj;wHUO4${
z`b+ascDV7#Vv|nHP590wA<&4KmUdWLc-1#uc2cd78$-0tenmw=Gb3bt@stP|>N5-Y
zgYTM;P2lzT3-z61jO1sd7%U*0em>RZedqzn6F!_dxl?1v+#}*?^0;K3o~8QjSfRAJ
z*XNs9qWD5eop-WN0&5VTNEkf!jVvq1?cs_EOF=g!5-S;GVbJrZxvw5`<6L2M
z!@AYzU*1>SnXeRZ&ZlRy{L|CMkVLn5``P3$_%X{~1^Tzb9nFG}W~
zZK{77x=CWIwSMTnd5XRgol5vCb$v-#(o#I4eVX`fD6xpAjkC*){@Q`aSv<~125Xx)
zM5%L9w3RIAUvkIMsQ6ruLHp3>CXPV;XpwPH=_tj6sP3mldccq8$RiYEPJN{69RJX(
z_4na_`yJ-Tmv(JEkF5>Ytm`tlA7j7Z;1^eQckWCE>&q)Ip(K>w9s;ws*1be#JK6u}
zJ5-?Ms>w<3Sej>KT&Vr)=)6U5aYiNVW5AgK(Fc>Wgb}m#Gq|d*Ky+
z=8S>3V#aD?9-uis>${A|!sy6kM~@g3nW?fOXq*(;LOS;&KOz)OX*hghJh)sHfsSEr
zf+#Q{1diN;8KF-a%Wvav`5sqpe&U=)mY1`0o_5~WOA#q(a|>~cnU&b<0MnFLs8R8z
z3=@#yW_Tba4tH6hm-|lr<0U*`WzL6%Nt^ybqFOD
z{hZ@*p_UKx{&@aopWXU4{APAF=G&yVZg}rB{prEKnDuzS;`3Nj@Llg|_2hL&4MWhk
zvC4CteG?nyD6YfOx9NY5;g7k!9j*x-(uCCUz<(b1m1c8ZyKu5XhReR}BhmGaau*dR
zA+J1O<6gG6^4FuW)RM@(@5Q_IDdmhK(EiI-;s}a6uo~Kr>S+LMcCh@>`Elsectf?_
z6Mxbr|EL1VL+l`Ck>l9~Zhug@RxxKOj;^^mH6a1RHYYc5uPmH+q}1+GyL@0QcSdF%
z>Uckl9~rxcH!dw37uOo3a%>f<#1JZp6qW~EUtJTYq5LqEQ&fL$E8{yBgA&3S$
zDq9lGa-J3Dl*+p|Ofm&NJ*3g}?|LXk`%7u7cE%&LUp$D4_u6xW?JruaEqh$;R(Q#8
zt!b;epWGnNgn|d42fcI56BVwb+{#iU{GF4wF@S6_LhD}Hr%$hdY&_ILe)X3V#vo}>7Av!6=btxJIZ7oc$PJ9^7-(M
z*zk*9s1|^vw+a4kO^SL(9x?tg@PKcqFX;MKi*QkR2v}SsQ8)4|c4L0r*d0w5M%dZC
zs79SU4A~W>QQS5?eBiHT0(eK5g04RqCd(;U=^B|}9#!K%4jHQ4RaRQb_<$DD0Us;n
zTZ#eBheDVYmiFxrLX;c$PnLp#Oob<79-2`@=zHeekxA1^|FWsvfX
zYFeXH?ope#4uqNvD*t{6<=S|*cLb7$oW%n5@D+=@twtR1<`Z*kmcm!WgbI&%wt%U8
z?Nw-};Tm&5uGPh#=*-hL&vltw%Ra}x1Z3=$l$v>hN*<+mUkJ_kn0DZ)~?Naobbxh7TsKTd!Q;3rqtP2dL+;EBveq0%h%X9
zt5d^NUsZpmj7>!{6jbF^tg`ZmRa<5&-_E}rCD?)C<~4Yej~@sMyob$X9rMzRM&~`%
zdrHWiX%#v}uW
z-ODg$t)))2GCZ%v^nwnI9l;>%pKa;~oiC(C4VBsu#DB>cAF>jzrYCwWfMo=gdMvpO
zDuA3YYwO(H+W7OodJbJQQX5QpdaLfOM{~snzjl|E3*Q1$8WYl}p*KLV_(D!gOW1ZA^*}jnf6X#xb|Oe%X@n(y
ztt3v%!wqe}vhY5ReV)IA9;#_5RU3eJMBpY`Vd7p$Fb+G087
zAq!p1yn1Uy+?YClgp7AE^B8#7hM
zD_;a9UDqbpzCgFfXdG_6Uf`8vr4!Y3wb$oWLVnW|&@qQ~vAxud{k+!v;8k7S03
zP%3=2U_kQ(MJ5!mSvPh(B=oCM-KUQc-Sq=0LvkL+`3H#~rdk2ey5O~mENo=@`gBmvv+ve
z`Oxane=JmqO3ENo?mU{%9DG!%B}Ls;)RiuNAxWDHFs6jv}oG~c_ap>3*d>jGV(dy;V12b5)U7j*MH2cfJ!u3bmBWcd%wisLE+f>#8%
zSVRFGc@jtAq9Q_`bV;o346O#-`$vXyV}YQv={=XN
z5BWw0+*I9_nP4NY1n@@+1*2l1y9_l2X_l6~0WXAN1FL@y-HSb!nPyRajg<`DI#uAj
z7!QRvxOYhbleKHImO*B5Kjkf(i;o@+jjOk>V!9HECP$R!DnIU$L`jo_c~;?oXd}m=
zxrE|fNP>*>rS(d}yWH>KThny6f$NE=zvKnr-*Z0v?-=IHW?53d1Gypo;)y1kq3USS
z8oiRl<>WvzxX
z6d;BkcUMESQ>|OK!$OjUd`5xymabYM?o(lqItY>j;3g&pk_k+NoGwNa)mx1th#CcC>aQC`fi?q~Xxf2dO$r^`^aQ1bcAxv>}##UGaSi
zPAh3>wm0|XH%F7dol;bvP9eH%wZd`+fM
z{^5A)l$qgAyQV3h<_T~x%j4lP*CWE8{Osv2hx1agqjN(WOu5(vhL#N{JQei-jD5#Zlv5{!dx9a|C#v_1E*SNwnYS(06s
z><*9_VHmfRcH(h&x_ws&wqJSRO~P{YR+DfYcf2n8AA1?9-d*=SaLUU?M8Wzp4qEzR
zB@|kL_6!?A!D?H4jx0OUGh)t(N^ZUCF<9-S(GFcY8IEoZHS-@_K92%r{drOuK`era8
zr>df+r&~4#Svlu<&7!$>Jpx8t<{orZL2kINL;Xk08-=>ya^PFV=At7~ts#Y`i|0Bj
zHi%u#6+}3_)D*A3j-C2PZC%
z)b*N`-SkW4#0XzAM_??gk(RpZ?E}^L=$e+TpUJpBeqD^F1*Vy35yHXUM?`eW*%DRF
zzW%OdcYcfQkyb8%Sj)~5Ebap$I-N
z^(0Q$*4s)V1$dOV2;d`5#tFk67Yg
zL|5>tKFxapjltq#zMqfVjDw>h4*UPucM;M4Jr|nxCHKx;aP{2Xr64b~7BAnwHYsvR
z+ew(NIXQYOa7giMx5Mm^`2h>7EQ`!^_K)19ar)RlO>*rher{@%>HSI-Ro-H+&2e3m#R
z`@I~q`q;6(K4L~BeMd!_>-#|`pXQ>4HB+`qd%ZJV7sazeA+kk}PWz}B^fRS7
zv%ots`ch=0OzfNB*@C_V7j-0i!}0=}2MjB-9yPu(W5o9}b8;c+xXQ?m)eq)Ua46Y-
ziv^|zUS=q?;xkX^m;!>KBRNW`cZOc7p9m{!)cZb*yiWR{X6Oo0wyyYHzY{1J6)94Q
zmu0Lscykf#bNLKAMp`xfu%Y!PaMauJiqLj5?dVQEZ2Kgn
z{yXB?ps}ZP2|EV0W31`YS^lc&lBF59maKo^CIcq1$D<|4zURw-&UdK=fL6X|*_w0q
zF)=oOR}}VR8R=U97cb|qL`Tmtbh7t7Y<7RAq>_(JEgp}6DU8)7}||^=+6@PtOA~%C#w;_b8b_;r@xuDid0cv2pFe-nSQM5#{1}?
zE72eMb-;_=K~fNV`+V*iJ23^8Tj$WmJ_X+xXLV;FE$d>6neHHIi0WC(ImQ7D{eS&x
zlajZJ7(rRVR+ia*Lq*SL`G%y)`qYYsmI;M-yQa`-?3${P8~va;}%K@E&JCN
zl`R!S+x0re_Ng(K2e{X7Nw@%Nd_?TS0T}<(eEr3+N;`{0Fu|y(F_jqU!$Y5ggU68sX`k#o|?bu&39i*JLbW;7LK
z)j9$YrnMQ{N6>t`)j?o&?Y-l1aPQR@dFS2AF$tP4k%%9NNrT~pR#z9o(&xO;xSYW7$s&`IPjw4EklqG-m8&+#SI
zH$HrvOECjM;r3KU$F)0eev+k{9k(!_KwGx}X|i`_3gs
z0ooKc_d(X_Pj7$(leCKEn{QkVesp|RXVHfdI4c>%vC?3bVd+fCu6)P(@UtaNuD8rH
zQ|w$svVox08#)Lhh+?+!(CEhr&0XkYIQUu9_2z_j8?1Wys_Gz!MS(~Ew=eZ+#MXJK
zp+kyTcKh5NU-*=S`u5Cu=hdl-#oAKCc#gd9uKC`20xq!s_v1M^cR@{&`S*GldMiv$
zYp1j57$MLq_%@V-xhjEQ<&DSjig|DbE4%B}IBm|k87zsIA>WZVVYxy(wmez^{SA7u
zEwV1tsevyj&Gx8&!6@NqBoTuPhc_r$J@*hY-j>OG^_dKBgy}Iw*W83p?q2z>Byc6*
z(|;l@Q_lguUYY3&BP6;Z58V)L4qvlj4EJ#rav60d4dHB^JKzqVo?iX&+xKNwADT&s
z!c?-N-0TY^yT`n~E#1^#fg*0mohTBVbcJOvq>{Uvn&+TfRK1;6W>hsn`h#3+zmf5w
zzNmMl;a$<{ibeZD5N8yDc&oC5F!2-KqBqwmX95s7ycOE&x
z84GI)_^_X>LB@_%$h$YgPOEZy9Ys1GtpfG)J^@e8aA#)3MBBMbt*Q(LpSo2bq4a?W
zk{UQ;64iVeul@b-LZh^FCEJ6G*
zAi>chasNued5`g{At$we3$fBj4drX3~%NU-~T+$xq|#-j%+n9p1#8aod7dz
zzp-B<8`wK*y+HTG4lg%{9~ybNu$`=VPt~z5fk*g=enn@VZYyWF(!rgUrZe^YHq{oZ
zF$|vh@VEAn=Z{^$4evx`fHjXk>2;&NO8TB1X{<+;Oi5tt=ab;R_jhV>Xq0?$`fF28
zdYcxvGMQJM>5EY-pG+=46!zDyBW>~?gnVO|1sny*5I_#tIvlm(?qWd*E
zVPEz4CKw_;uz04LvIwnWBB2#@!Sc4lL$}yN)igof
zCWTN3&tAQPlC<4Ba28Uok;DoN=MboV(v}%jb}%859v*^@U{kWSm~}AU@Z89pqe_ag
zdQ_x&b3O^xw!Ip9t+^vV2woC)z45t0(F87Ovd{Q1N$QLRqS-&vVXY}7>f_H>LzH-s
z3@qpL|E4s(SVWCAhjiMN@81ai=?)=ihCDLJTV3&Dzj-6cE9>cL?`~&rr(o@CEsZV7
OYh<9VW*kio<@AJ&~pyMT?
zWk=*kmy8ih^|?ba%|daq3o*nZTUby*O_di#t&u9bLE`drpaf$*rlf$&l0_qU=tDD=
zl=Zwi=W1~al_+^=!V*h+)(n@ZXcGt{By$93dBW~xcSyh(3^5<7&Acyfp!K$6|
zgWMw?bd&Ly5X;ioean{V`kK;B)Z3@kkT>?{A*8cSA@g=IGesa5ga{zsau5
zplcIrvT7Qtx#p}s{2Qwj(O#Y?j00kh(Y*6$mD?xf5}xPw%IVBPUNJ1x}@@Sv|0dPU57dZo6;V9fpDT~6w2d$U7KEvT+)
zp`4Y3=$mlc8KCU9mY5qqKQ$Pi0`p
zZi}JYUn6$hut6|*@jcj&aM^|_LPo~oZ+c1*O--mqKb-J@{hren@!^1<3}eu;kZNih
z2;_Q$hl94%d_4D;Lnckj@sOhpm)Ps7uk0R=+T%3YV%7hCuwA(H$3fUYJ^z)#38T!f
zn?lZ7)-IE!KC&d{Htc=F#JvEh(}B&s3x*6IfAK@eQupIW#UjBd36sv!%dz7SvEzC$9NU8O@9bCC3UU%vAw0YHJRJ-w$C6f33G2*7O9A`S$0G`-R{lQ$1vhK
zm*EM4VTQUC?Sk_Xkjdh5Y~Pdk!O87uQT52POmpL^|85>KoVC%~(6_~A<3g(os2{0z
z(Q&geq&ffYflC?rUOm{4&t3g)B&zX{)|s(P->rq3CZzjKzpas9u-aY0@Oq*qPcU5F
z%nPNLK%4)g^)Ix-szuWLp3-$h%xGaNrP%MYx2wE?6!!$Hbcbgbm7eGb?4-|Zp6UXL
z27ff>V!LaZ2~wJO8IM|yFMrLU0O-T~<07oz{jq$@biamJm2vwSPN#O8yq!&IeO2kW
z7s~p%MPojn&HmeJa5t}dLO>pKTO(_nU8|z0=yRxLnpzI!o86Cpf@4q<>-Z|uyU~xG
zEe|@GM)vO7))8cTgWa<0s}$&P%1lA_*DX{XzO(JbkB(*E{Jz0e*RO*%rvV13O)lo8
z6GxN@cFg3j3b-lx!OxbYSQlW-%R)XBIv%g*P%29nsgS=vv
zM!j9k9`w<35hwq4Tm%-u3)i@H!;Z`>i0c!en9%LiNj0MKhyJyF{{vOXiKlhN9`~%t
zC9RjpgAx~@kh8|U<4sb)1xns@%7=b{q^#+HW_`TT9>eSO$r10*h;1?mnYdKf6CDfV
z!tJd^7<;_^j%XrzwRyekhp4GXr^Rx2{l6Tq-yW}qK}YU7ciYfKE)q%6$6rzK(}N2T
zJk%FuzLhc-Xw+riFNK^Fp+<7ERKc^*3`h!}JI2EKjcAQtI*!x%X;Gt?)4
zLv`U>uzIzjHNnukk?7qeEJ%UaG;*FZh&Q
zgTWbnQ4Te)O#h8|{g2wPN~kL{7%e
z%1xGcYbif(r2)GPs0~emzuNmg^`{Waf>P$s$L);wpU*_+YUXMwtqq4TW-&hJRHfKw
zSM2bSEP){6us?a#;Uq7f=x46MDLY1V6Mua+(=Ua?aTGY5^e3@|8VZBux!+vbZQxry
zmD9@X@ZSS7qF8MhDxRZv@2{SYZ>i=kc`BY-ym`3ex^LAffO4Z|8B@XBt#m05stTT-
zq3@{EMLW;g+*_ws`OO>e_EJQ2inST9wmW_U`FtaPMdWXumFFo{a+4?<8M`tV#wp6i
z3(uP#_Rbo%{d@k67?XD0ezvMu@}p$5C&{rp-c>c@EDDfwnTR+`MwC`cK2MIT%iv{G
zU360q%M(2&188Ds+W{SQyS56RnMh
z8$lu!0o`&VNPcZP)+0qfvP96HPX|p%Sl`KQK_BUNU2?|b_laZ_b58q0GIQ%WR7X7$
z@oMy;Kl$W3_XFEogjL(=J=3~}X`6to4mr~baDX2#uy$GYlu3|Fh>pmasp$`taC}_y
zF8d&9axD5pYW6yGSm924;_^LpHUFw4*(AAd9O)ZeJz-x95QXbiUV#^SzXFMHmeHvI
zdJ7)2SatJsR`>l>)VY-eQ3vJ-v0kKF(^gJ{(UAtpAxisU1pgb~1NoDrfwwQxH_Av8
z$!?Dq0OkQojN@LC&vQ}N(Dusv(>JRL_G-=5zxvDs(@`oUR;e=8UU%fko3dt*@8n>|
zF)U!c{Ox!V_}w;+cjLU{ZC3UQ0$*ZnVR>^gTEY<{`0kNF1Nw#HCqy%0HC+?^ZpU@8
z{WFS_`r=DwHR{6?beD)1$*btZRc3$9qxg%KE7`p#St$bcnu?6oD
z0`Vcu1ktMm2H@8>rCM!ktWBhHGJ40qtCAz0;9`710&WkA5x|BMg=ePiJ2%LI$$T&7
zgCAz~&sj;4d?Y>5%vZPb1;6k7xO0amu
zm)Ve(7%_V4#>C$DJXBG=xLA^yP#5#BK$jHWgOyHwMZBn@X;R2zzg881M;vltHqk%6
zb>~1oeSMS~vvQ1wu1X?{2yx5-tE}ainA1nj&0P7CCm>UBu&;~+h(2|-^Nisfc*Qw+
z6%$^~s5LU*ql}bf)QjaOE1PL)hdE~vLCUYJ5{`y6gf$&mC
zqv^W2V6R+>;3kQyP8{o!b2!EN$`|u)^HxRVwI}{}ij^s}rq#Oo=v&-rQw=4Zi5~x;`xp5G)roKI7_IH&_y=4e6AzS*M^@0BUba_3JCw_Rk5DFpDcK|ID;%v+hC(Ofb)$ErgSMafhgAF;qSTeUod-=BB-;%#fX
zG}N*ZOFwiqpr!W$!4QhX1-AjRjW5u*Y^44I4hr%2d{tJP$*8-)VHdd&59J|eW#m9L
zdKMGq+HHfXEDE{*;hS54Ax~o0r1u*7gy}u|7s*X~Mlw+SQ$
zUM1t=-Myu*m{7loA`U@utEOzxUFSKL*I3Ab4{Xjs&wFhE*ZBKei!!}VRg9zKBL1{R
zhc#onaICX9m%KCJrYdY7V8eZ>?e+h-$SZ%$|9Sp5|DgY`q>v|hi4f`%_2*PE7s!QeWSq$C}I%dc#)GZ`1ZQwAv1mP?f`g$b5f`hl=Fr!DexlWmEp
z21UJLm9)b{9~eWqzvhC*ov}@>fGqxuNl&28%d0rhvP$a@yF9U3wvdh
zQic??AMK=k@pc5NQmy}-Qwx
zX!CN}JQ^VAIApHpZoMgpnrhLxB8s+*Gr{86+_DXuq`QsR*62()+Qv+WdVA$nABIg;
zz$4#4m3TkOYQR3F?o)~BpJolynZEwf(AiYWbslh7@74nQ1z>(~Hsu+coDeyGe@s(&
z#+s9sl5JBFCwHTPQSwa36V)^8vLjj#8NuADs!-5~%OY4XcsXkNt&)KgKPX4`{G%Ch
z{p}^EuyP8!{TFYTl9_RFDDwvR;GqaLetK5V;D!7-l4*BeP-5RQ1N
zuFZS=Q%Bq3ML>e*tHGMZq=;MSb2NZcjv4iosQNh9S}mMCW^m%n*BG!!`WrW)caX(wqpM<${zQQN){D3J@h!Q>u)SIk
z$lDQXHJEb4Ik>tLvGHNX@}l$7Uy9|5s{V4uc<&+8qI1CJPL;PvI3mU4vT2^f6>Mm5
z`}#s#!fGO9UO>&|k^rpF@G^q`!}ZVFA1xjLxXCX00y9w>=GH~M&yD-|3K*ADff-Y`
z?tC*d_`tcSF#bdfw!A*$+2DVEvPoHq56X^^ydbJP0!+AcN-Ukh@I$kHFOstJOX
z39zr>PdJvG(IN*KwTO@=umyTP2q6^RxFRP;wxC{%^*ozJ;Ua4^eXzM!)a6RZBMGnQOa$<7E7&Jv
zhVcwIu~kB?E6iP!!`isqHVJ+|CkJWiz@XwQg7z4If;$6cNOsDFMW*gcZ_TvG-GYpf
zltaRi`#mvrBJ9_|gz3DJGrc2m?pPMT@TZ6tj)@a=0NDlI*7paIW;>BDBA_79vR@IbrjXaqV8$e#4`0`WU?Q!yxN%;5`oo^4XBg4^#yTDqugvk$jTu+DuW&+z`FzYl77cq?f<;RvsAZR)kT`GZPwDLT)#2
zR$Mdko0=am4r+PFfpiIlA1Ay?q!$W$VWU3TcMB@>xZu$>!$c$?jlbBzzhWMvvF#K#
z=W}|@S9mGxnPn2-rucD2d^O@t(cauy>nh4~{RPZy<~|+@Bp8kvl3ye2+6MPOsT6C=
zQJL)4dD-O}l)}ap&MKKI2~LSNq~zjk7!C#^8NsmjS}A7$dAAH}DlF7QpmDtalPXvE
zq3(QRn_rBe^fZ5Goms5)0CD!4S#*IDSrM;3Tk~~sxzoz!0q(2<-9Bt_TE?bjnlu!r
z#&z;@1ByG`7c|VmooL^*2U>_X!vVL3;A^-18^8lIM`NWw<9AbkeB-YK&~!Sjfmepl
zZ_i<>BLpW<@co4Qq!i-Ke38Rzwp2R9Gd}jUv^TJc$$nEBxO@50@731NPcSEtH%9kj
z#)tuwgZEmn07HmjZgk6OtOwnPTY-C{ae#(%Se~2sIPYnH`25uLJe^orIvVaZwf-X9
zaGQ@c)yA!m8Ee9%qQ&GqWjW1;HuU_elicUnbq}y2u`-{e|K-y_J{|QJVA?zI5i4W9
zZ~sept)MY5W`cFzPUS`F^>mz(oWOKZJQyG*_^Lz*#f_;>+6Q}9um6vc
zL>n9YXf+i0(w6MOa`BiX{fqtTFsZ|h`=EaSs3=;2qVn?4CQ-q&&#&chA+Ee`jT0gc
z`G8ffJk$)bfN;kh<|k;C#HOSf%KDMXEUlz;wyLAB%+%j}T5Oq^+i{a-`<{xDH_!~7
z=35bgQbMdwKw0Rj?(QK%v|I;V@l;}>fg62`l8hU=%g|tmPlfKyP|+uxpk1TW@XAYk
zi9Hy|j0CP62mX!l`uNF_D}C2=458$u1>&26pgl~>?;Bu9LAd;9*fEN~N<%5-qgW~A
zjI(ODxI#}zTTCJeI
z%VK%RN3zotx)*-v<#0V7D<(t5L44+LKLZBhSm4?d*Nw{5&Vc6X%&V}EMP~^D^1N9`_(C-_Es??XOO_(8V7v$)<^
z4@S+nr%iHsj&EH@vFtGa0^u)LUvXC>@N8y2?~OiXO~Z>3S!2)C_|tqRf{Q^uKZT02
zRcXYkQns*Yl2oEwrROd$HjRn_X^I=lKBt$f-aQ&mcz7haw-T8BuLN&fKX@H<`BxJ@
zFU#L47}qYA%T<%GnE4uDH7xYmTcU@5p=w>6ZBv})RvSKS$AVR*y>fe$zoh&3Bhi}t
zB!)+P>)KkvO9pEpb7brIobp~gFN>h!|G`_}zrnWl|M)i9OJsI|sGm-e+PsE2!Q7-q
zPIO2`FYdRToDBM9Jd4IGUZjOwo>X3$#UM#;@uZxBMMMm%-FZTI!9%7h&~CE3jPN~e
z-_5!dyiNV$Gt!wFJjnPiX{G)@7DBN%C*S71e12I1dKWN&yiKphZ;Yejr~Tdqg0tsK>E|0TLac{T2vXeSf7FiplMWf0K|KPcPXqY3
zP~|yCJ*tlUcD^hdUH96E-Y=<4BWF=jhARmmDLke!-*5@>GSllQo8{1}VCQth6aom*_}zt6DfNzop<1
znh*5Dh)`d{ynKshF
zEOLm$x>UmTjw6)rC98YB`aLYtaX!tD2B_Q}PzU8LioMuV%0vENH@b*?e}A16jnd}B?V^g^PEOy&%@KLMN}?B
zaBuEcoaKOo#5?D{`rza{rp((>A?oo#SYw_{b&+E_N?2ho_uTr+H(gjOo)`6KD<$}_
z!=XS;|EF>6ZNVE}bb^5$PN3^CQL5A7*Y0?#m48MmqEy}lRcv5AD?HX>X#
zIl}ZN`=6{`{AMx;Q_esq<~$3CnBqM2V3!Jm3T^rP
z;+ag^vNkPV8h|Ki_ARN-v?j5z@Qo9w$RJ*qt;4ASw@Uf4+;%sm&8v^4e++90F9*9(
zk<41ZoyOBry}47Iyxq|mDKm@`(CWOHl{ECzC;o!>Tfd0^Q^nj&)QO4wrF0q)#kg8<
zNnmYva~XFG64&KZ;Sq?%b@JbNI+YA}ryiSJ;sitY#z2dssN_oO#N_#2(u$7p0)u{d
zzCLi#n&7lZV5a`!7otS6#!&&JnD7bFSL>O|3VTkr`27+7P-~kL)fbO?DQ<}`_Ui#kJ766*bL=!wcceciIJB-WIcge~Q!33xtQEzMQF$
zrrly3BhcR6`iRb85A55a2X?I0VyOidwMsQ_kX3GF!&?$>A((Z821-#F9#H8U+=6{;
zL*Dyr7@n~cYsHe?P7PRueL#2Cm^UeJuX)x;YrM*95x($RS{*;Ef>V^w$m5_ct#=t-
zF^}a)7;L4YI8KqBn4O$7cGyuv^J^*?=ri8fdfcGi0Qx~Elv|>dW-zjQ!WP<2
zyO2JlXLAjCvTgDttIbwwsYfE*(11Us4~4KWM?BQ)81Xk5#z1j?RdxK|0c<1E5{oR@
z&3e5$G$On%pYEb~`&p2^jJuzQTYA}cj&c_yfz0&QiBEF{uI_1rvY@*
z)X+9IU(CD6f%CK!GR*Yfr|Fu1Ad7*X6htk!a=se9r~U$`(|RIQSfa7@en?M;D*Sgy
z_r6a()#!URdF`$5ueKqvqrzL@9^=sn==fZdHPQ0r6Mag75W9cAyxk3UaJS=Au=
z>Gr1;ub{r9bIWHd+CZY{O4{#w{ZSeIW`eTucPn{>j7d0;nY8$jdfO@X6O^cAf1m2+
z!-_(Wk1Wq^|2#WHDVab-#qTe^P11zgE~>$`(-^$ITu@DiqV?x&e#MQ|@spMgJK+no
zORVrK;F4bLh$)@TJyD(T`B`GwB|EQnnrsv?eAj-1a{vxJf9FkHRq8A}J4{$?h~92p
zV`h~j&Qe#d$}0bq%Ja>W*|6xW5zHl)Vb~*Qwcq89*~A=!pLKrJqo;vc
z6$lQCAvqDXybD^VolS)|lR$i7DqHBTpntaj+s*5|H%j567YZvQyr^!ui8sr|VFSYl
zklLRE-Tb9@am4`r9HCn}lLyP2-4ZkJ+0C|fC1L3*ll8xsS_3*y2NrgYQmFS9N4x`|
z>8WRPG*Met&Wlg_&;yKh3YF5bnD;adN-_f}QjT0BKhK^pTbm#x`?n6i>wkScs-)!x
zrYXD#o!PP7vN*OD!yi;5v?5@_J@fO!AHC)l&tT|oBu|yfuU#M3$`3rRFm=fPmCid}
z?=Jhb?Y87;;&FD~s=LrVKpLdFs2^ZGe*A=23ggVLMP*~;0QwX0-OIVR&jYrqwnNSQo*-ME1a-NakdSQyzH%w0f-#KstFbyVwT!ZzhvCxC~)i-6D?4JqeW~}
zxb~q`>kzl-@-LOg`6@>#cjpb%k9{=3LUxsaZO%^qDM{t{lIySNz8|yv6I!M|pcHTv
z19!LjE{>hoFFcuDsk2+~5g*=2?2$6~U9Q%X%5<;|r?T5S-`#0p0)R+@k**oBW4!|1&n}2c*#}5B@M}6h^#t
zRc0zN8H$mKqtdX+d3{?*i^)Z6hDY_egtp+XZk(Lrm*2GE)S2zK3s$E;(|Oh)Cnn=o
zr{h8hy&gx%oY2xv`HC+rRLW!H!$Q6R)0R?bIp{V&=2J@!uOBIjS0#Z~?J+?}eB<0h
z+`kRYsR&#E{gTU=4bc61L=#6|vGAY{-G;0JX9ZQZ+nZ=uYy|46b@GU^-%d+~rv>Th
zIDU~t6kDI);eISx|1rvyQj#oix~TCM?rcmc>=`L}h?4rior*D~eX=6Jo+fDvQaKWr
z#Ng?{XZX|1CE!=M$Zh>Ps4F>I?y-=A(MMC``)FqKWgc
zlM$NAAFYH#*r>lUtMk_yeCiI+k4{slmGPtiE7MZZb%XlI&Ll
zy4Dp^AH_`_05Xc1q!9DYtsG(NwY9k}DnCA-{Az)${!UglSIlb1
zQ&EHC7?|0lRz52iOspucbvUWbBf4G4OiCcy3O`*4P@cVP`XNsgnLfB_e)YSK_|Wr;
zRrKX-!u>)xtnmEb8!t_R+oZkw%lY(mh(PSTutnuJ%Ib~W4H@8qcZ+KOrP#8N4ZE}=
z>Dmg@(>0mBt*9=%zmGfGkp1V`+DHZ#o5v=O_sI>~5<1prCD#1mRp$w%$d-Sr-MK!V
z{XK!C0GX!fQ}WxN;Np>QTGJfspA0Hp)~0oE{erDWG`$(SW$W{SapAk=*tO?~|y1$8NL#b{llDTt)(^Q9)*nw(<}U4*O27yMM^S94K2d
z^=QYS+eQ&-g`S}IDE7F~-O?|m*}uR}V2nqt0X&u#oV=&1AnmAHI_RZtXnhP<493ZZ
zxYjt^>t@U>KUawO%O>cIj0Y34+^GDg_%Ac&iL9Icu;X^CJt!{!Qo>;A*jH85u!xx<
zd+9!+h>S=|w-dLVh`}sj-(b>My6<{!
zeF4=%j_nd--_CEZT!Ow@q#M8960*~V0+0`D{#gcpZn0leoDP3mewOtg?9G@HRP-Ox
z`(f$P##lD&z|8oe=o7Dit48R&fp0SO@*~bCv*^cl6MQDa)>|&ce;I+g5=j}3BEg5^
zZ{Iq5GcK3u8#?4j2k8uDZ&-hkb}(#&}#7PPH-_9JK2@*5i=+3j(ZW
zf*l-qI^3T?ZIhCBp9(ogYfYdom2bOuRB$9kZ3n<@A49FEEdR8O3mjoIRjW@XhNcGu
z|4~7Y{`h=_QQj^>c8@ujkMG68&v&O0xNLO>se>b^FjYs-Uf&D7MY;(hTr`rh7oEEY
zAMg_rmogkSJu*ve&)uV59tb2Hm;)g3!x6IyYnzuBl`{eM^%f?`e@*;q;oz@OVf4VtA+GU>glK|sy?-?*0Koa
z9bbx2q3^;`S1#RG{CbOOBDTHd4V0QYv25y{lr>?}H&|Bj#7i=|E*=l&7C>C5aDYB`
z7AY48-^^!^_c$B!%L5^0>Vyji((*dK1djhbDt!>*~%RLcmTHaL*Vp=*uW8YfFY
zar1V3!BI{+q#WCx+2Z@>R$(4*(g?X>1Fwe7ciDKp#m=l0H6ahOn6&KKh&XvS1@e&X
z8s)EBg2_^tf7{ANG^bdTINwfU#J;g3bSp-V
zTWw-_
z#Lt9B3k&W$BS+^l4tD%x!8(m^FW`HxdquC`2O;1PASq4)6o6#8z*zT&F${DSNxK*R
z0koL8lamCKHe?r_V9i~CE<$hz-`Kt#f$1$3%TM5d<##qLJ7
za8`%*NUf{Bxa#fcF!f7_b1ArnIQ3fcSakP|zu6M$L@vec4rAKYu2;^dnoUdGSQ`K2
zEvIJ?^5-3?fhw?<-np9t@p_v1IF;>KU``QDxj9j^~u_Ek?_3?kRw03{eAxUA-pHx1}-ErIE=p)k37bJMS%MX3Qyn^ix%@Q;*4i)&)+WXI7t?--O$2V_@L(uLzOt7goH=83ye_y@O{i$
z?X$I_KA~V-2#Wr3vKsv@toMlvodL|N*U#=8llLi|x1v7~<{{2@QIX1qf_Rw9tW0O*
zAZZQJw6)m_wihD5WDiOS9&Ty-o7Z)+Kd&pW0HhZ_I2vVsalLU6SQ~U!2j-gNwd1{w
zn&ypxxVvO6?K;P7vPiYt`EVa|3UYw=e9l77zy3Chpjhc!%B-ncQJkEqlgU}HLp2YE
z^tH#yr;KagThUVe^xnd(OE!%GQMUd2s1n*NY1|!0LueM}!70XFj1ZG=wMPrQSHhD9
z0IFUXBV`&hP|`RFEKxzv=E5vppa@wniXd}P>eqNZB1O}whQp1tzdYVAF+Pr3!=FCc
zUp|ifX0flID|^RGX!?Ga%u#tK$Xo|*g3d;kp)J-lqj1xX@cVlllqJPuJ4gFW?-tc0
zwz9{?HUT-0i;``eddJccO~a6h_u-$4
z$Z%1E{P{J~2E^J|CzqV6zCj;q04JNMqm{VU2g?SET#5H2J)}0T2^ct0#G$JZ9HS|(&53GbV5xDdMy?oCetL*eA
zbNLDFyu)8qzk}u?=tR*YD^Ov&*3u6-QmbKmdvKKJwcj(A5O}y1Y86rGngy
zX?AubBGE5_Lb3lJ{C~U$#@mf@qAr*t^n;QB?+f!sK-pZQznQRv`2PS(=IdSsUWr7n
zs+2h+$Vw?lsY)qHN&75W<{coV2D^CEZ2!xY|BHpHQno@l6$ql7`^rGKRm*&R0;D>r
zNlEE|DOpNme7*hr-2(!aEn4Aj7r4UR-%cVra4hPS)UU+X@8`QBc!|5`e`|pptCIca
zU*!}9yIohi`TzHZ|68$ttFN7;ocjuQZ+D+n|9#nCLY1mg!`NT4QbK#Z-2;}ot_t>Z
zXYJJhQ_|jfh@9_wpXgR`BH&|E)Y}Hd{J`RflqtdgY}w
zsh?LZ^LCd~*X90ecm}tV8vGZ#kfi$7v`m#-(;GQSc}f4u^5vJT`X5Ud`#Px?RvhB~
z)xXMoKT*8_tK9wm+kxrHSH>El$XgOO|P(`R#`OqbW)oH*b@al>bs+i2|wd
z|Ml1Zi9aOk&ipB_&GsKkX*Ni#AeKbF@-Z)H%Y*;=ue8+vTm4&E#8>*i>t9_a_J8kR
z+Abt5TW#@wZYSfd`saUnKP%!|*&qLcKLz}U03~vKd9koc{sr3vx!(f(Pjde|zN8+B
z-1q;WSHk}#@FjARc1Y+FyChIT|4;Tw=n{Pr{6E1z|7fqoeu=!KK8Zewyo4@k$5%m*
z#C}QtOXMW4i#aw&V5D^#8;!!d8u7(PVI1$^L#oPv=T!5gbUMTO+20+A3spaMS
z2<0sW{!~F;l?{ZLs~JcHF_*~(TEs;N{iGt;+37Q{$4-Zb^~$__sGx5en?NG2Z$F;C
zX(}~9%w0OdTWMy9xa=W(`Kxz%{FNWg5HaVN-^BROKJxgc3Df`)
z7v75kDKTfM@|NM9bg1LR+@3j_3?5zfp3!rbsWbe~<$VlyjaiP0#9Yoy8XytZIT38&
zTWk+8T-{%v;U4ov4C=bHW6)Qop7C$|rI{z@RQFjiSROSA4HR+C-W7~*HIqUSF}HqX
zD#N$y|6+UQ*E_Ip#t8aUu1=FK^bu~I%#N~~cijEa?9gQbqCW^V6DqGPp
zV$MTla24^x2hK*!DdtW!eDNW?T|vz>g6H<${6?^(Lr53m_g|BNtwO{t(bg~dP5fR9
zE)u=#FF$6dgG`8+_y!ZYF*^f~Dl=$qQNy78o(arOi&-j|Kw>T-z?s>(BIg!skNtqv
zjGxr~3F|Md;WEtrA328@G+#6f6HvriX#HUNHe9*jLHHY-;q
zlrxjh7_ZM@;wOGT*}R39>!Xa}gTB7C8qCgT6EKZXF1tjF(G`yI?G2jA%P(ks#_V^g
zo(M%EE~d{NCV$h5U+?DMN*MlVD6e;CAHJOC2aqAZ8?%Gzb86x75yHOKu{%%a<54j;
zkxC9n@xbBnhvvSXl
z^Yn{BJS?N_2KvmH;}UarjXYjX0CZgvgFVLuoIjb!B*;tRM`kw4Gjkz#RFJzYKuMfHvzPQq^e+?I
z=`H9>5%_Glp%0r6lm95#Ia82#6X+5>tAz3rJ0u%I1X$Q@2s$XI^D~5Xh6O@$y
zPkagefuL8?e-gT+KO}nW8AHq!3G%dhsoy2`vFY@0{qxWMlGrcNCxJ}O|IiMJ9sfHy
zi5>q*j+qVnBz8*he?m!l$#_WWm)I*AHwl!`CFTAJ|C9cI%1P>x==~>Of=lZAPxC~=
zm&|ubJ0x~U>XDR})F+`!>iH*>l$YrLCtrg9bKE6#NqNaQ!71XOm}_7R(SOD7Z2QDG
zN#bqV35yUoB-FBjTP2^kYJ~bL3{wp1~J+m@DYj6zmuz;HGS0W4)8RM*SPuf)R0i
z9#PB<7Wi#!L&83aD2eoDvyKsSOIgRmFH-o2@|vYQebXbP*x3FRO4d-OWpV&Bx;db2MHz48=bIxgP
zam3$f7|-}7vsSTsORU-$zux}@JdlX%rM_T0!4u1snVnzG-DdLIPoglOB5u~PFnDAU
zS5P@ekMtVowKD$kbN5+1R5rYX>1$`{#^Ux)pIRT0{NELmnEjI`EoX83k<0Q}l89rK
z>bem>Jgu-N!N7KU2MIP6=Z+`-@36pBir2Gc^jLhnzdHQ80_YC+=A@C_ql1TSNiO$5
z=?KD$r^!wvXd2+ZPa5?5y`?I=c<KxUdV7J&CtiLi&n_(r1xqfmU)5*U4z?TMghGg%|aSt4b?`EUDg5(R0
zy1%4$86J?UYy{o#V7pp^5if6zX$EX=csr8n)8BD~^{-9S9_%z?u9MOhHZN@hUo*Uy
z>s~hApT3U70uynwTW#1muDt9mb~q7t;LaH~UsY->*!k*Mb9)?=McljxrympSeJz*i
zDO$3P&7<}oH!}XEUQyHjBJz{4&}WB~yRz
z`X+UqOq2Lcfnx9B#5o3+Hqm5RBQ
z++Q}{J0m8r^;GesnCZEuf11fnS(Ch<Oc>S)rD@wdGs2Zg*_
zhY}`tb0$x}Y|HQ4?Q=m!`4GV$o**+IZz%Yu3nn~2+lr?TAafyq1-(gf+ekg2ybMpD
zK8%N>u-b^ew~)uj5ej_ES1Ksx)?MXs&nJ9)qt^0x=~7<)f)kIQ3FZ4`OCUu`BJRR4
zzF%6ad4BCHUY>Se(u=`na2$~#e?Y)n1vrz21N7ShzL-$p5~xh>2YRZgI{|kJ{en$|
z;e7@8S&lEa`z;T}Lce(kex9;6R6d#=NcUl7Lcv`{Ws;Im^Y_}Ju`
z{92)1?t*+O4IA{ZVTPQ<{yHLqPOE_+1OEnDHISFoFTwv4WZ@L#{}W$AKg%jYepY~z
zdL{iO!F#g;@N0q`YYy~dlf*v}$BxB-lDJ$_{-03N@2nW)C3gH1O2&_M*Z&{~C&TLh
zPx>VGv1Rvf_DbX>{mrH!_$!5W((Xy^lC)ED{zCE&|KJ}5D6x|*56H9QKKO7Fth}rM
zCFdE*xKPwb`XzC>lCTch_VI7+m9$@C=PO}8g$VVZ66TY(P+p032;vqB3!&#DEBLSZ
z#=>gYXU7c2xLxJtD0@cjlC=M;P)>sxEavh@@%$Zv+$MoPURbY1YyxAQ_2BgtvIQyP
z3WfFUBJAVN-FZFFBgtSfry|7p>9oB9UvQo;zeAqqU%pN&6Zy3nJpIH3UT)=h9-q-D
zl*^`$5pmZh@%XlpJYRknkB?}fnE?N=j|>)bt!sGxtw^4qHIX_}#5p+f_3bs{>8{n(
zP!Z=O?6(D*dAsIaq7{R{mU5tND^Zyu3-Vnaz0px3ZhB}5mJIT2{T^N;
zJnm>MIuw3x_x&G)$G>c2{OKxEnCc?#PsKEBg<>wI?k2Vd5ofC*gBd60)DI2Al0$wk
zVG3pp_Q?}|nS}QnG6@}xeBahiXgKnBxA$rjJ|L$v0u12YMC%C3OSg?9dSmY@U&1F1
zkX=rAwam|X6#vTglGcz1-dcI)CGiJImtsH>pSSUpU$$;)4(UrC^`-+r9U27Y6uLde+PA}&@c-*gb@O4`=jmV>|Bs{R{96}GeX)us|@^3AEE~418SP3g*^u>zJA|tKI|O$
zX-$kK*C6JWwH6@6z`5-d
zS4#D}oSGW(4e;!fZA)lAY}nl^k>uKR8j&$U9JFq*56P9wY!0LI(Vf@1uSoA}MH4HM
z@1<0}hU8+sZ|Llo^bldkMz1w~p_ojI;JzlGb@FyLT#uD^2YpMwceDUyb
zDUHY6_MB}8$Qj+=piAe)z9%#6sNH3qul1+;zg&*6r+SoXda?Vfu67gIddadqz}COr
zmkeZFaBmcCWJ~Q+e8-=|XS&~F>mlMv7L#jI{4|>E%sMt5E=DkT=HpOS2Urk2hMhAn
z*3@ckt}U;ymi)8QgPr93ItuiFj1nVk5RnDxu>KV&G@wF^a-IJf`g{Zwc#s0E(y
z331dff0>78UtePM^Sb;w$ISxUF8Cjo^XCCI7MP2<6Mq&l`bPtF5BQFG6v?4~Z}8`$
zlXVm*p#O6_F#9`n_@$8h{l5-cU4sy&)z%kUVDDOuAkb0f;ycckZT!b+0Y5Ba&+5Gd@Nr4``3Aoi2-L}Ecm8^90S0aIQSBrbsz9fjPb8~
zD>l6Urk_?H)mJD$3C>W^Y1NS(;|0DkZ8N}Wl~K7^LV+(3a0Q_~NpE@jQGx!B4Fu-1
zfRCo-i*jONJ|MVZIFe)dhdhBGpDN79-fV%0IZ52FA;`ZIe(UFjAm@5%eeu
z^*k5i``5qtdTt5xxIl>C4R7=1kHql&JIny&G1x&#%ncRn>C6t~B2I4+FSmLwF~poq
zUmjP^;PnI|dL#PT-_$W8u5uiW4BiWLASxmL-7vg2p?A;G352?4&=)p
zqA(Zye%Ac{C--$b)2H>Er;F_Q__@OmnvlrTj?8BKprd0LJo1v?AKlt00fD@aH;>EZ
z(#plT;oVnOZWo&fIR7W|db-Z!^J?0qR*c_xe-eZ459x#{;--G!+tqP1!b{YfbIF>~
zJI--p`~6opYev@%BgKesZarXpuNQQ%5p%oCW-&iFy=(}>XUUypP*t-p8jkpAZyXH4
zeZAHJ=2yFyj6uhUxk;sc7@w0bKnIAq@1oBbal~6PM^TZOyZF1A=}FkC#pGJM?`7p0
z9JgR5ARikz5*HGPw|%v-F^agM32(4Mk)OVF4-FD=>h5bg6F#@K|15$Zo11h9=DR+A
zM*QgOo%9G_QJRuKc*!8E?i9~#im}6t5_1`2W0*g5R@KCb4|(jiW#fo{K0gr;Vvz6N
zscp>YcMX`nZPE9!Gl;pXR%dca-<&|zLlh^aUl_QX@QxexrxRaV`SVXIZy2*Of#Q(9
zAvedf_{QWMwg&h^F9*iI;8s#c_<)BdUKB^Za>?jMu=AQN*s2f*`FlPiI7i3DpW?d#
zu_4o`zN6dZ`{)CXlx@93^tamT`zXFFdeQcg;@M}NE6mJ*7waW!Q~d3gl>UbF_$JJA
zr1OLE#`M!vKGU-E>c_x;4|Z4jjyQAwYsE1X$K|K*Fe8|8Hd?(J{Dw8hdPM_{8o1*M
z#fRbjR_1*Ko^WsEHj4AQTrP2~2P~g{&ngyt-~A1~y8tWX?LUtJv>D}c>l65KbE{__
z20s18wQy6!!5^*66ILKD{}E_^x;yB%c9~2)0BC02vHc$KJwmwW*S_H1V(Zi$7V^k5
z`EawST^b#HF6sjQJn3<(74UCPME9@I^JdIQ6{Vjzr;OgLHw^V{dsH(<9`XOHwyb;p
zpx1|++bPJ1xf}B*eBP*leE5=AXG2jwA$DUAil47dU!`q^n3s7cAOdb+d~|!lA-6R&cm!F
zpo_kjIC`Po2G6uVYzP0*sMrxx!2hb2UT^~STeM#qP5oI=)G4$W{dBTlwRI@;%I+L!
zBMmuC|H`bwU${po$;_~azC!17X|I2XxMNZ6t$U)})JLCEH$bnp&8*+=0Y_b25poXY
z9Kzo?`PquOHy3SsukMa|zrFa`S5?df^dGx<82U|gY;VLB=+!&hZE!OB<)zi9X?Z9Y
zxPZ^kPqcWjlg^dR^vc)Czo5$e7sY$
z?nW5&-fk&sq56eSS$*u>G4SO>@4d+fef?=qQ##LG
zxSIQe?s;k>Ra|Jjy4e;Y5EXOtH%pBo={`ox7Y%P_m$h+K?_8HxUr<17M
z-R5YvPmVYE49|;qO?h-+pD7;NTos**=W=dCCv(Youp-{<=((Gl(LGj(f{bU6`Gt-VbH?
z>S}WAB5uzVG!xL#=oYh6-!g~c?S@ZaP~+$fCg0`#MusPSE<;F&d4H*n*u05)ZP;e_SYLyPV#Cootd8syBomt6wNGR{h3#HACWxb^-4>GrpT{YCm}+|
zyUvBHS%22WRk3^Hb@%dFxjYS!A;0rAs)JnMeje`KhG7BS!HHvZxoFDR;#`|V0jQXp
z9fM;I@X6O2R@@f3G3Dm!wNkU+wt+h
zsBi;Dzpi`$1o+hoTjqb23a&Bw!R!u9K75)H!}kgA8!Xm=4EYo?3-Ez5*wBCQIgba<
z;O(joo>1O0W=7
z4e@m;%me;Fjc;dfx_$-S9jl+{ODK&1ybDbi;7lCwG?76cPh{ZtFzSRG3iusDfm75+
zd`aFvUZA(fY9=~+egepb74n$OtX#Ix4wsue?jqolcb6H0-43UCzT!vT?q+r{5pg?7
zJ@g(C%3Zk3%NaxyLChVdVM2XlX?XyD{EElRe0kiN!c*|;h4Q!A0a3&a5#-yh^YUrE
z_;u3#7LPy2tvu=LF4%p#2aiw2ZcFqMeIED2DkfYpny<&Gm?kv*zb;QNn#iwD-#~tz
z^Z@!s@QJm}TxRCjyKD%gMCcy6W?Z`55D%oemr~+9vq3d
zJCPPjlpmZl%5E0r-wmS-Oo+cfU^PyxBJS$9HMJB6sW~2*NAb+j(6Q0vx4TsM>{AB5
zc2TDoil-x1#2o7geEOt*6W;?44lBR~iHNgH-eybwvazE{DCMCQlI{i=0ryUBJVo(h
z>z%RH6c0>2dtZy{uM7XKOL4`~38Ehlf%}%XcPT^sxAky)KWg{+;%i=%hkem@^W$Cc
zr&(v2=m4*@adn)EIPl81%o`ME%eOpm(+6CzE_Op3%A361Wjh}6@6>u{{b
zC}qUuO;a+~gP)$_STd^<=4*|$hSCk>V+^)S?W;yUX-swJWEJRJeWSAB0M@fcsF`TC
znCo@L)cq0aS+;v!8?~qF2N#nEsPBHC>dz^#qtkKEs2tce@1k+{-M}vtYkyvc`Zkz1
zW&c2Xc06!9R)u!AjJ}-Q9sILP91d)3Y2+$K8|Z2s#XZ$13)}mMv*V_vcT0x+Mz!M;
z?;&0{N^rDKmTTZ-mkv=#(Qf2M`z|$mSdVkx%wHW|Yr>z*ZEzn_hkT9NKF#9=b==dl
z&urd5uIC;G4!v8S0Q&5>D>DtnT)XK<<9yH`yDpw-><2qv7^IDeMm+`MyK88k>s2V?
zK^xZHrCEc-fHyAsw7&ot`LNfx2EdEu%JnpF-WfkWL-!y{Q(u(O{C{Y6@nX<7oDWSG
zE~mIZc3bf@x`#>VqchhN{FmcwU-kt)ecZfp(WN-owVj<5f${Kj&xq8-JbCx)=`>%
zyfm%1rQz*v(K?l1U8zdzW<-Bc0^y!f3w7u|=HALj)|BrL9z7$6-p#BN7@c2Xe^84?$9)|I|XLqQp8}RRa0vD3L
z6_-*v&^q5TtXH@e@Tg6z&DP;udf(RW58Yq246OZ5_gZgMS|YbVE?+t}P8)dDn&i$@
zzSE7aEMD%d5m-R`M!WlekpRbfJbRSwvr~P{42bS}KgE{l+&wv#Z{ecmGk7`7gUR>)
zDn{Z3_X(fFFA)qnQ9Xj>hb0A+Q~ASvR$L<3>&c|E-9Uf2bR@e+oY!;vM3T3-JxrbK
zDe(Nl?yv52ReNbpZEsTRW5X8OylPZzMN-)mV63#Q&9d;fSyg$S1Kl`N5|u
z<`y>+?*0%Q^Q`tE4vzlLhAWcRL~
zy(a4uJuD~`8GhVnE;D2KrmC1i1kiW~x=fGRVHb~M9r(FA6nG#Gq|cv#YYFde75ylK
ziF*JFfKY&*YpHc>n#lFT|mD
zdr&wJd0VOuxF2Q%K^e?$qJPup`AW!c5#2#}N8$3A$G`XGx6-i$_{D9!-0syp{j23`Mj!l+?_Ym-X_DL1(8ge*
z5ch?Wnb0#>$P42Vmf?42@b>)0=utUY>VCjb;e0YRipOIGdCO-!UPa+H9?2gdNmDs=zJ2>19~3j@$1E~2OmeaPU7op
zQ041aSiGCkde#lb-`d2-d6`3K0HN0{o2{#Y=U15i7h$1a*dT)yeGf@4TH4{yJCVNpCm6IL~^Is566tfcL%5FgE#o+
zhpN{JKV+7K1UJ45{FDSkkp~j{Xi)w}Cju2!A^46N@_+Ri9Zv
zat7hPti0{u<3~uJMn;eY$r)s9_aV45MQt3tGtiLNX6Nn)#deWwf4&yCoZ>ReO?%QP
z|CC;Iv*8rrwZb7v6!#vqcN|Lg=++GUMsnjLXZ3##d|{_BFOruvus|jN{_9G_I*KQ=
zQ{KsvAIcrl5~T+`R-+|KXI?2@q*~(H}kW;zR>LOr&&F2%YfNpX%U^LZpsxhRCCGg8Tf1a`iu6aK8
z>Pz6gvzLYqgx~&FqNr*BeA@5xPsRfF4bd8L3jQ8-t~{xOnYDO
zMSc9$oWRdX`Rq9gamsMr?l-92C7sMlJ3K@lwkkJ48}!^Yot*k1Uf5YOFnJW>R^3#Y
zUIi_Pi?WZ|zRAXYoBYJ)Lns$Nb#*?)w`11YxWq#~)+%LZf7I*nIA(M`V4ndWZu9}a
zT;4WlF7l@{CR`4(gFVOkZ+qPi@mPJ}k3(x#A-=QzdE`FYcizi-k4!D{^v}}We{{z8
z4u(rS`))!0-~Hv5w_h8$Ao;DoDsLf=d&{?TD&oWAU#9Bn=^(y7t~2JD2ga#ygX(5~
z^hb95quH9N5tc-D;l|51^F4XFi$+bi%+;rQ%Slh2J@nO&7zd=Ty@
zEKRmrT>Zi|oC^0Az4^jj7&lK>A)%4mdHjXrDCm`&y*OmQZ41{bn|>zx%ok3}|Mbe#
z-(NU2wI7?d*3@xBq&Kaf@wT3mzo4q;uoLl&PEz<&Q^fImyj@P9Kj-b8_QTY?j&q1L
z`+ELt9aj@8C%5B8BiF;a&^L4x&I_g`y2A$Je#UQMzl**#+*$35Prv5Ya0^*oV7?n_--6%S(`B~2@Haz#8jcCWtCoBnwJ>{I&Yez}O_EJdep
zfbQ~2ah?^~;h?-h@2xj~gm`KHz-UW7!WW#FKqer~f>PdH7AL+k;uC&;9Aa
zrJI2pco`p}cU?V~+x4hNJbpwmM)w=W=g3h7M|wZhxvKOC<i%fHwJGS}Bgdc(y|Tt-H2!>+f{~Jwf~6qwZK2d;av?Nc-&C{udsW
zVx6AUF^-P|t`(kFy
zX(Ct2&QqJu6&sUWM{`-E=8(78Fb|1j%)6Q}=Kp$~C~>2E!k#83NZ4b42};>Wat+qM
z?~&ZKie>f$&voe4h1%tD#Lt=fWs3SPmKOL-^YSH
zng73{&*$OYWgVHn?)D9zk)WLa`kqWq-@Yr8%MtPS!khj3GC7@gS*#!CEmvgzSF4va
z^W%$6zcBvfR8IysS&TSLcG}EchQuxWW5HwuVDJkUW-)u#1q37F!#mGg&)NI5s_fgy
zwBr7^s)gxw&Rxa&tz+93hQ}qqor9l$0LKRC={=H4%K|zTt{g==#?AH+utsVx67lF
zzei|kM-P)Q{}%%Q-ki$I?b$&AE&6j7Z)bHIC8pt50(t(`-@N>HAwGR($iF-K)sq5Z
zd~e7;>=bi4LVk7jRty8^Yod95mWOezL;7?1_*nT3Z&#Pee7yQwiHAWpYuGq;-ND~~
zxE?&o%8iO%!0ho3J;3I%%l92@Ud+CZ=?c9|78Nr%IAkg>w>};X!hGmImW_MTDtaJ|
z??(1w`Ji3O`GoM5qgmZPD1x)XN_-t06_VyTdU-ve`
zpUty?+YL-E+be?MAD_HKhhd$|+(kka`+)8WHowHP_ry=gASvmJnU@
z_ht=Ruai4nHX(fJDo+-74m$75;_9V~rt~4Y%hs=}>3)A$d&+2I{>@21N5fCpn~x;?
z{q7#}gkPU?{3FGAV|d1L^4_w+laofb(5O7kzq%usCbb
zyAaTiR8ICKf3sZY=y!Tou-0Py7K%fkI1JoQ{ju26H}Ed#tKO_wP45l*f64Aaexci&
zpna9V*KaX=LwSmr(!*!0;aA$_AKp!Q$NqD+4;cp7b7g)UVvVbrJq|{s_bA6JYn8UHyjL$NjgT@|(U*Q$%?8b~;+I5B5j>%uBTe@Gq4!
zJD+HWc{4=M*c|PvG#M49j=0%RM{lgknGkj7d^E0xlY6PS`+Qy8X**RWqn?EM7y&uiY@@j(2xW#46QYs{xH
z^*h!-$9!@O{rajF;~u0mapZ!h80YmbEfQewQSXiAfsiXwOg>NzeH%B{ZutxO+$tTD
zvFM+Ga98J~8hrPhm?x(XIqM6rH0q=>xn%*i@giI^w3?+Y{R-r(I;hWJS)
zR0@3ar4V{WQ0a
z|LHe>@hWOhV%?4xw0{N8Y1u;eG&;L`yHega^Lj`l?Z-bu*1^N!K6>#6mM`8n&fa%0
za7~5T)wEyTUw>yT<(J*w{~1bhu0!L-P(ArJUc0FNs}*C|J|9}{!~Dq?l>-PQ5VxPL
zWcO4Cp^K3)!G7J#iRCqm*7@zM+`TkkICA*wZ)Pl>EIr4+ch5N#&g7J{5gwx4$xjL_U;II5jV#G;
znybY0#XX8-a$|+>R{d4;n4S*fnwgw=r$i=KtqymHavmkF493-f4SeteK3-ZriO03Z
z@p;?GtYfU)!EHR7hW}E
zamAFId^~;i(q&f8YApZmGCpqvt8ba+9CjYaPJ6@n9(!@!ih4bJ&SmlmW@jeXDJH_y~^HgC7oSg?Fn!f0*)@uxT*
ze?c(oLO;eIdE_^X-_pW1;zc<8sG%(e0{g}B8|VnUW3tgeAcy@T;pQR2BfirYc+j(B
zKrqe6jj#8drhHVF8~gN#zT7_@H~a8Y@9kK8wpr0aitrWZH>{*QTt`dwJ|t)0q?keU
zBb^W90TJFKCN;DB<4=dTDQIaq{q-b^4LBnZ(IlT
zKDzc9t)o-3I7}$?_v)c|QH(sy!u|L_4EgP`SKMgdxpeH~C%R{PTHc0FmBrlcrbWjq
zfaCRMkTLL&p<5b;W1kj19o(Jv32zUJsdS#y9haCz`8&g)1Lj7+{c6&0eFB`jan~T)
zhZXc~y=Wi0*5l`6+K1*o(0f=6__N$KmGW-SMkG9-dW)8%-Rujz<#%-TV&IkC?5rrR
zakF%_kjACFGI;CT+>$Mjz94ax6p(#j}>AIpoM|1BQnzc27Q_*8`W
zSfM8V&UxfT4nMxGi+Vcry|VKl^tu;LlX(kyv!Tb{_JaSi`(=65z8n0^t0;EQJbnAa
zOGWUjau@G*i3C4d(cddX3GXotCwKD0z3tdk(PZj3n>q2DRpD2g>XmjD;Jsz7vHSMD
z35f4jZs=9v(a1e|Rr~I81>)h;`aMVE-gwFQd6sh!KiGek8r3Hk_iu7$HMz4}IOUf$
z>+ZmBCXbwX*l`%v@lrpp7qHvP+GpEJXZT^)>sc8pU%5x>OUE>fM85chZu11(4}XeR
z^%`2z$emR5nz2*d%w1?2(zp8hS5D7hX1L}^_|ck?N2|b3`}pC&mE}!bg-)Ay;^ePf
z@PtLCcQ$_E{1O-J*Hg!Ph<8UeC#E5P9jxx1aH)lp4b^nFO#8xRsV)1_&*>|-qW8Ez
z9S7HNTbtX=48GKI*HU{nH~((r*2q{__cF)3!VWo)4x{~}wF2&UZ>r{A)>fQ&U{T9C
zoIY?yze^*xv1!X+edM+D)@?rfaX8{eqpX#iT&uaDBH4#aq-(gnv3GU6?$&WP+n1>r
zRyT4tACF|;`vpeyKD|i4np-$PQ{m+JYOdbWyTay54QKWsF!-)cJvY+bW0|3I1Lv=F+1tp8PntnyYx!me@o`rNb^~wX{@nN2+Iu48
z;V!s-D_)BEV`gzQ;RNQ5m2$vwV~j)p&@$t_BjDfjK4`2(ygOFus+u?5-y6?z<`B<~
za9z2^5cB7SQ}fFR#BZIRS~reD{xmN;&x*xxyEfa7C
z^WOHT&Xk+LT@G{`R)T&WopWYa5bOxjG|Z&MOs;bYEir99lJ(G%lnKm5KwGlcjH%R_WYe$<&8%jmoodFX2r
zoeRC9yRrL^uJB1KLMJD%EYM9;b^M^(3GF
zaQ>zvfL0q#?@R}rvFS!w5#;ohHXNrow$)=?C)#H>Sbo7z9$-J|`SuS%WxwC-oc#2v
zeFN3E>0uy?pHxp*Fh8I@=|nKm2L)IvQ$5yaw=sWqdD9{m=ZIfS#&6!?-Hn^u4wCEF
zWjDM3@ffg?`HMXJM=ZY=dT)z8#ie=v4cBP@Z&`e6CF$F%Joq7%OPTq~kD!fffBY0Z
zzHb`pyOQKz1}tQ8;FZq99Z6nhf&U4rufouX`Pt7FQ3xP#@6h8D^V23OaO1$er}FX3
zxYMT0pZ5r&mvP8HubIuq36EhW@PawE3~rY3Q6hZt70=hOaxv90jNg7mEQ52W
z-ei6_*h7i=$BrN8v3ke7kYaF*)E;KfALCK%{nMEf*37R*e63-AU3Tgn)*q^S6&c+(
ziH~3W6fxXrmz}&5t7popMJ%3pyv>-|G5ug93}w#|G$xZg8p
z{P&v>)}TENhRw(0ca&H^M-1OJn&`TXYhKel^1MHd%}e6t-2!-hyArGC71S-aVw^I
zQTfWqFMkQ%KiY2pJJ2^BH+n+x!;Rn08MIGytTSi7<8&h4T#x!8E@${Dy5HPayyVt6
z$Qkv2HIwB0n+h>eai8#gSvd8BXI^G|ioaIvYi~0L^!FvriQj-{wZ*Y_@YIQxV_b)bGE{l<%jbCgS$C7%R;W(
z0K+|(;7`t9-5(zVIX~CDutf0F_qnuDUVZobG@HGs_tn^vaogah*2QMd+YEZwP_vsm
zAa_K4cO=QB%vTZJft-_e>5Wpn(=z|az061b^L7OM+6?`P0nvxMb;R#l|FCJUO+{QA
zxzE9=w2_+}f8(CR7wl`!@1ER*zd5V7Rk|GcoR&*@&wSz$-@V-Bql^4cJ9F`}P~30X
z%S8pzKJ%j7NZg2cN$vcJRI5b9G08)vCXPis+7ms^8kRN#w7WC%rA)L2IYzTPd3H{{<
zX7)K*#~BxYKXxy^k+bvad(E~N;>hCH1NC9=GHt_kQ_$Y=H$Qze>r={p01*rAeM%axzVyRg8Zf&1cWky49(wYjqJM2s5l
zlQ!&5bVFR((f{#s^~74vs{D@Cr@Qssq_Vv~{1ML?eN<6zkNED+-)rjhorkAS()HoV
zwVaK)>8j(`>$%`{*CI9}zjbWGnTr01Hipm(Y
z`$WRxw&&n~cspYN#qT;{x!T7|5XWZA>i5O?g#~V2?~8ssuHZSx3vwZ=XUtfJald-=
zocrY>e1{e25?$W{yN;zsRYA`0==x!ekh|aXGJ^7gP91KaTn|58Hur>ACh}yPv>$JZ
zf}G4rvm2G*@3wu~emcoVYMc(K#JYVt>ZLK-@74R^sLhbG9{0N10QJQNo-eI|JuW4W
z=I(%@JTD)Rk-iK|1=2u;#1e{bZ)qxb|-fc;JI6sjTMl)
zud7)@`$N5G75jaN2QNmguY(=>tEJqRKra4#hf-S4wkQ80u?_z@->{bQL;KE5mi+`f
z`<&f+2vx|H=mq5
z$nxD!x7lTo+|4al7m(cT$mGYQH?rlZH0@Ip-+S+;eJJ=_s~YV~wU5T)0FHey^3qd!
zj~Anph66S7xoO4kXdlwvwtW!EjZJOG-owq`nKFvqhGAT``uwH7))?==lxKw$^A^SJNIJIU6Q+>KM@WN-*2UEs3!dT@j-0=pFO4hNs>F;B?$=x?4S1i
z+5Or`_c!dEexcU_77yrbFo`F*nDP-RB=>XYYg}02`<^OSmiGzn{+9WDhiCjdw|x%$
zyQ2G};2t5j_8$Lx0io~Jn7^Dk={ECcRv-C$ndFBs3wph(1DXG=s9M3|i`lp6p%v~e
z3wS^NIDo%Lt5mCD{`}mlX4bCGWs0o)C};j2E%rU{&pkrAFnuaL*08)#<;o}~_t+87
z_@H-IwmGv;|2sYG$M?@gPOQE1)>E0k9DeB<^Otk4Uu6B|GdXfBwI{Qyb$6O?qt{xn
zJcGP%HOp6YyWqm=@o<=ENpc^O3R>t~zOP2cknnRCIhT|KlUf$>NJgxE;?+5)-P}o5?
z$n{*XSdQ`+Za>vG-;~kyD77bQ?9?Z_Aos+zQxTmz%+$K7J%gM{ut(fl)a(ClaIGWgBi_%+qH&a2
zuxmy#;Dr_0-HRcYzT9A(1oZx5PEjwFv
z4t{)_QI{UU@MmEmKi8?_U6GA=$eCNNP29Qhx3_nLKR3;?UHcCHAk(aebq3CnTb#}{
zTi`wUFaMC^Umjtf-5VNI2EWl^&Yks%@DoSk^&WG0m#%uT_uQQa@w=;&jl(o?Ut%gZ
zu(-qCS}r{GREycwdTzAQ^X#|qm%CmCn74z!v~RsOO%DDb`R&V^EAF-2es`DJ-GTMo
zj;-NN?f2o{spZ4z$v9URN9nv7fp+zZZ2PdaxrQ4&!*->TRXw-fQ}0I#`eWe9r)GWd
z9x5~0p_>8p?VPycmttiNcQ|g?YWtjeuIK&0ZPD-W9#3UlSeiP1uVL}wC{w(H@m0L}
ztGIV9SD0Cwao@6m%iq|kO&jsc?@@oH7o(qM+^SeyjBz`6blH`4@R!;NeNI@T-BJF}
z`^2Dsx{d7|Ziw+KlsQ|}h<>o!?P^Z+PA47k5s=#^MNBqpb~F$^XqyD|$osMYsCx
zv84TFs&YAA45ObC9(TQG9a^u4cmXqA)
z16$A1J<@b-Pj+70x1#~SeJJAoZkmJ_%7v;`j~R&FQw-@uy>tG8q;=?oJt3~
zE~IDJQ8#vel#c4nN5qX?8wr>e%}|;RcU1nBP*8E5k3CqJMtu
zXZe$NW%PqF_`X7=ocaAePPdrc_Y^A+va@^93_L)AKW~<0am4(PzAUcryK2Dv&hpXG
z%uiZ|!M#E5WZrP*FOQt=&HT=F+cQkx&IrGM
zdhJ~+E5GF(uYc(5QijV#>M>})Sc&<|xgY5VcJL13hdn!2Kfhzi&eIn!r`b@yB~GAU
zGQoLpg(GXfnZgX_|98w4F@OHxUa#fEAJDXc#U-i{8cbflrjqr?w()1#eM)@2x*5q`
z+tHKN|6}P=X1_`_e~z#8pecvk6!!*q#~~pj=KeO@940t9R#~5ZH{pYCYdPiVoHzAGMaU0WI<6vkDstj0n#XGD
z`&{XL#hG1uW|aZ9*Tv6%0nWNK;3>fw-meTupH7@?XM$3{Iz1%+zR&{UGGNHBb!;BJ
zb2`oViCu=)k^X~k!bgyN=l8N0P~7{9RN1{tpZPL2gx^Z5tfc;}eBD@2`VLB+H79?m
znQ1K|yV{*MT0-}^qhi14nPY!Tn0_*LGxodEpY8qVd~kPuw+BXmT?Wb-FMwRxpkH|!
z@Eb`3ue!A1TzGf%ue~Jq-QDmW?I#Bl-+w23=4Fq++kp@882w-d_Nnr3>FoEMueKfZ
zB)bN$S-+Of1K~ag=g~Vq@1`!|G~k;I$3>DKfAqo6ne-g}ZoiNAv!2JK-wpwO#(SjH
zA;2T?lV-2@g8f=+!)W90_+2t7vsWfKcRwtCw%~LazAv-7Gwu}KBaDvmNrC)`6=t@-
zaNezQE4%Rken|Vo85L#R=O3!tcR`fbh<9PqdsE;aLOcGrIT`)1ZFuRs-q35bc+BHM
zocpC$Tr1sdh5MqY?oJu7$1d;G99h88(F^0~{qxyv{TvVAd>GPbRdf;hC%kdg--&vC
zqf{Qu!hY`O)UaVR^lmje{&^qdE~`y2*!~>nWUU`o4vi=uxW({!UNaY8x^8yX^%m|^
zmzIRjpIW#Xav8rhty;Ly{hwW@esAR74{5*9cO&u@DNPlBYg@Qr4evz*kG60-6XIh}
z+qZB5#Wywwqn&a#-+xc(i+AJoBd2zdY2r?;8apX&)EBP!_ixoxDb1XN+Q7>G$OHHf
zt{+gWi}&80R}6UO55G2X!{4rZnz>!i7j%*?ZQ*XkXerph?ikktY0*aXeq3cyLOT3t
zy2jAMj!j%d2l3}g5uYI#p
zE~({Irkql=@~-6)^0YeFI@EH@51f9WU0uWN3=Z|_mQ%}Zw_4!7C995`ZPV$}<1@9K
zdaK>i(bH=2z1A6(mN&IrhYbmrPmXNhdhQ-DzxsG1=eF4T!q;wfoVs`0)DM-l+~hRd
z&ObF9xZ!#lGWT#lWU*=MUK6~7-h6CvheD@1Zi2?2`xYDOFwaXAN6v2K9(pEnHcs^W%zGitfENkOkV^={zgzDcb}K>IJ{1-HiF9nR0`5xYL&{fB97&eDF<
z>bTOy1~Ri;8@Z0vhGW}$Ag>Ykt1KG%gtRNE_Xk5R@r%OM6%QM^*_Ky-bwb={p}X{y
zG~&9=$2DC~BJPnY3hsUs>!{z?+}@9IpE4|a$Jh3_2fMd0=yMOmQynuR{&3)H>1!$v
zh8>kXgC_?Y;{SVQyZglg)U)-h+H+~#Uz|8@pn4l}>wH!pXvKW9ZSgZ|c!A$DKX}r`
z8uQrY#1Dn5C|^7#==C}9cZg<{(|at>8B01mKz!Q2%c(1rCyVlEUu6%d&_iyn0{9<~
zfBQmlL+wo;yJv{werg4pJ_J8!hvJ3b;9F1l-L4P#eqBF>=mNG?jPCm!aR0^nB`Va0=^2@p9$TO57hNYo$H!0l6xtUB$F7Z7JHZBp>`k`(9=~
z1iZQI&?b`aS+80-AN#_&^48u7Ci?z~9*_S*%E
z&XC+Stuw!<{X4ECJ|q6@pq=5&KcvLop#8Aub6YIoZrO{QNWNf4f3`n<&v-w8_=6sM
zv-7FN^H=($H+f}sSJI==^?$MV=21C!-~Vu>L84UVkfBftMMO9vO)@qR6`~BGGA5xr
zqGSqX$dnW^D-;<*8c4V@M95fO4OCY}rsuWK`+cqNXWi@f`RBKu=lSc~TKE03&)(;p
zefF958Q$kw6Ge2hZ~FQqH(u!nU+11sXvgRMIo{{3DZQk(fBF7`d*~zX-`qTWobSgw
zOV)9}W6o-`jwEM#&@h(LE9fQV@n?>`&gYe8uK7IOCM;ElkJ3k
zkhU7meBS&y$Y(sI6a09QI_dT6W*DD$wWmnnFkoD&n9Ti`jUB#ny#G}A{x83*AD
zJTCX_>s(H2(laBFH{ROB!v~1(B9`8Sd7$61o#%NziC7hroHBL+2&cv%82p&6C?@DB
zP7gWl0BcbA6x0%h|3lY_L7&3c@kCEBf}00Sh4(|@E9iPMa4D=u;p!0rt*0XB^9$Y&
z@%b|Z;NL10==I|S`3*C$4Fmq|PxPPx&ilK)=jnV{YRUZGL1o<2rT%QH~!_1%W|Aa07T1zMYZ+|em?7TN4l@CX@b6OvO!S}v;d`kdGAz9h;KlLea)&z<
zZd$Ws7=@>0JM^OV<~(I@9_@cPZ7}5Rt3JLv*CW-*nEQdB>TGy9y=yMxpb&m_oi9)C
zSHwxq&&Zp{+uhdRwtOE@kT{l=YFbk+x2`O@629B
z&cD`&3V{0#@%o(3o}9|t^|s-Ix!*r#`{yFkGv&q^C)y{8X+CfgotyUAzxPmo&_xzs
zulxb3*go?kf9&bj-u%AWr@8wslHQ{VC%z^B$LzYR%nowKg;wh+KZE!u$z+E%Nt59C
zVm|0-w}$-D52t>4y#fZA{G3DjAK1774IcY7(8h3@#*UcN+gX9-=e*G-~Xryj^U_R!HPP6k0OfY}A$DVzA5q_7>
zY(HDivlxH%4NS>@vy5|%qHd)}lN}G8Y_LyZ8R4Im
zg?TyUXH>j(^BVSe|A(0-4KmiZ>&;RL=I5g`-yOPxc{y$KnUg=^#~GElM83g1TXF4q
z`jjCxEVJJS#dgSN(3O6jeUX0FvNGMLn7{ODM=#LEJo|o;>RIA!)VP
zul(*b&{R@jOxxwGMxoPv9CviAGdwC<~I=-jcRCTYV%b*&z(q>Gw
z-q}jltirkBwO=I*|5<$dh;}9JNffV4K2^aA9_XxbSy;gummmIU`Kpqgdc3AUavXfC
zw6S+oE16}7b2;YteyzhOoxE#h70gHP`r@gMRV-0$bk4Sp*gMeId6%SJ#RBKHkMF5p
z&CaemJmqdlH9PvczUTZg)$EC5;NohdYL;Yh#%*3F)X(v~@47+n?oku3CiKAla;uwb
z`W03)4=ZmE#dptxEK*TXw*vClN1WO|C%b}On-*_w-M)%>H|^V$wFl!^@fh78d`H;-
zyT#(psOPlU%=b%ho@kr5*w7~%=OKp;^Y$X2#F-vS2Jn-5E%>W-5bLrxJH9VjfPKLF
z)vxV(f&U;)@m(R}U$A^qN&V)~OruTI?=D}iGEW9S=T+j=67<*I(Kj`Mu`X~~@%)h~
z`0cwodyj(r=Ft~_A3**5nKLX`iv7I#sq@O5(GFk5_w2YA{-ph$u{*b7Uvp3YNVU^g
zFZC=Q;o1xRd3SWrJR9`aVY5!!(R!#dbdI?u`sJDd9lvjZzQe2xf=Ca$4c9&nL_Q<^
zJ1-amIc;mEX9@bZZNQ;v;P?4CIl%_uZ;ZCsoC2n_`ren;o1PteKBoIAKDlQ9qyqc6
z4)CY_wrh#A-H5;V)pI+#Z`bK`K_i_9Z~hda;14^v?A+@>`(n>)>>_AiELAu0?@90@
z-?h%Bb!63NNFl#q}P>^UpD)^$Lu2TpN;%5f&596j%RFU0UPGd`k@PS
zPFNU8`)1xjxqg(d_pUvBAImN8~yAi`BJxqD$jyJ5Q@kiyBJj#aA0f`R{pb+(h9n{SM;8
z8SK-k-Qx4=-)&bh0Yk4H=4S}2nm&!D@xSi_O>2t(CjUI&|B@!qi?2AhSeU{0aTYZH
z8ASQ-`&>Jb}eBQq3{F~30V;&yC
z#unDs(>lU|!M>V(FdRAT-wmD0=jYQ0`g4D;^pvpAgO5<26aUScRPMJ{l{Vw^?}x5>
z9KRoY%IEn37w>aD?j{QRR}*x|VZ-^(f<0WX-GvHV-o8aDPe1Fe5%+U9nwE3^V;2X`
zf8BK%$C>S1`2C~?$6Y*q!<)jrl|}V&oImiwG9K@aY6bVV)^?!>p|QTWY0ddnDQ2AS
z)fL4?_=2uDrUc$iP2~L1--P|MX|aMIrLQFTU5-|_x!%d*`P@fKxMzqLqSV0SY2dVo
z=o3ScEim)1aIdJjw!n7}KsE^fF$L8Kd_!&}>`+JQgT8fJpdaoK=oIYcQutpYf!sZX
zyr)QKkh=#u1E;RUHx$EXKuZdrO4%X&3}p}c{2BrIvk0Isit@AzNc=~n7UWNmy9lfx
z0Q!jOnTY8rqjpH1zQQCNKT0UqT@_-$&kGXlqR>-V2lVc1#KX-*dt1=+w~)&b{r%}j
z1bdf#4ddanx&gpng4;Bdf5Bx^0P|rd!5(WrYxD34MgspQRXg|-v3p7K^ROBux}KRGSX@lis3TsNkSF~8?%pg_3)a()=!$IK0T$J_UzAMbg))JYP~
zRS!O)g$3>f)NkhTM(p{(>CoPbkrUQeYlZYrMZM+xoAz*rNk0D@3Iu;;MEFm_gC2F9
zA9|G*9`FbU8;3PGJv3I>XRa*D=d|UBR$T8x
zuV39FIiFS$6{p4$iodbt
zlKq6X{RR8}+VG3p;Yi2LJYG!D5Z+(aC)>Rtx#clh*>v7|S?3dfKev@}*EughFZ|{8
z$QAjWTfcAnBVhLnas2lrvTZvm(>d&Iv!G{m-ulvd_$u139KA2)2c6sG4{?hjJ=wA-
zyex=)0OP7vG;V5jZskbwJyz=6d^N^?Q|QUq$IjR{Cr3Kl*j{jvS;6^RQ_g4Q7dR)DC$Vg2(nAT_U9OCFfJC`DA2w59qn5mS_u8^
zx>>Uf_)|xF74;g1^SbLf57R4CaX#jCLu-T#e&*@{_i`Yo7unT#7W|I8##(Vl;rDE|
z3QIjvgmqEb*m2}%tQnoBHv;Joy%v)(4)g!&L1s^1!>>u-tU2yH{PEC|b<+nReYdwi
z3hpBRYfHLs@I$<|H+uA-eJ8cz56yz`0t+qw{{Ru}@KO!TWYN^cr;SSjuGV
z6JM`uIU4JYS(C=x$-=(u%=cTRIsLH?YwgzOd3Thvw6W$d{Kef{)aHGM|1~vAA@e%!
zW7R5!AJE6Sy5-vmYe%DgJKATp)q$TUUD-4%3G1FN!>1e@gmdmUJ-eq2gWfCFp51m5
ze%&6S}Ac?D1aNQWg51y_KX{g7uI|;8rQ!<8y2EQU5da
z9=%*^;Bv(O_UwRKE94jY;!-7j_jk83zkVO|ec_nOPUD_>bWvJI^7Axf2aIJOiQo@9{JQoY=#|AQyFLewNXrUt32Zjluw5_ky|2vQMdR)3
zwoj(genX3|No$CHKHg+3?{FZEZ!O&2=TUgdCG+kyPW{=h
znNIW5v_s20i65fKmXn`+{#$}GjjPQA4&mW@jMuV-E@ZDW^Ui2fxYvqT8H9)0UumNB
z<_9FUq5F2{muKbCJ(U%X2|a0CH~M8Zi{xIOn{}M*_IBpQF!JjwVt1R9eQk}-;+G9j
zzMfzB_7W%!u&G|p=eYu}*reG
z9Q`yqD(Iad>?40VhG`o4wDCZ-Aw55wf)ah;i{M`d72w(dXsaQDU;5%0NYD-k1b(ks
zOF3O#A;|4JC+IcjlaP)dP6tSS$xd8L00xNfQ9qAHcF1qXI)NYGDEO%=iL|kd^%%aw
zAnf8V@Mm}m@UTZaipZ(XV%)We>tX5CAzFe4fmI;0sZnFAE??-an;-#Vn84MWpLQ
z>AoV;twiF}YKr*e^%C7tq(em7j@)MO>DR9admR+y9C7=G=(pd5
zdTB@7;fTMTD~R!y%nrJO+<3@6$Pnbe5ea$#RS)>$Kpzz4cZl-Y
zyg*p*j}+1wFWR?{C`Vp9rQd;?7vej;7VvAd&@L~<2<
z9|(_MBD9|&fkL~Sv5G7PzpKXrjx%h9beGaH8}cUlLVNg;DA>bIL&&d>gTU`ON|0CC
zEzsMqqnlECkA!?2sXK#C6x-ufe}OK^6y)RQVA~q}xY5NtzrHiwIbM9?iUcq&zEb7m
z+T5`n-x9s~&T1~7X}Xxp?cEjy1<}9P&PIVS-*uj-Md+_3ti!w4Kj-qnM|yMlY)`o5
zkk^`bf{!QTvIKowc~XPNx#$A8<>2?cC!`;zBkVUkHW9{)V~Rq5@>3D$ib;>Td}GJf
zXb2b&Ruv-SZ)|<&#eI}cf^Bb4qGjItTC^@|)8xqSQ}4|bzGG>*Dᒥ|pGNOE(J
z+|eg`pOx=dn$I?c+~s=4E^FZH9OaF!e4pPuVmz03|0cY%7c`bC0PECe^neZehx8^Y
z4Cf+KZovSUADjkVB4pu*x!lw-%edZ8?Cx<|GAD)0&+5@4hUDHDHS?nPk4rY(+(hev
zcX5Ze)4X}w-@nWZXlLB=5uJ~mEP0ke=O5p@p5*t)*5Av{p?NT4%zFIr3-;~eTzmBe
zzOYNgzzzGaPne5M=dX7%4NkmLP=KWD8!BD`Zr
z;zEj7G((e@N2WSx>p{@V%!Ur5^7Jq*$3=AfKE}cuo^-zQe#G_xRNl{lPmhp)c_;bk
z*{<*tdpc~{_6X+_qi*Ic_yoVK>`zo5NATMgBuy9#95-U(r+n!7e5&i_X5hEp75f0Z!^p%1N`wR88PD^-%i*68?B2x&m|Plx+-C)LOQJvulvs#Kz{5f*R4s(
z(1(5S?0pMpF(;LuSFew+Q>XQ;$rRN`bRJa@(BeM%OO2;y^`iaJmT6b}Yy{4|mUs0i
zaAVNY&a@uOe4Ljx400|(tNKg;S_Ui%r*&eVqMZ5TpjYm(iyjAoBMcWxZvyjI4wP*L
z|M1iz&Bc)Kryb--@4nq^osvMmH`86McNXcnkEO8e2dekJtGXNze4J1%1e$FF3j4
z8{H>cGT*8T@x!yixF33U#M4crui5umeE&Z#&0+}o?FTk=GokTe!^G_E)UVq)Y>B0E
zUVnD)J=NREJ{EYu6#GF*miPby>%#1x-lRvTYk@X|dMkBTklefe?dOr)Pg`M}c<|*E
zc2dzVdzg+SzTs$G8v#Eh&7_#pS)&%{LGk0d>0^RJJ$BnyMd7AdCg(`bVNhr8x3@gt
z1{a0iJO798PvzdjunxN4Uwrn0c#fAH_;~we>{dP>%x>8NQ#Z~BU(f@uSf`i%;_*s7
zFbsg6cm4||9-JR!s&hYotzRR5AH}+T0$dxk<3sj*pJ+mtKuqkoH%
zsg3aaTS}8|@cBXe+9^K2&B_Yle*4alL)_2*z22YmwSHh(f^?2d#G
zxd`V6&iM5pqMQ4|4B$VvhuWZX#|Zf@+=uBE;a?As2a5YYD}{O8v_hD#AFUGno&lQOBU`}z;<*DiFZi!08{~JB066T8
zfCW@+@HdO$2Sj{F!3gKYhCI&(Gu0+iQ`V1BAq?s;+Q|qU2EY7){
zsYrJaY5ub&h%aA9@nj&+mn{e%Aj(&Y^}?532$$POE{{Qur@x060KZcw(sJY^;_>E%
z^c%!<_bn0PSx%z>>~q#o!EmoiP3S-UhLQ#0x9|*czDmmn$Z3h|&Q;=ir(KrNPtJ;Z
zritw-UPX}C!>CH>v|lgaUv5ymLnF%H7Ue6&bzP;n-q|hkkBag)M0%Roe|8!O>Dz<|
z{UhI+5`zCdQW#gxoF2z%U$uO$cU_U7XW3?IIoJ;q@8ee-ddum?2p5iL^#S0wNfgq5
zbdxNJ`!`xtAe=WGq>h7qD6#yzCkpb7`vv)BIzsvxiv)en^NPa#KT-cgS~elQm-MU<
z($OE^$my-IL7bjD_%%n}H+T6sz3op|zAvLxEX9Ha^ZD0gLEc~$mn-an&o>dSc-fw>
z19JPc;qy)Z7lIyLmSC9zzR7ADpwOR|cjofOCt@Igdk+@ExSTjbDA(_HFS-2R9>V-_
zX~|U{uLXALNM6@hu!q8G0o{{s@$jfi#V8Q=XA?GHM+Ez+BO**m{*ukdF@)DQ*EEnH
zZ8ZIM1o;ySDu-hLhyU49MMC_B3UfZ6W=sxvL-V)n`!s&8vbxrV&#&W4%`w5?+-X-8
z*Jt?gP2Gq-5p;#`*Tt;STnvLE{@5Yhex9-Iu(N^l=ifG*e|a`Nc#3ya?dU;J
zoEOx;=i&N|+xYry<}hut9};yZ@w}pz=9K)Ob_++kfkHA%9!hJ&3O_V{-RmqQ`#jle3sP@IH%`
z`Y)n?939e+{GK;Q7ID2=d4FC`@#}o=@%=-eS*LA@R=*dhLiy~wl;}~n_Qg=KO^8e9&mme^284m;67Jk7q(C|2-eC
zcfPlYc2H{Zk}!W*z&z5=&^JAV&LiupyjoK`xU{iPZ|}OcQ~Lf%H@*@!
z>ky_(xOqf(zHZ!pZT-dq#QW2~cpdHYHmEqK=zwqd*>a{SFnfyoVe;?eD{kdeyyR}n
zu2H&L(#C4a_ruU3I;8j1Wl8EpSGy*lVPoBGb!9eT>(qCbNS|4y-@J&QU8{qaZqZ&f
zXTUSS{IRiSAeAFZV-$}UwlF1+>fvm+L=DQ%*Zt@iqI(?~x{lKO^P<9n;(6q`-eKYpS#_SmwR>5f+V8vKFV51n3r>O5b2q|
zV^Ar@@1+?TPvJe^{>2s>)}QPfMl{@C?jie3|0K=zF>p-p_vo?}u|_k1#;vT>Z;&Z1Cc{
zxnKvLkJFXSm_TqYo0!AXd+mRh?^j5&ygC1$ZgVjZVx5=JlIL@@JDx`b?NhuJ10vRi
zH?ML2r@TBqPCYlY;^XZk9h%^9-e8Mv3G807kdHfeibDDLImvfE4{xyu<^+FGr>Q)A
zvG+4Ro)vZz=HoYOkS+Mv>uCbVe!@UuT%TKbny)YRwi5jL$dScdu4S0;z4^(bR-E2d
zAiS58tiGMo&E}Nybe8MW!~{Pvq!SPSt{ukFDEXa(u9_LuLlH{~+)iy9ji(_`A-d-wN~^KY@QP0OF8$
zG7#u78bbPN;(qA5Ab}s!Ti}l;_aE{vxdRCQbe)i1S26x;+)^U`sTTq!rwD$DZGkXP
z-W2_X7Wo3-y;#5r_&SW@xllC#(`Xt-xvE6JeFvEv^fIbG;4RFOgw%RT9)l#OuaP>S
z!ygDZlRK<(Ou$BPZQ>mscz#-An9REv}jIv4)_XMKRdgNK`oh)JCDxj7)ugMphw
zJ{H}a&*4APk^Al3JP7|cax)^FoBu!f^7Q2>PggE4r{(v4o`OE7`j`Dn|
zSN&%_%GU*kTmhWdi}fv+`?p@axxjAyM0vSA{w*h`Km98|dHBDPS3lDIZ~4m0A+L`c
zqMrCzfS1Pyqan3}fnNmrkl4=U`aKl&u;3Zv{E{m
z#r)*$N8X;~@*6~XBQf2#B2B9a(o23W)>WkG6qE8z%oQ+sh#;qJBG7jDwO@*7*$mSG
z95nQs-A(&VMY90iaj#SOg<{ppSsD{9+--xhNV?Fn%?V<*`4W
zDx}wPqktn;2=Z0qY2k_YW@ZWcR9zGJ>-7b`KW~^=SI~5dbX0?BVTgMIYyY6BVO&=@
z2+P0@Mfe$IVCCJvsCsZuwk0{2t4$4Ll7yG~w6n7_A->sg|lix!6cn39e>UgAsofC&C;Nkl8coENn7
zfc*faZKwd8W63Uad%i1#`w6;g@eec<>^pC?<#M}2$8kGnJ}^duz`2HEFZ5`<`_lV1
zUr*J!&rT;=>(Y8x@{ikGTZ$bze2-_|hQA|QdgwO)K2Cl5qm?A56P&z-@SsH%Y8dZf
zx7x?kKk-(NV`5w`8Ya#=E+}9~#C@7^b-X^G4?p*k=;m=@7|P+N*bDbJRqS!R0r}&J
zXZSkt*x9XoJU6;*p7fYo@p{hJlvnEe2pm)>#w1Xe0_XCp(EGxYLW2!vi-&_;QA&dzb>Nqt8|<&
z)Zu#orLJ7hqX~yF65u}jqf;0;Fpexg!pnPCNgW*y-@#q|$$yVE`p7ywkc@MbeX|de
z{5Wa<7lcEP7_TFo&~e2HLjV2_aM<7vZ|dKfaL;lquYe6Umv}oJKS2pS68Cv>YPj62
z85+F)t@Eew_7mnj
zgK=c25dN)?fad3f{FZbVXp4!0p8KenNT=-$f&Yt!VbGVvd3UmkknX7I0>3*QAAo;a
z^f!#q3n@R@I6<$dd4hbF=(k^_VH*6o;=E~kPT=$93(~2fY5;acZIT@Q8Vk`|u!1NEa4I(Nn8#)my3LE4$xd(#CHaDeDC;fsnR`!Hwfi_HJ$Im|n#)J*gU+6^@dJL(q;mU_fKPaDH9pd8Ki&l|
z?q0lmHRSFq=Sn)-e_b%C^81gGN9Q4@YT*=%XdFd&vsbW8NL}_QYfFEZ4eIlCR;=cX~FjI~fsu$f+8%J%7peS12t
zg00`xEX(G#lyzTrYfGwkIkUN|C(WNy#>V$)W*6V9j9Isc9=+~KCENDYHSEos3by5h
z>wzjiDf5`2|EEn;8Ebr8Y-TjEjJZ!NJ@nww7k1>nWq83${NBprbKSpOkuvudRmJnQ
z%GtA4e;S54lrgnGFMMqd6|?52PBd?Rt&}mZKaOjQ%h|Ylb3YE9UB=XHY*f7bN?Fk1
zSE}7>OIf3bO7=4AGG<$~BIN>p-_K@Q%|EeD6>RbMidfm)O18Ui`=rGhW$aQHrM8RI
z%h5s>u
zGS+Hukii(}7rmi);F_dTW;kX?*TTS3HdQObZjM1I(|@l0(y3)Bvzx87zI16Ni`U*W
zXjx?iD_wfA^{kgtyenln;s|~R@%LFX(>3^gns*gb?_a1X$323*gX<5MvwKcgFRK17
zV-Y&Ksm2S+nD3PB!CfrN*vM@IcI}rS-Fj)!d*I&=F}^t!rEK=xEq+(qmN91s?UwF%
zk3IE)YLMfJGIq}`qVdh%GB(zGOt0b%W$gZP_4@aLWvoqU%ZA}TW$d3rW23tK_`+OL
ze$Jd$@P+w$uv7cgN}0*lGS59@N?G|erLc)!rL0XKm6$N|a{L}pUfFZueDGSQBre$VjvHmZ#Moz}-v8}U`YX81-;EoA|kFD%}zDP80*8<+5UEAxOqaQwESNoqFY0;^ibaue{9GoWl?o4+RVx+W$jO0_M4tp%B16uUf6~9>A7LhsP*}!tROFO
z!l-W*Oiv{%SbJ#&Q^_=1vgntT?bW!rqP#}RMz?G!nbRO;`^w!;6yGmrT{X==x+Ik2
zd-H+edos$|@JQv>2O2Rh%sj2wwUv~aB=6dqf_B!?V%*Q}sNbKra+>Q+C}(<`az2&b
zENA)s4)o5fEN9J*f6vbhFK1TQb|s8SDrau>4?hk%hfCPHFKh=Axy>A*jNSRIwTQxnjoLLl|i*1Ga%AHf+Kjc6;t39&7dDzNw
zW?W@fc6@6&yPmxMvz4!uX)S5fK@uuu>oWaQx5i7Ef1+OPLd+9R&nD+BHL75N4~(SE
zJf(O~>!SAKXeoR4uubUnd@1YPp?F9u)aQXc+C7+F1#8jDEoS(1DLXPxGtYIM6u)y{
zkY0%KPtADC&E>WgY`|>ebmv7C%&5amtK0fgX4Rl)+1g3Uq!tgiWkf-b6?!J8ilwaa
z*%z$c@O!kmTZeAnSdQO`**88mr=02RpZv0_kCa_Kod5RdKPdM?jY)TNrEF>L^pvwh
z@H?>Uhq!Kat6-W}JpEkpU4QR;ts4dhU>;TIbfh4ml3nO`WARSI3brUe{r&ePcqgED
zrthua73`~lQ9FhC_}$y`QGZ&Uu4FHtKVKc>SHb3`DhFh3t6+Of+ItWFSi$brc{(eB+PB@dbyo&9(bmd0ct!ma>TX0`uNV)Tk#!E
zgEYE7-nTke8}%k&49)}Wg4#5(YG(ND?9hQns@Z}W!NwaAn(x68-oCEtP_xU^n
z=fu{pAL%Gpvkq2I^c3e;v#_WOBR&qm_eggYuG``Jua3btG_~O0sh7mAU0hMc>P^}+
zZOdxbb-h#R3%tV+(PhAAm33Gz&b(pbmxue#r7PTjEvsT54}a<6lZp2p%4`)>*HyF1
zg&I|F&9Gl%QhdK@J??FG>o(^B^gNWHx;$|y&J9*NeYVQSeI&M`-=JFD1C_ZL41G%P
zP4}4*20hy<9?LH0mNj^fZddbd?c2`7xpkY|
zD-Fnh$3g4wRVp>?=#@!*RJUMV(|hGNW!T}cEcRs=yg%FJ*6zl*>-gTwv8zt6hxm>n
z`B=a4uW+yMOl*V{?|RI1J)c_DrIr=04m7Fn0RKB-&Y8@?waih~FGae)hVA$4tz!JJ
zh6NqG+hSyO4IA^bZh?kMEt_!m$KusywX9~(k9(QEHSD~noy&~i8n*0S%-)ZuYVh8G
zquz+p8aDOk@peIYN2FwEQ^PJbypJ-Z$UDQdhJ9=*;)~2KI`xn~+z-
z1}v`fsC|z0vrctv#kP@}7WK94!JSbKn@}##HWq~q4z+B~S{381r)!z%
zkrfw=_3GG2g-!9M)^+UXJBPck{?@R4-!5P9HmYUD6a3w^r`NKtyIv=^PpW0-{nr}*
zc~;Au<1F5+%c)_TzdwDgrdZ4RWM5HxvZR)EdRM)y^^IEQGiuMO^0&22>CY*}@D(*|
z%7v!~R~$#V|LpVV_M(R6?sgb;L#>uYcbI)D%%+w_23_<~hTdxyxHbDWzK+?~hlVZN
zU&pkRgY!T1$L|7l9Tl=@b{*@Qa`RO9xH`6S>B(^natHim%gx>ESe?hVExlLb_l#`E
zDMiQCvD?QsZnS<{%c?HUUl;PBmZ^@sF#5^hI_B0ZNk3*w9n&!lJ`umOj_ux`pzHLm
zmermuYdxq%9lN#j;iZJpu*cHh8s=Vg_wV)
z+sCaOjP$~4&G%F$*0G7J_H{MCU&lsFmZYApu46rYKIKWh>)1lS9zPmkC&iOi-QR@5
zjt>5gN&a=n_w>$$>2+*uw^M&^T&QJ}4o%Wnatn5u7%?}X2JhF{|H!{{v<}~~Uy(hm
zgq+{d7cEP%FF1O4=eDbB@ZET=>+ixE_VtbTQ|GxgtZmu_oqD|2oU*%A!E8qzs|?hg
zv@xoVt$o-(*%|gfYNh+4C+wg+yexk$AI+ptN
zTX;r89rG!AKcq-m#+1XBSS0z>F~hGG25;=^*sxWu-Dc0NW406GOYKkLou1!b{dcF=
zv5fp73wJb=u_tpUB~1QX$0AHiw^e1-u{}3lsI2QCV}1tr)6?)SqvnFcYhK~|;GX4+
z6JKf7utgVQ!!$~8Pkw6bqQibzf8VqFn~L@8wuKrF&y?%fykbkepSpEyo$-<~&7ZYw
zwP$0OOQ`RWKYU+qM861GZCTV(MaE{2Q5}5+`9x2xIPErF#3f
z8NEx4WNh%weQ8J3WbDW7Gg;MmhcY)-qSLYr?R2HZnqhr$58}Ye_V4cE{>$pM|;FJ{f@!#6dG~qVd+0PBX{mlAdpKEdd$n|)SQoGbPTDxOC-b358
z=0>87Nk&;*x{h{G<5>N;G_00s8ILn*@w$e&+#dT}^MwrSAf+2a)aqI2ne&pj_Vp~S
znf?irVEk@cfM?>Ao%O8SLv>A$12VQ_()0B9Au{+y4}HoOqd#r6%5YwYcSz&DC;gr+
zV;@dD?!50c?6J8n^qE3E-hnF}7V|{Lf~zLJygpvW1~sz@-`@q}Ov(NQwk`2HdT}QT
zoOj?n!295d{<~{gE{@hZbog^7p`*Ox2F~*GIrQo=c4Zi8FSi|)AwsXlrzCgXKJ8~
zb+OzVbRk~GY_+cM{5uHsv&%nnekU0-sQ#ios98OGW#_IER*s^(^skY$wIm^=x+dnB@E2>sfi57~A7ahVPVIDweOUV@_2W
z*N>&aKEsnY@4O^q-MUylZ{;9kJLh*U*TOjYkMiKApgtHc%#J?&@DA<6D6%$2B4Y)o
zO7bts>R9rltGD-#hF_pw^ym7@8s^x#v-gpqGUlc3>^^vgjMe03HB3mZW7GBzoZ+;l
zmgS{3+x7r@d*Axg^L;rqd4_H|5H^EZdNF^R@7KId+CDN-iv&e%``6)_D>ZQ
z6cyC)AO3`mQh+G0C~4*EzegTum8W&H4|xhzGMHz=OoH)B!@Mt{~P7${x`eG^^&J6&tD!cFNa(odA@R#%MBLw5-LW*28gtUh;qH;@^U-L`SN&j
zIk_HAV*KGE%HzxPGZn+XafXB~73q#5E);Q{$Un&yldvX{mY2&wl#}cA;a~CP^(fbO
zoXBqyu}VzOR-`$J`jV%6jaP((wHNs-MEow-OQc7Pq<=_Q0dJrZHeZaV$14JMM>z^b@gs^LW7|EVHv9H*=JLoB9fPT+C;Ssz8T4CPgHywBLmXxqaM?rqqb|L&%Ux7C7F7RihQb8r`
zb`Mfq!j3Ez_!(JLa0zRjAkYT$M{_!S8?`(M3w$H+69NQ&@eu?ge3n^njtkyUjY-(B
zDuExGD#-8tBcRD_YH{Fu2{_(Z(C_dW0dMG#fh5e%#f4+i)%P6dSy4?(SYLy+oZjI4
zo6{e&R2^ZaI`U=nu4B%G%=vj0RLKJCx~jy;ZFw
z7tAuDn1qdJ_|Emb)65w|0_@eP!jN##26eP73A@%bHHWazrkAJ?3DbNW!u2Rr4CV29
z?sn(twpeK5z~z%(qoqh#FKHnvRKj*zYI6C4)rJ_FB&>3ac@M(hoqD2WNmz4RRi5vi
zA@WPtjA1pbMF
zVy;KYcO4!ceNB)%rzP<9)(Gf0=n>}+{4|T>>=no!>9jscGlYcQiv|FdEGv0=i$1^M
z>Gn_;^#6KFkMm2uqgg`k++!LcG4B5G=H-5rBJ|sb8@F)&sb1MUet{~?0J)-g0AU9K
zRj62?rNjn44-v2{nG^IeKG8~;6L&NutXLdpcg70k*IFTz>nv3-!?@FLYrB#9!(z*fW6Gpr=wWP+!zL
zijoCgKmfcAYY_g$C_tzz%E{;5X`;OUSb;x+Pv8<}OvQrya8bU8h(mS@dP{W${$j3%s^Bs=T8*TS48SH
z6pyS!NV5!Kh*+L?B90fsxw*jS<^m2C%NZ}0>%1tBL5inCt&7r26Wd2uo)N}fF5SAgv;~!{yOu3|7jO;IXUuX4n5>{ll$3nz2*7H>3_@1^OK`oUM~0F%0Z(r*^#?{K)D?6
zMxf>T$!WQM@^&CkU(Uzm$J3LS``>n!>m`pbuMfFBd359}M|rw(J>}`k`EvdEFo1aS
z^yK;cTTYb+pdX1Sk1y9pKEBxVj3tb`F0v2JYJ`3w%EznU+yD}WNtg5Ii+p7f---C0
zD<)y#BHdg>IXzB{Z^b(#{49}<7UhyfdJQia#(P_0NZ5CAJe|cmjD(Hl1CNAdISc&u
z`vsJU;|7h=6mN)#Ioz-kwq=|U{zOdgu*m-{<~O^Kz>iQDa3D9Jgq`LUCt+`S!^C_b
z>gBdaklQ#@z+u_~I*7E+J<0&%+GOgX5*ECK4374{M3Bo9=dCzhA>1R3TBd}}?MoTL
zUsM+2TNROl$X}TZCt)W#l426(_*Y0bVXweXX)mNxZ%YNneAt^>GUk=_R59>Zlmz~{
zqXMlwu@?^?n$()(?2S~>5;jkb3=My(zu+fFy?5sH?-eqRt;dpqFpt(v;Fx(umE-5%
zw4%a%-`xfoVck*ro`*Ybq!uS(5e~te4lGsR_}4B27L~A#(*imFjz<6wZ@
zLb-fLErG%k*04hGlg57+{9LUKOSybw&RmW`FOQ=D(EreJp8s1uk-$&>Bcz+hdh_sW
zT?}D330o1NuSY0Jw}as&_}$b_C@}nb9Zz&=30t^jV=1Ac#Z8`Wr~J_{jD&sA(BS^;
z!ZD%f!xGjecdUfas`&)2=cXbPR22Lavk;zckeUgmK(yoU4*BGdUTB!c%bkC@7v~>*
z?1+|tc|9u+hLy1D@7s40Mn7BaMA+IXf$M2E(iANhc3k_oJ^AC`lzjRSx@vE|PvIJR
zd;3!Q!QZ1DEPx?dn+6j-QbX3gD=^#nZC8@llYH7p;m0F3f1-Fkwduns{6K|@9{Klr
zA@7En0gb(f78BofS56DUf*VQC$*;b>HNKqWkE~z1YA`T;q}xJD@6fPPOzp70cdh9V
z;K?PsxP5e{wQJEA^m&&Da108OaoY1A_Cmj0dKM
z?g=&oems)7x7J+3j%uas+YDMeEEOw333Hp?Bb?H6A5ey&9rJ&r@p!_eQ8Uo+By7*l
zHA-ZM57%6-k-mwGLKaf^lYxU32@O_eT%hJPkoD#M^uzr3x(17y!;i{|Q58M-YXUI&T#jfyb
zvQt`AUJ#Xc_3_W+jDUwF9}iHv4%>tII_aX~9^Nja=IWz^!~dP1%-icSp9XkjSoimA
zKaIj~x{cxC3J+)T_K@+Xn#<>O?S&Ce!gM0sDo8%qwTibdhwgVVGGaWbdB)dEW1pp7
zB|SCPw?K!LFx6=WQj#}a5zOV7(}iUue|AA;0+sjO)96yFub_<^Y$?Nn-EAa8>*|bN<2;l{x4yD
zY7;7?t9%E+klRk%E5OBz1bWPSfeuy?=yhUvAx#I6I}NK5mXP|u@Kr+j+S4!%T4jw8
z9-=8=cVz(+uLyd7@(|{iDbAFTgt1s*9A6hJ#Q(g2CUA^z;`n@Lh>(6GdINp|RC>h``QNLO;H|QoYd=*s_=#3&>FUH#-%HNzK@E>zW2>uaGcaYZ;)43_e
z%Y7o~J&&3j_+K0axmW0=MDG{hW>9V$X)jo__P~Da#z0!;djM!xf%$^k!E{JhguJz+%MInWI^v0(Va-p+${oeY
zF36wY8Ntq?y)bFv5A%o^ADxBMay@x+;LFR!iw&9=7l_8n<>jGZ@
zVt5v>P>i2qJGH`zxcd>gLuWGfAfxkahq3^g#FPG_}fA$
zL--@SVZh%N<&}AdgPfNj7cz?!f}btw;m8}7ggtc-_zU?00`un!VH|%HA%w5HBcz{M
zBhYCF$sm|F6)^1pS3DwvOPC9<2+XfOg50dGLiku8A>Tbag>VIUZ6t3yTuArmL}A?i
zy@(Q!uooQ!dD8?TzSjn-D9rE1WFV~9T&bq8zp$wX$M)@n`AYXT)j0gx2U9uD_xr(d
zLpCXda*m~zD`Eb$9S(GMUdl1oyAm0pzgLt%A?*7e6y*QjfqMX2;yD2dNm#Q-i;w}n~`83dh
zhkLdz0d-Lm(7{2ah&L-66Ypr>oZWc`3&
zeqdU4R(H^&3fCQW1O9&0V$nB*D`>Z!VhuVwSG{PH1=gd{-jAAtzUG>$H4Yf}eEDPp
z;7ISmrhg&7>t&1BEuan0wmrEF7~bbmF_r(4O3yF7feFV4CWin&Pu`q27U=1={q=Li
zYp`~oJsk9&#($o80pEr0RV_q#`xQ^uCIfw~6Qg4hek9>ZhiKroSNntJBE0Re)%Dt-
zyQlr~m^>K$;^>|`_CwL{vcC7aQ9NRD0ks!36lKNv(wF28X*AiwD7fRXi+70PUD7nMUbf
zGJW;44X~IcmA!$!Ba;S(nFH;O%GS0u#W?3RV%8_nJz|>t!%z=*7TNYMGDH8Ko`8-4
zyZgPrFbp`w>603@6P*p~;;20&yd8)w64c*_`=)(>^A^>ABK^(7A}$UFW<;N}Aii6)
zFC0(otIaG;7!OqMP!wpI?|#9`kZGxlU}8Z+Q21&-2y-I@#SEb
zY(Bo^_O0UM<~a*bKHe!-x8v&@*R1WBQ6;QJr&FBoHbIz&Y*C
zkz|P^N>Y(f(IR&d5<-itS&CMYq7+(02`!Q$^=nazXjR!lB>VE%N~vg15TRjQ9jNoM|0UWNiA?%8T`Sv$?H>zMp9z5NW()_IKAs449hW`h0kfm(0$u~;pAm4kz}k^81329X)xei4*E2h0ro&A`
z9!1>)?IVbx!f)E~)-S`Lm)Z~q<{Q5{@Fm+sp`ajmsi0_-2T=)A4(+e*j!RF6T
z!-KV7`{zHVch{H>YB(<+LNRbhBHxaQW)P`8jFvEkng!0NA>x9(K?YxcF}tvUU)jmm
z8*RwPO^*DyEWI1uLcht1ub*>(-_CItzMlO#ybjcJk^FO%Z5E~<^7Ib8Mxf<0{`0kw
zE1BGEvX-HO9T6Sy+$b^-e)}|kp6?PpfP9xkU@~u6wH
zyZ%=`fQtrjnmPTWj#HxP2mX)$z9JE_-@x;i2FAT{fd3*{#_D?rpcgt|7p+>7=lwra
z7?5=a+WX(QXjhOv?FvGZ0gM}H&vZa$BI^G)PTGB>pD|#E)PT&EBg+5x+=_wz`Ul?k
z|K8uOf%feO##K46ei8@PspLR?>_EHz1MP$iv~!KNANrwe525@2>==+E2G)&YB|V6F
z9gtTK^z)fbKpj`e1`Pl4hu@CpK>z*9e7R!a{k)>>MePJqM~El1-GLX0z5*4K`ErFL
zf1N${hC4^Ob3DJ?yN<6XQo)x!qWSGeKjQnR-|zVK$|?N%l1+U1p&EamZcMRc?J4;2
z>lY9`Mf;uPzJRla^Y`(GJcI|x-bMWO+xPSLwb4kv{AUke-$1#S|NKCnL~sn;E1sLb
zlr|`M!s-Nu>yC9Z6zf~X(4mO$cNa&{L?Z4zsJc}mJUfoQyh0qDC(GKgk0@h$9lOXd^xwqU
zvG(JJZ-gfU-`Li|o{t~>p6Q+3bc5MbrQ^u_y6AN!3(_mycO{PSLZdQP2Hvyee%9WK
z?#(!0fhR0jE=kz7dL#}W;8`*0%SnFpN*yZ>{OU0NLqMQ=$0SyM)BcAjJ`@S%B^t^i8nOLF5XG;ymo0uAn9FOv8sXe#BVy`g~9KOu3`DyjwG3v
z#A~O&o9a#V!Gp)kQGL{T-@SyY;qITQoj=D%)Dz$O@O)e+=?zQk)1Y>ntM?#Phq#^8
zdYkG?%L>2D1?n^iqfz*Kk&m;8pUk-QsEv5i^96GpiI2(nzwIqz{Oh^>NXWr1k}<^Z
z)4T+q)AN%ReOCWKeNg&e)m+Fo=R7z~d~EvS>sJ>86<4&h|3-aQ_5P|QkY~zV$#DU%
z+^>Dc(+RTB&RJ68K;btvhaG|E7l(yJ00TqnB_i;grcUsR&p>&t1y4LtzR+y)AUX%Q
z_vb%%LivjJ%Y(bY_vV>f+ccy6yP(zQe<%;udQ-Lss9-MrU5w5X*E6e5%*Xjr^Tt~r
z$P>lSMr;CaKYq+o#1wo!WU|ef_2BJcnsN#KN_AYKhvxjeS>O|@Ms^<1yR~n**>|+}
zv23yLPUw}|-8dkL{)TBxOg*E8bAm(5&yA3OeUOxDM0?#@4w?T=tK;@eUCP@u5xiB=
ztMEMf-BzsXvKsY^%(koirwKlPcYuH|%zAm#{g!$cX-_dyycF#@c?eO5h-wh;Q?}c7qgidWN&KXwQb&UQ3D-=49
zm|{F7oueN)jj7`*D|-XA$77vYSGzw#KNr8%{5pj3$bS0!Yckr~kx`#3a+k+tkN%Un
zuLA4p?_=er$v78%yYBIr$K$k$dNW+tg2&kgH?F~a=F8dc(||orlJi4rU|+fY;^p0F
zuW6k28(+wg%7)JbF)kP1pGL1?k6(^S!amf8>{5Oli2AZ?Q)=maGV)&HW((N4+WTNJ
zop0P*mBsa-|GM{n?G5Pf@#C?5oi1>n_X_%sG8=2fpcnSC+iWK4_p03L^@aY!FUPOa
zd=^U@+NnEX9-6$>@-RPBH`R8_VjMiHi)l_6=cT{4IqlFZf116U_EqST2|MZhlzjbX
zGu@j!nL4?d{6Xn$`%feXJ*@vi`{9N9d^_65Usk)e)Bb&+P+&}Q)x+P5XN!7SJDWMs}=geGTS?^WKvM
z%ufbZ{bv4c&ha>7Do+>bdPaIH96AxRAB!8c-tcuzZ7Q08VkXOi2Z6G@@O_lN_Sy$r>J69cbT#Ues?`L(}
z#}LX%0)Tn@Jk=v
zqP)}#;Sm_c@%3#yu$&-w1de9-qW&U_+Y4ih*f?XBW09tcvPoT{9z`~^iGQM?F!w`*MGEtuV49xulM-v9;SEgArVQOlk@ocdn_;wkWGH@
z=X1!?6>OZxDnePgNHKpNbb@dkhu)P6{&PdXK1jA2;HT_54>CL6087XWV09bcj(=)=
z`4LtPm9HUs2kacs+aS)j`~1Ls1P$3^HA@AiscF5k$vM{S^=IaG}2j}71@s)l?3Z;$W+EgvAGtw3ycCbQQ7*KR!^-j`@)H4UBin5q|x1qR}Y7V#dcRNj_H8@zEODP}YtYUycN8Bl#>X
z3*f(4K4vBJ^8<6R%cwkl!0uB6>uGN+zkIATA0H3gJIX}zpZ9*qFCQetpMQ;X{(bIA
zc0fa%WCyxBZnPG^ep&(l{f9@6VEsi5d{--Vlt0gLEz~jmZ4kfyi^W(b?;bXeq5ES4
zh8j)${g|sp3j*`|b_UyzH`f|7d8If_B>dkue!u4`-I;tXgg@TEReZbMA7i^g?@s~W
z?#iPQOfJ4R9R^`t$Gn9i@=k$t07CvV`3;kWq8gZ7{z4iC;oNK4$>gsUehd}fuV#AL
z)#I^5ke3>>it*-QdbVs|_i53I7V$D{6_Xo0zheU-j&+D-kiUyx`kncWxU6;z06gu<
zCQKaebKla<5$?Tvmuw-_n)DkJ0e>(BX-A;f{4dNe3WwQRlWe}|d>P@3wav^w%W75*dfn&Nf{mic|CKrIO6TM
zzBFl*{IE%*i|Tc6>r5m0&7tFh-fPHP7$odWWCWi^fdR?c)(Q|S$>aA3P6ZaYKCcTE3i7ekAFrx7T
zJ^x8Uubc9(t}e=hwgENQR|U)i{@&Ssmw3eDJBwoJxta%}ChLLQCoQ|T3voV1qBw+j
zL9mDb`>v;_PMczEA>sKM4owqULG3
D?uZ^YFf=S!%!6bHLM`dL@e>J#71t
zAgt$=o%`MU8R9!;Ozyt{nDIF8E9P^3+?&6efMwIK-O>XzJsYNN1L&(DJO;Px#i`Hz
zcz=X8etB{hQ2N&uk>h{@YKwTVy*2SpfGF-?t4>cB%%`#2*DTyVtsO0kP5`EVcOJxg
z?70)xx|sgA^u0}ifE~6&yQ~1YXI#T^KV99u`xMr@?#6^|DN4hWW4;!UjFD9&}mCQ>A?IRO#Lu$9?;o|
zpPa|5_Bq<%`kJ^(#NXj*Rq9`?m(Gokbj5b%M^5F^n9q(kkIDb0I<{*a{{N3<&CA;Y
z*fV|eZQMUg_4bX*0qW`Lxk~^aeY}rrY$05f_qEy_@T`9P0eqiQ;NSGC_&YOVMTzyZ
zAii0PGYhu|ty}Da?begu&v<1CI8R
z`(NBF{J%&bcMUwRWJ4`KV*R&FbrX03L;nbjb;9Gu`?BXkYru`y*4)PN2l98k#^X@m
zRiB(!ey$aD!R4O#Xyfx9_%rB@N?Q!!JJa~y905$`*It3^Pd%P<4bL~@g6Z{}04M1y
z_v3QPe?KMP6U{DM$-(JS7nR7pRE?=t9awL7X;9dP=SlIP;8&bJ=E?~&k3(HL$ox+>
zaej{f@1*fUr1My%`Q!8p
zjo*kKT>S1r4-OCOx{`Y$e`M+Rqx?*I$vWdv6;AwB1HVP=!}TafaLBsBcV!$&U$cJb
z70w@jt$>twrG-BphXX`D61j6~aTBO8Fh3o2$hsQc-9pyi?ej0b!0~sCb`rhm)Ez#la#-0`DKFn
z67NYlPojoMSR_-BDNn){szL8gj*O~#N>xbLgHcY5QIri5GD;l_PgDzo(q$V#IxrKa7abMa!h^T4D~YW6NXfXZd35GyOmKW&4%gcfnyNsfVqf<(n;^
zI$JN>Z&}QEdBOaD@G)wVU)GD_W@f*T$msu(OuOK5WkPqH**E-T^7k|K
z2Q$ci@2H5O|IOt0XVO112*w#w-f|`#oMcEClyHJK7=*(F5BX7#2p;Jo
z<)35t{>{*>-AUtl@+h8S%3I9zukaxn?{7l!W*Mzt;x=7>IeFB6vqo$+;k#txeX<_c
zZm_?J+p$fRwo^i#+D&qj={e^-EqXrc`Ob~V4J)R;8Yx0X`uM`uXfFgnrBP>7;#e6WU-%!+opsszffYp4T{p!)u>75jrC-
zl|yTJsGr<*wPBc=U@sT)A^RRFejCE)_f3NYOM|u$+}s~X5DAszDqVvk?1XwM9+LC_S4yLVkE1OvF_`X5vof(`9+|j728TaS5F-6n`k}uVhg;%|iC$zdA(7e{
zmwOO@*~1cF4`BV)sNuIg*7H-B9J0pz$h`|BezEH%KOe;PR_|Xv6~~KP@Gr*gd^mDv
z0mhH-%-&%=)#r9C`Ho&`ZonN(&wtj5oHNb3^5iZK|66fL0i%}r6taKokUBdL^JQxC
zdOpSq(X%@+IzRIb$Cx)mk&KINsh{|8zOOQ3FttH%e7i&yUa{p&aYchf{Rt;Bv&bboJqS8gE*Wuu$B4VH|dT
zerXIw(HI9`jIzVGZ(|%CDvZFGd@<`AM$?sZCgO6e#+Fv&aHz<1SX#l}q1#r1@j#a}
zd;$*ky%EhMJwehLmS8w{Q(6q7AN?5JAq<(@d6Hc)
z$EC%S@)mTsk#bff?IHi$rDD2{xve`%=x+8$koEM{(o#~-Mv=87pLz$qH(e%@PQo4L
zR3El{O4sr0rlKVO1Dm%by>u*HcN6-#lwP5q;G$(Tzw$h)ug`X(>+6!ev|lRz#1gvt
zxG|)>(ZAH6<>Y?qS5OmA>-!{gg7Dd6O3PpM>^sSKT#?$_pUF~tS9Lipf5Af<4#=n^
zbYYC0T;UZ>KXZfnN8FC0=~j&Xc{Gm7hv2akZ(XK*>|R0hJz(VI+ch+uCzYcAB-*~y
zjWnz?kzyf}eqV=%e+y8)H}li-mJd+*xQ5aHYZyDb&k&P;3B~b@U6Zv7cQShQJ;pEE
zh>^RxjNZW7&#y50uzMmchqbpaV&wTmhW;0$Z)-27bUzq-yEl`+he?0Q=mYy0KcEPv
z-`p9xe1;AV^+~@UX7mX5`!FzRNj&_S1Yt2H*vZhtrk8|Sy%dBR3A6bYFnq&cA@Qu<
z&gzdKTuA(Uro5{ExQS-qXr
z>;Fs7*2Cs!k&OrE7g7$ZufZfHVHQD%k}yo(|3Q{d)=tm%4@(Ecgr4PxMYcS)ewOe5
z?o-%wHsAk+EFWw-TmFBM)!SJ;p5>cuCmYZ9153}A%l0!HX8C68Vbj_FpQU5tSw8=-
zeqrhVFFH0Ki|n`rAx-+3-RB@C?9Plkwp^B9HXpnH@@MGT{$}}Q+rf@AHq4gyUu64%
z?LW3WmX76{4YT7VhuQbT(t|K0^?zZI^`{&DP2*X+z$xZDbddzzXJpFK
z`$F!_zCz2C((hp8LF@+F&O4S=&J{3pDa^VHMtV}ORR~43<5aGIk(9*SGwnLW^v4Fq
z?s9R2$^jK7|174z`_&v
zC!+Ug1*Wv(_@mdu$vY*bdrp&lPajjeWA!t7F2pHXOZaN)lZCA`*bkHc5jx%W1tebf
zlNkwX2-5G-S+pmT`c5WLyI;Ums?UtH=u`UIJc0)~b74xs`AN(ll3!v+B*B$u=sldW
zD*>e6QZ5z}eTm0tg%Z~PWPI;Glm$KX7KfCxR))SGz_U@1@EJeLfncJR
z6{#=f5kC*1;KGjNrcW?fC-y!?dAE1cw0igAUmJK?k3)0^abAf5;!3f^@%qZx}a+-KrtjCYK~NIzX2PsgFs
z$9_`Z_4T2I--rVGuFcPO>aXo6M#s%kemTNtmv}nKf3ov1!9bo=QlG=FA(HMW>OkUc
zm8%KAtvtU7_8oSt!S$zII~$DA?$66Gj2;q?h@LEaY%JMNFZ(SQgX0%Cs>6~1`=`?Z
zB>y6TbEJG9i8xZeao``q@3a(V@xYXX7x2WB{Njh`cs4jr$76S-0^zG&VI~;|4nL-n|JO2r
z`kUIGIY#oI(pMq%8Y$8B%knApgQ@uSjHGkDXum#Dqwgi{4yNnD(WKWT-M5G8+x!2}
z`P$X#M&e}`)A^COj{2<^a_Kyt5JK^*u03j@;g72+vi9SEr2STmUd;N*Y_p{CvW#EH5=LI>GYHPGB;UDH6uTI{Sh=al$XU1q
zM(8BXC=N5>aZG-2P9y2N82Z~vH0;mBM>BNGnJ_%=MCf-i;T9%eB*PaBevs$EP^CIo6gE_7~CWt9$z5{iy}c*|6}d^usD+V6%3zXOe0}dPiD)x
z%fz$wu;svDCG@QR$kxy57i@mEK9-)|%j)C_a;frUOo4nGxgCPo*2skmOMb(>ukBH)W>C(PcLmbg;Eybn
z%VXE-+lc<{F_;qat{KTsU3%Hx3GTPgtiCu8-Y4hqUp|<61-{3%)#}NlW6dZ@W|L;<
zRW90k=-0{=sGqMfvcVeq<6qT~peEe+n=J19biV+c%b$7@mapB4?zl_65(Gcp(yw{{
zBz2}v7FLrcatqNL$1;cs7Y(DTSoBOi@9=&YY(?`^nOx$;bd5AT5n
zBr&Dw*_4I#=uxJA+L7IL=;in*kJzX>^wasZT4G8AntjvHX*1T7u_^by(E|f?LKPi?=}Md#3L6c-7@2#b4DAQ112`0
zgHM7SohCG)S4Cfqd>fk4eRWRciMw1>vKl=VjjThl)`m)biFGLE{QTGXOX^Y3=r*+y
zdk)H}Jt23kj)Sg+SIyrfS&tgr8h(1Y)FYpsDF%{4_2`zfwBdM-I%Mm!vvZ)X7Kw@6
z&wkZfhXjj#7tXvtK<7Uv-r=I3!jAJj7ID$*uhTVi*Ege~wSV_s
zw`@X->V<5oavD)xq~4ha4IDJOAh!Ng5(f=Ne`(&mjDx&`GUdj))uS`^18oAL^=N)U
zd*QO$TI5kwxpY-;EqZX>@_~s_9rBqMDD(PN9a=VSzDH7T9lTrnQ8N842ff|6<9%li
z2k~4}&K0rapf6qnRhsYX5$DQo_(WtqQjU3N{dxu$Ntc|o?hkE74`x}%mpC_}zhN=y
z#Ul;K_1lolMuP@4FIjGH+UrJi*VfnPok}B84%yllw6p;o+ZH%M8phM#AKCAM>*~>$
zAIm3>13uM-{NKvf)FGYgx&rGj)SJX305!=>;I`p7n-AbcQ=;sWVBK|4$
zNR4ZrB4tsJyWq3?;O4Jb;#w=zUjK_8<3`e-@@dWMl@9?)%M&ZE;3zI7Ow*LKI+eIj4g$G1UgF;
z#*OQ)N5@x*$-Ix}pyE$=9vg-=qNNGNsVf7U(c8;BHc#xi=mcDVoqn$ll{_tQ4GXVB
z-}Y>Y-u$Ny4MYdJFNFW4sPImy>mv@znou=|-=-e*KRbMMEBsG~9xe^N4efU;R$2E0
z_}Vf~sO3#V9a3sZK5BNQ4%Ms}-uevqHtI63w;!!TH;=WiF1X4;Pn=WNw~IHRlv9cC
zpTIg3F2FVAi-7ge)2L})4;MXNb9!R@B`&hbaH^NO3E%Ua9&DTGz(pNuw#T2qdZoVY
z{>MYS&FHn9!K&(>u)7?-gKGD$rSTnnhrxVY(c8CX4d~FZK+V_)4mu;*)z$fwgXTSwx}vm_
zgHAMX^fpiBAl-Pr$YT&*mU&WqQVIvDebt?My_AD8o+(SMH{l?gF|UsFLpv+OLVwm?
zfpy`~g_hxT4wCE@I&1u%gLcb55V#!3K_3;88@-{NZ4;;4ErWefn@a
z)F0&&vacBaC)cuyG=A*{)Y!H^B=B$}>a937ZvH?Mnvq_e?)<13jr`*puSw*fy$7U)
z<4ZWm$5=+yaX-kF%(tmi4mF~HbqgzPr$hTzNO8x2yu5G0BiYdh?KpGW@QE1ERlO=I
zvTZ=RgNqH{MZ)?qcYed20uGu#%U#B0fP>mTm2R)v(SXc`Jf=0iZ9pG_RaYPT+kg)H
z`Q&>?bI`jd;S;J-VZ5yl%BrFdMIrYs&lRpaXqvzDfKrZUG2nuFk(5HC
zCKR=#a8%N_3AsP^oswMHgw8q@44pdFj5evy(5dZ(?*V_4>N8SlLIr2Wl=AyEAtz3&
z!NY0rPQs)UhZY`fMz@O%&R%=~{Tn{vkj|dL_1OhkH$7d$zhSfcsbmeLoE*g1_ql_2El7pzr)uZ-1e=
z0sJsqk4>;N2R&+)&h5F-&gymXePiKXSecO6c>%a*rS;EEsR6!=!n0}Wjg@9xG=sBe
zqmUrT)u%yKX7JwWdh3`!p4P2MKl9+zJ@=ZCw^~d`+yqz`h2ql-^SS80pImdUKa_uU
z%oqW9ufai1;M@M#X4H0TTiWk|W>hxqj{i=0XLYtj;tE$)xW}{nPmWC>=&>64EAYEo
zLpv`Thr_)s`_*^r-Sgp{4I3j~{fBV>&N{h2_a&U8ZxcP`34D6#FFvtsaVt{ocRjw!
zvlZ?5)SIwvS1X)fYA1BTJ8DVI9?sstXM*6FUe^Nn{_nI;Zhyl;j~(1|aiV)08p+(6
zq1xRB@4C1<&j)?a-2UBI7x>=Zg9G!TeO7%(>Zvs?N<3}o)q;s7%?@qIa8l*>Q$cX=
zXL|Vjn{eMOFv@_xDz+8*|4Zy^*Z2vzo)nHWgFt1nq}mi-2r;g
zDz!54PSkGOIQZau8`?8y?0jZ>J5rrjzEXZl2U@MLd}q6U2dY~yd8yT-135c>zwErX
z4Xtzb9^Dt!hCYaV$Xooc4Mn|PJMj5II}+^G%4p>8K>J!H`hP$Cj@nfc_m*vMLvN%&
zd>(B_>2bO|kE+_y4$-Cdf7LqBao*R?#Vs9Z)z4u~$zvTTJnY>z6DUu(P@RqE27t%FbpT1tD8>#ZnvDDtxjot+xnbaiMfmF9mtCif-fga7*
zn0<-cfgZdO^;bRCiB4pR*B1_SBD4MwVQuLS)Kp#~RQ9?9>D1qS$rs&;mZ)oNM|GX(
z>#S?iX{ZZvcNh!HW`f`Mn2MFxV&MJzZ(}28skS2n#SKzo_q&k#gHPvO1bUE^!?h>-
z&$J;i$);I?;vMMYnP1s&ySh+JMbQh*Lp^AZqV9>SuX|86Y#+=Ib)yqe4+VqLy5as?
z;pHjC-6)KAJ^$sJZgg4U!yV_LZlu24bH&{6ov3J;a$LZ|F7!FtN-CnH3$1+PZ|8fd
z8_n0c?QpoO8~xynM2=n6g?@e2nc%0~jbtq5+ibP(K@yuzhplkxLAvrw^`$@dpau{7
ztSLKs(EHUk7SH=SP-MHdl=-$+6m0c5QCk=0?_MG2bsFHOY~=9FJ-G)Zn5m|_yY--F
z5~m{8o`!x`zW?%Xax1#-K5a+b4!8$e(Rol~v>R=C^&rj0vj<(wlL%>E*^4aFCS)x9
z(1WtBKM9@G(GA}n?;agH*p1}>M4Vp({p7>bb+B5o2mPp4d9^OH2g#3eT+Bv$5KqH#
zzwHh^NJhFTJYTsN&D$mOXf|&zDmo>)p>vJ61nun&Fv|PVjSfYg^?qL4jTG>1)Bw^^hVcFQ`kXIH
z_3&KOllCkp>CLLQ3Cg0ISmo9aPOAqX8)Xh5T0#hBKGev_&c5Dty;SV-f>aV
z)swj0g+ea+R#oP9A&%I#@k<|dAuoyJk#Uk;Xj)p0qH|s+S~>If_mt)yw636HmelVa
z6an(vM57mt-*}--(W@76PD@Cc1ofiV-~5qFU>DkQ?}d-vvo6#d;-_PbfKZrW3|TH(1Kv{3ujM3d@H#3QG%-E&Mg
zTH4=Knr+jI%$o+J-U<9b@@_Xh-f(-7&$sq_3;lc1{2182I8di%(~G%jSe%rx^85wSiOlSqX+4%t2(HZ+KU2f8XnyK)QdDF
zI(YU7b)dL6u4RSCzoRHA-c2q%esJIzJ1
z`8E54V18`6>@Z;ThKqcEH7FFp_;N}x*|cj2zEgYRaSdm4E0V8Wc}Fq~ItwO
z*meKp@-2t^NpT~Qp}=R0g^{qvG^p3Ca@v^GZuEU*vxV}CE>zp};atb*PE^nPqbD&M
z`ul3#P)sJ!CqUxgUX-%)o^$Tjk^dvXwiY
zD`kC0X2|qn(+u!CDl^WohWW1VA^x=@s};W2b>gGGF|7Y@_BFM}eu@{+&iIpv!@B?Wlg*L0exN_yKJF-{h2U5Fye*8hXA85(QCS?x%
zU(2FAY>ijI|FCCk$tV3@bTURS)5@y{DOI_y6291j-i|Bl{Pnp9&1+wJ=gjF|bijDF
z$bu`q$XfN1qQsFl6h8~eSHZgmTaRl<`)%bShZx_IOc$^#d<^`{I|A=~eVZO{0qb|V
z`j4FSBb`Y9#zyfCquYL?uM2s7#fajEM|UhRB>~GGquv##DwV
zqNvp8(;zeFz4pGY@ADkzd!Off&iTF0Kj--4Uahs)-h1tR?Q5;~UVC4kd!wC=ZBsQp
zH3j@q^1riCUBODMrv^pT^H{XlEnu$C
zLVpD<4Fv^#6iTm>wa)@yKM#NZxwE`IhA;N^@Eb0vn%kgHssF~je_tQ(V0Vu>|E&ee
ztV{Nvf7MeF?Vhp3&F{ZA{NJkmTYpWYdLG^$3p^GE{P%T#ie0KJ3}AmLDTw1;;NkBz
zBOut&@${{9Ao$xw}_i)*b3e
z{Wet4rg;vSyTC)CMLXfY9#5Yp3Vr^-3rXG2Q)t>(!S1~v)tBa1iPv`z_&=@v+p2G~
z3?Wwd*ZivTd7^&(13Y~H+Xth~J7a@T;g*Qzkl&<#4H|GkG-nkgswDc=|87K7uOW>Y
zR8dipHPBdWPpN@80Sa$!k)HHVQxubOxn`_~x`HVyQerL9`gX(o(Gn%c&23Up@bmJK
z8c?wNufP6R{2{$}Js&wNrt#4^wrFGMgDawUAqFy8R3d#i@%;mx-QO<#~T(}|f%~^Dp3v!W{
z+Nl(I9);w>6lRc2kjf={;85{TE({U%H;espxFHuPKvMY;u{>TZKQ7AGiBw-KKO*uc
zMY=PIR*Fn`Cfe!71}+oa8UB_F^>3-7OwgInX)Rk=GNHFvUyZ~lw-WhP1L+UBFkFOHC_M4{j^1hy&zV#rqA{QU0*aAajf`&sL
z)4On~Ei1QE9LnhC@VDqdCio1V1XmyvjNe%^dFaoLOuuNrSjLZth+}#wd5;;@O*zWS
z)4$LY!Sh-&mGOHf}X*VnRkT0PIiD=8C(=?1d$
zDEA$#Tv@-Am50UBO2oRhbifml3ofmWRg->9_i3uc?;Ld)Pf#Z0cJG6jBo~sMbulo^
zOSdRizskmu)whf1cBF(hVZ2e*3)WwoIu&N8`98uGa^dRwZA`ATfFPV)nB1Bcf?U|L
z<|nf=qL~&eH($!R*3QO$wBTey
z|L*hg0Of*ES_Ri56V7=l;nkA~1EyIr{RyL&;ejHKWpg`yV^Iy|dWw8qCHCUaeXkyu
ze|Oem?WczFb@J1hud6o9VaSCGjWHdlpSJ%VqXide!D9Ul;p?vICok_M=j*GPLnl_g
zH~AXV4Zq4B0%G(#
zlBO55mWBcOaaz9M4QBFjP5s5kIZu?|!BZ!_&m#ZBfv@ivhQs8-3g)5AnhCt$+ZM*GJ+E-agFuO^yz%{O#;JFjOYARd0-tA`^}p
z(~eLsSRJOFnOtafkK;SldC!<$o`wz6d*sXU?>(clti6f5XEM6rPA#h!v&6`c=z^wb$fW-UsS;-VbhnsqSP>k@)c1ZoN9DfP&5<~i3(X>5uyHNDb>`1^$aP1=m63ea
z$c5EE!6zy{d7=njZ_5VLD)7^6J>x4tP4^ElA-%yiE1yz(HuBnXqHk~K;?)Aadi8?M
zk5j~BYvLyc{ABWmkEXD3o7lW$^Sts)UnZYlzw!jN*V3VcjlBTB+URNZ}
z#)FUx^LENd{D6G-S&aw3L0_-_3`8s!7HLLYCH>QHJh3q3!d9PW&&ggZyM5<~x2PKG
z+5ptB&#SJ)7w;uUFSdf~CZ`~H;KRTTGe2aU6)?mAu^><1^G#)~BruFkE0O!Q=zD5fv>!}3A-VQccd
zcy5vC9&|z{h#{Z6p8J74!TlINw_htpThVq2dX8rN{nfAI^>k9ngUW=|Z&)s%y*KlA
zj>K?(Fl7kLL%SZ$Pcl0_Zu5O@O&CQWnV>iZ!$SEuJwo!Z*LWe*%XsI(=$&Q+_>pG`
zp~Qq-XxlQ6=@o6`cDBLlCHdEUh*3XZ-1jtKfk`Gfx8izzu-d7-hTIDD{lxviJCTpj
z<93dr;i7yJbp!oUkv~b>H}F@EQ|1WoQ#RLQ+amBrk-S_*9RIjZl&L~K!G$7-On5n!
zx1aTp&r4DjuYXhA&pjeF=shJd`u|O4gQwj%@sC71*)|XS1+jd!SpGxg$BW~1r|PI*
zDV8q~%fE}$HA@9xrxlGuwh1*?d#pTs_dr04k1z7A;X*$49
zp<>i)AyO<_CTGn-&K?K!q_|FApX%WyVD%(?##rbVigJnnTfa)Ie^#8gk)l2u2HIth
z7gWLl)(?4q{$gB~=HawBZcnkkv=5Qmp;bzbUJRdJ?|*yIkTWJ$egh
z{-k-5hv6>;VX2d>2*`xJcxOi1Zp8xJ=k3mK(D~J@Sj9pJOl=^UvHr
z+I`D57&75ErH{e0a0hB7w$nna_ntW{neeVB*IOpWo0)9lkOvp}R!w-h@h}>QOb8d_
zNz_Zu$5NUQ?XT&^`5uAP0nRJpIj<}F*%Y?Hg5MQ!|Ee8;dC>bD!1c06@pc;RBtvDw
zld&APcg*8BrBIAZ_r`Gj>Ts^_*_~FDOz>`ZlhH6`UVo6NKULgsw0gz$r*@|S11F52
z36%?K!)Qgzgogp#Zr^YUz;fY53dil|dQk+!b5ZB<&~l0y<12k3gV19y)iPOdyCeryz{HWj6W8d!+67#wJ=OByqkCPY8Eg~%lnW~`u%kX(3j%Y(^>W!5mgmb;s<{-RxHcck+9
zgG&%{WWu+tkKlUc!p1=ykB4?_j)b~g2zs^XJJH>mpRn|CUa|i>8&`xZ1QUYum`tP2
zL<_U?ZxRjca0#!DOc;A<1C})KY8w-*6uFRb?D=MrSC0+dOt`agKrX|zSv4^wB%f6I
z1rc5*thD?#p7c%I95W`|ZM%OE9<*Fo+%HC-=#*_ejuXu@J%5&PVOXuPF7`-fQz
zZ&)@2-`?)XTsio;^GqXef(Gamy(iq>FVuB!Ba~~T2m5~nukUWSCl7SovkJ8dkhfY`
z^|T0l-0P%o{lK3t={;Q+d{p(czAYQz{p)HTcL;RBmCG$_A#YJqeo_f~#}v-GlO6Tm
zUovQZ-cASs1ds_Wd&Vtk4!(h(!6}m4*ObK5xZ)qEu;<@$c+P&3$DTf$p9T5RqY(?g
zf_`rAVNC6Xv?z3n$5dYSJ8U@A%mg&(tXmbeYv`rFm!6;6s@3o$@EfYt
z7l?m#a3fnEp5~rN5d%N!dc7yPSwDJF<-)MnbKz;^LV2$$wx3AbU4u7UCfuF)xHrj_
zI&T8vl?hHt&n^+2n63_2jeJD$Js@bThwa0xX+DhgyI&=F+mWl;dYxD|uPN!9nCN#^
zhkR9<*0UP$PCZ|z)4UAXr0PQRzI~k@o4+H-U8N@YX@8^x!IqbdrX?m3L
zYp)6j4LGkmr~prkJpT=2R{mChJS#sx+J*6YyGJuScbr!Pgpv!di^##tgwgNUvvRWm
z*V%i0aly2KB(EMag6)U4?5<$)Pih?)ZL;P7JU7lu(i0C8-EnmNbJ8EtKjA*nag%3PJhnx6lUHq59O9`d3lr16fxyO
z(}5HyWP*8kDYK*CJukPqF8bGcLehBdITTTm&zTg;=B;i8IewWi{WjNA6bKO_9&EY6
zmrtF^{$TI4g+B&xgTnK+mBO_nQ^o+zy{R5JPUT9%^WJPkYY)U_cxDyFfw)
z=#QEKZQqrbUrFcuCOytuI`DBW>A?99>v(yvIp_QDJ@Jmjg8wd#D^o0|
z>;jc%i{&Ar{z!OXk`EC1U{OC>)Mw!i?U_XJcAL|1ppQk%=owMpOw8}4H|NjkpqPhK
z6>&)0cD(**tV*h%AlgCEo$UQYV#rx@;HCB{T9KSJ4@$F6`gmEGoMwgi;i4T(3_Y4P
zsz=sPIeWa|BffL_22oxt>PhXJqtdT#&J4x4uLo6ZdR*S?g&&3~Jp7h?HtkxKSTG>Px52!_a7FG9Tdj9)Ry6=W#fBPlisVwId~P2XRX+VKnlNb^d!r-
zznt?DiBP%V**Uux(bkbCS%2=g_&uQ$TZds{(od-zsXu=
z*O#8Y%&s4wnz4RUUX=|XJpIn%k2BR%4?W8IIXUaMC-K9~?sX^pwCeHF9K!9vBYj?x
zd`G6U58<)acJ2sCa$(koARWTDepg;}ybgM(nQs)~bTs!He$M8(hgKjA?gm+);{&UYBIQO7`
z%388x`o_;w(m~gSZF_nYRO7AcyabdxT%4SBOc{Cjz@*?)kl#uUeAf>3`)#;xlnnk?
z^A`uJz*hxCO~?Yh*wHSK@U>1{*`UoR&r>OA8`D$|dZ>)n8Cu949p1$uD)kg{M)9dlL@&-HRHnOvV{+ul~_m+cB)m~&5
zuZp~S`_KDt|B?xrLyVp8qh7aqyD6t?u>ZYuxL0X2)Z3uHdTyOeSfhRCOY|}9*DtIZ
z5r+1rtZ8tr1nsncTKc>b_?Tv|0tE1*x9)#y3jUbyW~<{rf#Vah0$M}f@{Z?&W1xly
z{hAW4j`>hi^a{M@go-i`MY&+S>H2}Ln6LCbY8rb`e)rk-u(qJ~5AqJahu((Pv&P$j
zZ<7?YT^r-ibyeLd18uNl&Ic2eFY32;-Di|*8J2b4_7r*2BOOC-S7Bf1eqz%`=y?Q%
zI-RV?`PHwjYlnajT&&)j+W)TGuB#5H`;mgDv@YA;RJ}vbSLt#6I$9q(dqSR7LcY5B
zU=?}~A`HLi)4F|-b9DU|&}NN0f2H+Q`m|RSox8+&p8rMrg_eD;-B*TOWlY2-qB+I0
zaL|Fc?X~FZbMR|B`yA8<-Rp0G4GQw@V`|@%-kv}2eKkPCOl|acU_LgS*mm9T6V0zUjE@0u|E~W$=66yZ8^Cd5URFA~(R(w?(Ik@IORu-(JJ;a+P0!PXv^NbkxOOV+-_Jc?|{7n!*+`&%9%V2*QM<1UQei64uIg?xw7C8pmw
zln@`*kw!412S=`8dXwf}(x7@38I#%g?^%vu@*f@eJ;lobp@?L-Z|K>E={3v>V0ng|
z>nmCN&u=ybWPpEMvIv0#e#LqJ6PlM%=Wg{SzUZeO0w30!wMrqiv+m9XKwO-+hn!$`
zY;eMM6>=3t9(SHr5n_@Fk;`~Ib9=vo*)ukgkR|qI4m_@n3FrHlEj#$WX-{GfvolBy
z=Ar%3g}gkgD8G5x}Je7;`XKw1cTH_|UK+Ua8ib~bY1%};Np
zfB%jW41&L~zsGppd*O_qCZ22epy`5s_Y72ro)2b=_{%g5@SXZ`JI9K7y%plVDfJOA
zS831n%Sw5??e~(8KlLUr4>#uJ5eqmUzlJ~0Dh~o2@YB@Sd#*RXfYVQw{h6I-
zmHGHg7IFRJT0Soe=Mr#|3k&Hnp#CXwKafP#A#X=w$UBPsX_{Wh{l#%f7POp?d&
zdZi3#%Z0IG{dZ!yw2!i<;iA3fV&2nS#7|119Wr{{=zpGgUsK+b>win+a#MmYsGo$l
zkMzb93tmm660T(|2rPX_Zp@c&Qkrr%LUyYZZ{c
z7v+@AAbCrXm)bu_=}^d55CPpTmg|Z9X_05sgmQ^8G1`}Kw=^z^O68LMZ23VSNg@1`
z|JOMdn^v^TrXBJtV!u1s#NfPGPhP-N#yT|bSWDU{GQkjJGO;=YuE(Kg{APeG*92ec{3I5%oWQUi|b@gQ*M`!1*dmJ
zJ1JXB^}kZu7~}2Om-BI?H+aSvXmnXd35#Ed1nUcc!
zh^L$%N@*g)s~Npr1~?44(b+P_Pi{&lF~F^-<}f~GRSom|KB;FIuicm1GsbBL
z<2&eoWcnM*8?tyARzQvp{>7G73i=)Q5EBAFy=f~>UVy948NndnN##nmZ`i5J_ZJ;5
zZes1ce^mL7%6C^8!Zpc-59%ZE>d6G#^aKnTc(=2PjQBgh^^ov~-!&WJOw{$;0TzFD
zc}KAQfb8XGco@w8+7lQUaQj=cScxZXQg|_P(6nkW6cRC^g3bMjqu#9^9oZa@1D4P)B>v4*?-DP
z>aVB4bgW$D&wgz0Px@Uh)b%9X?mXYI)n}BOH=Q(tl{N1O
zPaYcBp@i`J%hKgrU0wJ6}YiM%P{^p*OfJZnZh6_K&UI3Fi*z
z?PVI-9OstJ$7k+Bzf-S9>JiR%8DUl73cby9-i})VYV+Gs*zy_s6?dn3X?n0bt-!h`
z+I0&awEGF{zoRv}$12pXpK$c>wo$<66iB&U)n#e<|WCd;h1eYMf#c*bO0b1e$>MZ$lK1n&iq^OA?`
z{7}sG4Kx@IfBbMF^JDF%{$z5$vv7ZCfBb_R%#V(J=u$%Cd{p2+o8*hx<{v&oAIM)(
zl?hdkFufR0*eMXumIqM``9Kj*K7;osUiO2>;TIyl-zehIToJd_i}60>A#d+SDv#G~
zdh&MlkX@p7MnrI$>O=??e(x7060y#gVcUiBkc}`Kd=*_U1)pcn_d(-Kc^s@8&*x>l
zCD)5=#`|j~;u3Qa2bEB_Xy+A8H`XUmPo9c)6jh()I`S
zyy!8Y+zeiTcr&04;`8(P^UN3VqX7f1*q_q25qdq+$N6JV~`1*Kk535FbDC@;swz)_!{*q&1;GVF`V(hx-n8+)u)H&W*5h&pk^kJ)
zd=86G2Gjh^s9tXOZwx28U7Th><+Cr3no9R1^4oofBfx(8Nh2qs;kLa}D1Ln}^G>1s
zPxr%%A_<EbrKh>U~}j_J`t^XH3U1;)h(yN~QQarpblZ+d;c-nck4(4_D7!N$r@7
z8HgX6K)hG9T1Ii`MXUPUaFV}U_lWun95`?=+1D)X_hlKi^W(#a#D0j|lSB3ld<$MhvwtMz0Ugp)_1Zu$P(?K*
z3^?Y+_U|9dGe0q_tUxn*AHL$@fY-*
zZrnNS3A|O;%iydh@Sxt&0b3rl$8$Sq&@t2u`RDPClOlj8b8b#uGa3Ckss?SPyy#tz
z@t@ji$%Q|LUV%?uRtXi|jUOCD`<>mIx!x_72|LGX?YV++T(_;W*EB$0b@-yz`U9(k
zrfRzf)uG?Vt5@sJ0{-1wsrbHWKj4JNW(9Tkkxw>JFEDIUDRiCKds%W4qvy1|hf$JMvNx%isXJuJRf3Oi0^
z%dHy1fBEPKx)@gr4_j*JHCv1Gl+_>9ZGk7RKCRiX7q~e2i`AIxUw|9`yiSb)p4>Cx
zXL$6|D$ILChok6!(!$rPjaC~R0)IpOg1smZ@*SbRBrGuxl4TfT$
ztgm90*%kMyPQ5+*JE>f-%kH&J+w2qa0jtJDET|G*RMx(DRYv>s&I4BsLf&P0fsQHg
zWI>+aF}9XeHY75NRX
zJ57mURQ2j|pz5uw$+pz;#tT*${EH0dX#_tg=|N0(*7XG>Y
z3Z~cpWgIK7TX=xQxfK87ES~)^N&`^9cS1XpSU$r}oqmxGd5~ndb;w71nlt*c*Q%yu
z=h#6**nHSr*F#{H39$w}SpMQ;&1DvcW+h+4O&aWbPOJf>!@cmL7m1{I@Af+&7MxG+
z`2{2boZiWn+0mkefDF!UlQ~|9n+Gw<=h|T#1-V;C9^ZeI1MY*53nPGr^9cs9krO)(bI?mJ9bG*Csd4E>V!|)K(5AYev_@FVa
zd8GHW?>Y9|2b$CGBjNn3ngAB=y)Bu;^qT5H4Ed9Kh#~hDaTJmdjPHf7r%1lXnd6qH
zQC#m7qA=CF@qm{nl9}LtICDJIH=gTv3F7joFs>IxVzigvj{rULdXqUHPe0E9c?|>H
z*hlr_eBzJwOfSTP48#55`y3ZK9^kmoSj;CLqQ`*pX4v-s1?@vv=hD3~@iv;vvV07KU8@V=vcxMcWbBv9w&o
zk6(CuiS)Q2pCY!`pC#~-Z`#K7OpN(_&7DgVk8`i-SeDQ~-I^W%{EjYv&Mh4pv*$R{
zmE+mv>Xez63nS8BF+FQDz8_kvi}VEKN3?kRksmp3eV$Ul^qM+wym#tFJgcwp2(JzF
z4~*yQEh@B->7CN|Wb(!uxr`64;q4gF=V8#-aiN=4I1g+-jq(0o!_gu3lj}w^KGKJN
zkpbtGGx+yuhRMFg#7D`FUnSaVBGP%F%D?dW8>s28{tS1{NvL~9yp!2pCQpr8j0r+M
z=GRuHult;KFmmC3qz&_fKX%>i#rTuEnLk_*#^>wHJid?5nqmH$>Sqo+8bW%-nW@aL
zlzctN<|%cr3l;$KD3g8I{(sH>XWvNg)Pm>Oi2+9$e#c5deksQgpPu4eckC(50M4_X
zJ()>6I><|Dz1NPIkdvDp%`odV$Kd_g`iH
zq;}M-_M}(eS8M$}$hAC69jV>5Ew16@3io5J)wfbTtv7zVNUry5&5|>a8!e1^LVT}3
z9j;J(FpTc!LwwQsc_*pfOeg=Dgb!Ojbj~dVov|P+kn&gd-&Nh|{JH4Rx-7(+~;puA(7
zvma0AfzLO&Z*BxSYwDjznTWq#(?9fH3*JlhUO&Qs=$tg9cv?I_3m?
zx@7V;T=~Jh{sluLyFzbL!q=+x;5#pQ{gLAI#DK-ot$>?qyi2WKp#T2zxoO)#SM0V5
zRtG-VHC=Pd3qzb+og9~`i9F`7t16Mzh^yW5HpZ?+dmHZrKYFqu-Xm8xC`hID|mo=fvKP#rnuQwYDEU%M^IU;@UM;;M0+x#=e=3
zdkOXXwCa<9mzyk5F7{IgE?GPwGHHF4a5O!pW+3GHFR$Z+N#O8n7qer4SI6`@eLPH#
zbCc#*2CwZ0JgKqDJPAy)loAIuA!m1N%9}M-c(gPdTc5zY2C+&298r8}e+$$Mtid=c44PNjRs+
z$3Y?IfhS{&y8mpAerrtI-5U#ims@S`Y=K_j!tG|XeuM&9sR!1Bp-r1E8o+xkS0yWJ
zLT~EMq}k!nyO-QyGp*aKVt;u7aIBGS*#5TA%d~3Mt}E=c2;Jdu5#@Emwr`|$IdGHh
zRd2L=p~TIP@QabrR`%XbcQ1D%zBV(I;X&)-K=rQRFPa&&rF$q}tPeaP{5Rs&=Lo`m
zJvJ&LV~6~pO+{Hp@DbfpY#)M7m|`D7_SAGxgX6*d?^DZ$)BCL1WBQv};7xUcRO#N4
z_K%TgNzb6^n_twP@@T`JwEnwAB{y6Q`ZMj=`=g+KpDzF)Vg2P~j3a!tx?xQ=wcByo
z{zHUc{ur-9#KJv^S>w~l&jcLX#Qf49^+Y5nfFoTa@rzE#7Z-M7dCF1!Pa~rSJap6=
zKn3`)7afS>y~;~&L3-b>-YF+}eLKlu$}`D0E0;A|jd((OZ!auh{(OU?7xU*^d}lGe_?5Oy&wgu11TdWc
zDl}wyiG=hxrnmgQCDU8pni6hUFC+N(-QmF_5#VJ)(3K-hFKkQ@(^H-3#Nyo>4}MSn
zrH?toW7n?}kOJ;3^=0LgyOgtda`_kmAmod-bG)|q9LFIcjy!&DX@#(cew1b$V0zjW
zlmL|r*W%GELU^e13E4mw*vuH>IB#fxvDb4C;0P!IIhuq&hbEBTi*W&pmr*6dl%}^Pc`R!
z_X`*%_!aC29q=6rH4i;SF|TJ$Fc184V@x0P8piSQ-a(Wme##C`J9}}xLw$I;<7?h-
zW)UCH*F;P&+WBys&%7rGu{9PLkg&vEA1R*l&6$$yDu
zhH^(=-p_M_E#UpTbYZmO;WW11BA;=b6Lf;x*LM-`ceFX5_g2&IFkE7_g+JG&%lPw*
zYRmE3l$bEqUVCf$f*R*F58N5=TUNr>k@?dPj5lcP&dLwy2VW<>j1IGzym2$SiHrMX
zv(7V|v-fU3Zj|7j*SaAWi7pxNkga21#p4Vwy=?Bq*8ecwC_FIudFLW^dcU%4X~ThY
zz|v7n?~le?OcZ`E%+(GPg}l=~>q|uUBrn54LVhBB1grmg|Yq|3bXGaNL9Z;faoCaL|D7KZXTZ6D?a&g`2xL{~htt
zl>Euhi>>_0PpwEPxk>G22bHq>ll#`_Ge0{g)~|;A-wD@kPM)BJA#0fb`8N8PEw!`4
z;Pqgt_kQEU1kzI)72AdGEebmCpHqKLde|N%y<+2XmX}MrVKScj8|v6}D&_TltE{_vFU^5?(HaV|uhO!aHs<399krh2W{gkD;u%ToGI%t5u4T36_|
zx$ay)81c+VzoCH%%I&{6^l6CtyS5u&Qb2x1$0}MO6mi@zXa9^cBlzhJRnH=!cgdp6
zy9eq;9(+)=9p|HwP8m(ZfpebNKe6%DM%?wN9{dwH!C>~(q{}E@Z&>V1areoKiTzHZ
zUh(h5($|BlgrV~%yxf5P@?Af!)hPtdY*&494C;;g{;GuV*@KD`fiZU~h1BBy>eFG5
zWusZ
zc;Ku9qxRRSVqIE|8>n{_&!tW6)HPE<-~Z_sWQKAx#e1>sFs`(vo{pO^-%f4Mtf6&3
zHMgtIBhXIUzdusEEfWkHU(~z>+^m}3=;b*)uQm4rza4;ne9G>PbUtq#G$|s;9P{1A
zra#@&t#f!LY=*vv{r+2VSO*QtpPcB9^{DYvJ?w-t{7ka{wv8R}UaUSD)VLD(Z(aJK
zVbDwZJg@Z_%)fSpW_>>N&AvToOz+qI4O4Mv3iliX#@i9TQ8R0iP4}-ac^}nX0QtMp
z;l+2sC!hZEkly1~t}=EnBU@paG3m`#bpGfLJp~n`hLk@K+}y&I-shw4Eg9|^u>Bz}
zpyGY>Nvnv3{mw4;`d`HOw(idL?Et;iE!G^NcD`&WSweQXEj}Uqg1zDWa*}ePFAr#s
zU&z3FWcDD7_9IgU9M7eF)COPE(PZz({lD?$0C1+|BIdUZLxNbmx!d|C&-*{ve2~to
zWlFO`$*zZEvYG$glua9Y{GLQaL1&V;uT5=F`?k2Ynwi8G-VQ-#LMA+lJ;3~P;jNE|
zwD3!#Vjt4I4T}SQ?I^F7x3;7`$)kF^{-pUSy5X)y_8bq3yB`R?|Dm5LWaqw_>2Bn2
z=lhNROgQ$*jPNqjdt*_J#1z((V>l8N(0dZY;?B7SJl=8Sp9xU{Jc&p|LZH#8}W$WACQMW
zVeKr{r-U2uTp^Bw&~B;@x5vhb?QjX@_kDfc
zhO+Xf2j4MzOV064=0mRMca`J3QK5FMd`#S7MziD49NPPIf%jXHhG_-gbp;=Ph6`RR
z*caZH>y_@{?TtS!;>Mn%Soy?7xON3S|7DceLSFx#D_cLUycaY4a$N5ldtN5(uU(_|
zZ(mhs{cI>(%k&~$b~3%(m&e{vd5}v2TNjTvk7DH$Kg?$HrIJIxL=6AD@IBL;ovecs
zVfeLmgV_42?PhX0l$6H7;dU7n}&(RKJj`-rdK_Pei;VmQ2xnmo$k{7#M;X+
z-Hiu{ee3PMY~2@s(!)V6@&SJusFMEj^s_H%U3&NXjm$01FYAJt->EtnqDy)g-1?3p
zy(ew7bm+ayZsxv$&iT@MrtYHh@S6=;zG|RbL>B37ovXNy+Do6X<3thoZ9VoSkY1Cu
z*AGzn0QCiI-&ES5+b`I;O6`Zg_^Pv5WXit_O^K^(g8
zc`J|Zd+C^kUuyw9tLQCCZ^5hi9N4!Gbj__P)>ojXGHYQ}68Ijc;#bl42dl4*EF^m!
z^@o~I13h-q)%y+fvKP!W-i-eB3bt>z0Uh-&$ao9#IbLe}nvX{wBcSJ~w%4G))hgp^
zd*}}t*klvU(-Xh1rVW8Vj-I;s_#SYD!C`~evc13~n-!ik0X{l6*hW_oPNJCCQR3`g8BZ8~N&^nBWu_;`H>K6v70T8I93dwLveg}5BBWZ0qleu#@_
zjk;Z;yovhS19zZTkzAPR#{1<
zHFCtC(-&l6&_C4i?DZCimvUpRPd30)HJkc2Y1$rfc$M}v!r*e|*V#KGU@VQ4c
z5$89QX?+MC3H%$)~kF=MC&Pzj5Z1HRj8A-}20p(ChHX`lJ{1T2C%frFcH3E^~4ZwC8p)p#}Xe
zz?V^@O!h!;cw^;kZ|E&=(D_~=^yKDWw^Dv`kbTT7ikDV<)bwfnhbC#b)4jCwDO#85
zz4MvY+Q}7sa#*7XI*%XV7|@gQqX$3w&!YUxu5Le8klqpn!(F!ED<<4N@(OfSNm^oJ$7h3{
z2ubbt6ts8Nl&7TU`7TtO^bC*Z;-{dHpXwaikKX%?gR?BD{cgY7-J$yL0#lj)7~S~V
zN8(kxH%TVF8t`jY8|wGz!1Zn^^7b?i@_GAmav9TK>V~uf^y2X=8q{9L$5zUSz>6w4K#PrTJ{rrmRcNX5*k=_d1t*l>}68*9m{L5)M(~G;ZposVq
z!;alZZ~9kV){f^W`cfKs%^&=C;)_?tvOMRz+3zlisv~rIy4uIy3$~t@DK^i+w0g9yO1zA^8GDTMyEE
zl`88>yvI8mS8A{C0F#hgkjHz~VIpz=GrpF=+{A^o#s6Yd6H2@_srMrYIlXsm4uh&Cr^WRe!XX5IpU1M%&a!M!T(8V
zgsYeE8!4vm=v+qYh3=i5pchK!B?KWZ%xM}dr@Xt0T}q%9^eTeSzAggaptW7Q8K9fT
z4tHvd_I8%-nMV2U$ID9g(mv(zW2+$Yr{UKZE$;@o!-|&=ufQMgvb36#gm!n8+OasMu=?fAH0V7)W-!R;bQRWzP}Cp&
zDm;F;;XV3wezkUf4EmkgtV`7doP&ISJGIFEXO-}*_o3O9s6S$@%5O8o4TZ!*H_AX$
zT%HzGBaiU0W^W_EjrjgfIM{o1=`*_Ap^mFA~`>Z~&Be`cu>`27R!g3GCI`kK~
z(ZqN|V3ja_gq)S>fqJ;G2Zn(;{7T9dZoF$yg|LF*R>^d?iD)nx^))xQVxa(FMwXsh}!l3&>Or#
z?G)QrTc1r*hTV}4wi(ILbG$#l|7hq<@9=C>81(kmH}-#q_MRDRTh$)znc2q%K81W^
zb-YqC^lsPLwV^z~ME$daEcpR?xWBQh0z22Zw!MAWJn#d`n&3B?alY~O*#jESmCp5@X@7oi
zQ=TgMnb{uEaCpF7IdON%9~iX$!S(}}^fs_~nZJG*0)||e>Ed{s^xi7;eoMTab<7^}
z8@F9DnBIno!;tXCJ+1fahmju{J#0`V@h6|R-9maB_a10Nym{gz1PsKXt%l5hr4Rmf
znRrk4l>MZ)PCJs}gC(Z;nJmax8!xe<_DnBS`xAdpp^)XJGj=$&B))pfO%@-cR{mu1
za)Mb6<6YH;Ge15hlix#$JD$z_xBgLWmKS+ab&>7+dTyV@^cdjh=w4Vk$#bk*B_+9@{HX+wrZEul$yAig@(97OP9l;M|@H|QP
zRY-$D{#YZ7>D%7rd6Kn{kT!u{V-+$~F5Fh7gCgXsOg%Ahyca51u8>a~fix2M@9p`#
zRj$6k_z9Ei*>h4qcgunFGA(Yg{r|z}17k@4`uW9`v@VlwZ#_ft^2>up3>Wm-JiR^X
zzxBMZjP$zi9?!lT)44HrE4_a&?H*%B5zm}6*!hRaG#$2HBX{>@^RutyS~GgjI;qcN
z93(wfXE`
z;nfaNJs(4FxBZ91WOuR7;Thy#)*pIwssZG2p^x--!GEeuQ~5yqk;!LTy!B%q
zm6xr!0i+82p;3OD^fUsmn%09Km#%k;;&^yYhz5;2_Ue@rEpcfNM&QVSdKksD{+O~<@B&v^2E3FISB20l3gdH?c@
zD~5qz|0mbd6nuW!(-aNp)uk!fH$;3XE^4u(+z9uueO9RV4XP3v99ZIi4flS}cTb$W
zuf9fz@l7b*yt7uAF=+K_b$kzZuBLxr;CtjLDr*x;s;h)Q!~eux>sc)vv0uMMv#wT{
zo9&Y9g*?UdzU31iVO%HQj*J|Fv%B~91P#5{M|Qhf9^^uPMLrd`K82VI`KxV#ki
z!57#UhZa@|3gQiE?l%UwU!Akx
zV(RE}VPbFXTb;6S9^rIMv2=8$FhE}zu<;gj$&>i4|cud0OA>h~t8;(Y1T!_9>&
zaKG4epUSFvp?;1|=769>Cv-8A?WR~cai{JiJ<;1R%a2_;UO
z?J++^#X;Q`W4;c}sU2j3=lssF%zie`Cps$V9JqmdEH`IqjMt^-Z>lK2hUYRh?b*Xm
zc#jUQ%DZBK=kP1^`iLi}ztc$vzwm%_fFAFA)4hAI6@lw6f!8gq`%#Q_qx<1Z!$YvQ
zQHRODx}n^0?BPMQ9uo(8eW!ZObc`PFhn*(f_morqpx?uhi`#>)+7k7Q?&D{-%zd*F
z{My(W4Z`aWP5jSqM?3zDR*&2UObpsXd99fjHQ2r9Idkm3l3#pvV`ekTUre9)>?6JJE5`SW
zwgvxCZPj+#x8%p_edFrmGPhCKZzkg^~20G!DuN$@V)Az$d
z!cF(;#tb0+X#GDo?t(YDaE{$q+u2~Y9?5f`W!8`%ajomezMoWh{3)313@+B$K>dBF
zZJI-T=)8W+&pYey?L_kF<8A86&Z9$>r&GD!#YaruM(HX72yl7pkqpOlSL=J5a9Miu
z0e)2P`W+iBqTN=8JR-dG77HY!s*jYmtBY@%jq|l$8R~*Tm%j)&7j%INqvyBalW2LXAGur7H
z|DC0-tCz8RI2lX6Gx>@`Wh}nNWg~2$y+)Qfj1P@&g~SB*Bm4Qi$lhCSFu8#{zbB&7
zj6Q6_bEr)Mg24A}h7Vc187arHKkB_+agXVnMe;bDo3oOg%d`u~VDiZMlURQti&wF@
zn*V43<0sywU!s6tX@%`9+G#$YfCcczICKa4r7w@qTgOddyplV@BlIk`gMhB3pYed6
zo%lQLX}56<55CZk^JpSNe>&D%S2mE#JQCP6>X
z>6!)fPUGoPzvm}#-bMjwDacomxuAKN4Jy|b@zI^8oZn}{={X`O$B#iVef~ofSoetP
zRNofpkmxqCe3+;|n#7Qg5&5Mw47Ag!KgTtxmpEQLc#e-ZHxiE><)woOsUhF!$j3FQ
zF&|ej{ckrQS23f31CLyp&*;@9yq`C8tb}quH;7T*NP*9vRUV(eF3q{z0O*^_Ymn6=
z{na`&0oX_F~euu4D(e{+Kd2EVj*Bjat}H-zOl|2)j`)7-*E2d9onBAI=UDe48+@J^gmX?KD6@d3*s}Xdoe$gc;NDC(jQ?s8wR4@YEy;}
zHEt;51wg)N;42_=+m3P4z%_an+HGryQaiX?DZTkD&peh
z-cRVhVSZ4`61uOrDzDL!rQkn(KAA`N9sBDSpWfaW{;t%CeTU*{x~ZD{W&bTMii8sn
zo;|yG1bCly)w$omXS-UvQ=ACv-MJ&R|MhF#3ZffwOe|@fg?Aj|=y#uVyMJT2R%W+`
z+3{|2Df?MBAGRG;ZJ9cbsXS&7a-uNdEQRl{JclH@(k&Kl=c5
zMwZ7fitAQ+F|FpIJnp=wMiK6r=+CW;c$h7wq5H4x?eN2;9FBQ@JlAWiBZtRK2vopZUEu?`J$DPL$j-ebUnwdwcBchL#-xe
zZW>!7oDGmQYnO)Srq<{BkbQVwO$z2z0@t~j_sPC&RV@@>8ECb=wpxfWKUo%f-P6cp=fZ4!|Cq3UTnz4m_}x{x^L`5Qj}sND8o^FC`;S`Ffcx}Mt&EMsxz^_)
zvX>v9<9?OXx~G{Qh;xg_s?hgDf99GN4+hS8q}Rq_3C5)vSrALVpU^XBt^7uWOgNI~
z`l_wXC!uh3wvl|mCn4-_(6U08Lx^`e($pM>Zc+9kEomBPo8BOe;JtrX@o(NT78
zUMaM`yixV*r%J)Y)3J~KK-@>#+r+#m9dUO>(tv5Fssz(7X9Fj;LEd2S)d8u%g_+kB
z#|*(dt@Ccu*#UW>`Bpv?C!D~3W|L~pHsHQ5?VWnGgS=Gp>4Rs$T^DQSR0IHbE1N87
zS&DHOgfve0ig8YvS2JG+xL-@J&CNZ))n7hsdiEN5ukTjz=R*7|g@y}if496szdsTh
zd=BS$Lw{P?_r>^^**>&gMEGmMF8j3mxaXg?X!v!^gWIs{8qs*37l)Q^dWU{?-?_AY
zCf29#ko5~Ak|h>-^5m-Dvj`V0h+jV09(3lDs(pPh54XK{Jc$RM?7O3{HsL(~;5cf^c&TyLlnK|hX;&lNbki&-OgKGFz)5C8*c4J{nL|%
z2GISpWj!6&4?=mH)L6Hx;Jd|ace{=Cy~5?|VLFdhj4kg;`P)qMmdbSR?P2cLEbfyTB$im-^*nNe*Q`^*P!Ww$7
zGGa%#kY2*?heK(fIx3(|$Jd~0Q@=c>e(O!jmeG8y7VdQ+zn9efPY1$jfv)yR5uiCv
z+G=!;eCFf8y@X>G|4e#A?YMl
z-n`Y%Aw=^#H^-Mh=tuDq%k!I8CNux~Av!9T>e+7B9729h=b_;~;&at%*m*)Tqk87w
z7dtor=-~U^SJ9)$PnI0-@sRlHcITP@)@eSH;ls(fHb4@HgPJdy-|ye;!UW<2-!)_L
zC%yITRT(WGF+VBH1VvDv}{&$W%hc
zOc|4TNEs`0k~xV;$Sg#3t4x`xyNS9BCH&rN?`xmub2`uSI^Wmt^*iUUr$4-{z1CiP
z?X{=1_qDIzm1(fBr`mzCJy@
ziS0V{_Pg_uuRCu-O?my!_=@jf-2W`r)edF|YhizVw-7#$Pdb~&&v%-1#W4i(TOYyo
z24I-Bc<;{1d1nsE|2ady@PhsI*bvTVkA`zSHM@v>$1PVQpul@*c8;81ueCVmvfu|%
z@T)c+V?b}I<^;#SGevw;silzfq-(3-x3?DQLAi~%exKI2I6ll2@#6H&Fc;+<$$&WI
zU+wD3=>?A{(80X^L&P=ubW8=gc6;Ckq?ccfZUD@kg60Pv!B;zpzlT0w0{XS``%K&J
z(ZL$_E6Z{G1pWi8x=>=YrFrls$<;+GalK6ieK1YLdsV^(J&@cF
zJv*>FAo=^G26RgTpf4pOy}uG)29f!o$G;N%dx~5KN=Euhg>FRE0KJaf1f+in$(y8?
zq9LMpDD)~tKRPYvb2j)0HaJbuDaku2$djRmyh8eTcA(5ZT9LEH(>>x(qGkX#!Du8s
z9(YTN=(pP`;Kup0!cU;<4SR9V=4q=FbPr_)#Sc(U!S
zmX%3g^MH{EkTJhq>WcyqR}8$&4**9kKavu#6`#_yBGkHB{WVrMDPstir
z&QE%|pO+U9-G|eaYU6kT`P&CYW)ps#w4Kw{0u4Cdt&WgiU26`f51l{H?WxtOEvGlX
zzR2b84d}yZ?P2@TL-0JJn-jV)?!EgSTTk|0YPD?&tuI$S^7(Vxm(%7%<$!KF!M8yG
zP+f8!4S;ncBADB&_dFO66!0ATV*zFi-1j}~+K$dUCMN4}KO}xFH361cQd&+jvVq4!6;B0KX7?gW$E_sE(SgyB_d;7t_7C%J6~wFW-vcX;J>wVTlDZo&w?(|3{b^d(58#78OJ409+hFE||#;{??0Q50zkZ1FX}u@2Donf}Cj(08~>{!4=xe19z@=fTS}
z_})RxwJC15zq04+wqB_B5%0BmUv=n;c=2a^SBLIjU;NYWc{=L3@@8ve`i^cwrAcjGqWrU6=QimB
zeRaPsD)Gnp^_VHc8xH_IAoi->VZ>+a_PQRhM!iNf*_va7d=oWI&XfIBwp28&f%NV%
zcfB2f4&l>Io&p-5%Nq3xap&ipgC-WBlXPqOOvS!v@@>oQ4tO8Y;M$+@9k9<(H`r70
zJnWpf@$n(r=j6T8&WeVeH9~jHvqC(q_3A|RW!P_c49e)0hVtAtcnqQM$sH^A@I@`i
z&;EL(>R{OGG)a?aigwgnUvmfTvv9t`Vnl3VSm(VbE^}O
z+nCW~(Q9Dgt9|;k-<$pT!HQeJrTuM22Lg|F4PQGQ?;X5&((ii{eCI>|MVA|Qa6a4K
zdfShV@Z&`%ZhaK^~cM5#pxox^kf&7HTRU_#6tl6YqFFv4Nqb3_YYYq8N)ADvd0zb9JNv&((
zf3JH+hsH$}Z@r5nfTc}CZiNHSwD{hC9{34E&Kpp_TeGeGFX~sbj_KfsFwvfNk87U5
z{>3H3c4h_mck|(|zUxpjX0LI23wqO|TI+8k|Hh)I`82LmxBZ+>e7*j0_~k>y^XEDS(zto()62{qSi9vO
z{3I5hBky?IjOM%b#ie{+usN%#Nb^J1wC)?}Imr#DYL8-pE|IN%lb*9bD&mt+IJa82
zR-fc*t>3(N9Wdyr`;p7Q3A@wt$^IqHToNgsZqPPrDxI558#~yT;(vptmJdizwTTW+
zq~~x-B7aU)F(9B7%{Ot{*E1=eH@IQyL-sDYt6zu8nS9|W77DzFHd&lMj4oY;15(^0
z={!lF%9+)<6`z;Cch=+c@WPeTA5p$Wg@ca~zMWg?0pWGqeyd;1=ij!j-u!%Z
zcU#XCn$ODLx7?a
z=+^z{Lk);WJihUD;HSR@kN@Z0&*SUXvEB&x!T+$V2VYNpwTk$Th6bs4Q_*;Gu9%mKI9$1^XHs|d!cuFvWQn*
zwrIHAH062gajy@8zYfzs^2byYae6UM`zha3e>4lytAF+50Gd;H0lA8!1l?Iz*kOmK1C(yvMqvGT!T;V-(9@2HcQzNV$1
zpDXcr-Lt~Zol3s`=Y^b`vM;DhZiXG_Vg)wFZAhZ|&ncsPI;AJN>Mwzxu=^zXBoz<7
z3jvV-%mvcdqjwP91*3_Oe|`h$cc|LLr(~oLROpJ-3`pOml+#+tzfIvw{hWjxKMn?R
zG3dEQ*+ElONqRX6nv9g^l||0BIDU(&Mt=Dfgv
z?OXU~;gJ8T;?I;7r1@XJi%zH2RIfd}K%9Fh{+X`qTPQjwd5L`60eOEVeLFV*=V}Vg
z)q!8u9@-61Ika0KeZ0AW)SHRs08}Y*9*8L&OM9P+(=^b8Xf?k`Sw@{#Dv%U<^*
zIx5+yS=J#lnis2ksf`=OlkX9rYiBdQ~?%h-knWp4oQp(Y9B>QGE@Zh;OI&
z?jXs}89eDD#W|ltE&2DU$Aql*qW#I~-{LzmVc};#QvPmT)20ww^c?9#*em4yKC-87
zwBIUo*lD7+eM;|#Bz&p(a2@j7-aF-F2sC$bUQ6+1!RNE@+W_ynKdCVuXwl%(ba&uB
z*X*XiJ{m-k$WH=K22Z4^bR=
zrgu+Ul6M$#;xXMz*cP}nlJ={r6DxQ-hMJb%rnqZMw>jCgKackLVorAIU;TWG^f^bE
z|2>EF1JmPQ(*4VRwJMc516Li+;{FdF@`?AC+xvzrqkZo=i^9g#e#QY$4${4%>+g>7
z=kYZ^uMMMi$&WR`4+&yl_{*g~l{@sv-3xS{J$daqp~C>$tBTjnB6xtTT-Z-wp30sozZ8oG~{G*rM&8T;l(}9P3Z@DH_tp
zl5o`U+oMVE;BVm<$Af?6iT{slz#xm2Rj9pU)vc}*|9(a$fB(M1#z}26!0+<0?|#Z3
z+P(A+@vT}6Yft`cx9CiJO8474r!hU>8*s_^8PUr>wQolJ?-nXPjw=kUf)xPuzinNf
z?2LbHffH4XBfS!S&afx?3LnS)vrb~^!?}p-mMv6{m-YvKzjgIP@jH-{zNBI$!+R37
zmUDf~!xfPf{1%w>kZ_6JTK@jPq^t&NlJoOR#|nt~;mCVTcv#0y`tf<=@?Jwe4^<8i
z;N{kukL?EdllmL+^QmyxLcZ?w{80fb9_GX6xV8^{`WtHS`FBR%XoNg?t{#BT+yTuh
zxbgI#sW|3=+>Z1xKHp^?GUa+FEI7%}$%1}pasJbVYdQa2i8iNmIz;h$-0B^S1qjdG
z(#7uu%uxO0a?A9ZbNwTqb>*0mDDGnp))#Skj5(@EJNT|QNBm_YLkeE8=tw!^|d*U?GE@i&Wd>Y
zf-VIL*oS?gzyRm7x1MnOzgDJ$SM1BxqMXbFa3|>P))cs4-5g3632=Yuji6o6K|Sae
zwEO~hmOu=0%~w&Ng>!=If$bDA$j}-csLi3-EM!ra`>&k}|3Aq|
zc^akU=b!&UpR`}Lr}Q@$hrQBnnJ(*JN$H35b!gIWrO+37L*w}?BMrD`%nd~RNMexV
zpX)~YBxM{lQ}WCG4@KRSueBmC_YYWnc>ZLC&xbkmsT7)`W740llt;4$(Xw7Nnu$(T
zi7%t!c07uPeDl9AqzNa$V38yLR}qRdwl6@5z;c_XYkBu~GNLh{9v
z1Ue%sqx69;k9qp-pQ0(<8OzJ-Q5
z49c{!;IO6Mf`V)z+t^4%^;ZJ`Wl6XFG#gy0Y
z_<~v-R}CT$;eD$H6rK=FOgY%+JYLT2Iq{Af
z7Vj1H6!f%4X{Z^V*JRc}&qiG1MI#jZi9xPh{@zTn@A_ci!u1^*wH_^n^8}j+EX=r1
zJ}SF9#X0-79{5D@X8gQw{Jzz>$tR2HeV*{I6YG%Nz3gnPEQr_NoA0Ik1Ecr6C49Va
z^=OLYoHx-44bD#XA4Z8G@&$IBg%Q=ja^=XA_z(no^(A1K<7LMSl1kGG-yk
zKP_=V#x4S
zSTS*)Q@>vdrFZHw3=M~KnurA_2~P#o#gu{bvydsyBo`9=9vv6YIo7C($*w~osZA&z
zO%2c4NBe@jxYSUx(Dx26MbY>Rh+Qkp7VKcZs*99Q@ou=
zXKzLbf$#l28s|G~$>L%Rdw+S8pGiK>XB5ttg@+YL;J1
zcBbZeZ6i8;M0Zo-{~521sTO|U*qhU_J-)b+onyvE40AIDi*uejgW@1+aEIHy0^
zo{xhzJ<;6A&rZ?GgYSj7ROb7D;Ch{S`}K6O;pvl812`7oS5`^BrsI9C=UVY*-flH}
ziFs_+R53r^T`AsMd$WZmJj`pyr}2DuHMD0xT43QX<-2q@Zi@Vs@byVbU$+zh-17_OAm7vgu^;%b2hD->F?MhZuyRFO
z(Q$s)Smd+6e3;Ms#|kJi!}}LUD5Ar7ydmQ1Mq@;|9eKcpb~_;G809?Ow++N0AA3w-
zJKV}5`b9gaMLFR-fWUhMJdi|x-;QLYn;Q!|(kF|2iZvTjlp%#`P5R
zj!?nhsPHu^L2q&rb`Qj8q4NFKi}ZfVdTMZ4&=YM1zb+j+l6~YB$PXp}_E6Sgc`teZ
zRR`&J34r`MHuBBGs3tnSsX$siN#8;Rk5e-Ey%c)=ZNaysZVNejnnQdGh3`)!=t#_7
z#Q&i9jc!R(dM!mRnMlxguo`gt6`FQ|L~}79CQ~klNsJ>pHRofHa3pV$9xQ^KmiO9A
z6`Fbj@nyWq-30ogl3rgynP2kdKH#Im$E1fp^eaU#2iTXd(9#ZGU8Hky$V-&+JPi9c
zd?_bU)?dXs>9TUN-m)A?
zOF3z$q@}-QePupLmo2X>O|uO7t!#Q(`hTiV+Eq4xS%1m?lA)B3AO?Sk+Y~xop*bL3
z`hgDv(6YbE{f6XA`LbBH{IYy$xAbROep#At+u^s0%Dl6d8-U*%9lf%5jC@f~V^#@_O9RuZ)w>EB%N?4u4qx6nVL?lkV1c`Op^v
z$8{6>k_>4BgZI30$>X?Z(}+BRd4Km>SdRA;K1X9^!1s4A@QRi~|Ia+C5cUNZ$WX)$SA~4PeF&!@ufG=nT!tu>@aPPo$9|qzf4@$D
z!1L9OB?sa;3||1gu`T?72>aE1+UeuoS5Qrk;Xj+AAiVF8O)m`L9I4wq9{0pm^x|==
z->|oRC_Nir=^#1X8uY>s-m9xMn&+R@q&vqM4Ub@hhI^}rusr~l+8J`ZGR24I8y4{e
zjzfIz{TN*z?-TG1A)XKKRC9iyl^K^GZZ@A|ott5(7~&MVUIC2w8_VT3R~yXnM%&X|
z-`B!yZqJhQQ3$B8kGIX@@o>|rvr$p}{>TV(uBY}+yuOHXw;h?!?X0)oneV4Q1Zea9
zOhy;_B{jS!++Z|D6y}FTMchs&hf0_Vah^U_m(!Q^=tKkcI1$Oq@4Y;j>ksq3#j(~O
zD~@)dySUw^V_+8a+4d2?W4%6;IwpSK;#+Mlr<)zeY5zB#9Q*I5U(Utzoy%gM+B0n@
zwP|r~NPS@*h?;;Mm+rj6dg;2izX7
zuJ5@W&Hhy3dRA%ZA_v}g))oEu!dr2l#n5p+Pe0=@iR;-=S@h?UYN9{d**S5$o;?-6
zN7jA_UN=R(epE&`1I9N#!1HHhyYg~ZXNvhNG7Q^#@SWGCbAFCC%t850CsX8zxNwV@
z59Tm&52jXMnvifGjXwa#eXHdZ_~Lx)x)>jal;44S!UIM;hoZ;3;IH2-+HGryKF|L-
zQN;5_xJ^p@fk%-Hy|3Dd@iR6-jC-def*)!m>}rIkZzR`9`QG%q7b4#vdW-=*$*zL-
zr)dassY-ku<^VSMy*rBZGkifoKTHwn(LDt&r(uA6b(Hw~3X!0Tl=HDE)SRG~Zxr(W
z7;U6CgQhE_zvl}$&J6|%J15X~4g5C`1=iyMEcOf3%t+T(){Oy=1ib>Ih4h3`b-}Ns
ztOMaZ5XSR6h3`vj@HZ*Q-GF@b)*a=yRPxDn5Svb({*K&)d{gPM2hf%P$lZ^8brqVs
z5p?cN!GEgc%UAT^B{-gsy908k6nc-s=hcEB`!iBG+U^xsfunNOA@^?p_4B%iwv`FJ+y;mrYBrc1v{
zep!_AWz);j(mvjepkLr>0^pRL7x
z+4gs|qkK@``Q_4j-UOB(pK!6#&1_~^pY8O)`-RWqR;(=Ah;z6g%aCAvZ)4-9_thp`
z!*ko@XI*qA=dw}r;$LgU=CQ(~T~5BkcfX2_G;ZVJkFEF9KJCZ-=Kb!um4Dgbz0Qz2
z6VG+UytDhZUk#M&p}n=rHux`KeiM7wPq?RhZ^;&u0L)XFn|ogJ%43`A_3hKxJ(u-7
z^V+jUW;XkK^Ng9mTs9Wt=i(sR{0sM-ot|I!(x6H&%YC__lfmn|(ZDQRPQj?AEHAxs_NB%St{qCu)BlTd?(I$KEQ$
zbDd4~n-NA4zHKV26pUjw;J_Wm$%HhUJo_<)vC+&pu1}3XghE1yuq2Q>4Ml0kHwiR
zp=AH<wjmm(S8FCRtd~xTc_y8+SJZul{{Zf7@VHL
z^4otqao(>?`hlNWtVL;6gSFkWSm03>zYhIgZ_I>_
zMjf-+<@efw&+BKgWkXh3nlw?fPg?OU6VGR|Mfs807O-=u-MxcFCYkJ3xsO*eDrK^d
zZyNl#Q8AOPa@g|wUj0m#()rhcZk;ljooQ$b^FEoZ+o;(eI(cNW%=wE>-E
zte$N+=Rh{g*sK@!&^V8U4)pSNKZSMT;$KTY+|zIUvBFO0tLPuCPQLoERL#cU++^}2
zFq>JudU3wPp{!?n4bx<>{>IK~|j#aZf^WU`t+NfFcv-O=8w^Oss
zxX1uU^s}t>jmtIAQL~M4mu?);Q?pIWp1EATlgCzFEMD*7p2xaw-0ytWCXbE(Xl8se
z4R-72?QPu=_41y!u;@b$n_DZY&bb*mtk$jjH
z|0F*_%^EFV+U&_$HS_y+FKm5ZHLGT~!`%(zv2n!>RY%rSv$NNRr|ouCv!|8s8x;nt
z+4Zx#X6oKjvwIPh*Y#%)NI_=f^i2w
zWU}yPj%Us5tC@Apt_B~#U)tnVD)qrf9z_PPJbg
z+M#AUll6C9IjUxFcU8-K7M95tJ{rDf>9|bhvGz+1XN}eAJn%r19Cp<^ebcAwYE}^EmOAULnq5sAx@%>nEPN+k
z_fHtU*K{Ci^0ZeQvRLozw4v(`sM(3JdYgBs)GTVu@%z8+vsm5jA0kf2!H%l^lM}yV
z-0C)r{__U?ufen0j%b$_Y{0U76E*YrRxrDYubMHRUkxIjshMs>;G<_xa@p{zTZ)2R
zbJ^B4!%vva$ze0A9&VD;DTghN$cb@s%V9OlCZ-!V&tZK$;twaKX0uiAV%ik!&t~N|
zM%TZPmCb&icyjK0cYHq};lpv0jah8zH=7w5iCN5SYp>n@v$NUt$9H?wK|6j}mHeXT
zs2nyX`C;psg<0&WzWW*0E}OM59=S7TLpDq5H-B8!(rmWLtJ%_BeRJ5FoZD~by;3vn
zjMJNTn`JSbZtZ$4!@6nk+tDv3G@Jd>(wh2Joz3pn#!>l*92V#maeo88S1@++yluv3
z^VoCO<&OI??t34*AI|XofXIznv+{f7u=CmNTP@p?!=BIIl%H}Zm#u0zW_l$p+^4+I
z`&>5Su68H7bu=BD!)mgtJ0@Y=B?bj1D9
z9sL6)jR3#Z!69+BSRYG`7H*&BgZnxzvmXUr!g{&zq~`#{fqs)eZnwvKR`-WJ8!@XM
zzOP}`;eFQZJjR~Xts7#3-v`U8VO)TCC~WfM#>0B!`x>f}T1B^W+43qo4hOW(V_Q0u
zkMYXLV`JY8buq+z@49keVItngs&=Fm8*(<66}j0T{#6I-!MpsSr7?M|O{$GS4_A7R
zE-Tpa1fJVA8*_<%=O)jt+hAS9XCqGeHu#;1`y@LItY$vJy-R}@Uv}fWU;)h?UoVH9
z$L&`ScWH<7o!f~9M;x%PE7#sM(+2m-3ohOmribq&It&jP0Xr=&FED#DANObj)RTX{
z!2Y>q6SDxDe0FHoo}zuL^4VC8!`?B7_io&G4IgBU=dIetRaQqL&fIs?dr6&qR#0w2
z;%1i5RHdn(m!HIYfWKYeR)xQ^clONbH3IuytJuP^q1d+tmhYeQ6!)9HH#%yI_XH!G
z?VQ^ixc5{s
zrmw}m!J^r`wAFaete=j0KK36&zwT`qy}f|hPPK1re6@ge$N^GE^iEer{8exJ|g
z7gRa|LM(6jm(wuA?k_ez_SdkYm@OOk25Z>2Gn{XfL?i3Oee>5p?!0juQLszH-a`OMnviRFL}1+1~p
zXO}Un3fPy1^IDyOJu0UW8*1DrU@zbL)*by6@1<#aeaRe&=jzvd%&+drXPO~vR~TDp
znCspFKbxm(*!<3&UN&>dXN|*F?yIX)!0v4P*uAr-hCL~N_vdB(Lgv1&g862vLbmsX
zrq!u~8rE>zIBWN-8uruSa`?de8W#WHwr=_p4Xg0OX>rV34SPLu)$oaV1+1s%x`c1q
z;Cl}B9^s1b{W+LtK8@C}()?38m0xSv<9UwnlHv>4+T$+4zD64MC2j13ei0frrP1Ch
zYeF^bw8!&iH!eZmKk`d)r9w7%u8sLa_@jdLQRfXkaQ~xkZ{uyFF>WsESk`pH{ezP|
z?Pqo`WSv9PUOMR&GV2OPEeG2bur)3N{Z`Gy_hFhmYHx>qj8D0PXRCbBu*9$D(~24u
zvg#J!$E9p8WH0Ku=UzlTe}1f3-6%!FCVFcYbbSxG@2yPhs5Pv`rb>f4{L-*aKf>=+
z=vBxjEMEB5&8?7i^4(v#>pBe!-+p6SXN`tk?dJQq_X7=EF{Zi6)vN+$c?HA%SOK$g
z``))sPkfhD`{Gku%(r3L#c{z0^Vl1!!pSev^H}evlN6i&F|WzuPNTMT^Ca1^abVo+PkTB%3}@d@BZQO=a-->m17a;
zS2I5h*rn4~!(vsn3!98AWD)0P9}96W#IHyNm%M}hb~U4Z$D+Ph`n+d34u$NF>*@2(
zkp=kuv-s`H9u}}BUW57!eOJKteeK!SH>H5N|N8W7Zdd{P{B2mtp_+KUz1sV>ZdJ^$
z{c9N#lqo%nDW=H;F`zk(dFZw?umW7-Jo%2SViW_=Ii
zJ@FBj-(H%5bNj`O*X{2O`|qxdJL0He>rFf7?&_~$8Hoo9vb8j9kKd}4($oT`fA?}1
zcjrRZ`eNH4m1`l+i+=A|zpaolt8#9p4+@!o$G0o~JTGKBrk!^6JyF22-s=}Vx>dkZ
zMzr!CkN)_kcK1I#G@xP%zec!>}O@rD3Lc=c9wedOT
zQOF8*rXF6l1^w32e!Ac9eD>95ZKt%G`27{DWnC;8?nkSg=ex({u~(~`6g4`O$NU!6
zFa2JnkVUC|tDjz_VF`g}DjjT6z@n$VnCbrqzuOUEXXF})=P`}vJn3&*#MJW(=en2|
zv1cz&9qepX#GZ#wy>~sokeP)fU$}p+kWKooU$k;lA*3Ofz2)JVg>194$Bz-K3fcS)?N=ItKW=Q-OV#=nvcA1OoIZ$gxVcJN^=>yb
zti#*E3$+6B*pQrvUR4A0n1RoRUhBj2SbV2ng?2CUm_w!e=dF&zh}}o|KpXb`0nIz+h-*+vF}}?Td)0%+$8?Rx{FN}{=dp=}EWgI{!*?!g
z*xNZ7t&=e?xkon7WSCd)tSw3X_^FT`%YN3|y&~Ev?6sjTD`K~NzHQw)tB9@J)BBF=
z!Xnn7#NPDVQw{6Ae%8;ZaLfZY&s}T1T*E3wT-?*cN5j?_?a|}~XxOHL_fx(jZhz#R
zGh-F@OHH1&iE5Ob$4YO0``8BGtr}jj{T8D)c+aGRS&O^}Y5nOik%
z!=-w2XX1OUd)!)Yd2&v}v>o4uUA8P_v3}X+af1ul>*nrmSIzU;jmLA|u6PN5CXOhs
zg72?u^{i%8bwM6;+TZR>hon5_J0$q*sr4Fm$mx8V#TBf}R~IM0ZjjGT`WavOgL!(?
zsE!-k;C=hG^Dlo}jQ-T%k$!8dZbd96P2+n4{ytPI^k})i8s<`E;H+Aj0v7eN!{1`e
z$E7B3`YkwC$WDDrpQwAZkd;K6&#MCe#ttpYG#gaFYM!2M9E^GmFv(3lHKmYUdeidq
zHLOE5tRuQF$2#85KgMCf@LZfPixcLRN#ew@kAVN{-^yCL{Qu?Zl&i78eeN{(i8H3m
z8{ajvvLaDVs|sCc(-Qv>DDvgR3GrS*YDugBK5^={U|Qmgx{sD=4K35Z;s9Bz5>km@
zzC8WzycQ^-phCIp^dJs@?{0HY{;8*EAzoI!gZOGs3(l#)H9?oR`^+g?^-)PW*%lhK
zv=(&L(xT^T|L6b8EB0#9>mC2MKT+1d_h+{jj}%k>|MTa;apJ81f7hSnOX<(q`2YVm
z{`~j;GFeZz&gk#o^G#UH*f-w9`W2OIqsV$*7m)+mw-$e3`F@!k2bP
zzC>9+sZZ8R@+EfW|4}iCQcjj5{aRL^^pDgd>mlV$6upv`auPc#d|40aKdDDzS$kx<
z)FV;aBlSpor5&<n+QX`TkQqqcP@q9m
zfwg!CQn3Zqgk3*x2s{=~EvmwI9jJmT=5bii`Cdkx-rDO23RW?lDkA;I3}MHp^3=jA
zHr#`Ho{H%Wq?S@)zfToVv2Txs{=${ipeknUV#RUgMPH6Tou~pT*1Qt+5Ec8qk$RAd
zP1#G1M0)MpyxfhpG3!-&FhIp3T$4rqwbWBo?9^aUevoTRt|w$xJWsbRCdaB+pv!lzr^bOWPS4f2
za5`-bRSf*-OiouzImP+&mV0t~dj)&0Z~O2BOa&@7sm%aXK*jo841=RpY)HB5+^z%r
z9e6(9S2m3(-6qLOmr(WoFqcc*W{n2x&k{sor0dAVi{HI@^Y>|>cFu^vngDU
z%23qjL)YP){xV9mLsvGC+cRAy{Mg${wDf2DDVrvWQaJrKLe*o#0LbPN?Myqz+PaD0-}Ji?519z5T8Eiy>O
z{NjIdeUJ81hsHQa6a6OSv*=gr9fWCRx^bCeAX9f4A=EcWLK|Ei~~uhb{=
zw^HnMQRHcMCOc>p5lTPGe6k*#{g3h`U*;2r{XgyF-4pqwyo^iA>XUM^JSi_x)>mmD#
z^s~$_%aQeydUzP}kM@#!rQK3q+AH}Ir5|Mb%JQXMQcl_-+f~X*yJi0^D=*8J_Deft
zdD72vJW4sKN9LFLWVwp$C-uwz(nd)SQt&WW{I78@{nJk2A5d_SBKM0I
zq+*Urx~yLVg`Ua{#(JsH4HX=zpuK{lc|nL96}_*O^qGqOT?*~a8vt=CH%P_WaL1@v
zKV`jXs?Y&eN6}#6@q#M2ya-n=+AbwTWoh`g!RII$R4s0tT
z!&I!9qHm|d@4*`Y>w{u<-RnYMEhTQyRqX1^8%D)OCyR2gzZ3eF6_A5e?6ISuA8Szs
zRV>Mu3{kO#uOGq#Dz?~=6d>N>3o7;_r9!TBG*w8&+~?EALd7a4QGqH}$4T&WCQ{E+
zv1_}@F)G&Lt`Eocji>^M?-x@AfysiORZh^$*S_ZZUL{oK{B|X@lEAJ+3UClN>QWCy
zeTKi}7~PQq5A4%*6FHq3I-1i9CelK$V$ME0;c*o^(cO@zS6p<7%WccD=Cno8cNnZ<
zQ{Qh#*Tnvy=}RRx~H9
z?bLZ3;e~l!<`O@m*XNfez(ITK*P!%+x2JKt9_)LJ9;RX+mh0z}+{1RZ<%ljF8H_0b
z{@rp4Q_3YTLdD`vJoh3z{N&RoLf19txn1#7uW|mdDwiiw`nYu`a!H>@tm+Mw
z>wW&bKhcYlI^QCGYrA1?L{FT$t_9I)hjj8SfF~2neI;C2&gnFzcQo)Q=?t8c?){Ph
zI}D9m*asMGvUVo%-zO$7B%IbQ3sVC6SM4I+9*4pgqeH1!Nlw`my7BUhm9Wgb2FS*67Fy7$Nm0w{1uLYO_yS1s95UB+m}p%r>%9k
z->T{EM2Ew^xP>3D$6~kCJ481wclNmvaL^-JDwGBfURdYbJUP8T$Oh=HYIS88wN?Ut$S#`!juKAeBzeFdJL+C84P
zR|6e$Ucc}uG(llMch-sX7u-0<+w1htcDj`QyY^~C{P2ILWZpl!8f5T%c@-S^zN1q7
z)W6hz%X~ZraJlSb7*Hyve`H~OLbul`TyM|G=suvI?xdL%=MffS9`M*L#`A{OVtne@
zRps&_om{!xnb9C2UoSymbY-Xk{jS1UZs)lLV*kHzo#6l8AoQFL#xQ`Kj_oxp73k+x
z9&`G1^`;zyRZbiyZAEtmKmWorK91Eb8}fREZ~w^Ct50vjab(vFjuqF7@;Ch!*p5&1
zDt4AH5Gqy`y_@S>Cg!I;Lj|qIXrc5*7=46Z=q-fpH7E`^wovFZuP4$kOb~dpSfu-Q
z(dKgZ`wM@0E)jgcd4j)0BkI3(n+}(Yut0Z%-VI}g{vti$m%d+x{4Q#C@a+}Uq3ob<
z5F2=1LFW;Iwov#l;C13VD|97APJ_`*>2rw<97fYW&_$6~DRk~s%BW%*9l>v?l%Ge<
zh4gwDrG$SMqT2&U&J*&l;)MQ&D+K>`Re=!%kQ=3-p@L|1{QWCF)>G`{W*{An&H2;|
zNse!8fpj`RG&(UyzO4g|$(z$Me!0sts#p`n4}2Mh9Q78;&&>q3RQie7@Tu5+YCfcY
zR_fhYDUW&!rAzy`+dy-(fs)UcbI{U0o(;4_S#N2ltS2|~A9CCb&=|D%Q!z|}9A$Z&
zM7q@XMXBd(MXr^C(jMMyke7B!`(?S3#$<{=jQ=voNu*1Z`KT9=yhO>DDD}$v$aINi
z>m%D!`dP}!ejxjgY;R1K`1@BslJ%42N#x!4ANEVVvV2*d>?e{hQTj#FQeM(!QTk7|
zk1R*Z(P;mt-DUMj|4RF1{{K`DK3$=Hvi|?6J$M-U4}VGhvYmK%3I4(|_R0FlID50g
zpRV*f9u^~??Dx$S{#FH3c?DE#s6tCSMk;iy!k6+Dm2_7Ht0-8ipp5?;Ddq1{^2_u3
zS=E#kQOga&vh_Rjh)tPBq{ISjAG-
z(}aXLbpjcL{fD#AA7x4oP_cD-BER30N}Rs^O7M4&7wI=@3*4VZ4#0U-{R{}GSZ;G6
zS5hSS-|7pxC2v@qqgid@@>lMOa_Vad{*?kjk5T;F&oY|pGkY)iclkht-7|&UJ7cOC
z&I2Pv95cqB76`P%$}t=tpQnWn^N5XzFU~Z$hOUJ3N_{%{LtI`Q$mJ3boZ+-_W;Zwj
z=kmdGxm?so9~7iwO@qBS|9Q7=oOV&yM}<}F=A9I@G~%A|x8O*ud)j+Af1>>Y&Y!I|
z!BD`tda9EXVThSAJg;Jg7Lk~$5s$Zt<8eol&3ztk7JENIO%cDhYGXtEl!y{^d7PtO
zxjTe#)Z@xLjy%=oS}Dc9^*V3DkW#U`)9chFte0@rlhA%&_IT3M?t7sw#c3t_PZ4Qg
zefss&nfQ@2=W2L4=Wc6Lys1v=(%uaCe9me#44(H4FRe`I`~2l@!r%8!{~$fHOMfDw
zK)-6!VK(LW&)s&C_yv{ksfj_NG07(?Yv1Cw;>z<_C5G&fH+Qj^g?uZR;(x1P-6u>lmf$1$cN8
z{%O!=Mi1aj^TvtofD40euCE6i;kTXF_vy{;V~M|X^K)B@w>JlMHtr3KA3k^DeBiqC
zkn?k;KO~6?~{SeDjGj)4Ex*l?ifyT
z?7>@~Wxy6&S5%CF9;+_<9jW}Z)A2k$9{tFJkCP3XW@83KdlUwgCp^_^(I3KZCFSmu
z{Z%83F(V-!G>PT=f>+)zc)z}TCpnkwcAd0YpXxU+d0TywD@?gii!i}^2&QcCwbr|m
zf837LecTb~GWLrv$)DZPa2)aH7dI~&Q}Jydq2huMB~n_Imtb6PHIS~l{@1I
z?H4DXdeDaWyDB{8(+!
z`O#5azDND_vq-M|?Vr4!N2`70e5)UeINoy_#c}?DrwEa;ubo_IK=Q8!ui^Rc9o6u8
zqFTKWKJL3{TRx}$(J=qhWRkx+X$cLZTz+HZ{kf_+ksJ0v@
z+4SZ(rps2WD418GXr)6O;3(1^N?{)8jJ>oV;CY(Ikfwy)W(zypJX^=puYJhka!o!z
z;`HWZkdW)sK&&_H3X(xP%@OHxJv(0s&46@m2f?p(O7Od2S54`$D+Jwa8b!L8-!cR}
zc!l71ye;T<8wEY~ns}~vnzrYV&pR&oN6HDl(R_+nRP3_(c_iRDgkN<|+dalI0Qy6i
z@UNYb@M}43k>3fui}=4OJFt|NQ{WJ+0)$VO3T#5#G0-=@8NVm37~E3SsifN~*hImO)V#=_q?C`&
z!)Y16ZBS^hEMa$L9;jpfR`h>R{1&PBky?lJadkk-MKpIOkT(+$iz-hi?+~BEzd@;w
zHyik6^-7w%9eioGESH;ybZ$2Ay;9F%O1VDM1Z5mO<&MVu!7GgO
zAw|wlky}F3AM)2$`dK+;T)k1or44tiifw8rv?1z0Z?{y0VcBJU6w$76)twb=qc
z)ARxPWeUG1O<$lZAQ~e)^iJUNQh_U{3BD5@_kn+T2N|kjKlnu)#2H@&zd41yNS{4I
zpw~7bSNEMBPdCCTEAfM{`V!`!AqQZd>q`bB?n3lNXtInPj_2`d0ATSgk#F`EbfRfLQvK!_)l=K{;-lEH}QHc*ei
zI+8;zf%`{47I1p^M{)i?XK6mChfetl	Q*sc=qTdhwad54{n>Y5U`DoNiU(i?O0&
zwdz&m>46Uib99fY!O>$|S2zIAO}gr%ORCt#ziW89->RA%>reCGIPt4JDu#Gy&<7|$
z-1d3^*Z*SYBA);C#ArAQ@n&1P8G+{jskS`c9JQz?9DsP!;0tyzI9EGg4K0jwvxVJ$
z(YpRomGYY6wI$0}VCW$J8K&XoOq)0q4#fHU5BIu+$rfGpNzS<49b5##yq_FCiqZ`~
zwMNBMEbHmowS>#|T)-7B^yAePUl83bO$RF^;`1B%dx?I!G<-YJ_E*iCQQTcOWPJ_N
z`?mkW^%N(qdg6J9;;C&*n^YtDlGd9`?16Q?JJq5%KQZw3MbfV`&m)}jpBg_cbr8^J
z`NhkDz~=07SBi68|K5(Gcq`~`#{@dxzW7Af!3k(KIOX0!;2O2QCB@zShU|!>xUR(c
zuq)A>7cc2r7kIGo>V#m#okjKrn%1Ca>CN6wajQf8#8=IMjXcYFdjK6{FYN1!xb@}M
zwVvZZ*N*aS+Ys1bk#7qhplwY<&mYj!ZjDd;mCkq$r7Ens6zM&+%lD}UwE2C>suuF^
zy*JyZxFhmsIBuS5h<(r*kjW!A=(sntgbLT#2@K<{!9;yN4*m7U#@xs7;$!{cQ
z{i~X_ggr~#tGwQe{f))AL4}wd
z4k-VZWy%KRi~BOqjrNf~s*p>xj|;tZE06X;`+_ci7zjKxr^Ud1K<^ooJM;s7+IX|L
z9k8VPi00I9C)IB{pX3MEOsTgE`i8GNu;ULf!{cu60Z3o6xF#M5VV#KaPbc}1pZ0;Gz$p9uL;wR&A10sg7-U7t|DHCt)uU}A@Pvj3y)Y4HE+2bQxIL2mT5
zH17|<^q5r549LxX>C=kjZ6`cFNaIjD>5V;bzS1sZ>SfzT8Aj`s_u
zeV9}2oIvuA#(X!QzZ%?*;q&8_iP-id{Y%n!J}(+uWN>v8nWrw9l8*TWQ9so3^|0AR)7y4;TJ*Erq*-E0gk*eZ5MWyLYtg(4cvLpKxnIuKWF
z5vQ|%!W^X6nM9El_HpYd0Km9y+l$ZdOFz>^L7Y#tisI?<6NMdJ{Z4W}*)shl#Q%G9
z4X&e-DCu7=3Omp6z!J~f6uz4xe_q)~-7PQlL>?CMF@psClH3LT*A<-lhITUW
zBUc3f@&jSSAz?!9m9qbCrj+AC$&jbsLP*{qlySB$6$g3*0gz_{k{5~Q;U(~mqK}7{
zpwSsPU+SZ%isUmn1M53yVBJ^r^JWHL?t7NK6!nbd1))DEejcUxB}Jj7eyTO;=WYdZ
z_W)&kQ0oz&Mh_tdDM$V|5cD~vJ*1r@6@CK+&ncK_PXEEWqSSY%Vt+q{E>P^9QB}y%
z=?%#@Q|!K`)VpUZk=`&x)RRUP$&FO>-B9$I7>V=)6z+karr3E|`cLVHOZkQt^VmHh
zC+)nh>|6KKG=Tgbl?0mUi*ido2yC`Yq&MOZR&h@EoD9IZ%N9(Jz)MOwDMy9e!?7aY
zCmrgbST|aTc_vdC#{=dHxzA68od0g2&)_>%2=U)D!JpYp@K2@+{_o9#Z>-#_*_cO;
z!#dT79Ep8EBXR)tZ&_mAI_7WBX|r3jQ^vSn)Rg1p?ISq0S-%ej;C@ieHOPqh;V70T
zVAC+F5Y{DUF)z%#*^Ei<
zMGb^_s+lv#=0ipL>nZ(F5XMz|gss5jiV>W@X*W3<`>3qN977u(=KFy5qrEx3$-gbP
zW7Rip9(TXfn&wLKs~=vPNtiPGjU(Zt_TMn_aZg4678*dsreC0|V3yED8$&Wley3aW!1efHBpiX(a~dRnRvTvL10cG?HLcr|{t7vibzwr`j>=yzAGQz_24
ztY_bvqhsDGE+2lQO}*ngfL
z;-3)@JH7#bm3qkvPv~hC_Ph23;Nl()?@U76>T3TZEzcNs#+;fR3AwAS0y=F4e`;Lk
zHx#$qA2%v&ZLVTRE05W3V6S4o0;BJofd0WfQwpDeKh4B0EY}Ql>pQ=TjIgge_;G8a
z@|YKV?VU%0erG>0b~f~7Z_s!yhMacJG#ihMWY%G9hcBMcdwO?qLWlaO|E)Vc>LV_`
zI$!grS~Ii{^O(KjMKXIbZp8LG9a7l2cB$#roUmWi)a_RvcK&GiP(K?ObGA-g9?;2X
zgY^x+AzT$@n=l${WdSbzhdYxs1TStWXa}^o;U~T)L>fx
z_78P;_5NrBKdjw;!=g6it4^HVJOuSz-Mnk1y#w&PX8jM7l#MtyGYuP?fO;SI+0!`^
z`;B&;hn8NnM!g0VMonD)lua^e@YVAo%1^miH)sUTTi$do39F5EI`!M`tRwn~n^P;Y
z!*$EsKMwFqh0ae_Qa?NR{qh)!ud`WV4E00bol6^0e`ObTIM8#M4ud-PZx3vk-FYmH
zi=$0fEh2tt`QH}bf&Ou>Hz`iN;#&U&#kCzOrP@-z4!Gu+*c6y@?yzCJ4ff$yH$G&5
zerP-25?_2{PrjE6Xpee+NDm8*1AR6nH~Yv0#M2os>ovlDYtGNrcB4CCAJY5V>gVu_
zN643zrXYmM1V^;mtcDt~TP$HsIN(UBL+&7<*S^Vj+^jh~LM
z-c6$PW?iEAc&|NHg%t_=nSm{gX#e~pbn{^H!}zU*2Pl1%qa!9ntkW~+A%wtrp^^KO
z@vzg~M%#hLMLo;cD`@=vFS_14uE+0$nd>_AW|2!YpIoEy8b=Gw~&pG$~DzPth
zHm!4I`Pp}=f84!=H%mZIdQ?t$5>R9Q$4Rt5$vPFahW6vbJI+Z{zfMmc_pV$?Ef2sDl82d%r~#c+hWpvk_qtr*6Jyc1JmmWqx&Cp9#wkkfM`a@Z;RtqlBWq
zjS_?}W?o?B#7a#;z`^?a5Q%^TKUn(gDDBrG3XWt@yHxKxJf!yC4((h|`GYU@12w_=
z4HW4jEb*_yi*?w?t1gNny{A8nmQlL*fRGHrnO-mrAb)PFnz(gE9{ffnb{K0ggfc?D;wrRlb5pt})
zEDa*2G+$9IupYZNLb<&F;1Qf65#Gq-;&y8lr{~KEB+ezMxyV19$l?5x7%s?b@(^Z#
z^;KN@MllX+>Dn6UDaxL#o!@o+m>e;d2va3bHV?;Yz!N&${zpVuh{dLx@fQ&!bleo5*b7Br^{2e}Enk0bvJUtf`mhn(!h
zp%9q^c{{({3jT9G^<2Mb>2mc?*v09E^y9DutCF6ZugLM5R$~s`?r`bf`RA0kuDk~f>Sx5!uL+o8#~HxQ#s=|?AUxF?U(OC_3v^dmGJz**G2z+L?D
z_&R~pTQ13^kEdyj{BKem%EMcdJmWSMgz?Ph^ySZT>8|@YdE6s12-Ke=cheSIKB5XIKA+jphwW3G=Ll5>Z4q`!wKpT_^GE{KZeW%
zt%5wWVFAO@BRRP$j5-W{I(IHZ=}ChaN-0^fb&?Zd!Q`JUQVc6k`!dY=W`~Lp*Ke)F
zh~s@&>pmtQIm?w>CdG~S;HDgwuJ%y}9fNa@6|(I4>kc%r;MYSO83w+j3q$x@#ZgSp
zc?*t1Az!~F&+=ED-@>pl@-iBPIDCvgyhHoGzGQyUEn_uXw+crlGkf}^3r#4$XZhq?
zgvoczSb57%M`P;Xyu8R!jnMX9a3G=jN8dU^{XP55360#6(J}DLLD^XN;6t{Wto)V*
z%HN4kR_>dGh9d6lzRc>ona+I|x>Qtk9;LUMx2lug`~J6_NwysHRFiPb=A#%e>~B7;
zk0s1jTlJ9QYKQ)%5|X=T$KZ=t`0Ha~Tgd-L$8O4|{N7IKrLw@gZ*}b{U-+-te8RWB
zeZ~}5V!j1FPz7q_S)L<)cVu=!KZ=94v;UJKK6yZGt~JShw~s9+{i%QV+^+^68fkN%
z;#*!#L^$)~l|sIBZsfgc{49!J$x|MxDI*RSWzDFdaw{gxc}wwK_*B?4ifeO2<6?+k
z7IuAjG8KBy1E2pEhphi6Z3ywu7b(y3KOxR#8hCG0L;f4p%K4&*-_^U_yH+6o`j`is
z+<;0ih8pKV&!VK|@F?)HcB7)o&xq^CT^1&};JG<^Gy79~oi<#JX9HQzEo{m=Jh$3n
zo6`0%I2ZZ%sO>6bm*}5;&d}=(>GM4ZJu!p+Z{~p~_^O<#HXe*PQ=q?a7t)(zU7J=w
z@2j4Hx+G-f(~d6BQJfz(Ppfb-yK*N&Tt;9T0wQ!FvZ96T&@Uhb<072wzNmU;$O@DkQL+O*@j
z&bc~LE-3f?Wsg-8!6#pa#2Ki+sNnT9H(b_HtLC+ho$B~uAndd|aQ6{-;GD462W#ap
z537VX%zakD+t6$|Xu!Q1-rUi{fPOW04iS?|HvWE}p>ph@N4X
zKDm~cKS%Sb)Xxgu$M`La3>qqU;|`DenzRn*&hfV%S>3MY4SHr5xKpm0mvrjFuhW|_
zAGPxqy_sLdlaY>&KGs{!)9z~VeXL)_OMO-29{8=87qa-;`-bdt-ty}EdQN%hw<#T;
z2RVVC{9352hV~_|uz47VbKsub69K7`WxVZst`F>=0p8}(tnxAx{i{@2Z;$rAe&p|2
zjQ*A@9=GsP&vRbWo3j0b##Hd;+PE6mOH}b}b_H&IA_acC^wImgx(eRGi|=ASxa9MS
zO3p>9UoPTJ?I*VM?}rNBu#n>ovO}tPtJk+J@=nFL<^J(|FeR5~`pL;J+3Fc@@fJtX
z&7NhvE6Zox@5Q;h{;=a?6sy4peJ{QKu@vu5*c!f9!+3;=)DO6idGf5`_rbNWOEdPu
z&HI>72S>=21$bg#B;sm%(i-|x4|&K!FMP(h#AM8y%l}dxJfUCm@u6v6V+F4}MZ7T(
z{J!+Ju91R1_Thb_CqBkF8Ol!`Pki63OiO9edGNyZyJDj-uX3`!yqXEVUjM{l?H}wj
zI`0@J&^jm-?(TNqU&R}?)$*Le6`b=b)jMhr0Y5L;Z~b}>`tkdMh1uZuR~(Zh>HK=y
ziKqqXK&{69ns%_eBedF|*5#&7iFtG{<2fue^CmFrrLt(vvNwzIMC5QdqmauGE8Po^}qo=!Rt`tZ4LmsRdK>qc}&FAO8qW+~r&Q39b
z-2S#8i1FTaj>qZwbgh1p^!&tX>AFGCTb+JvGxfv5C1*#Ie!P|A8sh8!9o+qw?!7cr
z2D?!CA_|)n3C;iZA`pY`@8#lQQ2B2bulkR(U~yrxiY$m4_~8g`7GGRHn6o(b>6;IW
zbCPF{vi2tze)>o4bdKKzW&*$Xw4RldpFsyW@Sn!*qU67|L&C*K=9LZ@NoeY=ZBKZj
z#br35PSk~YbAWqVrmduQJ{{ASMCB%^e~+g8;#pR|Np>@gXLdx+Ok#2O<0^zpq+k7y
z<0B8tIKJ>P_W(=JYRhJ*6T`*#T_PMG2{=hi4eQ58mi0q{>q&+U6S@7s>V;hV7d?rv
z3BAnzf20Y;Es3Bae%VPdeU+-M%pTQ|x=fC-I|D`p|2^s}hy>z|*=QD5A6I0vep6jh
z!1M<3&)1g_W(0@+*d>tsF`VN;2kA2#$ZcP^beBR@3t1k$NO*9DGQ*%j
z+<5NkL^VjyvEcNaz!pdznEa8UHPIKyJ&4AHPIoZOK&xVo_eYMJfeu9+ofN>*&#mJ2
zZ}oGy{`J#D^N=rE#g%_B0V3qlxE&@8PToC`UANlVieXip4ZcgO%7O&)ZqRJqy
zzfCT3>Fd98dI$OMQ*CgW!OC^n!s(^cH4y5_?Z>rqF#Dhc?Oni)>$7@0mj4zki1Z|$
zahS?KS6vg%>9sas+9JQn1KTa2;b9q0K8$TQWLF1nKfA0T6(!-^#Rtb~z+YX%7&d!t
z;jp>^1Br8O*^5lidKqW$%%7OvNN;NjX83y>w)K$n!h+d+7{3Vg3-V1J35K_iY+|S<
zbeie4Y&yi`1LJXB2EEqx+;|N?X2a6w7JX!TBKx`Tjn-c%XL`IRe_#Oo-XJRs6wd3E
zPx}$(O*o1xmJyVhTUbdY@dJ2@(S|I1BNW8`5tC*1swxkbr_%F0aq(Yu=d%^nH@y(m#QcO@^i0e&v7Ez@3;@E
z&$#Fk79`fysb!eS@Mk?1sl*2_8I7@}a{I2DuOjSsEolt#!*j*sw~-%rYM8A@yi&?U
zq92ue=t4;w>CK#wca{9dl;@?D#0#&K{4HAqoUG_Royrwj5;BYOUlwK7lfKT2wnp-k
zfg6vVO$6TdZE_%9t1@vVJ2y)F^HG(`-xGcG67kWAKCgDx0uw4M<%y35>8g(*|F*=%
zRga!CRBjQW^7pA2B*`LPO>Ta>k@)7y0Euynf#%uGqYB}#TW^JnbRzCP4_%%r1w6MU
zyMf~R6qO_?vb(aeB4ZHHrv2VDN$}3)#uKckBQCd2YdiN7>2e3IdDHv3lShwVydH8_
z{zYkVlrz@3CDU6G^|+a=cR>Ah8ha(epl4+eRM-!4L)W{KW%T@_1?TC0yf^gEeNj9Y
z{H1Bi5AcEVm|qjhp=Yx@g0~ybb*oxU*@yGEep&84n&6Y~suxE|AYMLvGfXoI=VROM
z+&gmz{BN#IX~^kZ@VPg>E`H#>tNV<;T}Qcc4`$y91AjPix-;N1^5=C{Xw3#6a&OtM
z?kx>oXOQWB2K+Tb@4c!J_(?=+YnX091<&sOk1=oWRPoXsbF^Md;rybvTzEIm1G8>C
zRPct~`wmagxYbv|lbAhCYVx0I-luWN3#N$T{?tJ6ZYKDU*tai7Wrl($q>kGZtbzV0
z`QrI5zJ_;uhJM>7rz&2bL3fDBX`IIfE|@%JIL?zYPu_ZIi1F*J_3VdsjSrj@>i-D!
zu748t^fBUAhVO$TsMliu@gb$)6^V9MJHOjue-@qcxgLC5v3F*EDPa#?|!j0PVIioF}_p9D59LXV}iZE#Pmh#(6VZ(T=`He^<|+2VOMp
ziT8Z3Qr?`KtAfv=Us~t-`m2Y5uf*N%cIvL+Wv#m%Qi7LHs^D)kv1&^l<
zhu+VBnOTzL#DOUUXP7VI9WR
zY}8{dTj(jKDlT3OJzvvh_5P@TO{w$Vam&yTS9cl*!p^wne)fIXhtx{f>E=LB`pmw^
z-O!u)Fi?}ur$@`&0u#YGesSo+keCCpGteB<%j7wB>mp^w-s7
zOA?>_x;|{33NZem&~rK;EMBO6lG=CpRC_bMFLLwq4f_R1w;re!rv&`&xXqH*=jiwA
zFAPM!!Tq74H-O$BN6#j`7wI}NwEpyG-evYgHZHpsi2NHja_|N+zHf7B_jE=6ly1f4
ze%`d9N2;Hi4qKF4K_l>S2K56*Lb1LbD>@E6_@}InMV*cAPE9sS{b9lb{HI>(WZ02Z6pFghvffwJGJiE;Fjy5BVK<-I9+mEm;!bX~~vim%zH$#ur
zdt*P{(8GS$zm9|*tIH(`GbXKI{h({6$Ku|ONFp>i=L=I}{AJ^~HLQHS9MD;$xA=1W
zueF2YFEg%OV|rfYXIXylQ4|Zk#m~(d&icaXwbk8YdUhIxEPYrQ7ypmHw`B5yby^H3
zq5aROZWN>J^gwu00=JIcj!0I3~hyiSo^kL
z9mK}t9(y4b`-I0_Jtr5So1j0bYa!E%JR!q)M$9yEmfl><**|qRH(uIJL@eQt43gRT
z+B7MA9s=3Vo6DD-&Xv1d_ztscd08hIFXnUhGB$r6KL5$aAe^>la;q1D~rwJ<$67gH|dzK^rd|JXICl+i-JGd5(=fvVf-b1+H%Hw=FjV5cEu@)
z#u3j5v^HhwdSkDl1FZt$n`xxC=3b#ArJLQw}ig@cH%SHHFl@TZ@BJwph$e?$(h#!$=~Yof~V5Cm}l|S
zW2Bd(Gthzj#^c%n14-^m8X##2JbTo^R26#m>At<>50jviP`7*d+l9boDzk>t{o-Yn%f-Y`Iux^>PC)!=R8>6cj(n@^?rrmd
znWabM=^pS-=>&~Kh#wKWt_TyLxNesBMLef*Chkoi{C-jUvwv?PTP3%xNQJz>dFD%L
z3EX>aE4;G?qh2
z&XTE+i`%=MYw*1Mq1@<^72;&bza{G?&&7Pxw)?1EUBSzd+dO{k#R{I$m!N63u(NFa
zjr)_qU-CZPn5Q9v`OORNNCxjRYKk!}Ppjb>UfxnRL=xp&M}0mKi1tT~vK-f2!P6SK
za6*5)KlonaaevL*FL)URgO+_*g1BUuJ?`HuoJVJ$PaPD8d*s-*w1lNqJmui5*ZIF+
z@G8=yGsojx??7;@Rl%zY-m^D8dDb0}XP>ef9bC=Rf3)J&G>LNF-Pv8ARE}XBpFD_A
z#Pg<|pNFa6s^%@ZxLr1CM+I;HTu+se{R(&^t}gD>TTsfo<1tIo6Zgcex;jeVcB36D
zJG(7~zz2*+Xa?bYNMk~8;K1HO-pi{2dW*9vc!km4uhymE{6i$S>nG0DTHR0SjR%jk
z$i4h;eQ!Rmt+j5f*XA<(Kbd9sE;iv_Y|I?f6Y~+@o=z81#JQd4)&XuEQy%lSTwO8o
z+|B}C`oqr`bH`TjzID_}?!Y;mX1n%;>G}8$GqKb^@D!Zc!cMtu1#|2uL_=Ys0(tUSAk=gm}XpijNKl5^-*X(zvo&3w5(_1u&U$mKSl^PA)
zx?Ez9J?zw2BPZejxlH$n6!{s=w!^E*kG%IaawR^r?CL|E6!b%!Zu7_iuy@_N@0%|n
z{p;epMasasy7Wdme?Jp9A>Ix4zLj!T)rNjjo966xq^~(YerYxI8Y8+JjUaE#J$H!w
z;HC>#KTthIfA62CdX*p4F+VG3SAc+vedgtScFvdKzUmXrqi9=@nDP4+p
z179j&{2TTtkv$1D_B#w==TT9S7vx79qLe=qf2+7|o}dA=v|3w2{irQ@$%E>TT(u4X
zAKx`zbYlFf@!>NTe|CL&%;J+zQfCn5pKD52A?!Yr#Q4
z&s7P+%=D9t53L!Kz{)G!t1yw$^DV!+Qae?WXG@YC^M;2TOPmJ=?X@C2kkro7NA`=|
zFdMS>su{}E&ZevT?^3^*We?v)dI{g$z%($<`Z%5gzBG8v?AftSp0g_qVG!vf=X3mN
zq|tSz*Or*d((R@zG4yKT^lTp^Oheylj~LVYrKZQ)nR9_VH(6H0@eXTed!|=vdk;hd
zep#90CFZ8zS-PJ>9DBafZZLa3&ZysavOBCm4#W)c#eOO4XZ7q!j2}d=h-P~3u9!Z^
z|9l?Z0o=9KjbUop4p!bG$%(9=&p+bqm~=~;oHu+=J75
zt+|HjEf1Dt<@tFeFg_x@0o9|v0W)M7e(mqW@SiCiP@%u2Uo$&T@PDTzx`r#Su9Yj#
zX08y^JJ&!TU?N^Au3@s{2w%ouf`axko^kqKgBRt8{~
zeO9qsNBP<^6VcIlAIlw|y8`zItYQ5YP>a{XAcuvXVtT6ce&Qw>=b$t7&|sY3|M`pt
z!LRk3!RA|7N%U%xb!IsSlHTPz=b4^pwl%9iXsC-elb&qAX?N1wAb({Zt!uHzfljnP
zJem3Jt}ZY(peCB?xxbTli11jHeJ7P?x7GjMI^e6t6AqD{w@ZC1<2XfL
z{}z`#r}xStjk0#FhW`xcuB{A(za6~c#*akgPfqQbNA}dXmwu82-g#zr%Nlx0k7K?D
zfM48B|FG^c(yxCH(5L;cvBl-X_K?$r*NnV@a>i8t?UA1VztivAz%rx{*|y8733}b>
zicdE}KA7*Qqzt{2KEwB9kHb1%x-8}s(v5sx`@V;szU0*z!H|`z<>y*J@1|#m!Z+xt
z$*z5C4*sE5_f+f^{PVa~yJPBk@W1Eu-*mjL-~}yu_gtbKIP0cGq9XS9`@LSi#(C8k
z(|74Q;Bgt7n(k+Qtl-`MK6QymBi_gA>-C8mE;#J-|j`^G^YD>^E8
zCu9zqghp5K*4Lb?Jpny!{exf35Pv<#7p}O0^RL;yK^k)ZDtOJu9EH8^R`JfsXXf8V
zy>($et$v6jdC5|`X(;c<-X$XxepK)-rB_Y39$v*WC}^QarfKv_SdnG_q5+2F?bqi1*?BHzZ)bz5leWvIx9mzQXI-
z4bbbEBEFx_KYB%fCW%8Y`^xbZuHe0IelIx_SGU2FtJmr6pkY#>k_mJowI!98fIM_*g
zdRb1sq_^AY@)^>b{A3IB!;9Z!O3^;;^HrP46xX{GYBrMIkm?CXNYDAPh81C9NZMEG
z$E1Ki2!NQk2gk1^y=UdGj#B=X*^?azjn``74J`DZ=2;dmt23rCzps#L&f-h2=`QBq
zU)s5_^P#;df5y{uZ)@UM9QpV6j4sJ>L5584o9*+@RG!S8gNl^CEp-)(Cn=j
zw^+PW*P7ZxdMfog3Z&Q5s=)N(l<19r%)3W8)&~x%l;|bBcc*`XXkZ-;!*&$7tM4q+
z+qu1j>Gi!sctQH4sUsQwvE<@qi)eqApA)#7wP$WCXcqL|p5pYP()^gbyC1zl20y^t
z%=G_`Lo#HG3nLf~IKPo$@WH1{FZIi1hM{}7daK2xS-yy+KSS4CyuOX++rlIn-s$7c
zDSkR}=Oqg-xidY7n=Y(AvpW0?A@pA@ied3G^q4sc#=Lv$#o}fCX{|t#V>AD;cxgE1
z5<6Ege({)7dz<2SA@XYkJY^hG`k|nzIT1$u7q9-#D8S%yw1r1rEf3$
z(g(eea}pAN!6TZkz5Jnscy_%dKxaJSh?v3I<3r(}W2Qg7vkdX~rj??Q_ns==Er-#c
zda(Z<8~SkLTwIwz-qR^rrWGjzzr3jNRWt5QT5mW%
zt%5ygvSRv04#7B0ekoFuU(GWO^qWE7WoD}0DBX{HjmS|)WEC{9zZ#ueI|%0jIX7IK
z7pYhC)-C*1RR`W7AwB-&d&Hv*)Bk3F1E1(R+M5^IT*0e58SOIqOC@i|l9O}hbeHjV
z8;k1oAzmKZxctK=#7j-<0(V6z%+FhoHk%_}E?zfob<~hDUdheD3P%y=H)^Nc`nink
zDxdTk_d{~}n!g`FuQRiQ&e=tKGwQuCfymoIO^!!h~yE72|bXCmNQ5xfY@=E^B=;g3~g?~^d
z&KHcfo=zzPPl&u~cylK7E*PbD)4kyo)zw!b5ziNN3|?Y`{=IHd)Nl#qbxoAkZbZGG
z`rp8(X*iGj?o;|0cCIv^bmI)<=-r*td5E(QjAb`_LGNyeTsobnZCDqoHVAlFGJXr4
z(@DLIm!Nnlwe-p&{0KJV<{rWj0y<%CgGR1S18DA2~ue5&9oK*=u!;G3VI!`c8v{a>f
z)2d<*&^}?PpH)9v&yPaQ-TpzZQ1^7laLBjvYdy_@&2#=*(0UoMWMi*BFzZU)dy1E{
z4FVTZy!<{QEP(vW(}g-m$v^!5GTMRAI%U`z()(KEf=|7$ero(zFhBIRF`DvS+U4Nb
z!SkKkmk~ecKjBOx=^bd8Z%cYMDs5*e|Lv5y^9jvscNLS~NU4cqN$+v0?P}7yY^|S2
z`N2W;EFKJuRJcicb#aEjs9nE0yAw&zLg&K)(lZNdmnFQvVI_-~CP(DQl6)e~oW;vo
z^%rzW7C%wM;>|^~jY1@kNH{D<__!yD=|yij%=9Y#;a-vcbkI-652mU5v;B9%zVQZ>
z?(%mE)3aKD?I2{e{k?48E%Rvui|5}?FJa}GrGzlO;Cu7}H^w`1P&?UuH(QsTqo*}X
zFdnh=&rhaTwxU0aQ|02^zI#C2DE6E};tGZnG^|*CE|*`hc)98&r?);=jpdtnarY9L
zqkl5JmFK6k_}lve*{D~{gF{c;(vj>+-2)VDmSx!Cz~v8tRg_+=!lB?j=ayIFSbF>~
zE`7{5t{kV`oSY5zN6&daQ)ZYJo6X{-<7YdR2)}yAo5f3~?Ms$!mbE{p
z+mrFH35f$(Jbzm}*Mjm*q`giN3dNjb?aFss&&n^+YGHDjs2G#o%6oCJf%Pjc!Q{k=
zRt&f1Nin?r_!vW(cjwu?OY%ny8}ySJ7qRpkJ^z^8-nN&?IzMDVT(EA1ci;!2v98+c
zhLJ4(!cLdw;fq?eW3+$WA$FjM;-yk|#6ilxAubs~;~dtteF^FLX|&qY`X1)$^I8R1
zsgupl8@8M9ZYBl?KWXC+#$4X1s?D(!=6
z6fcto?wUCUesb-p=7c@)*F$7AfhGGUAf1`rE1S
z<69>b_osdLqO|^6{&?^6ftSwE_y@Q*vTv1NjQs4+E`}%2F2~7-JS1^nQT1fdLpy2o
z%gogoVx<+lDf70pKR~%1BjmbiUpalx#^LUe&s-n(O&H~_cibDF>I`1AO>)j6&
z*6kaYI~F`ASx7Hy{^}~8hWHq%yo73Avt(Q4{bLopmi)6PY@a^iZ9Vt3XUwMu*nh=-
zSSyNs@gb3L*M?>*N=br`OYh(6kxA%h|SbDc@_f722
zoa=&@1XS_9^@?vwML!KW6Tfk-Io|si*`5+Ah4&qTo=6PXhwlZJE*WJ%rJAQ#s995X^PSeWz)m4mx!+Yy(H=gp;6w9A~xLL^4d>}vX-QEh`uFFRp
zJn)`#vdhr3Mc@s60nd$I=oaxJ=Q~BI?=9zTa^?C4K%xiU9|>OV
zWpnOvDNku6v2R^=uIL5yx(gaC4x>LFn0~e|45;Akc(yrZ`#H>0
z>Ew?dIygV{G0d2R`-`9BzNlJ(=Y0FqvSpuG6>sAFGsC9)pnt~9SUwN@HmgA5l?nLe
zhU_;M->^Pbbjw)y4gznj)ND)uuXLZ+9K8zjX`}E56*}L4SmFAH-aj%_o?5;H8Ry%bQFEs96SefFCbv%2qdN1m+4$Ayh51yl#b7Zv>
z^qq!EZnp*A+ME}zMSRR=mSs8Wi4dBQc@E`>AAJ)RjQs5lLS?o{KN)^a#2WMTRn$(8
zV(_3u{o~ITL#`2vl`fcs(wDLcLhJ&0`iOFgl>-^
z>`DBhc>;D$nUHsp;F
z{#^E$*1b`CnKkK&4*Zqb4!eVI-^E4+{=j~If(6iSL?eDl4&#z2Bu91(JDs(+3gwuU
zw@o5G)3;LS6yZ(1a5M5dx6cO}klfVs!+oA;2?+pgTi4L$BB-XBOK+3)Hc93;aZr}P&ge`hrzpZRr9e+>j+
zymu;ITtIqnySIrL0{6Y{lOesATfawBe#@2KD9X=DyM+LZ^F^^Yq0D|?!;K^lkSt(4
zW#0Got|ZUea$Sn>`A(gSgezYRM}$Uy=?-M^s{FkHi?@67^4UIDsJS?j@)OjX8Lu$$
zo6Gos_O&NWzNnSP;_lXO>-Uk~h^yT!KF3Y^&h(NJYnh(vrSM6VKYaTy987?(S$n9G
z9A9L@;`;mU*Q}foE5(@p&FteWj`g*jWa*|tM_9VuH6s>Z53G}Ba`yWoR<2>*6J~!*
zCyvok?)0Gucfh#dB@DM}l`?yt3Y#;TcN&i)|L=Mf4=iiLHXpb!XeHCLS6|NLP#KQj
zw5XyxkiTsg_k7$fh>#~yGVmEMttWXinFU;$%ppi5lSlFOr}6dgYjEkC
z!AePg{&Mb|$ac0PlcyXQ!7#{gB*TIkZyC>7WybYSY6$27p36`6yKDfhvO#3+N
zw*^&Pxu2uC_6hU9S985)#PshDgc|f81p|NvIL#w`xq#EZyN(;bcqK9v_kw)>n6Zn~
zFHGa~bNs16Sl9HrHSlSGw*#S7UMh$q;_ZKFAa0o7ql$+UD#)1I5HGlQVgMe%xq#IJ
zR0u!3tC-E7jZ<8izSjYbxs<-QFIm5Bn?Mz88d+TkM|4>7e
z;eBgohN@Lq-pE&NTFc5QbK}N2c{$gAMq}++e*26vR$gJqNM_Hey)TZEzbVpjNFzTc
zcl0d&a|-SST;|8q`Y)dKO0}BO&HGo=I{LA@2roo|-#B;HQ+>a;XRIV%bb0@P7CPsw
zN}^w&L44|KW;`^|G9L!Ok0>i#Bb-|15JBZHnbv|C1b_7LH6|{eAN7&7Z)U7*6P17b
z;&eyyccu-2G1dsC-?*pZf8i
z{p2|$OV90jR0DhdmZ)PXf%h#qRz~fLtr%WP^JAKrrimqR@a^(dbgtBGa%Fi5&M`Lq
zrxD|bdjVeN4TT^HjO+2Y{cU%HkLkW!=eN0fVftrbKC%Y((cWP@ANxd6;;#LctOAL_N#?cp?}muf8lWOfZ@?0
z1HU2vo!Wrs8(~MbwQW;9>{i?PNqjom`CO_0UuT?CxP7bpNBx~$xJ`xPTYdY7Wz?=Y
z5vBv2@O+p2>IHI|@E0#{jyv+Sig!L-cDYS{1-?Ubem7uL2`_byna|4kMZEAN%k(A|Z!uIz#znUxG@{PuSXI$GkFUdf|2k@k|0juU>ytFUBj7`Nj+YcX|Z~;8A`Q^)X^{%Bn%gQ-{?sqGA=Ds3<7jQ0S
zn=?&lH0GP3jeJXq1rWKVQs7ajh_QQ
zx6q?}ChR;B@Kj#~
zpD{F14|c72rl3^}JK{b(U0egcTdTL?7u9Rz@Z#YF;G@l*3jZ*!mieMevhWH
z>O!jLm+c)F%Fl3joI|Ll+`)L@++6uYveQjN0|zxYAH3ceN?4(Gncd?HeSN%?WMezw
zS7b-Yl;y3|zU#$SGL#y?Zw7q7C-%bYFNBl({&Aj8h)UmXc*I5lfdG@0Yzt)pR@$cLjLsG
zm)Uduof5rB&gmV^&cX8TgzJ*L?K<2r@TSHF77r_roB~6D9clE-&UjDaFvsWa37arI
z$#VLj2j^h!MB3vSW{aba0ca+c#
zkT-5k=j?ck;eu=@4-u$y80G-?(7$T}zMCV>>UH&ZXL7^`u3k?M9A6^;eG;cXJlBND
zU$r@&RQCiT@>z3$L#CtqfO79JO@Koqx%}J!4vl_u`EDtk+&dHW4th^lqI%$r3NC-j
zS}y;5KMvnulvq9i^nUX37d02M3w0x~nE;5F%oyrWv5;|D&2Yvj4jHXOx)X^&aXxtqMe4q{mpO3~e0{+B*qaeKZN|>Qx
zBK0i%_a?3%o_yfub2$-0O*aU>t&!uDktw;%*n$Q$FlTIGSAumPj#*n
zTfYsRYK(tx=)3oq{L$DaM*_)@cAwvm2NCb~(#8!u@2@ZO8)LoJ@2}$g
zLGwwjKPsMnXXy#kUSmPR&;BR+nRvJP#`IQ-FB50gInz18-Hf?RN$T42~w5H>&J5*^?J64hIFh_NZfm<2?t%k_Ck8
z
zf3+JUh}W2oEdKBu`5pEC@uZ)9`N)Bjkarb6QXPTvUd4<27QcsnjTzP}zXJQ@
zknTB7*mp9{iO;|2((uK-afGgS^l?02d2Qd99Prm?>-8f776<
zCCK*+znHiR`nM8$V!eSfT1#gHqdz~rGuuw-_A4e&rgmO<;5VM;yHsn=uWaD&Jtw>9
zzI~fdioaw{DR`(zdF{vurAK0VA57@gYpcr@yhF%4~(z;iLlDVocH
zGxYcGumw(EaA)NjV9~SH1L=Lh8R1_XPeEVHVeR18KPY
zUe%wY^_Co66FhDK`o)tsqz~uf(`JdUR7Zbaiq?H(58j(1zGvhc>~|U!mTaAZdwr+X
z|KiWWuAfuRC_SBseWv!qC_nV;z`MUAyO1t6s`O6}o{Qc0?3oPuankR?%SK4Q9d~Fl
zoo8-N@sc?UJKcYMo!O4`2h}l0F9CCPmL_~e`h_<`o=ZUQl66yf`=e3l9E-%*lkzc(X2UPk^q@xn@t>sg}MUmQ7~e`hrMbF8*lix%ov_WbX8F7g|f
zsfttm)yKClw1K^v;nfSBP)_+s^=0&ZpvLlFOR2muD?0*)0fp2>gRGDq?D1ke`GYqS
zfiVk#H*d+O)BQBhr#YMajlcIj_Iuc|wf!$ULO%5VxDd@3y-BCNh>s3kc3qs-^NNKz
zmuTGtq`hk$2-(B^3VUBX_Wj&nlyC4o(uw>;QC-8PZ_s<@pYiSo@W7e%TjwA>pl+Tk
z=>>`XJwfNIX0!DDX}{7IHy^*m3BTa>?-==;VAGU>Nx&8EiXmj5#J;2K|NQ*=JKmA}
z&$w5WL&-lExSH;ve7CEz7YKI_IQ*OHFNhj}7o!m8@(*33bh{$%|CHW2{+hkV<#wfW
z7Ukde{-Z$lOrGIjOnSp>S_TldiTJD|{P1_n37TIjW>>$Fd}iIn-Gt?#!`@Tih%L#56+Kdf`zkXK(7{V)&R~8s
zai=912=*TzK8q73##{t*!Mr@bkHxjGi?Ug~oU|nl0T2A_xOV{I(4lu&IkJnm->Z1n
zeuBxoC&o-)y|@4cV7+b~z~cJF`WYY$nBTYLSvfO}ZCTuYw5XfK``>$Jg4uw#)X6iu
zVmw~4IPYmfL=^8M#2PVuBa>Wa*RI{%IojMlMJ8L;c`~d%>&om%r_Vs~+&c4d48^rL
zzWi|XXeJx(AI9zA@#fH8)0)ZJ><4*qFHwW)kw5wx0O*Ie$OtF)`{oaFMO8#!5HHKrT%0xdbceRTZ|`8H+~
zVa
(UXeEQKw
z-FKdYaejF-Ml%BazOGd~wi$NlsGJ??!lj{}yQ((#zw8*S2Z0B5H48BQgHcL1{_(h#
zLT(9(o9ekd!BMY&In;9NKN;5Pu0vi?C$Zp)0e(j`%{XEO=411>u|8?wlWmI@n9acW
z6yzs2HH^kO9{?V%6%`^
z*g63{J#;3~I^P?xAaw`kKW>;FuMXoMRavzEH1uQlDth%pUdZh!IS;*Y0hfBGedlqR
zdj)Ie!QRvP?vujiArDazlr4t!rs>DKs)Y69B_G0b!h5O&)e_p`!)&Zx2s25@B<9=uYT2^_C-#y1u`#a*Uxe_?K{VRyw|j!SFYZ@PZs#J
z=wt=$=ON1FD&$YRwg?B4-r%=u>u4NK@~dWP!0ysKv+!MzSBK1%kq6qGi#a|Y_$|)h
zJI%`?=W%m8uuiQD1pGHcKQUh@!3gF3hw{DXTs>osZubGm6KuW~=uE&qw&;5xOtNM8
znWfNwylj>8O6ZBDg@~L+`Jm5&i^%@f2`^L0-u|~wSbo8ai?L(!c0Qju|8)^=nql|W
z?Qf`DS^v}&!ehUaz$Cy2<3hnO>bS3k4bR9wseL%Titu}!=65Q$loS|B_($J0o!VJv
z_2_#v&~x3&Bh>$!d+D-TK=CE6nlw(?`w#tSUQ{bm?!5uF`Po}I0OO=~{GI@ODjd0=
z)&s9?UktS?SN0Qx7wx6sb3vfy^P?<2W|WR#aVhLm4#QC;CCjP)Ve&;5kB+ANV{v9`
zHeMgfH(mB(@`c$2gGqiL^p3@qNs|Y$^6|P8SUV1x7GQFSw=W_#&^V`RzPn4~@^jkf
zL*-5PZz8bboFQ`_0UCCt_Z}p@BOg*v(tEE+mSuJ;+jxjmeb>8rqX?r?AF_CA5xkq>
zi{V5Zajrg2%nRRF4dU~0*TeLs7529;!X_~7z7azi&xs7nninvf(z}PnTUYlZ%x=H+
zeBP?ElL#I9E8*X7XsGpUXv_%8FiC7TzU3L=7a+=<5%**RruK%F7U
z#gk_+Jt^m>ti7)r+?akr*)%3=4n{TfmL~y#%>sPfF3HE1S9~5a^A2C0nuTtlmvWRZ
z7Z&p6SU*`-el`QcLH+P3etkQo6(I**M7O{d+uWF)@&=+z4joH`7vs9a=RM&qSQe1E
zd3=4zt?f+T_C5dn5H$hI0`>fe4(WA(#MwliFME`jM4e>}+eX0YNEhQ}gA*}m>=
z3u5)+$_Y$9;Qbd9fP3guYGCO2KDp!{v$M<*pEX17Q!!(*ikS_=`p{=sSy<)SKb?Qi0R!p70U7n-zUsw`*`g5<`Jal7axZe06wztpA)rntE6`s?eAGN
z>zUuUa$&Oy)lZr!y_xW^XEHu$0beSpVEqU=9*(Bx%0tfJHv};*tAwMD^T(
zZSS{sy8$&e2ee6`z1KUI*X@9;KBeb{GoIHO(X{6Z;@gA?Q&*~?eun1PIu{k(-|sm*
zLk0DvlbwRM0;8uM8{`FkCw=VG;$4%#qXLhvj~$Eq#qk@B&LjReSL%7ZMVvQW+ctG1
z>)Z;FXN)@lu(ZE
z%ys47?gannxn}O23Oi*C3;sl+U)P7y;-6q&`sGO>AHgHmZd}p-8+@cGey&!^V(^wf
zb1KG7kX%-{%O|b21mQlpMk1W+?5tz1|-Fc}rk|@H-xNc}e1_;hzz2
z7uJ2N9NfeiNG+}^jD{UWiQ#w|T#~BlK?6Co_sMs@qj^2|y0Y<>JNRVA?n5Hif8bta
zx8bxu;XKYOrpzS=`jY~c~iw|2aE?nZU2+d=PbjH(1(u`NV!h4tA+PFKMSH*2f
z4HAj#;Bhw%n#SHfgn0GaTmSYT@U(>-XC-S6nJouWt-Yb@X5Sit-80d{v%BGc`U_#+oP^LaW&+Ex9v&9JGU2@
z#kk?zkZsr_^A`Ij!EK@LRp`Zgra#&ay_|oFZsbRjU&Tem!rr2Np2y-)F8J|wSOPG2
z@_}3HP#@`=S4Da+G#xT%-=9sHe5GtYz9aOkZeI<#*t=)B3ij8A)0g)T#lA9rZ*fuz
zcBa}I1QeiM#kqc;YH+^$Dm3q06z(xT_YZkNeoaHMTcrVL<8!NjB{1fM!BRSZ46=4P
zNB&S$*Qk&DWm3Me!(m{};AfMZfl;yRS|?MvgMbwIyJGIoIP!}*9~UIkIagbwrHuUT
z1%n1d4d6DHH=Bu{%D83W2M)N`srAx&1v`~ziOOAvtX(G)MfX@DUoy{=A9b>r=0^TD
z@n6U?Y51ScE@!87%-i;tnn!D)*Zn#njr_cwtLtl;pMc4_qsWi1TJ|lU+V7mdWk1=S
zH+_yYq5HR5#wXW13;(40rLl2;2o-l{`%=B|6RFXJV!^}Y2>tsmg=hlzK2Kub^(kF{
z*iGdYH~jDkFZT8CmT@FE`DSO*Je2k=4byib9bIxBnvW)7R
zhYP3!BNlySc^B#CS@jg3Udo3IBK!Rptzq#t_=N%EmCu}xu(-Z*)*Z$Jism3~0=?my
zV_BSDXnTU8ix&UBIAE9(tJm`V!u0J@%e+bdU*owYgn>3y>^!h`=y}G2b)J4^`azDf
z_R_cxip<$bdh@K3LHuxjmgKSc+`Yw=dN~vFoVyB4p~b0ElZ8fZ;S0Q
z80zTZ4cucb!BszEWksvD1YV2=bv*Q-DdUYTRt+>y^d~Be_;-vk8NciI^ey`
ze8}dZB9^`w!FeZR1JheAqR7zjm?}dJiEhTjOCnmBzVm5Z2SERvbr`GX*_>wbl$S@A
zQh9b(D;xjWWhG2ss+eQ4_m?G1uR0WuLC-Wwi=kZye;mF3B1|4E%RfgIE#r^NWD8%f
zk9`mgKQ$Z2M?A0aWH{6R8i?tH>}kU1vEPUCdG@t&^a}#y1I8a^_4lgP7s#o&WqcwpU9_mx4`QpH#=T(sbj0Od
zIPhRS9g?#m{M{M6ne>L;OIb_)ZOzX~_=F37t^JuB)%z*VVEfu?m^dZ~`TO)u59R|s
z{Qp{!UUuKn+oYE?`08bnV_a8#CXAMSewzGI-r--k_`>_T`~+_f`yloGS87Khd<{F-
zcYls-UItm}krK<_Y)ozpCco(@=I2lTT`%3jJ`i~EzaACx$48z`9!L3%M&rG1)K2NG
zzOt!6ZI$lfdw~hg>=jHPAC|B%@&PuMZ&h=HUtcWyyQd9&^6%uE9^S}Dy?46J>j%%<
zF)A^Z@){9Nwioi@&+GLEUpo!n_uS@tS_z&T^-i^E67;V89ks+2^|w+_1xiETLQc+m
zAM$%`cMOg$b47f-IH)%O{kNDGZ=8ZZ+B#CM670gz*?Iqsv_t>>gO+8t^SFCA
zJ>)xDz~e{A2E5PN<)rINkY{jT@M%51GZ^Nuz3bE5dag@n(=6M#
z22N)1?fR8_z}Nde9}_}<sa5i0`bUP-|ZuKnrZ1^?`Hv}Tu$)vk}(qXoSF3PsokYK&iUAagHN2zk(>dA|f#{NQnJ
zW`-X5TOV;VG3OK(nUWwA7yn!c}lKlaE*-S;=c2^dsG0x2>-@2Ilm#uR6(HI`LL}kc`
z9L%4y%d}u`@GG%s=|$1^?{OX-*n7vzALxZ*vyxN1PmV%5Sw3c$VE*
z?Ms-C@!E-wcPTGDG|9*Tcq2J#LLvA|jL|F8Z&(*;jzuD4>UiAeU!ozqrho^W3mUx<
zeA4;L@VA|q_rrhv7SF`G{}h_HU}Gin&2KWhhe^?W%=;r&$S)hKzCF~3b##659$9g;
z8_@Vu!AcZ7PB*As5j=Iuf9ot94UzYk8u>W{{h4hS&EYWqsADaC>R5NvhE}W=0?!(f
z80T$`^q-b#nd<@^iid(X%KoAfL0$GXofLc-t59
zr+=0yaKyLt22C7Fe!0B#IsQZd_@SJWU@K7YtwqN|pmoE3PwHPoPi7SHL+?E?GpPO9
z(!-AS0DX1|=8~VvtGJjn9(ZQZ>ITwJC|Ra@QPrr1qI7a2G
zPkhIhw)lQsBAD@^r#^oQXna$2YY|X!jy5;EPI|gCXR!Ej_sJE;`&>d(d#K-UH{JC~
zZ(9|YLVCBy)U!OwbO#3z3Gih56}E&{H?OgH>QeE7#n&fpfVsUPf!?Xh{B&V`|r^RG3jJL)zLiye@
zK3`mfulPuA#ViuumV3?G&5z=jPr`RuqKtRJbv{J;NXBtB;j>B+rUM8X*Np;O*r
zpvZ>|`EL#3m6dKGRKLo&PmtF8%Z2yEsJue=6^Jsa(1SM1fcGy3d?Gu)&HiID3;6!!5hwB=
z^2Lfzh_77y+#!AqI6lZ+f&AUG*Unx#
zDE=BN^>@HeimxqAdWO6~{1S!rwy0kscZa(IycBryY(4Y}wp&C$MI7rldj60H
zy}FM&!|%Y(Zx1691i@dn>|7`?5^?Irz29d>LDq`v=?+gv9@1VvT@ud^cdD7ygLw00
zW)?RX@vmU_e^I|I;XjWG{TEgPemhiaOFrxq8fCsN4E_2&4j%gn@n*(zoe8HPx1X_H
z9aX{O%tbRdc7V5CNL`(`-V}az-xJ}7$jixCSw84Tyoj;frhVr%c=y5RDMyiqI5%@h
z(3`P4JCUz>74bY?_$Ag~+4gk#<&E5kpW_W@|AYNcPkl>$#N+lAHqI}&gFNTI
z$>zs0cwAFy(Oehbe2(WdWL@n~#0hDS1#{GR+|!`Tr9;Qmb9SBUM+dpra?$QRJI$^p
zaXWvXG}??j-OTw*$IN@q;}nzr`_MYCfs6RFYx5ZSLe6>Y=@EZ<{&}l3`^@YQfH%4>
zuK9``v!_Q)L7LeACf_|f~|FKLP@-;aF9J?+e`
zjyIjL4n+iqeRI6a?Y}MU;BY(-@9*O~QJre;&DG9f5hZx
zT$;lf%^YJ^*ZdfN2c>GXw0m8n*gGO>6Mylz=3*hk1gxjO=i~$|kylL)v+I{aey#WQq~p!V
zx8At2-TAl-*0H7DC>P|jhDUaK)gcdeD*EQ&YOL3i_Ugm4+7QQQjw!x;4RLdKhmp7f
z#m5;lo3Y;B${GBdjQ(7AWnCA`ui%E7}Bl^%J-#pt^ZeRraUb1XaZ{KX~Y_3nqu?Z&<^6V^-~0=xGveen1+)?@kArJNkr
z@1iX#H^Z?HYA4sIn?X)XJ>5m;q^%c91)t)(27TM#7sIfh!`0H%qM_$j9JKfZ^jfw5
z920`x`pGFj&tpCBH8>(l`{C~d&1=+-l9Jug9Mnfn@DZoq1twk8k14_P1w9hK8^Jf*
z+wYhV58x%Zy*&ZGnHzsC>jUZ!Y?XAp13L%j93Pad51zKF?Rg~faehy48ePP`7QSVv
z{v6M}3tS-a0qB0DC4&6V7Qrc3>Hgq~^{8(0cVBBCq>;b$Z@W5Y18`BuJYOf^pc5H8
zDZkbw_8cEh;@neiwOk*#Jj}$V7k2t|?>t6)sjyy+@sk=i)g`1Sp6&RH{C1aC;P+kF
z|5~B-6CxpRzihg@56``n9%D^@zVY1=0cuZ6C@MJvdUvFk&s~jiohe_my%l;xLR*H@
z`Ei?GjPh^Y
zV=l1YJ+91;;n93;kL|6e=QKw8j3ZR|uCb2d(t=g(_voB#dMlR2`NKEvxRL(azeZ;%
z4)t2(vOLDu*FFoV{XFk$EItpuCCPY8MAQR@=XJ+1{^J!;#Nu_JEdAmKd@?hW#Z%q-
zK!^0wFpI`hO+o?^GlMUeu~0%vU70%SjIQ?9Z$1S)cri0ZFGoc^+V7&J(v;LOuYz9%odp&8-^n#yOvieZhA*@~hxxbiR
zhvy(%tm6C@6~cJ$kU&+&U(PiyVfJl&1;FSoah{Z3@8nNKvG8-CrI4a2}!k}j{L
zb{yodf^dSDyf>Ll
z=Ns4DmKxfJ9!v8N5`U5HJ8*X+@UZd598=)R9w%oGs1Q2*CE2Z+Gyfi)3ske!>^B1+
zzcn9!0sHUFIK_>`bI!dL%P0S{&9`dVvd4$n4)6~J$OsMb+3gWPv+WFPIPmp>L(k-frO1sh3kL;A3c1n3=W
zneyEm?HU`({~$Yh41O5b!Y>3IH*=i>|NY@jU(E}gfBir2J(dXlkEx+))Sj#LxtwLt
z^K2O}zX)-|b+ntb9sHu(sCidZp;uFVq>%hCk)p>a)Nf)SvTI2onPa%RPVa3w)!&~Be{Xki!f29dA4f5;4i9foP?iG%C8=|_Za7dh5*M@B|p-Mt2Wa+-eYE!Y+5
zVf3e50^h-2zqijvte*3jC;xMeG>>~ZO8Hy2H_o>*FB8fTXHD;9yBgwtX-iA<`C{;b
z8=hGKpFh-a2El_aWo?1KyWs411NpSa`ul$+$>F`9w;x~m8|ztf&s*1BwcN@r(xQIb
zA8}{G0(%txtKnvAC~h5@48Ga0VNo{Xwa&2&-BQG%zMY95oxocbd+k*0k*wz4$J^FC
z48r=J-DekX2ES;WcPRt+W5IVky-c>{ams7&e3BF|$KUsA>#-2T`rWFt#xE4>@aMK+
zS)X~_pl4I0?e^zz_Nh(zo3cu{wC$f;)#G@aXzrx)nseaW_x{aq6F}Z6{g%9XS2ia<
zO7FGx_7ZOE5RpFZLp*Nd??YYnSHL5iQogkgLp<#bdF6oluW2uHD+3=9HMZ#*g?Kqn
zF>_fJ;-y!9#g0LEpFhog=PJG7aSJCbTwR6t>2PD#q!{F1Hiy6I9SyxfL*5zDJ-4CQ
zm$Ku}c${%(ia!thb>^erX;;CgMog_cBn}?X`(pgcg~06Z`<=7C^0?}L#aY@i;JwKj
zC)eV8xbm-Z&0nyNg4zuWdEnhQFUyUy>EUt0j&ncFm%=$?cjjs*=-ECl%Jjy%?9yxz
zH-z5j;}3sjAzmK%IAq=m*jr?OOq$l8ke>Qfia$2BIzMPX4Q>`qNumJQrC1C*u=Z`
ztP|sx(29=og_7&}R%1*vcdW*F*CP_i>hW{#OH|SN?olYg*u{*KL
z>kafOv`2}@KpvI*B482l<(UN!DPHzEE?7@~t?p3NLh>`}doE^D9FkA|Yq|j_QBYAl
z1vo`>#0v6TqAu?qlAlYT{Hv3A!IgQp?jHl1Rq~{$zGb+tKl!zk03AQ_%M~6LsigN^
zaCYBW-~^%H%2Xbwc(;@CCpV=PhmzlVqNFmH`b`eXyO9h$SijMg{M3mN2PczW!==`ADiG-Kedn8yg@o$dixt$Kc#7&a(hIA%38!+yR<%KddRkd5kFs^BEsK|a
zJ8D_H+USb#i2AOu-z;8Ak27cS(njJY(=&b%!RqT@IkUJL^xl){P0YNH3n#pn#BWSb
zet9_43v35%K>4zX>sdUXz0w>Xgdvae+mh+oJ#J=tu75fiCSR3hdROX_nVm5^=|eJ{
z$ESF(cJxkqF};k1F09}1b0OoY{(;m7cJK4VJ(uOD+B=`fQ~B4T*=p1u&sYV27#;Ia
zEQSxea6WEO^r8B|%~J&kM~%F^oXXulhy0!ol=!~%EXma|`TkVD=j`V{Bu5{*#PTHm
zcRnQ1J<8Rq^KfAY|EU+m-fPyB4%TmQ>A$g5{-V-$CH1Qi+0W)@R;C&fQ&=ya7RH3L
z2m6YV-k1+oEwnBj&E9>Zc->GIFH7s*RNwL%t;6z@p2Cj66Nmg?oW{P@zGRm``zUDZ
zj%4EFAG6n(5pTG=QdZ9vm>eV?Zw$RNs~?zKKo$zvtvL~>`};>W*%Q*YOL93-{ooJb
zpYTWH8W&%oe$T1*Y@vQ-2FH)2eyi5zv=I*AB{W2{*L>>WG^Gy
zcP49v>pXz@8+4x?!G53neU!pCwA+|rGid_+!O~s#tX`vh&#TA0CiqwDeel=4yB~ZCs^zBN>ECj7C-V3i3O9$1#`#aeRQx3R>(oA-zpM$oZ1ak=qKqo;
z-Xq)GCt-*iib=^Uq!7pZmVa7?cx3lfTWN3>+Kt*P8n>~W6Wcwzy&({Brz%EU5$91y
z;US6j;6Eok$N4(8!k?W!&{aOThPyB`#o5yl@_5x%FVK#Cnwa583FHk%uJ0B?-ow~I
zEmP_(j~lnNCTd4#HaEfA|5oR=5^m^)=M7sCF9pmS4L>1XrYutGTZni$CZnL?*}P0H
z#vt&&uHItqt#|rq%cVT7^N@bQfjc-S4!LyO5$*5Xq`AXN_Ad8Tc>1LUtBX1Ri;J=r
zd-AyF2bMu0Jea{+0za?_ZbGY53qw7_D%ef1S)-{bly#6WhZ=07Feg`Qlm3jUZj}wZG87qx=
zIb!yjA(ogw*&qK2uR^@E=nrdJhxfU5u5Gh1)J<-68(F5YkXLh+yXpx4nKDRC6>{;TTnZ|@QR9&0~1b02zsQT{((;2v^I=BYFl
z^y}JuK;r@GmrdDOybA4Rv?R!M0B|
z4KSYL+Iv@;K(F@1%)fppzqV{n`4i}=N4H))4tbZ9W~(do^xo%e3W1*3y2e6^pI3FZ
zzG;G%GMw|M}2eI{`RpxyaqEz$k&`D=A*aR>hl=zwmYOoKEplc!ua&ikFQY
zuUNc1eQwRg>yWSQAKyjippI+{XF4}X7uy?jrLML{T-qZ@wp@bs4;Ff_ass-8al=~n2WAWHme=mzS
z+p301Q~kBD1B`!oMcJ`<`TdnWi#3ch{$w={eq>
z#N^(zG<_;J>K)4R6e&VTvp^0@U&^p`g9t;F=b;SsF61%&B+Lrw1**?9~j>t_AtXDpvls2Dqj>USSn;zsE4!W0K^+_Nu#5l!*Q?^y9{DoGU
z|NG2%hDKf)n~yV7#1ly0_-q%O2d#bX4BO^DW_BO>cVP-WuQ*8n6M(q=gg`M6p4Xx;yE-aCxqbiLkeZac}pOxe9i&%LL;q`%Fh6hD~2c;71+
z(>_>tQvV6fi}lK7j#NMW!H<=Mj!JWoFhL&C*I^m0dl5+)De6ai$$FOG2%oeY2`0qD
zVzU*b=TcJ4zDqv0?)4(d2i$4&mW+kp`W<{IobDq6-q@MF14eH?AhHbca`VHT&vqft
z>se-1O1yq_sGj?IjEgr-(BB3Aqy0fwkw5(8xJ=Vw(~)mo@!Ck<5_+So+SuQZ)9Kf=
zA-nqO6HgQX(^DOXSLoqfQ#H|)_>FDa_0zPU^5dSpro4lXi}asFwBLHT_YK{*g#36=
zNAqR-xn4sR=-F~~9?e^;S^gHf56RU#(RhS-K|zSiV0WB9M&BsBi1=do^;4^fIGx`<
zhWZKBa|tg79f`y5!){$qJo{4x`}%C$#J!_JAfxIkf@2JY9l)!!ll
zs<_Icf)&wR7FQ_eFR()QC
z@j+`YCx7zrqzzASUv_Wzlc@7NE^lUgUQ`dp?Xz-X4el+1!!Ia%DT5c>c^m(BeF67<
z!ot~Wdh@tVg99~6ax%GyW<{sU>>MusuJNJeU-CKAhj|_v&mVFhmpncGFS>y1uTpCN
zvObqPdbaxS-(3Y<)5{MP%a1+ce!P^GyeU@9smtZPx^cRYi=1;kYKv&c9a*7VuCrV%Cary7}CUj=u9J-_Y
z_;Q;vPS7~>mvUzb7oR`RtnXwE=RN6mJ}&~}TfP0B3i2IsEzhSF@5trS9#~6%*;UTD
zYAnhnewp5oX3)4E@iwo!^&!6NSzvcQ(&tPew|T=^iNN_aT>JXh9rsV*d*VTzwnxD8
z1Xj#>{T}PjpnF=h_|#f%iRV76J>VIBoi~+QZ4pmJ?=Eu~1^-|hV61_>k=BSExvApy
z+&lH#vW+g_3)V|IZz;fkpQwypj(dc!9)JAWxx*Ts$<;4k4elkRQ<-~6s?b-@7ZxlkcivkmLps>9?7ofpFW
zWoLXv{zhdTZ^liuFFVF~>I;3uljF~~A4fZ;ulka`vF|&arfu+8gme1W=fw@s^P6{T
zS}5jaliMSC+UJY%Za0SkcdYQ*m;_ArP)Hnt_EQJhUfu$^;nN?E{LMyVO%BUAp(LqGnbe#GUX{%WdA5ApqGiE3TSPkfWUKJEzWPn!AckwiOJ
zLND%70Vbx^29iIG>^=GEGV&_(b2pU=La#_j`wQhu)Gv<`ut5E$vtJ(4IZJuwhKYoF
zxkKFKfY!IwjfgiiJuC8}b6e-9$K7A5T&!|WHV$hp(?}_zoK{3i-%*EqCC#UfYpELz9
zTz83I{>%a9p#G;FUw%FWv;uO=U0iEvw52Ukkw?*;)9#`zVS?vDdN24w}K-?96iwChU
zjht9JA6{0NV`sZ|66*4wu_3zy%iWA%EI{@Zx+}digA2V0#JU
zo95Fc*nBBfPiEsAvFikz_q5GdF%bMNZc@4~@i%*~ZKo80eHNt?T!EgsnP&0uCq4(C
zY#9uHCz-V#U%o?+GuTD`Vp4D58}buPU1vgxrFWxQ&sGMCEq+}>kQ_~A5XFM%k;fr;Hp
zrzLX`Clo_lu0Zd;?U|da;CIvVhSaZs-J7(%A`c*bjDEXz{t*kDC*NPQ@It$R^ZjRj
zM*WseA;`Em*P)S+-{rjH(Y!P+^S-Q9o@lypWr*r
z@~hGp?2xA!?lIE)6ZoXg^dGBEfG5wa^Byz=_f^u10=BF{oQ_P~B)+m3`)pmc?{7tKluuIkgn8#CEF2ilfV4_@Dp~o{1vLXfO(uz
zI-PSt9;kS)_G?Ke9ozr@wY?6=%P~d1tyix^VcT*Z+fXUoOt!*f>Ez3xxdy!^41&GawlT%ju@TK<63eC>BY}S+_zNu)fJ5Q
z>Z;p%;!!PkDl0Z!1m9u!RPGmDWRLsS*{`QaiQzkm%hA^_K>nWPC0VD0_g|T7_D&GL
zbGv(Q_8vVRr@E}CAV;T`bNrAfE2Ub^-79k4s`8_pJO9A^^N4??T=wIjC>!LV^j*K-
z+n-d+y*+#C+pksC+>>FkiP~xvocz%}wxvhQIo(tJBc0FkxML+-Yjdk=xjFfvq1qUS
znU+uXTkTrTa&)ayqA>DTFK?M=|K@Rl%2|IN8uPe>sY{3TZsT#eO1Ivm;(n{r&?-N2
zEB22`TKeu5d^hd3Yxv|O9ydE@WK-pE+*64P)+MjR{p-DRs?OjoR+{A=KL5(Nmz8Ia
z6;3GSG-i9xJBRsCIzIO9xp?G9J$Ljt9l*ZU93DPVpd9P|KaZIgs<^by2RrKzcdv9Ef5hd*sn2S5C5^6WpX8zJc=FIB9wHz8HNssP_Q7%Qj-%Yva(^cAzeP9J8d1K>gSX`VcFaR~*j#5D?6)bx
zsR~=rZ_T57W6AHn_Y=A$4t*8BGK*^1DU)+}=pD#S0git*14k}UHYC0%Yj4s=a{7m?
zBFfY0EQyfZ1zGmV1tVGDk6Tqyeo9;OLGr!)<((U6i32Dc8(!lxi9Wh6tz1w?p-1Ezqao3
zA8J2aDqWj!qTj;~^7kKHi-r@9SoZ}43i%l?ar~h*=-kIcQDhQuI(@!_lXgos7TgFiPFUQ0(zE`&>
zndO0ee$Hm`CrEHO`@8u90jAeTZ}@Pl)%5&?HA5L63CJAI;&OAhi%OaaRa8Euw!r~2$AlNsMJPpn|^S@fbJi`$Ng`<(qoLej%u$Bp#6lQ8c0<|<`Z0?vv*-Sfrp9y0oP;J36n94gulSL2#?6{`D{N%OEP`s
zP3TN!0S~G3ap8Ro7qSVd1KS7)$t++hnGci<=Hp{GOfTe;PU;xTC%s>ehb7UgVZsGMTT^vi7AnO()<9;`g*0DpZn#q;yP
zg`fEAzoD#z>2DUgi&uvAuqmABhX;wV@9M&bWGx~d>3mG9i1Ph^o_UTSJ@@nX))M+(
z`QFO>PDy4l$r)W1`^YcqWaYAU9Ikg`g}`5VdM_gM8*Gh(3)aQu-51GERZrpf=XKtZ
z?KiXj;aHLIYjO!JZ@i|lqLt(XlUz2Qu^qcvyIYp&uz8LCaS;bT_(=g999+=P6u*{X
zK+AXet>h=U(0?bbfH6}IY`hT{+V{;$7!N+V|HPt5HXg;6F7nHwQFbx1k)5n*8f*&;tKlvIAo?_i(JANs6iE!BKwcX%Ft7Bs)$YK1E
z^DXb}1HUP?IS^%nylhf{o#PAG=^yUBNF42~4c168gDk36cti#`p|VQOSrvYA+wM{Q
zn5P3pYESD3<&q|!g#PypeZBN|kpsJj#_Jd%pXgVbS#zY8dl??3aZnlew4VF4GLV;>
z8M=77KJG&g4;y|%KN|B_yTI=w?4NT`)^-}s9}}(?YWM-)w%4CI3;T8pP6#Z+yrd}?
z>BOPkpv1F#Mgi}i4J+}-_zWH>tG`G8XEfg3aKQV}xMtP!0emQL*sBoYm)9o$3CTqL
zuZ@-N_ONecLg(SHK&kRsr+sja`697=(|?$k8b{Icbr{c#{VtzJ+*Gl%`xk2|$
zM;9Hz?<^En4W983>$-5K`$C-m_WOIiFhD!WzYI;v^YFbmmmKpK@3FW0tJ6;8HOq{g
zq^w}4*z%C`^2kH#Nj>kM4L)rj)3yEwc-Y*+Aic3u@qF%xuvCn5p`o^l1D@Y|#`3r-
z&OcR$c8zz0T_pj7wwwhX>sGAa3Y@i6_gx6=%=Zn`zX|(-M5}w;fLCSiI2;6~R2MI!
z?;M;4&5TvTynR^}XV?gI9MXLGWdjfV=vIqL43As1=gpxOVerT5&r9BIKpvp=o6=rwRDECE4HOj_ODGx2*r<{^=WdzjP(nmSRFW|gLgu0YQ7UDghs?u0
z*F0pN=afoB=}Q_EdG|i|_PoF6eV+IJ^Lf_VYp=8R+SA(m?7I(2m*BE7LjGQM{`q|h
zmgD`$Ffy)O;m1y?f&6HZb?{5J(EiCRV^N5^z271n)0v;9|V$S{EP&AH;3;kz++G$^3{EeNEW|eM%pA>`phTFe9MB~5P
zVNNDA&tSC{o6&fCx6*Vk2C?(>URyMdpILJ@81a`pws@mBcO+w18b32neSseoh0l4cCyw-qB%Q27@|&f|8^i08DB&R(>|32?$eHC*4
zeOpL{0U8{Ka`H)gs$%8Ha-WddPR2RDLv$2A&rilLcnmzA1m&mWG&Uh5erb}=IlzCB
zOfM5>Bjb@{ToU9wFZC;j8OiBRbCB`!Xv-0oOU=HE04629j#bRe;Yl^(atZlTAu9GB^FrS;Nj(qBn&p`_fSTl^%7=K~VM
z>)_*{9??2m68YwLl2}#_ivjs@*)TiAZXAX|5WjVBlDKe}p0Dx+J^co{t`DYv)uPi$
z97-e|!uU@14l{cCsdC7Ma&|DvL%4|kHVTm`l&(+RgPtBMNXK#*ouYIe1Q3&vdWcyl
z8^qnH7!V&Y@}Ve{Oh*9usSJ9FLE9Xo^Nkra>YDEtes^p0Um`N6i<~Jf?Isai{BdLCYSlb`8EHaSxjhR0!Aaru6lS
znGa6#J!R3
z52g;Hr2NpTIp}#f-(BY;(|;YI`)6YNX_DT29)|NU{dJ=ybS#`_lFpHQGkJRdzvhV=
z2Aq$Y!$<`4s3XF4O8eJ$X#D=Q61|{-{VO7Y2l1N@{5XW>iHv6dVU*5yLV_IMi-K-h
zqjXgphhZdtKtN|VN|(&cv_kT63)ZmaVLUg8k>g)*SPnTphs~TAgs9g!>wM3Si5GaguL3DJNACPL^*gYdk;VhZxKF
zBCPrJmmTe6PwWdK=ew}Fuv=)Ku7+28x(4#Q*2+(DLR#g@dh%RQ#JMUTWQRk%D)}DN
zw*Q_hdQVXzq<9I*XHbIHqIE*6)=zRCSvryS5;L(D1ayU#iCBS?V
zdTLQ_>q69JQS>Un;MA3Pbv$|0J
z>AH*XkMpn{j2@QOgt&fU^>Yb`IpW3o$WE!J((fuDD!HGzA_#Hk0WVf5i0!`*Eui>h
zT^sLd;)Bm>_^M4x)sUjq4V+OdW>9+vAo9g^TtNG`juLYA2eM9QPP>=6*(o7UynK(x9#$c*ZB%QrqV6j!F{#&&W-w8m;b^FXxd40wa;
z*&YUP=sh31n#*w%hiunE2g=|&R449EH;faL;|@RA4(-)?S>OB(VpZ&UH)m*vjo_6(
zub^D|iL(A%kX~A$bSe;{s!y7eDa5Qz_I&veO_fje#6r1&yiF0vUZz1I1t*9qO%YM{
z5Eqv$@=im{GCXiW45HXN(JW-IYxf@qZiqTh_8dp+;NXh|v7q=W
zo_3HE)zkUvxhd4%BsYe!qIP!I{pcvF9}{nX@m;VSt&O(_rC~dIJ;54iTHl{b!+L9up2wjwy%f
z`+LYg{p&&SBEW?|AO+KK0%~Exq9Pe0qw`-t!I<78u8}&C0xmz9-h*2R)
zzKBj*ymJNV@tm8kLG{bLl}}Xx;-%%luP9&95ALRY5c9=b#1Q>zX!I1aPjU^jGUC7X
za3}NMvH1{+_^QgIa0Y|zV4!&|!oCOmwg~N>c*BVRK4+sFJHl5ArsO`Vxm?)P5^jKJW$=Rx0JwENyc#_eVrV*-dTeVhJ&*W{9Na=w!O
zeurPcmJg1Wv<{PZJAYgRpkjG&v7dDV37*&4=T1q}=E2A*7sG*gDb=8F4w}bAX{P
z$!DFTpZi>mJ|_V6sC0~w&ygWNgb3Sr(0Y=;_&}e;UrH!Yh3lK?cvAniIv55(`nMll
z-~AY2k_MA-+J}6B`{N|;*?fb6qgyE8
zh5aHB77L=W!(9>!Z^6eyntByQo^V|Cqo;cs(rKSb6zIYB#@NSR?1kzf{a%6|Ke1Op
z64JcZ^!0`EM_6n~<77Y$_qB@Y>F&r3$lsAde_rnodi*DOhtAJHVG2x_drQ|l%Xlxa
zPn9lru!A9oj)g!uOAb1HypL|@SL9yEA26chQ)D)zy^*^i8lr7?h@FOXnq01+bmTsW
zQb-NNdKlFpJkD50cp|q#`W+*^8lifh2I8aIL`VX}fATQcBKiOFk;VHTxqta#BPZqlh+iGi0G@;%b@bS4I0oq_-S(8Zu%GVHNu&_WC}Vqi2w-!P;8KrV7PY=;cHn@K^q
z9)|x4K>W;A^!{`WMwN)>hNpZGYIxCUr}K1r%yAdVzs#`nCh0)9PJ*K%l6SsB$7<4}
zu)Q+MZ@}=Q2wFx!xgA_|v|zlC+V_u-*f1Rq_9Wg_q$5A092%o8B;QFJl^DiV%BU&8
z_(-pXM8QaUeDBY=*U|DJyo42S3!ni3dki^x`>F=XeagSw`5E>9*Jac)5L>obPYzWeO+O5x{IPCBAPWXZf_D@}PT{M55
zDL2`I_|0Ea$hfQ_=@;}IocDqS$#VLa+QJqA^~yU$+SjOdzs3#8ri
z0#{@Z9q5#}65)gD{7Lkj*}^0hM2B`6l6GW0QGo`+aV2O#8(~q;2yB^fyw;6|6%YOP
zAer==tr~s(D$Du?wkWusE^dci62_59k3XV#*+clMt|&ygudf17e4?fs8-e)eWuMM>
z!8l9F>lhP?e|W`u;Rp=(8C9wGP=2-5*>Hk|>!W+TaoZq{Zk=6?_)W6vdyqdixEv(k
zrz`x(Z$`7F`V~me`>`=H
zzQ~G~Wk+&eKFrw&AMy21Biz&dXT=(b`%AleP<*z*z4!3fvXUqNVqP%co)W5N=!*dAa-X9(vaMf`fD60y{Xb+j9iZjM{=R^IH|2z
zgY10Cw0I2
z^kaC_lLkcl@>TgDl%9~#M|k7E3uOCmA6?EO=~(4VgtdiL?+~)YMkOMQ45<8w^cEhR
zy=4hesVfA2fED(SwP|bu5RHxn@}qSBRQF_5PNn=uET~;s)^1Tl>&x3J#q20<_;h7lIMrjdJo7;xFCqUPAt;%%Y#~nA_&-gwoU9+h9Zh&uJeua7OD+
zl`~=L$R0Vp>@R4&P;+(s3rd&KS=)x5SHHO7Cc+o13vi?_^!pzw(!YA;ih3j`w(Tc5
z?k(S|kVpK?hRG|4?=fBPiSWqQ0#1ZPcP)%G;Q!;>K6N5&W_Ks^-3_=;jsqIG7BB*V
z?cqy33^CyPA*raE)YrOmJ)(DXY%)X`!zln~O6W)HE;4_;#q2E$KyH5DAIuI#;@7r_&u~qmmHr|j?>54A3u}GbS%e@#J0hcCy>6R!%fml2hfBL`?Rj|xc*fSBq4vLjV}3IZAJPz<48Fl
z$+vq&pVu3F;WPl{ascd}5I<~&!#%`EA3E06z+yxCvk{$cB@fa-yUs&5Kz`{3s2*a!
zH8dMy*9!;`ZSLSC_H@(JFT2sxo#N@~Mci~WmxJzw^1IyW`WmC@{JR_Jh~9dj^l;VP
zB)V>)+u`0v=llPl>n;9FM|WR(`r=x8x&<8VQT{qq9T0Dm0UumXchhN~<8-W&prbi#
zMMzE@nFBF{5nr5utr^ihC>((Z|2#?}89qU}0J#z32y!dLXh#0Ch=jB?awEi3WS|b;
zM=|Vpg^CaPqR34U4H$Ax8g%+5gBC!=gXwTEAQ5f`lK2U^1@ezEXxPa~nu~#23}j~L
zX<*Q#dtpA*noxde1{N?79)ctJs1{K=SsaL*44N!1r2nOp=0o})U!NgI0_2m$hWL;E
zC;#vH|MZjYg>wJW$?Aagzje;P{Qo~B!(1r;PY+q_|9S4eeE;;r!4dx8|K}KJ%D^B-
z`B@nC31qXxvSMGo1;$iT+8TjD_G7!!yQHTNK
zE^ay&@t^|1^^!AE2>GO=VO&Wnfa8ueU5+^l83^M;^ci`GFnA&H9qB+gZpqN;jx%)B
zK*J-HL${40{I|~B&9J+kkIp~vl&&{|q3;c&-;tlwh568K3(9YRhINS1fv9D{^GuUS
zF77YAP^(3PuAB91s3Fl2{yil9YFxLIr^7
z=CwhC9;bi=<%K2l;AS=ynF8=?K#I7~kRQu|zEe?n4`ElGy4_muqr?B;=p|
zLZ|n5)BPXujy}&=>TZE0g7Z%GW#|$3UW7{s76_i_avCG!hnuZyFCv=gbtcPmV+yr!
z7!OF%%VSu52zm;}FFPY(LE*YL6D~&}`dAB)czCA?i8?M`WI5$`Opw@p7=|@auHR0H
zOus9@O=6n17sH*Bqh#-bf0H$L3Az$nOQ1IIMC-{I1XWP6&_8
zZ_PreqT%C*@TE*298%!CPW=YEG(1P<>rcjCS2=jMA--hRU9ucQzfEBmh2#B{A&GH0
z7hsErarOo!*dU;uPtC9a!g%R?H!1H*BS`=LiQG>j<=!+Y|K?N~Y;f@0-ugS96o)_YuA6KtJcril#1w!+1K(1p1j(nO$_TF=UN5>VKa}{B$aw#4Znbtv}@Fzt~C2b5*&wHVn
zMq;7I2$|39JKgWX$2>^Cnzg{?9h9%Op!aiLaZi#TpCe2nw+g(*0H!Yk6!^jTcdrqt
zPfk#BC!!ymMKd^Dw{$v_{L?8nNxG%Bm888#;Bp_PyT#Oy^ap-666ZM;NQ~FiBQZ=G
zRuhyPI|aK5#G`9qaUqr$LAOFIGN!LXzMuF->c18biw*hmk01$A+KfJ4^Wt>=$NaM-
zf9V1JIh9wTTcF%Eg0Am6W1jAxWg_j<$uJ<(Gu4GjdO~y$^hG&A?%J2
zKa+tQI%fj2LpsipE_cNlHWNsWO;->L8T@k$J;R6%`ArP^
zBO`w(gLY%23o%fefkO;EsSJ7x12q^32RrzK@f`zED@8O}EQtTo;b1}XQEw;hV?6iY
zbN;3CFw#*kMslQkA);1?Xw-@jlJ0`|FTWOp4<{)y9S%k$lGOp_V5cN$oPngfAfJ2;
z#DDZZyZ_Oo`2WcL(^vdgGwEt+Gw_|63(JS+cVpzOw!H;CK(Keks+tM-f3&-5g^JBx
zpr+v5`ILkzaO3Z~jlsuX0gk1JCjgpQ<)Kp(Cf29FH)306H56{PbRs3dtg1-lBiaPP-ms%@W+?m{Qx~vxj_}o_X
zu!i?>J~*^}g9|*D^6=r{cKH9f!@u<%`e&O!@|?d=OlBMCySBWP8Q%%&6`uZ8354_N
zf%y2+j4t4zbyR4=tPcF%_&(Yyv<0|1?g{an>jaUO!wN?4;eF2s7F_*e;eO&|yZ6Yo
zI^gC~%dR9{4K!;Xdnxsmg0Zkel^Wu?K*UfWd~a$7@ZB9y^~$CRSWQf8?7LqJ96A@5
z*ab>~oQA~Qy0jelJr<|myYQMK;tUxP>+4MGDZ0#%f-3-pooV}G`BS%-u^oM+q
zB~e%APt65}fi}gCd)vYBRGH%RQ>8%Xo?iVW-cm4=@k}P_WC;iX+3(ZmlE8d43r%mg
zBMAIT$ZPD#sF?KWs$}|63L@%^t20G&f#mmK3(D_w;LT~j6Mv8f>g9@SAFj>3w}W{gC22+Q+P|s5{s`Wzv)1PYa^WX6KPwFU$p#YFeGZN+W&w?s
z?6rO%8w_<1J=S+i12x@jD{z$zAoA1Uco2U!peifHs_5l_)5BNRiWcU8T63AX1>H>W
z*f!9(zAO{GtzPT4f2k0tEee}uO;iJ>?Q!|XOdCM{pt!;2#Y`Zn)y|bVl>u@t6g_*)
zn*~leT8lMA7l0qXkLQTXlmRX&I}6{?9B{1e(k5xS91up4D*Gsq3A%?58AL(d?N>7
z3xynmc{$*+jzb*vPAb?|->o!U76n43$|@@*Zv%^nuerFjnZ8nF0$c-Ofa0u-+@lPYm612OB@9yrEd0^}Es9hJfhL0XyEZO!EZcz!YS
z
ziFA^VST+ck6y!D0NCQ9G))b~!CV}8}i9ZBraUjE@k~isg2-uk!6+=Dl22!JRRcxCB
zfDu2u`SJ06@Wo-`?MS6P(D6NS`oC8MkV(&flL6bk_{pK_#f_z4YJN5;M}Z1Dn@m+N
zt*8Vi4Nmn2ttkX?MZs0AdilU{qS-B5>?IJt-XHADo(If#)+z+D6@s$Rr_sVL8K5-b
zMo`2|I*3lyviDBN0y)g_|INI~1yQZiPnfy#L7HvPjgwAsfX_xsME?0Ruz=;%Qt_6YTTQ
z+cgYKc(|te?WhFf2Mzw)Ur-LZBL#5L$Wp-OTKneQ_X>EA(}=DUyvOLy!;J#5MFk+d
z{m-XsZ3Q5V|0*W+sub)B+;@4~no1CF@+e})tO_Kxwiz`n2e7JDV
zm@A(Esu9<&uUH=j?=^MnzCZCC_#Kh$`rMrhd|z%}yA|HQ_{BEHt?nEZ>@DQ;ZHTM_
zHghH?B~R7Bc+6VyXnP&-#wF~|Y-|KyGau=4OT7Y_^Cu*QZdL)|Q?8RAC^f*nnGjaE
zRSzy#dAo;ge+5b^J#YSIZ2~iV3atO>RDyHo?_Npws0L2!t}y?qss}pV>*lWBYyug*
z{39+0TELTzocBkl1Q>l0U4~0k0k?8C-Pz_k@aj$WUxB+#;8?NRH}IhaSb2(vm9kO6
z!>cwNCtNDQwY4Yy@I9&nZ_;MIJ>+NxMIu}EH}JH9z?#j
zM>*TU$Cz`kkHh-`%2|$B*zbe?*Xf^nz46B@FfO#=tGZ?jC^#|DL`Zai;HwF7#r>V2
zaaZN{OH#rqc%cj|cT_UDXLzu?v}>hxbD|sMVEePQdp&+%j%5
zpId;z^SpukpdH?Ws+bb+rxRQ-obgDI?E&qz4HA1mFYsxXwQ5~#1z+4_9Lz6wz-ge*
zs1|+)<j5;|cMi%cdO?3@@ZTZ5ULY-{wp-H<#w-2e;%v~)*P^*2e?RwvGxgR|
z8tnaG#b%@0u&#b^&a-GDn{5Dm_%GYnM5qruzWZ{pq_+YcSC-C5@jfyLU=U|69rfSMzyzZi1D|;KR?~+xNHBI6B
z8Rn&*@87}my0%|;eZA8IPA^KGZHVdt9_a>`GhiGQtWuNN?A8l*x_^1>!PEzS*k1VQ
z`K%9|tV%Nfl-virzyGQE=+p;PWgiZ{XBq&+^q6Ce*8o_#G16?q{y~t+^XO~A$wBbm
z#Nm$K=0OnmsHDj%Y7h*5QpK*R4S|sH>UDpXhX9|@-s0Pu
zzR!7FGWOsSe1Gvsv0f8?f2RKAiq#J!y1;$Fmi|WveqW9?zj?{B2Pki85%$4)K*R0C
zm3&v=dFa;nPO|1*K-MPa+2r3&z`P*YzQL^zoUZ=0*7{5z5S0mkKbX@Cy4qcwKjJ;$
zj{uYUW_Vvhii(17k-w%oYulqs0?rl*4{Z7DtBhvrT>P}$m
zs&n9CUI$3>z8p9=+zwh!rcsLH+QB}IV@QUTExC(Z~buZM&Uyoq+eNB#*5V
z{UrGcWEp*l63B-4(b+8a3sqFZ?>yd0Lx!Lv;3
z#3mR=FA*2Va^QV?->0{MEca60AVs05YB_{ScQoImcQg4vUKUWTn
zHO`cMEb2OUQ(?~l#C
z1Y8aKURmbkgI9e#tG`53!MmP>$v(q!V1Ccd`L$*#2xL`RkTfg?s^fdl=9lCG|3{6*
zb7g5j-Y!SAc}Fo=!8fYe6k7l|v?R}KT*w8|ax>NqiZ6kIyknA~XCBxtCD@or5
zfS-uV9J6l%g*&s=TSMEx8L#N2&@w70SWH)+JzENHJU?rf@wEt;i@n|V>TMagd&$jc
zcR&T$%N%mr)DEr>&dDgtiX?+$0sA61zDWYrvFFwHd`<*Q*#aMRb|wN1Q&W}V^?AVX
z8rR}~*I$D9XO>FOZl-~;_?u_W>7)WN-Y4(t&Sb!Q`;28Cu862`O|+XUig{7-D4_aU
zWXqEZ#RH9(c~hQMgqbVOu72^fBGJ2!WxYomIJR{r(m0?M*k|{Qng_Q7ipSw9fj)ST
zV7zoO|56W-K6v}SZ&U?HZo2*Ki+BatwDqdvM2aMHL1mdX
z`^Mp1(5JBT!CK!upno%aO#5dZ@H{%pe_%sYg^GehVz8ifg;41EwWBY5L1yzp{Lv8?
zAYuRa*Te%)pd@#-pw6im#(5=zjdEQefc@0Hwz*31{8MpF@{c6&((FOD#$r17wo`mZ
zWqLjQ-Z@`DMsfu(Z`E{v=#&MFw}|9z{hR`3qNeSJ=d*zNc&6fLZvy;2{bl=IFA_jX
zYpGG@N4Rb{B$0EeAFgNlPw7`pWdkRz$p*oQJTP`dBgktxAME^mqSXe^1^apLE&uR`
z>#i%yQsqpU@ch>+uQ>rKfFkqn@J
ziS}#L@f472CR%-OI2AD0a+uex%K{2VJCky&bKt#eT-(lD<$_YV_GSJPxghqryS8>`
zCWtu_E}(ZW4RBf{pEc#k0DleH>Lq&3kkyzOMo-3m!irI#vh2
z#}rL*|K1L8!#63cV{p7Pq8a}8rWc6#EJbaT$^rvFK1WJ;=YhzSyw?4mRB&Fcs8$rV
zt4D`2-b+1h0M0+m0xTqAqqLFe0`Of#E!psjak`Km`8@I13$
znr;)oskJ}|7NW5iS}iaT5v1b74u!2
z3b^*Wb<}sHgRDpAP8;2900uv<^iVsh0J~=4H;H2}!GNv)K8XiOKyzGbPhe;!I2HLc
zIS`KrEX&0E$Q^J!!6}w++5-0pUOI2y(wPgyG9RRf`(^{LVhPS0CeZ%>p0D*TgX^&y
zMSAOQ<%6xBwD&a}nPAOf>D3vcuzjrX9vlOSK-Q?%WZW(qykDSgGhBKKOhVKfAFqa{
z7h~()uJ(#uOv3n2HBmZ`G+^6&KRb>+sls|MJe-&|ZO2$u*bbg;RKd7>Jz6x>JF)Kx
z{1TMR28_+vW78mW4aPIIZ{3(oBX-|N%2VDw4ErF@d`~ptCH3$`!@er%!
zXI*_DBMmcq!ci`Ah`^Q-c_l&gOUxx|MU!Yu1=dje>=OIJOR7|RvMOlk#ESDn^3&{#
zF%LC8!RKEpFizpT_PMvU*lv&R3%f@Ru=pePPk2Y_u+jA+Yc~Apz~mZ)_coh6#=xJC
zj{VGO7)bw6aLc?7+pY17*6=L`v%Fy1(ziC7%2e#2)flTl{l3{QFZ5~<)r%vdX>(T%
z#wMUGswi|1Q?UsxT3kq_j#a$)Fz}`bQ!aOCyQk2Ic}k{SIeWMqD|d~)F4dGxt#0`K
z=+65NEJ^uRjqv3z?4yuR^Vt4YY~P=e7;CO-%=bC-^-gVTET?q5`^HECwrit&nEI1C
zOxmdM08dXLcJ=#*u4CRGuopqwbo?BvvGkI=k_9@QSi;SVrV1r_)Z~`KJ2;nWv5P9S
z{{p)js9cdQtnKAZn99kCf>jd5n9p2ZI*Vo^c63JYd|M9<`+B%Wc~is;Cf;`SRQBZ@
z4BoHty*8l)(>nMkG2v)G<`}t!17{1P{#Q?Hus-pVD&*Z-t9-79+7x);p5s4ieou
zQ%hxA{$g*G&X-Fc`auw!f|Q`%B$d7I`}6
z4-@`4a>&Qh_UE;vN
z@7a6TF5sOJ0&J`N0*Oy02c!xOwD1FaEZF324iY&V1UW^|@51F)u{xZSG{Zp$nBlO$
zOGvqKI#++!BQ7X^l+3Wg@S59Px6E`85UTIOEI74I2=0T%t55#%B#z7ek{v%`irc0u
zwGsiU#D~8f?|gU^@XBjP0;HWRap8vY#J6nWc#?Ty6~5A*P>h;USA1uH-_&+2nBFZy
z>|t|syT#*2_-d4#n|_l=Y^WRQkJEWUn7G`P{=Iw~myO>Pbk{n7STFCdJGm)^xRvwN
zT-{@k2wFCq%WN{jWk*wOQ`YMd`}WgXyH%_S%?Hm9JyLWhEPGOT?XBM9F5mf{_A)0C
z(g#&O|FC+2zkRC_ta3e($S`M*&nu(i2iMhiswTZAHof`q3_6;y@NNjY&v3J(_0~?E-K0?e*%Slp7nTBKvIK;h$;e?)#;EBLszY6Hk3zM{(cb
z;$xG(nKI@fbY{G23vKA*)XCi`9e6=R*yDL84vMoPg~vi@C9UG(@AzX!j?t#4b$C?f
z6=?^A7x!Jddz9v`^ipMY=0S?3@N4Jt6=t-?Mlltq8%Jr|UXB}Cop7ZD7yMlmW>cVP
zpVYV9!=prL_&mNrS3r}tkGaA>``jVgR&l%9ibi|N?BDb={I)W*BKPnMo|3{8zmf9|
zgHHQs?vJ+#7rI$dSZW1Wm1B3%4qV!A(mbI=dAq%AB%)M-7SH$GnnzuiBBmL8^Td8<
zT5PL4_s-L2Xi^-U=cc7>Y09)ie$~yk6qc2tPC>EPX_xpOC9J%CpORuBWv4wIPO+oh
zsIy;VO54OLJvPfLLvtE<+niQnNppP?KQB7$}1WhxtOXcR|6^fZou0-C-
zvy@21b)z!Gb?xe_Bg;q4BNKvd~jyB~x?xviGH@NA%;RNkr!CXOyt}-k~
zR_VNqJgv#+N#o-i+i49^276cxtZ1dOF&kQwj#He%#VvNl>Qk;N$k=TbP@zOMOQa?W
z2~kS!AMOjd{*Y1{v|ymW#(^SaB)h$E!zJ1o>sx#?W+t>@l>t68>>$Oxg!}6Q{|7Xi
z9nCCZElIT4LstVNn!{;a7o6@02j8X{tGjcpPjaI19Mb(Mu}Y4@TB!KpuOc($#v74F
z5$RvV>eXz6lcTJZ6K}S26Mx-}P+-)czF)Ci&U*MdxxuqlNrd~5yMZrBj1&kljK(G!_C^IR?p+qpg4Oxg}m?%0U!)|haid701Hrsf(`
zcr6TcZL7s-J^eJngQc3Zi@Qu8vE0?CJV?uZ#OXXuXi0hHGyV62IB=6^E&HV^!d>m;
z_AMWh@%?T3>Vkh0iCd+avKb?+lpX5}2jXu2!sSKOzvUWd5~`kBoBH~K@rkUYVB4B-
z{7b8$U5>|Qnrc=3d4I}SJ;Xdr^d!_bf5EVw%$6Q?l2z#yPLpnZ(D5u+t
za?W1orZlr#tgGw(PRz2Nd(Eyhh?^>|`dlqpODvVBM&GP3piKUJafG8pjUv1l|6R*|
z1;qzH99sPK1(z??cX8y-CSr84qRho3#JAP%M`ykm;K_!IpWYVZ#HvIq
zN?nc`KGR?Nl8GY%clbQ1r?kP1Sl;;c?Iu2Hyh3>5$%QY+h`&`MpPZujaDk~UR=TIp
z6Hl3L?+i!{!8Ok63hI3}Ca8|T1y_oi;{0*f2Yx=(AWk@B*(OIc5fKS@ugM)rBUIAU
zdc`05*2q0d
z@Flb>P>CIQn)B)ErHai&vaHo}f#fml`C(!?W?umQTjfXfb1gmm^hn|>19KIk+$DZv
z!f6G(zV6^&i%SQI(YHH9C2m9#KiAAI9rgFZZJiF>`qFuxIR12KCg+?sevz`}Z<*yu
z++{(rvWk_5khQ2bkY6!@UB01aziDSFCKcw8w(4aHCaxJm1a=K$hGx$$q-=eMNv%H<
zV#YN^JvZLOwv^e1_1Fn|36{OYRxii0P_&XTX8T1$=_{*npNyi%XVZRTQ*uKM@2Gv$
ziu(Bny^or)W1V3qW?RZJTSDWKk7Qvy%Fn6{JA4_p8#JG`8SHmMn?J`YVVHj_mp$(>D_7?h`JNOGDTf
zHWROv##z{JrJRW+nQDw{LFV@qTO2lj^0t;6XC?M@1bf?^R*H2hz8%~
zvuA0q#xN1VU)7#pv28zPOn5~)uoc0@o%SA;n8XHwm&GzU*zJpRm#^>1#Hbkw_O4tD
z)GPj-{`)R`$JXul%T{S2sMS&W1C+8(?6ySTaG=5v7T$bGsCI8DX0^2L=4gBg#x415
z@s~{rHs7Tq*5y=!g`SIFk+Sm%c3$c}lYn;xhDUh`^6YEFj5ah{bj9Tll
zR_q){jaDf(v6}Ns#kX3_@9FUmr)8?Bfm`n%_Tz1!TAivpbk4Sxx=!qz#;FZ?RK@Tk
zvi9&fSM|k(y^^!B{lu@<;Fv^Az{#$75Tsy2MaO-Deq6)u^c~(){%rtb-YUE+@v<{E
zzH-aQe(eS<;I+L=aCsHhro`8}Wv&pLZs^sJ)sLm_m{8qx^nMN2L99F)p_N2+{;$T~
z)I1W)pV#s)63L?q>?&3D}l=ZdLLkE4;F3eP7wX2H9*q9nIjet|nugo-K-?hH8
z`}oyhr*08vk9S*P_tt`hIEOMU<5Fi6uVEQBF6zKJ#g~Esp}NMc7b>Z>XVnW+;QEK9
zG4kspp$ux=2Um6jMH*GyfAi2C$tzf7ty$gIC*9bJX|?L^83G$^Sk54h)KJy%29|Ym;
zq2XiBDtZL=ms@RSUK9T^sThBG-wC26xjS{9-5x(tJapPCDT*-Ctx}0n(7}&2uzpn;
z*B~lWsx!S+PvE^fsGFGF?TI71!>m=;-pBVuK3cQU&46h2Ow=(}b71
zpy2i;SDX(U2%HFuAbg&SpOh%vf&VvpwY@&okVuX{2kiQ-@$_wpABI1L5IVaG3&r9>
z3A4TjYB$DS;M1{8M@$uC2>hwwXwP&U?&EcAk?CS7@z;QjUDEP3{w1dS=>?G#;uyzV
z@8vg<_^@xV!jl*G2{xuPZl4Wb61=x-^Kw`|;-d2feNiEv;sUj0Z~Ud5A&hhLQKG5wiKZ1Q{CKDQwnm-q6Ne5B_~Tt48?DF3sD*iU^X
z*iQ5l(HHJte&&;lo1Fc=d*g=~oP+r-=Vtj*;`h*@*TlYY+~bYjp;eE*;qgx4NDrz+L;6&ENi`*I2L^5ys
z!Q-dfao3_K8t1J#{M|}EK@TPcT3dc!!JNZ){H*cH>Id)Vh?UW0>X{+GMl{O{!
zdUBzTmvZ)Lx`mPqA8p&9Ez5F-2xaz8@Z7DIBQ*QKpTi1X`zYR4)wILX+7!EY%@&=#
zsXfifLnr6i#U{7*<@r{{n4A
zs_Ua$PcKrY4^5ohXAnmDOErz%$%0er(hO^rjFo6Ta$y$f6Bd-|TK}?$}N)RF^i+
zdt19O%91vh8$*>{(4g4*C~xI&c}O{9&B76H6F^~SH_f}T^nh|OJmN3EYZ&e2vykOM
ztz1gpqJf6NAAYydP73s_a&*aePEs
zgFXCj?s_RLhUr58S`#0NduI_d%e8B?mt%jPiG2&B3?@a2E*>nU*~w-f?oNA3Yv;+@
z8$I!mcJNBqu4^ihl+-gumrM1MX-?W!@foc3lpu+1qba>K$|B#nTb~0mMASf;Vtozw#|*JS*e{$;i!HmkRDu2OAZaBJows2*<)yH
zZ_dYDF(XCEm(iJ^rJSSO(rhlGO?Vvi_w380g$zh(J7opXu6*9@JkRMz+iL8h*&XFi
zTVKn(!D1qbHZ67%|8g*rVsFSE&CL-_q2jT-_S=V149@TIiMi`UIq_xos(xNIMfr70
zlG`C~3dMZfy`LIK^DVo3Bkc4;TFj$YC2i7{w4-M3(+Q6zY2|XyO$3wMX-=!=^Ax92
zDGdirOkc}{Q&eB>*>Qd{f~M*{IdMRFnr35lK#ZMhjMBE_#D6UG08o@h{-x_^qUw>d(wv|Ptz_UbIn&EBD`=s!)kKD)EC
zRM(A?VkJ;{_V{HQ#ZKe<2VW`5(#AhSn@n^m`6Yfk*NdN{-Q>G|13P<(7Ch$WIT__n
zF`kjSYMiJ>8(kLO*VCy=o9Lpd4~1Q&7*#&KVG;D0=JClpKIvl=Ex{oz&qKn6()s3X
z%9`QJv=KsvRaq~Tro8>BUc;Fe6sL@g0PiD3w2N(RC!7BZrwDKPp~CSeh9cG0WRxft
zPfIy-Y?EAy3dM0nt)6#et^_S#bV$RXbLJCXdWz{J=3-n4h&+MXM
zhl$`#jk_puTeMa>e|bS`H4)W&E2gMsIWDll(;
z`j8u*6s)l_@~t`XVehKp?}9A2=44=T|BMw*i=EFilu0CxSc>0kI_yjQWcN~O-{piq
zi$9X~+S7`-IMMWHMtLg{<39PG^K}W{!aDnLcrXO7OkLQ@bJK@Nz1tqS&*B=+8!j$T
z-(f_ISjCon?;a!ABczDWhfN6?%i>E`?T>LQjIv@APc&gZ{@|+Q)d;-E@cLOjmOMN!
zad{nAWG`{8k#%$Vt4iF&oQ=bHcP>FG5+W8(ClF2FSN@D;Ig2NcZ|^BBR3P|nhS!U(
zV#cHX+fWrZx`f&7u;>&T5GHu@qu&Xi)WpxI6KQMCsNyvRB^5u)#R->bFP1GIx$(g3
zyH@s;t;0h&1m#b1T_WauVmTFZ^@&}bEMHFK9KyrR{a%&(?j($4R+X?F|3+PHewBOw
zn`qqFwZDP2`#iqjCA488*O1s)m^+^do^~9
z=AROApBrD6kKUhqdM!?2xfbYic#L{+hv+RYp0`+o?mNAaj3%tpc`N(%E_K4f+p}!9
z?^bvVd07(IqqX?m=flU!ivLh=e_DGc&fyE@c<`Op6Q-?r+sN(FD?(a?Ui!11hv{Ot
zBF}?NtreWaqpijpR`hKou1Xa=5zT??DD#WKWASm=O11Wf$^#u(==S&Bw*y+SbgR{?
zv_Jut!Ls|XK3g|7YBs{p>r{yu?sZgkm8-;dFtt!0(;6_n@oQ7>v$L?2#kt@!FB7TL
z6_$US^j~5jCLukGu6L+PX8f<3{o|-MEK7UmJ8h_zS46})+J9kTue+AMnfGEd3c8mI
zH|0@9cgjxoIV5Ae&)bcXChlMcMbB+NuxCbeFh!RE63GPoO~
zvaNr*m!lPP9dF&oK2k&#V2#h6KiY~lR*KX$W!GbmEGHeW8s%Yk#5d=AUT?t$HQPj5slimWg3gy(F
z&eyNs+E1efecQC=<&iAvyH>)rM6U(=e56;(V_hufct`jCHq$1mM{jSe#5OxB-g>$(
z!#Ioje;B&%e<0r|nmtp9kWFNd?7Yv5WY28L9%XNmvXfL+S!G3KB_k5v=QXp}*C?a1
zL$=JQum9lr;eMWb&b{ZJb2Jr1f3J3=;b&-H*GR#8#fIi1VWO<+hHDEc=SmA&6z3~~
zDHG4NH}HmgZN+iYVsiZ?B?D_?ii(tv3dk=HH;Da_Ar9UDp*XENN;WA^fQ4Jcb2sEN
zA_lyF(j30DG-!QdIA!QIstC@>CbmlrHQ4o8=9+)qSFAB}i4^9bM|cS@ZhTW~Q*>#w
z6i=M|uYt+Ow0nqSO7XV09pC$}e;Q8oH=;|F!r1L^l}WenttsA7W@zXp{?Kr);^t_`
zK&|4wFBPb-RqY#)vP{4+-4e-yur;bg_SZK8{>c(WpvTy-=`D}0zaH^V)LL{;QJ
z$>%SJ)6J0}`r@k{7P82N*X}t~(-*KexSWDp0x`1N@WM-<+W;}qG<-e1>5Y^((h$Br
z!-@6UuIgd7v{>T5{EQnkLda!v-b?c97MSOc!P#3n;nj@Tx0
zEz*2c!uXaW_$!oLkRVAkm!KsAVLyB0tW%#Aa#VR6|3`fTyLDWl;bi87^c2;u4Xj6E
za)vhhQ>XNhiS*g#JIrR-wtH3g(6R$2Ce?K)OA?Gws=i-hoXE#c4<{MS^z~sC#7>(P
zrf-nPwsbcdZBCF(vg0if25Th|&rfTmf93{H2Jm=Db_JTZ!@)H
z+MYw@@0bRWJwt|67NQ!2!+VgskShY~;8Jl{lgY()s^4uTT^~aBwSF=`c}fn5?I$h0
z9hNbv$u%k|&IN?%Mor_h)OF0K?k3CJ19EUF-7(R%n+HmLP~)JvvyVjFagUzXU&R;{
z{X#AWtRaGO*?wCePv)2x9c5R}3&VxUiSU>1TbNgH(W_TSqezSaqtSEbENoRJpy7MR
z8JIP&7GoQ68cN@Nw-=heip<<${dY~E8o81C%1kYE85=P;H$+*$0WzKK9vN=Y0!idHWxi2$=xY9`XV$v{%%r1+kz!z&kJ0*3h2fKS|YB9c~pm_M!@=^N|<`d>9Crc%IT?hD)_>^uC7R!3#mK|135ct
zL5D@?mCAdaF!yJp?`!USV4Jlq_h=>^#wwqkHoo);WN7_a3sc76Tej`1@#qBP-erRm
z-;$vbix
z`&+g9b=8G1*Z(JHVD*iw;Ta||pjWVwg>~Nd`wSldL=HM!UI6=oqaEQTo`TK2&^dH
zp~DpvgbFJE<3c6fgl{}Ak;gMzQ|<;C8lb$s*QZuQ4A3;*fz{!$eX#33wC=sN1wLK2
z+$tz<2P<)=b`p2zK+S5zjWCNJKnBqofW~=Ke_==&n;z={A~C}ICWHrKMOw9xHZ~8P!0Y}Qu>L7
zHNy4_UV6-gBksfwY`eFk`lOws>Y9e07%i
z#i-K*m+nA`R_<27UvjzRYWgI|c@i1fL(GDb89yv9?;}Nr7gIJ9dbS{mUTmPkwLbXI
zS8%4LX9dj6F0P&XT>z(MB1%$Jli{=L&jO9Q?ctiTOu$)PB~aiVynFpn3Dy>)ge)IE
zfI2=GdmrO8poyGSs{U;X5k+nn{%Uzp>X-TDtg#YYE3mlZ%F_gGJ+eAu&9mXxgLOdr
z!3P9*kW$}iF@!b6H7d>5OabTCH?!$mU%^0SPdWci4w#f^LG2jQVNpVI!jsJaa9e-8
z<3>R+%%kuREKXShaThg~Ro+a4bqUudyIYw+k-+AXe{&i?^A|HP=V^ejpw6#G7!u7kS<#K5u5e4Va|Iu
zjAX46sx?*ZVvB=hlc0E_?1Yy)-A
zLrX~&HnkBN@a<*QpM1#`?1AHjD+w+b)@-tfh}Y&KTVHE=nhRa9;M>oE{DBd6sIGVB
z^||}VhP1^|pGq<2ZKI6YIpkuJHz&};ps=8;HGNz2qmnhwU19rs)R)h|r}(w9n@$$h>69{V0}F0g-2F+}Vq
zUMH<1>R5TqbEbxlN@Tq}ILz%^4&sSpV?J0oHb%f#$I$DIhzuP$%3EAQMsw=UZff&j
z`ww1o?GPL*=7C-td?2r3>!zK>TgLo`#tvl<#{Nmgze5Wi$03B6y<@n6mLCD8-$M{`
ziDFhU@{`CG=FT^p
z=W4c?ts)hfGMz4V>>f{)N+gUlE`PcfP{xX7WgJzrG!kKjQ+KWzQsd<4|rHGM~37-CgXXE41ehD|9&~q710c-FZ-S7qUP+=
zpI%>7>@-fFA$WBjVN^7_c}wL^gAhNHa+Od?gS&fLj6?2M#T>`;7e|t&8$Rm2W~%?m
zj@fmWjIoX7DP9S&3_+64#61seR8+Wj
z=9D!n0g^B)IIZJGiIL{&^d5%uBd@Ecd^;@dkmQGz=FGwTSfz7&xB!t5^6|Vn)2R(<
ztS~WdNpIW~x$#bR@s@)vRv#x|_acn~F^RrB!y0q}yUDfie%U|(;S(>sQ7fW@ZJzO`
z3t|YswA{M?T8on*l(#aacZ$vIF-W@+UrUs?C+mqa&<_&R9O`={v(m`&EL0(XQiimb65gG
zR?{<`U7w8I5)Y;&_sT)?*lEaq+B9MmpGETR!rviwYALI^%;i`){~`2REJ0YUE?CI+
zf50Gn-W$17eTavb!4H$n3heu%g9;Jx9?UQAAwh8P0FtzzmAZI$0*jre)B1ez2w{oS
ze!oji1lHf2x+nbKHYRX&N>}fA1@X5c!D%~wW9JOezRPFn0J`UEpRod9hxH5Fvz|kU
z-Na_kAeUE4J=lwH~wK%rsbKw;7H&
zNMs2Uh!adtnxh9tD{bp>4`e{{qi1JwW9Z=ObE7FP>wP5WVf>@5cXQbJovBw{9rV!m
z#oI7RGDq0`+IwI$@+>eyqBUY|dElR5nkerc2_UFHllWrnI+V>&h>+Fvfbn;=nBWg>
zxXgZUc#_}>2-*!2tSgX)dN+rMcGhk{sf#whB3A7|gG+Vbo5zRPnCilm->MiWASWKp
zB~*bFRfN7o&aQy}-TRIB9bQPpC-#foQwzApL`M4F(uJvlm#Cjtn1ixsg?x6dK~OiOVyN-F$H$iT(^@foPJl+3zh
zk_R1^7AVi%$H9tzIlYJMBm_h?tedwUgAs+{Ihy~T!E4HMdVPxdVE#}wt^=Yvm6
zpipN1P%@Jke>i(Ea!dCwBs;#qSTDZ?$4hzjL#(I3h`|NA`pi@K%6LC#3`s$kPsV>jl9f=7YbqT0HS1c2BziIQ+yvXogA5b2QT<(<
z-yLPVz`wieg4sDVTQaBPXVV1u);)T&MTZ41WxSfF`&9SjMfGuI&b+8o>7oDfGa@wA
zPxf=zX(9aDeTlfdi^}L4w~a6}szbRLC}!
z(=X!**`uU;Z@$3WGAb4#GQ?=JlcR7rm-b0s(_JdVb63$9MnZceS>mJE~tX_dwZ56C6ov)%}|Io$Ep8(HHZ!4!SxB0)jm}FAsJ(|eHMScbN_j(_FsUM+pfKhnS!~A
zSqaDI4&Ykl@%``Zbf{Ct^rg#2RH!KD*y$%>WO!|zjONF{Bajx_c7a!i7=8KmDan^v
zSybzp{T*hO3#d)K@)l<>5vtilN;$T{jMI!qjQl1&ho`25%D#_q$ItsRzh%C{gfl5=
zPO&{#MrW^mzQ2B14F5G!_(I{7Fzzm>nPGj{22yw_YJb{%0bD!j3=&sQ_E=mEY3}YN
zu*mEg9KMwUT6A)%3zqhvdgJH*Q#xNEjnGMknQjK4mhZYg@h%x;8*4R8oL7hM7d-Wo
z3J6fI>&>g-Y(pUGb)go$KrIZ1#J}n8r@%he^p8kU7(CnAxa<3d5_edpZB1XB2992h
zye~b!0r`py7n|-_Kworhqu2iyD*mZn;G+`8tE-E^v@jKWuAlhVaag!QFpDGJqFZ(k3o7
z1nBw}TU}Z(EVlknl=aeea9`sQLFBO#P}m#_Ew3_yS82>In?nouuA9s8U2`mmuCwOi
zb-xYMPHYKETsM)-sk_4(-|Ycgh;B&%S`X7SN6n~xGa=imqWzv(65ycGotc8(yp@->%7;T0&)K)Ro*O#sMd$Jq0VrNAv2>Z(xRJourd
z)rQn802rgk@6)4^~8*#BgQE|KCUxOe`~E!rFkxS!C}Px|Hn+f*1In2nr7
z-hC30dvDN%al6iWlG#>a29>gtLCs7c)XbvXa&#Rzj!ep<@0`T2kA(-fvIdc6U7=eA
zt~D5eIKwb$A0I4@l+M_TSjHm$lJGVmbBMB?OKfSzB!(mrHQ6i=Ad0e%!p}x7gM)_1
zrzP2^p;_ozE4jaYC;P4CgQKDuWbAU7Psa2*W-fzjH__XJL=%1Pos|oq|F7-0SW7n8
zC3W<&?=b`L3G0j?T0ITLpGo=GNxw%F6K%|;1K%OO*GM~ky5lenUb@lN-vP*ry5ox^
z$F4|nkqpT^OBXh{P~!UA>@B7xKJ&=_{!=6***Q_bGyXttF$A9=Jpev)wg
z5^?9q?LIyD0u!CM_91^h7h!7_&ENGn2WvR(X?IO#u=H0Z6@k)&2#>z{18T?5n2p;H
znJk_ygu1wwpTdE$TGcB{k{+W3iha^ZNj42P(*uHaPh1d+tyU2;L?FcVlVhBb`
zEqo(tt4)!FJ@+~C`wG}cS=IGRM54&KrO&nsu&JT_@3?uNWZVhg{Z+8OVudxRKJE)3
zx`wQLNo~8cERX3I#e(iJHbhL=ZS@e#L-_V=7e0sbAPw|}=x&fcMq(v#AuvD-36uQR
zI5?|$!XuS?n2(#0CGR^@XQ`ecLGqlpGt%#4hQncf<2ruGplRzL6B%=ik+xNiahnjU
zVZK6H+{A<&N7LFrb>zm}cyDYrts_V@JtEGj6oh>Ep&0txe?#%9rgpidDg~CxoTh<{
z(jlkk#wmqf@nio2cw|!ieXyPa5YPHxR`E`qyZej!6TXmBqSC!ej43cEG>%7*VqFPL
z+s3R>hym$K<3#bZh+giq632E)gp^fS_9LoA-S@c>$Ud*T&lN_&K!|U3(;)M?e+OvOTPk;DMq%+0^NI9`UPN4
z9vO?gZ1f0zTUCJVEZS;Idkta@m!{U5ZmD6#1{Y5`y4*%KYTWr+G6J#7fs6ess?U)2
zSL>c`@ga#0SS4-NIg^uB*v
zDAEc+6n}YoA1^{oDd>j#gJ3!k`FdU7K&2JC@+3e`lkf`?5fRg4!#aoU4!OUd_aXr)
zK-}YT4=1<{KPg`Dnn2twGv9Uyf5(c|^*A+Ke|{ycRyym@9c9
z>)q%{?9t{c?_FIs;B!rRQ;*;nVT*QC+0JAFcF!sXbJFQyr>gg{P{IYkA=S^-kRb%0
z2!3SL|6~Eh>TZw_{wEC1S8J%YSxdvFC+|I0Nmanjk^Zfqa$9(XnaQ(<&I#UfqI3egbX{#_>XFd5}8#ZJ%OR1zu9;
zk$x^~00Iw1ZG2@S}VnyBb8pL%zy34uU)q>D1R>
zoWMP82PN8}P_Pm3!i8lZ0U8{+tBo^OgNb;B6d%fVz!V-_GoG6X2$g(}&Aw;AontnC
z_q+KZRop*)W%?6*{+iB};(G~HFlF20wsM5xz3i>T50k-qg8*p|bsqF9d+1jGz7iC%
z?TX1LECM@@+wM%pHK01akJvCH7xoEhezl#90(8H+v!#Wz(`m7jf8Cy
zXfg^J>a&bOki{_jU+FjS;8I1=&2#fGCZGl14JE*Xf=R{dw4mi@9kc4KouP?b?6RpqOLdWt2@}wAuM}43chAw_
zZ>`-P(UtPzNmp;nY>%qrvqB8S2K1zqG``
z-%E(g>t`BH-10+lYT(SP8HP%VY3)7sv_y#tb8nd__@hIs`+n|Sw)mYT;-jKqN&J(A
z?da`dAAI455YgCQe|$*hFvn?B4?kViVVc;ljGDaDxECOO5vPq$r98kmP&8qQSBy>;
z{Yv>;mSprgKD)O1Hu|$P&Tzi8+Sy12by{6$tkI-KgE>rp7I&tjcQjg&8+HDuK#~Tl
zZI~S{qNsg`v6UV*-)OISKrD}nN~P$xaeY8jz^d|Dr+jqn=UYjSK^t7RfTryWvl{+7
z-P2&y-5!nS-t6$#Z9vD&zeZ{=zDJFy#-zFND=5JrE%^?Y7S1sG2`2ybL(yKULu(Ey
zl#}T=5WCNatK=*xSuGsE!#78?OJaM#$@S7OBGpigssB8?}X;}*}H&JPk7iP
z)p1z5IJx{WXA6GsX1ysd_!GF^Z6PClhN8rzq+TfBdHipa+V|&=2LM*hvyzBVpu_!5
zM$0r*_=T{M+G~4`=vmdio3Vr2@B&$+$)yDaoU_2Z{05s4`XY~>eV~jIoqy4KDM>u&
zgom(QKYDJ1+IA^9b(ZSm9_w?1M_e2z`MZ3kx4DXF+GET47UNmS6s3NA{mBSiBG!BI
zva1$&q{i@L+0P-{muEs7{|ce*fR4P))+T7mx)Ha?Gy)Q`>rC6bo8iwzzP}66;RnLaHQ2hf{bP#d{CTLJLMQ@mm99;8011r#Gz+o<7v)-g}k_ay?I9L!SHv
zTBvk^wI~(-W-*NYkMuAIcL`fYWV=C@Ziq$+?+i?QG=7^a7o6m1`FhM$QQ{k3vmYHg
z{=lp?LPSpH5EzCMD~=FvLFE1yF@CLlxW>C?(^?t}K9z6&<e4Vu80Ugk4n(K;Z@aC9@n
zT?tyXY<-oTD!VQ~p;@Pt
zHK`FSy-v%G^8J6v#r2DWLG9g`HFfI1ht76n4t-MoZiEw1>WYuncJ*MBXF_zASo*Qu
zJVEE%<2{J`n!S3X)F2kryGJz0$_EV?v)c9wu}Ky&$VTq)k+$rg=(vaJPybp#?pGK5w-h7{>qV-Y
zvg|KIy_4nK*aRi`)~1)TV77%BnY#aRZ}^R6`V$^;i*_S33pKJK6p6@Br(uz%%?Rvt
zjuaoi^c^JXV&O0Tzn<7RKaQ78zJr)M)6F5CLRTy@KeaS1I~)7
zyo1P%NY6WHjw6w=w+gZZA`9+rG0-p>>*D-6SU{ArW!_77m)t1g=l(aN8et&X5
z6ly8?`CcOO0dKLp%Tc%EAr1gwg3^)?{9CCHN<@H_m1~f+(Y6FtMmi~
z3b2DNgII?RfY4}*<>>gzV;Ouo1w1Pn2*b{=f{2HK*ny&e4We3wtTew1zb|cq_))J}
zB~vLOZyKvUFWk9-$q28kJt;{-_Hy-30?{k5LJs#E(tS3Fvf}TiSkQI)l_2fY~V>}@5?
zUc@FIrrdj6m5B9eEtk%BdSJh+=gG4((~)nU$|b5(!&t&c`eK(X2GC!>OzX{1iA2Y?
z#Yc`5VO}YuhDq<=BVq69oajbkk#ix+HUEwdu*yEyyVc_mQTnX@;rL+#W)kz`W>i}@
zazcV0w5jZ3>NGl+5HTU3lYff+!Pib~Ese;ZtaToFpgU>rExwJNs$ccXmY@S?v&~B1
zJF0@+o!_66^S@%zy2fPhkv&8sKRx@uMG}}7@NfI6upk&@ntrhqp$ta9V$UzvG6O{i
zcY2o
z0>r!pW7pQi;4jNq-QMrMppfr174@+Q*p^70GaTlGKg3$UpQ~jDZFI>zPrDVsb{%)#
zX3+j-vAu8
z91*7^Tfo`h>=`Xv&fsmo=b2>9I4F?z>Zf>HBq+*noP4Tb3?3-#NFMQVfpA5~;n&?;
zScF_ck6>dYEG!y!YRvo(@M~@5o?i+C^@92#X4Bqq!TfMA*fJ1&4J9)g`057rJO|&M
zn3tXWr(zq~p$`C`s2W}0#c+6F{B$@iEFSnxKhoKvt%3&<$oWKq5wK&G`Qmb9IN-;4
zzl;wgg9e9+O~=3j*mEHAetMu8Xtqg@Fdl9Km1$B5*2gK}lHMgs6PsLc!1Cw*!LtwW
zpfzfUxM~tirMw;}dCRRR!Yb(Yf2kWn;Hx@VCM$0{w!j
z_?$6u|1Cy3boC<@BCJAyZyn6arYO^)0nfD#opEiv*g<>$E{Qd&CPboSrQO(=?qG@znCkdH-EjEt&B?K
zipXq;TjBMaz5G!sL3o@>VMMt6EUY)D{Fa|9il)DQc{R^M24~>B!EJAN16TV=yCHeu
zG3sbq`X8a8CVqKlP`Pu#0&QuC5eRN}#+zcURH&8QMQv|BS&upK%itduENM$tM`eFl
ztlyUmL9K7BGZMe`!+-ZDuIPuU;1=6FSr_TN(f{ri_0*K8;^PPEdNQqUIQ5w$GLh{F
zyyKfX?9<)PIs
zyn^{b03Rd!5Uz8z9_N+&C$=NefCsXlcinl=h%ZNc9L3Jpqsm6jjWh};ev5mTLiR=Y
z@Zi%|%8O}i@#Rx>!s1_j(14ecV=JK$b)WAz=<2RSwYlz+{C)WVO);RdP`?#|zYyKh
z<9h3gZ&UE&GEYwY3TG>DN9-edTEb9R#jy~5OeJ{nih3k^(^)k8D8=?)n-a*goAj)WOnEs;U
z6B;RTK5LNS#7E^vM7hU_f28ARsgcy9K$NbP_56~^&68Yt6^5#C9o%~JMIeVXH%jI<
zd(FgO26eEPs=XxYjU!6RQG+d(==Iw+nTm%hXcPUX`>p}1_*A0*Z$3wElzR7V3jFJi
z`xSM9wO_ut<;}B?GHJEZ(ac9Rx7g%Pa-eq`+Do-?MUy=nCV^bkt+IdPKT|vO;&5}1
zQt@rP^5t@Nwx~5a)-GY`QZ0CrvpDvljJF)Me<=Qk{KUV~YrvK8fF}_5Bj~8`v(iA7
z9@RhB{$Y#$M-|pk!zhFHM^5y`?9t$REtVhaZB|e2EUNZDo%|OuHzz&xY&T(!$4K^P
zA8GVsqgL0>2X3_UjJbMVJQ05X4sC6&+BBdoEz97;7J%R87vX#N_3%vV39_9*MtoUF
zU`9)s5$#&3|B%f>f?s^Z`Ou%b3EoF}220K9@k3f#nFc`>-2BtPvj9FK^!t?};`h|#
z=<@&u>&K6r29zp4@dA_Zm*u?07AH!haS=%Pt6@H$ILy(!D#2-(`5M
zDA4;Ge)Stp+SJ71@Q}zAlCBTn`Y@YyX+SZ&`pd;?%_JSX6+Q>LjAj56Gl9%y>lR3p
z8ZT!y`w}KJiyN-i!$0EF2#=Y4pDVxmBeZrI0}={mL`8N9Y~!3Z^)H
zV~z#CQ68Coph1d@8;;Kp6)gdS`CMJ=F9Q(0Mj0A@wI3248?|UnyaK^*Ij&46gahBQ
zq%C_HXGpr8bF_Kh90a9gu*!udZC4_=u_(Z2ISzZ#*y`fX-f@ndOeejcVUjnMRIw5wD4s?IM-+R7;5ONhg
z(a7RU#|HkKqqzSB0B`r6yPQYDV16`KzHyrpF4cWwJwLjQ=sesZUcT2QV
z80Z{2zc9c06A5z8FYexyfS%@CKk~a00mT`Qr6y5h7@80;s(p;YnxXZh%ks*gN;S&?
z5xoW#%j2G;#BU;B+%te=*bk&j>a=vW7lvJzvdB{VuN3KKFpA~AmW)KHCl%Th4+Pj*UkuOt${`1uj1Z(1@EEn$wnVIRa^TDMG_
zB+Vgb!UdS3t<|6&Nv7hY(-M|*cSOw9y+=
zaE8g_Z0tFO-@OjsB1GY?&_mg$ofw_X9nXmTd93Zse3k--0pfI_xogetA+~HN_d9|4
z8B#phCwe}=8jFc%9c&z`K%TmBO&~UCgV*3fwQ9`a!*ql#~}P_%mf9
z1FSK8PleMFmp8`43DgCcM8+on$ax$Q=wN%xDnbjwSH=?piEFXMWvM@!EL(_)^gK;-
z3IUXex4-3JauL*+`ZNu(CmfDJGHUBUG#GxzoUPqNS_>MMZjbW-g%=HsQhI-p
zPOZYSAf;VQWvln_XVVc9b9>0?`qD9WhId?rFrLK
zk$7$XA)pEC2RA?Hf|kDph5cPx@aS3D1JghgAohCiVfsb@d}@-t(AJv^_NT~a6g<^|
zO|nxJW2*|(3+|P8yr&Q1MyY$Vv%KNnUes?+)mOk|>W_u&gf+PPdnnUVR0m4r(0rga=jIIcf8vi9G31>9uZ_(Evf5}4=Qc_lmc3%+S)P{#EB
z0Edgh<)vzzxKG27Nb@gZG-lnHjdXt=SS*qGp4#YwOK!1ab0%Lv#4A=8UGh_S)1^Fv
zsuqCKT~Sy39Ylwo%198%`$L9L_tf@W**}ZoFYa009gx6-^9HP+`Rk&FMZL?556+`8
zwNl)HA?q;r^xwZAWEN0nfvCXSl6Y&An7lrd6$&(!4xSSz-~=|tkJ3^MP|aPhePqA_
zH?s_5-!=9{^@DX}fB3vaDNA*y`{*v=XWDyK?&j;DF=r%e#1d?A+XzbH6z5>n#$-H~
z!T!++eh5Ac)l)(Zoa53*(y!u1j~5@ZAu2eoYzKl)~K2PxPVu^B|7o#fJiSy#dr_oLC#a!|gGhOu(`
zF-4f6yXBF=OUvcxcrdoKWs`x*sgSIls&~gXw%i6U+exD>9Q!hYV&>=vHqBP~mJjHa
zQxVA?RRyT|Wvx3iX_rwUtbR^4+7>62_qj^JqK$gTPwep3)}o#_^}dIWm!S`R{CSp*
z!qK8aAbG8d&k#C%LMx(w>-h0ip|2EeqG>B*QF^a;Cc#YhxP>Ff
zi*I_K=)HQj$Jb*l(C^0EhEIOV;A<>ioP?tMIE$P3{jUx@D4+DR?(LsvaP=jJrEw}t
zR7{`NRv97TmAjHl3N0^DzN617q3k6%I^c_h%Vdu1bHz
zJ+uc%_jl5;sK-L-or2TjEbZ_W^EK<2w~jz~n@yKd`!rN#{_Bz-kpvQ~t~l2$y$8Q!
zOIo@+S|Fp0-cw=q3+S^DiZ!>eZFu#7!ItyG5G44TnZ;#00U|D?DM$-_g>(N(6yAPO
zK(QM^J&HCv@XWH+jk2v{NO8ZHJpJb`pr*<$Tqj?E#K-evQ9=!n>Ip4fx?44{+m)6a
zQ@IQIZdfwsv3tX{d@fDnPfq|v;p$3FSsVP1Ymbp?KMU*zPOHgjyMh0bya*%kAhVvP%2FE;3ocV9|nksUnK~>-hfR`BiZo}1|W4vK0#bt
z4Ip_CfoHxbfQy`p{uCYl@cZ87k8Z{Bpe6dSC*qt5%v@M6T=J5IPFH6b0`{+fy7zy-
z1!Sl|!TwznQsF3I?qdAu>tkl9E2t;9(;^M8oRhm`b>=eAUfkqge|s55J%3CK+#>+t
z7W$>xP7Xwv63I4mDnKjOgbJ3Qm%(WvhRAkqQTT1+L32*m6Yy;>&@x%n`{cftb%v|Z
z2tG2qTmL)YDhMWPQ}ZSLjx3#}JXLud4oYlwc4I1{fZH2yAKF1nD8Kr*I)(8HAXO$8
zI2_XgUp-7kt>mu4A=yTEjmMpsYz(y=f5;A!a5T1Y{Cg7bKqUih
z20Iwd6jw0rKt-kCoMnUoc-xuWoWoch*2SJxenV(ICR9ExT!z22Q$L^M`-?4XY>c=m
zt|J%D2}Oyr&STRpYWzZGgNWph9m<|fF*q}FZ+`ecPPi-WBHQum7_nEf+CNfX!YDi|
zJ_Y`m!D#-B^HDe=;IEksHhNVEO8Fd)o1Ng%>38l<{1+z>s>%-Y$@Ei@wNU8xm&PE3
z^lhG^Va+4Ny>G+sF8ToT&sQ*KX{^HNv+2rd*6F|w6@#uL=?jESrL$_K^d;74=XKuX
z;|s*(GtUEoJGF>Xj+7t=;nInI{g-`5Q}v17H6!+=)CY`vJ6f^PvIVhahe|PIo0yT5
z+w;+ui$EkHqK{9l8GRI9Gc?y>X;#Z4
zf&yFcNjqXb2Z@5(TQ@C}q(gx->cadd{wDD2O1}NhT@>5`ZqE8`+R#B!V$QsY3#ce<
z5iA?`gZ=K}0Z8j}Qbnmp8e>k7OV&A5E7}#9E5$CD$GQWL?>4nBkCNau9smD+?S=x4
z;ZgHxhXD9mq5MpwM-;G%xWgFdRR#wh5B;!v+7I4**RW}>^aB(if#h#sB3NG~Y@f}^
zgX1%Ktr-T5p!uW7i>X&DU~nPjZtUU%u$oIkdL!TkxJc*9__DJciWNj%x_9LZ5LIie
z35xj&etkbQ@!uW+-`d-0dYS)&&$GE5lU2Wh=a}X8Z{G!oT1B#Xo+rUyOdh{s9I60=
zx{BuZ+m+Dv&-PswqGqu2QNc5KWCt#GMP^NE?}8>by@TMPkDzP2XCc+57O-zHX8xu9
z3g3n+IV>NSP7)cfJWk=K`Kjy)Z=Xd+n)`W_G$in*;>Nckv-0>c_f7k&1>|V<
zZ7S`ztkXcTMUF1MwHqLsV;;XJ0q!^bGSx!N9{uM!Z^&Q7f}i(cQwVS7LYL_ukEc$E
z;c-O*g5(#BP>&8%@=nbw`0cC;&YUuKR6zYm>u-`cjtE7OvwJF|MPr-h
zclXeL=gm;!Yaf0sFF4_~Mn0Vcx9*`MOO*k^=W@`r?SBh<-?Y$(+Nj5uP6eTV*=vRG
zO^4uj(`ZeZmSs?3v#!B-gBm=Pu-n9l*#@7lPBs`33&txhF}P_-IN(c797a!`TA=1(
z;7O!pBi`QS5cb{oI(lFCMS;LGOO(^bUv1bd6}@<2Sh0b*8vkgY`Aq6nBkrB5&OOxo
z1l?M%x$-Y84()w26*E$vk2}AWD?1S;uN^b}!ATsBE6ZW#Nd
zlXw>5h3N{1$kST<8kI$TgXw{I>CCqCi4jsyQtuNqY4P#5&*XQJ(#g9f2VK{(CQ^Lb}95Rd+*v96d4
zQRe;hk24eXD1qt=ezo@Z=xM$BwCS+~bfhu5jg?piAJHFEkxs(Vb;UAOvM&(r7HQT<
zYO6)Z5-$&VzdOk_37$D?R+W#YAQ}U)XK=LV15briDnyk*Q6uR~G#O!q^
z1P`SbbL$n=Kpj+7h^ZCz(24YlsKMK6`0fYO%8y?uQKSF#CK?O*@f)9AR{@PL{*_FT
zf;(9b4LNK2xcQ?J`iuEr9&flN{-7-XDAT6^tx6L1#>@lJD(8#i`*-_1s=3PcK^Hs#$A2lVsB=fDQVB7&%e#t-<7E6p`I>}=C7ijUFVeVZ=
zCP%bLAWF>0iWEN+v}jB?A&L*&K0h;;#*4NziPu*Zli}s1lLrpZS5Fk4;{fN9cK|c&KxZhkGe>@Fx`S0iZd6&Z5nI6;Z_miRVClJpWQVycTreb=n
z=<&1>4tkE2VYsUL*=beeE0Fjq;4)j*3#m$6d3+MfffnreB6fxWk6Xh&e99-pdvn+%
zej3aH>kS`|;*KszI83bKJJ10u%L&iQ$|r&*QtNh2?z?>f4y^t#sPeaqkcT|=s|2zMPt4PEyzQJRBLg01QT+QAx52u02zH?Zt)4f)*!at
zo~la)s24qHq)EF1dHBC;F|(Qh8d6sL(poU|>3z7t&U_ElWG1}LYET6AadwI6RVZ9~
z!)eq{b`^X(Y`(YHatBU7Y*fD)nG8&S7Ti|QRRSs#0zG3i;_!7@Sar3c9C&t!y9K)#
z!;SxBw5!IR0*%xM)}8|jAj6nF%kZxh3`%n~j=O;X=S`yVaC>veIdPZ%Nz+3R@{jOS
zqrEh!t6J=|mXw3MrF<^B_3D7XCR~ZY)B!4}`-=4S#R1`7>B_v(B+wb;2p*0mfEaE*
z=WyBAu+phS)!Fzn*pi~3C%ZiY^nS4u(Tl$XUO)3JA6|}v`Ya)>)&7>Cj%)ZCs$T(W
zxlwm
z9DER#HsrE-2=D*QwrA%^0=F*A-g@}`DXffdxjM|+3|4N+t}1Cg0qK_+2{X_7!{SFS
zdB)qKfNX!XKgV$Gh2P8rk#fZWw}c&d^(oEA3yWmvnVglMfz154
zM$ss}>J&dN>mxq&u4@t;u*f-3=KqAz$-9DLd&W?X
z`!UdKWKHz*41@{=G&jHYwu9g5L)7R0ZNq0D+-aWVi=n|WM5l>Wz5-$~gU7#q55hb4
z$y`I?Goab3z~s)^BPjQR$E;?V2_1eAnwm&E3wn<4Pz#WMhba?_{@uxc0aI;b-|8D$
zT(Qr!RAiX}6>$wn%#@VGvq&;R=go9buCjXy_IajwkYClnR+SfOv7)6qp7aFu%HB
f8BlDZgvA0Wlr^&H+o5|+oQcEVBk|43n*}eP)1ic<-?YjN=Rj3bCMRz=C;o8lzbGjy
zW7LL}f|iRU7*F=w=-{PhLi6GmG`*s^(6Vr;7u>;;c!vii#a^f$I>;r)QLm|CfhmOb;IR>7}9ubz-;ucy6J}Y^QDFl(liw-oFJ>KNS_kL|fZj1OA~3BThP;pv`z5D|-k_R_#mU9dn3w(%%KD6W^zr~@GygisNK^DRSvv$3
z&gB{r$grw0Lf=6;NH>0>Y`Y%r&d)AmvxdnWkiVyz}e&zi?z1&QKW$ozj^B+d9`2
z$B#Q8!#e}MBDDrMacGeFMP~<|f;N#s-5bzI{SJ!)(*U5>HDk=XRRaB%(zwc;o8dBB
zZqjL&eQ;guhD+DWZ_v9Y){(~`4@w2ER^6qm0zVd*TNQgT__jNef~sQ|m`BxciWzjk
zO`0Jc((Dq*O?WkuM!o`AjVE;vx>rIUeUF{cjwK-ZXLk1t(G2{9g`Bthn+Bx)d54q>
z^FfEvH6h(o_3$}8FO0kP4yNC1p#O5E6u4=*w%@xS2>S}O`0fwhf{wwRYa9+yU}o$}
znLuv?T!2Omj*_{+-NaUKc8CwuxgmTKA3Wgb+FsHZ#xNlG=?8U8bHI}?g;1V
zhzJUJg8)^;{kHgC9|%PnII;%Z0Nu83l>%c8e0@W}Ot8ow9)C{1;~wY>Y~DCNM=>`@
z`2}lD?Xd;|@$p-q6&he<-&Vd|kr$B0lV<+rdjUtIcEqR36LLQP7O)=Y336{JX{ne;
zfI|+Kv_p~WP?Y~+Fma$DY~#MCSjjF4Lhn)S1ck}MtbkC(#=U!hJ)fjCy+juBB>cXb
zoGcDA%zoxRToMB#>#C16E{Vb?Y{Y7Z72d!JqwwbqQ3a=^@){-UBp^+MgX}qJJ`nh5
zcwoYh9Db=Pc>AQt38Z{FoX_n5pk_wQ;1|tBn90Kgpg1R3VZ0c7hG7XK>yHk*pX~z-
z-Iz!j@}xj?0UyuBV^K)q$l+%=$p>Cp3`KjGFhPaw_4|AsUy;m@S)OXvT^NnTPA22E
zPsrH(pt|qZ8H8ovO(DlR2UNHp+F$ds8~Jy{%l_i145@qmbNJ!8$V`XG-3-Nyc
zvmY_tLKIk@-cA2jhYd9uk_Uu*KxEXil5A<~F<$q&=Q8vy$WnfpmsmFqkm^a_47ojl
z%}nMiyHAW^r?US#^u#VA$}#s
zb$5~1YQ|spEHA*@`7iTG7@0xr>FxXGduPECRI=Liq=hqw)jxKYxWO<*$m$kl_x^VaZn;QyC|N{-WW~#K<)4m_
zf!l{Et`X#L_vCd9DLVs>l)RE1hFD-Y@9n2A`@P{GGNurl17a90JDAM9CJm@;ZCk}8
zv|!xnshP8DcR7I))T
z0OVLD?bBmU1%N$#X&!zC(_f->^h)l-;$LyH&h}m>zWgDRODF!M@ZL%~;`|oqzZWJT
ztvL%=R#!jnw#UFG=fa&`n?(5EQL5MAfBE3lja-9K??DI;8s`&k{{jSCXYxG)f%q?NEly0vJ6`NLRY`2bw~P1A#ONRH1A>=9SM7pq39L(ij|sId=^0?!TA?
zmwH)b6*S55V|U7{5)RVnq50hj!_YCn!lWK`M|%{~h&t77sLX;*1x%!-e;+P+6zc`D
zjlqecxw~1rqd>^!yNiFxN09o?_sfUP1-KeS_7eTUfMyJ6FL%p{;Uuql%7rh_fVs*6
z*I@OZ;5X4|tMI8~C_wro>e4zt+S~mymokkDAMq2O;=Raw;L~YVtN>iklO^s}j_nU!M;B#m`yRdhLMUl(bOa
zL9SIdWq#iI)$Oh0#$#BuUq
zw^sqt)Ob8~@|F;hB3im+my!9&9`{o?9{ymLfbI-8X@0=PQ74ZKD@!VE+{~lYwKc;O
zeM`PrdsoySzvu9XxyCyTy`nVtdYtVE8s)mj`u#@~swHNm^X!Q}+7u<55_(nz?-V02
zR&*T%mp*$$She}0jOn~unac*~l#Pao%qcbeOyGYNjS_r#{@H&v?dj)GvG7p)Ec;yC
zTpL#xPkoAq%BgOMy4v8^zcn6i8C^lSn3cKyE0D(TUwUPw=~0fSkt!^D*yiF9XCI&c
zrgRzKiqt&I_sIZNdl?{glzkoVtk>MRl2C~gL|W+*{OfdRD?$*_bQo1nY8jo0ml
zJX-VwakH#Fho%q3OmtAv;G_%lKde0Wz+2IvCOf@eFzqPiBiMHlpLzI&q37=f{NBar
zbdQjA5W36u+i+kBW
z0dPvN)=U2*OwPLzwR*Q4P6?<4{Ciai4E-e3CLAlE*IhH+qo6i;c%!-gSLg(!3L%_&
z2;YNvCu4%iSR23;eOIV$vIypQ@?NwI+l7i*J4NunMqqmT^H9`G8E|mBnf0oo3{rJ?
zpQ7O|13U3dA$JBRVMFqWsZPfRNK5?g0utMyh_kLbCHWKRs@1vhcf14qZthH;n*R;t
zMF!_fMUH{i8uz!!uwj_1(XW;k_!PWazj>D9upi`anA+~z)k0Y_EHkn&{A9nMj;C}8
z07AFRhtg?InKg?uTp
z|C;OpuQ6fpjf6^IpF+ZMMbIDAR59y=RZrMVWJTYNI)ahlkl^-f<}i>*m(bw-C!i$x
zqETM!IW#3F>DJx#hFDyefg({bC|~?(^GrVyE;`w%+|%y^84Plp>?3cWfCanV&R0L^
zxv*z&nL7$}MzIO;DkVV6%va~}+ioD^OsTm4ggVgZx@;TeE)L^$H#teaaf4J%USi(1
zMJ!(^h_L0RBe3$ddpv(?2BU40HKGt?hgWTT+G!uK0QnDsT{LU-Fgks|h_LxKh=FJ}Kzbynd4J6F$UwEf2dBd9ES45|MjYv`z<3`sOR)Km|*oo0aamVAo8#1#Ow
zDtOPnD+6;iul$Ye76q|JUMa4gkKnhjNxVX_Oz_j#)68`fb&zP_pZJPi1*Y%Nbsvw(
zf=!+XK9T(~%(+DBrzh1Kra>MA2&M@@*R@k+EBpM=Qf`#%(2TuuX*NeRB>%$qdbp*=r2)<>0M=KVO>1zDF(!YFh1#R$!s5hH2Y+RmjA~pMd`a
zh`}eX$A1qil8|5Q+UllmZ?OW2q&|w%Wr(_hH@PBlHReS|`D2joEIfloTg!hSgsVqe
z`i_cP;O?=^r_erAn7t^H(4HO%!t^vn-+oAg$HAi@=7cv_C>eNup-K~oh|mgWO&LNe
zyvgL|q%RoCp!9#c77A2pY1;mC<$^lziv-`^ItL8PdpF-yT!6^;f6KCjT+mAN-`;X&
z0JwPc`Ply19!A$~?;&uK4=}S;D+ir_kqt#i!Fu&Hc*@)h4@13R07p@>#w{**s!de8
zj)W618WB)DdB_Kicqx4kt}6k4J{m>OHYXsMl=$+{GzNMIy;r}Xaw4GGZJL4X5}^2_
zx|+g!9LDg6ng65v4aUtMWZkND1CB47BhEBFgTjX`a;;A9K!c};<0w`4Nimf9P}k!r
zc-l}WZ_n)pY?91vQWw{PT0Vuc)W3cqk3Qdh^g1ycm|8a~63X?!>P1QqVFNc=J&xwrU~9)^HR#yS)2o}7RZCFwI4l1b4zq0oJ0nWCS$1;*+xUeFHHo2H2T6(=vv0R83
zPn)%xDtI*usgIP#*$Fm*C6Vh{SWJW;afx|JI&z|9TbhQO9nxq@*`1sDXcaj8@J^!0
zr2+D+imaRd90c)C;<-4UljD5KDUW$iv!bS`$AJ4$Gr;IrhgvVUK&RpQ{a2}-fMK+Z
zE4E-5h{ohQJGijoft?S|4)9FC-zqA<%PYsh&gDAV^``GIoIPA&C4C26?Ikhk_vb+q
z>jd`5MpV)MFn;S4BBK*0kbvntk3HJ|?_6vV5{=81=l{ns>yElO*Hx@7aG(cnbZEp&
z6gN#^y-CM$1&y-qc*oOm1D}2Ks_XKT2RIc+p&~)*1w6}J&}2VS3}x&n=Ja$`!Kqu2
zBY6c|l+(#0j@I83H{h&^2_2(H|5EziOo=DQD|06Ard?t|Uq9*AxX7l9PY2E2Qwj7&
zZ?jCw@ucve(}A^rjou66p23e^u*b@x;)l1sUAV4@i}M%p-S!H>rRLRn&Mn+R-Nenp
z4uq>vm2XU16pbIydNfY9=t&bw8Er9zI5(k%Z$u3GrhM_`Hlh;KfM|4S>aaiI#P{1E
zI|0mnF;vBMt7yKW8IAGd=s7p*f)g%q?l)b1fM1=ie060y2jvOdwRT@^Ew;EVmpstHRf9=??S_voPU#-6$2ldvYeR99z#NM@_AJU)G+wceDepkX-
zIfooj4*BdbF?kF0THzlvwX|30wX}ESaj)Cakh=wi1->nK*@8(Jr*JbaxEB7tVWbf+
zU_acM_pQROEvZn~gqP!j47v16TP^s9l9BP$*k+tY{Y*i31jPGQE`6Wlt;XLAh>Beb
zF2ldPqDtvAYQ-;6zCLVqZNXQ%HOw#eG~l-mm@Ypte~XiZoJkxtO2tQGU;Bq^wBQAQ
z?B}?doAGis;yZ;ZWq8Glnib(oNJMsG04I=*ByHG$(>(DAcJOzbg+KTw?NY^s^=FtuO@d3HR|uxU
z4O2Wt^;&;eeHmU`^_WC^{wbbq!z?Af?1EmSNlhoS)5d*e9_#hJGsQzc_E^c>dy1Pa
z?-%11k6xeff3;k%jQ$Si9ekyK8|BDh|7jLtkMpCCH~4d?(Wq-YKOVkOMBQId
z8&q6Z#sz*&y|ya0#re8}u=Pi3_^f@chRvqoD;Z(
zpRN#8x$G-}N-fa+jC_9{4c3|Pq`$_Fmv=s`2_d3IBby~j?@Fno4lO|sL<(h4t^ke<
z&LMjIH=9Fw?8*I=;XAtfYWNs*2C{i!D)MM?W80H>DsDUu3u<}wZ5omxODW4>3rsu~
zUWh5*gBG5@RkYu#;#5qmL4q%6@Hd~G$3DxGq33sU9E|gJVT#0C)^PDD@LT&lg^1cV
zEG@}a&tEv$5T13h4sI7Xdz`Yq6nO85N`a0s|0^CCm{N8#y&w?-8;AK*`c2KnmdS|HHr
zAZy-V0|SCdH9TaGAcbvNWP)}hJe?jnVCK{a3eRX&SxQvEC(jNCpV8-m9iG;)_18;)
z*Co-YJzxmTUSCRfZTJ9d5;WEq{^Wv;7NvuNkq98bEy&kxG!8<0oxiU~yoSPhn11@F
zI7r=}FCikA2&TGBw4$j~pz#O`-^rE$|4^`R$}g2bF|VjaU#ch|p+G|uOKkwLjppKO
zO!uJHU5^yOphn1kd>o(bbRTZJiXNU~3I&qwH|V~MdO_=!uHN^B<{*D^C$~8J9pJp1
zBq0zV185M6H&XeoFiVm#PR`35w1^3t(qxK2Nd$!7sPThz#10>gyd$CW_9_|AH#0C&
z(ReZawFCE*3xVkQ=!sc-(9;gb-0G_wl&IM1wrRDQa(>pC@j
zA6o+uwTKz%wwwnTe`bT3fLt}g9`RASW^Wz)@hsg}@bOh-WwR<$W-<&}*#9LY)DwqA
zYa5p`T`E8dE6uvs$Yu~u#}gJ7sVO$_!c?hv+6!~Oc9l+|<{t7gu+Tb^H3w5rdzSfe
zVi-$QNL(b+aKqBrsVF~&L?BuHpV^jY5-=v`rq?k)3K2)S$uZ~ZZbZ_|f;=$cJS=Gy
zd2i-*9tisC{v$Xe2t}tvl>{@?fNmACLMmbl`Xtzt=xo-p-b%$wBjJn4^3@WFnwsC3
zho(;!xA|X8wWn^;XF&sq-<2vberVhVJUK8rXvyLcyF$
z4?jSUrzw7?z7!zcZ)FKJh2(*)5Y5lO>6hW(w0oSqy>1|otmM!8gA>0x;UcF#?FD!>
z#M;}`ni~pdim>OMkq4#f6@RzgtYJ&~7jW%+G}I&`*-&YhgJxkp_JYYu;N{8-Im3@y
zFa-(olw)-O*CZG|zB90b+4bd?E#4E5Ld>X)p5-ez=K2uu?EX)vSK7?G|Kk7zOZeK=
z7;>PW;AK~z&RS@2`I^u}mrB4`<=fxwPzx2me^0LaoB~F?(&^+){(uLems@?klK~um
z89*(P4Ud?%eyg(-0`qFi6r15T$g4zJ=Dzj=KKJK-!DEwjvOo5TM}<5EAB94$jr_@l
zI?p-xf0;Ld^?8TZ!uWo`k+k1l-x>+mFVk%Ckb6P@!s6+Ye+gj2`q@08Tmwv6@Ai8U
zkHP2&-G9gw3)&^`I~F>}kJqr@vv7Pbg+{V3=LICH<1tyaJT^Z$?lIE>}`XVM+4->)r@#%!1KZx
zLqYs$xkEvRL@d5V-3;c4B
zC-%p~@Z`S#+i79S4z)j`-ak#3gSsS&J|W)dLJ`N0{$8U0p?gv%oV+0$JX2SI_uRNO
zx-KIAkHJ#~T_GY+y=z#Gdf#ZMDtRb^<2U_oam(|gmS^9LY=-gU3=|F(A8RadX9L-}
z^U06!^QzR(u@l`%%<39Rf8j+`exr_onTs9|&Q7ob=2j>Wt@FE;AC3xT7`P#el4x7?
zj5&$58ZI@X0UQaHQSOL?RKg+~Ji+P4Q~N0&y!5z;=xJ96db8p|^AW{I^s0@SM)eks
z>YW|5r8(D%PWqCFFuOFPsn5ht#HKRzU4y%hu3A0%DA&0o&94>ZlXm)6hcu&^&q5Ys
zY@1Oc4xwthxCgjajdoLvqytVW%4u-r=85lD4PO!wXhOgCr2b$B}t7%=$S4oZ!fYMI3;hz;F%*ZU8aOX;A%<&b5e@+X3t?{T9A63*+
zG`U=dC(K>?R`anM54+;e@$o_-{!dNyO6sd@T(I9k_s*FToWN^q+IP7a*S~5Z@b<|I
zT*lQs*Hzg8ozdU+XwwQq`!gQaTI9UOSNIw{I+`CKTI)DYN7J{7Qa0f~iGxEobYJ0qM&etiABLf8>wiob9yp;RWZ5BEbr`btQlu&O!2qvbPYV@Irs@?)^vR!6c1t=GP1UH
zLLCSiV&of8+#-j=*}zU6H}N>Mbxw{zw>MlvZ8km88&Tme7RLm^i}{};FP85B=sILWVla6~(t
zirRL|9Pxmzmx^~fxbSBEbW^=pRrD-Nz<#L1$?yMoTCKcpjW;fwy)Q3hfcM&JcOBo8
zL%q8k_=S!r@t|U6Ct*Pze9tJ}Op;s+AIW*hR-!A3%AIl-NH#i+jt~FVb_!?2U6pR}
zH7QY`55||{EyUDL^vP@4WHIFUqY1f-RZMhvO-YNdBpW5#>2pc{=;sU!R(KSuQKEt0
zoe%qVN&f=gAF+bI_F+a3^)ptU3zl#(sc>z7P^xx=JvQUTws#zkxUZ>uithehwz}atz5+BB8;}x4g6gsgUb~
z6W0~35||OQwIbdT4OCthh4v>|!uRWP>#}jduqyEBZ6v=G(wh4TDFr+O+W%TI{hae{RtdaDUvo|PaYk){PbnS(1aw|rpU1o7YwdpP`OIN2kQ$&j|guGg7$sJv=L5j*f+Wtfd4%QCRUuhgTM{Anrp(O
zYc2||k_*_X5L6;JXOOybWD^s0td+>Op#a+pz8$B1WI;Baux!N~7ucZw`u*C`F_tSu
zTGD*!52EYRlKbjl9C44ZOf}9B2jgmgJG;04VN&g^sa-v`6WjWP#!lnd_9A_*(}X8DXf-xax-JvEBT$E#-%C
zi=;X%-b+LHLvtB2!kduJwV^daer;s>>f9&Z6mCq_*bycXa3Fgsr{_s)&mv`lq{lU#C89GERrHB0NrdCbz_pS~!yD8fvu6Q_UU1_rc41bG|pW5oZu-{<*MG#B4@~aGP!&Z*~h#`GL!8Hu0NQ=Y}|`+mG8Pnb4{G&^Bl;}k{x0y`nuA?
zWAyOVmu9x%g%YH*#hv$VZxL2?;K>;)Uxgg!eG`hr`!TDt<~d9nJ=pBPL6~e>EH<%;
z8a<~^KI!q;VkIii!U*m-R`T?n^d-G%TIi6eM&>fR&)0gQi0_w8-LBj~EWMkeN-ZZI
zDR;KJ$p4}ev%UJK!s#uBU8d#SyD<9*VWlqnYb51`eH(22k#A>+9G}AnYDeB82SUn*
z28oSWitX3QMO|S?^eOd|Eu#eZV{?4fOHvU|Y0!{OOPc}l%A{}-1_O|xe&OvsZwlZq
zLh~sUo5aE@;+|oy%h*5P+f;^vJiuRZnCsJn|G>P{C%t=*GZ3>719i%x@0fO1@S2v?
zCc+$anSphM4a#3CuASZ&h7P=*+m@lT*p2!uDNCwfNQYaW;)j(rjLe*VvVWZ(P;8Iv
zl{Ya!=Qj<$g%3?$EE@h-0k3^8;Z)*LhO11Ix;0ta;CDrV`@pm@G-sei--|M3A$K=#1d7&zJRHRPMO5zSP`)A>U({3=G74ULcAK70)o6s&R@11dm6jx}**_8{!t?JT?Kd13-5=2vu5cmxkT
zuJmv4J_9___q&UD-$0j5(GM)spTNIH64qwFVUPg2M9-T=L5{ZiJB2d}prOK%FtFDj
z>>Zz)&)u(u*S+v>I&L3(a-OETxD)aj5M7YfOV)boJ{p|6_vhN|WoI
zjd#RTDNDA8dWOM5^7VC6nhg+aJjX{Ez7NNLx`aO~6UP5h`9HU4GsJ)7xPxi;Z75^q
z#XFt73Ho|$G&z<#V3k5h)f>tS=uAr3vtKGgsPT*7mbRlM@MrDprFy=9zU7h70I*4oa5jiZ~qI?rR*?C`6CAHP%jAxm1+!S$Oafq5PF`>sfa)Fi=4e31JEod8@fI=2E52fVguDP!)Yki5=OEM&_s5t
zJ@0_`C}yHMO{ZRjs%<|yV~k^n_4L98q&J74j-T+
z)chKuE8%D;g9W2rxGy@^w7kJpZ-QqZ7kJ=wPB`<}ft=Vk7yN#1ee~0l`OJbLHnjYP
zF|KRpuHpaG2&J&XwlYt691MA0^G(yY&}9Qr4j!9p=&gQQ$&Ol8bi0zuIWWrzpR?d$
zPyVEZM@YLWrp)o94gTE;JqF6?JHfz-A6+sikmdj4bDWC8A`%7Co=BAQ>l2%ioMgP|
z#c<~M%qx6QxKnlUOaaaqNvA!f>5FoJ?cDt~2Q=L2*#6XxH}15$tN8js4DOb#XNcsPm~-`mISDl=@hIC09}n?R%nq
z>RPNI+P3Z&FL_ZGXCv&Rwb#+c*>qMsCU>sj#xK9nyjqe%l|%c1oV@_LvB)=jMobjf
zQ>VO#t*PM6pG0CedQ|c4nIP3#`+U6l(xbeUQLJ^5Q4Ma&kXqm@m
z{D+4LwMKs$eSrUaT9j^4?}=v({l}FgZ-WY2MN~TIYTywbua_zqtZ_?)%b#g4+o02W
zHl034mgol^(pSiHS^VIAf0MEKB|N5Cb=o&V8r>aXCR7a4!Fe?dE*D4Nz!@F}iFf`u
zf}5d|tp%KN=wa`@_0g+@sFL;R{?bD&oZZ(B`1_pjx4Eg`ArEBGZvl)8tx+;~=wYa_
zU4}e98mHW5M`nZjE#JCwEaZv9-rOf&iV$34aLZ5Y^vU;`7&EA{NaMDYss7}Ow)j<{
zK0bMV9o#BXp@Vl^0$o4S8j1;70+nSC>qpZ_aS2{HT|Fm@Hi^EKig=`eQY~KO8wlJ1
zmRjFOiL^;^wlRHHH@0<0tQ>%u1UL6x
z=`XM~L6%2xO%abj!_Y`KOoG29k`6>`8Bo$xHc#=RS;&;FW+B%;2?x?i1-@A-;Lj{>
zu6Okk<6)c|Nh9iXs5RC6UABZnSd;N%qW1bMw9me6_u~g8?jx%$ZzXa7JY>%NjJ-Dq
zD)iUmJQgP**Wt9Cqi!F#pC55~sc9cvnxU4ESStq%RgC%f9u7g~7o?rki>)ALI!i4_
zDGv}{UcUQHeh%i*E;*e9h2Yc)34`f#2VhE+ei=Vp0*QENW`J-S1>hpZFIbnl}Q3br!PIj;n!3t%OC?`8=2uIj>)wlL)K_ZkMmG
zd4U4jhR12&y}^hQf`-
z`2b|X5BQg+5xfg_hKe(Xz)MWUoYTh#27i|sTU2+3Ka8)1UwG#TqQb7N&fau@
ze@&99vzBCm%b9FnCsj{4O_D%y>6|KftO%YIzn{hOTg)tf>>#k~*q%;3Bmh1P@DZM9
zQUhyd+!o~|!oW<>#73i21d7<7V;UqC1J(xeg|OHMe366?Z>KYY*R5QF-udECOhG?Q
zYFhw2;B(k~Ps;^&vVmXiiZ@J`?V|ef*qE^?d>Q2G@TB?&0PpRm8cagq)^
zc_z)(6+eMId+eXhq&klYzvZUNuvte8x{{dQ7fOOF1{!%GC-XeUNm<2|wuBK66a>we
zd`BEwA4@JPk7Ly%9ydBU`C$ab2S(KkJD99P-}?)ne($Q-Rf0%Ct5D`#&0#EiyT8ou!lBKVcgm+x3>k#>B#a^W`%Akd^p1*S{L
z)h9()_*=eUl{OSKH&r?jf{a^*4})K0S%*u3Xy!Un(RnpSYxg-8#S?MP;zcgT+!%|s
zuVf?s5yA`8w_acgTGx&(h-b0QkK7lI^h&WQVtFwQ?H5Q<1yxw_#bhj(lxdku+83$K
z?=ou$A4GJ1v9vl8l^OD
zVqA_)a`oZ$jlRPgk#@gEl-ZTt(x}gWYn(JPA^R{wh;Y1qAI!Wafp~`;4{*!0DYpes
z6{Xd5H3k)mN5+?YQ`YB^ikTzcYn-Lawa`pIhXuR~El8prY3%0ayZ*}am$I*)$~m{7
z`Nj)&*K2eGNst~h(&4m*g+|HW5p5qEEs;{ErI(664hXTvcfJFMU@X`8ak8#c0U~_%
z5eHRzKf)>zlK%C&IVRiU7x?wDIkMT)Qzo-#kFC$nof;O5Ms`?-VmE$eBD!TiJ$nvW
z5%XH&b&E(DOw!iUIiFJlImg-co}t(UJ4#$(WiO4#R?NoAnA}byUNQw~yENR$ZcMOb
zE3*tXl^3N^i6Rl1j1Pl)tCHBWpwksPJ?V{=2cm16wOokA0)OTfp#>HM
zEeC!Qx?<$7@9**5TtZrQ!@u!4&SUNs%ZzGsf057fYZ{q5RPc6F36D`CD@ZH0WY{r(
zim}xHQ*pch29s5h?GZDmN2ETVRV?^4gMDkCYFS++hHZhngYVyDVnY;;^^c=+5r2{p
z0=6^7Shf$h7>#)k^0ni`(plD6r18$%h=^y;k-Ftv#`ByRSeYxyVXDz0
zOni-QKg^>IakOZ>b9p2OE4oo;#_O4Z6xOALyFbmt{$vnGa>+mhJv627?V7`6$h5K@
zGY^qTs`eyfVlF@|8g@cImWC0!drP@Rw?TMfVKv3RKA3gn&fR=W0rNatSS%bF09hxI
zD?jxGc;BTs={lVJ+$KuBr7?}jKflf85%(8+eb4CT-`FGdImdTqk$b>-9WV#if}@oE58g(Qq++9CmY?(rLkkBKmP
z>a3gNWIA{$QfH>kkHeOt`Tj8H&v5?Fqq(GgCvc=9=<0jP3tn*us32Jn0iSJGwvfRL
z_`-%}c}c4V#;%jWgIlW5b4`$${JH_yRn;dp?=XiBv*PFU8zTT+2A`9aVJbM#)$>c*~GF!d!B@wgBKsVi-Mt6%v8v~F%}@k;Cf_vQ<%@u`Kt
zcmwsCd|U+6uoim-Z8QfGE*XJ8}53i0S^*^7CgumT&i^aUK^L6!rtag3g^NXU|H>r
zZ|KAjxEK&bq3O2^z6Y;9RXLQ$1vNvTeXeCg-;TcxHXIQ~Npng!txo1<{g~4lX{WnD
zWF|eWg7Yd2&FU#uj9|f;FGxtNpJGDybjV|d9&qC|h1&beMN!~(rixyR=sNhprgRBR
z&w;m);EUbx81&s6?=j6{K$p_)^x7LI!cJr^H
z%~^6hZ(X{J^vSO+|+(+LBuMtDC{
zD%p(ub)2A3qv>Lw41Pwdtni!jX;j%dk6SO69@Suu(`=4DkJD4u4@&%y!)-|={m26q
z@pQ$4w>lOQc=wkoUEXmXJni9F=7lHp=uX;C$3J8&XpM{Ar!vDcsDRrR`$h9T5aS|H
zd+Xi`Q1N0Y@8G%8ehh4=9$J<60Xy~9#T
zhZEqVTWgg3_{ZK)`r4aZxZ~~LLt(u<_^S85WN^$GJkG`Hwr&1d)bLrsZnvixs@N}h
zIq8=q+WO`~K+MdE?n$7zN5PaSFdqLWL#~F3PDZN#IJZwopS+4Q({G8#&7>i1snoaQ)ts#z|-E
zxf7#AaOS$Y`w}-KQRBIx41uc>sHMZOhT+a(4pM~E3sV^3?snB@o%hfz9M?i;kiZk*>Hzdztxj)oA0-PL_
zWnJE_f+FGvMU~tV_{M`$S+m=W=q3N^Y@zcEXtxW=Y%Aw1Sy5z{1S
z#WU;sUkE8sJDUu>2CEYeocv;zB<&b{m@nIuOV;SoFgnwctD?b$V0^YkrJ3jH@5BVdX
zOn6C4Bwk`Q2{fp8f1Tws1n*0iyghXuL4mpD!A;w6XkHgSsc;ktXdW-Fzg4k?4R+gO
zW`VA7mH+Xfc7s3Mj*gYj90&xwGQv4tn!ZqFdVf0yYf6mzHcU@~_0xP#Dfy-tgF*37mNnK57v>hd5-hN$0w(VOB&3Gu5R#$Uy)h-yIt<(DlXX
zhJfD>WQlL<;IGpnrjtM&oXGthSu`uv=g=O(Zsxf^CtKu((c0YC#D-QeviL3CoZ|0@
zg4KMHsq`ooZ>>A9^RfjQh}x5!eaZwh-HhLe#2q7VhdwlA8qQ+BxI=FyYkfwju5Vuu
zuC75`jbpbxdf0$5UvNgmnQeq4Sme}{)fA@bV0e*xun!@8skfE*rwBWGvmP5*x{Ev_
zE7u*`Pee}ZoO>TYUWT!Ca9{kQ`3{jSb}ltJxwhOOBdy~N`-bgGZgR5fRA5^i`+k-5
z*+{8==uM6BWbFQd+`ute2qM+Ka+2!oM~Xs+!c!cIk%j5K??YcwuzsE)D}vL}h?%|O
ze-t;&u#OPraeOBgyEH?B_QaQA%*(_6-i7xOA!1AW^A^t7Xs{9DQ4)Z;6NWp7+_2qSU;c`^
z*dSjBro+sLOtG8iIb`&twULI!yG`ejI7Dym^tk4M9nxL7Ygqv=V;}$YWmmpbMtJaq
zZlV4eW$uul684;aSQ~lAXBQn{5%twEyLo<(<~zdQrRCOT}U3hb~4
zBz^JrCbtmx3=6{_K^L)o4Qt5>u7ApU>$=p;EusjmKx_%;_twVmt{Ge5;)BY^EACCt
z9KSVMOZ4VGTHH|PZ0lL|SKCxx3{;gLo{Li+8&-eBK{2TOUmO3I-C|5xb$rU|ZC{h}
zkX=Z;UD$zgHKE<~IkD76Eo&VPY^tU4oAd$egNS-%nzTg;p9dw%j75z48$NxFuZ^Yq
zPe+6(m(e(kP0?~S_C25?va{__)_+Oy(Y?4?*)o*vUbEe}vdyL5pPb=Y%4yE+N|a~1
zmG}O$E%EvESy@y@(lg|Fk@7F!M$L!{gGe4w%f>~#w3#1dJZYj
z3{}&(e;qsAPYuWXqOfV1lI8iAJIXr+r6d&p?Kie{&qQ-Rp+FAUlbVbV1hGr=_>Vh|
z1{fn%T5M^0kFwU;v7dU^MwCB~M%-B`|JhhHmG|X1ofhGry&^}{tAKcrRGm(_SE5XX
z^KnY&n=13Fvb7a_pH>d@3w3cl>D!cew;`X_CxPXt)zRKH`mX$m|Q_u-x|HY1RQ{aN(H+6{OS%>Xq_@BM?I>G$eL{gyP_
zJpnxoE=cwGyl_Q$T=h;zs4HR@S6Q}5GcO~_rnt-JMonz4@wB(jrXC{p=bjkR3ukPv
z$jsKm7n#Y^-gvM9T0kbJ$#?-
z17hJ4TV>b?F_-Ru(r$}xr0i~I&%($umZEkoy_FwbX{H|?x%`$
zqrSbzYOl4dz10~+lFBrr-`p<3#Jbw1r}koy^Qslf^RJV!l=MC{#VH5*MKH3|`RXI)
zeqV-At-S|P(xUn{BN2(X?>&3>`uZ)*R#mU%mx&8@`&2v9cCj95ku*E?ZhR4YpdZxE
zW5WsmC+y3ksrugk6EYQ*GA5EFN|HwI*-DWHB_vA5NaiU+iJ~%OPEtq_3Yq77&beH(
z3>h+2#%Pwvkkt1~_m9tet>5~sWv$nGhW+ek@3Z%_pJ$)zUgzVesZQQb=&8!^f$O6*
zL_lfTs~_{1pz@z$LRbIVL-l@JHh!F8N9uoy$-Gw=f?l|`lpR|sgZ#9AzmI#x5$NK+
zR~L4^xdhcJ$EY06g=s^LVyjLKOv1mLhv==vOP~;e`D1BK3P_2b;?wbRb;NqUIQ*kq
z0L&3*QWR5`|3VbGkds
zzeLvXobN6eW_hV1u80%QZ%6vuBT)0nhskealM!o=Y3`xUMC9kX
zk{1fSM#v5J8UJq&jG^S$w+!EYfuMiCgau#dU5EAvNKM4#pF-HrACRx_(1K(HvZXV>
z=^($H{C`^+9YluRr<`_%*&!#73i9@f+Ceps*B`kmLxbckOl!MdDM9MGf=l_MbfFi=
zzpA{+IgiyZyQ3b?1|t5Vaq-_>vCr9Q70<|7e1V<~Ol1lUyh4T&ip)CQlcDUANZ
zG+Cu0y7XT~GiT$F3kt8UO~@xB(F?uno~)=uG~GQoUR~8jw2M6HB`pb1UXQ++Ey~P81_jp5I6O%c<
z+v1>My{mnDps*38`qyRUa*hMzlUxc-vr~MpCQ|gu14ppkWQ8^Eww9(6n;?VxUny^uSE!Z+id(dUtVQz$m>C8s}KMa>J%M
zBp2$N8(A`j`0(q7bw~Y#T-G?Ps5Dpxl3MlEZs_uW(0=Lm+=3d&^Elu?G@1b+aZk=o
z<&_}p$BQ(pa;uRxMPCiy*c#;7+qB<5$I~HmiEA4bTpvLFQg=EoV(&e4WC+k1mcyR
zZ_FE43B??AalZ323-Z!)d!c>(4wAFBE`#Tg50uOXqbuvep=rftQ}2vQWY;67;JbBn
zB#Zgmyw>6YaxDDU>!0rek@W|Hk9^L3h#dDQY1h0r4mtHJH%xFRBM&2v|69AG5ee4Z
zFzS8dJ|wNK=bq?s4^n!xnPYp-L!{iRC;Rh!31aqMOPu>z45A=9krdD#4(+8eS{42F
zLt>Wy3VGu1A>8_JEm}DuA=NF1V=OsuLpQP{e9V0=BLPk)ugIJ=Kvs?JFn`fnhq^W%Fx={-3#qs+_t2k6L|jUx1Cp`tY2g0pY@Je1*wbBXpuZ}i1Cl-3w2K#;JpiK2A{7=KzqNY8OIi#giKYMWDj{w!8J#u
zJpU~hhN@ob4t|pg<()UzN02M2eBMWr~hV%rB+K>ZzuMaQu3ovW9*8(yS6Lu;N-%XA)>hTiP78A^*@j|lWk^K8pp4J}<<
zaYd=G8jfhpR$r4vqk*~u@xOc)V0-+a!%FOJ4@qOa%6kuKVI
zRlnVhwy)v(m~C(Ok5to6c|||aJ1PX_tv^(-
z8$VsacnTl%C}HmsegOxMaCr$7r_tQ8pMmgUONF%`M+W~p(g$a+DBGJ8moY>_n{50^Bm5^`
z>hSMz3z}EIHtl%05v(`*Bq7u26iqw)$0kSRJGAhf?E;lGmNYH5?HhSuZCIXL+V|@w
zHQLpctD=tyiNL2lCVlSdUxB&L=qSq`J`ab_eM%D(1~e%}`mFAoZE%15K%1ye2i*E
z&wwUmyw&V)pE9ifKB{2z2PvA%pji1cLm}GcGtgZdy)!gP|EdRazME+OMsMVD>a2p_
zmB*xtIxeHFHrydBDK$=?pbJ_ZuUCRwHm@>T^LsV?OsDYt>wp=xDuD)bW!Dkl$#20v`uE$gC2dK(+ECkP&mUTCY8M_T
zdj80MrS_{y@qVU$s+!$auAVM2hMLAI>DLJoF?8477EkG{7PUbxotM)Lrka0J(7q#A
zThw~|J{gI(*Q+%?V3Zli8`H1m2Mo&BWYS;pR%e`cxj^J$-&qGQ!c%Re{N?*21$3o(wR
z^G1iHHTTb`b-P&PVQ-;nha4`SDvfVZ+rG`_%CFEmH726~_)|EF}
z=-LGtJ2j2!=x1ohB_O$MHND}SO>2s(=mOI(dP`R~t0_Edcko-lJ~CIivava+icY(7
z+3TO$LTZkqe86?*ETN^qXtf!2i2#_-K=
zYl;p`e57|hEsQfP`J%Sld*6KD{aLyV@8Pq@Jw$0KJM))cRo+E2|8r{1aCn~@=cM-D
zywK0|^l!<>Z!Z0*#-5$j(UGOSN-{>R&$GO-NTOy
z^nu~+FK*_qrRCq#6XR=@fu#fWPYv%3g&&&NCc1nMr!}%$&KQlv!`@||>Da$|(NOC%
z+$zgpc<9bIKNd^oRaf?dmHVWd-<7c*J#wWwBwY1lcJ+JbsCSm)RPmFgc3nMviMBIAjZ_z&Q(czy~
zjiE`by1A0J`5DZ+EkEEEXBREsl_qNkQEni?}F?KvW+!qlS3x&
zMs^5sNkgX&8y$Kjy9`+orjfDm^aPZ&b@Ahvx)!7y#ppQyixV2jh(cez;Y2n@FW!OL
z#$Z)gywhoD3zGR~jmnSgS=zJFwC@wwr{TN70|r02=4oJZPdS_IcbXo%`tN7CBG3>R
znH%d7LL%jd+S$gxz%8}jH~EkB!XdjGtIsjnkSZGm`!%`qw7i!=G1s`U-z7e*ckmo^
z2ul7aY&ECz5V1)MBfM819K
z@oszNhB#%u`qE&11B(ie(!1)=I(~zRF#ZmRY+7M@fM#nKh
zBd9zB9(OOdhF1J(Gk$!X7qYHeKeT0?Hd40fXJGj8orr*W?AUa<7IH>+gUZXlN0IgY
zw!x)IcqV);zG{WkB6hQEaZA`pumDf@p5g(3L^G>5YDPDt~)
z$ZJ+lAm|#s`0}qx4QTjpOySWm8)!;PG1oWlB(hSyh5c8KCo-ks>Tea|hCuRD>~eQ5
zA+=-MnJeGuAU18PpJP4Ep^t+BW4l5;poc$PAD;AehAexXw~E}ff?kEk^0EIlLMknv
zq_J1JA-X04dwP9*km|MN3rF9(AamblzC>|YBRS!3%%@ulAn_k1f`|7cL(n!Wx^!SR
zQYhuQU&E~uvGSIKnV*dm-
zcJxJF3j5Z6(7J?3c!i}#N9!X-f30XIFIz%8)?E10nC1n2T_MZYS#k|BDt-6r*(EE;
zARCD~e&rOTwLHe=+yPUh(0F^y;oq)EAK%Jk?Q{?1(gL5T!Jmr=BROg2hW07sYlqju
zEgn-y
z>Z%pN3cis@r1pWXX6c*Q?-NR7Cc*a55`*ojQFL!eD&jE{7C8x&9NUwD;6Lx`^9P&pSX3s#n%p{KN
zSIUr-z-Jfqs0Osi*Ot=aht(0Kdk(GKx)V!24o*sh_&`RcMq5{I+KZ&U*Ey3VdI~X}
zxlOaqIS3UFt%)=)wt^IY^1&(GT9Eyg_db!8YRHrE8#_#12_g1VKg&v$_CS9*HY>xh
zDzcv;8U0051*u5l$+FzL7P2axI*_q>InM!S$$fq3srq43$NTi0}+w@;65H;Nh&iwGD$epO|qa8c>A!+WVi{1*{$iCIf
zHRbH9Xs0`oBTgG9;q~T*mphu;X$dWdHtF}jr@fiGky&EL14V5Mq?=4s(v0N0U%x!p
zL>n?4ZmN{ZqUq|n$3G0Jhm$3lFnGs~e2tD;zVUV)eAKse)1{_TIDM;j+>h`)SiE3<
zJUJhso!X(a?7J~HHI@VypLT6jde8(vF5Gl=YqcNEbE{^2N5w4~;vO(0JQDz)
z{UzfpobO9>tLZjf?_Nctzcu@DP{NUx?r#=l7H|W$3E6*fH{O&TpCUbk_8uc^b2Bs~W^;U-UUjyMDS6K6mH|+%0v^T|Gb%Zjk+YPdV-^O>pVp
zV2p*$#c?oz)^EZ%`#e#P*7+)1*{(+et||C=ePQKZ+UteJ{|s0v`~*=Q9OP!I?4bgSCnKXkyzdqZ@6v!t4TZ->;~0(Q+gvrgrGs&;*yw
zu^aRo(XMyCYkB9m7k-kaAfNkKnsz5^xL1BclqT!vea}hd5^X3isNP=8fEIGEsV^r+
z1)eR=Z3^SuNE0d!-8Yc^Pt79kYZ`2?M{{x9xY}c#1kGuB*$c;UemFiN?wh+T2aV^C
z`SPU~r|D{r3*Gd;N^rEyT;Hvd74Vn1eV6keO{*nT^4`*K7^Q!+xM;Vjvq(*2X8lz*
zO-b5x(Ixp7aSmGBhqw33l85P=&l=vI2!EsYM^5Opo?$cniopkaRp`Cij(}r)3X!~U
zfpBEM!O)OeM75HI&-z#Nh=XgrlP`_Xy}hH9Q+IK|QU@w;x`;1@3yeaW(nN;T_?8(r
zZokq%ci644^w!aOx`yFbnSEf9UYj9pCO6eU*9n-}GFeorrY{|`vX!%5&1D!l9rCtV
z%__0TpFO!_gKpVPUxEqa3aYt-tzA8*zTt)XY7ipZ4t<jb`23o%o6kMQ05LM0o#YqgckACHUB3vn(g5hpZ2tkeFoa#cSt;<;?`&6n3G`vUeOb21I+r#_JN
zy{!zLT%%yUVRt&JCBN(=+k_{maptvuFF%AzTV7Z7)@(yZpAKCS8$o&@ZDf3pNce>z!Ja`7lQru{HqLghZ1zgpvYl8hNxX1-J_vZ@{Ea@C6Xb<~49
zavhCgoGGBx^+NUukzlaWSN4m5Z4mm};Yszrco=k<#^+nqmw;^#I^~+n(?M<)0v~?n
z0(1ivmQ-&xLo4ai
zgw|RCBvJ90`++EMP}EIiJfee(&s2juj1lnWnM(EY*9cgyI29qI1cTC?XID49Vt@m!
zx06n7t_B~<{wU3KG621mZWd~RfYWnl7m+is!Pezfdk=)P0iIlOhx2D@fdaRKA@3Ik
za7L#4Pbe|Krt%A$o90`=j&;a%=~x@+zjab-M^F=3+fiM>t5*rsf3}?GZm0y(BPA!I
zHdmvCi6Mzu7xU2#JfG7pfo5Pmvur=&$N+bA8V<7|2vDmViC(!60WuHDs`vRKVBz;Q
z{kKKs;EBEeGKbgsAXZn;CYvuC)X$5W{*_0-lt8zJT
z9Ig>b5h(>DRv!~@x}sq9hc#=CR>L5SF}QbJ2?Zv+!J5M{DA4A~xTR)^0^`EyeH&9M
z!PsG!<3pI;0oNSKQ`seT+k*mM}Gj*%$4T`Y--!Jrk1RG8SbX0Ox0)`3OUl}6?
zcr^Oy4gb}0(5GfAmCZ&6vIpSl*(3yfI=J-x%LMG}Oir9UvRW7gy7zn8cU(Zhw>Jp?
zZ_JRbv3!%l8jFO|bb!`JhWJ2*~|@dhTi!3f^;~@xKnBV0S~r&};?*^xQr|
zMkWZzXTN^aIS>K26A#|mJ`01h%JV8$6cFGT`#0D%7y+DXt3O;VK!8U}@b1a?2tdxo
zpbh#AAolxc?G7>Q`)v;CB*`yDL7>9&oy+VI@WlDded!9!-_07Xa%WKBmMSxMCjtSJ
zn>-}rK2!kFP_93mdWGPGs=z|aha8|ebQM;3j({bm^QJ?c2>2U1dLgg|1_=``#s`zj
zLALQ-(UU19;I>kZ^{W5`a7-#Lmq|l_f7iW|4cNMM`Kfm1*wYFSraji%cef0PM)OY?
zE+SxuN|V(74Ji21Cexnts{#yFc=D~=Pz=gOo1-tiC;-cR$`sBAVdJy$5^Iw}z?F#K
z%}LA(Ai#MWHGNbD_C$S#Jq62v{h-cp)6QJ5`Kx+q(U)8h*ONW=K{*?w1{8I?qCWu|
z9(!Yao9_XMDmj~^?h>F^ysX*wbun0T*NmpJEgz%`ZEJPRe+C8|hqJfoKLVLnf^!4D
zWq?+3DjL}FK2F5pnvn1iaO0UfpzR@d-u-9`
zc*c_p`qk=H*4t(Pu73h+`x{cgQMy*PnnW4s4|}X`d8h=mpWpo_$G-q*|30V{{_!Q)
z?FwR6-_8UNBSGp*_ZZ+TKVssx(G#RS+QMFRD;dq%&QPUw=c4{wojt-&OW-
zDjwVCZl{-@3Vxk?bWK;$z4yCw
zP*1)LI>Vt5{j+Xi52GLp2u^-GZT#yIF#PQ4${rJo>K&MOo@>cNr$>!Eu9uaddu}N0
zt@SHFwDh*Jhv-ug;(=gPCWuxAxo#@MXzr@9C|Q@nWO?Bg=@
zc*Zf2E&eFFJfTNhNudhmc3FJA#F&XnE;$?JceeyhTs<}r@wftQK6l_-=AlZ|9z=yt
zYgeP@g_=PE2{kBUJbJ?9Z6Uf(PEzE|bQ#*2o4P->y9yP~Zhx2bwHAHCk^ISks~#29
zbYuF?l%Ov--R4&m!sx@KyS_J7tI;3&Hp!os*Q33)mXp=D8qn2SYCRpUW8b6ne9zj7
z*HvhIbEj;o4io(-z~HEAtU{NRN%*UGRHBlH62o_2XQKS7S^FchnP_KfeTs!bHF_pm
z!g%H<6YchOej>V?iI$4=MsFUu&Bf
z=EgUoZx6Isr7FEbgUrV}ZPzuU1N9QzTvwXWg#|X_g1kz!HU0d%v=vn-*Rq4VAIw*x
z;_S}(jvFh{+78?P1P2Cc!$g91$urTo>6Y#&RVMmENQmqFS|&Oh2Bn4WtU_Dc68b`;
zE76|IWw-7Y)u8#_ZF=cFHK^CX^9vmLHK_BEybnL0R--qVPEHdKm}p4Op1&ox8__S~
zK_}a78qqf4P0rER8&Kr+vn$!Db!g#%-ioo)wJ7gC(L13w&1lKL$o+FCn$iAfewl@j
zO{n3c&NpTe4QNT%c3b+-I&}H>L=iC`?Du%#n0dcm7}dToqQP~w6wNamS}kLogCe>?
zdbh@&p&w(o(qBHPM7{rvwYlh_DEE!icFo%5=g<>4{2Vp!|1Acr@1rW`w5|@G@dtwYx;R#9#{ntG@gV23LR5PqZ0Vk|Tr_7>
z-*LuZGWym$_O^!o8PE_MEYbWf20$U#^VK6t(SVD^{lU3~=vwIkk^RvbC^-LfOp`Mh
z75=0fFpPKuFAl4J(N7WdL1UW3i{di0Z`Z(Kn$b)2n#G^bnfftko|W2%4fiZS%?0iK
zJQ*=y_;^kdbID_%aq)D^&-QfSo+96QZEZHdeicTUtY
z7gg=f4S68CNK@VWbTM#UE^NE0r3C1Vomh3SJO&sF8+@TPm=4bIiu>zip`gOLD^F6O
z60AIF`pKG619|XZ0s>6HK4WX2
zh~+yQekB^mZ$rQa*}%W?J(XZ0OFW+g`#zC_x$2@5fz^O+C)hBE^|ya=`@G^I1eg@l
zoj&?kg2!`Ps%Ctu0dG{py{HQ{fK3LqDl@AB!kZ$~t7WSIP)!I-%&h{et;{E{4>5uJ
zr|PbR^-R#uc`xV8(<&f1z57Luc@=QJFaPZ#4-=?$NV*(3#ssbFn-*o4RDp{7Qfp$=
zF+cl@Zk5Mj?VdYd=isddF9lM5M&&U9@~nQ)?;;bpuGgMhZBq?4ch+qxPOJfr_kR|4
z>eYcUr-DlM#yY?iWxCVzN*!SMi?4ABsR2A!*NxZ@)dIEAv&KEQ>cL%8g^)*W^`Nup
zvs}Xo?E62SR-WrZB?wzH=Ph`%68tQ0|JnSn
z5@hBiYYN`20;bBVJgY^jfV=1>o7_PLu=)D2(L=Ek_&OGy8U{>YlJU&vqahPK>6+BM
z8ps4XXA`{G*qFd9^z+GDzA7N_fl={067xGD>r!@j9e7aIq|;nk3mo>|a9ojF4f5)^
zBG&X&0q<8JR!X-X{9xY{ckD$SxbWk9?&XL&V0HOpT5nAafCP(ss?w@JsM=jo-ik_a
zMkM2wRtW+k3jaiY6e|VSFMj@$(47Y=Zt(>FLmvTPl%K)m?0{fPpcPYjf
z%YmOuvi)Vw0&vyV-v>?31m2^D!z+xc!0iDJEB>1d&@&em`aX*eBGhCB$M{RZ>Wzci
zOtcV8z1F+%y0!|ux_gVuYi%X?EN%U^^msYY-g@Bm_|gJ!d!m5tZ$u8*0*)3dZDN8|
zoxUZ^0<12%sk=Ms00Rhwd<{H-eLvyQP|AVT^@Sj{U|TEr8wJv^8~x%>p}@}cx1NiC
zB#^54@YYSy8+;hfb6$H#8O(jCztB;f3ChDO#WO~-0J19QbKKg8ptOH(*GFYP)IKkL
zS3*cM`e)(rW7pUs5X4_}SSvIWY>9!@c6ddB#f*_8(NlNO7afbf@zaT@#j7pC29_^@
z%FDIyM+-B6XVz8=8KW2wzHIb^QGFEpdB+MSy7no$Z~VBrmuDfkZdfe){?JP>$27O#
z+7u5uU$i_{&5cF3ZsvBEbj?O9lv2;EO14F-udn2cHV8s(?6l}zb$3y4y~KW3u{)S%
zEY1e@KLP^z?&5JsG`i5aRLNl~5uHw8ug+G_Kwkv$GX`W|qB$|HA6>EhX?jqO7IYvH
z-T%Al#b48Oln!{DZm-BgdoRUB_Vg5@?A5)TSiC^9UkmelSv*BU4(zr4YMzF!(m!h_
zcmD-i=zjSY!ypGelr8oMd_EslAGXy^C@Mw6CzQ$?v&zxU
zmxA>IYs*l>b~~?2>mtB#-YxsVoEX%{{^}^_Tp9Yv;lnKZ2#kLC@h6tYo`DvfXcQ^N
z&bhHSrFRxT=c8+a&3yPJDp0YkUC*7KBdD0Qv+YJs6dm#1=i_2kg38ZV@N)ktN0pRc
zd=&mjM_n`5Z#NyHqc;lqTTai@(G_+1u{nQo(Lo>1Sj(^ZsKKy;H0R4=bSOK$LD;DR
z^%oJc6di(5p~4PlsnIf2bvbWt`Nd-N(dPT>o?^c*R-$9le4Mu&O
zDZ7M$Hu&==$xb7v^{=!?>y}rd2fG~P4lK}7n|kI~i@pNXe)CEnMKJ{Xe{7{PwNw-x
zMk=S`Vi{<@bY;_<&kQsr*z#MT3j;j^_naSyfYIUIh*P^5I%*VLK4Q6p_vTy;OL^&TANB#pDSUBzghvR_)ay%|7<22
z=&{(v5poaBzWn0nQ)-A#~J83$FGA@zF5Ate9h4PKMXW}ZS&5$UKC}#
zJZtjO1xCd#1})iYQGu?zTf(@bQI3vG|4cR3qNBoZo`fl8!RW5AqC76uO6>n0{yj9r
z3QhN8Z`-Fss?f{lUYaC(RidYavYO>47-+YX_}=d&NubT6s*T&x2F+2{EQo-MQOzR(
zHHvHul&fCoPUl+$?Ux*FP1nssb!X%I_^~`%L_Py;*K8~_yN{sggEt#DT%x0qJBr$k
znh?~{y1ec7W&{;4j1;PlM$tcQ+D>BL3^aA6v986b3N*iJby7}!DmoKwUi;_gO|a$Y
zh>js<#xF%v>a8LJO?sriJm@_GJ@uDky8aA`I=obu`_)QE&$Mom7-L7!J9cb7;#j@#
zXed;%S_Yb1Q
z4*l5Q7|8h#=KhCf>6K9Y|6(4+FQkyAU-lpU{Qt;V`b8AKfWrU!!?MrPWBJFz?Ekc5
z*Ub{f8`XP)sDq~N%2|Z_;0_gas1ammLDv6I%S9DCyURr$D$J{
zJ=XlP{Q2)VnE$cI@*n+A`&f!**=O~~LKa`Zkz6ON`T6g-zfg95IFWjO6x~E&1cgf}
zJ013<-T=jqqwqV0J1JaF={r;O1ch!C22gSVidLm)R(p4fwzebtlcf4(b0cZyB@#Q_
zNqlHXVtgivDpb3Nl%1zk`^{84jwQUia0iKJElFfJlUPC7Yu6{~
zh2TD7Ki<7T()%bm??sYcSxaKw%0WU;vFI&teQC$K1MoWM!XasmgR4H5WuJqNaw$ADosLeT3E{3S47&VxYD>-pFc9|LyG
zYJwhoM2_R&OCN&n)0HCH2g;V0;(D{2e-ixd&UFOtbJ#-2s~?{t@WHe!wuQ%lXnz?Y
zml0V)@ONfiBj{t1-GsdMhA?5TIcO!>PSli;f9xa2Ay;>Xj@zk|^MY~wXQoH^ss1ev
z+d^YtL9MJ7M|NW)qW#>i94(x78ssJPn$8Xode^r0VOwhqyna|j^w+cfUlL9$>&_AF
zQ8WaEH59Sf12?Z3$0_IXC)x)z*|mZTXYO)ee%V&*cj-Co$7Ej
z2>8{Eqk@;f101{BgRw0&21d@cZNf2#R>OniI_?B4WMT7jw{?V|$8W~tm>IJeW{06O
zGj5X;h8?!YALD#ZX9FyRje+jsX)F|ufvsyk5ICMxP59jy??{Y8SnLM_mk;Pi6Mn|@
zT);xp7*KyzN64pBJ&5^gO?yZ5pTF@cF^;KLK`baE7{^p7|~Z%Lu(O-GF8yQ97g_t#IkP7S9o{?w4fQOIoU4X%H?^DLo%LZy^=
zz9aT61g2kqPN3Xe2Z2MIj}r6UzVkdm^IjqM;g|7W1TA*4v~$mp^Imm*IC}9p5g|Cfco&;3M?Q*OKFXBtq(o+hh^`
zuRJ4*9f4y2#6BkUcWRLPeaNzappW;HcHhuRe;Rd39K1#%mpO?askqjYLekMx+&x2K
zgD1)7?xgS+8P~GCNcxTyiC?I=?6G5n=>J|Y*?;mi6cDEluc`>gy0TeQSHWKx(x
z;V^~Rq9Ej~`~xEh`W%IB6gE@Hvdh9d)Hoxl{?Ai-;}j-R_`sC3Yewm>rueM>6aV8s
z%WsyPm6zsF?O1Yb5flEhkmWzN`3OF%Jr>LenuRQVZ1WTR>lCI@$jTR4<7D*He~|MXb>vh4qN9$4*JezNA3rOz5S
zs~yWW|fq6l^$3Zo;Uv0J=LokSNefWt#c)4KN86F{c4--T4(ZF0wBK2hPLUWO
z#?zEQ=FbYYq&)l^DZeB`(&=nee_P)Z`eiDl+^LJ?cWfe2>o9qaFO6#?^!BYM&o#FN
zQa_NI2iZijU*UN&53Z%2pT85?o>@uC-Iml7c61cB5~xu4gh15|WWN1xnv`eA1rhw_
z-{sg69s|}BJOp}}RuTM>ucZFtUh@3E%Atxau`!Tm97V`8O9Tmh_Y*M$%>a)H+Oo_a
z+rneu%xgV@zS`hL$hD4|5;RY+E4I|ez@b$GggqXO6oTGwDMI97Tz$(3n(Ht(k+-uo
z2oSV{?i{vt$3R?g7PgefK!(jUL6=VT)Znz6NcRC88O5`N-BIRAqF?)w*F?Vhcf{kT
zzb)m_J&Wh-+2Jc>Xl5Xy|`WmU&%DyeyZyI5}Y64!1&~h
z;qi#KAvi65|Mv$RdF&Ura>eWjJmRJGckO6`!{f`LmOjLK*I7;m&hdYJuh3hELk?c5b{M9C4J@
z?!rRr81Tr*eHw_N;#ZAN?ijv{k?P0&oGwY_z~wPx0ui|W!H#3*__zeezTQ8N;jy`o
z$v8f;d;HrJL-#KgzwrJxMoN$2?e?GIA?l-g4;x}#?min%(2EMo5nP^XDTak^tPVCa
zz*43$aQ|opvCj0A?_x&)RzCtziKD04C!#LfyeBjOr~i!W5Oyr~{3g~B^R!tq&KEy6
zcp0BJw@@+cDrF4BmzsFvd^Z6>EQQ17f6Pk==YNQSWN<$>ls%cm=eytLtvfEi?(gk_
zKc|udNjkWlE7w8^KgYP5iT$IMn@Eh~(*sbif2IAPis+j1%U;H=`SA3Rzh%B-yV?9O>ug5t9BwoojojbFBXg$w!<>J>E1D_fhqk0U+gvsXU_f
z>^GwS;xY2P5?3Yl8?8t><_$@25Flw02XbBxj*#``rU7zZUH}pwYm)X`se7~4R6qZy
z`UiWcM1MI{KEOwv>xU>>l*%Jm*g&1{*S*Ph$EbO6qx4z#j*neQKXp!!dNI_zNmA#0
z36)pWz9;1)lzs$NNB^Mer=DY^oaN^uR$fCP_RtCcS$S0sMNd=6%8Oa~77P2R@gJk~
z`Y6PL6k(5*53=$XL27*4C^;83Z+9qKi5gb`^}LNJ`|On77Yd(J`YgX#{jd<*L`45A
zWaYbM6d&6(gq($}ypx42J_}jxEvR;^{FOCttbCZ&AM2jkj?!n<3#__=RiCitnU%M*
zkfnzm+Jv91elX&H{AT&b@|!h&7N6BGi)QurUwbT?CC3hVq92w$R==$IW>WG8)bqfC
zJ)zHAg+rx|7cE7cyQ?q8It`|tZye*c%U+fCuTHR(r+Jc+FPog~U`G1YFI+Rqx3q+T~A
zSEb~Ksr?#iM9Pao$^9Kf-Qycl`&)pDoatdf_gKf*VL2@FL@s@|R&^
zzR*h2E*~2i-!>ML^tq#?-b(4$gq|`dDc8&*Sh`fDaaEfT})-FusM;k5=5`G-%WW$#17?`sQA@ZV@
z%iRrkW1$_(^RMh!?TsP(
z-97!doWI#e8%H(w@7Q;2VE6mN8ijcN<`i_t8_%bu`?84nIm!LwB9RYqZ?nhqIk+6v
zw8pT);C2h%?&0q)HN2lC=HHrdd8*M0i%S^VX?-4vz)*pX{YDIiCyo{s;Q8=g*-kp1
z=juD5J((DaU6KqA$B>J!>(`b(Y`oE@=kfZ%-DuhEix{RqzVZbhA1~V)w#OL8|IqI}
zilGq7;qHOqsohh3W*F8v?#agcyK!(?02hWen#a!yV5ljxCcp$kw}_lcyq^&#SsqCo
z-=F)jh@tm%Owh5nn7>~w-z8x**TZzpVayNa(}%U*7+>7A=#n(X&(SJ|@_WXBl&{pX
zPK;i^X3?-0(`(}X7k=hFR#)U73omZR#yiN|^RyGI%VJkc9wE0-JEX
zjp8#Ucztx@+R00}-OXMclepccTd^6qe&*@(gZMhsaCx!!DJ_rfsl
z=~M{b4`a$>0p|y@FBVQ<<4o5%AB#Van=b43gkbzf>Dd1jU?>@O=?^|`-{=RUxL-A+
zgBo~0O+b7E_p2v=6R}^GP6!h5Bi%CK3GQdguO+iMGIq2Qe4D)&h`2}R-bc`-x6;*c
zKeHJ(GjV(0{FV}N;Fd}Q5qD-k$Pjg-gNU0JzK)wXX@j`EUAn`bxPA$E6pi-}_ULuu
zxSx6WIsTl^G%LgyVR-q1WFpS*E;>TgS=w2*|KPO0_-P_;NXl#@{JGNbjEIx3g2;10
z_{DF6zUtdR)_o4e1ij6UeBYU^xrx}fi4r>wzu>^4#RNdqVUR
zQ1pe+=Tdq_+W$t@Mb<{7zSic?grD7^!h}Ar?F^x(J6THLF@!uPdR9P0oNv0KO3aT`
zOC_Q2Hf=!YX@4Z^OjYXM%zl_Whx4iTP)n(NBIzV~-j(bk=l4Y*Ij(R4l5a`X17E0o
z#(5)oPCm+`&hagz{_o$UeOe^hZpN8>j}}u;%CnA=`f+mPe2o7j>58433BO~{|0Dda
zaUkvMd?Eb_+d$?4yQsWon0k-&kn%^5LhQ1GXdgk1=eq?-KcMO^K;>`PAw=l;)seWF
zdT*3T)m;!}|23sQ6+-%HeuC5wy-8v!RUgGsdE4*+skd<@IS)(ANWCr{5=Bpw=s@`y
zN%?_2Y{DNFrc?AIN^cp3SP-D*heV7Z=tZjjQ=sTk3X>^h<;_LZb8n*TNm6nNingZc
zUsV0{k-9g2N#&nWlwJyj*e3ZOzghV|tG?r-^stBfKYCbDAm}U#8>x9_wP*RuYR9s7
zj>@}PG)sK3AFO_UQ|sXowH`K7
zc4DZ0j!^p3)IR$6gtYgL(yOE7UX*<-X(9aao+8&rEoCQ#(yONQ0;v4n)P~$A7pXeL
zB7$t+q(tIYKXP9=W|R5sy5lh!A14$^dOU-SXWuOBiE(e_B>9i%4
zr^z^YuPK6%ThFc|P+FAC@BdvQ-y^gf`AYEhWXW^lcYDsiq!yQFFk`ZDjNWKToU*opIK({*p+Tm~nZ
zM7-V~vYE)&6YMGpJMQy82|p||_=t1pYM~8wRe_!V3u2ZyUU>5h+cMZaczh(0*Z*Cg
zf`y7PP;lba5RQsBcUj`+_OVas51g)WNWX)lo+JA_j&PUHQ5<B~MCisR=X0qm+}3{-DfLfD(ou_Nk+T%B-dyj{HPCu|E~
zpU?TT>NrmCtCBFm^-X`*0y|6&`T4~_t)tPnpO_Woq>rJWq`z}M#xFakwfO;-ucv%9FYU+j
z@}mCd`|-T<r
z-LVUkFU0D;y^OW%{rKYf(cCfc_x{lsyl$!QJp2N$4|YHJp89_2I1tQ{9L^2G^6R@3
zi8FysD*tim8c}!}s{`sC~*WvS-{n^9nFg8!j7R%zVSl#g5arU<(
zruS6oq6fhIu7hK33o&~KI#vdr!sPF&p4Dez{VdpS`fwSut9K+-diN0a{F)o2r7->r
zAN5Av!{b2mfJE&4aQje&Du%Oq+6h-M%(VO@im&I?
zWBO6wvFA|x$Ke^it|HikRqHWK@Z>tQ1H)s#n*Io2_&{wzV*4w^SeVe`9c#TVJ@Z?OHDo-oG5^!6#9Gn~ch
z!y27Ce0V(vJ&Vi1>pTbDF+qI)e$Diq#`krg&^zLNoAnc}jWHOVQstg?4MX7Xoq+G_
z+3k-faJ}Sh<8yesbt8d``!KZX5AC+VaQDznX;%#Wgd|;Yy&oLQ{^I-FQ9_|I0K*FY
zy`hE}F7e*TxPam1#VK7}kLjI4)E7J!rv&@4@iy|fc;oxabk9NT3J!};VI{J7J4Ly3
z-|+o?Rw|TechKkyins5MsKTxuvFHEOortR^yxfTRefGd%A`UNSO1#45x$7Pi_c!Lx
zN#v(zVpo-zpDJG6I2K*-y@=xka=it|-N$Yc_c3ltWpOy&t^cI~A74((qp$e5Z03B3
z`b;1DHYc2Jj_#Pn_3RBYL-6|pN%IXc_S>u86U^P=$PN{D_%D%;y;uBUjwd
zLc^6$aCxid6)fe%?g3isns7X<3S(E$*nQA0@*Zwa%mm?wmKK>uY~~?x=q{O;2qR*I
zJZGA$1K}rRU9@tByl*r3Ox6Jm^d;fn9qOF6OCt3@7LfVGTR`RwdmjBD>bOVz--vN4
zy(jY#esA)9ovO1m!4E6;B;M0~IlYdU=k2_Pg#8m6KNIuvb(JikH?WS(U%q`J{nn@M
zb&uU3^N@4uBt91+^M^Z5WZo3UOY$RLkSK3J;*LP_IX0_P^PxiOOAL^-$y-vdM0FL>
z|Mqgy|If*!`~~&-x3Cpt-WFu_g3wF#W+Tv)Mty#9DOry)srkNooU}Vk)kWB)ATjUK
zRNnfIq7SK({CMhfBi@w$aw=bC<=e}DlX^AO=Mn{}&*?p<^5$78-_A@V{X7{;>N`=N
z>+_}V(Wj{UYrpH{I7haV;|zF3>I>{8afq_Z%I~q{h+J2c-(FOIvlQQ(!fpy#?;Er!
zx<7zyFGaOy)nC}6Anb0W`V&_qX%A{1#;LrVgL==%El$dRo+kZnFem9s%C013hgXi|
zdj*i|S<{T9b*Q@1m3m*8aDrS9`^w3ClT51pC91Afq4a&XlI`C|k^7{M^3yhvCIFoq2cr2!u~HX1=TZiVb3`>|=oNlGSYDVR
z-v6K2Mc&`NHftm9r^4fRTH*3e$8><_zfa-Lw|M*WcK7{oWctivr8ai{sFn!fwEX{Z
zb>;C`J>TC}BrQUltt_b&B1zIMku{2thzg}7Bq@7BiL7NwsSpYYA!I8{rBWmm**%uB
zq!KFMbLKvuKYqV|yk9eO?wNDvnS0KA&dj}7T-@<|hTEbp;@5Uc=~Q98XlEgbXA622
zW?{TpY9_ye#4EqFbS#fNJO)c!&?8^ygyr)#StZEZgZ;T=k)Pzr%fk{H)*ri9U%>WJ
z!KTOP{|nOPnvl1r%8wn^2K*u%D~b7c{-*j#Y`^6DClvet@#*MsF
zVEFU276m{9*&keKfbE99?0A-BTLzTbfAx8M0=a{8#cm(Sf;gD<>VyR7xx)79zCE${2YJ)iHR^7_ug3$deq{k4I|@f
z`LtbK>+>jsl4{NM-2is@-PtIF>#dPj2c=}7KjkxWfqZcPN_zynKLVc=eShHhEU@dM
z<(3;t!|;0=ri~m0do}0H?A3+)5w|j1gS=nS;@!c+^}Uo$#H=xHKNvr;Z!3QM06x>T
z;8_-o(~^%n**5@RZx#*9I5yZ#oiECmoHgA;aXBfT-uD9R|04Q*#ViK(+i$x?N;TZ~
z>WV`-FS;ln%c6{+N8J=>qmQJAV;>bF-9H{H!JziMFX~l+>)P%Q%T<9r4mX+}d*JV7
z{Ou{u^XvB$TyT9Tx=3*Z_ho*&g%wExI-bgLHU~6zj`G5JbL(QwO608+|B+x^f1Y-BTVGHxjhdlqSt0G%re!SAZasua%Ky^o45v=!Q
zXX{8K|IOXIq8OjInrB6u5p|_opW|~={W$dtKDTx1_uishOIX4f^5%KrxhwHGs4}>9
znLPikRcjGt>_<;51O3rnj{IRzW_Q{_#z9#qb1{yOY*|fb%~Vfr4}+g;?+BMb6u
z)QNNGF9HWPRP6z*IIyVjCSXa^ZGmsFzO{ZU9D5te@iJU@W&jh@HD@}y5
zt{7Vv>UEr$vjpvZCA>Ef{YFW`r+UP$BYwosUUgXxM@RZ7_exQ+9&x+ZPVy~U+MCIK
z##yDb_b9KJl?|^X!1~$2%n`@oP)w#g>ZOA839sEeqDk@|C2Mk*-G%!3+wI6cV)G3<
zR+Qh6t>1$7{N$`7`&Tl3dVG|7tuSMlrDp`2z#Er}1hht`w&^o2PK6kLe>beTNi^X{56@o4vgOVh3>-z0nfBVzx&E|N#`ojF9}*MUw$
zdf(Mkq6NoM`qifaluJyUL*z93G9aZ2>+S6##14fk`~=TmsU!WlOVDw>FP`SZhF|K4
zJw;4@8FiI@Z)c!^<|84&H2-|j^p1=}Tl;CEXHc5vAGXZ*?B21_`Maj)oHf}0spQ8b
zE|;$TLi9I3qxnP6l1y^_*zh3|Prn2#A?p)02K>8BxK5{(W?IOWT=}OP8OmqS1(Xj)HFsT~3dqdHcIGIzO8pN!L#Z(`d%z@e6j?
z6FVGjH)e{4j)ww#tzt)%s8-J>xyjLrwoe<(-f^eSl{IVX+o@BMB1cR4A*L;4$)`$>3U
zd^nx=m6}JNr_8K4M%v%Iyd=+GbJJ2dDg<$~C7tM%l%&_!k~<4&x#z=#Zw6Rol67UU
z?tC)e_Ocj~c|Ce6o9G#zbAjX;49+b6iu#Na?~?i__LD>p!!w0k7bQ-g1GjHfVxU|t
z%Qm9-b5IbO_ncXOh~2H#5(I1OhxVcW>erGaavhsyljo$y^KKkouXlT5^wib6hGB&nx|{7%rj)K~~`D7E1q#@EB*7YGk5
zuFr3BfwF0>!9mP-2`+P9e;BZg^S%eh=Q#a@T;!AbJ>4V^?6h>sFcZo$G5dX0Al~On
zZY{_27=O%LiroQ~GmiM;bw!nx8<78bD|F?Wft-j$*a9mk+om}$KM$C7?sNh20)tog
zY{dZAo|^Uf55&EFnR_1_LfNR{%ezm2tp!itJ%G5TEMFHT0D1&vI+p$Hg?x7VowH@o
zUMlV{KMJ_B;G>KL$ldYMI&2B?U-nT!_zjSI~8J)hs7bJstWq2WvSztPU&)8egCnDGt5D9r)_Ri&f
zps%R*&GG=~x9sYxKT@DKe#lcA`!#CLm~jDKA^iFAC>O}(98U4;r?U9>qm`%Pp~zP-2^#72kxy&l@OP7SQ^i8F(@CY_G0WZ4sW^FLPFoU{#q==107Ut{X>eI%n`YFc`qk(Ilz%X+E!*)r{i-Y}?^-`OhG$3`i3g{mCIrtvwn5QPWy9>SpYWsJvqMOn?
zXH?(f0D00zJH8mO!!Bk=yBmx@r>axUVWloAnxz!3Gk1N?`u^#3_EW=rE0nC%cddHG7$lDh^6W^wP
zf8iBGeS-<%b(wU)R7sL1B-Fo>X}c_4VQY4ZM$G-D)Us
z>&5~GMQP})IbRC=w(^;cG^~HZ6aIAX8}FjFbyW_Wi|nPAYMEvZN;9Z9RhM`B-+&*=
z(EUA8)<)^SWI@cX94`s(^{qDa&569V=`|(hw
z6g3jiF9j5D(Flb4(bH^C@4_7J?gy~D|Q|I
zrRo_kiEzNQr@<#sj%WG^c^~CRBfBB``Kkl#c8CjZZ<7lId{AYrfqu%vbl+*@zeA}l
zr;(q!9J{WH{#<QkA)zANGBD$&zA+PnWUyzCN_NVn94};u*>g)_v
zKwAZ$y3J63*;aW!`pvKhCoQi4?#g}vM{i+VE&hxms+iY5?*Q7twwl&_G6mAjJupC@%yvE62i9!?3NO_-rR8#&$)MN
z((&C;o}lPCrB>|cw)!B+&oe?=iQR{uJtutO{3AMFdm?zL8OmK0c?d@bz<;Z*Ci-|K
zEDXGg4oY9Nb{k1
zCSSA}p?TMO@f)N)w6c(()l-_!J>;d^n@-YvOXK%9V$b{)`-z-K=p?}<6Lg;aG8?_0
zzfyfA;o;v4g-E+g3(e;pr|9!aGUiuF{fHs`J%^do^tr3X2~kpSZBM^v!)HSCW>e;S
zgW~nH{RzKme)nnxoxd!xn@{eOcN@)1eYk1fY0pK=g|kc%y-8d&KU{1>w~H{pBXBQ(
zmWz!2OXPYk((8&{d^8^&XP$ov{d&6Ci7wJWfkFf_4A}i`=W}61S_xj5xMo}ED7(dabzdI
zds68`qIc%|cOuudh+gjmJu@V7i@C=da2~nW50m+`&XYbL$5~1dc~1@>(%(j*PopT8
zrfx#yHkV|RdFelDL&|yHSIIno7=E6-k0vUZO6FVcV?L5cZ0tT3jn^L#i@S`^QTHCn
z16Y=r3I2k3Vy+(PFL*0&5|+*92N2#V{gB5Q{nEJU_Zh4&*w;?-l&yFA$+?Dx=Zid0
zUx%gY4En=<%g^Mzfc3_7f0Y09F`w|&2mK}F`r_t&&dC4XUT`PnbG(e
zNE!WW^AgYRswOyxFBxYz4011fwD^R7B}?w=k7X>
z>jN4n4prR-blGuS<0r^%-1N@(HI&1P;%7GjMy^mfX9RKRL)L}y`w-8nHsrFtfjD)e
zw(hP0^lw&U^v@6E9IIuY#e&?Qzs38W1OHPQcHgcJ@$uLl*SJBDtN&H0-wEyidbqFS
z_%y~F=DLEu3bD-_xT_$}9Em>I1abmf=8YZ&Iqv|AbDB`jT`}=I4&vRi@7?Mz;5#K^dq5kgP>Tgz{KXsrpg&*1<=~MGlK(A+x<;k_z!9UkbKEb@|rrF=`$J7AN=-oIG
z1Lp>!{s`K%1J6pfTX-zRxR#Fd42Bmj;YoCe+#PgpM
zSKnpyQj6z(Uno!vJmitY4L#_OgORNyIt}?u@$`brtAO7Oio9B~lRsL?z>>GZA?MIU=gfu9=gn19nnc$tC?-e
z!O}xXkKA4L5`Ooggx9jO=Q5~*>m!TK0pBWpYW8zSMhA6OnTz#UaTn!Z)$hWu4ZqV_
zN9VfL3@UXb%Z1_@rdme*s;G#=m7A}y4^*KO<-Q!
z-fWuW0P>Ho6!$%R4)ez9&t9Bo5of5{hol46j*E!Mgx
zQ3^b^ruN2A1?W4hxM$sNu(zSLz2Q3W*sv+-?*%abeqPKrmIO?SP)Vmi&-#A@oOrJ1
zFC&+_3}|5aZ}C;Y0_h#ahd^$plt6(Vl==M*JVO7o-&%J44M2~RqdKNwZ_u}Bb$yUC
zEn(S;JZXXvxkwlAQ283}`+$zTB4v(XXPb(x+bob{W&iZ+0o0#h{HenHKSQ;7FbnYb
zwfj5xfS)g*ZUH#FB7LzrZK;qRP{ju{ozpi4^^~t%T`TYv}e(JYr<#ViWL&RV&LUQ
z&IwGsxJh_YQpb+d*uO%AG|6jluYB?W?X2^@L(Ufrv&N7(^rkH!8SRGuseTCC?_Hzc
z^R5Yxi^1|E#oI)Gxy~;uluP@#mJ{ne&
z;lvMY*Ve4Mg=l)ch{Q3Ei3Xy#+Vdfa&wdYNNSqtYIzYi{3!@1N^V8=p=e8~-{RQ`KChLL3;sVmHl5>e*s&g2zJEggc@TSrFX2N5{
zh0TawKJHh9zZeXPld^(I6&Z&?aZ4_oe7W0RcPrP_IjN>(8
zyo2a9lJz3Lo7>JLQs%Q8A^m@|xv~xQ9eVSY*fIA;2${En3k3;}84%kjfbDZTcOJs$
zE>CQn@TD7JR)oKe%dwDoEqC~#3(C2NxCUUq^K3`vVR<)C^*$_XpHA1q@{{GK*5Gp!
zZJzrW&qY0c6!#a~9rCnY5kF5IhF6MVUCg#93*~luy$!@X9527%9IRhzy2l*l4h~(1
zPd0%c`Rhvd9U>xlFQebtm_Fwww!d2JlZNwuo7Bdq_1cWw4(lwcC;Se0xvVSm^^UH5rmbJyN>C)@Fd4JLc{|{-QEGmt!|z_3x8xxZW-H
zgm0Ka9AmeWtw4Tdz*SL+e5kuscP-i@#c^2>;|qUf{YQO}d+2$87W$ptZ~Q;I0!m69
zbwIxKFso5a4bc3mUcwK+Y{3OH*zanV44q2A;Z0vZ)q@{34&`C11HYQPtf4v-_)wB<
zp8Qj2UzaqXUIXY{%wymSa-$h8ho&IT)maXGAA`8#AQC5x`;}JXtWA*+pQOEcFZeWmr~8k>`gtYqx=hex-Qwf=0^*^P(n-gEtZ?3vwKYu(cvH{n0=5%-
zdZ~!-mrbHdVf~%6sWNyR;@;kqhmz(osJcv^kjAT^?}dQw$=~0o{ZBS))@^dAA_Kj8v5dYUuIA}eodi#U87Wkj&jzuLttm6@EXOhurHbP=CNT>FQvsN
z)343=N?ko`bMadK7-hAfYb-|;*5QR)W*>t**u9$vLww5zsN{1JU8Qrzs3zWF6&>|a
zYB+s`pv&K0>W=;j>vKx|RB6BIhri{+)Jrb*x+%+1s`KFFlY~`cROWIQo?8bysK-l`
z74(3=_zV1wp7XhvGQAM7dQ=7SYA#DSI#?Lgac@s2YGntNCB~ChuHH$B+#KF~ZK8+D
zi{Do8Mv+0CoPU&2Zor_bA7m$usI*hLP6AV|G9A<}x%*-&WnI+F@aMD(RlSsMYGEDn
z=~WZfBJbXOq1G%ieEM@iJJovE=il$94!G|lVciS6sh2C3Mo#bOp)L-nd3%2TLRlVN
zw^?-P3uUh_DSKI?gIZm*!}Y*mCw2X)iNpBkE^5wi!4}Q|i2uvZoU;S|vV76I3JQ3O
z=*@N0l`wCdwZc9`1Akefl%OtY%Aok0b8?5__kC`f*!>Cii_7;Gx_kpZm!917I-&`l
ze^IVo62RXgpX)pjf&J~#OR;H=Z{YbiA6UE#?BB{0YO`
zz9J%T11oN@R_Wb5SDC@-3B!$BB)x1S)b1~dc
z%bU3svM|3ka#;7T2077@Bqcm&+b@>Jst36Ch19#NkZ<-$yy}4E&C&I@-hiGF;djya
zJh$y!s)T;zU*P@#J*fA3+8Q+s&qd(SY{E+x8kaizLw(pG+p8x*|K0{)L%|~8uVN)t
zn8)%w^v6gDp5OJaEO++8bGP8@ku@(tU%5-{Gvql-e~wxsA56+yWQTq!Y+N%6{a{G(
z@-Fn7=kwbZArISgM!p``QN{Y}cA!7hS6iEb?e>=%C(z$3sJ&T<{&4-6uOm0A5<8tb~4V^6(lS^qcc|c-qj<9`tz?Fy8``kC!IaFh=6kL&D7JoD>TB|PV%|C(PI5A5G;)xvhi@LAsxpC?;I
zBPu*dA^E)*qIQ4LKSx`=JcGRRyr?C7au)Jw0Shjm+_I8fDa7k%hx#yYIQN_OYV_ax
z=OqvxC~T-_jPi?Z4UVAwdo$_x|Ki^>2+zq0t{`!G{DL%zzxfO4{7DY)Y!c5K%pVi}
zCeQMeOV-~+dsQak4NkvDQjla{AJ(DNMe@}
zD|3?mJe24>
z*reNMqEBfRM?K0(&uJt4MlR2sj9*D)9pMX4Wx5aH?`CoMsV~ay_%`wp@sa_Z7u1V0
zBIh6`^u-9z`5as5g!Ab!%ilhfQ(U!&?6=0uHALmiPXR~o^51sFO8}XqG&HImL-6QLcr3Ojgaei8=h#
z8g8Z)j=!IwXut=2z9qcu|Lz4G$T1}Qt4;QQDo}oXL;HTT=hDSWa*l2D+M;^~P}Z8_
z$u0zZuySVs`iYqN-)(W-WF?`v68k@#u_kRFpib|%bsX@#hIB1CiE=xgG>oyo#mAEZ
zaU4#nZ*kF
z6|2@pJqDb!=zjS<@OzbYL9-3PzvZiqhuA_K(Kwr*hU-a|qZ3BHAUCliYTrAM+xqVN
z#@`@kmbBmh1NcD|iw)eiAm?VdPCyRireJ1l1<9xn4z3?;UJ$=>Yl^xiAs>;q>R)*r$o2Z2IQt#szMXoJ5CL+U3&-u>
zVw_v!rsNIr?0lH|eLIK;KF2n&Ye2k8ZLai=m;&DS$~5bo2kbkx)V=bCI9#T2TG|`p
zjH-IA_bP~g>&-T|`G8!vY*UEzF~}F#NiQwl55G%PM`{z~SF9fTdzeGNHAZrgcd^els#PGtlrKLdZ5T(v&{?r-%?>P
zQ@4-!Y_o)Xgzv%M|MY<0b$;mUvje^(+P`3vj~m1%4YrDrat8IM@xVOOdtKDUx5mQP
z9D1q8*RM2B!22Rn={K$E^3iwX0Gk$9@uw~@><8qpoV*|
z4Mzh1yTP5m_#4>u>$mP%+ee+0fb86$zjJ%2h|Qi)RAF2b)`{M~{Ri@i+FCh9z$9vMaF>#zN?%%tqs3}f=EDeP&urdO<$eg2lxA+U{Qh({9Y_$
zd*!Snily*7s>J(q^X9@lTySB39Pke*qYmEPFyCU^4h}WJd^%UkXV3)WU%BzPu;?Rr
zeq1l79f0{0z0lHb1Kj8Q{YEL~pwDskk_0zE);p?uI}72x8=EB8C6FguUEO=y1N7-<
zmka5Gp6;XS>ncF+KD(0eFU9a4O6;teXt;kBc8_s?khA5iSBn6-zKk1FrXXj{{m&;H
z=ATinV$%%RYgcRCvI~&&w%LFiV1vWmA>8LWGvhp`4KSd^URnXp}ULZ_*v~
z589(cT{yqjt@Q~s1O4|8Z1LX&aGY${1qn%gv-sRi^
z9DF8o1mnBm&r=!b=PnAfRiJ<29MjQ6zqI|)bq};tCuvCe3gGgX>sQflz7q0!iuJm?
zY-S-J;anImfMr%8Wdqb3`_%ghmS1n*ZHRs--{W;N%E|Ic{J?U$x3*Cdpj+Yc2UyP`
zQV@rJXg$}YF#5ULLql8e`YLU9`fHgL04CC=)FIc;>kSZfmfs4Yr4lDz+g`f6624^z6hK!avvp
zxinBt`^Yj=Y@Ztt+JU(0fXXez^RaOx{=WC3^^fNordrrKU$XRUjdWZVXXCMsJBMLAqFw&+YxeqgC)uEsszNBa}%AI
zALU76Va4(emiX%^mzuba@P?L?TQ^|+iiev2@Ok->xMu|0bI!KkMZB*xpZrd5mX^V%
zgW-J0h%))^$hoKCghza2yGHKss;F2N%17jJllN0)Un}pz--|tdbrr@-m#1~j7%x4p
zeEo^@FN$~gFviO*1Kw&|;rW`aE0&G<5O32?IXpKM6KL`s@2_dgrg+S!G|dUGum-s<
z4eNEt*R5v#n0*FN%&nj7FK62bdmMx^d#vcDV36xAoVtenI;@TFz<$}|pAKQaJ1$1z?W
z)7n#w@$wei(vcV7f4At}9I}LX=^)zt?J3A*_t-g?gWTf1>U`}WCoo5M=r8!|&G!Yq
zcS8JV(mk>23&l^D)&Z$M;%W
zH|DBgzr89W{9PdDSfIXR6ZGqJ*z)~ykTXiCgimtA_*K>25&-Pcvivy%a%D+#BOPGf
z=$U2^s0{H^*KAbR9qhTLoqvFWc=^pIt!WbC~S`uxJ0#(go6
z|32B_7GE_$2^?t?r{KO_7iL)81U-zY+?^GmH|jvBhMX6~-z|y!Zc+VIpZ1Sgvw>%%
zv`rskfw;vbv2-If0{cle>5rAbzHDx@aBWqH=U4czDctL%Ec{P*|48em*0G+PV;Bzk
zboYq-Oo*4!EKNqXV2ATx#*&h=oz(j|KFQ9u-Bi=9k5PBe!0&J*ZlU}l1{KN{|Gopp
zA^5aD>m8m>ilfwQ{OLM4htu*od95eN@of7s2W89<~01--~i`_l5ZQ=tD?atV$=9XmCM~WknBletGDL3y=@!5f~8O
z2jiW))#QgX?2nCAm|yBBhV@&a`@8!PFZT@VmU%*AOZ6tcmCISC5Y$I
zf=|DCfgGc2DJLh$ozyG8lMHdc?%YI>8k|G(<`5aDKyHa@*}pq*pN_Xa%i(@w%xTZ~
zdB7uXT68_Bh4V$Z5u65U(67v^m5cPDy>FWZ$0g9~a`Eh@e$ac`GO&vs=Ap8!k~7B3
z;8rmUjF(PoZN?batk(pnVLr|DYSH^~D1T;b$Te@vMlga=u}54+4p%D-kXhb`(Lq(q1-J8wq>^gow5=SqyKqd{N@+-D<|PNhVfmpfrn6L`JB!bBdFJCc|UpTL8C6L)T1Y?pE!Fh#%o+(eed%al)U
zBwpTUTqp5TMHtJ-liM{L;g=8?i-1m8>7nIIJ$ga=ui^G}>cdqN<5yJ;NkTYl`prV(R=HV`8|oXkcO&_e8KK%w*dFftgv7C1?3U$N4t>5v810{Kj!b%+uEF*47V#L2&0#RMfi>F-7ylAI=SN4w(|(f75Ne(!C~)>tAp
za@mZ;Ra3s}Wc}`QE1txs+E4U;$Vx%_`!zoI=yNB_rRewD?nG`P^3BY1^K!H3{gnK+
zFGO$M`F8{v?6h5P#_4k;pY-VWAs;IBl6JX*e1dKbe+e&$jh<}8`{GMqK=f~spGA0;
z*vBEVejP~iSb*bJe7%GR_}V%ft&>`>9Azq4d*l25LC>Y`ALdb9OwsuACO^&|C^zZcOr
z(y@-%dHUikVn^kOay!b6o%wBn@zUzC&n?v7q!Fx$^Da{_Tg?b?3uD#lI>51A_agB3
zEk2keg=lEG;Uvy~(}US_@O;IwAhmlsfbqj7WdBCaui6jwEBfCcJmqwHI^kbS?%(Xe
zb$+~<6tQdY#}JvnK1~umC|`PY=qBnHH2C9)_Dq|6uEKi8%H3xX`*vl^Vtrs~&~+T&
z_NMq!)F+U)$qL_}sX38(7N1*@%-0!sF7aBZ{=&x)pV}wX_-i0u<~Qu*!#tmrSCYjO
zz<)m5mSMj0@#doC;Se{T6u4jagZ&C_JMn%mXdjr
zRcyk)gM8AO^Kxn+SG(flAEj>~_jB5K3gWUwpY7D|q+V)&;h+6@E})|GQ)xep!}Ns>
z4NoDTeE32BvAij;BS0_9+lWCGUV9*^2zHp)@Yo)N`%6A`#+(~?hKjXG&ujP&#ZUIu
z#r?3Z&paKl>n`*ge?mpF81!1mo)ew{y^2q--WdeBu=3ln8?s>k#CB1v%ZGmIh1kz)
zi&MW+*^9-Nvb`FnHZRLH?HW2m`K3jfC4d`G{tJ3cdeB~XX
zw)@MhU^op^ydJMMDmH(kikp|MS`ELSR>!9MlC$CaAe@<}SO@wk3C{kk(y9?Els`>V
z;rv%BU+kb$4V-sAW?bJ{xtu{Q9hx1Xz8B)B+{=o9wO=WjV@|VA36D@J`7L9SU2xvw
z-@8m_Ne1;!c=M%bBmz>7+haIf=aN
z>!7$^3YizHc2en=O~>57bW#Qn*%x|ycTzV#>K6)Fby9I#RvwdO?WC-p^rZ}zc2e{G
zJMZMCL4NHsk
z7jyFlDey|LQ(?sr(P
zJ~FIRh4UYbEDog(IKSMmgQLfs>ZR%)&%E3Z=Zr(#hn43+K0#7Pa$YFpdxTDD7hO&3
zr946}1g63BGB?40t0T<6Rl79iiNHMI4z_G9fblIb{?Hi#`zhNNu6q;l6V|uSbh?6p
zPrdph`~ctQY&sw_uMy_YOuK61UD!uC8DReD9q^YcH8WOCvgm(
zw~r6DvF?HSCHS}^Ck^KD`AkFWKDduwo;3Xx(4MQvHlzddx9Yrd*t&8!H=-pNkpS~}
z!Z<%?8lG>9)tbLtK{M?mQZcFjAI^-q6J|=3|PokFKPek!EAIN&x`@|S|
z-%wLibqm@py!Z1LY_IropYWT`JPDGIzqIiH;T_vHZ6CmRHnWsu5j1e0^b>?%#9y`Hg~w7bm-
zBXQo+cQYy98wyzf63f&x5py(C3m(
zI)h0+izVsvBo8O)_t8Yz>2t^FfqkUkJ*BAxIrQoFea!dqC{eo1AxK|uT|u7_MayFT|Ew|>+MZnI_XH=I@57yCewQPMgVrbgl0FyH^5-D2GoI}xxnHSO
ziwS?J_j^yq)85+jCyw__mI<-b!+tiof6Z6VNqK#Mr@6?=7UyPj68#Bkd=)
z)AG9?(7dm1tuEo8o7-zi`>&1kxtcd~=C%p5a4KLySSB1BJL^}mzSS4{^
zVeq9f=B#he6UP)_6J1Cp1EJR+Ean}j*(W%DnXYj=XP`s$g|2
z;JMMPb!>ojts7#vAYNQ+G=718SLvT|_&LBs{n8o8uh(;LSzH3RGpi(~9&qAl3l|TZ
zXIKy^|I2Y0qP?NUj*AwDyZ6YeAmk*fT}f+^92q7L4y-zQ`^6l_WJt)n3}VsU3JrifG#QQCSA$Q6B1ig}jx|(*B!A7BZ+=
zQS4t*_6|^6GdSj{!uw&q5fVOH7Vw@}zICuQXE&uR`B~&8>{AW6g-Mm1=%QF!)x$HY
zyD1q#+o$&;yD5{96gk#EUDP4>n_oAjcTt8}cAFC0yQnX$Uss68!+U1;Oycgb^-yDr
zKU8v0cT>|7AHHj|^iuiC{(T!tdnpr+u(C?=Uico{&7h~;y;SgAk2>jzUTVRy9{YT-
zJ0$zv^Y_62;@(_#V$12J5*IphrabMXT(>^oC-4N~oA}q0(gF<1a)Z*lB#6&DVU*-CW*9wSQVF?+E9lZ*yE;
z;Nk@9!Oc;!jvNe%L-g(DAmDHB3x-?2J?W*^M)UukOoZoXo`?J0UyvW=ofRC}2>YFC
zi}uX{?1&$aKLY2bH#+aK%>_GdjT>!HgL$%m&8+Mq%&!m|YX^OpPs;uCUbeyYMmm;^
z!|=R5bNvxH09bQ&ofDqZ4x6J;F%0tSMI2h&;Pn;_uYAHYoF+7akIU_S1jQ9?XGy&Fb!{#Zcd-v46o6
zKu(>Pf%trTh?we#0ZwtIN#gr$P3n(quzk`;ao1TuT{)I4Il#rA+}E_hIKJG+&5e1n
zl{+)4#o#&eNX{=ny-O}F@^bOH$G2#w<9a@0`5PhhKi{vls$*VE
zwC}+y^uI!$OI0y`u}$;;lLg#n!01H(`sCl22K4(SVouV?iK`ZWBgsCK%=A2K^#7YOD<-ksgLiT>_G8MSLh@uP0uK|h
zU18h4kBF8x-lbvxg=;EIQ0|WYoU>T}*~YmU@yGsSBz}rIX7^xw%FV@#vA%A}D2dzO
zmVG4gvQoh#7VF=({ZK?ZmlqhHz_NX35Q)z|TyqKEJ6^#;;`>><&sVYC$5x&2)-2<8
z5`S-m(Erb%Ak~+|dEYm7#12P3I)BCG>qO*Qw{sA?Ox5T(9&kR4=)a~ni}2in7n!8}
z@-qp-1HwY|3EKP0CS$+%*jb2vzCn7QOVHYlv`f8xM)<&Xk8HwYQ`4>zdz++%Nqa=&
zHq!3+faaUg4|b4tcd-kkeexj9_fFlQc~p~QDXE`(Kc3iqcr(3UcDkL;*KKm5^LKSB
z^m|~lJ89mk4!Y
zoo)|fwl8DKFJkH|VtyyWjQM_`;4po@=(Y=ej@!|S{=V*kh)~jxMkEKp1Ij@!o-aPa|6c)Rbbx1>Fy
zmHvLCgh)0ir{B0h^e_C*OZ3W^43m1J3-tc*+;sYVK3xtv&ulN~M&!m$4-vi6;g8Rv
z9hFO#ko|wvtYg~fj}ix0av@K>@9*4#?PoZzM<9-bJt61Wy(BA1UY(I9`V{4UtTQ9}
z^u2bH^~;MKaq^rp3}+L0*RnHD(61aypx?(-YPdtj?RUHs$=5&dua80f*1mds5T6u|
z5j%JPrPqxT5d-8tY3pO
zDd;yl*34!_dCjOSmz{u3=7RIEe~DxV@_+O=tuJmsf2iwgl!E$ICzK|90Bx>(dWCWY
zoiSwmaz!qac@na4tQ7U9a80emIAC1Ha;6LOFTcZ&=0F^;Ju$Ta&zo?vJ^!Zzyq2{^
zI?@OD%ys_q1DPN{yWqSB#@QuaEH;>*)jIfm%}a=%uhYIrp99`oVKm080`=Xy4^wvl
zUDeZSaUWSD`}Yg{E)|1E>7q8oaouY>N`s+3pmevRHlUsHF&VVyeTzjIo^#(8T(D9d
z@c45*9=v}Z&jm@B0duujHsH8;tE!yE@p6`U9E0b;1lG-!7lSzcX7F7}E%2!+VVPIr
zU{6^lcdSM@jI+h=O>%Ghsn3zmFK+?4!<(nfUjy%uv6_6B2fXL>)=(*-JosIfcF3*=
z`^v??`7uEM*+s1lg&?2RwOB|S#%EJfq~RMl*L&lqd;CqA8`~%2?#WK7a*MVGSy0Mcv=sVW2xpD{ap?;55hKs>|^Izxk
z;~)^ff6`CoEPHIA6Vng*ZB3Ey{S1mHwEN>F9|mD(4!MV+4
z&sWbAWKc~p_m(Kax$&dTcSIFM8B}|b=L0wRj#$qlkFcX)=SRK}PZ`+HF4o*@7+C@9
zDKly79N0(Y^q(&C1)dk-caHZd>@)j)nDwj))<2!j31!`2w+pB9R4MSfz~nb&DqiVA7q=t`m4-VRZA4?wBO9e
z$_{pvZFrw?8^&Albgh~e{N6k&2a|Mw|H{WJIz~YI$IG>?cwRwzYRT6bn8$YR
zaWr^RmkRbPvziZSjl%ky)&8Rk*c%_Cvd;kanZHEytLTHB7k@YiZvnggxUchd!S$>c
zn}^(?zq1mN<@nwohQz$$4-eUT)rsApJD9H(a^(u2
zpq?xGfDtcT-zvEC0Y1NOMl%-{0VYyTR~`YrnS7Il&ue|7xn&;cmo3SXM*g23a^pz~
zU_v0fD*6Ym^)n&39%OLw)b26H}F@G!NC;tS;`znsgUJ`Ox?D&**R8+Gfu;0TeS>^AO9r
z$-YZp0fx@)_eUOVoSa7TF+rwFZk&O#O<8CF`k5y^GR>Gbp<0I=G45>gd*+PoE``gb
z(Vset`Ea8jmghhC7US5q=`1Vc{cUG2zrgnMk$L3*sFqJ?W?(;7v!`n@@6o-$MGyV8
zg^MRpAILXP&0m7(ZXZ8`TY@CnnSu(FkUW&`^@;Q`i?^-qo@piTh`M)^gN2^J^lwv)9AM11cTs|QN
zkLwMiotlAteOP9Z-hBmee0gj=;@pEaBtPbpvS}@@OTzXmW@7!tLgNs`8vVP?h?WgR5ygVjlgp>;o(fjC5nI6P$@wIdw*)NxVzp(BleO<&In*Vieq4|3%
z&m5xvTk;2@$2OPdiF(Zc1IHm#PTIpYXgM7r`uw{4CYtvLzoz-6e<=N3jWx{gQ3ub^
zJhtfst?z&>y&fuCR7Ls`V}3v5yDr_%UrWnN_0#%flj!~YhfnDCL;ZBS0h0%pG4tIQ
znS6Yn4c(q`fxbSC$%D((=`t){Nx$RvG&=sr@0t1V8Rqxy;+g9X7(^lulZGG+0>SId
zbpVh=Qiel%1YuGToM7Tj=6V8tG*SJ%ug_qyPVDJsei!<`|9kq0
zDbE%_kE;)}|8LB3q?qHV$=n~X8>02FGVfG_!_akF-mH~g=d9dAe@BF;m!9vnVd_L)=rL_)bTmC*xF>I1ojyvQ)A8d8@RLJcC+U16VpIr&
z$h95GA()r3mgse?H74?Z8~w=ogiC?8w^oW?CuKF|5qZ8}`9xkVlAYKUc7tATjXTlL
zle7z6jxS9o^5%)tM6SNvs}lPU5r{BATzg!?8U6iq+nj2QPi8N@-7$_d^gSYWZW}TU
zLb==b^+`M}kVyE5?Oh|=Q&2wRTLQ5&&+rAwOFaA`NAmp6XU`J(EoqEo)L-(8mGr+w
z{Roj4+HOYjKI4@)NWLZR%7WwAkBU3}f8L*G=@Yx>%vh7_KFFyNdAG`5U6gxWqs4a+
zaD%jURS4k6LoezvZc1PB41N!2WgK9S`y;o*vdDLA_ikLj3ge)X?_D2^bM9F^J3;|H
z6tdYX0T*|yO2oLu^I4_>+nuB($a$$4=dn~=N4$TfPxQ@k6e%P3ALGx!`*Iu8pF(>S
z)FK!d&lYg{k@Im*3mfiY{q{aj8N}9rPh|ZiRTU70{cv@ctD(KaFC0R60bix+F69Lt
z6Ides_zm!q<@rUcqye?Z9v$HVexKpwe(^HM#cODfr~z`WKW?55JZ1eD!&DdK!nMb~
z%0v0))rY6h{*bA($g2d(C4d
zUgws;I*Il=t=_wn8!$@jfHd-n*(wj#qMn9Vsr&|jneJbg;CP0~hyDr#yq$9~H3WEK
zV@>29Rp5Uc&n;*(1m4CflxoEW{UyI$
z2Wazm+0kx5r3~)5+|X~zW&!p0AQx}HLah+&8>MVdaYMPusVr$7;Bt;5>G&Ry$!K09
ze*Y!JJaVrI@aT|rDznZ52Gy_M5sdu>EH1|Rae&c$-wLpO){CXcU+3>K|Bm@yWyv^m
z+&7yaIQkLi#Sd`~4xGPXFQcF0I=9*Jqm(*e<8HlM8vx4>1rP5-Jp8LA25{~AQ%$vi
zyn1Di-yuGW2!9LXy6E8A9l5}>$1lpI+TyyUG}~+v>aT0~*mc%--Z?;(^K(}HhWm{=F&@1O$_oyqEQtVIe&hOc
z9A~kO?0FjjTSvYO;e3^DR@j8=57)RvGrZ3&cY+S%a~m#``Q!uO`x&?32Y?^7zbN7J
zLx(=hZ*HbFX{A-u*dgo2Q_xVaFLN0^_WI-zWQwO9Ow8dQnZuGk}@H(_Q$S#63&Yz~|)Qfyr*XPubC%yc&NPO{loksFOBC@@N55M+*
zbqDS2yX@_W`kT2WNS>%8t>iP77q6Uq6VZK(L;&Ke3-tfrs+wm);zONO0Eura74{R}
zZZvp~@caUYB9bR-vZCL^G%AZBa-aXS5dS~zQ(lYue(zjG;_j@xG19(P`zx_CIHiY_
zjor=2K7u4CE8*LmE1!_-@5FVI_NEw`Z)+#khNC`~>OW+BmWCUX@u)4Mdq_^U3{->3ZHnHNO_vS*X>4KU_YX7wFJL^g|Fep8y8(0G
z0bV*K_jQ%Yd!3oQc1VY*kNG`36=pqHkkpI#bS#
zIX=To{BOTz>>#bTbQO&(%=__RUI!mxA^pB#UUz5z|Fr|00BMKAU<3u3^5M*N+<*1|
z*#VnjMD7?9cmB72VV3{9&;P~~4lNV;+sxlXvz4Bgt3>E|ywZv;cX-mzQGW`(j|DGr
zk?Z7qXspqv@%|u?oisLlrS+#Dr^{EF^UawlA9{(dkBp(Q
zmDz5}l$T=a=Vd-Wjrp|PtU>zu@XVt3qecd4zhbdWir6jXtor!;!Y@z)!
z%fJ}Xo8_NNW9|@<7u3}u76&9q`bL!sRH>+fcs~<&*B{%|<@u=2lAhjp;4+G33_`qOT!U0z7Yv3wcjQnhN5=
zvUGY+4&JYsB44i^pvXh|{~COiPq~A(<&7%JmcI9z7J>S2Olr6F(DG4DIMQJ;&DngXBE0L0=NJJ?jrHvvjmPjj-
zk|aw}DoLTF?O9u%o_e01BqHfIckX@u_*EGeEg$NH`jaSn}nAMuSo%nBH$#TEP#Cj&2mK;gw
z{U~56i}1S!7ai9i+U31cHJtFzL=G;-_xe476ZoDV?H6qlMEmPhvo#6*>Q3}~5cnv2
zvpk{CZO&Eo0|>J&!r=Ao9oLm#Gl-|J(N^ov5$5
zdS*4zzYj#tl@j_F$Sqn*@O3)PP7(E&EH3dPp2Gs~53>n*>vMd2LTCOFj}fv{Fu=&~sew)&inDEz8P2BA>C{x^ki%6ScM|0vrA%
zEow%1IzO{10ipYWW)WgsRm&!Z1|STOdniKme{;BWI#J)tZ!gk`by07+QtEz$qv4f@
zi2AoblfnBGCdxsnM7wsyEZar&^TqFSxZOc!rp^Sdsdq<_z~luB@O@69S4=Jt`B|s;
z+7anuD_Xe(#`KkF5PB;-tGhwy>3`Hzljw&d_Cqy<9;#0k`ymb7;Y!ZTP5ex<2&Nq@eAOYY0P>c55SAIl}<=vnK%!L<01)F%cIHt
zJlFkFI6dPE9V2^MG3KvzAnR{yCHLo!D3gBqgt(=czon7%&l6XW^JKyRIjA?KHE
z6CH7Rdn1Oiyn_Lm|KkplFA+!1p90yBSguKzJZ~|mP5SMHBV@ZR5=pt&;#0Al<2-U6
zqGyrwbk7=cUVR)%(y2Sh^~LCBa-PmLA@jS?NZN+-Bh=56`}uFF{ZO4QlK=Px>1T{l
z{@S%QWO`vQS^iKV$-m@CqMa#;u8QRT=tBq+aGax2M~+)3tBq5SqV
z%5VR2icHVwAW@Cl-w&hwCiIgz?$0BXU#Y%1Z*GX`x^Ln&J!fMoi2XVmd{mmlUT9{d=K*EcB;u|3^-k9}O~`
zPnchrPguWDU!h*exxn(meE%N`pNCLRSkHe^D2I#-E?-Cs*9Cv6^x4#R9U&hXAuKP{
zONhux`=56Gw;e)#g#9jTH_C+b3Hic)K+YLX7p|-RD<^D^P~X+mxHYBxdw=SlqF!sV
zoPiJ$h3mf{s{B=y-qqB2R-xWY;d4gLFV-WTLO)9WHdS94l`eeVLizPnI&w<>CqJ4D
zFnyiE9~8z=?~9`*$)DU#VkLE-lwk;I54k?%dcb-s$^REi;)?Yo>MSPh(FmJdN7eh2
z>3-BXzAWnd)f(y?+j?r5gR*O@squhjqo0k}GrH&!p3hpdNc%^7L)tf&dh+{H`lUi#zH$K@&oB3f
z$#Umwk?q_0g|u^S=ew|+wIlhxD_{{RS294Bt6@jV-z*`&$9||K_jm3tBjt)7khpjU
ziK-V!`Pbc~op0hcqalg*K@Ur#p@a7M78c=oHf5*i#>gxhuO
zyD`=)|8oVN*B|Sw$9_U;nkE|R=zlEPGnl`N`kpkle+ACJTeb#cSp>Q7RqI2}pN?X8tN#LK}#?=IFYth5)yMAad_EXjwurCwo|3>`r`GOBoD-{VkSgaiD8FYe-b9Y;H
zI?gvDpN`uZx|;WikTYALhxJ(G?68`!2pW}-Nzq#?L6aGI=
zrz(3DA*aN-U`U`%T2%^x#^ICkbCgh##`TD#AO1+ByU4d;zi8;$$1s9U$ccz0uzHfq
zEdr|wUmz2R&MS@XuOe_#u?Aizj#m9>B51vdZrIOj8DE1;67pNkJ+2TKsrxIGz|$w)
zkO@KmHy{6r_hHM|gyVB^(xsR1|3G3l{^I|l0=z%j5dQKa-Fs}p58vxTuHPahq;db|
z4jn;TOz3&lmEiUJHt6srXm^E9tf%EbEaq>mR>k|t7dOUp2)-Zl$N+)A4o}5;Dk#oI
zQw%yE`OF0C>nT%&pR;C&awL&%>GK=wxobuM?w7Tas_%&O)w^WB5omgh*FvDRLsK20
zr;f725yBsTFRo`zr0=qu97EvGsT0=`XmHWWg}}rQhrST_8?WKBoPeM!JS=g!
zVdtXn6SVbGFI;}C!Oaf@ZLp^t>;FBZ7?0DMC>8wvJ^LMn%kOeE!{c~D@@A~hkbNS~
z?-D@9e{iyRf%#1uD{;HPHyUH<&p2F8Skgs|O>sA|o|daVV}7RyX(whJJ&XDJkI8ez
z%U8X^dYWk3<8-rSmod&d?SawJ)*aVxWhnD=#DMxMm%ROF~FH?d?zQm!+MYf$Ogz*T0hA0iv#uc3YfI*1txDEO*-sl5gHZ&Wq;Fvns+B1BkLckB+LJmN%CE7NTgHe
zU>+5ddKom5@3?z)zglcwfP&U~_*>(u<4K<%$H+(|SiDN*IfQnWWk%Tsy)MfXyOCLNp)y*wDXR6cPEh5jCTi7_7;2aHw}
z|NMXQqd|oE!hGQry_rH})G$AbLNrNYdK-np{d{D6aXNZgF(RXYv5Z2Yzb-_f{zCs)
zSig`)FEh?B)K7>)IU$W+_W!9D5tzRAMe%vVWeyl^7jKi$>;{&YEB#&oI^XB=;MP2J
zzIO5n6`kqm{P3qtO;)BMAGF$v#x_0V!oyARQ4ynPUR=N#J6UBWIuAR;$;Czxg67HS+YuvCJ)nKe;l|(r4!rT{
znVEF22kt&Pv*XTe4rJ=j7qs|vgT3shg|27NeQl|d^fhlffl>R`SK=l*k4xv4B-o?4
z(RE>w#kMF;*7~_iJWtZV>94HQ#^w%?diw5g@hJ|71m+w*g~~atFw6JVHV%s89r^L=
z85*?jyA{hyV#9>JorOIcy5YSV7Xp>
z!46cvj$7=mJ`}IPeel?>@IW-Lo}GDh^fn7T=R2%^^oR}Z-Evs
zJ9@(kog>d*9dSu)6B~Tw#R_LRbwKIAm>n)Xov=-`JHuuK_46j#cJDW6KQYcvE8!v=
zUU)UVyr9Yg$(TCE3IhhrTAom|eUJucuWP8TMf%(R_R}8oXTik#WBPGDOz^Q^y)7k@
z0paUfb{tqshwUGqaM#gka5Yv{T_k`FIS#M2r@v&Oc+@8z>7n>vqU-KudMPkq<=#jg
z*O_#qWE5Z8>e@Q)8Vz-ICaGdjj$m^
ze$D41I!I3nxaVzy;&g7irDbBw0D)JJTBj=w80O;JRMGR0aF(|k6QzS;$!puU=zhPz
zXU`W--^@b$<-I|HHf-1~cq}N(r9suDbxmK>QT%apPU~+uRF6hrxZgeoq;7ZGwL_8#
zzQqB~E%J0YS@X5`>313o)}CvLZ>NE1aZ2}u)2Ke_{p+KQXb}0^pEVA(|J!ZzH!@#o
zpmXfZOx3w`uvnh|xAr2{lA|NY{<|r76)s
ztjgk=hA0iTf6~c+@5YAnLrV|uV6lMzky&hCWOhxR_{WOYWiKY6}
zn2q-di4{bo58X@dugy(&S3wWI2zn^>{-_B$N=}<
zy}VCt=y@M4J+yR;2BCL`{$`faz%qI3P;ey;3NBU^x1#$P<1Y>Dl}tds;Y;24TLJ?=ltXkXtkUz${HV*dLuPSg@WB
z)0)jTAGW4L*XRDu-}-crJ0#lu89hhcmVPSLkCpp5K;LyAw(I;B~59GY@mVK1N!=O77d(tx;wl^<8FMxna4L8
znIQE;z2E}+|F<<{W2(td8l3J(9`^~2lNp+EeB(|U*yL8fFIPk3(9oh`X(t0*>y>78
zqw%7fmYpw(>X~#XDetHfimyM~rX9GD4&I$*x797_(7f!RWF#8L9dZ%<-v7|^3Gu8^
z$w%_+AAx2e3@A%~^z&mL1HSzFG-k7e4z{3c7OO#r_%F_T98KtOI&ghW(-AsYtL3z@
z0_gDUuluKxP8K{1caT#$!h#IN6qyuJbPwmVQMUOo6aCM-b6)jdCh(i=3*V!7Ia*3^
zO1zQ*&Q*U7y3rW0VnObK5oso5PS}uJq|Ah#(AV4$Q>2gfSLug!sQe2Dwq_qro3*a;Yt*j
z=SQW~QWqv13R~44wTlT=?aB6cpD-b)^uFRRB_>S1`!l*zo(ZyYLxm5&qjD}}wpL^#
zyK!dTt9NM(sE!zSv3@)g`egWJBdA@+4-}qM+kx&6T^jwQSOd{KgZwsGCN$@BHfqgd
z!p2p;31e%SP?3^ry(^dr_pC)fe?sws+9HpNHaucNs&^Ep`Xstn)UwLtN-+bnBh6QT
zVixL9F!}UQ$_7>G~B&u5sv}tl_idPR!oR)-rW@I!vudRlXsy~OnCD|
zB0>S_*~9Nr=b?5Ry9_36HAHb5>o){7?_q+X)Zt)S5sG)@;F33%hV;mT_cw-F1J
ze{4~i=*(8dh$`
zV1fC1&&58jY#6HFzQQS<4GK>t&e25isVBUMzT1D01yj!pTpH1RtX75p%Cqd*@UvVc
zS2CFm?%zC&+r3zDP;21o>j)O?KJ~oTxrGH$WuB5*Mr_p1`L(`*Y?yW2F>*iBb58x+
ztaX+wh^WXPlYPtr_TB3984K8u$e5_5VaJBQ&D$4_pUZ+35+}Ar7^C~oC>!pDu*V%RW`o3O*3$gzY{*+2p_E?BhN^+I!AIym3(nbKj|7QMI90MUs5uqI
zflAkY5g)^b-~K0pwdicPfn_d%m%ev>-t{2
zX2Znke++ZCc7T1sn-g^<9q9g!nB}p39iXP%(VdIhzhp+{>bA>lu*$fSnu_8dnCqXN
zX%o@`dsF4gu2gkEcv9tAJ$*LNC!g8hew+>UGp24WLVAAJvYRWpyaV=+(70@u4k*jb
zU0r*#6Pz#ayw_jb3BMTJ95K%>fYgV2Occ-HV2IAmnQl<|i2h$2GRUY9zl#178@msj&JL&?ns0j6tP`wCs|{VJqVLg-`Wy1l
z|6g+9?{_w#XqQ?ihFLwe9dK^QStVPh6U7}V*78Sj7|Iu~Q0cTs-<=zKf*aD&`Z~jC
z;;?fE=<}b>cgg4guL}$6_BeM!=d&5RXfx5i4gGqG;$F0GF-OlW)V&+5u0H={{07Al
zKj`&h6Y>-FO6yk-yhHKO6mR_*3Fp8Ur$;yR;=AG7>0tdqQ4T0gp+(q!X-_8JeSNgZK!4|3st>ZC-Y&Asq6DyMLLelMsU
zQIhet?StlTGkJ$g`#?6pxS4g93(mhTGqs1g5aBpmDUjX^GImpgmCp4+&{2(B57`I5Xh6m3!_k~TE
z%!iF10ydQG=7a6e9pOu#^}*)&$>*$F`e5q3h{%M?JV@DOXh{Bg
zk{(x4^?q2hX#88n9sN*pwZo#klLuWR=DvLyJXlcSmyyHd!4pM)(+)pA$eqfK&`RV(
z`^PVP0_ZWKJ#cE|z-YF5QSba7
zc+On%Z)GtDx|1~4mwrU{$vp1wq)pwx=$*M@|4bf)s5>@=Y4Ctq(Iv8FtPh<_wh8|8
zu@@%7hn6S~7n(&_wBtxmoA5u|ub%D)`(f+%ufzn9oGIfJb6kMVG4S6=#Rx$3`UA0_
z@qCc#GW@;rCc>Al=HCyZ@wwE%{5Be&SDWgTYqI!o<*muN-6)^@t*e>S4hmrGz}3hm
zLjgo@ZBf?QAb`emONXoU1aQ}I-lrFq-B74Bu3jOC16zOBR;!!wU})O#Ju`*?tXB4k
zlwK0R8d^7ZzD6&YTV}_kX!Af(rA_{;uK@gegWYsW1TZ=IPlf9t0qj)d=Ds;2fTVMJ
z;xe}dkk$PB&~DUkhP#(|K3XY&LcuuhLVE!?MT`~yyv2tpu9I8MU-QAey}IgpNIwh(
z?OwS>7nO72UJrAp03=tK1Tswp&~?pu!RL>Q_JIj-AzI+gX$-6jd?q<~AZ)rZ&Ht1e4=R8+mq_^Vm&zJ$!
zuXz)9D(A(a`^%tda~_5dumExGey4-ZOP+^0E&_QDh9^f=QN
z4*VNh5`5XV8@4H&`s{}8YgO1F{e5n|0K$)_(`Q%-VAf4L*DxRaqa3Q)nW_*e`%Vw}y-%ngGoH4tU%_&+*W`#Az`W18_LcYuIz!
z0F;crIG|RD^y#va^tKj2>%+8o2h^X%&c#b^N~3-;Djw`W$Rr1z=J0+hBWw
z0FGPE@hi9{fS>{0<&mcZ@OX+*(O9hjyq@ILt{pc3`&Mq1YC+}La7-1fBaqylw59VJ
z1hBE?(DXZ~|GeGB6&XMJ!OB@a%hw#ui;2EzC)phExawFG`4o+Z7tP+Yssu16zD|bG
zB!Gc{GKneW{lHv3Jz*Hx{pNX9w>}>hz=F<43-;|3z`fFEk_DW8*d7xqcfgwu^Y54X
zZF~3QdgD6^uj+6&zWjIJ#daz@0?gHfL9Hs
g%nL^Z5SZbx_sh3_P~;e{%CPB!=ZzK%cn%!+KRLE5M*si-

literal 205696
zcmeFZd039$_di;qfuay8qCt|OjG1yrMJXb)Qc_7O^H4HGgovc5Od(U5=OGeBid2#`
zDWbXO(Lhl-ueI;{`Fua;eZJq{=lWgeT<4#2{9(0*z4qGUT6^t%Kl?1rrzmw)?ODVHICI9^u{`u?3Fp>n9@RtaGNDFYU057%U|B#kalOmp)Kvx#@_v6(`Ef?hG
z*jZSRh(x~x3dR0^@c;21u(X&ole%Dm&=2wgOcCaffU>zpe=}hT@&5*t%-0B6UWr7n
zqLdLM$VkabDN4ypNw0ME@LngS47+&KO#fxd|H(o{DN~_bM+l;v+nNo|>pfPjTqo5{
zSxTxmn3APrw#s|;TDNuUJeGR7S#0ofTWcW^9d3>~CH2em^{-y#j^I
z`>%3xg56FVo!9>NhW}f!f2*&Pq@0_Vo44D__5Xd@7NJT-sWI#?87ZN?-frubJFWL!
z?Z(>M6--Hc=dN70e(eUA^&8f@xlV@^S}FB!_5OSLWgcF9dAWZpPnw4=AH}LeIZ3@8
zq`Ff-ulMkFlTzu&{cCtec9I(T7rT(8`b>Y@^G{~ia+30r{*~d&yRQFVmhO6UdeXz9
z)7*dcuL9psRPVa=Zma*>fzjkEV~vpGEfLJ2eUtu4OY?*iJSruDGQz(4-?fO!b)Yr_
zwrkgpDUcWR$w&(wAXS`1a*}^t1u_X&c8zvZl+tG`8K#Nmw=?>WrYLRPyiHP4YnQK*
zD3G%J*I)ln{2^I)M(MmZ(|;(X*&wlkSQ7c-H@u)J5B~MPwABAw{ac*PSNfm&w;vPx
zzxFR}7m}839QOa*PR3jH&;NNpE8<$%A0qCz0RIT^mjK%Y$d?xjtK`2RWB(O#t^Ywz
zqW3@PX%h5EQ
zy`V=@uLMf?l6uMnx&JgCyb}_0|FlP9mjr$n^hxTG&?Qh}?{G#CbG%Ow|7)Bj`n3dp
zUp95c+-pHEA6SaHj{>eI$Y~3BPXW&p@IC^*UBLNWNX$JIa1DWfN5EYL+)2Qv3ivhw
zUn1bQSVKgdHS0JL7tOq~m{VsB7jgSpJSygXR8T?Q
zRPpkvLis5I-%5~IWdkARrn3$db6JhN{6?Xl^aw?Lb!;F++&)bnf1*GN#N0hWpAwru
zB5vVep6)fB8X)G3&hof^5MOSu0gtO{Q3Yadastg95m(xt=eNw_`Lz~2o@&U;%{QuJ
za(_Pa{H(K7v4|@v2rokmtOzEvsBJj}zuaSWCx+n|9WPF?CV<0tA;
zC?e*<&Yxj;=#m!3|NiVGqyO31i^2M)U5r2f2;UDQ4WFO^V$NiWHD&41(DaT{K5XL`6ry#3Zwx|b7upwA&z
zuC?noW{*>?H7l1OH>Z;L;}0EVe1kx3R^Q)rBc`W8*%uuy;{5fT(6M5!F={bpqL|Ze
z-h+-2b0JofONig&_-w?SV(xg$iIs%wB%9k2R5`B3%6AOd(u4Rii*3IU-Lob4Lo1+k
zFQfZJf4j<^*=f7pPE361-rbm;Qd@YqGqIB4L$6I@cD^r_N5_e|bgNm+PN|hoSbHAJ
z_%i;v%mUV5$5J#|{a>|H7*x112@_Dn4R5Jt`g%K$T}Jp#C&O<9LkEmgB)#{h!_fnF
zeNn;WnqtPf5&qK6{}RDf@>3@h{PE)82!e6f^bpgCIghQ|nEcVtv=fQA$$e-?7IEum
zrZc!@0(lk@xBdGgW@k(6PKMWaMKe(T&k`jD`&nLR^zX-SFnx7T4l{b;D1P3`3F{(c
z?+iwl+Skb1mFmBM!R+8oj9&G9DAVKFe=(aMkw(XLsGjbR+q3>q%ur$a%Ocibfr+_&
zqWg@Vr3wH$55MWc^vqbuufOFYetpXS0U7eiL8qDAjMo_qYMtTfR_Z*wtONDXlOJf$
zbuTB!|u=0Dw
z$~j?Hk^IsvJbaAVNc#E<`}ELS9?v2Gx)zB+{*r)8>@=il1o|M_Rw1{HP~fqG9L*}C
zKPR(54-(`TQu9IQ<8bl6^glVp(><{&iT_OSPuDTa2p5qW=y`aChYI(3x`W_%A_Tvv
zP1`u+XA1HSgaW5oKu{7lNa7a>pLTQ7FNw#PnUH5>&@T$*9}7?tC$M3F92*uuH=&(g
zg1&QtJe#J_gG~(oFfRlXF#_KS4?TlKzy)N$3(>Vwc1YN&OPK1k&zC
z^Gh;sC3Z;kOQ1xLgfFS*e-}#Zl*}W^cuVw1_!4``E7AO^VGI$+wgKQG0cYDL@I`_g
z+ZKUK+9Q#d(CY;`wymMvY&(KHBLjXG^qyu7|F8P~*>1`DO%v9uWWSK?lah15azURG
zYk-LR#2O~zUJCTff<3p{K!`Y7VIR9G=7J)i@3`lczq8Mg%Yl=$m4arr~|}Y*>OG&wqgS-=7zTMpklp<7c-+SPEQ2=K;h2G&?g~{@6trB;t;JAJ6d0*C!eO;hjkguN!v^9U*!`1iL&ZyA-=tW%NxQi4d(Xx4gBU>y_TSJLw&dkpT<|S~G&|6@6&3COxJ8t3615kZ5TU
zwJR(+?p-bD3!Yw46#>3Z-m|n0P-n<;!xq5bI|Et&b_|GR@x1)@P&P0B*uH0Yqxxw!
z-jhDA!~#UT+G7Aa$9X&c#14o3DCaVpuU>wGu~mz?DM}aE`LMSAhd0DOaPd6T<0co$
zuYWC&
zN+h`vH9C)|-N9y>X(Tswdm*zk%DEmcPsGj8Ph{!g%kOxV}aT0LWlfF#;LCSM5MBGmACb%RKw`(DW9rf7UD`j$*R=`})7x{L<&WCfz
zPX$D<$WuL=%+}NNzPU_(qT)>^7rCr1iuelct66*ZSg&LC*AH-H>s9T}1P1H2QUodD
zMmLiK!MZqzWr}hYkHH46WA~0{m)wUXLMdgTT{5|FZsSElFi~O6XXzf
zIS)&4iJZh+vF+Q8NkdxRcIagt`Wc(Cn5aM+KO6-)>1E+(3
zB96bYL|#T%XS53wU847&?JpMQ6HAYv{x~LpxKk)E&ngmeJ`@&0Pazv15eucNB`VoZSAN)G&
zP!TtmZD?Z7H-HQla~+j={yExSfiJqvm)qNc=l^^{BY=G4T%K-h#mhxo@wj{qk7p%P
z$G{(0^Y~hh=Wjg1Z|3<90X)BuU1W&3lhgV7CK&Sc#!6}^;w)jm
zRXV`i)&4Ob|Esh+%J>es{Qi6(mKu(D?F}_d%uU=+2MrOY>{-lk#b7?ZPoB}4;qnU$
z7+-q_Z&%Kx0}TJ(LMJ#e=i@>%3w{I97vV!@kVC`tm?|@
zOY>iak;Hjz$TWtRo*ah`MV#4dCo3OOrHhV2{=D@YwjRWZGP+L)4?9?i4i$56lHW8D
zUK7&H_@Sr&!f?ciG1IXX;(R6_k1ZPex)~mT0JmtGfFrSp^X_Gh8H4+tgiDtRSDopG
zjuvzAS7caut8>xngx{Go+L`zvx;|S7u9Rw;N_6FY)~g90HuC8T!ac@3v8VX=_tl^-
z9e_J#a5=<(+G{NaRK(p`G`DAxk~cur?07HdO>)`
z`wiPDek-uJ@POiJQOsu6A7SpY^NBxMWh<_nMBKMm6(LQ4o0ZmfC;7IGIaop>Zb&yD
zRnn)Y5+6%({jF7}RaL;B`E1jb9)Ppglq{!sc>DoI7x>lSrEbcAzSmyq
z$G73Uy-`nx;(vMnfv+h3H_Uk4m+YRJ>3&BI@W@|_D+QoGXs#_FeV#!LhV{Tl{@z_r
zdPa;N5fKSIbnfX3X8=2$Q~p5peZQAJ~Z`oH!32+;5RjeP3~`b~w!H`YmuIX$ltODukg
zxED(m&kS$IID9p1`VRir+P?awfCc*BQp*5Kq+DI8f7+I*|Dbl}Elr8J2wWjTrI^kE
zZ)aD}=?MC5gXAsLz9X{hd`Q0HhP;CvMzo)Nar^T
zpSxPQu=BNPZ;hYm&(k(z>+`{XlB;qx8E}My?A0tlhlf+j9bjkdXV+)lAXnerWu6Pl
zcQ=`Q$s6!esArotpiAVv2C{QhnaoN$kCa@R%;Mtx3Tr;W$|bw$xUwm3Is9NbAw+Y={?c3ZX(GqJ`=WyEH<8A<;PDaL6FTlb}%lmbPoYc2?
z13EAITvPU-b|=@a#Q`7lp`Y&*s;BRU!R-EO;~vn;F@;&Hd_TDz~(BfSUl|9
z0KzMcDwmMF`%C+L!lxa+h7A<)heB#A&FjOj6yMXhiXG$N`7kc|ibljA(BDX#U_r7u
z+c#D;US{@OHF(YTL&qwLR4}g``S^bIz(TgKJu2h(Wpy9&)FRI0lrAgh9Zm-k)YH6-
z?T-iN005uRoic2=S1Ivic4*9^xJS@2JkjWKl
z^5-1!5Nx~PH?QH(17Z2R+{6t_8U6k+-hUOnq)1T2U5ckj1>^IJ0tFG*{zqqKfBJd8
zommGdGRFS0xsdT~kMsU&U&02q&+pA!!usddd*0uqclBd(8Xf6ChkZn-f9D3E!0m-N
zVIo>fdegAG6Wyg4(+0Tvex5FOmZzHv=YcUq2H!h@_x~?(Nus)F4t
z?1J5=i41s%)IwfKI2W3cnV_$t;-DX(>VUK37+~UHo^FiYg7^(WoR%oaYmzwRZ2Rzb
z?)}W;UM#>BbKhAYk9-gfAL-x7%N-ZqAB0o0Kp!jg?-Ukria19igZ`D&K|iYwxCGKH
zBKmqk4o;iNvotd3EKK}2c^m3h(AN58J~3*=&V}-F&UX08y=J&BFHfSoUtJ%!Lj)85BI17l<1T612zZ7r`1RH
zWC@U&2V5d2InP-Nd|JJvmsS~-W787&5&@SN+H>kNFLz0xe`XC4aTx+`PRkeNez1zf
z+)6@$OZp*1ke9s2(MNPd^-1D>c|pEZ(DzE1Uypb5dUgtQv7kpzs3%2;@2|J;`d$e0
zIGw^x=pX!oFCVd&$FHB^*GUYrq{Nrp$M$9u2K}&P(rDxmFLt1j6>;yD@%X&lJcjSu*2wsKg?Soz
zl-HB6kQNy3{nPolJyh0)$z|s9`MGaJe7mfI`8fMd>Ubuv;y9D-_bpcAnY{Zhe!t)K
zocCL0r^6Us_U0Vs2fKHn3lT9_UK__?%uKrZ#<@W*2!_D_w_Ij^HA&hG9V6yCKhkG>
z*Q@F10L-IXCFn>IXJA1U!H*REV0wHK)tH=-)iG9XPDB`H0-Y~ot`H3Tq>YVH#C5aI
zMa#t;x8^Mxg#D>?wJPD=#|<8?fW>m?WZKz5MqW
z#g`8ZOv)*q-PEZi@81=mwy6A6v#Q%@{m4~bc{h-3bney$#T=l&~&j495H>uQV
zr9aBAvtRQk`w!yv$#*LUv=?(@(#JVEK`vssgX|Eoh;v>wb>dFgGwRl}o24?y+ooT=
z?T2#jSLE21WQ({-?RrK$L45^5LEd*z-;db2%l34|I5#K1_;4NdPU-To3(EbzrhLiQ
zPR#98$+NhF_`Y!dxTTut_ms&uKH8zaE34B}Zb47o>~|+Nf}X7EYwUt{f1g&p;t2RL
zY6trpf`5KN2lWf6zq{VE9@JkaHhL{CM?c*P^oidHy{*P}1N9)+$s~Eg=N1uH{dsZ8
zTnvHSQ*B!B2s5i(rZcg`N@NAShBO(L>W
zVaXLw>Tl1{_1meRJw1LrVfq(|E|J{NaB}##hda@W*|~o|?{{{NqZbVD!!A(_rtf*m
z@5k&Rl!)_^wPN(J^K)4F5@QNDkf+J}fRGUWvALYtt+9!Q)}QXLtKh&z
z+|;wltem0($dFGiM0KDCP+AOd+95uEDA~+0T-KLA2fmM@02Jqfy%bsE97E47f!|83
zU{Loc%mUtV053nIKOZM#W>bb7>+yA0=GW&0!;L_$YZQhR^mf8Me%vk{MxWXe!|v&R
zUN>d_SL)^yM)$1g%;cvVP#}*yOD0MQ{bC9%anC)Iw_js8&$m6r^LL(x
z+0Zv{kvo$+Alr>W{Y*Z7dh(jL$Kn)UPka$yKAf5Zy?O)y=`>Ax!Ua6&EsqydxCXkj
zAa6%%fSU_)Bc|}>4+!Ulg=8l9ErR@HG7tD7L%#m!6gHy15;6<$HKH2gYtirlckar!
zv*&YOez}kzcB5-W$SKiufgJ00z*|HH-HOP-Ut!b<*OB0a0%z5N&yI({j|ucHSj|*!
zgdoqtZSYxI9q_i$jy2DCJ+$gcevDxEMZxY_S9$*HQr_-Hc927!RM1P=8!8w1h?iRv
zNCdo-q+vom7PLHozsu$E;#E98ZWc8N{!1vIi0m!lLj-xH1Rk&K%deBci9DV{$8YEx
zF4*nghsX2UQG>C*wR!xoa2|gY$k#I?hk91b*|7thnAC6
zMS2bP+?mPO8@Y{df4ez+xebG}!9X6*gx_~7UFl#V;uQPyal70%K97IInhb=0eENt%
z=YhN(PEm9q5^;mmyD+-_!5a+UelY_RMZ}#+*vxP(WKjvf*M{&CdYm7uLW2;$78)@>
z;5c53**T{Bb;e(Rzdyr=C_11a$TPII_9J+85goAbE>nlM_xsW>Om28VZ>A?9=?z92
z`Oqb)Y`;*u1vdrUI6ax^?Pe8+DJ0@HZdE}73vqiu8k19(u|kK4xmnYuVxX`O_ieyH
z;QhVNx%~taN7KbK;*E+I%$_w5A7XK4?yRf8*E<=>gmAzgO=Gbc0TDEu53yE=Ie2}dE~FGo_zm7c%sT+
z%s4SOaIW8d!l!xoV}>G5ZFjOO#U~x)9F`EjAZQ9sf+CJ9nvSo%#9UE8Loa$?v0G)o
zJmm+6_s*M7`FH=cs;{Q06>k8=-1WT=wotx&&7Bi^kij2^`m?%eR2RN!T*?==b$|9#23=Sc0|YnZZ{^01W#$+<7VcM+!z
zqWD`iL*wmC#DPa`Kcv$6;cTYjN^QUamV+$+p!~*R*P6*_V%)c`GR663a?RRSn!MFM7)ps5E?_V42X+(LG422!H
z5tk25NO%^6xKbzPcJUg-bDe@`%%(W&*#z}s3&evr-Si^z)3RRR>S?biO74k2-ZH=xD6mh9Tu4`D-AdbBEF@Ien^0j|v*Vrf^
zj_zoq+-r}B^YD&POi{tPCA{$D>;%M7`BrPjmE*n2A#p=xM==+$V!v0XC=quwOJP#%
zBAhETH^;}Lo=2PgpHROZ3=D|KM13<~>|SsJc67QFoOmC0^$(nx5CQz?wJV*kv@cU7Fx
zjMZ;{ys75KYz;P*4ij-buMJbT(#5;(cDEx>p+C;dUwmya>~#6QB6<(%NlaOTFM!0{
za*ZBIbUw5ZeVjsZsEXW&?sowPhB)cg0AAhwqM7E+OsmTm=^o^--LMjx|8GyK|K0c#
z?f3HPKyiPG>;xCOhw)xs;@}Q`-w6SUgMd%lRCeu9KF*Q;yCQaCJaScwO}b;AoZD(>
z_yOZ?V%WSo7;r~KiLwXg&B3p?-!|u=Upm*#cf&a-#`ai$f9QGnW7$Sa^h;dNN&9IX
zeS8(h*5l1x!Ov)&*6gWNqIHwfYwi)kUlwiCs|2q0D`p~{D~22@x8dkN`{{ZBA-98nh4!rVW
zBfGC_>DkSQ^hd-5uzW*e=if?%+xEN3_Sp$OmD;qQ^q70Zl<1L@J0U_ro?+WU2E)Z}
zOn%*zL}cvYm(qP75cJF|HX-@7+|ELh?{k0A1A^z`d)%UPV1-;d+h;o@+e{<*6VX%f
zB?anj=z>f&^7OhIIYz+c1FU<}xO;oV&ZT{3QHMi6NiKeo4LmOLId`+*h>&l56UE|W
z+Y2SkFDx87g4w^vof2>Ge-rrcevYaAVE*Cwt67YmmQR5o-XkadVtz+$i2FB^xA$vh
z{=xDVMPMRskU2##cyCr8$NWXBrsg-Y`vKP@f#@NH@8B@;-tuz?1jxv1PMOU7*YknS
zs??r`W@fA(SM2sjh9CKWHA7gwX;Ezk0%-Khcr9kfi2XR$fp5`|0uS_?kVl-V#{2hy
zt_{qdb@Y4{{M8|-2J)xf_;{qeGmnS2{AP5i4ZPjrhY*K+Xc^DH5rX3ha9`6NterWD
zUaVc+uC-(IQ^I|C_Ki%2do)jG{{mj$XWd
zY81|cZ$;G+9kYSpaLjI^w_|~$n3EIiv(w<^7r@(++*o}+Zr|Sr=0Scpr4s=SDJ%is
znT8K|MaCh9U;Bt-5$KvOFcb8ps(hRkdx)1)U(eGmM}J~+?qB$RoeD2a<&M@+CLHHn
zA@1`dGofdUkQc@wlHm*I@b)MP@nk#dev}Jffil){AdgG()f15&Bz_i!+ko$dxIE|@
zUvEnqZ&!0So^N?ysOL7OE9Cby^L#ZgZB~Ddp#P@uuDh}mZ?|?a!W`JE9Y_ZtF?Z%1
z-@Y~#z8{M6c|1>%0%qiocIYrT>?BXucjxQ3>Bqydw=q4Sr%Nw8CU>w8A4euk=j$_I
z4}38G+v$e~(BHdN+4_1Iz~_BC+`Y!iZR|w{7u*-^uVwuG^&?>b?vonn;DdO>_&9^^
z-@Tcfy&}RK(6?)>VNk7sj~{;+CosO^5?S>N
zZ5(OfN8B@viNE0JKKyV3^1ZIIEZ)0w1J8JY+iP~h1jGI3617c)@6vdJ4HbU)Q5ma$
zRH)2T!e_gl97X&gAvrJ<`+N6gDwMxj@YfsxI`S7wXR&?xQ|=>{7c%Rac~xhe
zqx_ResVMg<;Ns?A@)Y;>OOGE(_9*Lb`bBaknhkoNfY-)YSde_hv`*tG{~CYT!k6O7
z3G$grGh~qqK&Yry^;$os?pPfOu%{@je(*Ud=os)7H^5divn+H<84Hi8=0J{8LQZWGh
zckfG!7a)K7ElhKT1?(BO`0|TRh{q1yw`ga-9`WSn@xv*v7gyvpqpbq>mv3bz{87a^
zigRe~eL%!rD_LPVr=fB0l_<|6uN_-iVuvt%H`j!f%~>v?FW-
z@<*A$cUnH;Ua;5fpW%qlyi9wzRdz$X=&4$;ULA4GuKb-}r`2)g^0vXD+BhE;8OOv9
z#{8A53*A##%dHsh)Tu@N2e-nww^Prhja>UCzs@_nu#adotnF1+%Vj5Jt5$46yz=Mx
znUKc~+_+Dbou%^|xWxJ`zTc13aypqyuOve6mF{=kr9z2wfD7LH;>N;S4=`Yb7jzdz<>{h*lzgUh)zZ-0f^o6ET}&8EKJ%qzL@z8^kyyI##5U%9}e
z^=9$D+V4{$;}+4Hu6%ar
zb+^MT%;V(7kq>9$yz+hj)Q26WBK{Rsb?c)Ex|iF|F15Ii%Fi@uz&vXB5?x#b{UM82
zcbZ*T!`+y%t>=zh?0c<;`lzVm9NE@+i_BLMXKA>>;TiNP+qe$7+*ZYnbH9FW-kn-*
zXzix)hWg0M{9O8==m*`)j{X^5igh~A?yMBz$pQ9nzFojPY*L&5b{^{MpQ7%%AGmIM
z-YI(5Rb(2!t_tz^bmy>NjTj$%X)$*h=-D+=rzn4ZZPBSr+BZ&UPTz48>qf3|+$YN4
zd(^B-+6#E$dr33J@o61G*U)|=bKuWCI^V7EZTO^tb@6b_(0#PeW&~Eo=VP6=O&b#)
z1iXL5^C*V`*l}&E(L?m>g86&QheE$c&Z6i~v<|j!-#QF(&yV>Q(R$0#YSg52s=Vhf
zUD{vXe%ON#b&wZZpvBJPBSfc-sXZ(E-pZr6Q_pzh9XfB^w?6%r_R}`6oA^)wc_i-1
z1VzBs@{EI|r`L`Zk7?g5A1hl#`ii2i1=9Z6<`I^c4196>nK|^nZES}ai#*T|nhd^8
z`{vn`xxY#N%359DXMmo|ueQ1ao-z~lNQ2z5$3JJ%K2)}46D~yY{eAK$8QRB#T-C-A
ztnIu2ju!VLZwDb_f!-u9md9ui%}yY^J*Zuxvx@fAFBUTNa1?n6gIZ6Le|qQ4eCn5pSDrI}tJh^2^WSD}6nNnKAw5Or
zx7*3uGC1n&N#>vLUD0H6>&K}vxvyflP3W8EJBUG1{x#MQX5Mnl|JrnsW`6woum;9o
z7U|C5>|;~TQaufWS0HfWU=>Y8nMa9w2C>`fO@9)
z@UQi(-?mQ(X1MKPxO4D>qi}3MdFP3ITzE%_7bB2GB6`$4xO>2(bBY-BAHc`IAI_nB
zLI2wZGXT|nx-$6f<`xFym2s?q{@lG3Xd>PZr$i|H;R|g9g!mr81lJzGPpsze%i20GMlU{O&gdyyEg1f|l#iRuDXjL;GBaG@z
z50h}O;sOAuXUEH}4WWP*`KbB4oz2~l{sP}*gTQa&<*PdK`UdE$F#ZZh3W#z3Wgm88
z{}S@6=YlW{p!*--^%Z_IWw_EqKK^aW=k>S?dD1J&eB2tg$%l>Osw7NXlrLX*g~8Y7
z9a(+(<;PJF-?upgv3a5K6VnyA>VO;u%`VPja-J)K(I9;H#c^!hRW7_`ddpU0`Ji3<
z!VDN}9uUCfvbN|Ul80ZlQe<)!eW`{x6--<9ft2WQePmUiuaJ^9&CP{9O%XJiP>Xs7m?hjRGAg5z5%QE6aJ#Dna%qH
zy{0mK8#8aQ^|tkIF^fB!Y4+`M#HdRpB_zD_h>^#uHL#Gt^BjmYQyTKVgeBI3Vy&p#!s
zMc!`tl|mh<<+a?m!>6YYt;2cXWv_sK$lvwU*M7B64ez0bF3#MEc<-{ukl4--YPlVs
z9rI$~f48Q$JvN2i>6hRC$;173+`W&BTc#tgvFNwG0phFRUN!ZH-eDeCJAJs{3H#Q2
z*$-N1Uyj$GAzHVwe`LqE&MNxBS$>xF&_R2Y?0(q|f}M44VQGU94?i0_y7OAVkfzCf
zibvLSnz}o5{B*0i{?;xR@?O_+!Mi`YtQio
z?BYMlq?YsVHP-Z5Kb#K^B(xny|CmPgy|@1}#=FOpsTBd0++yRe9_@P8a2r;CJ~tM8
z!zGb*4|~FX?ZWPwh_|dy-#AhiRKaDc*EPL+Ud7E1-m_lU8-9M#iUG&;kiUIdtfGs!
zO(UvFqx+Z&&UTqu=%9v5&Pn_Cl+7||ul1bt0>oJ}+r{_afc)}#zg^XjO3;3{-)Wan
z->KH>+ckhD_b#?Z8CUVFdFEMU$j>0@qP@ZpKo`>`7A-l(yAvaL9=?v!u(RchYmqf@y^J1b>0H@k`NB?gZnS=MpS|0$@QSdEL3}E=`^$UUm(0tx
z%jQ5Xyz8g&JAfBzB}P!bbmH=%F7kj1+nC_4
zI<9?B``6Ix9tY{3rhoc)C(8T2emk?4_G2rbV0qf-X6v}IeDV4BC2K|kPcOPuPWx3q
zbx-zgW&O0?#w6GAn7}nujG705HDtHq_43P9zFZ}QCN%+VE*m;q}j}`T1RsX_uRwhH?Jw*M+AZU=3RzNf4$8o=8u0mIWxJU
zDwz2sQQZ3&_{b2vXerO++GhnY
zIT_)*)q658n4WFx>zG`>hNDbwiw4{w%H_UyVsKdn*udq}_;~4s6^~0V;PbZKUtMD5
z1|8z*`{`$~V2AZZ6hpbqg?zjiH&l<&XWZxGtP{a}`Fr2_a`yRrzV6GTp)9VL@`{hA
z$J~9y%H^B$?=J6Dn6UcxYs_Qk0qKG~#-Hec>sHizlv~K^?G)dM#V>D$Ut{p8-Ug=k
zM0zvhI}}f1_(7Rg)(?kQnCO$epO$7Z`h@ZrCf7Q2J*%(DUW1KK;nXR}^x`}-U=h=M
zVKpUG;a{ddW%#@eM_IXS-2rAq_v~BD;)-_`Jtj_r+hQ$A?T=AY@W5@u;
z58Wigp#ST&coB~C^V9hl2>3&@o$-Wk>aT@B4(HQ(k53a`k{4@5_*R`OzBC_$8fV8*
zKI+=x6Iw(sA9)!!`-so4o3i+w|!q;-b?q$wki+D1Ue9^N%t8fqPYS++IfX)P@iqdvtoIjQ%OrEzZ0Ioo%X
zHt^TUoxjmL$|1I6S9OHkh~3lMY!NSPt~46Z0%%b&NVO9DDI2nsCt%-~U-EwBAJ{#!
z74aSNW718_$KJS$^G^?x)%`&iXXks-`QGKQcvlF@E9?r-cp*X_q4{0NT*MP6o+~fZ
zMm=}Ggod1iUWK2BkCZ?@Zu)-bKJZ_I4~XmM_rkltN`nQh@SD9%+qcVxU;VMRvv?=?
z8q+*C`O1sA6*jW^9=Nw1IzDnG`2{1f<~2F^)z~Vp16g=)87tl2BKQd6Er&i9`EIpb
zM(F8=XGQoOf<0zuX5rrW%z!V7ridRR-k6?^cnCl0({Y)hV?CE~yQgjd{ATZ_jgbe&
zpgoI*PWy8U`3O(rM^_y19sT2qhQYueQ+KfCkQF
z>GowgxPM*~{UPdI4BksAZ=E6Ypq@*(v8aAtOanJ$TFTLhwvF7RzA@JoM1kOmsMQDkj+2z%4@k-OKSt4viTmP%PwmS%#B-l#Uf!;Q`J)hK
zFlr0px5rDKB~PaOY3f(`_K1^R-L}?#fc@c_<(wzvZU1cAd|*xu_sMucic$>n+TyxN
z4S3gTKK1FY>By@+UF_vmgL!S8ndhW;tcu%xFZ0>e4Ygd$t+@GDwUIa9ygo1p@yd7Y
zS+dsO;WtaHMJ2E&D*bNYEX?cL{zmFO)3s;rgy%}9^AF2{o%O0Z)OPAr&3OFD&ehvcI7z%KF}Gwp4PK*k=>@x
zz^9D&yIhQRl}cyYJ;S=F`dWfdcaZ1lT6nz&;H7K7y|zPsaE^uFbsGhzzKUb0bs!H+RbAS~xM_;zJnZ-GywC(V155`&B{V>UWy&lHye@c4qV*cWT
z(HoZE)3SVGL2;?^^Ry?l|KAKtSws4Eb{LdOi9Aa`~Lam
zEDkKVGTN54)6fl{*yFuLHC@nReK-dqfWY}h>O1q(6{$HwR(OdoL|X&B%${rUV>z~$+z-f3qh;6fJfcvjFa0>h8_
z^7+CcIS8{sk4@>wS~LvPLUXdXp(oz3QD)x8`xzbCFe%*Ne1XwaUwW()f+{U8X!u=(-cjiGRE_{~_%sH+N8gk99xA
zPnzUb8_S1$0ljLq>|Kf}?YTF{zz-@?yGeQVQL7gniA23~-|b!;1wS=s`IUM5L7!>4sdFgg
zl*3kzCb?H%zm0hbIqtao;xfF`TD?&_GZpn8DwF;d2>lD?TRi)8!S_1nP2T38K^z;F
zQhYM6mW%m4%HFO)#2KY+YP%?;G+t+h6su(>aKDT$hx_~XkIG!XVgIWcz3&3-o~NnW?tHQs_-OH=NqsTzB4YfsU~h+{%O7T;z25Prn+*q4aUs=SGy9@Wir$+U@Y=n3-`16wNZ}5j3oV6!up}i?`R*Ll)FZ-yJ
z>b+?9#`db>EbwPEP1n~4{jiD*{63AVeG-!A&YYw4r?y^K}u_EWvXRV#E*pXQPkk1Jr0
z{gHY0hhgX6wrYIPhITyOr1c4WCq4UKlmqayb!Y4P0SufG_#)9eWe3av`Tu
zdGjyD1D;d!K9Ic${#N**W!wXOf%(8s=#91{_{shXdl$QM?DWZW4=0w}ltcUGsN%CM
z-|h2YDo*Oi=agM?r1p4q3Vus^i(3y#(>@h#zT`OVL%Ma+?44VH@n9Ul(H@z*@8~_=
z{)#vpsBuonc=Cnzp-;-k@r&t*QmdTXw%D0)}v@^E2=``uVamiL)@{WJ5w2WpWP0{_Jz{#{Y)
zX1GVleQ&SM{H02ZBJ-DtZZDYJo2R^gemWCw3FUHbtz&W>XL+#r;o)0f29S4Z&oX!^qMy+_1!XZo7HuVs0m(({2#uA2(~Uf%4R
z5z`xZmmc=x-QFY{*50irXEA@7ed-nSm(r82vHp5H?wUEZr)GK2-ZbBmmyTe0hO{|#
zEMIXqXf~@y?&?-!lGE9=qk+!lGNCfMgrAPnW_h>zop+hs#`T{#qL=pSslw!B`TWYz
zt&^BPk2ILX#=TBDc?!ur(Hb$F>NV&;i^UOvnO@9p@w}Pn7~I>gsbuqhYw4-=G+t{Q
z^KtWt_`cY86#2_3+7~|4{Oh5q7fJrI^Y1QuXYwo6q1wFU)iO
z$Og)9#Pl6_Vgu;s#-FaBdn8@Io_6%keBIs%cCM%?{r-;fE)Tl8sqO=P+n`;}RPVBw
zZR1E^t>(!rs`tkEk!z{mGus@4oI$@ce%4f~N8x77C2HT1``aA|h7SF3jNa=im+N#J
z2wbfBB0C5$aQ3<&@|U{(%%+jQ{4k(XzzgiZwqKkxQ{X=`IyHMA2VHmU!dhp@sVP-n
z9SS_+g@d0nZ(l)w7=G`{maq&4k>C*mmU
z?92_=@AIoCZr8#&_=8WWr2^!~{|t4>fL}_rfAiS826@oJ!O|V!XB3V0^?3w8?$u7k
zc|H7@@9-{1yr02eth|4*sA|OzFq(Zk1eA+?V2W8>%_0rS5(*k(gI$
zqcUu9u2!9N@Wptv%luYptGKR$>%A(uzSp>F&Zqylc@fdbt5jKP8sI%tfr=x{n-$cQ8vz?seF!SIK2gGx?fi
zT*Hk|4rt#O@ymCE2USbZPkS%_om_x%+m}~-X*2w#`>g?g#=&3SwUi&d7yaWsUv;-G
z#?Qce|IJ$T!=EM%om}+aK%c<=9g#|@BSE)CgQmJfRp
zzdgKz`o-0lE3fJG>`1)N{?~Um9Q4M;(-f{fn3p`ZY
zwmbRDH>R3SO1p5#BZbhdt=`=N`|m+vG0*J0GC$9rJ!oVM2Mi-4c_
zu8*G<#<|00OLzKCrS|ZW{^ZY%Q*K?RbHSad9VXHFJUQilImPukBf}afPFT4tD}(Nr
z6z+X)q3=tkk3F`B^aRvA+D`jM&Bqrug#S*>3MT*8VMC_^x-WXFew^K(OwCEai(&ju
z#--zKBvXki%14wRa>da7*+w{Pq2kG$$TEfnc
zb?O6{U(Gprz?|fKM=S{;xx>S5u=Al!weDBKP3CuFe#=Gq*)x*MO*_u=Cl~FSn19q5
zS-|{$U~VFl3z?yh9}L2~LCtxkRQ`indlpAbQyRqLisgfJnco?>aS!vC6HVdXAm`sO
zmifywEBZ6PGs-@Wuh$F5bnxwR;~BJ$O^t=c3&<6ge
zyBx#g%JBRFc%~Eoo@ciTe||TK=Ji(GMEU~db{X;I9xdSQnj!rDsl}IOR{qWxUVrnV
ze1w8o%dxmcr&d$}GmIFIaZy8c=LjBf|j&v64=}<6}wLiEk1t!Qd
zKB;5=JVwFIllb|G+gMyO>_%56KfOm27oRcs!8>hMUFz!*KN<3k24_uAvT4@C0BYbJ>(C1?$&lj
z!?3?ynDHR$IQF~P<0c#Ee6WAC+1vhrNz)JXc7$B{cbTNF@EaL_MuxQDT=;3q-K!+$
zD!b!0?I-#@{A9>akE^#T4FO*N!FbRd>{FRXMubw_($#T_7uhRo_i!bh2lQS&SxxW!
z9G`9&bRPKeb6fUNJQmOt>Ogvj`aTV#{p^!{b^a*ep|j1BP6Kiq3-9k~!2R*7E-6aC
z@Xn)?bi_n@N8zOOP0kJMt0V67
zUFHWC9DI&Ey_0$IIrxWciwYUzG^P^^Nvm$K4#E6Hsj8?
zD9F8=K4M%z7V6tHe!;>fe6Q9eu(2_*j+4s{C>@nh&z+q1GFkb1J(s#=j=$m9dagiK
z*3#lvE$3#rzTe4R@ZYh^4qT|L=d$izs9$}to@?9xrRv<&dhT$r&n#cG^YgxO5&DA=
zZ&qzC(`xsFJJw~wYq4nq=U~@&rFwK7=l`f-fi>~~##Y&0+5NEYf;Z(Cu0@=cA-Y=-
zUdPpnQ+8eZRL=!%J*s96yZ@vfPYUag{FdMG@rx0+Tp4=7)pzC(?r@BbYL-5
z=XWO7a4{Z@nWETQu5;?4#H+okI62kpwdaZ}xer5ThPHIadlZdn9rj2*d
zmS0Uw&)ZgUiDCObo!C~z9Siqc>^#4gGq)YsVK?HTr&j(>*4o(r6%K1F&8g%PJN~s&
z(yQU7n>|lDiuMoq@~A8j?{F@Z?rEvU`;YMfzc-Jv!~5lRYPnufL92iFL7c0w
z=kH$R6Wo?1^%)5{tCih*dZyyNjnuLU-4ORFXDFMuLtOVErubDH;+_*Xg7JhD{%gpL
z7q4-ja>=}7bZ6Xy1-r|%^+r5(Gqk_HDfmUt17t_Sj$dz+R_)Y9UTfNhK`y8#)_D0r
zY206o4eYJ^0&2!7tiqbDf8m-AU_J;ghY{&FE#v2OCN
ztn@6#{h{_BhvYtxTb^|;cRJ^lL43Ir`*%>iX}1q0
z(0({4S8^&d6)Be6EGZR0^gmK*(z}_{k
z`PR2D$vv@X`ajrv^Jpx;Hhxqg85(F%hEfs@hz3cuqd|jaQE4D)ph>feCY6e$Xb>qw
zXrMWTXrh5qkJ6w?10F;1JksoZu6^In?|s*Iz3)0}opsLn>$TSR`{^3?zOQ}l>Dv3=
z&%XIO_vS@iKJP0fUAO0U_HE$%3vpZXd3>YPUf7S1i}UAkN8#py?Mcu1e1`*+e_y2v
zo_^!-bUv^6KYhUS)v2_%B|DRS#&LQ?TQk0%PJEik_dz1IOy=`uc>KJvl+UoT#r*xv
zudQ47yer!*+|O}BeG-pf?!EuQ(IQv){hw2@g)kqd4CV82Tzf1}VOKjz3O`4@`tC0-
z*KiQxp_}_J@cEI2VA_Y?tas!2JXzH)fzz}9jNzDXGN0@HmCxmLOob5FL{CGM}4BE#<
zpwsBH0^lbh`XT-%I-U%~2VptiGDx6rHxca7q_7O~&J?C0T^@FYh~7ZO0c|=R+c2OH
zO~v&rz|ubNIX~XHKaXFua{akq`)If2?ILOFdanO-DZ&S&uMLKqkZyHWC{O>l|65+q
z>W_r?EN_bu-a@WBSJ0ofU>BzcbX|rAW87JSW&^zBiR~HWw=)XcCZLZO;aC#rb49{?
ztK`ODgWt3#*x+YQ6>#4aDiGEa1>{)#9=)3f$HgriZxVm6%CZ5(Z`5{7S88vGPc3fo
zbF_<0dHaeRV#4hRH|fgbz&=H8yqsxW7Vf2f6Fymp`!uE`a`}s;J_uB?Zx=U&?}HV^
z$o(jP4~$a8|2p;%w@3Z*eZh{!H+h_T;(<5c$Fm8-wk7mNwWJEgdfge_5cIV4bG%*G
zb)1w&^s!lKpX8tQ;L5=)wSe~;LD=DHq!GB7v!8IKP0rVN8pQlY`-UOpm_9Y*fx)s!0M{^
zp7i|EkgZOnucB9hz!B$ktIg+={uhrI5i}0&^$X(t_TlP6-f!0rnEr*^y;1El<-e%`
zZxo?touR4f3zWz1HCllABJ{k^2y@IIlU3e4Ohw$K^WMXH;2Dfxxh{(+esg<$ZXuO3
zs^0%Ioe%D;HrIR^=84YfYKg5dkKOT5yG3%#Udjt;z1Pd?@qyEz+mGyVhkkGN^_$me
zN8~@$rJ%LA~dusBnn#@&*~7qDTA@xIu-
zj^>4BzTZk{K2jchJfk;o$*^bA;h49S`h`Dm#y&-R>5|pHn3s1i{i3F(Uc)rE>hAn1
zle2qH54*m`{M=#unZk6;%kS5_O!I{tCH%uf=zqKlt@2X5+kM#9aqM{0+dLo271a33}H$9mNM!@cuKmNu?_6
z8Sia5Q5)}99B$a14~C!WZ_j-liuf~l%!{`hptrTWkudLie3^l``%d#hx
z2fb`3V*^+AXnV=2mZj}{Knnf*>)%DCB4(V)LOhR)HwcSTVxIUyK=Jb
zc*HFS_dVM;5%Ja<>qGBe9>F?YlD=3I@s&^5#+fR8Ygq27ZtHE%R5G(+%1w)WDp_<^
z#fcNzl`PV6*r`>?70iF@ys-NVD_C_~(*cezD_QF$N2foGuVljm+OE%NQpvOjef_J4
z->(Jql@F;Yt6+|&I%drrUB#lSO1*>HSE2qj%+F|7vC)5&T9|dLW>b5cez^OtnnfSW
zDsmrM&4zppx2ft@&DG^}9=-wkM@RK5SOWj<@KqFaDnIuny~jhcPR6nL}>#!0Pu9=%+?a
zmLJ3K-Ifiz|D*!@dE*AG|1k;e@K!|rnuCZZf7Ez)2*$opD8pcy4BMr4@~<@6L7DPH|e5g~c|Bn(&1Zw}`C&rEy&X4(
zEu{2Y8XxIM`-OW#)2%3%ZFQQkYs&V;=UUm7bGfelC@xp(xq{=H
z8O~LdzQ>AvJpc4d!nu{~-$%GS@xlt8?#zM;9&d$pp2qd6FWYl@@?|S7&(lY-VYgRn
zyru;DpGn~Ij2VG@oA=DEH*k>ICN
zxQFx+c+%B
zg6IZ%?Fjv&wEYOYgVhZ23sq2UpdIKla-hS^gmr+!J2TE#qMtv4f10kpg`T?@@2kcM
z{x(qG%K7t*0FW>8hC1{=ULpfvmM{ZJUjCZZ{IDf)p@-7{4}-$
zA$QHDg#~_pWlakOoQu{j;dI{O#ds3prd2}zfrs93`2vP;82qm4xIPQ|XEtIO0Y)#B
zae3ecT6iG-NPo)d18S2w-J<#k$FQa99GA7FmVo`iC}I6FI$GFg9#fsi<>AkJal702
zU&|uBs@IkyDPQZPgncxwbuQ~NhT_yucSiI5aND(+ICzHmVYL(m!u&VBT^P|`{s&ZP
zoV&Do1Gn?IS0CP=++R8JetO0_|RV
z?&TEFF$u%kKLT!O8?%kdbAEN1GJg-dv)?m%-{K
zHwz*@Y^JrA#!ZcuCZqZL+1VFTo`PSut+5L||6r=CKj}*rEMG_ZN#iQ&XXASq?xry4SAo_knUK7%<3s=JrG%y}_)DNWhdcCj}=zw|LKYe>G?L*2|
z&z9#St~I^4-cao|_8;$_kEFPL+U5?*6h|MiI(>)AW3^4w#trjl{n>8@jS=^ro>nm<
zA97=x(Ie>heJSr}tI)oQ|C}E7J3v32;^8XlT6*(I7e{=`QqGH4~1z)OtQxW&4m{s4mME(u0K0m&N
z=SP)vsq;a)LnAvpqu*!s_h@o!E}ql(V0kw~ycbTJJbU|f=*dT4>sJZB)UxE96#8i?
zmkqIn#DzEl%)moCeIniu
zTDa9QtT&F2^fJi7zAW=Dl;>IEJ+*7o+-IH8p7f4~C}F+e8(N$A6Y*E$kUvgWaUQEA
zP{*kQ-m72Io;1|~^=tC|z$P8UdB;uiQWLPwIM&r2#`|S=-VXuzY3pF|X;qG%Y
zb}Hk9x>RHGXY{`qPP-+5aWt;XB*?LW&XhjqcvyNbe=~wVbdw0($*?i){Cy
z9(qpeaO51y)wI6VH2Qs@n)jz3hc8#N58vl~IVi1WE}ve_sDhu47#;cGk9Er9;(Dv`
z(02;zan(Eu<3*#%JFH(}y*XjK(qjCcz+}Ru8dcbvHc=(|E!IQnC)RwghMhldjMe`H
zyK`$N%2ptKRfhj_Ej-V&>w&NIyT8ub6{Uw@ug&6-uBWh0G;6)nl;XU$zs7hvL+&)<
z$98(Z`P=B&fqO{ra6V>R9Ozz?dRlx3?r?kPn6#4pXAG*m0`BXo2Ypwbk
z?>}z8mA8XkT3c*~`l(cgr1jByE77&Kh~V2!;IF6wk2~$0{#>
znDG<&dP}eMxeR{E+)K~se8uo}4+E+H{`Q!6Cm!uw^?O^NPN3(#JhYAaYedB8Fh``j
zGiS;idLJ_S%(^Y~{&?5;v4Qmcm7nKUt)%rpXRRHx4uMYS**tnEC
zG~R9=F~*hl8zv`TSxr^W`Z}F}@Ih^EH9{jcHsT*xnQ$K*6~CY)w4*ZF#opR+?X~e>;v(Cu1Gy
zdlX+bp#2poAOBB1+l~!4>{qOv&*zPTOm}-qzqDUU58{t?tK$1ohXPOCp!~NFWRnOJ
z5B78?`I#%_e7;P65Y2zL*exlK&%0|MW#NZp7@sVg;D8wJ6ZKpsp>#zhZ+N_YPcMka
z8|w~cUnRYU7tY^E{`8I?0tCbxejj*TT`=nje=n;wWy3|{pK5r`?YN><%j5gQ)8%|V
zUgIw8x5XU0Qb>9O`wijpNxBp`A>MJs>woZP`%dTc^OxmmJpbl-c-;?v@be=a_Zc4J
z`kjsl_qclIn)32lJzk867W+>hv$_22z^r?yhl&#dyMBFdR^E@eiMsu4)Q;AoE-V39Twuh`U(mZaW1R?#}0tu{e-x*SsX~v
zmXU&7x?nk{&*Ky)@q5uWIPy98QOKvWrJz4Q6vq-EKet1Oi+_bd4SHJw1bIxo5T`zl
zNA-b!l)@pP-cmt6d#0d2aF9Tk>!G@k->L+GPFN?%M~iW=Dpqfl-gXGw0=+RfwNCVJ
z3L8Nqc;TpyRU7exhy>q`)Iig%wnR6OTR;z?We<>N3*=(Rc{ZfS|M1>OTnC_W;t%6F
zWdk{#DkR!ar0Ydgu#68Hb->k9k=
z0;ETf#MAR@KR`d=1wtG)Sm2Kl{o7sCyjJ{^+kQ
zJn!2`LGC$3&@bL6(DoVVrpRy09pO0#sXK$-C$`7WO9lG!4ME>;5w@)%zqk1nf1Y;3
zG>&Eiosj|Bw~-nj*PeGUe?#>5yMA2X&tx&zn|6LT3`9F@>VX1bJhB|yf^cYCVI3Y?
z`HbtYi8JN;8`mN%hkp96L_VGbqzU$VcnIsHz3vFhAur7q^50pvpQqCpD2x}HzlC*r
zxsE_DwWk`zc<`nL8UlXTx$zzzjD4^1mktrXuczfqq9^=a+nm-#O%|L_qj}&-@M3Of
z=cXH>ka7
zDMtf_(|%le=tvGD=M8(jsKLHncrITD-J0wAh1L-&Z?aEOewsBpP3ieM
zJ-rW+{)EF3od_kRI_)X{;maLH)4Y~3VJ+X!x_ilN4e5Is2e}fa|DIe$`ld0PZ~*}B
z+d9&-F~vQ8m!3|KMSK#_rQYW(_yr+nuFV3r>CsZ>FxE#?-|ksW>lpi1n{8-47aaO$
z?;xPgGM$no;P1F}b@Uy?Nj=Z1pQZBc4GZ<7{Y1Hc$rOqM{4!q7p?yQI>p31Spj$4R
zz<*biTXt#S2k?te{Kg3{><2&C)tT&D-`a9V6lj}TRa+|0RGrj8l+WgeJDlnLio=-R
zJg#U_;2cNsWuD6{gARxj7sz}j=F>Tu#MuKsA}&+u;?{RGX483?H9wZ>eg}O){$)Z28_7Md(K%t~{#`qd1GbsEB)}8=ENfN!F|g~q=Th4!
zV5jr*o?iz_eOK28L%y-Qa?8cgKRCSeSi1MEAf+f)74hxrRyT6VPG9q_jj4UKcRp21
z=dGuXuAW2jYeTTXV5%3>fyxu9Eq2@J{z-(&@hbYG$E4u1~4bZLEOtZOo_-6;J9e&)%}
z^LB%8xj!K-3mCC(#JmddZ&qx6qz8Lfs@b-r=O(TlYewzzm{Ch^!&8!k_6gM>9yWCsu%cYev}NPegA&bJ|N=2de_`+3dtjen*AmJMQ
z%fd7EP=B`9o_3JRSwFHRP9|di?)*<@8b=ppTi^!}I4|A2i6`0NapUe#!j9U;t4ME(
zT9f&t_ic3zW?-Dl)yc+AD)#BSju}jHd0{YScDxVS-L;tVF+Q?l2BqIWz5^yW_$MJ0
z7yV+tJFI&O>HTyvDmUDtvFOek0nOMdfo>;Q%}(5oMR=KDVvSE}@L}{F%Kz1_k(iJ$
zKL`0@Vu#=Vm?I!X9Huh-A>p;W`OOTJT3zj|LpQdZJLHE5Zd{1Rd9ImAXUvq3*K0kb462#>NJ=HlKV?Y4|)(fG+
z`PDZb)m+}TLlZ7v)mw;bTY6T%AiZYGJ8}O6=?>!lxI9#tuPb6GvOzoBJcG-15{3Ee
z4_1G~clSawLHdhnnC3t~m?peGSZIc63iPabLR>B#iD?A1)IoS|KnQ-WgmfNJB|Lrq
z)xx~~PAbgT25W`5$9#?Oe5-pDX(F!cDab$Z4}W0)?5v>o%g&$6o39n>>r;fVKioS{
zhyxabQEo`f70B_i2d@7~eEw~5
z-#n5B?%4Me<@__9(7#T}fOfA1d`QK{^8!WwQ4!w}8+=}D=u;~onuph*c{mQ*UF7pH
z8GKHHuRxw1_|yw1J$Z$muaGNHp~sggkds&bT~9$P^c1Mz^Wy(oItBkHSHyR+MgI&E
zX=Q3AuxEhiC-ti&z`Vr`!u%oXnTvEgk>=!atl1
z{qtgaT+)QUfB8dE4lV{?v5uluIN4jbM9?>!D$vRFngV+BaT|{+y))sC&`(lCJ2S-g6sIcan_^U@d>#Y{_?=fA;tNs#wy0kzuIrle3PyY*
z$|>5X^w&gsy4Zh~cMM|?33^h2QBb2k9Kb9$wAM&LE3h$}*7;t^B^8HZ2
ze1WgJ5&vF)p3PB#+ax;`8Z~suizk{w~Om
zQzIJX${y*L>P+IOJBhwDh4W-fBl!9(X1W%S8>}t%uc7(;!v3*yDNZ#R
z-#vl$OUtEO^(YP*)4a_!iqBjP=Xz59&qvM11{KyX*|tekKEDl{=TN?y6IvTmylw9r
z!q=BudThc(kNvHUyCzb)NdLaSi5^_}+nnN_QJyooUHR{8S5o?ufp_`-VW0EIMi71O
z_6}8gj@}@@my}+6tKoFQq_|$*Bwy9T06hce6r2X^CbS*8r~GLp+wm
z0vn3w4Z8rmuHAqALty-Lud%cac%J&+pW;%@IXf$ewo*4aK;@0MXqiRjy7;R5W3qGV
z%ZiUwpGGG8|I%~5&s@-r?8+*gn$4eE*6ulB2cMYvWZ%sN6&vV%WIMHGEh$~c{k=`8
z+(joBM3J8Ar*|rpzx}%hUkEiT_Ov2=m2B!r`HpAjw>^TszW<7iw9mV6a_j{i@NaC)
zb2kU-wmx`_;`?X~rH7R6a8Tei%J*=?>uP$gvD83^?A|;6j5^Uz#~i#ua^H}%9)v1}
zZ_krG8h>8SBDv$TidfR$Fx=Cc_;$fV
zHh}W$m;SCVr7H;Q*NpPBmgYUB{B$38OzR8zsg_}3t$~TR5=K+I8uPjIGsX8uo-Z{a
z|G#ZMyPDGH%GXrWx+t^xxN=&rU+=LV8!oUrv&A8@XHCwMa@udOxij_{(QTU>@p2n3
zntg)k3hx|mLTBk^es6?LvjL@)erC?DG*b>~Q}(CL?@b1^TnDEc+$n^sXo
zg>%{cD3HWH_*Eh9UTY-Gv)3-t1c!e5MwnNhZ58IVGmixR{KpikVP7%?$0dMMo(So$
z)e3UeePDxsVwNxuE-j_V5bNtaA#Qib74lV%5cHnrQNWDzXQwFep?ja;W?+jFL0)Aj
z(7_)mFvEG0Hv+xZN06Ud1~&MUy9l(GrjWl9EoYGLhP8s+p9ga2-xU5senTkVnEic?-RZ@>R|k<6^3Tq%Y7WvsC
zJ|;H!%|v-NW?!P|_C!jLL5|aRMI0x_hZHT4o|Bl*Ad%)_G194v=`m^Xe5e*l|E)-K
zfF4C3#J3XhgDB@_fX~@Lh2A|;uHbXGK+eN*3K`ZiA=qc7YU3muRmtuV@^cDFkkc;6L#ropE
z=_&MSwNCz0_({S4C-P>AeE*w&6y;E?1Jgx24{`@!-H*|b>bvh};dy~#J6Aj>Pqd>S
zXVAG5L2ngbNaFjKq8!08{%}4`q}@eYEevs()86H;`8na#ALwb
ziuTt*r0Fdt>9yw#5AlSJpqJW1;JfS;#x>VQ&a=7*^UKD3VO-8UKm!r>Ap`Y!zS7o3oK_tmJZG&7wM4u>I7=0W{XyfGoHnt<
zYXI=?4F?ner%M{>B}CGJkMkT
z48lDKv2i^8hII>g`i%Z{Tt6h-0weEN)~B-tmp@C4
z>_=tZGb8*qAgl%Po8K!uN_1t{yfuXBGp)G2)0KjGoU&?L!wcd|E+3T;PHZ~Pn^0y_
z#_crIzK0fu{g@0rp6+ji@Ez+nIXjyXf1oCfxY)Ni9gGUVdy%%6xIZmgOF7+Y@OLy6
z_^rhZqtyP&OM*?c{?OpZiyRUj&9|ocjrUUqJXfO&u*6?}LVElo5J9qc3k@U)J;_Mm(%v`W9nK
z#P1_7jppm)Ins9A&S^Qq_hpaRE#&r2Z}N)QvuEou80z4+C_`>%eONR`0-VG9nataJ
zK$~N{yec2m(c#cfo$t0IJG0*|NTYcC=|J}=(jU;?43P`$*#rPw%QK#LqqQh5HUv0+KjCQR4;2k(;jYeA@On%`tqi@Lv0w
z$1g6wvHlCUt3VCQGT2)gC!8yux1ogFx&OUC=ik0a10vdev)vrEz7OT;m((FS_&pX0
z<7QEsFy0ggQ-Fx~5rb)_!ta(Y_u$xnJ=CGURQwKKhgh&7x9bIWf!{V)u(#S7#RY9i
z*G>XMbcE-4V>M3t0c!djxAYLk?WZ(Mfxk{X-yW-iWJqUwf+n)B?Cu3LJKzIjVVu0z
zpC&Y{&okjp=uMa?$j2`d#=&J`e5pMZYS2@^Dm;JhPjp}CAGa0wWo82QOA(&8ypurN
zju-3ZYSj7^56S)D1%cZzH4y}#o2+9Wf3o(ij
z+EcNC3VAzGPQ8TmwMh*5C9%A(L^%dGE?*>KGn&4kKZM*3tfT5hx-cR^V-VxGQnY)y
zNb~Rlde$PHLCHXSQS|{aICE4y4~t?>b2lL0NHJY^5i!VdId=!-oJ2lOqWm#u{Oi0C
zXG4#>5qia<{D_FWIFNHUL5{%?e`qHnDroLr=yCTz&c#4Qdc|}88)6W}-@oGW|E9;A
z&418SJWr9%=GuhhN8yND`_FS#Z-@9)^{n=9++(?O!n)b>VW(=4=cjf!|7>6-G3$0#
z#@f1W9^Rs&hBd2GnLECxvS&m>o27`;+=fkAX^r-Mb9}?Y+t>$nOZ~OT
z`2$-ST%u+stz>rTS(BTWVP7!%O0pyF&3=%7KBNwM+kA~LMml|FtvW8tbP20u62}F)
z8fMs!YqGHHfFD#z1$=0JkAAw
zE`4D5A`0hNFYg~dZ*>LJz1rkO_sf;cu|8p4lay-KbH~*BR)siEUo-Sif9Naq&njEy
zT*=1PtT5`lu?qM8I>xwb;hg4LgIIUmk2zqTQBl$-?5pg)GWuI^C2Ki%^mX&D6>Ol~
zXvc$AeaV|K6>PNe-5o)=|6)v&cTR4exM$q`WW3*`
zGM3TZ-
z^Te~y_zv*E$;S)JS+%O!#ST}cY>#GOucyt*ne~#_R*lD&vD-CjO=NM!tn%rt36*K3
z?E0pi&KF*nvs)dmoZjzVhW*jy%zsfS>w3{~E2}MK^Yt@r-R#TQU+=$1Gx2@Dzgs?h
zUgCs%IcBt4(bKDvg=Gfr3DPWMZwE)d@lr2i9j0HH?Cb>lHRLk6Z8^KXdH>45s&Y2B
zX`lG6E@iAXbZehp1In09?%u(DEXvs6P40by>PuO#&06l?o|m#wPPW^}dX%y8Q5~9@
znU&$5s?$GSk1k^kC-1(Rw6Ba!SoOkrO>`ME+p1(-0Q+`be_^}yL@Apvb^0^)<)utH
z)^qgWE~RW)*4l$_v`X0v!>y5QMJ3y{wSIGcWd-Zo=K1}%xUb~w_^1Jo@f}3n)P$&R
z_&!Z`X`xMAO*!jhaAMl$qvb5r`Fp_o-|$C~`^3)+%GjOS>CZl({;WT_Wh6*SS!%`6
zGiAWa6IthrD@vL5hYR0J+LW;Ylgdu*?vC##ZaTPZSYjFL*Dg7~=wKNe={>W{{jFsz
zJ8Hs_YrbWyB(i>COD~i!x^cH&KR)C8j>k(qAAM#AT$|pztX9ev-t*IVJG7K(M6{U~
zJF67;dhfd0)w&$tNi9BkFAepcGHi9kpfXmjw?Mf;uZ-=u-2CCfj%7@#;&9dJzVM&j
zsn}7i%2;OL@x}4KOIZ)8;mScKWz4?UoKM;)$GAd+tD&$nmjyp~W>>}x_M67^o`(9l
zUAH+2b_GW#CZx*=0XcfhTTJ=-#N;?+@Q?3?B>zsezH?C=4geU5fu`PKUAJo7RZ|52%N
zO+y(geDN~9+nsV&6!ZJq>p5kt_k>XcD`4M<(GM+4oy*vy3#+{7keC;^=uYJrKN+R0M7?gD&)rhC
zJw0Ho$Nf^4v-O2*6x!!mUDNuod|ZWZ(q{FRuNBPwoXd!<%PUx2hs`prKT>9Yrs-(s
zS}F59IQH%9Z&Eh!VrWP~PB{x>H$v^=%GvX`5^K+la&{s3n(FNOa;Cn)s7tAql&NX9
ziOO17#;&Gntv0_}#!fx|I^T9&IV)Ige_kuAoSi(xvOZOoGkJ%`DQVlvS>C5fi~5}a
z-+7!)>y&bqQ!;GYJPF$GH4Vd;xOXRL(C5&zuS!{MZvXm-+)~!;X0~O{&r-Hl|M}^5
z8_HO#ShdF)m&#awd!PHMu=nQ`iMA|L%4&SJB}{aZvhV@DPAnfGWd{lejred<%08TE
z`QkxSDcew+UcaNUl+AKY2|D|@oE^Jy;KS-IxIb`!%|ajla&|d8WKurv9W0L@`?Nq$
z%6?C&T~Rht%Hr4S-TGoG#dp_Nxa~tesnhGiX3NW&x96W`Z<|Zm^S*y3P4<+s79o!w
zBqvMR@*|^CuEa=L$beJ7n;t7;U(JFee-@OXe%4=0t0-eynx9v$btq>6%kq2e+fa^s
z^y2lSuHjyx=gVv7D@mD_V>`DQ@LQ#ukyliIDeH7~iZsVf%DTOpvQZcI9&5N~K5Sk&
zdzfq2+q0mY&DrjF(#THA{gXE)r
zZFeszTf4qzGy54*rr9Lo`Q{`kYg5SQd>32RDQ`K9_dS}%yENLcX>GQwe_|^`7
z+xJU5U{g73mv~vl0QL3FbRW-8hG
zs82n;q7*m%3RbphQgi$1pj}5hP7aW=3;l)$KgIYLJY-ga?}!Sf<#Q`~+M)^;Iev1(
zwT@DDYumXZZPfqI(4O*(yWx++%`BperL6V6yn6$1A5)__ksG1{%h`t5Xq$a^%h`&p
zv4u^$Nm}7G4Y+m{?uwtCF@vyAntgMl=%j~GYo0K{ZhMzj-0uqf(@TC
zSX~9*%jm3ec#WoYCA%D}xovw=C2MTtcQX8aCEKLZL#e$oz=q?*v%_bcho!vRSk`2LB^xws#lwROtJv1$WupyVRIyLhq1vA`s@cWZD{3{Kl`Ks)o=dOSeprYI^{hp@%`U8lhunW*}jqI-p_b~a=IE#X9KF(p}o=T
zb5~ZeET^lr*;&9C-jfeZZP!|*8u3>5y^Zjlyoi`-BR=%{
z!lpzxRgS{%ziiu?EY(JQcRRYzOP`7=*6fHpzGeSvCfy|sFTy#LPXt&xX0o81Q^N9AJtgw&Uc1>Xu-2Y}*xbW#y*m*qd
z(}t+!_`S97z-X&{ycccm+tLB|R-Y4k>hdk_)nrv)X{u1_dGQ9m@P7PZy#@5)YLoM@r
z`ZS@wT`ij$;Nw@-ua=!Uuy$JU;Tm?c??u;+c{QxWYmcmdbq%xW)b`dZm0A{SbYsCx
zt6J7*ds_5^c{OZ(M*B6M>uQ*^Q(=KLxrX`W9b7!Sw1&NNNsl}Jr-seHb0TM_8shZp
z+E*@^*RZCOnpfK|sbPULrp~vBtzj*C-cm8XU&FjIFSTw8Ke*q&I8wg^>j>ptXMLO2
zuqmO(esvmOgY#B<`Yv9Idr6K?i;V4v=O+(ccfhBXIWFCQ;(k*Z>u%$Jb$5FiTlZ!6
zR_RyVJJM~wS3b&RQWAA0V^l4Rmuoy4aH^I)$r#uvP+!Jgev2F!WiMk7Y?pWY-2gk1
zD%H1i!@VYB4^3D&y_Q8N=TFu@QOiE0#EnsYR?F`FE^IgBZVf9Na_d03QY|xcSM9pUzwQTF-qlp*Z)UwP=-;4`Z)v%-8%@UL1aqo%$^LdAhYFM%N+lc;6YuWBw
zndcr1t7Yx?ZQYg7SjOUuZo2mxD`Pi?wYq%oxQtDkyWj7Xg^X!-NW~YEWK845nN?Gq
zWGua3(xr#cv%EO2tW>A>21T?j|no?STpMJlpGmbJVSq$
ztXjsD)!I*&!9TrpFUU3nw>x}nRJ=*X`g}>L3|uN>UOv6&`%cHbgT8*_)6dtk6VrA!
zU7B6XjGN8-uomBMn)atro38NpE6WzcXIIJCj+aUH8ej1H+@Rh4UTsEuy}fUgFTRVp
zW=Foh9=_-Fqwv<;NUS^eKOGuq5BrKY)VprlDPxZE=_lVGl(8z??UQSr;lExxehhCS
zWAl5|T<>2ZV`n?88kD3aXC6%_H&^{5V}_5Vxlih4?BMZ*M-HmWnbVZ#w!^$+Ozqp}
z9OIEPHpTMqoj-GAETF?-yE#cRHs|8y9@S|wHa^03|2h>pTYN9%g<2yy`Zm!kh~@{kO`I{0<9n(?LX
zIB##(($XRo185;RIth{OU^XryFI&8i0{Wg8-C#BWf|VnhSYt!ih3?>E47NlJwu(G
z22YDdz4qVe{rtL&)fBvP8)GeJ>t1GG2<#ze16fIV3w1dgIeFdFEfq5Mq2u3{v&&?x
z$(jYKZF}N(7e5ZPY>Rbx-P14i9W`a_Q%Qws8-qF)9#gn8?yj5#^|Q3~HLhjHyX>FX
zM_0z$YZT^Bz9na}wU2*mtJbkg8oht`Wa2rO$9DdlBWK#?!)&H}lCwP*+NHT3m9w0~
z!;5k@q8#e$GaC}+?4aWdr8(DdFVZ>#i_%=QkCHuWXLgjc$me?J>i5W*f8I}t%?W(3
z%P(ZJ(H9vDNm;&jRA<~u7*`lF0pUh>A=aDWWo6TLq-sz8o*^S8U_7dd-x=~5sutYfpj
zytKG%QO91Jt;&3^Q^!6P*0<{%C1>{aoqTr0%2|S}W1A_g@QFs_d7=&SM#`{ia+
ztZ!K3yh{EC1AqLkSvz%9H{CXMtZ2g0Ni$*J?2w;ZYS7Qln67KPY)dWcye72p>&qIp
zG|gni5ZLp*|=+A=fG1d+XSt@BZ7u56M}LeNkxa
zdieM5%X5BHGmUhTtp?6jz~!#z2^Z~N%mAf#XF
z_0aBTO$}R{IM`$$#`S{(3dam8uupRko8UnTI{LWyP>a~9g^(uexWaCA%>qv
zYN^z*cI_AaG5aoOzH1J?J&%6n;x_kebG=%YcWm&roD%T*&kMpc6rqALVd=0IXn5IM^+))!KSm@t0$uV
zwjKJKaIKq+{p=O26A3??mrq(hMj~fl6EALfiS`pxurh8K?gRV!s!^|bD{I*I*aKJO
z12Db^hd+11c=2sY%RU9CWvu_*+LWQ|F&=O6HA}+pYg#G^Z@M=<5_TAe2=IUY-$Y3`
zS<$FvBhA$w%NKggS>&~1wsjfb?uAw(B{lrV|H0EVQs@g8&?rR-$^Q>3Lr(hIMobAW
z)bKY%sgI^op9a1Ut%Q#+HRbZgjT`f9n~M1=4t*h{Gu3Q0T5LJmqwtFbLTf$ROh{lCVdy-G^L&h5YcFFff#`?I%EzG%q*
zfBrl+OW14wf9g;2rNW=f@bCX`{P~~#Wfdg62T)S_&+rH
zv=ULFr@&q!-$q31f9P=%M3YNdPO-D^k>eH
zuvMbIzK9D%+#t%&@rsbJUm`t1l$(gCu=}G(&l6Gc98RL16m~V@21{72nBQv=euPgr)HUOV~;=T~qNnvv@^HSVdDxAYt3NBPGm`I~sP0
zPm)7SnK72yipDn
zT*3^F3-lZ>2Tqp-QV)`_-){wZ^d?fA3k(CQ_m8*h&Uzvb+jcKHi
zus=*d-`;|KJ&p-j(1RQ#Vg6QAIIeE-p5qahDkw;povuHpRqp=b^umHps0j(aOHB@z
zu;J>$bJ~7$=CrRl)wG1Ioe&Ozgtd1So>TMMk;^|6YjYe@ev0e$*)l#zbldzdb-BB3@uB-wBM=m=la%t;^kcaxep44
z{y6nM94=v>8x2KEld#vPo^iX~zb5i@zpmccP5f)Qo;wLObu#z$0XiC%F0uwro>o;v
z{MElcJ|uozVxXrb@U&7S&$nO6MsBz4&lv76{oVnb-Z4?1m-EG&Ok+yFFCY^QNWz9*
zxW?^Ul}U6g*u&;dKtcJuN&?7hMB8DH$n{V+k}97pY~%0J0I
zua_x2|M$nF7-A)?e-SlY3ENdKxzh`HYITwp=>>1`;`MMX&V3@u?cTWN5l%=n;CA0?
zr^)SZ7$MZdm1r$a{|=`ST*6XxmU4T>4N2toul^19g6}lXm*YJrLEm|Q6}NZw??v1W
zY(sMS-*fMIx}tU4IKBK`A}_~_G#_3cYbOrm`uXYWxm;bE$n82-{wIo_^I=S6S3)Jd
z`ksVODr*N4dZ*0b`n?`P9rUXqLcg2WKr?`ZIXDRXe$<^n&v;YJ?Qndl!|8?$L2sJ2
zARoC(z(H;gx!l*#gX4-tcs$atO`;h}!fr(ZfO35)FK=hH*F4{cBi3+!>V6s^CCud=
z&8QNVpc2OWiEXGmFZYm%a;~4Ei{=M;W0h1{rvfM2=@==+ospr17e<46+#;6eorq2%
zpOes2#6Ol|Ib+3grHJ}LqCN&c{Gp$V?V|-%187eIpuec^CzhAH3w)j(kgNS0D){JB
zJUwqd|3M#v2$wsEsL3q6{xU7k)DS||0e&Zo`U{w$eTOT^KJ!Hluu!g0u_1+KPuv1
zg0yqn4cX@tH}SNNbeQpsUq#i
z4VJJHL<6MvTO3bk@`6j)Fw7>z&zLM=?I8hE#c`u0k&x?&`L*F4LBevz2>ebigzT;N1{BJHvo(?qFv9zh5UkM3phwyKzotax=jy|u)BN#Az`hS
zlfxzK_##2?j#!T176Sj>E;3ZYniHYdU%`M8e9N4}>8SR`~us=SR6wi$nQ?sHI6*
zfk7jV3DeWyQOuJSzFgjV(lX9hN&e31Q|dyw8s0C0!4mc}ehAOsRb9&EYZfo(`nij#
zMPgkfPeK7CY^!Dhe}3IUY7m%L^?fh4`4o?a@9)ja{W!1>mpeq*btAdF?-#T@^A>~(iYbU052{Vr$YH0&hUY3NR8*%ySYnnZP@Bc)-A^Go~
zE?xQqpI@^Z-W|BRyi=4Bu)p1QJ0)NrrDic>ff4skx|#s(XUBxpTEo8%SMKZpy=C+<
zRkE)|@%wF*UqsnH49ST5!e5|=OIX#TV`z91mYdMM5&7ZdkV99=-ji`X(DTqg(pokq
z9J!~*SxVn%qOU*YXSRCEO)8(KbHGXBUs~8=3+e6Mbn-l*$(`R009mElKyeSDIrwepf!X0b^}`9+<%1f
z&Gfy&*D;wE8+g0a=-Uk)T*6`(o#5@Yo%QK&l>e<4xvs=#SxY#7R{znwJ$&l@mFtJD
zbsbA`%jp*@NdL#|6TE#jSXZGVNZ1DD0z|wL)=_2iIkL0g+Xv{dh(F%HK!<|=lPY;W
zc1_-_AiXt?UA9npFT})rp!&)Q-rtz?K0F)6%R4^cAhu{E>{-inZqE+8Z`}Ta`&#h+
znUsO%1-VLb0jIAt&gA_?xx7E`Up5CHaenxTHe9dic3+Myu1x3cy*M$1_oH3LDMG*3
z6!rzwe428-nJjx3<$rLIy&YjA`4!%ejyTWY{m^|OZFpe6Cq`~e<=pimJB;X#*LBAe
ze!Q=akVV3*M)bv0FJWB`?s9v6s_F9l2X~{H8114j+zjkJSQt-Y9tz{?v;bk89@`GZ
zhTLCQplvc}qCk82JeT7nM~WCEESjJ0&Y1Y*uB~d?uUHN
zB+3Z=?0_(iYa9~NJ9on{3x1KV;LlH^g#5Sd5%TFbP?+BeRR!9Hl0jca*#P&6_I)9D
zgYF{YTB;_{n?*gvI%=D!zh;ske}Pen^z3N51L}$CbH#M#PXxQI#C3D|XhE+Sy_EEK
zk$FHz9=JOb`P&^cMyX{9&Cartd7$
z3cD1~;p)(%S|xq14pfwn7auei0~IuPD`*EXor15>Q{?+lw3E9Z`snQV`vx(+>x$?u
zVy@_a6|uhz5@|&{3KMB7UV)hZ#Qqf}@@e!Y`*Os7ks!9SY#)K2F2=zZc|kCai}a*P
zg4{b*z<%6ui0^m<#C{NWEcOdq3G$np=z)mOc*RKATTx#b(Hil;%@R=8oeYw&C88Ze
zcmt8J$J{{@wwy0?5NA9W#_>l%Lb{DNh5U1B1iCzm9D;d73DXWRMjSULh~vZI`GQ_r
zJE{n@r)9!(zl8{V?OXv%1_=2coFt6fuJfpcWB-T3Y2eLRA${dmswl*J-N-=_c7LWm
zN7p@6!x(S0h573Eoi3bindQo{V!{uOUG{l#G@nf^N5a&)3;DI$zns(WSS22Wc&b3y
zubgTr=x=jHI03%WAXm4%1^Egrd!-GFxS;Lmk!!CP<@$%iAGsA%NW~^|!
zL}=P;GFm$Jrw?k3Aza_os~KVckia8^-nH>#DIPhd`*H=vj}C92&!;%b*rL4>@%J1$
zQb_c0C!0>4fm79nb|AjT=kW2wPmN8(Rw&+sjdATwwBz#BL54u%JsKKBU)t}zfb>7=
zB&Skbd++1MP>O4#_F*vM+|f&xmvejnEE7a$K8{1TVUO5&1i~G-OsgH
zMsfQBqoW5XejWTekn4r-8g;r2=*?yE-+v%3zvI#*pcCkY`PcGYfJQTu`g{d{Y@fIT
z_MjJ;?vL2sSHhB(&b-|ew1>TgK)D6I=-EH>_2ky~SdG!qGa_<;pp@sJymAWTq0lS@d`6B_p|BX4f
zPXa?-P9`4!zuAP6^1Z-Yw_C273;yRRVZC)gA0F4zbYefuOWL+>BW)4io>*WX+za#Y
z#wOpkK+oawwyOPhcz^i9znc6McU;|e9_91eU1u85ym);j`7`ob*1WF30L{K_%z?Il
z-Brjhmy&43jQ0h4K^Mrc=dAKEbYcCr=H6-Y|DdGs6g}X9@ey6!fQ{O^!8
z4RgHTaeI|V{&0Jj?jMDE2&&49D7Heq7*tnNeQSP;$ru2%dvB~s?IdaZl}IY*rCqNu
zgd={F_)aGOFKw9pmh7)`S2<1VzIVwB`jgyF<2M3NwC6z{3uxWeD8}~&)q78yNiE3_
z4qJ2d$v?-=O@B=J7PzK{y@Nexhc<&Kj)XTna6
zl*-5>z=vo4E{*|zrp7WAO6Rw`Rh=F1yniHK%t+YuX=xaU(0;S}Y#{tK=cPK~lFT1Y
zgpq0)FUj9e=ezL!T4}C_fgAJ1yDJCuZVuBxoUSoI7*w+?jKpIWr%GTr*I{f%x}1oJaRLCLJX_y{pev>NuKOnz}j&z=KZ(!vzb0r?&7(IybbdmYlqX~RGh^^=2)b(0Z)ax52^+X+$j<2)k_X&T^|NEB21vAi9zjDRD5
z(8p7}sae#=Jfs55_iSE2J&iA~+KpnA@1(E{9I<&k!>xV$DSt!(Dwk!6$QrKAme(_h
z+r`QgB8eH+a1A*;&q#omAvxkD%m%)!G-sHvK|}-lEeA|r;G6@Por|y8Qea}tL;pG4m97+7$l?|bL>y~pmp_CXSu{*%7>bJC8N`}ftKBoXcH>}zLV
zUpqH9^5wJp*qiA0^q;w!-3cRCz>I?*J-e0Fe|kBdmEYh0O`P(e2j?2Cr?`Dfpn{DE^4R|!WBpp#54%UaVO8QG
zAByLrpLF<=UcOz6fb_EcZ^lx)G7evHAcB1T@|mR*ff<65%_R3X`3aAQ`E~EBA7Og5
zNjtTZc|N0w_?Fg(s!r1TB9b?e+D-6|OQ-s~LbEPY{jo9oUJ@UA@FVmg2fQS)Y7y}h
z5zpxj#5X;1e|i!h+axt*;zRJSQp1hL(vV|=3bufl<2}se^I~eXX%eskmufMy`EDpdoBM-MZem#qSnOS~^H
z?E;?{Jt(i0=-0Uq*!Wr2b`bj8y2D{?)FSL-wr4kaLFRsctJn;-krDWOe$I@=
z4&d!qhl@Y$8jW+qAp*6+dEl%4e#tvP&tm&~)ef{bIo15fcIcfma?_@M4m@*kiJc02
z-gLilL;u!3N_(r(Ue)^6ilk}izphJR=LEdhii*}m^c(l$z_j(KpDYrXenA8K(Pbjz
zitxO|m&wz|ufTc!?5Qhcmm*%rc6a-zfG6(v6Sqb`N>F9o>vFo3tzsc
z3+?UI_4}wb7k1>BYk1@RG`bp|Oa=dM>^Y?720Pb}n|qDs$*Ez}E|0_b_G}jmNP^vQ
zFHbJ=X)LQs_`fMbiqN+M63%Rx#j~fLrB}s`s)Kb`;-Q^hFKp!ExUDQ#Ar^b?o1Z
zxlZu2AMVwAVm!@)MT4fn9-G^4k1Ao`kf(2+w4=SBrBV*tAn%H>)cb>WXHT>YY=T{e
zXB9N|qyAP{s>yEDt2*YY6R$1|X<23sJ0Iojm81R5O6mNl+0fVgI?Le(^#7c6$)ok7
zS~py&0yw^0Y4{-MZM`QYtBLw`$3W{U4dLfvg#VRcTLio+
z7jT07PgIbZfW|9%UnK$r)?KqOC?ELPzScxSr^kM!^j_X6TF&IB#QK{dQHAsA#7m%@
z*f*Zt<4n)jv_58WWl#tmh+}^)_K?Mk87U$R#|$iH@n@j9FpD>Bk9M%~uk#18a?PB%
zES}8J@nN#dO+{v>M~N|uHx5h7SpCrzQ&{_}f6@&BtS>cfP&wfBnYb}l!<`-dkd1?4
zPYG+MMz;k-0`cSCni{gRVe%wLlHDGsFuxuqp^diSQeFe-Zti_^jn_p)(u
zQR4Z@xZh|7^=Xtg1UmHXqh65tkSqT2<&#$OJjLo6uOD%OFF${l*Yg$Sak;J>d(Nuw
zKEtpBy#DURy#6E1dTRH~z(9uY9}|%Te`3TI`H>YE2FS@j`SFY_Tg%>4tb9K!k1pZI
z;kOUA>c8~qT^h{W
zoz^!VetrB`7_5r)UJ^Y8t~Inb5%ROF}5#z`I&^0>RfnVIu%O;+Dee)DvuDZ*A|Ed$1
z?$G;Sz{|^#RV6vDkvhga*~8acjT^({qeE%J)ZqVV&}KOD6~7**vI~pg2mYFDJvuop
zVfAf&`-5G`GSYLAb?X_**7EN=%8Iv3s1(Z$^*IH+-Kl4WFgf_gEEoj-d78fmZtJ;~
zm3L*l#e?8qwslPQ6CVzP5XXoAVDdL%FNWONjZ9Dck_x5>@`Mr2j5l{n^89=F9Bai3!2ZV`7$~g!Ng|jbSm)<7wi9|QcVHmE
zn|*Guew6QiW`1$-o;Ai)p0NL1DWUH5m&`x6$n9cw%>D46A(cP2e2pcCxUufaHuC$j
zc^%lfVVu=7ji^4aMuOR)9bdbM$~{{ABZ;@i@5!G|a`fkMKS^)!AERj`{|RmuC4cPI
z^tNd^&^)8ll;W1r%OTgO{$zjiE2Q5XwoRAHUGy!sQ5>pP@_Io1FAj-i{bnp2jsd55
zlZZ1y#QV>gYlzoo_CEC|y_U3jnBh1-5pHRv=l2S3{+th7(qaA23uyfD=B|an@Xxbz
zh(`>4b3UA&oAI!9r3-MUR>!x!i1V6;f5V9vq$Wxzj3K>iflt&)>=Y>uc
zc-7zPgbEqRT08$5QoKBvl0Sv|Hw(TP+z+Vs`=IA-VE&Y~*#m)o(VJ^l1CRClSx5XM
z#JuzNM&K|RyR)ysi{?wbcT0w>FEi%#BiIr1#cOR1|cwCi*4Htus-~UbJ7x$+tkcHr;{Y;7duORT&!3};{S>SKU
z5AQfTqn*F;H%E;MLOk88)U+J^k5C;xpLn8Uw&U4y@R{zWB!i(^7%#5Jb|QFnr^{8_
zEWCG3W08zD>b-6MZo7r&UVBR&{sKPz$ZuJ;yD-{`Syz7={f@P}SK~Lw`SbewnTLxp4p}=26M6)kPS4m#O^;e`kk%@@8TQrO#cPL;#IA^}XN$hXp-W0su13Q}+gb5FU
z-HQ$8`8D9ZPJYtv=ay5=IYliTcqc)?*$&@zqShQdTIIXfFpOvXf8NQ`%9tk!9#h|~
zFXztXWj74IQ^|>yTBl2hBagY<&&K)?_6u*L3|&THA1t05wf6KQ?qgQ2Shz1hJblOHKKfqpNh9O_&H+%k1|NIAZ%%Rr+x;*Y)`^v)^+g`U>n01f+r
zNvQ*lLT|(3=AJhiAw7yLzctEvOWxp?D;qm!_I3)r1?jClJjQ607}mwNs3
zr}aHBrC}4TJEPR}W?IifPfR;G0+C}}<uIB;$UQx3YU6d;rA6Q5G
zYK4($KMq39+;m5E#(13Tmpcd%C~$6kbU5|D{_?mHQ-O(}&uP$mK6~5l
zN#kNI|0S2!|FI!!S6Tu0nE&Y-4@~vF7fkcvo8^rNYB&6m#Aw2T5L_1omMNcL@o}=H
z6pKsR>#%)8xwd!NTEelPlUO|3u(5~58S7X3SlmfWb7FGYz(irvYybX^#g(`v!mM0d
zPldHJynY3k9QfLSU0diqv;I`yr1v@y7~)RlYoybM5c-XeM}S5?M)Lh3(raxwdxqvm
z>O*;E_pES-!BlVWp*@oD$k=;qyo+}PFqD)*+6v`|juZ1j-0R2lamPKW%$^6E5w1|a
zE?1YK+NKE%tDY}rDDoqa#amOSv`UqypISS^7uvk`9^6{rwPe3za!iMJIAbL
z?Vs+!wi5E#1|ou(pB@vzsIdPQt7ZF@BZD-V-ZPg5)?V>;{yfC40%03^W;ATT8A$+O
zn<$TV!+6}+%<~ZYEMA_Ni*BHoeu9^a?(?$n))A~cE}IA&^4?*5eF>!%As1bMnZOC|
z4$RKRxdWN(q(p=lemRSu$7ZiEEg+lE;q?!AP=Xit)mr)e!_@JZ7LZ-8VtNCYTLFM-
zef)DoAD^_vSEll1TKsrkrq_hbwp&1pzI{pZa=xE!!+1OAU*+|h$sFiCAI_t&1#j2s
zh5YwhV8y@ZxHf(q%S?Elxbgr$AI_Z3V0Mew@y~b7=k0hH#Ph}CRxCHPKf@l=61do&
z=Z}fw_<5vW%)j5$t$e#{6KSJ`dGvr@6#gKdw|nCu9{=2sV&!ALUt@X?pYrpj%Y_o?
z$m6ytGE`$HEbz+_OW8a!Ri+aozh-DOJa+w4Ff^BWJ#Eyq&*y>v|v!XQbL!?f;inEhdXbIEqU
zqx9UNX=mQ64$x{R85W18$6VSPyKLw7cd2OfmZ}dm!3-Z8Bl}Zpiwba*34p{a`it
z$rZ%6r>&aL$D{tfM#m(Zv3P#;@Khz#+Xy0qynuUdOc+gh$`vCIZr=-D*SqV0^pP>x
zSFcdsa31krAuBERE#myhn%ut(^c+P7qs_Z}>+Y4Jn6^eI$+;&r#el&
ziQs*E=Nol_k2GrR9eB+M`;$>G9hF9c$IdMiAB4Pf(uJYwCkNDW(JjUu2H*B?*v@%omo?+f5Ckf|Du1V`#t0SzG%s}J%+sGpR2B8z!S?thW|5_#lG$r
zXL@mYEw?IgX~J#WYEF8F|Fc)^$cr5c-RK%3;QS6QRnrH5z0}bwT@_u+4K4ps{jT*P
z_j0hg*1FYkRmzFic0MZS%EpP1naoqJXTyr(%T@wz?8ipzrt!Lo-Kg~^+
z!rsvKORvnE?sI0F&M5UCTFljUD;;)Ei(k(no_r}vK7skuc_`D>74zU%qWVt%3dGr(
zu6gVK2)KSXM9);CeB`8<&msEY5s!7vn;~ECmpgvHel0g)vGUXvXTc{&9B}TG22YAV
zAa8;DM^8J#i<|M?U#v`XIxDE)Itsi4j(iqy9nLZbvP1EHf9zv2gfKt#gbgoZpY-8{
zyU~(~k2t9(tM>l9}{xIN%(`exLxO^Gfcy^G_k707RGAD8pI6#JQ*|4e%~
zLLR+QZ^2lsuahUd;`(D<$v#+~KMZy*>mT^H2<@g72+gm?{_f)djqPDL$9SIa@|67A
zd)uojb-??I&BEk=<~HQ6ru|1t&+!EEhpUGLb&z19Vimme;e`BjSVR|RGwbnWYPpX>-I*;
zr9oCtt~pHSSXn)87tA4RHW}HIzx})`+GRNW5BKQ4%XN&~k(Q1l)zA|R-ACKsGQAF*Z)7=qFb}z`BH=IzXypr)r$+zP=sDAB_z1@VL6biRd{j()f
z3WURl43H&Ml-=pC0o<8?-iq{=j(*rqgT#DDF?O&xsC|pjWVZG>g+!gHJPjCBvT=t4NP#^|gn;GX2F#HS@kOeP!ikfn-nI?1g(sFJasz5I^vw2Li?e>?gSEk^KC4
z#!|9#?Wn&X%s3ZO5;i5gKWzdVUq!PB6Ow(^U>5M$%HND{#Fim_5BYjEngtd*^8EAY
zgG^RGy5S?kwls8udNVDakBwW;>n|OCkBx)Y$@xrA`_>kw#|_}~`Wquun0(m(7th1p
zUNQaHBpe5zy$MDKSbc?QG?TBrK5k0o{rsBQ``=mjnCYwBn(BS`SS!#Y#;IbXP===zsMKE2{|0Uej-$o;Po_Q
z+L&JBh2yOJsz8;YnIg=A-mVOu|IT_(7l4qDi8#aT)EKPD>Nj7hKU6Z?*HEX{+z6TpO
zoL|kmSVVe%EzU6hJK^hR##`ki4l?=7noVr}6{KY`-q$|oP(OOE=FbzxC%a}yu=#yu
zQ)CRSv*>a?)}MLa2ewYLEu667!T3L&zmjm&8DS(+v7WcJZ6g0>zjzwE?-)1ao&(h{
zDcr!;_2;m~$)uO#ygqLcFtt|*69Rl}>5fd&bD#a{63KR1>pl|-jOEUeKf3O71qWX^
zFS__aiPrVTjCXC+&f?1@$H*_7Tri<&4bZCRk}LVc7%e+&NHISfhHfMOo_?-i6Y;o0
zqsuN9z-yrk$5K9H`ozwS)K0y6@dL`UxX)QTArKh1a^)!}_`?%{yABf1)D>C2m)Z%L
zbhMY=&#$KaE&0c>pQWUyBR*&kYX~%k9R4=9f%2W##Y2xR1J3JuuwXIJD6hP&199rF
ztLFQqkVC(Vcgg^JCmE>|-&{QBVH){q_chBTE`tZ|5udVC5Au(aPix$Pzh4iY=>Wfe
zEo$Sa7Vyct1=rmj!I!iHE{_sI|7IzfU&(&I%A~xz@aOaFvzMPm{>aq%f=xP}Gustz
ztOh>#bx4-UI@FH}ylOH6`fq}!b?-)gPc=%XTg@K4ImLX%PU_z#^Z87~2Th$Jj~(&c
zw>^j3f1}*wRAqiJ;=9fDNqXPGNBv4R4>%2e)R6czD+}_PL+Xd7qrdk9&hFiS_@0)d
z`S2`w!q0&7>7&fCUweP=RoZ_7Zui7kug+I=zv0_$Y4@jGXhpDnnH}z#7;IkMunO}q
zQc8L98v&={FQ?e7UCTB1s!Dy0spDeAhL0Y+5BH_lE%>?~{oNWUo$S>q;1(VEcV~Y7p0j)JzT6m2l(740yhN
zSS@#R>oL)bj|JSWjn>zbqmU02o}oXXR=|bKU(u5-e3uJWkMh`||A@0(5)s!qsG19m
zxjS12?0;^Wx*Fkpc2G0?uB>sd~?3
zoH$nrDXf?2BfqUge%a5)ZdVuPOQ^p1uR&;cnBl;V)dP{w6_0M015fRU+;hQN2l+!T
z?6p7o`-PGz9{V2a9uB`V9F
zO_u(BcLsXBPZr3n2FCn4J%jAg%#&q)QbEvqo%~$9@jrFq1)DAQ=L7(cukyM*8K__X
zNsG$cA|9rbpSyV4b{^#gG+iRMWCK+eT74yds>8N+erD(Qj@NM
z%|P9ftj~HtgH_WUssH?w79)ut-fmp4N$n>LN<8};D4)5dfc#Y3puZQ!0S8-M%qG2p
z+4hK}hy!OdwvfI?g_#lcyVNDdm{2sbPL8l%V8CQt-cP6gOIE#MJT333k~`H$^md66
zK8yQnO1Nk94h8aO1Ba!u-%p>IzcQT4U*@*`qBxLLL^pepUw)Hzm)kymeGAx9};8rzd72Edn{P=N8kGy=B7m3I}Z7Oyyrw
z+87VmHtOv(lHb+MVR7CpU?J4i*DU2`go{ZN(
zxn~g1xAfBa=Wov(%=BjL0zE)|RtM5(q~{{b@VhT)0b~`k*(?v&?pMIx$L4YsE5A+0
zm#8-!QNsAc_mkXGl7H_z$Lf7*A2ZC2<$sS=5nar_*Dd`^41aGf;CY9ZC40{|wu9Js
zKf1{O4oXaY9?MgyE7~wSl|_;EM*GTV^_g4{j2Q>nH+fbL&4-fKo$Napit*3uXt%R|
za^D?f{C&TMh!*J$lpK$QF?dE?GhP(m#VTnNVW)z8h4YPu)J=lAW@2ac*}Te@*hfn?Vb}~ne|0%>Q&$mBmWWP?=CM9xkUXW?4EIq^imcp
zgu4M>T$vZ44~!j`s73j;%-L_+Mgpx9r;o4$hQ_}f`38P%*?xztd+__Nw?2FzeO)`5
z7}C4uoghj6qbs~rH@)8|
z41xa9LHRlxAn$6f6$*o_CA~+&68_Yp%_Jxb{3Kf7v-COg246N^+GK+IM~_=~Tmin*
zSrS?cy>4|G{pW~dGQky>OQ4r&F?4M@>@148^0Ghp%Yl)9@P7cnZ~V8=c>v^j&f5n
ziNX8Yxo`2EwcD)rBM%Yh=lkXd`e|`8{QeR7n)vC)gC$>s*IgINt~Yzi`QJ&O`luKA
zx6+uloCgA~X6#mP@tcT)+`qjESpsg4pWVE8{UWX?RpzEgC)PoEz$Clzi0k*Sm~0ze
z%e}T89Bpq`&E+hftR=N73BRXWkt7?8dyxu0T_ax!xPQZQMW-&T<4Pm4SLe#!=RUZV
zE58l%Ds=b%RoxHgvH>nHJk*oxxPY7enwnGBm=e82lsso$BAb
z+EmU~uQz6^MjDqO>=R~Xn#Z{p_(=VJ{)pQsr1rpLZaJsvF!gJu+f}aN(%#e*(QDj6
zo3^!++;4Ju4+olC%V1ob#n|Ac8mX5f_v1u+hxatmzt?=uslceDfWo|3glG<
z@+k8>!RCuO0y8`
zej&TnC>s35_O`@qS)QPF&VoQ^5>nme$zf__ZjP^XPCF=ymq@EzGsG3e>m=XhMee17NI(f=&Ux9;B%Pwi~BsyLU2`oaCug`A
z6I)0=+}|Gu(%}2L{CT;1p8EtUkBYJxM5w4dYzVcJr6E3<^qNyoCQ`X&`7L&Dsm0h}
z1j+s#r&m+{ZQjKVCNv(iL$3*Fy!@))*V1$PpFEWbQ@>B$OmRtR+p1f%PrhE0$>My_
z+Y|@Vf3a?I9L3A~M~YY;L-LTvVrpM{?o}2aJ!^(B-Vz#6!0>A4SjK;vK0ILYdTzg4
zEUt@y&C_%4Z`LxsB~vX~Kh-v0^{G7eeauuspJ@F27?{;EjNz|Y6Vu4fjHk+sZ-(2S
zWBpDMlVSZ-mb_!UN>Zh+mFyf~WC)^z{A}`996aLQdG~?Qgq&YCD;N6^!R$~T%JY{;
z`#&@LrcXG?_{5vSM21n4%b1;!>nPy`|6Y!;545oP#@_3|LMctE~lHa$2I>u=~?*Nw#6Q5!Kt1x+KU<~w)
zDjw@ld-=5=j;w{=#2wpr=p$}ejddNd1b$J@CrV>H^gj4Hy}JXwn98@44gp_SpHFo}
zdk$yx_ugL)-t*y}$1Lc%WzF*Xi~5a2eq0^|Y!dnI^b7m_`!^>?+nMA1=!n6=Hab@n
zn{jMB`qNr@GOh)7j&+}!SqZ(tL8}bHEMT9egPu=!d71nX)roHl0(QfixjrX
z8T@LIn&tY1-?5)xZ+Umm$12Y2ZPARF?XY)X!tqAr(=>)3D6){n_kNJ&sPGTtCO`FF
zi(fSt+P{0mgPjk#0f)@D&yBC*=9e{u550^$&d;j11&G)8bOvM>BMzMiJesEn-qIC)
zZuje9m7M8e1J!%J=&$XE#kr-J&#EsBvT!~oniw-QU}qtx+CJa5dr%qHy{Gf?e)x_V
zErnkDkr$bHbwc460avG|8&q&GkDF0{{KuNy$K1$JE#XP01>B_TpT#SY*OIm@3TqR^
zeXi_ttpUI9a0e$|8n9~DV=l8Pc>ByyoKOAY-qc*dJpXHu|9uerm&=;zRv7;$g#>US8Ma8l3<*1v57y*ZaP
zQr$2w!-R)_S_r+0n6v%vAYQ5r_+V%Tdo6>vO47bVFvw#n#h+r~m+iEkf>IS)wNQU5
zHR4+~aEnm74*k7{ti*BVk0KErxHy$dZ+)U*Dj)A~Y
zXV$$q22?yHee(z2f5Ok=h;7h|cJ|RSLHTdl@{~5TD{U%~L3&-8W4=g2@6gKGn;f7w
zXyxMT#5=Znx^H|7y{bQBB_kkn>a{x!frgbScPU;@yyxITeyy=zkRJIN;Z0$;DGu#B
zm~BYBD~E8``@W(a-|FPK@mIWHV&;hQ8y_3B^O?x1|fB;D1X
z@UJC3alne3@?)vot9n2F)Fk}cef|FAw{9-(WO=0@W`;K}Lyl+;w(XNOu;`^N-r&LKyN+Csr^$aiHgN#mk`KN5@lt
zC%+y&Nb$1!x(F_IArB}b?oQ7+?l%`DY&}`GhRQQGab1gmtHv3}k*uh2VF%UE(wK;W
zK^`*qC=Lw4Kgwn&(m9GErEUzZqqy2NDHDMX$14{RZ@3=2)Ot10Wkh1r66jUlo4H~+7VRmk%YCNj
z*ZmeNrTM={YO^-2JJsgd?HRA32M5d*kC7jk);uj-
z0RI~7Ev6rZc$u&A)T#)2m&&}y?SbC4y@Muhgx+Dx__9Cnd%YryH}6LLo4IDT=Pc-%
zF7o+R0zJ3UhnLDh@6Ms@;pd^(8nf*65qI#M4a$ZU&(Q>Or%lJT3htU{rKfZV{ElhyO-CvCQMu0P;}DM;`i1%5zk@u5px%Donlf(u
zMfH-{-RMtt{5gB{_(!d3eF
z-TsU79gpR6-m4zLy4Y$Ftp$0@oc;5}R_Ad;o
zH@3tz3Aj?xMaQKPFKzzV?lr>ref#BWwhr+!CbheKGrni4ae4L<%mg
zluUk&b4XXQv9V*(Z=B^+g96lx6+Aw=4(-}iZ|jwX?6+AUzaH~c%Am48#mkho>rJ#y
zd_tA$D6ZwFqFQ0_mx!`4nJ@ou;YkK-a
zuQ0f+nBr&akM^c#&|4z3CoC4SrIVMq57x(;bxM-)(ED7p6F-my&;9A~AN{`ipX0H@Z-lK6P4f56B^M2Ak$jM2{mi7$>@sSmz=M{CGkhb7dr(!&{jH?c!PHJj@+fs-eG{;$h*Q-6>SA{C3nEdhb`ZyWL5)@HuBm^`kNzcaWcstox0P
zEBJhZ35!G1-eu{M{I=`HRMP7$lCmVdN9n;~RQ{o=h{Xq~)*BqjKW`1mA$y(-T)=qO
zi;nnkDpzU_XZIL-A5}8{KPg`B2DNu2?KZnNU|fBA59ujP5bsZ@b}7YxP|K&8{Z6_6
z1!4C4)v&BU7H`!14H{1MSB(xc{*gD_oW)B|M=KUDYuvW7czJsKeikofRP&Tc-(pTT
zEAI*|W_qE=u>C}NYRfft&hTUIAeN^%Ac`~#%9U@JGPH9RXPEhJKg0F$MND6!@Bga@*^HP`J&rcbTv;m%Xwp`2Xdh=uwlh@uwIt1lw@({j(86TY)iWu;}yBl{{|09jB
zTtJ%+)%Q$yT|_v`s9TK6Wo4ZP)4aL2q=n@(wmKYDr1~?$mJWos>SsBSURTKHFp5u0
zM?aoV<%PX6QwVow^XF1VvG>_{^k}z6k$$s1j@f{B$FFBtcli;sTQMVXGL<*lzQ=$d
ze<%5v*=sywHp_n)P2lf81zDG0pgE_z@ma^G2@8b)0Nm88dDBS9}3plg=H#-;UGQ0MN
z^q<~+&-hJD)~zM94ia@f)zY{m#7kIHeW{D8C1Fxn4-zKGzutPXhUR_koSYHV&oMh;
zmfu(>{p2?3jjgI%OL`;Kd-qU%+wXKs$_MONlrsDz{8owk%INiw?>k6`zX$H_p1fu?
z;^p&MQ{LMlUY5F9sL;Or?4kQ3&*Qzcyx$7!;6Fy`tyS0oKN;Sz(SI89tyjcTR;+;D
z*BMvJh`(?Pj&3Krl6U)XcY(Dd($k5l0iQk-@Y{Tg+E{2+8x#kf1h<+v0>-{@U0>JzfRCbJZL<;P7Lu`
zYPu4BjgZ4h`GW9jV`_~3mt5?IXFI)8D
z{!!$|S9=v+9ee@t#L?fsbyWp-c-`+U^4tTQv&f(6lDo~dY-)XNFgyqMB_(42{>9yEnoae%&+L3HPI}}A@QUtXOFxx@f5%H4
zd>sCm6N#?+`{7eQmpiTPOs!v)v3D3P0ry8
z-9n{*Y`M!B?s2hJ`%}nWD0H3ZRB)Sf5+41x>{cG9H*ciyuTi;Uwc)SIU2{;%0YgH#^NJ;rff%KB>ts<~BXF_L?kL&iQrAZjnsJIIM`h
z{XG|XC$kV&_Z6kwm~Y1vTYfy|MDNf0s&=-Ddse!c%RLN#qJ8wmc;q_{9k>G})>iba50(@WnNp^nETrls3J~1r7y`Do6-g?tx?{nSH=Nfq}s^TWRxm|qg
z6wcLFbTuCb&kL1_Py2}9C&umiv!nl%YOZp6nN9$BhRdnYFK<>OzUtreupEhXsl6q1
zGVYr=K0aS~WpFKbwp=m4&JKC;33Ht?`__Awk
z4eAwSDqbJMeU3AkJA+oD{ZnBUV?wb$m;5;Q`WyI*SW<4BH~6OZ9jz#B%x5{v^-JDh
zertyRiKG2ORq#`0d~SU0TukfiXA4Tgvx
z>oVn^La#+_)2RIz7xBF>#?m@peCFGm1HhN(76c^$g=ZFch@kzgqo-Zn4p~(vgd=~`
z-FRYX9M;WDn|YNju)F8-;P1ZBySh89pAPz||05BZk9sw^bM3!@*?Jd#(tVeEGvX#4
zM?LpvVBj#cBc}Y?e=JaGdbJ<<)9S3Uf0vL~Nq9b?rXTbM?eh8b4bQF9Fz>e<^(vnm
z@6$fZ`{@l8!e5e0?PP(k6=HPBpMRQm*@^hZb2tBX+UHN6UaC}t=Vzv!9z{GS^}e8x
z>Wzl$_E4PHTE%IRe?DZpVHo+XjJ+OX%z=gv7o?Ma>-FtuBtN_`K_`gx{z%V~^#sQM
z>c$`X!2V~Sumkn$reP#W{&}d{6l;?0UyN5HzNF)I@GtG(2BsF*Q9I{1Y)d389y0s{
z#T{3R7cpe#qq6-g$0CB7r>8dIto4y5H;r
zB7l8>>ax(P%B{QPYOezAsrH7DZ&EB2|*
z{Ws9_o#(SzJ3^7l?7ZR1>zDWw%7`O0kRSv+@eZfEs&PZO9ta_$fo-xC5gq)9KO
zP?6bhlFRcVM}M9N9M$>B^tQWzEjoSb~
zKT2Z&o6W~DoS(dtq25KlyvGXWpuWhQm!m~NDNU#*R?2i{aciHyf-$v3Ke8iaXbye$EV@V$J
z$Bg}bT?dV`OGy?lFjgVVehN3tlQH{snx{&#glB&V|X`(sCczz*N##R0qs{DLxPN&Kz*%`LxCz}GGz
z2KGReqKt7-@F$xS9QOCZuTFlxwu1c1+UkJ4V?WoTQ9$~-
z>!kCPfMq`ilu&s?c0>p1FUx*mLh}5@2WQdxSB$sC1VaA9Vx~wC{PNDI$91ys_peIE
zl%43KFTRyFrt)GYaY&UdKGV~Vv+pk{-zuWmCPSYHA54_	)kDWdU}d+
z)NO5fZOeCL^Hsr?tX5~k%p`qGC!BbWxzzS;>Ey%@7^xg(B83XB|}@pC-Wb=
z$tLq^In7ntPtLgDUdSdHl|ao}&R5)bK|oFg7kb8J!K>N~ZtKU`q|1@ud#+{r*2o{U
zR=Ju8*}!k>`p<~PJR&&NKwN_^KNYn-PwrnhEm+~q#(FRZ=&>LF(%J>{i_Mj1Ex
z%I6h&IZwEOp*#NV+h4^k_|dHSGP07Zu3eSlWn9XgJJ#SB691T+Jl6b$SbhapC;!59
z&Y((eiMr*qHsp&6Br_DV_h5e1zPn{yAD(6aHG`OGDt>#+4#8-SN6majF
z7EVlBgnYq{ekPy&@cWs3D}&1qaS!FqeAli>^n0r&eY+*%%%%Rb5$g`cz
ze(fKP`^uut;$Ho#xnm&z(;mz^RJJ&waE+z!jE0(rfn?
zaQ}svtVzN7)~ic)3dg+QrzXc<@Ws8MO!e@aQ<88md-!0h@}Y>kSv>w)jho^+B{-`+n9p(K60$+d0Wi
zi4|N{ddZHmWB5CB4=M|i!8bMSMHji_yff$Y8;8XotGUQ&Kl9D-_ZP-E-pC2F0)JWC
z{PXi9+>aIct@8-$LoxJ<@3=7BM_wD!p^bI*TKoj?7VfxAWKPtD&PO25PWvhmGnA?$Eg{4m
zyfUZ%!?QY=-wS^AJ30XU6o;CM3`IXP1P_}#9GO4xZS`2K(_$g&-Odpv-P
zX1FdOzWAwl`#+LbFJC3l$MZw}SP$O|`GMq$#Up?L5s^p9&)BR|5+;A`xKT${7n6TtEF3(YaKy;U&*bmJqVtCmzPR}X1PbS*XQdF(z%#9m
z)l>QDUokANb#UQbX_CkFyR(VV(0j{8YG3@8AB!VCr)RNqpfy?wwp1_4{!lUIgKv^5
zy-)CU2SsXs-nlc3@9BQN2tok673Q({v%9Mme|Zb-8R}gny~^xWEUvpN_G5gc{pT-Qw%@?0DW;Ta7dt^iR(|!O(uNKf^=zNL!)2@d{`la2Z`62Oh8h
z0Ih_09CMUMvD4@VXZ-9636$y8OO%cj3n>EC;19Sk?;pt
z7h$jr&u4|m49IlLn{a)fjLTI{U-+J5a?jnNO#jG!{`a7ZxE{FKn!15B^P--fNXa;c%1k;vD_o^~JH=T9
z&yds&_P_Algn-39deHMwiU?Yd=K>XTuIWPmNzUdN+MmqH6WS#QoU}bk-W=E>z3A;)
z=>J!?Fl-ojzOZHCq17l?ki4{p;*nv;pIa1WnT;#=YY{xRM(_bu)neV@E$yGJ1GXm)&jTmku#
zsvdy
zfJu$vIdQO0!MWVC7~}GM);9SwXm`YJp~(us0f&#;?7;i!4xOL>0sZ&WZ*#Z8_juaV
ztlSDdr2FBBgcEpQtjxPFw^8pm?0LQg?3*JMUE2nHsd;s%2lg?QPv4wKz__?d&3fWY
z^XBRq=Ln3?9FOe4PCWOtDx=`_aqKHknU&AEhk3fpO~?TIzbUJyj+%pZTH1YA+`&FU
zCOl@xKYT}npEf1Ck=J}Z*up_jG2P=
zW^S?*$OjW`D=SIb5IcwN;Yf$BhIAHlJ=Qb~3W8;i!f7lr;`C`!x
z*cX={XSE(^?EiP+A>hY%?z%CszkB)Bl+hTsugy2JjN3WiS+Q8rI)A@qcw4;@vei1SW;1Nq??
z-^5S)x#ux{woZ=<9-{u>k)3+Quw!QT$H6-2XL$W{eK%m$+!5SR^na~PH-qw^|Hsvp
z$JOwKeL~sWP-H2|QYu*r&D_(E2rb%ZvsS3IkcxIuQYtA*C?zf07wva$n^u+feUT_B
zLb4QjpELLRyzl3of4-mRInO!soc(*w%$!jw^-R4EGV1E2pVM${^V!LScvo)5rhR~L
zYpuOwIDCVs$_ZTFkr#&RK0rSFYTLKr@mfi1n?@<%+m_>b+<w@+=
zs5r8C4y0r0`S{E`2)`LUcoV7w9DpA^UDk$iLx~fKDyuO$h
zd*KF_ziYhi#_1bhgV&1zDH(ev-y7ffJ;&>xyJX@7Lq@1*mcS;*cu%pVe^`OX=W6*PY-xw?mz
zb>j~#!|}Lif(Gl#4Q6yW*UCjSA33*bBh3pcxsvB#JDsn5G{4Vy3d(i!mXk4-ymQ!F)
zJW{0noX=QD%l+##Xq52Agc`;pJ0%(?Dlw6SdGhE5cKAFtZ;xBf=C$$WY+l*2!iX*}
zt|xx9-2;ymG|qnx%P5F%@d9oK7*Twl#=fuXY2;%*-|*gkv_1?M`wK
z?`~rm4ppR)js`NW1+16i;J^5XarQtDWQs;*4$|&{UWOyj~NT)-`0?JJ1fnxPzBzpg)D1FCf1W8|582=>yBf
z`fS=E7;ea&ezn+s{&i!wgAH};c9TgTxQ+KO
z=hF5IIpuH%*g1|jw2xb~+6g*>$z($}=kIatM$Rr|svQ`vN5oSV$J0g6iFr@YdkKf!AN9lFWN=?NjSnvC3QUT8sJ;l>Q
zINWDcjvn8IF592Q;mJ;3@Zv4-GmW>>I9z_lFnxZ>`XF6UNP7J9R8FDC=ZYUX
z@Iejmc*(nPz##CL;z1~QI7cLwq{lbHypOiu``yGE>vV+Hsu3ef+*XSAKA
z1Z>lTOin#qfzkKFr>huGz875h6XsR0=n|jNvRDedZNk*z$~5q-d5Xcaw*#Ls_%YJY5AwFJa{T4W
zz+2nbe>?$N1(ENU!Y=tKS58>5w;{jWj%uNN(qG&jr0)G%7m>
z`N>}WK6Dr0u97o}6!^&}^(-qLF#JKxgdU{B-5r0%9qhB>A{6`qyX(|W;Ca9B_3jVtG>Qdu?
zT7hEzm%xwYt;i%wz;^S&CFZagX^bSXJ~vgXe^Et~6BM+3jr-;$)@4QLy$lj{uj%}T8sa&Y^(?SEi5$RYb?HUvRFf<qJ-EU9H{hc|?n#`F<2%>sVP0H+Y^FXJpmwsml{#Rc(?fq;zjK(2
ztZ_blCGT)y9z3Fn_MMc96pPOplx0i+SiO$IFhm-@P-UE;>W`OU%;R
zg6)5fJn0?+e46!|3-flJnF$gru$_wifE(cZc%C)>fR~n87T|J;e#$L??^mK7&WCxo
z&rH!nxZWqvrJJq=TsGw%wHZ)o(FFVm7T#xD`Gyy_gP_&IKXLnBku+Y6^}_slVHm5Q
zl$2rrdb_9nae40_3pj!OKW1@W9QVin{Dt1)ey#n%iXQL$gx4Iz^>pngNuM`tt*GpR
z_hoWjJw0&$+x9mEUbuw)xfMT3as3|R^4_lmsJr6M7><`mEv&}l&Echu>#$5L?K_J7
za~J)%7wenH?CAJ%rw9G8{=l9dn8BbQ+E?>nye2Gu9^=VF*I*)m_=*mTG5T4s&w_Yl6EyItu#9im
ze~kk&=dfJxy+aV=@rL8kDrW`GKVkZnK)-qrDz&xb{`Pe
zHnt}?_%Ho`cz~n-0IeR|?NneR-EBaY(qhYnciG6I9ftQ6u?XSxas?2xWGoj0E5>Mx
z0dN@y?Ko)8iQgv04u?(&7mQ;ZT+G214(4$DIm40HaFAZcK)i>X^m;k|$X1L!5l%jo}i5=hUh=
z$FOg!jGF@V>+Ll(h7{tK1>Z#};|S0%@3F@T#|~WcQ16Vw^2YZdlmrD!I^FE^bMvrkT>(O$GfliF)7HV57~CULu|Wh
zEp|EY#gCI9d{@5j28>tIx6tG5e&^j_2t4oL^@SJ%T=tq^j2LyG{oM2pUPA=gsBi=l
zfO%&H{I(iUP_mr|V|5NMo$lfE8#D^se?X`AJYJGcKPn*s3Id)V;NJ)UZW&_Rmqk0$
zvW6{QNWy%*=sGwGd?&$%);oTZps`Yr-9BUjKGCv`Q5tRc+eOM4^RU0-Ct=|R@1YPg
zH1Pb{V;lt=Nic6SnPZrLM98&4$w7WAC1}6q1R6t&0Ke0S4;T}krKMy0x#CtZg2BEj
z^9Sws>y`LH81Uz$$e9?Yh1}oc`#9vVpZ}g}JxcrW)Jho~g!cw?y45hwn8iN-QfW~r
zR6Ovi3HE-|3t@9Qy}#1%J3G+lJiN*0%gQy+qz
z2$;7U>}#<;IP_Bz=5-oj%V_sz5~h_RgH9MzxOK%K|_Ih)a`7?_{7At2jlUMZPPgZp}J}M`wQXeDEj-I
zq;mdv9Ig{3kcIKo*&htNf1LBV1%AK{`P{h~hFEAPZ=0dv!n)>L;A>oduEqTVI6q!?
zE4Z=0C1*#fv3=n4Wh~4?r_EIxaXI|wEWQ)_7uTnJ6XT&R%^xssxg+V0(|u;&VujP?
zGv%SnaYt5k1(tce%dTP+^_A4Z_}TDlp?4(orEFz2b<8#(e;`E3pOCOB2NUadpr*Osx
zAaGEgmtP*l`Y&RI(-xCw|Xld#d_ho>OZ(V%3iSdM@C!RY;ky!UnMXCSP#12
zvBT@m9aqmC!2T@YQU8e73yXArf5hQ25`1;|e%D&YT`}_93np=T%QsEJ&M2%8?B|zXE1T=$E4nbbPhnU!1Ui52B`F
z0)hJQ4xz_W)}Oa9)xtWW_rY?E-#!G>e0?VGvQ=2#joM*`f_|5whalElUdpHaFn5DvOhAsm4BI<^;TPHIujMD`
zco73$v>#vbb{2#O`9U@T<N+ki$d+4Jo!73l6D6B*EL06$Hyqj5Wj*RN~_5%f=8V7dV0+RI*N
zY|Iy?^_tb}dA;u{Oal=9g8|(W5T7oNk$IJkxuuY8kQ?;a@)dhbsG*%2Loop6o`K{6
z+s}gAfCG^LfK|6hHa4-ttuC{}ou076lLXmVFAv2F_L7e5^lC%ddT(AfKA*$JsC_cD
zoyl^xA6~6&z2kRwx}~FRbn#$^&zs8*H-*t2r(cGv1JEB@A;yU&Ho_snv`lSeqY<5^y0Wp9`hg;rtvs
zeGbm#;1~yeIsJl03dpuqPPZzL$*N=-d&U>g!+h4`Y641-t!wnL6oCU5hSO;8W+Z&4l?F0F+
zFx$^EJRO0&2R{x5Tx^PaI_$%ogzYWByH1?^*ntT0?T>8P8c&}fTlTW&ncRoVAxE%J
zJI4Y_1ok18ib96qxkq;w%^%Q@(Q{b-vcrzfPe8{JI)1qVJ3kI96`-WxdC-bGkWt_@
zF|Zr~d~SwYIJ_qzZa`z11JD|f(^`gTe7F;64ah&?)Gv(9D+Os>qU=iRd*0&>72y4M
z%xQf(AG>_rU9@R=YB~Eo$03_S+OLYSq+x6yC)*Q@k!SMe5{%Rqz6Ok|rDdWqZ^?4{
zy%Ni^&O9m@H=o5L1I+h-0w6=MA4-KT4yF+*xSxu#e}U3=IPOi)O!nEVQAC
zwqK)2(dGL`V=IlJdq!ycJIC{(!NL0$l0|fVehU|e3IaZqkV2P_?LB=5EIUpoxMD+@*FeT{Sai`ggtb7`%ukp_Yr$Aqk!=z=rLWNnU_D%<)Sj3
zLg!!MG29-)UwOO%1?-3Al*5Pz>${$hr!n?B;t?PA*I!7}{(W8JPxoU%>oU5$Pu8W<
z>A5d5fB_nwn_bPL?PE(TX#eNsVdetQ*{TuUK2{8k(D#c{c}AnliEcXHA^|sAHq&sX
z%T*9B?;zgxQ{6PW7`V`SyIe6E+ZEs#0|@U_!o&}rw{F*?<175wxE{-wy}W7tnz~E0
zUOmQ*mM!aR+43J)?t|Tt$WmIqBD{#kR6!*goeyc!*tZU<3G51Vp_>3Uazl0jO|zj`
z0k7z@*CE_`Kk4+FqaoX%{}>7)pgw%vhVdZD)<4WXMavWLcTcQ;WDUiFkz%KJU=N5O
z_x+kp`$r~d)AGp!Vl;})gKh=kq4b0U^-tgkz;AV-I|A~Iu@OgoX#$w
z7Ds-~NhgsbFXf;b2XSk_@p?IkXDckz*#e}OGk|n40OH<`!~gA{t8!%MG_-yR2MaiU
z(IVKB96ep!Ak)PKsLj!9a1aI|xc--S)71eo?j=}{TRld)IspHL|BLs}KRVkGPMZVL
z?g0MNx70N#XJ2}W(xnO=@4|b#E7q$!b}^b!bF`frKDVj!{3aEHYIrZ%YI46sN-_E?
zXS7LGvkHCT>qwS`bGV1?rKc9bcYmADuQCsS=h5z5ODbBDi;>i{#Sw$)GBjU$^QU!*
zb?9+I?w@<`T$t?AmH)IK_A{<{TYc1PL_FWl43BrVqLnVMjNWB8qEJ0)t_$!zriWZl
zu)sq&m+Fw^lDyznI45*>3fGp`C_G5y)_nUmq`1sKVm!25nV1r&7D&(+=Mr
zw8875z(L-Als#%xd|#p!8F(yvl3E7i$1>Jb*S=ihjRpcml$O_`SFhz?EC9GCmu^`M0&mCf@#At
zbg<+^_#>AZBzWfq&+4%Tw5{cBXxKm_YIxt+?iB&!*Xk8IVYX%Hy^%dyv%Um9P3dt&
z9eK!hc4du%L>lVm{;6>;ArTe)SEFZbS&d3m_16XXzkqYMyBHm!d1%?ipuvEsROFg4
z=bo(pGo*8BkNp2Q^vhGSiM75fM(Cud^WyD0*a&|JRLvlYGb@LD-{#~(<=L@uc
z=AWs2m4L*EQjLJe^`QlH%97sdiS07hDuS`b4pHFvNE
z>vu}4Zali_)^@mQGy$E?yOcBy=do+OJy3P}NCs-L(?x3&(&0amIt7Kx(vfw~fww9p
zX=vT`Z5;~7)6fT1D;sW&D5R`QnDVtpp*N9xCl5Z3MPG!NrN-RP(6fedp31CesBLhy
zht}2{6!6En;7sy!I5+9CUqWLFyzf#rbYWp4x;l7g(e%$mRLSjV_N5{jefhfU*UGp=
z6uUohc)~If*&fgTB5aTZ=XDXmuG^E5Z~xkDcDIty86RbRTiax$`)I$#jb*9ue9FC|
zXgn3^hQA%H4^BnRCBxs{)zZ-D*;?m)E7DNf0)tm(lPRbzHJ_6Hos4RHx=zeQDahA7
z`;dm~Gc^YW<2h7<{)=+)?KG-DQvdDUH}6jp^1t;yuJ%JBa^yEGUb8G2b?a1~
z^$UHDetF96yCjv5T8#J|(*07=#uNRQNQG2n)OOdlS|karK=vjRGoPWm9U3nrOykkl
z{)r!Ajwwj>^O&DgP72aHBRaILBn3r((r@CiNk&Nqo8LcOn}SXm?H`JEPDS#OHntBl
zQ<3{SIWqrNJbJOtVT&U(1a(eSFFP*ni()o1*KAx=fw0^M~!7uUs8jN)Qn
zJ$`J?LjHl1JUc5XRADl^@WX|CG%dAcll$^q^mdkYpD>*J@7BRJQ2hHj5}Ua@vYQ*)
z+0{Ii>aTIAW9go`CGzp8Wb|pdR`4_AKdN+V`{@J}ds``5O(zkR263&swj~ADVXkry
z#8c3Rn)5aCD$kI8hWv4JDhAHc-1}*NK@_t8m>`;R+Xw9`5;)s=>@xb+@v)qz&kIFf
z$=s$B7l_{V)qJvAb^+Pez2J+iqY(K~!2HVM0_1rrOmO}3JXDn_xj9vp3I7K*{WZc>
zglx8_?%lDaDk>Uae(ruy;*T-v$`|gJ$v(MMt)f%GE
zBdHdB`+@}Y`@H#m1+`?=9a;Art+30j$gIjkqr;1Ds&W^>cWIRpIme1o_cY%!TaIR6!9mR$V#zn~VK%&^8zGO67_q$lJF9n?+GrLvwJQFpYUI4G!=A%IG
zzP}=w6dJePUol(2A2A;3eSSLh813F|X;#~m2KRUVdrc%4&P_lY))_O=R%X5TbZ{~H
zQFZoZ-_cU!T-s%PtF8=954%QY^TIghb)dy{Qx$UhM2z{l7NZLv`D<69Qe+lXCfVds
zj`%}HV?T&iq5S!~&AsPUBmYSy?`HKP^mbWxVY+h(TKZ09zD!9uS~9=vahF>)I@z$}
zQs?#>^sV>WJxwNs+^137x{bvsE@q;O
zX3RAyc15UF@`PFE{WA373gyzm{}Np_lU2`HQinWkWG&vxR3ZJ6KRxDNFVWB7y-iQ~
zU!k&ntG#B!`2a`5-S&jb!T0jZMMF&{s!)kiX#=5JgI?za?c1=a9!V$+%xi3GKr$?K
zBiUm3Kdso#&bQlY(1JT+3%04(p~b)5jyBC~K%U)F4jRC7ip~kk?HU5!86l)sI#Pp9
ztm$!ngA6FR0G<(@Pu*Zwq_I%`bKXjYC)@B3kmG~Q;SmE=e=KIQIAr;9GKV#
z|3OLiZ)Cij(}E<-);HLRwW5tzn;&LK!hd(?P5;^j=W104<>oGmgYUi6P6iZ~!+*fy
z1A}Ag8j)L&xY401_)h9TwYr7?e2>&}?8q(PRm)D02J(*ZK9_CKwTJLs*h1ZO>m`E_
z?s@)=q*XJzZ0EaU?+o})Nt|wer)CSP&B}LEhW`s6CM!P(f}bxee+2#+Z9$#aI?nB1
z+=fOy=LvNq*e@&m
z=rZ^Qo|CtifB$r=85u6>Tv-y*jPl=$SB0HzMl4FuyYg}i(w!Bb<2bt&zSpv@a1L)p
z?H{{levAd)X6DTI)wUJhZ~fCaIJ+Go;@;>(*LKA7z9MkJ)(-SW=t6z6ZU^ksNHy&d
z>Ohwh;0=|H!49h0?D=|op!^$$(`?L-^actrHf??Uq1rH(f-J5iH`NccP-SQn}a
zab=kU54P*PzYqRD)9+yUb!1~BYUPvd6jX=vAu{Lg{cO^VwmGdjZUg`6OON##USbK)
zqlK;fWluCBR)paN;%@`;zcRW;;BqT!m^(Y{;_+7WgUdv+Ewu&Bj{Usp9odYG66AIb
zLwS9Vt10ze-GRi9&e3%lXh#oTShgIHZAaOk88Ly~`%trBm#8uq>PrmO0J_y6ix`*W>8BU*ozt6TeW1Nu4a@=VsJ6{)Qg
zU0S!Q6#=rT
zDVkcep=h&SJ)9q$esEVx9jX$oURL+98qM2&_9bahf#h2>^e-xv!T6+m?#}fFWEUYn
z&}&jv&E8viX6D-}I3G4}6qQz>-Odu%OWPY!LHd~^uc!tjC2zhe>v|nJ
z@pyKld}9^LZn%5YQLGZ3jXo9A0Df-!{zZy{?=zDsWlE)&HXu2#VM9iK4SH}p?cB$W
zRjA_9hwW`KrHHFv_}cr
zKc9P6#99{G)Rx~ab0r^@+JESIwyY9;KdG@xVNE%z))!kmtqk}7(IFYb&xHRuw5|Vi
zHwS%gjS|@m=kw06dmf%~uN=vpai8&+D1rZEk~y1i!94SB=WDS!S?GIPRZxs^0lIX1
zZ2j-xJoMc}swW{n8*SS0WYbceGW15^xMe}nD>$d)dBzZ&&%ZVOt^eyi1?aEU&TPxC
z>1bkfJ@@&POeEJ6>$fbNi8O^vA|4+qKsWpE1V2*FL;43LEMDp6px*s&91rKFp{-Hd
zu4?8yLwtSO5o*9sm$_~<^@w?nHk&$ajy#iwY>bab^Y2YZd$b;X{p6Z~zPOxGLGf8g
zRM3C&3Gm`8O9z(n-%LPf?E-(SiFZZ%>HhBw-`y#kmlFO~MeW
zZ>0)!*RgEJy1Ot?QWHTEavQYTCyf?DLS}HJM*xD{|2JNne-f$$Vt(
zxzy%WP$3fBZ+p?@TmqUZZLdgN9E&`M7bh%8i$VWAIda^h;VB}0U6*t&e~NA&(!6f6
zI0I>FwE121NJqt&ZSwNspP_HQN$#3w;?V+iYy#e
zsVy&YdsKMi$Wdm3P()!+QCFPl@S{Sfb0%kicrP=ksCz=G3d=6-R^aw
ziKu;{`r*Baz%Qw`VvG?TFFXA~yy9i+M*wt#(OLowaUBBT%4bZrkH=fM;%K;j($@CQv1?S
zo??*q3DD_Hfkpeg(C(tGakU$Es0vo3^ZAc575^RoPz
z=*h})-y6`bSk>cK<-;maQu<-rIlG=AzwzSYBXZB9ndaC
z*YE~#KS6?<^v^uL@&wiW5&IQk8-wOnOwBUXeTvpl9_A|qThXO=>wlhw@0Y)In1&I@
zUZBCjo?j^o;!&Vsgy+Q?SPvL$<$1eRperk*dJZ-gqisb3F={Zb9NH^qnCTaT0%rzi
z2|>9^)IBIXOFTgyo4t35i$|d)%`+Z^SHvUv0}~>8uhJ0TO3&Y8b}1-O=k%;!2G5}!
z6s`ITvQXHYnMxX-ndt3ouZxG4C87K-Cx-pzMAT|!A=*kjMQ6N7rJNI?C_Qbv?a2E_
z$kEGJOtdLGdi>_Q)d4x9sZJH@2F3}l>vJZNRb%$o6PdBC}K5CaC@6c><
zOY<6Zx7_41S9~Hm@a!zVq
zqdAuoUZ$cg57)o*xRQzFixZM!2bk!s=X&c8&x(+0E>QJ#yRqvF}oJI~j$;D0>L
zmvfZsk?4Yn&{&-UqK0{EjpL^>V!O(!|H%Ep#7-U`3-7#A!u-C?_lIe(2Fy5F-!6hP&;`3GsdjM(Fh-;`V1br?a*-#G{H#citQR%)HXwNy~rbFgGmx
zY*ZGSN`y_Ri>ZD%#5`zqd+D4n_lYaqCj0&@e@0aA(@z~gZ^JAP4V)S+$s=x$J#;)N
zT1+hT?0t6lbrSR3w1-#2;A7@5;j|^!A2t#V_62sqUWr7yPtk0@?YV^3{+PbMuTq&-
zuJ%uSBpV0`q1WjbG+z)(f{gvol4F=Z#H&}&H!USXth3$z3T6=Ys~&%!(yJwIy12dC
z9$ZhHB0jFzI8aWwDJxBX`<_CSZd=Red+G%tkXyJ)<#`gJ@jNj$jLK)u9l98)@id%i
zvFMk2Gw(Ixk-UIQGx#~TiRIkXl*O!Rigq}BFr1mX;<@b48Aq9+JCrsrH!LL@9^A;^
zwX&EnIJ%>LsYVR*W5D0$Gw)vzUC~FFzTaOG-D3(RGiKBipDk|qa1Z4XIU5TS7M^1g
zhK8pi=HHAbGRvC`nzxh?{AvXWpT6Z1&YL5r8V|l9dfqNipnQvnC$VzVJM627J}rI4
z^~-#S8P;A8BW_m_m0BBT320X_op0LzI9OjvSi3tnM4iYZir-W_M6ne8K
zrK+776?o|JGl3U`|Aq~pe?2NDmMVS}8hSa%yyX5SJUc_48h;n~J<@)hIM|sSu+w0a
z$+S)NEid~{ybdZ2c&8*lNxNtBNf+!O3zgG$2Pu4D-jEiOcyRtJA#11c^=|eL=F7(X
zOS`uUk_FpaZZ_}OK}9*wH50k_hiHGWTtX{og4y^-Rx+kzCb?u&y{4m=2-R%xx#jmK
zlKg7%OMgOcnpqOGO=@avHd$Fo{e3vhM|JGwS#@`g9BFAb`=$t=6178e*7v=?RjIF+
zCO3Y5zmM$uq`Tq8CJjpQ#L_8E(=(*dl|VP$JvXTKk=w_FSE-WaA7#(G9oj)r3m+!z
zCZ$RHp^yLxfn#LB8tH~zE&&wZuXDyP%QUHOt1ETJ6N;qbtzWtVFSk+eqthBY^$npM1^RH8!3c)vHqAMdz3WT`D>Zzs2b&U_e=Tl?dK@ZA7h(d
zg*j3MfA-(5xi~;7v_yu#sCY^pSx{!@XKjo*?yMZUU;ZaOd{Gf5jjq^R!
z$levTvRg9AVby}A(mcF9aoHsMo
z>#icVIw-Bbo83SimA){iXKau(?@4|+64g&V+j=5dmG37dVak>J`qx+TgnrY(pNT)H
zPmAJ=uTJtYnpanC-#BX<%Tjlp4bQh`GN)wW#$PL1sh#dB)~>pX8NW|*v10jnSgR#(
zMqlLlk7X*Ye0C>S7d4WT?(MLCkX&n4I!Eux9M<|=u~WS~T&!*J^|`IP#Tl%rmiM>&
z`pJaa(}>G;oU$<8=X0TbKI6*qf{@8bAqJVdzUfxca+XxbHcMOkX7Zt_u$H9;AH(={q2_cEX5tq~i4X
z>5Gai#YvIna}PK$f@oh%VU6sJoQnR<`;Oi@zk>Cj7cQ>%!U?!NwLF
zhK}*2kfxiPSsHUM%kYSaF{%pg`SqNUV}*78I{nStlu@p8{QSkNb*$y-#L<->_A>5e
zoDUjKRbpNK{Ms;AQG=m&U@+n{y2RRev{YX0(Q%gQja!a4#LlzquJ5_4M4e~w7o|_!
z54*zR6}!{${%#;cKiGJv^G^_CbGfSDOKT$*SK};VI&U*ewwb3t%F=|@EcgD#-|=*Y
zrpTR~hn>MJEon`2@}nhV>XgO;S(#IewOlzHMR!Or*6VH9?Ri~-(KF!o{`WmehPB!I
z<({`>8IG&2{P;CnhsEzH87nyX3`?@qq
z)@yX;!09`)7?U90%k$~r3Klx6
z^hi^!g*y6qm)^a1pUKc9aVPhoC9E?(6U9!ck_-_8|G}a`C5EYRPNm=AT-L7PP=^yn
z0t|!G54PCWZD)*{9M;_|XU}@Kav|S@j6UP9lDEF+NpaS(g@HxERfkws+`}H3PFf7P
z4aeP=w|}7SGcE`weE&`j3VZTCxmrxUoMdRr?298q4edq>&cslEhwd46*3M(}7L5%C
z`u!w#`EH-ewn?JuF5I$lIC7KZ`PrcV!8Vu_F@0Os>LA2&Z!zT>&U%iI!cxV-=
zR(xqxWNjkVTN0S`($t&!=r?r3+*_VeGVd|j;4jE1@by?b^?rhSfB2&IoO2!IZ4;+-
zIJ}K&m+@Uxn4--XW^vznSfs-E(|F1y&3Ojn-V-q%1@93`hNURCGT;;GiELaql=`(XNO6;2I!C_8EwbkI3jAF{MMR3M-Gtu*CPz`(}KxsD`$N^PhO^^-&rS+
zvo?{gZ49?Oi~p$k*^s^W%M~f_kKWhsXzV{Kw
ztj&2S0Wb1T&!u%MgEYy${T4GEoen`^MCG)v|AhsY)*M?p%HPnJ#%!%IDALWlJa`;K%8x^@D_sqQnwjo_yk}
zdx~0kSqveju3IAB-AfE8p2_947$BbIJCyG9c+YH-ncKK|pq{XER6j!%BoS|vSN*)d
zn@ObirRMID;35lRUcrw{fE=%~l&LG5O&#(K~rAVcX
zo;CI_;G^cdx^(6b{3hI%){oDv`NBN3CHzpfUq4ZPKg|D7oC^7==F!Z*7E08`*BcZD
z<}N26WP40jG|ZqB*Hb?ya(F54x&`*eE?op)R_W`qT}6bcqV*XKmvUlAZ5EyZ9k?Xq=EZP&zwgJ?jB3K_QF~L}mP}&P%b7=iM8*^6
zHJ#tMFG?V4d6b@QyMK!que!UmS+AQ=|8z@CNHmTRmEI5?)09p0?@h{3*#c*2UAi7(
zxz>~Lt8KEp_so;|Mdxt+@0-M>8{+TyQftzDc(psF=aThC?@&PoTRIkp@u;5NJ@#0j#W3ziHgkf0UNuM|UgtV`7>cyR3m{*=YX)brAnEsI~q>xl6GpBN1
zdrSW#<|BmzH{HMY5E4hTl+C|$Q(sP#BRf5|P?HxzA}+peAr4;@K5i0QN>mdGl)_*>
zlX3U7xx>J0a&ErWs)Tv+2By2>P}Wtd_(rcia3OxZK~ck6o}s(${PmwnT>s3Qtn(;mKzB+suFka@Y`
z0p-l(5?$97MUIU{_SEw#J<(Tp$rJoHx;QZ;W_
zh3Y7svU#pjw06rNxn{EH-NBo2Hl$}{=wk&8VcWc=DPP1}j9l$FAM>ez58Rj_T}Rt=FhYJvRD``f(J$k+v8
z<%TH_$dZ@G8~E1bQ9oN;v#f>Rky
z!<|
zZ|yc^t+76J>_>nXW8D@rRsDc{tVCN$rp>8H+4TSgUQ+8Ro{TdB@}(Svvy*o?P-JSV~v;
zENAc3V0329S07($!a5>nJ6P6Y!3yfKmH&3zj-h^5Pfk_%Fw4eI*0b`e3CkwOZt7v)
zL59;6C69=bdko$xt#2Zv7vpc)SbCa=4O~K)8ZV+p}{Mg;5pfJXxZQFfL7~W@y{oJNtbTf}NSBAC6
z@61((WLnce+Yx(K8~^^v&9D3!6+I4_yOi=+W94I1NaG`xY?fJe&+t9g+aBi|dw1Mr
zbmX85wtHh)$*k!OZi3|u8O24KTgF%nk;|DA%3isQ`F}U&ot33n*FzRdkKL|iL~Zih
zrxVu23QxNvvsz)0r6F#lYE{c*Xr1-YWSq=l{rB*gqHI$uV?^W)1NwM8x-dvGU8zF1DpzW63QSa=RSt$qJCM*l~MSEbFH4
z9c>@sy9{Ybfwc6>5XR{KALnu;{TUzjseO3;#+Fg;ci+aZsD!b?lYh}@%vFZeEQd88
z9HLl(V-=@`cHUzZ%0HF&K5xpptLodZH|{-arSQGDn3?q~r&T1yw=kX|VIIwVs}RIk
z$#dadzIHI{))K9d&B`OJk+o4?D_8e3#$9x8NNyQmxQBnUi|OuU%y3kcR%}RTgq@Gc
z3=s5Tm&}mm;Fv15)oWv9vGO2QWZRMv}8WD+`N|O&Uoc~VioGiY~P&xGdQR_?#
z)}RZI!7iz6Hce-}?Pwyp$*y#4_1s)cxNe`}^F7jL5eixO0|0VRa8AHJ#Gk
z$8i5M9C|H&55w{`oc2?laG;6WVCZ6_|Oh3(%Z`imc?Z5%X$X1bPi8(T?)X#Z|s*j}^
zC)#UT)L%!kwE8kDy(L0f7T=tF)%0~)y~97H<-hG^=rjuPIFQ+1;R$
zu6cTUB*c%AO;W8*2c@q_eI(7epXVRcsV9G{_jR>g>>_hww$@DO))EQX1+z)6$u|cg
z=Km#D)CXyOcr8XL$9umzWO0zB?nrXAY40bGoRI1}9J7Jy66)PAoFhmYwH`gfQd&UH
z_E|HQ^1_T-y3PKoK!z3-ky3D9G(?^hdnBIz{>pmFV|tc=`rZlV$aOKt&y}HM*XI(O
zXYY*3eqWZ|2{{jz(a
zcWe8Jg;|@6A_&dG?IH%lzb>SbIT!KRqB$Vso{Tj*OAhP5=Ng?
z^HH{%18EPdc&XA|_x$+-`w9PhVSnOP@(Ec3Nhwvw7sQk3gB_@>o~SwLd{TWm{Eh97
zK*YN_)r3?;V5Dwq5usXI-KzMZh*%fa|K9LXHF1v5cxZk~1|j=;e%i}B>C9T0?XLXB
z(ad}`-CnZCkGWZ0`u^FjM@+E?5+*a0-I=LJWkX7xei2)bmsP*p+(LX@=joUfna(^I
zW}5s`D~ovL_F~GL`w>wi(lBp@oh4x;FZA`pn@NKEQrAu1C#6h6YqnkPktQNH@QYon
zRV8sZEq=$yi4119n6Av%T`!0Xo%Oeobqyi7@b7P~>!n2GjtvRg2kMA6;jKnlheDVq
zSKkanb~B7=YCLjG9oKkA%*N2*(1ATudGzI$a?O_C_AHMZ<36%O2&`3_kTFob>G+d
ze$F{H%{J#QBPU;LnrozY9aiSQHZPP~Gk#EhrNCXGtDfGG(;Oi^opbj~t3p(r(hG0B
zo@R@;`*W9c&SFa?e!s=wltQaWoz^bpQ1g9?{~G197ZpU&rO8+x0xbJaV_&K$12ViW
zQ)*oNq1o>BJKN7ilL{-QM100o?LD>^Hq-ZC_FT_iTd3B
zsX1!HBI!2ktb(IYrR}csQS*g05{BcmLYV8z(@YaQKNX5TQaLkal{M!Ol(2f;uTc0D
z{60L9eqDie(0jP^zyR5N^@5V)78~YZEPGB1WkdcxefO)Ck`HsOoOcdNxQ_VGuBAF>
z8Y7XE?5-i=G6?IlY2Wn^tk`Wn{&x85G%}0d$I`@f5$2M7?9RLkaxI49+6);d*5-F*
zS2pr2rjZtGAG3cM*{-jai2G%NUHOC$tSSa!doLsHI+lpBP+G*nCr1ns=l!qyPopAs
z@}+LM2(?F|g-e51n}d<4v{NkI@DAd~_4?5wff=^z6?2yEmMvnhpq~oYL$IyZg%?!?
zTF5PX`5x<&+nB1==^Jc6Z7>t|mG?X3K8TXPXZYGg0VaGSV%xK=AFHX+{L-LajClLr
zOf&S##_#}Ur!mI@B;eWh&R%vSR%^AV6-N9Exv3O(rW?t@zWVe0w6x7Y`u52L)Orgr
zFOjX3dh-^<)6AOVKfxJHuczT=3|R)Eo6}aLR-26lnkAu=xg|)q9LJnAV<)D|o;9a0
zID{zHKYBi}SBvmv+@(}l3BnrO%&G-7^DsSXdTS!nFNlCc&+)$sWw0nzH)N={iY3b6
zik|e#2(PYU&RWzj?D|Yq{kJD%;7)dypj_o8*r0pr_^$q6L@u_Pe3fkjd(E72O7hB2
z1m$QA9h%!joI0K?>{tlF=G>tzyIb4X&k|?C<@OPTM8b6Msz^HaSLY>vEPe*6H_5N?
zqEyiIpIrZm35mPyaZ}`4EyA|oq;_
zN-hZonAY+lT$b@7aJ;-YEWVGx#7o^s)+i^K{qM$NvA7h>tP`>czf2FjW@9EdBKZF%G
zuR4>thXc##zVkVp$>0N<#(ptr8cf)GpOavg0`8exdBpz8g{Si9Ud46SgQ6Az^>+`u
zVMBhKnK)emAmlAEtL{&OsdCCTn%oV*G?$`MLlJ}av7+ZiE>1%Akzhys?K7DCcifL6
zJ00}>Evl4cuY%$CygR!ZieN44cm30)J;3@vyzcwcFtBQN#75xs5>ip-&
znPlksf^_Y`Qy4m0Cr07Hs=jMOy%(&q<;=nHXeGvZh%5dgyA(Z|U2}9c&9e!lOkVV$X
zg-SU1ysgc>ii@t_e5q)piQtsd0jY27@3&v!*|~S$WdkN*pH0rw578f>
zGSzU*NdhCj65FBHS#$!szh0-x@a+O4B)6`PDSrWxbk~}#%92227_P8-PXyg2R&s`gdMx4RJL
z9Zg2~#5KYN^1H1nWxaraL7aS9HU~b@UFS^K$%gG8cAipo*Mi}_euZHJ2vd$)+eTCc
zf)AiN=yDN#_7n&xFp`L(VnLTqHj<+HNYLpIea_;I+fX=)!}z;*KWuceJ9n;W9o*jZ
z=200bgo=8mD@AJ0Ac3%lVeq0g40PR|x}K;A#5l$>ivHaP&3^J3J*{{Lt5ZL*%jg3)vOOcF15n74XPM*mmC-)9Ym(|!4>%4_jrWMG2LZjeCIU
z^xyDE3ne(v>!EUyQ4V%Ctt_2ymjRh%4{Mwm|01VT?~(A4q-pi)>%GTKeni1k$IyvZ52!jd_kb2Q}|BBVAixCQY`Ike$Mn1kVIVvXdn70fk_^8Tb$vL(
zqCF3cdojuY`BzzBJ5e%ZM+
z(|~y9o9KSm8WwRXzCzm%!@hCzN|C(ILn=!A?8P71V`|%a2QP&TFiG0+Uz&7*h^S(A
z@#o#wSl6)m2X%`)3^83)(Z3vmFg{od;iz%JF3I^c(xu+PKIJPNkGOXrKP(>jhd(Pu
zbh~#81Kbj^!PtSv*KUR&40ACfN?)}wKD{t$-pn=Zs+aB7C)#Z6&4ymd_SZtBJOuvD
z5Y59Jdk8Q_#Wcj7_MsH{1`*8ERjxa=@B`CTRb=^UGK7d8y4PhqXu}+2S>-YXy%2Hz
z>X9Wj59CZNfz_LF9qdI7|F-hOw2uoXowMODC
zMDK7_-K#Z@zh{}r(8HW`Sy{Y?&
zSMtQMhnfEJl52|!XAuEQ%aJ@}%jW?*<%&Hr@Py);c)ba>5F56ak*kL=36VYAl~KdO
zbmXk=R8}M8BNs1aP39usL!?t88iKH$@g8E$4=#w#YR^b>h5)kuq?vnE=n}TKnx4!n
z^`FAw>1oMor8mv>ueOIvwaXP$_G!QMd>vQFd`f%zg%BZDcW}|9hKvAHeXhIX@^L}o
z+xSJM^O23s?W(HDWN)!%aFbI$yn!2W1mPnzbyUcXRHv}e*7xR{Vxac9*Mh>!tp%P{
z;&z48aEdO=LKmx$S8in2X2F8j4Bf@UNReA~#~4)xF&uwL>6PHt;&~_B^vGi+I9;q_s*I*H%zs8Sv
zI|(h9D)3+lWbE;hX5t9X7e05*r^Q&e!sxiAn=!JSdsVLc-95}}GFB4&a7F4#y0(9P
zOu~ZmgQ{QFv}5_20a{Huk=P4H@?U{hZX>htaVj74?2z#AroH=EDdx_oGDc=yh;``I
z{b65wh`rKn{!FVFgm4z!5l?*^kNGDgtxT2HAS>@alO3NcK{#6SydL-^AVu}@JoOFF
zFtszWV}_2oNEo~S)n9j8u%L&=!*z;fi1zM7a&4wc?12)$uEg?NB&A{ecGk+o-W@PZ8|Hwv9}kiM<992RXUTHSbAG(Xs7(AlDJ{FzV;+Zidr<
zh)fBw@7lq55^qwEZLcHY4};A}+IBIgmx6^a8NVsD}b*xRpx1IzaZnTAbK3k
zV&s+2UaTjtBljQ81gr#}fiCj|w0HaGkl(r{LYXczh((2SZL9AmB*!=+l2+me_TV6^
zt>xP}z+PJ=p5`qL>O>XkB*N%mj|T6YjM-nLbZg_{+^Z!_9kmjf>^KKUsw$Lq2=2k1
zLZKg6FfEY2)*_Mok7IauL#)ldR~&e*;mg;?^r1)JCup2}$UI-K1%HH6&+iP#
zfpni_e#(1Nuu_xFwvPS|Y`5V|K!R+5CrVJ)v~+;I2~r*k+ZF|MhdXS-M9T2X!F#?_
zHV*)g^9R*W34Ad6r~0>jV@JBUM}%=IQd3-5J=s3U)v&>g3`KVJI3#R085L^`3D(9
zxS)oAm#O42y!^YHjY)hPuKeAPE;gD2-=)M>Ofyg6e&3;N4}TN*k6~v0&ZUe0cCMm|
z9tHP++lJ|B|@Env5T$G%ezJTuu*yrxz
z5@=8Bp*v{=Ct6QOaw8()JG?JveyQ^&2QI1>v_Hinigs$=*)C2nMAwB6I*6Q=a21AH
zmkqD;Xpik9_v*Px;C=in<4~Oim;d^ecOqIFwG`gldNRO`QY)*rN#veJcOGxwpQIAR
zr(WJTJ$*?DrDeGMt(!p}4;$bJo}$pf4bNZx*dRoN|M?(b>ggtfeiQJY-(}IjcP}tM
zj2%?QSs2a?Ni=DpX4;yfCab*oam%bcKc_72YAyDLE)T)qo91?X&38kkwwC!;scq0r
z<_h<~K{@m!+k$X*(-prv(YwfF9Ey5V7LLf;F{4?Z61Qej67b5GU7t-)i=pmR&QvjP
z=iUDUHs`EH=0c@MI6i3m*t(jf=6A^
zlDluQ3>DAsk(GkeXiC+vp7wn$bo#o@*CokoXp(JyWSIRmlyg+KUnul43Y*s4ZR|+!
z13KPmhY)*I_QtnSZ80U(bd~wl+*uPmu}Ss0AQ3n2H@j@SN~D1X_?$I!r%uK*rt%qb
zk3;Z|i`T?ZC4HRxUo!YPku8q%UX_X%xrcuxF^CKAZbQv4
z$5$hG7a_qriqy*7uTWTvW~X9zA0q0vM{Ehs;(w}qXF^4fL8>bwPr1cEopnzbo3idh
zu8v63QTB6a;}!Ee?w*wBRi1V6P&ZP%<(4@vVRi)elniVs4T;cP(k8*v;WDWA*~Ldg
z%&h3J0H^3H^V6ua5LMmyG&AnEee0(+$9deD`D24qnj`*fijhKllL<$EIC;NIQbM^L
zhe}u3MR6tn_OiEUh45&BqndBSAHn?krI4BjqX26BBo$L`2EjDE?Nq*RVaIY=&QWSE
zkhB-xcHiEE)5XrQTNY!m>ynR5d`>zr3EJ4Xkn#+i?V!KA$e{^aB@{_3DhSYDpRXI9
zllcPhba$nDH|yc%tM`w_<)6cCfs5-Y=}+JX{sOMw!xZ?1t`b_+xp{D!jw^C3bQw^!+lxJ*uCMrw2gTO3Lo%o&#AOUEk)(7y-eu)1Rx-
zrGO@&j{9NF3qw-d!!GzR12U1De3BZUkpy^Gen`_CDnF4jC$s&F(MQC{Xqa;VmJbY<
zb|RVK;%}KnC`$>N$<$vu{j~$FdoN#Q9*V=)1@*~h`|Qwk23eS`;%EwG5>3L>j_&~H);k*?9+1PK`2~;lynW1k+C45>i>HS2
z+`oV3N(^9arnZC)6$=PUiiA+w%r8v5Vou`G8EfER=y>(yh!vD6^rIst7ht?v$L$YK
z=|NT$!G+}oYUquXTyPyKM@XVab-%inAz9|4cbZ@{_KbxyxMR)}!BsbVxI`WzAGcn5
z+f(*nk0D%1Gkt{xw)-{~`^6)j#C?Y(`oY+sjS+Y1rX^A~Y5jZd@dWbs&vFox_gkdO
zHKt!7J{2qV&bZY2I}VAN)t?A*3d91+#edJ0Y+}hIH^>_H@)1XY39T<*vXDJ5(K@=3
zOl&oP_TlzY9&%aVlR@#pdFYvGZR~2YfSJ$~pLj_PAp$b3%5&Bo7;{rUNi=ICl0mZA
z<3C}5<&~?Ovsaj6PfHXSe5|C9!$+2CuTQ0;y5&8L}y)*{iEEPC`@WA_er3(?_YB~n2S=iBB}
z2;{KHsaMZK3ojr{=grKW@Rvx*v39mPDL1ljEkqnrE$}N%p0oMLyYFxmW2b-e|fdL__-I>@ZLA_SRomC(L{0%CmTiL
zeRMRsjeHPhpNI3)=7HFHdicPaStK&8=acwJI~N;yquA+wY8oq(2Hb6F&k-N{P-9Vw
zbZj>8M*M(A5wfk&*v@3zjVa4~?`oyKfoZzEe
za6Y%kB=W}Rl}w884A$arKEBlc9pTSxSmPllf;nM^!IZA&;00DiKRxF}Bs@f!wDHpy
z#Cq+RMUQ#`Q$(W6?|c4-?5y>CaQv(UZn=cT`-p8LL1DHxMHy=`0XbK+q<9J|Fdq~Q
zG`IkA(v}Zm@BN1?5meRNq%wghA7{~{XXl`@$Nj8{a8|G-_M3hnQV?of?OG$NFoAsz
z8cmJgg@8vw$yF~KDL5`2Fi1|W4DgjnEJ%kSl3w?Bo3E#-aDcJ
zb@o)ZNYg|h+0(b=vy!*L$d`juGPXGI!LR$#S>!7CPZi^5T2hA9bfp;Tj(IXsEk8K7y(n)ljNI&@WB(yaC^04LN`e{N4T
zKwaNQ@Z-5ssOEA3t){yN{hF24qJp0Rr_*kqok?H9IAks&v!ELM7%%xjFTDbqlbF;h
zuh)TjH#)n`kUW?}z43t9HWaiA6m)AykAi!nWf?yYs==E3tXet!6wn`;9#uX+0`HD-
zIbJWA24`L?Hysjwg9(G?g6!@D_y|YkC%LXEz?dY_Wp-r=u0{TPHH|hwnqy=nojwI_
zbZ4Wa(RmYoZ+>;|dG{wkK~qjcO4|pn_qa}+m0W_1+b_DpnGZpylh{f^A{Sa5RBkC!
z&;`m0D|wbLeuJfFqpt+@?tol|!)xJTbMVW%pLIn6;`oGf-uf;-6-q(duzm2I23PDq
z84z7$LrGo_DZW)h@uTU31~owz^bLo1-Gg2hw773Ml7jp+&gzPGm)4P?l)AxvX=<0qBbn0E({7aldx^u3Jp-)x*5r=!6=EM3yQ|dks(Lr4snc7J`CnTU!UsVyLMQn+EEpi%zk+j0!v>!YMvC7ZJy~
zqn~IdUoUw)K|3q6_rz9B(5@FN@-n6#=rw|R6=`WJJhZj-Txm89Zb8la=PaiuPT-ZP
zb?dArzG0DLF+{F|n0E-I+ethbsi{HR^hF;epdL;*^%YB
zhv(3vk|!4mD6gWQlGZJxFT6vW9yRcd*%Y9zJbKgm0~Wa6+qZX;X;tyHg5m1&`_|~p
zDgF7r+s&w_hzQr6^$Jv^{^!k_5jnJd^PDDYj|NU!xE~&{_Z9r?UW>ue_i`Q0jf9NOhCfwFRZWJ*P-G)1AE=wZqWGCTZPVN0_wI2
zH7BKPLyAsy(0TDY_$Xizk^BNhpGUfcDl;v43x7tC9k}xFN2!Bvm3hy3TmrF2wvW};wt&|g^+HR!uONzUh4yPl8?;K`sq}tQ3ARj)48mcc1aF2p4J5Y45|Uy}q)49HP|swv0$%
zns6YUTxSPpZdtjBk)4A}Jx@6Kcv%1W2p*++VN75!kd0#YB{R%VEiRR`vx92vmnYaY
zEh%?9CkOr$I`s!W1tWc7)UiIa*
zub9<$$;0I8&q)3y+7Cp%z1XULN4-FIC$jJwBkLOB1b-UOS8f#bVcqd3RLz$LuroJb
z9InswAx*#ekKSAv!qgeaGqX8);XM7dTZiXQgDTRA#wV_+n8ti*i(6(F=1zJi?)!y#
zF>5lPYVF^cR9++q^E9
ztR?^Pa5^6PlN3N(W+Fe|Xd5f?F>bPI+QmEz_3Ji8dl9)d@hi(`5|QV3J&Y*+1YHs&d=brid@W;p&KND*&cX$cA*|r~^AE&}53rcDa@HI9K}e6Wa!$0YKbHCQ>43hT
zFCseGb#Lv)1Y$Q!BKq^jE9^&%C}-A@GbX)uV^gr!BzM(T^*
zi)>zYBbYKm-G)|Tec`?i*{lUfq+vx`a{fWt;Iskb|!guj0
zLm_rf@4>OoFM!Oey%YKAaus8scG&aTQb%;iXVmXUdSRP8S88e%YmoZH|JazM4H4tV
zg`D-wipWZ0XcHQ0hMCp}_8OHXA^WkNv=@u2Ff&3nN#V~Hh+cuiXD_ul%sbuyg
zBt=*FMX6v7Hs&X^D-^bb-3#@s84~bD1_|V!*jc~CX6+wDakiBrzAiQ4abce@Eyi~S
zO0QQim;cy@IbssALfYSD4cpGx@_$6}U?2^t$v64TxiXBYue}s_m`o44I@`bMovlW+
zsW}Yz`(I-<_2<6l7FHmS9468%RHBfcLZ-OB-}@N%Osgc}1Vm<^oj5tiHDg>wn?^so
zdy$HsoGE^lKUmGpfs|@NL6EtS$l>=d-}#)<)a&8LZ%Be4RpEfx4o2LQEyO5B2Z|r5
zy0_R}2c&tG+pn=P?97Y5g)z!|$QT>awYWwMEgpUL)DixdcY6^wTK!ZBycRw6v!{v~
z%zV{)ZkTfxK7HF6|T)Sq||G
zkRS%f+2Ulb+oI64k&(J<(hcbPa*#Is6$T%xW&0ThIH7ibZ;x~_J7AYCE;pZ&1B;P!
zLV3;RP-S$nJBmydJmG0`?5WX)!laXb(%XyyRIF>vd*KP=;+{NyUKS4X3J&<(`E&uF
z_|kM(vI%rG=MvJ|wFB(B<+i6Zqv7SRHV=tXL%?{i))$}01|Z6=s&iL{1L%isJc>M%
zf^kI0yn8Yd0?*1^()LXM403h6dqkJKfQq0C$$wKWaH_!OWb~F7P$Z>(7~lR7-jsUM
zyF8@?{2G*wbloFB`r?!9k6c0U(I>Q@_;C#A+fGVSIa>p{*ta_$9u5PL&Yef?5Cr&L
z^95gzKLZXawzLcWg|L-~xo2sp4V;o#Yx=mi1)ftG^gZ=_4vsF#O&2`K1Bsl9y^C4z
zV3epCxn|uI5O{mre{A3fShSY3U}ByC^(t?98R2L6F`pr*k!J)@b3F~nTI_%Wt}j-D
zUQyv{ChNB{VGD3R&qQ3tQV;EmKmSzPZ3iU3R_4~^|G>sKqd^g?`yhvq_9Sq<4KQ_T
z2`wAdfqo9^ql==GFcor=Y;aSfY0rL|i29tyX~miX61sLkh&eCortJYF%knu#6-SH)
ztBt-=j%C32D8uaTN6O*jd<-zkZXe9}+%ex#C&Mkz#ELzq|IegqC@F(-^U1h6X1r1Bxh*u(4oGRUgcT!TKLO@Mq2Wp=IGg~w(`AtI$Zg4
zJ|pwS1+;`t;4qtA7!TYjnGrA5NApn;n|=*_^!YaA6DtYNQRsR2j-=)-?lhR>c{1vZ3tUp{@OI(FUu8K7i8(w+O+*s<56%i=hjmDd5oGQ>$yP1m}Em#FW@
zU$I{`74gW$UPYGcP58noeOg!HW}NwwF~d=03(k=y>7d8igwkmDM%)1vsPgD%q?3vt
zUwY)pGPi7jSDH#)&gpbTWgM4h&V)d8Yb^e_udf>Q3!8AOO^-kubWM}AZ2fTO=9I&N
zng@9QV8}b##AY-|k&s=q`8`VT<{k^tzrLOs4S^TjDj}$u%JTh$mIwHE0da|UnGjuH
zlZhLn!_ZRU;-$HaI&_usW8e*gH)xBlXzF>1T=Zo^26>h`M7NW_Eq$?sXamiWieXm+
zdgWFDL#uH!Wo)Gi!07g5Dj^ZD=Fj(FkT
z=k%I2H=Io^;#0vHEwm^^h_{(j78T6BwK7_!iAUrH=*0`_)f~V=7&b=Xz#9y
z*x6Z2G~BN4u{Db{dSusJ-}#pYZxf-Cd1$o((f)56b%$GkI`NpV$Z8AvX4z8syGfzv
zECeW@)m}um+vaq{V^8DUJsB|%)#d@JkVbllErY8qef0fLZsPxrT|XaxVZd=JLkc2g
z29zr1#cyX;VmzC=jr}8KD{PaLG1W6ThdJjSps~TO+`kJ?
z$MHUSu8Bi&?Lh01o_FB36poPoxBs>Pi;|G2KnKQ|`*hTt!gys}M^GT=A%VXWvI3oh`@
zp=&~e1m|q0ZJ~Yp1Dq}TWnDTp2m>kN>Hl6EfNm-&vo2Hlz*}nM{k_B>FwZaX!$QUm
zrdL>sTxT{0=No%|Mg~Yg)>b)XJ3JR!cu4$aKIH;MTaq0~E?5D9r$Qah8v1av@yyV=
zs|bMF8T{jPwZKlYP$Qv21x7e=-Pa6rho=9z9G@W0K<_iLs+yb)EM#VRvk-{`PX)Y2
zEvOufc8>pW5ln#Enhagm240{pO-H~Gx{%+K%iBW*CMJ3E!q9UWt?VY{5~)ugL|&JjwHfdaGmxxF8+-OMEbb@os_fyy(~ShIa(I*Z5}olQ!51A3I+uo
zxV|HCZ<0%Ox5VLpXFnV|_59008O@x&C}9Bi%(tdw4^VievhPobtP)67(C0}KQGw62
zBZEidwh)t<^A&?nW)TJ{#jZ6M3?seaAVD--ij;NN2Yy$2hEQhD9M%*JArmFnx3xzm
zu+w)0`zhP0f
z`bQr^pmL?8hN&328T^Kg`}i{aqQ$}S{24LWUr(%^uz8037Tb@gU>m}?z7MvtCM_X<
z`nT-0?y5rd|Gq3MnEb$^8)o|`?G6xKZ93$$*l%pQMo3LgZ4+T7xhVhkydx5EjpY2D
z|8g+lTIc%vZm$uu!=>vMvE3MNA;&`Tn{U_x;f2JH7j+TQ5h3#++bGP!t3iN;Eg5N`
zqYQXeSc^5@?ctuus6sriCZr3k>{E(4F-+1n~E?6^LbYPWj5Yq8X>{M`3HRen@`sOd`UxYuk?4~eX1`=xZ
z^TrNm8uD57twaG~AvShl4X;0eNRhw}x_0?2(8J_K_w?(r%YxRz9v8L|vuCk?pS~b~
z_J7Ak>zqoU)yMVTNZ>bw(jlnZjPWftT<$t{{&gp|LCbeol1B{Ymiwl>
zu1Nwu=35ozWUGi1h3Un^SD!H#*ZwNwXI;pDnn!ZxJzL17gbaPviA!KH$5G5j^%zlp
zYsmXk_7C>iDnF&M_6Vu-604aw=>^U4ug)vqd-rKRbX=?fw1m;Ik0X=
z0$M3EVU_kbetw~TN|cogZxL`!*YV|p
zPW7nj?ED(|mPx#U=;$>lWPQyURox0K9H_XC+UwzGo{S=kItYvdU5RSX)xtwH%hihK
zv!G{6?%ub>F`#3Y!j`!<3%rukq6yBNl_%O2q
zpk5uDabxau=tnA9R;5Eyysil$+})=|Yel$KGDgMmciax1L)O|TUm1P(@28BY+Aqo|
z(D(~xdZmiH;fvs|IZeL!0|`7JPROv~&|rMYAyg;wz~SqF$3qn=7qUyEDcp~ig~
zevz+T;AHrEk0(UD8*0!fP0W
zPW%ZBS*ozbbMWqu`+}M%!@7K+u}uS7N9wT6&QpUj?~IMjdIh5hkdsp@rU=@6x!*_@7mO5k7pjRJkGQ-X+u}^;P!Nk2APgrCRM>DpNF{
z@@(y~;U!eW-*%%(UjW~gBeC1!XGBwC8b>wghJjo=F_SQ}2HLml&AXo?hwerhe6mra
z!}*^%O8JUyf#mlmw$qHK&hXPTfP3cEO?R_UH{Wz
z^}&(}tyEb#p}*;ZvJleCm_T#Xy+G!8$@)6}xqiNCF=8DieXyd>C1%LMqz{5oyh;Wxnar$XXTegG2l?AFjmcfvGL|Cr2}Z%~DE|A^GIoCGtB&C7*?V=%GG1*spGL(2uG
z;^eJ=!_*TCG3%ZGAftPOnD5FT5Nxn5OF_8|zo%+NY4bP3E={#Z*^aefcaP7HL){++
zu*Kc&VReCgOzD9N4RIj&{kFtv#Yb3p{bR?`el`d@PHI@ud!xJ8-e)%)1RVh`;r1JR?uxW+b1B=;)kGK@g@Y*A%mR}gOKSPw^M9K9SA7*
zm8HurgtvSLrmI>!;N83=%CgrnV8xb^bCbak3Wj+<)N_)79u*=6Tzm3BK+Ev0f4VYU
zZ6;xy6AA^LzPy)6GnwJrsy@+^b}3lO$SBl&MiS5zERvJGmW0{DH}sgDg27DKB&mj#
zEP%rnwuW4C&~U!6v*EiW;9|OzP|Ga>AG_pe+V#bO+sR2MgCZ_qmZipjrsNiMdQZwv
z=Y0*_km(&u&7MYxvF)pW2!la&n#H;Yhju?DCH2lmoiEtI3yeRzVLmjC4vOsnpWM#@N{JYmv
zq@q{G#(OawTMJ8*Bb%*`3qt?R{KE{SC)35>AtMV5-%E&xS(!*hS>ThM
zhU3p03TWnG#@4KnVO#SrKyf>M#<1%0DI#LIn?wxf0n
zk^5ctk;~^4G?RZQ5-@!USe8eJjNJ2IPLwiCoCos$w
zD)Tae2VmYy8qR%8a2hWoenZFr3@2q9I`yvtiZ%6|r;9caEsD)2pHKo!X>2{w#OhGA
zMDq1BLIY5Gljh=(lk2~EBQ6HjnGks4h6m;m#KicQ7hk^pGn!5u6=l{VrJi61sUApQz}y
z01NMLzn^5S15%}5g?B$i06m7c@yd~z;1bDSM$^GcI3=Yjg(0J0uaP(Uq&JmRkAu+E0*qF|8x(eLy_RRisTn*{$
zt!IKM+kioD8rD1f8~*I+DtWW;8+h;Ee=sv!2e>yrUn$wF2e*EbslEYY(6%|Z>QI;*
zZIUFkPD@>diHClk&E}hcY~XA9*NopV{ww#1<_Bn;yXC#~^v&Y_FnzZ4P|?
znv!^Mrx!e%((&Uf0J!zBx7ePFHL8EBE|#&61)uV6&vh8!LSpVaNB`QT~X7l;a37{2SqP^5K0Yv?Rq@g`_f^u*G>&roUY({%3_47
zH?Ut9{cel1yI*E|U=x6PoPRbg$e)WIFf&Ue=4zr>RA_`7$-U9Dwi(W!cKvYSfg+2z
z@vG>jr#Q!lq&l2RSDRBp*aH9H^QGykj1TU`l#?^CZjEOJ`=06Typ2kU)Gkn8Zo%0N
zuXN1({O2zQqbZ!3j9E;s&Q7&dqLb
z6&zoWuX0>US2=9O?~@Rhh#UOdwIHjTyTnw3F0Zr=iZ(=|d4dw}rfxjKzp(p=N449d
zcXvhJM59e8t-4&T{G&Sbpl{~%&dm(;=9L)4Gr=3D@m;GO0ABcWq9B9aJcvGU{Q4|s
zstFA-ea61jQGsToD{1=U2`HQgr(7qcaO0SG>A#6M8vcyajdc{FvNU4GMIY-?Z%b!R
zqOx+7wOoL1r@8=ru2WHLbp}Tr?j4T#r$F?%`N)5s1C6MB==b(^s~VJh#hrsRw-W6v
zD5s1neuO$$%{LDlSfIMeCbRwTuj2(=vr>^twsW+dS5U%t10dqJAAbM3zptBs
zI$G73u(F8UL>K#uEs`Io;(LcW)6u;Ys7Le19(z|lynL+T+YK5wd{fmv?^U8K>d1N0
z?q08mdMJpRa|CJNj~HVV`&UA0!aG@$gBIuuMR;ZitaJz}k;5kBvCKLvPXtJdb$9Q9bIjrCA6cS{;3s;yW=B-t_8R
ze=f-sC?Get*k;%Ql2Os*wI?_6U(Gr_n+|OFk&Sg{yeU1p@k1xMOq2*O=5!DoGR%f$
zE7@B^Aq;46N2lWrHAOs!P?qB$Y9CxY37^%DB0@WzchhEB&*INuOEnSmJycNJiMR9y
z34R*2TvV76!Sf8GSn|`j(R-u{H$~r&;wpqIhCOK;FiZFx#!LGFc5lcn{T{9aU;WN)
zyS#k~uS<_bOQrdP{mewqM80+4d^IJ
zsBe-k^G0M0jLk)BQ&)b1hf{@o-;*jqkL_dL&NKA*nb?NZtU^NkRdeLgoZb@X(#Z|7
z>FI%~lPYb|LtXINtHr@``9#qE+V*AZC4WFs3RI)DOyI}92>k&;6(Ci7L8HP^2rTeZ
zgtF+of+9xRujAg?K`29gx=hjr+@dhLVo7fdv6;Kx2%Q$lzuLlMpj-v?w~sW&
zj8blMoHO(^NF*3`y$|kRoH}&rvV;
zfG28r=E&C`yiy<~+!TV)TQolD#G(eMB_0@+HM_(8J4VTKYc6myk}9W{Ee4RLn`@+J
zor9l#rG#EI=7sw#)C&Kj=)C{Adfzx6lD%hEvNE$}o%4{>-!&^AKvftxXyLGUeD)C-$Nw&3C}6w8}ry+3jfpQ>H2r+suA4ZETuwQB{!zLU
zp=I8$x?(Va`R&@hso)|3PEof5JTrK~p9|HCwOZ3i?TPiH;_GMZLdV;u4_;3q8;1`H
zgbkLlsYiktBE1llmIyO$p3bkBFe;;;=g{{PT#s97uh)a^}$02y!M32r7)HfsoiO
zN8@e<@J@D%;lZjDY;0_)KGH@&8WzJ+Yh?lR&u7&JwI+Z_FNK?g_R_#(G?dIuP7Yo&
z=%L}k)qzY`=3CcWjxc9S>E{F<2Wks>^CNqofW=H!)4m_^z;V79SRsWlQ(HH$#rzYv
z{Ixjsr0548=UW_-TzCoguZtB`%0$Al5wVs|?^{5Av+(X;s~XV7ni@YDSpg`DC~8Lg
z^TCkRT%2%XCiHk9rTU~e7jO@V9?Qj7LGhlYHWk(a@G-rlpqMZd$b|ePk+({KBrFH*
zyJw$($UCmTq^8nfaM}A;N@v=^TIVXShGGgJDDZBDYxm)G86DEHay}6L@T?}RdM|i0smOE8pKG+Xu^!yasSw-(rzdl40XPb@tuJqZGanS#pahafC=9C*b)3%L8QtBf8U!mOXA
zaxCji=PzB-
zA-*KPY_w5_nkGQ_+7usA@#Lw0>xDK1+nm04a{r_IWVBDF!+`28ls&w1T>|9^>T0U^
zt&I;9aEY3cc%T;VJd|=Kj>{pFw~;8vBCHT04TZ#0Non)5e0@f&nz1aAc5c5hRApT#nw
z(|!>dG1c6tbf>*vbG9U|9jzdCzFrUgVkwi;Hm{7YrkGVrlT@QIH(EQg%%7q0hemHd
zx4YrebFa&XHPz6fVD*1B5+>-ob(h0mqt$3jnU34V?JV?4P@+&%vL&9Eb9=Pl)XCoD
zL+10~&j#qXaO1&1k4p4;jeWyf$xBq~bu4LjlM%YA8U2~8NFQ%Z>Y%D>(m+F*m^m4(
z@==tLgnCsk70uVXuxi9-j`pf3)J`gE;k>Fd?APwdqPG9`UjMt5hO!HOugJI*gYMB@
z+I>?mgIYwMd6WHB6~Dso!K#|7jH*N%3TnJDL(^|sd&yDqXN&fmX#Dklm*Z|mbUgyGy_R|&FE+g-f7gQn-B;$0{a5!H&K8!M
zGRkY9;ud1=4W#1ew~*$B^y`fH=?}mlQ(zk;Ty2%SyL|{ASC|ZH28yG4+CHqsd)%n)
z+89zLaR@yHw=7ugYT*2qPGDaBACR0y_mx&$1-%sA^r69@2klL3jrvtUi|^)_^WvMo
zfhGHs-Y=Wmp#Mm%KIbVt9=w@E_XzC6CN4F>F71=w>4IrJLhu37e4%(ptlk38blRL!
z(%ysLcHS)T_Wgp7dYA^Qm<9owL2JirQ3)xH+L{yZwnNm9HNDE^A1FNcoDk&Bz+>)9
z8HG3UVcOy}HhglYlGex}&HWF9QJFfVWg`AK_YFCCbKJ#t7YDF&GzYk8p`k(
zB~W`ivqet#4Uqc5<@z=@07fRo8u|Zs(y!p0w(xh11PjkqHv<1_fxDmOtu9IBfiuB;
zBTr7gvnNuB=k4pd+pT*B_j0%I=h`9L*P??8i7Di&fYIs{GSi}
zDv|y8-mojko7tT`$J79i1gFcty!M0T-{?%92lxWpt)Ku6=n5OSCJtQttwAXBOclLS
z3*@aCA7QTb0%at2kL!!Pz*jlc#HijAI(H~njfvQEXzh8E#q@VXx(@kIl;*dgr2
z)&!5Dhpp
zK^H;=F<7&t!XQ!S4Ge`09D9RQ0o{^gIYF}ojH7tNeoclS{3fkV-U=XrvUL9y6ja;+
z?x+#}ybX7Bk?eCqs-tQCzD1buJ9
z9%MbPzvWhk>@ScC3H)xtUQ!cmiw?CT$&C!tUUf8}mV`7l%yt4ZRAb>3Tph#u^s~(W
zq^u%@%nOdXswaI_OB>#Id3#84&*wz7OS9Oytgli}Z?0km8P^%-b@!19VOG~9tMhO_
zM17fanh6km+;q*EWCQsm-xf7JXrYd_{-Npy4_IY77>P5x0iEyoRJ}QR3O;J=#S%qv
zgF`>#dU+KA82|CjZRQFYFekwj_2{lSB$G9hOIMEs17_b;uB#m(he&JZ#mgk{w&adh
zq8>GP8YLBzf0hNl@4gf({M{Rx>lSjhI-iDAewAM&iDbYc>7Cgapb1Bu3R}+}d4R(e
zE{Zc1ad3IWwMmNk2zzmVm3j)Og83JV8iW=%p?5-*(BhpSaAx;`!Dd!6oHm+i(8$UJ
z!1E6?$36VgGVvG@n!|kdWZH
zADZzA6f(Za`L84f256k4UP|zSVZDnyH?vcKb?%_g?3)hwdCYNZ_u?E-SV5Lx+X>I{
zjUa=mS0cQ+97*lsPz(tDz5W}#Hw3?R9p{)BE`s`N9b7jFLV%6xp|G6JTkxs**tmDL
z81@9~$Q1~0z$SvA6|xKoH1|Sz3IE+;aEYWV-g|urN@#wNXl|GT?^!J6Ej>tZ0lQM}
z@+;D)kY7vbyBNW4EQI8BQcII2MuOH$8(^?2ma<(s1%&ww`;g
zHu(7LHgafk73F^RvUj0W8ds^TKAd!`#$B|fPFO;D__x(RIXdb(_^*+s(?%cl(KCV4
zL~Z#7_^!<7XE|}TxcSSG!=RUWxV_T%Dm-5meJ|lJBJHDxK7UD*6fkLqYpT%PsGH>dWTp6V18>Ft0}Zml$DBY9Mh{pJ3h7EV-LclG>t
zavI!qu%2QFVq}=kasSgtttCwn1bFed__==CJ8EZzf5H1|`hZ@h=LKDW|~
zysF_TFP>ZTxwYUy^U8HH#~Ro*6Y|&AuM2h$3?5AcPr!z;fn(a%axiat(iSeX0rWKe
zY>LxmP}G5OF+Fr2_FZ!=d}P@QV)h)_FU(be(JfA~Q}l_4NXC3
zs`9?MlYF0n#Qox^#CNbu|E)Jlo&p~dc|Z9w(E~Js*!!c_e*lB+>s&GykHNjmW3^`C
zpW&4zeShxoH1I3r%*EQn0dRp6*VV9XgoBSBtV|b&!F;x4vLRc4uri}%C^6~@HHhbF
z9~jvH;&rKURhv?%swJt>loV2Q>x{r^DONY?b+mm;XkYuRF7z!Jvzj;ouY^le
z74c@Pej-
z=0>91EU@#^@@x+ECUT09@`?~~6eQLLAbg4ep4EF1RVtwXE|Fc^pWc^&)oBj*?tc;k
zr!1FPG`$`}u}R7i3J)fD=Wz~4mx(%f_=8pPBfSb#bzS1}n2-gc115VW-^VZ~9?J&6
zu!YHR{^2b7b_$?2ZmaIc0`RMRSVc_SdC*T?#?oQkj|FVq=Lu9=#(Wfw==Z~R5c9vE
z;zmiC;m4NgcmLkVLB+E8iR^$nY;Dr&({O?d^1I+6RB*JI~~RU4J>UU9Q4Z-BW`Ve%BAoW~)cK`XtMX&k}*D=eqpQqn{uuCK0#J
zTzQS9e3y<@|MmuXQtr~WTUw9R-Ls7wB4LAzS^+Y=S%i@J-_yVL*EE5(tgNW>v?=UO
zR(EgCiU0>vCw}_;bSQm3@71eYve38j>_@>y4M6Op)s7q(|#+3vYY}
zZY6uCItbjrvp>Q&zjS55{KeH;MvoEz%=N|NsCpsu6RCfv(x##L)gwc-KL#+ujs4-N
zG)-tZnZ@q
zB#_n@1UR*fmcD=Q1mdSQ*EcC-fpb~*<_1vNTXVo#rj=hZ-HNPxp~N`R;bmkp25vH40Qa;
zjHdEQ@R`e@vU%rNPdtU3La|%zpm6R^-VLM!Dy`q9pAY{4uDgzOA2ojlPXhlvkvC_>
zV=g2a6DXbp=D+FRbZ8w1@>vq8tX=a^aTOdB^u=1h?f^40JW2&fC9^ci+
zq=vW#u_nE}paV*gmh@dlEeijzyMJ+r#0701>)UjVKgk~-94EqB6i>0?iCyNoit@*t
z={ft~4O}Uo>)E%ihqzkF;g>e=^Ej_cyY^$Ri|AXit+-()6A1KXp(JG)6f4%@Ym$bf(*_q=xgA3Cz9_9E)d;8XB_5@vJulq1fM*oO$8P>
z|4amL{<=tn1_GSG!=DRnFjqhU*Z&r|Gdk
zbLJ-b*h&ge-~B+3;GB18uKjH`j&PPFz@K
zGxml{2VUxWck5DL3r=ZB#3o|#8h_e77_4fPivNsl=!wwmz~f@f32j;0@hZ)W1)(Zc
z_{tN6U1rW3FQBd`@>>f)!$S{`c7C?uBXcGG9=A`hErt|HnHTc$)x9v|S5!eL^=3ta
zY_S9ScCXsp!?_x_i!eWSkIu&(vbBx}O%VLad}vCKxFy=|b&6$w-4yT7b<@1xT!l|Z
z`z|^!r{PQ{MVdS7chSN7oj#5ZS~y`t`18?|eUU&)1k+kb8cxXfiwxXV$BP8Wqebn^
z(5`FDo`lbD;zaaCLhCsv{J{OOJ>{wpbm`^JrMhu@^ryZRdpze&yv
zw}|Y^9;@L$zn@BA(UC^icyAlF(Q@M%vUUdr!DM*u1DD`g0dw4dqfkuRR{|x!Jx$12
zaSnZ9Y9I1kmmP0cAMkf0q(uo#m3%%)sH0**lYB;{GU&qE-Tk>Cdi;H__~Y4^-@yt+
z{mQ_XW5C>R>9uuP9u0^n`getb2iLCKU7c-Z^LOGW#z+>V*{uiP}qzvN!mdu12aw+>unT_r+$Cc?d3
zS%+Yspq6B)N*!FMzEI@AJ^@@lYS}|qCiJ-Z<_ag_Z?K$B`1k?SEHn+M`@lNX2YhfNQL{OvYe1@W<+}7#fd*!N2dwV{23lT)gB}
zPW==P>D2I(ynh7A7{p71*DK*H?Sj@2Wh8K6znj!ZuMcexg$etaLZFwJ_EE!wR(QTB
z?fB-M7K2+mtK4;zV
zTmyba;tH-arNXOY4F9!^s6%*hX0coT8n9-Od8hPM11_%rjQ&@y2;>K@LAk
z`k+Z8(B9azeOL7eQRJWI94_U9%l#I@9p%!1%RBghg3k#^-xB;uu&M>hn3SArl*OR(
z=79Ii5(i+fNO`vOYNX)At+
z(OF3mUQl|Ago$pG{{6j;C9(KV)p%b+rV3glKYj^CzB4R7NbZTl_|@I+WlFq4nDpcS
zdXUZ{T%)2h7KbL-TBnDn>y#J9II4dur6vUN;uSA=$(W0kSK0pTnf{EucjD%6y5fpm
zi4i{?^&lMKzgnzmG#ihpl&rf@&Xgjb8iL5KcK0G5$gP~(9-V{po8!K2Zs)*8bK8MD
zxgaF#o@eDxQv=Tr4khbFY(Z3EXKa@FHWpqRrDf^8ihLkyHg|gY0}ClgV3$hV#FW=N
zUW4D4LE5aaTdL+Lf-@wOJU;ae^FK|c5hAmRbl8raV*O18*`6~JQ58zV;`}Y02Fpq8
zt3Zom!<|`#Z`5ivc6bc~1^2#GMv#GOMeD5MC*+{xN{On9;~Qk{{SPl=%gjGb5!yjUJR^6gN9f0;_eK3;YJ{RHeWD$#Xtf4!jjfaVpV
zV6J;sL_{7u_@2dXlA#0rS8wE~_q&1@Qhonvo=yhi%nDj(D9%H#-aubcM;<6_XplNh
zFAs)=z1U79A5iIczFX^h6bv$dE~z{$2kjjE;s~;pfP)&XSW~|y9Lujg!_4IfHZ0B?
z(Un-iGSQ=+ZjT9QIjyM5#5@UPxoA0gqZiw2S<6bat$$|1cpnA4Y(FifC
znz!QSwLl`xmB!*mBNRGsqS&&W3|?Ce5>-b01bOMY2g>(O?o#ejJ#ux>A>3yA2+SQjzgk5#UkZ(`#?jvY^|Arp)d1
z7g6=Ga2N_QdNM@jpkgW
zjj){sU_NmwW|U+bqB8GhX1_6^eqqZBZz8!-??Gi#QA$hPmh;2Qgu0vPle`D7hM(J`
zf?VI5op^K6B{6@?SKmLP(S?2rukTr*UhJiF8!I+A^%l{Kci*kiB)Yn)3tv(6Yw&4k
zXK~_LCx~kI=n}zi5{=35@)A&yN%x6*K49I4nwb8@7PmWkwpK@wfj)nsNtEsxi6Ba
zl+kbxk_VvP2JhrFB?5guC&jR)i#6>%Xy>E%X&PtVqd{xe#ybB_2DgICL>Q5Bk+xK>Sbp$+wf%G_3J7)sM;*d-}(AJwZbbnty)i4z7zQ6$+t
z!Ke4Gyh&^>L&vl7{hNZ{qR+mx6l~4j#ffsyd)QnFN4@KwfA}qM4>umZNbFI16}`U1
zZeUb>;<0ak=+$|*4NvNBBTJ5E=(I*H7^rLM$d2U`{HuUFttu>{@HlovR|`6FH`Sx3|&
zJ8^AMJrup8^-!??Wg$+HZuc=VI}aE2zHmzZjWK@QsM)!}7mSA9dbN1ZfdgIA=wjO6
z!SPf+u9#b-ZMbawFP$Bo0zBg0XKV9^p=d#|-09k#<`Iww%`?EfNzWig#)pTy$&g_=^%UA>@KS2jm
znuy^^;WQ=vV($2Bk>0r%r9Al6d&;sqi=!Ei{auXC*IpI>C
z?}j%=xN(aEgJt_is_2@Ch`>{46!!@euSwms#`%k0Q-2iJ$NhGh_B2A}(0gMdFDMSk
zaEpi6XM%)y@qrs#UkWKSaY~mBPpT`DXpKFiH?s*Px|8wCJ|>(IFNxtaWk*O+&1+kc
z+>&Z2u~8?PMkooMYC-Y*E;Ah-oAdJXRW>s8bC17W_0lYidUArDsJV#
z!Kmmh9xnYBz^UDoslmDgE@AW+a-yezaa)fu)t5$Sbg6)5iZB-_yPv`JPwoPNA7pO*
z8T$t6`IatTrK<%Oq(^*gZWO_QTjL`MeVM@NlU7-FN(9Wi(zl)ylnPT`sCnMis)TL1
zK}kU$qQDtOU-jKIOBfsc)hs6N0c?!)5sfT<1EYt?HKGDCK{63~FU_`1sH@P+u$=LLL3m?JHi8%71}V
zoyG__OI`C*D$#(X$08HG3bf##UrQ~exep9xAgA<4^E1
zWUBuiN8Zr5M-FF-gV7@K!1Db+7^&J-W>#bXt1^n~G0_=CPOgf*OBd#_q@#lhtw;gr
zVo6Tc+4UcG!Pg_~W7jy+-o8am#yNuZ628*<8~F)|>dVMEt$7~I!6yv{`YQW`S$K=hZx*{TH8%b*~-p6WO>9)KB(vfZ7
zamMnfHY8zy%V6z-79!awmnxdVgT3Z<|F$O1fxMu8L_FunhWx&sGBE$82pLhkpK?)#
z1DiTWaNJUG4x{aM7`;L(hEzzI5HpM;kMqO0$aw(D2dcFg;+m7AF(ZoCsnSZU97BjQ8MxK(W~mfG^
zT&_f-Ie%yp^_5|cgtO@v05-
z5&o*aHXCEEQw_TEUmLQU!;qJ7?k%#gpe8n{qJm5whs0`@24E47;tPq2A0cJ!JnkY{
zwOC<0@qd+>7!X
zgoz`REOTfC4WjC3-a>=rizUZ=>b+3im_T+yFXRDwxC636JBj@>CN
z>i$?pjOIF)uUMDRD)exwV}3v?%1hF?^}QWe&SDa`b_ea1p5<+bWM+-ohXi8D@IIfq0_v*ycqA1LV&b
z8}lPWX20jrn6#usSI)uk|4StW*XP(f!UKEQCx~LoO87
z^b%kW28qB_h(iCBT^&rn(X7R?(FZl9tnE)dv2yXRWYd4oIsA(P2M
zv2Z-;g!N-l1GP3d-U#z3!%vP|AOB@*fkctn{j9G>aDKhVvX3(W3|}@5;&Y1z52C_c
z_~nkVA2JW-A6tk)TaD~f{pG44uD{7`iqH*OCB*%SqjQBl7CW5Vx&V}A-=cm>DghJu
z&fn1d&IjCdSFP-CSc7CXq7-Aj1c(mO)h)HoL7aSp*d%NTyb!)_G}E;O9XIF+39D&Q
zpXE?}R!)BO#$Ta^{{2Q+v-p*9y9feNM;&jkgpbhswZYQxtvSH>d%{ZY$q`WA?^wP5
z=ov^Vy8fjgt`Pn#I(%^3uLe}8at&s^9D<>(eI%znR^XfD!_Ni0kKxT!ioXKqGC;sH
z8A!}i1c_JF@Au4fgE2g7a5>~Ns324&914kqmNeiUE>G#T86^!5;Z}!
z7t$AqG~NS-yG=%ge`V2GpY}z2O+`HSm^t2^j~49`Bon?(Zi@fx{;Ia;?}Q8eT)*|}
z=Ll#yfYzk4zktqZwIS1I|De``0mH5u5&YqH^p(s{H}T>NHp%%eyO4`Co8m&w4w%4J
z?g?>xfQL0Zt%Wq_QJaNZB(}Q3C}D(p>>$}XcpLxETSj;jnm+tOeEwoQydk`pHP6h5
ztD5xQ`a{NmKNFf8h*lniUTs{pJB;6fgNv-3mUkYcps_PLI+?3|?BL^1-zL#V4~@`Zz}P
z(XY;0k3CUK{G4IKTMBbBT#cRlO5b}~G~-R=CG(qS(eVbB6z@P~95Y&S%cU?y^^+(U
zoR+si4{>Wv=6OZ@q%qz9F#(_*l9MNxTzY)^Fj618X@=jBuhB}$eubVNyL0?2upH(4
zO2N3RR)#KF-fUvNn1^0fW&eKeZz_6srb{A7QU7Fqw|SnZ7K{r1+?^?Fdx&2BC#}f(
zBMjxJ1k3CZzUU*}AFJ`u1ZSa2
1IgQp+u*#Cd-b}#lJ4gY10kCa5kjvE@`f8+wi
zKV=%CL*gif4v7_7$U3g4I%|$1vptj@C;i9#X-bQ%7FCpYnS{Ull_9RM^zv?4k0$Q*
z;xMoAw*bl+1C_{cD5K5_t|r-?GH9~4?_}(8DtdD!it$@-1Zqqsv^k!ejB}BW&!3qq
zzypkp(ub&D;rBePvJ$j>(b#5_`^4`YQ3AdxpI^J)_|cn!xXFjncy(7!9hdk+T$BCH
zaOYzq9KlI6&V5(KX^?=uAFob0bF7=S$-!1=SFgry&6EwwcEwkYT1E}sadou+5G#nX
z&Dx|12+881+?mc%+FCe|jb!id-K+TaY{8kaMJZH#R-iuK;R31<>PgsrQ4BAxr--K5
zR>2#qUVa(>q>9(S4u~ssD8>)0K1UgA=Hcz%#k@v}?NI{fO2VRy0MyHRnTa~h3fKO|
zly~OKL!2Q}*|fCT6E_n0H+o0G1~vDR*bcsS8NW++qrigE8n+X*c6g|5gMNsM_HR0}
zM0-C_n|*#Rd&2)^I_|W%gbNE_F5d{3M(?jEIF|Wq1`m>X_
zgtYO<^||zvaS62AmG#Mbx1StQ1{~Rt(3gT1J30b#i$Z_Uv)0B%p4ghNr3DL^M
z5x99jMmA|%5l;`7(VHDVjVFsWOv^VgpbTWMp35H0K?&`@25SFJK@!c!7pbikaK6ir
z)x8FY@JNF&fr!g=XmahhKIeo(Xe2AjOJ^_#i7l_QDlL%Vy9%DTo9F=`@fG_e8a4#}
z?b(^TtWCj>LY=+!SN;PY5~2JqZT~=ZngZ9sRy80aM`F1j55t6v$H(cboxog(r65r`
zA6%ckRQK$}hsw~PVe6Xp0lsRwvH15>54iT|Zg9tFB`hZ-q3eK^@Hg>-&+lvfP=>8q
zn>A?)Ji7L`>6%6(%vfKIBK=+f*}odQ<$P}f-wAjWI|KG>IgO2{m59a!2vMPg?BVU)*Z{|~21VR@Zi
zD@$-Hcon`PTJ}d2#9Zk2HEwW(QqpCjGY`B#%|=MI${P&G#UwXRNT0A;S$*ww&j!Pz
z)&}bD6ZgT4;eDd}!OoDEfyOlWQ7!m7r*nYa^MOC$R&12IGi+~>@^&n70ypRn*U3yB
z;cPltb?&Gv(CXYWl~D79Z0ynTQ(UUR@xL^QnW;IfW>EwL|3TmvuO53C(*2<{##Wtmj^fD^LL;O!(vaL8~j&aqe=%2G4B
zz1+P3j6wsnu@nCD)uxHfp&#Dxb?wDFt0UK7IlY#4(ViUObqMYcX%>V-;oZD~e7ne#
zTn=*=(YugSUbw1FSsaE8$8yO@@dL|s{UF*v4k+A~>S^Cb2eSJ_3Y9M?LSr{}VX3cl
zU=StmNQ;_4{H_ueengis|8pL6)zU%Qal
zYNg3PajZZjD>?S_<0nRXlrzOUjIf)glDfQ|>#ZJ@?&Ft<+FKh|@@=T3CY7J3{Pg}$;-J;n9{DXc*j2ECjoM-Zo$p?O9FYicMfiFwJTE^
zCAyUB{@2=m+I4dN>8x_sQe#ox)PAdE>PyF&|2Q$D>BDaO)8nn_s{(uM70b%Zvq@hk
znwMM0Np{*5?h_-Y`_C-+<4dhM!|Km!BPp@7ow1^+YYNDb6H4jUEEw8
z#aX0|Onr+rMh43o6-LC)Uq;?&fQg2z8O)uDdn3}x8qsLlCLucxM~D<1&QXs@Vp;KwoOy$_tzOrU$k6ZH$i%T@xBux|
znAqU9o9f|R>>jgXw35~uV&IwETy3?C-BSO>CNZ~(SgmT-9BflSk3X_SCN8Ytfwhau
zxv(@0<&Bd35?F-wc1q6tr`?R4U!x(
z!X$E;qyp=zyrtY`+K24F{eJKKQ5-@+w5rP+mx*M$zCZnu={aU4V%PBMdLfb#$L+5Z
z+JWut)o$Lt(S@|hGY`r1=VG5-mHh4RJV&e#l51*{^RdmhZ9fC0A(F>X7Kwu%hr%#{{XqwE%9>$SE3Ce&pMh6BE
zEU9slmM7KWe;c_7xd>{VI7UX%pLAJ-64I%Le>PH&_ncs37-5
zQ*GeM`_LRooyjl}(zC94!3<65mELk-N
zo)JkTc3a$R!_xhx(4~pz{>)*e#fD+uQfAm_zL2J$UeyM3HkTx3I
zGVS9B+|8YXjVGACj40J)U9~HKdt$sk{hTihJ(sof^i?nnt9N^Rv*99e8Ktc;R$zi+
zpX1ogcv(RD59>(p=G#Ckw)uhe`*^6JNJ&^ivkV10Yv&$@ZGuO|Y_qy@Js>9Cd12j^
z82ut+NpiM(9e&@RRJrTb32UbI@>ZG~fz!h#C0~~SNJ5hTEB@+Fppd#^&#kft9&EE-
z6HO}x1i!*R$~`HA|E#rcFcF7Cg~H@rJHJ6FpUtmQCtDA@9omv}8DBs)Dis^A+&obH
z=dr+j;aJetFF4;y(GK!?XJ2=wltby{;tYe@2-u
zX*0C{o?ZgG_vlQ8m&f46oA)@6vl-AA^@;6Envdap`$N$yT;E}sEFg+
zh=w2T&~~@7li)EkS0`?5Gy>|i)4NV7t3dn@!%%to7M!GGU`(HQ2O6Xue8!@;K}u|8
zfLoU#zFO<9cvr~)f9S?&FEJp4n;L46IJr@xvadBxOGMG5RW5x565;3Y7dEYAObc@O
z_gd4?x(AB*>;7+DW|k88sDjww^f)h0Z@Ig6Cxsr3qa1?Y$xt11gnb8LQO8sB31{#JYg76sEMWXz#{FKcnruk-Mk#ugWe#x5c#8{mjRWrR
z!jHA<+n{9cns;o+A)IhuTm2GDi<=!>Gz^O;!T%d6LD%FC;k6}I#r}Cp^nqBbvc)fY
zbYP6jwz-xLSM>Iom!cHFvpj<-y!N>9*4c@8_y2hD{uf?r3Nh4p_KJ^IZZR8b`QtVh
z&q=N)nq3QjxG0I@Ne1b^zWxAzyExWmjJrVfKZA=5#1qiUDl8|{ZvjGHWu8%rHOLCx
zl_x02z%v0;ABES0;85uXyUUMpNXN$igiK`{X6*WJa$enm^fD^7oj(sC)mU+c;m8WS
z(_|la>(34lcV|
zR#ELUFDX#%u-A?KKaRk{JEhgY=e^LrEYx=1$$yQ*&6)A~1@*hAQb0yt*Cy3LsS-xKcn
z>5N=caf?A%#GPG&*5SbS>9UW^p!Z2`UrPe$EWnZK@MHPPeh}18`-L^37u=2hE#^5{
z2~Bcu6H!(sLDtXO8^P!{NUDpwRdsFzG>(ih#C!F?>t#>6!van`vy#;VN<`TpMaqBm
zYW^p{u`@HotJe$MzyH%S((iz;&6IV!wGu(s!C3SZzWgp^U06
zG&M+G3Ji3CjqXp`cU%47>KW}to1p;Uviy5-RKpkQW{OWgX%ho%&ztwzu?`
z)Y}nocG(CP>`1|HiaD&JzfST9>r+Z&#^=G8x8zaI)tvD6DDj|ounV+(`bjXqN*1c4
zLSBuxC4k<5Su{T}A5>%CD9R!|4WR2khIU~$Xq;F?9`sulTBUk-^Tvq+&ozn5sXRQ8
zeMR3~6sHARkCZYA(@}UkMv=zCk_ntHFLG4!UqTS$i1Rdew=k2rQ5u7)J*25qCRf<@
z;t4M`c)!m38$#09{HFHKDmG6#bQuGH4MQ*L@@P7!=3yX@>-({GFu)JE~
zn|(ymf%00?bRv>hub??~rV8U)DOlIkC_#!5dczxNF}7sa9bgeQgAGm|_1S3EURLj&*#X4kn=W)h56_J(!v}-C+t{(%o@nQI69fz*O+V;-cv>)
z!IA+woF-`Ptut0wdgp8X8-MJIqsr0l
zyLODF{|D!#tYYk(`DS==(nBPbeVK5x(+^`Vr9G{3#srH`Xi
z{GtD$ju}ed2GYc{R3AB4^d)8G9^Yl%3=TT(9?)iG1
zI#RL6va$#|{~4PjHI}5Gh(v|3lE_YS=}UidmUnk>DzQ`0{;)KhU4!GD9(;4
zyT?oKkS5Qzicpt^Hc{>
zvwzd<^f9IO?8NWf@01(0JO3J&4k^P&@bK>UZ>@$GR*uTcX%XbSh~$}G1!Vj5f|u1)
zg|dK27sCs&r>)N-MLM-+r5TOLit_CJMDVDrT{(R1
zigrNja4Y>>b4F>y8KjXa*3zYf7x{V+Q6wb!9IO0>7fm##BT~Yw#LQDKvGc*zyDL47
zNbUL7O%)mpD}NE+Y$tXBBiIw;n)qsrjOa!ckoGuXLwCQD^phqa(swFvdODXO<|#V|
zIrlUXLYF@lDN|Q4{XD$e}LA33?etGP{MC>QJL>;7S?bVm?S`93LR)5?4ryjC~a?to>
zI%8kR@4UUGnS>yxRW#LpzelK}%z`xvMzLVn`0GCN67qB|M(I`nJ*;_8$=_Q-2iB`7
zx}Tq|L&U`8X`Z$~%uSPZh4@x4lFC5I|9W^Gdq+``5YI*gFB(rzJCmeh>yN6UuHSo!
z#D&%zGB1^5@fVh+9%>IES5DP`epXe6eN8$O<;5C<)Me+(-FuUaMX<#ZrMl-LP8V!z
zvI^g03EEf5kB9q^slcZMqNEXs>C+5&14l5IVIf-V3o>(r%aAFnv
zA7@`4P38B+Eo3ZdqDV!AqNIW9-sdDmC{&1~k|CmG$~=UkC?c6M3njCR_de&Eh0G-h
zjVMY=5ta9u?jOJJTJL(-djI;YefH<n-wnZiA=-x9DwO7JNEUC$Q7@(xBJok9rTJF@v*1zN`){jK
zm=0~%IeyFDpub;y#$-BH5KX!%9260vhNw6tefK;WiD-nDB#wQ^?R1-DG+6c4iqkPfAjAG@_zGYO-B(ra?)z~RLmq828%K~PRzNaaU?;`uqKh_m|5*a+uvtg{oLxT%G=?d7l#cu}v)8kv&
zcybWYa2Ja@jeYN3Cx71HwD3D*d)~jzKkB=XuFf4=HCpw^@tOmt9{v2eWaoTBr
zJF)XISCSsTTVpONeI9;Oow$S8p6c$Duwo(pqRJ^
zvyt$!d3tG48S2HasV-Akhh{FF(TR?(MxO1}c2Jsb7G<9~Ez{n)f}^
zfKDtcP1xd)jwtK&IopXRp(3$^?{|sUqP1dMRfMH(pugpMZhH+pK!j9hiq>xqLN7;h
z4syqZqaJUy?-j&UBZnpzmF)WQ7`Yz{M4Ekq(ee47$ze@Tq@VAF^GxGyq~>U(m;S?Q
zl+R?mprVO^PASJ84Y!CzgG;rytG@|C`8soyYl`AgyDt5$Tcbyipxc-4G%rp;l{j2!
zx0RdG+g!XjN
z@#E7Ph(3p6^z$}zlxx3+>)<&9^obXjr1DNabQ?2q#}a2Pq%KKiV~5-hbhvtP^1Rt5
zG;YDr*@UC|sQvbjzE5M8BJbAgy?4ew58RerzkkEEr%bMIyZ@c>mO!4aaW@`#qK6pT
zs7e?G{$Sn!nrgg4qKJ*5Sc1phzjV>2z{5Q9;b_M1Bg}gsJMVCnie8tE(
zJuR~D{=_;F)N}n16n(oG@pv=%KC06K9g^OBC&la+eP(0ZN8=GG9$85*yb@qja-W%L{K-ArkAtqh@ZtV7l=1R|NFEVR{WDjU0FUPLC*w7AfFe
zkDLzwFfe=OBU7S%>+I2Uy-b-!C6y;ny`<~*xT;ld>S0>^*zh5VM--J8zakV`5JL~V
zFHrgT&~v)gM407*3!O}s>9R{%AL{7)-(}dw>WCme1|Pf-&;Le;ic(2Z#Y4>EEdl8V
z{&vt+4X0EWt}STN^F@0oB<=eKO5_fFn>+vJ-;*KL}uO5S#d
zxmLSiYq41%{f=zDymp~EbIbj-S+lxwX2HqX953&DW|5rs*T}pu`kVHGvSYVhm`60;
ziL7mPVpiBMeAH`P%`9(Tf4G>iFVnM?#zbSRP9}^1i79RvU9$SqIr_1?%3lq_O_=?e^;_Cc>eGF>{@uI0J(w=bd-;4=odrF!
z_j2C@rWP|%=Mlf^9va>BrEct^9#Q7)7p5wFuBd`{OtU
zKPxih*BCf6=-muE6{qj%X~&rwX=$P=g=TcYwaBtnpOlz)okzvdI6gX$=Tvx@paMN0
zfL?dwtvTJg{pH0v4kP+L%Zg&%3T5W7ex$5g-+Fq=r0Fd`-WjtPoy(@KW+sZ$_xaX8Tfh*Yr+g1Pv5oTs%^v+
zY#R3~#q6iy_h}c7*X=2@^`KST{8NpytYbu^ulBa8EMcsE^vBFwu#s^sW6h?Wyd4bY
z`s#|`Ee{yN^6xw+CQ=x#{Qu6VU8-kTpJ_ebv*IozCFkMQM}58KF=uex
zB)4oR>Uf
z+&yr6Z^fHhhQsF-J=oV78SA-555x&JGwx?A^Z!;EVQe{fGSPYae)>?5LKqC#$Bebq
zDj0Otr0>1B|4^ylIVNwFriF8c88bdl<9L(c5TkW>oNC~{DTaRH&FNM8d~}Yi^M$YpiOofu^_gv?|jHs5KVIXPFFR4rw`j)+x@UnI@M{_BR%eC0I5b0pZ-
z$NCFxc2DJEm!*Fg3r?TuxRSY&{wZ2ipzEWE<|Q>@QzRCxY2eanW@OaCU~
znEuuhkteEI^xO&8bpyMY%vz4u$JnKmCbIfzDIpnqHCayLNp3RC86zxI)oyYyF4
zo0fN`XV4QCF9PlnI`pLCK4V2uufH!v>3Sb7i|?%*WX`sq`98uJp__!k
z-XN}(=%%8%qx%F!(96@-V(6w!!?iGE;Ah+
z?(Z%231OPwT(aA-DV9ED5MT^1KV?4YyAk)^wTo#Rx0CysVF2@-nVI&ssWAGY3v~J9
zE78pSs*!^^zw79d+D>0H{l76CXa@_xX@L3}O3$WCY9X8LWGtg)4kDX-*H&EDH$^z4
zom3B=bHu(!!lm%1T@KB7%Mg!WD21$PZ2dAR%Y$AHVTznfKZ2|>)Vk(*NE69O>gqW(
zvk(dWrl|b9cOklROS-GgM@8wZrp=?1L!)dzzZL8(wfm2MTnfj2fz;F7{
zt)t3W=3nV|-r35#m#jjDWjDz@dn$sKtPXEn*gwYH=i54b<76N6_n>EXaP*d4*~yQNxc)td9^Ibb{<6av)slCSaI(6B{q6yWr@B%EnpaV0edY5yG~8v{
zOz!kCr1W7!y}&|kWYEDVL2sD}GUD%-u!~`hc)VxGD#Z#QCCXV{{t{Z~ft`GQ4#&5n
zOsgk-KPxoRk6WY+5iT7xAYYWg@%*OP{{06L&c@IM{_FOm^PFAz*f$o?SWB^Q
z_VT7^>~sSM_pLi~|3t6$D6cD>d7`PV#XFJ*9nmYzZL3~!WBuSS;NQ%fS7&AvFWkX
zL+~|bf;yYI$lkl?gN}_eP9D{?K?RL!XQ%HNpo*4*cQr0pAb(F3
z=fN0x8fzD&E^8_C=AI^WI464CtIyqEWiKGJv&Z1q(;gI3=%`W!Rsg!V!h*O29k
zj%dhb<^8cj2fqEem!YMPE>z8iiUMblJDIDOA5(Bd)*a_>9k+5q=!;MK21VDQ_f^Jk
z^J-L}Ggr@j7H$kdlP+I4`e%&~di0MH5
zRKE4NZM*6)^7xWd?c{;Oh?Rx0jWqVV04wAeqZcABB2_t)=R@A?LtlS!Sy6p!GrEJe
zSfBsLCM3_^DkNg(6|^U3i%V~x9Xh_6CB3Ue39Sg+!R;;}fOz^z-_@1fidstyt+i`)
zM8sD1ot+VNMEdUVT24zDIiS5AIX?4y
zV^pXo(v#B%K5SM+t+t1b?Gn>R|87Ytmwd7xN!?Z{Hd=->%AmLH<=6U5aG|aB4ZJ%RFGM(g#C9X0HK
z4XvVyY=@lg9Cs{_65F(fcJ`%=nuZ!`eh!=ngA~
z_I!+~WLAyGZ!r>kO<(hK!y<*9EasxdYZ3dr9hh5IAcahYyYw)z&J>y9B&Mp6t5^PGQcK*AGo4*{FNT)r(hZ6d;=a8Jq<1a9
z)^jM~INf1K#D%yHb*7TtDVI0ms&wUDqvAKVZeThbcvax4afKPd7kndkpFUIdiu|R+
zDZA-1pDuL{=_xR=Z*_f;c{IB}jXl8!dCyf&dbJm|^Fbya182bT=HJf-N;
zpIXKgxW(u(4e>|T-Heu_51C@W
zJ~A>2m#+pN&A4hs0RuABIy!`)bRF$UwoSyfo@&Mc
zt=qSll8+h9^F^NTH@st%zw4?_XT~$Go!%YtR3Bx$af;#}C@N#Dta_P|>R8I~n0u;j
zoRpRD<$6o>dQa*|s4KIOZD?i0H`(FYhYTmu$RaVf~A9*444Tbli)yu^*xln85
zL$>pFTOgXKzEt}y9c*Z~A9V@TgBQ5Nl&>wWgX`Q9{gs95VRywu(7pRl;N+IuM#b9d
zz+9fYrB2=tEc~N-$!xq6`#xKa<{tT{aK-bZi@aJZU@=lDeql!zT$tM=cjS#5sM_`<
zkyqv|1RcT>Yv>(tD!?}TgKQSOb^NRlYwsiI*{E+WdgwaXpHw337SaRV-Y@7J)#`?t
zowTt9`!PRfi&h6aKZL28RRt13R-ouxHCkoD2F1i8wTOYoyKM?}mLXCK&
zLYcr&D15DaOBpa2_*b{HItw_A7cvqO9Ki(vCnw{fGf*&3;m3?K11!$JHoHQk80Z`N
ztDOHE3(QZMyFa`u2Ni2-oL*;n!{t6}a%B3dfaozT$$^^?%ar0yJHi*EO1w3L_O081!c84kIasA
zg7>Rc1(l*YfRCJpLduZ_z*tk6wdf}cJoA?nx0hsr4wp~`;evKhTmJBe_e2MkSl>vyXv3QeWbW??^pQrv
z$68z4)LST6@=5)txDFFsW%f)pD^~(Fd!>QDtIEO2ph~#$0t8aD#R^kXKlNGX}ko<%`qf|;aH
zD+)>^H>WN8T?wR>1h)B$L-6Ktl)BqC6!a|-GAZqVAU$W@=EsX!ARwQ$E8*#5p#NMg
z;?B|nka@_rW4*ZKJfj$!#1zxQAd)uf4
z0?S=WAz8~I`1>w&j;osqJfHDiIMBubd&|Bl(=ISU>afL22`&^A&!NJL_p$)%@R&%l
zKMUwy4USPiSq;9NA71{Rry9uJYI|#D%mP6d0w3=3sRWL%Qa(K3Vt|mi8dLXV6r7Qn
zczO9g1cDpRcKs89;B$(u#plxyXw@wGT8sIkrknY44zn*@rm?W?6a)jHc+SM6v^MQQV_&l18
z0;cKb6_q_GkW8Jw*-+2KcoCl?Q!2sN>BIha(#rsLOTJnf^Yb!9eYa(zz{b2)t{I!R
zTUad<0%=v?MVi}{=V29KXHrud;s+H4OQw0jO+#BD;R{*tLom2alR{*0&WB<&y6@Xnw
z5Uue~1z>IEqBlibazXzu%a**VRPed%7wCU_7u;W;r2DnE3_J`OJCW5{3eJ~56#TAO
z1ZI{fsUIkQ0B$Nfs6I4I0NW1n>rD7mfbNE%)>qD@V3E-B!=>c~pz0Cw=g8hnuq@d9
zQxaz~NL;Wm$NNMvD3Iq(j!!QFhG&1#4ha+hxqbzkG}~;De#^^yWpg@6|9&Rq!DEruUUrjx!;#ewe$T7h!MfjmysIf+!_waN+)J1NzioYs@?x(u0x$4v>G$Ve{MM?2cGAUP74Fx=zRyK|YX(Z-orcb-p{WY!{Yq~3
zV{Z*yK6PI^d%OXDX606G;%$O=Eyh;({49eXo&Wv)S;T}s#sSA__S8Z3sXyV(i<)2<
ztS|g`?J2xkWNiJ&rv^HhJgA%LtbwnyE6hia)WTiLN{ltFHIRdI_ZGRXYIt3wQJ>ql
z7JfPGY~+_)3tw#hsW-c+4whJPt>yV%3l|uT@3U5^g{rRc>%QNvhly=Vm)R*bKvUjJ
zN*~-B;PFytUT&ENs1_T$^LS-FlwToJvJdlP^mDZ6vBghef9Fx()Wl}ExaLHrZcj55
zY0kPc7T63&|ERRsyFY~?83A;c_-2?NzBfsDTPtJ^4vht_c?PeXJE{J|As2gd4+=AB(c*C*h@X_DM_j@CJd^=C)v?mLeqQ|05`$5T=wi;e#wLkSZ4&jEb2hQzUmGHIPe8|cb
zB``C3lW!Xr_LG_8dhtzBwJ>vav0~kqYRJQ&%{+e1fD1LmJDg}`@bV6uk{;}T-DZLM
zS5dPDC~9jl(l}NJPu}Es$un6EHBHaRh($7>zK&?aqv&$@YVe=N$aEw0Di&S1#=VY;sgPzzOvUAgP;M4Zs#xEHmL+8aOHTNM)e|?RpB{<($#>h6i}Dd({ccG!&lQvwdfLX$4F)&o|j=g8i<8(w6`Z
zgD6N(JB8EW$MK@?C}Q+?wqM>2>O{e0nDXBMzti2+rJN%*7Ibin7(RBA0*2<|s-Q%*fv3eX!@-^lYYOt3C$yLkEp?
ztDXQ(;eXxYt1H2{bf;L(K}#@x_D|~G)$!nk<%4g{D~iCUC!L=@c~%1Nh{}`!H3qn8
zdwP6XNiMkFT9y3Tz7z~xGO$IleDCM2SMU0#S)jEzl$8=z119^ny?pke47lvca`=+W
z0OGH|pDf9NAe8gT7d<3ll_S2Ft-7ZuCUKkp-L6HG=g`f&R}!vXx+A1fTUKMHIw4I1TK5Sq&t3ENvMswLqerrFt*04s3pO
z>%_)X2GBV8IHlReN)_8ZdvZsCed62O5Ijq}(*G2j0Oh{-LL8fXSL&M$0zU
z0D9l21%*#)K-XMnBj;c(Si`S$Q%I^72yzedK1;3vIy^7@63lDBUyDjdVV+tLdL(uA
zTb){kYBr
z$Etx_$M+4lovOh-janDC*=k^2X(1^USpyb+z2iC|QUiLhUsTKtn83(7qvov4p1v%<+5?%DvSHR6;TyFOO&h
z`+T(o*VZ(E$$|Fno4Ix1#2sC!zWy5EzDPrAr%@AlWjB#Go!JOpRy%AEz1;|eoQ)Q`
zG}nV|K~Tvfr3MWB$ZS_ms|IJ5MK(3$qd>^Ara-b~a
zuV+UB2j%Qay2LH8R8MqQLTv<)Rk*Oh+SeUyUmvt){kv`86#s|^@1q=WOfS=XGnVH|
znNJ@1^(PK+|2-kY)8PvRgnspv21i2Vc3F!}>=W>W=lHEtaXFyXWn+hW@ExEQlk8+^
z7YvIBqiY?0Cc-4E^`d7_=Yg}{tL&AsvO%1JxO{_l6qui1n}5IR4xCH-Ws<)#4T3@6
z5>LMpFj9OE-Dr~s)V3uojyezrZtRe}bEP{PZdK%T?{Un97drEfK8d^l?f?15A592>
zN+(w_Px#$}d+&SG5c$k7ckH
z`D~+Qo(7qv;me)RX2PXs&TdQze+2)0D0?1uG#{RtymsK4R{@;Ac>`{z<-!85io~kZ
zMNmef^-m#M4!5oPA{~}%-6>x%g?dkp?6Y8RJjY_w;S&R-;Nc*C7E1`{1R1g-)qa@mk&`mcx;=}Gkyp^
zwRxnET9mUpyN)&49mKCK7c16hJ$l)j?t7
zMcD6-zL@EGTnbe*I&Vc^tb!ZX8-e1tOlVMJ$zwcH0d?B{@E2K?!Z#0{#FUSfLJsqY
zg9(c(;lD?If8;GIVZMj^E9n6UOV%$5IxvYsk$*XokEU7Bm-WgsuA2cHO1rKsep3u<
zR`lPHS%<p>K@93
ze7iq!Z{5X&*EhWLY_4L$Ii9;tODi!uHnXKV2220``vF%@yV2V_d;lxY{S^?iQNx{y){WV%7V#%XX5z%s-YCCGWF13
z7WAJISsBm=;l-WpnfEN2utZyEuZT$%Og`rP^$Vzko;>JgeaXn~3ulCs?rYnsn4mP7-hnyLWw}`8;%-
zcfFg9mO_R2KkKglgU~&4b(&5m3L`{Y8o;qa*jXATJCR=kBQJQ+(hf7=bV*~)lqG}?
zZ5+kjA}r{$POaU0Eel2k6d8`bWx<@79^2*-?02Vh{Q8W}Kq$1a->~)+6W$gT;pd59
z!f!)<2uq&{FN%c}Wd<=}x40eHe-wo(uGhVVm=J#H)A36!VnNAA*W9rO469at8~t>F
z0ju~5s`}edIF>Fpohyez(_6a6%Wh%6XVrCm^FezSjQ9209wk!+bI^0{z12sIuS!ZjsG0A6RfSV-5FD9SG$U%MVSm7_jJ;lUWfD3J*_b
zoE6DuLc?2MDkb7r(CXCpqMCK}ux#D0Y28cpaJSyR%mFWruB-?H{|GU61WAyIr=-Ms|B_JM8|j%h~?3>p!I0
zW&6vn_g^`Y;<5c;zxF1jad$`n;ti|3NV+>pUZQi;|uZyoZXU94<`iom5CU-|z*Yzvnbe;QC8134GE+
z>d)*Zx+O3Jpu3@|cFH8R|!As`o$3p5bP>~}2
z=PsE@!sS(`XNh{n{w%_uE85OPy{oQUQ*pgQ*bJc``SBMEE!g-8(g{21z6Mw*90q7*
z+z^h*>Gv9O6gWHf3`hMkiC7$`(gP-NY`NergJbgc_53*c3QS?4X&899^t{J$x@A)w
zj?lFv%nrj%rF&%@Fbp2ppNyCDzEjM?vFyfIECddNl20V6Y;7R?UebJ;@Y6!&3k%ol
zE$}7$EYvc>LentV;`)rxPYXvA0Z5r$)4afFYfQGukBu3=J_=vf#cd1xn7*VZp?(>
z>m4a2p6~CQZ3J#Y9}=kisGGpDJx7W8P`TnpqSvP9cHhI_u||BkCVpPp-$cX5{*;vmn>2*9^IC1XUgrJdOf#Jsn9S
zeMI>=uSmilJ4JGRN$QgEMEW|(4-O{nT`w6Y=5ZgV7Ew=4PmrjGaFgSh$xHIH
zI4TJL^`39QwiayOZA&Kjzh3_%(D7~&fnNtnyDd!8AFrb%E}`P-LUU5qw0}kLS7wlM
z6%}{4QaIyAmgn`6C_P2;Bi%{4n2M|ER9rry_@1yg98R|XVgbopPQ}&l`lLKNnnv*4
zP_kd%)+D+fC-n=dI6Z7i$~UO}XokWw6v|P4uAug#j~7UO5LM3RnN$3|j-;Sob6vC
zRiEu2JFmeeE#WVlhb_YYkuiagvB~^D$nF=m$cb`xe#Op@*nBKl6MA0?Gbv=(W4Fil
zhn+VB|Hm%2Xb3y$6te5F%h}__Ms~m0h*c))v;F%2MRvd0cG>-4m$Q*Q&g^kzm$Q*w
z@4x=C`@_zcb11*r?Xms-?|fq0{eSh7&0~)@o6ok-_UFII?gzUZ3%0~~vY#h=U9frV
z=ks4ayM4C*?0JK2D*y9**gW?9Ve{Djvd4*yZ2#HF9=CSNE_>dv?P19S(cTz^w$%DS
zDVhEJ{_AH96*t)S*n0N7$C3)dt_c;#u+52(+0UE({MhXtrSj%JipQS!>~nw!RSr**
zcEYKA+>0vbq5QEtOX?TOkSIdg?V{q?N-8dlS&{m|5E3=dlX#Hg|D^PBfn<5J9*Ik+
zePlDGU+hVie>_CumoO3osC{J4mz392`$cX)DWAAR#@knw)s-%8*H(Bl^N8*$kxsPvcY$kY1R+IbMm477v
z%QcdhluEXH{Q(E|h=##I>U=U_N7moaKT^bR4Pab8aFTO*^4jvhge=$5h8?AcL0_v5t>8Z^$j_?6Ak=m+dCP4Rr||mOy8aw^dGL+}
zS{NhBO?*uE}uWdYPb9Jv`sG
z?ojGDkKxz?P8(Ycm)Os2!Ts@Zie2V};StS+!1IevYxFxS!oei{fxO=xSFBUhl>g)k`?8
z{v%KDq^z@YgD_cY^)e1u4Ao{2f5!d1rqRQR*Nb*w+`##r-TlsZzufoVy?zQqW#OBx
zICdYmT74G7zgrIsC_7_s{a^Y|h
zD}3By-uedO{j$vB%X
zPsYRfouup_wwmzA*}sS23mhlo>A$5U?(HJu?2#n$T-iDQnc!V}LfTp4M#ke2C6a%e
zqn~{4dYJ?-)Qr4mD4rqDQ6tu*ET`6qZN0T{-z&E`nl{q$y=>O;tD?!x!C)k8_Dy1K;jApsSg4qey8^3A39%%
z_Kn8L{VHw`$uG1d<-}f6{l)cr?q@SCQki2MW+}2b3dGx84O=`V{nvnX%)O~v-b&r3XTHmWrka`R1Ug@z58Sh(fk>!8GNEDJJ
zvE7}FQ$?TN5#!-kOU8{29AvyRC?n-MO_CR~wUgj=aFP18xuib%E-7bQ^b&gRNb-KX
zYJ}9gxRCnL3UdGW%}2JU(n|VKU`&?lQ1Q0=8tFGfu7$AI>flJ=t+ix)-oC1VkbQan
z5?D2Jiol$M_lbS8ZZ?&W)dgn=gbif`&N+~NrHI%Na_ISm*dxdCkxsH-OP-ScEEgc}
z+tVC(Vj&C5FR~c~f9r~F0{0Hh5%_%D7_sjO)^`(m!#ag<28Dgx$u&u7qs5HG#;N
zpG^L3#qF(c+<{GL?Ebkjj?fQjE+F`!7b@%V^6NPcg#1iU1e=<};B3ke;m3Is4q~5g
zIiE%3MW@~7opC)+XaSMGoZp^}EdeaA32!6x{}%Pf;&MIX3Xxx`Ux~q%#4vFF*-zv#
zde*D3kTeXclO4a{=({KAERKw2nb=Yu27aPLF*u$tUxOWOv3xpX=_(wf3=Ov9==G^t
z1J6ei{bglwIdk`kFkFrriCc!}m2R#-p1Wa4J1Tw)*Ngie!LA;LLC%oKGhF{vDP5}Jg`yO@0E-(h|{&2_5wz{*qS65=JXa$a*k
z*TR=rKOWuq(}T$!J{I@-FkZJymYCt-FtEzpez~LzJI6>v{gj?z>~qzd^9?ZGi_5%2
zM=^90tG{Z9q4e-RaeRClzb%b(#&E-dLDQ8Oy5IhG4<8qrSn1gS3}2kJF2KjD`l{|$
zoUgp^d$yauu#%Vyxo@l>+j&WVSlzVeqOsCCnSo`hqA9n-r@6R
zd;H61IG?XHS_bE-uf0@)pN|5HyBu-5wKjS`al7i}+?hBZr24(V=VAPEO=&R&zxC!q1D%R^99H
zew*3pzQ_HV$$O@bw=;8M$9vo_t}VjEdbvCyOT>@p_M}wYPkp|WUpQVX?jy<%?Xo1|
zUT^4nLN5LBfR6il0NM2zw*okQkB_GL4&wIyw&%O!
ze9s*!k$C&R!&Q55T)#NH1fSP`c8%RI!4OdnPQ=UoK1>nkERWr^Ib62=ZArw9o5ouS
zf6l~a5OGq}gglQFZJs6M4Kd|JT(dJQC*-Fe$@`ftkEDsXC&T4Wloy`gPsB&n8ghS7
zy0?>MNbF*{lbTrxtQi`H9_e0cgZ(f+Xt5rTg#t&Bwv1xwEsDrtfy#4zDKifBlXQ%B)?0J^e6oX
zDI4G0Nc3lg!93wN-I=tnK0^8ACHZ_8OjKXfpziO&K6V!PihB|lc
zqw+Uw6C&z?MiRGD?~NW)=Pnv$|2f596iWKJ`v}Po4Kjc{lx*QJ?v77sE=&|1Of`(D19Y$Ue-H8
z>bWVq_b5M3Qu1GFJ|s}{VFP7{{XS2J;{T-9(OfENub<-8QhImFo+o9`{Rg>D>M4KH
zDPBFr3#9UUk@Ms_v8K);XTnMQ?bLmXm>;>WbaKgjc0n+yXTOi_q~1GpY$JIuBuQRy
zEy?R%L1M@s5;Lj$^0-nm-*SY6-O*{
z$arRS*`DAn;U;-eO!7Uy=}|Hcigbk&diSa|1Xe60^ZPlcXN2s!cbveigXBF_<)sE<
zU#X(z>92V5{^fKdBzQ4`M~Qf1@99inltwLCewd86rXl2e0wq;ZqI_4&W-{J#9wYb?
zSIBcomC^%3|2c=7XfOA7P&dAwBLvk5p518`qMW&P0kPkdkN+d`_PhD1*iyjudv|kU
zpWU&6o6tMAxfA@>oe`ITHrVf
z{yoO=Jzp4+*Z1~jVmIz^hkM!q$Df5Ch&+C;)Lw2p54qb+zlGyMM+5At10x6t;h4oiM2;x;Q75>Y(o>ygBoW|aDC&2kFvPmZ)_Kz#M^VQ
z&lxSmaN>}93x2;_hRb7%ueD(K`R@lFT;^N*=>%>k)&dd1?Mv0~
z{au8Y-}4ULgJF{Q!wXheepVj?4c#%>?&%hO9Cd=c`tZD1++FAK6sF%4;%9XYlX>$u
zI2^}NFfUXc&-cDBxx4Hnh8rGe8~9`S>F=isiuW+Iv*fh=iIs<6_Rqre?Q_LdJBqOK
z+;55Mu~@!d)mNwd8q3Rt8am$McFiXvJSs80{QBnXPnh0r8_mwKV;ER2P}5oU0n6k1
zpS!eTJcXYoQ~4OijGH8$9>AXGxpSfB-B=#q;#+&|`7ofzmAuWs{51v@d2X1#$>cfh
z9aetB`)7#H5SI6H4}EeS9R?=@9(Z5E+DmYYe!ahU7?{x7+L-ULJWho>PVo(vpMLzE
ze!pcH%s0n#e#iV-v~G5*3Z`#@vQAc5y;9lvbA|=j^JznZl`(&0L+lpgS2V1LlvU@#80h?;OVFxmf$=XiP7qaqi`PtY3dv*Cs2l_A5%x4?n;VaC(3f*f=>`
z>8X6c&JF4JcYcb(csr%vavs6_z9Te~R*Tu|Q*mE$5Yrd0i^xpF+8Mb&yh$Fj+hZ){
zvF0syu6eX7Rs<`r{(I`@)`KHpFP}{9A!WN4dojGW
zAw2L3hH7`s`0)97C{sCR3VROHcY@~ddDSz!4{5?sdriMBZa;tjj`1ZJ7SkuSl`(s|
z-FL4CVg1v$X4M?T%H_D07C2(#VWu^HuN#vOmF*TS#r&?g^W14I_S~a6w;2Vz7zQgB
z+ze%5ycmz(lsW8tn7p)p3x1xvvu<-9e$JEaezFu_pRzUbzwmWl`$It$@0Z27IOQk|
z$8;7v@x~C88OP)6I_JJQ5qC~>Tm8oC=_E3nc4Igqlkw~v&QlDKb;6L#<|q+wdLlgk
z;r4dNPhG;#Tk*0luARWpYgy})(-`vpZ9a*&tDY6P6z6I8Uy$g>`dejtivKF+w@Jg%
zBz%1QtBn=$dL4hQ=kWF2oBv4Uh!d?Z{vHI4IxItQ?n0(<_e&BF2V{w^D{3+LS$
z9(Td@4aV7gct7^W{UP?BcK)Bl{?Q$Di11q_n|x1wtJ9Ux$NLx)pBt=lizH+T7ij|f
z%Yz7h%!9Lp{6e!Z9=C5|o=faQV|P*s{jg6aF%EO_mxz5xa@3bV&$H|A;{5AE3y6KH
z>g+MXj`n;WF`nUPtsQYczx%Dju57XUqI2FUICA`95#tFx8nM)D7)&l9@8LT9CW!iO
zI%FQ9%1fet7@3#EjfxTas$udRcru+l7ybB2-nYFPAz9-H
zMBsihZ}3t1N}Ml`hKv*ae8KyIu+KbOL*Q~ZDWZJiLx18uU8?jNVw^X%84>m+PJJfE
z#lKpH;MGcy`HRJ8(r+4duWNjj%%9Bmk@#{Y>DQzKnKyX}kmb>jNZesaqF4a=976$V
z&tV72pB^OTW3NeGx9(D+{pU-Cr;FC;QpGmGsxImE?=6lGsJrZKnLSq!9bi
z0MQRms=aBd+=s$m3b9E_@N_7A8A!_OsQT=47Mr($YEOO($-6?0!zU^)=cV2=3W}5Z
zqGP1r?dGJ6Qg&r2I}@^GxnKY}pS91DvOaaLbfn%F<{TlPuUHj%pPok5zet^Hw^ICD
z3S|8ak>on*ru;NXB+I)LN#uA)qWGmrqW`R7Qa(?e$Cvyh+wogPuG1Ccq&`QST%TOe
z$^2f(jr9A%HxhdcD~WXyTuAoYc#3?^SpsCe_zLnK#ioJ0&n|r1M6~aa_ldxaGvqz~
z&Nt+}4*&V%#QJ-jN1g)|my_$eVRn|_yUKhc*<*>iSt0l0&(Kr
z#(7yEv5!Pu{?v-^7ZI(~#QXm`cN!6&vz9ggAFi%E9;@g3R|-XhN+Og>MI@C9T`5$G
zLMY0XR4NssvV@9AvV=s63Mp%%2!$kDWzD`ncG9S(ha1bQ6De+%OXU10C?G=vBh#ukwE<#A4d3+Ypwv;J4p>C+8y6cFTN3-ac$l
zDT(}~d)o&1Bn9+TQgK0f4V}QZX#a(z?^h#l=c_d~#W=3xGEs4izYm*jogn^r&ShWJ
z;}*+2Gvq;5<}AF(%YC_CRAGGBjy1c>8Sr1(@rTHdTLn9=Vtka(nA=(({Jq-KloI3@
z4ee(l&|ii=)O=3(NuAG@50DPB9G*7|kVUM(#T+m}wo(;w@FF)k=Q47=x^o+(9oYHZ
z4S)~!EJ?nL{?};1sSRF$3p=}uF#gI-*>Xk$de&sI{{fyP>{;*o6Z7XEn;!=_-SV2x
z3ZDOw@{t+-Iy;u{MnpC7E>upge+pYOQD82H%1(v#8!z$<_BZ_1npd{XgRB@3=+0?SVo
z`~bPS6*~)Sx`99J)f7TL_xc=9X*2Xs;J3{yd_aD^q>0{D;MW`N8NS)T2j$gPPhT>I
zeIQBQg2Uh7{LO-$LM^4x51g9=^aeU8gJ^{#iAE6j-I6d9IS=~zO!LECFFU9?>|dg~
zGGH7C53To^f^$@#hE}$kfS%J`13N|-RLPS2^?abmPW8YF91r3Zmx8xJdx1(fN4enr
zeN2?bYJgAbXgAJU0D6tKU8|Ap@1Xkbgd|yjzO&Eh$f?8oHmc&}=~}I?
zpg$+qLxFh=D%_-b?Yl3)qi;U2{`$3rN*~yz9+c2ZiI?C0ecuY&d)T(*=4uAjD!xj_
z5}w=oJt#33^eHQ}x$Xe%9RI2qjpMxdc>>vA%CbzHgFG}WljGBBz>ybS8U}!MJPRJ-
zxRJMV=5_-V{Qfxr*Pon|oo>xPfafkKSo8?ed=qw$$^q5({hM0>{bC@|q?89~o+ceR
z_Ed;J8yYw5g13J>R1
zqnw10{Rr~rduzUm;yiGEw)8H}=bY@a4T!h6vknSC{ldBv%X}c+8aFY3`%~Q+x0Yc4
zd<{2E!TKXkxh1~)TqrntQlE
zX2mD*at7r(uI!qEei=XFmx=4JQn#cgtamv2AK}ZY2}UH|ASPM*6YYo158CcuzESxL
z^<2;}m#jxBL_0{l#bWo?IFwt%AN2$A{Xdr`w3jz^j#yzo7>tH(#(pr9Izf1ChSN5(
z4&jL2C-@BV-`;RPjpf^v*FZ=W)@7FGexW|n54+{Dy%W3=hNXbN9OyWq^Ge*RJ$@ME2p1`LNf|QlgLiJlg;J->1LZF)E|^h^H6LKj)tQLHglI3$6ESCJ*XSpue-T
z{Fh1U_Zni$$My>c-x9x^EZ9KmXDOihgR}4(BDdIuj<59;@RRk)SQY(!Rdg}!|MML#
z6TQbD(|QTk(7dEEmF6{n4%5i5PNQ%Ut!EaK7iF)d^YxkMC>HIc)4dOA-d-9_$7i$s
z==}B-G}<$He84VCqUR}2kCjXRRe*am!g63l_%yV0zOx|A4+~2sxj645f?pKKHqV@gnyPDffUS1|h
z_m?#n-5)(JbXvHM)>m^S-H(fm>GIE7qNJadCTTfWL0S)ICwjf5EILfTk0E{DaG;px
zg{v;n@72DOe%}~HdcAU#pM}VmWYTgmU+8-6p3$fsM#qCyS!9Tucpl9ob7JWB_Jz{F
z%indJM6ZWOM(Mcf6FTmj!R|?()6AvMb(#efkojj&xbzj72b`}-`Ph^VL3Zc&q#oHF
z{e*9Bt$$9|mCk#)$hhPwp}(u^EKMZ!h<|k=afbSD0=-zDSpGATulASz-qArZiaZzC
zu1V_6usn5t@&m+3j?M^_)03-n!L--fnjXaS^PgTuJZN`~
z)boA!S8|@>!jPK@?qlD*@OK2ueV6sQA?oEFokWc8(-K7Vjxn}IIc}pQ^8C#;PI{Qn
zE?h#sSG-$ZM9vj&w-mgB@xdG(Gs16kzMt5N_GqX6fdJ&AT?1a~$nTbaiBb{*j7`5e
zgZ{O(dpZZ@{QWl8SOb#amXhvPn_n{0{o+NYxNtz
zFH*ua^5E~q71L7iJVrRDVdix}hxOZ?@ww;OLt@DPZcqC59s@b2pRX1hL3+f?Qs5dO
zUqfFS#*KH%e>7eN*vk22(G2)Kj)5Z?*v{RYyviEDW6O6GyaK<*2LE(s3DmPE;q}k4
zcIaoeDZVTycjWB2{Tpz3fKQSb$eC@qv%?7dzwhPP#o-|5&080S@ytoBUz+aVUwdau
zI&K3#Ri>8NQVjCR+jc5Vy@Gl@xlD?o{x6b#9+Njlcxc?SLgTLp}s@Ma(XzuejHlZsR)b7pkCx1vneO52enr!H#F7j93`VM?3
zt=rqHZ6$;9vmW}gmX$$yT=&yEHPBA&7}ZH1d)q@5nX?V+h%+eRO{;g9%P^>yf29sI
zWVcg4^*@EbUf4yMMcLOx92lh5e&paw)`0k{B1fldRy#GK+w^%!34er~0tPae9~d;$FYPNjcdpvU`W7VW3d{|O!C|DJDap){Y+%KhsDJuvP5{)NQ6}uO7A13~i-i3ns4p_|Qf*eXYH)
zHmQMXa7&FZnP{L=xgKWu`au28i{^7#d>VxMPl|S
zL#5-wx3InnDEJk^&7d;*xWlhy!g_3uU*39H|8OKbzqvHrLVY&mjW3(9^`t&J68hkfQe=HNXZt*%ioZMbCLb
z=nm|L&84lGG~Rj`C|Z9z(ksGsn!x6L92iIvl-)IDXde--y*fX{vKI6?fMK;|cYj5|!(wO&V5x0B39xeUP)CzO*|z5g#hU*r?qg8tyG
zcn1&mi*tfIiC1TdzJseVU|);hnG5B-c0M6~+AJ`N#A#-}D--|koI&#!>1)@Bzn&Mp
zl<@6K`OAo$T=X#U=Vvt4Nck1x%>;`Cc91xY<-TqbpBHniBYOY+oxwi^KW38BK3N2
zr}<%qDP8VpO4sM_M%O17{+Gz{<4FIWQ1M>RtPMh{*8-(CZ!eKTbq0Xk}*^$~6ws_gEOO4I}kb
zX)6)=>#h!@J(bx%f1{jc&rveY);>ri?J?V92pAFf6YqX3Y<1P87Ac-Ru
z9y18U=f|frJ#Zeabn-U9w7r*G1LBt$dc6E<`9s!~xktVc-Z?+wD6vbLP9rlYSMSnA
z_Fo=Wm1kr6+PX_Ou)af$db4o8+N-y5V%p;H+gn(!o&SvRRq;u=S(rbNFLMR?pLV?i
zsjsx{g$c?XJ`=nZ%fem>-^hxESrbqZ3E)0l<_S
z&vs(F-tpyv7)Sh@6-VMJJR%_pXlK>^rX}_PJ{x;5RttFdPNgpHZ`Lm9K8f{w{<-`h
z@>F}1?XM043f$-zxd8U_Vx|5M^f&%K0XvXSCM@zTM;;^g3
zMH^z9q5Mzyc@lS(Z4Cc<4eH~R+tn|e2XPelGw-WFF2~~2E(?&Gs}%oM71E>YyBeQ>
zzstBB8<7p?0aVpr&J;oZ&#M-~#!&y6hT2t&p`6M;&WHtl6{P~=hXY`L@nHUtHPoXv
zyJN9`KAamkcSbf0&J8Sx`TVsMc$Vhp_Y**?be%V&kbayw%8Jdg@e
zSr2#lmG6af!3S&}K8|juqGoORJ3AfX(=~n{3(`~2jOM&sb$*rQ8j`(T_!UP-Wh(?@k&J=e2_zm*E_?=gBD)lRWl+}L_{74Wzt
zUwWmBI;i`L)-2EkzFfX;CM4@Y3$?f6lY`-kc1lgWJ$Y1!K@B9|DvkhsFE&NjcYf%g
zHU)VlsQNZjb5<)>vCV3OTXuE!<-qSgC~#^i2M2=+4ttq>75LVJlY>$oubZf7nKL~u
z=`GYe`@IgV`{DNqy0)ZZ9q=co3@R;B|GN{!b$Q(0T8V#z?-8HzaAbo%kH5`V-4Fa_dbvX48{jXpTK5Lu!S8o%
z@%z+T;P=ZU*m)~}4|*>Cv9TS-Q_Ic3`AQ&HxK%JsAAZ-V;Az$`5XW3;YQs4P;?TO;
z9lMVM?>sTdx>y&+MeB&V#|4m=`}tP)bq?@?VQTk$7-zozt()$a0-S|O%Y4n$oe(1N>x!*eQynlRp
zgM$;`p3z4=XqSGcUG+ozmMyn{eE0g!bZRr&vB`6rwmt+@OV)@)`)2smYaQmt=dN->
zyJfg3`vcmg#tPrRae%U3*;820xZq-CgjUqDQw9CukH;Z
zen|R``E%5}Pa@YC^QYaW+Y!4WgM$z&?ms28XaW>5VC+V+;VnK7^2ON^hiV_wsKOhk5L@)s~6*?h+ox=J5JJD
zM(OybmHYl8l>cbA^F3nfhquIU2W@;pb;k?gh6#PIfd6wnBImP&
zJ}-Y!ljeKc8)+VtwTR|3q9JtLw(9R4q6hyII^WWo=0OWfX}&dUgpTuGot#Vb8~@!&
z(6NVpKKL+Qzepz?ulvQqMfBwp7$tI*oPh*&meA)e53&f5_VjwTk@di}p;S_C>E=$*
zJ=urouc+2Sc#~H^3E?p|ZTh5Mfyx5HU&>5Zlk~#jPoy6X7h90?i-EsVN&d`;Fp(2n
zuTR?Z3z;J6=ElT+?631eyGgyG{Ep;zOTQOM(iUocq7uzM;q77y{lVBy`0fU4l>6Es<&NojQJh?uu3G;|
z2h)<-(FZYocd%zY&YPze9&9D?5pu_EUwCuSUPn?btm
zQQ~~ef0s9N9OY`CkBeiPyW$%eN8d(cZlc}c8@U{cKbG=i?i=J+n+JXrBOluI&R+!e*)|;{iT))%r%B^5$gvw&
ziK5-H{i{DZaqpPY}j{|0$-tK~l&ivO
z!Qbf}xD{#V3wCz?*<<<8PWkj>S$ELq;NQ2LH$Z+edBjWs+G8XgP=3L~pmc=x{fh&+
z^$9+^1i+8vx4rR-2LGo1rBpOSZW+8)Wipw`rx{v2+73O8I_EQes
z$DZ1Ne^cL36LWPxgUUN}Qtg}{#4q|@oICHupicLk@Y~+@o0^g9Y?@=jpoUi~vJ3XY
z{G^z74t(4;}isjWchz6gwyX!l7=h=Zj|+_G8ysf!vtao#qG
zeUM`5=os0%=Qouewso%PzjjKl+^VTpu9Mo-VfQyDtDkZ?VXY!+^qXo>S9HC#W{~n~
zc)mU4L=!c*(=p#3_=}>1_Qg515Z~JvlRP5Npw3lk__fYrP_x-?ub#cKiE58DykEYz
znc_IT^G;^tCKR8rc7z`a>5RE*TK
zJGb)MsTGm;h4#SjaAyCzoVwx$N=c!hV}QGndi5aj&v;c6^+rKTu76Q0rT$*ekyW{k
z>ODQz)ShnWCOmEpU+tv^9jBWJEY)#7~;Afzs*#$pq-q<0bO!19@Rb=Z|R2gjcmK;
z_hEi%^6Nased-6H@1IG-_^Ov=A6f@;CO1+yi$b};s<_z)0q>agzq}9eX1O1$E@N7~
z?#4rmck49>gyDRanHkxHcH~s&J0o?-*YCOTq#x$R6{DOav=8Tg>6+bwd>_9ZkIqB=
zE|VgL%RfMTP;`ME#<3V*YORG}{)+v`k!go{x94$*PysyGzhU
zIB-n&2(Amqrw$0<^E2&Feb?19Dz6H6=8dudNfE8+2h~1mqRK*>(}~dfA_tQC+l4`_>V7W
z*EKeaK1aNE>ImUaYCRFFQICkTPMa{l^^AG}V$|e0;?ESGDG;6`@l$FH{ek*@<$YM*
zl{0q;ahsn}AfkFe6p8N%J`MhZ{XZTTWrMtP=;rwltY6=SmFzEZ=)Dt1d4Hil9T+#1
zQy&#a`)zYMj_^RWn`%c<{`NLyQ`A3RUYz(@lTDq3=j^-qiTLS@{t_gA0Vf?l$@w^s
z`1ACDw}ijFZP-kB!a;U^A}77pkRY``jl`RLI?j{$!ZUdt^4t+s+MnOq(NE;kBqJ}Q
zo{!)6RU%54Y$fedK6OMsz=)0Tsr_dKol#H8-lN2?e=3Y4etmamIO&gAt{#$B>ZJKg
z)u1oYE7aAIv?FV}o$!hD{S>K}W2Gr6XGGHRr*$_j5V?)TbbKiGj}~cnjwd&f>tXIY
zXR*r?Imu(e6E26%`@v@&q&#qukii3i5tgqpYOd
z_7xomyPWik)VFBuAHqwW<9-r;^Tfb{^xr_?N5U6s7A-cx-%XKm+!gCPtln9P*u&hn
z4!B0W#Pp&JbHa1BzuH82eN@482j)l18<73hPe*cyo+B;!$FM$;7w`|BSaI>
z(|oD(d%$W_>r}K8_b!Q^!gzh%!uZwLzQ=)qU_8e*XWPwxEHGbdLSI^g4|~$T
z@{SRc7z+WYoOq2yr!TuL_^{R6>I;fK@Z_{#f+bK_-
zq@AwdXE<&c%@P9t$L(Lc*%#zG<0nFzfd^d*STvD!6ygAlX&-+;eC0;Rkm4a|*WGl}
zba_Ag?zQ*62*UZV;kGwJ3v3yb&BVKvYd?2T+slf2oZ$Vs?GISrgL89%8$DG*oMGQ1
z#_wh;ODE;$yyuBF#3RmZyb_=c{BF}
zE8ast=S4e=3O;Y9LTV+R?M!Z^&Q#V`)fCnaA3>#O4-{Iy0-({ZJfF@KU`q=AuD5;H
zgepP#nNx?vGvRl<;61WpQ8LWy!28P{0spYvt0a2}#@n~D;<_prPuU9{l&YZr2U>kt
zRwTfF+*A22$6@?Ly1Kf{z@`_?U#BWynj^4!o{~i&P0G`k1xn|cWc$?fn4#(
zBDV)H{vs#;)XjjtMz4NXsRFW(H+R_qnkonP;yxFD&-}Udfa6E_H_L*2R_?pVgOFa8
z_wp&)hlo{nZ9ag`3s%@1gZj1qUGUfja=keT4{$yk8?iaz`PKZga4+=V$1=Uxbl`o?
z8}44l0dhw6uQoeCzL1{_FV;KsE&deBh4;<7-VAzv;OXQN8)$8$Ehdjs_hGq2j|2^d#;#2f7<>tUA+
z%zqPfW**wD`IhcWFrBbv*%7R_@{2_pra2-gEwn>|)^*h=7v;n;f@x<5<)e{+joKSt
zVgBg@PadHiisIlBMmv`#UcD2aciWUM9}W2O%R~|8%Zw#Rp`6?D6S|nrlQrFqa=8O}
zA&9kao)P=KMJt)ShXShuS?6!Kqfhdsci-#B_WON@iQlmG)gXRqTJhEkw5vDGMTj4Q
zf2wDUax85Lj+o!=eEApR%++JW{)>LzL;P~|)K62CQ<1khMautet4HLUmqOO*mUh8}
zf7GS5??t&oH~ICjT!-U96QZuz&M?HK*&)RLT0Jf&{_=nUJMou`M(F*Z)c#K5FO}jV
zh`&7WiN3edY($dC&F7%crF%V6BK|kC=xR0AyI`visjoYC1CcAuruUHs?UPA;g})qX
zQI0BO5IKY6^!X(nLwY}LMzfX3ImTW2j&hGLog#kpLP#z_?`(?H>$%Z@wEN~m1$kei
zO?t#%9#g$R`YUH^9`TolV)@8Dh$q%
zp)D96jlMd2Hl~aByn2Xou_D|*5x4SxlfnG>0N)v$FDx+!2QWWNs_z-1TEi&$-DUK|
zlu+)?NFD0^)PE=jvG`nmnl51c{a~LwK!*^agkX^WVji|<80^f$
zb1&ARzx?yW_i-WE->v~uHaxGKv$5n)2FQ822IOXg+~+y}Od3E=Bjx&|*&vsn>*3o4
z{v)x)&!qw6a+IHrMuFVM66L&ckQ*5N&9edI;+F`VpPLDGHcaej5Xi|!+o+6!oI>;*v#u;)P%}H;Uw@wg`s_~UUI6RT
zY#E)0zEIEDySVTIFNkmRa`H!(cT$(#-cAR>`dIOkjvWugJ(kw;PD+7&o#dTbn-B>0
zE>6CFcTpGhq;=^lQF!ldm+o5yK|SMlA1C~TdZqI=X1Y3n|9?`KQS8@ARabr<*?>Hw
zE5fb`{9?S2Q`VIM&|fxttrPh34Tt}hsKdDz={XN_OkOlohwl$1u*bAgIoqWs4?F<7
zs9DF74F0l_m$47eljwOIEN;KrOkL*v#p!I`O35D#_!MLVzq_rT!FFCapS1HrSRM4k
znU8$ae>j_|*NI{8iX~gAzH2ost3actr_0|B>N`@d{QB1v?BiH&98-vB
zrJn1~GN}Ft>)pFk_4WT?KkDdQaToB9@2+)hKO)~uWg8!O^GBqO%H5dkco*UU6%yHl
zn$X{&t^uVous^m___|F)I{1QvIju!PoKMb8_0QnE|FM|eszBC!h@;6Pj;IqiNUx?X_m^91Al&T
z(?*6P$VDnvjIe^-nnQLcBf#(H2VG}VE`k0Y(p#>9Ja3!fX$XEKSn{1Z(G+3ih-PptgagRwbVaPi?jgEVt35K
zeWB>A^k;_wulmeVSPXJAGv}Tdpq!!NYqWcs@sIUTk1rZ7@lk-&mD71x&yvK@ZD==2
zD^4E9^n!WGd(ltUaOPcz1T_Mvl)E+9yAtJMic|!{0p*qik@eaK*R(Ng_v6W1gXrIHtxS|cxwBjz
z5?+9gS)LEjzB<2BJ%{px;T<*Ma0>T@$eUX9e3D^J#kIcBp$
zFrB(+4)I&ur=QqiyOxw4iJy3~7u8_7=&MxX$Br*HBI{GR%kx&Cp4ks%9%K3HN3CQZ
zX26*>3e)E^#R&fxP%19LB2-%-iZ4xlf>`H4f>P%u8mfca#5LJg2jQyh+k!y
z|B$TTulocO|0LE(?}zj+`$6Q6Mbqa_POYTBZ}VThjmQr_rq73*W}YjNs&63jc8=c(
zu5YIOINK;KXL^wS9`dgoeNKMz!#g79ck?gd1*&Yr75HBABJ}sG;Re4+9AeykD_Kt$
z>3&><{U#FnmBdNnPg|4szm~WCDLyBw97@jXehZ6-D_&sU%bp_jT6)m?B3#(L?RIZX?PGy-hGge;LfXFcj;r{bRBn$6a-h>2_^E#d9H0Jx63p~XBwU+FdIG(PKXf$>&(`=Hy%+p?Aq?Zx#8FVDa&?BDI`
zp}(pDy+m9tCWHP9?v9SVhIO}vOvcgcfMzQ5ZzxQ{y38g)X;dHD_Z6{7gma(#r5=2P
zAiwrX#(rgx+gon-WLqDs_wQGGO@m*aS+jQDqsVruUPL}<57^z+jr;E~pdYw1*70OQ
zJo*0UXOnkRu+B_Bapa3OgW7P2O4CzTejka^jDoEZ4+?KRLIQr?9h=TGM4ZfBTDG
zDpqTHjdWo@#bxF2_n^Q4C6!*VZ9z;QmC9~%@*3E&6JyCy5OIa&D76T0SlyOH&eN*9LaRMFAgteLW#eUBm7(o7ZYGT{zd+eR@|9oY2mwo|sf7RN+{Tc`@3
ztUv0EW-8;K!w2JktyIKTrwnC?BZyR<=$?Y}(@OD+Cit~lsEOSLMT!nBl$GWcF;lsA
zia%xaf(`JYGGqT!TCk5W;H`Yl!@QY#9umab&(%h~FY9Z-^F^}#o-$!W@ZFi1zwBN(
z|FK9dv8xl#FZ(V1HE2M!Q#aR~U!@1*?O)ql11^XsNUjw#^M-g2dvNZjYv>SL
z6wH@|u++1cVEla#K67;ijDs*w-Pftmzn0G)Eb)f@ls7!VtNceHzW(XogeUN+i=Aix
z_5(lRxVnhF0>&G@zsU5z^)Tpw_t3rTWGD*d$yY4%55V}%)Baa2kqz_uhs0hyA9a!I(f4VX
z-)>(nwp)W->FcpN+s9nQ+A+!^J
zO;Jb|%HNOdJh2^cQH72%%KLLO9L1qr-DBk^C+ijk(kfF^dj^%G^&Thwc)2k=8q8~XOmkn3I0neS$dW7=xCW~ay
zKK@EgAm1B?Z%icd=BsLPeOpPW{RBK1lv
z68Vk#ht#hp>3IPXr6_-JpBssT2>{Do(iq9{5CPDo2q3xdJ9Py%U^#6}*96UtI&(HQDd_&t~
zJ4s(SAVK(+Z}=MWoWmLVo-HSTWg?fQ_kzep@2B}o`WVe$YRl;JAZ2UmbIH~fUPOk>Wt_tN_*;jDBV<}!yRdESIOl>E+Xx#{!M_WQDkeBpLlzP6a=qiiAv
z3IDWgEF|Shvh=x{3eK-2{dI`|>0e&|0fK^;==U1Tqu*mvnAX2L=QELaVm(XbjQIk|
z?^V6l-5!5;_jmUQPf!>fuEFv4cym*L366)xO=NtHd;dv9IT5{$gh%ZRawG5gu2gIv
z&YL?KZ%Dn3T$^9D%?`QZ&SD~H_KD_I2Uqj0#PF@r6L(dVD
zBEXH0^w(Sg`&j(4c`fp>;>PzmINyz`y)K~MnuAt%kpFmWHf{Y2*v+~B9qQQ=tzd%o
zp|Gx_7U$`8HQvq0M_2Yg*su?9s&a=Q>S=i|KojMUXi0xT`FHz7ZsGi&`Rz7Xe%Il24o*oL#KVJm=C5w{#UJiC&w&s`a
z4EQ&jRlF^8J;9%;lq6GOf3p8}0CxcRv7f6eqOl)N
zw76J40UU3dXS*D*uBM_E&&Me2>S0HIRPp+07uwPEeY3}q_pNJc`4kN3^PPir4&c|`
z9Y0vnUq~BOpxq6axPH$b&`IS3xfg2Wj?mSQkk)h)vabNF6}jcY2Im>h@zi~7`~&uP
zV$ZTU5P!HMx}$O7?@mg@CF-Se)^F3bmlr9OKH&E(-vq_UK>ui^u1>)HexvBavijg>-`MS!6}k;PGpX;kFX$;Q
zG#yt4a<#(e<3EEw;Wm3+&qF)X4_u$EhwsR;ioP9phko*jWwrYY`xR}hqgO3Je!EZy
z$7YB-bhRgUF9hDv@xbGt@7w|Etm8uVaK(P=r2VDgDXCtnJu
z$N;`CQD44RrHlIb?81qjv>vMF6#pQ6^hQ|-#TiylgCARcFL46;Er*L`k)}EDh!BA)
z+q_On_r}|kE5y4giB&8tTgPDCF~4t(n(-&p)8#MQLVUh19==Z4;S{TATS?xKYRlc$9|ZeUMYG=*%b#nZjBEbw{qUugDqXV0<*8pQ
zRh9Ak?eugD)%#=aWl{Kk?d|gu5fzOsRH|Lj&efaSs4ptL7oN^(qq5Z3Rxg@vrBi|>6t~KiIc2v{@574>UE!RxB=;lj3zuNM!Zv;P
z(p&~rcx>sXd%)kk_}e+U(}0h%`A!c;!1*fkrq*ZBpWOoo!u){G7xue)%m=JEbhywI
z&P`XUmKmgi9$!P{w3K0-n4LeA;s)a@L-p7z70
z@-aW~S)<2eb8tSd%65A=4)%cMhR8U^TQ!~Lbl`chmG^8W#zD`N<~<)YKo7R01V5aI
zU(4<-PK0@4c06Uc9@2TOl6)diE>}EbkOcXs0xk?Jh5Vd{OBbgBvRpiV2j};m75?$7
z0Hq_0r7_;7y{p~|%LgRH_g)2jWbyOOCP0Z&wmWsuk4GLPkBq>)>uFRkT?O-~`OAkS
ztk=DG*b(iCgwhEOyw`BkmXhK^NNe6Lyz2uvd&l%rHNZZewYm=gKNK!HC_i3>LJamvzhIXHg1}Zr+f8$yHTD>gEBDB9Y_G#c;s@5#car51si%o+4BPK+NlL)>V9x
zO#G_&N}nfa_s^YKXohI9CT|kU3r6OU^9w!H4iYCba*vC^@^5dBRw8mQ8;r*G7jb;n
zMSYAnv0cUd4Qs5c5&i#~6aTr(<9QpFD=%KS6!Q~y4H3V6M7DzX%MXUvgE4>qgNCiB
z=jBH_=P;dD=t=x@tkwd;_kNz9P5irhbGoI~Bar8N9AH}mo
z54TTrTrA%65-I;p(f6cMTk1%f*TxYdN0BtOz1*!4yFIkwB8
zGmG#D*f=fg65mvfxAh$l%NYK
zuQsCj-la!0k9uyEN%9RN9}_)ucG5ges*#S@$zP)5cS;KMcVG)gndg|T=yOGdj&wP<
zCLOmsz(=2te#*S}?14Yc!`ztna4lut^AJ?e(yn#}vv*kb8+yqVyF)kD=ek;t@R#?Do>>ltpxZYucS8a%=9>dfro}
zdFdlDI_(@m%X4mAWFJ3?V_NMECgYp?V;dQljIZ>5s$(B78AsgD%!ob--r?jq?OX54
zaD1-Rr|Y+1{XohUdTAd1*sh3-Bi-hcgtz-tm5}m~@)3e!>k|oHa&RH_A3MTF>V0{z
zpXBekRX}im6#YG)0S6stUcc9t$f@&k6TMeU2V5odw@#7l|BnUP?MHhgV9CvbJeBc=
z-UnZ78|;g6r6URCJo`)DAre;?`{0>@abh`gw?TN>IG
z{v~IKzO|piN&Ass){%JqnHhThdqYs+Ft&4Eet_t?Ot6OZLt$12d2S{`lhoh1`A`7L
zeeCLWHwJvZZoUNC-9xqvAr(LutMXF<;0Ltt>g#cW-!xGDq=0$~Ti(pH27IzY!yDtC
zNlRXZW4zMZQTQwF+q8yGG^3v77gRgYuG*$_zC^p}E+9M`<=9g&qGP>6Dxt-7TDO!X(oI2qWO`Vs37*}a4O-@kQj
z&YA6?=PZp`dGX-K)Bc8Q1%N*C_N!Fnf!AJM)#LpD_{^r?F0P7MZi0pUoLr>0=%bvYUf}66!;EOq03_(=*QRgpp^mjcXwMQrhzrw_jF^$P_Vlw>&1Rz)~B?jiB*#5NHI
zb?K5}i#^;UCRO~q^bF{E;gj}!N!ZU0KWaPI_Y=gkxzu+h!9HrLZRQ3S;Cc0uO>rM!
zpP6gjzstX1{nMCi+SUqsr%3w0%>-W8!Qr5@64qPn)5+JpK>k%%-%?zEi0^*4v=QEW
zv5ksMD&Vp!EIVJoc={&wit*Z!K`~B8wrRre!7ChbviCTQ2U|Pdiz)4t!=P{(X1`#&O5vZI&V+pFEg&8NUx?Tajqe8qiMZg@ujr
zeQl?XG&`&~4*inuab))$hy!m3+EV!o`XS-MryDhP@cTQg;aM%qw|cY@vnO$}wu@chNMpEIsQd$IW^z9JvD5A*kD1%3Ua
z6;uUaohUw-H-&cQ=*Bl!mqPjD2&V*3NK2g_Ua=jX`&(voAO+q-!v2Fe-k)7+L@DON
z`tjfb%|g(Cdc#=bP*>4$UIB9XSB^ZhAMzcm#gfc!tSH8K4qV0E6hJlY3-$7wAG&@>ZM0sZBtCmz!Zdc24^T8-z^_8FN+ZH98rB_+K0kuvdLTDdV+OECD`u0~OB%xl;%e$JffB25>A%0`qHiZc+*ZU<$zP}nQ_)6kZ
z7YFayqg=b!^7)93kGH_TkbwSNlSlmJx*gYIFyAoKhU`Np9W;P{4F~J3AB+yn5Bjd|
zi8v|y=qF;2yBUd(IcFJcz;#KP+@?2}?<{xp9%9Ck~d^u*7~!aC8jT3cu(}$
z;GRX!OR0V>CH`2ZM3vOnx6t1U<-}?X2>(A7>P-BvM2af$$5w|Q5I&rveUQ|jzeAhk
z?^m@W*cvUEhR-*%)BCqYTUf~Zwe8xUi22%bCrG))tKEdBNP4*uY&qj~>SAHFReGw;vnRj14O3u$?YPP#tlNP0g%JdG|l>ZHpxnLId)8SmcMPxJ8^GrBz9g?>JY
z$%C`^(rNf-3FQ5Tnf(7Eli#N?`|chw*Bu~CjWaWzZ~KCN?sf;gJ_}^(*~+{ZI*s}NTI!hc8*bD6>cVUvE^jCL
zQOtha&)gr->7n!C@?28>-@Vy|%;!o^)A@h)(KvjQ9+!tX==G)GBKltL(GvO|j@I3W
z$b67{NXs9sr`I{b8uYjnR^dj~8Qadc0|+(c@k4K@X98HAC08
zG>(?Pa)^dlOM?}(m<9gg2cxzxi4$hnc8
zQ(2W*uBpZL66NQOKOuUS+UJwF1n(0W66dcjup#n=Ok#9Q2_%Fkds&kdFlxwOTaMDOqcQ<9F4IY8v!Z51Q?hT1=uv77*TE;*0#dw~1`
zKQhp7&WqIT`~jHL?n(A>wk{pAwSaVfxy@?yg9oHO-$XwbYSy~@J|K^d!G>JG8`AO-
z=(ldh3Vp)z@2ZwXsE@)knU}bZDtQ}iivGQ_uP+ndKbJ3of$vqQ#5#p?xA)BHLVw0-
zZ=jC;_olKAiDy-PXOY5kr9Wm|h!&QvzSz#xht}Pw@A(qxTRechbFKF+0v=P>_T)wh
z@Dgo-1!p7x85_A}*?`~g`MCVK2gu3Fy7nsrvMzkOFBW*pmawx9>L4dr6E#;B(rLNV
zr&0g$%!!x)K*q8+rpN=58X^TSZrCF%djrq0^0`>|pgydLp~Hs(S7-ke!{;74Oig2c
zq~gI{3jkRnhu7izn6Yo$jP;z{e1$_3Fh6LUAol0uLBGFu0PhwYTY3+8V%wT~ixq(X
zm8LYrYXNVYD}4Og9B6Ny+od}bAjjg*crOonO1`=(eHq@@B1~)@4`9&vf(mnxPb$$^
zumQ^Ty=>a>yMSb~2QP6yUXpkA=`MKw5#!aSXMk-xGglyw^_<`^SO&ODDz8Ek&`-y(
z0C|+i#ejU|M>DBjnl}K`!&TKVtzl%|EDfkUwz7H<+L@L)cAg8;waM?wHv#T`+@R44
zC^l^I5AOjS58_w+0dgXLPA-2B`mHoyqrL#rdzCVLB>{8hyoka3do3T&uT}&3jqM$}
zy1=8&3~#Yq11#t@QT2lQ*q%41zMtq-8ox6J>|5!u7iAca7ev9)l
zK5>&(vk}MT+o1B^4}irQ%XsVoop#+>gXg$%T=j!hA>F>Re3>uc^7jTg*w1TsoJrgQ
zcr>e`8^>$e%A;Fx{h^i{p^xtwCo5or^ERY;e?~drS<_uUuK;s`-;jHZquYzV+CjQ^
zh1?1pS0*-=iEW_oswmN2Q;?qC3o^7jX?mo^*)w5;iTX%6`Iiv1w|K0L$bgoj4{vQk8S
zCmtPA#ot4HEO?C#;0ZfM3(k|fY{4oxPqz04w&QztRsOw>dV3zKU$#Z){go%HNTom&S>v3plBYaq+t@$bHnLu50
z!2GgtF%l<&a8NC#PoJAl&V58(5xtFR$DJ?T5&Q1y693TAd5icrj}`iax2Md$MtHtb
z$p;cAOFY0v+HK4UB615C))4zIRYU(jl^r~5iQmomFi6T@sMG(~#5JUiq)&V56T5rt
z?<~T%tz^>3^WVLiN%_WSG~agnUhn|*`rJtG?-zw0CH-+B@idXU{bm!Eb_cZCe616%d<)O{TZ6KKS`k5m7WeK^@}W^nD;N2ZKKOGGU@tnh11X9Ve-cto9Oh*F#4RyVm(?;hC%1w*hb^0
zLo_BZ&ms5g(rK34G%6d?c#g?a8>Z>B<}DhlneqKoOx_EJT8aMFTbXjqeTQGn|1aXh
z?%t%ryLyY4Us|^Z&BHW|qU^oaBQsA-J3=e~-D2`|o}J!~fPv
zKn{sKTsA-u{t*06|Nq|Wzjag?Q{VVHdR+FcpvSR^5uLu{K+mI#ar8b`2Xo!B-kGlV
zy9SNk-8AY<(Ca8}rk-|8c^2mVh%I|)xjW4J8egAfKF=I)woLiE%>PfB7etp=G0TrL
z<;9ux!==mQ{hq#~QG1G>AO9lh{iwB*v|TCZ5hrqN=jeTsl9lv(y&k3apDMo6^K>is
z2O@WF`6hx@_vn6*VXnJ3O49Q#(u=kmr=QUKFn5{jy;&b=`RQkLKZ-N|KapW0t&an9
zoo&R_zn3{53!~`yAN2kwX*Zv>gxH}DVP9fTmxR;%DvI}ns?Z+ShRzUwk$s-NCvu@r
z0g<=L-9hZ;RJ$eFf05&3CH{tY3*9e;-SoWyZHJ|ZJpZ3@;wO0ZtYAvO{zyA*m-jUe
z5Wv#w+8~Wjh
ztU1$&9(BgAksnAI=;tComQT|hMEOfk8Dt#=#JT{}949mbP`+uOb0*3=NJxX}0G=4O
zI}h>AsT88mho=MMnD*LvqXkj^9v^xCU0v_ui2OS~;^zW&dmmvsC`5&nTLB7BD@W
zT8{F6*h2Q8|NkSb$%cF?%cz0eQ>m@YI*9!G@Zm8CrNH}#8j$Z~n)}SzG2PL-C>h^3
z_-1t*t{Z!{(Em5?(&kT3u^;c+ov%STfuH|Z*OkZB_%vL{ZYB
zg|ejd2`QoyO0+1IqEd)N8zD=JLXsjP6%uzX-R|u!B`J#EJo7yFpYQAS`{%x2Gv~~i
zGtZqlb7r1-&dj$a(!GDVHN-mUN|Ez9BL80BkPLx(kIWJXx>ubX-rp^ly6G*Ep1P}f
z<7I^P`QbR8Q*m6$Vi}ZnJaY2SLWFmmm+P7#^ycL|65kEHU7TA=tY?nth|UsvKax^o
z6MmQ6@E?7mU98;`;|TvO*}fj%Cy`bqf$#Z=)VVd8Xn*nQvp61Rk(HDmq31|@xFVq+
z@LuW>>krjvyLtk>CrZt4rZqI|%~P7R{|(?=_A5cSOuA8H_;
zZ_n4bXN3OOC8wwn^tgDlaH9VEXDj`Q|DieZkpjWb`1jJC;BzcD?;-RQGCpt#`ezmI
z#YDLbPxlRkey{nqO$6p-X_ymw{-fF7C**}_RrM0(o+!!Uc%n_m3la%=W;Llx+7aI4
zF8Y#=@J~Us1ktaKTa=s%LpbkCjRf)h8ObI$i26QeIo}}GMNcc1T_x7*UtTuvBkI@E
z;@K0=b(N1TI$D7Cy$zOaA<|pU&cp3q`o&}$k@juBwur!o+t)87#$9xx1KyWc*7tQI
z@?8wrTmlznKGi4m?)J;PMCiG%T*-)d4lQ>Cjf5UYg683IL03q`(RW9cqmY-dIs_F%i(6P}Lgu}gx=2{}ijXJY$Y
zob?0SN!c;|q#f3~hwYJx&Q_fN-SQUBXIql{$m37y;(UhoU2M-PlF9iveK$F8rFq$6
z`q6JIa5-a$!Rrf&tE4|4FG$1n|NDLg*B@B58S@==e~!~tiNA1~{^mKB^CyYi&(qx_
zgY)ZC=@?Hx+3C$a%Whnk>JQx<}E2@*|>8k^A{&)PAV(car~Z73pUTQ~uf#
z9Wwt8kF4+aLy{ijNuu`}5-S&y`;))SNIoOVUnr;UiJ3zA&6Dj(I=U1WKNoe%Z@)qL
z32otI{;f_DwW$4l^pO_kXHkB6AmvwTt|aqkQs>(XD0(sFw~PH2j*P0+kk0hz@z-_K5xX
zMylK_3Q_0AbTqgzqE3wQIYmE8rN#cc*zXm~6{peQ#O1{PvDlv$zkd+RMNF6vbux_N
zw79&OPrRN%#sKr7L5ERXpSb-0hvIdJm@d};e^JaQ{{O`Ki`P|X(#Q46Q~suyU#y?_
zxmHqqCYapIHC@<;$n|&`VCtZ%6s}A=EuZ
zLMJl6Hj+fKUdO3&T9n?esQ#=$jZ3jyeD*
zCJ^S&r0@rY$<%mjqt@Zx-$=ea7fJiz8%f&3>4D^W!^xhc|4SxOgDR&%ec!@mk?W||
z!zBG6b&jtnf=p{u=h#kB>ox=GJRu5Nf%O}q_)niB(-M?j)1vx=E2ZxdYJ5si_M!MZ
z$#3+6wC@I{q+Qm_Cf7OF`pNU1ks0xLKB>iemXUn^w@BQx
znZ$zAB>(AN($3pde#Z0gbw~LsLXXGu9^rZRY%1dgk^Y=JH-^B-G!3k0ZK*PDmoCQ&
zKgZ?tIy|qxOE<=T!h%vG)P>PHgNn1e@dl}nrKLJ9f6eYSDd#x3?!|LMI$WQ>AO~Z}T1(7#Fph=wv5=aFOc(O2mfyhlT#R&%wh`&E
z#&=x^T$G)!N#L?&=D2B>xA3c?VpXc|eI-IUAZb4fiXg%@K5A%r*k@uGV
z{yPB;dE}oNOCXbi{HK4}cs*5rQRx7YzQ?QCLHNgs3+~M#{PV`85@m$n+)?3*zmNOF
zxVeJh%l~=Qg1}bE$JqpCvBu;7asMT0WD3yx9@zs;1V#tbuwQg!YBT;nJ#L;!Ch{w8
zcp;O8-v7LNj7$;ok43%>1U~wv$09H|YTy%rADg#hKX06WBQhOmo%vz+IRdY`4n`Ab
z-mo2+0JNT%HF%Z4Ne52r5c;cJihW4q^zi
z8Tk2D{5ycQc#!|xUlmK>d$q100#!@9u%0T`DVRR3S_|(d|M{H8A?V{jALt`6#!?FF
z>GNA2?*qFRT4Q||9(siT=M^W_5(qwrDZ^OLJO4uQbIIAM^^(Xh%bfR_zcaPX1O^9__Yii6{>1%Ru;)8o=g)uoE{mYIOr3%IV^Fd+
z?nj@z-r;_xukVlbyuWoa{-1mws^k7S&cYJc*Vmwq`+=nDX=y@Eeu@D9&$FGBL`3@a
zQ9CSm^Te}gYDfF=mH%P+TRX2HQ;y<$2fx7j3%69@e){0D1|EOE*@;;G$I-R8AHO?g
zhxNJWmx;^o2qoh`G@Ddm`cwbqxLp&X&te=WNX2qKXU1ZrbIJGG+jTx*`jH;ePMivf
z!t~^a0AE}&zap_TCjE&oN;rjKc^Bt4dlIP!3)b3$=Xe@bdy_iex
zdp-H0i}?lAd68u59=_3gq@DbCnY61N`f0d)?sL+P3!6x^i6rfG!eMg1>|w+VEbs9P
z(y#Y!Bkk^Odvd;V3nuNhYGo?svuY#X1DrTV+DR)q=~us2C*^LJA?bhG$$2q5l$6&>
z?L(HSk@8&kka`X?NP6N^5_$7TymXgDjcMdO+d=sc@6^fsTh~eSrRLoa!DRiv>Ph)S
z`6S)jnM68u4#u>C)XTb!yc28ZBZ#Fe=3aNV#v#I&_9JRkr-$C-(Q0;1dNv5l*
zdHoDEZ;zw=A4h6lp0$eP8=&^%|5EdIE9H-f?;l4Wwc!74Q3#3XB^XXKD8J+!r5e*)U;`1?+
zDS6ka{AU!RNd)t4qUh*l7*3->f)RBhjAFkQ86}+FKp`>)I9))YSiS?5MuQ2{k+H;x
z1}jGNgHf!97{z?zdc}w)FUQSEs17PebR2e+Fdj{%a}#S+DLUt=n9<`ZqbZWfbkpB$q&Z
zDcT1byYFZDv~GCjGybx`ix2Z2Tn_mV%m+a+2qPsz3+hb49w_T8?
ztg|uuayL{ja1#jKyW!QsOoLU<9B2xCe>LkM4fLyfU;ENJ;je>2f$~WX9NMiY{T#_@
zw0!w9e**{1{!L$38bgCU-><4%%3y=S;#>b>ji;2=pD@)*lQ<
zE(^wlsYv*zbi$UUJiQgQU7+{Y;&3~P*P!uep5(^}wBI#%M_EY@3#RHm%jMo?gI+ZE
z;8hfVqGiU#Lvi=ezFSIez&>=2JVg{sm$hSqsQ7+@{`O9EpJm#Ler^}YTJDvDKWJV%
zSO3%YIr`44;qUy|I5rd<%y3TBV!`g(!4CP=45)d#e16Ig8gvA-&QLl|17)Q{-QNzg
zV5h>2O`RMjw0oQ0(oJN5Ov|_LEjn}r!e7b_j~Di5
z4aUbOIK4ydU$mojr?8C%!^aN3Ge+%8$(4-F$w2zn)Uy3c=rE8GoVUS_4w*7V&ultr
zuw!qob}`aJWyP~VcNH2e-Dt7XVk`|JHU7*|+{uPGCG8*iEEcqva=hdiOxSmL&XIX2
zp85*zw2M;cej0Qxz(sdD6Mauu8}Cue0HgR__n&Jqp!IvLzNIrA%nYYFB#zJ^)W?dc
zypjgTi%%Pr8_?kI@eh|@dD9^1c&
zaJg7I#JuvV8Qx8Y={aWc@(;j
zF=GAdr$$vYxa4$ysnU5G*f+ad4SuCTm|(@plu`6t8I8FL?`bepQ5;*vr$PFe^B2A^
zqC>|6sZs?aI!rqsuD-~S4hy4>N2xl|!D&i{ew{fT4r{Fa_d<~lcP;GS&zGUY1G(&#
zC&tqu;h=28v{)M4Y_(@wJ)uF8FE>znGHUnoFqLpEIy5z=C7fGA2l>PRtISp!1gO{h
z-(u0grO#mX6DtOMZLkfCY-55$$C9DT==x1sD8~Svx
z9->Y3v!;Wp&9_u;ARRcVL0?9~ken~;{tS1q;FTB4G%1jU^2a)Ak7a>JzHM*u5EI1*
z2@N$FWr9-s;r&lhyqp`-d2XVY43OSXuBcCAfWjf}>^eCn%u>HP@AMKTRLp%hoW6z*
zwcNU6HSf^9o5S~hyx7Np-Rln?AK%7+W4#wFd4p!kC+dD^HS$M5V9+;xEtnu3r>?
zJc|jT3ZvzZI~dTSzv_059TQHTSfqS-9}~rSbbJ0Wp8;*#vOnvh_UmmAv(ex)AZh;8
zGX|Sb{Nuvd(6j?c&f>EgsWMC`d!VpW5$PGR_o9>#wR?wEl<#H>RNt4eiXA(du!PZT
z-d4s0O+C-wf37p(TjIYtn{AlT9CC~khW^)O21Sp?BRwOX97C$SneZ=e_S%qe6enqX
zmY<&!6D}UswbdLA!00R3(b{^i*yB5;#RQ-u(35pAKwXblHi~2q9?gEoJ
z=)3vVcW(HYuwc23mPbJf(DkqJD?6ccNksh4
zvQF5w+9t_esS7%@eja&vf(=*gmSsMD%7#PIg=cQtcYl!i`-`K-Bry*awaSCs?o5H?gB
zR!l5Mdd6%iil4d+>DirryUVK+vVSVAYD2;6^KK4Y{EhsuWrtf$ChbA*U9&_d*eITX
zvf0yVZ+CV>p7lYF$HFf7VsDgl|5z6+v651_ZGrYbwAQ?6MBlG1>lz+2_B@JHwzKR+
zPgf`S{J2mSWzz*RI`eF|OQG-ZTI~KSMc;qzXs(|772S82*H5$MyzPYBzSidpW}|po
zKJ@Mo6o+A}kL4AXJNiEJ*XWY>H&9#^e&?Z)ZJn^}zoJU-o1Jhq?|J2}ZCzm7+7j9+
zi@t-AbiBF32l=&+^L;XRb%W9Gn8!NLyP^2|;tqpw4qTG!%9N_-fSzn^O-(!pc4hy0
zVVH)#Ytt_C=#xNxqgGGu#?KrW)z7rjeB1+qy;GmcqxcrBlU|ibT6V*6ze~@R%eo=J
z+-!bBG6&2Lbmslq*aIh}6YL!-dthUqM6gx?+PBdhJH=%feaDr~bFVH!_rGkv8Xt!0
zpP+m}QJ2SoHRDeOTkl8h3JQ2wfzE-dKPhY<{?rW{DvVF8Kzhxyh|7pCwI?ZA
z9GK#9vFpXd9!UDTDYJ7G4;HG*EcVdjfvM~Yr$a$JC~B|Jub9t8=R-dY7@g+Af;?e#
zRwEZ&L*eq~5iaa2^Xt*v%7ffDwyU$tdSFm3Hs6iW0|rJv(OoQDIH@}O%-C8k*pKY+
z(iI{(J|8DM7I2W?dVETUSr6D6XYP+d_v19o%?wu2;(~U8X6<@Z&*t_c1$&=yAfYLv
z)VTx20~vJA?K0?rei_b~x9lG9^{lthG~^PXJDmR@EpNHrPE0^+Mw%}rJd=d|xCk^}BzqJ<%mm2#Ht>c5r#J&p;()sYn
zscqhh0WQ?eoAW8!jt58k+UkU*JSbRQb2P}c7Y6gPXM{cL1)5XqzU(M2cqpg4kD&M{
zoaZi6dTBg}T57+nD5@89G9v6(*Y&~$hc!9v>$q@NyK~#tGA@YvCDT0d1Y@Q
zET3NC?BgJWWohLx$=U)q>*%^|!8!rN%^K(9mmz>hto({`#|7Ydde)t|mjWpKS;2Qu
z7J{T*>s{?`J}92cidF4G|96GDWyN9vRG$v-dpF95SI=Xbj=e?wLh_|z^9?>&)IJ~f
zt>wc)&r5x7`lx;z=}0TK5IT(?ZuUm^Go|m{XKhB~z|dEkWTz|_XaWj2|3gGlCx3_n-1hDXd>XIoA0{H9v(I5sr7sITnb@mE;^nLe)
z$95Bi5WO@vA!LgXEN-?BX0H~4voJ@)$3O^=Q%r4MI0+%tYyCA*Cm(eFn)Pl)dP>?(
zsVk--y#;Ueg~0--ps}4lqzmBU8_!--FA-EJ1e!|vNZtSv`+VXgVE{}wdyK->SBtr3R{*JCxd4In+gU;AFEcT5OBXRsDTO%%bK
z1395@LqsqXzrWHbNd(JfE(8{&3P3|miuF2M0Fr0x{O$z^plws?7Hw|C8Gef
zqa-|d`4JJsJ(;XKxK9M3&!S^bpy$t4G%2I6ME7J>8s1bf7eRuvcVvV;ij#A(dS%)%
z4#aQ2RB5}G4-xg7HYu`1pn6HM&N@*9+Z|QL_96SVZm_6xp$Q)*-Kb-K3=%IF4(a5GpfmEPy}yG9#m{p(!d)(coUbMSrMZh>`0bJ4
zwkrZKJYKJ}0^QH`<7d_QfJh;VhoZueGZR6O%xc?E7ZGer7~`O~Rs_B?Xl#cF5jbSd
z^~muMfy#-xs6f6DEVB-#okQ)ad6r~qaZ3aX6JASL%l1Le?%`_P1QDFeavq&!B7}z3
zk7vz)%7^&!hd$XSkzC)aYd7VJV1R!==24pn7MnR3p5cgq@2<31<&_9dN$=i&XoCno
zm&c7;g7gTKZ!!@fy=y`yYL}tsI%Cs=;;!=|Xx#8(8Z`6uGdI#L?
zYbop%LQV8U7tj(xO#i58Zn_A5-TO9a^92!9Pj$_&3rGE9v3;R+lL&6i9z9nvwh#Wi
zomp}+3iZzu54PB#|MQqK6b`nc{`>0Xm6hoEU8|HCNfZb{W`T^ej0+Ej(
zOT2TIKSJ$#zwX3Py$Jr~R6Vc#C_?vhb$v^!7Q*e#K4U+(@Zs&D3mWqSMKHfRUmOY>li+#94@N$YdWV~Mh1zziZ-Mq&K|G;Q%?Q_WPC_V2tL-uB+S=!O&nL?0|S3gyo
zzyk@r^68WI_CU`9)t!21{LL}2F3>+9f<@JKKi$6x!H(0nrFly)=sHg~n6rZe{|8N2
BfrbD8

diff --git a/previews/PR1624/tutorials/out/solution_000030.vtu b/previews/PR1624/tutorials/out/solution_000030.vtu
index 1056a82c4c87a6d4d54d40463de74a969eaa557f..d459fcdbb7cd02579363f37d4d62bddd2e2731e2 100644
GIT binary patch
delta 12125
zcmV-jFQU-klmX+E0kEY50XCBi1TB-(133XPlMMu90XDO81lSgTdkg>o00000c$^iN
zc|4U{7st1$FBAGH=LZmVzMWzhb5TeYXt}(Nu
zL7gJ$lFX(3yzBjIf6lXqwf9=T^kr{L7%_7v|uP_%38JjKzKMFr_FZa$XmW$#FF
za-qEZ3kQnNT}khMxB+92v7NUTjE+0pE@z*k*uGee(|VTT-5E2J>1QaGw7B(pDV*1>
za<(J)G{uTL3Ez&j4EEwNxLCu-;MK)*4>!(GEHuXIbseQx{|T3E@JEWHCh|O1y{5R`
z#eCYXlj4qu9RW_w6mRjpX3bj<_ZwZr_30_aCuh<&&g4^nyrM_B+~)zsbWoo+R2-3u
zIb758OAtwW&teHoGkDuUClRTS3~s#`mloK=VE=?&zOzjXHZ(tb{Cf?9f7y%&DLrPe
zOnqzcTqc9_{Q6sxVi>$k#cp!QoxzEh^khQy7|fCV`KUD*k;y*YXBMuAJQc2vF+2tK
zE-MARv_Ry46nmKWvN4z`%w-=P%Roe+ynG&iKR{$O+{N4~4({tEU!)B8etYlb
z8NF6SzBlSh-)=-iqgk?LydKtH`SnJu0TFZ7-S_APA^|*Cam>Gn-1)dDaRk;GhzT2?
zX@+|%y*X{)0`?pQnO6ACWl&%q3iG4Iey>ttKV4IHkRSHvuF74T1?yI-KhcSY(T=}3
z^Bs(TxhH44Dq-JF8|T#vurAw^$I9jzB7t1SA0Jd8;``z94(l>Rls_|)s-=jCEkrb4
z0Do{A68PUdN2EFDv`SMJB4_zh4%foG?x)R~f5Ln|%kYv1>@#ps7oE#T!G#u-TNs3Iku8cFQxk
zJ0+C6`#Z(yqmDxr%&hV|=x`kj-2BNHNlp)=t6M+UL-evjGhL!s$vBUVt&f=6b3cBCcCkDwew=
zvS{ku+6|D`IV^GG)M`X5$FobF;oL+%Jqa_={r$)4-#-8kq{Cik@m`^SSkBRJb5aGx
zaoJ*VTO}Dhu}*qyPa=Z@Tr0z$FGs}dXwS7s;Em*?!;;bAh(y!=7g8mNluVzO^=v|9
zZDstp1LW;Z|A6f`A0cv3vCZ@p=>K1d=k|v$Q0ye~-AW;s;>@ljYZGAxOYL{@&PZag
zww9_-hX^84`(MoPTO+c6T_J63;SM56<)*y*YT^Bj>1aFfMe2e2tY#%5F03L^)mw;E
z|CF(81unMn2~OBTT}XR+agXmUiZ`s2+!jZ{Pr_F_HEu9i^jcYdqA0{s>enfcvv9AF
zZtoJ{+n_|VmJsmh=g_e=-f-{ytcgHxh{N21W>mZr#O0^OTOmn*M656Ri1dN(j6|94
zXwX+@fk5~^4#ib+9ZKSU5MQtTGAua1IM__^m3u9qZQX7E)p`+c?9yiRn9r8nZXr?UCKfiDK-qZ{x95(Vw*?>PJJuI--55j?Qmsq
zWyy98mH%_V;a!M-vj_0`?=zCoz$r~WCC@a7C)Y=}ckP9UOmSPZI>NpA6ngnRS%`e(
z|JH4GoWZ?aLe)1Qk0$ocyz+LUIBdItQA`8HCpIaFsJJn>-cU=tL<;K88`p!4xt
zo%>zjzRvk6f@9#9*6}SH>0Lxd*zB5aZODVYC)|C?7##F}Z@AW_If^5~;@s>YU$@n0
zutJ?FE*s*UyxC6i5y`YVn+V{Q&@nyL9Z(l!`4Tc9ZsC){X$QgY34c6(m<)DyIA?cX
z@k6BN%god|(1H9Db(69q47QwWOl|}|_P%8Ozhw)k@WSs3p&7xOhT
zm}||v$PmziikVxVlPFqE!smUvD7`?y!H&
z{?6}3o)q6LyU|+!{8_n4A#ya8VnLll%8I}P(SuVdGr)m_LS6F^@XL+1+};b2pK33|
z+3t{+FZC3?9)iBl;HwpqpcCEU!t!DZ1~>C?hmFC1y>GHAG{b;XoTn}4*f5q#+dZfT
z{*>GZ)A|Z_V(vxMEd~BW#jhXA1AFy?%)b55H`J@C-#-j`GU9gJc@Xk#Qk|r`265*X1v35VF!CqJYpl1*0DA(`x);P$Abs1t8T>~gi7`$@F
zuZrS--elD-FX)rvoB1<8Ut@5l@qnxr)RpQ`M-v2n!|$ou``X?RZ{$7rH`M)KYh*th
z13Pt#Ec>s5j%DHgjPq}n_M
z`62e_vApvPuG7%Bi-Nj!cKeqc2X#am*@EwXGQiHYTL&u~;61XN_Y5oo|7kt-Yx$7N
z;7ud*&S@Kh5;KZfJ1$E=G{-9WVB*l^X0XYi5^EV+g6*f)`j)}R?Nb(`_=c^V4
zK8VkO5%+ILKo7r~H4lGG0v`U=-SNf}`jGdYE~ZN%-`(y9-h(_F)jKI?3EW_DkDbpB
zggQ9VCGZ0JSAp9-7M~)ZuRZa@)_$6Q;_|Gev7zA%zT4TN#0@%_*eJkL1aaVbmYp{M
zoUQO|nw!r-L`d1)*TE5y2<;%9rqzfP84Ik>^Z2ALVvt!>F-^@^(CfVd8wf9
zZi%VV=_HD?J62!m5@GO-%5*a$sCOF1>V>bsPZ_BW@f7-{Uq?)=@<3NZ(GGim5}@y~
zZc#0g1+FOaS#X9|A`+`L?Y0g4VBcdZRNg`HjNt{_^uH*!X4!RkK|MCh5(}58q*z=3
z@ev{L@7e)BhmG)#y0*zmKj4?;^CQ}>z^}T8jSWwLkChqj4_F0=tmu?k)Cc|3r_LJJ
zMc}uvUHwV@zyXatHr!@q6gLomv71Stx2mz1rd$8tPY8JAOGCe5rP%#83hWq(Jvp%j
zxOYyo>hqWza6nD__|YeDUmYIH);2^!RksBhK)vIv5#$>>0e#y2AHiv(3}z=ydB%d?
zLaH1$h2I1|noRhpK>jpU&R>_vpji3$s|`MB6wmP*=#J|!*jrkSQv-H?{LLB)V&OgP
zKPnu=!EYf2QL351%a{`NPuu%~EOd8$~>JN)q~g&iwU$opbL!-+Ryb+;hKO$||p2=f@a!zQK9%i^m!BO*b}c
zfx*pg9_zWs8KpW={Qg|K(eMGdIRi`*@zKYj%v`lePb;9Oi!rj<{k3fBQ_COE%aRmoh7}Fvh4^fB5qr
ztY7iCaQq9H8#8RAjJP{P!#A0L4i5+a`~_?^cNkd-#%twC)?wbk(hIwae__j&pR&>b
z?+FGS`5WIkZTqWv3G9y+o<&|DgI#Kz3EH$dynhG|ms-7Z0M-2Cjh28qI)!ga5bwCV
zJ$ER`dE)yIAuv(dD2Es0fzz68bgpe0&n?9fpC{SN!Uq1Ps1X%2%nR9C`L6@`M+Z+!
z8<-(h`Rz||zAvW70uI6)Tya
z!CP*!fmPs(HkDCBuvert>N=>&p0m}(x)TlZu4lnrqV+=_VAKLj`abBCSGHIc@wN-_
zscZ)gHubLUhX3uFt6St@AJG}?REAwDbArzl6dzu`zk>E<;kOs2fpo@R}BEu@4J<(0^6RXF29HPA}w3HDKFD~QX4tpPfEO&
zHw~RsF^f|d!Or;}fAJ!T_BnAW-IJKG@tq>LqHCXoEGVFGczgq$vwN0FI=M1_*on@c
zXx_Fc5wu|UJSN>d)vK0}!uY9r5swh$mDze^Apm|=Rb!1&%16|VH4ni$kMaV3P-*r{
zaw+CDuYc@H^+_^awa=FHcbUN~2KSjddG>&B6tYM85I6bee{-fG$VwL2u>+h|?R-xA
z8Bpv}AYEh*GzqJNvPG&C*k(ivf(zu~`$w2zCjT94{MsQ~MEIjoD7;JmyItU6lm
z`IB^|m$7a)<#V3p#|`qmm)l+izflE;u{vxE#&D7q?0%I;CRelW
z)2bCsLCf~~e_e+3I$!*)ENdi}nx^gH;<;d31q)-f+=n%{=~oYP;p3GB<8Y!rq+tk$rN
z#!H2c7SKLhwdUMXz{>(cT`k~!Q{`)PPL9}v@}zg}_-*;r4??xA-2TS+xeXenKfuG8
zId)XXe~ieT>EamQ@bi2#=|AUx?rIEBG3Smn>7$>ITPKoq!qGQ#4P=!IjeZ1O)93%7
z{iKJ~m=hJNGGxZ!cb6CpB;9M|ukIU%{V7Fdfxhc>u_>GGX>Ry01$0k4lf9qOd7J&u
z&ot6K${7wM9fyC{?p;pfZspY!cWRT`4aWhlUKM|
zLl`ex#E$z!{U$qLRsjC}Un`gEz;0U7b@wr7s^i~5`)kV*SEKvh{V5{4jOrV<-y@&i
ze}l8^sR`0aqP(yK#+~*$%LsvWmkdV+;aB#_)}}mVGQR})z<;};?FPLg!~3`CB!QP|
zWo%OvS}ZxmksHh
z>sqDt-@#wb?iMkDee&#wvWt#jdDxw~YqN&I1^KT3_8Bg`
z)spgCkj(HB1j`N{c-0Sok4Hxs)mQ0_R$o169J#5V^uci4b%pv>d6mrI3(zK0jF0wT
z`!VY=^|^})f3*_-r{VIA>e*oyxrh2f^tuC!sQ(;ZCi{)@&J2ydN9W(GS!Z#Yf6g_-
zc59_R=*crY1&-+Pvq&ep%Z?S&JmVc+pH|cU{`yqAg7&rRa_~2dPx`!M&w^iLawSO*
zSq25OpGk+7uaD7rOlPvx1n@38E)rELpm#8H?z9g44Kl87o9Mk2*&#^rOXapaSqi?>
z+P;_eI}o~AiT)eXj7{CCj%KWUe-#Cc54Bn~Q(jq0!~9g2hENG@C2+Y!>Up}CJxAlE
zXx`w5IJY(6{nxw+)CYa-or5Xv`%__+#AGGEe7YBBt~WoS?{_~tL
zhSI%i)(hT1b@1)9cA)%BiI1j`j$N+b^`klkKe{tVdaSs-qo4HYd#L&Le+i2BHf>qSdMPpLMh=RveLn*#gP_0oj)+a3+
z&n4aOT&woLiTWU;;sf0;e+90ZNq^GG-^Uzpg9rZ!y+QA&Q@`!p1#qq-qAeUems%C>
z3}#yfrELTo{UYa|BTgP4#~A7CpUv;;sQzg!zIM^n2bES#l!CIyER~Of&+}3Pv+2J0
z)z3adKR>FkYHsdMbqc98Ama)tV*^Y6DGe^FgHt$dP5FM@)a
zFD)q#3;rCU;H~u5i$?#|LI1BJ#M{#@<8=1Hq?<{R}DE&Yl6U@+4&=rH)_jfb1*9UYuD`Aqsd
zoqbe@^qkQvYES(oVS{rv>BCodGL!09lO`6IP5o7h&GrTte`mJv(*Ium@d(AW)HnQ|
zoT7ZXOr2Jt?>j17l5wYd@W_}&@6~VBA{Pn~_oID{+aB0L^QI2ePg6{OWhQ_(8}zG4
zheqsb%_F3nVzNpirbn+R@Rg+
zn+z{&*fJN)e}A%Iw{NeqsK)y14hA}3K#r%znmggIoHXg*4#pb|`G|oJGlFy^=^crd
z<(8xK4`>I8g6Fw@+f6$CI`H&?7QL^YJAV(Neqm$jPv_|i;rS2+E-hX!tOEZyPvPzq
z(p6tt$6c!DUW39TV8+R~8tNzB(^A(-humfS*7V5scZ|qgrryToL!!R%S*_sWp;9lv_bXkn=ah7|^M4Z;)qPO%8
zHcZUhH)8&);-OIijQ<~&MgS?3d_5R{w+a9N00000c$^)Vc|6tW7sqYct}RO#5*lL4
zzBQJyyD2K8v6bA=AY>_pNtEnmN|w51Ng``P_HH2|gA>{nLvUs`K~kUA8O$-iL}Wo4iW?s`aXM=K|im=
zasD+}hm)M^CkgwOy3@}{L4THim+QGXXtzH57mopqQx$(&KMS^rPgb@U@K#_qAqCi(
z-7rFce(fP)6C6j=IrPgR;4$91U?MOx_d?iSV5DsFL?O`Ia@6l5P=|P5Wj}DQ+hS}U
z(9cS@$rZ?7b7uE9SXVl!RHPPoY1m_I9k^2}pPwK4E2jhXm|*9PZysiU1^a1ZcdR~8
zIUsxfGPLt&-?4iD{I$d`#1@!JIP-T6^b;c^yn=wtBtF_lXs2gf@0tfr(hL@kLi>Ph
zcss_ArAG6b6i}t~*h6-pxP00h6JQOKIHv>mpwRhPwI1ShNbW(f9xXAlX0{L@gI%NXyuYnFBv7~If5r~+L1szOKqFZg}qkpX@RAyBV@9c;60G{mOCg9MD6_p9=U3ZcUL_ac^>QB6YiHebbg$lqo6oSznpxhXD
ztr}2Rzm+KnY%j3AR|WG2u6_`X0!9bxA1VMo^AxsyhwBR1RHXtX^vbATfmP&j&M272
z!7ts@3*=kb{pvSh59x*^o^z-|%kmh|O7!Kb6~;NgByJSug}w3)#d9^e8fxX^{U>e`
zcS3(}j?cCY)Y;d6v_)>Pt0^blb^~v!`Oe}!jeRTlfctJgc~u7E@$mTkz8qi?#|2Ry
z=zrnZZjHL}CEwKh1nuvB?o{JMy>>5AsvH#w&i?
zzrlDP*BYK1Vcya5(tF9k=ME`U%-`{;jkBop+(QYYIREf}REo_$yoWnC^%a2*cf8Nl
z!1#`+oTU?B|I3vb?gaM5kSE$sV8>Qi2BJ>f6wBw%qh1T#JkY<*hg*XfuZcBg`~=3;
zaSsu7opyIx<}nbPe?f@4fvh>V=wRaayD}c_wd!};(&xH``
zmf<|(VyAb17-0QS5Uacml+BD}!~GoIymrHQtWDcVB>;K09(|7l`ggcRVt#$7W3}W1
zW;gFH8$my+nkC)Y9FtZC?iUnL`~~YpeN(j#=2M%+L!8gJlI??e?3QJ<
z6YIx8HYmYu_rB3E*pdOv_qr>J=Wd|KJw-oPRsG2#z*JL1
z2FA~SRYY|i87=0oodo0l`*v0X{q$B7Ze9TP74e#4UU?jKb5ut^maN{g04cI2uW+B?
zJR9q)z{bxJI;a<$PRb_6>!3PK9_uTGl#qz)dls5xwxUjoUX%6#?Q};S7U5ny1j~dH
zuwF~W1*7hT3Z^P|0B`xNJ;gdx-e;1Ad8QkG%~d@NG|w#SMtg@QnPU;Si)U{)7uGKo
zHa_&zo{^Sw6x!Pz!qVh`t6PLy5vVWv#ja-Hn?U~qUw~%qdHI-^J(?f+u#Qk~Hj1L|
zJ+B*c
z=wIl+*J)W@2wW&peTVr@ZIu@>1zHHsUqh+K9B8gNVo(nKH=7D=&H-;6$?RhQ2isz~P(Ou_1Y$96s%K}b&_#u3gb`B31d(p?x~@FFV=}*YozEFVLrZ|{CxrI2PJ8V1cTmR
zb*=H^9M1Qb?cqcH32%2LmjLl5JbS4}S207*Wz|6n~>S?ltXh4J#Fz**GQwIDUxcWB?=-$Bd-=Dz&<57a|{-8wZM
z-|ujj`E*6_`@o7b-wfDXKJoI;fiASlam-f(Pn}}aTeNe%ttqs>(DRz{!F*4PpEbh#
z*HaR~cXCo%umtyaj2@xIVBLKn8HDFfvvL){c>k}R~Xa)0sotS?&yO+w?fPQ`Xop>)VxX%&Q4?mD(~?xdb+a!0v+gr7GhX^8;*xe5n}5v0E*F`!v263TK`E
z{sVZvwd4lw({P=8a~#I`Jz{=B-JcA;9fR-QsSN@Lo`;>RE`{%&Ps3Ffj7wFc@1P!x
z|Dt#JNgFU_dfz(w-=S~Rf%iTBc)k_&t(Cfe1nU~T!_Pehb&>pd+Yey+Jt0AE{7!3@
zk;nU>eAf@R2QJBL2w@(7cbR7`p?=Nir|hskh6G*C!hB7v9q~`cypE^UrQy4+wsZvR
zLGAFm7=C9k+9XQQ&-c<}y0~7mu1OHqhtkvWpYYsT%8|LauE^vgQaCW#nr^3sI#0jQ
zihft})QAlMGlnT$Pw~D_Y;!va93>818H0Gwk6Wtv0!h1nDp&!33v82gTY=YfH3Bd$
z{tZReVn7vU?k%iCG^dy5Jutub%!nMm<0(OZ_2YYBuy(Nl^&S?G(1iK(@m|SKZJ>Rf
zC5;U0rCc#GaRNp-oh$qXEOakdMxA(vC~jbVD)=YE0qdt=&6mpufPv%k3b=kv)`o)x
zaPj>Oe%wc{%I80StV^>|b}JZ{y9ZcJLxDzy#|GShtWq{YsQ2`-$Rl!CXUaoemoc7K
z*gUXaUY3on#P3Yr`E$aU2MPDr(lL)i<(29^f$Wyr|J4CS}iCr$%9m7|hb
z;eMmF%3FNFp7*SLi*@Oh$t`U`7%%jM$KwFl5iN-WSl4KOc~?n7z%P047+AM&`zR;;
zfOCu1nqOWAz93uXV4W%(U`}*k-A~%zC=KHueeRD#o$I|Hy@2sqAg2t3!gx7)NB=3{
z_K_>>_`N*U>nVocHG5;O$@&A!lGClrfy!OQdv60HnQRt#KAJ^5(a!r$i(Tw+tz96+&c=%Gj^TEm5HoNt2Yi
z4NAr?T|%O>&pv1GZ73n*Z~e7C=d;(@?_TeEp6~mN<)uAT_{PJLgC(Coj_?rCw8~=T
zFb{?8ho;7V@Gv|1jH3Q`e;&%HZMJLsdDyD9C3%45p*~>c{3MzOyfk|trG#tkCmznK{pzr5QGs%y>oIJx2OK3bKJ2d143))IWcUm^7I
zPYS1Qm#gUWDEP@fi*0P7@J2(*W$il(+XB`7?VeD;e1$&g0t$6ve=>VT(ka{@sxQqA
zqHyELTNB=LAr|~Qk*Z%SK-N;(V_#wgNVw}KwcJ^NVSauDZ;b$}8y}?1R3N`EFS!*w
zO#rjb$1ejZ8oM7%|AYF6#vT*xXInC891k;dNU)_bKYKyg;h%gg&z~isxtxWa_=DWc
z_Y9itX5K##!(c&(fBreS?F^1H!nr#)Gsu*0{OGfefpq=%hBd1gPz9nt5|=T^HPCBQ
zm1K~X{739RokH|wxDW136QUs?)AXa95MSS>J5?BPm=z@+uRg@$+&hQO&t9{T8Z4b}
zRLUYb^SPH~35zXa?8)9oEIxBFzy2W4ohL8etEeN-YqPALf7G(D$d9?K{+Pw5B1s!0
zu#l^<=6qFIl%J|QDVxT@XU^%`Z;>2sX!Uu+u!pXGu7R+RoaM=v^GDj$S!@^!L2=xA8
zxe^Q-W%VdQp%ATA0h-+lh45dnS#nXa0Q34@`JRgsKxC?Xy;Z&dI-yJ2rmYv^@zKCO
zx2FO$e`wm89q|%k@WHe9rpZFYwwUyKY$LjDm{o26S%72PHkO617oc%du#WR-8o4_!ZCjM*I1gy-dY@bpjvP}9<*Rw?jtY_3Lx#V7&e>B8qh7d=e?ydY&nL&e*W~zrgivzb+%E~)g
zoZs6oE=AsJvChJ}BAr9L-V9w0e-72rO3&VtbH(@Vp1c2V!lj*s6_I}A{ZU07jsZ0G
zEj@gmO%tHiC^pQqLWoBv>>7L;7$_&lIjp4GR=5EC~93t-%l26I(J
z^`a?6ONqNi1@fWn$dvUI(|G9jf8Alud?DU{l>3VdWH6CYWLufVqUA}ov7HMC?{NFt
zBGM1l?m^)Z6Ald_v$iS(vq-!Um;A|$!B*B*X)eXX>VqLibgU_KPTiVTyn&B}KZ7KK
z&eHHvkkTJg6k_a2&G+yNgp-#xr>E4gFbLRJ;1tTi&!h0ve^DF`o-LA>f9YUh?lsRe
zpp@`7py{Z+z7RV;tM=U`y5miB&w0Oy!s0HAOIG6)ny1XipYNklKc;VZZh;Ufu1}m_
zKW3oxu}ZgagvGU!JD1d_aB%5*RC(Bu!{uvgh6b)IKCDW6B4{GMaHMfMp!kWI5J9-2t)K_?Uk@)w;rGmQ7#S|u!#~Z#6@Nm+%ATRJS
z1y3_|VW%Yx>Y4v^KNBH_=UV0uD6?2mXSJxCd~e9`XVsb*!oh~%f7T~|v)J%G+&YXC
zB1WyTvALH<;;vA2+m(E*nZIs^A@TFN8$n(PgFJ9XLBn#56sG4ohPuWRKD-vYZJ8rP
zdEw_TM#or$YCrFrPUe8-yGR*#Cl0%YrN_QVut+;?bzoFd2&tOsj}(5Rk@KhBJ-tW@
zo1VThmL>i1s7e~Kf0m+fBHyrfs}3Kh3tVO|mJ=Xl%g=-DMhv`tR>hjK1OSpLavcr811tAX6ON~FOuvp$P(dHb>6
zTRjIcUl|$>LYcY=!nM)V!j1b#Zt*I&=!_+6T99Nq2!H)Cmxd74Cr0%UVUp>z2Fc?W6kXl0OS8f~cSV8#X5Rjg3yobYN#bsp~
z;)})i*S0s4e?Cpj{xzs9P9blts@rwKyS-bLw+r7S=mHC%&Zs^=INm9qN@X#?eqNCUxE6o=q
zPmP>u)ONR_@H61SVLgtAt5;R>l*qj&?l|7|#+$;;f1neyN3{76S+BCQ)R4weudY3v
z5dvtJ_Zu|56(X_E)zg2NLDOCXHD_|3;#*lNUB3$=Xh($
zc?w>A(#GMWZ<-m-HdcHd>bBOdP~A+Seqv;`8~L7-ttYm4Xw%T|&>VYwLV!(wcOb1>
zh;N}Lf0deq+iwp{{p=weicZmzDgGwJ6`dXOTYUt$k+ouY+5{h$2S@wvkUVIRHMuKM
zi9(Zk#o&7~U&;qmnV|I)4y=&3J`H=l}0g{57CppA4>_R
z4FN;&`Hq
zMq}PCLr_Sx>>qzbbnUwPR8n{x4TTb|w5wzeNxeu{4%y2gpY66GQ%LPJ)$HH2x
z>)?xA2JzFCd(1?HNdMThy~d5k0(BEzkzooxMjNNrlDV<#Rq?>@L=XP8J+}kN{p@e+
ze~YsxI(S>_khp;8tzgZVRx{C;w^QO~%8>YGLR5Js@sG<3?~%-DByXwLI4C@1Fllt*
zQz5DI!^$B3D>;wc=GU7dnkm@mrp7{y!mAs-=R=4NwikWB{o73njNhcsKSb9bd_HFP
zl5>>D{Hyhl@GJ5U7-x~36??BOe}L3cf8U!vFhc4a;Vt}K^cn-hhM@UBN#9I?aBdXyK^ILwFr$*~eIo&aj;$LwTk
z7`#*qZ)P+(q}_}Xyd`zMnqIn7^^Jwn1^G(GlfewzLkkOi2`5w4=oXTHl2WPFe_lk_
zrFD5n?~r~?R!Hcz5T7P3u{|~FM?uT)@n8hShiAinqvmb_!q~e?YCQ}TPB5JZNX{%<
zH9P7NnS-x97Z<1IvADSNEli2ttZyo{nKjX{b#A-WL~_CT?BAWFiJo&RCmY?#D$x<`
zK%4WD
zuI+?B((jeu{#YkOw*BGC?0Gai^2F*Fhf}y{;^&r5^!0y@H~?e{ARs(1L2z(oZe(R-
TL}7GcJ_4UaIrC9p

delta 12123
zcmV-hFQm}plmXzB0kEY50yQ_24FoKc(*rpHFp~`gWC1a=as=2Ge|QW4000000C=1g
zmw7x^O%%t+@{~%RQpu9Yo_Z5)QkfAV*^;GHN+m@I6_R8tYqVG*@?@u1gpj?G?0b?H
zQY1x-hxBUhd-`iW_uiQ~Gw1xy_naHg!72NXQ@nX+!KSfO6o*z6=@DSh&1nmxrUtZ%*&g#Fso
zPqg1VM)B&6gtbA8!RPsm&s4J*yr}S3rtBoe3&z+lTt_K3D&;vE^q%4?-*Y_{zoPiL
zixtkJlFxBEHd&fuuChVsFN4DMI_{h&Drk
z$MY(Ft=5R_e^(ghKW~M|*EK`oc@~Jco{p^1HAf_0vRCCFGelBxljJ!gM6T@~;*HWq
zBvh=ZH&hLg%+`91QZYnk0!Ed>$`H}A%Ip91FCxzsSxuEGi2S=*gs&tS5ofIu?k*1O
zONWIJynh^Y>Z@EDBEPH36&Ge8^6^*p-m!E<*kxtYfA}sU-@;t1Y+?}^_gYz?2In3~
zYPC0PMuc0hFBe^ph;E~F)A&JaNrZy*L=%
z1PU|Wf5Mo1XsWXU)~&X6UNQso?s)Rq*gi%ifXD3p-Eu@i2F`CdT!P5PkBqcNF(M)}
z;q|A%AH%h`1YSQwq(0}EdP5c>4lIso4Se7EVYT*J_@2c!IqL!IwmWEw|GJNe*|9`s
z4%~m=aXTruC`3lGyndd&h)B!yd}MhVkv-kKe*u{eh$Nmhx?cuyG}`20-W?408ra0k
zJ%z|28FAbBf$-i!NzE`6k%a;=3j6_xNNq1;kHC8Vld54&mWXU~xi9Cfj)->nQ8z&e
zMC`qks-5`}N$yeX>z!b5L$$ZsAP*v2*(wo6QivRR+0wRhHpGDw$c!r@GEx8U;yJ3|
zfB%BnKb|Te@=ChsR_lC3Bx`o^4fcQ@AMt2LMlyJ`)RvQE!r)gShxVq+F!)}|F43=S
z2CvQSQ%{rxUGJ`(G~LT!s*pU)oy}mKWqr=;jTme%Hcx-%7=uNvBL-GmFnCAb=Z4KI
z8QjVV=I#1M@%2&1q4F|{_q!Me2t`vIf0FDSZ*qy^yyDg7h36@5U0|s?$=5qefAQn=
zq)whRidkD1h^-i^p@d4p5CAWm~Q
zH*7CwBT`c{yUOYbBJJ(Hl^&o!zsS@&d$2dOaF6)2I}}STvRM}YiQ*R?*~{bf89YnE
z$*QH8!6|l2CF4X85!Ba|k28UMe>0jglktc694icZ3;NSuw716z_EpF--}?ggy`eB<
zyc_gy&^DzvdJ%ZkAhYr^@ToyD!=y`?!3H0LIVb!W{AsW=L1YFvl5Tr7WeXywR2Qu&
za|b&UC&gAkUO!?>nx#}BVl$py>Z=FmLBKNtc6g?Xd
znOhMz?f`ikamjD}jR%MruW7M30{U-|dUo9OH1J&Xn~m~4iZeSC51WfHSZ0TdcX}d&
z^>j3R+C>qO-%&Rya2OFaf92G%nRwt>nFas$8bp@Yr=j)07vZ~_Q`!}XoM#t^Yup4*
z{gyw_3|xH168e4=>H_D{nXMOZQY^AedR;69y@~jDXkBCQve1(I3F5#3ng5P>oPcv(
z?(!}IzO_j;>Iegm-VNJ8^k$@(7P4RPT6qa7LNgho>yEe~8$g@e%C--R%))
zx}rc|b$N5cwhvOAt<C0e#uO0Gi*uOf+QfQwG+;`D}#DZKznzk1&F7k2n-m>b3R5z2#4n<@Sf6dJq1
zmBEjS)@!K)2kac)e_nR>K&0`Cy>t|CN{6N5nF{e_2Hf7Z+u0b9#+a_OnJ5hXNz47krwG`W~QWjNr`)glEvPcGzhp$DWl|biX
zH#>H?z@{LY#8Ff$qK=sRb3
zUGN3of0~?F20G9!)if{J1Nr!?KDi$FaPED_i(1I9=m!^X9r2=AUqNz#(GG
z#dmakEAXT^zU11AJmAlwRmu^gDWD&{U21E92jT`3oJrt7Y`(tLW$;T>%e@z;AwO|z
zn1VaxWtZU^uT0SQaqM3%4LVUD&MzypX7F=9-jFdke|HkQTss6fHT`bSTP@Cd~!lg2KYF>sH^wwzVKRi%cZ8guC90w*-0D-wXGR|7Q~CnzqL
zSi3Lx6oc!vjO-$zZaJ_2l;fa@NV9^_+jOvVfAgk6xx?T4tlru;>#zSF`8EyQWAOTs
zY3I}x3>Nh7A1E$>{kA6;B|%nO0NfDc9XpjBK=JrU
z=iEBzU*|=4TYm_Le)!;zqsO@vw`U2)e*}jyIHRLUg%@=2O?EC{0mK14&d%)v&X#&M
z{F=^z{z1+CqJtyU3Ee=wh9#hXv$@MNJRsls8QI}C6!$I`{IdzT&STM;n*#c7lbR^z
zCPJNUUvi;Sl);>eG|Syk@3hP`^Iw9W?xr}zQRtT@_n6z{g04oQ9JbzqzUO3yLKM#(>^}Djipa
z-Jm$k{JW1jQJ>8n!d@PF~=GJ}0+juo~=Ku&m9Ce}Vfb&MSA2
z1iyvmMQUUKFRvGA2CoDisM;<{C;`1yS*xW@gRY+j+5`@Qj(FFNK2Qe^9CkC;3xv^N
zMe%DZh)-b7&zd&qALhx;@4vJY>buE8w+!IV&ETMqB9{?arByq11nM^bN*@jj`rNtg
z&24u;*Gz_AQtxs^ZdvO|8K1C(|B&C!uc9(2{J)Lpzw>v@V8G_xaNSJ@u@R1^P|k?YYu(h
zjrFn*OE}-ayqIBQRm9yIa#qn4bbc`K=dWOwh11vN;C1Z+sRqn5Dvy8LQ3iW`<*9?)
z!TZ92NB_ci!t4%Muwj3)^eFWNbqC8$vOv3br*{v)q4I-ooIo}I%f<^p9Y)D3HsYN)
z;>Z^Q&Y$@HLj+7wHO^g#@vup)b~=~Z_NP`7h%bD@)6xe1f1#KjRJGjj
zGg!=$J6!{IwyTdAf&EL{qOO9g2WRbeVV!H!Mz;u1ORRCo9Xvm8FMSVmE~xlLgYqFH
zpuPpvSNyR0E&QFbm)38D?ZX)8R)t+Jb5g(zlphu}r~-#9U3P!UWBj;M__zVs=#ZI1
zapFgv73VQ7{w`jop3c*{b;lC;O%lfMtN}+uoxZdpFHUl{cEPa61~x8>p*l(J3TJ_)
zY?e+e#krDMUkU~mUwU0YdF!3i2mrJ1vDhoX_M$YwyNEB+y5TM5WvXxGS|0e%CtoR;
zgifjq%hI@E=YD?=f0hCUj3s1v5R=xv*$j&H?3I)Q#gz|n6hObXGlry-WX_?(bpB+E
zcJ5?w?_lpk(#>PtTFI3d_ikM39*n#iwjG@hfPc4!$@UR&GU}S_1F+S-vUmxoKJ)EH
zIp(#kdFV#iA=*X(7^1lM=#i}oHHUo`n-Ss)X)fYxFMvb27c4%eoFh9
z-rS>1x_I)XSyU6$F4c%4U2HX#&I|>8js8uheVmum&T9mv1FSDAVO^pmPvR!9#;elf
zC+Sl+Q+_Ax@Hu-wK2YGD(v}+dO{=&}G+`fR4yPQ1J*J*#x*qn4CSKhzkgxJ|4(07;
zdhKR2(7b=MvD6MUD-FFudU1Mn^2klBOVj8d8mDs`OIJDLyr~5#3opRd?OIsy1M`@=
zKkAl(&9BFjbis_4`(H!po(}Px-%C3B(mm6M`MX@Crpv)SAEc}DDPO??+nvExlIcsr
z!S4;+)^8AheI?tR^lu#^zrz^o3#WD1X`sD`-b8-_-P1^Aeab_q@anO6tV`-GZwUgw
z42^644d!aU?f4J8)fTFDAJqI!zWh7Y&*^gmFX=T-%iEn|CA0ra%=IWciLhC9;Mm~a$8FS}p
zKevPHEQl&KnKGmBdq@tPCf#oc{Vr8~p1$jPp*e@{X~EfFi|L*=-|&7y=WX;q
zH~kV6%pE>WIu8D>^Fa{fuUVC~6xXOkT#|qGUr{M>m+C82G4k#!*uXQG4cQHuoc}&xT<5tX?<}}uz}!b1{>AyR!E}%PUaBTgKaIG^
zXFr7T0`9^1Phdq(z>E;-=D%t|2KA!_J$D|G9vJ>zw7<@5iEVV>yFP_SSHPbdYT$of
z2ri2loEQTIk~fM<(*E|i$cTV-2}WNB;9uvHqeFShH2fCe1OK(A_G|Qx3>$1>q>xVP
zmF)T9|8jK!FU1#WQl6wb+dQc%yAB=`R&tF))FAvB8i_`9
zPWd)zy*Ka|4ZdA!3fm{*eR>?om*#(+PxZSpFIq_X(U{)6XpZWezdC!E?&-#!|L-$g
zbb}S;x8MfTQ<&oH-}j=Q^x)nVO7&HJrQO#^?^L8>Kj}l&dB;WSS7kLa1JA$%S>gh;
z|N4*F52??^B|T~*_HK8bqI!0jNA9M+a87<7H}xOZV!0{GyGcm&T{^#kR)c@#8L(-3
zkktnNJ`euk32;Dni9P8=XR&Vy&HGi&^V2Gj{peHuGEiYhV$c-EIX=$^XTZ6bd@0hy
zWBuZp&!j`E4qrOYf$8jRLU1;-YQhwst+c*q5$RQX
z%O2WqcgQ*w`fs>sVs?b;c))+YP+b}0KidwrP+o0ShL=!XUW7>MsF2Pi)6UVoeD8I6
zCCwXnAJ38n?{zFpqCV*7=n_P6U!4lACMKx(71F&3x!O`h-+%MO)|<{%q_Rqg;%8;M
zh0wj~&<#?cI{0>5J5hedB}P(7$F^7R_)(os=G~qpJw8uV>nDAlvTc94HBI%uu5Uql
z)Ge_TqH}H09B-m?h##oUrF>PqmA*-FkEOFSU%-~o4A-XfT>Z)0WkdaWrgE1oXe|7^
zk=}Df&501oCnJ{UnLb#?_}$Th`a!HDoBI7H;o=q4&jn)A#EHTuG(sr;vcP)d7V3YC
zIr*e}jn&)kAEtMOS@nOO?w2xe-MBy4@t3djEpY$eA=l^~b?vvCjRR-8!rRY+XVPlU
zx`0_$f$3}MzWYVaJw==c0$ih{vybcEG*JCBT8}$KgTX4x#>zoyUn^BF(nUeq=^Q$T
zU*pUZ^z#5c4a=M0%pc1m6sXUck9m_mQf3F!I>8A~*NfCY5C4C5E05~Jf6ym|^dc;*
z_1ud3&b&XD7?_aJcEK3)&gAJmDn&elP8rWXz&!#+qDi29jr~P7DA;)?Y!Y1L=08C1
z=w|OIAJT=^k)BHpf}xY)sB(hL5jszvi9NknzO_r^N)Y!S$2!(-%Kx016ZO-ZrUwj@z#C0^
zHKaq+!CI}Oq_bD`b`HpwkF)$jb1*-8SZW*TzARgi?wLz~csunM+px(Q`me~+Dm`dT
z@4ii@r!{|UsW^+D_ORQx)L7PHeWH^-;~O~r$V6)^{N>}O{ae6f;~^h$@L^^kLyGQG
zv>cxjc&b{}0
z@44rE&biNX_5NL@M@uFR2Y;@q=Hi?*lr#;VRN|O4)Ga9(eMaNIEw6nOMV$G6D;RuZ%{8i(t3o&p0gVI@|dV%rdDdLKw%$xRf{%KR!5zr<%$4`hT)G^?m
zPxB%Y`vvA`AAV+oR(OM2hCq-5rll(zvD19^OsZPrR^loNRN?
*eoNp^%
zEU_h}s*jK6HD8GsVckn@tJPM-ohvH*Er=;;Cr_^<2Fb>}%OZN1^?SV~YFpe^*+^XH
zG#!>s^f%Y3btH~S5OCrV3Y2nf)8)|jvE6xpeNLkrtD$S$gG
z4bvmaA4~asj>d&j5^e7g*W}usuz{a^M}99uoP&eh{fO*n{_eLlK9g{{?kZ-=PT1j!TlDp%tOKjS8_KF+U
zsq5}$z@Mbww&@{18Md2$f{8ozwYUv&e#>~b57D=(vrC2eeOP4_{-5!B&%1{B&j~)3
zJMAy1@m=dS^0D$E-w5r$ZDqGU_T8!fEoza*v!eIkbtYOc7XrPBRf0=f$B20bI`<`M
zU9y8`_A=t!j!v0Q8u#T*>8U05&ziHQXuoJXNniMpGNt~&ofxlw7@Va*WYzE)tR?FW`|!gDI?u6+?8ozP*cmQsqJ%m;eJC4salmn>`Y8H^^@N)sjcfeUadUy}k*$|R
zMA4VRoL4y$Rr~adwh=|Yj~)+kav#it|tk^H20gLxOe$zcp>~;RP|=AA|@H@
zPauAO2Ue-hgO|m;wW4X=(0d;Z`01i1QuhV=&0b-Qx^ms&v`-!Vlx_Ze39(Mr=qb+W
zO}Dl@kNpRNw4s;X%{2>%*KYN02KwtG=g27R?~`SeR8QkKvKyS&6Zh%#@13T+HVfqo
zN1|U#h5JMI0+}C*ef?w{qnAB<)(Ed
zDxCcAbI(Lv>JIexy{F^kpqC~5e+Ch!M^ClX5g+?{Z~cd8_$)mGb=j%;mLGkj>sqxa
zbbt7=ArJ01ZEJiQeTa{TRg8XmR@_-r2l-4Ju|ZtVybddcu2}Xn5n6cPqFD}DcYnoy
z*sH5Ff4$kPGz+@URegba@2+R8GA3FHeM%D{a(w+T(@OJ&me2B`kBD-PD)`sN(DFDy
zyi?qFwV4o&HTyOd(0o*Fmi2LBiRWee-{ReQ=%?
zf%iMa;d8tq`E4<4&;N;RUe5@|Bh*QE@et~DqnmaP^mf+1(#Dv^|I&5;rwJ+Bsv=HUI{bo^rCk;>EH>3VyTG{!Wh(kw4GxG98Za4t(eIdTq@ip1*
zK`z_jkAdl@0r;z*uYVN33ku(*CZaCQ{GHa~J$%^4-dIE97g*a&d6CD}?jJMh{&TrY
za)yX^$A?GIkCtt5uZ6DSpS&t&BVT+=9gr`@HTy2jAWn=tF~sq?TFPO6ychC5cE4{X
z`q$@P!8wJOc@~Cfo#(x&Ug&R!%VvnoTf#y-^gHcN-5Lh+
z@#v#o;4ah^LqiyK%rZ%TW<$RwqlawKAN~B!C8J)W%KN&siHTHJ&gR-7^
zG5pT>(kPJ&Kl`Nzb+BKxj*%bwL-yf_Uff$tIXDgbN{rld4kTW)9JSqmIK-cMe
zWn!<0H+yPY3W*X2mN^|H4qChn9i(`F9WqlnN_3O|reIFIZWFD4Q%?-i(Kv>cnJnKC%-K9VoI`KH6_yheZ>rBF4^iRpMf6i?s`VBD@u>Zs4AA3y^hu2pG
zaLz9!p8ujRO@`Qi&LS?iwsIH;KyUiHUpnD^ld=|u-s1;@x67dp3j!R!(tbsuoUZ7X
zVX_yB@jFxS^KlW>LFnzdc+{~EL#fgYbzrLXZv}A`gL4JmA;zZ&i!ta-IV6UI@*B3H
z;K@<4Kf4t_M_+0%`cq2?`#s=w-AeZPCs8la*Lu>=I|~zkd(&S`pl^qJDo4)Hy?M$_
z&dn2RnP#czQ$;VQ-ZkOg(HpC!Y5m=S&T!~_!|VQ&h|d%=_GJLAmm6*BJVX@l3tfue
z%VX_sV)$L7a_OS1H!(LR-m(Do(UP<7I`P~Trz!5&WqQnO1+5>?OfQ!qMkh8tdleoB$`C3l4b*KQ6b+K`)_t0#l$6pW!Vg7BiAqHpXrPXa7uQgzqybS8Bjlnv5oE}qHrKr*+mrdb~f0X=oNgjp5f!d<$Arx-z
ze`P6HB!0*=zW>_^UiP7?k=Tf$U!?c)52|5ET(qB7UR8+H2>?=|3T^x*3MpqT>|95bB^xo#(
z>-ij{Wj+?!X>s@*Zu&-bC5P3gzYJ{L&EZ_x4XMyCEYA5z2P%_&3PU~_kH~we=l9b6
zb;*7LotWf4MHjBGSZVgUu5Lj>c05mtAAg7(>BgwC>%nUgPqdRLYIxfl^f#)?`LYUj?`{64^c=LCtp6f{=E?tQV
z^c$k#Ij((Al1C$T@l)DtCyj~oG(C+HXTkYH{H2MSX%RNz}ey8;C-^GdHERSK56iz%A3be|2HR
zcXTqF6r{ze2uo$qjb=%j$;`hA+3Kf2YryAeMqlClp0J+o)BwmWmh6bJ=g|M@~
z*i%zzWG7T>Pors!RGwMv%?UAXe{U7GZn7BuPqvr;Po2RhvjwT%$}Gb3H18L;v3Rh(
zdy*{q@4Dr-4rM$J`;8~-&)>tL>Zn@98*;CdF25N&|D|wmb52=w0F8o}+|~_&LhPBl
z?*f}HLYrA!gio0m&yG0N?XF{BkQDE{`~i!=<14hEJ8|I4l-MK@y-2rCe;d#W<}hc~
zj);+;gky9U?IrnhGwBHN&7u%pd1Wf^1WoRD!m$P_r<@25YNS%7FYfdme
zN_<#$l6y=r4ebrg{f>Mge@X&+ou?~`!EaPLzy&iHJ)P@Veun7gagDjtMh-o7P
zAJg0W#QQBdw1iDvs}f3dbTK~ZlQjddj%qU~0Sxzs?bmal@L~LpME)uoGy6lNL(+r@
zR*^LsP!(f(+1uX8iwyp`WWh^$%R)bJN485i2QTlOdJ{1`zi4$85biog+o!*oLwNjVd&z!;LdCev$kQD{)c!UxJvU2?tL~3ot4kRe
zHdgBA46(>OwppV#g@etvlJb2UINZ3QZMwppMYCc0V^M<`Wn8PCkD3VU=Nda+l%$ZJ
zoATo8D2b!z{Vnnxe}%Npd66G>(^Mc|liAjGNxI$8)|j7(9i?SLboin{PRgPxyNs
zbycQd3h`GD+qKsVNq>0TbA>7GKJK-}Wa3iR6*G-W~x`{j)QJOG&-0
zb;NCULbO)wnHXS6d}D@PR*yOhllS&>+R1qn2ft{&jpZO$e;4}s@xLrAdm|kpI5Cp6
z>+733g-F{HuH$GxW0vCb$)-flMz=zI6Z?oR%|Zs1>M6)y-w^JeAjG6z_jKWKd{FN9S7{c7C+&9)%ZQ=;c4CRjA3~h}pVK|#XbQ$p
zs>~Hgeq1Y0e+)Uu5`MBw`D^uP@Uk~f`%_7Tgf;zpTg(`E?>3CHVo4wC+xAI`a9(v$
z!K9pHA?JAAX%g|(hVO@)9LWAz-%k~WETZ9@QZwzwbPD?ZF7qpJnpqnDi@%vT|L70xwp)MTO4I@;ty^l9+5SbxNye}i{y#QBaDtfxa{cp(0B#^e6BIufT|KVWu}D)F7)b9I%5
zX{h%>%8$$q9ZG={j7Z$fQ{TNW=rN#pJ+c}key-a{>tf{y@QY$q#7nur?;QMV`Ta*aP
zin~`dyb>e1%iU+sAcHU4S7^JE`#ipLM)TWJF^1nC9d)ZGe&(EaFgJ}xyVHg{8W$++
z=#n#!BzaRm?YhpM7NBY^f3cPYg~rh#BM)*O-sdB0ymf`JXkL^KUF7KCRoc
z#_#4$77bH2CUtCP(ELuX+PqeT`$As%f0zZC_u^AK{vM+6qs%8*mFT74jGv!CxOitS
z*kTF_v3A`fC4_5lzvL5<@j}ciT$p~1^r5NGcIYhUpEzwFO{(3oE%m~Olw(#5D
zXV)1dPgL))mK5V+W5cGm9zx94vDB9wq_ER$_4qQQjh*+@5V6%8(#^L+7ot$|D9tW*0x@
zkT^p76$#(TeH1OK|Bh;+;G~}#2Pxu*w>mF`5e`=7_U8R@oA4Gew)+L)x@LFd*-mng
z%GeJJ9})jL^cT#}kU1;jUUOCte~F{Io!2u&;`j%`PE>ftrJ3DI7DS)*>t1q9_$znxZna*q)mL-{C^@F57$9@<33Pe!g9deR?%b-9z
zvWZ#1A?9|B=oN`mZ*}QzH#W-_>WH5&wl*~g34Me+WG7p(arcyN0U?_=XNT$rm93eBADr6`{A{r${
zG#>{IN~!m|_U}0E`+Cp!y59Sb_xZg4JkR;ZxvaI+E8%1g_aNd6%!p*tirT2A@FT3L~b)V`ETs{J4GzxfQj`z?|)ynNz$aMj0yjjyo_YLZjNi6
z9lY0jI`Z*$M4@!Nv)$Kvukl&#?X$*l#S~~^lrsO)?!VS|a&Z;vEB#A-vaIK9$eTkw
zX}bzC8Z^(|E^dx89kkhheV&nRWJb2Y3rU;LS{_^I+@*#|>r3ZXUZ}sq`~S7J%f2ov
z;wGJ9|2@Ad!aPyCYrP#k|LunvAT-7Yp(OM~(jD@f^q;J(AUGkgGEyop@vHxCL{v|K
z#tbSaC&vvaOYF(ZN+v+&%X89`{?}0=Cgqm>UmR6sgucmhOSHb1V*Y4}lH(ToBqOuN
z*+XhT#`eGd^}pgD(tS59O|WME52Lgiq`jabss5MOf}*)V{_Fp;GXHe4c={ThQklOht{ZH*n$1An>Pd%w!!KsUxblg%sX}_fW
zcS*l~N~lyWwIh6KF;kP2OY2MfAuTVHl<$|&ZxSlNC^4&+@KSvarewsd9dA&~M3QoH
zm@<${=jpRVen3LIOUge=_|p=q&nHmC5_!i(EJLD~BH`Qf2|;@u1pDQZdPWkxnG#;`
zMq>7mcU;UqNbFvh=-uNVRKz|=^z+`(KSXS%M1QM9zlaZ1%>MEZAYzlSQ)pbve+c>(
zJPM21W!@1H+tNuW*9oEqM9kuzpjR!iyI7(>URRLM4yPR;VrSY4^)5^7_LbP#HAK*p
zkERZZm}Rhp*Aw(*LIi&EB!TbB0k4=@2MWC1ETKH#Md16d68KYL^yNj&%T%yCL!AaH
zW?LT9M2pzc5TXBfJCQ`(H4?f9hgohv>>;V>Xc{QPb
zk1o&(6SKba)^Ymb`zu~P&rqH7jhUI8KRHpC^Tq=f;0ub_vXQx5uB)?|)9Lg4d3mcZ
zZDCl%Qk;u9-~FvQr#5|UIPEuX5T}Z@)3Kys=UxjYNW@;MyyrCdumdljzrT{VN$ESCRof?*tF&PCx<=asL!b#oM%C0m_RYBpEUt1
z0`op1ncE+ck!wcsj~@?XqC`wJ^)PmXm_3?qj|meo*D-VO<;ARa&G=VDFKw@{Ao-jn
zj#y!07FKGJO8ka(I|mXU6KP>VRAsggK9HDody_&W%VJw$F8KoFR@|9l}
z6c9BTVcUhuYi|_3CVp1=vQVPVU8=AZ#Vq^#Gav~uv$PDg&KU+r
zvRnQ!ocBBOO&XV%*^Bx5`T9$P+Znq!9au@sbS7wFLm*C=#_{#HyI>|?Z@uEI`8=Jr
zd}+n4U#?wvw*
zf$Xdq@s#(c@%I-_$9=(afnJ~E_xZjGIY16g%w}&Qk1k@d5#;bh?0&!$F8`{6ZlT=g
zWDb}2b^5~lnV_c4-JbI<6S3T&Zd%Q!{wUr%odeoeek@JBwWP=o&S
zG?+vE!*WHOTKfrd?Ph^Kibu03_xBZePgN`%)C=*X$R}c{ONH_WF+zFRZh=4gMHr9e
zWh^`B&y2?OLcc8y7ktqn3=6bwUkLA~!wM9moDCG}ElCypWWr!!9#fYJB?4{zlR6}3f1U}v-AF;tnZg9*o1F!DPYKVr1LO-xOm;{Oe4&JY
zM_~r?-4Zz;9?Fv><+c)eIDJ;gU(siUzJ)|CB%G8``EroZcnZtV*Okm$CkY+@
zOqiGZ^x0sShfPpVNq_c8>PJZI^5Me{9~ShuS@66&pfe=>!Di>}@#%-0e-7wNd3{Ma
zhXs&h^K*Uvd7;lg6XZhrK+L*I%ITm_{gmP|ICox89$uZ~6D0g-30+NM$oolnPGDc!
z4{85tb<_Mv#{nn8d0rj${}Yvdr~gzgl}mA~bRM{Qv?nc>(p<@Ul=g>DKidB%uH~Nx
z^`-MB?KfZkkW0rUl}qjalk(=EFQxx1m-7FdFRA>WRJtyt?MkV1ywd(Y;sOzq&Wp5M
zYEQZj{3LqPb?{HUN=bPjuPA05Zel#rd6updN|?|(aFon1d2`}FO6p73)i4R~DA^xt
zC3;dkFWtAdrFz`3n2nOyy&}Gb&{;ZG0UT;bH7@#!Lo2@U<)<6fu|CIQv)dK|n7@uJH=~QX}
zale2DCSr|z;zTT&;3w+c?kmWBj#9&jKNf;K@1-FBOyLrGS0#Sb5;6McN;|HSF8!{lgiUOGpfTf{S
zdP|1~ao(nJI;ZuM&@Ac?lc5zVVx6*1aC!P|L%0$#JDxX}+o`l!%;gufHgLU4f{iG*
z8s9dL>TUdS6f0NE6f~7Eks`LU(hL&`UgjRhagSa(S`z>ImI5Xo=h4ycc>Tp@3LMXv
zzc<0k5i>K@K~_ZbZ(l~}5V4@qpW#YGtexxEOro>9%-%}nUGt2siRwL?%W?LAhXK4F
zN-Z|FB%kr@$SBgkF28v@;n08yEiuIFZl8daEM~t7-lz~i^vSn{#J_M-?Jftp@xrO4
z#2Z$KvkCtt##;>`ygt;vG>hcX(?5SFoI5u1$5E2I?+^Y$_T4PbomB)c@7PJJ1pIrg
zO=)SMrp4Ybsos&ud`H6f{_9m72;c7=XMVU5d}>cc`=6l0ngT~tdyh4Ywvye3gS!Tj
z-JN3|=hYGQ+5d;+9Yd@BkiAVUA&aSe+JM%^G>+S%ox!akZ;o7t6Qh{DZ_3sndD4BE
zEA^l;*$!>hAouAoM(!ziQ^%6Iw?NyoS>9hkdyjbK(--A7aUCwlfmayP6lwrkQyBbn
zAmoD|s>xJ?54pT;B;kA6=B0he{+Onv*kHizYHBXj|8iwpxF|8}e4za}n$P0d3I@by
zUvwTw^zJ!_(L|HZPpqYWy3~&uMfF!D-!Gy0IV-pI49&}{L@NMi5o>ez?fQqH4#oPL
zvq9U-B=B|lqEpfGR;V|-woQL($K(FI{z{+=({rC{g1YGVKj;Gby29#4H_-V$b=oxk
z#JHuysGrfATmBM#r#KOaP0W^bcEE-ev2UZUe<%9LtjkjBr&_SuHEZJ=s{z`&3necfr?nsfNQ!We|bH%j{3i0-ho(>uUIyy1Ib;)
z)x6#AJ5^L@KU`d%ypZ~RaKP{7!+bjk
z{cboQ*!_f4HOVJM2=)w*4dZt1lnV8p-h#VCxydms1K7VeP^cf~x{J$Go3MQ#FT0QB
z1HH{VggB`FRM6kR0VVc1g(>LO)AWIQP&kDCWXXBLOmcp?8!5=m_<;g`nYstPG7hw`
z&n$)d1D6W4|0|)qQDXnQ1c&S2B_IR;+)kK}m74|q+@*s4JxM?A`4h4ev;GqOx++2M
zqr~3eN5cG^mGB-^4E;7_2KMa6X}B0r?Aw9{*e@4<<3>LnPzg&w(cL
z8xnc8MBkpe4gHDOZKVH7g7bDt<}F2HzcqbEl&{Pa`Z-Rb|5+kWlHjP}$ZC+@P>Fr%
zIf}j>$$2->9&Z-ZTOy}jLH&^8fBLHbsxO@f1VJvBo*QVDP(513q>rG+d3VXUI4p#m
zb|vXa$2~yOPXmeGNJ+Vu#7>Sx{?B%JG3-hEi6k1gD?Nuu%cWF$ekuMR^!Rdu-GP$!
zrQ@e;0ri`9HH~k*gpZVr^Sy-UpBd%S`HGo_o+
zulJJmHc4XVH*Is+xuphohw|9o0{{Dsp#MNu;OEa4`1>nqWFpq@k-#t25X$}E3cMk0
zL)d$4C(!Qj=ETqB9|(C;$vmW53vwAm9kQpnM&K2?2>kN|fmbmV{M`~GfzG@t(9HvB
zAi!B81^$f_IReD_#zCB}-Y=9pOZ@Bx?tswmQ3CGv{7M^4%$gi%$B5X2U4s7BpMt&m
z)q*{fz}FZsaL{S`(8ycc(aHpV`Xlhx4SAejytkH@U!Et}pWjF;27cq}XI?(P^E%Fl
zZJmOZh4Z!oT{r+oDbNWLd5CU8z2W`7^Y)Hi>Wif$Vj2^Jc=`4%NSA_lZu^$=zt1Q0
z_@vU$lgF!B1tM;@!@C!}eD%J5Tz}ZDDV+bBLU--Jj~=VAQjoVjLpxK%b~qLDa&bFT
zE&9dKk9F(-v2cw{jfeQGUMnwafTHO0|e#BL4J
zfy)syjjS&GoX~W|8V5HqJ9{Chobc`6Y5U<}fV;CI3rK%h^GqZFffxTwawR_HNd^+2
z*dP5K;zNnq`Hpvuh`K~j3P;2;1M1DFe1)Fl4Z^`c)mB)N-TbSsT}kg%a>jJh3r-p^
zh4c*W=jEIhz<3*prbN0(eKMtQ=8_`4OLo+DLU3D1SUsxT*f
zSO3k=l<@qL6pN<=mT+$uX({4fo;6>ht8}2q^
zU$^n@u>-tq{SMH5l@F?sEhD!O!ia;TZ`kClm2uCFD_?wnP!$
zoN4>xTovT`>zcpc1HJk*`{qH&=cnIU<_x+#et7Q&$a5nc!;?X?^ZLIGfxI}tS<46Q
zf7>zg<%_=}_N6|#xd`R0JL>XreLD23KIA-q3p%|0)pPRb@5=-CfTLCsIAO>+f3
ze(#%^rwaIpmaF3w#Vm8;tq^a>1Jw^4pA7r?z
z*H3n@3$2eK%Z5e~eQWdWY7*q{XSbeC>nNwU=3m0oGfoVgRR{UgHnG3xJG>co_AA-h
z-Mv!Tt8h*#Ji9rD`Wf^JFe
zkTtnQ`_xl)fD7pdyQXsgFtd9iPNK-;$80{@7Ifjwn)Os}GW7CZ(vMpHcs23A%GTSg
z1K&|RdmYU~Y0uJa^j#Ce2iKGR_2a65yhY5){tl3=m}Mx{z>y>GGh_)uKk_LDPjY{$
zGjk`W%bj=SlYI0ob?#radiCOd^wzYM+<&^xD&hW9Oa3IMk;;pBK4x?;8Sa03n_6-^
zeLD^1{xSc+@=4TQee!AU&jK#};`UyeHgLaO*P;Q)BVx}NKX^y}CD-8N3F5a8xX%5X
zN!V6CZ<~+4h$H#qMJIVXeH{o80I#>kF&6UtOG12e*A?Q_rO`rOsHCe9hxX~g-9mqx
zJ_U9WGyf@^2NJud@&2To_TciQOAk1`^_-BPn8hmI;=InwUA(_@X6@tSy3yB*%a2$s
z;k4q%8?L``wPgy~*WWgS&)=X{NnF4C<>Ls%xR;Ef8wTW8Ckb&f?VS*xPtq_@ejQ(#
z$`w2?Oz^r_I&ePxlQ6E}N%(A#Tk-@c@_F_u6w>T#sb{
z{ns@8Xz#5A5B=ak6mb;MkmRX;P=lTkb`Q~x5}egmg5&y^3G$82m>%d?UK8qd4;SQq
z{2>X>c@q3rH-P{c)`OkUe*6bPKjE`5Z)eQ}K9bS{(2Kf=@DA$qMzH_eMzHJHOQ_G6
z1KO41(syJA`q*Un5B`-{@Czk4(?~L24(}l6!-n1#iT*Xo__y#xKKwT|i}DK+yQd{|
zU4S6}BRNO+q0a|N@)%ksq)#k(4wu18&l?k{
z82oBUKVl_%gf2-QAmNoH>VKB>tAZc2
z5Z^}$c94X`f5_WO^qvs>M7@{r4kYiz4-_J1@LQ0ZlNj<C{m%%S{y8d7mn*
z!@czoL%!jq(B21n0SrFp8!aT{C6H+*eqyMA^Up1(0byPBOy%XR7YTk)jFc_U#fKD<>&62a=%oRW{o2{{OrOj+lZcOc*x_X#u0jf
zC1#6e|K1BLx|T|pf!%@Jm0Sz
zXMcJWj}M6V3==6qCt_j6u6?Pz#pqEM@$(%{U}cHf4(IF~;@j+v(o6&HJU|Hl;1D!Vk$&HP!Rpli^0?nF}_*rE^5Zf)!PS-(&RbJSZ<-ua!28
zaF=DKNf*Kwinm3XDkxWeVw6BQnCYXM|fruezKeyn3s$=lu5IW1AHX6~SkXbnquUt+`_U@Iv6xjPnn(sQ(wPE%da@r<7M$-8~TCw)Bdj53$)k4)@!IA-pVR3x;A0GL_WP?
z44gVNxn;;mP)Em8f99b4fZ3coBY@k-z6_n#3bEMAL8Le;*GhEVTpQ@NmTKx7*(Jv9}|KP;_OR$q_-))}-^ff!V1Z$w5yEX00<^V5G
zGe1*kgz}6cvzU{RE7-RwOKuRcJ42$s8Uyz)o789eAB=OB;k(He)%b41J4{#iYF~-?q^m1;8SU@3R4bQt
z!o9)rpnEm5ktdHE+b0wKcRA5<<H_HMNIs!M{sr8_<(>;jcNF77ZL^xPVo
zDX-><+2oBkMg&d3x%s=T#)USRw*wk}ZHp1NFD$5jiSbQb`^oHW7~YE<2%fCxAZ89l
zcZfA1FDSNVM$^HS)g^fw52`^uo9YJYI<$?zzR$2`uMFeH)I%Zc|L`X2>9;z92A
z@t}VvFwGjO83}7Z(F~qK@0JnWKMRk4T9c)(otJNQ148kUQZS1f6=q
z^70D({m4rfM(;LE#dpq&x-=8xkm-1?!U#WIWqWuwfvAM7kv`^HTe&jdC7;QwROTbz48LSpIVnwR)LUzkaIH_0{P>6=LgXF
z`T4#4d*mm2XgmmSsmFUpy({~kU|qD)%8I9X9r020*#PAaqX)g|1Uffknk$`S$Az^8
zlmhA#RU#CP(_*huuno70T+=T9)bO;mel+73Dw`Ao`cMfr%n4KEgwe1`p?#njHR
z6Md9uUd+Zv$BqKOWSzl9@)sX$4eiL!9d&>Aq{x%2l)0(=w_<7}&k1D5rW<7Yk
zF|c+)0LOjWr-htmIA7uM<;!a=etroWY0u-`*7vyp0=TD%-U}cAKMBhA{Sf#?%p-Ue5VVL53DWvT
za>MDRmqJ`w2r20)gW`wmg>i^VOi!ykDh{F#V8!u_NG$_Zf%!b9wsh
z4?K=;{F}z*Cv8o++2?&0Nmg)?}0oW~f>Uzy*E
z&zJFQca8(>49Ylu@T(ul_4@fa@%fol^InhaKfm#o^ZD0A06xew>AvLiexOXS;{+V0@>0<2_W;#UZ{0{j^zc_c
z0=_z>h+)Cb-}eIVC5P^V_xanvsr#QGuJ0##cX8ed!$SFi7{CtLY1i45_q(b63oeh;
zx8!nLWfRVC*d?6TtR?4^-vk#?uNkYF?5v^B0zQR~i{LFk2;)05PLQvZoQoVK=f@71
z9nv4ar4`rDlH%Wug8UW-{K&sZ^4GHr1pH-NDA;%R5on8|ps#sA;J=0o_%W<49n|2L
zCJOW5c>|vh{HlBb4`)&}@H!+04VU2d)zoe9TUH2q^i^p*{2CAPYMN&7Cte78r&t7EAFyvx@Scn=BA|9n-@yJIhW-;icHFad|y33}5O2<2e}
zhfzP0o@=ALvXMrBcXr1Ge&a)-+$>F)_aFK~dlSv|dA*Gqf*pfWL9f+wVLrqnVcp2<
z33&>=P#|GISYgP%1t!oWq$rl1J}1Af{C7zTRCv@NWM30$R|Bs};6<-^-31zLe}(rU-fXGWkb1@!>t{y1keg
z9D#95&qxwWa1>sdoJ%_zFG`?w;<~GI(k>~!RPe-M3mc)U!F7we3OsjILh~TEtz)1
zA2=_=$t{!kH{r!YDZYNvyvFmAuFf7}!vCg^+a3H2{Nqs;t55NB*ahdUTA<12lotn}
zyljf8c@X&4x?{@zf*O7GY<>xO_Q`3+E})0LceqS>T>qv$<#ex5dVSvZSLMj#);_-K
zkA6D6(7ROte3Wkde)wSEtE`b0EaIDp^^pHH?OX=d>voF?w!kkX?PNlSp&z>1*VRYD
zj*QcdzLuH50c&HE|8~avk8C-W@sN+nJ@L&2_{Ha9S<^+}f4{Bc=iLWxPmAq%QZ^BI
z*3RsiHQF5$88(OV#ipAE*1u7}dGf}nuR5R6j)t;A|J8UW_-?)3JG8s;&*7`nfa69l
zuTQQ4o`0U$d*Rn-z;A!S1JTOaHTy;0?v5hB*TrpPGA_9ycl
zw?lnjNDb?#S?_axG}cRkZa~_6{N86pn8tA(5$h1@UAe=lnkl@$IeGa08Wu0digPz%
zJc@@qOQIyXklu{VqA$
zrwyHFhs-MJOgM5v%oZPa%y0cDgY2`=vq}hlK>I{|uwo9a>mldEYNvyKRX$i0qKV%%
z{QbSE3+5qVQOuP**w^p>!MZo`o_&fj4k@{m7mv3^1ObAV>B9G$1T
zzKoqh;|-c}{|DjGA6>M{Y5khJrllUlIvhTr-*ejEy|xXArSm{br+IS;_usFx+ja@-
z_IMkYjhi7?EGY@pgueNK&7r5jZ+>l@kO)9Pu*X&$d~-noKKteiGSd^Z8`aY#X);Q
z$WQ6DifK>&ZOrzB6{NQ`?Zh;qz8yUCFGAkmyT=rRFPhQ
zx@p8#@D`J`!Y_fY-}^b8<|VjuF&;9(|L%x2rF_IMU)3Ty|J%hW1RVyyEbql8@`qc`
z4O~b1UiXsP(K|)!#oY?1KbD&p@VGM0P#qaN%cQS<;5`B<##?uv#~+#O(4lnh
zaDP(4?`L%e{^jv8#CG>#(vQrVHkb5+<_Fb(1>a%Dd>|y?DUlzKBMOm8X~g$5RI?$y
zK9}uye`Zvz;r(fSRh!4_U4bKbe0_3Hh_k;^Ui17}=Gg08{yD23r-q-7aQ%;WtvT;K
z&Y6$LGw&zY_fi$^d-iUh!}ZkM9`bmv|85WOr)ShLF86fQ;O7F5dviI?Xq+{Q=jk3^
z?!1xu*_htU<8r6*9#u5X-62W5-xqcwEe3v5LH@2xz*Rd`Pz=6roq(g{
z7gOdM@5cykfDY>}y09Zl**u@?t4nxgFch_c*_CI<_zD`kH*?>#cilz#r%>dXHrc`SeEN{9rYffI0FR
z>d)|DD1U9ldAUezd+=Iw#&G`r(xIGBF8GNP4bJm@H*B7bDz)sLUs`z8LkLVBS)igD7wzFuW;jQnOtji+lVzj8qCO)!;fZlH@u
zoHx#npG|tICbsG1|LRU24ySh0U$y0UGHGP~J1QUFF2I@W#LY_!rFbwh=F{^Ipk-O_
z`Tg$W;~D!&FLI*cA*!F{*Y*YZr@hk;8jzl^TDc|lv-`r5E%r?=k$scrM$SQ;A9N`&XdiIF5Z|~ZuE^^)HLD&Ax`%wM+fl1az=wU-Oiy$}xqU|c>~7!(
zOw#Hyr40A}118(|8IJjSKKfBK@}f3>KV{?sADX^PTxeZ|d&;nV>2mGFY|7;KL3KNT
zXP3RX*9Gm_53<-Y7X7n+wAkGOxJxNp_HdU@zyW8r%@4nhdH>t=@fPx{itG18#X^6X
z>W(=c?Qu>Xou2yMMZ^q)CKYuvqJ8L}&3lEYzFgoyc6*JA>Q2ZdcnzeW95Vvwy4V!CqXlVcC
zVs_f_>d|8XxF0slUhDJ`<7gVMv{SK`ZMoYeto4Ix=6z)7vMs3JFTd`=)cPu>?|rj*
zz?f?GcDQS>nq4h>FgfDpm3h@H>0`Fch(Ev`@#Crw+g7u_dj?p!EUslQBTLULh^}Fq
zcN+u^R;^{No}4HQ_NZYu7uP;L7+1p*2L{VM1HR1;t=iFNOf9S0wa(5*r-rSv^LjT}
zy@GX8Yj2_!P|m)b8*H>erGgC$nd*6LTNPXIa8>)E@whM8^T1#CZaKTJE1I+JOBu^I
zHf)mnT*jWwF36rgx{|fiT7Tx%ohnw=y;t~>!R1W<+U<{vs!Ca0%IElimJ)XFTJKDm
zQ)Mhi;mWP8niVYS_>t$^J$0_ljzxHX;c*((I-rP3D(D_LH<*Rb;WzDl?j*j
zW&$hE2#2Vo1
zH!&+`9>TsmKfLu%I{y_qci&6(Hu%rkVg_6;<5My>3i$id8;ft7u}{pyqaw!2dcI;2{7cPxvKXG2%y64C#^tYt@!Gg=^*WpIHigVP@h1S(K
z(Vw;!ha2g<{h)Z!Tsr3;yJ@(!9K5T;kZ;Z4ZIxOdC7fA2{M97#f3b$++zH43E&p9a
z=Ma&GICv1~#gOyrJt5Z`<+79R+wP?4b|{0~f3I?U25qgNck%g#N7SC%LBB(kZ;z>tv=4+{PqUfB
z$ZvZMPS`{K{L3?)4-_AKyPf_`@xd$Z6#_Bxi*0*m(79ysyWJZpKDd;<=l6)+w|-ej
z{c`EQi|41mPOW-Mal_%y!|~+z|N5nslO2ObRb<+bA93$=gX~?rzvw#kb4ubGJ@U_i
z&T9+kJTRi5LsyD3YgcJxkexiaagDRUm&)|>rS|1}gq@-|vSnFd1jWf3A7h@k&I_2u
z^Ch1$_9D@YxH3G9bFl?-cHhw`B3Li^0O6J
zb%~dApcf#RhqSemNbhRyUK|kMmv6Z(Cq2vXV1B=0+jO`O=}pa}2L!;2%R_lTwnhp#
z_2@jn63B=17vi}8Fe>M`Gyl0Rm$y$C!pGO=%~Q_%Js0wcC1-t+@kJgK*Ih(yzTvol
zcHZn4eowDi-kwunSE`L^gPyOI2j|~>0n7)l^Q4gTsfHlX3vd;1
z=G*e=oWGxlV;=M>3Q{;9UIaJ=zThyL1#O*R!{zJC@v|rJ&AvkZIcmCa-!Me+JuesI
zAu!n=;T^=!7xsSye41v2X-B<_!4bS%FTkDis-jT>ZhMKe8_HXT;5wf4LI96oXXN(<
z0Lb`WkvBP>yH)psw|}!5u#L(?^7wo=)OO^!vn#7Z1ISN4R_5)WnRkSj-;fi=Gi1{m
zu2&tZFo^UHKi$pSUpnClt9D*D@Fsm1*}@4_u4L}a@0S$3%)Syn9ah9*$q%mB
z7_Lrwa?ZMZ-`h{>garjZP!Xy|?JOsd=zPasJA_VpX}nE(F>_6
z@cIj~7wCe@PSRGO_S<;(!C$}=vu9>~`_g-YN5;edj068vwaY#)(5}mE@YV``wB=w4
z?K=nmt`q5<@?!Z*TV{jbyS-Nfoug-O$Otp
zYzDqc>zA85_?_xUDmx>tcrAUhGZVayPO4`r{PPW?Z9NAeZax2Ac1X1b`6%t>sy7nw
zp60Kvi!Ivko!4u$HfZSZJJ)=m*Xt5vmvDdFb$!RZmQRrvGLX^z3Y^)bZofVNb~oJF
zR`d&Ts+U%7@x32d2j&GYw?qHr`R0P%=x6BG`Yq|`*R5+a&c}huMD%RB1RODCX}dAA
z`(j+j9c7OLC!`!`VEYgs7ad!%Jqx({bwG_n1@c`z4u@;3_5e;ZwjSe=hkTsH5{(^*
z?|O>9)n1oy9$FTu-+w9cWgcBF2PoFCx0`Qv{El(2i46HPX&CNb)pq##Is?BOjSuXb
zkKae;)X!`IUR1nyKFk34*zd3Vt;`-`*8Z1zcVFO*%Re+f9U1+Ll`rUT-)(Xon>J~D
zmmJ*dUwU()Mh5e7seaOR9{#NQ(j}|TA+Xf;&b+v5hI<F@Il!|y21YCzbUemASwC$GP=WRQ0~`p$LA1%nbcZ1CEvse?+`)ww_Z+*(w|ZX1T_9WAS5
z^+6BJd|lxW&irm&2fJaebH4n+I*QLUEL(%Ta>v_=tOxLm`oqv+wpbVIU!C84a3s!!
zt^4n{18y8KvUZy;&L0Ln+b<5o_g)!hQA_#q(VDhRZa4=Fdp)+7GH~RJ45QYv`2Kf#
zC*(N5&eSxuhi8G$43hVpyMuh@+0}WC>A;ca;`6RwN4vQf_C>$NcmIBTiStI3Up#$$
z<@v1`e{ZJ_tMs6!cK+f3v~#6xLeKtR;ZF?jb|^r9l!w2W;(&E|>FC$T$FM)ruND2G
z{T8Q}88ZO%OG8fjKCJWXg=62P;`=w$q?lAdZ`Hy9jb|X=-8#&48uVH_9i0{q{_Vb~
zb!%a_ePQe;Iq>(jCS3W7{j}w{!}eh4y|ptqa|QApXUeP7fKO}ZE_~sF@h$CjBGChO
zZ=QFa6A!)7souKuzCmf-Ui%2V>+zTpm79xs9e!i|{`J6Ro(HCG^Fz6)=u>ho=&5IK
zFT29-(FwYH=pLk@sQ485t>V<;2bAxd7?g_ugnVc=yD}MkgrSNz`IoasHu{7YkJqwL
z!Z~fqr*|fNwyIyjP(#o~10zm60?qzbwNMp$fhPCONw2g1Mj5g@aL9P2SkM>m=k5&!
z?R{;=9rA}pQ&vTi-`!OgZ%g?<0D>Ttjwtrv>u--Q1QQ09g2a!2X3QRDQL{*O}-!&v!3KmFR#c#`C2@c196`dDx{AObr
z$8lYmSh%0JvcU5__Oj^0s0Vi4~d>V)#C8!#O3-N&8bdb6u@`8o)B7%QfJ
z{Pt|c$8|FE3ePL=ZF`F6^KOsKNT>1(mu9Y^?-x43ozG|V*BUrOqA9&W&|*d2LDbivbPOX}yQ
zQS~~z`E12lT`ln@(qrgjLKJAke+dH0e*;vd$P(gPe{LE6muoIV#5*q
z!aalozuyG39?xYx>dEx5=X7ht?Gzx;F}Ze#q)z^0y9cch_oz
z%6ALALH^KT)!bzk;E}~MYJk7|AREPcz@HBtbjyx@cVaMZauogUX@GT-vmW?}xqI|?
zLr=xFSNbaCqbA;2Gch0KcQY(>DSwvrexWED{F?0A#_iy{IV%(`fdAfQGw|CW@V^wj
z_9~!!=ZxK%m%&?a`-2xoc&Btd^vuLk*v)Ua&HbC=O*^`;$~YQ$ITwCfx$TxYwTO4G
zKHT~kmV|ZF`_@`z;FqS4Td%%=f8JO*H1#avQP|T9SsT#aQ1_y`>_D84JN-$$hw)Ue
zsC(2I`GE~vOYQEJAYN=eE-C^3sOi|IdiQ?FZ%^&ef_irEc0WA^9F*_4a)dVG-qHNE
zk6$7VsMNYgjIOL=Yi@fz{04i`*Z!=wL%(|TK0S3l;@+uK#|)l@A}{;IvzJ;rey4N!
zZ>s1s-aQ`qI&@5L+&^i4|IsT##M0_FE*TpPe828_E9)6G%;}NA@V=JF9~|mCGz#-s
z+U}*7QV#NKQGa6Xx7ITGsS_Vv4X&BOX(z=N(9VUH7b)G${1Hh5iT4O82fJ6W;3n#~F^
zJ!gL7C+k(YD>AO7nptk`xl8t24Vz{cQNMLt4LddP;x?rxJwEBUa~1NEAwm(GhgLfaC2r2I}xZqW8m--HqoWv?3Ml{?6uA!w_OfptW0#h
zL9VcpZQ1f_^XB=u|2DI&&D1Ypwa3D;n&e8D!>72>8#k3Q+hk)yuzo9%_+ZPU^%afJ@mZRrVZ?lF)AE!A2_zW)%x@*
z-2d9w<(g?>9iKfLx(Bi~R$oE>*4x9$Pjvq}t4`Y8G
zA9<@WANy~$QNMFJ(3^LrqB=SQ^~d>LzlwGGtvW~MIrhtn`ll17LNBOkYvE~7JMBo9
zeb^6^LJtK6V?W35dNG6UAB
z>+N3kQaX=3k2y5HBlcI~w)pZVXy3fGXD#L7zU+B1iSR;BLs$~!yDt8Gn?d)wot*sZ
zn!x8|+{D8pynm@%i;N2JR;0>8(i`mh_9prJ*ub?)x}Y(1njlMCv4{MF8y4@^Ba|euXFs-aH69Vm2du&%=1mJz1HwN$GokNId1Cd+nd)j
zy|~?j`u%JBKCZvDYQ_PUf1w;9JkCq
z32-us7n96u%PI6R5;tY57=LDYGGc@a^$?;|+=
z&b>RY3$3$_{zW0QF8XZ}PowhPv+*-ojIZ-`_r65eH%&Q5IA*Lx$8$7q^KF-FkX}iR
zKY%B`%c*z#JCtVrj3d1jey;}7x|?Oy+)Q>&MfZl0-u}e+0;=aed}v>4SF?9d{9+03
zNj@eOk>2)DeKXSQQCSs7dR4z;Cy?H|)lrA(9Ay4%g%aT-tz9QR5uO^J_TGy2b-#zb
z7mfgL(#lPHF6igbn@!YyVqJ85(w`B&&A1eL`u_V)j{_fY$1s@AOPhQ{JZRsmQ^5BL+N+`0eA*;E1Lcp*E9#bmH@|20
zquk4Q=Gdc^b{
z@J!3>T|eMgGB!uGR)pQqO+I5|pm!rHWz`m(=QS2JE|A5!dcmrqS9EV#yLv~~G3af*
zHaB$&^hVf5U5$XA%{_xDTj5_zn)~awfnMXwe%+FxcTl_blqK}C)?~c&$GN}kX40cU
z$R90z8*si3`J&>U=be@C4r+YqNw^r=_
zQU~wT^$dNL_u%i-Wu@90=i!|3+?V}a44f0)x>H}g|2k-y^Vz*8&Rq@Gb6z4(pnG-I
zTbmU>+3?Pvm-UaWVFovgM?FUz-M`UkPIfoso9s^2YU~8goAgO<3vsc28()|^h$5%&gdKJeIW8}5DPrdit``iWo1jn->AQo}mWTCGrodpA{&uP+@P
z%b2!}+*+$;Rcv~YyhRq;8RI(h{7um0g|R71no5|<<7C5E24yV7*eSntX(it8bXpRE
ze8G!;%hvBgdww6wOh-nSvYscq#fMcuyO#6-jvc-@}(p!x4wSm3M4b)dy~F-{2;q
z`{-|JN$teI|BpgEvaud|`u@_Zf&L~>>-j5zKVo!;+*8Lpzp&`w-78?PAW_kX&K2jC
zL+!$VhkANki2sfG2{~tEG7#@6*E5Ya9e~HzG=B7_-#>j9yH2Zw9?Og#@eq8^yh2qG
z)}c|-&y(F?XLx(puSCmkZ#zrx{B+JVeS3g@JTx7xBZK{sb0PKYN5~cbs&z1h-glos
zl@s6x#w?g_2D`!TvLn)=_o~`%sx9PO9os#j{BFAHZJWE`XIx6z=LmfB>_>KQde0iQ
zd~L!X=oe0k@^yw@!lh~HxsZE4>3I4#^a@%uR(6J5Q#Jegd+0e`@Qrf^Z~1K249Xv`
zneG+59W>Z1jQfMMri|efFW(jTwvd15GQNBx`MV>&XXICcch!uTN_mz%Im_d8ztu6=
z<|UnT7HcjUPyVsJT2bp|puw3NT+%^XjDC+Kzo=|-+Lr9OZ=K5DXK46;HVFhj?A499
zBv6O`#ka`Mb~s=;f&ACO8MEJ$U0a#AXDBa}IrmopjraSUZ6`@@z|7J?l)s;RWJmz{
zxg9+&b|k;=ylr3>_1D{NI`@~Y2JPW_pYyNV^0?`7zl{G~JL9>-2NPb%+`5S0HyAt_
z!tY7+dW7+KpIzdUNaNX5dWrw8#?0zyH2M3XA8zq{%uuIx*3?eb<}W`tJ)kC(a)
z%}pfN>?PcTjU6h)%gn(hQ%K(Bj3&>Un8c0ddaFMojRm`k$L{j{#t-l32Bf$DZ6Mcs
zp%}yQmig1WT<^)PQmz*fTG)Z~3`Z>E?Y7xTe`*#uPU$6&(^Eb>@i^-r;V_Hj>z*oe
zd~oqHSn!n*9XTDGn9KD#+&pue5=g%Pu$e8z%X-4>EOea%&l>x|8p
zsgr(VVdi+!b62p%LqPn#yjRjeikCBg^$Dc)Yv1B@g5sr%?T$GVFBh0Eb9KeO+PJr4
z2IUR>XI9D+?ws;^9lxgMm??9(FdurJEFS(f?24MVqi*ioF-;Z&Q7`HSG_{U~L
zmrg(M9$o$8n~6C0kN&br+z0nOlk!p^p4<#C$9d>ttQ>lWB*WSlIrmVfD`%EVU31t*ouQy8$1B
z={HTSK>sY7jfYn5_{q9|ymBw0qJ}MUNocJ1Dr4gp=x5wTJl}BiZzv)er{@J8ZGoS%%-OAFt|GU&;nf
zWlBng6-*8}w4TMNM9T_2Krs
z^UOA|XKQ*@WeDP0RF!XY4BgK)?Goo>z6X3L4-dmS`TkTpB?RkcL0-}HHn2B6Y+Tt1
z=v_IfSThOd`S_k&b_W0-=vZ6Dq+=cTl0UoY5$ITz;++k!tKnR!L+|_CMvq^bfqv<~
zoppuc<*I;l=ja|{yl=hHFs$oUw{IsIL+;h$Rht042aAUKdV;^w$Dlv?`5ebeYd`22
zhsv8#zVQC&MIWz0ug#9IWuqYh|gChT!g%(aDo!~g97E3TdskxeSIfm7pU3ikvAz`
ze$6%H@v_5g!`>99rf$sNP4UvfZXjMx!+(6)UHTYwcIFcP-hGbZnL5%_>lb~D{8+|=
zaMwMcDHEJ;JpvtkSV^7y=QW+sK+;HeTww9c(a10Da(c{auAy^Ca(
zXg&tTmkcI9_Rv;gzaMC;6WSMwKx1O`8|fUB|8|5Km3K1p=kY~nhzq_&^mi%`8vbH>4n%7_8f7j}v!Z}0IyJkv1XvDcNzBPXMh5l_jGl%%>V~^jH
z|7;m~ZyM>TpQz{Y?UH&VE+h~aUVm9gdZ9YU8|i#lWF^Dx?THoso0^U%Fe6Uc>Gd
zJS4&2MOePppVrM&8$1UF9eI2Yzu%Y}Zo~V1QgtA|H@T$z=px1OoO%5@9{IE#X&Ts3
z&lLK3+GGY_5A|*b29Z3;U6{XClhb+qA3q0R;Pm@i`6tv)Ncl-l-L`%3pm-^o`6_|d
zLFb!G*U>%1)^Wc)>D=Ar`^0dHQzC^t{J}lo{C$sc@q_n2%e{|LzU*fDsFHr*!+-B6
z+XMRW(IK8c$uOW_t^!w|jxr%?(%iI;_JxaQOn(~3D_dt@k~hsA&GBAh*hM~nRwttM
z2)C*R%paNuy|UkZFi`l1{lkKZR!8SOqI9Sr+`>#2-%Y%n_$gHZ4f=&BXWLZKg2>`u`VuZyrx|^!1OTNpopHB}Ii&2}$FY
zk_stG^C(Gakf9=_c~lxzBqe1mghG=>l9VF~(P%ozagHXH-+S%Paqn}#ujhCFxUYNf
zUynaptzoad_PqAm`?F8sxs5^ez1OwQM+&z=?(rF~u+pFKM^iQgUc)(rjo*(x?~%{n
zT#wm$kZ-tmk$?AIw)FACbjZ*8Gwj7V?4vIEWo^F?xnRY?M=3tB@7Z?;`fjD+h~6YA
z^pAbp=A$d-Dc6P-(<%PpaJq%!WAM|zbjct;NaFe>xd8q^;a+3sMEDP_R&6fd%KN~k+8EHfBG#_YG60o-JNz8_d^z|9J}}j`{^v3saCU7
zF%Mcgc;9uydfayGkbaY7Y^m3SiRWSGM}r}uzs2xF!`xk7VZZ+B@P(N@!m*C@3i27)
zAN}s~&zsv(uDst56x{(w4LZ6;6YJU#-^va4Sf_U9_BgZaJig;tG&;c^sgQY9>+@VDwyV<|gV
zqkLQCrG$mNIZ-{QxQeZHKVDJ!yoyB`i%lo*s$%Ai^JjH1sA6k!J6x!5S;aJe8RoR<
zC1#_J7Q7rHu3}wlzF(Q9T*d15&iXXkO3ajxW-PijUCd^T
zDX<=Q2lr9lJ6&wIrHb9?aqXS|*D7}4?U)@Zm&9!4x9Lf(H;b`vX!BYd{!Egu`FeKQ)<|fZ;A~o(SALRKloYp!@juU`j~t0Z>mmR>~Ct&srvR8Q!Ho`fJN;8#G6tzyA)%U;g?l*1r>bC&pf>
zWnTU-*3^{XKEqYhxGdz?abkILD$ZeMD)cmd1V82J=tX00;QmSb@zvS}7$-$9x*m4~
zJ!rVAy%+K`DW7w>JNkQzbyYue;Wq{R*xW7;`KYh!b6*{Cvf`lpq!{pD7Hz$xj(*J-n@F
zL-9Hrr&oLGz}MG`TSN0w_akHUf{@SSi=n3$A)mIdogPnu{3rbF^O)|raa)+Nh8
z?>gCR9ND$DU9Y+?pjVxH_@3r>NlnWgDJaQUpR`2zrD2?-8wAXlsq&(B?I
ze>w4Z!FZM6crVbmZ|bgI0$5jh5ieFj?_c|2!^_CC)2mkz#;{QN`Vvt$|32i5=K>-4ut;ex-W
zHJ-;O3XXd5_(Z1xSvXk1eT8BBaUlfve$qwU4~UT_aTsm3o2UEdUCHgW)WPx#xz!i1
za{SNk+Z;c;>j?M1a*yJAG4xp;z%d5oG7SX$h>n2E)Vp!H;3W`)9D4zC0B_bH8^|yD
zSj_R(PLnviK<*dlNe%)uq~j9E&o#jD5$JtZ2=u8UkU@_t67b;`Fdz8ZF$l{*uBQa<
z3*hI72wv_3Q&)0-=$BSDuQ!!vYq`H*Y(CG3((TJk=5imsT)BUEFKRotTSF4S>+`p4
zyambW9+zptG?kDB^Wy(|ElZc~0T}
z;bG>+_5Vs8!q1VuujurTV-s7ng`{ZZdNsJ-zdyJSkqA<+yeL`BVDK26~cQ
zyO1{7d_H($^OVN%@}{0V9$9nkLI?+iS1m~8uw__ul#fSauefvL`cx@aabqaE0>MJc|ug2dH-wh
z6L)2<-cSDZ;`IhCd%>^0+RG?>;$7@3{GOz?cB*If2|L$I{p(q1-GkUwo9fxp1A{J_
zwXJ94J3rm1)or-9(=GDGK-@E)
zlwdPqc%6j3@7wpqo1YREKXKXcxZ@HQC!PJU=o)^f!eP-EmjQ@x?R#QVtRP_lUpkbC
z|4LZS5hJ528{}6Vxp97BHQQ}jnfxZEn(bSW-aE;-nw8udFf;lM?z7zfP~o2{W`?!N
z7L~(m*t$~bs*knROz&uL&v?YImc9Cu+ka#=%M6}+;^rqYv(oH;!auHtZJ*zxZr9ox
zrq#*iP0Z(N*7Ck=sj^EoJAHNkM)Nh*EPb=<`@Jn{Sv#Y1ZK@yF;2!tvx8tm8@Lkuy
zh`G;e*sP-p+S@k6uTp-P#>UpNG?7Jlk3Ti|9x1}@CGPKN21l74{Zh+x&i3~6IF9ci
z#SPulipA{Mubx_`BgD+Ru{73eq?m<`>{_!Ac3brgH5e0J%a)~lduNaD<-DXm$L7x!
zvz?Jwe9jwIvo_sLl23)<-pc!sk0xojuPDyG9OhchE>3lQubEZNJd|c^D7+6l+ML`S
z0sp8l@tTh*zIS_(pDDZVT+1G3R5^~f#JKr)_EB5J`d^2ocgJXqc_=ruf>M@9Ta+
z&w_QEa_4~m@upz{&C_R!wH53D%|`WiDgg}eZK?ej{FjsOzM=WjtnXzd@+TE4S}r5~
z@u?pptieA%@~;`KQv+Vx7SsL4N{gco^v-t5k~ceP9&WUC7n8p@??}r!@+W)eUCN`l
zU_i-CAM(3m?rR;Uxc#`Vtvoxzt_Q*GR;vO}9&u=09{9$38}JLGxQ|e&G>hU)|5UQV
z41lkG`O*C;@XteAA@G9l89%*r&Ub_`Qyw
z2&47y*0)H0&Z4|^`AzaiqfD0lA^!Z;rX{?*U;j-ey`hgk_!B&ztl65@*FoRiX3+ZT
zd(Rs$MxmW1w&LGsXIbYDAbP}*P5bG-L7-Pm4f(CZWvB8ey=%+ci-^DTmGFDAu|Bt)
z=)L>Jvo~`8BGm2mLXvBLN^LdC`7H?zqWCFsb&XS8jHibmAKTK$J>nk*7J0i
z3tDi#U8)w>_jk`}?uWHJ>&*F^Ucmi@y&oHpE%N&k?8D_EJU(!L_?C$_=c~BW0}EIm
zqH+8Lxqa1WHsH~Z`*Z!BV;Z^M-QVze6ZrM~2mKH~ZydqprY{8<^lR|WDc$@m0e?o<
z!9fqK6Y{aZtz^OtuLy82T7%@5`oaAHT`fiMn@v%hM4y0KCm8uskXPR?;Ga+%M91MO
zht5)gKMheYqQ5yR;4EIi=fDdj{VoIm$IH`2;mpWyHI1qMQq;JS8!4H4y
zf-od-{RslxYzHyOO_?je{(S{|X2inX2Ho8a$Fh(Q(i7_QM<+pVXo8@3_&h;w{_iea
z{#z!(bC3`AKo}V5xAqYF+vFK1bBXSd?#<)FnMUh)f3STxhx?z0eT8>e8V`4UOLBc=
zQ@sghxe0OFNl(1F{!Y7v{+8RmCklkPV69k2a$9#(1K>VG!VRH(=H?u~H2)db-=bYt
zzJEBqS+M_f3APbPH`A3m0^WJqbe7B2+J#-HMcUG+3S`X~R0qPU<#PdfyvaSmzr0v!$S
z@(y$EPx7yiuCyWfpBC=?9Wtfo__;LDA3f{M?JN9xhL`X5_CI`IPD%;9^Ac%C!C03TeK&-m2dOiUZ^)A9zdmu4WB$jn;;L
z`!MO^iAccBM^Epd`OjvRG>+aUQ#{u+W&rSsgUv!#0#|oCtl0(rtgmGi|9;Uf%eNi*
z?K?W>Ig|e3rv`1N^U>)WifUH@=8SLoj^^3zF21o;zJrg>R**e?J7%ZT`mxKdPwaib
zD~TsQ5WTUwUW?}cUim+8V1a&DQa_XY;C{BR)Up7R4wdEWVO$cj4oHKXezd8jiU2B2-C%HDodw~DL
z;H}~);A<;-vr_24SwccdE43n^Fu~AP#;Qu^*t2ZJX
z{>S&L;oV=q!n)87)2-T>I!*1%%#bDSSmFS7Ms41c0}NU@g;akSz`<@LX?53>23RnZ1=
z_FJcVnt-o!&@*uh;$nN3-YFOjdt7Z(wjYO|q`MVR=KswE9e{AKOd9
zmbt6HK>V(YsJ})BE%*n2=LU|4{}rsVE?bIzkh5x4+DqK8
z_O46Q(I`aRt%p(eT-q{#Ax4qa08l
zp&OhJE2CX(rk?V!K>iE8*;^NrDd`0ne$`!1ue)w0S#DeYHj*Rq5@lU{sU
zgLUWCNMGMEI2Wm$>K~Pe^DD;!4{S9d&KvQyU8SDPP^qR|eYgyiii}NRj)G}jF)pnWc@cTRGe%k@R
zc;$$Lhkfxqr2Q_n8?LbPyEL@T%3{R#r`a8Ti*c7avz^sF8h^=cohKmwCmETn2KEk1
z4>U9fy&`T`E4u&H#w0n^8Rb&GZ!Wfn|N8lo$up{VgCVI)L%`pC=TwUl+;{u9@>N(F
z%C%}?&jx3df05|$y_YdK2hr1P-H7iZ@|Daye9(^b!|u0Qgnq03Jncdg^m=@9a@PU9
zX1A5J82#m}qHWZ4jPntF^>((y_%&ElXieujg|4F7yMX<@%XNEV-+rlPj!sL=V`jf?
z#Zy4{nEmUM3+%d*czeb)*z?%+#qX!UpS1IDa}9juQPT^2&`)}pp5AK=Sk!-l%5m6v
z^VF~#IvAIYUzhvSyxXTF^ZGTEgB6strMTpdErT}H!_M1x^;0H8KHD(2lFpsVrli%9
z|3B`mhrJWUW?A=}z)C1)Osuwv^(;1G*Om(mcK1Vg6RSPpde`y_9gP
zcZb_k+{yLSg3SawISyV;@sw^Wlg81yWm?!eht@g6DUC7FfD_KQSw0p}WyCXO^4l+I
zH{k*??!TTe{YmrO!i0Y(>Ap_TnJMbE;O}j*W&`=v8Y*8e?T&Jj2dN!c{
zz?fd-KhI8d+e-OGbo=5!>xIIQtvjebTSQISMtYaN9%Vu6fv3|p8#411LoXXN%~Cw&gJLow+5`cMeB;w>WsTY
zSJ>#kiR4ag8ugaatwt1v_ASAgVt!6JE_%u`;)nNt&Cj!Y4Zrk;giE3EiFYeJk+A&*yS{se~MG+}prwp#%qv2;us3EcbDKKpe=Rhn3z;*=uNb6FLKQ+6{
z?Ub&?HU|7o?t=f^{T$pU;9c-50VIDtYZ`|(_{xOvQ8Tbz0YBui;1|vg0SbKP`Qepke?&
zNNzLmb5w1B8wmiGQFh?Jk>lOv@^#$={wQLDPo*JwGdb=c*I!X8@Db$Yd>l@27%7*d
zuOuj4BvmuuyWc{+GB2SX-p>~BUPVH^?UhaDa@)8<{0{ppL4SRikY3}SAU|&uZFmsZ
z)Dr0A^%Aa6)s1`>&lBL5LJR}Y7d;mEt0|lgewRD}UxnUC{2BCYBj|(s3j9Cv{aq&=
z0q?9Q$c>l3Q|;c16yQEd2?=66ouA0lPkoDLFMzjm$>!r%a}&0qz;7+43dMIbX=xmP
z_?HR-mtXcysE;k%sbTRx+-0FY+N^xX@o%v)T>cRaYw&9=fg-(~e1ADlUS9#h
zT{-`!lnRD(>2g8uLkpq4YC7Wj1JcP-%sF&e8p_A_teHi89&s>O!S_X%I-cO;S}8xG
z3(cE{ZaVY%WXF};eEcf*5aJDX-FET&CJzFaR8fAvg8jMP%cZ_N|8Ngs{2UwGjn6~%
zKD)3ZMZX^+jAwJp1zi4$Zyzpy_Fpouk1pyn(POdR-=E3(!M3@aURWf&%dA(m0UZJH
z=r$*KzQI0jc;E&7``Us?^6xDiH+CRgnqi$y>*P9p?7lr5=1ys`mE>FZ8rYWjtCx)A=W%tV
zGx<45>%^fv4uV+YSkiB~yD^yLYfp|UA^HBxXL5P1&gJ}EFt+nTCz9)D*ENXtC%WqX
zo9SHQ{j?tX{*d4JP9vAr=L><|Z)ssY_p&(p@(cWht+|l_6es(;MYN38$!(sx)pY)G
zK*{3_h#w399*|}eJGo^Fw)U%Q7s935q
zmiWGVUCjtDt5)--`?B^6+@on7-sbciFGm!TeL!*^$2+Z{`#hUseZ>?{
zyEioOhYa)&%TET@Azt`!%zd|W@bAwZ`&>%j1rLdyJ?kC(s(?zP^E!|ZYbAO~amn9F
z-}ZZfZ?vO>RVe(q=$yu>X2AEk3{$23t!%=F@&8b+QBOjxrvvZ(MtLRqr5nO4hqQ#h
z=hA1#N?N}iiWfIhyz}tn@Wph0J~zpWzk_fr^Uo0p=qntz@b|5UZJe~f1pGs7-K`f8
zz1Nry*1%)$lx-RSxTf{AeH!que@t8%z6UVW?U;f$_A#QV8yp|PFRQFxqpVH-q)B&6
zI)_?W*0fg&e7CsasSNuX=S};cq(biG3-_>i(7*Ya)Et1FdR4Y3Qs5t2-SFwT7;n&7axR7%la&_ffl*q63AHA|S-5YS&-_dR)_{Sd4@Q`31XwD=T$xf5|HRH|$
z4mWZB&>wma)YV;{0~n_|?1D4k(%22#$j;rmqt&UNGw#iikRNqcy)sD$FsfnlOiH(D
zN`4O6se5sd-ekZ_$)9g0LVoe(nzv-<_kBkL#{oLe&pNsR_4a+^yRtd(d#y(K$-V#v
zuJ5N>4tskC&stWDdNaHWd
zbJ^~Qa1+3$lI1niKz}l$gVSI@bxlVP@^=f{Ux=r;x4UD|8OK8nrRo6i?H<*R
z>I3Le-cDZ;^fLouX45@$FQehrTA*uQF3qKR@OxqJCloI~7%)?l?t`72wy8!NaM{+%
z^<|*1c(uKpkMlPJKhiw-UQNB4;tY%
z#_Q1QUD9Y=4=~A(rMS!3SwGv;zG1s+Z#|lSS7%JI?gbcp<9Q1i_&GjZ@DLg9=T%hr
zkbSlfhE~(Ovm+tsSq0#lR4x2;9==ahTU$)?%Wv0Dg9+bpWHJ^s%m-VqSd!j}zhq5>
zXIQ`D^82so@p-*h;GdC1|8=e|mGt}0H;AP4g=(QwiJnjr*_)v4qJ@m?*j6%6pUPwK
zA!Zxx7ltfbh?O7v(`~6s2yWYW<`JczU~@`|{Q94n*YMMOcsEDl!~O1|7i^PBkL{@B
z0)nfXT4II9_qAjCacI15@MogORr>JttojIkJ{0^JK}LNDDin7~C4SpT`e6vfvk%h`
zE#p1k%iDOI&{)G)gZTa?#XL@@|7QmlR`@F}_J|B)9JiNpz4z9R=Ihg%*G1fqE{mPQ
z*Y$ZjHu81;)bJi$@0$-YuE+mq5Vzx3;7Pu||BGwj>+HOKJ-8iX4QjaG{=RYw=U?pF
zlCRSrWWqY1RinW19@epZU$Cnnp5wds`f__3m*H9j>@)r|nB!H8o48)*cl-D{oabu7
z`HO~;e*~aKF*CBTc@QWjw3;Ig~
zg!Jq51o|O(r9`)qLwo+ihgkoH34A*$Hqw_D338QP1$;gM(4P<)>6`t0s%4^w@&DjH
ztsKvm%k`7vE#&wkIS!`{fB4>5j`Qse@MeAZIEnM=Uy_5f#~~L-ez@es@wM{w=*%4F
z$^MI8@_H%#Iyu}d_wTh|2=wCvNCDyt+XcMUE}`A{@dN0#a{bQKEYLGszJE}l<_CW)
zRWs(_5f^{}36dxqO
zAJi{EhUgb_@tGdrM#C#G6#nk?=nTRg^v`1=z`3o)MAS6)%h!ba`i_y$dH!n}y->lp
zcWTt?J;{amU*LLjx}N5IgCjWR0p0M_F<#!Ii$|d)u=}b
zmjn4*=N{srO~jja*P4)jKP2*t4e`5_yxl;!%hMj`Y28~eFkkH+V6R&_<4Jzt-1;*l
zSC%s!8)2NQzG*6@avo_H;7;pl4_P}@AkHIw`sNYrFe#Jo|1Qja?n!v>YFGX~Mbg4=
z=-Jr6i>
z)!(ka35Jh88(R)J9etIfX8=9^u06E^^oDJp@9zVyJHG#eRQOpP{)OWQim~2K-#C}P
zizzj&TuS)Q>ES&nE@`K6-jwd4dH2lZ`#-a|Tf51Bt4Nq-P4T;lQ`hl*=A7fbXVd=V
zec0oZWZ(5K<^1QZR`
z=5+M6J)I-9%<7|0%K%%t8nS>(8&@tEGS{
zE4#IzxUAnF+XJL0vg^v}-vQTeJ@|?2jPodvDKO0)9P<@QQ{`nQ!+w44*Fc-_mkIs%=t*=?l-itBf)2*^%s+!qttd5Q9r*i
zy&e%EeE0G)^Ejo`Ua4+Ra7Tv9yzbz?8FX<&UqFvn0Y}L1l_X1t(?0#-cg>->px26@
zt{^>|a~991`*pVr_vX`n{Y8d~Hu<^HpL%3a|D1pQvl`j?=0Sx&+39(H%oIu&+DdyS
z^%v#+b(^SMC4&_IlAR^*4{KAo7asrDllobB%H|K$k6JI8G?wb+Xnb$}j@OFTmIKK@
z-tqK%9i@K|@z$Q?f6J2Iklge22m8}KYgzo!^vi(0t`qlCzD^U2?I`{c7TC9j^e%`A
z71Ovdd8uws?J=Xs@GJEja-FFV(XSk0`IZOE|Xu8K!KJ(Wtct|i+
zL*0<%?|m!QBYawh)Q$ShOshwH9x_XO%I}{J94z~gnvnajW
zm%)3f9L6aF4idEb7F9}emMvWasJ>FqP-wG`HCcDeB>sBWGeZfE-}oYw^4qKU>LTH`
z-o-o)b7Fh;Gs5?@n8WSZVbfys<$gj%eGn6D+yeEbKFwB+;C=;$+8P;fuU
z72793_ZxTlyv)L1^K~d}n;xI%ZSq$fA^zq0rC3<-uIrXnd|n=Ai_fuut2b@r~S`{2dn_Sz=>c)r7?4a3Tharo#oBDv@5LVNUicGgfY&LKV~+d$
zp5y?-Z~i0jt?6e(fIp#Uq@id2VgX(nE%*Zlzl8L8a=-BTO*9MS+O2!X;nj)Q&H-1J
z-$$3xwQJx#)CIW~O2Yo)a|?0=5pShq1mx%9EXbXq@Fwsw?5YW7BFaW^|!b1OR)g4cyjs(6
zXr
zL|)EUQ-t&gf^dDYgd!d9&d{GOFXtKg{-L>CeESBuIr4PH@_tkKNJxKLu4lQNzgMn@
zzEU8&1Lfl_^@*_GYd?TWiu)hO1^<4dw-8^5Hs8G6f?TS$ASb424}8;g!ahnhU*KQj1xK9is(>p@6XZ_d*Lg_ZLQBAZ-xBzZ
zqXj$@x1*?@bmZxm#L)=FdptXY^U24(`=Uo-9_(`+Er53fBia@aG`H!+nUmy{?#5~zMXJ~8XEVk+6!@#&01wV-|M4mIbHi6^(ee|)|Gk;-k16=
z)Z?fgg5E{zh5d;yAIKQLiGuu=V?sTMf6~N@cwQt0I1n#$9D$mH-?^+BEd;;qTL#|`
z?OT&MpKyue414kiw#-}4JY9m
zfd2D7o}UK{>EoPDbUT0HdwBClZ@C`*1Tn9ly$w~|ubBTPg8Ms~?qAS@anHP4CtZRT
zyL;cE^b5yS^`p4$0D}by6rXK7=_@jO9(ccX>=5rRqQcIcCxGKGhK30~dPV-}_VYc6&1MZ69P|>caaGe+_CWU4ZmsKEcZ;rHhF^EN@m6LG^xpcs;)_m^y>f
z3$zkqKnttD9&>0+|e_wV#}1RsYTJ5Ky1dC}aC9y@h;|0_zb=J!E692r(X_NI+~t{>&RKdHRd_
z_&gJQ$3PJucNY1IJReKlM?8JkHh(xY4UORSqTNZv+jY{hc#c~>!}S{2uRSP&+g0O(
zYK6W&i3554?EiPQYCz8{Oa$l=f)b2xu=dwZU)bt{+&e*5JZ-hiE|
zg><(Z=*9^A-{-rz-1J!TL~visqbtXKO30DJJ=Cznd>&ZXx<99P>n-^A`$kRV_^w0b
z_+g!YF@)n=S0CcIMhcQa|Kn@{POYPWAkO<@FLLO%N0@i*-C+jg3R(&F{k0eH+Lb$YDS~Y$5P1bWu#;!>vm=EWItP8=829l=PW-3H${({UNzQ&(VDWt8%am
z0N%mr9LeQA6w0lpCgk%YTUf^&F-s8tnmm600Hc=Djl*g~bRFu}fO{~j5iIoq8~gV9kU(DUppn>
zz2x!w-*Wj{dHi|BL4mJJ?h4X{Rtx%f&@v2s6RDx{$nEf^ZUUUXx+VT1IiI`+qWe>E
zK^IfC0rKq*=)(vF9w>*F*rgJVNs_0N$|0{NN=Ia*>(O11!;dy`d?5kQd9ffjU7l_$
zcQ|p+nwkfEOL;zNgaW7C8bQA8_%D2FMWS=Fp^sM|pq)HDCX4?Izgdn~BhoeBe`?lC
zyGJS)7IU5sgPr5brX649jFPgAzOi$o9i&V&XjGQCrGz~ibJ?>x0sA
znlIg@%qL()@4*kG?9}wfD^zXjnNr8eE&iI*eJnS-{0r!(p6P!AOzPRz_?cC@&m^o)
z&)VTOzAyY)*grVNRl?LI?Sn@o!5{6kZ)(g}37d86;wtSLe0T6iW^?2|;?ys0EVlWI
z`y~Tim)ZA~uol(YhN&y6S!5I&v!$?xy?d#jqBlaqx=2eZBvY}jb*RX*D5ZXGw9e(y
zXLiwYQt{Ca)of+^31w0EeluiJH)B0#+!q+!6r|fi!a67YJnWhFnO!?py2`;$%&bLg
zRg!W4wrkF8^_4o9PrQrFyT#*oE~X84pFUR1lpbYQn6Ah7fKLrlVqVv>g~>6t$MHRC
zg2L&@XYn{+ihl04XH+#y(&{|2XXhF=GW+h6cD-xat|I5WL*uY7H*oBDVl%xP+QPrU
zqM8LNwMh8Vs)ltnn$>sCuNqdcX3xTqX}FgzmVNrW68-AZwC$OPs@Ye&1)n}X7qj&4
zt<=pF#q8gMFPYb-Rk0Y)*~MO0Dp~0FQ_mhRLB6Uh?_QUP*^}bwj?bLMtkbbsHV<}G
zvF`~dErv)cS-V!Jj^6x(?-0w9^h}qFS;mGbry4B9?EXiWoJjvFc6PFN$@iC)O!Y=-
z*{pM4Sl%cHE4Pjn%(Ab0pGomm%)hYsvfJ<~_B8R1%CO){7VB&I(m=h0iO(I_^ZR2t
za}T$%f;}sp
z24)B2_X1-YDp={@N;dkPpGpsh3btq1hl1{7pnq7df5?;y#4n7@_6An6rgz(BRIB0r
z^u&f6uJbFH#=Y-~Uxwg!X4*yd+&2#Ipo?NkElVqyZBk#)XPYb73H{ac24t4A^nCwk
za{|iQ9>=5G9~~@b<89Zq-TbtIxf{M1^abU5(AIF(oxT+;sFPV`*UlAe{l={EJMp`P
zL!DbXXSK$Asv19Y{?wPVcj}uiwy!B?UinH77Imm#a~7-*+=kylJonS~!eF}!mUSx3
zeI}r&v75IQV3$qL^i>@3`#MZpw-@Tic9^9%Gp=B*)aKf+o>jpd#V1Pc+ElRDUwXKV
z1bt}RK>s463ieTd`@5RmeA5)Xya%La*tbb!fIa@fTJZyAqIqPu8
z&*fCN3T8fSW2Z@YH(vEz@7yYr3bspemX_&)3N~g>$NdN9RIs}zW+e38mQ1d!>kqy6l(K`%%`0%vN7wuX%B}I?
zq58HPdFk}LS9WP*3KlVsLKL1q@}5=6HYtCc@Wr>1nVhdY
z{}Au&FI1n^a(^7k|D-6jrgJ4L_VCnFHmYQYW6oO~f&6XJrF%Wg%UM|F$A)|Z1V
zGWz%U4oQ`4*V0gvG;t;KowsG-q5)N`W9zi;xl^j}J(EG*g4vajD6^Ilc3VRqvp*(cb)yXV3C7*Ds3evTaERmsxN
zmT#Y$P{}0svRCyPQN{lKDeV4fY86{G_0QfYFTCp${Wv^+K^5EjJI|@HZxvf4dvWBZ
zSrv10wtb%Jf_DYibv{*}T*aO*93eUTq>33HT6b$qWEC5p`#G!xdPc3dkP@;T?*v+S
z&RSlF_k$!)2gFYjv!ffdj{ce^W`px5oL-K136B@1og0UI77d@<7#38;D&rzlb>CO9
zx{7>_hK^#Ue(kZs+-s=U1unWFcPm+sEjs^x{jOyGPFs`~BcHoI+Q0fE{hEF6()xF<
zV*5XCP3k$TiY0HjdH4N!yg#HAA+AZUVoO}U*0ijI-Fvj#l;HPH16?lM&3jqJ&dc=b
zDhjGtLR_j+`RgjCu&ZT|72Y2j+_mZY-*v%{8{&+D9Fbqh%)&UX0&IPnw{rEoQbF0j9_2i`j8svjlTvG5c%Os@2->=m%xKPd+Q%5w=at9LH&QukpZoGEU>b2dgT!dQ)vi_YGA{
zw@6RiH4pWlxVGS!l9+X7F&?e(yR=L5GAu&>!mi94otsLk
znU&I(4NviYQ}Oe>A*PKehnrqE6RfjGzC19|y;RLEEF94)81G9vS^cPr#P|4atz*OI
z7m3+c-63|fhg7o-9gRBw^s8o@p2iuiPr$nXUHVy
zFaP$|*0A#WwR+kn_#Ko8v$xY=XOcmYN+N!LZHs5JSfg7F8+tC~=S066c3FGuHHYLH
zc6?Y;ATz0EzKWS22CS@RW7aI)qx-I!eV_R{SZrRy2H1D}_0YA3iRWn>`kLWAfEGQL
z4#)cimMPmR>35F@A8GOWE8dfOw7Mq1#})V5$7QP>wy0%K`gFLSf$zT_{dhI@fivzI
z#kHEe2KQpC_jsEmO~<|E+Tv|DC)Bc%jH|m2?yP0^Ge#w^P{nuqm7e3aX2Gw@8Dcjd
z_fjjBALs<5z8Vf|JG{rc3;8}NRB`-LXgrEg$o@#r>=
zHxg@Dc0$vb>T$JfU+>{+5iem!)v&tQ9@uZqyMEabcFs{c_;cB1y#Ha~WZW8l;UVRD
z^;b7Y*z2a3%l+^UhR&(ebBwKUUr1|azRVZjnWv4Pr2G+n{i0`2HcrLw!#q)_*4Qm!
zCwKKUx(Pe2YYxomHX7%Z>qCRO9ml!uF{wk{PuxFo>#7GwjBWKu+nfSBU3ci@|HQq>
z$!e)p2j*kn(Q4ke+-TgBj9mKSWFhL${PxVRD!89H%<^op9eyX|QJLkzbe!us)&IRP
z3ilD;XtvSyM!cqYe1un&guPsLdt6McgsC`83tsv_!kj0EPtABQVcUjliB5f&u%V;>
zZtk~T!p!47R@^)&VWnO3XXu1W*vI^>mp8qUunA){jYw`uRKw8E`V!{8@3Z3K@e+3T
zNZFNdYb302#T#oe^v_(qvwrh)37frUUi(PcIbi96Ln|5)XOgVjUq1@^Ps~-A>m*^n
zG?rf<=!17otm}>}^ujqF^X~s89OsT_Z%3c|g7ysR-S1`_yqEhvvhLMD2^%!?bKTK@
z5;m^GDvcD)de-&zys<6E*RvIaOuoD-sAm@^6>N{Old|&cOoi*ep?A*1m1q0bvliNB
z`W{p2S^ba5
zZ}g9}1!gvQr$pVRi_YO-3G?rgKX4b`>77`b?Htzv`giC3J))0y^vu1xcMq&*Jze!b
zZOX;_DHlV_(tF}PKIgl(Uf=84E}8nSyLi8ET=CH=?c4S2{L~DklV9uE@Lw}No*sht
zS-ce0=GB0{$LmExbUm{km9;}
z!BtZxuRL1MuGUVm-Fdp6h57!p&T6V>8@uco)KXK*Qr~@C(SE#?rNr*+us})1ZpVJ|
ztxS}%lVhHA{Db;l7Wpj5dm#3M8^TYm8H#rbd#-ux=`Ur&r;ooK6@m8)2Qi;-|LWO`
z57*-UDM*=N_TJ0SA;0QqpFQIN|4jAL8>0g~OGbV^)l16irsYj6)sr&CgKxFs&7|yp
z(#fA`YosjK+H+O-ZqWT!4E8%y&pfMIeLZ)wo^8rIyJ_*(de;8RHbOIaVAllpP6Ti?`ipjWJvjo8$8M;h>(U^Jd
z=v65j-9Q!BGk*U`{u(fV^A+q>z{vs-gd=z^A@5$jz8bx75h}mj=k9#VuyNr
zx#Nb;%rq&xC$lmJhmy~tN&ahjtS<0;cB=uR+Tgp1v-IzRcw2TeQ?V_dnO2*!tacotzN5)+C
zOe(BNlCi1E8;=>Eld&1a-IbL>WvriWL+|BlWo+QM@Mm@oGPd%?{>Hq6GIl+CeyuoD
z%I7&G1sC$>U_K+o{Ff
zZ{uChTo>Q7t}|sU{>&|G>foHYKGl%366
zVtnzTl+Ac<-daCI#*~(C8t8Zi`n!iNbMTbmUHLG@*6kYDgpPZfJWFJ(qy5*6ho@!i
zhEvDt&nVwb^TMecGGr{`xxJ6CNXCYoD(Mt|2jxynUDts0e?6p2)}vg(FWdeaa36Z7
z8178K?|vlp8943CRT=BG-n@59{9cGw)U(=rR~fU4FL+i2dVyj<*fZoe`?gw_p*1pg
z>!91SaXC_Ed?+ub&1WgwEINH}r78hiOwMYHRn$Edp$(V7_#=53V
z8QX88s<>JvV`Wd)WiB$2u?|B!*_7{>vaXGb?ruCUV}DW)>BwHnSk$_dkxR`Q*oMqc
zvDq#SY~acv&&oG9u=St(+&V_fScv9{->a|5nDw>r$9FEtnB-09S1q*X)|c8J)BW)K
zXbpjXHdn~lcZFBtQ&lo{x#xUk%WS-_KKa+#VKHdup`8!UOOvr4ov&S5(gl3a&(nS0
zO7YH{vC)-Yu&b%<>$lHj?5=b}z|`lkH^xcl?^=wf3(wlA*h5dX=JLt54eZ)@yPLi@
zW$f_t0P6#=V_jC;(r=3zn1N^QYp;0?OzD#AQ7PV6pR)R9)4y9XyyrAsAw6BjSc}FM
z>G*w=>OD%&M|#Vc=ZVFlO|dej?Ki;t+9$M&z2?z&Q8HHhbm+jP2{PvXc;;|@^y8iN
zHPua}c-Q^wu}&pux7S`5p6`#7v7jH>g-4)&Q~J)CFTG^UKGR>~-cQE7D^6;t7|Ga@
zT|Kk*VBR_O+vnZhco}mnTAgtT{W2=eXWF*_{66o}6qC7qWGu1&k*yY!rEKnNCHv#(
zAJaay>($>`#%BFoH`hg7#$H9F`FWm@GP97Zl@Ft(csFM2)Z*(h_G!e5xMy2rOwTY#
z<>Of?^9&DN8GfgpC8eGX9iol(U=rKCHVEVF^TI`cX3N+v)rpD}?|h{?7x#3W|yX
zAy_~anqdp@iPN!>zA){Qs{<4QlMGTprGJ|J76=p-2dDED9P=GHUIG^z{|Ed}
z4g60H{7()1PYwL{H89qQ{zPn)9C85sKj9cTfAs%GZjzjCD2LXCS9@fbtBJBZu}}frzb^
z7jBxLHd+|h^1=@a6C6$#2)hs7qQE7zMGlAAGl3`j#CBLXRrW2@(w0qJv9Y<*KHx+
zI}!ngP8aA6Y63i9FF@x@WUz=;_Y&mP3kABvVgcX2hYS@l)j}E>B38Rz;Je7n;XOdW
z@4Xga_xpmr*Efaq5xk>`*pb-+U)xH6+aC|Xe?<7b7(qTnMWD~%4Jcy2zEK7urdCBG
z1ocrvJzc~O4Ho1jCj|bKeN=D}Q#eNzC}IV*)UqP>B4j9st;49|M69=S7Y-9ukHat#
zJ9$WuyZc%wpP#(GBDM+i{9_xq{3=ro4iC?uDL}+xOa^iow1o;JVi%5PaCrBH3y1sr
z_2%$OFY4JMHrZQ&L#q=a7%XBfqet`n^Sr2MiI~IPQjQ-oKFx87<~Ix#5fe8a;q?Bs
z$|$girJZiaafOiI=t&|r;_zuyh=_R^sG$HN_N7Y+*FW*;OfDap`iRpte1@Y1MC`Wq
zc1!^xwyW=9o{!hVNt~ZumV+K9Vr+~xdWML_nNGlv7qOqar*S#+Nn3e7HsLB5(jxYH
z+C)c!E6nP6I-Rj8sAv)14I7J!7P0H6uJH7S`<_Hai&$x_dN#opsy*isUt7P5!_Hsy
zOo`qm>+hoxfI90nl!!kvrvn;X#G
zRR}7Kc;1enZLcnA1mCQwUPSPv#lsc^!@SCQzIUc>k0v}@!G0&@@7A|}GUY$NV>B94
z#H?gNwgfvQe*aCw~=IvDX
z&j>Sth@FTodO`YcOuP1;;JK`O=#c2Y+vj6Iig54c24(^g(-W755ge=5ZHY19LhZOm
zge$+O=}0hUhEfxykBuxqLt-9Oc%4M$(24NvK>VC@{Tv8ZymfC+a@}6f<^5)tl?}J6
z-GMi}U;U~-&dc#l&xqqcmM!Atef0haW+K?}t%1|E-wonWNqGn_Po?G=Ua#}UwdC#k
zYnv6+1_C|D->ycs(V$H=rY8{Z(z!CFn3nhu4dlSLt7Bx8+gK
zF;j?`e)KCh;%}V$YB9lGr$^?J9#zGT+>ZLe#k@bSoCY&McX^C%19YRXHpOMe#sa(z4;>^ThqF9QT?pjLT!6cX5309brD|sN{f&1mh@E
zjrZ3vDn5MNI-Whs>u>&UWC#AWrHA6l&NDFw)CgaEBbxUYjX_5^U%R&sw`c8W6a)Ot
zUV{FV+fV~M7NdjUuM#v9U>AQ{Kv1sp0v@nPNLRNTYCzw!3IK2_Mi0S8&lBWd=%8DHzbM;-!w?NY{#Fttz&LY9@c2o_b9Y=A18@LJbGvs9Oc{KuJ5aLh7XtWXJ;J?7;bj^C3
z{qbhG=K90rh(D~qay#40aV!p;&bvGK&Gp*s-#7DlHv^v+_rJjA`ZSa4;Q;(*I8ILI
z-3jz&yqVt&|FfLUa?SShVTSaBq$#hvxG3!{UcO
zw67dC_m6IJoDW;@+sW~{a@gEY7RvD*a(!OS>Ey7v{MY3;pDv+qynLMK$=fB1H>ikh
z;|+>&+L{uOUn9WgdTAb)&HnGGU(``Cugb^YOnJX`lG|e>*LP%&pl2~}P|UY-{vqCw
z@Z)&Hi&$ZILGCjF=>5eD06&-;06*ZnkY6`Nf$zu%nuz(z`O-cDeGYFZ5%c*=BT&SK
za>GQ-^s&GnxmnQj$&3mPKOmPzGUo9bG7R&>Qh{GGO2F62$WRee`Aj`Q#7dk5eBefb
ze=>n8O2p2pQbmYZldC0%Id(L$Vc(!F@SkrL@>6US^pEM>g7X{m95|Frp%DRp>iJF;&;y9>88Z#U2X(nMM6E7dvN|H(>ENxtkUH9Ts~`uo+)Bq
zQ)uN7vChkBWyZQvwSwyz^N$)#!~#!L3)sr#*Y!z9iLkHC
z{*DfS{i$@96TyC2yHT+sc5k>dm!I|P5E=-6--nhw|IxQ^ttEd%6z=3j^h@JhI}kK<
zFjzuxWYo1!#NXrSft3;a+K|o}1m}e|E+f9{k{;-Z*sp1xSxNMc*QR?BbkJNHO!QR&
z$=rX7Pj6dK^vE^KpAlXEN5**aum2ZUUmlLt_q|PNL?I>Od+JocU<1a(=Q}cO!@#_dVRJ59GVF>X{7in(ud^*}%-(x&5z)(_7$IS&Q#t&=2Tr%@74Xz5H#8F)&8qB8WBQl(Pj?|B2!bZ|c{Ef7cN*
z5D(7Ij;8XYDNhV({)gH(s!@M6LpPl#eU7EqaB(Im-ACVZciR7V~NSDSC_DBRS>lGq&z$Du*M~<9U9be_D6CjP1WAw$&x1
zzx?wE+t(pYRcv4Xn(4&Kb1h}zaO=5W@aZMU4u#mx+mqSH>jg0Tiid}o
z{g+^3h&b2x&S3s-dzn77zcpbY^M_|6KQVuJd4mMAx17N1hY!3JJ@*o$CCojBYJnN=
z3>h}h=Z!CZF|R+Glh6#xWtw>y5y~5PO&@Q3>N9!uDsSP%$F-e29I)-i|)$IAM
zZem2(zwlCoOF}6C)pJh6_Arc)20;Guh!+p1ujlpajF~%6|9~g^lNs!OFd_#(GL0vX
zpk|6+B!?VHf3h2JN-C@$3qw5L1JnE93Qd
zdj`**e=W~GQi^BSb(@zzQnQw0_Imj|{pxU@y?Q?{9($4*>N`nlU>Qv_&~8wcqiP`!
z%;2F4O)q3-20S-dE`g{>`V)ii(_l7F?i;L!4Fl!N=yd=uJmTT8!S}vmuzoA-PAYe#
zWedzy;=R8eUOf2>Eeq%iM0uD3Z%Oj-!G8RYr?S@pJ*yVTx(Bo#><^nJ$n-QlzjP3>
zDVZ!WSYN`R{J-%k49X^h7(%ZB&wm;mSLvX>Yw-F1owF)zcy_ev$bS3aJXa6)=YRR?
z|MeTrli4wVJ-tPGE^-k6*B{;h{@*y(4VM4^=O$v8^!)$y!v$^NSuV?6S)YCqwKR2EySHY{1ejV5qA;%`Do~t983)$>3RREsW!o%CP
zJiQ3KGwI#0@??c#p5I*L%aacxdQkbg!S!h;z>_z+^88KX;J#H+2HRCV_gm`H;0Uf~pEqTbwCzS-SdNM-HSic8u
z$}xTC)*`qX#QUjqz=EGBrJE!0KZ!7AFD9;o39jef*p6cHeY%Y*D}S6jmpw10ObZ9`
zMbsSw!8&bdSxeZi?xstqf7OijbGVWNUT8i3udsj>;RL=mW^Y-!1UmqEvFtsFAh_py
zw|4`fvQj-b--hpp>#4`z6}YgOYY#z10
z=-8*T#J`oMd~T!XhCVfU_Yr(_(@0Y_;(7Nb95N&x-ESd%PZnqzFz+bwTE|nar>_Fv
z8Sf7v{@3}iB|scFxptNf_4~-mg&u{#tt)2TZzi6r@t*H4aCy9b&j{#umCcruS%~pH
zzP9=M{Ce(}Vby4P1@KL=tN|&Mzuh~lz7zIa6ca*%AfKQ3RXr3K86_RU4}I2wZN?$M
z$Sossyr6&4!6j}2nm$RF*Mr`FO_aDkq-H
z)gAoSz^3&ve*)(rjv8LR)Q|qWwY!k(riSyig2%}TC|^BYkDGx0nHr8&|1cZ&DwodW
zErP$Ve=EKncDY5$nJkN80Sm)|+fW;}3z=&*Cr!1A{0)vJJo7DEPTy{Qhf`$FsH`KRj_Xx&X*
zFmykyx9j6f&(Qns5o>#|1B|Y&I=&33s$X$r3EKa7dy!Bv3@Eebs?7w9w>@0;R{)+HlNsY1i1A!(KhY5i%sFuOsV?kJZMMBc`)uvY
zvKH!pfaFCuShO>(9gzd)u=d&~Gl7Zam)mIluh#Pfb;5n#vH3@--<@-_;J|V2N=jc$
z>-^TZPwQxW9^Y4Op!(+5UEfOfr&k>NME2pqu8M|0=OYu<2vu&~*-dubg5ImNKd-zn
z^(XtQy-^8D=x5lMgrfD3x7mEoFav%uR%)U88@dG}$Z!0YhvOvhl+qFA2MY}qL51qM
z*AjQt2sKV@V}3Dcyu&e)z4uo!zy0v%Is_=ZzhFJ)XAKe*n4ev|wkU%7`Oi6E7L`{z
z8nW_(fy9{bewBH7mIK)b+o_*?d2yRae|daAdtS8Jg0-hEdY8@L@Rj>HDnGzq%I+2K
z6uV@Qd~@G}v4oq(M}Y!ie+lv8-~tUy6J)~-UK~w#A*P7)-S!;}J(jmJ|Gr`th6{b-
zIk;PI-y?(-345L_VR6yImJS%mGnn(@yfsd_
zB>QJ!dVt&bF+IRz3!E87wi2U6-2RT&1{7ArYXi1#;qixO&1YCYZrqSy`lokv7_M9^
z&+rXB2K$dYN?AXD?BvPQw)6PHf_XfC@c8XSW|#G4Hd_~B1H5tVK8v{MJx4a}@P~>6PcAowsW|i4dH&AY>LIe=&bPaD@
z;h{Wx1wUSSs~Hd9pA=$t3-|Em?_(>^zQ`HVi{~YO@a!I)H(_$>MM;MD26%Ss_;~ZI
zHxI7`_SvQYpg5_4)l?18Y*2nOD5sG*^f$D4*gB|Z!-u|Q(0=_`UU>)6Cg=^Zn+X4x
zFZUUIuI(1D9|nhb<^SmQ!EQahZlL^Ndm~bmnLOkjPcJ>#e*6TUylXcPw-Vqv_BfDr
z2iP{K@0!aiXEYZ2afA9>gJ?;w6Z+d=7lbDXfHi~od9dH-=`}&WVgnD=z{*I@qt^j6
z$F3smCi)4~HRZj>MT6_&!C?P|2iN6@F+97rO5S`Z|Kr&w8}anfB0Rhh#6#tgy!F3$
zupJ9+p1w?qhjzIq<|7r+*Ysh=$;&t$Foc{myIGBn6K_9~;Z-f5=v(
z&mey^;`yWP6};!CS@Y_(eMuVx{K$F}!;f<78D<`$7m0ZFf!FScw^mG^d&-32%XxHy
z1rKYZ10CY=QVWJB+g+J{i_u}cDdejp=!OP-;G`S|jB~KY6jpBENE2AkY3I5zKK^D@
z4zueRKaANKbd6LV>s3_qUcQ!6q^zsGXz9%@(6R-%*i9rF*3B{Z0m&H7gx
zr@4j7pG2!Sl6`2TvOLxMytA;4WGl(p>!@6!uFHb>%%{2?TEufAt2f5T11~nF>?Qs-
zxjDO<%9W)ft`L7)vNJqtCvcUm%W8`6vAOmdWcTFnhz!Q>w)}D;emLjstf|BUTdtkj
zM?78eeBnlF|KOOhOQ@eN<%xZyAF3cew-5Z`Ww!1cdj6fST`tYjrns2LRPJIa?K%;7
z=*hze*1+nnIp(8)I^Fm6&4E`}?J^v|^P?(5ll&kL#0tgef=AhWF*=z8+2gmmQ$Nae
z<4QG`K%Nm8qgMy?Qr)lA0{-~?dya7`%FRkHZJZAM*wvd&J_9#O4e|L1URJI5Wd96B
zte3B`+K*x9`S^fmF3J^T_kJUO_UCYr`CHh{{d;PTiaP9S0?(gE`iLX@
z{}VMIo+O9&#XXxfXK6j>^Uu=y-7LJXvIvE~30NmzzHfW#i1Ss$^4bu2#P_x7ayH<{
z&piF3f>E!__@ie+(SOsxqi;<0a9+#RoVk<(ABuEeZG-lvB&(^MLwi}G0avc0ogD_=
zGILSBD#>Y28_o?DocZcX8SL}O+Zva&@%@zO=f+4OPH_o+eWEp@Yi$SECyBwd)0Hh&+Q3*3e+eL5A?-+t5#izh{1Df
zM~~TkzqF1!V4`Mcu(Xjgu?l;*{zL5MR_m^yKW+tYj;udY$2GZZ6nE~g<6ez7QTdB`2tOzMCJgU+
zOP`Iw3C#atSMQ1IP`_-WvZWFFabaCquK}>z+lieMw#7v9%R_bz`G!wvu%2^z6{}%C
zvdX`Z_OVijgCF^W?dQKOr}CWGM{aWH=R?(tZ}T9}c={@60#HXQ1569^BKo<2&I^Yk
z+{f{wzy9JuO+Qgz%B$cm2k)1qcJ6E$C?P*|e!*YHm$xm<+DCGgo{t9E9hN>X
zLjLj6-1@_0ZyKd{ips4oZ{JKf@qJhi`9-sRRbFKOwNdW_*&BFV1tEtYXgtL3RRum}
z1<-kNrrtZlxj>IN-x``1#r^Xxkl$S8Ec1qBkN5mjy@Bqcqh8qpr`zljmIP`}J~fo`
zBd*F*ztXv_Nn|??8hBoR7mHgzLnINf@qXXY2epWQM{l#Z`uq33JtUv3{v=9hqVR&n
z!QwO95qIe29@i=%h#
zaeWN?o11yx5sYloV)ftu)W_z>I&?0W8hFKnEJMOBuP8QOHm$bgYyrNd(j>-zygMpb{ZZNfm|cJkkKZK=^LU~*W{2zr
zSRxwd070JY?#x5W7@nPv5U;-zDUAyIHT3#`{nQPh^$Hz^sUy%$$jTdd_E!seytcv#qeA_fVd(vlIeF(;}~)$dE=fi_`SK|c&K4l
zLCpibX?Q^V1$Bnfb9wb?)AU0>WzbG(@SNm-f~P;flxMfogC|Q5@@Inxp4=|U>%adj
z-g~_Mm)Bok6JGi5!RP#9dGFt9E|w?SGZ@CRtH|e#vqYRHThnJFD1YsHh~cs^ymg@l
zFGKoWA-wW!C3IlHd7&0G5UBE$*N++FdG!S9@$5uHc+hwumEK&aQGkHZ1y~Uk<8FnHk;W!
zOJ2v;?X129II$tW7tEW-M}^B-y+$*xv+}0p(`2dq(n4FDu<+fwhZysdhbF9K{_UIR
zCDw1nJMuUXz(0L3Uq$|?*-s4{3wg)6eK@irPPou4@-gxb@Wk0s0==`fP~cwWv3d6I=KoZ0x-E`GJ1Wa)?eUevF==_a_MLcE$#
zmQ1o!QC$I*FLV25MzYd}3?xXvAB&Xtk-zV_6!?<*xv)5GFO|3LoDD~Xd}m#SImOje
zngc^cfK^v3IC}0^Z6Xp|h?{>k6DYpio;fX24w(E>aBe-~sMLffNyOV{8B1NG@=2Rx
zcafi7lWRMI;`7Gu?yn94rJQA*DSw{1WOpF7w7PzL^l3J5<3P|Zig&ks?Qj1>dsis
z(|0dQCr@pZ1y3tD8!aIK{oZ>k_=&&n2;Dy}3*{y{M~kH(i#@+Q?&T8rldQR0G9jPa
zC&>Q_<^Qf;eogW7@?-xvny)*RCxjxv2Xdc<{LqH{$f<49JJ^{|+}d9ZSzaWi-4ybT
zqJ7?y;2BMI|NTA*6fROtIFEj~>sB|bLoO;EI{5?m%8bC>p9jogx9-Q@ui#VCuFkhk
zL+@L1Iwcc4i9dUTuZS(y`PHa7TfiIl)vmhtXC?UOlvlZ{=76sx$mzetJByqhp12oKBPSm#cU_HDPSoeDK>TlNFvV2jBkK_Ahl7QfC+%0%Om?B&IzRV$0=TCUoEB=j
z6})Qd#$pwm6U0MTEV#6&hHG&d9&QkgI1+tI-FbZ-H?VQE(+$rCZg0_y)%Me>xTkwv
zJ#6)EaTyN|27a1T$HiOR$}rzo#|?{oCnV?Gz$NdQRx>2|A=mUyf9){)IPOSLf39J0
z9TzKXI(5pQI;_(T{+^h}u|;tWUv%)k4=BVg057k+S7JW;NgdamqoI0XdmUHr)fZ4U
z8g@@Uge@}$U%TRc=O^q%P6(f@xKPWrL`2_t=~BZDb6P5>7X3rwFKRiTzY8zkTMvJ^H8iVDqKdPx
zG&~S3QOcQC#lJjKT*~FI7gI^yP{Rp_wq_j9tK%*N`uuPPpUhI$xoL*=bII%e?nbPa
zW5c?-Y5gs;JHo!ZI)3cSK?%IinUMt%PUz2RgNM_JCx4!`S&QsNzE^&DhjpKMta8+D
z%=2Yk{~BFg@asLdm;2%#=8NRAN)fE1=?`jR6!1R9l;2LDcMInLg?)3sUqyRvMgI42
zzS}%qF{1(Ju=gg5SAJQIb^dm>ulJWaZda<;sE?oVe$B@m(nmXQGyX(}4MqDAi3{Zo
z;5Sy?P9omet87+C=g<1c$qQ!yhieavCqHpiICkiA;AqRAZyWJ^@1Gnu+UMO}2NgB|
zr)O^z5(j#@xejOooo`y^#9_TZ-twxN&Jh7s3oq0DuUz~6ktZPR89QM<&5MwRwg)};FU~}65^#6p#+$>?&(Xt&jP`{cUwYts%A-uZ
zw_k8~;2mzojN2#xu_usHVjr}#mV*N?r93jx&I
z|C;g4alI0Zx6gZa7fc2IVQ!Kw`R4-D`^XgHzUNNTMY4C?$bmp0KM^6q@sUCc1C0^uHV+8g@t4EL(`#Y`S^0o{WvlaB5KM=j#<
z@~rSK#tX_X6M;lNFy`70Y`floj=)bVNwI#sb$!U}^1DD|
zpii9y5g0WA#lV&Scs$U`j>li!)?r&iUoglsx2Hpd{2Z%_@Yn_(f8MgumB|k#p;_n`
z9RO_t%5LM0PwgCU{ORc~tRJGUWmx&clj|5Z7T|LO*h~N8@x(AkFSeeA1F_7ZH~Fi}
zu#*xH9b!>gwe*ey7#iNN#&PhJbzK?KTmc*X#o}2`t4*$C#
z4lfqpOI&^{M(8fxhcEB&J*&&jxg@`D!LbnN714wagSSk6&quf@B?`n8-&Z7t;6Q+J
z^si&<(nUrXClL5IxAXRdn-^WOrS+M%zMJKVdynp5`^i3k78VS6Nva`Rr%?-3Y{>3Y
z=a}cz{)nlOd#T(fY1MxuyI0@BM1aS&=oyiJ);|>ZgY+p&H}sQ!;b#vV^x%K~#eAfF
z|2uSvAAL`;Qu@_Q+D8+n+l7$b%L|KtlHJ|;K4ZymPX83Wg7&>}>Ca_kr;;#&eQ$Cg
z=+AA^?~$CYue?g=_-Gye?bvpg%@F`vN{Ssokrf(@K@19V5@eN=VsvaXO-U`
zqI`$1pll=J&tp@Y=T!c7TWKNP>q~7;(02pg8gufF4D1YtuMzwLyXf;prsB}2uKJ$m
zMdgRSx|g6_@uc`;;-Tt^B0IvtCwG|1y&ng8gnD195AdNueGlb@b)Ok;>O|ZN7`t9u
z5_!gxaZ(x@&_7%t}cs2!SttaDD01isvPl)JhhB2bj3yhYsL_>f1use@7f2V^Q01
zKMQ$COUHuq`uOfZU?DdLd~)Z)oyo*QHU6aDR{+0rb)S}?@}Q1;Bvd7=h&9t(_)0mX!TpE|)Dd`b1u8!Mi{o?nUn|0i{
zXYs-P3Jsi=-F6A#q=(#tt@Eyz>s;d$-n6X08DGPdevz|%bH0H)=DM!*8u;RzyL;9w
zNlfEvgf_ga*U#tlTb8~1bh?3y4k#R$vAdBQ@-+FuqEUsM(qW@8uXiW8(;ZXWt2G~T
z{i!M&C;eB$Mb7#;{`vX>ZZ5a$x5$G_+?sL+-$&(lxGZbFiw$9yxj+jU+u%j%oPyxl
z-nfZrT>a4By>{lexth6&kE0hPatqrYjZ&-3;KF*3Z&7?x#qnLtwca_cl(Ws!DOsLf
z$~|z=(q3m%!{r@ZI&EP|9XCZoJ;Q&Xj1#DBX-ZNn=2jL2Zu6g9g!up6Grs>Zr){K}
zVEeO%dl$aCNkpTVTRro}%JkcL++VSvgB;q?i9efkxv!8*N)%JE3oqq*KX{KW^DX55
zb_ZL(bZU
zx^c%G>Jj{YcyP6>q^`yi1IHaoui>1P!RNFR`Q^U5y(+JM)^WA7m2SO`&E~$A%^fFj@ewB~urK2I
zs5-98@saD$O7PP&KR>%-U!K#5it0~D=Tf6<|5;fVauObb-voEna2mH_KT74{{`jn&
z{>CuyA0vDDKJ2Hh3o6armLu+Jx&2uS{;tk7ss4xkn|*W1)nx3;9Pd`m%{XTr%1GE4
zxdP*Df1TF`Kartil=&X}bza2i-b2{09-qTBTRQ4Esq(ZlyRa{cqeSE5alf#7;M&jQ
zupj&Aw(L;sL=iPCyVzBR!Tw>(8~@w(A?_E>{dUyL#lFuGZJm8T
z1AfbO;s>i~@VmPfp6`*uxwN-oVFUVmyISq&M4TrA+jq_xg?^qa=*gseg3Bwvxe{+b
zw>a_GdSL68;+E0C;HK=UbYJvaPi+dFqrJnm*ZhHfq(j+Gx+k-ixh5zLOtAQ`copxU6qqUCHD3|f6sq^72k-Ag#fuMN=x@vcqoa#4Z+j2A#Ymuh
zS+0i;_2-yL@cnnBFU^`^jB(0dvyj+^^6`DwSH1ymw!C7y8g}(R($xeY7srfQOZ;)g
zBYpg`4E`ea-2#deN;`JTQQRnY{F^8Yd~T59NBywfeAl1i$I3rztH^Hil}!m0r`neP
zcBS|x_-@r?;_I$4qgi>l(NqvZ%*TTTx9PduEvNLUpIvuzAL&BAs8za{?CVZ#7ovNE
zSR;vciaR&=4QZx(Db36;pULk(yqxW{7P$FzU-)$3xQ4c$)NV^uK`Zrhb1uRHFlFe?
z3543S$FqFVK+!M8Pvz!1+mIbM`;QBid!6C&+1!CV#shwB#?O(UZ{FcIhvs48S82ur
zmpv^5A;SDUaYo_`_pMbntp1RV`SYpY50v(;pmF_jf4hjrIrnlw4%zjU^Zs6gu+$he
zdjDIVOpK;Dx?J9r{q83I^O8ui`y1($PZ<3;mGNUebz|1w@n`!O?^$F^zhuVziA`nl
zgpVf~&oanTWIShh=Rc<3EOUcrcg32?8+0{?(Q|iPo+BZNb8X2f#`i{Lk7m!sy1!#~
z^FBo)5smjO<|0aVs&`f~e$^1g`~IqUQ9PTM*_Mu5sJ!{(*O7#i40!h%)!l83w0=br=T75|-$exU4SLtUYs_xt{S~ZTiJA(gkCuJQ=3({P<4hj8>kh+1Kh`mx7o&28
z$w_9s@3+Q!%*KaT_>OHs%rugZKTBFgn369iMyPNs(TDKM
z@Q`G(+nLjh4>G`;-DaL6{gjkn$i(BG=b>*s@x2@NO$lW8GqnSR9C=rtIc&YVxZAUP
zH6MS$O5&Zt@oHhT?;@j|Sibd~=UA2xm92@w3kRQmcI6q_t?bV%Ccnb{dyI*|??oz1
z1IfR{-IAI~@~1^-glWHQ+cjr(E&RX%_os2BpL4!-Jl)srNK>(){r~=}=H*#Hmo2Nl
z-GrU*72j6!mjb@;Ly1oc1@%351zvLznmM29`4b`h3x5Abt(pB0=(pSGyVHGM^!BIf
zQJ!?<2VNS=Mpq>@E)#+SkC%2)6ri
zCiMxoAy-^}o#%CKe&VH78g4j8$B!`hq*cdBI66Cd{BGdBmY4o_s;>m+hokw={lfD#
zQlEVY-(Sad>-XhLBagZ8ZJMcd4(7Q#;JMrWdtCL|Qj_d4`|`!g%|o`tKjB=Dn_dhF
zY2f-_?`|Huu9mBPza>t5ObR!2>xRmS19v!e{o36B{L?v$xTirEbPKsJ{vj8y?yTXy
z{Rv)>{SZ7vbVE9#HQuX^J8{YVXI~xs!k$03KwR^751u)`O;3>-jEI#8b2958f1DKblG;KES#W^KU4v2ahT{
zy-MT$TI4}}#oSC5*Ku6Lv>WptAvWh>sX!Ib%J51v<^59R?&RJ9|Kxj9vZf1m|CT#141(U#=jQqC
z^gNfj%mnhpcl`m4hB#NYkL%ileeE)KLSH-l#454l3zlKOociy|-LtU!?Ky;B$OiW`
zV)GOxU_brW?`pmk^{$JN%;r2?5B9jb|m(b(0E0c
zw`e!f-2A~9>|4=X-JW5nS8CFUDmn+h>i79#1k4S)FG+k+P59q&^0!|yK8-9zzxF#U
z9Nq!@&KpvHHbS56SuHXRxHmW~m3Wuw6^q}=u&bKVqJ9H*ws!3|w?O}Du9_v;30+i>
zA^+Tvt{8s>_6i9f*<(T@|+3r7wN;P+ljw~oHrDrdJ;}6M3di-s#d#87;n7nK9z5N
z;MAl6+%zJ2BH0zpZD72q@t@=r8t<9lS%Ng4DKktwSo~30$?nZs&m3iWlytuN56Moc
z|GCLH;O}Sd2dO@ds-)$lS6;d?hT_-i+p+9-B5^xlsm_1C?c%oc!?#h!TDuamJtN#g8HF1lc9$Ad?TIgBC*GFn$lqXVspJX5*KO
z>}Nb@f#owM&pMpVc&PQd`Hb%!IDDTyC$rFp_2c+AMaEk`6JAS-eX-#
z5#u?lzI7Ds$^HeG
zH_$rTRqV?6Rk)X5HR-)J=u42Ddi$8~)ZXKi__;d36hEDxRPH2F&-!zzhyKD5c!tUx
zd*UZjH_n7pyNjZFatUoxhFzoe_hqr~X?rE+2
z5_`$6wR+dKDZqALLx+vPRr%^pEhzV!eJW)u{KaCAW)mtm?_B+X{Em43S#dj{gp1K^
z2iTnq7x;Mqc4J?+yy%8(c&Rh+Cd!ND-iFZcnWRI6->AZ`X?`7Zc{coAm`mnN@*nyG
zt6MJu&xsngdBQIF+U`3e;P*DA_WFs#&T{wjO=ZvzDX=#uKeaDb{+d3}B4yKLy05W~
zoe@kI>fU(U;*-Le`+T`ll@)CX>53Em8{puwV&R*U&0r^Gro4$VL%jqKC?#I
z9sT@W*!?pCJZYPqvQ}|1?m^!v-Po1@p6ssvLo2q9o13Ej*JgMFSC!L}FltUMH_I%3
zpSD#RH|J~h@W$;moVw_$BhRfGxSj6PXRh6ga=96eD$U3fjTzb~=9`ZHPkm)$(Z?F@
z%fu6Z$H+Eua#|F(B6m6tJ?qKk`Wd7x>U?u9{gjmoQF$yUu*%CzTS4Rz6AQ1P$Y#+<1C}
z^Eo`(JlidY`+W7?x3}TtoQ9WZWUp8aC-zfn;FfeM*JCkx!JL7coPhS1*ZUsc=6dt|
z1DEyZbCo9b2~#$gaI>Cm?kcND;i@CVtdd9?dB|IIB=2&?IvZK~)>(!R@m28AGJC9#9T{I8&u)FfnO!#V$VI+M?X%^2xd`~j
z(*@3Nf7EeSSCjiCXF_lG>h<>GBJPh!YNz?+C&)iGgbX_d{!*K~=Z`V=*J3f3BVBb|
z#H#(T9`k2%xs?T5&UY1a1D9KzEAQ8G>IJKGPM2bzr>ZoYAYZoSeL;?YS{fIcc76V$
zc?I0Nq3xS53+gw>FkrIuy(vYo-)%LrnRt+}o!as5u)kOu
znQsj_LejvW_|vc3dwyy`@3q|1q8{@5=>2ck0sq<$krgAkN7`!{u-54A=cVvF^LmcU
zEy4J1{n+_q8s^Q?-1imb{qBzQOdx-+cXs}(^^pI3mzAD{aj9LOGCu%vuEgELiNL9<
z-J5p9?o+79{J+4+ZMmrwKi;XvCNlogFt>=}#7nUf3yK@Ts=^;cfR)UZ%pV9%J665>}@dQaJR{R=cj8Q+l)+OV1GQ+%;Y@&Vdi
z(qvzu2ifb^_wQsc>wM&tJaBPRiXg?gjO<|iMODP%HD`s0U+%G0{7v^VXA=a=W&(TD
zRwDrf|8`kt4fW@;f9M(N$CajNVM5)Q62^NJb+$7;F?D8~4B4;xx?m5L_o?WAgh@D
zeWi)v5=iub_lPueh9IzO&xm9@z&9C~lGx)_B?y)%SFtTg%7G=C^{=D(*o^h#O
zG~+L+MV*YlJU9`^_=`l3J>z#eLwNUyIt2?@{f^T%vb>Lh<9@~)e~$PM#0)%D@;2j(
zOG2)&@^O9}8IRg`Vh!V=;Z;xBeE9D8F^lXSclERPbU{LkwO{U7JcjhMdveTae7ocq
zG2Ui*)p8N(KMvsMuD}IBkC8ybdL0;YlI)7A>tB&9akK)25$k+v0{dOIg0J36daugb
zr@zqqx)l_}&LNf+FBrera5ViS&5Q7_SZP`heq$D%A%2oClg3Z>`~JpcPOm+DiH_YZnvko^?rXs_amy6zCzJ&&-_
z#d&^W-tm(wVecUmu3?P&a=xwie!T(X-=_0P$pydPNs5orzk+xyIU7}si%gkL%E
z_XoG>|ICA(P_OmK)o9gmm2>y!C3rLyT;!C?
z|I}+^oE4(iL#2_=da!i)_KABNIBrjfox{#@t|NL%+V;2SIjuGO>>TUUILVg=1sXF7
zxR)yr>L$zAa3-5R%v*Vr?njPXe;h==SJ;znG&GG{-5PK`Gx|QKoiEn=#>$F#R`eDRl?(?(f3+89laa#M&?98;AJ$^T&>Fc>hfTPM3}#PgMDQWc-hEocrfL8)1Wfd^(%4TVDcs
zinn|0w`2ZA&$W6BV_mh22aR*YxRoq|{!9&pU&vl5s*QG+c^S10$No6&SsXZ4AO7U)
z!1ZLvuRSG7#vop1y8o*r{ta=Q0H-{AxA?=D=*
zC;vRta@EIqC=V<4@u2T8j!fAsvH|)%bBrHS`DjtyYO1$y!qeL;P`)GH+@8MM<)7bM
zY6gAwc60ne6!44AM;!TE+wqn7lse{5=XS7jvXcw{Rx1BCZt-Wr9ILz@8c&PCe@SG2Vq*SLs?Wl2nGluldHK?i
z;?^F~+*iauYIQH*C(+o)JNQ15oyT85S@12!yeC***SKFxmB)jA!|XCC$YwuW_kq
zFUw=Zbbk=1{yC>=JtRB+fB=?{Sg}?KAEcrGkH5?%l+fS6?2%guj^JEBs+6``h%OuD~4n;9wE9ObP95*n%!8!BOPT$gy-@D>}lO7
z`Zep*_ym>iCy+jB&%5U|{$?ZbN{W}mn*DL3hIQfSAxw4|%bx!ty9>9@sFU4I-L%s~
zft{1$Ka$~?&QukOYAC|^B|@w%Si68r@+e1~~yIOA2>
z>7P@`PCL2tKIH>O$Tse$=Nr=-@COkv{@~FK47D`*DDJENzW;#M|8yb#W{Q_%ul?Aq
z3)$#p&jpH?jmnb$=)0Y}a(R;}e_Fozd2A=-{;)d{t5H70-2X1sXVUK1q!0b-L#xEw
zVRzrR?eROvscUpnO;LVe_e)`^<_bq{5KRs*UE#l;}Kdn{7`~4#x7E-$#ev3*_
zyLNkhUa3QWrS5L@BlyXL5vT8sfxc(E>Izz~S{G&&)4n{ayLT=9{g-DRzcvxS;9usv
zT@rrie9P{wNo0R`#wI1i%U!?U9y@|~`F`o_4Emkt_OTjlh|5kBlJo&eO
zGTLKxkk>d{mUP7demkgd&JRiW-+9J1e`y{z@yo5Dzn|iCwEbL`CHB3jQ(V6&%41{d
z9wVM_I`dw9B>ax~`c2FI5zlYQx+nO+PXEBY_tu7(kG)4WRbxD+U%H&$0skJapDGQ0
z(9!EMN)d72QLH2H1I~lePr6F@&|c`-72a96#~J(VzH-Fw2Chc*MBCTB<=pQW6Z^5o
z7r3o;rvlw3;d?2of1@v5tm71GUPo4JZQ%T3?nR%UUdk1^{t7A%i{L8vWVa^^!+)pU
z&6pim$8D-B*)U%9x{`2SK@raj^H+;-SB2N80q&6o*Gp5gxfiNe_fB!k_rUYCIxc~PKyGbKne$}az^U6ScKDc;NUKs2=ewkTF;`_cY
zvp+i_E@=&~@w^DT(T?X%UTOkwmJQZg9gh7rwQBLC@!*3>l|uBp4&5dFvd+k#`G+LO
z4Ecq;r^KnNHaoGOUps6v6vN;75pKEs0qgfa^*sjhi2G7j_AAP|@LjCY&zjk$h<|JT
zh@BC~`uTGqiNn27`MD`8RDfsyc}`bHywsbKx_=+`hu(=9=@c)IdD_kThIPyL$aL*U
z=>Iib?wgDCI^?GQlcCVBXns567VO+EgbV+MT+|eKmi$Iz)VStO*u_{2+`J2UAK%nb
z(XiXCddX}Y*0;`N&O{RWPuDV?Q*^=OM~=Sjg}B-G-|ED(!0}l*;S~3a)y5h=!nj`=
zbGx>K2kj5m*)`Qaqw@ipDA$=__5oM%n(a=r7c7Zfjt
z99v^c@sgX|ZAkIbly79dEy{!H|Ao*wV@PTsAK9(Dbae>%3)d4OmXwEe{#9kV66Kw;
z;`uj$8+22XD4xHb8rwbhq&
z+5M4^d`lQn_><+o4{vW__2}64vp74G_^JXcnj>j$<;=cBDDO{~Wfk`<0Cg=`XfGKV~G;uUIR{(7`=~;p;{C{1)X4
zcJQPXc30iiqDy_+{;(=OmjfVhNL%#b}Y8
z#s&9ybzsd_Qw54+Ql}>~Ug5jsI1&t4_g7x?5!Szoa3?HQmK#g){=qa?#&43>zG3}J
zU2uerQ}u$bJ=w*`1+X}~x19b03;fKgpKRW0H=Jkr7Ma9*k)&T|Ie8=Xf7l0n{W5
z`vvo7;~aJIcTj>y49+!EZJJXtE}Lg3E>Qh)t1Em*!LO|DC_eT9_GX{jGRGtCm)&_E
zTZ(f2e35YShfY(qf~ddBTg{$?p&w^oR!`V47x#SpXNQGBKT%AleFrctuICNiTWeZP
z^`4A=Y#T^isf0YL^Ac0Nw5|9aG`k{yBYqF|P)SPR@1X{+$o$^;nos3iw8YJb-*u-s
zRWAu2=~~=#uAFpF7xxe~Yu!fYbVHs!ZBn3OEmu)G+pwx9jr(!2Y?>?1Q64Y6+Vc@7
z4vAkh+A|z}(R=UP+1J4{XJ-nYk!j>kcbY{V76H%5AHDK1;(CvzY&1W3N`%Uz8)o%&
zoaWz-pr{t`d*>rcGckV8_ew$avr%vF)JaW<6Yrd7oG(Nk+0WM9dR<;EH(FQfN^|i&
z&Q15liY-&JIP+Wg-aRqS;u2^1IX9^0aia4)>I5F-aCIKjt=i=?IopDf`TtTgxOHP4
z+Lx_;z%3WQ^y;N;9{#?s-oJaN9&#&6>W-&j9?!jW3Yp^lh%0{+H%0eF4mUz~&G*|;
z#oXu0J$9Xw^10^C(~sTT-^hIzyz7*Zg!5&|1d%IC9&!H?W?Aca7INNFN7BA2m2>B!
zOn)ib)Zur_{XIVl8#tYg$~3F{m7MjqEqlGs=W}uY?&m5b7=Ww&miwtS@%;Hqsd{(|3_J|vo)y@6FcOZpA(X-y{E#!ty
zG#-|H>jC#(_)4{$b}si*)h9r|vW$EDpQpG+PAM0=a=N=+RuOm6_S7rw^g`~)>x#Kb
z0=QRbkK4Xfq>|IkOLMIFubj)Y)Rz2yxr7Vd$yf#4a
z|8Sxd}625Rn5c2$UnErxE(ag<-}Aj{+9lh&#g(ET=crDnv*iNP*M#Bf84)4
zaPk%K?6q4g^?C}q4_5lmyU&($LZ|=y)7Xgnv>PvnPl`hxzxH#=N34(HJ9@8&nN)Iu
zE)UQe?n}WX-
zZcfpxdx&x6uB;QmdTpHe++BGx@&t>|&t9YhKKgy(IbZBMae0eA%4gJxol}+_2Y+;V
zzJ{?n{G&kWCP|cEi5nSk8U5U8*dda@hrGY*zpw)A+vi5Z+-k8uQnQ3iPl4x)w5ezY
zVPAi3?{>71#W^ff^;$vJIES~UJ*PPZsJAcAQFY-ry{>iEbqk%aO*0mml
zo%ad5%{~~Xqf3b9X~^5|3zXfb{umlw`~q3k!f_YH@gnWb`Oi@Q_IST;5uj$^SDPWw
z58YW&N^$7T-3?1ApISXF`z8JTn%GmLJjibqD9h?~pq_gHMO(=4cjj-%Apia7{=E2^
z(0|Zrw4(EwSznhQ#Sf*d@9o4dl){W&C;-LR*c>oHd7Af#rL!UT{VdMX15QeQjQ=!;
zel@qgr1%!}E<}pr&8~YnWy^slhy4;Ezb`Ck@__tmr-#{YvYR;RlP|@SmD4;em4I!N
zV=JgU)npmF_rLNcbAX=nZkN6=38-(S)<|*X^i1J%^!!DGmQ7?oCgOuO;r_F~7E_$+
z_mQ=xxa@ymZZPS!*E+B`D;`$=o#InshpPtVRc78BDM)dstz!lIj?y}6?MW*C^xMpn
zQ0MTxb=1xV2_Kf1czv^j>
zt3M+)nengsNp8%}fKQm=`{lfG?$!nk!}E!C4RZ-Y=FMTUKrF{lSqRrX&^KluU|8Vk
z%lOtnG`a!3Gkt~voZka_2NVbd?EuP2VH$w;ZMYr*N_+=x0lpfG>lxrM9bP*pH}k&N
zX_ggbyv2H7AiGx@9=LH5++fe^CEup^b@)2(?^FJJ!^h_B`H*+)z9!bNOPlQO
z`9~fnJOAtvY@UWpS7F$94Cxr?=f8Di^HKgSj`5FNNgFnvNm?Duu6EiA*1uVXbD6!~
z>irDw&vRkr#@pYsa^0c020?xHVKwY|?p)wpi&6TE}{{hBFuQ#uD=_kVO`6W3iuKvmTIKOP9(?%}
z8Wj4E$(`V<*&F{sMc`a+cykoiKa~{o9hXyV7;*@Q`)xbF0b{V(zoQsOtxrg1V2`Wdy}UT^6W?ettIgm!b|yQU>MVqp@9}-%X2c9vA*2L+VNrVmlfaJ
zt?@id3@Bs5_1ny9-huttHR$DofTI>ij>iBFeELSh`SorQ_s9C#x4#_3@&<2zxH$yQ
zamMFX+XCL+et6$m@OxK2UkFbUgK;j}aLm*g&e6izW2*bQQDQdFTEWX*D9RLWC^?*41PSIDdUJcIiIU%dc&!dJ3>`P*mg(r2FIH|IQ5jf2cuiRvg&lvJL5W
zLU=Q~Tk&fEHUpN#85S0(oj9vOgh1+toTzhF0FP8JdmYN&(rbl)QD
z>00#T=mL}KP4HWbr$;%Ly3zBwH09b$u%8&7(OUZue-F<#x*N5jV^`DO4}%@-KmQon
z0blLt3Z&t+jyrJ)*41A_!7)zPTG6|pvIy-TO(>+<&3R^T3(`1nQ#QY%1KIAUTDtRf!uN2K
zyFAV6(a%X?ju@6EbZM>lIktOk=yCw-T;ZP%L{q(&{3xpnu_);M{2AJY{CS$jhpL*<
zbg>bFNQlC!)E1)GVjUt7}w&t)oZw!DV-jZ9e<
zR?WcqPU+Lz@8KM$*jjAg9O$+L11$HCb)fl#(%|YJZHUYe<|%Zm2el-n-+B2I_R9oso`1ZVBA?w)cso3Aqae=4MZk3%XDd)!#J-
z^&rLk^Fw)oViF7(E(K}%L0?o;z>?Q4a558~NQTzVb>?HZc=@o8%_dLaBw
zH1BN#vOaU|`s%A~DCN_KF_LN*I9e674W7=_nvQt?~D=$jCR8FzN{Uq9t(7#
zDvH;LO=&lh$u3qGFKS0ow#`f%^V`v`_mPi&-s(bH?d0MYFn+eUXl3Tuz&w3(Qs3!H
z7m_avcZJr30>*G=RCj{kxxr|&;wPlv(AkiR
z*8x3N?X_-z7Y7yIF#+>Q7^yn1J{divU}NeBF8={;#M|3ikWP
zN(yG!&mP-u$&Am}gmSus@bgBdI;9Y7ztzZq0DjIt8e!Ur+o$6RFY(+Yf$>zL9>{!O
zZz|yUFG=|<3V_!r_m}WG_nne9i}}IiDb#ET_~7OA2wpFY*3}c|7jBI6%DDX0kXyv_
zn2n1OUIHN3FzP1bbt}dGB-}&>_OMnE=M>SbtBLm^@}B=K#=O&>{qX@@PITD8bGV$n
z`qEo)d*-Z3Aj)UhmUbNTO~$W5Q#ih?I;0%OTZw+O!s#}DXtTt4?Q;+b$Jb8b8*jlr
zk#oekPp|WYI*y;qJ1&6nfcsi)Z2w~}y#cJp*rIHK!yWamZp3&!9}zs|O+-4uuc&Wh
z!*Tfcqh|=-w7zJT;4>l`_C&ruy}=r|+-+|gjIiFD-%P|gQc1Ql!OQ;DsSxkmXr&(_
zc+s)Ww?zEUC7>gqAF~%|E}*;d6@q_OdCLo8c_9Fj1CDdSu_Yk$&h>;{Lb3Y@SyX|3
zAM9y9TOf$Q+jHInu>>B;RY0VB`7NHn{Zu-Clof{TA>L~honIcI(0OPu3APOozTpsE
ze%u7vgDl)b*Pkkb<1~=}elsT!|7$XqWq(65K&L3W%&~lr@MBLpZ~bFH4~Lg=2tDCV
zbX(&Hr`lv&Mv(+CXo_oHprSn!V*xg{gP#2(qfXpYa61em<
znn2U%^mEY%>d;*fZ!!e{SWV(1(6JHdA;?Dfb#lN8I3>im!H(XJ@fYdsO_3=i%0Kh_
zfgO&&(r3eq(R+31af~-_y(jh)FJcn_Y&VqtrU%w8HLjm_iULo2?{}%myD?LL?qTH90M#Ok~xH+4UmCw2p`d*6WU?7}V
zj)*}Ifgh}_A=+8Oit%YKZm3I@c6qKcZ`*8dvjKzuK2z3KK>>@RY43vBUo%;?~`XKMhX>dyIKJ^eI0
zXi)Hv$!V>4eqOA&OvU;?U1n3TA8ANB4jT_xF9f4Cv7XDK6Jfu6xPdwjcc%`@;&#d0
zc!kKv>CY-|qWqAzwipL>@E>l0f36$T#`Zdl=amjdJ+iB-}+7fV+xA*v4&}SALZk_6ob6~U2M_G9+t!solU_HLQC2dqooSnIkH))$vi!C&y4b-%U#
zwGrUuRa5u0`^LM`vB*p(k+0q8BI}Q%R5-sme^9gu^UFIT27TM%JhMvusU-gXIzCV*
zz+=9XMOgR8Uv4QNi$XeCMxR%Ze)RN0
z&0z-^mqpf3WQ3r<#s%)kf?cm&eptb+3+P+eF5NAVA1O;%xOO9qhePadIZc39_-kj(
z0Dm(6durV?SpVD1L{wCP@0^kiIie1H=P$>tU-iJJgr_`r|619No?iYU&;Ots#nt{u
z)mYnuj+?ZEFwFHJBg=mtR-1d#;cMa+a$Q|0yWH1bZ(BEF|2_Oq4$7-Yt1b2Ffai+u
z%o%g%p*@w0qf*7;xkF|}jVC{h!_s$CeVovqvi72_+QbqIR7RcuwSOo
z>LI+JB%~cL%Les{alS^1gZdeau70QkcF(OoV0L{y%;Upu`|V|5ehG#K%0Yj#EQmZt
zFn$~#6pSHgrweMkw_9%NMG_pg_g@1Dp%-%3ijC@nEBuqiqJ`w4%~nZWpARF&eJL{9++D^6@;3
z+xKVh4@E-zTOk`wImoBs{Xp<5;CF$O{~VQJ{$6zczwz9yuaVe`TZx)2}I;N=l39v)mesb;|%!}Dc`JoPI&#Lh2UfO^@
zE~8c{&=2BK_g%MvyzX**%U)>rF|Ac&V^FWgO0}ImVDF~xv6x!upK@`(Z6h%L96TfM
z{egC1*Noq^8PX{Qz2V*u@7vj#ZiwFk6`)85+@{MV<6@T<_iF*!XQMZmkejn@|)gYsq+11DNxKCkLO%v1vQY+H&KwSe;a
zAFz18g7~uu2mghIPO4n?rx3$^D2)7J>Qt=|!Um
zD%UrGc;;@BuLj60Y?bE60m~&4ObIb`T;MO-EMRNl>fDZCIVP$IpvSfkBZ(p`>FvJM>&YpL3=1@zKoj(`HpDR
z$9c$q-B1%c?p)l?dl1XLcPDzV-St-O{M{rdXaTnrPk2%4InhnDU
zy~^9WiE>Ad3kYMq5T*bkp3^9p&Y!yF37NTkklIL`fgftDBX~>BAuA$2#lnGz
zzgtp5AWsmT&pvndA?&Erx=NJux0sE{_eKCz1M*ql_mbdcmsip4I#miZ62jf>F`7J+RF=*2D2!^>fmW0^~fj_WZi*x?%Rp*VMDr%zc$5s!W?wG??=a1Tj}^;kP$0Db
zxjyJv@Jm_vVrSr!%kAKkNgq#KlJxySlRG?4f_OF#8=}0t1=v1d`79jk0|upIgTOxo
z;R?a|btvu_Sl<}c2;9#mNubwPdjH3!(C3rhIr_d=b%?(2vf4=3f1Slf=(ndW68Txm
z&AZ|J{2mm+h7a~3Nhhvjx$nm?T#y3iQkgFqwgakmt%4SVbLqd&hQFg>yR2|aOAJW)>1Los3>cKzcV(Joe7zY+0#-&~0J_C0$Dx%@92
zKR`YSQomU-GG4(0A27^^2=5GKB=qeY%ZT}$HDC!FYVgk@bt|!*1tAe9@IFdyozNld
zr{9$Z*=+=@iGC;K1-L%q3Lmc5f&CKTFu<=Vg#=;0JgUwRjMoYKsGC3e0DFzqui<=)
zR2Yc<4B&bC0qc)Hvt5bzPlmN}95`L(!ykvIAtiR)=p*{BB
zpYQJ?{Gucqs{qz-vxe(j0c~ur{K0y>O7E?)y$^ULK2E?oI3yd)(Fg0-jx_z-Nf2%+
ze`!_+*7XzibwWeHupQKm*8m7zjJ$pVFSp*
z&2QMSy$gzQwb=i=l}qyDb^P&~3`YDMCCX*iX&=ZJT9n`q?m<3Wx)IlNMTza6{eX96
zCi?Mu&?zEjh1=Odaqa`I*M}sQa%{IyvOF2*zxQ#)YgfQzr*&mlfoG2{Eljh+zU8!P
zWe5K5aka>zG$-&UyYi}=nt!_yzm2loD&Ql+<)3E`0AGoY8@(
zL7$ly13b`wdKR1qV1K|KIXQ71%9(N)Og;@fbydqoDRDsB)?dGTAm0s=?0!zrKa#so
z@LPlYHq_k}_n++3(Befv&Qw8aEA-32uqUe&;9O`W@t(eUPx}ddDF17>crAV(3iW5`
z<--2$Nx=rqV<6YRqnesR`h?bfk=cOJtQiRy`L$mA*8t{jeO$W8t<>ea?WPp?>_=Qnx*DI{046hO*~#Q#IJbb=WK3ICX)FU
zO1rZ2JBV-mW@)bra;Bl$v#WrEJqI7-dNFO0YAk{F-zu!^htE3(S_&I)f&KfvI{5@3
zU0UJn9o$|F{rl^hVVsnBPE;j9_^V&MwkH6CZv=%#Li_{00_u6do0NmcHbJ=j?G=}?
z-9Ihb;yoe!+@qt9v47#~{jDqvSQVD;mjI}Ilivo<({>h_YJ5)PzV-b>DM0VCr#DnF
z_Pk~>2He--+sFeOa$Njb9q3vcoZr%r$dWV3#jVsET12B5X
zK2RHw?^nWZEx>(zXut`uw2L`Q9I(*&wk-B1F~y?a_W*`8nYCfRQs3IriT&?ECdE}a
zp803yAM96#)ojMFtfW~=_{%Qc)nC*>4lI(_!{g&x{_6*LyoJuKRm6VDgL||O``4Ry
zZl+?iJ;BzD{nYjILbtJO{_6W#jB9=D4&wGH?9rRXe$;iOF-;Coz{{-&hu>A&9gFju
zkXC+z^A{60*opTSYQNQp|Mx2Y_g4_d-`MpGPGZ4d?>kHIYKvw|J{+zWEcX$Y6Y@9<
zm;|g(aD6&P-`&+l7;i^iIfXGfKX?SEzhpNLObXi1Q-@e@M+F1Kuq@V*vmM79pMMQR
z1;)+0;QbgMUDhUe`+?BQ#Clk^pss=K%nwc_?D(^l^90t5%r{$)vGJBAp?967j#x+M
z1^tiU@CzEV!1N&9*)k&ha=rnaIKw#l%VB^~A)15WJAOOh7!BmspnhUKrX*Gn^42^Z
zcwqz9>4%qz^;%&liHM&P6(i(}#)U+^^Odgy69YTa8i{%;X|5sSqu6tZ_`la~5*SZ?
zsgCtNm%JzJz3&Lebs)1Dj}Y-I{F4bRQrSSne>vGp!J{3E9fZh=|X;FGAqpO**gpP2(c$9$9HaglBD}^ReC%I&XJtfndnzeiEHm
zp&x4rd2J)m1PK3LM2}y+9}{rscV_4wK(zZGQSR49dO3+fAVPQq@kfd<|2KRi;yr-H
z;P5@1^!zjfpxHsby$fgsAn#^q7C^-cXg)wmR(d}EtssItFHXnT_Vj!#c%a%4{;q%?
zj&G&Iax69*(8Hd-5BWDomv8FQ@e(^7ANbSJ2%7`?QaC%nVQday@G`IE#nl5DNDYCS
z%ZOVKhciQ~Vysz4HY|b+gNs1fWh5SRKsd*;-tpyd=w$f8Io>i#E#qliEQABOBv5}D
z@$L+V!{8@SYB|3(L_!!BFzayqjZbvDMe*@b|mwypQLpU6s6KK1<-}u2xmmgq&_#?|mG#kiwa5n)?8q@K=`hYVa!arev_!rCV
z9=?(;tNP%M0{^NmLEzQXIs|f*)5jU3L@qJjKjCA2(0huvwSXHh0RW~k(d#)UD@@2z
zgY=#3EmIut?8ztjyHYW+5e=+{~Dr)x8A4A_RI3&OSt8s9jEE>#54r?$tYdk
zdlGj9e*c?Z{;}1U2|Mqko5P3#9xXmb_@Nu$x5Gq%_wHJb5j=UbG%RBvkFJO91K`o`
zxFcb_@9rbgx6fQB(r*>CCc^ho1qm#3IYeN8F76QUV_NtEb>LYmHWKoHz6OC?>gj$!
zbdFxn;J9os6rOXd(C1$YS&q;P!=Lp-cx=fU0;`Gy3A?ghtT~LsBXxJJz{vb@y+1}%
z=@ghr;J0q1#pCtu$oB_OaY*l!K-lYP`4cJv`>r2Hf-rtgF$u;Psn`ky1D_Z+BkV8{
z&LYbBZL9$sK0I#9j$kw_izng@(_y;-viO;&bi0|x2>;5ayNw5jA0Xiw2j^KlhD5yS
zYc`@CY-7g=|LGXC=@^dR(olXHuWwAURke8CJ?+mqhWX>27eepx`csgu=!|(x+DC0L
z5Z1e61~4UopCe0{f>8c1=iONUbIBPP2_TpFcjIuzF4K$n|h-XG%{iVO>iG0L!B8mD4Gd+V2gXaz*
z>M#OfeQYEV|8w9nafXi(cu9%T+5;pAUu$Ogvk`ca+d;wYGa$!*ylfr@@zk%($8kF8
z<9ZU9m$~nzDPo>8r|AWM$PIkVSK!kVz&F-rVz^!8vwnkTh4+g(j2~eB6RoJ4j_cdt
z&o|Bs*rLn)67!iWiq|Hwz5L~}Ot_sMYAGGV`lYNcdU&6utE#V$(>W`D^~QQa^H08E
ze0ji20_V^AX5%=n*NgSuQr`hXL@nw!0iJ2v+g=Ct2#ozcgv&F{)_#V=oee@x`T%C5
zO`|S=f)!^4upOZn_dZ?+$Qb>(hzIcZ_VhLpKt=M%AZyaYds`NodVS6AHJ{}b;U
zucYp=!1ZqUCG!lYTNA_dVk=-aO)&wFi$)=TxVRK}Xe{YA=3U%h!b%J4r^qJao{5|9FTBF_dEu=|!g?FhRc~Rwr09B772C;Uadj8QuZe@nxF1^8
zeuv_EypqeB!unGuR}Nu5e6djAO$XR}kW@8l4DxB2@$=z;OKs<`CjNmt2K-DO_H1o$yhNTNfMFVtb8+IM!iQ=atLA`Widet|w5n
zVFtHr(lcE+X@h*@l320*M(hrRAL)3fp@HKMH5<(0_-!1gHsJW5^8ZqCejMw54Pz|a
z`cfY2HQ&^CCi-hP&m`7&SfNDt*(ZMuM6h1D*wODezWt(qB9@2SU2bDNP6Hue0MNb%
z=ZSUaM7P>DTu$A3KEfaS`UJp20qw(ef#9u)KgKCoZedzIhSRa1i(QH3-kwWDc}znG
z2*3T?<{nWG?mrg^zortuK%}eq4Ac0Kj;62`BQgzyFBvhv&b;wT2Mxb)tsIHNQ&w&wB>@n;g~Jmm?TM&W!{iO_k^N4P}~%Llf06R6pJkU*P2
zSSBI-z)+kV@TduW-L`qnN62ivm~nzXw1;8<{SY)4V37$kfeae3-2{1(49yRC
z@&i-{kn$b67f@f8&bPFV(bLHlLi0oT$38l5*s2fB1+uIHbU)T#glYp`CWs=eqcAJ5
zoUBJjQFxmI%Ql!60rJMt!&?Z#4D0Y}x}5Qpoeu-|icxwW9o?4WiD(E9#Ki%gv!i4Evi`_&dN@oZ;`5hfyxYS1hn5ldE|zVU@#1p5
z@Myr{xVJD8*#HvT7r-sc;RJyEpFWZN|HydM;Pn6L{eMVTfc+Yw0P|}ZiRwc7|MJDX
zfb+{*MsSXV%(`4pp5=O8g&ZwI1nbRmJRV&*9wr}wlgs+8
z%kqDA{ipw5yZo2G$Flx^>Ho8r2!?$BYftm#{1l(y7}!@o!2z(}PNK^r)!0BduQsNS
z_jo~i{3W6Tz|Sw2vwwNNsQG~&za6{_PUpDX4shs0$oT6C9In(v&tC*jGmssY!{Or~
zLhf6Z?fmHSMGT+^AIlP0i2-Es<@E2c`v95$6s{S}dq28AnJK~(5zgK4;}F2!GtZ$b
zVgK@(?q@hfaL>YcAEVckbJ>67bMX@48Su0h$4~qO04#u~!WcK6rOP?YxP{@|y@M{l
zs;0*Wy`alg*>qX>o;0DSo%SI>Z`71KV5cwI3(%YpFe{KzWk-~V!0Izz=
zNZ^jQ&j`EZdk?@+1^=>T4vqxCKfdWO#b|oqC$SHRsa@lX*L(T#7EK(UGD)}Nn79Uv
zFgPbO_Jbh`?~ye+!$JV_MG@2iE1aTf5zM{H8vB
z0?WAR?GRIgI{^Fu`JFgMwY~Tt2J8vSBJ35P7$)L(FV+xtN4(D@P<^44BnNt|Psy`b
zf8}V#Q_M%TZT?-ry#I!qgFfbATgcxt@Vf8%=xhYWI8s9h*1vOC10HZ$317p|SJQzVxPv3CxAaHJ3DG44Mz8|BUCE7`8!HyNn
zHlxDe`GL2*a)SMe370QQ2GpL>U~Zl$Xr#sDKe;+$EiLRSd)
z&%UYV-&o)G#25pvcLc{2(XJgAU5s!z&0bv#xX&Wk!r2mIKa5arnK%`%E$V
zMavTHdLL=CV_8zbh7+Uq3qNLzmv8Ln#@M1O3o{tbXUZ-cW2`4v6ZyywcNt>YX6mvU
z#t3!zyK%tN=JTq!ooTa?r|^EEa4d}#^KBbr*-G3WtIw_k4+ZVMF}(!qpHouo#Q43%
z!wt8y;G2X%j4LXwz@foDY|k%P>fpKm77sYlf_S0;%L7^fwDi1B9giT*zl
zhV$Q@aQK;5&4iunr(O~L(N$~?jsnV?I6#c69a|WQ`62Pw9XuEG?~zBagoEGAg3sy!
zW%WaJvAwF{6^^)_^CUbrV|;I&;(+bg{P1uW#w{It6S4jtQ_~2Hm50_7{oKJaM~wdm
zZoS0#-W-+y4hZH=?Yj*aWr{8m^Rn6a8Zn;Q8XpmI=`}$y96$RWUgHN$$fVoTSeZ=t
zn-5|tgn#JY!9cWoaLjqaf1MR(B<6?X;%lN_3a$NwA5b%)KR0j5bJNH9Nj|nB>NjMt
zneZ=yqo=_Uf*Ve=Z?)0!LzFiSeub@dOdS!WE_q
z=#2_v#slN0fbJ*LuYkEA|BwhJ(*5uOb3oo_4&4X%Hq@KI_>HqpoXKVnYg*WD{k)N@_K5E1WIM_-S2zi}nxwF7veg7-XQVE6#u
z+yTuG=u%7fpC6?0i|N2KaWev{w9xhJBJxf9iCcg;9j%?@RP@`!~|%54d?Dd?hXhkoar?&~-T-C+;?o
ziO&}RTd)WSmwOV3|CJQW{kV8QyvxIAyDZz{;y_LXFM^Sbn;r1Tvfi=fe173(gYbt0
z!H0RfY%kk#KHbat{Kx-8usY~7F2@ti1~O(7INWR*>z5H;+91LS+63|cA<+yVn=i+&
zUe*U^Lg*9bLOdR2IA5iJh6|QK1&I0P;Iw*HCE`dZ9-p|-fwB}Qo2j}z$kQxyMkfR6
zQ5E&JaQFq18uGU?RH;S`A%5;bV@+toD5GzicsIHuwQ8M+Bb+^BDBy=V7*PV)YYUGSZg#QVc{`r-e<52a}<_MtR$
z!IQpCFkf~L7HHUZ!9Dn0f^1!bhA+b$+KH}uJYjD&A41NLb}ZCt_QU_B
zyiYmzxCbruw4eQPw+EfPnSSiv>rP}ceD7q~zb-UZI?r$;yBo1OZ2jIQ+=FDCY=0)m
zb)(CCGVIn1a3;2Ui@Qi#JMuql7k@^j3)x(p<~ji1PfEUWn7`iIjk?Q*-cMFG!}qE7
zhgW(uppPWc)>@w$6!zwDzSSuFFQAO$TmlCTWd~LZG?aFt^tWT$^Cw%-&U;&e!bj`S
z6-&co?6pRx4rtCRWjDZ_hC=CtMt}+<@Ek#=9;e)$o^N{I6i^|sDRFv^nd#{~xA!4(4nk*hn
zM~^;4?R^mn4=rZyYgq*rqGP9A*T+azqFpmORT_rH=oKYerv7y%vU=GRWFVZ5zFcaT
zIm4WR9!R-lm6VjDx9z-Ek5r4{f84tt3LMHsww6rntNU`$#9bw7LuL+Y<*hYxlP*B-
zr)c)d%mql}`GeRa2D#|%*%#(>8?un=N(r?+y18h@hnBSie{#{b*N;?n(+kk^FAD-W
zeR+snxFaCjNc?o%FC*S6$
zGDEp&cJET4dsH?$#uHlIb|@RU&c}UNip)lLRFX|&#IjJ%UDlA4oGi2=z?e2>l#3YB
zuUcG|$wPB*H{4wpoR0b@(hXemfS7rM?&BAm^!-Z#m7f
zQS{f%4ju>K{&$xXLU~HL$goAiLgIEVI%RUsp)w*Dt@U+Rd_u`Xn!Ry`DWZ8uN$AGH
zBfVTCwn;Rv+9wCO#u)EUT7>$Zy{q@kDI2|Lxt{YK{_iJ7%;#dSRSru2c9?_nR1Ol!
zyd&wQlZ`fiWo=Rw&qK9|JJtG|_lIJX`1h#1d8L&t)RLiux0jr@U>9WwbSVyxr)?ciLbaMYocCkXE2U-j|b
zgl;y{G>Bzg!<&hiB^+eWynu8Z24xI`*(m${UhlDpZ1hF^f&hn34r(qfKC9x9gDw<=
ze|oTX%ZeJ?OST{ekQ$GoD8_gK}@XbfPVKJTt5qap_3?+&6
zT|U}(M&_+hTp{`{f7s0MY!Uig8Of1nmxXeF_>)c@%tm`wDceYl<)Y21J<6Ry`6ypZ
z=alK)0+h^rcAPqu
zvbz$JZ)})?_*U2smlr0W+s@j^lR
zAo>KcD7mg|n211ShboTfeDFsd>juajzD3A(?&eLmh$2*RDL>L!s~Ba-e>q)!qXf13
z{L+t5c!myzR!Ccg7NEiz;U8Cd3z5pR69XO3i&1i?$HWlVGt}vR*x*WLIm)`?p8BpN
z8%3=CpyB%?8wuCx8H|eMBmc%ffqk0ADBP)VxZ_4Cy8G?@sQ#TO6dbWQUz!?;TEbnm
z56DF$&CPF)jj1N1G@)UoR5^QbW
zm7}`XBrBJ#m53+!&*k1Z8oHxi4Hp`|K<6(UTq{;y1OF3l?Cq3PirlZVz6&{Bh6FQa
zS^HxvP-JU^vzUJs5?jaUw7`&r34gaA@s{+CI!*@BW8uY^N?y9+szZNOOd3dnK1KIB?`S)w*Mn*
zEsA(rbIp)ehmLO&32E8Zgyg8jt3&evHP{5V{VT$UVuRHW1*Ry);VqrpZLdS7D7V5Su<
z+!??L@utS!vcCZRqX|PWFM0c4YD_
ze!OUN7ZShhbj1O_r;&IPbF=w3y#K-O$4;|sN6b$8vbVl=paZW;q#|;=P#jBy;L9iQ
z-eyC5YxnIQlq1<)y?JE^%1i6oxbu1^BESFsS1qL*6$!K18&ve5@Yz!@$d$dwHH4aR
z9PB)Mf;QYgm4}!7CK~De*wwgCX?r
zPKMfD{$V7Us-!SN>OrS`^e-~E^dO&#tbU#|z39@yn#{^mz3AedXNP=jMVX6H6GE?02M@(wB={Ag%I}Z+yJiT@Dj2!7?i@lA
z^Q#Z-SUZGvn1-dP2@NAJyY;qx&cn$0kb?Nll_RJmsKVUIas&-57zXI6jUfIPKY9O%
z_M+e2vMEc@&YLU+Ic1h$pL4Na>;U{POhN>&{d>5F+)0yN$Lnt|3LKu8KGEKbq<0J2
zM=`_yHy<&z7s!O?(%d&>h2T3Y!>NC>!x95%$Yk^JmGAly|H7WnA?*D~EEc)5boZj@
z=blAtSN5VY3d5TYu=~-vvo#UTuTb~i;?+_g2a!RM=CEVb0MgY?3JEIfhyQIl%-N+q
zg1XW5>dqg-XsTXcxb@QzlHBQ*CV%b~x__8);N##R`Yg0$aC{cN3uK+--PYKJUfuvTE&F|iXJUr0DUT?zNJMI>khL;a#6J>~+)-DvzfsqzYpTitrQ
zRgU|*(8a&H&gbF(2Gf&zua*|~pxW!)FGLsNy-H2V6#?)a);YDS2@-d^kwQ+AlhgBl
z#4b}Z_DjDXr7p5bo>1#UR@qO6?K*mpyS}Sm_g~mwT`;4rcsYoYmab1~>>Na`C&nfQ
z^ZSu*sh{M%KfUmORn1|FoV`d@zRFc4pc64BJn%he(19|JYBC2rZ-M`%88DAH(u~~Y
z3jfAmXh!?L_(_`|>Oy4J-z_Zs@ITMzC;1yQThVEqTXX3`&8U(jA!2I>_rm?I`^oOz
z4ez@?W50c)3ngFB8PQd5M=eo2p9j^N(T=J=vwXHqNK9Bd%ynfCI=!o#jj0~)bBSDY
z(Eo8K^4LDgreNNReBTLN^rSVR9UF8hc^i6>xv4&jQ5Kx5JQsKqN$EoTU!yhOJGCLI
zq)%H!Pcu^SOZBuEXhc?0dAjnB^~j&>)VA8W22H(oqkbu`LSoLYKfl1eG8-r6{z>aL
zqNdB!sXSTrD4i#w_TX$aGF>Bmw=?PmD!Ey-j%G(giVu}<-8l0vT4+m&vP@qgi>!zo(Q+;QKAP(n|e&WU`~S;l%srNWuF1-;(2%Xzitc
zSNFM8pzZprGFVoZp+5UVLM}Y;|KO!HpST#P=%V$bzV7mJ^mK7>v74zBmA_CJQCwew
zddB^8=i5urjlN&8AEHZU=g(w&V$2Yq>^za$|);m?Y-Hxp3gD(lbcwPXS7}d?mH)T_)Pdd2*>@V;T~@^D;akAQ@dq(Yf(Ws{qLy8lAXon~y%8
z_~#s?fAUd856N$GAr)!8y=pJ?Q=sK0}nSB7sJBl
z(5pJJpKbn$DE-2c)z;2rWO41x2=}EV^!S_pVrp+7`j=iXTY4xARd{6|oHPzW8sx3x
znF%pSvGdd3N^&A<&2ib4tXqO)Ypm<{-N;38mrlZWS4iag9VP6+l_o3H67wr)(Tg7Ofh4#0fxkOpYLzIFFr3dlw
ze^!Q>H=8!WKBw%p}&xc4vA#6mnNQ-ryGwtGs5=O@Fc@`Kx58Z
zLNn0|_3a^|$(d;XkC!7RzysNL7oCVal#Ys2+oi)=GLWmb|G(*y6cniEEjF|*9hu2J
z)()b;cu?D+t?ZD8c#U^O@?XzF4W>Llm-o@eHwmm_*jFL!04
zpP%2}TlWd}HMtEX{_FD4s!NJl>~5Ke!gl+0RTb<9rGA`!ZUXy{hbg5!
z?_qzVY!GsnHxJQlvvw^LunjCaR;)}QJ$cpLZWUQ
zvb;gg-(j1ExZhgt@na}PU1vqF3{91yPN64KN}Dr~$(e(-ZY-IoN-X@3{B$ZJuTh{#
zh9)7ts>7D13@L~ydS`Dua~`_q%FA{O=Gl}q$LJm)fE*+%d4B#*M*^aq4ix2PPeI?*PKj$wmLcW~v|g3WEO>8hP29X}2I2|&>{G3r
zi|q8b1V(mdqHoW_JRgpyAVH1|b8$x#QC~*6d+k^{+Vx@6X#SR5)N%J|8`JI_#I>Hm
zH&{3i73CNi@V?DQI!mn)cffxPa9npgZ;^`z+8iR-d-Bn^BNO@0l{_@)nXBlqEepPT
z{uXL{Jri|3t;
zX%(A`Zg4eq+{({KMTS3e}kufJoBr`K?xj*s$oa+kk>f%N8|=7(pI{!oitF3Haq>Pdnyd)&}HlO
zHMvVk2lgM|mBjFXG^3GGlD{8bUR?bCWXtAfBuk-f30u#$lZ3W!zMDOLt@6V$bLM&8
zZj$Erq_XkeYEq8xPL$L6jFji`x9ebWBS~VtsWj=0Iw|${IWc}xR;77BKvanG9a7X!
z^F`AAq{_Mei;M}+!$|9SpX#%zdyvpW#KHP9gVd9AI&%7ZDk(~=Wa5r%Gbv9fQiyl1
zgtT~WwEFnDQqq@f#>Pw1RFZx9$@Gq=*_FjAzfSmv7Liybzilr`4y|-~rEw$WS|`b%
zIO?VpYZK|4u$h9|(JGR0xo{x!#cEPeK}c%RGhb3ooW@S3>1L9c<#0mc=}yu*CY>4$
z!GOw0Zxc=Dq>M^dBgNv6S6-3U>vn3~O}$Ub%s=?7O(wB&`y+SFn|Z|~eW#NIDgN7CS$3@`l!fORsp^ZfFjeqT<>DU6-LXe$q)U~@wRolKNMXmMtwm;jkYuuw
z9v6hRk>vc9rSH+GmBDPhom*aakbXbX^@<$st1SB@`XWAj6M3t{#>hQ6d8C_v65NVn
zx=6lf(29#K9hHSGo7TT^nIR21j)$%GmL)eCCk=lIe@9}pR&Z33si~YVUQACr`V&bGJ32azflzTT+-^PTv6k~nHZu*zR4l7^6Q%&
zWJjgvTKq?)sm@YLvZ^NXWM11m3(){FwPwdlksJI+$kZJb^NpoD$u74BPPAz4q9z;p
ziXRkKBS&)QJg})Wrxqo%+9>|trf$QML{n{l}
z$?aRdwQMl=BFmdvIQcbdQwzBzlGYr9chf$1hs80I>~}YYBdj`#TJ$hzQ$vp?Ns1qKi$tBHsS_>5sv(#gL;Y0VMd&u{(N9ggKm3H}!Wh}<>c1QM
z>OZJVKG;HK*vKV6?#4@#lUl8!=)ID{08Omlp}9HYIPGq0n~D$j
zHdU*9)g|z-mUyrixO3JP8x1_H9sHG+HKHL0}JD;ZU
z?+o|5_+Cn{Oq8GFqez@iNGBUP3=hC2v2)W)?Yc!#B-}T`R7uTj7d%0BA
z@hzA_9ZWmx_$!c-WVYgf)o}+3L&FzysT5O6@U?x3{#j;}is2~V<|hF(!4d9nT*rcG
zYU}OR$QOFh67;iwxo5l53T&TF	fEy;zd+aX*<#Q)No%?K|yB6X=<^zxibdrQ%7u
zKy=St$`3u8Ur{zUDC;|}-YY5ZrEpG$3vi$8qI|zEYm{7FKuMa^E%MY$pd?xN&MY+~
zQzSl-eeas;Ql`ILKm}!yOM!B#3@0ypG677ajqcqdEW0b+e
zl`5y|j3{%~FICdbglTUlp5JsdW2F?N%O=Mk*-P;!$&9$UIMYn>pJXtbnbQ(vrA$T>
zRVW*{CUjnj$kNnyC*B(ySEpreeKdOEW)P(}S~hz!)`(K3*1nMV&9&Ho>3d
zKovT=ng6Qe4RYAcJN2V#i_~l1U#^)A`$erN7diXmdkguTz$VYD;w98`fd*&n1R}_4
zJOTlpt|}Du!p0u+=QHFAtCUBfDf861-RlH@ub(2HGMl@9si%`F@Sz}(&+Ho6^46R}
ztfCoJI@!l1&qtjs8!w(V-NH+5dYmnLYKIzCcNI^YlTZM)Y4OMGBW5ddw;6+GUe5vQ
z*UM{V`7L*oWk0R@YJ6)iS?d$`ruRAiWO0@;hcI(@>L;7G4o9?4k;4veo%||Ie`U#*NT8avGx$)GY^bAJ)&pC}PSl2)n$Vb|_sLcKMtk2!AEDZ8+OtKpnvX2w
zJmB=LiG}R7ErO>jmxIcGApW(k2p8EaZS0%KC9Kr-r((_utlCEvz4u;0-GWT!43aGwmgA=~M5I0PbsZ*Y>@AT0*UwAV
znvok4pKzvXt2iJ0({4d67PHi47ttW!W3PPMy;qUCT4`8kJc);z+xbH3dTumXVd?3S
zj=A&Hq6X_TXZ^*=Y4`pl?=#y^{w&h**)q$LdPy}mXshyia_e=aKAVu&q{a9thkLVK
zmCAqaAJ)AzKr;5p+|Lr2N;+>KIc(?1MYRy{j;5BsCanlQz4d-w3F(-8ptXLSL!|*H
z^Zyt+?|-h|FpS$Y5Lr=?Eqj+f=Q&Bp$gX63%O+%HW<`-bDjCUMDIuHl94jMxlvT)1
z$ll}YKR7=;uh+Tm>we$Yh0MrO_1*H
z3pi!>5EGjjR%R;WLaZm26;&xtAkF9O&qO^xgXw1R%ecBTBU=_)!s)IqHS2i~5aNku
zg2B|=`HtEE!cixBvY_fK;q$oqx0HlhLh0=+2|CUp!bDD*<9J>hVKY_JF7jSIVM&mY
z{7?oXxL)^LiuhYfI4HlZe6=-vhLWtKAt~Qh`BnUkd
za99)?=wL>yV(#6kbXuw^0)t9rNtxlLErGTRA^$1zicXTo*RTSxN~)OdQIE
zgFS@9`qUfC373!?Z@sUOFZR|*9F11}hjtNE*P4q4Bs*$un3hz=DgP$y=P<6^|HOlB
z{kqOgabuDo`H=MeB5!+5uI6{e4ZEL&KYdHD>xXA+j8ASQ2!0bm_H6g^h*@T8Yz#a<
zKgyXQEYL@}Qzk6d+*wZiH*QObY{tj)&ZvrEnZH{|GMG8B6{gpTQdNwI$@S6d)84FD
zfI(yEyCp$naon`q@18T3^;_Y3-tc9_<&c$m-~}Icz~6Q~r9%vnTwavYey)gR>D_ef
zFtoy^2Cj%IP|09TCj^6(XGM_TMO#l)?nq(|FNrUXsv^j%LFaXjJPj;5_zLgA)$54;
z?D&j;zBI;U$Xqg|C5N<^xEi+WT*H`R*rpdk&5-A>RFZ@41MFJ*7t@4Of=IaJ8&TtX
z?3jCL!G~EFV=O64GClQBA1h=w_S}3Ph%xpx3Ooq$Le9*&ddtu~!^FEICeIj0ARcxn
zwPoJrW1RNhuU+px!UWw5AKf_biqM72>Xd2*Vrl90I&6~}$nxrkdXtt4gyWF_J(XJs
zc2tlY7f$VtxUb$0bIEbU@^}h8X?O!MLsL;tvA4b0uaHaAT;lP_8O1JvhngW6efJmg7xIIn%gex}eA>pV|KF88NY4Mb&OKfiR)r#&l1
zvNx4#eZ`kCO0&1K;%)s{dD8Fq_ZAzFLR!}RN>GbstXe(EdtQba*q-`j-N*qC#A3gR
zXxAct-Tuoh9`C|7agj8S{SkzDopNm8!6x?Gxb0T*%Uw)N*0q2QAH&MGxkPK~eh|A97&vAnLE@G#ytLC>m|u#Ky!EveBs!?Xf=qiAOO7b*j#NCr)beK7Y>(LC^X_-`
z!abt!tbAd-TAKiP5fJ9vL?#Z&w#7ev{w)WF^sLA(_lrV0H^Fm)nVSOctv;@K
zD-F1+_5R8fw=yWYT{51dtOk1y+wi0C``~=@+Q(TXXDD6e<7BaB1@|R1HG>pXffj3e
z@VO!`(BF}=KmI@)1{``VzKY9(!HE}^tgdLmlP48#-3qb>7l^b>O9vhS@p(jw#Na&)
z;T5PgXN!UH)vdzHp$;%1+Pa=`%My^ZaZd#8MZx`RI}vgcZ-G2}-!OKp=cdv-?draG
z1N&v_xE(%kI6uzUy)BvqLYRuFY3t+QImdTgy>dkm@HzaNO$mVp7yk3j@Jj$gRvW<~
zbcHaY%PzN|CIhZM_IUCtp$?pvt}v}XTL%YvukBm0mxAzS;pLT#A$WhWQ6fct3an+2
znMJlOA9tZ4+HB;7(B%c?o8QzB$fV8k8@GLfEXF_l@8@EmQLX8d(4|rEb1jNuY&{29
zdG|WjFO|SjX{Y7jH>E%#q&Al2OAjziY^E`qtArbdBwr+Vl3`0hM-;>gz;eCTsYJFI
z@VsBOHs#GER1+whzqsE9MSLsYaJbaMNTF@t-o!5OTjR6oe(qn`tJgVRleGt$Cva@&
zI}v&uDjPWYZy8|UqhE!P9Kb~2c8YFr6Li6sY+`M!=*95#hohR~@Y{-1#-&d`f#eT+
zA79}f5Rk@mW;8PwFd8*4r1~>KM?MSph45(Ts6FtirK1pfnJ=u6OVmQ;WRk)KhcY;E
zCC`XvuNZz@IQ3s?UnaDbF#DH!trs$e(Wvd-$$>3rf$D4b8bHdTc!J6pnN@A(4eh=$wRdLjYSN%PKo#gUNe?TreNGk$=%a?n37A_mYbaZ#pYPav_A7~x|y
z1Yyl~Q7E1So+@IG{72rwOSi_JD44v4SN!6y6prt}tD8OS46}m}7vZX5Eb0RvY;Wk!
zd~AVjtyID<8oQxnbLWyrWE9{{(RwnT{0ta;yelJbr3?3YhUqzP3&T*+!N=DgNdnm$
zS2-lqeTE#Zq{w!AD0D=_^vu{*Gb0B5H8hFGHoVS3t|upqHFpzpb8c9OR=
z{95ksyT9cOeU!9CQew3M%f3dM_6H6S(NQCudM*jFM^{;kXT*W#g73V%t-&B~qBCS?
z&jp_2@YV@>tp-Yz&W(1a^}sf{472{{dGIXL{y@y#9H47pvFy>U4W3Gpg)SI+L#o^l
z-)5Vy03jw5vZ+_ANXz(w;-7DuSfJ?_*{|gjNX*^Cm6o+uEbfY4pX~`A*m%W?X{&+`
z*qve5@D%@!FteKI6kTY*a!5^@XHGps@=u8tV<(%DhQXqqkOWRxXliBb`Q#TuUwJ@1
z`m+lg>fkIrq5U11Cs`HDP`?9~5Gdm1aRWr|*zzesM!2jjX1^{-0&bE!Y4$l#z#IdP
zqjw$tFu%U%+$T##cv@hefvH&m?5C>T@(~b$V)41;j6U<&HTkmc3ZdUvWN!QWgFCM=
z)oHE(U4|&cEcoL{=cz=@wfc7Wm}n+4OR;ZjS5StmWN}If$$i8wi;%^Y?WH1*T&Az@
zvd3cE26X(bB%zq69djYQ{s1z-u1BjDjUgKD4f&@@OR$8=h93`4eMD+}Y1y;>I$_$H
zUSS#KRB+_U!VP-b9E_-tUUI>%710!G?GTshz$Oy7l(Pn#5z7maCe1lV$URkUgz?-j
zWYpgC!olbwA|kf;mGbr!hCkBw>^b(6E-SE|81MUl@VR%BTbin49ar&QIVm$t+}G*$
zojG&lH&1NN{c#KIelwZbsi-iFJ~f9l-Z%jH)1F~DMs^=_o0_;@aq1H0U1BWqJ@z4T
zb8TmzNH`SRU6ngz@Q%R>JdlzUau+0Cl`gmMVJzd0nh9m+
zky|yGziys^c$uT4d6!;dDJl-_Z(gb*E6U7I!bw$-wN)m~
zRA7m9xo*B|xXXnU9|ZZITv{O1=J5vZ&JNd{4a~4E^!Z6}D^t3p!yH%B$7@*sr1(5m
z?>*V?q%%R-2$In)jjtl)Gs{ZEsy5azznU{xJLn-aE2dpOlD>rSZ*(P}idnDO-8B|*
zTWTkqna-%@KKq>za^`f*dF*@5Sx$o|M8R|jtM!)Y1uAx=x1D$OHTzCYLN|{}sK+G1
z$#ik>b@l?G^OPK**)+rCHGXvUy^uuyyfZzkv%OEqWUxo+=c%y_;;doGOlCxVwZ!A%d4&Cw|F(Rl<5HBRQdh7``ME=KgfE9zLFs*QF2q7rtJ^T%`+e#riK`Vg7r
z6nrVca0hb+zg_!!3=r109>YaVE?8Xn`rV_RQY7qS?!|5F1*UQFy^3z@GX$DB$(^ct
ziBSlcTX|+ABY`qxf9vhjk+)QD?5SP+5u}SV*2};ht4Wq~Dm3yzMz@7o*S&qQ+pbn?
zEE~L=M#|rL+g=SuhTKzCDcIeZ5?8AS)Mwfc>?=(UgqqZuzKXYNS4{KMeDZ!GcP1dAm`Z4kE_Z8|3Td+CN7aG!To|x&`I|-Rh5E-_|cb}RrVG%@^
zjV%|>Ly=4|){Cy+kv|UM&diS*Fq5mE4oTA(k~v&b_<6Mfk@897@1^4el8$4&rTzn$
z%cLcPMS2Sotp3#8D?9Sbr!Br+b9J8H0C0B%CE-zMgTTVsE#$12|1;}ZQJC_(G_SMtGU(TvpTQ9^
z7*s=ivPSv35e+d
zhGNC5GXJT-4B?CPst2in<3n_yTevqQdl0uNCFulm>a9+N17|qG^0O?N{~6E|A~V~f
zO@Ur29m-~0-oRx1&5w=D=g|DQ0{wIDSWu?AZvBNO0~%DtmM1Sk*l^1-??Y5MAotE3
zZkEi4{&USjI^7>Z3J}ZPJlzPZLiaN^`Tl~?s-LFSFSLZQ;P8-dFvkZx580
z3ayv^JP)*07-kacmVgjrf((s)Iq($O;Go!oP+T;nY;ol$P+S}RcjDT2=$!o8K`3Ai
z=Iejuzq{#M9fsmADXYoZ&i3R@((Jsh4s|$p_$EaXGVo
z$9K4GZl}=1Hv$f29#;%BQK40XB0{vO#5i?${3F^{B20RX~2hKBzK@1RsvzkGY^nf$sRzxN5|oLK*NXMY#f6
zJmK&)OEyP}CVZ?cG;+9vbFfEWda1#NSJXz=Z`>Y(=W~TR
zA8PCYt0JGm2?=&Q2wb;oI^aP=69@~>Y(>$~A`fk;KNP4Jdwxz!A|D==+H}o%nQaCg8{1Nozn41-`QwF+U&2gkSSJdrM3}7bOg2ZGV-2h(pG3e-$kg
zoMS_Y`F0;Ae!nPcHfWF-SAXLtJ{(4lzwjmJU$I()66Dh(B@|kD!~?sLQsTa;1W=+Dd#J3~4~P4$*z;97;M&@MJsaIa5WTwd=KN?Q6b*nM
zI^!C_7;|jyY;`vd3r#IgRCdW%rhcfw3k
zz1Hgt0{AtWMm^O?CLG<+{dALf17xJV^Lmce0ax3$x6N5C&@6?6ciBw<9r;8Ow#_Jn
z`*z$S*8g-4_j^i)6)&=)En}JtSTO}k;~O$7xz`U@I-a-=kKmAtLvYf+IRV^!I=Gxk
z6acce3>`G?KZObB+E~Phd%?|a&E+3X3Glv?)w(3-XQ<8ReIda)6xfTudbhfM2WzSFap#!qnc!K{?zjlgLND0TRVcB^D+}na>n4%
zg-xQx24286>6U(@eG418@{vTcL=HY)=8@~POoDGVX;m46{2{V^?rX9=KVY)B6UxVn
z0`r+zVlo>oU|g9eAh%Wr_F{fqep?g^zBm|s3Yw3ETzqHMH}l*;*jSaUs<9JLY-1D+
zX`_cGUs@x+QxU=Gok^`nsk_L`b%7@tev_Cdm!7C?@G6GQ1#Z6-7Y0cYH@iGKda=_Pqrww!#||T?4z-%x7i``a3s4bDhhu;bq`
zuS-z6osa7J#{p!!u;Y_K_y{tm5gzhhZWzl9<8fdY??_O5DU%v(ad<|H8NZP%`;Of5aHFAZ2byBSVcG{y#M4OrlzUaedoXi
z`SF%R;a08-h6}66mEMlVI&S1x-}?I+`6sp*Rb^(25$WLLslToupH}WSDvsJ9Q~ejv
zN4nXV{btJ&?U^)eLMGsL+QnxG^ID=r5XmEKW{FYv(8Mv4{E-<~
zfm|f$@S4ZjvUKF#sSLB5?N2cK+!Ag!e`e$&-#`POHV>k5>vawSxG~38yBkKvoJju7
zKJ~^rQ)JJoE%tBkO^mx=K$1Lz93i>4M}HrZAY2BVy`1+rF`VgyjG(m|cAAmaH(gi{
zG4W0}y>#g^ru?H=C0d#i>HqhUWH6i`G3h)vB)%Mm89j29vXgp*-HBOy#@MKhbSz16
z@>wfj-+ILyE*5iQXX4IVsyPH9;cCIzcVP@>3Y9Dv8Lg093g2?2M+%s?|Kvexk13`m
zV(FPcN{&!$4BRkGBEs(9UE#|3NQ_Vt@un;pks?bGlkZE8jF0yL##L8tBoW3PjO!Ez
zS8A4h(dY-@fL#wcwy+td{{@
zXgv1wcOQy%hsk}$C~LIxc&SV=>+nGu`2s0KvhuR3WUVS<_V%55PDTjkaZW1HgX$sX
z_4e*OtGGWhtMI{KD_IPgLtk&dxaxZB8%!xpOXeU)auI2XMn4c5iiqH(e}Tw{{0-Oe
z3;u}BJEy-n>v7oI<2M!T*m_rSZuO0TOzuO*z2&$
zMqOOOes7nJW$0`o0yhSyuU(~s#9tKlDc%YJ5nGc;^(<15v!HEU{ox3k$ntcPPNf3F
z!Ek!RT@V&(qE20;WQmOo*U*P$_|>5L3%qr)%zE~GhWx697yfedktvEbEf@PS8;P3Z+L5E4r7J5*^8-m=%aeY6d{(h{{UW;RNU$AD@VZ8$k4Wca*+{
z9{9AmG5$MH9j3_oI4b-4Km%_1`{kKQVC1m+Ywj^eqH5g5F8Ig}h(1Vp-}%b}hO%2Q
z?4%?Bdn2mq=$Oxt<#c}dOZgN~cuFj5UHw>5jfqJp|I7qY`w@=&&SmiNiztK4;5r~f
zd*%0=7n$I+nE3Ts){J96)^e(OzYt{e3I=Ul?t_XCmXhyM{{UH|8-fDbWx$SKPn*al
z0s4@bbiIy_1SN5~M>mMRz>iar>Z6zb0g0}r+i(9C!NAkSp}c9IfRzBb)>)^DW6nbj
zbLH;>8VWBW2(2%n*H`+Z2&Y(Jfh3K&>=#0t@0TrL#gwa4JG0^5SmvShlXR4UNT23
zKr7WRkA0m4-_)2Kwvl5&U5&DIZyAcCv!@9ok6Fmjhs+bC-#Pc;@qZ7Q8R08PfFt*E(r$799BRKvDTHSxHoS
zJl(ZIkp^$+Gry|(U>|y?AGHLlZ-9N?>_;zFucA^uiD0tEnOs0@+dRS=Uz90C`#tlP{sXI1t+c7
zjQ3%*!9AJBDLtBu(9gT`;)7Lm_?0t~Ht%UU(C^RKmHrdeLa$I+#3(Zu;g3yNGklXx
z@h1(T`q{l_@p1B>`IG+2sM@_N|Gq?+;111c3NC)CxY@|~oZG|y&{FH~na%In(UI>Y
zip=eI@oe*V-o~+J=n&=B&{Jne6bUvI{27Dba!NHCn`|8D!#MRz=f!nV{jMf)JB3wf
zN5xnjZx)ZY6`1nhy!HmSA7A{CJ-7+GQ)g3+cZ1PB->_&(+7w(}fq+{S6yglg-}N+!
zv+xJ%qqo;^Q`|C(`kUPLecZ8cp?7TbCaTFsZ`V94iFP%A*bz73!mIa68j>O&qCI1z
z{NcC(O4`dht6nPxBim2f8OlRXZRWIT!
z*E&c0c-3%{bJoU`&+g!Hi{5uSb~#Z(G!{s4GtS
z&0|#Nv3_ZdOb*`kuwqomHx*By)cyKO#}0iolJHloAQIIT*4w@;^a4%eeud2|a^T)e
z3im$?iQ`kO6J>1V+;}5;;zY+3E2imiJ~~Wi
z&_Acarw$(f^W<_sM^Xel+V?r}(R9al*$6dU;oEK8B2xh2+{>7^P0p#DFyt^DoiuO)FY;yVUH{gCLb;kQ;2Mi2Y?fedg92dXcBBqW3y^q5-XX^Z*#nk@jNVEU({?BW2`%xi?
z(F;m7ef$)%i5CJs)7yfAHSDz8drgKJq9T?F1ivKEGJ4
z_JT!WMFt^@SunJ{^+QB)8gN@`y8zs%aTWgELN}c(c`n_$wT+}sunq(XVC7`
zl>bhwEzrEgmZdL22gaR_9^v$g&?Sk7TGj40NSQ6HR&jU?zA-cW+wJgydtW~ch%9J>
zOHVQ<5_M$363s`isWE-X`T?G==lKZ22<+jD-@`!~hyF$B&pyz1B8^pB%obcZ*+%{2
z%RS%}tht|IQUFHMqhnRl%E60WwyfLnu~1SiBIL^#N)W{8`)`po5+q5nkG9fiz=iJR
zyODI4q1ezLzC>qc@Q#a5DX#4Vv}xH>bka7!*vN^`zc0`M7&C*ucJJw9p7p7^3kd_v
z&z;YcZlnfRJHjI_pFF_cf97xg_QepA+jB?XsX7meg(Q^xT^WFX@cQhlaR!)GmoUTj
zmkwk#jMhFVQUC!~vuOIPHsqUpZRPum3)q(HuL`8BAuaro;*xrNuqO&Xg+O83?IHHH0I)(4o
zzO}2hyjSNhb+1ZbVKLQC=h|2hz=8ye1$;ix_b_4UtqhK0i=m^rY@4b9?PG|vo^d~ie!85
zy~{>55wq$ysd_?E$hY!{yidK_80_^U9ohX4i6%p|OxxP;qT+7Rc
zn3d2^STJZI_YY6;={6c+nvv_yd3F8>b5?CZ&*MZyk;QVC;Cl<<`eaGUc+3gKyGIHQ
z%Qzz)e3@6K+oLdkyPKob?m5`u97
z0(qD3s0$0;B9ES5<@lYHkChauQ5uExBVz@<%yC+En4PRh+q+xY$n(%L%Wdl-Y(D5D
z-!mH=32D}Cc+L46dzwbdIB~TeIbb-Rgwch>VQ!meM3*
zUHL=l%_1|nzG#vv$MFj(l0TzckD|1o$C(L!0xI1T55e`jhsP4$yk*8#cw;i(MF40KQ2?z<&QDptGWbf&@~j{&2p0=CztsdkJ+0ZaE}
zeY^j4IIr|<=h72nU@{0r+*#tlz$>#eg`c#+hFar1Tiy=h(;<8*`?)Ig8~w}ZPvi&q
z)NFqSu!KSRYCXgInaW^_>8bmNc_Zj`t(>a6&K`7|#jNMr#lZ8ME3ZbV@4-D+`nIM3
zXCQHZ=P!APBw%5R{3@|%4ez{P49OXc1vmFKNlkqk0Vk{dSprWC@HV)^J8AG7wzdCZ
z5uJSj7BlvFe>}4X0jWe!v=`!`yF{lBD@7vkzhPl2C;A?`)I_4QWjVlg?8Cz=N)WQg
zW$_d^PQ#xs+eJgIrSoHowE^p>G}EX7|FM2RQB`7w(E_us5t!w!*c3%G!Lk}G(#lNFz~d2BhVB5Tyl*tU66%y@jSR2
zeVPq_T%PnWKhF~X-F060@fQ))AVSv7Eie_gP5k!U&9x3Mb{uT|55I?+o;;M;^Nu;T
zmE?!x!b4GSUu(KUj|w~)U3`<&Q-`Zlk4Sgo5%Nf?}6%
zy1(a2&;(me(jC%5^sx`IqV@C}G?UX|=!*RvoK0@~AiB{8=h69j1{mRJsak8-TXPJ3
zX;j+B^EDr>Hg1>pw(vl=@+R7CpE=_V#`2e?7aP$dbK1!(|LV}noUCaf-(vLe)DKDh
zwP-Z>jy)+_VS@isI|y63dljub9nI79;xdZgXRJ@Ub{@6klBnF1U5DAxn!^(ojQBN^
z^W^$s9O&1#rQ_PvJgC;#Z(imFdR&R|xyYZ}bZC(G72e^$XVD@r3b!3Kdwh`ZX4XC%
z54r&*jdq_&qYOw~{NC|tHs9;h_wEUjqLt@m8fyN!;jlw5itWA`UeTO8#_WF&r7dW9
zc;S{H{{F0B=}LndK5r+m4lONk{fMW1`qkbz&Api{LW2&9Qc4qBZr?*Sg{Lo2y}E~=
z=22X#+h@ct@7Q#oY3c^c-8b2}*GECF%lD9Y;}KG@A>t0{W$8~B8R|L6~X!p!=(cd3(1Xf|KN-iJ3FXv@AD_rA~$^!z3H
z!r@~ZbON_5QI{?7BTeNS!;vN~HfUBX;QkMMbRIkBkk5<~)BhQzKb}wRS@N0>E-i!Q
zR64uOum9n_s(2FBGkN@@`05r*t0=0jw)~kcm=QPKEA0F)`YcZ2q!e1D-USE)gIB6M
zD*<_0U7QnXDjXKqRT8}N0u=30-R0AN4%C?t_-SGiO8Ot)Yepn19cOLgVO{)fVfkKW)KhNNa;$`L5=l4Jn
zFR$q5?=?V9;AY|<!2t31P_-k(uGO|fRWVW+#F>C_|>R2f^=RGh@qj7eOU*T-6vj|
z@cIJ#XaD2fZ;ghfjV_1ITu;HyEK%$08e7;Uf8D9-77cvihXlN)ECYGYi;*neDNxNL
z49(M)g5f4^H$yh<0H?+kuc#|t5PQIxF)uX41fs!VA{J
zEmb&j`e687)A5?8;5xF`z)^`@KzfKO9W=Rqb;n=UFwvN&sQXV`*C#5J%8lG
z1H)4-dL3A!Ng8$ZiYN%@i@H{ovWkS+hEC+2ox{GYlU9CvIgSiI$QS>5wF5J2QsfuN
zWCu#+yX9woF$1El*@%ayRxzhN@yf!50fc_fPC-lZC*s?gtyTE)79!!=9kssXg5+dr
zBmiMAY&G+-s^J-9Brf#r*hcvh^83|87FhQTd8%GKD`@i;6PI4MIW3rm(AsCVyNFg`
z26ma7_N(ofgqnKfWOO_tEJUHZYwUtO$tYxS`4NJM#0G6Jg*IUF376gY!zh3;Nms=Z
zLkYrop4XVEv=ou0HJhb$XvB)uJZvc2rw~fvy#>|dyT;Q4t^H977;^VA{{u(18pJ`Y
zcUy^O3z@b#*`mpJh*Yw92b?~=h49D+y?@@bij}XXm9c-^LB1#FnCSK>sWkB$BGQ}GQHB|v+2chl37n6o$}Di`$fs=8yCT~iz~rDjxIr8
zfra1|d0{X{xGu=WcneNvmLHxFL!h&qBx-bl4QNWlD4t&AhFe+ct*3Yx%H9E=40AWE4iur0
zMK%ja(*=vPFZ_5L{NNyE=m&nTLU5#4o3YGi3c~ABwdVEjfxc3e01tu#45S-Na&d_S
zG3QA(S2AjWFP~h&$t*`8wc+}uKFbi?c+6o$`@|E1h8?OL=}h455naJ3odcK!16|wm
z-vQ?m4&jNwE)ZDC
zdU!di4qlmX3lsNm1hlzA=fksG;Aa7YnFm~bVA=Yyu!h(a6nPR=CV-LR-qcTb-?rC+
zb}~B)qhmg~-=)g3*nSKcNv&4VL>|JhI5l_IbPC*x;rXY&p9@gY_LY}W*en>X$>*-9
z{{#IJiGTg>*n~Oi)^*l)JSef0d&IB8bg&`PJI!QX56WJ~(a~ZP&@Y0uQk#txczv|eABu+jrN*sga?ngeF4W`o)9B7&
zaeTR~MfJlZk6dSl-n?d_FW
z_<<;)^a=N`vU*9QDYOFoFRc~vu9dXxIR!m*6Sjn~mDu58QE&bhE4{#Zot7Dsv&7KV
zWjD`$S$Xv16_*s}>T{LM8f`vsQ4>Gogi|N~WcUZ1(YPTHI!_bok2<#J@$RF1+fk!z1#l83iKC@rQIp=iXEt>s(hU-IbymJoNs#6gJR^
zhGx~jQR%8fyNYISuG$o%CzN!eD}-avpjOeMq*O!Pkh{yr|2P+E`AsejYkLD4Ve|g`
zEt7JT%QgBa{bxGr^e;5Uc|067;yMxd^iC7HJNZ@jNofOGH%mf2*K~~enaj;L`+Y&L
zQvZ^s+&s=B+(90SWHq8%2Gcz+)9cWf))s~LDG)6>ae?BkWHoxSHgP^Lq!RryiVx1O
zH=q*gV}hURYf;YZ)TBmsh|0>p-}aisP#=IUVUE#c|6_rl6(1P+J@ybC
zIB<86NmfP?;w|`quK=}KIV(T=Isx^(*giy}Y>$Vx|5G>}r;8&B#mW~&EKm#H=KDcX
zmFU7rmydhlg=i-DRx`01h;EJFy?kNU6+a9=KN>T61OF-1qVmmUN?9pntSoU?4uh`b0@z+f}V_R;2S#=tc*4?&9N@qod*&>@>w=?0POxsQC
z9yX|h#tC*?a%J>Y?VOkVHv*oK;^#=@#DP
zRraJN-@t}Br`2e)6iyhQjm>BkKwTo`{)h{?;42Kd+b`)PQHLP6liS&3@APxiRppcgBNKz<|`nVi>3M!4JTUC)mE#tu?f}v
z8ehEp_6Ifx50p10%z&R7I_8A~y|C=;1If7bV|7`t$H(5ujH)=7(bqW=AUAG{f!)^5<>;ssTBF;;2zxEbKkHRICyi1dm9CgnvEOhi6nu
zZZw`(2L369-5>X^!A-1|VUU^^kh&_
zDfOp4=Xhaj8`*`Ho)9P-(Kf39#`rUOTS4x$G*Qpaoq^v9_dZ0Go&=VcDP9(?ZeW^8F&NJ?L*PM>4@y0A6XcTAyU?@N6r^3IJpxU;?FQ`BV22b5;8DUpHz6)xgdKkt8TV01FqLg;9-M1=>
zbLO1D=vCStCkaGKCE3VQt*VhKUeWT&g-&exCR4+R+Xk}vd%`Zml^Epc^3A??tVLvr
z+r8R{oG^!_pO4)+(lCw7)s!o(14zuAZ69bP0<(0t`pj)wkn9`g(!V)MvFxi7D7k(C
zqH0G{U@F~*d_~y|L>NiIzc&F+UxfM*U%}1`6yiUzHPPF#R6LUi{~_t>6PZoSYlX(e
z;@1RrH-5Q?p?VxCb?toZ{rVp!HFl}O=4cD?%wMF_|I7d{kt>Q!Ix|9bZdJ9k6ee){
zuE?t>0WLT#cuOZ~MI0>8Wr)~3(}R2Rzh1uQGY3bu-U0JR41gkk$0*f@2?~`d3A()G
z0)K_A_)E~6@Q%glCsL^v$2`Hm3XK+i@JW-5j#Ziqj#Q5M<=Tq?qna>j_lvjTnQk(#
zDJ4%ZmF5=CRUi&Nx%KAsZ!p49t2@*oBlO@U#fxSaL>h)junxJ{hk;n;3;mTQq(E}<
zr(EtU2|!(cd?0vN0Z!YX`$<2|Ky_Bo>e`sSyznPCFo4u=QkrjCDucEMe4Rv!ULcXwiGJzYibz-Z9i#pFRb7Z1TXt&ppqvOrhv
z(7QC9I>?fP$n_ph9M^sf$kEzW04DErq6@Wc@Z?6~_%hE3_(65)Tvyc&)UkX)uk?i!
zkK!J6SZ=ET&0l9I9PU=Z_$cavLyl(PnVAcBcxT~z+O+n)vRz
z?1V_qi8^dx0F*Z-ZE>rwK+E`$5x>DR=*#1Lzbk)hfn-IGk+*jgkQvoy^;mR-e+*#2
zsP#JhuGMB0FmeG^5K#C+{qGMPGHrOh&%XkU-JG7(G3-G)k$FieK?>B4hgRb}sRVjE
zr@Xc@o&t{*`%zNlz>Z3=OqIyp6~YND)SDYiC~9VX%bVq~75eV015L8rNpubi=s4u>
z1^jhdSqAnLxZLS7zF;*eG_vSr!S`h&R5zR7^^DgAly}S`O|hR0zgl)U#;!06o~2Z`
zNw{*M#$2q={S{P>dklvwH5C3r8KY-+-3_TxI>$4hU6mUbzjQA!=vYq#Jv|V4U7&_L
zKVpa|b|pi}!Yh+~&vE0cU0E&PpK_r8GjyKuSbkv~mqe1XGs|9;l3kv2%&dRNmc92D
zA|tbsl9jC}BO-e~=UBdY^DQXi%$-$&iz9j`U<+iHQ&7uft^5n@9zs!s>&wJbuhIpoZlT<
zJsV|-O!LF`nzvhiY=pxXeTHO|`j7CPI36h@IgCRU9*K$@nZpU8692*bc3Ad)A|v09*c1>*v)6-S6M24*mT5FFfIiKYLv54lXc>~-u~N3>`nNAvI5g?V+))L
z_t;qLs)6(cRV*xD(_ob{MX>L%4|dqKyD}B*hJ`c;QiH=SaO+zTznV-ve0cuB?vK%2
z$Qe1JeL>0rH?U6+Sz3AEr&A5~WO1Z*9vy3EkX18W9Q1nTU{MUM&Z}A%JCg3(ya_@-
z4IS|iRdv2|F@`{@V)1$}g3;1XZqyEtJa)PlUfF(3gn>smPsLi(Kz{%KrY65&C^kqZ
zR??1P!}ahY2V4u2%Q<5?m9k-7!K{)9yBCak{!eB4k~enfzSwyI1zDF*iJ!)Mv$<>*(h~TmhVq2K
zFMC`yD4=Ttm0?)b3VUm&1gUq`DWv1~WjyUS09hh@NWSAZqwJtS{OXOPMMcjE7)MXL
zr|W43=R0nt7mi)U#|qf$2F^2L14114;Pn8;EhWtcDKho7+^poyg
zc+yLpl@sH}*P`CBG|)i^7Gc@zUA$hhL7;
zKzi$XAF;f7kVW&>XV%~sYBsv&|BQSU?AKL^RgG!kglqAynY-BWYK){#z-w0Mqq@A*
zFn0vA9%#fjyyno~?gmbKO%vRHG~=Rqy%HWK``xi9aR$ac5KC0Cr@?=FUKo6xXTc{}
z8;_q;*ax4|=jCX#+reHpbq)7g2fE1OwL1N`5xDp6cu2j;LSyG=3<}4V0Ug^cnL|f2
z_^AxYkm}i?xxv4z0CI;DF-25tVCg(AjT75pXGQI-#Z@%LY_^%al*WR?09n3`nZYw34
z=(A*gRXH*z78360IVwrH7rCWxfxp1T{(1SUk!
z_#gUig=Baa7P+wkFum*EDd##zP>vJ(S3ysY1S9=Y7U;P^v+3Grx7(rUYuJ>@(-BSN
z^lIZj-)sW#7S%0_|L+14+nHAow;>0sZPMbRLJ{ahO6Cnj9SF?822Fg<^Gj!-{VPBBljLafg8dnBXxo3J-|;YQ^RTGY7aa9YQbAIT*B
zK4`1;nV_|obd+HFkoY|zF7^x}5G+jNe2xM0+IyMwn2!G8tkNWlR
z5)VpAKR-V}1}ZPanyL595DkZ`Rk}MAkn(c9&=KENg8K4RPDUDbq-*k+?$;_mxHlcU
z#r9GWF+Y9k`HVsprIkNr=1m_bD2&8k-jU}(e`GF>s85K1x%{J{yEEFTNME$H`>-|8
zlq*u6@}dRw3gMIr)uQ0~h4KwSZ8I3A^pPha`pJ;WE+^o0y
z`&l6RXS2t$R9Fm{yzW-L?$8FMS?b|tA5Fov*Ss~)cHB{cMxA^|{vGf;!RpL`+ImnY
zmHe_Q(-BCr!jo-mhM-T?;nqAB{cB1r=<;U82+}fXi3)6n9!0
zqM!XD$*J@Nh)3_&%ESC`z{N*9>`qc25v5Xl)UX^GR?vCIl+xD-uKe5Whn3mf&8z$m~6Z149+yaYR*X?-gvKOx@}hpU+eIN%D4*8xFmF8JIj
z*tAz<6Zln1U7-!xLw8H7ig$n0!lG-FJk?!i@T`Ai5&H*j%x#-NtI_clZ98UlQTda>
zZ0qe?!Y2-4$&GeWM1&Dj=PoZujv2z$(J^Z4o12KHM&PT?X*T%4(puRfPZ;O*j#b;;
zP=@q<45M><&M@utN%h>)voMc?&db(S5R)tJtRD^DM|zSLrYq}qP_S61Ykwg7B`LY8p`BYWl>&6BJzW;}vZB*eC
z&%`x(DntCWNSRt$*al8sjgrq=N`b`o1L4*6;;=7ZuxOfL3oV{~bhkW58{a-OIZ3Fw
zPV(x;&Y6fl!+B)=vwuxoU}3Jz_x~b&u_Ks?SNQKH%%BhPj&-?*`NqvdUP`p$w2Gj!
z&6mQU_c!?ETU`o%S-jF&+nNRg_B@N?Q#w`;MBHoxnWP>c^eT^jESZN{o+-KmzA5Dw2{`=iU*
zf_qOT7&e(S;2Y`_Ulxn%aNwQ8nR1KGINh<|CakX>M}GcR%d~?b_4mHtmg^YaWGb>e
zT~rI3b5v6az1gtkUk~#i4u81Ey{7WL6~kjmbYH(Jw?czs>Iqd%jgWTn=xl){$wNGM
z&1a}39p1Pl`MW%o+SJq0zz`PM&@nU^1r%ect%c_!n2g1E;3|5g{6?6pVtr%e4$0U1Ay?Sz#a+0m
zK})22;*X`SF}Spi>*0z~x1$GJ8{xIg)jmcCL%0OGy=kk-HyXwjHLyu?w266Hc_4Njj9biTNr#%oh6L
zsF(^avrTqPmS41(FX;L`E4+?em~)b39D9-1_%i8zSA0eX&M
z!v}B-iJmExtI=&jE`>dJHJ!Pk0BK`7`s_E-0Rp+7k1rr~yIISfna|*h`tOQ+(D(TQT$6LOv_sFUnZ%-8qpz`0`h$=KInV8!(>;Dmb$AbYkJT`HG^Sm-S}
zoS(%3xAd(GhEvy($y9eIPewM%)#PUs@2dsm{U1dgT0+nX)*mal?c%a6CQ9iH}6=R&ofkapYhiOZ88!_uwSNiybk_wb$m~_W{o-)-f&%eKzi@D
zeSSH3IDBJmsOp
zfYL4eed#w(0>@)4F<$u!XsX%kVaBcn$mF(^zQaYz-Tf!>kM7t(RJK>~oAn$GcxI9D
zf$ygX$}kDDcN4P#`M=?$$^$aMCS@uWcH|)1Av#fX#2*EJSLk)vM^B@I#Dee>7tF!y
zQ02>JiW>wr_4t$Y6=djB_lZs4L0VuSc|oAW=mKi4*B;XlvjRg4Vg^ZdcEB^}3vVX3
zKIwa0GxqfGMnZKnv8}6*fc6=E_C<;!5GlC8XS?bO&b@g$6D_NTTAp3Xdcq9Q^`j4D
z?0lktl+PYVIA1ofbe&$~o%aM+j=TOKIr@>vAIi~r4@W>LI6Kox=Z&NeRo#&%@woTN
z1(X7RID!9`{igDDuc2fv@mv;pa}a26qEuI7K=Sg$X*xj@i1SIZDuOR)C21_j+*cKmo(U
zvG<5Gah7t&cpbQOOr^WWj3R%&S1%j<>0sj!16dsfH_mwA(8cQ{1$WDh`RHY=Vdk&L
zt3rd@ASSK0VR>&IXx^ayGqiUEKMGjub<8^rS83%UKAY=95+EuZEkXrr{(V|{B)oxm
zJ1YZ|3|7HC1?{zF1v*$wzr};RRUp-dycmrP1-zB!8F0vA4ShMCHvU)e8_-nnO#LqM
z3v?aX{r2wgC7AK0+C5V74@fOLX6vyr09u}7Lo3Z4#1#01$$Og*8jdC8Ls1@>^~wGg
z8JjI!ap!2u|F|vsMtz_Y4Oc1$M=UNZ%#y2ZY{2zgM!|exW>BTjCiG#EEgs6g
zWYhJ-8~QVGY5pPg10F1XdhJhCEuJhn`!kf<3?|wOehSaD#7RaAx;ZXh@O$XJ>ygJF
z;bUpPI!aCb41&!>b1^ps0lHGE>gaDwVmHuSzCjDl)(
zu}8y@8wbAw4p7jR-A^lrm+=T$W3=-}4G5?fdDK&1-~`bH)CIp+eDv6&BwkWhfbLfti&vZZj^
z%S-fo!esd9-=exkQXVO(#J=)BWeNO3!o^34R{;_`jPsoY&f)Mr;ttbI7OcSJqG`5d
zhQ(L&i4U99u~EwH*-ljnDAe|mHuDW7Y2TVGut7S1k6vr4{OzTLLwcGh!jCIqS0JrS
z`+yrBD!X`G)oLBkJXNm|y3T=tU35X@FC842*t_FQPQc@STHL(G0x(}Zi#_eY57bCI
z{bbmZ6>scVD@b&-qOJ*wlBsT1*v6OcH%6M{NDG|J=NSBm+-aANJ+)o~FQ=~DPLE^2
zbV^mX^d0CSs~B?z>&P>->9U&VOY**bd2#>T1Je0^pJskFGlv&Tr8U3WtXl$0t)AVd
zg1>@E;gx_J)?ZKsdD^L2tr2i0ha;G>|0FIiyva9h*oq`ZBlrT%=YbVd+#-YXG@9ZM
z)QTu;0i`afcX>%X<_nqem-Kl9K>Ll4ZQz&Bz}r7_)h}!YJ?5~?YclTyN7?Kt6`VIf
zLDv4yjlcqsZg(#=n5Pk49lrEX^HLR9ShA`o8_Y&4-Jh%amsdf=5s5SVZ|Z^Xe4U|R
zL@ja`{5TW?O99KpU_V~UJ4k4|yl+H(3XB~&pRt$O25x1%kKXJpMwYitPdQqrfH^xp
z)0Y}iprZs7wW{_4&bRL+xQ$zYipbncv&9^=aIS8%`$Ghn{$aIc^Zqs{bFOPXyf6&%
z7%Y!8FCh?&Y_5*F6rktQ!(2UkB(L{L8RxLGXmLo$^O{^1`aD&zxe$39-P%mnKa}Q1^8JPIU8gcc1POVT
z1JZ^_p2oCg-pmtx5Bc|iOZNskwI#EHuc(7K$#Ki2{GZ
z`BhU)JQ3yTb5MPV7R>&66_WVmBKX00?SOr+8VWj;EFaDg4*a=&`ljr!Beva#PxG$>
zP!XCNL{TXY7)1K^S^iu>H;?VU*OUB7P&l;YnXdke80Oqn0aNw~0YOVvS2Xw$<>6Wd
zP9IC4K48eqRr{G>@=I!)E@PKyaVYCa`N(0=duO?kaQiHZQhJa!>_Xyvwqu1kJ^5#XYHDwelXZ8tA&$)Rs*A4>mxK2K+
zC>9jv(y%TVq5vjqrb3Q9F-Gr%>s}<=QUI&Ee-ojyG(di8XxIHNH;PkFd!rVm0ZN-c
zT^h@^MOpF(*ng)vf^~}e;JQ81+=IRr3?8=vp{i~r35Sf4tDeeJk18WDVM}ZAip(1L
zzG0Y`(AGeVsMf_YlH|ptNeJs8@#y?I{f{GKZII*#!iB#}9Kdhj-LR9=5+U{aL^Si~
zIigkQtFlgVe)Qg7`TpqP6ku#c$y@qe7YR7dq&NT7LjSHdZ>T?Z0#Dr<#d0N+QEkSh
zH~kXDK$eqsk?lV-v_vPda(c)RDbeW6U3%&Z9CIDV=s>70mlt0nHk%3^m&6Iqeb-?P?I2a-Lz_?$14;@
zu{4KPO1kcw60?ASF!cfRm}lVP#W!hA^%xBpv}bd@n+N~&o3}plY#D#|#{Aa!FOEax8F9tfXD!7M4dvm7;`
z-qPcjCDg`%566<=@X=o#ePvGI(iw?hJgEd{N@*
zfGXZO?(RE%XA@cReS07KX&p?7oa~-pCBq!C&fcloZ1DGsQ`#*V3fO49sw9^4H@Y4D
zo3WPpH#)>WZ?51+3)A&?>737t<5s=V5W5cmi|Q**1=}_u*5fOsd7|{VHGL?$nSl?I
z9akWCZd1lA^Q;y($n|h#&{P7;b4eV@MEPfwR|X4q)G8VWUxhZ=%r`RJ!mw+o*|FPY
zDVTWj;MEpSW6XRfb-#O41g0fOx$C=8!)4}EI~N;_F!kS-^p**Cd~JL9zZjws&QQs?
z(V}|=ZuYO&=SiC5liXzNJ!}u)^fae=dQdKGZ80hzG&CUPBq&cRS{5Y2kPahKpRi+D%~E2e1!gIVZm_xT4l~|W4D7YszHLvZo3$VXTnj*9IB^FoZymn}!0z2(})UJ%o!MEj)$gFEw<65;k;RDUK
zkb1%YhRN4DJZ>PB>h`P}FP9pM6+X_z9zEUBv1b8p?tia#p3?<>7?wqvXBDvb+{Jo^
zi;_6kW2B-o`2;+%Y{+~2P!Z}4im#`CO^tnZbfu$&bnymjveeUmB6x6ExmxWWFR2$b
z(v(~9FfKgF>D*92j`ch!w$p;8q4}zz+J^u|I5g_w@~DIf_kKNZmk4am*<2I)^fucR%7G9gTgR?`9AFtGY`n~@AO~x
z7b>K*?DqwcBg)t({xh{_(rFl_XY??bNf=J=zV*BLj1K3%Jd;RR>Of@P>^>t;ra)B2
z3FqggBS@*O{p0iRg@A93O*JZt665@Sw_684gIU6e;%M6t$z!@e)Vcc>#f@|fvR!`#
zWd4|iw9l|YrBqb*FLem~<@tOJNzWp+Q^PpKoWyT0Pfq@$tV1sAqx&Kt}Hz`(>l!I767IAZPmx&7Br`
z5PD-BjD=<0SY}v3v3d#Wo5@Y!ktA$mZq!_rn+LeWsH++!F(c
zCJv9D6G$9?SBpqc*h|3s;l$g#x7kQA`^ibxoA*FTe}(adC`TZ$YhYz8-HAR`q(*aB
zSEEawoP*b#Q-P6OLDl7=5F|NvIdga_4Aty~*xE__MAh1-=gvv@qUmxO`M9}8uwx~B
zph~V3c?eu8h9uuF$Eto~NyY}Eqs@!Be_{v;oQQpM9JPWJ!|$du#7fjIW~AxO>5sCe
zUo9MyjRqf+{lE2JaYTi)pS>pKnUS7sBEuVbc@V}YZz-Csf}+)`TYF?o(I_4J#=Ms|
zve?fLinh=I(;9CRINEv9A@}4{l59nJgV$SS+LWx+;KF=d*U;V}
z@Vs@6hIjlFqJX*)YnJ~3XF2lB$SPN)EX1tnYp_OOb#M0Z_{0pHIXdk$d-xEKjBOt2
zhYm1DsLqF(s)B6f*x@wui1^ji;_~g1T|#rF+(D%2$eC*H%tF){t83LNAzXu=TYbL_l6WIY
zDOEkBEL+ZWLc$5K6p@Kr%%%Xr$dvExEV7_%kiwbroC-QrF!YOIKo{)J`AnP*bw}by
zK8~MRPDTy338U470$`<t8R9Dm~03=p*i+rTKzlNkYG;S4?YYl2?>eJzWkw_2a0D-8%B&jM=VO@x>U1W;FUnEYN`ECz_%^2Za|zx
zzw&lbDHCrX=nY@*pfD(+9)6DI$nN3xFmGf_&A)wY0{*L(vA
zB#s}`wERoj55Mx$>^}s#s)UF3%ek;);W1;WK~*ez&8_?!P-FJ?
zet%arbe#|mZg6CWG_OxT`$(mUH5z*iWs1z9!@coYL$M$%WmQ_f`|lx?(sg)k7=bK*$7B-E?vl7(<{H1Y;qOflr@ss28)<;MJfn{eNVLM8%W8V%N88|2y$2`4MQ_2w
zEe2HLzsjMgU1a|0Q4D`N_|+eLjo{lm8Bm$06V5T}Ip&SYVxDqgaigaXvHXph$PWt5
zaLw+5PRZ+5XrkJvnZ3{k!`A4lxpUH>bvztU+lt2*^Y(8?sFcEFhiAFpNF1?vA8EP$
zt{&dk$`kJ)dxf0_4?f@!>cGC@0c@Jit(c-{`cm%KCfvJoW?;Rx9-CNquCx|)Vk|Ga
z-~OlrKQ7yLKkM0w<+a9~g_T-x->d*Z=y4M^SoWKH-Q9^bT_w^gdfV~c9F!+@xD7Yx
zGd1{d6LIMC4972J4VYds>Ph5A2ewH_I?e&xu_C&zK=q>)=kM4Z#S67KO&oB~D3s#7
zr1+_ljdm6Ah+#jaJk8P9DUV7qkpz{Z6p>`g}=abCX$uVu+%;&lX1-2_
z85OqG4hQ@JDjNAGb}YZ5Us>8ED7z8eld4+QRQm>2O5gQMuN{Krn8JSX*gNEL%fjVo
zNDY!beDQ9U^CW1{4EpLc$puGuTQ7Glz5?Q|r*DO3&!U`r7k;Qxz5_pEz=%cFYt$Kj
zNe2)9LeU&L!o>v>=;6sGpS`o~;69a(7xUSdXgcBN+Ux8r#6EImm?B^Uc{Bt@M3!a(
zSL$=fkd&LFziKV}c)b!G^{xK16;uEa+hLQc$b7KXuHEvzr5i{d`A*JLk%{6B9NDYy
z`k5D^GcgjR>&>aJN)b93=O@k3V@nfNXX9_@%
z*M1)Ka6zlvRyiLZ>=QHz3IbndT!053P4RGI8Su4qG*OYf3(AIi)ii4Tk&{?}yVis~
zSg9>x-ebH1rsZClXHw*V`NwB3J&g||GQ4RgtULxnn
zd5FY~*e|WKJD{B}Bh0rANj|P{uzylV6>sCdC10nAmzm~nJ0`9XCa+%W?w|Nel$4Abd~%B&
zyrFT*jgCDJR4MBYD$w^4=UyGLATJ#!-1>bm_Q=T@Vpz5V`_=AW1itLkwN>GafS|}U
z;Qg_RI7JbYgW
zsQdY(X8BMZvHRXJ^O3BN1i@6=bsmz(Om4|;vQy5U_`Kt!T1CbK;>Xnz+gW`&&kOO-C=2JSf7M#rO0<#+wuxwn9I|@_mhK%IpqG
zvSTbn}q0qJ7e2I6zJNh)I1#zBQOr)f%J|kM;astp;ql~rf`;)@7NlAu+ImVm(EU3LireV@tmgEQ
zP(_dP&!{-$9y`Uf7?lh1KiE_*aKAvG3j9NAt`C4t?lGqiyi34eH`~;H@)lb9^uoV4
zDjK+%_qk?yK0r+h*B!1TKLO2(;qPrqMuC0}A7zJ}8xr`(5u<6Ufu^MdZ0bb3f!VC4
zlgiO-)O4VKY|FS74URnWxM9eMrdapo8EOmwi>FKZ{+cuT8`r_{vo#5PI9bAPqw^4H
zUv@aEevW`7;-7O3gQAjE#y
zT_GPwwe!?+y2Ct}1Nvu`4R#_cVLc;{GXr4h4%-!|^&T0#x%tGh={NXdloozX;vhUO
z!K4xNJ_t1PU6Y_ZUWHzU9g`Hg*aprk7ZH28rYxP%HzezKf#TFw1xSnror~3x
z=(nCPLvCn35;xftj9mEyJ|7mpITX2vLaKMX6dF`8SB8m;aDgslouTaYm@~x!T3^1w
zGpV~b0Q>BYY8UV+u%ub2xI%14KQjc6v-350%9Xnbrv};zqCfy|?-p#7OxXO}UG2<%i{Q<{0Hn0Pz6MuCkHX6k&kn
zG$O0LJUsZ=n(CTUk35_>RaN}(g9?n6nGo>0MUAIxHiE_<4Whxq4{aY)4nfhF(uu4F
zDLgH_{=Be)42w;?U^#Oq5xzMcxc9Cq5ls(1}Gb>%f!=V7(Tw$=)FH(_-tDG&9F`HQo2f-Y2kb>#CaB_(*R
zJ@v#N12?Sq-@gIb#fPxAN3zq-z82z~L^g#SBWN_*l;Hl*5|@s5lj=^~AiL_{LwDXr
zV|$ie9y0Prc=%hM`81^=emOMac5%T5e!m9gvNqkZ>Y2g!@18_M$LB|`nu~hF+esQ;
z95-LWX(x`FV1rh8_}$frm;D{^)}7!l5Bl0*4dZ=zJlX*@1gwuvyzGSeA&r|7f*nwY
z?u;p=bt}x~a=9MX*aihT)w1pLT47aN{+
z)f1Vm9a!dHYoPk)4m{Tq{w(xwJ1%4#pNSyl^Uz9hvq%-kY#h1;SH
zzqX0F6hZPr+$EnLXk%=|Rb2T|>vc_7h}^AxR)B~ByX10b1QE;sAwg@=&A3M9`F(Pl
z8hr0+N%JAAS}gfPWWbv=r!t`0;v1TLiQ_lD8H4-^G4HAf)A!FWv7Z8U3twgx?w0qA
zd^lE#XKwrUi_PZY)yeYJ)Yu3-W!~TxPt}1#(zfE&&$r<_C-U`7rV%!?Gqd36ufXbG
z$Rte1v+yLlGDm?{8?M*F+5b(o;SZ+zE$1y7FlTvnZy8+H
zmqyt2&U)72%&_H(IJU=lM=!;AQb7UE$eS_0(I)xSUO!dxr!KC5cBf#*zI?vW
zL3DbSBKQq;2(H2k3nP?aa!M6zuU`$pN_3M1;5~dIVHAw5$p-6tj~9x6JYG}$M-7to-(uJXY@<4q~Nl@*J53FMe`<91Q#o{M>89q@31JWN7t#)x`-QvSQttbX*7q
ze)zt7>lHaPuS&_Gq5h5PT-ZKz?SBRg#Y5%aWx7D`iyG!WHbFR*T9z))_6>9#(gvCA
z%V_=7;%d;bIS_kA>GB_|PNcYSW69&d9%#4-T8pmRRYuD$uz42H1u
z*_^|jDDO?+xd6vKG}gyV_vrH*6wHvOWXabFED0SZCWZBAvPz3eXY~#kirVIA-1-cL
z&%Efq6Ey^K+$$gGXx1PZ)(($kwQK;@)w-shqyyf}zJO1udEnB*fy<9(yHJ4m>muu~
zkHE@O*(yVY7x-{u;XVO0&Pj=7NpNx0EhL(ScMUD@GHDkn>1np2Kx8*bCRP`;m-M{fzOvg
zEfc4c{tIO=_#ohCKaC|43bihxjmt#C?9@%cBM#t4bl_{2Y;EAGaOZp!(^YhQDf!@k
z9eUv4f!}im
za7xJHnvc#8g4)VHySfl33hYeKoj12YP3)iU{dC{~1cKbkxiMO>B&PK5?BC;P6bc$E
z71DqZ?|3?^w+4Vlu0oJzg#jq2I0QWwq(@Xu{d*q_4g+{B@rBOnFQO`?SCE1bHzFS4tthWDizl$h
z))dM7Eg`bch#iac(ItdufAElLk|8p7M?F^{s1a@oRiChb9YlQUGP2YtRZkccKmCrI
z&YH*>u6J!|QJJW6N?cmwKLf&hnz^}}cW%VCdpBAIEP9A^o98;bwpNHUDuY+t6-fI$
zJGQnbA%BR89pkU$Ztw$R@kWdL_q0g9^_*BH$#tS)i>~uL^(x`p^sD5q#!ccrzxHa2
z0|R(>E^fKYRv29@O6Yl%{E|>}So!AN|0;?6nB&NrbSvTIf}=%_-4CKc;DR5u2m=tU
zHtu2Wtt7k~{p)Nz@SK>^YasoZqKuIJmz86}dyt@>eImqhpB9xIA024^_Ldla&6Fr@
zLh|Zm+&X_YYn({d|5QaoXMylS%PNqmpAigyn<$SZ^%__F)YWMm7Dpf`hWg5oGPq6=
zuzcAn5NQgistZUy27g>@zDv7GqFImRcV(<^35V$_vpZfw~Ln8EEv7`_h?%cTXe#sehB^M(N(XYfGb;24>E|QXn6?bcq7)s+4rv|dV{-r0c^>!F~H9M(Lbg10#tp5V~*$NJD|{2oo7`y
zgX(kcTsx6`A4yf6GonyU0jJvbI=u9spr`L)e3K^;*xh`RWpa1|7_IrtXl7X>{sjrW
z26k4o?`KhATjc->nW@}3CRSJ+`SiV+CuMZqgiq)zrlX?W7Bt+MdD*zW2
zj*Q95d%%kPiX6DCf%u6Bsy?KOW
z0X^rEIc@Xn6_7kdzG_q5i5gkllgbAA0O{5-?-o{q`Z!W2ar|L?lU^c$-P{gS4)G<}
zaNEOW-ALUpBMx|u8^oQ~i-D&Tg!ip^!XU%&mGs|-#c;k(ZGa|epLMm#=3?hbY3!3q
zT!{N=2pQ9OFRS?Y!Gag@O8udHkfSeUyeoqrqX(#Rb%_rinw+q1ysU-y7y84-@>TI;
z*(2E!M_I6>>qKPHdp20?(ox?#!-S`XUs6mJ0eHj5yG!f!QTQ&IF~XZ+9Q{#lJfTps
zi701F`&SFs0O!D^{*uQ$nCZvM;R7VTHd*j4{YGIFY`vFoS9mc88(n9gD=l~ej}Het
z{$AV;D{>NRP*My0LnmSUYsennzRUllp5z%m)_t-0_B$VJyLXSw(~Sgp
zz)<6g-&=1Z;mZSI)&XxL;OkCT!{MPIn7ODh;ZLiFw-_p7%6>Rwsxa1!mMCBR(`Z^F
zO5F`RM=L*9CFKu8Z7%WUfAWUEjA%1d^IYKAYgxnly9Th7F!eD$@EEpua>7~s>3O^n
zcBXKlIu~9Edhs;F?luhdIL>C6lY#}ds=a!>pW|(x&p#AAD)DZ&$kTA^Yi=t6E{9^ahXed%`CqbN{m_mMH4uY2i~FAQOPft}-{aVuEgFK%(7(-ykD
zyW%gsa~6y6DmM&@wEQ&e#hID_`~5Cn)iRJ
z@%#Dgr9|s$e6^=rm52W^#>b3Mlv6O~Tw8Qmi!a2f?_>Ec85H3f>xU1UFK1!Nv6!Q}
zh7S0#c8fFXH*fg%^hudliZ}6ugaqwj4Iex?VZQjq)EchFGin+5*}~h-k1qJjU4)D%
zrQ0Xm2Fg;szv)724-Y79P5*9KL+?*|HP8DS#?Y~%KV$~`m10ET^C>Y+TF5_
ze%1_EoZLPVe@+W$dUBigjrqXjYVX)WSr7RAA)i{5k~sF*JxInas)8fi%8h;;7lazk
zvo|EpC_;y|K_}(DtC-sK`i88MB9@yuW^N;P9_~GUa@X`27c`IFJ@oyr9+sb;IYMR4
z4HcMfK4s=6Ks}(bwJ|`814cYDqk6gUFU4K&oUeLNQ|UnUcAz@sOzf>?G7-cM)8d_$
z#+;b#q3MZzRaVHllg<5(lt#8)cZa0sM_7_XD@aYJTLf$I%5zW}upWw~zL20ELii0Ep+fce_ft9xAIpra@*
zp#k#4;I9LV(Uphbn$bJ=FP-bi>3ogZeZE-`D)4x|iSQO&c|j3RQ9}VI8o!*QUF}A6
zBR{Dw8}uOWM}DIf;$7fd(T~X^aaD-NRDb2J&l(cdET`!48$>HlbY-eEF<2CNo~4^#
zimr#1IjT!PLedIWh7agS{X$;3eT$|2AS|7)GO+q3`gHXiwaA48aBa|(UM%)8*c^&#
z6=NI*EVFWXLHs>vYvg&)x^@8Y={TblA({yqm;}F{9(an_q$pDKN%^b6)dyqrE}aON
zp3Swyw?M>3S%2%lcyOog_vjb4TqK+k``(wf1)cF2CW;OhAWN}>>lyxcz_3QZb0qDH
zDycmlU$b^aqT`P$FE89hv^PHw6lNrXZ=x|j7H=3Kv&3&@p#i!mA^uOV^tdBP5wZNH
zxS9uY?G8JP;G5`}rk2xFQ(ZLhV!P8R$qYy&GgAE(vqECY$_Zfw#mG(TpTkpoTeN(T
zL2=gF4bPs|Jy4P-+%hdh#Hm{`%wUPolo!4BTI9)4<^~J#W_kPe>_|THx_s7OTSUv_%+BP(kGK-{
zt!vA8L2PXMKL33-L^XM|f-3zmC=f9_^|jXw6vi*gr7X*XnX0?<(pn^+7q3WHHy;yF
zSZ`kJa{Neeaw*H%C-s&pXw!-)+Y5rlbEoNPzA%$|leB2X9%lj{D$vHXEpUGr&@fs6+cZ+Y>l>@E{)
zAEz%FoS!AIa>)esg^v>TZWu0RE))|6-Fr&3f*C=uuv)i*=1+ow@`vu*(G$eq2ci!m
zl^z2Br@1&ui9A9_5Wf?$+ag}K4~i=qTPJ?~rAbkGWSEdN%=)tWZ9TDLoRVWcf=G09
z_Hk|=v?nTaX`U+>H6#=p`D?L?n-P0@`L`|2{0Se24T2s?H4(c9^Vb?r9wN4gd=czu
z6(-31RIQQTmLlrvKOYK|Fej+Z7CN#yS`)&{)imZR`3Meu?_c>ZuoF}7nR+oF<0ag2
z7S}hYClG&(L{&d(_b23EJ`fwt%}tchDBY(jVIYicnRhTmF%v^#Ha&tv&k%0BvGdTS
z^d;)Gc|pcDcEa|Yc9whEc#CgE>Q!2*m6m9my+aD;1qrLqt}5Nq@+T~0(U{FFbrM;2
z-oIV=OC)G0w07v6>mVj5%ZNS}YbX4zuQVOL@{7PB_wvmS_kBWk#G-oSxd+6{ya^+0
zdl>}$Nlm)TwUziSx!S)ltAk)1#bIpV7EGYJ=+=12-iO%z`c=|1IX?mkE>Idvk0k8Z
zj+0yDG!T2s>+@zQgNe^e83!-l^dtP;)L_(V3nB{qIm3iFdWY7BhId>0uVzM*QUoYtL3Nj|8;Irw^EZx8F&nePU=pA}fp3!5XX|zGHuzoPgMwbjM~c5c_xYJPe+@
z-h!q7DLlh^n-4$17tSUqRfzEl6)5C3H6bcYV!Kl_-(Z$~NBgI0myi=`ImdSr^$Y9^
z&b%LV5g)kN#Q|<#MxoM|r|qZ}9u4H+tbyu1K|-2QImxC}^{%A0D!=uF8=)8nsnfnO0VOf>+Mj
zuvhT3pc+TFupD0)ME#Wuf-UW?;RS-5es_kRN3(xvrtKTn!mp{eTyYI}f?rAX$qO(E
z$K|{0D@FUAQ5WtwEbRnw?&Gb|`?X!pXoEzx*K+3rv}jB4@qbsdaAf4B;(;a?e8~j>4M;&L`;#<(^D^Ly%C{D-$f%!F-D*o*-z
zuU0jd2}{AOrw8FGMWT1l0`U=i&t;;=4zvWOSjY%^HgIfc&RPTY@%cEGX=zWqtM
zJK%sr>E8X=e13%0re@KzT3uLh;Mad9tf(*
zN9iAbFXs(?i@&xs-Ij>aL-Dt5JSrjLuuOgr%9wf{{M;dM=F!|GDEKci_>cwyjif+C
zh1E51GO@14Ae#lI^emNZvXBC~g@?b2a|wYtfz5Y|cOQfAcS>cfOf$e@uA!BS+^^A;
z$^0HKS$;U3#@Ov_yak9|sun!Zxe;8;uUOUFtOS>s%3Df<3eem?`z|_UYy}+~qSoKt
zxeY#yj24+3eN-w`*#r8VTrVBY{(`reE5I`j#Q(!f`*+R|`-WYOyz$FllpErV
zErwZ2{4my&bihLQ8VJ7w{w4f!0pb!NV_ttPK*V32`o5JDzc;G#>OuB63k=*o53w5
zIG)_U^99eM+iuP$F94anRJHZg>A}?|^Xbux;vl{&U8nKME->EMIkb3y54;pbE9=q@
z0lIaQhkoSl0T17v-nS&f57jf!E9ziB%+1~`&LArNW>=Nnp7lU+7l%stsr7(kZ16|P
zQ9hvdEv@(E>&@U~V5xU;(LcQHHRsk}BfPNF{)x@*5Pq2D68e7AI{*zBbfod4Ch*jz
zdDEo?L#R+O5RzGO7?9{CzV`(a&zslySg)9T7D@Ano^#UY;n0kmale!3~{p{?#`4J;*w{Y#sT;PTYrTRWG
zuZZuX&NMyO7g)p}hH83UwcZ9V9^+H^`6d&6+4e8%*Ev>jMm8%o13PDNq_{QR@b5v+^{Yn@9V{2Zm2dWz-Bo01Lc{uTjz1)J9_qN
z^|UU=3%Tr9)|ZC-#Uo313`>XnLl2!E=IDFWjgB2`vnjm(0f(cCdgq!pfzjv_o}~}z
zLGQfn;tAsan-1q5M=EFy<3|?`M?U!T5>3jlWEvW2t@~g{iBCBY6OkrDz{XM
z{$9(&h4r`VSii_ZC&$lK8u2~FUu%wBJ9gz1e%dI>WGuf2O;;6OYE3Fc4H9l+47F)^
zw(q}~fse81JK<#wL6iF^$+AF+o;4L6S7}#&Wsr+oHu4pw#w4Skb0qDQS=Xb9&Fi=8
zM#tgxRrDw1C38@N=*6Ybjm`Lr*Dn+ON*C^a=sW^L5=E;@LV07-a5=={R$MYC^>5Cs&);0X!QAR9~IVSit!&UloVWYpj06p
z-|6*kF+E-aJ+kwk^+o9`sPp-h6&ZK||M+NM3g6R{sMW(NW>Vy3{C?RLn>xu0Xsgs>
zK+tnu+&0zX13iy09}YaXJ@<51aeAag(3n4o<{ADMuK5(4|36e4FIDa~53m=+}ZO
z3m$7jbc2KMpc?A|l&8b(hjM{1F7n;4nk$Ow9G@!9Lm0=9fE#nF;6X3OabZN@#=#=w
zShjcL&OCM$t#!IZ;#wrm8$S`H;@gKEwK(@CW>+IJJ=T_@nVE>hZC*cU6flU~J?M5y
zO5-cW>iuuK&gXihJoJdmsjw@EN7`i}FZD)DG9cP<{ez#_>D4rUQ?@zm&&RzExsGj!
z*n1U@;mi_j@pralBIgV2<3fTvA~T1XS`HuhGgXeUcyivfV0(_R*VryFxujyD#`l#E
z={h5c{@X|5?=&J8?(MYpQY=I~!}5HrRU$B_IxjwjBi=}h5a-0~G8cjo6{xB#-UzpEmyd<*99qd(VPy&b7}y-I3xEy^YNIueNs@Snq)g|U@s|tuX4_$;N
z*8nNDdgSj&e!E(-@YS#}`x=s-qyE!Wn?(w(mlW1}=+!;H)66%XnGq-BYOMjq~x*=hUikL=WF~
zH|-*IL1F8lXS=YR&CzGrgKr|=zWA!WG<3t9LyE2ZM!XP_y=^>$Hv=%HIxY>L_-9zb
z>(cb4nU+qfr#!wzF&bVJJ=iL;)CKK=VLBZ+8w@4I~afg%>+bJVO%v)l9D+iA9Pm;?HUCt-`*O#A=SYeLzCahoU&ccJx?M^qnB*G{lAZ
zT;u1`1Weze?pVZP8iKL+w+TDdVAkW~6ARkYNImwPW9OJBW~y1Lz}Ivc(co+5e4{Ia
z5L>aZ)vWi#!p&^=zgFtO^p9UWT33DvN%HnS(X4O+DTy;KjwU%`2b?|##ZN~d1DiTG
zs_3;LtT|EQQ_apuxzi*9?7cB3x*eJ-VL`~>&cG_evS(Pp%lm&vy<4yXt-t1D+#Kj~
zcM~=~!i|65;NpAf+b*cI4soiNYQhYC
zr`ESTltxvndUoRTWyqVwi|fmewqd=c4~kbr`jAhzRWdlfe87&bw2A=-VxH?s7uQ&N
ziJhu+e)GJgAK8}km?e#W5^GDoyJ7RKb*Spn#jfmqJg9V~n~is>Gu}9~^%1*?Gs<$^
zV19SU4Sd4&c(LV1XLQ>yk<(q-Md;ptXNEck6!0TQIWr5~!YfS^WZt*$MgyX6ewQ^7#jCk?WWUM?
zgUff(jjJl6(i}zeXCsc|Ki>K34Vk*(_Q$*bVD8zdmTuJ6b>`1dI(_92J593DLS7MM
z)4dw}xbnt4jvwFf^c1TQJF`Sw^|Mv2Onw?#<&`h+=a~wste=7pZ>>kmt9XZpY8}v^
z4TYwu_-I(TnavT3EH(K$+`5KxxBv{=QYlq9s
zn?Kh3mV}xoY>OVkiqU9>{Z{{s6H!(~E2Q6-4fZroX>PUS0!w$hzsU)2fXM;(1uq{E
zf!oT|o0r_nP;(QO3?`8ooI$@@(w6fX>OYso_qV>D@IBoJj=8j;
zzq@@$FIiXMq+|Ze5$#?0mu;?RkFS4%KJWCo|7^!qTyUj%h%TcY_dQkT^njB%XIp06
zg0ZF$KRLKDm~l}AcF)Ei7ryZauS?0E*llkXGifn&oL`TuR&aSvubG|){UQDlD8)9C_#VaWjXN3jiKM!=xXo39z}KI
zUypHH4WQ!PL8;N%sd(U-Z$A#L8^FUpHaw7dIfnO%@+6*18pJnAG~c+t)PsMV-9g{Z
zApp4e;u4(ZIDqP&ed@fn^f2>f0C3k4gsj`cIh9Td5$AL^Gjv9Nz>U0|Wgd)n;29Fb
z`?e01qes+L&g3{eSJ!`dz7KYQ3?pXZYP@Wz1Uc!+>XZe#Psw$5QI){b9Qf|c%U$U
zVX~IlS2VHlXCVK&uP9f&w%|Luzqt2>op&+wb
zug3-WXFd7jC-7y{Na0xiG1RQFWzXHIest`kzxAcF4fq@R>k0R+7vm4MG~~YztwWJ4
zx%vDDLwM;cOWF6g2k})yxnfy~RGg{XB*O1i4k|ZrXIpAVF0Qeo;M135&FFmt;a+i*
zKJ+W|kq>q^nouy-(~exp#1B1o-u8Md1Aor!YAvn(m)M7od7Ocd5%#7e_P;z@gfA)d
zZJZ~KqbL*kESA}TYrctg_19Vd0)K>%923vFC+In@eSvis
zZ&NjH&rs-lO5hy6WXdjjqb&*LP?nya(>6fuc-q(3vKHbyJ{vETq!**wBbu6T*q@^t
zH%ZpK*NVc~#u*}%!-DbG-r+{~bc^uuL;-po@El)Q;kMY?k%$Jg7uks4yocY2IeUH}
zDH1<%!lX{%Off#>{e~^q<34`zb9b$#9q8l!}9$pHLC-&`~m=r2^4noI6
znzPrFJkYME))$YiJB!bqma&zzKZ&*?0FGy!KyA0Yh`+6LhS-1l5qqtOEe;Q_>(cpi
z43#wLKD7Rq0ug@^n%L;#b0nc{-v+VP^MK6zc6U4(DGdfrN5XZu2ji_k|wa~%h%X(
zf%+kd6g5T^JD(BEaL54tw(5H#VpIsN7n9k%3FF1>-rsFL-Nu2=Z)6@cx9G!WXM)_v
z9M7P)zQuj#_LN4wm@FQ$?-n55|8Y0+F5o~TiGz0>{D(2Oa)(nUvT`WbtXWO2>Mv|V
z=%*5gH`CbHh~rrp!wfQDSj?H~KZ%X}EHv)^x)Y~2QXXS#8N>XY-}z)8e~(nm2X5w&
z@4@mNV}IXrE=S70h?h-XV@IKHqRn2tN#t%vqv2>-FBW;~RP9I5gan^iCc-FR><{x3
zizCJ>nB$cli_1Y@u&TY!4cojrkud{}uN^A|7^&se`B+;L))d}iV$JXqJK}F?IV-e)
z-L}!-p0@5l${IH({YovybgtW<%#(hBwXVxc{dDUSRzH6E;T7Wi6;EcX#)!;Zw%y8QuNe{d+}LryDAUim7L
zk{dTE9GH)#HJsheT%Lh3cJDlpmwy*IryuHJ?`(~I2}GT4D%oMF^=BaRwgjuRPki>z
zJ_V^c{rsN7$rD&O(+R(KdKTEJ%;Mv}=RA>HZZ=2xz00FliGvChf=LJc&YaJ_7cDmZaSIIyL`v{Vg^#=H5xL-{+dU?TD{;^unv<4N|)RZ^kPdrA9OWKGit!bo)LB2NuH(n!%qZ}>MiZ
zCp}FKxz3UFR?V>XjLv@ECN*nWTl-y_yx6s)yII&+-jPfebj+3S*Q(u;YkSvW)JWpW
zG|ulgtWi^X(s{w%iW51=a&oX@E>}%v@8^qrGDD=V(+#3LH+$5sI^w4pHhd?&T*nk^
z_(TG8jyuV@p98_X#jZ(t1R7%&v#V1qG0sS{{$5Pz{Ub~w|8(&ou?Fm0h6{VZp5s`)
z@0PK_Doy0P!afxf;>_?YDdmEhF-t^3>Uc4SS{T;hHg34XEi8tb3qI^hdm>mF>}FV
zSFdPBBPjs3f7H>M8
z@`n$VpK;GB8s37lcR=y=gH0Iwq7H
z+;=yk%lQ$e%UAc}+afcy-IgaY{OFpSI~_|($P5czidod
ziKy0yT|h&w8Lqn8y>h=P3B9nhgD1f9CN7@3k^W&V@%w0eCO-I`IEVR6?MJm76gOZi
z8^sO1@l#`WM>PL>qx;Pdo(?Rq#`8Z~{XD;X9WSEq0iJ3Uf(=;m1tkh>+W`!a6!rYr`8!8qXpKYejYhDQKz4~`nhLLp;->Y
zu@}Pw(1Z5I4@7po}{RJ
zREFc9&_HDp`UqKVXXDMseV0ZVC$x)k`6g_g_K#fr1j{b(c3V&UOU>Pc&g^jXAuq?S
zpL9>rC)=XW*sW%wtV?Y^$s1EqN%I@*y$4P3xqAU>Tpi(fFygBdc|H~2qOof>qbUKG
zvzHzGBOOJY-^G*`zUYVA-SO%Ry%~Y-{`M_TfY?VtS+26dSmF*kSU0vIXWAVVczf=+
zN7E2~?!n5HrjSnj*ZN1%#^SYT%kEN>+|&W|+K}V<-(hGvnAqpV*0jummR5i
z_!c31p0iPCZ0A)*-wnm68{_Hx&Hbh5m)9q~H270d^mj62hgbx9=ce9z4Of4h>9&0<
z_joEE(#*))Vvvr9n!=Wd`v%~PuT>>r_ug&YdTdQm9r
zabHs*Q4f?|bCIpJ-vVXhO5$1~-lgKXvUk03nE@`N6;ymj=RA53e;?0Z_zvyXs5xcN
zJA!IFTHIVI(1gBeI3$H;IOM}uIF9|dg1UTqk)Jtyso!&K2k6pw+xRD4BqdL-#rxM()Qszs!{Pz(C^$8
z+)M93PAKuck_mivs*ujh=)S8c=47a!Q({1&W23qrPKaNUhZ
z?;A>`UlotXFE=pi|Bb(cH=liIyL3JrP4=AUO`;D(J&s464Hv$J2S`aq@H0B%5rbA7
z10p8qlfL(rvmBT3<^2QRCHdEJqmd7(77Fy}I^)@-rM`2h;kFUGM@}N>sXlIw)n|v$
zQRj&Jvt9OR`A9B*-~mlE>RhnxhZiUD^VQRS23m@^dH)jg?}^K3z3#=UXUwkPu8sqJ
z^7_Q@k<+fJ2=7LxEpW#5}o(G`zvxY_s@G~I^w63ZB_H~IShD#xs2nj-Ew$YoXoM_3J#P(
zqVbh;FYzwXL8vb;+=(qh##@H;tBA#oO?h)&8mJ!U$*oS!9B6yN>{(WcQS5-l?aO7Z
zzmbDFY*p2{KQQL&j<3r#j-#UK?@so~E?}Sc+O%@r=fJu2w!ZxHjs^X3If<>I@&{H(
zY8DV)zY$f=lKWy^umKn4%CK|~tw&0jq&mfxyAbQ@+k8hXK45>h#6+BZ!$|CVE->$B
z@CI38HvV?KuK|(w{EpgdH(_Cy6Dl{2;>eB((ZHcQ>(H-qgL(~O(U|)F-n97DI_zhR
z16|Dx9C_6=!1r|bb8I+_Hyr13B=<=F8Cp5;itSV%GLe
z!v1>ZzFl;s$Ab?lfb$zzTSceu%$KN7jcoN~>!mVtq|dZ~B^sBHHswqN0I~$U_lTmT@y4Tf5We^BEo_Hnuq*tZ@{=S>vDI&Z<9QZw*jD;>x$Xd8WF%%$^Hhx^7PE)t
zVO)kGHgNWx{*PDJkU+s3KOCQ>BHr6tIo>2)$4>QpkV$$fjxb9aw}(SlM2#-}Z!Rk%
z7I25feffMc63N}ck-+JS-S4U{-_EO!Tssgn{3+c5dAxD_(g&{?khzNDVSQd<
z?81?X3LTwYSl*WFX2eTfm?!a4)8va&NVa+F!q5hB>>%4=d*twn8vL6P#i}ceIBV|D
zfqPsK8CfA??LckJ;nu}nLvCBK&d!p)l5!E`JL242cvl9~nfFcX{bz}k|2jG2+M$Wj
zGnHmM&gxc6`?TC|tFRTBKET`dnMWA2nRU9r8lj3ATzbB|-dqyX%C8OlB(f1fV~?)8
zwYo^kK-hCn@_!+De{_DwE4Uq7Ss#ibB>iy9G%sxPk$#YKE_WG!r
zjGkZpaxM>o*{B%pdije~SSEaT)o+g`
zx@5JfiG@z4M`ui_u@oO0mZ~2ob+_0^=WukWog)=B3C|JVpB=w5yg=;BaNvgB(J}c9
z=<{9Vl0{1}$UnKeFCN&z$A&|V0sCqpH#-l%(2sf;G(0%_L8t;ccWr8^H}i**0t`1#
zJ@o-1%POx&4z)nK6duu20*x?edg?>qQZ0NmetB9YwFF)paj%c0OM=;M4{O5N`(UuV
zD%YjB4mgqfQ;5y40*1bIIXaV60>6Nd{qL=^VD2-6D?_AF*o2<^8!XoYr@}O^Z)Y!s
zCFWDf`uj>?iOSMll_`JdS<;BIc=o}X+M7#RulgZxN1HqgsT$g5`Se!>zJOQJ8IJFs
zk70brFYi|6F5oT8WveEW4PFV^2l(*Z1_~>UxdKPhA>(1++SkPD0&+A0Wlp^X-*vRC
z2EKLz`)jofci8j5z{v^C3HxZksUc-&kk$+ycTdEx(AR_4xvB{_bW6dl
z$Y(xQU(-PUL96|T3;lp8n^L>i`3ArcuO6Vr4MA(xv|;&`Qt&fJS@phD5*TK>DK5F$
z3A~TGlDxd50g!5Ix2v9Q25zd=M?9}Jf%1l7PCnF(tF9`9!N483Ury0;_OBb*JuT`oXYMP1fIPF|+
z7B+$8hNpDn>RrTloB^-)ZEAx8hr`+Yb34GXrc2B&Y|S9wr;f;GS4*srBba1(jgFN0;+rNj1
z{m_37naOiBgA>oEJzCuB!5Sd
z;JWQKpiq!OJ8f$XkPVKB8C$}E#-Wh{MxGjAQ})3))Sw2~W1`YSe;^oo{IjhWgFvIU
z_<%?>1lLYz$zATl!F4UAsAFF-Fn`-9qL?2CO56AL9l4EzhN~?6c_J7%Rr>8_x&sb=
za?D>>-V1>ZyVTBWtPm_5DE9Rq!ogvy@2Q&ze|c7aHgpmFO5)Vy`}_nPJlgr-iGDo>
zNc%5W{S_hFy@Ih@kEsS&zsc(e?1tbjICzPLsQ-6@^X=vp!mcoM+{_LKqvH>+dLZ?t?RbJGi4l9Y8$~7LJu_=-TISp5FYJ##hvi8VCM09^9UTonyVY{sKr71
zUGvnH4>+L9#h*D7_a(n0h%;ye2RlyQ7IS}%1JWw2mM*}-)W>H7w-a&jaRylqT~pH`MElgveo?@2kQ17GO`IJA&Kh67wMydP#>YZgBJ3l+
z8%}%q0>!|t7eafT%B#U>j*rBH!79+VyIcA!q1U1}DE+h^2aO{f=}*s-K=kkSExKfSc9gR~OFXz;J=@By$-FoH#Tw#fMdZ-MeoXMz)p!KS^%k70F^?dNTa;
zfDZ-^m|oBB(y9is+mgaxFO&g3=RJXZ4aGp|nWTL0-y(4CiE5$7A_#QU#n2
z`~QmSDF+p!dE1AAOF?E+fA0N)BH*FfM%UF-1t##z^#(lUAU(b9a%Xe_&=FSiIOC88
zCJX$Nx|pLuY3?t(*kctye<3__Gj}QYvThc+8j}Z%QeD35CZ&QIQ{%=d>qme?PmF(+
zw;24o&Hp9mMIl&rQz&%I&j$61rK)*9<3Z20*l3CIFd!6j{Kxj0BJhkwG0eMd4F|LDY;3=P#yD|4;ARU0y$GAS@oH-q>!j$DjyGf;8p1sWKS7g%11S
z5(S)Z-ly+8X$~0mvCO921ehK6_nSw20j&J~G~an!8FUSLCMb753|Pd}RCVprf}N+X
zPlVl1gP&sLs}v0jVHI~TM-fj6OqEXezxg@_aJhfTlh|$uxDE+?9`eqB7OC8yrWs4&
zvg(sod$lUqNteig8N`CKUE+L)E~ygdHKd=LeV7ipjQVbgx|hOsL{F}4pc+OWl&L(l
zxg6%#o64LGsf66GTbQu{2;Caib$s7f4|(;rtQs~qKqg&}^b*-JD0eV5*lVByrpD*u
z8EiH1`pDqR?~DzQq0T8%L9
zY7J9pWiw2~${c?%SHbBJS7olB7!-cKQRb6#9X!D(xARGJ6RdvJ694FOD-`OP+OOYM
z4X^SYyXJDQ1|~M!H)T5!{yg0;{ZX?88oj=3Ka(#(z5
zXY(7N-F8L)>EL<@FL~Rv6xBfs!JAmamqw^nx^g4%coXbCdY&osPa_n}yLGdYt&w;?
z@%f!4odzh_L)tvu)(mSUe^yVou12iLcc)xoOQf_a9Q>!I!5cNyo5>){*D@>k+jbuhw-
zOVgUC4)&cqPakI90CT4C-C30luu|+JyKh+o{L%c@TK0E6td-X3DVC^*RZ7l}LJzb+
zlU+B?q!RuL$`xEn>1`&y?_Q`<4QqnaQB4mL{2O5`OWM~c$(K;YyK_6oj+bx(1}-`*
zx5Ls0aU_%Xt#H5kdTEV|EzmmaN7`rAE;t&tMQ7dCE_fgm$&X3tgz|Ft%@zpzb$5go
z7A4x?$3uq?h;6Tf@uoi*l0|Bu?E!t=Hr`4I{=Mkbt0{(tk19B`iwof$2Y$)(;`NZ5
zA%1p?NiDQE6<$_5Uj;9(i$5~$QUZm{*gnaSiXf|t^i=1=M!0$BuCwnl>tSz%;3g)O
z8d#9Y(0Pof60X~*FT<@<3OQ7s_GX#1z~p`BWH&T6!HKW)ZOL}^Ff2bS*n0{>)*~;<
zErm%?=cUif?SgI4rxCZip4kG+GaoZuIoSxm(J!7|eqIYNJgyl2egwj61{vocye@>B
z&n9NP`~3_q^9iS0^F_h?50amZ_6CBa8UEMG!b!jn_tHliilBjrj!0~3E@TX=g$@C6
zaI4}ULzq?zgBzLwq{aW%+%NZuJVRH$1^q
z6&em~3}iXAl}M0fcOi56av2Oe5bm|~@HwpYnUMW68w6XA%j@S@`vCs8F*+=VA^fTR
zD2RVw6+Cj3d&o?r5FYakdgW>n1BFx;&oZ=}g2)xs;c%TKVC~+vD|~Y{NIxvXm!MM&
zdP8o{G#sr24x0wb-se;T4`>zZz^!^RJ@$?pb%_(ZJK;4tsHP%DXhHGUIk_>Yn~+D@d2#$
zhPyfUBm+|0#wamqVjVag`Q~MCCGe^nlIl{&fY@!bld5CH{&l@iO2Szl0qvx3l0W=%
zz{q~3Rbkh1AU|W-9k(9?29suADkCa^!Fl4;(ETLv-m387k6jSF%2nd%kg5fFuSR(_
zU)6!#A_Xj4KU9IIR*Sz_fNyj@7nH^`QNsW5QxfBd8wJiHP0V
z48%4JL|l1N2QF=ScF-!i4&3e95fc$r2QFhndFlPNV4dz&`>mvE@T^ZjPglAgJn(lq+X4>Sus*4wZv#Ow>?w!++raT7Z@=dlw}CdLH`zt$Rlv5P
z=+xcawP0e1dDvc~9=y|Ds4ODddAGFt9;gFB`ng9ze)4tTNYau&OJ^sw$ZezLnez(v!6GTaPyZldpRpj7orE(Nz+dl!PBLLaeSoqAyRJ5X*k
zyB4foXa2Fv3IiNZ*eo;8m4i?A#>31XioxJ1wJgrl4dB45a=Uw@b>QN^7n0-WA^2*K
zm8ufs*C|kT_PA*oP|%p@2+Jq~_NEuFh0qm&ryGn^#ARZE0*jkPlY>8$&z}%IvhOiG
z_Itkm1`!Y0>j*?flsp5u8*iI9m`8*BMXTCVORiAPQ1oWot|-_}pW5QHQ~Mhwr<3^&vx#QoTK4A!lTN|=JqoskZkr1?K7#UZ%tS1<$3n|>Q7Q4s8F1&&
zyKP_g7edFXi%mG4
zV9yq{sjVgnaEkS4wOiO@NHWW7kP|L!~dRnRip5ozEZY!;X
zvWiwda||zFOdH<@A)#XUBK1oPzFGn;m}4L1v6MqujdvP)tL5;ff3mu6Q!$ME?37g<
zQw|pm-)1GrR>EX4y)K1o6|j_P!0!3S3RqTuwXs?0hABRP#j*jQ-KVyvr1W?!J&;l$!(>)D^Bzj+3BFDf9Fq5gE=g&9>P0#CED#_xo&Pr3Mo$za~k-zo3
zXVuW~dDwgVQVc$vKA!J&qXz24GK_QOK^S9h+xGED4IHFz33+H$1K&=!9Tx7#VcWLU
zM_f!8yp}b8VYd+peqW4{(mh%Y(`
z>eyuQc5u}|X1z-*&D=F`YI)~gI?WnbciZlWl%h(U3;IO}SON`p`#RN(do)>>G3ynESW
zq85s5blvzYwFcfBQu^%tEEBC^%#na{9ku~t_iT>n2;}F_>^LqWJyB&V?+i1^b
z(gRgT^gMrfe1K&RgT&tyDEJQx|0A#2mDAGy%OYBO35{#^mH)AS@gMz~eJL%yn8r1K
z@@VpZ^R3Bi_G@>s_;}`j3Aknw#K(9-}dtMo8nD{hD6~9jWELq3N?|?4WV29UjwU4I0O3d`RP2
z8b@uY<+<5WY3xL$4^8fIp~@X$RA$ifiPHRFrO6eve0#1_?fh)1!S@AFVzT%`dGks-Dw`%6;Bco;RVgI+n`5g?E(xJ}*j@<7s|equKurAE4+r
z)Adofa-V8vMl1K!^CW
z!&G_XEmc1IyN6=8>)Bfh&qWPWxL=0qH}7{UcTbN|(yg?p_I-Vm6xmn4ox?(n<6VPQ0+3xIw>-?^qoTWBW)DE+fMbz!;MG&_Ss;#4wX0+xF8@7I>}5N+eb_RME#{7i78_WeC<3%(HBo^6I00)an43^
zD4DXG%qg;(b72ix&Z`q8hTs%v?zN-%wR1;=k|QBA9KDuKPJg_fnHY*wKoqJNk=ZsK
zSW4DcUA}V6iJ9lGpHFsk|I1lcYH^LR$~b9%pag$+S^#injof-|yl8_DUC<@;342;R}Y
zfs*ZiBdYXd=GmXECYLwBV@ym*ME#GuDeanY;oEVtJkG;;p3L=?S19eLZF-KFLZ`qd
zU(aV`=Eq;8T>ss7QzGQ@;-BuCAg7;K>*^)b@7!q$r5etBBkT3rf3=a*Uno9TCsUUv
zTY=2>OH&49h7)g3UL{x``9#rzpcL=xI6H!`#g1p%6Eq?jMv>V+J4ESc!YWL}lsg3;
z`Dau5jk^qW{Ou7Grrc-Zqu^|^{;|3f#qS>*Jt^rq?d_EQ7H9T_QjXr9Tm59apZ|6=
zl3D%iF{S^FNc&Ry+s{dBN;~Deh$g1=DRB73V@m%TkezEK%N|E(wl^8kHm;yHYUL$1MMg0vTGjWiunatC7SSkLBZ=<95
zB}V^`V!yvU@;KK)oMcYSel5pfg6%Yh9quTZ=YSy@wQaUaCJwO;5hJLa|eS
z-%H`WSJZh_qM52c^ouHg*vm?}o`ZE8Dcr$Ht+ywbTAzK~drJO{zn3X~a=fP2BVWTz
zvFBE8rs!20sqN_dimJERP0jz~HkJ5QDy^NV95bW(arzcjp0}nFT%&TvoXX?6)bdWy
z)~yq?_3|2x^*5<@*~wHg^-}GxW>95UEh-mj>silFu!#2}^QkEO}PVNMjehn7xs28w*0#>2Gqfg9BHwe??xmQGBD6#I7C_2}NA
z%C5Ba`3|jJ{?Ysxr*TD&>Ze!&)&4ClU4-UOFD?IRTKTCouEkGl@epxSDfJuC{CQ53
ziB3#OUyB3R;;mVlpCA6CUn_5|{05r7fo4xk;*|1;MS#L^8fyZm@(-HbCmQ`}{`Avi
zA`>OwTHL!9x2|#RI*7@NVn=jN3W-VQ|71d-$ZK)qn!INJU#^w6)(&fN|609k`9o-a
zuBF%0>LDgGN_}f__nLmKz1B!fh7`Ls{o3`c`SaiFSkn`W4JF^2-I||k@>>0Cey{nv
zmYwMCf3MIZ-zfFMX;!}W|ql*&9JdO*ZBgpmFsBNOeFVF6e5kq(iOlX@>;uB|wNMb6R0z8b_fE-8qy`AbH%L5;7
zQ|h@>@PQb*Q$UDWZBvOZZ$0^}hOAea%W$NW
z=+azB#O6|Y`;OfG+&
ztmvv2!GRCodW;Av1o{N15tP(xR43=N6|Lm6CD>KH;O0i~<1b!4a=fqUFuv^)!8o7u
zi{x@zpK4*`a-&%-KalMm6cZxN2`=CJn?v?5v8N)JTwcM=D?7;cgF%z?WIxk|zrwo&
zzi_>bCYSFfUqLMGL_6>eUA81xhNpFh5X`&PbgisI>WcdIiDp
zgJ&=AAsE(de>7!a3P?X}N~k4dr}p_W@^eTz>GCJCKldDwL2~&#)+x2*cKvtnlOS0Z
zjIrB7_WyawuLIWzwhB*iljXZjxy)n+;_*^smTI`9T_;#*X(UN*?=<$g|M!8RHfG8K4TVK4y7He7U(yZXU}>C@l-m-?8fR=dm0iE%^zNyFoWSVVY)tfP;*yA+@q8&w?8^8w
zW%qn?DZ4H9{SUDhjZ;O${$%(U6Ett?BdEHLviqlV))Cq8?LE@YL^Ea2wR$Q0eq;}o
zAL4x~orU{Z(qQAm9pHV{zXkT8Sge7elniRZHoxH
zD^dBFigwBmU8m>4c12TujE9H9=kKXJlAkc8_mzpVznQxz|F*iH(wnfQ#v4{o;iE7r
z9*WkZ=7X7y(i4o|L&hbxk800tJ=MN=JmpWXL{Zmkezzp|qhn3^pV-ZLL@xFiC)hGV
z>5FSpp6%UEfA;7m_%npa0KO;h`brk@RuK&}CSZr2Ib~
z42zw}ujuW8@gg!Tp8vsWx_pby54Y0wx#{}<@!wl?Jr?>nTj=_|bp3n(>C@#Ly8eH0
z$A9s|e|+1MUJnn~|BUCqxc0wzB931FUwdHei2i?Z3K%aULvo8CI46StUH649i_=k$
zKJEaz{9k)OV#sy>&A$OXPM4s!``_>W-~BwKm)FvtgC6>Q8_~5SwQo(Y
z2QotX)26QvF>6W=r}Jq&dN~9A`XoBv?+B#yUee?4s8f_|Kwoc*=iM_)Os&uMEmZlHNDB8TQ7CnlLPc)sIi)?KWHx;Y1L*Uh`;U?f
z>GojgFs1*>m+EKd03}zZQ}%m9GKK2|=ytD+%I}CLQsqjDls!+dlp=P|H}n^oHy^s4
zmG-0Jl7DRzq&!L%5xtUn%D#yDQ2KIaRQpl;srCj2(}@3>&7$mJOgUvoH7uz3f!&(o
z^M1FgiM{6+qIkXe85-&5xeT(&+
zwp7kxRP|T`CanjRmkE-*?NZGN2<>}NV~I^4#`zGFM9dfT`R1FktfA~0iLuDhs~e+{
zuBZq`r0(mDd2!z!Jur1}o$7v)f2TPKv|xTBsa+?F`A50Eo;BvdEqpuwVV)TmdMy->
zS0FB#6<^1cxc@NL%RXqb7xUrt`<-04-SfI?zj6K8lV`=4SM%tjQ~H2Wo1WSk0sdRo
zBaHc9v{k-5=65VP$Hq+opH^Scz&vf_$Af?*KgmTGfHnb6o|t#Xjn_UI1U|l@RfGF3$nM8_3J~Ddz%xHe
z0Yz-c2{h;^wv$N
zby^AZrQdH}g!$&h!3rZ|z+(N$*iC@@ZJ)1=g8Jmv2iM@^6X?>$_uue5s+R55+6Bm$
zF{kDRUFz!_{le*3D`))AbehvEX9%!}e
zgYj`%3aM^@>ut7i{&@%GISiw$Y#$-sDE&1QH3;+j@#dYd0f;N^1%`5f-kzObPA>y|
z@?Po*=Fxn!{=Y&1Uq0`3djt4vpf+PK)UR1Jk<@);pwL1>VPNA1}Dvjd(a&x$%o^#9u$zWk--4UQ#Va>6+=G;SHAh)0y$3Q
zVrw(B`+zqk;wIqr>n)8YfCcOS1>*H%Rk@Q9uakRL_LB4E#2l5@p=&DTbqv8U$#%b2wp##d9$zaI@Z&QJYWv}Gi_a?
zy9@B2(_T#+r*ei=CmsS+i|*sT0eGW+5kFqHUB8wYo`l~`f<~L+2K~+=HJ56bC(gJj
zxnjUY{Q>fL-9A_SmVw*9n`gy{*KN)r$s-s!)u}j1^o+3u9%ssYg&>Y|w(I2jV|*VP
zXN>hS*$jyt$gMJR!q1;?$k}YH&v1qFEPnnO1H=OHbGq|;DzPJhQr6#_c!Yn2svJe#8x3BUJ>;wAQVe$_Cs%YC0M!*RXT$3;L%;5p6YBX-p4r3$g1VilFd
zekNUXC8+5vNAfWvw|)?rxyypI!+LKgP?a7e5|XHc`?XzI+k@rWvNA}a_8_}QpNL&9
zKR*i8s0a0kMh0N}QqcQEAIq=u6o`HEU^FB1rpA=uh|90ObtQIwL5zx9gk8$XxE#tU
z{}!50`7yh6Dlc>+j`Bkv-m;T%$;49mCpkT;d~ebfG9K?0+N2#Ho(O_M)4PfOrOUxY
z?v}A43px!oYHSGpzy^x
ziU&p4QRDr_K$RyYQn-n}f0I*KM*6oar+Cm#9%?>34pQy^qsN!($0;6Y!d*-By_Zn!
zXFj9!HUv}GN6~rXe|cm-I!_Iv^HxZTkn1_=JX3(4Uv8l5{l~{k>H6mM`&Tuiu4AFc
zJr;ERlXOh7q3$#8KBf27n8J5gD1=RA((kc0YM%3!Q~LGvc`Krihy4zv_uo00kM!~S
z(DA>0vLHIIzCW$maozHL
zOrGobnHNNF#E;iQhc+U@?TYuus71LG1M%Hl}bae@bS)
zfc^2^=@+qBPG+(HjPYFSCV!0D-+!#Ye#+X13y!)0&$*jOc(xJ4ec~rYCxuA=R~_;h
zaeb~+S_r-_qlfc7MhW)(P>gT)9G%BF!uj|y_Pcp~M(}6>uUISp9LqG@seXJt=V!@g
z?8j&1d}x?gcyhbeVE*%LE58HQ7u4zK$9Qnn(F}auE{`ZqT(1|r%IyST;%LJzT)uf*
zz-6o_ejuq6xBJ0WkIav?*K-fd+i$P`NAkL1_t{C_x+1fv9P^K^dVOOHz?19>2QWVq
z{@nin&(D=DX?JmbSeS$#=D+PH*}{DQe;mBJ=@ekEd2B!CaRJ-p-{JZJmrd4~-(Tok
zcLDQYR&kZsWZ)ygq8`rpzJj)f`{6j^+O*Q!e84MSCBGw}{;geYSQN<3e60qUuc8Is
zv$p`n3jBv0p?-dCbhaMwGB4pp9+JRIwVMRVIhAu=TmizMSHE_xE#}ifUKwdO07r+j
zVx@q8N_y_eDuQ}9Kd-7xKyeX)7|h?Mc+~^^fS0qJ)n`csJ*Fd!mWhB%7LD5q!1e1s
zSoS1*0iLqiJjNM#?fduf=f$Ub(9q;cfh%yGtzdBCQ^4NIkfFkj`Y6}FSWpYvml
zgx^B{$4|Z~hjtc;?a|Hwy~>h<+kN0Z{F#c6tO36KG`RO~Ka9g6%XsZ_z+2vEsoZBc
zr}g;vt5)Fsf`ZXAHgKP%={Gi{HUN)g|H%6juD_llWHZ_ayszfZS?yZ*T~xfq9TXoC4Zuh2E~3C4GBiI*w#&-C=xfH@$0=sT~a
zaNSS-nz$)w51skAX08M31BO@m0!U1SFg{kt;jY%mR(ly0%VGB$;LW=Vrj*)fC{Izt?ik!3qhIEiqtn2j2VE;%FT?zMNeUVYk$b@V}bGM?-0207RGzw&4bem
zP*1yktZfExg<0GX&hH--n9L|@hxzDCi1`BRvh7BHi3D7imccxr3H5e+^S*FF`+xYI
zH#F2hy!qUi%K_%`m$bOp0cV*Mf
zP`+Zp(gQyqXOe#2I|-Omy^a~%4gN}lb=Y1AoqX1D8^$>?FH$fFa!b)&25dhXE^HdY
z_P{IZ^m1$m%yl+P;B~}ZTtAEB@!9pweAQ4MA11K`??0V2y|o!X--jLN%zhx@!Q}{lg
z6|Oml%LA$}*W$QnBWrFkjwdw|5?14KX--K;+>hu6hEi;Yi?%KoXvcDF*J2z;Hx>vl
zH3ghWmS2qHL{^S1&Ny!M2)2Y2E#yTyXv9BN^m7t_)taZ4kIP-OqDDh`)Tj`VNUVj;*mJel5%<;49X9u6dd4w{^=+QhqEuk@&*{QbyIdep&DWH_qoI
zX&RFI8$g~y_y}QnmdPgZwqVL#3<=@)YcaeBVaPf;8tiK}eHQ^7BO{sIY%Z$^>@7~9h
z-Hm_8zpYove8s;lB5~_PXabr4bhU2Mj<4_*(!O#&mB+L@Mdwkw==(^q6#q&q6(@T2
z_bI+Ky_Mo&jejV;&zmV8CYd-+{JeH2#jgggQgK)Q-jC#d?m1BVYMC~TWZpK}n2>T=
zWvXB1KNOGB<$q4fHKGrfXCJe8k*ju6NcssBO0k8gC|E*(h^E6-8mD!0fZdfS-k_uob7kI7T@J)hf1eZVCJ^7|S4
zQF@+A)cmRLq~_zJDz#s0dUA&7E7RljL0yU$|E{EdpD~9*Qtw_?L)P8ro7B2_qeq=n
z)v2NM7qNN}y$>Ky;%8_Tqon;M%J?^qZBFM
zc<0gw_zvF&aWgI5YEz2e9yzZGGD*xiBu1lVK!|ZfEtj8J_P3&iYDHA#WB5MAM
z*b#@dp?7h;;c6)xYzNw=`N%jG#iEEke7diJ*wtyDJQ~)!kg#nY+r7)%&XM`qnR1yt
zcS)>w;&A=3aAR?7FQpW>lYG?U`lJ|Ke)8E&B*rz{&lCIT{AcMc+~2)JDIVCrxM@w3
zbKM-LPr;TL`0G1Dwpf3UT3Z*kuZlmS{qT7dy*>UR_`3L|E(h^AMEfW9W4|(bKK3j2
z8%k-aok@UlZ%U5hcJgiRGh=;M!EAEwrtt1Y3*4@J5UR&~c<@suIfvtx6Z?+X@gR*!
z+|I(@_HfL#o#&AGZjub^{@b}vuu_pR~
z>~Z#zE%w8?G8Z$kzn3|8VvZB)AJ_15JAvOe8Pk<{0_D1>#ds^=mV24b?*P~Fa~{Mz
z=Zcm1Ks@lO_gk3Fa6bn$c0RZc7{|Q*AI>-G@I*yChIYC*i#Cn`{}B}AV8Hp@(Mwm$
zQvhW;CMJiWz9D$#n-%oCan{@73Gk`Cm9p9rP(PxzZ1;Dl*RPwdGKTtpb5hM&_`0sL
z+>vVFGjS$<>!E)C4WV7)z>gY#Uv3@&zSJD^&@=HBoWs8;>Ee?M^HG%bNEzzC$8A*U
zf_7HR(oAOo4L&@2wIK)A=iT)Ad|nzVcx@7h{e75jSIkD>FG6wbzs-7k(CLBlh0+md
z@1Nlr+xrmzY{|^hf_|;#rF9u#yi)}_R&Ed{D7?umk9-ciNwO-F^Ih6_E^B!l}o%G{@m$+`px&dEqPAMNpT&rm4!@^y4?hQ0_ILqi+pF6Cp
z&&wYCIoF5!Kc9A5YBh?&xXO?Dou?tQ`uDA$uhWpi!B^{ARss*55c2kZGlcFGyAOXX
z7)5~w$1mv_L;2`JkNX{%_n*>XvTz?!Ea^ubfPc=vTxvK2<7X=>(%a4ee5}sXq;os0
zpWMltlO@1E)Z%|12EM&e6}n0p;)!>AdiL}WccEpp{aX$_??O+S_iqrX??#zV+&S!E
z9_wnXhJ;yY$RW_2Cup)0U0&xVY@pnU)SL!{3)Xa@+p~o#9h=}B5pB0lJp&B|MvR2r
z=jcRx9YW$ZTkx?J?*1;XchdRWjILS+kuLuj}L|cV3pl
z{T2&v<&RuVLov#;j1OR)#2SBbI{@tqTzTo93-`I|bzRU9+|N&1#?6N!G$dNT()8CF
z8v3y{TSyJ!4a#Vf-z(wAVI8WmQ1
z@xDaUbXl<{;L+cO%s76Zc(l|HuTS-1j_Y_orQEqbU=iqZ^{qh>AU8!wiUb2j?2cW5
zdH)&~?%Nd*Zx)-^=Gw!3ol(5JI}qllILGkAW03t7pB%-!{nziiyS72xRLrDv?J2~w
zheJm{Hp04P?cc!e19}rRYz1?Ge5iF8kF%Da!4%JL>bTr8Y^VIa!;`QbWAQx+M*v_x
z|CCC;0?8q40r)J=KEgq1KE92qlOAi`$Dp`u{VYWonINmR;io1Oh$N$sr
z>N3nACm#8`3){o-YO_>qm+Jau-7B@?L{q&=SoWp>Bj_w!0^K#R(`VyXp
znsuo+uuN0@qlWbyE4+!H70@atSSdMb6<>TslYcX;BOxNPBW8sr-GrYbj~(p9;mR
zWv);>>$3&Lqsq40lX23PQT(dgpW^Ki`zYR*bCcrtiXU7^J08P3h<>Uo#n(M!D1P?J
zg5v$3`Kn1g<0z~`^qY=RJkF{*n#gzMsrL$n=XR6(Y2Lhuj5q)6ani2E!FZzAmP(zQ
zUPeFfwW5XUr*c2V7Xy2#@h@km;;AE^RC(iZikDjKqEL&0;**<3sQJ*1rub^dYHA+t
zY@}qglDdDHu^puUzbce`rH{%Bmwcl3vpS|+NquhL7BcREtJHg$`_!m-=1SfqslTZ!
zNAT;TzXXTNsQW5WqTW{;I~+mEV}fJJefPtMT=9D4ilpKI-G(_*uP#Xae!6Q?h}`26
zP2#eHm0U#Lu)>vGA2z^8>klD%a^VplaeaKMUphwi6*{E+L#rhw#-9O99wg0ecvZh~3!d{DC}&
z%4_z$$9Aijf#Rn(RJc!Ly~1T7X81YK-;tw(?V$EZuPU|!&;8!0;d*IFtLqr;FW1{(
zJzB%@%eb9`AOB3_ez(~V24iGpFxZLhfDT6t9O3FgD_;01YmfDY
z^;cEnb{5HZT49;JVAmY>Gb1b8z3&3*g)SAvcGK_jP7hrEk(pf{x3}?ba|rgQDP0o<
z#eo0LJ+#N=Yo~ivu-)HRr<#uK=Q4E;(MN#nd56~KfxSh3g$p>Z6!Lr
zv3hVS`1u=OwN~Z>uQ0s+Qrxc|c-$g`7J&%J*9(kvBtyMf(v?pqfj5+#oY?gQ{N?RB
zF7LIKkZ1ICcinb~hN60HZk55gj>^$3r^|Z4k3VjBDh2uBn33N;ry!sFnnCxC`y>r{
zMwRs*2fwUlzwar|Tc4@WJ>n01Il$rai4)8;)R8^2cpP|~nYkE;5aidS_fBR+K>wRp
zc7J312mGPuk&S^G@b4p$s}c{>P+G?)cexVC3#U(O=?K%%+a-HP&iR9X)Hxtv;YdU0
z3-5jn-#dzsY-qc)Ee%;Jd%H@w($MPzdA
zlO7BqlylYn@`^EJTVTVsYv(XpE%WgEaAz;t#ru72LFy2CtKHZAed8EnKW}z8^2jJ+
z8POLlYZ*mq8`f~UON^qYuIqJ_Jpeo7vnupQ(J`5!hbqg)(2`v7Kh-ZsP@PU0kDz2H
z(%XI~F6(j!iuX18S9Z9912tYc}aX2dy2&+c0wP
z?P^EcA|(ULguBp6>5OCbce_!F3w!xHi4NrRD|0gkYdd-skYmlV{TC{4uF~?!Xh+Le
zi4OJxf8}tA?_>9FM<>b}4Az+bLZ@<%v3+~nirghy&l)WIgm*6kwAnA|7s|c-
zVU^^|4#?jQMz+cWpFGdDyh8`2XvR^(S`yJ%ve-9
zSF2Do?*Fad?!#Pw8w=hljl=j%neVmd!hP1vJ=lulhUeBkUGD+S-&$}EL4POP8fiyC
zp5L88@(>=I`OUB&sf_W+#{Oa70lU@MU&O@B1md`4>6x?A*guFa@#@0m$}3A&<8ej3
z9BIaQ;>Eq)nD5%N=_zBoyZ>-TB5vpGx?YmEh|69KpWFbud^XY&kJIB-0VmE!9P@nb
zg!%n-*MSD?X9^xKHo@(<96?VjV4QMG8^uXnl+HBe4&|3O9X*2mq(zA?J06EDhxd7m
z7Y6=z;&B8twGuzS`$9AE->=5k5k4_?e~RRHlKXZMKEPgKL-^@TzY^gCs^1O}zaIN(
zR11$Y&)*^uxA%-mJ09bPh}FceYh4ab#qyK-Z|Qj4S$tt{asNKbA7yYkn-=w6NRudh
z8ysZSh8i7`m+AgT-Us5F)a1wIE?HfD7{yMok-YxUp$jC>ZScX0j3X(X;;Cn<3kd)D
z{FusL|4ObVyfa&xdXN4~pDek*&%0xZUh>oo!Pu2YVsZOXZ5^at^<7bf7X>a*@6!*c
zijnKa`LZLi-pV;)!o!Vq1j+ueOo9O!N7Nj(FYo!ViOgF-RveiRT5}s|?_+lbX@^0@
zi0~w>e2PyFuA|1AFHD^q`g!dD8JDXCbzV5T?E|R0xrc~cwqqN?kS;DvRF^x455y;&4c@gB`6gnZ!Iz;)}d0
zmbcVBzJ+n*Ckx4M7I*Ao#LsiYneYkRp6IrcR*a+ky`-J_i-qv21H_XXT5K@y%-CTt
zg85uPuZS|%YaVYR-{COJeRT)7`&g0Yfc4(AtqH>RP2%mH4&49ub_MeNj#9%Qb$s3S
zn`tD@UOIjM8s>AK8HP=8`N%1^T>RWEpHWD`<6`;i{Q}z&6A_J|0zk7X&-Cyzum-!AJ#n}W_n$PtGKKBPt+2Yqn4f-#{qP;Se2RKx(qk|7iOTRSkHYu
z4E&q4=r^f|8t|W-V+Y>Ef&U*os9b<~CWFjGHt2`_z8D+_?OhpNSnUq{HB{}f`bzL)
z7FR=OykVSI|8=^)g!@xz|C68t{>fso?QKcm5m`#b6O7`(8!|Nba=@QXetpj;26;gq
zUv0%o;Fk}4`nM{=`2m}4tN89#(9q4p<5BC1;k@^Fw1vic8rqe*wrc|TWSGTnA+38f
z#AdX0!1Uq>I-1_*aUA?mNNDTOm^=;TB-+1r)C9iI%&gyirx)cM>~A{%Wfbi-SfnbY
zNkdQ5`Zr9hq#@zRZS}X5Aa6QY<`!%@fb!#h?OLrrj^>O{I*b!bSB_dv~3$e(7~?8wf%3+GjILfh86j3JtJ
zk675rQRHm<_fdFL8=8C39AX&Sj`$q%q^d2tk=n0U?R}eQXjU~p=Z*jk{fd$b>2z&F
z;TF6@=S14kXUn9kjE!B$!me6ooDuTG`<4dQ?w}zzPN}Pd0&U3q@od>5{a?syM{2!Q
zXeX+dx%2e;H8>aUo$~v+Amnuqc*ghr1-=*hdFM{8HuO~O{JDQp9jNO4+p`+CyV0N6
z0ioy;cpm56cjfoBBAbSimwn$`k;y?ni!Qr%WIm9wd~JOf+Oy)_y1Ou+ot?|K`v^h*
zc}Fd6-UDCYE|%_qyt0PA+_6+RpYe9;>yCN&z0bt>xQ?mOP`{Bt2siM?!d8~3Fvw%7
z8kfWd!0-M34T^jXylCQREqkXfocH+V5_kuAEGB3dIKeZz2>JH#R>Q3T`xe&hy
zq?}AXFa-WkY5A-&%-8L^I+p}s9jB_S`->pX*&~{N^9Bc{YB^kAb-({qKlS***)RJ?=yMP%!@!T(K&dV7L*^pm9uDwyZQC`s32d#GP^
zvO5vzC0A%_$uAL9#T{z*eg9YU4AF8u)JC`zeIk3Lvxa22&z+6_IL2O5p{@#Cw
z?eogh30rYHvm!sp`zKna+<&nA(;@l|uK$z%)*t5&B!cdV;qrU$jp{Haf3G0?EA5fd
zYkWUBJ3P+>GBniTvha0g;>7B80
zzmxoMeP*r?;W@#34wC$O;-n*)cg?reM4n!qN#;?Gp@ihYR%~5P=E+)NFDX}QOd|cU
ze$XWGz^mULWIuMl_Z~9euFStK;(q+?sP8N^zn87Ve8qWfxEJ0Z*uHSlnAC6P4Z$cB
z7lQEb=;PCJ5wG7*U*{Nb`DK^Y2(L5mJyivG-6@t%h2i>(ca_C4&p7wC!XguJqKwf4
z>*bBTBz))^-?bmOKaHN_$vBP(YP`lD1?VW3pNq>MZIsNz?HybF*vfaBtc{69jrrj`kt>xZFX)W)0vC*>=<4fcz`H&YFN7=}sA#{{sAlb3jxt
z9Lm>~8@-eU411KW(1_7r^3pKaVGE5XqtcLv8|?hrg!$Q_Qj0JCfZp2`7q10W+>{x0
z5&Dyevp?Vme1@lC6Ri%)8R8d7*+V~9y|({;1NG|e>>IP7yhdpzLi{z1+oJJfFvvHb
z#jU{aE2-QZirWV~$51xzupsz5QHhT#ACke2R7)*g2mD29Z{TDE#O(!PFZSOCf3#AN
ztwt5}|HW+l5#32c@h&;*XyBioqeG!v!SDEaiYq;@0{QZDo8??Iw3%yPdMYdM6PNm5
zd%!O)F-zeKf%{Sm|MEiv{GP43$Im427j;udzwGTC)(-*?=yB*6zT^t={%6LQWpHjWgY{Tt
z1=}c^42a`$fpdd9`1+0TT^3h_NW2oTE$MZKFMv?VW1^s2~;QVHx
zV|8F&H`-x)MfiwBAL6|f`L8T`44v@I6cm^mMw^=QqmM#nK|CS
z$zz{e(bHw-lKvOkQDx+snCi+d#9-diRY~QmeRB9An-TUlE@QE?dU~a+lQ~-
z9VokCRD8{fHdNjq=8MHc%b)7&fC(5kkSt7kobp*I^>ob?rKLtT;r
zpTyYP&~<}R_kl%i$X?%2{`2`Z)UjE>qlxnu%DWUd{sef|iF0aSIVE9z$Zei0f^}0F
z*S$^oCD=6v)42<<-uAg|E55$ag?toWU22DQtK*Yb`InQ19{7HEe)BoRA-Sy=m0`VR
z2XFDeFWG^H5)*HoXy`&~TXsEN4ZL^Z?1f!c(C*c1D_llPz)y2j=NReyLRXG0|B-dN
z1Jy0=j$F+SapD<17O%}Tq$d90q18XI8^XbPvU6?drp#PTyiYr-4YaFQlJ7>jSC{ex
za?nsi{`wzlfyb(SmSE8x0Y6=bCenewWWTtigwNwvF@{b}!1HD8d3NGzh`^VrpV`i0UUJ3OU2L(00fqti1uD&P8_g`)c
zy#PEmqbm4*Cd4aGxf`#U!h9X8i;B7p&tY`Gp~7A0SK`{I@DZ>Rcg#;c7J_k2XC*Ge
zd?D%5{C=G8I#``>^EjYy?5k{RJkIQ-n%x~-XF+{{2B`5
zTNJP>qL5<>aNyWjpc3e@ya>t&1KC4g?M4n@Nz{tvivgnoCzJI610tm+v3+kk
zKSI8P8PzcipV))Edg^j=t|%#eE5doccfp@aal9&bK)Vn7qkDe}*J3~97!ySDFQwX<
zCos>HmYP40^G>YG>Mr7Tma&(9#d@N>C*c(v;F}Zgr&~pX+Aif}s=gmy!QW$P;g;+g#|B&?+JK-;l
zKiLQ`68OzP{QSh@3Bq4)?cPT6iZx2XU+{Rm{G5nCPS=Ydc^tF*)H$`NIR+oxZklZ}
z$u~u|>JZ+r%*m(ZX?H*hjo|y^
ze4Tp>`_+I&#?rX{Pv%Sy#<=w80gUZzzX^X)ET!^VLECx>f8px9Me-syy9-Ib0X|Hm
z-Od*jKOFzNl-ysL3KQW=qDiM5aeIle41{NOmQ(w!g|XE8K{Aqe{>Fmj9VA9k%IGOkx=e~|I&unv)Z$n?X8Y*=qS
zr@s-#Qik+!jEbV`r0{%d=NP=g>+DMJof^C@*2(Fw#`RIsTSp)Qt7_O0OKp`XwB>4H&h>*o?4V*l6tpUU!7!!{MhNdwhSDf8TIWyI0-1R+d2L#tanNGl$$p}&ZI5Zw+1}qZK{jc
zJG1n?Mg4%2a_NJIK<~FyYa5BaU=Ji1OEL=~&g6Mf7t#y7YBbu_IwuqQVNI~R3;IEQ+f*RHcs9zPD!h!cWeAz}uC!xKVY=ytRZy+ui6LVQo4)#!NwH6!b
z#f8~*w1S?#eP_)c(EFod;}HRRm#oZWN`VJe`ma`c2ln2;wK@UY`6-#1RXag1=vRoQ
z8N~fee*Ye~LOoNFe*g>EZ{g9%f&lPq&HPM9cR=3fPwLaN8o*y>Jd_#-X~XEsM93k$
zuPQOXFyyut-rLaEk(##yz2@@Ip|^X{+{B+WO`cJ-aizxDU$CF~H?gH9gd6fhGeJ8R
zTYx$$idvC=e*Bv?<2H!@kSU&u$Snj<$w#0c0%8`{X!8Mvt94i+R@jv>7EC5UFeW%
znUnoEyl-J3^|liDSHSifkNKKk=-H~!Sp~&*RHA7+qRQWebRR5e?u>(dSw_3v^}s7<
zTed&6JkpLNU5;k&Jk^fWTHGX!UUi~p>M~y+7{YoBpWZk67x*`4Ut^8Z8siwB9*;|t%uaWj$7n@lLY=*cE5V(O&%Ibb`Z=E0{?59aG_ur){{%NaR35;
z`6N}n;5$4YH?q$tbW70C6{jHo?NFXP`>{3(_OG3{uC||qJW3;z^6e&AXU?aOw+|`6
zdqK90U-kdM^PDzz^I9GFKfZQRA*i43+7fyU@<|#Oo!{TX{KNQcL?ROnrH4wK_z3MK
zvHH&ag7u@AsZ{C@e8`|ueY_C-dE6OYv-hkppSPZddSE_#R&@t{|0(0a&q590A#o|a
z`?kP!*J^8j7{hwq!Jaf453H`VO`1<6inmyU#9FT^i`pJ
zyTpT^oq(^4RbR0I9nw*hy+eMW1p;r!6mA1Z%w+?Q2rRNe@B5-0u$DT8b=)Uh7thXl+oEMUE?t2=1e
zKIw087{%}5pgQEZ6J*us7i&EMn>O$;;PbrOX_A3hf9}S6L!1{%z2q+|33BPD=?QF)
z{k>ndW4#}2i)FDLn{M9m6Z^|-F$qiTFMXbhDPa5j?%OoU`=Es>;y3+DUhrd{pyjlW
z9pj~Vx!qWAeCxB1)OU084X`dAX@e`IU4!+Np*
z4B^#e@I!{7#9wA|J|g}y&h9humv;vTh`(Iu`@0nDtDVs&d6Ph~7Q&-yUrv#H&usyJ
zlHZut2q)iBJfN_a_(>@xU81Ko6-xGHmW&S(z5a$T@9}*a^?xJ&Jbr{m`pv)gi1e$y
zcdi=uw`lhT;&)vN*$I!3vE?NBm0$@9!+uK;zn_yAV~_PqV=rdnc4f?;i{kqV6aVCm
z%dHY(e`0^B^7AMucvcr-j0|bJN`|)gxh0k
z{!Y#z9rxfPzi+Yik)2p?!sGU6AmH+`eDYqZe!9H~?srR#0_oR3{17wN6R0Q?!aUj0
z)`9TQl*A$OyAREEl64|qUC@okSNS)43-;p*H#}2<0i~AR^2h6pVK&VJ|K2Co-q|z&
zcx8#;Ax2oA3|l(`LP4&}U)f*|*xB&D9QWTgH+#GT(1)F=8T&swdHxQZkEl<*sbvCK
zqVoMd_Ww+!OdZ{T>Rf6Uxk2x8kZ&>e=h}QS$@n?g^3wkYUOy7eOCzyAU*P+FhY$K)
z_VwFI{5#p;sK+k<9z3(#U350nm#zH)@caZmTHRx^q94X-lc5=7e
zrALDhSH$SP9)AgX6ZJuZXF<<**YbE&1LxB-lt;87e)^K9WKDziB)*?g!u$9|UAqn*
zf_~#3SBo2go>sNlA9IM8lElt%DM7v7jUgdnz)7(uS3ZJXa<;PaTR3ktAMh_S0rpK6
zf8E6P9{i=Pv&E}3;4f>NdlqcKU*6+zfAE>JA6<)h{7L`tDDqz|6|X5vLo3s6Fb%_b
zZJ7gHr|+MKadQlOpB!)f`rTlBZjlXY!oT~_Ry}Dcfr?RNSx~H21?O%w#m|RJIJKi1
zPECG=j83#YKxUm5_{(eZhRk;0FGDU}RZmN9M+FHlXH0C`QO+0MAT|F^r1|(02lp6HB&tkv7mY}nzW51l#@x4Zm(0WDas`>$VIGz;qjHN
zgL5wvh5XGNaDJ%m_iEdBpx2n(x8fn3hx(!uD+B)7sj=&f`k
zGtn;eLp=4V1o&&#Ket9pz)#*?E8VaX=4EUA^To@Sz&uwSu)PHSQeADh+7Et5XXl7X
zZSa@3zr=h!4)J_TOjTSE8x0jnma$EN-pav!EO_6NTV>L5ALvbGRxBBWb1h*9y5kPO
zKF*wjThRyLk>z3ERIfsNIY+#iUxVIY;eI0}C79ogov)?fciZ=zM`0G$Nvc^TvpcLC
zH*V9rKHyLH*tJUVf?nj~4Z2HUJ?2a@s}BM{SkEd`$^ksWvMZ*)6XN;aikbm)@Rzp1
zq08^VdLHrI&$|!$YhKF0f&FDf5t|7eybI`Ym
zhkEaqoEp)90w(tI*zZSeYqi}Cdi~WmH(9|rwOYN^Kf`?4`fQxzg7Sr&pc))6cnY80
zjq{bJ=7mA{y@*Zk-&jk6Z1VF2@s}+(*uG)=0k2PybDLo%&fzQ|Ul%!7rH<`Y|N3^nlgeU@~j8|#5=?x~xF&s&CUW0}PD
zE)ypgu>HMkl*f<9D^le~{H5i)$7$Fv)ZZ*VkI{MOlrgrio(KHNcdfeAMT9Scob2C5
z&c`e==`q22K3CVB!Fu%^34yqM=>qNJSnq|(JL2DHg)SuSRA-L0$9lGr9=W*x!?i8z
zv7SOH^*x1yTNKE71RhZl`1AzHeph`(-Bt0U!d&D8sz*B-o~Z^c3Ci$N=d>GU
z7>L~dBZ}y~u;U}`!Dp-pW(vs);Bhioc#t^yh{wZ0EPEalOvLjg7Rh%4$Gd7-GPiMg
zRE6txj00{T74SS=UiIrB_P;^hWtw>1upVDY`njOGl;kJ8PW2N$Vq%q_jjvn8#C9J0
z`%j!_NPgk!7wVk+q7_HTI1`f%Qn6lV+8%Q5Ct>o<)mbYb4A**lV@1-OXu-83#=vLyLA*4Hj_xJr0S(WjL-U-5x8OAKRY
z5cfKDc#h-}mFGsl9z1;Xag`RtJKDu@*_z;|1*FZn9>O`wxT`Be@O>@56YQJ~dVJ@q
zHeUeyD7*J((r!2>Vdx=p<9ibRADyl2he3}$-#MZL@KuPCDULt3nteOk4f@|TcGxaJ
z99EyqXDS8t`5jf7zJTA_l#Q?*6AEwYJOTQ=i&(k+!A_l(@4JrUh_Kbc#*%=e=l-m~
zeCY(w(LZ}(9Ba3-NoxZyILs@^uyqRTP=fB_L^yZ3w>7Jz68u#__RUo%l=_iCZIb?t
zpQEU-{M7iHDzLZp<7O@^Xo%_bYYXWpI47{iVsmW?4XHOZ#4YrXpk0p3oG*agzx_z=
z*?2YZDv$O(PvN}7E6s;oMxJoq$K}*i(_-Lh;VagT!+lw4@i8(3ipVm~jsg!U_!)mp
zix18{k1+1AafAHl_tVPj;Jz#(!_ogd0-g1HjjEc-5N$IW`hPZ3xjCaq2tLaEknqQ+49b-r6Z`xcm3zy%P`(P
zjbIiif5Rnj(+jA-s_^GNp&|4m{79zKbsBmjw>;{W*8uu?Jz%W5cNh(1eN0`VJ%mQz
z>@<|I7)Bh^Kg#b(f!`2)|E^03elM=}>bUViWTCXYC|-65eNKH~J_Of0R^MbvgZJB}
zS@WbuD!LJqW^l|c>waW2aQf$Mk0G>PG5^Zv)_#wS6`vKJlY-TNZz_p$JgyB_2W)`W50Ifa&yMXYp@+{LSGc?V>{5#
z{WtF=!TD#au`~N_zi&rZ1J@ghPIVv?*5upoWxG&XPW9fB#tzgOY9?_A^mv~cM|#P1
zAakuK^(x6O)V--n%PSJjqnPF8ZPaZ??cyrcIa}J%-p%pMW!t-uGS6GdLiqhoZq%sI
zgFKH~uhhmj0`Tn$f3wJq_uG-2e!q(GKo_!{Gf&O~K4U8%Sf2qr=k=1*`KhPvsPswi
ziqcD+$Ub&QSfVA^vw+{OQjm9v-dXrW7}ov8(m%7|tX(KS`Nge+U%FAA?+b}s;ESqC
zy$NZ+m+f@qcoZSeb)lh|bsg+)DQW#sa4CZ4B}Uhi19;Jgj2coO<6{PTKmB=DHz#^uS!#DFi!dYdD7uFf@B
z1_d+%ud4F0u7T&);wOJfJn(bo@0ay?py>t-Y&=r}ags-3364*;Jr%f64bO?p=5Gq7@ce4=R&T|2$&oc@?NWGd=KAAzi9$cX
z;8k#5($=kWNw`Y{Z)@d74g1E
z#5KE@m4F{k&d;jB^$jcXW$%N$ZM4iF1@OT0dl_ti`rDSB;(>9Fwx-ZHK@Qk$`x@IV
zHcgijHNdaOG;U%))8FUpCIS67N9#)d2D$jnm((MGu4h~vu-%O~*N~6>f$<%~2l8nXl^;)3@vZa;Tjh3CCAFw7X&PYxEyVL$QevOYVOeadPkaJ;y(
zwO|PM!@MMM2-g>V*lmzAF?rECVbF(f%>lEwA(1*9XAsw*(RU*9}r=|
zMTDm)Udv_vqp}sCM@qqP~aBrDs6$GQB%C5k91{{&vpVgUlOi2NTH~4SRhj
z`#JhQsqc*xb@`L=1;tYTZ<1Pfjp)bCQ2!UAF-GyjE16zIug3lg;b9xOR*?0=t^D#a
z9;a?yG+7tTr3C_bJv0mFHDP`j6?>YzSEaGCAvF@@dHp6XtiN@N=QLhtk2u$n|GoIR
zH#QE-Uz|!_+yWdqqw)~zwF#VlfHB*%f$*bMr4^)_~)L%AeI_-Q4+cimWHJ
zGli_fPD>Gzf2tGLTa4SiD!YlyTiYTV!Ve#A)gkSlHb5_0-{507qH!Q
zZ(-Pr&%6GB15l|MGmrcIGdz&t9Tu?|Y{m-n+4#Fvj0;3NrK1yask~S$@0j
zYOr5^5+6_(*!RCXj!c-qIRa_vcf9z!!SB!h3uS@&jJOR)Zi9Z}&0nI)(9Z_tkI%xu
zzp-p|uWN_;hM3MWF3?YyFg$k#WW{&l=h?uIR5}I!#^2jLb6nDj%!`8<-%S`V!?4Y3
zJnlnZ^!H&qYc&z^X$k1FmV{i85C#77>e!v0>29A_qUrL*Xz69W7Yj|v={&4vAhs!$V&=RL3wX51Wo4EVL>+IpX4IA@At%5Ap5t^lqcg
zgGi};jg^_&AhO-YS*1KafGQ&NvkbWgkkh>7K8p(jDCvpA)bZy-XkPYY1Ivj)^vh9P
zM^k42X?BD*@j+fH)KNC($N7E~9o;VLF3^vZeq`=v7#l#r6P!n9ga*-%5KDu?yneJ=
zZA(E_Pd`$-!j=BY5ctCTi#A91L0+tSjxp?0KMFN^{#~xKA6*ahS<=JPkN#O^p1AD`
z_Wks%+D1mm3x5$t^4-1Yp5v$F-=qDAr)2GD{;fVFXb?GasJ0V*^y14aita?+r*2H$
z-qMMT^=Lmo0zcz=vY|P*umkOQV=bf1*9GsBReqZJ)``}84@dm)>qOhOW>?ROcB0Y!
zGSidaI?!H|!RzNDx{yb5Fz=_sUFdOhy?$#$CwhJ|>}9GUe1{>WLgcGaC;act!L#Pw
zU1(!AYrbuI7fQ^~Obe0jLML~`>OIZsLch|(oF51<(qgH!5DINVWzCW`lt>{Vu6=jivZv*DX6EUS?;pQ^
z?)x<}XJ*dKooD7f=b7is`+X1l*R`PMu>rl@LTyOz(URY%FSnzBaBqW5_+Bo&e3RQb
z(-xGx;`_T2saB+J_I9k*s~x#dsHA)j1v`>bkcpxt#GiR@X^_^6E?z93({R5XofJ(_
zO^0!r8&}R51s;APe1x?@uN?`zv^5Wr0K4g2x2SX9;d`Wynn$H!ysfH>J$nZ1Y@4>L
z&pEIk-j5o;31q{#s+_waf)9A&j?k~dPhfo2cy{Gf03U4mFF$OxQ#8eH{&>z`iE)VehgF@cioB`IPhp(t}IS
z3T}Y?Q^{xE^-Ex%_3oV5(+jY#3_AUJ=@86!PT|p#+aY||K;s4*@`)LqkE?`y?qx0$
znsAOSC+3M_8^j9^cMoBnbyoA6CJW}v75?T`E8+RENo5_x^XSWFgME=8_jGA^_8Uk)
zTKHRVCCGQk{`hGH?J8^!ikk^`v2jzUK{<%$l3!M{7Ulz>r%pdSra{N<#t}dOe>=zEpVhx~7N81pd*RLeqJAv)*^}BM0@%sch3s$@!;N6K?hcLhEK`Y7kCSmW7
z@5S?d>|=%1I3Me7sMvWx<*~3t{QiPxbipm$-W^i4nmB!~;Qh}SUCd+)t^jH#8BD|L
zs`6sfLM(TMx5FH-+h$QO!9_0EQy=PYv;v%N?;nHtq|EMVHaOkm#nZ3FSY8XT-?=CFOH{nM3S9Kb-vyv7STv6ZTkc{MvXmM$f_L
zB;BLHiR9B>_#fd_LxZRvr;BBuPsDhBHRU@pJM0N>n6@v-2-h1gUqS41T))L>%rn
zkyTF#-yB&={cSC&rsu?Khl+^Y@cJ0S`>sBq-`UOkQbpt^W>LFmNe%VmVEvS0!hEWy
znkdD}D^wr35_-=3Llf0UkGX&8@fBL`wk6d5sIZvI3EBQ8^+eW$lJU^N=_LRE2Z57h
zyjlPJ?=sfEzJybNpBJfj`XpXUP>Jx*-2p$zc(aJ;yN>1NZk1KS==7EP^^X5lLE87e
z=QkJ|?@GxR&flVVqh9^=E-ME+scLsEX)j6X%VJf%HNB>lEwFu}E(
zLWo>%mot&m%TIcW<%XOu2jX!*_e}9V9G>?xP!Z$$ss8QQJ}*4;Pn;H@gWb{{IDX*O
z3-W%`!a2JR&l@He^26}CE2{>}*_bzf@RhrZ<%WBoL}8SbdNPhNa+?nE6Sjv>@&SHs
zXE<*sdj2sSA^mgA&zZC@7=0r9g3?KQjImsy$bRy^+*LQL0+&;%HAwn*{q9_H|C)I9
zO))&Lyif0ZiRY`zA^&D<2YA-;*Si{U??!p~VBoiWUuquV@7PigS!G;>c>Q~;4r&6n
zdEZb>1?C#g!3sTXAa(l_n-&L
z#*;B`pY!$JihID5#ga>{!oEPg@<&280>A#7KBOB1_RKl@hi_DXp5k9?ui*DbGjttf
zg<*XPSNY}d1naGam^hyttm6;0_peKW_RkNDormwiSTX&3aSVjl-<&2R59_SY>*-@b
zfcFg2#k@ez)VqbL>j2MORXL39DAD__=Hqo>rsH!Hd$3R4;@x;TrWe+~$2oA-bUQjf
z@i(|_8~9Do`*uCDqZdVFNWDtX>_Id9zZ|ZG^GV0WHCp3=CppSxt+IR$^SRm0*+Fm)
z{cP;#Gg!-JHdn!BuJGSF|_TW9x`{>;R6=~4E
zu5~@4s=&*S^}DlI0H4^^H*Mt+;8{mb){kF;eNc9RZqYfATR-c=fqLLQ7oN!&c7k0@
zS*gHjJB?lFpkcFi!oE
z|N9(>|1gqoSPb(F-vVa=@S{-w5QU0P0Pkur+0+a6>T^n2X^Wsgv;}7jxWNDD5zunc
z8tQ+TG{67m>P~b%NRWTdJ{Zr{uU@BKYDXmsPhJ#EZAVR-qQ%=^b)p3`C-k$8;l1}u
zqnHgr&|jP#hO=ARk>&rUYsTLF6ZzAo^!{R8{AnIF=$z`p2Lk<$9}3Gk#6
zTXp3a*hg>r?B)V`x_H^BbVGVz!kC~JjH6#AHg2swzz+JjH+cD77;hB+N
zKJ<^S(cr{WLrzdL9xDt3VV_jEfS`}eSKFiq%kQ-t>!
zzTw~Zr^9nzIOY@L2G4In-VEbI@ch3iX_M#yEU5VN?jqz{>{L171oN16(2v^_pnv!x
z>kC?-S6F}i>NGeX4p%#{JvHIEs-hVnD#$&J_0vnd
zSbhu8drE99wm%K7o*|CyDfPnoikR2)-0sK^^`$UF(}XITxnp;uVe8
z=UZzx4`Mz0#R`Z&x_6BW!|{4^lB?<~PS?*1q3>z0-z4wVqb6-}z_|X%Y
z2c(=wygx{OvvF!)1zz!A!;v^a$mJyfY~;H+x9$2lA7r=LGT@KNsR2)L+}?DRkcu
zX-Vb(-KUbs+>7#yxi947AL`%t7Q_F8n0-UTA)2q0@q5f>?mfK6+=mj%+{bd*k;+vu
z@?nhM=m91j4hfQWfiaxmzkAce3}|>NlitGAZ^0mZ>`3H!8Nbs$M`*vU-AntyFQ4uk
zxFo4v{>1Ksq~FprXg>ZNy3e_lLjTuKay0*xSlZty%)N}iji~$_W`A^MF|*&$r~8cA
zt7*DV2*rFYio(o&GiFO@fAc(}@rlg-?sFy$zYC>$42seB2(A|sNxNF-zo}k(nEmAeroUG3^^^J&ZKBJ*lW2bhUZ#Fe-Zk40edAxB
zCC|5NMhKZl_PpK@i1jvMP0q)7Dw_W*=FPt1o5*}(wBRP)M}=k*d%)L=mgIYk><-lj
zSl;Xj^@Gy4eGkz)(u_?IwN57TM}Jvf!*Va~(DO6SzXFLK+`XejZ-wI_1RsaG*I+rX
zzH=mfc6L08FP7>h{l;as?FCLB^f*rH9XaeX4bSfz>-G|V^?PHle#Cm~UT5#X@9zwD
z?z372n6@vG%oj3I{bb)2Jz0MWp65(eHo}#;z+2>9r7i*Pn3{VUmmjAtN&4aJ!!|O{
zRhl#^VEM>^spPv_S^mxKSgycAp1dcySg@`dhXdYbuEO*2wMW$^SUz54!7Uu0A%2d?
z?+VE$?*#+RgT2VSEj+0o@1rj6xp@B`psj{>_fJ?~Cpns#+=sBU$MEGKSkGrB==@y=
z>%3#D-q)KD@7tati0xawdK!VjfR?OPhw-}1aXZqU2=bnjw-7(3MuN8O_*{sLPlD+!(ATu6&Jx=tcP9?K-3vTm)cfO}*}$(HBM)t5gM8B6(RTd5
z>`nVUjCra}RDm(}57?Nb?#Y4pH2rNuPOu-_o~I~+*VCFab_qO?E~q_Spc>@WUq3wV
z3-N**)2i`(Fk{gTZCGE`kDVIZ0T;@58ZQI+?kB4XUxPf?nMQg1eT~)M^^W&{1K(0i
zyX6WzdsAMk#z79Wqcd4-2=+_k)6_bX;s3}KYy7hV%K6l7^K%#QwMrw`7Bkof=!Q-T
zZw0xb9bIytfPWpDsrO<8^z?|T*ee6LIa=r|zAt2@Xw8>7P~VtE#k(tjqw{2){(?S^
ze$IZ_AEBj#FBjfV|4?3U-WD
z1YW9V?N~h(?B8{&a_D@{vXUjm-ocJ{Xr)E6p}
z`C|j%0l9Z&$>+>q-co-;a5*dvxiGS@2%r+R4la%u;!RW1b)uB
zcxBGudC>LC)<7>nQHd!hmIJEitx`4tv{Ev)#s6Pv{^bjOrB`P1>d6S>IS=pmrkk;fUffFfUdUGaRV6f@)evHbn?Rb*a&
z8CFH?-mPo0<#GEf%VyZ%^omFk;vZy+yaF57uls%-xd$bCZr~i;o^93s%7+2(FAtf9
z=jF2F#&88Xw6nqKEY7br==d6^pH=rU!{Hsqm1LiwJVWj{F0UshnAjD$j<=KVD%@K|
z$hpMMx@q@t`U0&=vYrL{sg>ifVvP0x#_BD7CotyasFC&^yY!0iYt|fdVkeZoYF>rw
z+ZK~X>~u2@)xX2?tS_scVZ3uGkF1NXb^MB0u6|=0;n6qV9U=C8p8joQop1Rtjg))U
zu!59R6EL6XkvPqftnW#Ssb6KgglR+%`SKLP+e6Kkk#uuU7Fmbg%|=N(ye+xe!7!jf*nx4rA;k@TbQ1&Mr&$ZCT8DLV^s
zyVlp7AnUcGLma7xe+Kn8J-3_km|Tf;k{+>_+NWcZC|~0LkBgME_hJ$$KT{`|P~twKI*RpZ}afaGodSbyFGt)U#PsPT*-ANiWl-ye+7MIk%ER`Ci|0TE2}m
zJ?C=n4NdQjr@VA!1=Wk^0>uZPXgMK&X#NQ1enB-unm)nscHR+MfAM=7Z+e|#lyrWlg5`aVR#6L+%E>Rn0FL3jy7a2bQq48N~p_G(brNb~7wP>jx}
z_-ZEAqu>_Z*Pd}UA$s~JCXwe89j4z0Bw6PZJHgQ3m>@h4n6%P+ZgLa4iM~ggUJx|P
zTSw+SL)$pAFNv5;%k|}@&+E>sbbqO;LZ8#fteZr>{W9%02^LN7jG)g|QX6fj_pKtL
z&){B~Uc#d3#a?tjH>F94=>5Ql>YrI$cLVEr()$K%fni-&%{__n&b61FB>vvwEW96C
zfA1C9x6~BAA>}81mLT(Ca+bNe6bVup6*wF*Uyke(XK|f`
zsRj5#e_1fe*XJ&Q>)TisOWr>ncqUwh;};b7kaeyl(=Qq8;qhNe9Nv#8tnNCn8!&U7
zxb9RyF72hw`2HH+lXk8{z?0e~a?UOX^vPV_f&Z6Ll^wH<0Ut^xko(!JUr4v${m#dY
z(E}oYn$M1tec{%<rIOv;O~3Vda7<>eP>7s5_{hY_jX^*D{f2|=EMEX&Ce5z&navOaFDSD
zJ$c4A73f3wcYVvYFlfiqx&hyKz|bdee60bkOZQb`ek&{Ac2WTFywWEHWk7L`&2`N0
z9+bzfwgkN0S=NGiS@*`VaLhw3-Bm*{@0<3z>;&EqB}*mn;rB>&DPt|zKDj1$P#o*S
z4;RE;0JK4|8&6r1)u06(&^~Lw6x}#
z=M8#3(RsHV^Ki3%nu88zYR%k)9cT6rZoNdOTov3-Fr`
zn|u!@$vy0W$LBP$*eAf9FEjs
zeD1$4$N(D+>P~
zYsylH-@!Qpmn?8S-vtc2u)gtuhUC7j%7(S%dv3+2vqZ3b_5SoGtY?n?$K`n3h)ms2
z_}hh=BcwcmLVB)N_jAV|EZ4Aw?-SO8b+d6h#-yFx8*n>r$V}42sNC)gCQR5Tdg|$6
z`L3|Nw{bl^nO$T)b*^)2!|9#_kIis>Nse)*IN!GYH#;yM?R(;ZG2;}q52NYaC$bG>9#r@>NceA+!A@egAE^}~
z^@>hX0g46t=^^eCj1tS1koqFlY|AP
zdJ(;wX1yYMaIQy_^(KbrG%07wlHWwG+sTNen>M)6dW@*Q8{bM=PUMVC5+A1VgXBLg
zM0vR2OltS{OQGxZQ7gKR2RPC5JWuved*i@SYHv3kqIPCD3_!|TP)hlS>Em1y7O|oE
z_r%e3m!*_HWY^Jsghk>jBKKC0+NU2yQ9fPbOL=pVBi&!zbE15E=mX_fV%1cxX&LQz
zFD05E9i-`kZ>b(qIkeuS^Ay#&Xuh2?G+lwQKd0=Z>2Agp&t*_OUyRc5FXrBPudOt^
zjNzv`$y6_NmFl5pMbnQn{7x>GhOaXAb(MG;u4Qm2v4-X0d)A#>FAU8w~ehB{4?_YiY!vFLV
zWb~ZEv=5v+l6<8nC|xxt7uFr1
zasziL`Xy8OUZ&gw48CX3i)m*&vwr|%1kndbGC>PQZ#V=*!Vi=v0tq8wS0-JZDMydd
zyMuWT=BiAewz<)K0QKi0-2Vn$w8F|iYr@fzBP5MPyw~e&3mF4^p%iBGWC+m!f
zpD8KV`&%nXf84#Ftj{G=RuO&TR(X)}?6}_%`AR!UBG0!aC?1!Sw_lv>cTk_+C#>(*
z^RZL0zR|sdyqG^labn2+M9)7)38!y)JrIra)g?S4=lb41=skqPXB&6~ae89hZe5&z
zn|UE^pVBiIoNi+wJBG2r>iRv59W%;~VYKi$MAqqN{YR>B_;7&sMT}PdO@wFtR!k=4
zbch!dUQl5ElAI546I$7W<$jvP5dJay<|z4YXH;bLB#u|@DyqT!-Xcr&D7Md<4J$}t
zyY`A2v%Q#)rkrZ4!2GN-v?u}Voi>>_0rN52u_#rH_3zZ%Fe<-z083EyPtshk#9)@@h>N!CoF*Dub7Se#Wst=(!+PquzajUWb{h4FV4lk1Dyhf3d;PPuA2H_6
zv}nb8?mTfn8P|8)m3t2^&&hU;0{$Ol$K#R&xh({97!ym
zAmhr?T9%BneesG>_3P=7*l*!g?2q((|oXTHVOFTPk&*JTEI>og(j%zbf25kIT#Gy!{wsqXOkqbuY8`
zko^ZpZPgJMzS@&|}?IZaALnbaN;vFI#InAb^gePB@)h^?8qhI{X!?nb8sZal{_DI+mr}TRykro`dez@
zKJwfvAL1kJ$a^eK%1wQHjp(mm)<&NHAB%)ZdgQc3^4w`HV&&s{Ms7ur{GU$M5R^}l
zC;I~K0IzbKzUJyi68^cyp3EP-8T9=6&4VFCzb{KClX+M5zeOay-~qKW@wn%Z_%%ZA
z1be*|NxIBMFOo0sNC2~_lC?AwD`$OcFw-*y+C;cG%
z3&*o!M6T4I^81``^qh>^*c&2e$W8YH3SSPA@ED7(58Bh|`Z>j_ki;Vqs^={iI`10o
zqVs;RGu1CmfQH3`>G}L)%5*)OctzJ^xzm)FPu!yMov&Y$azmvl?;mlc>BFrweIBCg
zOVfJ#{vlE$o5-);LiO`er0p)eL@`M{pQO7yr|aqOI=Y@;)S>09UqZuk`RF>Fe}IO=
zx@r31beiuPa}FnUB`sGeil!$pd?AdnqYE(S^F8m=bjM+uzl-6?51Dj}OEiA(K`Qqv
zl-65mK;vH=q8J=W(enaLpMH;qP4-eub)yJIGt$lheToVU?{8svJ6sY<(w!On%J8%+
z3~%Q!yh@DWQSA(B`_p_$Ouje<KQrMti3
z9u^;hw;7CNa0`llo5W>qWWEIb&}sdyvDT16eQm!Jlg(!D>OzvIzME&swICV)P@W%~+t7H>+<5y;IOiMKC}H~={QgK4Jbrrz
z-kYaNUTNCUivAnCS+|y}4b^0?_>wZ(hJL2EwVV(IKCoM*NNfk#Wl#LE-*4554$OI3
zGuGdV42Kiy<~6jTUQ4G`6Di;)&1UjC+F)P&I%|?J`0WbyP53hxzGJNM5S_e`w*}p4
zmFL-I)r>A4I-_~2s0k%c|E?8m3-?9Gty&z`-Gba|mu&ugqy?oN{d{xmRx?U-%zW5B
z7yM`Jinxj5_N9T@S(!~}1E2In{<>t&IugCQJ$%Ffqn$UR96@4-AlRlBt{Qi1)BZ>?2{{8EDBg)<+GBzuSbey`Op+%Qhhk{m#itby>(mH%wzA_+ek(J?r4Y
zwnilDJ-bC1{G`r)Az-aw2KW3O|8Kr%aU+^OJj>WEt`W7pjV&2_*@((lIK4J{0RG)K
z3b2-VHKK&e{armSkU#D1&4`3XwC||TTaT(nBpvc)GzI*vckR|S7ZPG2mx?`3o(ow>
zC2@OZ=o%JE{drZ$WE~6Tx;kB|>u*F~gnpkDTmkJmr(!funuQ9-6Pql@8j-r>%!r=R
zM)aRzXl-w8BQn0QXSwa)MpSDPcY6`|eJ=>e88L(Z=V|!w<5QuZl-8RXd`W_Q{TCJ*
z7B-?n7m+(Z1{;ydgH5vQlvzl+^Kz<&EDHq%WXq&MKV4Y#v}^@03vsK|zBgRYLc&9#
z61>Y<=!}Yr=1REVXV+~f`wWQhJAUs-iXsc8%cx%JmVoxKtSlOYS!n9g{f(!gJfZmV
z)mMibQLg$}%~J3ye`Ct7keBd3?Fn%5a}Z!5)n{_fTE;B&C-1`CGjc36;=jc1?M4jX`?tL6A%0m3Rdw$8Pga3Qk4NZHlv(WSRXFJcwvCszLmA3vbSm@{$b=h}o
zS!lc<6P>VUp+z@eW*g0JLdTbGc|Cl(2|4lV2|T~wgg7M}tyvb(pGRjzxt4(*bL+Ny
zt2ZIRUuW!-_BA0pqYsm#&NLxs*LBg2ZqS~OQfVLIUQ4IT%benOHzD3sffQ|T@T02}
z7+w?y_hzcqra6F~PqiH-O4l^O{bXNbOe5jB7LYCBTiA>i*>3!)s@;s%@PC+)O@Myi
zJIBOG490bo`a-M3CiL~l)Bv-k;Advb{ip86&G3EHdY`Lx%fY}6Mnmz
zkqT#5SABdFYL}EA|5n`H^V&u
zz1J?w?QTXa8M)%E?#)Pat6pq(YBMra=b2sdy%{-+vQ~*MY(ZQaN+MffUNRRqmFfs<
zM*Lq|lM@GCyCL
z4fMp7b`HL2Gp=9NS^iNQ}o0vkR7;X+hU+zuPSNv;}#OaeuhV2jz-yUz^$q|C>>GdX`WNVyR!wE;4UH
zH)7ANRdj1Xbs>foJfN4a)y#WV&|m&>zdntCzGY+nzK!rdxN%NzO44pYGafyea`#mW
z+Wci**2FNZe}jeFi?v&k%{RGAQMcf`e|4$YX7Dqavs16^aA^y=5-2><#c4rn^77j&
z*SDgV23Mx;gL}sol{n4W2L7efGb2~df$#Aze9gYHyRrq{U(^+McyTLwp-_GM!SPn~
z&gRpB-21I)y^~G#Wzchl;;V+#&=!=~KCUI#*n(c&>7Tk~O)K(xo|aN(*NRTKevz%#
zZ$V9QV;Rl|ThOEEYt=%a@AlS=
z43eEQmczZh(=7A!!del3*d57P^`N)NqoPuSHuOTZaYI){8*=3dd~|bRJK|kqfB!q^
z`6RBXL~3-Cg+U{o;dU%t?)Bvx19vPZTo)~
zlzr_)cTCl@Pd)5He0ldO%uaWs=;4#k-u88)fME-zQP6W@{;Ws%zP;zJGVXzYJJIBG
z-gon>y3i}XAgvOOZj`%1RmcSNOt)O36anWVFFh*Xp&ST)8;Yl{C|B-6Lmn2{$#5Uw
zW~=Q%3#Wj;nvj}%M%%%T_N|~;5x(1OH3)udejfaX_r12wp4^2B!prqp20h64{u4)e
z;L&PQRz}V4;J-voRiGC9$0`op9F=1Cpq;gIqElPC;as}vr0AdB=+>R*%2LWbD5Jsh
zn998#^yE{KiGLI1n{;hHN3$DUpRw}b*#2(Bt#J6!&Ae`;vGmup%&k3WubXkYolg(?
zIxV~F59le~blk8U+81f%TC2RH8&y^=EQx&CjpDW?xt@gjw05RVy85#dZCqTH9BA8x
z+)AZ;9}RaQ@o&2nJgvLY-V4TFc8T5S(FWTalE1;<$Q+*yR@%KNEH~?VtYhXsw!F_6=;}cuxd9#$YkHA8Yg_jszg{%0H>-y~q7RkU>-1Nw>_^^R
zUtEQ6^`H}zFVsa=^nm~OYwq4>dr>wD)_o<}hpK%xam!uoL+O)m$;r2MBgMx*hvdL7
zOOcEGQXaluq;y!NS@&5l+A%!AI`Ov`C2d}?t39k6m1cUh-3NarynjwzGGz6jahGe?
zBSBBwRj;0`f7*-KHx3CFz3)fyT<0d*J^GM9eb6J1UwtThtM&UF=YFJmK5E`iG4Qh&
zz5V-|_CA!iNlnqltq+9>Ic&L7(}z~IOk<5_#bn_`TT6cNDmi&4B$nLH5MHj(-Bt4sDp?kC+O+TLD
z83pABYYarbbLvO^+RgbhjrtL8sX{bYm8nS3`T>%pLBSy0#ycOp#Ha6vIa4J2h9Aeq^Hv
z&o40{fOl<9gU16l%G>66!s0O-J>ABVtFT}rQNyoVswdcJ@*3@TD)wyTmH8`fUN{>y
z%hdfum29+N&rj!vr}|Mu+ZEGjsCTyfbi+xoA9m2Z`1_Y4j%XFR|S$KuY~=P{%EJKk=F}c9+6Dd>Rn^jt@%Q#4
zk1Zqjo{NB9wH>cF9AKmL(Q`#r$!s)!*{tE(E7*5yzL*&7W1~Y)sw9j;*hr})_2PmD
z&`$>y_s_3^cGTvpZu`SVa31c$kJAHawesZWzx?ZFS1c+Rm!F3#cbq1@0>zHFdLnFd}!TZ1RJpp7VnrWIe@~-
z>u0~Y4ELHI)z#3vKY)ghtv9*3jg6jbeDT=`{h2q%yY#2Z0GgXsb8xfQ0Lp8hq`u|O
z0BU#lC|V2uv*}S+o0&BO=<9Qvnjg0~s5d29QhNgjEgDmEVY_lr=A-!?S=w-~{iV2#
zzFRmbpLf@-^tl{##@59^b8rBabo`y(sLVlyVtff7%Gu~tYUsJ466pUoPj6;e51^E@
zJa;Zm<{-yQMbp$4a?p}p^FND+51@kdEf%M92GHW(Jqt4`2GEz4KaPB251^1kLH^75
zIN*2Z{%O9H0aPzu9)I!70J@{`&Dy7M0PP7`^0jd`2RXev^mf6-0PxD?Clo^iSXNU@1-wXY-e^S;M=(jW2!b*E92i=(DqMQfgaAs|gNp3jEy;(CR6Ujlle++K<
z4COwW$@NXz=e(vS}LUFUip5i7t!2yT+}h5JonZqIlSy5>8wUZG(W
zs{S4I*PduMhk2n|<`Ta?5A=WGt3|&hz9a9l#$COR9CZ8E7w@}AIB3UEXr8<1cf>v?
zSNB2iJDig;vaov0L30Mi)fXP(pw}5?-a-y=AKZuFu$?gOPPRF!=1X&sYTKoP%kY1%
zX8*VJ^j#RQp9(H@K-eq(t6h=}2mLlq2{nWJxiI1m2U#^Ma?4rtGG
zF`Ib?(%(_?%+P6~EpR`a!1Y`A&v1}uVD#fWxYtG2u;2fTBK$wYdEE`r54MdCr}PiftUF`=wu3D2<7%am*Is)#_TKBe*50RUvxD8J7R@^}
zm!tnwRppfAH2FVe|MOS;Z>^ys7s+^8`CRcI@*<58X|jUw4|%zEa#Yq%lxvFiy9nmw
zyhXjq6Gx6D6`6e*6|4PU@_+aYIE)-No~B@$I1j2K%@)^>$nv#D|K`dv;r|CJTd(_6
z1S2xL8gk}bpeUyzry-{*C-3LsvnW7L6a5mpX8T`V`JZ~IA!jSrYYs)!^IW>jEzoDN
zUw~X|O*y%aD3oob5sMcs@%Ib}@R_&JbL6sxp8g|cs+M+WQ`Ww!(EgIe3s-u0di}>9
zDDfuw|NN_7GqK-umb>}?=Z^ndwSQ|*T~^O?q30q`zrg=o_m|kDhTI_jFGV?Vyo)>o
zyypb2T;j>ctBpe0c&GRU1o|&?4_xN&>EQ$|j8g93+Wq(X^L-Wy^_%@$eX?xk-Jdsy
zda`zv<#cGC1AP{G%4v0B|MhwLsmt~Ii&IF}{IFr_iu2nxFj;-s{3;6dJp%uiwO5Ai
z$;xm%$o@6IYQj8Gy8(foOa9}BF%}x*gJ>r7MC=ZooAjT&yda1dSUDM06wlTFG$N{}
zOk)OBP*C6oR7HD=^5O)@y}L_#vj4P2F)3GQCwgkg34K%KmS}yuVg6`|(!nkCNlwn+
zd$G)boWp
wm{TWc$wisbJ0aA4X|4$a+CVGW~at1w~td{MY}=%l-c~zwcfOjsAD@
z+l8zBKj)W@3)#q)Tm0`k$%J10^M3_DD`CI*KP2pjNPmj-yGZ|tRH!c%cgepfFa9fG
zzyBLOnce@+j;tRtD(l}r>&eQ$iS5ezD{DtqPex_(f7+4pO`;u{U77r!RDds1_Vpj*
zAkrbc!xDCze_#oFF6uoJc}?C?DbwT)N?DayZ-A&bMdXFk
zNXiQM1WMT((Qb;Uca1MJ*cbK7pU^)fY?7$IUDSWWCrZjxMZPn3ikzI3#gqy9{rNzp
zEW1+R{dI-%7GczYge6=O^?vp%oxu`7MNcBSpLJqMgg8f}YZ8>VSlO
z2p9O=-h%$(jRL>VQQ#BwsUay_7b5T@_=HH?vST?x!TQub3dT6F-e6Qo`g54siOzU66NP0~YcP@6Lr=9fh{6o!$-Jd
zg<;*E8~TX&J|i#mB!2eNO&F+@ZERYM4Qu*PTx=lorLOLt_Lb;`Xo0f{;XX^DnPI7&d(3N$Ne=O9+0+X_~Tk5(IoiTbl
zANM2EG#p71wmGzve?Pfa#e0aKVV87-=;0n6`TT`wC1PPp*qng{bwo4fJMi&qzHs6F
zt$CZ>hU^~r{-7n%=-$zsYHcV;Aby8`GsLdw?+BG`R6qSe@EW2Gx9=Pv`f0&xyGEj6
z*ZsbM-aD6={~Of!#5TzvP(68_4@9>}tFYiD?B}!={QFPT_u}7GJ4)Dpiyj@~{QM#7
zrKA^~9nbqcW9LII_X}#^>nFH(TRvWSm3Uw!DP!aHu_17NkIv@nPxHw*zTTE~8^!0T
zLqhm2vS;0LAzxoVYu2doarH^z^KtCMVcyQQ^pU5ipYL8hdPlV6>`-iQtb<$ExZS8d
zNl9cU?8yz@ANTBcoc2+o6J5&m0x$D@A=yR;nv^AHoAdEr@G#)=FJVqxezYUHh4RYj
zg#*FQe=Na*fA5zId-3@j>-?FI^ZUzeuJ5|{
z1Gl&K_?S)9PqPzhFNxkb+6E`DlqqhB<93q5I`e)#eJquqk8)yfIb>+0iNx+S`FyCizEVAzvp_U9tWJ
z`dlcF@)XLI^aS3Ej#=o3i~ej&3oJM2Wzgq_e5B|{vpBGjvNRt--iF);^xhHyolaut
z3*nZO(N`e;JXR~wt)d_2WdlGTg(51%3L;)E@5+=s2-IwW6Fi3%P7OGMs59
zmUCDDJNz?1Plm^2_GMH{)aRcA`ZD```cY4b+%e=6M7<#*^%48S;R*D(7?dvy&?Zq&
zMsWz?KM5l*Pn1`O-Tybrm)ZZKTvktpBW3#B{J+_k@vp@7D67Y(ANFPalg*=ym(9Cu
zo@G=vK3TcU-ao0VTsHszESK^BY)>Ya^;^cv`YZd+viX#?BcpVx()_|HaQe^oW%KVZ
zmdn1yHQ7+rJ
zvVG~!4NF;n(e5cxUqR$wisz6&9iwP(i)gp2SZ*rvqxiri%v;=dz4^wHu&HACTRw4!
zBSiZLMg4K&d1frylkJb8Kxx#TQXhf7Hx_97yEG8sXOVvyB=8vzr~$+eZw3A@!CLTl
z34X%P*KPveZVxpqVcUlZ@{w7B`~=Vy>D>^|(a0V2#ei!%3FY^#g>qYhsi;3`tiV^b
z6Y%qtN`e2ik_IASxy@+cIHwb7~j0rqBr|WhVmqa+=cdEtj{6ea`u3#V#1I
zgq6M#@bs)WJA4QUv#|ZeaZKNnQ@LJ;RdrmipZXonFLRILdU1y~a6a;-GpEV@K+s;R
z7pc6R&McnukNo?fqf&Opn;ba&q3$d$ckx=y^?s=hJzrCKN{?Mwxl-0Xy9Fi^
z_(*36$0tjbFLE4L+F_s_m2Yh?;W#pXLm{ufYF!hyF8tNNR#-VwcHJX_>or>LgzJ#7
z<&&#%Gy!j~c=?KGL$-4m=`EZv4=zs1UUzikIJ@@pI^K^dfp!ig&$_&3FzJt18@8En
z==AEZr-(msi(w^8*{9#{)QG>D-O81Cv+l3jDS$4GjddgbVP=~`!hhqAj_E~s-OnZO
z70D;P99~H{ckTn@Sd!bg&wo$$o1M6DNCo_!Ip4dMgCCkQw&E%1O3S)?RIjlrVLst|
z|8+5Q2;V1HY~K14ymPRfOD*V*6PK;2z00lphLPQ;D@XJoyA#S%-+Tpq`g+1ok`KP6
z{DbVh`&BrL%CC%lYDMEHS=DS~bI5a^S}rB~pOoU;^YNvgJo_ExmNidQH6dSod3M=N
z@C}~bY|n#goSM4yBWQ4KW?(mzU&ws0;VSq`!D`1XK+pXgKF|d6&$_lv6o;%xbp=F1
z{^?~!8?x{E_YpQ2<}Jv@i~7GV?hh8GlvQ_g0`QiyBS9Z5iLYI#)SYOtOQ*p^qjxp`
zM*Un|Q96L?2VK93z(B&1o2H$id5MfSu_WGlaDsa#==6)h?ggNi3Jcy*yI(uEo2-O-
z)%PZKr*<0H-{s9fdmUS?)DhIes8eNI&>e>(wzUWCQsw%w6=*}kgg(^IWsi3LCVEuP
zA4i~+z0@?uhLx~IWhOjcdcL6V9O~!L!h>_D{M6P9+o*iVrNu_XPusb?J5gVc2M9Re
zPx|#aPUV~SBzlot^|z}Z@d2L{cN4vG?|y&k=j08E{YYPNPt!@
z`ynR9VHWj!#_X*+q+izChVTEiM}y%)vCp*D@P57xxzEp;g>|O9UrR>`=Vg&8#Hd%8
zq|N&|eCb=xH#;TJYjeqgU_UGs&i$}=*bdN_^TlU_d_f6@4H~M7W*|2<5qS5n(-9d+
zSXS<6-rrHRgSnoykzjZCLm__Cd;Fe{*SxX|FSosYpZB--&m){abZ7^sqrHWG9IzAo
zkca38ABs5n+DSBz_Uc;_(m|Z_0@cv3QSrj~;^qnbB^!ax+$Hq4jkiEwY!mFNd=dQE
zg1v&hiXw8@QZ`yyD0jMo;}Z7$cVZbpo@67`Pu2_La<2h`|EY5>;JnQo!4LkrDd^|F
z7Wy-Q+#2ftq3Hr$z=5`eG3<8Y*K=SaWt4TH_6&JIA!U83d*DA-3-tXkLI37lfv$NV
z$csh$5^|5Q_v--x840VP^bn}(T0tH@Ptcz#_T$$sx*(LYWm$rLXqBM%Lu_C1sxUwC
zqJ4iVhQ215fn7e$)PIrBBR3AY9K09lWsCA-BHch@$Zf=N@5QPi{R^TzPt==E-G=-F
zzs{1f`veKnK!u=`E1^P^o`xOcOY!LMuMfn{ON0}h2L3+lbeLkJAPpgdNGThFa
zg*-r%%i=@&>eL_ls-(}m4PI50!wLRRaxVTi`^DmSTp*5%cN2Ds#rm>w_Z0itOVqO#
z%L7IGaH9C{U-1(*8|OKJoKFuZUlyReSx{ab?a8Q2k5CrblZ{_iE~9)op`L7fvhmAs
zJ`OefCuOTeJF((8afoueGJV;+_7LSxyZ|^(tS`G4kd0@#D321?VSlmS1X12m}QfblF9cfzS
zItZjJS6T2AA<;s4y?D-kqqHFGwCqPKLCQ=%T5uZZO*=rs21N+^{l5wJUMvvoZ5aOq
z14g`bfIhU8Nu~3+GWrVhJywsVn5;_ndJ?M$u&OwKZJseEKQpmE|7dgGpG6A^zN7{rMo}
zcB7r{@$&rQ9$er4>sZb|TOyAR!{3hb#YzG0Tx-e8yA*!nn|VKLg+_B703St2cF|Zma@l6j}3@U+)kM>|Gu#ciP!HU{Ydy??|=l}UP^Tq$CGQT
z{vZOCu+|+f+EDq{&*Lu;4xWAQ&q%VnV@A+IvUBNZ&Un&WzT(Ar(i`0+>>c6Zed}w^
z5)Ri!fqehIr#hEG
z%jTFaoo=Rvq#$EiIl=dye-<
zC42?_xc-vHWYAZ`M_WyYUKeTeHBO)tTg+v&ZWH_D|Db(h?)ha5?W0|7a$
zb*!?(A&L5Vu|-2F@vGgRl@s;2eiRT(!X!6pKNB@r)yj+PSUmdqgY=uiHlL$?x*%x1
z59y!DL6{8wUex9z6vs+@rtfb7+TBkM7noAEMsr~l>3166h3{*ldG}kc0ROhV(ma}n
zUtKOmYJ={Ilzt=odP}ZEQ9RQBMOi|=Sukb+1Tbhr$yTOJRcJ|P?ev*(W)c4o%HJnS5dDb
zb2$!dDfs1Pii5LFQ<4ZB@n-QKnXwJ*12o)^K)~Z
z$u*Ksuctr)^LAW_TgFeuv{HTXUO7vX04#9x!6vRZrn3g;|0>|Pg11n($^|F_s`_O;0UW@bg
zg~GTJoblNp&*ceHDLd&V=zGuF!R2x52w3AhDubJV9fKDHjBua7RM0CMSjOe%<0&A(
zye%r{bbqigAA#9~z!5LL)#3eaICG5ayU(WtCG42!bN=F0LYPuk)`VpM`Dua|;9rS&
zsDUS-k=GF8lU>13gWhiJ9-@jO&N3BoT)PTEehk?kvXg&Cus3CwAa~;rNf1Yi_;0)m
zO_+pLO%U3@_)^eUek;tIvlw6c>IyjUbt>FD^q)Qw>|Y)&*mdYE)UOh8<9#xN_GNe;
zn~d|@Ngeg=M4V|Mj#o#N)7PYW;UX;%#~+E^O!ZDsvnW48-2>0VVNk0zg8T>21(lo7
z=YzaCeQxlzqQ0ZJ9$cISxtTa`*yLRQBrOx@%i<>vm%($`3ws=%g2&;*^)6C1$W4fa
ze2vIA6U)cp>yvyE5l~rta9ZRQXc?k@h1g$fnnv(9#QH_FtihM_Ko0j|g9JO%#rkv#
zklkAZKcROQXovWzrUKuG!h6Vtzz8@<>`&`wg51zb_|6exKQu%;vis0AM+Lp_bgaRy
z6CDGfI%0p4UJK&_h<@_KDFRUS7V{1|cc(l>JU}`8pe+AErBTJfoeQA0IewU)NWU
z>y6Nj<9d_xgmcYXT{yoE&1i`a2HY0>5d)I4qnitP{b9$2_VV6t;Pnc^z4^ImlUc^?
z^(}1=LpXQb+VS#|I%|F|{jnI06aa8y`2|$O{d0@!99MTfTF2{~?v}$uOW6I%ecljl
z_PFdFwR`IHSMG;j7!Jn)CS@UqTyv>BeS5hn@qKM(jUc+ZOAQWCDVx9WB0eDS)aS9h
z|8q=zyOF$o?aF+Tr^IZ;%0m9nxT=u&yO!q+iLYAT)JT5!-C#p(EyQDk;uMIkyPCeA
z%1d_J;HnGxtC2I?5^ix>duA))kv_+Lk#ayjda((QKbqN=HzWSfQ#&0hZ+_mzvHy3yEvBjJzD!}SkQ
zem{Goxe?(reLp8D>0P@$7NI-vL3+p|iX&!KxZ)xM?~LB8&!`7q*J4O8;kd{TS+!K2
zxAAJF8p`+VO1?-q_!()n(C2;ENN@?Gr
zpqKCM>`b`Nx$C2OR>19}^m_FCqbz03I_}xDxCQd+v3VO~QE%?HU#eeG{wZi%%rhn6
zs9p9~4rw626>(&$v<|pV?rixf*!gY{@6MoqP20)79r}58UVWT1@bcE%UF8|dbr1AY
zPJle%-0>q1n()r8T}Fq2!2Q}k?mKazrR;o0$3W>6
zDeLI{>fK`WKXUZ`j%HgC=RULWdmay5>pl7R9mQHEbx#~M!&k}-rM;Cx&=231+Xlv?
zzv1`J*nC%%vfvwIe6C`?mQ-Z!s)3#9BW`rriTI`9#g4`u->{B79sbO~cxKjX&3}jX
zr7dhqo`YUKe!?^u-^K6VJo5tB^GX`D;|=VoZAe#&0-ctTzMBm@FZ|t?Vg6Qz-d6h(
zi}+DXv25#l+!Gz0erw1N;BE``lTR>@YbLU1v|h^eFJ$ZkfBZ+CjveUA+g3pl=%4ZA
z{`aSVF4}vvlLPc#9-Y%Ydq&_oXN*I)m%DR0zGwE;cuf_|cdnelf@QFu9ljy!6=>Yp869b#G>HXJ_
zCp+H^ADd9T)$>ggke8G_JKQqj7`~t8c!xzHkhj>KZQl#>Gmar96hGIjI+#S~gj;K=
zx)Q#7V!2YEn^+glvlp6-fc~8I)mjvn#xK`;qzn4I>FgYeV|6w;MN@k^XBGm{;auwf
zouAhkp;{#*Ka=VhPIg~8#Bl#`PNNqPApBm?x{jp(P-&tO(Mxyp?J2%FF4valWk#e7
zSw(W=L8EO+KjGA%G@@G-asnwX%5EuDp?t)jcjmK6o-?i)zpv5QbQysQ&gII6
za?!?vPQ=%Z<i
zpeD-OxqrEr*^9?7H@gb=${Edz00iJ?2JPbhNj)Tu=Y1C5xx~jccbO3!Eb`fnx4FMF
zIB*S)8Sl9wHoDP0VdBmgjLKV&w&w9%$fn_N{OJEgcOYoof3{!nf%LC5Ri}`C*vbps
zueH@z;`W1T3xU8T%>HmX?`OAV2l+gA(#G^d-pitrQ=?5ixZK6KfX6+H?+}tiUTG+0
zhH>ArmjD*x{mX4Qf2d~;k1H=1|KfHfeYWxPu*wO%{Ns@!oF6D{;{Mx6Zz0D4P2Vdy
zez0EClk4dQ&gb*fVVO=JvhSJng!8`XHQcXNIA!yBe;TI9=iO^&cV6$#L;`?_8*9-#
zXt{@gw|ZE>Jb0tM#$3-hM!-3C*9084`<|d@4KGT17wsrvi~NX>fUnl$c8|sz|6Je~
z$)j2DzV^7D1|4WlH!L`BZwc+r7>!{;-r@{k2jriR4d(qGtZ|Ra4~((n@)y4ba(?4s
z0?_aSo6_!%+YwCXM3FYkAP-pOSG%~lcgb$1GU%4q>VZfr?`8uI01gn6(`73TNc
z8^9yT^QaoA5s5+RwFI>vK-~s^#Y510O4AN`s)+OJX_~lP-zVrLCIaWp+9G@j
zyki)a#`8=rn=kY8&}Q#6ey(aRd5y>s`Ag{;WAYpK9^e;gpatia@qMQq<$!<p)
zun|c9vp?nGew-{dA$&aV&WJeDYqn!_G3k{J-a4Dg^8;@^A-}O})ZwIdpy`ovh$!K|
zuI;fQefd3={i)ya#qC1LZsDx)6_no`)zv$j^e(r)Y)p2#zt7!Cc5eQ8+>!iJ;wz2%
z&!7XeZfv4{{_U>2m)=X*J^N}#d2O|ip?@e3m=H1k0OdiC<_DKk-t+v1D}I#6A5a?o
zy#e%S=&Jz2zkw_LmFj>u61?Z`)B~S+Oy&G-lpAH3-KKXjukJtoO!?Nu4~v!t!{6Un
z&^4a&_32xVNeK_ETls1{jVI^fF3M0cg0
zP6KdTU}e#sI}-LKV!+~s^@ty~=kD@A`uic^K(N)4u>)YP2WD;a}~n8;bey&)Af`6Z5JOTsn5Q5pZEyp@qpr|}J#4WA^ZTIfP35D|3oW`hk@iVKMZh$QYeVhW;0d7PRN`lc>mkl?+UwH}
z^YCU*cR*actJ-_9u`BB9SLj*Z#CW!-<|GyyVqBg_LzDD@<9})%D?q$)tROZZ67~%x
z30u!04%=(nx$enl#95mYjZX}evJthl7wQpboDX?hJrHrq{guP#`2ok9?AtM^E9~Bg
z&7RjDRLh}d6z$90*YhUQet*_WU61xn=-!h@6d{l7x8XMJPs6aQauiRWdp36hjn~)E
z`ZMjHxSEv%={`j_Hm7osktYxZnva+oLLL`
zB)iiAdeCqGa`xtf;3u4Tn@{b3jyQiZAMtIu?Yc>a5YHyO>OOfM_De5o=ZGyRkIP+R
zkPSNO*wOaXPxp}e<21hu^@1kTIdy!g-Eumg=DpA2IBoTg1UJIjmaz^S>6}U*&;*Z)
z{GDr(2kE7aJ?u>MWR^(
zY3@7$deG1}m*%BURn}y}wHhs3SW`aYKxx`HivK4xHwuXX-%p|c8afYu_my8s`kAY=
zTF^Vie4F{m4CA}U&ftEf(=IJ_YUe|w1@|kb6Q*&$vFx`R04>fb*Oiy4f31%Q_q+1T
z&3Jj0`W~)d7BT4)&Er|i6d(xXEf2e=kp7w}&)`UL-d(-I{m<+Tn|o2*VPsIu?`PB8
zRCr$JRg1MTr0;Tc!c@}#-ZSVc-N#;h-WLc7zwav8$o)w0v-3}h*K5?VCp}lKiQLbw
zf9cQrv-3q4?yv2Z_T~Py@-{uBz`0xb2!IEEujN_JANtsx(}B>)tZ^}cW9P7@!7Lf*i-)Pu*Kb@N)AGDWJe&y!H_u*9j)<`vn=&?iY
zQ%FxUE+vP~zfpGvZliYVH|NI^^^LpqjLO@3_VOV+`zFuZOy}h?Glj==f8XurlNi#I
z?|*d%>6y(ru!s0X((F5QKCwDWd(wM(*Ax%vB`nazcPYug=y&7qsx@u%EyK5*l_mwj#NL&fNerdv@zVOFO3zFRF
z9rMsRXUZsFcVpUzwv@;1@-Y6{U-+*>r~T7fd`CWN-n71yj~TuB!&bt(swv8APC>t`
z%d^iJ;L`(-+syzUl=fi!K=9d_D-T>i{mM%nv)_V0@#g-RF`#`7zohhqA6GKr#LZ%VKuuE=2jw)LX;5fN$gO
zvD&->@354_Cbp&wN!-B@FcDl%W`r7IkpGCVr&z`=7{A${F|DorhKd^m?
zW?(DaAGDk|XR8n5mTaBnd-~%$r8+%xdV}?HGihlO@Z`~%{R|S4@x9x1PnQv5)4aTWTM;U&&gwS4!yOSHtQbsK&L$dg$y{=KUOYA70tt{uu5LKK=Qd
zq_FM_EA;Jn!RT%!n_y+OrN|ZMVwqHNcMrVVFuCGljP*L;?~Izd$K@=?r=N9)mS&H~UUw?_)1K>&37el%|0iLuv>J&X4?Z!`D
zukixZ!)k>6L9Fw2^Q;}eV}InVU{~H@{m(A_v!es{ZTz~2*Q=mscyNPGV?Ov#LtXuW
zuRHA>Fn163o#U@JpH@RZ(eOt&)%#@c>coJ{oi@%&ItKi`aPyAnwb&=6duOE$RRsPW
zZgfirdY7aokxCesq4eCsP1v8U&yLw{Q;T?Yo88HS=O7Px|E((9a?>qW(KfnKqHFu!%?^$*9V#4vSSA@T(
zIHXNk>y0L$>vXT`c7puL(GHvGzAfXAxmE?_1*^7mJgS!UM_UK;6?~%n=y&vg<{kliG{V-n-WVbW7}qX`#@Y^q`X&
zo!h^(R&S&8++ak%m*gL2I!5r{Gq^-Nfg?s-ne;W6;*zb|#;eFbG;{cj#5>}Fphq*w
zKUo__^8EDiGrw+;-{@L?c?6yN$$1Yd$e+7g*IoUiNKo}+qhR>3E!pPl3Gn$h{}
zY7qLC;sJ-U4?J!=STo=S*?H&|)-V}*r*|hVqxRE(Jv~T%jgyU-dzrw%JIP8bAB8T
z&3&iwknHVSv=NDJeBVcl0nl)c9@FFTK!{r^60-RF2L;`csK&jz{F}|BAJ{T)6Z
z(QfpYRS-8H3*dHM-3j41sq%w>C+!r3{vPgzX@j1o>0;i$wMzi=!5e1@xNwDtzi$Ts
zMnRr+9oJ20Z>lB2Jn-i}J>;+>i4{@m3tS6$T98o^Ml8F`i=&pDnmJ9m$_uQzQYx1-dfkk5A?wYD604zKSOPviL#tIFFq7z(%tezYx?
z2kbohvXblFaC9;uxpQt5Z{IcBfyW`Mg7EV$=$&{(FI;fHH1|jV?W_5wr?{R%n;twr
zd$f0~8kI+VeDjR#MD@9j2hVuNW{i*0tPi(VhB3x;4wvN-{ne6MZ%qF+UND?-GGwq7KIVuTR1K@^9LPo^v#Oe
z?-m*>p?#NrO=?N+lw)>=CrtsL(&f}7%8Q16oVJzD*{sl;eX0HWWG{Qlr!1^1oIM-z
zBM;}brThOi=eB>Kd}q}AHyz)gJnG7xdsHttPIm#_k17?`*H^;-bi4RAituRr7G7_Y
zq4y@9Iog0JaRRV2Rx;(
zQ#toK^m;`bj6wUeT#ioC2em$Lr?dikdoSd#jm7=(nikuojW-c5wA6K?_rwvNzid~*
zZdu+b|3>(!qT@~_31z^|3P#;-Kwl%$IAkOG*;IC67Ww(=znZC;pwr9-eoX<6$bFac
z=6pBgp_^K29s*A2oExGL4gYvEweZdx;A*`V%cAp;@9I5I-mGyk@Nx8IHvKi;0k-x(
zbpZa|ZO`N;ixU!7?bXG0h#TI8R`j@S+`NI+UwG2u1IDe>&gq%IIsC(St&de+`2FOX
zOk8=W5x>{S&U67@+PD75jy&L-
zyX#u(4n2>&kAwDmt5c0k#nH-LBea1Xy7cs&;e~ScclNn5gRlxVsf%6v)bJYiEVt_y
zOU#EtL9JXvXDMr!*)?4&tehR2)6FoWMFksibNt!xs%kc|?(EHHpUA(cOC-VBY
zURghLDrag27dKw(`GqOmnEqh-;VM>WCK*wA2*0y!n%?Yb+6e>1AGyB8tG42MM~@sSPwyD+ZT26ua4_&pwuX9h;K=i_
zzwXK5`$z5kaA6MYn9I*8KMZ^}qrBvM3i6q^Pll%DVxAAI9^NMv?Ka!KKJY2NyODb5
zwyRJ+x7F);`@--$&$6R&US?8eYjwA#C)&|5%5v%c5$_%J_BRxxKY2&u-R59j9#R{a
zwjcXLxpaCR?YDrOm1m4WZ#k+z*^YHS#45b*5%z;sykpNQ=vhtb^z9JjI;x#)oT1lk
zN3i2A@KLs#8UkQTsMeCQ=5#)P8+XH>>;&DSKg5IgU`!tX3;9r+05vMtF$(i0
zy0_^((_XL~kUk>1#p?Ja3MH}1}PO?n0r4>(}F8&Wy>lIZ;PpFKz~q}VZ)
zaNn23k8$A-yg6x)1IbgYZ*o6*ueKqac&A#s`|qGP>9wjZo#U2T_CHB3clhoh;+ORD
z8c%%WyBQq6{T*?IyKe+$xXEujZ&CxG7ez|deLO0T{{nsKYb?hyifP>nPPs^hvU94b5uFL{8&G3FWEbyU&8&hP4Z8UAB&>;
z@O#X5ee$`UU+rw3pDD2Tsz>_2L-Tq2!A^C&-mYbvI1aNqdy4mGMR!a)^d{&X;e1Tr
z#hf2}xGl#Ed+!T)sCyOC&CvUv30Q#fvR0Di~UeSzBND$)Qp_LBL=O#o8m8;GiNK+RyyFr6u4q_Wbv?p4
zrd5MC(|ygfO~YSOKi^+G%=er7jZ(az$KOp$?>CU>rhN%%bZ8#M8A9b{;=0k
zGw}Cw-G6|Y*y-NO0oIzmteSIWi!yv<>nD%Z=FV^m5!cF*mg;;Ey2#@Lpjj?%#$#O;d)STKb@S*)HIq
zIqRAaG=$&VKE18;efWVp#>ca6RMoIv=RTP?!JhJi-^OFmFQ>r3>)!Bt-`Q7g~$sRl@uqfLmuo^>wd%cG_cq$!v>V#{!&l3>gm?P8m2#>VMPz4
zT2^Q>#!B;XBU|TgV|nFlJu^)?zT;9%4Lh^PC$A28&{R1m)C76FW3yhSv@B>~;i1F-
ztdi6-y)Mavk7w30y&u8vhX1K&^O@bircVv*W8S0Mi@0xoq+#XXMg?|NI@{j;TF>r$
z`FOy_wt;E-jj5kIv4O?M9+r3=ZeTZ+-+p$P)_`@e>5h$MIWyVmkQdpboOM_$R~R{`
zg1PPfa6tEcHOq2+-ot-7-Y@j{=<~BrIVMb>{VENsPv|n$mWbYrhaj6|y
z&e|Py(Q=zy&gRH#KF&5PX9|za<)@!4XFpFy$N5eF!Uo^}^P#PCIg4+!Z4+o;&Z4JX
zzw>igIs0%j{{GiBhjInO=2G%5G;ePz8
zwQ1Y6SZ~|ZB+Gkss$qX>|F~Ks@9SIAq&^JmX}XE`5ba|8-cQ@@c@EZV&h86?*P2x_
zlj{Zd!Yiv;m-n46%o1|F@HB)}`C4qfv1z;2RtsA*-_FKmSM>{um_N~W4%I_bDM
z-Z6Acuv`LssxEol_8aoO#wwMMzdtHx>jIZe@mf~F4s5&IVZ^i=rfOMYqz=1JCv3Qz
z2^@PYahGom?tdf34hYr9I=)z35GVzHIr!YRJ;hNs7rxE-0%{ysHR3Gf4t6(tJ3}v_
z{r2rK*nf_O!LN(4|Caw5c&-q7SC6!LlZ^Kia_en|pT@oqUE`&G7yHaH@rUC$=xN`X
zem4EyoIl<&Ga
zbJcUY*BupoLXF;IWQ{)cknTAKZ>yd18uGMX?`M+U!`v3Bbgop_9<#_NVh3(mN36@_^pi
zDxULIw}bptVq$s@#(B86@=rRqt)0Bfx_~;_=}+DVy^gQLt0<0*kPft=^S$Z)o+dez
zE6!24O6U6#SN$Qx&v>A+iSm5&H>Ect|6-+P$A2f_n~~#5e(L0uRE}SKQZ7s&oT)J(
zVJ4mXUK)iNWT$=Xcb=zvF=Oa7vgep#8FCHs&nb5tXj~VpzbTV`@tE}sDUPoQ>$02t
z*}lPl&jo|`jF&$`_%Ktd@r3mC&wgA$^X6sigv1H(u)ka|(Lrvp9KW=l-<#v2vAV6f
zU;eFy^aAuxo=@Rz^4tZZEadsA{iI^)pN84@~0zG|dtJ!26Cn^%cY`
zg$J)9y>DYRIc~{TnA?HmAIHjZoLAu+!tqQ0wNj3AS{D80`IYq_+Hkz0d5r#22I3L>
z0^Z*(*X21r>*6_{C6U
zdVuGLM)VKj?MIAq=It66{_-Nd=!;i59y+n3XdVD2tm(!ac~i_sV8b-MBB>TvMOGGk`Y
z?_}S)cJ`xv-_vct4r;GMyJrJvU%i?U^m8QSXZt2Oj010SeYz9r|56yHNc&-V>+l;C
zXL(kg4z`6{>UXvNHunGKq1C%7K7Ei@yPoz#OWUw5^zJ-%ztVQXPc`iy+b@KkdEl0b
zFCkYS^}DS*_&}4NGh+z9XxF#83ID#g^_?z*pCaCvJ8Nwv>G$6Mb~$Lm^vkH*bfY9pDVTG-s{NTcKC=^K5nu&ezsg^()n&SL_zyeII&Fp?S+jL2pgo
z?*0o9_aBXtJ7k0Wk#@?ieND(01w9+{sWsj~&AIkMuL$>uW1rj(s}I9>^=)}A`Efnd
z%~nlp4t>`HKILz^V!nF+o_cXB-rsc7PDy)>{Orkbg@JP#SijuP9VX!YS5)oFJ^`I@
zKe5E`z)s`|s*7?ve)gzky}y|{iu*n9v{ue7rO0Cp-m#zn>*2O?=V=Yl4}5cO%zWSv-z@LnI(X;z
zu=kKr9P(s6$y>;>$U<91rPZByb<#==0lQxJq5_UFe0+0Lb);7|?Itwsw
zxxN~DN0h#d$pp`|)08DxhYL#PB(#T}77fXth`zhX5))BA^U<$w8R&<;C3_x7k8aRL0d
zOBtU$fp6}4j_+|B{GY#3=YK%|oyoUgZ|G%&m_I3kJh*qH65U5uraAewq57}w6Q4tG
z_OX4p{J@VHGt!yz$LF;6t=$Crq^Rq1IuEW*{%cA8(s}cszr=6qQoD-II~~^M?|krU
zhQ4>AJj)ELHL-NR)#;JKeahqfI52etouhAqnk&x(Ra?~GI~P<{&#FJ2i#`9&#svZ5
zp5%V!KR_>(23v-LPW@GvnFYGwOP@44XY02pkEZg=8_b@QU8M@wILZr+Ryn?k#;g0W
zD1r2@r=*zBId?XtYBim6s}gT%)48veWS&p`UHo|r_fx~KMYkpXky=acH%r2PSrASb
zGGGXPagMz3B^Q3*kf&(I?@78iZsY!5^+gE&VlML4!RfO}Z+O|RWIBJ3-cRHC7*$`*
zQDk?}w74SbM^!|@NYd;6Gry4ZJPIyn6I~qCejMqIwVyeT#$_{5&zAHSt-8VU0jDR_
z!NDT_8#tW5uNk`VI`@~W56W@BdSL5o?k~-LzUTh#VWGbh>EAt%KQBRcLg@c_0QTC
z@z4zKXLkL#PIC7YKmK=IF2`;Rr*-i{Yy2b9``~b3Ea_E_9<__~GW(__l3#UJ9@>-q
zpXs!-Nu=jv*!1=k;PN<&TiwqO0TQk^hs-v6qrwM&M>^((AnWG56C4|E%^S
zy_@G8n^U{A;$_SZ}3+qUW0IFSX7Ud)Q{5*bG8qKJR^M9q@Oq^
z=2mz8O1QrLo1jF>vwHiTYfI~)Zh`4Gx}P~c*KX-D=viGdZcgL*a_e>$wdXm$7ymmr
z**lWMs9lv_Gsmt+yGG0N7ScYPqdW%>D3M2uJZB|={>*_PN<+XeV7GhI`}=P5N)1}U
z?!KwXi|>FxF*hpK9_P}w+(7~J;4iz_roaCU|MzK~#*rO}7YsV7`*lYBj+RNjji7zy
zj$4;Q@1A9s1-UqP4^%eu)rOwi9W{GY?U8rbahmBXi
zNm*LKGK$|nJOzKQ*!@Y-rfXHKxjJh;3I1}L@^IbH@Rv59F4sx-R4_*iuk%l)Rxm$D
zd#zvls#wO=gQb6whsZEhOz}m0{LTJ>cX?I?dl=B$Sgmgba~{=Ze8k2o7TqSgb00r^
ze~0wx=9+jv61Jk`YD5K7>5;j~?E4q?J+P)u>wOhlvf3G$HYLIrDfvV1`vt7g~cPM%!ci1=2;dGlX>b&`1@$YjF-bQ&I}OuczpFiuYAB%YYy0h~8)xZZgGK-(nI&PDc
z9dZS9`@!&qX&$bL1~iuKRX~?jiJiqmG+nUHcq*o-+_~
zxeFSfu0t=U^WRlVz!%hRHllOB&msRSLD2iW$M1Rl6mFe@}
zZ(jYGSFsHJ)D0LCWr=#TG%_Q5f!6K6dYk-ynaSaEkHO#A)yZ)+`uSylnztA1m}K5@
zI1YLE%Sozq9-MtCzy1vP=YOw0*#g>X_4_mAFLit(xWDXg{ktpqDT@L7qR3xweK^~a~tS_sLtoEfa)mCZcXQ>VyQzY
z>0Jug{;L6W(6(QWt3l7F403-8IyF26f07<}NcvcX=HvA63{yJCqCRLu2Z26Z=6CD^
z=<-sRpA-kFuGg}m@*QU;JJ9*-yXTxc)$_J{jlU_6^YK~^&%+yfmRzKBd%KG!_m|zT
z#28RL!x^dkcSlp*>HmNWez9wF{O}9$OWaWx;*FwDJ*V@#Pwj1IvUhdnH}2oA>&D|k
z0`nd9aVF`FUK;U};=`Ew=G@-O)Al^SuWCFikMx4PyN@Kj#QXP`5N!yn;O_uhYwPg#
zTIc$3{}_>(z;VRVQhClVpE8I0$&jDrJgy!yJBG)}%NE3>klj8@w($OxRG#KE-jV(i
zIPTHcx95H}>%|JLzatglG31w@59E6H;-_%lD?F3)HtIp#ub1WHc?|U8@z;&1pTE_S
z-T;5>U3bnmJ_Ia+{AWMmo}}`J@J@VL$rY|wc18FdZDzwIu6OJGtQk~)=H*BHy@h$p
zW-x&BzVo<0G+*bFFP4$NwA;8WpY%>^fk9+C6cJH!gC9RuTr}y!B
zaJJt5mGlO68O8f8Utr1aO{xuhCX@e6-etz|h)E3K8|?fs6#D&nM?bzEa!2@?ko@4n
zt-O5LZsEQwxm#}x95{bd?G3WuFVg5H(FR>h?l0>j)@IQmMAi2CZa)6B1O_*XQ~VlnMt7mMWzNs%9x^zrIN8sp)?pmW-3#iC^RUV2Ri4J
zS@*T}`}F&LpU?BX_qoq~?!A9~&mXPUUTg2Y_S)0hYw!2o9yJX-o$#d1JBi$z_DJNO4}
zLbj3}P`nx7`Wx(fSY;4h~4RvF`M@^6oGgD@^mZOq?;
z@6OJ5543A@rGeQct=@LCS;`Kjy!!5j^C7oxdiLsu`S*6^io{;GvA*vxHs<>}tmFSi
zecx(>dHU6oWfy6E*l*DFa(rL+tF=Q?5!UNrQ$@kg&%yq@IN^M&0XXM#-&r*P<#Ld}
z@n9pMpZ49|-7&B2JMuTgjpq4XzL8rZVXxI##3!I!QL8WfQo;O?peVjN7y6A`;xK3z
z_}>3in7YRj>xrNX;n61~Z2RBPMdtz}Y^ll`r}C2$*5JEV)kfz
zLtpzm30pmAwQH%Ggzdp7(?vi=O
zUirZ9FE=ynXIVA-Ic>-!v(wlw_}jWd&j;t5-Fn?sdtQxwf^9CgasWA`
zMSe>}HS9xU_~N-3aT81~J<_ZEv-NHsgWXjedNyLxv1Wm4=p
z*fUKXJ}3QDhd-`ZrE*nVHM?^Qt99Sg$K|
zyqSgkW}Ue5?+*51W@kLS^8j|rztychlW_jz?9@XC4bV?I1u8#VhVV&)wRU+Uzey8p
zk9S9Ve=J({Cl_|p{Tri~M!K&;g|j6
z7{ZHvrfQR&a`p1i!#1FY<;~tf`y$&%H`bHAbT4t$;5_6rBc(i@>@llnL2u|>TjnxT
zNfh{oZfHC2BFdQ;=%^hBJovFfkPhg5MuaNl0k`qmf&&tmKYF$=azeOV-iom#*UnBu
zzkWgZ+4hBnG`@eFpE}`+TUrFg~
z!n=CMC)4Lp(pH7)Sqj_Sqi_2x=JN;0ngvNcz@C^$|?~>%(
zu|%)7@w6m*$kR#tdS>|NtOi+(7Cj05x!BtU2pZj`T0@2Ueub)M5PDJGZ3Sa!)
zu?OkFYs{aE6uxwl0-yhPk4@zM@U)a9HLCwk=U1e2Ii3#)CcZ`EV)^{9QY7Wy|J6S+
zJ5Tg?w}tPPBGU9~iEs4KgWP@zI=hVTe~kH3#-petT+iTf{=G;G_nCi{_
z6LKG)VPggJve#Z52!WmNEaG;6=I;~^bG>))ctv#F8GL7kV%h}XjDvB4{StJO<1T&&
zxc$|6KaLlJuM1vVAih0g1iYt;fLCmzy`jhFCY$}dmXX?zgc+!$J545;INYk
zf)Ty})+NFE^)P3^Z!ddiu+&`$56eL?!XJ?7x<46C9&*0eyWwLM9Ftz2huG+Q#MDA4J>(z#=s=>H4DfhwANwA-K*vNk%|Inmi
z-3HRQ(!s3n2=ecY8J~iSHPJqHPBFK_IOXhiD1-Xry;R-%)X&pC{(4J(nzhsQr;~rS
zDlGQG3c%rYcAF{xyQ>p}NN=@WV_Yen4<^^wlE2nX^ztv+Q-K3tdK2I8#W!)&9@=Sf
z;#2C!k2)MLAbmu*BpuBHoEv!a{z&A1Cwq4a!+P}lrPP}Wu!{z-?VM;!ehkf@N7iqL
zy>lQa&Wy(OirclVG0xvMx%cBT#_2;{#^$-gE(+*Tvc?$x_#c=1=k7v2X<9RqB0zV)
zuQaC%_8)BT65(SA<+?aYaz
zAx~!5L1&VGYGJ4i`MWBeiX!NHxp%eh8EXLxy!WLn0KfS>jq$-6*cX2{xl64o{07s$
z1wIdLV5?7wUS8h*Vbhv!(u%pS;`v`%I#oIJKRaUCD_36EaH7
zChWFX+F^z7&USd&PQD;!je1>ll!l0zS>dAC3~ezRtDq3xCar;$byPkVd8~oGxYpNq
ziLaPl%?$`lMY?&(ix&BmHn7CVt)nWW4eaSz7gYm4oKLj>R%eWJ#!}Om?}cA+?j}N0
z=f{r*=G;#)V922c_UU`XyfxS#`*)(F{cz_YYd)#uWWc7+x*h|
zQf_@MGuho{))n}#{(X~F>yN5spEkJXrz`{`
z)vXy<OAHcC%nlHrd?k7X!J+9qo6(>N4(gC|>bvnQAS2?7#1Z@w^&lyldf@vxc?o
z%M|D6n6ve)X5waiY6!nZeC6aL94A`O#A+}H8;^t1B~3KFM7MOYk5eb4gTI&i=fXPV
zvsAoHeFpquF^&hE?;(8QRPTS!z&Cq!oe6z+x5swk{LSdM?J~5&ej=UydeffH0liD*
z@#~a+$%R>pvjF8k<{c;kw9M-}@GB87JU?xJ?UQE=`HyM%RE+ZCcA6u
zu#O=#jy)I^y|W|a+W6!2Dmvfzp_7q2?Wez*uX~v2<~P&r$WDxS9eKe3`00|!bx(o6
z9o!0z7t9-Xl^n>Q^60AJSXy6?T+%^I=Mnc@OuqUZaD&6l0HQC_Sa_Dk`_lf&e4nMs
z@@o>=qjOCn{t(?UtY;CG_gbVP_gj^{DTAW~=kEd}iZs6t6FE(%d9_Zu3NN_8KXAAs
z|2}(y(qkS3OMlK3sjr!;XYc
zE^z1PUSH>Ga=WSP4SH}6>-9TE{M@Wzm*e3i_oCBk+^?d3DwOZ%M8v=0^rP>O3HC|*
zaBf$9|7@g6@w*3}{z3T$brS5I^6+QeZs>M&3EziO-LRSa+tPH`@OW2$3-_}*`)Tm>
z@*m>35z?L__-bV4V}6O_qSi=Lq&{(kCGw3wSMw
z&*q!}SEDwF-y#6!55gBD33jva&?9Cz7qJ`?-B5M*uA9imiNn7NCf(Lc)ch*
z_ke&ePZ03MumUOlRRjPTtW)AkdMDt1eV`VEhtstj2-oP0ZBXEetCcuzvc`$SP6q~X
zd$OPG9tMj7!T)z;c@pPecE^kJIbEfT@!-$xPZ`0ESq?K8>6wYza9n>W+>pS}b`#*M
z83+bn!+Zgz$>dQ_gt?7y?I1y(H~NA;Pj?j3?U5>^n`AHWZK&_U`PE+B;C8jiVYq=2
z|Lt7CuNms@1IwSr6-VwLzAi6(uUYOfhufdpiJf^nzT37N4rHPK9TIOMm>)0rX-_2(=J^*mwgh^22SNUG
z15-Jl+bC)XSVuM==X_6mbH&83A6&MI!Z%Kra6k5zc~#yNetg{IzBHap@2i(30bu{6`nrYEu7eS6d{e@$t4n@CCA)cUYg`=eW`(N72x*pQ;|LPW+`at8wu#
z_8-FT^6zaObWWk-aj$}b1($D3{7NbD&l#n|{q|kzXF#D?_sqD!_si9em-F!kuWO&YJ>XK#rjKfvS3BQDUntJKt^cj$fs<0qT!E(2zky?3Ya
zuUCiVPicSjZ>MF$w16wW%gl5Je!Am0lTKvM`h54N{GEc2D3RSBdLUp8rEk;6(2Mp*
z&-?e+aRYqZAzzqR9KI$H|Lpch%1NHReZM@U`NQ6$Zf81R`-5qv6kg($JdnnJgT2>V
z(>Ps}>CevxxcAX5&!q4T>6i5|ukSd~W_2<4_o{z}cQnE|)V|ZJgWU|VPKb;Mkb_;>
zyefN)KFZUa8M7c9^Nro)HnzGrhwAbC_wQQR3#aumyDGv@8NU92&UNH7p`xP674uxg
zu)xvpfXDR@9x0D``()K!zeliNf7Cxnn~3}m1&t`)2>Q`OCoUNSzqV%j(PE?@S(#?t
z3HHFR^UpJAAE;=?JM9e6l|9~>(7d>L{qz~+_txq)RC@>bhW*{$`)vK-^5tJW
zv&&7;Uy*`&DYeQ5Z`ZM!^s4E)XB!ynVWf5l_K#lSuI5u@e*|=JfA$9FKWq=aG|7X%
z`?>XzwN?!*s%=*N4(wA8{G-yQ7U}*T)L2aO=c;ZB|ID6a{nEyS-&?Hz?NO*YW}@4hxYq``8Hp9@B*A`
zt8;6=xCraHff*Sdi(yxG+&wxgAM2iR)vCL#Y*
zeXrPg>0!UaSgozucKESEnZD`ydKS~Z>C-aJdS-O@i(APm*m?Tni~LOC|It$DekmFD
zlm3Edz1qMZe(1XTN8B?Z?%#3lJLo52RHl_T_RrI|9y=(4Js75%rnvPbe{B!hl~df@v~Hswe>%-EJPEsZ*RMS$f6zW{v`mf2p1YP)
zZ7>k>FWy$uqaW&NZ`fDyIN+td4!ZstRL@4(n+5OAs$*TMn~rN=s$=WsC6BSGsb|$q
z8j-H4xS!Z^&CWA-@twiPLz?4a>e+pRn_3xtux?k@p5za^_>levm96+5vi77wx(DQZ
zV|;e*@&cTL2rbIZN56|X+tKYF?XL$vYaWaIuYcJ)yAJa9*siv9Ji=FejV!50J*GYL
zTj+*zMHPq*?P0%;UUs29o%2*ySh?~9=#kTZ78K&V+uG@V(d8)Dg*v%E^nFL}V$&(l
zQd!nU4Rw`;$%OIE-*_HM+}S~?$jW?re{D#-Qh
z*x~~=kmvIumi83*j+p-a&x8JVWz5GdXeYaG`z{&@I9T(9GTk4O9ar{O2mP{>O2tDO
zcZcY#Nj#5o#C^?GC%>fOuzd~RAm_CX(xVf>e_^6yGwnM`Gm7)b{(q76beaqB!I!$q
z*@E7r{P62<;CnK(4rm~~9eo-#$WFG%lDlFFn7weu8iIEc`Wr6+UF^PiyE5?3c4MxP
z{d_rp@QU{c|Ixa9nhD_9oRR|a6BkBjMbJ3?{=-KfI!`NkIH{QM4o0m8k>AOD`N%B<
zdlZMel0PLxVVnicTQ`>Nd`|$>VcqUEsc~8ZQB(
z-p4+Opz}JlN|HYHpkMTSwTA3+jm2wrQ+{)8r^|D@eCyzJ@@Mq((C{j2t~hjriQ-68U{}
ze_ENA(R_2DK<*&TgBJ&1cBb_1OK$Of_0o)PH^}dF)co~b3fIyNTTgu23Yz&8@7J}r
zZD`#Rs+@-h$>2B4pW;OHJ}X{uJ0M6zFW%w)jyC?56h0%$)PZ2t@S0)@-{&jh_Jl=7
z3Af*O86D^HKD^NL9?^~6bGUtUBa3ZC59)=T86n@HUIZvmnbx&^J
zSkBku@|nt~ar;a$Sj7FFuB-ZRx=n>Uw>yfzuH$@GX)n0lu9fRB>=FrR?W
zTMxB>zw-5D4o&%k9$0_bHgS5|W5F)$7y}gihh*;q>D?9V&&|h?Ji>eLg*ZrO%@M&K
zZlvW6!e>%+g#W}>q=e%V1&;qaDDV|iFv9sWA>h-YVgTPoW;5_qQXAk}0)SPN9q8|5
zcuyI>xSK#XB{JyNGCFzT37<#eA$~=%fa}QUv|6C>3lt1Kc^Tg&QZvHs{|S1P_Y(AA
zKVQHrldkvkpf8srMN8d@X8o4G*HcRw|OuJ}R-SKOk$6%`2MYJcG!#qaL~eaLO6
zibegO5%du{_dUno&yMH(2j8MwgC4;v1m}BY>&rdV-H_h5aurUmkqB~jpyvq@-efK4
z>w{8Hj@#sq=Pgw)=4w8k6sRWi{(EV_
z20pHQ&i2Lv59bQnt*Iucy*q@b+wp)um%~JMkNK;D9-LpU&Bk}cR}?v))BoLH!1+zT
z_vg6c#w4!4$i8#YVqy2rx8w9T896-MVhg=2g!5t-*Q4R!Ty^1go^N_YtS-qp_qa(M
z+4m+H>YWH5dEV+3&6AFcY!1>p-pS<)pb*$+vln;}92b9QDviI*2W@WB_#dWP|DZkK
z!Q7yGlz#S~8(a_jA~$e2H1U-$@sGdPuMN@lD(v}w-1pQ`e4k|g6jSa8X$qF--(x*l
zxh#hGt$dVx_X4?PoiBg(E^YTK^nI_}GUo+Qf
zBh54M=;=8hJ#vD4&^RFc7CHd37j);6l=U3zQHM%yy
zUlz@Fcsmp=y7=0_?rJdK3I;exD$Jr;RPb
z*KMkw?*{x>nAQ@C@4UD=z5(!lhI22nA0&?o!ioUX3`WgZi10UV?{cRCzXAhM3-Iyk
zoZ|kluUB|Ggzf@-YN6kD6V@?b&P|*B0Ct(^tbLFM?5bD4#*U|bsG=1fza+q?j@;k2
z5$l>gJIZr!gKw<1$n_1s~Lhpp!zic<^7VJYEZ;c)fz-RZ~qfh6Fc75*DO%3Vj
z1W9Y3LmsK8&h{k0UTyrMX+7O_#hePVzt+|o?Ipi}q59-!R)F8uZSxrj`q2kJcfQ+o1`V2kS8e94DMz?QxFf%pz3*Y;
z5*i2ZMF!N8pCNCWN#bqLqr`1C(Z0g1k%e!m{zAXs@h3jr*|USGzk23hzuN_Hb&`8L
z`CXXPgbuWBShFH#2#q^qfAyZ;7f`9kTLmfT-CtJRr}6NDaVtL>r!&M#wKVRyHobaU
z321!f-7ix#w+8!hKj8*7Z%mLF
z_X-POXkwkHz8!`szK`f+56>|Cc7_t3?$hOZn@FyzE`{8VKG$f)=XI}qPd?AT-+|8_
z5zow^iKk=xH;T(~!{IQW-|s#a?p0ntya$)VbzU8}+q8nqQO1@x!~fa{7$LI2M6;h0VGg?-16^(=GqBo6qC#mm71s!3Z+E@O{W!
zGGy?bvC188w|`B`lLw&eJis7VIU^-YT${q-QgbqxaSpJK9Pe1ik=lTdrDX-+
z>GgtNys=9;r#Ghx;mbY1^v%}LEj_o|HV
zgiP+MGJY%cF5;_{(Q#iY=f5n&QwT-;>Pi99?LCA?%jkn;_(z#MdpZyS-!04P`9y=z
z4nE>*K#KQXPiVLIb_a64*cAeQPka?dbpJd7U+wjS<0C$hfbdhpdL_D=Ox``R^`=DD
zu9|LOxBwlO*KmG&YQ~7SN+!3ztbOfI5!&|$Sv!h*Eu_1ixPYQC3))CtIX%90V
z{=EBp5HR0YeCN1xb`?4XzMJ@X9~~Os2cHw#eZHX*r<;#|%jHQDUE#QT=4&Jfd-0sm
zpNyZ8!Z2?Prxt{IKTidO{qY3T3BnEA3G@VCfgVEF4Z+~)q@Ku
z{Dw(1rem)#Ef&OwrSZ58Gb&1nIIAhxe;qLpA`T0PP{kL(^
z=2zBWu#itfE?`FQHRi-$9P1ZOe9s^K&LI5N
z$O|P@&Tau|8)!Z~;;=le$6OI7&%sHEP*i^5wSc*yUkXni^#EgOEA4W@jZ
zSl7vvpJ!`2ZjF%$c6CyNJI!kgPG@0gq5IThIuZZVa0Nc^q%N)5M)<<3jlT%Kif}(#
z0lu~q2X8$Fc+*=xY8k?hjqr)x4gA*OXMwk2XT2TRm%p3&#yN2&eHY_$w7nC_|1S9w
zf3KqG-6kVChvwB=m(NEr^N;Q%`z>uj)Kpqm^+^8Rp7>Mpedf^mqWx>t!zAC@n!2TA
z&vdpp-;4B6IB8y6T9-_FIHZcc?`i1k?@4l69=G8BzVLM^ugUKCHBncW?3q0ZnjdK$
zyVq*$ObO&%I$ihM2jJ5kIu;)TeU_^3?~{PX&UF;icTEPbyY#sOx_NUEUS5KosM_9(
z>_DC85A(>rdlX*2iO#)cY^v@_cG<1Nu~I6BV&MD_w7#49%kCzv-_FI2x2N+28XD!g
zRKAGt=X`%}UWt-6*@Lbjat&nvDP6KVNbA9noF#4~r>n~+HwmCySLJx`-_6e@
zEbl+L3HU3qQXR>;ByXJi1ip%3x{-2A!|+a}Rx1+~MbL3upg>VTk+6yER-T`O3a_UqkW^tQ;9a
za*o%#eT3@anr3h=)r+{tz7i_8QQZUn{`=76*NaHbbH{sqq;}q`_dmYQH5vTjAjPw)
z?uj7}cXk
z>HhC%z1}LtNQ3eZ4W9Ot+NbY>KfOuLy?2fAU={kcr->ECYqQnfj@rxbPh&Pvx#l|;
z{2_T8<$7z9UXRWF)sx!U(IcxrkzSVQelsP#tT^30m-w>+<_{wK*sM(QjqvWbU)U4>
z^2+CLiLdZLhrV>q+DCrV%`<=t&b#fTe4SV3%_9HDk4xHhloRDdBM-0ex19GkB38E_lD&Id$+{2Bh~-UJDsyA{%sG9T~v;383Td{
z#{b%0OnmMgri76kwqqu&B|ddGWj-Fo79G?lzPE?sPEme8*St7Mc^n2;YtS*mUnh^p(BqF%v@XI>P%8D<)X;@u_gyVN57E=l#tC1|Zgfd3X7^
zyv-+@&qLSjhVXGdM$IvV=o-1jm{{P?-|WigkHdz3e14PDTF1*<*t1!U=>AFaFrje1
z%B}+@I_!Jx^yYqhw=DyCzH3_zSEg{+KS$xo#eUpwJjZ}|w}0vL@jt}Mk(XoJ7DX
}i~S=AB@lj8VX6Uf}P#B8sP@
z^|^w>g6l#$o1%+_@Juvwl&?oTS$Npr!)_79+b{EH2K_)Xz*%cD%yFNG7a2zQ?%V9Gk5x?Cefv%3(lj3I;2z)7M-GslRWgGZrgbKJnW=q1;
z(5eaE!77;eew-EHeSDoj_yBHT!>^6nB%ErG;5V#dDPC5M0DF^q5x$GWMR*4p>`Z18
z@B!Sw$2wLPPTC>GO90=TZlTI5_{78(R1n0jH
z)_eP~n@)Iogkaz6uM_+WH>U65;f6W_zZYg7%2zQ^z+cMluQ!Sp#w}Y}KU|tE@a@tR
z_$JHN(f%I7I%>*0fgaLNpdXJD@C+LPPY)(R;Xl$4@U}Mv`qwc6u7lgvNshI${ySi}
zFuuy^2>X-y=5){x-$SfRKn#x#QP8Q%pxZq^ji^{u28G4>DS
zK?b~>BJgkCFX-XQcN%!{y?!(~IB*_qo-H&7`$Pf9eS!D3D1c)G=hh}(TR`~U!!Pa0
z9ylJpp7VPY>7oL0&dAS;!%VeuzK&`A=`x@9{{E}te0u|^!f`Ii%#Z6Oq#}nIGa6WVQb{v1ZbUXxxf9~}qF3*}?*_`gyxt`PaEY;%p!TZT**|-m<
zw*?B0?{{;L@27Z*BPZyPyn4G9%q08i5
z;%_pQGa&p{?0jpA-`u^Dm+!==^mL+Ye7Zc4=xc0ma=&7TYEd$U4?1myhJ$tXBH?>+
z1RPt+sn9q+&@)uyCN{`GPiez4)}}~48O^P4uc6ko7e3w#gFP1
zr%ry`yR9t}$UiG;+T=lgZBe4%8L~^~FYC+gw2KQh{3zZsqp{z~9xMBt7D!O<`6_*~
z(@yle@ru&xmVJ6W@ukmqeZ=!qbwmTi{YsxNZztHn{UW!!&M6shBK~U;*ZF%RA8I87
z2@f$Cy@a4gXb36*c1Kn-*XK>$lF@`q3(sLDgFo875hD%er-U8nNX`cfqFiWQ6ZGefV}8~4n-Lh?N8rH6)rcSl1$z9V>ORjmVs*A1|{L@@1cKS9r;DSUn16#tc{KOkGo
z<&4v#1E%m#Uw7o~XyZ&Z4wv+%0SMpu*QQFzuGh+We~-@1xKzx8Llu6>DesIa{a<4h
z(81xK^mxhNCD)CdqfGL)(mr^Q=&dFuayjavhNGclpZ-h*93AjK=?}t#ys%?$Mw^#9uutfA@+8T3>p;*pN>6cJfhvWEtoVGI;H5S>Qz$&v
zQqP#+*s)eTpAYRH@%Vd{{&2YC;RUW2wMQaeugX^wIlevV03SbI>0jV-7Ho!Ek=~6p
zgSdX4e8IE6D9`$6VSMyeCIb-Pxn4NQ;fJAfI6b5rwjU7hM=OX4dgD@bZ@_kSLcFt{
zbYcW{W!4VP_Z6?X6Mcu5iV&}W3^|-fdlt;cfeUi#JUqC!VBg1#8OQO_yUFmw{@^n`
zj(2e0$MKi95RLTkQh>2v$sveylkq1xY`2RHTkKb36-)4K8$mwp*#hpH~A!vwdGQhKU%+yeZ|vHL^t+aokzK%du`1^}h2
zW%4}~%5B(N$mirsVIJF#QG(JrCG+1uL9eBF)|hQ5eCBji7oh(Z0WaJFGU&4h3HTc-
z4)E-BA^n|(0=@;kgW?r_6!7J=906bPZ~>N15a*Z*gMR*-QXoK-BNTBO*hdl1%s1@k`Gz}x2Rg?_)4P>fi5=m*Sm49q3y!ao
zm5U$O0o_WL&pjdor_~xkzWe~Rk_5+=3i9B8iCz-IT-sL)PAKMfeZqeB+_5-AGN
zW4t6R=8D!Shja#s~ZvyNbPj!Fg7}!Y<&aJJ&_lC31{aP~Uc>}B8
zIk2i8-xrox>e`L>Xkh7CMO(B|upf1$>y^B+24;W7eATg9d^h^wbN!ce_!-^SDV-?8
zcsVR)a!UUO_9O83z^%@;%wfT~3I7V}*ruzCrxoBFw8a7MOUhGlZo_rbklDqkul*s;
z3!^@>)45yjrYhI6myeH(=$=l*0GJt)#Bgxsb}APwL6EK;aovU?<<8q@Q+@t
zef-y|mL1l+Z}&&3j+N=x&Gx}}g!WsShbBxxzf)>9S>}Ruh@s2e>V37WDbHt0O;!zC
z;nC6Cr&SFb(ec;Zgvr%xWtTr~W6xEwYL$yWyc}y;VAErN^M)FhlYTlUV?_-M(QF&k
z#=n~Nb6opvfTW5Q9(vI@wz-C_dX?~`qhk%Ts`xQ$3+~Mj_01Oz!u=Ep{ao2$Q5Bms
zF>*m-+!vP8BVp5uwv|j;X8&M7ay9O)yJ(U#w3?OZrypB+wu)_XRnVKMR>aP_6!?xP
zu3(w(qqpomRn0o|vbY;}p^C-d+Tq$~LnS+s`bXn-9_}yS81r84K_xq(ys+p)yGmx=
z@y8x3=Sucs>dhavFDu!?7oV0m;l2#ZRW<)Q%2%USOwtrN7$iRk8hb=@WG3RmTt7IcQOfLH1{&YWwajv62;lB0M;rTy&Dw(A1xGc?#
z3bv13o;NS7f=#~qEarDm1*?&$wRL}rds0I_ZF=ol$qr5O)<30I$>MVp>{@rOWW#TD
zSrCDD3;(FdU45>Id)b}4X;q0U@ZPUg|^sIoD2@{n1T
zZ15t_J+6QYKV5vi98kN}w+_n7E1B}o+5#)@X^2PnuQq))bPwNyGRkB4x
zU%nZDdNDmaq_VYRCA0nRGAG=ylKD6L#daT4$VTFG{H?lf2Ahw$)u+4@B&r<3xooinT0Vf}s8|F#0A`;?~R
zo_*!qkDYfURIUh#O{X0lQhlit<4e`re;bJ(S^<>sF%
z=2?8V#De;r_L9PI$g94gxBG|6O1Alp;-*ePxVL|2PU`PA)oj|vZMWv(UYZe&NsgZo
zf9COVhMFl=Z1bO9AG2zz@IB|+C9Yby7jW(hrO#Hl_b>GOjR~`>ShSdpyO5bd7pWw5<8&`Yd{@?q$2}2hGw>V!SQmbamgW6X-
zwnVx&$|EvX;+?7QO};~JRI}@wKdwu9Qq6ql8jaJrSdI7B&5~=7&c&VVMdQs@*VtC2gFx+8p;124CO3vp4byTst-SP!#e#<(Ah4@0a!+
z^{>yzjx{X(hfZ$B`6?DV{Ay&Yd(h9l?7@z|tC&jH9R|0MPwkqW%bXFv+$-$WjZW1}
zeE(LS%}nTPal$A#8tDHM6PxUHRf9?lJT%`j+ylnzh=~ulQYF
zHB0-e_Toc!H4E{Xqv(SBht94G?yNDhhPizi*HMh`yEogkdy{8T!=gI3PdnA2hP}^P
zHrTScnpvE8az5R@hN<06kNc%j!*=S+&DpX5?>EfbHphHq4NFY2ndjD2&AyH*FSlP&
z&BiKCT)w~q<7M?h^H#CQf8FPpsI%4VK+u>2Kh{(;N59-b%Acy)Ti?kIb5MTcl~21Y
zUXAhok;H6aLJd3B@zKo#_>R7y@|bUAQ8klZh<1&YuVLqktSeXB)v)YZs}(zwYnaig
zZnvU7)UfgoPtoDKF-rBXpU&!%fZ)5*^e`{F&4U0ch?P}Q{EscP%^Z3sH
ztj4!i#kFjd+v~y4aDP+o>2LOWziZgu4lzB4OsZuo>#GAaPS;``9Q^4h?w39xf24cN
zg&OuvVU&aQ#~SvaA?w5fy;?S9|Kp2)x7RXj_5NeMuGg}lj=fiuK+d4odp)e9YuF(R
zg9F3MYuNGpU%e{KYuVXHO{0T6Ygx=xw+rJ@zPf>WC(jM9VfPM?e6%VIa({3Ab#r44
zv)MOV^mIfmD{~I6zg~pz_aDS0RSvFWku##CUbvT_ht{BT7hxylAJTCZPsF>x^(*Zb
z7t}JNrH8+j_pW2+mo6)JimPJ{@29Tot%`STJ8mowwW(*v4xHH2BcYZR>G!&8Dz0Vn
z+X4sAUs1;bEglWuTVKa4qleAcHm+wEt=gZmf}B-RDN0Frf32h1bF+Tk>R45(b|%|i
z$J7-@>dj4s9ilkj!rHi&t?&KIWH9WktD1pF^xwm-@bWzIb$lJ0d8xyMERQ<2|9jyB
zf6IE-#X7Z25BC>nk6raXW-I(7+qMOj;XWy+#ak!(dEnletmrl&lj_+<@5xz@@%>k3
zb!^NXH|*c7dg1O)=VEmkMDu+#?QQ4<^efpJUf-O{0#Q_?(A5n
ztkA%^tJrUf^K4)pjQtkp;vTa4R%@``R@58zi0
zRot998h(n&eh=nthQHv?(Sg;M8d$#XN@oe~wd=Xm^4o&@4a{Tx<>0YJ4XpJ>$5Y>$
z@Q%$03tNqC$p7R_GjH4%dsX*UYVTNlZ`jpl{NlF_EM(qN*JAJ`^#0YUVQ2$;QldF%
zyF~-@-H|o6qdV^BnAoRA{wmUYe`A(&W&_KL3Mz|$oZI?L`!()2z3=#=Xa490_Gqd9
zWpkGXcEW7t5Uni@%(!M>PM|lPV@N9*9EbaGL$6rpeu1BV)q|2sC8X~;c)*WA4Xni1
zGU8-&1Dmw8bM~hJV)i4pG-Zs1m}vzrpY=XZ%wBIUZ`_D`>;mT>N-6!-z;*@3hefK1
z***CKA;wl>rascsu^>ataPgnFO?wHOyk}c{eGb+gCX$4aUmIAX(+RVDCT30kk7C_I
z#cX;HHM@20e!1Mh)Q1{lMF?64rUVD!$|rvv9X7
z6;9b=d`D+m5#CF}ijsSn&cQu=n$p4DKUd)%U=#n6HOXRTTwl~Hsa(u%b-R*#QcuF}
zIL!31sS`8PABMAzUJ1)Cy?`%2TeQ_gxyG##jTx;VW$!gN
zm}ewNSkn=!Y{xDVrWNe8sZ>kC1|0cvb&kG-84o=>B|il3Eu{|AGP@{YH*aXP9x4ZZ
zBi9o;kSB$2Ytx_^7u`p~7VW6~88ig={SjWumbj1b
zZ`kmfl@jLBa7gZMpoBfu9@1SJcy72X!Mo|9hJ|qwR~#wrAL_zU|W_?CiU&1K&gvc4V%RvkK(Ok@KD;4}CUN=*xv#OIY{A
z1Nzl%m#~rJYdc+wk+AXicdC~AOITv7yWQ5K{Nw7>ZyIGH-O&wuK0$9rXTq1uU6rs}
zzf02^yu_@`;pWeWfLf>CX6a^$*+YM~qY=N+o*Zr}CgQ%uIk6Fr_E#j#{PBl7DS!$=
zDMf+LCD^a}VzT3uguQ9?Cg-rBgvs5Rz2)Fr35$w6vrx4{!anppHPySnl-4bHg4AFXA)L(_?uo;
zvV@h#T<_cOwuFuT?z=k}?_G(4ivr$iN!cd1jFOZ3QsC1|+bc_0-n;8Riib21RtDJ$E%qf%?1lua^~TYCw5
zy_S42@&@!FGRuk^+eymwt)q$_&Bwi+2JJE*xl373^H5elRLY!M-P`ldOv-efp512?
zkj}mS<)4$3S#%4yn6nG*tm0MgOm_)$(n&h+g7z`#`HV05Z=j#1@a4B2
zp}kIP)BQEx*_f*AYQG~~$~LRL4cGROvR?NaFSP2_$f9}}G?bP~*%iG}6qszX9=Fa84k?C>-CGf2vmvu5jR4v@0?_#?`f+e?|Hv5K+LD=CvVIP5j(
zxs;j5?sMGwP0AMEby+fQ6z)gGYcK&p5~dzDy-q(G^<=g5kHZ@&+a0#^c!*^qo0Zn+
zmbjvkJyPEJ_L+AhOPyW6bH*hp?82g9?&qb23E@?)w7eHQe~j-UI#gT=3VKcz^9==8R@L`$o2{ubZt7
z-c8xII#X-}J((Z#J(<3!k#){kvE<0SM%LtaWaTC?@-I!(y8=C5v{$>~0=+jmRR#v5
z{atX6uT9x3We--*KK$W|lzmN5WO=C9^qmO?w^2^-vgAeaiI7k2^6&evrOYQ;@8HE^
zDXVS!Fg`F9?I}C(`ORxmmaUtHofxO9I+y2f*yHr~;s`20nDN~nI?e?V~+S6705egGgua95cJ(@0M$M2Xt
zfA1>Az2ZmDPV9y9E!6Ed{jh}P_u1Ty`t>sB%#jwixlx%QPzpZ^2?M+5((f&bCK
z|7hU9*MPAR{Y0z#-!4Wdp
zB40~>!)4*4|AXF=j)5$^g@59I@NpF7Zs9kR(OdFwp|`-6dZ^)+j)*0Erxq<@MLGh1k7ELT
ziXRnR#7uojfg-j-o)jozHm9kjiP+(AQk;n03GBjQjnN?pCSnGAv^j3_Rw$pt4MAQ<
z7a@FY6ctp&qUZPLuzo!CbP+SL7{uXk7b+0ycja>q_itaxVSiIq4)fdC3iM<-4o8^f
zLSPZo&M@Wqo2u^P;h~zJIPR-?gyY{&)}X71n9tcz9zI7~2?Z9h-TECkK6zjhT9SzU
zX8WKJ5zF!Fivoz)Fwa7szP$EK&Mz7Ah=(f-AB-9hu^~77&>}_b=gbhE&$CD4INh^X
z4qBLqO$u7vo8UtYhOR7PLmyjnzV$Qwcs{yTT}&xF%SLxO!S%Z&Jl?p7J5aQUsThrd
zqM@JQGdzCqx1@uHz(?K4dqL3pqV0U5Pl!0mq2Kld6ABmmu75fLFnvhRRz&}Mxd$p7
zdbKlz!bMCsyHcLulfqZ1KoRqttisc8_%U)C;e~PYx)S_Sa$A|;;cF^s1bZ}pT1;@+
zv4;u-n`gFJKrrBX++~8NGd2fM{-R;yQYrr#&M#1rBGye~{0xGz{U&@d0Q@{S>=xnv
z9Zv)p1J1f5**FSt*OSncVSq3FPK}~;;>?HN9!L1gzGxsK)_&`5D}qTcpNuE?FY_<2
zp9ce@xjygQ_4T9hJIfy~BzRV{2d|HY5th83wmo)3Ll!YxNi5f2u=T@2$|oxO0UD%;
zJ&Wy#4k=<*PZnb&0N>n{vjp#VpX@Lau#1M%W5U&M>vbj=>v*P_;tx6H+|v~B)#NER
zs2p}TXS5~y$!-_t5?u1+X;nIu-L~Pqs8mUoV4n`R9M6B+Z??=i%
zXXPDEH(&o@GKG)ctTu$8YflZXm-3!Z6p8Oz>5&GizqF>kYl-gkYyJ{~%k2Vlh_05R
z%;gCF`jO{zejCI<_~ke>8^A3W(5wJAF0keG;t+L|_p<>iB|QAf)m|K~wMOy4Xa90K
z@1LQis6OD?qIR5L?`k;@Cuvv#eQG;FFTdmkd8Skd>2F7^6TOjw0aax4mmgX)
z;TIGS696Qw5iF3&L#>bSH8Pkl
zH|M-PJlcaz?O3G>H@AIOTVR@
zE#+_FZ;=<18Ga&0)&)Th{tIl0$D1L-TlB)a5%3nss~`9SnVj8a>Cx;;>GqSs7WrH9
z<;@m!uFn4gIXiIPOaTAS?D3ZH7CBq+miSzKh|ik^pisd7on8yQr5pofa<-&1MTWP~
zTk7qkEWDQtwv?kq?--o%gT7?-*Mehk=i$6rf=;6b*T1Yib(G<=Wss~iqPxgoh%CKL
zGQ36pmioCU3s;cA@v?rRE320bt|$@n;|&#dt|BqQevr|5G~lta_zPv>=08b5jKi}2
zH%lfjCSe|KAxm%X90A|R8={DPlhF@w1tK4=SP{$YF7Ra&K)SzW<3T5L0WXn_S4-su
z{E3?YGi3C`yg{KJW$B&!OFaVPzMUYaJgfr3H~I*8TT3c9?2nh!BhX)JNifX+P6GX#
zv4GG0PJ+Vjs3*;fSo{hB|F=e<>&eE|?5@JNX|aG>xQMl)=@PJX5Gf4yr?-$_9Ipu2
z>5TqCzPILbIB^!WG}QkM8YwVuJgM^iaC$#5q
zw%ZEMf9a?d$Ia^n`Sxg#;$e?E2=R@TescVgP7e+%WaDT_G*IxhR(9pkO-YNVrSHhQ7Mah~*Zlpn$NWPu%46={ZfDZhpF%)2p)gqDx>sm_{uX
z>ub3j4g;+;c|KcO*`j44{T@XaIz(*U;A$uo?Q4fCPiJHs!5*vhFF=7r>`<Y_3@r!cXkUp206SM@GP;V0&3@nk#fn&D$U4qH;-nuc2=@02HJ<+|
z9SaY#N6ua=_M-5Z>$f@(%vb#2KycLRU@Ub-Om&M=0YT@%u~<5bScb}!r9{_FxWety
zf9;H1DBQL5!X|=yd!K-2VPBv1gn@$jv9G6?!qtY?KP7yFf4wQ$*FRP+N}}+pUdfsi
zUbH>h>Qn})vAsguVW{EvP^MSJ#
z?!V{`Ec%k2KN%QeKX)^==Xa-On;y`0+qzupM>)B@g4*dj^a3s(ac=ogT}Jf^n{K!f
zj}Ee5scHfg{0Y<51}a)DH6S#xmoGE|`rkI}P3t}YV0^Kjl&sqy?K2+$q47l-MbEYY9SKPc^wFXrzwBft_0muy&c2CQI67$xq
z?}u`;60N>dAXl1%o}+QdzgN0N^VT7~JCv5lkhbydN
zZ~Dx3BfUx^t@-r4)yDPhWS
zYQUMnCw^)GL(RYRrTH3^Xdy=ZhFV^mcmZ}=J$&fc09@s${dpkD8$S9)$^j#FziwOz
zG%!gTNcwN;`(Vd`_o_bgC%dxMY#a&m=RQY@2Kkrg3+B-L-#_3vne1Jee4#svf&)0to%d|r2y;DpMNxatLuR+q`A(o8
zElT~nPmfik@_^6`1A6X6iq|=MFV|(+h?T&&I_)j=Uadt9$5T7oBrlGp^|#(*Dl!Wg
zPvZl=?)xVm<@;k--vxZ%-(32D?>~#9-GyZ5jhA=$zAnou(s
z?MdzV%SKw0yg#awG`1s}{S1PZ#G+l$x
zzcbc>*40JV$ssf@qj3!rXuLnJC9bFXTD9pOgl-S&xc~Z=D#QImM$>X`zbX%$@GA#I
zxO}_yAh+{J$A4V!gct={@FydR@K{G1h4CoxvmPH;aezH$xY?$AMrCiTxIz-4v
z_6$Qrt77TrUc=#6;qM@B=lX`e2e|%qtsPwNi-#QdyJeAv+}|xsp2+nLMP_h+SdiJo
z{o#azL%9ALJ)s@{P)qG@-6TYnL&AQ$-Qgxt{&+Yk;Qh~m`vvwJg6@HTUJCKZG77|q
zOMQjoGgTC}`)Iu|K01CB;=I`2+-?m~VY^D*Bl&aHFDX!CzHd>W2aleD=>}fAD|}8X
zl$cc&+tDiQKXT?~F3-#n+Ifip_0qG2IDr!}tlRlQ`Hr{{-=qum-RBAAV`M&_``KTJ
zi(-W1xinXZWfXQ%zqUtz1(^leY^plPloFx6^X~}tMSA$fMkS%%$tH8IC+91im*4k<
zc)W(x(C+{3c|gO2ypq&_6MDwizzHPI5j4Fh=heXTJ@wd}yu8Fji1tKN
zQ2)KB9X=h9XVbI-4`H_vZrvb6>z?)(VRusbMq0MOR2|`ah}tQX^XT=V{Kfzwp6=ma
zNSg5Sfdr^G?Lq!JAanD8{~s6c4l;iXNbG{1!=~gosHeT_Ju+XWs7KJ|^|n2Do?ZjW
zn|j8T-&2m@%FAhWQhk++5dYur+Mao?>Y1MbJunC+m?H_n<-#Hua!?&--}QBlqjUtv%S*gZVx0eSXir
zzSc9YaXsrJot7Q!NFE@>&Yt;O)-wv3{p}(=isXqy6~2GBqdKZAw)Rlil(k4%q$z%M{~eLbb6pw}TO
z>~|Eu0RvyTEY#QLF9dw%v#>wcbMz+R_j>$)*(c$C9`{vPZ;O9$Nc9cpsR5j8WP0M=
zW1;_;W=lH+c3O#WR2(&#>vbzib2&jqSide)mCJb*b{vh*-^GBcSj}g5E)RDd&T)O6
z(BEWM3;oTa$s!mK{C7Mp2;|*&Ba8s8KhQ>~Vy&ScdA;sAI`F{H1o~nJ;T)3lkjrB>
z_UH9tPyH|;=x^$8Pk8FxeO^EFY&LLbj%9Yz$
zBx;WrT7|!3Gt7x_QvHxIgwr=Qj7Y6T3kwE#{n==eXO$1L~GWPvXeeg5uL@B*-RZLd;PN_2L=LUose1pDI
z!?n4GAn);xw>VDixg8J{Mfs$E8J@?0oudu~??U}j!C=&Xlk|A12>aeVxzOIK34S)Ab+@4zUTTA
zG);tYWHK=`^MmnYkC@O}>fm1WraP+92%~9enUK`41
zYvHv4{nnsc;Hn#V9l$f!hH$wxZYIZXb_i=IU!ev7ZuThRcJ^%$%3s~2d3n6HJ;%PH
zN*u2)AI8_kvR}e+U7trM1?=lV2xGtr$|pD;&Jf!9VxJdHm{rn$OlAOgXk&VTa~BH7_5HX||4opvUf)iLx>3YnDP9Zbuk@8rfAl&`
zFUmWA2=!WCSaVr4Svam=zlC~sV#0ZL953Vxw{3*_1F0C#Kcr@WOMB$X9yy5AQGT74
z9q|8rmwF+&zDNJEtWa(nDnv`7b*R_tX=g_dE=UyG;j~j&|BqfD^cK7lcO$
zfHghHkI|6Bc*BSfJ?~@cWT9R*y$;md_3-R}L_Z;KvY`&K|9aL%W{>@nw46~cDkH?b
zkA(AK&@I%zyg*ofO-_g*hlFSV)shLs(vSk3VcRx6$V_z
z7VZ@C=ehR6Jhpc#O(5=3oi@VZVqN=Wa6KnUaZC*CmyqM_Y_k52fgnC6-LxPqOs18F
zbu&;np7ZXs0TA!nr*l2?rY$&dfOjaZ;_YlO%Hr*AuZ+h)z!%5W;z9755h>U)h=-BS
z)Cu?BKaG-Jodr~Xml{VcqnHJjq>fQtQJDGvF~$mY*493suj;;lFJE8hET8$G`vDNUB}vzH&YuS-0^jp~;w`#&Z9VSgNSs9lTyoOpcqxtDKE
z^=3KWa1g;bE4>Yf=NNp}y+VBNe9RH=$-pyAt%~ZuIIRpL{#JGJ>@rW_3+p2`6yI;k
z_nSm|C;z73rTXRyB`4zdX7MjWmjOjfo^B_eo+&YZDfOS=d(fEdlsz}Flgfj|r=R@`
z{t*6Brk0);|67zz^AwR8Swi(6*2xDb0y85ASUUr69qdyr39P-WGS~s=dr0q47oJyY
zJ6#(Dd8X*42eZMWj>>E-%7nb&SD3^<)Tdg{@tFpBLi(N&0^k?pFtcXx$HYhXGjE~(
zzG#8;D3sR@zv|ioob3AF#b)p_tvst*O&y$9UU@Bf4863y`I2(G2}v2_LLI=9qxh{&;REiFsHo
ze&vbH4cPnmmsYQ1*k86Q&hGzP$&Tz&b%4bB^X|=%T)hO>UarW$@NNytC{NB`0t*}
z1?+6zxKoF23Rr+k$C3rGudZFT=_1~9ozatl`(bydZ@*_}Q11G27`?Amk_xyhb2j0MXQ!5oUY{$G*NqkNT
zLp#B4GrmlO{?6~4^D{8cWhxTkBeCA*>P85xajrXW`Ece(==b@2xHSs)Ty5TXu`l{;%{8V-~=UiA5{_%?HZm
zZQ%D%uZt4?XhTkXo3n=U2;I83^q)Zg?S{X}w2v1Qzd26+Ai?+5OsapLX(_D@J6GCJH9@&*DN#mP+
zJUxKg-Ig8HNcuCf;}l50spuF|qnOW{+w;wVQ&sye9t)hcMW>pci<)xu9Qn=XTeBNT
z{`8^sp+A=!ZZ8KuRV-3e0*-N9-fKGW*RP3MzlF+!xBnYJ_;lV29tUr1ri&BIQ{!UEV@%cGVHgwC(W^7t73
zTb0Mxy(_MnQ+f3=sp;gOEuQ5d@rLu#N9!so_j!Bo9gTDImPQ$plln|@GXW}h%m^X<
zN|p4VM6owxrj!j;Eit4Jws-%Cm1IVvvoB$vaRr}H=&z9)<8
zSM2-1%OAPj;Ifn37A{{spULBB^Sc-jBKWBWTwepFx=ncdK{Gq~{KUET22-PZ(L_ta
zkEg@=da9(^@>7-{BC@IAy3R>C`P@Cix3w`3vu;YAwIt#)C-``nV~0jKv=$*ULWKiJ?Di!
zW;3~LD*-c6u1slGU_u@Mc;>B8&b8{zw|}pp~x#76OP;evvA%Q4Ms7_3#ogcKMfD~u4i1SV}$KZqv=Qa@E*N!L=zz&I3mO=
z(}jAQeT4FepF(?0&In}PSHgp_|wQJE?0cj=W=nA1?SP}H=1A=_{ls)3>Z9aKHaq8+@~(!
zvVFcT7C7RDYC3=3%9<7)?j45Eg(vdAo9LGp$cILFa%E5`M~Xz_g{z;CGsRw?^_eT7-=s{s1LI`qg*CXzVlJ2hr}}Ahn0hHp@Hw00;`%dfnD9Nwotq?K65RJ^1J`mMEoiS
zUuhhwqC@%lqXz3w%tQSSpLv%lPAnSHv5>xdN%9`MMjbrOY0^~ZVefv3%H
zq-H>l@Tp5`gkG)g%p(sVi#>0CYzw)e_s;JF!HZJATx&lHR1h=FI1f8KzbSQ0hP<~e
zLcRffrDKg!itIA*km%&$U%{txJr}EnqTDGm!892>DM4-U3GLc5xIu-EI6E$0g
zIe~9qjnTQ@4h*)N9xlBF=h7`J*R22_n(%q|B17=HS!QD_`washXm&j=4&MEwIw?0@
zx0Y$|x`10N${ruhF=MNvMp1iuFf(6kX^5tCc4ethDK
zN~R*~uYF_*&cy{o{j*0rW>#|r!3xd=~>fn7Q
z4~{ABl?eVD6|lKA8oa^E>0B1hcV9l;@e$x0W)>aZx5WbMOnOVs%8vpTdgxHZ^H!{v
zb1sh;pr6HB?oW>4eHSl0Flh9A{65C_@iyX(4&l4ctp<)?G-?{1Z$19G4kkZgtf=s3
z7BC>g$Mh+lPivpXd2OFh)}!rEekgi?{2<`%M>&7!9A2YflyL>?J>BHGLNoToTZa`v
z4k)*{tX#7VIMSib<}SwnV4cJtU(ENISJR_X@&3>JJU%2B>vrABgpT!)4~q`v_a&oC
z5}vJhL;hf+>9*mR7cc)k_n%^%m-~&pI0|~F%G=aRfW}wDRcE99PSe%(^qx72~J&E+Yp1k$z159x~mrZ`ZEbu%C56(X(zZVe>
zn$vFt;k5zpc^p$ou<;`~XH`1?E_2$crh?@(56gf2z)v9&ca)bbAib$$?sNX^@?(VX+@bb^$Kg`XavrzG3|+@@
z&H@b{f6J%Ka9*)_TPZJ}sDxu4+Uc95#pC}-RWBY-Yl4zNSm3W`_UG+xc~1lj^Zyse
zCE&xg37*vd%8QG6dA-Um9=8Jr$A@NJI@mjXkN?rlVYMS77YBM
zqbgsg4xSk*)oYBH$ONh#v(ZR^*?vw=Ziay~DQy{!#fv
z+v#QGZys9KHPgN~6>-6b>+sVR@#?gX+`|%&k=_%%oOaT4)lHBizZqrr9E2PGx?<;S
z(o?Dq)hLtc-P8}(&gMQ8FYqLa<@q!Bs$4-l*j^PGx(snO
z@RzuW0`jaX#k1c8A`WUA1-<3@<~{P^%HWGPFVv|#$Md^B*?lbl4`^ifO?1Hr%cq6S
z`i(dq=GPKsoR0UfCN&~yE_k`sXTOo)ljm1S-6bBXDc0yd1^iA)x&QScSpvqy^VMe{
zPiP#F^ma7zm4jyuSk}c5O~bkujYOWfxbfiLcHE?NaN_NM6t<|id$!wcP>3@6V1T1Yw
z|3hN81k6}GdW-7VYL=N$X!R;SpH03$H@@~t3>%}rEOzLX3U;W^F}>%J)$Ge=&2?Gp
z1WdlyV7CP+$*lIO^oUlsT&A0AyXe)KYUbJ=58f
zD&4A`&-T~v9)3np!L**)c55uV&$_uk@3DtpxOd!WO|$E@qaukh1q#jJJe
z-P=y-#Vq)be8&-&3f#+unkg3w*q`TVuh^dw7JmEaOc%okY>Gje{bkhxHtNQuA@%Ietp=acn`0f!2u5L>8|o@q1f&3Ui%hcKddVfayqs5dVJkl!5&U+X)B6>KQRmJGk*%cGoO;w`s$Z}jXkooZSJLXW`EN?#xk;q
z-8!7QkZA;*+*
zkRR}B`ndE2exFiS`RPCGCpl+5FVETd?nooD*COzD5sx{2-(mlbR4?ediGA59exq&~
z&RJvE_8)xK47{oof2AIN;_+JV)x_7ELgnpuVZUCt`5FA;i-28ht5WyEzOd`1{_-mB
z7s{0aYWG3^^#zCFV%YBnu6pl%aIaF-H~eN}ALLJy?pzn(JP_z`IU*nTHFL9iMP_2(
z%QdWW%uNN){g{;MJOX^gTy5D8W#o&kChJzi-u^q5St;Q>k**-KQxbO8SC*%qM*B^>
zO4kx^ziMZ9a0#%%r@B@OIK^giXg2IL(#TVzb9C9%B(o0an=Fp=afa-s^-6-yD)PJrnnL_qBy`Qeugz~z0W64DrXT+FsgFI00|5|p@TcF{y*Y+0B)7kgIpf}{3
zTIUxMe+-W}FofcVT4M&!ufJXupiOZjPcAOHKX8vi2R>~;|FT`U$%g$`G#^tudFShL
zjpCGLQ~U;sZ}a9)Rwce(_Dq7;Th^&Apz==XX-V|lmTCEn?8H9?D4qj(_@j1X(jR_r
zU_ZJy(1||xiQ|ES+_7H?jWo#`ws
zfaGIdu?mFV#}s(JXk#V)uny(x_qdQ=`4B=7c58cw+DV=m_b`cLt5B*-~eLh$7b
z^5>CC8YB<>Rw70iZ@dnPFTAf1XWsrRiNywF_t#t7&1hWO%5_s{oc;8RGe~d8lu4X7
zggsB1MDO24b9y+%(bAW*rATk*>$#z%C$F(Km+*JXEzXapv{-R_m*4!$c~9t)C!Aj`
zkb}9@
z9v||77v0YoMe?Zbi3+q|LN5I_s)Qd%dbas0l@EJ6Zm=HEHOt+F_P<3&8Gdkue9FTu
zSL2~~?Z}Q+@|R=YzTZQ9a=Oy^whcfB!;sGLs4uiP`tK+Fen@_(uPDl|41UM&^L7kv
zTTA}*Z+dpYd-xL
zYrhTM10T}==C~X66O{~oenDPx#!j33cJ1ZA1IxhArng-kOTV|1U8pU07kVpvr-f{W
zyw!e~V_*2=O#BJUWaLq9loVfIkNN@ks+2RpC-cN2)O``Z)C-&ft|A}l`{(QLH)-&9
zSLR!(qkoxBhc%AS>&hzGR|{VG^PA6v70d8`cZxfQ+=AZ;l+9Uzxc#T(sBK@wCF`@l
z9~7cJHmce0_A1QhR(}q_8W{I28HenLfiVEEBNVN1scM-d@eF
zJhXC)iY;e-6xNB(ToKDO^B3-l-GJZuO=-+Dm?mKB+Iy>q|3ZtaNZ
zZzpoKwEO&+xJ|%D?VNk8@ek~M74Knr5BaER5BhHky32AhrzlOx^2zPxJNB$>R5@!a
z?(Sc7yqdZF{X4GNs*;_sGtPcE>=s*~qIUXs$8EOE_w9^b`%>8c@B3Qg=j5@%(qq{-
zJu6tby0=eyKKylcy3dtU_zqnVcwx}GWM(lbcJL^c!}25J1I!DbFa;THU+bOV@oD8Q
zp_mVmJHu^sG!t3Vpp!dCT}orwY@<@j%Ti{%)L5r`C%(goo_V)hfP0COX=OL#Z!z=5
zo}&btlh}rlA3F~AdBCy{p0|2%wF2jqeu29OVxJo*rpbQ1$x_D(#(wU&$$mRJhJ`=6
z&#HgLCe+P+!uHwj6?4%Q;O`2{wxr+2x?`^vJFDV+r`$WR8Q*_wlhRz#jQ4(HTacD6
z*3sAkbBpHxFdlKYH9x_Np7`8&c1I3*{{7RY^u_sU&ZQIevEW5pCm9zG9gHtJJT}es
zItjjXL{B}gAL>2FU2y{sGSXaIy$IpOkH{?_*tm`86RQ>*meXL47K~
zqfXjyk;<`xAA4{wZ-TLaUEdvhWONbE#ZB&ADRZ$-y#_zL*Ck+C?VJA|gFTrM!GEQ(
zUk>Dq8f^*w67b*b*kG*F_=auLkAc}{F;$fJo98{#ECu{ie0ikKH|QC<9@ja9@{jsP
zSJ&hDu}4=LTSIQ~lU(`C65j!g$nEQceI2_=iq@!b4>OB*rWW$u90#hfc8g&6Nq>1dHkj6CiF^29G(ydy%P1HTW%;{Zsxg+
z^x~Dik0SqEb>Ua!71T%1*!SEK=orx|=7@5)=QZ44FU?8O>j$~Pt9-#KjO)xoGHGT^TR!=Lp6$^=iaA^tKWtb9H3msS4L`;p)O^xwpA^84>3C*L9b
zy)!P0>Tm2n_tXfua(I*?>FErA$$8Vs&BKS$crU41NYi+x9XsYj@h5$d$yUPU#7n08T-^L}H67+mkb2fiYZ~Sg
iY&SUcSSSwQf<*Y^l
zy+@CyhdED~`A(SE+cWk)pSKHQ64Ipia&F>mLc>HyH9|}MFLMc%{l{@$U9)=rc9P9^
zd+wq4*lt_Mc}`yR=UG%PF0qaCp`<TlFm56jWOr57C|s-buPkcm6(qj%$-Q|KPcdftu&)ZV^R{rEg9
zEp1*-_p}l*4Sz{*Xqw{MVZc2RC9Ca$mmcogOuviTP*P;E8vf#$#~N3vKc_kNDfu1s
zvKM1l02SO$cDOR+o5M?xbez&$X30NUyeup<>uQ5cR_aByI0i^eyzcH$=31k
zcaL7q{wa<6Z*kF!qksnz=IU*OUferPM@jg-`mV24>d=#DsJALbx#sYlzg?l1Kd9{2
zTwvM8azDDS3Aw0xC<{ELaPIWJndsks)pw`0@JAh64r
z6xf5eq)9ZtDS%%;sUGB$e-`{Cb!}=a`Ww9c%*;24R|{-jyic)3yqGXa*}7K1Bu|Pg
zPpGY6)5Kghbs>(<(aQB5hd4U@wB5W3$gjP%4OSS4yqd=Zt?V0fYuHoyGjbw<;H|st
zuPeyI{*ITHX0UT_ztcv$!3*~F%CzhEsDiCOwr1#s82G0MqfLHM0_JiScDcS}%xCV^(*Mo$=4Tz^4AVShnEYZC0!9_-x4aGB)F)o%gA+>Fh^P?X%N?
z(aiAE<^>uCDQx*W@w3BML^C%9v6|eAsZ8qQg6seNy~A81LhZ7i-((|dpN^FFPGnm`
zPM$c@e21NXdi2#I*;}m1zSVSKzgsNU#V^vN{65Qgc2CmP_6b|}c%F;7KKS01GmD>B
z-ejkT#@_z*OcZu)57rVO^f;M33Hr%KsAy|pdl23D|a#n&I3h9|Iq8)_XU|KeG$
zUPMOUf+V(CQ-9t2zqu^C?VeVaTOr%1u6(ysa0_{|{L>rWCa^-gY3lRC;#qb04~O#q
zl3BlWC5b_)cUkw)F-Mn|B(RJdlYO3zPhjmKOk4hQJgYPPZ^Q7B3HU#d-{#qh-p2n|
zydbyKP{96I$c%2INY}kS9Pgrk0x30Mx?8kR~g07-&oFCjXg9k9|)8}id!%u*}
z*dE^J9gcmjGj4P3U+jymfu`+evEPam&dsFnk}VDEl+>{wt^Tf8*?|3eLE8M*e5`Ma
zp-KL~8U@VeLgv)Z8u;%(TDa%s%h
z^n?#fkjHXaa-rxy=--dNbYvIgLX!=li(qfFbW4>Y+6y?dB~};v_+am)PN6e!UaAjT
zZiIgI|BU&%8Fo%|Nv}KwyYw9V57W7Dhwu1y2cVMJWw)+k7pch^R
zdwxQH(ZKY4XUGSCqy`gzs{JQdVS;jzy)Rc*K|Z75*SHwC^2J0Ad0?F8mrxRgNgLu-D)b&3lC_hkwwPIwhPUq*2wD#yG*4ru$+
zX(!#!=o()?OK~Tmc`yDz65`E&T}voVl{$Bxm;=0G^rn#bmF4pkd#bPdp~d-*SB=F=
zYEMSFO+Fj_4*UM2jP9NEOeeRK{#TnZA!C59=5qb$93JI1E_w>|zVtdMP5kor<M$qp^Ku>OSWONuz}d6nwH`Gki{vKr|RTWPd~
z>i5fbaK0j%Yc-u@yVAa#e{3n7eT(!uvMqSLA9&`-5UPKAwH65`yswf*&YON)z32J0
z`4MyOlm1{q_Iirbw$oLS=z+cYFL`@i3AOxpo%W|b@VrR+%>xrjuTe1f1nDKG$aCJM
za#n%gGrnp(%lVL9knneoPFlM;f4S2#nDe{jEywviVr=Gk-oCS%6VLms)?dqccbAOQ#UNN($h7OeQ)J2{gp*h;v6s@5#NGdXm2%PzxelyiRdEjrZK9
z%jWc67fAeWr}t$gFyBY||FsJL&Q<-hXGch{DA05mt%r8Un^%dSJk{FYi}ZK+&Pb#B
z4Ig|skM5r`kLPRarAB;EMESOUU1LcvvhwJ5(tEEmX$|$iVbfW8>UX^Lz2CHsEt_NA
zNKZGu-jwwI=sb%dy`9n>wZwOhPM!RO_S3$mzs!%gws1Yx
zW&4E-$9KTLNF?nK?2q!$B)z+1fvp=f7SVnWGJdsiBJ?I#9`Q^Bnp>^57eV>FL&rO4
zpY7b}hd-Q(-#0C59j=D@L1|;{ZLy!%iC?Lp-@T~I{gY#;zgRgYjrM=t%3H5#KbRhm
z)Aok{G1_UI)ehNe-XUMwcNd2Bj-~TbT*DO)x`&*!&RerT{7;^=|8%PF%Gl8q4>@|b
zwZU=t5jC&C-jCp4i`rTu9z=kzNNbcCq1{18emz?Oy{yY$YOW*ie?X!&{R7U?Sx@fl
zT6G-zWaZD8xj4^%Q4%{~0sZP^%lqbNFSqWWakm}53;JArQ+FMB*UBTkO>ka+uyaX^
z0mgN^x73+b`1^z7hYo)Uy%j03@*~mzu8DP31Hg}-UK=qf33kqrs2)BF=ijr5zcmx>
zk$2nI_V_L8hyEJn{VEdny4|p`h!rq{r}t)@ZO40P-}?R#@+Mu?Y!3NRbW*^)Cl7K@T3gN5x<7ZyGA(6Y
zv-@OuXh$)F?jx^dFiw#tv+N(@dpGo0nYOi><%@Ne|MnA^+7D~)5=~|~1s%c8=d)PKf=jcW
zZg|4dy4NX0t;V_YkM;#S=m(AW@;uUfn_2yC`gFGV9+MP%I!IIEAk~@-1{zb
zvoO!Z`<~|f+gm-At^J{T$~^imo85U;>(rMLwj%JR{47-g+wJ=OYGWwQ*Ta)y27gUu
zBlPc?jfuR=JR=<@E!jrfFUD+uUhTc1rUS8GE#CKQn~L`z9r-m7d7^z61GB%DB3{4D+wBTF
zR(~^IGIt34&av<3Juv^tY+Q#7*43;jl3$!K?hAsCTQ#t54?29=HUs@ee|h-Sc;hhTgS|CoWH+=d<|iy=IgfR$SbpZ-Mxlrz6-7
zdw%Vjk@W)Yx-iM?FVMR)X5R!!lt*;Uct0KbO%e0psBx~b(Z%2T!MVfN?mNki%HAu;
z@0|-0SkT{_upfKx3FTR?s~1fozAqyIq}dyHZ<}i~d*ZsSUK%ox{lA29BDMW(>4BCmlw9(q@pu1oG2S?|$(3
zE}p;V6#47nxf&MauXnzlBcStWNccM=(o5RcXEoK=KVYqdrS%m*Xgm*t`o(d5r?eo_
z6Re-pkLp?bdrOK#(!Foi6aU!owGcmv#(UZP2S4P;IXHGHE^N?$(uVbFY_Pr=LB5pX7?u*L$Xz53g;0S_Fp(w)i0E#a#sWU
z0>UNVeOzhXNdMStM&t85v-2vIw_h@?r}4KA@T;eIIbe6{angJDNKS_I<~g1GMS4y*
z{!Syk9e=ehhyfo)jr~A+@td0~h#$Vx(KRM}N{bKACcVjn)>Y8oHSxCiwSe><`JGxs
z?InFWdWqUwGiUf6(%bj$w}A4A!xENlr{^7g_wnzx%7--bd{A9-zupx0eWe7owEq8o
zaIU9#c~EqX%^b)^>F&=cUdHws`EN4HBaX)y^@E=`_ga7CD`eX(9|9~;&u%I~DeK)Qz6MMr3^(wW)B53{irMA8xK7U77WYh=5N$-rPLgM`&
zl60?9zXR&atf}97WfvY#Lb=Puq}B)UlRG`G)X1QGu5`F6t=C)WihZqtfBh8J&w-xV
zhpY0G*U8YF^G_9i=vPwN`FO}i8@wa65HB514Vioi@zUbj*L@RE?>tfSOc8LRRktCH
z$J8P4rxg5FOWoo~7vwc+-yXDniF`u8Tb3mn@V`?W9z8LH-mh<}EqkFC-%rW6(h>JD
z7A>*A<-i|~ZrN0cc-}C!=BFh5j?TlqNqZ2_$Gzzvb_sg-d)Jj~S%QCTYI<3U@tif;
zCi)3DU+t@^Ecijp=kb+-HhLZUe6w8skNn=rF;=t
zpXpV~Is#m#MZjXRe@4G{t7cgTd?YNF6|&pb0Xb1^N7+n^R>@4v%LX+iiPr}N
z?8={#P9wi+cJjgkbLo@C?DE5;=jV1GWRup+8XmhH{&Mu4(#1Xk*4NeNpE2}wCiEY$
zXJi@s*l;>xMPLZ)hz*O7?M`JsmG@n$zjT-NnsH)_sBa0oetSq}_m@gmlGM0t0OC}{
znAA(v-f1kZ>Gh8!|1@^8V&~c2*NfR7{fwu5DsZm$Srr=-%-cpSc8^&652OdD5Y2Csui4eTP?-&6h{K>D+SnJ>JvKL<93{i2G-JRg#~6!*?9w
zZ;xr&;CthWhLNWRVV{L$$j!jLQLEv#T?W9R1A1o~Azr%17M%0N{#YS$;55Zc@53Tf
z+pumAJ+`-yKzZ!2_ADK&SD*L|U&K)U$1+GH5qj2Hx72?_p72fM6#0#}elL^1Lr*h#
zLGc~PYgDH-he5BeNuLb^vA#cL>R2kFe0+A4LBbr&r*d4_F0AWdkNU84!1(mF7bxy)
zsB6wG!niv;Yu2s@4>}xn$C-X#(tFO!F3Mx(Yj}L04!x4v`lh3hgO8h3&jpUO8RdtHSY>nM*A@mWe%6ZJD5JU&DB)S9
zoo6H|?yVU8r<(4^jZ8LG4MTmGdcW(xfL11bbLl*zUoW?o>J!sl)9AkC`p$tJ^qja}
z!4&e_|Jm&tNPgU3e#~X^tFvv@dAyuzlMwRr
zQ>%YYBfWmkzkNyXjsGWpUOki8f)B`HmvwFo>2(dLJVpK6JwGf-dZS&Q;e!^OLmGFy
zB}`88w;;WbO?w*%<&u?n{yWyHhPPAmu#3l8|GsG-$&TfR()qc2#=nt=g2?<
zUpje!$Lo=$_*oP5^j@P|VB*)I90!;(mU<@?ke%Eq%t-9@DA@`@F>nJe#Y;$Cq2*c??ebU4mz`$P_58RhT^#XvaOuo
ztl##Q+ZEQmi;weEi<>LyDchT;Qv10%>3rT5&*IO7K^8xLlIL5b%kGDgd@n_1DcOH>
z-SGnA9o-A04d}TG>d%^J{RMS-UZ08esJO^>&veL3)s?HLUSi(P!?X`ZBu_cv1ng}&
zH2x~|%>3l7y08!R##`0Y!WV4WxcoW}H1qJjXX7760G|adO;;VRYXa`?(-l
z1SlppZs$Vq30rxsmD%to*+)N%Xdqtpo;gA6B;t91@5Sm~2Qkk@w=SQAKY1gFJe&vp
z#JCN|S0VoXovB=`fpb){Y5;p11-|8BRzUfNMwf*4ZlG$|Tx%)#l`~94&fqipCs2QT)S7!6fu?iJ
zl*7=jv{L&ZRoGGgaEat3_`F5iHfEBz59cw
zbhAtI=Tr&UJw2;MKkMMPCXC(x72`j=IZb1uCgO;d@#=-}>+DCDiWKt5Wi6F&Q*$cu
z|BWA2_jz-dO>@in;XU>qi>&jS-tKgdIjFl{dZVAiuCBWu=lv#wCBDpIUBlCu@nGgo7d*<>*37w&taw%%hO_6&J3RHBI8Q5Cz?3*UjXJuIz%
zu`Q34jr%x0=2kXaU9a;whGnwhS(?RSPfD1Vig|<=zGu$eQs6Z;y@1V_Q8&%@b{-pa
zuhGatynv1GC-pf??h$L13|=s~p_DysnZ4oSr9!5>#ACb7+Y%-=;zfJgs>ke+^5=jT
z-ygFmt2r-p+sjzs$>s$^@{5^P^!5IC4?SiIW#5-}&8TFavonWxn`Gec)vx__dTBZn
z)Wv!z9DT^F><7;{1U}*VIdS~n2Z*l^Rn7!iX0pp&(j8T8xolj^G3oc;p0GDj9y`V#
z#yTpKtR8s<{MJmn&+8v~tfEl2k7^k5GP;os6YK>np`Wtjh|BORPG80}Vtw@bH%Uyw
z`VsyQc}qsqo=W7|CM-_di~GwD7P*tbw~e|YHib!{J&y_3>;6=*nEqo8wIz|4sQYE`
zY8QBgqeieE_~<6xx+kjGZ~v6P)vAAjoyzCtowR`;TV7wWhwOj&Y5!XAfu;l3&)a~%
z)WHns{t)D?S3-S~P4bybuyaD?6tAlyI4609O}dYL`>ZbiAo2Kv(T`?_fakxpw9`6-
zef=Wn+$G!5(95chO@|$d87sBcqrBJTBkmgTJC52r-Q3|9-UOuA*2CV7Us?lg;5TKB
zv~pLW-E$N6jhPKUlhMyPknUN>^{o)4^Os><$UY;k(tgP4G8((Xfi7ZkFa4k=
zD;LQEFwU`K%E!}vO4y{(M_H(!yuUj9BjncJ&YLNYn|l`(JV*PlpI_;a1OAYn;Vg=B
z*|!XT8XErEA(<@%rV9>8EywtikIL^Nzh!*6-z@sRMM-auJNf;nNY7O6-*aDIR!8|N
z3u{L@pY6{*8cgw{S~u+z@e9AaFZh5JezauH7HiaxGyZ2n=P!--9_gk)gAtC0Np^nv
z_BF+~rfrv0=zJ&LJE(Lv(Dr7L9QpmH4J)%L4!juZ*0)Xy*Z`8jm?
zd}oTw`)kz?Q~7#BcOGZ=|9kn9;?#hVWk!@&@!G8}O>rnCdzvljNiVkL-zlWj*=-}e
zzLJv{Q$GRce0g4Cx^ElLlbksmY)$nl_1}0N;;vyKzsD#~xX5{q^ZPc=*Jg{ybDm);
z>(9UEP1M`LdCa>0DfoZ{>sws--RwTA{hYt-d-rN4m3y5AO#yyi0a^m=mzKrl`L&b
zQ%C$h(o5}-;O&;47XJMNH%VMyp!}5WYCa$1t_XiO8m#NW$FqOJ7p`ApY{u=YbsWR>
z>m_|TUd&s|>vPUE@%pq!bGUq2yplh^wd43~YF{x7bO?_9reBL*YSBget-PVKhMW?&+A^-
z{k+F{-L2av{zAB@`OZ(czbPoIzrg8!crN%D%ifW^jyOCmT!!G2-_OQEg~E8g?6wD^
z@W>(*1onqw1wwH6g^tbe$5Qw|U{`@Ib|2M0%}4mR3lRL%xc}?|4i|WH;xW#DD#cY2
z^Rar1fWeo5)1gde#-NW()stD;K+Fu-8&tvcY@KHKX!e9SMUf^({52T}(<#anXkng;BTeD}V}NrrTV7xa2>fIZC^+MFv7dM`I+yKq9i
zj76?HoO%}SZOdzZ2!?VwTW<8m`_XA*H~arWJO#^R_l>~*YnexJMS?z3)1+q|;LdNt
z{RFQEWq19X!S(Xrtu9j+(EF=@_rgoaC#b#8yr8KK*$I3uk(2I%d)wh8UHL{-_r#3r
z#opJbnB-A$rMn%CN4vIVM7E*wyZ0RK_JTd1Tzx5zw;i>z9p^eySB>gU8R+ScRUv8P
zE4@O%k57I^o}5eV$fME4`?VX)yZTZ0R=sRTwdrRY$`qSW$LCKr_`
zwX>x#uPXd;BMD44pn&|Vo+>Q$sQR1)<>284s6%jbi?7M?Ys+;0(n4a+U?{W@P$o%A;J`O@EWDUX|wN>Z*(d__I_SE)WT
zr`d>9Id@ln6l+6zo^z}}|F)u;pyY$s9(STFbe&^+3f$vq7Jj7~(So=hY_{5<1^;6k
zHhrPZ4EI{+W4B4cKFBtir0;$2pq&msD)TFCLxK@&cCl5rpz97tv$aRye@q+pj8FW6
za!tl6#lk*L&cvR*EJAIl(}SxaIjk9JKPzZu(Q88dU_+}GgRo9nsQfNcXvPRy+i_NHb_=TDO)q3=3;`{j<2O3c00ndlY+-*oa*k0yj
zQ9CkL)~}h%gL^_YFC_QpG^6ID+C%gHZ78ss@9k^oKfl7WqLZv(zv5Ho83&Ixq!~?8
z_H}AUXS+=}4?%y*e!eQmSQG3{zRPxT;5R?rDK%V`z}rKD*Olo5@4L}$|I`}JJ1G6x
zs=yCJ{jT0K1WXmI{lfaC75<-eMUV!3@)EhucJ2<`&(VB8-2m&KD`aVQ2I!AGYKp&X
z;l9{EAMmz8*@*#1
z;Cb8@7nF8^J=^88NwOct`^%mMCq7|Fr%fir0q)27_v$^z_s9k$!~^g?`-}N^n;gMD
zylb0{r2-%QxJGS01l)B+qoti-TpciH$x*}hgK;naR&d|X%$vOp+N;EAFjYqp>_>Dq
z7v@tb>}xnSKsc3;Qx#1R&!k&Wa3j>ynT7d%EufEu-R?6EU{^$#H*k`{UL;9M9orAc
zbENiFJcK_b|Ldh4#J4$q96lWk_XMl0S%U!A&I?jo!S3kApK8P3zZ0PDiNN!~Y1;NJ
zPJsO0T%*i@`_xSmFi#5gjuiO`sNb0`(g>LF{WUWeV4g/R^xs?kFy023eWX2^$n
z@FdS3e+;tq*71v9ARh|z5R((=>O<8qCkjzRf=3{&NoI>G;;{`y30_ubqpjPX81
z5cgJQe!#h-^BnlR@zQ$v0IdJ=*d-o(&M$jnss-=YX+2b!!}FQHH#tEI@T$1xe(WDP
z;X-^@?#yKN681lBQZR-2!Goaq20cLcV}#=n!!38}3eM-|OMc?}
zcwqxsME?7O{+Qr;Tn-NXiv1Uu{7P~B3HG0sI9|DDlPN~oIWH3S?^P9qFW$mDKYf+h
z_o?m}RmT2@U!CU0_|0*S_@6j)s%9?^H$8p$EO!5)>#-dpuQ||Nz#`jRf?qvMkAH&Y
zR>lhiZ(=y|o8U89@9c{oMr7alEy#Z3!rHfQQiQW112n|B`KR>=`4q)z9d-3G910QbVhUQ-g4qQ#K~fJy+M$c`FUk^AB?b8i>Oi+$4~P^A&+L
zcsmx{S09Du0&M&O0C-DfD}kOtSaGnguj%#m?<_sN6gzr7CSRl1x9rh8B7djBa9bR{
zD$AS$okGlqTe;u
zWfJ|SaUPB(!97mUlc*m--${b^7e&H$D!6z2(I@&_d_)Ez>y|kadM7=D)Em#Mb^?KL~hby9LK!w72
zPGLhM9-oPtbLH5+)K85Q_XaO||NVj8oeN5i;&40s)d>BsJLV~4cg?7V-FQ5<8~h~F
zky348!{IVwMTvTMyU#@I8;1&JsA7G2UAr12!0t;Ny*U3H?4`w&VImK>o7nx@IuG`^
zy-ng1g85+YK0aN%u04KpKEM?4v9*2HCkUUE{bA21kheFU_K(-PRf(dTsJbO(MJ{dh8LFYIS>?(<9i5B6cV@7`AV
z;6pn)ws+HO2`KlzD_O++J5*fAt_^zUlp^
zGgu#4Jk0U|9w0d+FYq6nUxt1k@8-#Xd+_S#Z^**De{8H4DG0(kAw%or2=FTEA^kC9
zd#Jy6(Is2q+_#rmd0rCh_5E&XE~LMGC|m=7=OWcBU9w_3oKwtS7ZWgoeHGi;-M@i9
z<;FiyFNOJEEK|iF0pDTxUA5~d@ExCO1-UZdQ_qG*M%IHK3i|fUnZfNyW9Qu}$xWST
zb*@b~|4b)}&RX80Cenpk^(5{owzVM!1;tuJsdhBE{BlGJ@_T*PdBLLv@>>!(o3a4*
zl>d)tgedfy^nRVwIuW_+LHKQXzysLMkYn*KX%NH
zA|$A%=+)bTEQGtzA3KeM9oAiNzG^A(;X)TWC^qQg*#_TXD|;w?L%$Po9X0Sj^9Sm2
zv##0gBb`WL|4Um5)lO7$<`L{P!v865rPbBjfnEnFQ_`U4e03Ci1@yyBe2Vrtkk8hd
zzT$Z3f6+s{3`%^!2YJG>te`)vQCX5b1^qN4II`dX#B)|InZ^78hK}%fk5m)+y0f1L6I?raETpbt1cUnhmvZep)$odc&!i
zHYA$q_HqaOzmPfgbX6XV6J7Nu+|9mV-)9ouR*kl!$YHn7GB4Uu*XkCo!<*qA-=I6E
ziw^W-hI!X@&~HR-9iL%(Op_m9qme$26bGPnlx;i+@E?*Y6$V^!j?1o($h>Mnbp
zg8hYxRyi-IUzz5Sz4^eqr=MM$GKTz0vTwyT!Fcx3pH?meeRQ4VUz~ybm?I9|$9#R&
zGQaw97@tYD-#)zt|M!AV|FZ?T_^H)PPw>}bmc1_y{mMl&Jl3(HkU|b^8UVOe&O3PlS1LdE{OinzCQwx&V*W1_wO{{5>OMZUp{~
z(_e}IJ8zHcw=e;B*|NjwcptZPl`KNwe?a8uDRYp0J>|dS{nRUWT;(_ctDO(qU>-fA
zUrFi#WM_Xmcp30i$H@`_z>3GMz9hi+tVv#qfU=i{_^Y8lc2Gnz7C>J8&f(hvq#t7U
zyc(w?m|FS75KwyiXa9CUw!nz1cwbBQ_`4uHPn@UL-NES`;ddX#_Frd7e;n^eYsyU2
zNq}4yw9ii-@L@1{H=eI9Uu+}JsWXlD_UeGlZ*yfU)`!WeUJUaF!RZfsupQQlaUt$s
zt-4S$fc4ng=Fx}a7vDc-jOWSm(Kog@+}qe{_@Ei^t?TXVc)zHsIS4L>L46pU=)m)g
z$6^>So_EUar*yDf?rGkM_0Eq^_s03{8Kj?+YQ5f?fYV8l*$Eq_uz&BAZG+RdkVqwX
z%9Z8YtFXI>TN5!~H!~g~c)PUWh7Z`Chuqmq#6ORTD%6Lbq7ueh)?tE|XN0!HAMQYZ
zuXtkLO?xqp;2}~wN(f#gb|Q)3Eqy8)|6#q>grxk&>4^>snd5XO{L8s8Zoe5%!gyd^
zFTuO`o_s!q<)5-&2>xZ?#Y6BO&6{@!ULDhZ0+=j>H^v_-gL8c4GXx*vFnUOYQ*)Lj
z@^yW{#gF|>1$_wrTgDFvf1!G5LKdj$Bly;IQ8W;0xHoRkMbv{g#gg!!B|dNr{k))%
z2)FCgD>WRh^R^eE$77kBL_V9UHxlXUUx#9Vzd%g~!OJ>+nGosNm%+9)xJ%leAn;4k
zB;kJbrzsJiJA}TEycnTH_z#ZJ*B|{iflfiZ2~{p4{d}d9gdEK+MxZ4lJ-_idX+r+-
z5t4g<7@x`??J0zEJ~!y-lw5>ARA$t57>xl
zIN$=XUW_iQ>FAb2kH-$RhTX|n1f0PDShW&gpNlU4z-of~ORN@PHIyDBc||TTpv$m%
zO1S&3_`6{BA>21;ZH%wi(sAYh&KT^|GWLLee~P{?y1Jb%|Geh_8sTw9uZQ;Y^mzPi
zbiLJnC=l@?c0jX2xE+gh_aW%zSnk8s4_M%e6@htQnx5Yvtn#q@94}`8H{oR=U}HBO
zYg}M?338|&UFOEi8jyc(qGRuEItuNk>uZqaOpLEPHt#`E7|+3LAwxKy_NDXaYh|6p
zxYV1*A9e;FvymR(w1?iGa%Xb-D|Dz8%ne
zkNp*lt90(cf)Cp3^5eT$4!8G$52Qf;#eoca0rkYzKu?1D`BO%Wlk?*N#J<1A-3E8;
zZuG6?1;%?5?og31k6hIv!nvG}Ch~E*EkdlrKC=8I>ZM8OH{t(k&Vle(QIRL)CKd&v
zKAlg^ZNTXW&%$vV;C>nU`rRXD6%jsoM>#Q`gKf^kf*Q`>pM8UeRlp9vczg!0qgX3M
ziG6un!&j&9{)lhmqQnirkAekUxLmJe$IUUeUDxu*c6lK6f-jyY_T1|HuoWJKdr
z>#be%b=__`fp0kc@F3G_ybm+XbCLtcyQn1&Jp}SkG$YPovKf8p<=COK8i!YWFF@FD
z-GSdQkinil65z(@@0o0iF|2WB8HYQ&+-8aAl^Z*`E`El2P)4uYrwisUssFy@>DOJDg2ekLPi(JQi7ej)L^}dMJThe#^=X&ky1p_{C+IC)SE1O~7A?
znpt>0(5t;*jO%&dy|M{BFU=1c6=A)tL=I7K`aDlhuXh6c96I;J198-yZ9TxH6BgG)+WJj`h
z0~YRb-=7ZgyzgkHa|6C&WiH0=XIvSx%NhrnC0?Hq&%-VfYLfW>0S-B{k3WO_b?SDR
zDCF}Z^9EbX7HDscHYQD2AMkR*)-U9)!zKV{hj;VsV?
z&*FZ+`1w}><{?K_9X%I7&pT}PRlk7Wi~MO_w1o8uIlHXW5%}dZ^^$HQC~x7Z+OXY#
ze?rWqFkkiAm;PxS5ZyiF_W{zeZ=W!D4tkm28pz)b`6zu_ow^^e;zRJ@Ucmo6^$vsp
zGOL|~sRH&tq8iHR^~(0^Vf1?R5pP$+pZFmuovGx5r_?iCgBx_s+jA
zw^?Cc77#eiDF^uA41)vS$Njr7;}`+!l6{IzTLYk9^bF)ggzbQRg{?AC_YON;QnH5n{)j+xNFv3j^hN`wfN(CTfpP9
zy#KI#y4v>QXYk*6@ijd44dMg
zgzvrzfEH#CSPTJ~pQcu`0P2;NMTrA)1=OZvo&IaNUG>7juOfQ5N;X`+A^-x}@4_GICU;cvWuv&&w!3ELGry@?iV
zfBXGeiT|}cdtS1P?aF>5lVKeG$H6qhUbbmm{7)HVp~uIyaR11l^}!!8VclEBDueCP
zLiu-?*VaRgxb#iPHMFuEijKZfh4|7*cGwxde^XOs5;hJJF*
z$L_mRDUpPon-U7g=|39ZasaO{zW$XV_P=Atnfb84m}Ef%4uA9XMS@rBt+v{V-JSFH
zf5-Vu@YDaFITjyD%)ftR-Wp){v|LXMjAugby~pu8rs3~~fRtP6#C$ui?=6I7VWHT)
z*k2%Fn3!K9wE{G;d-zvXAPmsXN;io4aCEa2Jg5NnNGX}nWA#}!Gwfgd{1gwyujkE(
za9p+ThI*Jy?%pWHH=K@EsAkbi#i+u!oqZ
z3m=yevKW`TAP%=G&z+dB13$zQ{xdv6ggkGSN0hsS*%ycljI*3tqMY-G>Gvh(w`LOl
z|9Raf@Z(IUG7eWgFh=N`Z194}r}x}@!k_160)f<{e1!io{VpP(UqA3hIGhV@b|U~%Ig!{Jbu$>F>^@`}ceO&{H
zASZ{@c~!#{CWtVv)Y5roY(CweNfQ%r=wFNU__PE62t7X4(eq)w2O{`aPwpm=pZ6Ex
zKf(yj3+_i)>FLS%K(&LMaS&();Gzgr3*f;rs6N2$8|dlyHG&B8cTqZ?u%oAQY6}z_
zyXVr~2UpQCZH3pC+R@h`+_QA|yBc)#WT9h(HyvSf8h*gHadLnkusVRmvn_zzaq$45
zGZRSAagg~}r)~wWf
zFe6>objJ+^_UNb>fr6HJ!2gbM=6>gZ!nE
zZU_GSp_jAQI|CGjdtaLL@fRgaAI}Zv6^Zbg8S4oAoXJb*73JIj8Nqz-ew+d0zYVi@
zFiHoXhLHw*HsMqZp5KD={Gi}q|4h>eeLFA0-xEPDP<6eB@qIjB0LFBUHpm$4veHRH
z5B|C=difdTV8I9XG>-)7WAu6h(-YvT9cu~y<V1k`7-(iV(odhvjU{n!&H
zi10lhScrNM|29C_&#npK)7alQ`Tj{fzdbS7Scm6bEd_x=%pdO+pwD>zNj~@{2J@Ih
zXRZ?8CwiHCR~5&bH`gcDcc`K6AP(Ow{>%-pr+kha>cIZD&!`Du3>mSu#Q1B1%NWm(
z>UUe7;dEt>e));{RkC4s!mBiC08yeP%uCYl7fv9_Y{A^2-P*aI6;5I&@L
zC+1}W)9m{&&k5SR@Kp%#f!3ATa6nh`e*(B(W}b_IWrh27mVqxZ|54UaO2OqdFJvQXhUk6Rp@I9KbIUq-oppjmM;phRwfXCb!#!ysPhy;q
zeLpv|Psv`2I
z%YYADPV!(qcqXM`#{kB)yIS5Bz{cYO^#XvBch(5w@*JM>Ov3!>$XPu>9A5Z{*j(K1$#u{>PDet$|oqGW=2Bg!fmcG6d%Ex>5UV
z9r1qAh}(a~INt3HpSQaJXZlhTaKG>}VW`2p@T<6OAm&}!heREILb*~?+6q`f)~)f3
zWCQfO=yM#G(}*k99`nW#&#g8%{f{w453!zWCy#hwJi^vN!sWQ;{Lcl~qm|hXaom4d
zm}+6Bh5X$Yu48<@$+RDrXR*8_3=xfhJ#_41#{xB2+ytDo#!u{fpV2vt9_AZC-xLtnraOvRw@ijDvhV@eL4NrTS4AmE4_g4HS6~Z_v3!?
z?9Pk_*30w$1>&BIlmDNcIKSTxCr)BK6rH^Z>uXoRKY>}9E
z%zo)eVLdq4aufF0^|AMMoK9fmHJCtPpFwn#g7N(%!!VBbQazjz%kszEiTrqCwF$f3
zBJqqU2XzZ9$H3o$d5MVk@n-^Im(O_Fz|0T(pg-aX`x_MK#EHW>^&BQ+_nUm1bTDrD
z#10b^@C{FW!Y*zOx}uNe2$41*0I+U5_=HI3dR8c54{O&95a|cV;)hP)zRl(t!oR*8
z%qhr;+CK<9ECJWSLEh#j0}~tAM{X}-Uf8^?fNocV==1GA`#3@#T{=dT&#vwo5$<~c
z{r_w+=acB^_Chm4IuY87Frh-b*F^yFfO9^HKw_R*RXRubuRcMy+s_jhi1{zKzLL=6
z2?v-N2(K|^PN0w(Gl6GQwh;c4ade(ieFT<05H5kA&U*r5pqL=fOScm^)rJ`$j6*+|
zCIK@>=zLoVM+fXhg6aZ!yY?Ca7ZhQ+2{M&}2_S^~
z3dI4`hu1SO^6jGYt+3Pdc+z=L{op>@P3H|7I#698SMG!6$MOC^u>qanmIcNX7!?>_
zY0tTZd4Csq)*E|$$!(0Qd?
z1b{mXBEtWFGG@UzJnl6ZaVufOy$9p}!a*k|E%?0uT^1LjE953B{*Hh@|rmSz~^u(jPcwhx~$HOYZ!Q9D_y3(q5Fqb
z(&d;8y1W3dBja$(33T_gm3rKQ)dlyzIdtSF8V2?!WqS!dSF_Up?;QDa8JZBz6HK;&
zf>5ugZ(3l~(D?xw;Cl+V!eM?~bArIqF$E(2iF2w@FxUqqE*`?V<(vrN-@HSe!2Do5
z;(!mvIuha68wnC}tT`U&U{CJQW%~p4^aEVNiS!=vKydK4^QGesK6<^GgK1XHX
z8H^t_Z^J|Y_9vZwAGD!2jL0WG(25AZu-*cC0;GRa9U27AZzkRm{!<>D1hy(@(d*$&
zG9fFj85PChHl)cwPlJ8lv2;S;fV+c)zpNTQIfZ#zBbUI?88rpGk67}1*A#ZY{qtT7
z=A#?OM6Y39ms#eZg?ZTArqNV9@4LR&dy3KYbax;QANUPF=m6`1zcHhjk8Wnly@%b`
zl5L=a!ns1{c5aOR7k*yHs1aTZJqY-ZydPK);K?6Ez+%Hah<8_rdg}jmZUdH|FR^aI
zcp=6W8Wi}X_U~ci0m9tS5hG9L0RdY0jkx^ZVgQzYfgr4Emjn
za66dp_(|MXRBU?<2XnA~db;Ki&i^gr(I}jLUD2^|j0qFVix^oCIl5ug5>;M}k(W6G
zIwY)r_e>G_x~#)5ngNe|-2fvScw5EGs~Al^!YnaTmr
z;YVjDSK)H+>Hk60>-;4L1MJ>5tf-2y=98HQ#_noEqW=hMUpk3p*)PhcFpi$zP1Nf>
zcU2ZF@7A_v#pv_%HV_NgN0-rH$Cw?k8w@4f>nry(#CSreib%)cUArEZb+6x0#5n2R
zLexjSO%e=LSdWh+Tj2FV?w3Y*I2_VvVJ^qz*r+nikNx+Vq!!|M3oZx2(!f4|p_3D?
z=cC8t{4j3Hw*`X+{Xua72oms9cW2nqg7c*$6&S$KFMZJ0h>nkBFAD5ZA
zfHBm)fzb2mw?3jhN(#<^p@4fP`NxQUm9c|~7$2g!&S1G<9aa|#8>XPgDfnAGpiRde
z4Xkh1!{PI|o{x>(7QyIY8e)(2uz0Q8hVkPo^*9_pr^DnaMi01`k8%HvX`=rt-s+~?
z0jp>*K)`p`e&WNJsOCb9%l@7_M1RT&2_@v6;`pLC@GcHs!rlJa453fg$ppgQcu5~7
z?8Evqj6}W1m|Z38fYJj-Vtg2=l(R2+@C>XhHuo|W&Ym+wdX+5_L_SGP1BCys3i^DM(B(+TChd5lf^}dNbRWQv@HK3V_uq07
z*en6ZAt2LmH3DvKpocpvM%O#L3%Vb;2g%(bP<#>qWWOM&X26g5+7RHkm3c#3fnKhr
z^E(N5K{hBB_)AAXHh}x!Rux7&y&)oB^L1eEKsHLkOc3^G{y?E0G!V?(Dh8p
zq{sh8rq|yRoFZX=F)^qHz?@Uie1Ime=(6l>dib{?;0*58I9WiQ_4M?O`04)NSNuz1
zb%_1{ckd)Yf$kr3o9^BSy%76L;^G04bm_aR|wsI
z@)}*H!KfhOt*i&8p%-EK%Mv}`q3v{;pflhOn+1gb*%fqOiN}VU4gC97;&);ZWF}lq
zfLve|F!ti?0Uxf!w_1^{aCYFH3RVQ;5nSznx+~#KR??YWslONaI0gJ+F;5_D1`yb}
zk`Cbw{=wK8ka5MIe?`WvjNOe^uxbTi(~@u}iUa<*HE}qix&cqF_^(@WCyEd5gt~yZ
zm*McMB!8UdzLJZsTYp7dIpyfKZtJfNS6a|9({^VgjZVb)K#zSn1$e2*`*5$GO0=-l
z5tXF)1}&`}syX(h4ozO}Wc(}Aj@s()u$rES{fZ|GhAMM4=%QMf2y0|BGTEsau#wt?
z^z*Y4e?MqNP6uNClrgp=$@|OJ*WsPN=W%RvCXaiNW=6<#!v1dLM$Hiuh5NuajA|UK
zY}-)oO(C|&^?hhnjBcs>`w?_ws(esh>sxaA#Ui@U!KV~q1mRv5liz@Y?YhnAwn?pO%c_-@0DtmQ+-raW%80
ztff+T=PRC;;M$4~<#gprzpOz@9x|N|4!%Kknwp!x+0hW?SrAX9V=2l#@Z(@A{C`j)
z+&43tDHiEW8Qj?0nT;%IO`Ah@SD^1r-Lr30bK!r_dv&rhpT4#n*zibwx#%_T9?!QI
z@6%8N{EP0`+G2DuVD=72cQ#7o`^w@oPepMYXq0w15AE3}-pt*bf@*TCw&*;}Lhr=#
zjx6}(!FMUdV+Hq?BdLLGCuze1lu7!~xj2-DM03Q7lm$}IQT3(eZ_KHv_{|!Qr55=9
zM#p`V3B>~R8=O-I$LxoxM^vrv6>KuqJ0ETj^C$8$$oF53Kj{r7`C+2|$P(;>dJOf<`u
zf7seK1Nmuru2T)mL*$Sfhp!ChpqY(S-j>J=Xyf04WvT
zjc1bQNJvK1n$S0i70F2Z5?esmn>4iMlFWZn=Q7~E7T!{~Kkz+v<`+hiqFKmH@01=N
z3Es=elnv*f%R(NE;o|&S*=S9u>krlMnP`V;-QHcX*@%J9f9Hn2EYwx{QM52P1IZO_
z-H~)W1Kn7!5dQ3C2AcR>aM?#R9X*J2mW#_xM~*WiomMBaQ0vD$d&;hCG>;x!7z#*1
zWuec)8G=)hx=32pk$aiwg?>bDL3bt^JNL!mVooMHtUf#;X`F#(zsG3?YG)!od-jUE
zhq6$*Zfd_=Ko)wgqSz$#6zctQ)y`f@HZt8~bT~#Z8)dWcq$Fx(A*D_JpBQdsqBU>K
z+aLUa@7*siiaa}?fmFxb%C2Q)pf)CRHDAk2WVY5W)6YB;J>S0j{z3H&B(U+2qlj2G
zeCLsG|5e^>WFC8~_F`@pl1z}hKxWNC=d;&cYCW5QnrdW7K8G{WhXbv?p7%46K6&zH
z>u5SMdJ#1Liaisl9I?y2wlfFKy@;~@?wrGTB_g~A@&OncfQ#+myWx@AlgeJ*hzj*Ey_UC$*f7_nGCdc
zU~JpGa1I)_9Q`-}>AJS;I%()zgs!k3^SbLFg2Zyhx
zAf-c^#vYF8sHFT{fgaCwRO%alSI@;7F;;!hRG;%lYx~D^md`{Y*O92Jm&vhc{BF!R
z$CqN1%4&2+eRCOV?>P`Ay{ZBYi8~9_slGwi_|-xxxU1m(RI76-UrN#N>h$kvd&&_*
zlk~N{Gc+WB5;R(vfi7J9fe1Y;==p@La!N0a|_fbKT|6GL&}bb7EB)S*-t-OBo#
zjp*wZme^f7&B!ZZpGfqHH|Td@e#T^C6>?TpN;c`MMJYP8F0si*RJQ5ni)O|aq(*g7
zJG89=F_Dc-1v%a#&W1GQ6suZ9dR_ah^L7)uZp?Mwczr8+J@8-d%w#!A{Cim|YU3Mp
zv|(>-S8EMAE*;#lFyJc%Sv+q466T>eq22U@1u}X7oxtgK7Ekdh4dda
zG@dr>M#Z(4&h63YM$Jc_DElpRp~Q!$jX&+{LC&9yWkTNdp!7$op|zKKQ7*rFsh2}9
z8o&Bbe~(BnGR!b|)4i=9nfqMN7JAT+Hs1c?xU_BnWgP0sD?c!RYF-Lag*FW!p^wzU
z!m%Zl&O0s|`eDcFS=3ST*LPQlmF
z1LQTimS9~ZwAr%?z60l>7GY8W-yvtbpqX*`Zx@o`+V$jEGkhN;;=JjTHBg?kBNqo!
zVZE;yYv2ptXKwV~3Ev41tyZD_XVwN=}zHgw{#a44J5&_j-F}(xr}QGS9fhox`0$?Gj+ROjCD%j0tu#6LcUq|pMHt?n
ziiYlhpRqtR4u|jPMF-?HuxXbm3N~~P4FFiNwQD7&g&lJeY2PM+_4@s
zC#id>PO%$3a1|a>Yw1MX%rDf>{)O|{YpzP$I{MHJ&!6f3GJQyGo9M>w>>ea0n_fA$
z+=cv=x{ys{ZrSXeR%GHRr}AbUffeo7;nN#QX|*jA6avqV@+FV&(sqqWhlp6zH>G3Uu?zcwVc
z#&jt{r5Q!9+3;Cku^xFgKfAeTS%*|NNc|dxaVh^cu5qdg)&+A5L2F*MBK5x3{FXC~
zC~S+^&Ya3RwB|+KMH%SV?$1^yjc3Asq29xGD+K9&Q8!+)YeH$O68o;S)T2+wxxF4-
zu0^}bHyxZf@D7#hCQpZ3y+u|_Cso(Ku0Tgc9t8BmcV)`0qH5mi)}p`b+FJu^-yw;x
z&Zp9IZxFNSfO=eTC3=*qt`&ZhhE9yLI7w*Np{)^2Pb~J_x7(=z^-$SNK>S8s`smz5?@ulF4_Q
zfQ5QgFJP+eHd%xG%{d>(8^1x>+TX7(#+Ra|4d?YX!aF;(El=M1`_Yiu
zmzZpHtQ0Mhug(sr6e5wqqnqkQkVbr5s$U)~$7kV=4!26Yh)V(_@
zb5W|8y`E2R9+HzyfiLV7AmeR*m)b=NkdfrW3mTKTC|KQDb8B1|Rg1hpO*xdB9J54$0u>A{#F`Bu3A?WWimY3?mV4R_0f!eF@M?pg$Y!Dc>aA*O-NDc4~d$fp?soC5##xc4Q)LC(onqVmWB}Q%dOG@NCq_
ze#bTFSQgsHcj~05R5qL|Kb2|xla1o)(?ICMmPi=^gw
z0`iFo`|9Koj~1UP?zzO~i^SrqL}wENQH4+AjVV$9nhIMj(QO)mQr*t%+a(=`Lf7rA
z%063&qAK^T?JovCZn}PcXJ8r%B=
zj_b9;J6y#q6~geIMr%>~3l_#S)YSaBRjx1#X=ZQUd;Vk^^4N1(I!Gi3HLm5-HQ>)i
z%i~+_9&?9qvQLvniZYS#Hbe5#kqi_aLt;9#H4|y9)|as#%0%^l9a(rbXQN*-8lh5~
z(viCpEi}wM1Jy8`-NIU!fih#aH$;SFAU2DETMY=-Covvs_Lj*gdSAhDxLbyT`=E`SF_PS9b5UlfNb=AVL8h8RSG(Gp#0xHi71rG*&|5uPexNhPad7)DnNg3
zlNvVu2D|YrI9LLKYB3DzXUqeI&+^~G#UM7`NVAu3GS
z=$wBCH!r+?v3MY=R>d3Etu2#VT;6A*t+$L~&-i8|&Fx~Fo*ILm!ui)E4y7X0_%WXx
zoq_Hwn5Cak%R$v1E@QH*vJqD$Nr?x-yQ!>J+*Jzil}+5WIesJ)88OC91jNK5yW`P&
zB7P?zlU-wCzZQUB`(Be@9#2QxUsp+v{Z2t4x!<;X8)hM&Co=ZY$!X|K*GPT;Vj}vt
zRVY|pF%AvXdPv8Pryx!lX*4IDg|?=+eF|5}M88jr`+IO^qx3zABdfmTpogy_90IIz
z&{++US}DCOG@fzZqOm3ig_zc-KfaTVb}Vx|&EA!cXg}v=Hr`7^J!M|eug0>F178t;
zQDHjb5Uaja54@9istq~c%JuseorwZ`
zn!e1zy8rcMOTYQmR5af2ts&W7g&uCb_x|BlxR14Q=4GNm8L6>6K+yfaYLfjmmrIfd
zf=GX^CwBI3c}FsBEgmJywvhI5syiJr^dpf*FK++4>m7+vi8I$!Gl$f$%WJIiaWYBM
z@{~uOeJkne!*>t6irmV>Ryj29&Z;P9x!=*M=}<-DVNfom@>P+{?yl|pW=AD4B>uYC
zJ(@vkWOH}xPe>#sGsgvNbFC&ZI(!UG?QSLQ={ffHWLPXI-s%cO7PS2MadCfl1>MyS2rj8eM5VBK*t9&*%y-}
z#-kp?UAK0VQywJL9s2y4lvzK(5b9D>e!nTLgT-Nx6c%}1>1o2Z^4b<%x%APWq?0?t
zF8QZ^D<3dl_u-cOcM^H_NTYz^Sb2e_F}I%Z45{Ev?hZNS|
znk(Njd)CWx_#dgbJYg5to!)XT(=W*6iXypX<`L`mkNe3QThrn;Em|a_0A>sg;alrEG?t=%Z>h*Eq)=Y}zlI-}7D*H@M^t-AMHSE?s`A`Rcs!Q#>_E4T3SxIEy-#nMIR2MnP
zO|h2d-*=d7*HAE+WH
zmAdh4t-4J%U8lF>intH;&KvpRGrfW2@3)$3H^oL!|Lr}L`Mdft^}i}Em(!WgsOBfe
z-s${(LU!stvgY`U5GvKuUHrq&7_#SuEoc10BdHS^SFL5|a>=ONxKL_CI#s2`Jl`ia
zn;iC3dp}c>FLn46d*~+9LGq@?-*=t2_mYn~y3x*@>!O-m_U$Q}e^2J>eRsLn??38<
zrnbwX-xO6vJilDcyp!B3)9~~1OgD9>cF~~q$zk$?a~g+g#HOjuH?;MxvTmWAym`J+f8zVQ&RB>g2HPrQX>Kirb2oRLaRyDjRe4i~sxyqq{_l=wrow=Ebce+y53FY217S}X*Mq=hb!Yu;1_
z4Nf!BPJCmDMP>&n0nGR3>PbJz4;o%Ju4QGSxc|LA=p4gI+sNIuMep1m$}w7h>%CG#
ziV#oGpMS48DYsLLLl+W+Xurz2vfB5rp^4Q0cz)>G3Chz{2{9IFbxLYOWeN)?AB8h3
zz(R_&o^~-gW?RP_5@jO8%lNF1HZAL7o(sBVLi@N{W%{Rp6irscakKFH9Tb(0rKBsp
zTC@prqN3BM6O^24eb3ICP(KnE`pL>%XE|h$eEP?}@ayA6M$n
z4%ks-Tm;W$R9vNQa_Kf@ZIyd=psFv8;YMo1W|?4Gh<0CkY@;o$e~Z&Tfm!?fpDWh}Cv>rpaKd$GDA9}3r>Qhw!tJet<3P0w-}
z;%P5@85jy@@@SJC-FyCPE~k9nYQoa6P(|}pA^-HB$)?@o*!E*7w~Lbe&}x7&?FH>o
zTlM7Zy}Pu%8+R=DTvwzVGTXVtbJB>ST#+03(^ZYK@%5Gu{{J3OazuAj5Affkblr-!
zQ?0P4>XD{ZDE?V*f!AjRX^fuWYT$CUTA^E=37U&_K6
zQQgZ8U6i`*;g3{2+9=J8)~~9!Z=os0M}S2wVP_
z#bdMNY
z3O9%Q)^ls4DgM{i<^J36LMxJa(jypPONj|#wP(-tqgxE=_e^6t6$0v6SDN-&U9)IttBTP|s2%TLm$OW$xgpXZs
z?ijeuzmiY$3dV%oQm17w?R)uefrFO9y^t6ougJi{N%DW^9Qy0
zP;|?8?^Bc#?~YyCzm#bjm1{)aF)7k4bUO@d?}}13jg%Oy)k`DGOtTf44*OFc{!NXF
zw7f#8I%a((ywZeL`=s_4rB$4^NtyXBgP|Oyq9-mp+WIF|X-9xtTvsyHOo#hLf_5xf
zb8z*v(&J5PI)X(c=7
zVtj8<8QI6~4tyM?uBlONYB}3MHa~h^{^DLEDrsG#f~VAFYS7!*S!**t^0&XrH_taL
zQY|lS-f|>yj{0+><153y26EzQiN9Ox3#eKPb5i`G0gsdADUy1}KyF-?iQS!*`%
z@e59Ju)~<`;44I3j=xBK-rz$$-+Sh;m!$=H{kk7Hs$*KzIb}70b#{lyXD^(7v@K4a
z{O(%9?tAk-r&RMlaF*Bs=&-i_whw{0M2jN155|JY1@
zH|os&H)cnO)-!!(NaiDxeSCZrf?sW=rU%M4*E4@0-F>E2be?4^`FZ4<69zgCRJ*56AKqk~
zr5bVh{$5^GCMV7L@O1ObP#;fec;7eVp!(CiFK{-5ksX*w4|8*_Qjeu_pA8fhCbx5K
z+VtePCfQA1BsC$@l**pgeri2C4>^C6-1^V221#*mk8<8s>MOtD(%YwbtCu9QvD5#0
zaUMxRWTK@+k&W6lQSCYZXNVLoBaV1@<4N@emrgLQF(Xat%XK#ob&-Dhe`QwL#YvWT
z8Gc6kKZdUR9m_5N%a#-}%E}6%%w)XJd6exdvZZ8YWsi(V6f(*RMcI2sM#lS`$IMEZ
z*+Nz%D_P;|KR7>}>pJ(jr{ppb54ab?a=eT1DVN9g3;arO?S3rTca8sl_d*=)ug`;l+NZD3FG0AiasAAQgwi7
z1kfOdZ`21WB@?rxkad`I{Zl^>{STtkpP
z%kufHNErdqDPi{;7$(pU9*Z()XdzJ8WbO)%HWA`>X@9MpX(T96&KT!keNX7p&SEYR
z$st;KQ4385W)a3En61PwmJ{!b$QWcRd?YS%`x<|HR!%ey=dl@znkK|Qw9nJjizD_j
z?7pcQ{XpPiuHw$DttGs^9262B`<9@rGn&_^eHIB|0ifVwD^acQPVhe+fGFOBoj9!R
zBkAjhz;@&jD)5?6aLmPJzacPLDVgFO*;Lfh;85@`@erjF(1ovA>pidKui%uCXwKYHVd46MHpLokhpoH
znz(pxkfG70f-p&*uKfENB1?(;Nm}M;E&Zmp@{OJvBvyw@-Z~Ph8z1
z>Xt6#K5r~1=G7TJcB;uDXl`y)zn!olQ14k@;%)6D1l{cZFXcTC(j-O|$GY&9nDa#o
zCLVPX#4i@Sy8Zhz(K)UrYU=$I;j>8C^g{G`Z1dDm%E8lV!q%#J!lS|#B5x9`pKqEX
zC>VGOiHH)HiUE-F0)~cw?rS%Z;K#(1LPiITJ9L*Q|czwX<=ks
zHgw%END#ZF@LDRdSseNBvryr5f(n-F>SXrS$rhvC^-NKmzl^0No-epDc?mHv3ThH3
zNMSX8X`I|w5G0?{HN8e?b&>{H9+_!^RG%=y(Pl5!y~u?H3nK{y)~4745BAdDU&a{u>-nBD51wMM
zQk~7v^C5D(;A%A_55iide@u=kzd*=8c5G8?7GXyFDR)))JuW50HBfbhf?v
zPqD?s6t3*KEaauUaHw*14WhkiLC&cijzyDcOqGp$Ass=_+^@O1VvUr&zK_sh~drEHqP)##NpW&5pt1Y
z>`fXIS*~v;Lbi}cf|q%psbgh9->@kwHuKci#p_mM<1U2mc)MWEVK?_ExI?rwthDGNV!c
z6AgzyoaCMseFZq3N;{NpQ9Nh@c1SIEdYB3m*;
z9lT|PIR~z#z;BvsS+gyrK;5oC{di{#7!L0eQ+x9ry5%JlX>N7FYNcffUAG2Ukshn{
z?D-df&(Bu4WG+Knltqc6dLQWLYOw4Llb{`niTfopYoJzj?Je)I0~prTBuTsc2b8Q`
z(3w2&Kc0?pQ1XQ0)|_%LU^OZEOXXPrhD;R}@|~1H-DB%)%rzF~
zD%(sL@0USir_hU!bsAt{C*@9VL^V7Yy2&j)^B(s8BH!z@%ZBP~Yx*^p`yj>fqc=`G
zdGI*-uo2WJ0$$UXo;1ddpxGq$z~gK=kfoXUp!tv#rCEQ`Yg6l4N?k8=&x~txfTM0#n;t^*)@B1PVw_V_iuulw^J4d;C%o
zobi48qASx6SVo*3GaS4PQ!Ipw$3=5ZT|}JPW1L!e_|Av&E9P0(2j!KZDl*X
zuL8gYhSX6)Tr4P87-4NMu!i|&qvda0Ay8v~doRl74d9@-K$Q~v7Lr}i+Z=ik56eD|
z{C!uy3r|@|CC+UQ!}Zf5eKpDbpvW`-qF7WrbpGcmGi~aDp`_Px=e?qVSdsaJbw&{Q
zUgvn--$)O7whXVFY`6%;I!L~J36}!bt{UBPcoqX6RtTqy-L--Gmxk?Q?QDUQxR=GN
zS^`8?rdY+2_UV^b&jn-3c3vcNtHcthkF8n6{@j1prw$>NnMc#Fnd1|
zO}_aChMwMh&HOSEunbSgsSG{?rzZ%{D#<+|`3OF@Ztx$V_uS51PVI#fHJa)31qE=R
zRgtJeMj+y;Q2F@uXjoLJ8`>rA{rF*z+<cQNjuqJ>$^TK*%)WRN8K<0?qshwFe7Ip#S23Crjfl
zLFdO@IivdXm=Sx^`G70a*oTzu%lp?7v0#?FDj{9bh?
zdB9PN@IuqL;~_*Jdj*zjb2sy0T@ljjS^h{b(iQty_Ip~$mKy3+Wh}oe$j3}q-%snQcOVBN98u!O
zyRn(ZN_yqoHss>5=%h-egu`KzaBVg`A#see?hu=iuLIwTACkkkSXWU1#2c!biU=qvhQEsBuz9&9wi#4O|pk$jVE8-T(*wI@anBup6~7m>AJvm
z;)oF@K-ud|dfO873>Aq>J?chApZj($*k&Pz{&V$z?<8QK^kP1<-HJd8Z|CTQjNHRg
zyvokowNznL_sBd9Pary$;y1J1N#w
zuKZXz+iHVXfB@1GqV_{Ag9l5n;l`_LIgoo2Obfcc=17-}oJn1b3|6>h!kRTbP1I5T
zd@6>S4(T#YBicA|V0xBmGJdAFF&56z>TDWKgh{qlWGRIe`#{|%pHu&fI7GFD8C~N?
z_QvFAKau!hOQ-99)7IHyuj|Pr%%ra)eGSz$^4oG)jEc)Sfog6nx_G_hyJr@{9CJ6d
zkTMJtZZC@6JFSVNneY7+kkdd`_|h0BbBm~eddY4sv=FH7NEq4~4iK7c9y}`19U$_IWYZiIVnFWs
zjryq{KZm#+KSi6fzDrau`jP$9{X0Ry(=}0?dWBFd{>R9C-wOLeOSpIOp%k(gQ>?}m
z{FlIVxp+H^|0JgN@aUYNB`cC46$@TB`u!M#!
zV{J|@B)2cG{+0*>mRoS0(LRhB2}?A;X*(#0y*k`BE&t|*IPfXiyqi;tLI0_&Y>A5L*)hzS2JCh`piQ&aYDj*p4mJ5#47K5K
z5s2CGh5H!^Rw4u6Dtw(Ureh&HJchxilMw1%=N;b{nV7@{qKj}%z2*ZXCd9M%ICBUl=Mv|FsDJWIkQRU{kjTmq}
zO7CeBgSR*H-kwYm26TV+kZ)q*aBuEK6?y+(L{utO#MX`t(2x85{Jg41X9>}rrd{(9+$&iR6W4{8xysdZf7`1czDu?*#p$F=PYG5pLFIhGUt)KgS0-FK5B(Fe>9=yV7t5K_jK?zB$IOFsXuf(rOX1w+(5PJ
zVf2BLI~1(x^>on+0yDIw!AAY5kS|?B)X~l#P)GmOaA*jI#E?4KC-QMXtFK^2;X)QH
z$RIs&yc|Vf!S-)v
z3m|#kE?IwkB&hK@jBs}8g=sYQi_zIj;Nt?r1r_2i@GXgk`-jyBAVU61sdpMe)1_Tf
zhoy1Q&-{&OEH?;aO`UEjdT+uMy9iI$oeCIoR)?umxC0~`E0#Xw{|=wDNmQv0yn_WS
zZvs~o_CU(XI9u5H70O*Gdpr7L5Z1R3aRn-m0Zzi35SMytRB`$2pU%g}@UdXqd>Wht
zm6V%k4@@J)?O$AYC9c7X?sqhcU%JJDM;~2q(7ACLUmS6)8~Sep7_mmlG`%Io@5P-G
zm^YwA0qGrQmqc2$q^Qhiy#EyLU_0qb$Mp%m{S<(_Px%af%P35Be)Htym;`;
z8@^K~RN3+Lr6x-Kd*9%sulB>IO8dauK}P7&#dEll&TYnAS^>1BV{zrkP7Kxi^jGgQ
zDHYm(ERE=yB#3ixy6u;aDWIB!dH2LMezb%Pv*k-B;Ia-Bk7Vys;&e>4#coL~_^Td{
z)SLWzsG~vQ3{W=56Oc2Poi!}*Ss?qB>Nqtn`{{Z;dC4*Siq?bjtZfQhF!aO{_s}Ms
z@o(wB$1@E|^$e@%PI2RrbS|n-hZylG&2Z)0%WI(Yqn6i`wQbPwR>$!&-w6CYpj2ZS
zCV@V=ZTVG5T?l<%OezGY=D}gkr04UyYtUBM*I7Sp4^;UX__Zr5;#XSCPpPsIaK-aI
zEOYod+=HiI>Rt#tPJQDFS-afktlfVn%^w3PsC&2Ba>?lo~AB?j3Rxw4^2hP6R=hfy(f>zz6
zAH+o8!LAr@NyY~e;0F>(GP+?1bT8q)^h`aF`xq5NX=Nk$`Rvf|Wm6T%Azv%}Fp>fv
z|HzoXkZcc@sP~^9v*&4AIjrGuRp}S0>wgv!!7uhTeDKIskpUWzM
z6iuNNWEyl_hL!OOX}X#}VaK(YjZU5_AiwSE|Mmg~&MPN8PH%ezDTWrtcz&3}X~LCc
z26I)gx|xwr|Bw%cyB#Z!_;Q5YXFaH^`=JG7>xqh&BCo;rYXioDY?4q|+Imk$h!*sJ
zYdAUVA_$TT1WAOYTmfy#?>mC=ra-c~vUNL808n=+44q%v#nv8id+IyN!Ls2FNL#&LHCxyoGiePc!yP|yt%U9A!Ll%$q_Bm^<+Be4scTs|UAC$z@d26yWvqmJ#XPSAjRJ
z=K2{685lt5Y18vs#2B6}p4e&_Ms5auohB*G!E(wJz6&2FVM0}%s~5*>k@epH6jwYK
zv6Z!T426@Ua%B9#J;rgxmMA|BJq^J)%wzE
zkW;f+_qf?Yv7^e>-0O+v*m<6e^?IuN2#=bMa+HUazdo^$J4nO$@Y)0AE-01_s8R;%jkoUt5t1PzsK#d
z$1UZ9Iy$ykx!>Rwk)tuh_CPuKDs2&RP`?wMaV`~89jx#px%Lv#-DSzW1pTq&w_3@q
zPt74D!>J#&mA@gkyUvj`A~<$R=S<*F#(cz4=-B;>jhP6e&DJ{7;fcBL^CrILVMW#z
zcircW&LgU!5)?dY{Fr+#10~Ub7YX8pZ_9cuk+U9kFDtv1u@AOXyFP#dQO*kfPgR2q
zF(P4fGSTJ5CLfv}9Gm+O9@q})nwXThscpnBmN&d-X9S^I_E!8Ryc;N@cvnI
z^dduA&Ks5|?3rTMw~Se)>#_((4?zxp@j8+F`r9kn|B)fLJsgANYc67C>@3$43=NUg
z+joxd-TyU_0-#&4?gwbyzz6bu;Xpmgm{dz*7j4m)#?;
zCWL)qC1*@MC50GW$-DW}OBahVr>&70zKlHhWY(v+4iL2&mg0r88dz29M;~!nePjtG
zv8=q|g%SU-8d}vxAr?_?$7tCNkyYDxeMU_-?B^0|bL-7BSh~xAOJ0s1ww9}k?woH!
zq*ZGf9m;u-fgGL5olzk~`tej{%u7{Fa^hLP^^1pyKlQ=c2-$boF}^CAU}{V3ul4tx
zC-0>ZxzbcDtVI(szY!fEmKBa|-89zwb;BH6Y~y-Ptq_RpYA!1Ma2G>_7MMC61Rh}7
zk~%O)G!L0Nerv_r;wv)Ac=O)CiC}~?#(H;%HV_%ntbk5G5;5n4Bdtr!tw`lX!AzeU
zrN~RWYTf<&{fPQGnM<3?-5By3sV}MSM2V>
z(g6)OB-irX{@X9%h~<=;D7H>zJU3k+|k-Zh$mNUp}AZ2G%^+BZt$8h!%h3->K^(7z@?>zJ5*{
zLiUB)QPh?boD6BE4ZLj$3nT`7Y&p5%K~U%k1?v+)5~s}yvERo&a&dO^zLSCMdNP}2
z3hKbH$*_IUvy1#w3LrdXxEken8*FznJ;ND-K|skl8;Y~~;AFq=
zJE0W?cuB0(s``Kz1j~n6>oqxn3c3TsWP%r*G5=ZkB*Pz6&6}4tEBQc}_gF^kT>!9C
zVo^1@@dW0s`KeB3CP6;Zwli;@nZPP?rMgKCeGuDYVHf-8I^+l~R4Kgw2*w8g{?AG6
z4IuOO`j0)|1&aMF4nk}l0m;)>4=TreV7AF6s)xx*KzihBL%;Sr`2C)WT0M6f;QQkL
zZTQMt`1+MV2Hoduz#m4_JY`l56<_gWb+
zA@l#S#n}%{z9*I#TMh#M6Z@BbTU3LL`1|iRLz3XCAQZAJ=D|wRVg%B0(j@Ymx{*Kz(?F6X6TJCfE?*n!Iqyuo5}chFP^*v4llVF
zMuo~D(sEZ_@Iot)gx`lt{u_alTVmWe1qqFW?m-YiFY{YqFdjO^O?dkWeTFN3
z8#}yOWiVDCZmf-g9bLY|6;ykLcz`UtJ%$7v)g>qcf+@Rdk&zq1Plu40TXLXoQzF@U?e@S9u)vICUl6Qm^+NG>Y9zN&GJZ-pb3r
zs(FkXmHti1O@8qvddu~JBZt5pyrRRlBrwntZ;`z8iI0N`&shGcvmdO6eyQP2{`kTI
zuYNbHI!VyP-!=-?N?y5(!r9xH%o%R9SG3Py{)-`gc3ZKjJHZOwpFGBXXzhwRelMe<
z?j_(KN=mBs6FAT_-N!#xs_CH)@1L|eNbbX->Vgx&Ca>{|7omcdNCNIg-##<^nH(=~
zAN`<18ixL?CO!?NPs3xhsWv1F%5V$qZ193M2mfkJO;Pjg-#HT|$N4|p!b2eDS8+`h
zExst}VplGWmR(G%(7(%tlSyxzUbQqwBO)Z*A2pkxmxT*|Wc;8;iG(@*uwEHdfpT(@
zsZIm^b?QRMVxAIC=V@gba6MV`_kRFj_S7ok_rx
z8`@Oz-p1l
z<gfaYjd4pPDF(Q4
z9!+paoE)w*KPz+V(nZvT{|(>Wv`G+I+ozH1auPST$miyNW^t^YmT>v##yY(zhGZ+GR5eV^iYR~27x
zj1De7u8MBAIXmUKLbZF5ZSe?d&Oyb^I=AFcsD%QPN
ze|&2%Dd};kwinlGGtzJ6UCGiB*KgUP@L0WjQ%Kq&@^BLsQTveR<
z8w2*=TbFiKJ%*>VIzwA30wDuC-N%9pWk82hL_U4upT3fyk(f=D3^VgKoD-YlKpNE?
z-wE^)uugask`hG(^ODEL2flv>KWVzdati%moFU0miQybb(&~QSCO-qHxK$nwyU^i}
zeG+2)__Cq*2Z4zEh%xZCK45XtVgR0v^3oltj{>V}TWoX>A^~yKi1xeeKizisvQJCJ
zO}MhNSN~n|Ca4#FuyJ5`6_(fxUj1@92);m!1#Sg8fRU$hsTooXKn#C(@hgoAjQyLx
zbN1E^@Dl%;uw&*7+H2L0y(T__s}l9AS<^b;ff1=!Fai$uKfb&3s1%$ydoHdq`2!#*O-g@B
zh=a3i0rW=wG{D!#>Y6Wo6iBMeHYwh}0dGxbsQ1$eLspgWVOs}Q&^fZ%q1ZqJPnb;c
z{e10*xU2y*pgV>wmHAsD(b~qT6s&
z&idTcTMn?uYo!+RfCQ|l7fmFutG%F65!AsXZM`D<
zIdy0|o?`X6;5XtFx=fBg{DD0F8Jt=^-hq)@NvamEVdN>`64*qgVZ#Y)2!1rVb7%>DO?wfMeY`bhj;v81uBYys`tOoAT$QbseIJK
znDL2Il_``xNYo5NC#3*}DUx5QcsL>eg(7to%HC}w1Cj+gYreh67d_MjwI0EgPv32@
z`#FW!K3(y?a90b8c%S@l_S9(@y~`I|z(5C%pGlOzf1Dh;T4!`LF>WD`5_GQzu=^vg
z=GAU3{dtQe85MYHZFwLc+#N1mKc|oJ0Y$T~X^vQlu}^jP{(I!l+9jW@*i__jsi&at0Wr0-Y?`sH)bzt0WLMvbXWMNox(a-_^3*_@6tA8>@F&GmnDYd$Z6kn%{t0YJxisB@qKl^nso*
zlLvPZvR1Z>GPyxmV0oR&6SpUb8_r35Wcnq52+!OiZ
zeYKk^J_u7SOVfBon~t;_tbaGeS-;})Um+iJ42HWCi!c|bTgo~i
z0|-O>>0Rod4VZ3_t=%onT*S|{eS^Zj9J@gl^=#Y@|Cf6puXpE}#>%E>kDoi&gm|&q
zCMH;vARil|=ti-8q$*3q%DUs<;~IH>egE2b#4W6MU~+{O)U3N59AlqAs7$#W$>Mvk
z%bq8T#3}oc8%i%dq|G4~|5&wx;}j{FTMzG*;rWfkV$%|CXE!k2VlvsZ**#>h%+_c|
z%`=mKjdnSU}cwvI}F8J?zan7aA;s29TtUW;)VJLMv0}MgL-5U@(4H
z&quTmqkG`TE%b&2C|)%?CuYL{H~wU=x`$l`S5mjF1DLtNkCQW9%QyJ{`8c%SXmFf|
z+wiLxY!?S+yoED4mjA&gQ+Gxvb}xfBS2az=<)uMj%~>6X_ZOgvY@gL9LrEaYWow}4
zrU$Kiw4RRUgn|5JqvnY_EWoGSN3~8@3@kbQ82j~55!UfXTzFz>3hIORbAK=;0=jK8
z?$P>x`h4+AyT29NNZQ_yi~=`JNG{WU@Z@&@U?DGBxkMcayKfxHCnc$YZ>enTN3(a}
z?n3L=zeFc+d)I}F-aZy43Dp~iSQx{hNi%1j0C%w29rBgy-{@$k)u%G&Wjh#5cby}$
zFAh93zSqkf&;r=a1bDvk#sY=VxAXD3!Eo{7N|nQK2w34_FF=BvL8JX2k>HlsP}^=`
z?ILM1cq34BX-*&&KKyOQmiR6YbWxv@{&g8bv7=n`ZKpXXm||fYEa?ZY+@!v%L74=L
zE!^m+LdpO|PI|+y@vks|eY0rYu?LnGP>~lLWI`)lBhbuT1sqGq8IN6Qfo;FC1x{su
z1Cef8=Olb*0N-l?+cQT!;DV#q#Nci(bTVChI~6$wrd5?WvPpiz*9k$ZuFp1Lo1m1O
zW@R_HX8gy8;NA&WGF1qbUL)YI5bfTJ19E&VB$779f)<@h_$4B%^$Yqv^$mQ%vJNJk
z%KJ`H?7|-+vsy+k*Fan9mEze`f~chdV|)qYZ01=#SpIhDL;TwS5=d0DnKF-_cwp#naCAZPYC?qIUz2YxP^O<30NiaxThB;-<4I
zS6NTepo#i~Z;MZ6!Kl&S`=b(vkY7(vIA&A?k2}1(roL;4^0(cv`S6V%Cw=8HNL40?
zzTFHY{lX)U9|qKnGygG0dza$Gj3Ted3*j2dIZWTK1bZs095VFF!8MyN$+$a{~O81$~HXx;ZqGH8m({d`N>
ztEgLO6wMz=33UF(R}-2tUz8ZxM9DXhgUhV@3uPf~ZOhGj`&40>ll$DJ*
z)2X$)T#4TQcD;h8s3z{C0(p(ax(+vR+Pti?zmZzF7Xffqh+jpwPZzk%g{GiA4`#W}
zizc9ctDQ1aHCCv3n(#xL5DUDPiz#%Geh*Lvw>_1dsYIopOKQ-NmZ7|#7=PWGOF*l~
zH~a@&3~)~@NhzYi{$C6*yZF8}jz-8tc2Zhn|8^j12^;_Vj=FYKWw#xC(BwxUAFl_w
z<0%CJ!Jjr;Q2L^eo%mWK8fu?67WU{p8pclbCuS=KjeeZ3CV^Yv`VCW^6WVg<$@1tb
zsTabi(8}oSjsgezIA`!_j`(kw!g2V;%8nVY>I)trp5R7JPO!#oZ1baRbL5dmlZ-g+
zPUU*81_S!Oo5S61@iZ#Q^_;Ov)d^pgdbgs?cOLDCK09Cd=<>h4D)U0mQhB`Y-Tkmd
zb}}?DuUm0erlQ6@*l2h>s>eUp2%lZ5&|)}b#@B1ezMO3D0mo0=elGWW
z5~^P5}(PSE2SDa0%~u6^M7;q-a5mk6fcH8;1Z^})c~@$aQ4
z2EkC!^UY<4MstX+j2L=NJ7b`N9eesAuOA?x
zJfmk{?*P|mw)o8oPePOZ*@Ejd)u24LRdCur4U#NbiZq!?!w*?RpHDlE;KajIK8tb>
z;WcMVgX>R6KyY@(z&qVkuy=;HCnhx+zO%cRa;nXaXbm|@-$7Q`*Vn`j?7=Mx_OBBndZZGAfnGW^JT68G*bRdy*)?=Ry5RKG4nD2
z6Ysp`ZGs;RskkuoLWv7>_`Y)|eI)@*1uo6Ttz3lvJ-73S)B4x>x}AO$MjZ<8JCD*?
zGuXl{g*Rl~_pgJOKiwYR8J36tMX7GJ-`mE5&U~IKFgONSdAL}t&6|w^y%1p;X0SooG0OPi*vQFWpSj(Bg~e*0Y~0^r7<(!!Z6=vbury@
zwaAOU%NB-%hX@oKo2WV5#%{Y+#ANqFI^YV;)Ge#TR|MlMk`Ihf5!|oo^#^u}W__
z%6}o-p|3^wVGy#HuCOF+^9sA}(`C3Slz}*BE5CgzQiExUpUW)S`-GW2Vr7X7eU0#q
zvn_d>dtf({cV#^$!;v%9H4R=-&DgcCNi5Xilz^{q(5aWc5?K`!{JdLJg_J0Zu$^^l
z!G1SiAh+q7MaFq-%zO5jfxA=cjS)o*IoM9I6BQ&PZromz$2ohjCrc!S*Xg$rKtqza
z@$bGwO?1peP;Vm&n^m2KpEfYF-^(>f_AWBJ>(ylS;5e-B($z|MFAf=d>Z>WsW|2gd
z>X@|sQiLI2=vukZ5Vlke$3|>L~
zxOsL)w9~d1yUFp*GGhEH^bJ$I1yL@LFU@+>=7O#{AEU9qYHt9{
zkCaQk9kYYJHOuu;n)jgLl8OVVkOBBUzL0QoMFq|~kYr`N)&mw*Pp$4WJcho`cE5wT
z%D`rr*y$DyOK|@nE&86V5s2h*|DCGr0*|*_GxvK$f#Z6xpf{@lh$Trdc%{1n35$4S
z0=o?unI#qRN4}8u1#w1DHXFzkUWuSn%>y=|nXfqIEhv(3?%l_dAx*&RD$a!rP}j!a
zQ9hXm=f5hr@bk_=A*C(DQ~&sg&PutJpp`3NV@F7qkn90t>a7$Lin1Vc&im%a*}XtK
z+p~PJK(Ue|pG%RZLgpJ8IP~7Nw=p;je
z|Ev}Hp@gHS-f~(TIsNMk98HxYs-kE3q~saZE~6?|q|=VK
zRdA+?#-&Jkee^TC{8(SMBR(QO{$GG~2%g%%J#Z^i9L<;#WL=lJif&?Mcu@RJoHvs8
z^l>1FekA=H*ID-(ZB-L!W9^VeY3mG*rTWOC_g|04EdDgb6Y`tR$lnb>w^c~<+8(Ck
z(s}gPxR~wGJ{^%8KbYL``HlD#bJQKBA^R56)nJE5`OAqZxisRJRy--Ep4y?YUD?w<
zp)q*&jo1MG(HOL$Q&9Y#N-fSF9)HqetpV3j5zNg~H^V0aA2mz-)j%75Yw?KoMWZuz
zv2r{#P53!C4}8oB$CC-)ujq(Iqj=%wu}HGf4!ee%KHWJbMk$g
zk4GEvN7jnUq|qO6V_k(z%JN41nPu14s%#>TQ-`uNsJ5UDmmC}d`Wn%wnF-bx_U}=i
zvm+ej;;|@0qoC46>}`DFt;WOZmu+ZqcheH_Q#1PNOt$^D`3E#C(Mut5JQMY|{`QOE
z+jDgJG%K%z;YT#6a^}u|RsZ@{t&!JqZ5S$;PC~{XSby?ps`6lsFVGDX;
zCqh`(+j^RgW4xJ*<`vuSN40#lMEGHKPu7e>Nx@8_?im@q7Z@
z5H%m3k9M2J&^k6H_)D3Hx)lk$3NWz5rQ%Iu2VK2zp^JU3-D}z?x0gP1wA4*>|GxiU
z($ji)b9#>CSD7f(?7v?AV-=S8nI9^*e=S<$9y?|i4ZoSA6r|5vozm1$#gnFa8|O<=
zNeTM+w)iA;`O?b%cNHg`&)<=^Em045Q&+k#chMRR_I8OpEnSNW)P=E9zbHfbJJFvR
zdrwhCv*fk5-3NI1kx;nH{54$V&M?07r5-&@)8j9bs6-zO6rH!i>{p^^jPcy?*bvVvSMadmD?RuA6GBxe-Gr_KCVX~}YB$7{96x#I!{fOE
zMRZfgr`%#f8UJhC{j1n?4xSLy5mL@(Lk%Ysp2>H!;PU>fAlSzqWhL2W@}yKlxyIL<
zd^!oZuy^7IM^|>V=o~S-hn@#LJYB~tRUm{%-?E6jTxE(@FTq
zf_Rh0W@g~~
zeJJ(GFZh6Q2L#=pwpc1IhZR#ruWHg7pg^`=xDT4HHOkXxiK9_Bn9gi`W<{}^%G^qB$MB_;XPmbZXTYbbT|f8!Mew=o_08y;
z{J2k)chl!DJm^-*DUJR!xxmLUZ_3mA51b_n-cI?o2b!eJaX+a(;IjFLLp#O{3b{B1
zAYM`McB#9-z=JSQwd>WX`ZEaTUo8JZcR3$g!#r8y`ErnEq?n%iFZV(D=&|E>vE-5qP~2D9+|ZOT8$7eq9e1pY)KRx=%ET%@N%o=f>o{
z+X8JcNy>2O>Uuqplwd71EsBGyzcxpLqC;UHJ8Ak;XJhCyb2{9XQw_wFQQYePs{qqq
z#PZ6~3jq1ALd2!5V6a5`oK63$BDl1*bi15E4VG#x)fJHf;8{nL5x^+`6W<@r+Vq6O
zqUXr`%~KD+X!dhS(JT&-ZT&Xm`m!oa^HCfNq0j{#{0>IXt^jwC^K_AqdHfEWqUq?=Mq#KAY
z(nJ4dc?XHA<6wC7bj|qhEvz~s&c*HNZJ^TKnL0(U49*n#z|Vn#@Da_un3N7C;P}_%
z@!raD@J{``vsAGz6nwKB*Ti!j*p+SE8i;0r>XKoV50PDLi}SOwwKXp&lb~#|Jj9R{
z(sWHL*LnoJq&cs*@fox2o#a>d-a`Jm9aPqY9|L&Arx82H2ITU6XqI`x4HHgc6{Aqe
zz?R?E@?RYqLbw}0ol0sZ0pePhZ!6ifBLf!UJ;F>?m~y}OMvi1Da$k#c>aO4RV|nFL}oaed-XeiAq#+msUs$r_~6LFofEac
zCBPx`9chCn`tZpesT*wU_kaU~ioDPtCZKXt`jFO-1!gsn5(b~~0o5HEnNwGlAp>^l
zMS7eyP%gG;{oN!4X4cO4$4K$PTMw->?>Jrpp4p~FVVpN$cW>mV7r__A>szW^D3t(r
zER!U8kFmg5mL3^NCK9S?)hb7
zB`8C(LKQt^1;)=LculOl04+k|wzMyuK*~`36m#lrP(rDc(1H_}-lA<4Wp^Kcpoqrn{&iA;VU#PVP_-htk
zhj;6s-4{CcIN>%RL(Jk+5MO}pbXcPH-95nc=;(UCT0fAO31k=U{|x_1`BGzpL!h1I
z37e_bI`olI^pfbOM^_>QNa&jyKy+8It<=RBkUDhkM0SiD>~~0f0G)nAyPcyWo9~>c
z*@j(pE6EZ}>Kp#9CAJQdzLTG9Vc&-!%579um=cZmc&MHKR{|Yty(&uc
z$Epk9xLG1AGi6IgQL=f?F(S#z-g|FCR@pR!>=j;QWJb2<94jOdGD@UWl$DYw>Ge0<
z&%NiK-}wH90@SUihxjpD=Y9Pa8XmY>bcwRe%Lv9aT2I--cGd`Il-HYd1iDk
z-SJdfT$@9x2PBF1-0#y2#&m|ds+up_G3!5A=p^e2)xIC(gvzgBlB$qKSL#T(JYAB>
z>z0Kbr3~^XFJb(W>XoU_fCUV;syKAs(E*o(W26@&BB56FuT)pFTFm$}@HEeZUK}%J
zcaLw%8NLs^c$2Q_KsP+>X}n)(4?jGgre&k<#QGz@JP%#z!I!SBtQ1k+g1v@U6sapS
zaA1c#m(`7Q_{AVaa^Y7m?qZ%{Il|qJBjd~-)3tTN|Nh0BHTECysSmrl=`Ur%Ko+%{
zym3G5n0wmOH`D{)=sv=_Kh_CXLaeGAq+8&_viRWVlLgRjow4exloMuQGEM$xjK9mW0$MGfwxM_Phkq0XX;KWFG2aelws<9m61&k%F3%D4
z5u}j1I`SCCR{gF#H=hDc2V5VubsvRGd+)t?7cu<$*R*B47sLP5^O|*W6HLwHE!kGe
zhqy8_;})k6+}Y+n&8_H*18t(3I{qlZ-^Zs5DTFUWH$z9W=uCh=v=jalwBW$hm({2(
z7kzQqkuz339coyL&yIeEpa$Q;kgT=6GkAx0_afmK7d~ct?#pjqcf3+sJ~}sIij9^}
z@!pJ4fb6*geXsxV;LLaB1^M;Ezp5>~65nSxZobsl`ighz~A6
zp{uW%PHn4WUB}??`A9$5_r06s_Vw$yxnr*_^glMZCH1>U(Z>Q7jI>sz8KT$##evDNdg=-eOe(yR`iuid8ug?Q2if6@7kALgA~i0Mv%DSb_W{M+<_#av
zFvaoibgfCPD)?ti01S~l1GyX3gOwd<@S8>dS?gt1%wFgFIaYZeJgMHH?#S-}cg1gn
zf8XpyN3@5Fg#WgI_;=D|WTScL47&@<)5%q!J-)Lj*4F_d)6Hu2MQ;A<50r9bqgzt@Zp}i9+nWpa_gQDQ)b}Q!BK{`EanNLX(
zsK^vCjGtKt0tpsQPTF0_A$|SY;&1`llmT=hNlZ35^|p1ZuBScO76-&h=aorzo?{<5GC4+T-zRZRT%98r!u
zE#e6^Lx=uJ6UG1JpmDSLqsNO9kcg{V|NS=}z__yS3$>mJ;yOG|`r@l5Dww}a$s>3d
zd3$Inhj^r+OGEV*>Av2e_WLR8WoB_r$Pva=^DuMPM{RX+i10=jfFL=5x1(=Fl
zp?Ia|g}8*?OlUhPqh$MJo&%j8d|QW6S@oPql~NLSx7H9xFUB3dp34J%%j~}_u{(oq
z#Jk7ve}SMs_i7qV3^NMTa$52+UnK0Yw@jMGF@q0XkJ0)U8+5&yHGnbb>eNw2t-VB1h!qpA~E6~@#ne1-`m|J!1r%Qcwg{5(b^;M
zt=5DRx-8ctrfa=HFgW)^-;0_9Wjl#v(r*ZW_~s_n#u{ama^!{lGZ8iPFp>LgR`wLZ
z|B_~*g8B(0TBDsKH!lnf;`Ep57j@9SYx~~e4I9Ax*0=ZSAmnD+(fkPyni
z-Lc!wsROC#b7yS-ysH)(IlU@dQal9sJ_*-L>DK@f!9#uVu^qtXj2$goM>kr2<`)II
zhQZ3=K>wSk=h5fXt1a}91cymnQ&8w`15{Bra_rm&5wmcI@%OB0;KgKZ!b$!MUHSmc
zRpoclX5M+)i!)0oKTC2yx8@5_3zln{jow7;Uxd4uc0U2iNt-5W4JnwCp&D{BYXS@=
zd68ccZvd@T(!VT_d6ujYn_T^|!BJ=dVL
zGHe91>Z}4NVz-g>8NEjz#f~5JqO_}}iiGgoictuUzbce&;?7`Vbc18TF?rhhA--*rPn_3xgL~r4
zfqx?XvH3~S=H5%;@YwwbEn}TT+(Fdam%Y@58O&;Lzn#7T@3pe|h*hT_7^pJ0C8{#v
zQGSOUiNs#~A3S*>TBHN_tjyO1WFO=_|2*RPT9OEvy?GfbMso3Z+t;%m>pghm+Jg0D
zK`XYbezGu~5dsfro_TTa{c(4
z4vp8sx-g{Dt!<^c*afLnQ@p>nw?P8^zlWUGmC$}QY01ATAY=bHfLE*AZhx>##Np0fow?otFLJB;b1)q2j);Dt
za90h0hia{c2$%h_uA3IILY)D6j(*?aqB4R%!s*jeg|%@v39-ieX%0Mmb1)a_8pC@3
zl-4^pyfOR!yn~a~K^|6udf@K|efURRcyXxwV7G*R$vkkg5Ms(?-qlWfT=!0FYCzlq
zAGG)Sy~pgKlsJXgz<+IUiQMVNzuS*thUxm1?E`<`Cg+}mPf-9~yp-x7|5_i%^53VC
z>}rE1oenD-360Re2(C>z<-l`osULv59S#d}$>QMi!0ylGI%HD~G2?$@zu%D&Fnav2
zc{oE5W@gs#P9CDcqx+j8ZC$Wp8PQCso;Ue
z4fbD#^VxC2cG50OlnZ>PBh(mkdn0;uWw1=4J=PvW&e{THlDfZMD-08{h(fST}3*q~>sjWa@+g~V2atJLC
zS^HeMISTT|eI$*=c<^LAhx|(JUvzxa1jDW`Ac$6C{E^ZmGWQc;n$vGbmCyfn9CqV@
z|Adw;2Oj=JEgadk9SomQwQ#Fp)xsh;MhF2H^2d>oxm`YqofzzXD-fMFzy!%G&t98o
z-$yfb>+ynXAAm0Hsr-zuAAp?s;eb*Qa}Ig|
zWyW#=xCz>~wUYJMXTBG0Qaphw^(ZKRD9R$I$6w+HVoa
z$aL5%r`;NHlUDqhUQ-4Bqjm@G{RO1_=*OSnLOL|KZKCLYAs)1_)t=7_lSlso6c;9B
zl~GS>X3(CM0(h8^WT}!OjL_ew^(HSP5yu&cpPOy|K(l+5e$SQ$8M1KLt1>Ael?HcL
z^GtQ%=J2W0xb_he3FK+XvhqhU4)bqlBfJ6q18p_FRtHodTGiTO%L|^f8duf%KSU`7
z9*R%t(ojgD^{}E#AZXPSU+uB5McE0IQPnBdsH%GOLU_m`;mIjgmVWOONbpdMXWxAx
z;9BD^JKC>_e!S!ARmnF3Ub211*j{WBmV*_xOD`TnQq+kgEx)+{E7LAlW~mbL&YGVQ
z{AvZT;eYa?>|~(ahKGZlmK?!DbB1i^XaL#qb7oJrgpplwLG_fV9dPxVtLD8$0;I^h
zCDSR%(UZm%swZ(rfmrf0Dbtst$cBee;)a+dc%-mS{q*b>LBm97Def@|I@{kYWIII*
zUVl-aFEy7yk&y`zBOErM@o(vK$4Uo~%k?0OL)rio(mrj~4e&)X*t2B6t+Ab2?<1}K+b%@Dej54xHrcf;O$gT##D
z@N{I4YROC{Rh*px*L8-uc@keVdcJElo+brkd_PEmuDAfhdqH1gw9V1q0Ie$~vX-Eq
zok{0xqak|dJz3W#c@IcT{9yad-~v2?&W19O8X?(9n)p%*U2ynhdS}*6cMwz4*IpT<
z52E{2|@={*2`U1WcJ&iymG^jv>s_3JwjB!d{4n(Dw>>(T92u2Ph%v5>xWt{$w)
zmMzMi8AnarJ*4X|euL`hP{SSedXUYdZJMQ2j*j6`3f+GVV4s7!=rC~|A9yO!&E&rw!#`SpvZ2E#*Y|Ib7yoTmcJ3eGK8$Oz&zwY^OEIkf
zA{k)g>@x!EQ66lwp-?O3CIyf97Fwz5*uujm7HuvGyFf&S_BP|CEpSyyrcqUb2A?S!
zf9uvK3gb@PKTUDN0Ls9BQ8y2A|Hs$2g&i5UP}PP|eBJ)~LGJQMw5S&YB$@PiS*3On
zhMmi^w>j`PAIwcYIc~9uoN8$I30z-*PQLSKrSf+`BB0wVPn3t!4E}FTKWqbVsH0Bi
z%}a2j<(2F5*l#qlyZA|4TmWuPsl77IIQTxYu$0Fs*+cwWG<{js1N^^vf_X_)~lL$tGy}SsKOXN(B<{F
zCEmo+C5#bzQaSil=Cio}^x9#z;iN^;p(jvu<^7D$$!y5_ygcRaCuiKMpT(FV?Tk;e
z+7}sTHbdn@UljT?8ew20Z)B1}KD4F!b&jz~4~OdnhIt7b_-sz>s(ZP$P?Frv{BUtC
zbfw#&Yy6i7MPj~GI;q&go;S|_oWf1;TY-C8Hx`N@pDp;#U-looypg3~F?=6NIO{u=
z*4RVVVz+>S4+O0CSJ>RvEgj~*!1b*6)8VH4j#p1^EY$1%Q+-Cm5SFwF7(C$7#8{K!
z_=iR}ocJl7{9~a6Ow?Ma36NBRIrn<_3LZ$|PZSQ^IRYe@V|zN&ht&8Fxfyx9M;k>@o}6HhrX#0Uqvj&WH`Jw*2gxL*%v
zEF!00#rC1VHlq-!X(xD{
zw?8WT^#P)o*(+Y1iv)Xar*}@ixdo1ro7k}~kAdfE_=(gS0&=9|iDDilh)2dc=sQUw
zXrK}5mF&C)Tv9oLriaQ=M(N&#zQuc}>f?HQF+%`2fqcfJXYEjL*Y4c(rXgY)l$G9-
z%R_@JaO7vL
zX|AIIR^+H6DCKzpw;XkR=nGx6XYOogPUW6OSU+MxN7D#2d=ElrDVGUrbhA4e_8*AZq3h`<%svsghreQV-DRRm
z^5XG#!+dCk&dGaBaGzk?Ktt_9Ljh{_j?Rx~v7#Zlx39=!m4F0Exq@=M2|639nEc0<
z0#t6QkG;4;1DMM%3J2WhK`+)a*QFCKgPG408)uU25%MZkFHdv^Vk7Ul1LNnt2gKA?Wcb<2!hRYauelfp$XW9l_9uH_O+SEYI-&ogAS;~QYub&N@_gabVM;d$;+={Le1>EPowPreY1_uuJ|6!N30IDN*cy)+Q$q!)B+
zUJvP!y=)U!)w5m&#?xycr@QE7
z$(I7KHUC@WoqF~G|6jUB`F$by{G7W+->DH%9xZ?4dEg%@?E0-8OUwgluR;X(Iv)ZJ
zv0}~gWsHpXNTeygeFDx)_p5sOw-C9C%8y%N_W?hz1UPu*=tHX2~ytZEwBP
z4i|V1EF@m#E?-+c;QG(po%v0M<
z<2*LmxHxyJ3*ds7^Ou0fW8_kFG}`zIBaS7DdPO6`54rv*TDXj=;=Z0);+5n2*kU0#
ztJhBw$Hb;kY3s{ipvINwlywzSZ@wpzM&87W;df{)tJ1L@V_GAbiwQO#O?qe$APj>Z
z1$?xBMGKXlT(5e0*aQzV%j&6bdtx%>Lyw|Ujd91lm&_fs+Hl(VL9`*cC6+SIva4oH
zhGGMAH$n^xpjl6Qe%xI{Y-ah<}%jUfgj;cGx#a)>PsW#SmCrlfl0>wyA
z9EX8btDkeq;`-oxhIRxc#XoNyW&%|};vTr+jZtEKSkLS7CjG-0mWtO`oRj~w=
zmQ~HtJTK!CYR~JyCmR=Ntt(b?=HourRvF5&6l^f_Xk4Dw5$8Q1rinC2KyraUe|cN#
zaTs;xa@}Md_RV>@WfA=Vb48>KRqEJcW>q$ElMZ`$lU;L9X{i~vG_+;Sl{R9_bv19s
z{|YckIK?$*rTXJkTp_zBN*l`0Amthup}pOnZ49qahKmj$Ts(sSQ+@p3ZE%4@FN=^_m+
zHfY43T~~%ay?1pJ%Mbd+Pgf0nmPzs3zc%)>RV?t2w7Of8qce2gl?j}r)`kuFZ(24<`Zb{;C&d{F9Yw#JcS;_AHt5a
z51%|GUjzmoPR$EFFG1x)W>bp0#=-fOE#AY_+WUQW}D*d$)(RQ3tOvDvp`*Xk+pzscNp7fZjO*JHbISnV6o=L{)6
z;k1Nijn&(&a)n{Fqltx)IXxaOyB>7Z?hBIkc3J8O`2n_0OQ(o1ucO-YWmWeJ+kt?x
z`tkJ-Pe3ZEsNC{II(of3H&nUp54bPO$|OCF1Dxak*`5P0fzbA~`l6flfS%;yTFZDo
zDjl45p$kt0Kb-~H!lRu54=1gjtXv=Z0JJ>rG&Z7I9@l4lGdI%|95i%EFYy*kom*m{!tI*O<>#n2(AGF3GXRVgIg_5tD
zFm$uMM0;B=1~X6>a1`R;j7N{r@A5P1#U}#L=jTO?3KwEQC--M+e?4b})^7gQqC40-
zN#B}2C8Y>nqVAaQc^6R?`?nrZ88h_3$W5N;rQt|AWEX!ApM-<4DN->Of_8)M8^7fZY@(7rD`_Sh^i`s@mi^Xr)3FX;^Q1~
zLhT;7N?fb#XmmwQ=O2DmPrC$a&qbWPlx_pC$1C+EkOZ!NG??$>slV;1YRf6Cp
zk1KtOROrKn)W>&Pl#om|13|fHfjGS07(c8e9rpPa;PWTJ)_QY?|mBjbR*J8mZ1dbUJ_MO
z6mJD~RM&Y*XI~;QxEAwKco!@xxn=d5ta
zMH0n)##yvz#xxN$zX4vaI8$+2Q{rg~=RA{$b8tZJLHzt#h{@->RZI7lkV96=h8#Hyz*cVU7=D})K5}@4l{j6=uNl4UjgBczoviCRDnmMzxgVr@4rx{bCJ_2s_`V)A!_>C?^luD9Aa%|DP1a|P&}&F)ukO$Z`1((lPosz?
zCZ`mAZ~e>?ex%lGkEzU;G{GkW+j#hXUD9DNuZ_w8bFra3N=`^tQ%+8nNltyCl_X28Qk
zy*4!mJJ&j@8bxK^kDy~_l0-~KC$vy_q(4M?u#T(9bs7X#<7
zcKAv%mxXk=6|#TfU;vU`@cBS`K`&J|WFM|5={gY&y>6XLemPzXr_H)_*5)zv-&5;~
zYeDdqOXE%?eIGOaOH{PQ=!tt4xh}Ybhhu5oGajLn1
z>iCad{8?|Jw+-}Q{kRlU{;MM-L{s_Tlqc!5+N6
z^~)?l2xC2phXvEqHTe3+mvXDb+RS8EyUVUcjujoA&l@MuT{SYCx18#OgNTq?MFZ`)e4^(5p|;muVOI0uc~;{O!=
z^@ReYmmS;l0FF%bVUkZ%#vQjr*bf>%(5As!_H^Vd67#^Zf*NlsWvc
zlwuS7kT8wd;ur@p>56KnxxS{mJn-6
z5XvrWLr;DQnQ$n40V90;>#3WRu>JF?-#TgUK$~TzPIO+ZxIL&XovYO?D
zoci>$T;HC8C}JJyvj+>PJbdS(EY%G7R9fm{_GAEQtyG(cjDAOrp7x$EN~TfT1>46w
zVm*MN_hvZt*?M&7xV3d!VIER(=sTJkw1vLzzT38}$_0Pk%^W4CY6d2?Q}nW1kCCP%
z`Mb)H5>U!NohN#u7}Prc{d|Ks2$12O2jz9S=x5H)&rk3Ap_WJYYS*Ij!GArq{}r3P
z0p`|ef5uBQ(Vv31@2FKX5ec*Aa-2jOcrbT*J+kgD@;v|{o|PFx?O%rET0hhy?umy@
zX$Sgw#EoYmFGdp37Hz%f7an@>3+#_?IE13Ks2C2;r|IC_3xVtoFE`}g{PjrZ^?kyO
zh|jg?_io^2K>Q3@YBiX?Qp?XGbq5GQ!p`yL0Cb0>=B&w#BN%?(#{-$QK~Tk!jCslj
z;96J`=YD}b68d_!Y2=e3Dlb>KYu;lBKGyu1#rP6Z-aGT?Ucr6TY^w7xws7w^@=m2^tJgFMje5apj6G$EBMbz~)!Pcg4(efaeE|tR>@f
zA~JYyW?Mc*Fe0#BsA79Z9R15`Ka0N;Zqx<8O^IOw#pJ=kxo;YXJ%bUASz0y3UvIp<
z=gPYXRjphM^i<143({X}>gSK3rX%-^Id!}UPV+0DUkQ{EaHZS@vBS+oiRR>f(foOW
zQ!MAJV*HZ0H!XjZmcx-4cUCTR`+)ColI-h|bF5_qVt8_VY0m_)pI6+;^T`i_2`__X
zdXy}97L{6rnAOpk<-q9O`v!oVOrW)|>>3)F@*M>!Nqs`4pPI4<=u(|CxwKH@ES)C41lDTjN
zC~{&CiG$sNr$@&$=qI)bgMEc=A7v=efLg-IZ=y_~gW=irv;bL@=-W{{*`x*<`M%lR
zQQ9UrUbm|lmaQb@SlOPmxVlOFUFH}R#UcjoxiHMoEf^g5SC2&NJrYnfP14n)Hwyrj
zfF*S=XAMek%vi9vHUbha)m@2^UjcUZx4tdUM4{d1@7=FP#)1)};YVAJ$%y5@fN-~{
z|G=0$R|g&O75H)W+C(V72TC}#F~**C8J(Pbf3xp^(^;mlV|0NbA2BAmt8gZe%?h?gG}{SNq;yT^J$+$B(a
z)lswg0}g{zR70(sUK}!r7g3gWs{w~-YJa>T{|N+cM5spMeWXB7$d{z2hF6*2d`R~Z
zz=~?~S$s}P_)W#-(E+6qAlKUWhhbw9U4A{|!K}Xkh}4JDSnQ`zxKnd)lp8N*475n|NIDeW1z|Hf53;`p)sS|ZT|_XgD#8zBO`|&=kx9xD~5wZmn{n^&Nm=`mk}Fr9w)WcfoLTf;U$J>ba&@#uzXc9`Z!der||tf
zsBpZt@UHhSib*^5&1+p17p$vDbvo(6Hl3jAoHR2m$DdPu?v)#Kns;R8xt9W)maaZ&
z^CO2A<)qbL|1N>PUng{5Z0&-S@s1?X_v+Y2GDcBx&;$O1$D7VCFu*G(P7iqQa^VZ&
z)ZPrWqOfFS+T`qz3QlXQ+4$u{z<*={-h2EY!5>U5xlXPz!dq_z5;^|x9O(Hcj6F*f
z;eut{Y{qFds3ntpJ4lcka~lMIW(a$RVpC!w)fg$EepKw>vur7B;C)>18Y=}p6I9L~
z-I@Ypv{N#~dQOyyF56#A4|2HDPCj%~fvu)OkFVN$;A1n~JS}VY
zA&EY{Uw}gsw7qyJFJaagmdB@LdRAED==hl*iUS_7=G*Rc=IdBI>Ciqo`8N~Wl2Cta
zk}$%TFVQUQFWW)maZQWGpPqP@GihVDFcxyhi=S^h=?f<+dEY$ft%tlE9b_zSU2vIM
zkCy%5|EsZ$RU&7!8+NnGYMW2?LRGqAZ#=@i&^@fiuhORLU}t+}lcBX6
zs$`W!b{BQQe0BN@q7hFZPtFD5lC^`~dVY_$)9u}m`+~szJy8t%bw1F)KiUN)8#k8J
zH&XFM*+Zw>xf9?*m%Up>eLD=~D;bC?Bf@(6G|rsNPFQyO!NW=4VEA{It5Ez>E@lLc_Bf35OzE;K1B6SL~S{?CkyJj9F_BzMH1;;3MwEV{gtd&wc2{
zUY;;t=64S!mRr(AFn8f!VS{Aiqg_~#Q-`Bexf>tTW`jJg@J1)wPBt@r)cwD3I%0?s+3knWV8OL_u%G|Ni$J9^oPrio*Vzy1#;Az?V
zg9Dw)N}6=%>FattCF+_N?O%q;+p!P#$9hb9vo_u`s{!-T#`Jxgc#QuGVJLn7t`JX~
zHcV&6MdCQ3-dYD`FE(5M*Y)sRH%_FDGK-%_xFUU$egAnK&MNCkfKz#x@i?jW#FcJ*
zt)4TB__iDWq9ft?VcCk`v&{xh(m%l2$z~}b5^ngRxmNI`atA)rb)vY-s~L-vE3q*h
zFT(Wgk!6!&N^rvXT@ba74}9g(aI=i696QVw)a(qV=++@eI3!`yF?`1sY7G*e
zE=C&QZf^!=!uv?vYfd8bV!#{nF2(g^c00q!rKpwYCL>I;F8}64y(0GI+rn|x4-Sr4
zSIe`m=kYh=$%;Ee#DbFM^n{Dkoc-V=1f#=K2Dzso!!DC%-95k}xt
zn!ij}xCU-CW;{lFS_N{dg=hFa_k$dj*^;4b1n50_pM$hr9aFxM^(5t#hJAO*KbqBu
z!1()(wVxzRv3^MP*wwTvcsu0feW}k<@JvV8e?R6n(b2;K--}5^G3RLCrDB2tKJ$4n
zx&P87xT>&uW0q|XRgG|$k&f_UHOiysyJ~o$i01>ADPARfc~w$w_l^ckq7;!oT6`GK
z=#P9Kv|z(Wr+Uxo8}LC7c_o*%?MraBq_>!VQ4FSBV132(;yC_zOmBBhfD{&fi2BTT
z{T|Uv^%Ypr0q$rBdzC!))>>5^^Q8WX
zdaJ{}HvC%zC(LH8W90jRkAW?h#0fzdJuLfx?c^8m)@6=Y@Z>5gwDXRMq5lBtRV6jY
z?fMYwY3a<96n}xNY{1zD`ekJFI^)$CS30`&-e>6iN(Z=_mK|Mnqz}0W)QYpX{zW4D
z=TC=!8b(g*M|89V`oQMf0q0xgEvW4pdyU)o+u*^ea`o}uMKGCrK-!Ca2~PSY%UfP~
zg7_wGU%hiVAHOUD*xgQ78o#v8?jAhtPWX6JoB+T9uIr?`m@|P8xyQVZG(vgTyJe!d{sR>cqm
zKoh?k=a#KN&l3u6%?V4e204%HGpB&!j>AXB(_>Lst*lb#k~--8N_+fNwJPX+*r5LI
zs5MGj%x!MHm5XkNyJt2}I)V8(dweHf2iU(qasjYhMLJK5`}_Lz!ElbO7C+G#eRz3&0VZm+GNaG!EIJXqpE^x9|5jkkP8)JZ&@yP~u~Y+8yfaSUD|NWYie6)*lpv@mhc=Ii{0C{wb=59{)iv3iBG)yd+^OV{`LoA>v`$=}TKi*yeoRQCYQrO<
zc|yjgwb?2{<$33I{oA7iu^%<7BZuhF5AWa`U%rkL53hG?A2NAKDAJSbWX+x;+HK`4
zpHTlys54>p5`4u3)Q`^gef>@aZfz+$;r{c8?92Q*#iS}Q_)_DZ<#rtv6P5H!PZfbw
zr7=xmH%X-5zBHv7K2E4kHD&11KMTSpM9B7Q?a9COo&H?aBj`(a9f9ux
z4G4X!ASoWof&P%%(k3ye1CrSqT1j6YwA#NKPe!><*jZO&X%Ac^p4khw<+=EokSf?l
zTnDF-8T0SOz>luryEA6&M4yT7UVc``#b^=bYGm0VeojEw$vHswN(Rj)h{w(p`XcUL
ziC-c@-XJzvf$Ke6JebkB`p0aj1W}z6ymxnh2K4R6*FRueK-uCq?T%F?p=86|K`X6v
z@U5pudeh`T)I+bZJL^jXItmeQ<94QiUE*4QvZF1cDKO-w-8qRq98-Tn`q>GHa#+|)
z!Xot9b|Q!SOa_R|q>TFgN*}29v)GHUUPTY)yN$0Lg+T9N-KT)wL{KJl$kQaa37J2Z
z(=}9fMT<&tNzWwxfzGeLJOdS>s90Q{rRZBUcPGQO*y32+in^ALwh*a-VKd9trd$|8*5^=b@i1_gS0V4h92Vo^8)RZxq
zB@wPgM9YOcbVskDCjG18?Y$k8t-Y4R;Y_3md>HcLBfN7@gV2rEH-9erpz
zQV&Q>J_nzY|NB}PTZu%*`vQ6MsPHq|md<-F4w#k^dX)&IT$`<@(woZMoV;-f<~O{pkk^XCr85|YLlLuy6gdq&Xkh;+Jl$TgU3K|EY>
zmmmHSjrhz~DuCM?EY~!*_+jj-h+mV|6?{%xx~{ui4aeJGckpCm#os+ESR)pX!(=Mg
zT9psXxSl%1o9PjNwPUn*RhVg^?fvCqb(Lup^o>*kMr@`3s
zg!TMYG{XVB*(b_>2|_*68S#=kh7f1a8ovMSgw1?(Yi`?%!e$i_(>GUab|>wyZ;U^V6_6Xw)bPMb5iKk;_Z;vXt?>%wdtc~S
zH9-8A5#>x@*jVba=vHebl@89A2I{tkKt40Eyj?V4}kRZn{m<)iF!T8OF)`U(l#H89&HToDWjo3nrFHljS2qzEF_-=WG
zV)iRRf=hSHuz%S_BTK__T-s_q+oPU`>!-{#2aFD$Gi&5g`DI^d=PzvVP&phAS#vqA
zYxv=QH4mD6b6dEem8=zb%^p6<*R!ruk%!Cl$4qM??4V~xXSpG*BjnL6jv;@yiG<2C
zSd9Zs@GWW&Me%Jt-0e0H%n>V$Rc_@`5Va1Fx6l4nqGP7`l~NSfKR+orJ^0nZdHyQA
z`bPS4?6I48C4_!9upk_YMw$(HxckGG-TDvw<}7gBO2o{n)D`Rwq6XVv`N788cg?@#
zyr9E^)&`60d8}FMYkOYwBAz{%K~857Jm?wlxGW!V@sv(|JeN9j6<;{NuF$Bej01<`
zCH>EygKQU_)z@h`p}_+(o`Apl*pq6kKG1{*hUF8&MR^I3&QIFwQa>#&NRGbqVUQC)
zu#fr{x1WL=gW+>xyy`KNgXfEyt*nW>%Y-U3?IN%xy-rA+%M>~QoP|C!
zxalIW02=r{Ry+dZAg>^0L$!$lo?>RIqhA|D(*y03AB~=&FgzewT{x&cC~WhWbeIDzgLpK|f{4uv@&;vL?hy&OFd
zv}X7;#HR=ib#A(sXDNaJe#Oz3@6ADgk7>s#bz@YzU1I;=Wj(N@%c-v0Ylp<=L!}d4
zOo7Debkl11apZQ9+=%2DJ6g|tx5)b99Pn!KbB;@L1(JuH{VIjI(al-Aeq|9pRDL|1
z_+FIxX)YDu
zxcXMja@Yb)m#uqmt}23mYr5_quLz*iA1X#i1eigb<)gBCt9io3G3Gcv9V4*P{-?av
zK@hy?(8!DX$bwR*FGg$Jq6Q(_27lRYhKZvW^e^&)^MKQvO4hCB8&NQahIPtfk$BbC
zk@txBBB7jkKUQdcmiX0r>e2`aFA5~p7S;W;N{q;#?aw*4KwwTH^S&PWig+d1(PwY9
zf>3fy%HBVW3DgOM3ae>;Blw?-k(7^{CTcO4)NQLhBm7Rd?`m|WkTAGe=2YUcL%f;1
zb4qdQ2hql|g6S#E7$KXpO3G!jg$SQ^5dMoK5{;&HIyeR$i9ak2^IZpx2=^339U3lJ
z5L=rnOTw%J2<+{@Unh#T6XnBHE<`p^5ZSb^lpm=UBIE>8Cmh?ABIf#9Q0ItP60S2S
z3fCmt65=>N^*wIpCuAELiR!m<5Wl(?r0kOL5t7R~%Ht0ch*Wa7qBPnA2s3}Le%fM5l<|6eQ^3pBt&wr-af(KOC0zYA+Ub5
zhd}$VOSj(QJHgsZqIy;#i9pN9Ulk9Mi67g(*2Yoh5&D_%Vr=uVJ5PUy5Ea&Q)m*P#
zBj`L@v}bA#CT3}Eop^W1j}Y=dUFRLo<@Y}R?3rX_dn2oal2y6yV=eqCL>prI^
z_JyxZ1>(}mg)=K(MN);8^s}sjFxw}mjb_qrAXCqiV_L-1u-Q$?dyujQOzz%!bFSYk
z_@=_|nIB}nAxr?z75v+UOb+NsdAif%<}Al|n=zijrxl&=e4d=bDjr;TPtWif*)iaK
z2r*v5hW)k*7Vrz8bRve>^Q>KXX+vA#j*I<>C32%!U8xD%zQU6lBnVlk-eT%JrDxf!XncKIsU1@bac#tXYUr5^}}Sz8xz!Ov={1M0}{Gapr>f
z_7dz?9{>H`rdA|fyDoP%cL=l6c5m8)|3I$2AeD}ZLi{gA$d}*DGaAyajpi@`BCP2{(CARq3V>34UEvUJFi@DEr_8^7^xP
zBbo8KCl8+NG2=(Sg&wx{@%|70&h5~)yOthbILV?=I(P}4U-^?;uE~vFYPek2qqr5<
zdQN{zqID;_!|p)#EjReGdbcMcUDQhgKi_<{U%m
zlQr-7-0{SXl_Lj(Uz|WUKZ)Zgu{($(Yct-#;rH?KPop8ggAQ}b}r=mCC#@dam5?|Fdq4E7OitGq_3OBb
z!@T_2l4vw6T9s!_z7bve_tHUYH!EOujM=r{kOxk^?9P92h7kzq=ql7nDnKUh=3TnA
zJAtj0%E0Gy^JwGJ)=hs(=kY9zt{r(AH@~PR7YtA0t&ML_T;s|7xx@ff5*%O%MoQb=hLhPHs
zKsvD7RRQ>on3c
z8C=aQwbpsg3hcuxZ#>wn3X${v?=+5fLik}<$NN%P$ct*MDoiy^sCL~
zg9cjL!N1wZiDpl~6Z<7vhMx5JgYT0wo}0`U249TL%P%Mk65oSpi74+AgDD!ts*@(f
zejH6da~JN0;2=FX+W&d_ztgfHJSS1i>%msg$@RG_2oVL92?5&K#CObBo3J-ah1$5p^{@rz9gVlqGu
z|D{ouw*^S8Ww9e++kn``ga7{NN&&V`bW1<6|4x)cEZzO@f)H@^lz)EC2u63@f4C|-
zj~AUnwB4L|VQk5`#dG4j$(T0HU)N4@0F{CEur2Mv(4cqHjc-95@~ajt+J7BHbr+#!
zX$l?Cb6+B2|^x
zCGcvTJjjo}U>#~I2qQLU@3|f&2%AMp?T$<>qmvFdQ>^zdp>0gVfk99NZssj!?2cW<
zJ9;t0S9jM?HiqU9wxoX4FPzD!#AgE6Hiz7sJBarPt^aMhbC(_%1M&RRpFZJdcG?~0
zGkk+T(%L>;y4HglGVHMN`S1e27F|R4IFSkdauYa_*7zO|++&=zM{aI&HWal_Kd8yv7Huy9ClZ{E_e9U|F{PZ_(9oBmETa@d-&y^)K>QDPh
zpKv>_eKo1=y7moJP3K+Tss{-j?9M;==w~Hs)k$AE?gD!R=irQ%nt`qnb!AEUb>
zrf%I~Ys0fy;`${j#
zB
zX5jp{fBCA29Yi;KR)v2?PNSD)-P-=Foy3z_&u}QN8K6~e9{*H#=i-gy28#`q^61L-
zfre{4PNS34ku#If7Jn)$czj!-8Hxt(lTMCvz>^Z>cf8zAy!TJPj;a=k;ChQG=kkOk
zal@H7r9Xdmqwim6#(6wFgC;Cwzntcg!k<(Qrf9KC;SL^ueSd39poc1lb(Au<;7t-y
z4s5?q;%o(Fvqpo0IA2=xrE+4Q6otoU_7sh6!Nc~JS#S%{qu;mrZL+vvg>J7t+)>7{
z7fsp!FXDT#B+h9clF5^_5iNEd&zHIK56L-KB^8DoLpwhud@lN~_(licW5~A6`zk8uF!#fC!Hbt?5QnubTOC$E
zVspuL6Z`4@!)+H6AFyyPV)y4R*{m_YLm02Mk>n!=G2w>Aus3Kaazxua)N+61#eKiTBzt#(O`p_M3DIvZ6h#Cy-!kqs!bdALW
zABQZ~?fvDwgq;VX#p-D<
zxHPTxBu(p^=AREF>yXUu0QUv0oO)9eYm5t<=Y(D9^D4;Cq0san94|@g4i-u-gMC^R
zcXNYz`Y_V>?yyG1qX?-|r1o?{YCo4=1#w2{oxzMR2+muqd|ij$H?dbQj=X3hsm^pLEReUgpOZN;XY
zG;=ov1tJ$X>rSK^U&dx-{?m>f2td+3YMIzw!mv@pkFPcs6k=oLY!#**>d3@k?WE$G
z2p+8u;Bvt8Du_>QR2gI2kgV5{_A(qyRlGj&lKTK1tg&{$?ffxo!Dj9
zmMo)ZD_S!fxfORyxFS{O=fA4&+JU%;YBxT4D~%1tb9>mQDOqGy+89N<(|ty1UAF)KIor)Hp`JQPcLcHBlxYHk7d?_5HQQ&~kU4|Q2sAbbHNiPLYWh@doZ`~g`ip1K1At>}X5NvlrSf%^USE$hs#
zM4E;iY)JB*7$$HE?B*Uutn#a3uf3kYBuFMJbDmPTwaep_l1XA8_qI>Wf}u?y2y=kKFBu&^Y6P&db>D_%h;$ceOh5ZDjC3
zm5Mf7^`)1hgC%EgcQWk818x)=Vn_M$T{A)pww$VHe6^6c(S0xcYtd})!_XkyK1JSB
zd`1rSxb{d+(McLt&N5e;-l>Rc9AaKAW6?&vKbVIOdOG31yn)VkHc{N3(`=mcJu%
=>(6US*abO=w7*{93II`;TEiU)@-%})OeG7s%sA02rJ
zY@df{vG~X$&P%0cZ!A^N*r%_&s(MNwhSC8Z3=>1m#R{gEbI*d#Fy2G1?rLc4(p}qB
zLuPPR-CcQ7*BpW3yzK_B(-Yu`6>XQucQd@h}@WC8l`=53DIG9hXj
z=@-CtlSr#!7E_Sa1s6{rI3gy-4q^@OFHU`AMU46xYTF^SV7&2)$YQ~JQtXGZ%>a!m
zO7ALhK#zq`E80`K&xZvZDpRkT7yM1~3M;A3_fL~kk-73(D1i8Utz&PTYY@>F!z}Ei
z9Y+>Fl2bE^t0B~1!+*YNpNTZhJv$V{7>SRf)1KXiR>X)UdEtz)J-O(pcEIYx9l|w6
zV_x`V8BzMx#JEb?kR-k(YN_g+CJUVkc<6&K5N@M7(W{%@WT-N0TIs)RQs?tL!*u?6
zB4aIlXNdJ7Vr0U|vEi6H8P0b1o%7%=qFlz8w)|Z(As?xid5KAtyj42x&ALmB9C*KY
zReD>KxJ?mx9=>~r%zPoz*niB97@idu9K1&cNSTORANKJ~k+NZF1GKNh$)!Zr5XOiX#I(%j3-9uLl6}v#Sli2f
z!sGt40!G1BQpS^mJ1d;xC%L(p_}En$ye)R{$fCc1ZYp138Fjb`zTM?qqU@)GJt7-N
zscv?V>-U~L?&aAemZ)@tFCAn;TD0SdBlh%2I3bgFW||axHciqP6mG980%9Au8SV1SP+7s6pFh&?A%`
z3^_US`Fm=hjILw$GxCqYHH#CyPZTSW_XQKpF&kR&jHSTc%Hcc+U23R{RyRYZPw;y6
z61G6R=ES8#v0310&v16m-_NAq8;P}~y=PHFh)^4&m)71LKNA<4fspW+G^
z!3B%%!-?gq=q?p|T<@PxVDY2ViTB$A&`hb0-^{%nxfs&3-8mctK&Pp6a77N-7V2z@
z7Mw@B_Be7Mv0p{y)Or=44D-Q-)Nf*~+d1gV#xCvlC&hs2y;5V2TQLx_v@E<;F^1?+
z!&xrTG*nBYQT;(W1H4x<1kJ?-h(qOw@@8ZPxYJp6Dm7plTepuAI^5GkKS_cdqUTI}orMHb(G|JK}tMgO~L+di=$KF<3VpRM#3ye*=kno-O^
zZ{B%{JKy^N$kbo6?0u_1i0j75PrL&7$F$LoB??EDZ+9ApNy>L%>)HRKZ+Q)UOZ{?D
z^wv%Y1?=9tQ1Wx}*YA`mN-N>%*0?n`BM&U)tXj8K?gOvni-Ecf`f^|
z=L>1SB{VOO&Gh5q4^YO}_1*TT83dHzs>Q9y;@+M3o2@8q-#ZE1LiKl9{^x19^wIruOaO-S
zYstC(^}>3~l9%R}0&x6|=K7r0K+JH%gMuCL!qIkn%Q*tvuqHPbF6zGyPit+n&*baD
zv8n@uHKGQ%Z|#W39#2zzm;17mthx!lIi@-_Z498!`|J5V$8}+K@{0UMtUhcW7Yb2M
z)rPrugg?i9Rf8X|WFBD8SH;U)2AQ=7H1K8J6a97b5UVJE=S-gF#9QX8CRyTBU~cdFY2Txi$o%Yo+%|>WQ03mc!_IeQuzBrcvG-=;
zxaPdVj!um!O1&V+kY5WYWPEh8!odb+_|o&_
z6X$Wx+nhz0)7JQ!e9lLUYiFU(hU4QO!dkGzi#eA7XK=rhq59g)dEB_q+lp&S6So^X
z%m4RY9#360x-cuPf){T}*WRl*gP;7{7pi7*0RCQDNC?o8fvdCbCS@@^lx%%phvBos
zki&h0eYd(>)$Va((+7tehH^w9jcAZ$ef@Fx>NF28BR16cl`YJ2Dd8OX?JvFF&G
z6Oco}|C`D`CQM}jiEi9hV)p
zk9wFWo}jM%GMsC`h3{fnu7UtIKRLVKKw&5BeVQgw=eq~qJh|J#r)(A(M{tR%_5MXW
z|788MWTuDZ8w>R3$ivu44^ytTKY+uC+yQ&Z4d8S4k;T6@T3lr4eEw#T0A9@|)!vQM
zVlz9jyKl3WLGGB*`>wyE$ju{Lq}hE6_5W&D6@V57h;hSC{9V$}{udadJ0W+p=(kIYyC3acK&0^%IGD^xFZ31%=F+O|BNu;Pc
zH^^5^jo++VE~J0!M}p+!k(91Z;G=rxSiE^Fn*P)+vhz$jqTZGu37?(Vae0m^!gL0u
zUApb=cWV?JRbv=2e9_xqF*K!CQGOebZ3ga6*fce-7H=kOxjOg_}swaTxHH~g@`itmr^BvpGF9GOj!7Yv*o)iwp
z`RIlb`ct6oUrYmYp&APRaM{zlP7N?5Hw4UEMuHQYyYDd>>wvX#Z~MlhhNzwUx0uCV
zZJ-jLvPSzu1no$tZjgT)0W{?W3&lN@z?VF`D#b5)==Xs^ts-$v&~6i`k=Jt)q1z3#
zT}79GIN9oWK2!~<-KV*%9?(IAKYcY|7sdr%MSK(I{VaguMi;yTh7Hik*Aogz^c@-d
z=mgC;-~tv!GM4+|*wIR^Tg?wCDlj8+^Y>cgIlynozd?Qx?*#q{G2Z%$QmuPQ&ZR=Hy)irn{_i}{X6}TZ}CEC;@ds`CAwxZ
z%dBG}`{XnsUwl(8HEfjpUBvnT8rBh}oVj@DJtHu_{Y!}q{X&>##`4M-43O_+qT&-g
z8i}^}!`t$=bBHI_1M-)%*2qwXoub3{K9U|yy;zcLCXN+tbcI|kAb(zZpMF)RkaRok
zyzd^*BH8D;AT`J}Nlfjq471%%BUiW03#G~DB~FD@-rKX(Nw=j
z3|h~SjpCm^-MryV7UjXn-UL=+@`mQWvwjWrOI~re)L;InUzuH@c{w0S_&Ccfr43vm
z05?eL&~V)RzemTb$BP(ZJk5hw31+1vMVF$
zKAd8hC3*>~q%7g_BL`41FZ~b0WFQx&7=u
zO5B?a<#;P~gRl=uzBpNTll-C4x>F?Bljy!ay~m#I2?-DEN3JD
z=08_9Cxjbx9*{XV$W8$^y}@@62%#UPr$&t*6JROaXZ1ZGzf&=lNq5SSzg&fizwlfj
zNauJdlb`oV55BmyaD!-q*1nHmlUgS>rB9AuuHPcdd{mo5_R<5sJu2Qgp~9&58J6tP
ze_e6%ux$MsHHFBPb+F;U_18szyvU=}PLLN(I
z($hAoA>?Q_t`|GElOGTG?d+NPLNqD~2^cj_6RMd9ChV=Ui3U#V$c&j{Qr|(9{?B)k
z;EZGlZyT8)zx_6Qa8$39e2}Vf#p6jZ`SHxiUL~Pq!sh(ooT_FH*}f1^7}fWNpvv%d
z^;}#aKU`_|Ej)f4mGhq2r%8PloR-xZN#!&~_IKXd@#lE}8cQ}`=HVwm!u6NU(kK;p
zd4vVg13O@-2>b&W;%#}(XW!ul+h!v0|l)t
z4m^O%q-UIFL>lc~^*X?`YJlut1w0t&`A8~u-wUsLK?P10HXP}w+=WsWf?n$-90!(L
zF9l@n0qC1*VDe0yB`zHLb$;IZ2WfKFNHFj-Jt*#ADZ0mg98G9NWte&?fd7`4CS)Hy
z1#h#Oio!M1&@OwnxHO@Bz%tByn|2?GI{qG}cDXl$+-?ws(Z~yE1vzw{@4bai2z8j*
z2>5{SLgL8`KOcg3v#$MNC!V3tZKl@>>rD{FBNORUd>t@k?0uFp?1&;^T7!;@J5W5v
zr0sPm4KP^JTuuF?419cN4aI;VI?cnv5N2fwULMFPn8UY_UM9zr>mU&gD~LURUZf7X
z7E_iU*PaH(0f+31?-`&Rno+LzM}5I2eG0f%91DJX6qqu_&7+}=Ar{YfQ(#9(?%V#m
zb7*y47{vx{14dJxqAe}~$VD?oQ7)GQCe{HjS=|O?b?oK`T#x`vfS0^#!YA}o#o>Wg
z^+#Y#FUUVT{v7B}H`=AsJV3!5{(-_zvOx2gUbAa#0@CIWO8Y!Kjidt<>`y#RMQ7bo
ze%z*s2O9cGUP}94BI4A`J^yJ|0e8L^&7L<#L0Qm)Qn;NCh#BMGN#8Tjji1ah8AFsj
z>bSao+`TsR;{)?RL(&lXyYn>Pak*XaRc+GGhVlc@f4tD_hkz^&rkBxwq$>^2_Wo^W
z&QyWhJD2?RgS(Mn&Tdf-9#&|q7I9I3mKLutELvpW-GO5zBuzt4?1ewy=*g4~cB0!A
zk5UreE~3+zgRb}ig-gk^G3-I2<36=SkGZdYII&{c1zOs9k0^c{1fl!HpU6I3
zMjD~7--Rfl3ulEL?V+JtPaAuesb6qY}hlf>|ut22bMU^PyA;Ez-E{%Prqy{=9H!
zqJl#PyATY!b=2B-u)y3J$p!CBhTl_6W`SFEZ@>REW5BJs-0Pe#cHpD6S>rd=
zYC+AVX!!KwI%qYpa{Kjm4eTc)6Lx?93DTgC)J^vZkh5Ya-+b-|TB>0Kjd=}pU@{-uhWZed;17XW$>FYpjDXO)eX$3qBJmx+ufu*jXr0{KgY&r!Xjf_!9TgqKw@
z(&giR==$wDV%QepF;m_~-dqKIl#?2G#?l2ky;%gU!AD~47fF=#^tvQH?JCN1JXD?L
zx{5;f|IDtU*+fTvzceze`-cAQD$_K(O@ix3-$noWyaJ};s+Vs`uY#4m@*G{vD`1tK
zs-xfU6HvA{4F7$m0Hr?NTwRhJM>tO3plIv^vOBrYnX%_6w0?2zsaTH$)DPc>%ADx&
z_eGJo)4v7r-=a;W4tY+TnSVB2$aouZ`*a8MnlfVk%DEYZXbv3FsXObJvy3bYuL|m2
z-2{iJ-wnK
z^HhrEI`}JUwDZNGx4`Y%6~&O=QAD+aIWLpt9V%^ReENm`10|8G)2N@#J05^a^_2}U}8Pw40dP8;m!8XK75z|`J%_Hiy1(i-$g#K2a|I8!$9#|X|
z&KLRk5%Cs0%XEl)i&VM|_ed3YgAaS24qO{-0TXOH&i~>YM+^9TUc_S!Xl|
zXp5G2m&leX!
zk1qh?{)%542C1>Kv9z-}Upv}T;GQySX#%h0OW&s2HKS?&VoqJo1Tauu=-8W6jcDoY
z_V~>D%)FEJqk^g$t3pX?wpe3JjvJqv9zRr7y<52U3wP=2x2jE|Vle&YfFZvu564nW%
z(BoH+KAiel3QVb@n^k_KgQ#@s@uCY}NY}RLjM=J>%4h
z25RqPDLQ&X9jKWdt%_~51(ngaj}obJsOZ)m4!>0qq(0brThTxXWC@qW#m0I8@918C
z*P}*gyqwkQ;Rb=So+q;f-&F?bdvcyO6v&{eCHuz8un;i*>OlRO?bApz%kOOIeQo40
z5VN3LtN{{3xYdr9pF(^Kq5tu61Ms}~Q6g
zew1op95@NJ(CjIGm!`+ufSup>Uw8~7=;?jJ;m)xW-K-otuyTS5bXR@%N-EYJpY60Yw_B`XzMnE9G_JQL8*F32#%NP{GMe#hoAAW6_Er=4Q(nlM2GlRHueeGR)H%Y&(j{|}h
zlVpPHp0mSxf~cC~D)avSyQI%YlHu;ZY2tih_O}I_aq`-Xj>PBSKH}ch-Dc7g^dK0%
zlsX*qk+^q_il;rlpH%EW@X4B`kq{KFCtSA!2p4H;YR$QIGOyrl0jK;VS@&~nurjQf
zV0tbmG7?%ydc0ULNuhs9+WuD~+CMQyTDrZ=@#OhHB&=Gz>v@<-1`X*p(^7ovZlg4I
zg}=)QO|$O?FMV>!d(+m>bdQY_7tMLbuS%DY%QvD^ohqD&Ee?GC(vf^(I|8`U?t4en
zetmwBTCt2^Ii1!xU-^Q(JAHU+iXnCax8PX2d*ce3nBrb|-NBW7B_!}ySuTeBb0d_x;#wRL{chVg@TDc$Y`K5kty7WY
zW|VBq%CjTB%T>$PIaLs9ezxMwF?kF1|AS^B29Ix681xm%1_w_#s1=}W6$B?WJfH>3L2ln*C)E1DGF9Me1V
zjhwD$P?s-r%B3P>WcD#md`MG#ad!Jzl};mVPU4
zX81VfbT-u6)^=v$Ruk#%R&>1H}g<4}<=*
zZ%UWQO@UUqAz5=`SAZ}2*@v6e0e>=Sx5MNuJ_pPBq&+gd_!q
zJ;x-9Nj|@(Rw;^R=0%gm&TzRNGAKo5PB46)kejuaryW@)%>87ejkfZL|9n0%7g?8+
zd246Hw9J}_$iHNYj^Z?V?)0rbb)!Z?{QgnPbwA4YM-QJq{wS3QzGLv^Fjp!0*85=V
z+Wm22%0jZZ=-W2Q!!r~=l1kZIY=6k#UwaP7Obh(8=eU9nt?czGW~KC*ax5i{I1<2Q
zrAE=x~l=Aq|
z$e`9S@UYYqbVr>9w+FhI#?SpFwOwDHb3^Im9@gC+JnNf;pW+SvI$cR59_19ZEPNRK
z{Z%zBWcLzKuF29X;04m%t!GttJ{g=mbhK2+vI)I>YjkU%{tXh9Y?`e!ybfem{=2_*
zTm#seZEyx_d4RP)JK6juEP&Txid|+j2dysZs$Lbnh9cNAt={Y~0KV2&tP_OekwLN0((+Cdld{bJ}}ww}lFuU;U!N%51dKoA%w`?hwPbfM?gKRY$=
z7lKPd%Jiq&<^WM}JAGSM=|*M&77$3|KntV9D8-fvi~G@6rvPM0V}h{dyATfYtphIhr;qm>oK(
zAs+b;eAM$_x3Jp+Ey@2?WLgX&iC2pr-zPiJ*;@a$6DsW>?#?Nztph_q_r!Xw_SHe~
zDqQh~`qft;Jz_H|cjYCT|9jSoUal4KU3}_8721ZLDWzo7-)%xUv`%E=D*m`4UQS4V#Wjy*s_slnx
znkVWi5ikh8MDDyUXWs!1F8SL|nmxrFC
zecDXQicgmitxDG;F6ng?$a3w(ebYr`*j6mMrCJJd7x{xO`Mv|&-iMyP+}n?eN_EEB
z{u@9{>4FzZSd!6xGQcgxwG!;JFBtkdkq)HpxHjyiss(6-M?Y&v1IP`Y3G6B;0ka>B
zeI1+bp|VG
zTVDEEz3++WSu5jq-@m8Gk6%(#Kq?BPov_r9>Wu)EMplzs&)g7Y=JoQX;20PX2$L%~
z@dD@z7Kibv52C8!PoG}zc!@q3m2zimRe`awP<4I2TyTIZbL|6lDbi_h+@xDiM|*9K
zDI^=d0G9{9EUD1NgNUlX8ap$R(E0wFF)prqV5LItjN(Wva4wrltd6;k;OFdrDmVPm
znIoJn`Z3`^&6uq~`zFO>ST`&a-xG@NKz_&9s#nod*Zu0K$Q9%{5`SmkIHFyRQe(fq
zMao
zdYJ+@fc}IlM>;xyKl-i`C?EmwiTn!k*r5#E)dJ%KPMV<=RdWwHK?9J|!M{$ds-T&M
zGlHJMfv91?>_fGn6_Os6wg|ee3W)PxJikAuctrvpNIMAfA^kV(QU~vPfQ9F4QN=CS
zz*~{Lhe9HfsA6}p?R#V;n|3H?|AORu8*S
z?b-vjkL!mz)=Hz2MGo&H5%yp*BfdRS#T?m0eyhGev=sz_!uoMAi1W)JWn^(Ql9W+Q@t
zjqXVm>(7s5pVBo37x(YvkF@?-8qeRvs=L}*+BRWIUhu;}*J}>6yYoTKs>2j1e>j(-
zjrmTDH6%H7xGa+k=bv?i44y_WTxt}4UV+z+@$%^Ow^CN&9f!8lXtnE
zUpy3p2+OzwkDnSmBa6C&Cv9)8lXiPqI^OC}ftYe(FBxwwCeK+{9di5a
zOFoO_=D#)dkwA*(A0i)(5uDYbZRIEHNHZqm7S`PV2;O7Yl{LbOh?!M`(VR!`h+l6G
zZbFAKLOr~4{}Yo+^1ob}{Q)LXME$orrC)aC5&41rujgfl2%NO{lDa@Hu^{|MS9W(Y
zxkAezN1OVDD6+L44hu6QrFzT5&hqDz=hOpTD+7{A#r(4`t%~jw15F&VEw`*mnuZ4r
z%Y)YB&WT*tKMH|FioQ@n$GI3nOMv!U=`RP8=BCI!w`Dov@Ap}%B@-)ROSAP-@0C2l
z`_U)$*h}FgG+C${4bvl^*Bn-OWM@qXq#d=g{?Cc5uCVX#U(F=WD+N3sxf(+HU!DB3
ztLhBdU48Y?Ew?K~juw6R#-;^19C^RZrpAk$@!KIjG37`yaZd^c(&-Rx_p&b3a>$VF
zIb*#!L$c&4eMz2lKT}e&g#PQ1c|C$NIq%cBA15iXL#&}liHYpy-XCN@xex5`nmB4^
za-Fyyc70KzT#>XLraRbY^|#(U;_<{B8yCUy`N2YcE)SV#vZzL1=t{bnNFc)`F7m~#
z&QfMVoDj<6Xr&PqCi8I+Ls#xDqBuMEEBlu^e9y7As%~E+c8&`gOxS9|bY+a3`AnU8RT$^>HW||!Pk7it`nwOUVNgHTMhHY)Pt0+7eXs_
zn$H%8lA!pB_A%*t#svn*-=_~J_?SZT3<9^32
zcf&tZiV_B=y5S^Wpq;lvC*08|`gS<`J<0;_oy@K@9ilZ9QZ!g;&kiuAIU
zR$>=qmH90AmhX;!{T9FS
zH%mKK@CLUoY%hjPcEQRWj~p#ax}XN%rSI7kKVYOyd-)mWPDmW)yx_>%4l{xdr<-gv
z!uv;C{&>bC$QL`#q4W+>cx5B*n(Cc!E$++leLNjd{<`-{->GI8YsOa9ZH3|1(r`k}
z4+L+n1iW01>7wvr;{PsBcfe<7B4ik~TVQWme1Dp313WDi(85+nLaHgw>A9^gc#d6`
zbt<$I?yc~a!mlU69I
zrfpt9*9hb0QeA@>8ej|W)mFJ~%Japx+8CG{An(E9&AVqY3{r7FxYq!|vb0!_tDXpM
zOiIj$q$4OLnKW@=6+usDMY^|j2>M?xSpTh#;64z!TTKAL2`QUMT5}Q>7aUPUF(lk|
zBtq$4B7!=2?#bu32v%xx^62{^ct!L}%ag|#&iSQxk~#>=^9eGfbs<<&@QR_9^8cbu
z9xhG(2xm&F|A<{tiGlDR^_0AAwnVvgM
zUx7spK`Le`-ECI{ozU1_>v|IAKIE#X++PptZ?*qwJdR-SNN#PEIf8|u2frNYqrAU?
zy^-(`B{%Wj%KM5{9h_(1zgNti@_lgz=DQJsj}<0PKU+pnebf1DL`6Nk{#Q}_UR({V
z)k;$x3PaE|gi3AJ8o}x3pPC~zC^_NnSV2+AI`}RzA@0mV6+Hfsb0+K`g8jXnD?Q2>
zN@O-z^H)&bKWpxz0e2l#`S|nk)$`R*HTL-DKMM7*#vN=dOw~f}w-S)Lxf;sf3Md$O
zSP6|rpL09XS5UYmWm9>vnHY4P>xdemCn<31Q6DGt~Ra;oW@#N8inr!t68d
z^cG_^P{-wxY5e;txGc|dBigJSjt{R>QN^`d0gjV1pMUy6Vlfzn$IG%7e{q#}nst^n$(2S_$}Qt}}uuEjZANre_?Zb?8(Iqa2c6Kfcq1CMl_R?miq*Jb_ZSM`ht(FF@JR9j)nrZREMT&p*=?><%{WfLT$}T3sV%83NgtKPf+4l%b
z)$Y^ZAMy&P81Stfx>thVN#4@vNi4_Qm5xKD7d5cXuDgueaxs|uP3gv8j$Dct%XQ~E
zdj*~~m~pO|s>W0PMJ?)v+W5Jy(B@g+|8SVQ^r@5o6=I-1&l`BK25UuUUrjeg_~G`M
zTOBpG@D4l0OZiL5SSFZtvYEoiktwa;-bKk#vw9nGzUjtTUSzA!`b-%v3BH!CGgOI3
zDg{pG>XF#y?46e)l$>@^!u?Bp`&;pryRmAeM>+1az58-;l5+ldFDYNrLfDumO)KkQ
zGq!OHxVVb?D4rXx#vbk8lS+;1v4Ah9UwvpJzBAWgLCM2nj|VeH==Zl{4;i{q
ziK~dhulTg6!{3Olq?T3rDbHEWn0*gDI>JUAtSc-2tgBs<1Y^&6j+^>OmV8q7G$=H*Lky
znxPfjzAZS6?&K%;;8y&D{l@q9fOgz_x8q|Sdj~FM@RU3Jtqn7Lp;0-qzYTk*&2?Ux
zZo_jfqR-E3bz+{k?{=Qm>BglrUmnBKF8rgPPUOaRC$?!4e(%20i6d28dnbN&B;dvc{EcK_+d*+0gkV%dA~Jng2WG2H+@!6Pnyvuyxp
zTz)-ywrl_=w*Ryi(`&&rOgcK>Q(ACI1Fijm`4;?jG~bs;tp)4uSjY{@Y{q)We7Y-`
z+VGEuck4Ty+we|~$m^R4d}zkG?k@OqP=)>(oM8?r5g*r
zjtD9+?ZUf?*;=E^2k=gsjj>Am0m|Op%}45Y`|;!nmrRGly*PKH-r}mzYg}~2sw41K
zGyeENL{~1T5$7hxC&rzs!=Ac5IS1dA;olyO?_Q)8ZRt4?Rl^_-*lL}qQ7-X)acqg)45uSvDz<0@3__Czc8-p$mMYT1Z4TRM*c&pIqR
zZ}m^GtP1aD7Bl&ExeI?aS;}y|)q#a)w90uZTCfPkm+SwX;)!c^f45UmGBRawaHw(@|eF~
zydwe26*%MdkAcu87l+QIJcDA-_W$H~UW#X)sb7BcvH%af@m{^|pNs|nGG9?p^MLo)
z4&*Fk#zUtg&*h@Ks_-Klw`bGFWtb)To7}nG8Th}lng}C0Q@rX)h%uK&Lt%;E_wV_z!c9VMiD6CmP
z&A0?JpQk?O7yK048vSHX6eX~8pyOmy%rhAAmgBdhM+RJ@;lAoEPy)HuX-@3xtc2@M
z7g;9F>ma#j>mF5nJp6H^*1(|aC4~B4K%h$rG*GI#rBz=AtsVab>Q~po>GSVs=0c6D4-Aug
z{IwW{wDaBP`dI_{yZ=s~nWCJ-0yo;P*C^n`pDV+~@d;4w!*?%-^kO)yNxc0aTMO?o
zOJ)w!VYnmIAt3ls9gJdFNn8@Br_@iP`+zEjubJCZc+NJ#9@aue0Sc$hterk#d$1OU
z`(F!np!i@4Yi-)EYd1ha8co-j%T2JaQc?Bghh|vMazoA}z8;>L{Vd4z4Z(zaC-lJ6
zMku#e=K6<~W;mAhbnC8XE9ANxU75B;Lg`%K
zs86;-@v1Y&Wqu*J_}ING`f?-Ga5gRRyWawZ`{ziOH*HWOJko%3p&d#c3x8i>&uM`+22QlX5jW$SeaG8iS$*!NDQblUUvbmR
zf0W;A)R6q=+Tn2C)d%`3?eJ2*$15?BP8h156>>eY10EI}`fjY&0gL-uBAjR1;gqZV
zi&Vuxq-yrtZVp4;89i`rGxv-k6a<&adhreu&JK79iNyP={e{FzX
z=St$*r5fO~&INn(UIZnk8yE9un_-1|jrmpYX1MXC{i4D`BV@B7UB8t!Ky2X^YGvI3
z8(6+1m>0Cbx2j{6OLtqK{_~suqv9>__U53>;~!1%u*#VY@5M&QmF~5We60gouATYl
z8Q%e2CK@iBKiv+&`WcB|jxBKdV(GTCWHTI1-zu0>?}l5t4u)gdUC=BRns4mxg!3sL
zRvWTyaCMJUWsqFV
zUvMPPn_sRD&g8I3i?mh1hnqDX*4(9V;QL?mLx-E;u?DC259b=;9hv+2i=PlgA-9Tb
zDfOM!K=Hih(F$l;9)I*ae+!Iiqce(BY=+j#0S4NE4Y24O8$VNL4cylm*lPcx0!F16
ze)!p40Q28kvxV}%f=x;a>TY|V!i%T>W8+kj!QoD&QoOgr@jI6*%N%t@Q2U+wU8&wI
zXshYhUqKlTcyBB-w!Xaz7v~|!zYvOFoKOt!l`4fnd2i|TFBib_c(!zB(^Pm);xmV(
zm=ARKhzj01^Z-8`QYIH=Dqu-L+_q&vG2Edf%GJb@2Dj7HocoQe;30Z3+rE?0xF-5F
z?U}d=$hU6l|GB>ehPl29v+mD?XU|yWEVV{J8}T?m
z*J7W-nJlbm%J90)s|53BhYOTHD8>zShNck`53x)8BUMe)ILx1aDL>-LOZ=^cg+ofD
zfC9WRR^Biv#)q{p$IxsA;-C$qV1CD^_?Bc3dt<>XY<;4{=wERzPIY>~q$^N>KRkX#
zAq_-fVOiVyX~|?fE>RsfBJ>IiCq*^=x|fFwrIHR0-7Uno%NgIb=tf|>u^M~Jnndic
z_g*i|Arq%|UzGSLoQ<8!%Q+7j{G1EyzwBC!|J$ix(bHCfe^V=~%{?f`EZoVu3=`#e*P-(3=Kphbc)%jIdQaRGA=jjejOW;hn;QS#^8V3w=U9%{tmfrE
zhY~nS`TD-ePZBsa_h#HFJfFZBN|^Y=X?ry1l6hi5@Y5Jh{&EYQ?Wu7b&9%or-qud!
zxcl&};LlIu7|ZbJIfueJg>G(XvWex~@F=>~IUyeYUSzpVnQS5_nxE%+gmV&SuicdY
z4g^GV>O4lz>emAg*t;rj=I(e-Pt}6d8-7U~@wTSIi&x!Dl4=;nCRofZseeal?#1Z@S
z;f`LZ6i!@DS#ZRsWR9zaNQhlXGUtZglC{g>^P>NDZ5B=YlFTt$IYxHz
z?j+87hcJx+yClvI&UaoCn#}olb0pAX5{J`KEfBRSkHa~?T1WE9ANYGmk*6&K<&!yf
znrRjDdXqV&Nf~1gXC!h~@BDZr$18yo+YsTIbT*MAl6q2b+T#>XNXCX2`sqm=6(QGY
zn^&c9eBBnxzxPhz^qx7l)B$*TCO0lWe+q{)=WDEkG3e<_=m=bs=WuRomi6=Lq;T|=
z*2;Nb{Kz@2d*Wu}uO!aIQXl=jGm|(aO4r)T9g;cieiMc4HBvcIQyz2Uqf$Agu4Ru#
zvpJkn!w&B^zd4*6Ed^X|8Hcm-WWmhA+7u4yUA1|Na2m(7XQc9CN&=_Wt3l(KOcG~~
zOTO_IqRRnsXcR#vH;f;x%mYk*f;=QiDO=uhX|zb>bqcM0;3&A%T;DQ9pZ
zIri!2s!}=btDK{nwxn|utH*o)Ez0C9td2jkMkb3>UNmvRFIaL;Zd41jKoQ)qY3Yd{JPH~p&Zm;>d
zoO<~^PdnlYIhhrU-+v9L=ENK>i#}Y|K;HZxzGdZCEP@i%`=|f^_^*A`|LMp5pZxm&
z%Cd|
zr9GoZY#EGTWodStB3AZvWa?#)Gg$A!U?tmb6D#kuV#;@JXR!YygE#(T(9E4dI~4|Z
zzGCo?2ZPTo8B{vW;L?K(p8Q=-_tU@m*|K(s?_uh@Qd#}rpHy!>JH83KPHAj^N;OP<
zry!&6#KZL4xPg%s_Az*5+%MYC=AZ_O-#;h{
zR*E<3epB3h^drT{aU&G33cRG~z2gkUs_V>r#`b)n@-x24u(caV{#OMmJMF(7i}JHY
z%~YONmql^uxH^iT2NzKN)mC$;o@HPyEnoYjlcG#mBE=_4v#H*;6^gKh8^{iySG3&6
z=rhIRE{!l{8B^{Z0F6cf*>qV}X)B2rti#Vtkrs+WSo_!9c
zI3U6Sm(V~G#y5kDXdng3FIFR-)_Dp?wSkBm1%eA`AUfl
zwqKDZ4M)6zmE-%d92RJNxt^)Ol#6wjq4~VF@IR#kxV^Wj}U{=<|S=gW4u_tT%00qcI
z3)HjvJ+}mL!GVNZXqSR*e-q-q)ve=EKcMM}3EDGW>+M0rMVGq_uv~fR<7Cv+FrP)A
z*Mv=eYS@pDxc?Qj<9F8XG?Z5gRUSsYm@nE#5p~wAY&r-iD^seE{j6DUP>rY{7O#Q*
zJ1nVjLriGA_tXba>B5G5U%-K<3b`i%1=jwuM19wInQ>IFdeUT+Pg`!I>uRBLl&%S0LPQCJuuBY&yHfqnRntgPC3uZ^sb0Qt81tGhE
zBxg8&L~K6PN6*zF1uJTo%7R|nUWmJww#)r=Fa-5@-_D@*9T&9ec@p@xhW58$@#q*_
zzlxNTy{IP#|0e)Z@J=*@@&>Z>%^9i>-x1OMW>M#O4fV^%%wLX}d0zWG>Rp%O^+T+U
z52yWJZ(dF@CMk*9Q8&c+r}x}yD$l>}OR>IX5!DMEVEkI>&hJcpawEm1ayk^3?Oj4q
z>^#%Yo2`uh)ExhbmWO}JpwD~BfClwXI=6Pva_hk^itk_Z(QzIR_&-8B7c2;){)Ma5
zP5q~Xo+@28o4d(W&wd#*PO%OXKWy2!aD~W
z63vC3L=HQW(28NHgWZ?0VPsGV%)6PIcC<
zK^CvF_>XrAv-WGTXmG#eeNRSv&td
z*MId8oT=Ww_6qF!t@uCvf9v*x)$d_>UpT9OnbntK>x)_Z$JhV${|{fX_26{q`2Xey
zhmy*0*eL#6hktg$CZ^^8>`!Io$1L7u5u7#E1E)mspB?}DQ)2CPWb21nG-ol3t^dqo
zI$Qn^Z?p1Iw*HtMGtL~g{1=NrLg@IZ>^kwW&o`KDcZjX8VX>1f=l5dT#j)*1*!q9#
z2TqpuYb(UyPxd?wvF%r|`)~hxrvA(T;|~Jayug&*k0*{X_0p`L8f4r5fj?S|{_n~q
z28;C=3`l0s_$GrB-Z1z=gTXWJ7%Z=6#=AU?kw;4zIi-=2O}{X*mPJ@Ki=W
zZYkpjxoz3BKQ$GGucuyR`uTi^LGeI_S2s>$c(mVsrrq&$roTr!nReoliL@WL*K=un
z@PU6ykL#zI!sHv_kC=Y0>9o@Nco$RZkIR02qk6#_On&g#o3Yo_)ryvnui~e1=T9+Z1#N%fmW~bT`Dx9i{HiI3SvUE^N9g>U9^_P`=0Cn#+fJCc8(>j{&~q8(N5Zp6eecVLt}f9)w~2xa=_{j)0~a
z59ZhcT1o$m*aIlq_pYhJy43V5=9=>;3WBHg;L
zsPCZtdT0yal=DlIivZg<*oh|rej9A{LH-|5rLi9SOMBh%!VFMHY&D-dpuvvs<8A_8
z$-gg$_U2gHl;L>e%=*pG0BXNp{^~N|7c2Fn`vFhKKWfALA)0e-3%1{4&rkCWC5>J2
ziXiu^#qBf#RBGm{@di}9TW^f>7T}8!n*gZrE!cB>IpjkV@-#w$c5&jRL{+c~5D_ywtMxK5_u*WCFDcD{GmJNXab
zl1?i{FDQ?cOn!X<@YL9C&i?^+Jjm;z?G9x~;XH~953fPo(RPBK!!rFjzG&a6*WITP
zw?*wahxT0G>vs?3uj2*>@p*JympcEH>2XI|nt>Cf4C
z3HSNyi??5)y@48*KowwLaxYQ8!lyf%`aR)z8R~D$JGWCmCOnhzOS(-9fI`5&SpIZ(2LakHP?O1XZaFjYuVJ|`_5WAJR#3lx?1l{0v$*R`k>}}7+Fq}(
znBtqiOkAxx&!FfG28*&7>{!R}0gY|Uyflt5c6aeH^+wH1-0q*t#9M_v=AN!tp3&F&
z!02;Z8U5{sSLwL26Pa}|{=(#)I!Bpxw5?^zNBiZd9V7sA*vYRljUww~t{BYUvCD~9E7)ofn0(~hb4HD|ED
zgLy8~KeOvm%&fyZHV>0z*R$W3;V%cJnR$x+Wc2D-p2q)GKH-la0esIkw#>yH1TK
znDQ1!1~b`l&#?X8y~LF7%4gQilU>&s8>aje`y7?GGID1YbI#5#VB}5L8SG*6fQODI#HeTu1nD!}*=gFB2Lk~0`X{Y*#`GtPKFX8eF_%=p4{
zm~-R!N08bVt}#gQO6q%xCrzbjdm_Y~fBP?lmh+y`KA1%
zakRa}Ge)oSTLP{3-}jo9TZPC%O4>j+K4JVv`PevG@8hRH>u-hppz$DW^&3c;Azs^l
zqq3=a2(1qhs;9DnhzxBPQzco3dM$6%==dMUs?)f0^7T>bf78o!;FVSb`EY&~z5mwG
zx=DF|gRT$N?=2{zJltF?nZ_}`26M`zPoK4;aVnD&L;1hyk~ms_^Idfa_GfcHnznl+
za`PF=0`jXe5w}-(gd#Q%2+_Eerg6I$<#9XwJ|XrusndJJfwUl+UkEQ9sl)Q?Cchkz
zHyrR9-HY+sHRxD3%HJoYe?>X-LgX5h4;?v9`NEAqe<|M>9S);;!`l#lI{u3&Z#bF_
z#I`eGAM&3^8WsE~dwuGl>v`qZt1ukz#>k&sM6G>>^2jST?cd^o^6MG3biS^uzUQI5
zbD`}e#HxHxapWWGI?7I=9n*^*jl+K3z4*5w8U&^>?+s>ESKbD>qeF!s`G@7jK@Dtw
z?eCN}M5E8gmtj9!p1K#U1DvC)w;uV;plFusUck#QqaR@Z<2S*No&oDa&xfuDG?hx!
zJq>v8-t2nhqm$+h@I3?U8@+pN5#TEw>06P&Rmeyjg$k0bK!(wiMt{D2l)dqfHVi}5R$g4HmKu^p2n8`iRdnxcr<^%MT^Kg^)
z3Ba2R&xw{pxo-c>7zxm8+&pnt4501d9GeKJPpO@iCs_*bmktcSOeul)AKym!$71=J
zt25>SR@LZuo`LZcmTn$@0sDKvA1GJ_d}-;Ky9c0OjZ154zW#vsKT`}Ie6E9aXq+`T
z($GNa4sUUfh5ptGyt?370Qr?o{v2NzM|4$K#K}MKKIUv>Zc`Sl)8drbyvdNST3TM7
z0d}PyfBQiNaJbL-MLhIxHE-|d0YKyD&osTEd~DiS2|?(`yWqk1MSxF53YX?W{q7pw
zH_=dEc;lGaU2Gq4ytKXo-dp?*+8+h$cA)!=gDm7bk2uXdSD}9L^x0kUfYvWYlY3!a
z>dPh8xq-d?o||{%LcjZpPN$ngzgOYU6`F%Qke;ERkq*yKA)xAO3hZkgL5|J026A6f
zy*3o;H_N_Lx(59OMH#1IK6mtayQ?N(w$MJK9e}11_7i;obx-kK#(lrEJIM4EU|xw(
zz6qdc{^w>TK>n9b^u5sMk!fEsZ#26$`YI4Gy!}qe8t6a%%d3()kZU&f7~=lin=sW_
z0nm2${;!w8j{14~rsv>1j=qJfIpFI}t`(c0UG(=oU7tbj7s<3pf_2!d@b;-GJXgPG
zPgh9;e)-G0>@w&n8-FU&KhX$gRnoKq(*1T;H!^`kDJe!f=(p2PP%Q-5Nfy`+7XNjcal>1dX?9qfK>
z)4zKXl&|wCk<$az5vUx8`dO1!I%2+4{r>AUT#u{U=XoJk9~lNp0{O4$X6moE-Qa=CSTa09<=_iC4(AA$;$%sN
z|0jN9c>eM~hpC-w4>J71>mif>$W892`gKDLKk;dcq;}4KzL>6qlOpq8sYduI)w_In
zEG-xIdPL`UDEbK1-x9r)mM3X3@n6!Dhw57{H3JF@{>FxRpY>*j4%PQ;e?{YVRp4@}
zSAT?g|KIh3;WrAa7<9kF@RQ5)8F{h>^F6_WOy+rIyH1!O_x42sv-gjz82YsFa$C-QD&vDH9)qiAo%>wp&gj9h#n&+-8P@wk41&Y!B
zuTD;*GW?x7imB}TsiQtj|7trKENA)s7B=6IVBe!`WA9_vvw6mOmLJb!zazVW
zPY`3r7NzY0G$rROf{O9ZX2aO$DBAf4gw
zyWiPTz5jge;gwATS@`lXt?v(G{9XO7bXtGp#{l)qlh-CwzbNT0#`G8VliJ~C^@N`D
zn7^*npNVJN(C;lG2IkZ8xBX$pZ{0hQp1-VwY4m>ks+2C}<#j81seh?IvV!)b7QTVT
zgS9E<3erQXX!w{2ukgXV&>szv}QhlYWQM{_;EZuODS1)3BXj
zxfGGB`L`$5fPGVwT0P(f-i
z-p3cumH&$OhaCAo+pykXwebp!n@M-OQ&E<-dE&bZaOKFYz_)MFFuysNnz?c6v&(G0&E`gP>GD&EVwrQg*?UX$zo
z#u#z###s8DP>b8(N#rG?%?kvPU!^WqTW>n-cI4BV&6#d61tqhV|ilf``gFBBr2sc2ari0Pq#z>yKVK0oLT^(e*03ur$vF
z^c<~=u89F(9W5AI6jlgvb4=;br2GcbD`O>b4(iE4Pq$;xzwWu5tA~N-ja|)K5e>Yp
z_{C%O!`1k{<@4}K94E=KBPEdBCgAATJ5Vt2C{wfV@Fr
zdEBG69=Nw}m_Gk+LIZgqV}I`8Ti_A!XR>WzU2J-e^LYV(y;Nvu{tC(;PjqqGQNtg<&w#bW2UVTZ6Y%UUyb2DZzRX2TBpCC
z(?~>T?9vc(=Mvwh{4eirH<9s6$ICifG?UDRq2~*iHNxN7pWgf5sYa5p;`inpu-8e(
zX5D$nM`W(mgeZa?*Y6OZC41ML{kaC#f7s&wtoAQ3UiZa&
zBN~ll;mpIw8xpxhf9AzLu}NIWYX*ND@_}{wc~oVU74YJ%{#rW2T#_{9epkwIF8SW3
z{74zvN4kDm;iw9Edv0<{U|&6H|IbN#0@%|oe12IcJQvBj9ykA@2GTm>Cp8|*g9WCq
zmIAwPtvUWz0qlG!n(y3N1K&SY_fBkt`Pr|#)9nxQ*6{9nXa@3`)UuA;&#pkOuN9vE7t8wvP|kPSh=c22
zT-19{0PMLew0jKZQLQe9>S(Y18Ksx#FSpulUy6RosaX0G`mvpITlq0BI=*U4Am&-B
z()(r5&vFA@qEP?mHH0kfc`Txc}}DUU{zSjWQ+&%U`hb?f576vAljW#7oUf?S?<|&#K?{?8V{$enx2CG*tRrC70Zw0+tN5H
z+&%U#%2&AW0}$nmZaqOi?(@L84`t~MD+Lh^Y_}}J`AU?A)A$!0P?wEyV(EoGSBwi?
zExlS;uk-S{A=;ywyRa8!cZ>2eIY8cy<6rRo+3&>8Dw`vM>tp<&|2f(fw_B+c
z$2r9_k>0b_-I_x4DW_QGyR%!L_EA2t^!1iUs2AGWN%cNMee#pQ5cu?cOf-lIjg@enR8Fac9qMe4b+uq>rMW&AxhizxBqpG8bjFmY0-2
zcrWUr@5PU8Gu1==FWYO~vHsnz@lO#)Z#PcF`3kLZqr9d?Xj(Lux2yC~+&_uoXPKt&
zsC;mNA?N-!w4a8DDyvN#$`GA1OZ)bswhvzIiuD%cJhzr-
zz1i<$_RnJK<6biJ*P9hf?HQI~=I5x-%vY;`;rqqQn0_vnTTpvnw=;ac@GrB@#V?qC
ze(hj*dSLKDs(0)NQ-69T!}t3`nSBtlfVppfd*~PC_pvXTdjjqK%(w^48EjeQPwmoL
zewFr@9?G1X4eOcveN&-Bv_7YMKixOSo7Pi4F5bb6KahBV*5}1D>n!wcEY-7l&ZXy(
z|Ev@3*IbHukMQS{1g)>1zKias{HHq};CWDfvX$<`{@c^(e!D(SpYFpQ#-;QgV@ID!
z7wT>5Rit{{7jd*79eYRG&r#JkpRnFXd7n3;PN@McH<0J2{k&~Sp}c&cPwgDm=k(w2
zMSm3LVtxnZHZAfQQTa(Kz4y-OTz(qm*+;jV;l0^srvhJ;mn|)*Mx44c%mp!Kq0kmQ
zH}}JqccLH9+Y~|VI9+;W57r-_$Zvu5%7*^aalCLXKQFx3E}Q#>^7=@=RQkT)+=IFr
z)L*FEdI$a9eV1Ybl%J)q-GTU`cs~#Nm&Nr~JFxzm?y1WdpRAMwMy>;1F&lRp$3J&V
zcs-7z|Ked5wzIr6*$4Ua0U4VDjGv!xsIEr8d@b5X0>7uo(zIKQe%D~a`-A9*Ej`0U
zUPD}Q5q%bh`9Q$h`>J@)_03m|7ybHs>&4UYJCF^W*@CMf4n?<3YsdLqc1ziX_vB%X
zsWnD`Wr44wVgZNzb}UQ-%zY)z;|W-+V!BTY(9gj-FA{KTzzcN|z@856qfdeVIMmcX#
zI<@bG_S0v%qJzS}xSX`}{odx;T*5c7U9A!7eLNSh;co`MASl}-n+SaD&9~Tb%elnlIR80^b}rGK
zu{kydc=(6pce{~yJibw6!^FKai@bl(BGw3etC_+1J9+;
z9M}!A@mj&ShsMA+TBp1)u;CKZ^-D#&fsYlOviv&)eCX*IPUhVxxSzBcc@Ut)B~@Fq
zVw#+qNxrO7JD&{j(r*uB_#t0Vn7{CXX>0>I@oT-(KJ!-6_FKm~eRVsz^C-gKA+3S@
zxu&-}cc7lAZ?j)>f24^BbeyjmROlovaaI2X3^o(D=92H?=Ya=o-(9dMtBzc%J^=q9
zy`9XMa_foW!geCRG}~kf@c7M@zXp~J)sw8jt@^9d>q%=uRCKK{%zt(Ff1e>g*nK8g
zCIp_FQ3Tg7Cb6C8^QM`UaK|>0Pp|)XRvfaY8_Dw
zow>f^S{>PU^<&05tvaF+Vjt%=y`G$w+naQyx}ID-XCosrR7*NW7Y(c`sU_=5tfyOz
zuO;3ww|1V~T}zHm;Q3v(xsIF|40#fuQcDizZV1sattI~ROO{4&ts%9u7iVhN*N{ue
z8_eZ4)RJos2j8`I0e{)P&~yXX%W0iGddCH7&4`PTNAWg?R`2)=??e1A
z2k(Y`ml@+fc;P$jj}N9>qlN28T8Yp;5>ij9G+dv&n!zR4J0n*b0V42y!@46
z_V7=+S~AksT<7wtj@&k}btAx&_0=9keU{@A<8O|WPr-h@r9?QnU}xBnaQO$YpQC!u
z4)f@>c5}Wb0tOYGxQO@p6`!wJaUdVm8fNTc4y>rFgg0QLSG
zy*Uj1hB&1uZUX&>9-|83psyygr8peUr9tYcdIgY2f7F+xKL$P>TJ+Ew=2a`dFiQu{
z!7|eqPtL;m*8f@CSP1Kr6w}=*0`s}4_eet|4tU}(_y^sP&kXrRN=Ad-JjXPR!r*;Y
z%K793IT(NCwH2erfRnc+y+S|r_UeB<$R`c@)N9e-mD#3*B>@_$lG$4U+m2aIMt^Ek
z-?tR=dcFzIk077Fl5^cB8*v
z?UDWv%lA92T!wzXEO8R$d##c!EAf8BcjR3m`tx%yPyUDXXEN-vQD61^2R&@xcqdQt
zBcR@zkEQ6xr8}P$V%{Xn`Jz5CH7oK3+?qRj^9G>4;#+&{xAj07%^N0E
z^Zms5^3ZGyeZR~%Y)0)rTeY3qBi!IZ@7WS$%_uL~VY9sj?a7Y5NqK(m%nHg^`);(*
z_`dFrGi|>yD(x>mCxc~C4^cm`EVux%+}J)8(R2Dl%HuyNYTm~AXrx`DzZV;5|Iv>1
z!3Rv}d!e`A&1oF(uX_8YoNo%$)=e%5y}TE>iuy
z4L+17eJL8IdT%}z(s>SSC`?9sWGtC`bq%vVs^2P;PuFMCaNJ3(|GL|O@*U~NlC)pz
z_M3Elt;&?Co~13rC*{%@-p4z_@X*Wx#!l0~sdRjf3G*HK%27tIr0e9R4X#$9>*XuUN9UKN
z%)AfMwqe@4rtYSC`TC<&Z~Pq%I`8&Rt7$)b$WdC~r{_gwfq8*+zQu*zl)nvryh6)8
zlbG-Ae%mv=HY&oG=`Z%@U0k=c9Ok~7cLsBxOthQm`U(9!M#uB@N|i^wS}$iRk8fjs
z*Co>(myPATGp8OvtRHhZ1oxeFva8@N
zp@w>TeQI*peuG5pW|a3=$O|IM2h~nH1^C5T!UN@p_m>)?e{tL4a
zwyF35TDYpw_rME&PTGWeysr(Xqn@|HuL$%H_K%G#d;qoQHK*e_ntMd+75dk2MaE6o
z&z8di3wHtbHUD%=2miLff6_4avsV2b<*z=H(L-o=zV;C}yw~V(ZcF_EdeeI1zGHj$
zn{Bo@p5Tj-{pi1IAKa)50rcP3^+yFTXRo0a`e8m%$KG{-bqjyq4g(Chcz+uDQRxOj
zr(B31v-evGqg|c_#{!lB4jC!(Her33^_)d8-U&^)V-65Meyb&0C4$_aI@4t(_~nQ1
zv&Wi%eE#Z~%fo=8F(>6VK)iP^UgLz{8Egv8E;s{vqn(pp#6ethGcK7G2YkzZvYYs-
ze(;ayb>%L>IG2)Nd_!JJ{&w!LfOr*Hkx(oJ`eu##Ee#pqcUO!Z_Zi~HhO4C~qM^U0
zzGJ~tKyUi&Vx1u1H{7Bk&OV46ejc@Z9)X=oWm`R?AWkOdiTs$A1NSl?Q_rK_wKI;-
zd~}0L9_{w;sON&8{M;Dp2>n{U^a;HQ@px?&Z}8wLE-^d#<@W`3_}*bpNrx}YM|bZR
zOQ8g~pRg~9t{cZC;jiPndEkCIAuA}y-K&m>I83E{TBp-TfCuafNm`K+&TJD(4N>WI0abk^tHAQ?g{f&_|B|6(n{|AiZ`-U>LIJfrFtxU
z#3dWjH5^@n;J(S)+=DN*f&BeCcq`{{C#i0|d)Zo~i_HFOxl+8Oo@{c?o>e%zo=gg|
z_^vQg(}Gg&u!PmwIFC!Qi}{AcHJ$u+%v&Xnj*(irw(nXzFz
z(G@YPuF`BHxfOSQ+P!EYGTt*1Y8^WX&v|R_(BMwuxl#0lvRnu8am{tBAM7Bani3h^
zJG)3z_dc)RS)F9R_+`zuqn)Hn>FS~T3p+`AR?wp}x$UGqAflkr2F95({*mH?PQn)@
zJ*<-5NyNVRxz^eDS6@M#hxG6Ep06Ow#LZWU4p)=VyU+Y0OR7oNZk3E3TdGLOy@eMy
z-K>DW6WMk6fn5cezRqXyCg77FdRiFKL~cC6JmUJxx?g#a4~Zzr
z<+fFm@q*lg8OLjhVC;nKCfFx(ekBRs-MJb?Ybcgh4lyXXq?IM_Mje0c@=FE;v)1I%mg^F={&di6y5?2E_Vhv8me5&VB2
z*zX%ws6JT>I5A(XX*uLw_Pi>uR{{_3c&|Kn54<6FQ_(*aZDg?6h0w<`Cr${#yH`$_v34n<+EC)
z@IHLUZ?RT=z-rr*kr1m!eww`#
z?dI{gMDL3$RJiS3uDyDh`GEYaA&gL?B-OTDptN`NoDN41;g8HaMr
zvxSuZ1bpXGUiHByh1&VEi}?=1w5pTxvB@bXD4s4fp#0=hh7Emxo*pvyC-$>?`*zx|
z@KYDs-^-0=l>cq7;nBqLz4mUT_v`h)#OZp7h6GW5pY!V!y(ioHuEqoBdm>Ml+8I1!
zKfR9|zdH9W_FpR{MfAXKGX4Cmq-k(n8
zt|`p-5u2v1r2FFD+26FkP5NO}PkZ(=s<+bKl&(*q!?i0opTnbtbp4**enamuTMEz7
zdrJEZ8M+UIUmvkTy|-1GL5RJFl4!qDL(kGtUYz_%1+msvQXTh&t#Qd~)`r@3XM~KdLcj@Z28#)_MoebE3i9-B$oJ
zKdf`b`X;YkW!S#9N=Om)Ha&P3=?*yIR%HPCi(4YkQt-TGy{e|)e`qcH4}PK#_nKDK
zfp~s9Vy*wf_i5J3#sN6~7ywn|gVK3AhpZ3eW
z0RHxMWJA&hDEDj}9>V)ZuC$S1H=yO}k<$;r|FwL2nbHo@DQ;zsN6bKI|)~hkB{gquvkT
zdjSb$2WPR?;muE
z>plkYTU>DY$7IkizV`dC_6aW8IrgNyH`r0QVbKhI=>N)>q|O6b(2vFZDc|5dzxSG$
z=ZYV>#Iu5b_6vx2Ar+bJGH{=`ApS{MAH1LU8JWLxsx+5;Ie11vG>c2(r@tKS1l}N4
zo6}+k_n0vy#Ve+!0q_5Nw)*A{F0o2pDEb@vDVwOoD-ZXWhb~>slLwyrSpE5qe6Z_O
z#kS44ySQYERnOVI#`PpbnU@$m=8}X{R}8WOV7~1ZVFA$p(}v)sI&(XTyU^;Ymixfx
zjxK*Kcn^4I>G;d(i-C`bJgv_C$|X_?H%ktPcM(}@1OB?uPU3O>o~ssoFRb6CmC;gD
zPacI|eznr3nGngs@00Vo$ksN$Q!btD#Mp>$k9|iyv1rfz`d@K9*~4#icSnCaxtUjI
zx9w9WDVyMP>)A5Ui{IaRU<&Mqk4t+yyX(o!B(?r^QeA{ENY%i4Sts%R+GG`Wu7miG
z1U=*8Wkhsvvru5B}WgswOiHd-?ly>&gB7x*v|az<%5tbUvZEf~>aZH0=R?
z_UdQ$>E3bj`LKKKZ4$=eD^BIdD$FgXX{GhxLL(4>}xequ-227+*eDU
zpWd#sd|D-0v2=(3o7M`lZN`;~HKi3KDl#F?k*Atyja6(GIbK8L6TjSWISb!^EV58I
zEy5*3pPKa!^um4a9pBIx;4g|Mn(5}S?|R&=#&SpCes0dvV5n*iSqC_l`UN4bFMY*E#YWh+{qv
zr2j0bC5!eY`u57ylUtA1Uz<9WOBQt*g;c}%&sLUHabv+xP8i($!Kj94+Pt4PPO*+G
zbA8}C5BBrxugRHf<+-Gzp!&vH;4Okr%~H0)zTYOk^35FJFTZ^yEIz>b(S1|?jt|b)
zkp;O6bm4p{qy!J=f}K2*ES{`^{Ty-1Yxe+@2df>BvxNPs-r{-$f8SdhP+4UG_n2p%
z-@UE~yu-TMymlAdk5)cXd6)<1S3ug=vZKhET
z<1ZAcnN<$#J1E~tSjDx4h!vM}f3c&wKxp5lkR@R5Wk{oVI4$)I{S40+uPp!}DSfUO5$
z`22u#c<-7mbg~M0sa*VP3+&fJzi=(`*X0W@XWjthQ4igN<13Arj6wTcp2aC(d|4_!
zJr?oDwD#v*^-(2_rvBmrt2Qggx!6?Qv&BSgRPnRrt+KJ`!{u{Z47&UJf
z{&fRd4|zgF6A#b_v=yq
z(qAo0?;#bW<7qoB+aFXd2p(tEtS6K+sm
zm7*d^`}6OKrSC)M?paC4Ynjaa?wI#ZOD48=TANMnpQ)i(hjtftMbdfkRGg)}$!?qp
z)mvBUX@UCsFF9ur^|JqZA;xOP(tGH~#_86DX(1>v}iGoS1=hee60J*`D)}Zk~$9j$g4L0T4=Tn&{eHv81uPhznoMs|JamsM+f(f|IDxDXph(Q
zJv^vaTs_0$E9h0P?V&vI&bmpI*Uqim9EAJ1w2pW9HsI4Q`sSYiGhM%qqJJoUW-`M7
zu;t>rZFc}e`n8tex%ho#WBU#8TaveQ>36Uqm7cBYQ2u_saDO)7E}I)x_&wOdab10S
zMeutRvB%nuATL@UbnY4Wfvg|AsrY-Hlq-!D7r<|J@;r<49)|nQMF$1&cU4
z`;QAz4djQxVRQUlm5;2Bq&PCd*O#*}5{s_Q4&Ofp^SFF-T##QQwpCVi`{?+5$`w)Na
z1UL76S`GKU%47Qt+*?VMj%Kt`Viyq@682Yw`#=RZu@<{K;BRtI%%AuN{C~>SE}?{W
z@+0v-ebv7mB=X?gWm|XDlO=-t>Msp5&sC1Is2Soqw`Q1rQM4Z@do6|=ihr!s*}7n`CxPSRR^)tcpEBR(LoMP7-wra
zp_8n&7(diD&_QGmo|_Rr)J|3%yt37Ke+SX?I-M(Yz60)!1WQLsJ4jK_Uc3KVJII~4
zrLUreE6K7wT_1$HE67>D)o;3iXQlU>i1$yeCi8y!+!^myLt;JlSQ_zG62n%{au4N7
z5=XrCcD=43n%dhgUDT=~N-cMK7w@PhInq9I=Q0JU+bfCe$4fb;Pb!JUI6Gy>AOK1MAAwS2+oS!WZ{%M?@+Q$`K
zqRHWTJjlx>d$h|w7yqh+yrXBO-?(aWXJq<+=|}2_@Z*c`)Zjd*-T%OMb{Uu0+t}Xy
z)?Z1S8%pj4OsFCv9lMMCl50q>%|5q$;K`9f^Npgu!hMBu)r8I~Rb+2QbLRF*RYY{&
zaig_!;r;pV=THALswd+;B(~Qag7?YicrE;Z=Q-vb-MK0e{F6a!WHG)k-aheKB=DC|
zh0wP#z!Uu+-+e4Q3h`U|Rg+69+?RiPFk2CLMVy1MR5S1$@5%D>q`}S@`v%S-xQELv
z@0-1~5Ar)rXNe~G4pzoQe~2IQ31{Er&#sW?{B?HH{tn|D)VRLGbtad5`)jE@2=-60?1i*O`XS~U*DgfR-=ATVRAL{AkL@X
zno~i~N@neeJ>Z9k#saZrpeJu67c>LvZBDA(a{~Q(IiU;a*A#Xh)N@z|vbB^Y1VE(LTtx
zHkf9+$HVx0{T{nvKKoR)kZc5eFle?8->t&uiX$$HHo7d{rU^6d4_oO{anhg?)xj#@7x@mjSaE<
z+Zj)3>^I4(oW3`Fa5S0ns`o!L={?!}(P?_9=XQHs9_k67d~zSs`8_r>p|XYxZfKm)J7v}7ndvJK9W2d
zNPnjRsRC=&%Ry
z@<0ayV~|xJfCfSy=$<3e$KHl+;{vd^)%7|58Hg5
zfq6?$_U=V#aK6QL7v87Z4JN|EAA?n0jm88Ovm4e^FI7wdkXZ53y;mm_m-1e
z`ZwbDaN9>Zyz2lZydG<7?%PboxwA2Bj_nz3-jLt*d|~SY7YHg3>*Jn
z9r(eQ6I~tmf&Utfj|^#ta-D-(UZ=p1UL0!EJqi6@PGU^-U$4Wb?JGrLhz#vgY&mM9t8h+BR0t0mrHngB?q;@
zuNT~$HR5AeqkofGgLDOudaZVKo{CcRA11-+}s
zt-FgMUIks>V|~JnOHLk-6!~$4OA-e?B0XM!|Fb?XaSq;R-gMkit?1EC+7kJ`@A=Y6
zO2@YMtrCUg+hy7#FjNgd?Fh>E)_@TRQMJwes#
zz)!~vmIlZ+65r}~e6rg+NXzRaF5ldClBO$hhaY|yvh)3(S<@h%F8MR{+3jnM#OdS?
z*96%P65Du3*sUrFsJXOTqUP4h$s)j`7Z{jT2R!_FsK3Z;b4BorktyMVoql(BzpZcb^
zp^7A~OOJkDTTP^oD!V3D*O4c84Gn~6!~W3l?RlqBP4p+8-}s`qiu^w9$GMqSO=N91
zYEM=JzTg-2^Zd1XGD9Y)aQCHZ;${7=e!!@jl=%wPoX@W&)#^4+X5`lr_d{<2Idb)6
z;-kilj-+bx)hoCo;6gR=&Z%y>Qc+EgS?{hdj;bZ%TjaPFW9y0Nix1sP-C+Oo7B4sA
z;gaE@Ny@RnfBh{u=6Mi@B?lB+rC>ktIIOsLR2K5JkNun&*hg<8z{=5{aO8=;j(&DUUj
zL$mkOi;TdJRZqLTY6_REj8VI_8~Ukv{7QN(^gGgKs&@Wo9SP6k_FHN|9vs~9z5(>I
z^5dgoVgDO1)N`2y{JQ4LD5oFJX|qG^@MnmZw=EusZh`%6yOPglJCrw`5Qs2_IFzi_
zeq9LeEoSD{-yTOJIDi
z#k$F>I1slI>_x7D>|j({?hbmFY#T040X>t#6ifWO5(2OJU5a7+`#WnRFi(rTvS=du
znen^6gkhZ2d9>&t`Zs~b8jqKtccjTTXAtDfqG6*ofLrsfilKkzyBOAq{`p4e1QSWn
zt2{B-i@!hN{W>FUJ(Nq!`I}-F4mkw|8B>QY|U%PD+WTBTtYvtHc9U!#+6T&1u7^XwKOlm
zeuJ+$UBY-Uv7J91dCQPQf(yordxL!Rd%B*X^Q(hFE)7~}j^jjY?k&Z6Y9H17jCw}r
z8WOO-DRz#&p@7Ei7xHnOKM#ZVp`Pl4x$USoVan_-)U%bFOygz2(I+%sZuIY>@%oHt
zIDL
zAAL{dBlu@0>TPR%dIHC>N^U-9;7=yO{ux3dxo=OB9Ky&|)nVsZc_Zq)^COcw6oz`x2GP_B}JN`}yPd
z`{zC%Gjm=uXXd_U&U2R68P8+H4xhdunu_I4n>mE20J`+`k@Ha7Y*qK;a@WQfox}2;
z1D}h10du&uTCp9{xpww6u1}6vi89vr?$4@V{6GKdwiLwtf#suk^##B`8}E#51-!$h
z+NTE?JhWaP+spl96H72Z5BhOR*B>W8DWB+@0Nl_nAU#N<;8|&R&T*K5nqrZ0gtL
zo<;BIj^F9`;jD*y4^;w_jlg>2*`)=s_QIz
zh~-#zob5cv(T@a@yr-_Y^&_qb@zeW3E_Bc1&LxoFzWFT;(=g7~2q;udu7dOEmC08t
zUHZ{*MQG?0^xMAW_afxL?w?gUIW+`wdiB$4@1R_Ml{;JShxMbmz@ir^ps(Kx6!o07Uhxt3oj!f7OaI9S-`%=8$cqdQAa)ggiTLH?X#0;}*S!ix(7!WL
z@^Mh!=l5cf3d@17o(Q=ew_y;mGB^Ct92-L)nS|7`*hY|=yzGNy?tY|ekoNr)*y(ke
ze5}?_22hPjPk|Zi^W52|C#@(jf|No_fBEi#?>y6Tm#FQAeTjsRPwp!ZAf9!tt$UY^
zBX!fGIeiPmXs^Oni9KL1pI!b?WqPFtsT^2t+|b>F;yQsJqQ5nk`=_#Eu2XKe36TMv>e8Q`5xqoalvrv4>=dy$^waOhD-8tSDzu)GB4Su*~)
zd+ciJL9@U4{jIX;=()z(3Wv3QXuWb>?Ybr!8d#Fqqs&1=TE_Oeu^)QSQyZ!4+vniC
zyU)6{aUo!*pY)v3?4_Y4odXZq`+9&k@Cxm3=|P-YL(zwi^djK_Tj{s;edz1zbaCHr
zy{Njs);E3^%%`kF2WG)u?qHM2Y65%7n;kRG4)$`UO7{BHmtOQ@X6*GLd9dq)Lc3c)
zZuyg|n#+Ob2Gft3)Pwz#6JR}m?RPJF-6#J@LL2OkHLwiJf3of_w0aUR~$ds%lbO?-XQ3wzFKUCA<$o5tP!ik=Xoqk
zR>)$1O(R^Y>gSZjb=zxmNURBVhV${(T9MTji~6g~vI&SsxSs{wf7lmylyJ|+u$4rAg^8KPCUQSe)XaNkMC2{Rz_C=o#M1!Vte_A
z^-nb(r~3jf6MNZTA3Bch6<4IdI3CwedL+C~2YflQVi?CimVLPr+tq_Q$Juax>l;JX
zV>@t1xx3^cpe^^59ghE3+3*G1|CQCQk~kdxc-vMiw@KYR2HXAjQd`J+e9>vYKy3GG
z1bwV<{+v$A<+$9V`E&eO&vs@XQJmf)E=%@@Zoh3v!uocFTND3lIVv{9UKY;$A@(wI
zxSQC^X>L(dT%J?cy>r<9EaogD^Y&C21do0()?j(xQK$`NG!p&Mi
z_CGYQPQAkMwnqZU{>PK`JI~|rme)>%H#l=El5!tfo0D)vvmpsjac!u^a*lNjKQfE;Fu{}hexwe&r59m~Eal!S`(;?Z*X`&OZXY|HC6_QiyK<}SehlEfi;Kv)pb?hi->@F@PO*0}{<>WpjoY`}
zJzN;?Z*azPHsko0=6}e#M9N`DB0pETlJJwZ^|r3~e=JhEK+gSmFXke6?=<}ye(teZ
z03Ek)Q>u|6&R3-;Jmv_XuEyzHJU@SV%4bn0;I~0D%PicpehZkPwynk@Nt6#7G{Pvi^SUR#bdq+!%v
z_;lL?aP_WU9z4IV`n*OQhkX(+73>3j4KqfRRAIj%HSJ5lao7iz?mjO3XaV@;!PON4
zU@xy9e0*;a)Mul@g{WDu=h+oAzr10-W8~xaaU9CK@bBaD2QW`AO*UPY2l>ua#Y|y8
z6bF7k>Ua&~3`Z%8lq-y*j_fAkxp}~A+Ipt&yoU9LToc}hPG3LgbOWB-^{C3Q4&?2p
z);;Bcah3H4`yp?L|DyUNQXAwWLSJzBLB4KFL(HAQ-iD2PcNK#^(ueQ5duzZt^9k2C
zhc>{vUV@s-1N#AVz@*cpA#M!SoZn-zqHY-F`bGweI`$*~s&%e>GZO0?w`#DM8
zDxnz5G1RQn;ccEaihhbLx_BJ+NoKA%c_a(>qq_DG1+U{UeyCL#7A6j&Gpn-WW
zMlnYnQ?NUpj|^^m!1({c;Qra?Fdxk~_8-zZ1bnPWSGf@UNbGTZqFe>_sM&f@z3DaZ
ztT)|=1$aQ1iSm)#hOkfbZcwi6Q6DOd9vX7w9!0-4M%{3d8%1uL?^NeM7(snT0iEA}
zj-VGw3-UbvBj`Qf9pg#iQRE&NzA?FX1o32)zxwfa1nvzL|5kH$1m&6BeK-c^pyNEI
zOuyV3Ll3W{tmEG?hQfU>RPwQpBBKi@AN@W(f`rVXnWx%@k?h=^Bfa^e1Owc|r*v7)Z6Y{)3;H2L;-
zL&6C7OWSOzylxaNyh!Z4@s5VB^`6=oRRinxtBV6YhH2-j~7aS|#9L10Ro1KX?
z^c?N-fR%7?SRKU$2*Xps-
z5KXtLwi5V?l*Y;@sw{MLDeR@N&>A|LbupV?A4fxd#qmJwS_~w^@LereCdlw
zKMU`PBlVyAG`|2pYE0I9BvuLjKm1Gf>6Sr1tGv4LJFFKy
zQeU5b74zs_{NE43`>UJiwL)?$*y$wxQk_W{r&qKk-vHj6r@DD-9q^@y0PmSLScfol
z%9P50{^6o=(kunuc>YC8EwL3FCr_cTn~)@c9M3Nu#~6pR+7s=0bWutW(Pf
znxx@=;{0yv1Mq%!V{eN%KqR*vvak3{vV_t`gw2UA3Xn=X~{W3Nas-fv91fyDf!5*ZGfQz^{4(o`j?EIUf7?A
zm*LQbN{Ih3w_`X2kVWv7KIThLzh3)`%Qr1@?ZW)!;qUSeO^7c)_wx<5Ggc*mMCgF*uc(>6@FV0%!Z
z*6NSPPlWz{!FHnb{QYJ&z#AiPWHCPz`6V%g?P*)*sfU>F3`!Z9V|&bA8h?Nt@PXpY
zb*#@llS%`3Kx1c-cAU>or{rmDmrDiQI3ocSURbqZJM;J*6S#&R}fZGbe-5CAf&(N0x1TJOzhCbn*#Ls`bt3#NlI{@l%+WEsUKQz~w}m
zN$cWx^M8x_Fv`!Wkn^n{vP%hXx#OxtFgAxeS6ezdLe@Ek1f!o~`QXdceIcLHsr`@m
zj3L5<6!R?zkGj5~Le@WySySJ?z4}Y}JvKADN%HkuB~pm<(GAcc_+gwn*K1fv@v%Fp
zl-~uButFlA_H-GMYq?B)9~Uq~@hnYG>bt#r3r0k)Ey{}Un7P??WdDRkHh}OOe>Ef0
zUWe(CWWS^9;f?pW9fIZ7k#o*IZevYYAK9f3!!hzL-%0jwSd}$~aJ%vHhmd@q4Am1J
zcgJn^EtdNtz4Q{sNmh>w80k}yM9!(UfwV`}A3+j+m(oP!g|BTT|My|m8zf%PQe?uCeKR;S`=0BXjPyd?Rxc(_3dLMEA89oWD
zmCKjZZCV=U`phO6B*(i!hEWzIJ(b1=bf*
z)_Y_Of}SlbBbydL-t(@}NBn*C_WC`6xPJ}r68`5l4D&6)r)ry%uB%}X8
zN(S1&Zlqb?1;*!~{w`BBup3WrUTYE^fpK4MnHCNB+_n|H65Q~-^DDQpc92&+T1Jkf)3|si-%&$9id;Shx#kR
zc}r9Nl32B}QFQWqa>Sz}BWS1QGG<-tAyi>>o#W46;KdvQ(~{Tlxd`Xv@+AXkuOp{=
z&fo|ttg>6R6nIv~V!ps{zrSn?Q6R<&X+la
ze6-}xNLvmee-&1VA-@sCzEk>yespmf>d5!ii;OvC>+4xwKG
zkJM$vhf(==-n(zOhfwH_cFUe!L*N&?-}w1B9r?yNj~9HTqjLW}@voxj==2)+F`Y0W#l+3p6
zf%niwWb4g7k6x6K&-`FuQ!mm;pKTbZ1b%eXNrgX%juPz;Y|oMCL(Y6vyg|dg=;egy
z`F_t{RPi<|>)kXRah%$qm(xW@r^l*B4%mPnh^Xa@t^)t^taz*RI;t0~Yt`!JQR_vq
zYfB;pe0q^kk(l1WQ5t$G#vjYVPDk}0j`ngsq9HrBxfqqhG{iN*Vw()zpkQ0HinKW40D%zCeTs%-ltq$Ag?0O&3}F~
z9k~nLuiKN_i}EX#+5=#|V0n>+o(b)DI>w4aVt|g+Htyb8$=!znUOd>~UOZxfjl>{3$GM?nNsL%R0SRz&H@JPTOh}_L+hlPp|+_oM2Mg(po}8lH-?x
ztsc;ksAllxNcisUeDF%04miK^-zJ$iOW`}Qj-kv`X(4np_?Z9hsSmyAbWe}Xk!;w1
zaIIa^0r_~|Z1{`U(Uqz-wzc2vL!oO?Pjxu-R}JRcJpuf*3ZoY8=cwE(z{0GmV-_LQIikAT2%s6=L?)(6Z(-mDp
zRxl2nsilXB!g^QgV9$UYj3=AAstziw?ngQfT3K6xZ>H5Ow%3I6LcVuBnFBsqBEz$`
z55`MF`V!Sm@Et;p{<3O2n4el}_QZ!lKG)~3(4}Gg(E9lH)DP%~)2^QDav@!-Okmm`
z#+kNewG!+HTiY<&7k?*U9>n8y0P34sMwh_n0ZpEKjPHl^+vf`BFF-kFlkAFjApQ1m
zT*qI)i?RYiBOq7emM22|F^D9{W`mqHmvSB_-)vi~Eg63!ohm?Y#G{hH~av
z_nA&Wd_eD(W^9M_$DhZr0B)JxJv{~dyXnZ+D|nsFEp^(14)`^5`)x@;r-A^-6M&5~
z{LN*6XO(t3i~%~wxXoa@5I(W>H!d$&zB()v@I<@i>v_P{Vk^SU0sp1{_=x#rSK5Pb
zUVz3Xw+gXc_VnI9xf`(Nm9H5_bqTY7I9*$K^%ZP?>@v?A;&Kblg{m+CT08yr!aQ;?
z<+U%i({*(c!I%$*Myc<@d}GauoTZr0@^=fJ!uI(2AKGWE*P}r}a?be9lh1Tq?~_X-
z^DYDC`9#=a-uUrMn*>IUs^=Rp?_{p_C+BjN7vjQj`7UQV)?@qq^26W@Z2uSkDb>dM
zo!b;a&b3@%&3D0aKN8G}aC&3Sz)pYylb6E$xp7g=QvjL}fP9NQa
z(@oaA{e`i{{2y88`f;(1oO>Nf6Ia6V@gHtHz?c?)hw!Z_X=8E@XQSg!ORV>whxr~j
zJ@#)AISRyT%z4he$%$g13+E_2=
zpccZX0uC&FjOF)?R+99*{*8p6NnE4+ogJ#6{CN4W{v;BsWQ@=Wpa(}r|{uC4#
z{>9b|QhppXv;Pq}17+%5wZjt%j|5Qi34v`SJ?qg|f`0@l|GrE5DZgbAgH(F3ASG|f
z@OM2ENAYkU#{J}lv6Ot)93{{8o#Mf53_r(q7rzpo#_}KKSHW%JDcQgDO?^iGhpKIH
zD{*^V4*5g$u-&9ZkZH>TX@^=KIl^n}G$=m3=qlyktp6-kp8^A_zUmXycYEI7e-Xag
zw1o0U@jQphcTS$Vr)1<6H_7)_^cuo%GkmH0VcUZ%NV)BoD1Q?dnkl~&&bgEyG~Vf7
zgik&m6(D?_&ooU5>ut0&VIIF9*VXqBKZni^%gKB|E7ANj`CkPl3178uovp<3YizfZ
z{o@9?S47Wt+EucyTUUFPv{S~m4ziCcVyhmF<+(+ck@UZYi^=~T^g5~-=ljBfjjRLv
zY&jx|!{ODGzqIvxqltWvt2AjhI@dSyzTP=@{0)}7_Aq@Oj|)X|9X*&=JC-O4U>-c6
zJ!EAFIC5}9qA-k$d~c7D^J%oKuIIRa@8R}&j^#fkotDu8>cFJeT>{)$~
z3k$U$vjIFapk{U#P{j7Ki7m*twvTvyh4K9%jYh}iG_f!bV>t(YxoAAz++ew2_63lK
zYnD|6aQ4izUJF3|x5wh}IH(&_Y_bn<58ukV&wvXT#v}1L(|h)8*I&%RcfT71RQ;a8
zIo!S`Y=m)4c@LHoRE@_M!4
z5A)JJZ#GXoh4#7LWc~r35AJ$RKLPWFjHi>Y?BRU0iTJ9B%Mc%Cm9~6+1k|^I@A}p?
zgGlRMcJ3S4Z}v5NAJN1&fJ)m-4DAE^(fhgW3-6R+UoofJq8<40Q1U%C<&48_$*V+NJj!Ed@-k1GI;hgB*!5afe!O%cgIClVTT665aI`}IV>fqWT&OCs2
zgqP`-6v20C&uhzeXMw$R+}V!(ILGim&e*j8@85?Ss|;9o`(mfAtG^W13*+Q}eT;{B
z>&3^`U!vjM&;FG0x-?jqV(p@dyoUL;OYp=L%%jp+j~a4--;JdH^~Jt0-&Dk$SWEFnt_z`$-T>B5mnCQd&v1W0DEI05q2@n5Bc)ynDJyu^l
z5ASKwp`<~0&up^;!scNAvwYJqQzx8Hzm>4j+aBsEP^K>#hWVYmdC*np7d5_7S7^|G
zGb1h7whqF2+Qp5&Wxjn#POM?SB=~JOf5elX8w>MTVYW&`KH%9B!$&q9>O~(EIl1W`
zeW>`;-;M9K!}pQvY}oh;dr_VEuU(CEy@=iKX?2e`_?L@XZ``mNzF+pA)AXC}Lu~f8
zi9>5(9qaee>#?`No^R_+KM3!&riIAmMbM5yv)1}6fdBfaZ#!E*0pF9J+^Wg-7rw73
zVfU#Sh4Z>0#*@y_FPYLT&4(ad_}afr?g#L(x^}MHux=WZ(wlw+c{
zN1w;={bHPxPB5H9Fxl%KF9rR7i?mXzA@FwXD@PNY-TRQ$wKG4bXZw)A*sjS@=)V#c
z8h<#Uzf9RoT}^?0&E6_Atqkw|-g56Xw`YN;ZSYonG7jr_Sx>F6KzZzr&2I6KuCHCU
zV-e`>)~Tlv2IJVPuD#1y>!6)0KP8SpfA(BB8bgErynmdx>m+#Hz{sJBy;k870
z1jZ5GOKGQBp*;DCF1Z?z8#|}h9=x_6g_r$xl!f>`FAX^A>R~;+D~j~oK=8b
zPPR$Ypl9GEwqMvT=&-CfyAt9f(qH*E0yem}F0}-FGaM&X4*3=B5FP&r?cme3RB#Q%
zZ}IfGgx8_v#Byi00UnUi`+O17<8HTG;Bx^7(~7p1K>VIxA=mIe`t9BORPea2?XjUz
z4{%CleDoz?tVVMHwpZ{HJ;V0(SDUw<4`8dF!GlkLu8yznSpnAkqc6hWXZ7=?d13pi
z8OO33+e5w3ynpI|y#no-nt+4scinJ$WByJ{Y!8>T^R32qP;D{OuD_sH^?Bd$0}$rW
z%dhkW9L&7XitT*yw)bwBzaKnc8;kRk519$YyuNU3n4AY=7jaL=>tQANPlGToU%5d_
zJ^+w!$J=Hco+}Au!Tg3xdiWtOmsw5tE4JHl_kXD1{4XWdAIAE6$?|&ObQ;GNvW}?V
zb9Z8pR1c?j>7W`{>RcJp9FqY7n~kc)zFI3`u9e1PR_#Ru7hr#8OUDZz4&o!A9%`DOO1EjVn=
zIz)K835O=(V={{;2=98@xK9nsMLpX=p1b@>m#j~>pW8tAy~iwdzjWG_AB4XhKc+;^
z^`BO;A>|rsP<;RUfDh5LzkY~3FL~|=Nf-RHn()^g2g)zSovlhF{5)U}$)D+r3<;0v
zdJ;aXT`fw&?~JK?Ih;KzN%+P!bCUnE9Ci}^u1VeBq1u*2o|`VFczp()LQhYM-}9cJ
z_&1L}<;U(1;~qryU`lS9aqkNYKNXhmqWoExGVW*TV%)V=?mto=`9|vg54IEuQr|P9
zlcfH)2R9R5y=Hw6k!$8{C72{b)idJ_weD{fGC|}sim3Ecywvm^$HT%K*o!V_)sDjU|U4`!=r?h2%LZa
zkBg*yBz23NyAckpCF4QPi>!EDj)l|-ctwHFR#NT9_hm2X*FOzSNdMIPdGb4!W8e6N
z=(E2hco@$Y%0rgzxDHqr5IC_1@bNFjU6{uVU3O+`06RCD-Fy>|zwufZiC_HFO^bB!
zI*PU9#~XMaWxeN*;eNm(orDv3T-#yg*MP^<+MT;!V!ik5H{Ns}kY~{oC1XH^U2X@l
zo4t7#}6o~zXbbu?9q)LS>S8Nc~Ttsyt$O%h6C8o
z^A(qZoemJTaQo4`59F-7Jwi_aAJ7Yy-GtYVj;j8Bgy)leD{I&I1HKbca!Chy#etb!
z5fHBbIB*HCLv3T#A^tfJdhxHs`$y_2LVodpp*ueO$N~Apj)*>-kMN<0_qbi(u@#DF
z1CBrDy*>hHJNMvH1nARQe<>~yc>U3miF@iVV7zY$V48*fkJ7_>^0#3gan#LoDQ6guPSe1&?gIJ2*{
zAM|xkYCnbj%(U~Xvf+CPt@O5)DTCgJ4QGXAKu?8Jqq_~DKIM@?&ksX-ssV3PDzxML
z*WYLs@UWDK)o0Li`R}_%aloUE({}muLpzE4pKVz(1nZ6M57^HGPqsedJgn_dH70nBIr5rdHtn_P(E{aV;t^3AzDYc-$Hqx
zEh~IGKwqEYd;7FNPs@;3+Ft;N4IdG{Q>&M=lK-zzXLo)_3JPz&?0
z)2Rx+XW+R94;v*Vp?#Y@~j0=$ZY6n4B+*)uP1jyx_X^8KOV=l-S0Q>fiWF*8Xi~u4Va=l0WV$mvWWzI
zptbKAo-o9P{jF
zdM-R0|!uH^Wx)>ME
zKiT}#W}M!#TF(p1zv;Sn9;Z)EygrA~QTnABuK%xq%oRBQ&-5K+9Wj2&*%7C^+cf>b
zSZAFlitGKTeQh0kvZacTe9!Avt3b|?{D^ZR_B>^U1F`QJeCJ8N9xET~V!awzmXdW|$s`w2
zj-K@zvThnZ_L1-y&Ak^0&zPp0lYDLu2a$5=t8Wubt(~HHTJk7)zT?eZ68|si9KlU_
zI)oqn-nfs170+f6o+9?mhJ*v{DL=>GS*1xhIGFOY+-N|pPuC~-6Fr
zE>9-;zI3@u>YpaKhsbp$=#zZTzsw=&jzZM=-@DI#5Iv(-jFI@M+#0YiI$g=+^f6z(k6hF;o+^eph@sP-0w4?47d7Mn?b$AQKf7JyjUK@9u^1u9{
zg?c_!g5t>nhbXy)omBYVX^OwYp=^@x1`i5N6)03@qi~QhuK+Sf(8Z{i&kag`r~W6B?|fSk!LRL9{gOVBo
zTckgzpS?`_-MruBOgw&YzN;L^xS{tP;ps7Yg(P3=Dr#KO^EgWM=K2;#;%$qgiTvtp
z>i?Vhc7nurF&B{hS+86t{mVX+hxlu>xKEuUV{#HF>GCi1h8+RT0N^TCVRSn)W|z935WKYw+k*5Gzv$&ZQ0D8;p{
z8e@Vdt2xep?WSzfZr3C2$oOV5Ia7tx$Cs7gz-Txd^#t3M@|UxuUOs+KWWM)qb|0B{
z70bzuVYzF8=ZKwO)x|oE?OSl#5(hkQSz7to7V~?R^LJO_b+n@7EcF@8OM*CeV}1X~
zi!fvR$>%vQiuD;ZPVmL`^p*Cr$N6^eXk3NkZ7X?qPr^
z8n)$eSkEh+qUUk^vK+Ta9PV&n*THs5T7zd5Ua#{BIgo_YCy$6_;PkXah2uCp@8`V+
zw{z5|t!FGDicxmmd}GVrLYpP`mX5(Q9!;d)=zjoyCZFQ
zM*?8%e;e&@0#1~Exr*m~nSNsdSiibVfvUJ3Y*Nh9MP#5+aPr?voJwaH(k
zwhi!M)$Xhg9Dj23eJ)^TpfVHYGo5-8>bM>It}UwF1vnSq=X4A3X4PE56~OR=A?Emh
zta^SwT?;Vlk;+|7Kn}f-4!jQF=#a>dpJ&a|SH=90?UIV!b-=4r9FaEwr=3+EM*tR9
zS)Rl^S^GrFb^Lx=>~)wp4%m?7vfdidDJxf*8T8D}kqkHpIKQE*UlQU^v7E`3#(X(B
zzvDUJi!F7!J0U(eN^&P3;P#mEMJ*8D_x$k?UVJw$myAA6d#P=f}
zulM%<{k8$?6Ti-7Gk%ZqtVT_+zJ|tjrndoyBX{1!@p@s*^;oYcp-)!$yotE;8yOsq
zpBul5d3Cx_(spCO>(y_V@whS>XXoPqSdeBU=?VJ0xo5Wu`|Qs;rVQzY
zRlL`P0l%syTPFcNZrjsU4k&9^=!w_4?-uXfgWr4J>zN|B-Bp$~ohAL#XC1=j(oe0`
z#p~?WUcsUG{gCQ^CWm>qTFG){tRL@H;hVUhyb16Ai1k?yMv*a{R0i?k^=aoy|C%-#n3=AGJD{Xo=gG>E_SV_&kcD;hZDR*U?gCCzhX<
zXj_BFox`131^EA5ek=MUPFIY43lssyFKIO{j4zYBIq?bFUctViM%-win2``|FsIY4;>UOya`xcrLj_kZ{0BHv>%l@*Zt
z!FGGCCid-)^DbgXl>ROm!t%p6U!B6w3$Jz?!g}bmzCD5Co1MMLI6g(cwjQT1g<%w@
z7dpw3ar=?Lw_q$szkPu4_8s~1grC1Zev`}#ZlCIk#_879rR5ke*i-A4VJ56kaX936
z0+A1_`287&f9Y$H^|Z*~`^3JrNo^v$JXUCK0;h8yb|>lI0J`ER
zTIF*|{%_Y&`Ps5k_Wc_lwZ6*T;zM|QnWj9E%jB~p`4}9f)~)+xb4mU=om72%HVu<-
zsU5|avUsU`AUPgVyxmQK`u_Ca-2KADVr23wV>
z`Ggqb-2BWmCAU_c;&0Cw`vZB4sQ9W4)bpPi{MF8YDqlgHLM9CcPZgq`&oH9GN(_E3
z7e<9WR46@b+^O&-#(sfv1QkAUoI;oclKQqY@H7Kc8S@1fMi^dZA_qSN;iVx6lVyVY
z82cFT=aMkIlmtr|2#3Z}GVcdKFk0E!6
zfu0O`IOIp3-_OtoIwJ}HuYLYk&#eske1<-^u2KI#jWLgkd_v71DrTwoJ@g9oe)qFd
z`Tgsq@M0$QoakL@9^j-KaFHkVRWC8?;NA@Z)41(>=^UIHM*1>{~(3J
zXQ}sXh_V0V$&DyjG7;3K4;zdS;H=PPZ#n}k<|QST{#
z?q?$BCrItLH6~H_Oqn=Kl5wJ3kLn-gjC()dB{vZHxO>wCAFxpSf2CrK`dd)-djE#%
z=kyOBiJXKwHGbS0rt+ilQ1z;4qS`}0jhc7WN0bnG^>x(uH%>+YWSo63_?YawtkiIv
z#N&3@ms)o`E*qKi5um#&f>CiX@-aZyb7Y{L3n8A95sKpTq|bG?R9C
zwU4qN8~Te${`^}hy{4E%Nc!((lzeOI6OzBGc0XCK>DH_x`IHP(?U?b2YX5D)`9z;f
z-OEY3K&mryHrAwXf`%r33_WjPUq0Uzx_Bnb5≤P&nsg-
zac-GN6`oJ!EU+xX^P^IJoqw3GoO#Ok8|%3*^ywGBbeASxHp~Yi&ICAN6YHrb
zwcZWq?_U#6{?~(MH<qXwn*k@Z$VfoWW!{V5?b84;X
zmmNPH=d<#V>^`C|%gjcsf5P#cy$qWaWCTQ2D
z@YT3p8~DQrDh3HY!tq0E7Zu|9LSJ@-8E&s7JG2gCJr4=_{l)n^t~M{g@x!6*JQz7v
zzubZ2+uz&U;BqD3XgA?}SJ%5<#dyMt>nV;mta?XyYVG8?V>mwGdJ1XBu**Ydak=#)
zorIq%96LeQjiqf?>ErlK@%OSYZV})hZ{CYbg;i_}v#W^FE(J9p#*k+V?qAmjDv2iYV)?HlRcIK5ft3>nvl`mPZ<{<{lg
zeM>3#D=9y<#D$!5D1Nt^v`_J7S5l7t?jSOL7yMXEFp23dLFeBZB;7aZ1quJEEG8Hh
zMcE7a%hY|-b2G9e{zZSJLSI=lAq0{N`i(B`~(v@DLcoLMXi%wDPpV(+Vqoj?NsVsL^Ec}
zZgnU&lX#{k%KrXUrR?#xDr#O3yp&2;&!EyvS5x+zPnEKFn;ui)&Yv`rPm3S*{6_|#
z6VIaTX`TR89``$nw>OtjJR!-1VKj+BwnK!DqY6OL|+kX;Q
z8%EL5Y#sk{22tKK6Ted~!*CzeSv8*PaL&3>zGHX8Ao^!FXl}|rgdQpWRIwBoLMQDu
ze>&nmhzdX5oE$wlfLfmDabGMMK-O~_$vN}=XmX=W5^L=MI{v4u=Ng;~9$J@r=Y$;{
z?XG>JHv6_8jU`x^v#uFHi+1lQsYx1y^Sz%vD^GT!B?s1g`HJXh+oM&_E|kDI;zq8;
zpTQ4kQ_gXlH41|$+n%19tMC=RKb=GOTzb$eo(=u3`LI8^Ec$acoO7tIPVYSq=e>i)
zO*cKO>P8v{^L!@*>BuR+OUgd04Of#^JxYFiHXhg3wo(^>|8Fq8C*
zLM>|ODCB+8sJJNH@3;AKJfAHM#eMDMiuu!nD*9*Fb1>16tbFgq7VjRkIDaW~ZXn#}
zcj8yy!#VKxwey;r)fezTUex*@cU})F)+wnwn)(fG-V*lLD6kVX=gazqV>?lc^9_UVPMv7?OnH^x4IsB+Kpl^b-R?k>_Slnicb3CvwWkPzqZI
ze&25(0sigaSKTv2bFXX{y02z>ci-|Z)Y)tIUS~rW;(b2bnyS%-%1&CmGllvd&r!G`
z4F248Go5C#_`1+uqcyXuis3%QQ`dRkwRWOEeo0A8f?a4hE7(d&81BUp8%gm7|L&GW
zGv0X{!OuHefoTVnvuODWQx8=rPsB$`QU=PYq2>9cGqeZx=^_M
z%QPGCH!t{E%WZ977vc>3a8+e%7uv+BzrS4q!tb`2nm~EQTboy(0DtnA{ml9W!Ebyv
zTSuM6qAsLX!?nB$^v&KcopE#(_^tOc5~=g;LbU}cBM;y{$Sm)%mxdx;=#4`bvm)H%
z^Oj?Dwi?>S;f7d%?CUOcPMdFh+vaYh-*e=G0Q`R`Ec~1%OS@3Ji>ssQ8fb?=J#YUn
zT}b<-%%#sDR~%bpZXFDIaIf2$R@{ZGXmuIyz+ZhleNz6WS~ofpe7O4QKDckPp6zSV
z<1S=+G-^O<7W~W?uAtE%{-i=)Ad|RRgyOEw&sff9BH`>0bLdY*XJa=SP;so&
zkc9rIc&V5z0Pg90rLFJ=`fo*iTh1oXGt}Nw{&jaZT4m@LtS8fh?!Ogy$OV3ATEtqu
za)7_Tx$irUqJByN8I7%n2(4t{smYjfXX+ebqh
zcFX>i2h-4jO)h(VBWQ?wBm6Y*uX9$%n>Ae&{19sGPpEQ%`veY)%G%ta0sq|o>^_w5Qq|wP
z7s}np+B5eZzOQCZS!90){0Oa+`E%JNhlWx<6_1ugL%Or{T~io`7770OF{wyHiOoZ6
zc|fnidlhTrq1^u#&*pK!a{>>qj$eiEufM*q{pVK)dNDuSXYh%Jl4MsuXnqO#e_y!i
z38<22xFHJeAz%Zy2#(-iXs>2MQw@Cg?L1+pV%-k;c25NLfPW&vwzWfk1vF&MajNb7
zAl#q2QX=gX6CG`6U4KGxmWH&Yei#?S_u*?d%ZsfjqoHvv1LjvyUXjqyoPD3RHqi4o5Aw8mhbL(>&`t{jB
zg7r5Iy%Rf_9wbbMd(}p+J95#{=j4pOEl_U8p1h^2z>m`QW!ZO)p`HQHJ@UO5Xh=2j
zfn#Yu4OxDwn5e9!p|>|SJuSHe`A*wqKZSPfu@aH*hjv(37nJG=_DpND{(~lHzqp`W
zZgnRb%3jV~En!1LN6#t@uJxp$R*qN$QB^w9Uo|gI!n}31|bYwZhzJt#{Gg?efWCkumLOZ5falf#8zH)~dXV}kbF+r$-dBwA2>x0B7e77%hW}_F_eT`L`i(4M_qDiqo!Xx14I#J+bO7Y1)vMDL$42UF5xzw1w9W3nC=S$o?y@<6WOKJizMIL
z-1de0apQOvsf+yVMH2q6#PV$ENORtE!Q~1a$s}ImwSjg}e;GP{aDOinSQW8bI;a=E
z`xuP3kftMfCMKahy5Rq6-}lmU*TCPv`_!rq@FVbFJ@rboQ%Jk
z2YxeV8hzw+;e3}*n&`6`n9o~gINdexN3U+P9owbfhdLhrwiStk@2HxyB)KGDzm4yy
zS1FwLGb}REIW6Ch;`2P>|Bmz`>tY9|k1BmgmEC^zp}Ia)P2XjC2K;SYaPNs{f^(*B
zDk;{t!C%3kmGt)9(q44*t4;+E0zV8(x*XxkFvP~Uczw}o*f+{!Pg~f~i~7pWWvsU7
zMULWckYFA72mL;^ltFa1RhxMieAnh9{&%}QHhNDesm||WU@rS0QwVbXG_Jp8@S?$#g@O1Nd#`DG9pyfZMqu`q}N)h=3Tnh&9C9iguSM+eYTjozB=LW5}L
zpy7kV=7Xs1E8Eko1mN{W?hf&zgJ{xgmQ#dh04c@KY3J(OIodmnCWqZ6AKHMw5xK^*uZo7yEw*jj
zv-S+5+J;@Py9|fXDa%V5g=NEtMfJT$Ps<=uHTL7T2fr*Qm6T)a28PhtW|6Y@e8b2u
zMVqNUdKeYjvao7+3?i|}(oi92Kc{5Dq;+S95O>a$o!6Bv_C7_E_vS|
z@`!sEsU0wg$|BP&9L0wa?MG{wF8D>crJ!7#!VG>-yaHQg?!Y`K&ppWJ?g%QNac7&a
z89`5;v9wmHjiA6Whl0TB5mftaO@Uq7FgkvI)ew{N2>P$At6o-O1o=hjv~_^rmE|JB
z4S^#gNSkw&f?CQj+yg#hlKX8KZ7$VZ!p0BtwE1`2J@<{Er?<;hIuvY{Q4
z!?eTScn>3SE?=$#Ab&qSJ7fUzjkR)LwC2JH%K4$B!nJl7MYv~~PKphqgVWu$zIo8j
zX{9Wgb|Xma_+jo{_eaqEtw{yPKaHUICp%6x)Qo_ikrmwEyGM}7g6vJctP!;N-Qknp
zwvVFu`zN*?=o&#&9lm+T``~$-D)(oCqsXDz^N#Na@Jq<^h_3Qw1UbE3R^@pT{1v))
zAJ;D*L1U*lX;;O-U(M~JWvfn%qCJh)>PeoX$VKk2QV#ejjIBScc{CC9tzVYlGdqGl
z9)9&FPG=Nlx|mA;0{N4oCS01&q5ftnA!$I!%XYKi@L!+qR
z`S#r@;J2AwgRN;O9Q+lo8k!Z)7)7g6eBN}e7(@Fd)4pue8ACjWwLTd%f&atUmBmKj
zM{)a>hb!JXj-uj@-May{N!ZCFE
zzzhHDCZp)fmF_q*i%}H*un_{9z$-{tXp$+$B=9Na#6K&
zW2o`cmEjriV|lFi>4OU^$I-zF-iuqezK)e3}0e@N+pNt>}7i3=M=$
zepibaLjvzp3KuUPM<)l%o1{g@P|oF>H@B#dp@_kJOaEw?_m_rEi~JlzQO(ODRtb(H
zi=*F~>I=q@gR$TS&CW5@vGyhN!qgaQNjoel@MjFMk1!jpfcqq}&)f;Iau`E4d94qh
z8AG|gPHYa)zLGA-w&qNXqGTbPyRF<~NJQXy&n1otq;My2*<*=uw0u&nS
zM3jyrl`!?}NVq@if_;sG%kUUtN)jGd`Y{IU-J+uJHOJBY{+b}xN8`x<&bs(AgK?zc
zTDZ?{%{YoMiy!}PIgWhCqucMmeWr#{n}z~jjHAUHD~me5!@U@8RYH~XV~981p_WH#
z9H}n#EAWjShja9&viM5Ik?U>o@OKq(-`KlooA>a4n)aBLt3x|?Ch)jF5*|mJez+U_
zbsL9s8$awn-Wx~3{C`Vi2F6hl@2&%3E8(87JgwWCJSUKj`LmO@aPRAXQ|9t}6(-Ov
zYxzyJG2=+UD*l>R**KaV9NltVYXW_kdKFPCK7mvYzVzdO`^S7`4Wjrf$I-)QUH!vw
zuh`~YyOfd^O`w$>ZS33@6UZTAdzW0y1nOxw=>G@pwC1_sqSz%9$T2V7@e-7G=;NW!
zj;a%AYPr`K(gnPm@MT#7=vl6FM5t;UxtF`k?3SBAR=zu}?DkKfB^9rmqX9*vY^=pX
z$C0J#Qv0!x3AEN;=&5|q1p1!$e*r!M!Tly9f{XA!|7k5kxWw>3;byC)K@;#lCElqa
zQMB+spp<^v+EDO6t)fR^b1v{dq2tt~Jyh^Nr_O(QZlv%(AzoW>x`*&Th5ppr6j<;-
zA#|~iD^c)2?wWcK)@txS0j5t?hfMH41IU6QE@SXN{|lP&dYAA&j*o)zY_{+}NusC3
z_NnkcHgJ-P?P_FMkY(|iR7PId^?L-hkI`!~B
zL<4dEwA=7MKhpz7LSgVf&DDl0oG3M%5U&LAWnH)+D{;%*qA-dTE@u2WOI_%1VseJH1g@8LAXEE?Ur1^561mEvJ>r({e
zNwDuf(xvuO{Sxp$`>=cpS4r?cK8=Y$<}~m>TvP1{mF(|7iKF1>n}qK_TS{mvBtY*!
z^w!`9NXG6zqpok>;_d7|^Gv;~lm+cSW?daNw`A`>18*8^mQ3$IsoulaZ1C+rY3k>J
z$p`H}1y}C3jKu6eC;`cVJB#l>U__s=Ue@kE<-L_X2zBj0UHDz6qa*A;sdSMOYeMTk
z1l9E?wbAcC>l*&6li%(?nX+M+t?cYSqPN#)N!aQ?mkJJ&&=cuD%f=$H#Z2%&i}6kF
f*7NQ^^%KH*gy8Hy`n87^?Gfrf%dwo3_4wsKtJ+u6

literal 205696
zcmeFZXoO+W$5$ZX5dfR!q
zcrLSJZF|se|9|iJf2#JM+M7I9&(Y1%-Eptie_wZ))1&~G7X23w7pK4O
zjt87=y?pmM(*BA-;aGpI_a5-t?``kpz2DKn6ttk1T>q)v|6G5!iyKpa!hfodmQ|b=
z)8?SwSi5{&({Y@=T-+VGrb?0j_&l1Exim*13XL`Y>Sv1kVfG*yt3NirJWPEDum9WH
zZF5^U7?nno|2e)BnQ_8)4|qB5`)@ytI@1{K#016@jyo7P@xO877{N`3J
z_{7Dv-`Qi#0N0xT_}BlAe~hg=)gs24?td|gvti5&EE?0VD_|6L8S)?hAIJ6o*7(+y
zGmZYQ##f4}{h!Ad!^K$74yyg{i)4&f|NZ~j)e;p#vO#ghQG7Y|bkApY*@h
z|F3$(occrm3p=!Ultsq$|8Vrj_%XXp1ljIr2bSz#`K*d_JdKjKee0
zh)p(f?AdYTo*X?V4&O&VAd5uOfnbqQ4xh>?7vr?^lVeAd26;A_&Y{cbz_Q4D+5tAX
zM+bsMj8&LW|&A@qXW@oo|2ccjEpIuyk8=fs6>%G(7D4mg|me8ZW^CM)yfDIJ~1$Q8r!1G0&Pc0JW!{!WJS|NPA13$lpTXepJuHg-}v
zAsD|TtOH|f%4e*)1Ovb(Z!RCBa#t&+|H^?ew7lTlF#PZKd@Zyq>|5W0AGnpH9U3Jx3PNdNl#E;24Xn_l~3Wrk2LT
zK(on~NJC1q=S=yGUWLZR|HtOfZyuLRA{Dp7VQ<0u4?WO&>ysUpcmP;(`X+Zf7
zV_BFWY{D9T2MB>p7Px;0lBmKq>lbw;^#N+EY|AN=I6lIrllz-`-8?Sbi(p6jnT&>{X7BK&m!c^crch
z?_eF$c98<^DL_xJ47wo4WSv`W-LX
zO@)BKChG-eK?H;S;i@2=|C1$G&PM*zKor$q`ST(~IvB?lz4W`b&K1u_esOW!IixpL
zR9lhW`F?&f)(e;g!wkIC+=xcx<>g~x0kTPT-KQLs8@Pzl@BU$yf)vVs&9SOMIb*7ak@}I9{*!TswjF4kdZPLT8iO>))t-lhSE)e#OWH(fK`PlN((p{M#%c
z@UY06uJ;f?*d(P)h^~*7?@rP65PVaV>W6ncq4b@2R6FY5cGjZhr)@N;{->E3K-t9X
zQMMe?cS^Z)kuG@W-Gll9uZm%(0RM!{sQ#q(Z*;!p29&#^UbIdIolk2eee+On{QAFi
z{zM&grgqhOF;KHfsoGEK$JBK&U4S<;gwFz$Pk`~yuyh{fv%fOs{rR}@u*mad_-vpz
zWxOlpjZHRCTKSKuZ+41_pWF99_duV2hd2G+rIXZYoQaI_qjKFP@Y%r5b#FuxY~nf1
zzYX=@3N_JwMWqT-{qiFl>A17F_EWtTZA?4k%3#<)KkPdYpcylv80WGEZE
z4ej6L$ThJ0P`-vEZ-Lc@<;}3lP+kS2i2Ow)Kre9c#m(gm?}6bRsUqw;6ssP7B}*IOAQX$H>oPcR+p;2~ZgnL;bHDehlwE@^2keD&^tW>4?sUK)j&S;A#No&6Cve6mawk?`6aPM@y>|#7pq(E$
zJwTq#;puPzKSptPMLh^Ulu`nGAcsekf#tMYP>+5_;K#nBJ(>qOZ5HIzO`x<{kmD@E
z`tF?Z)8*(>cR)U7A2vz&k4+44`GI_hqmQc@<$F*Jazr!8kJTHqKSsy;Gv?12Ppd&a
zY8LE_`9J0-H4pMJI#%9A1uTMS2HQupgmi2?{%d~?IsRpG^rSfaEe?Nxd&C!@YF+qraSSvf+GRj2<{XbQ(kXG!7rdp|YHLFUGNl(GT^<
z#zll9|8^Vy0l0%RPvbax`keWf`jsi40J{|G^>gI+IrDKhonUO@N(YWbdO7v&su+Dc
zgsIT3GVL(n4$it37GdP)q!=oSuodi=Ni)2q55rp`j0O26H->t`Zh-b$w=z7>1%}uB
zf{p<$et?d_Ib<@!=XEmce%v4Yl5Fx?4?7C`?Qy2Q+Xklo?OhBXwT`K`KB|=J$-ZLj
zg<<%E`T`vN-W(^;CD_KmcBmW$fHnVCe75u-$^ZOoPGEo+C`X$uv-~iMtO|
z&uASEAd74lSqGm8;z(Zs<(n@te*Z~k?A#iG?Geh4bm~yMD^@Z4*ojW2{5wB>VHSC~
z{xqe#s;^SL*B`G@enJ|+Rq)4&YaivmxQ9}Hx;@iRzGd!|uNqdRe!wq*9)D!xV}e3O~2O5@r$8-?yM6Y?62KB`hWI
zFSG#;5Ny(^n*0s<+^W5_ou{P^ls;6ErtQ1EFQN6F`gbRx-u8>e5CT}lR)YCX>FW7Z
zKk%Um#Z?u_190Sq`S>{jmOPsj&6z>%RMq8De{&uf!$1M9EwP59G@CqLeg>94iyT$@
zM(wDX>%a#Ae5f!PCIE}*p1Q$<_{rw^%0%RIwwcrOg!S^Xw4XA?4R8c!lT{yu{djLpzVX@1mNAKy!$1H
zbe2jSo*M${`!$h%-zimq_2-!cRUKPr_Yj;}sl
zxES&MoVi=g1W>=l@D3l)4g4Aqn&JHQgj?qi*l&_sm4p6>v@SZ{2Xx49>Caz4lMGWo
z;y7R7kJ*Cfv!}j|YB*l!cjcR)o=|p%Esj&bErVdRD_|g~gEVi~Q-w`HWrva-O@RJ-
z=9o4MXiZnFIDW5?V=T8WphK}kvvfeN5joA_A@C~$Z?bSas$0tTKLzR(;6pb
z2hfkuG6`59Y_i$5^b@u#`)}N1oEP!O)!rh_@qIoH{VeSYU5)*kb0G%~t}J2}*YyPJ
zt^X!O&l^IXOCO`3tKPghf%BuCTOtwr&(iZef_6OHgnpxbki*nhJTS|L2%n1M;4Oiy#t%AOAMc_@ds}
zz(W44M0^`kBm1MUCBr#${yx|mS>)`lEP5_W7HI>9hhhu=VT4b&~tu(FNzZ-@33PsDDX(BOEwb#JltkjVs*B
zcW69Fn50MJ0XJL~t{pyjN71l%xBu3ol5J4WoCHAwdal+$q-eX)Z2@oENhPWHXp2oQ=*uBFGA
zqh4cs5FOwA7Ya1q)Qx_i{_v?m3K{gSUz4Vk%Nvmt%yXI$gmsGP72s74@2Aw!8aH+p
zk>ljSzI=jeP`_j|LsOD8DIX}0NRLfky@qB$ziiX
zJdX|{H8u&#FQoFIf3X#qdfP?EJ_!fAi{;X{FB+gwc8QNvPhR1W2X;?0m{#E
z&I9{7d^Wldde1re%xD}o(A$E|fxL%PE{DfjkQY`l@<0xL8|0je9&_|dI+*gsSWEKX8>BlDW}5*{6-G1
z%4xTj!_zc4l&5p}vHl*w=>u{)tk8}kvj2&X!RZ9NF&#jNkH`Y=4XXqF+Q+GHz&Xcf
zaOOWRr<^tidaay#rjVt;dLa-+(2f{qyv}gS-*f7DaB%#L6h^Osqc;NR4C~i%>IZW2
z73>|%dinXBDSw009qe0i=HpN?BfqZCtd~N_mSMS;5<_RGGv#rUu^~2DJr5T&n`mFg
z56L1U)Il~mbDZJz#xr_puYb|<-CLM-mwKM@V_zo2&pw9^Lw<#ena9G5nevUNneyAQ
z*kQ=GUS{Hu-a!%Ct|w$2QNB+W2++?6o57#_%}l%L=}V|Qt`^`Y$p6}YrTjuSDatSW
zY(#m7A$v+w%a6UjWc8@;_QHQEK*00~}(L4oxOb&QoOc+Kh03S>(Ql3$=S#
zDHaMK&!>Qa0`{+))s(jJ<3Wo}KAyxAHj7N`u7nxQB6-jHC|`DX7cGzcF^TFQS)fPh
z8KpVUVaWRzDbx0sy-A{a(~B;{z_7@E3st&*sO&tcfO^XnUqZwI`-aw3KevmE>YdZ_
zoPgy8E(Ne80H0{Tq5D)xLIcJB;vufI+@ew#A`pv+PH={Z1n1D$=Paam2N&>T{iDy5
zX!%SDyl{Yf3TtUtI&7Hd1GN6G!zOS5W04i`8-`7>XN7@yPAD&Gl~iv@6P|>j-i+zxD8Ku8
z%aH+~Z?-(mM?bdst(V2}D0Ojjq^udDs;K9XF`;W3@Hf1+OCbIb`OtI)OoZ}v@`VKy!RKN)=wSpA-=hkAN2_FxZ{QzF4R95y`=+jeR#G1if_Qz2^Y?xi2EOp?EV`KuK&N;nd7O`4-Cc2Rg@%XaD{hv_Ysj&U763`+k>*w)MZ{gnjEvJCiRG)vc
z7v#T|PRV-!)Ud4PryR)NmU$R=fIPj~$$mu}u}%(=qu&Oq;%soq)+^jBYc`&ZDHj=%Q?=b<;ZZHWZ1_A2ba{&;*hZ$&>x*N(Hs^Mzi-$wFKQYmWT|Bms29*bDU|Dqz9@
z&tmxozW{zFf5RVKht+K*L%0sJHz?1-_#)&tA$SS+ITS`Dy@0QF6Z#PedRMmG(5(b|
zGr#c;>bEC;tU>!%W9Py~!y;4Xj}{TjBlJ`7bD|Z>56>5*@n!S547$G@bj1uM
z&F>tmwx#~EHZc3~u>MlI?DiWrsNbh-Oym3s)f;r*sd*?$cB-SiXus~Qg>4Yzx6WZA6Y>~?2#8>xDF*_y=`e%u
zqa{C4-r5zK0r@lvh**Rx2LTr3>mwO_;#T&9?yL1}^QrvkXV})DJmzCAwSTUS!6m-m
z>*>C`(JX@M+qlB^1bU0w=TSO2ZXwm5kv|_U{vrPDgl+*{65>krr>%iB5AcDnH&VTr
znXsLK|9fwO8pwql85%4BVvw6|WB8^9I8Fj@yq!7M8AyQ{(D#_Y=!x!Ns7*a4PGOvL
znEqzfz%+t-1{}Qg5!0T)k2ju9smc)J-=I7KaW?tA5w=e#uV2m7|E4*Ke)m_e&8a*l
zkg;FzsD<)A>7A4goaOk5a188RLNVA0hnl4@*OaX(0(pU$z9hu4izeg_FqWeyEObU*6B~9Z^hqnIwaoABQq=BK#6le!qe#Pf}&*
zrpF8pj{v}5VY%-4o%%=tU|5i{R=JeheHw}Dxw0xJM+g5P58#kBsiOAl$>
z5bDAXz&^1EKRk*xv%e$H(Okuy_eI(L!bgac~;t54}G{{d$p;9)@}kv+Cf44tYw=MA}YOP(F>1>EGgMT=k0R
zT8nmG=3^!Ua5UKm3kBleq9xU|-SPEXkk5{bkwVHf5KH4~X7QygESHkdfEf&V;UcYj
z$SWpEMq@r%(P28p<8NIgLQ%f#l`q|wC2q`v83i~XBO@F4t1suCQhfjKMUOn%`Q&*w
z4fTW8UB830Vy~hw?)w&(d?1s-BISqdHBe8vDQqV4BYEm?u-rGkZ!Y41xVU$ISZ?ZD
z`~vfX@3I$+!~HwhZ$k&->5&T)vRi}
z&;c76o9N25#bWzP&m{zrYA4xy;y5TY9iD{c2Y*bPh4?DNO3||uXcqUA^VLB2iteqK
zhH`QKUz;$0Tkz&ww-WGnvtC}>33MITv}d0{&ZqS3?jqp72GwT|04~cme5-uLh`0bIUbl$Y@Fz`peAG7$iKGUo9ec0h;Q{P&cC-c5%aBf;Y#AF3Ot
z^$YO%w0XWWbwU12>Fh;3mzd7|>vaUm7g=vxd8Y&J86KIHbwsg<`eD@+G4SV>x9uxI
z!1G`G9b`JduUU;LSB(n5d?{5r+6
zddTw?g{$Nnq5N*3$(cAd>F*4w{0{ZFR=B71TlWwpFV9T*^d7<&5z$n25B58=Um6kR
z@Xn`lw3C3pn|@}WQEPznp}@OKUtxStB(<%b3w|H0xG4|-<2w6gAI}=V{d`NyELOqz
zTo5ChlR=)h`3*Mt
zW~ei1zaE?HU$5vV@*VtX&Dd1D1@ha6hF^Njf%8?cN>3K}v+tUi#$E7#@LExwCVZct
z7u#lE0lBfcTjm$=v&QSSO*!=In@alIH1M~(+$zrt%B{a7x6T9q{w)4*qaEf+&~l;2
zK3%YGwVZ!%9)S0-b{R{GK|e_R-(C@@e_SUjGZFL_&GO9R3ju#_tG!s|0rN0()|xNR
zVLf)oN!$hfX>OOUN%6BugGDZXu?maK->_}7SywkvS*G{bd3Fpew?zR5cic(pK>T1()qtVY_k
zfWCNg7&1pJ!umb^Jl^lM&Y!!W2l|zywdipc(C$XjGx;FbGK!vn=P=Khm9-czP8R&^
zn+NR+ynEXv3H_Tbzi9su=nqUCs>Ak9HkZ7|^P{)Jm%t*J$H~|DtwXV2uOp1{UU2^k
zH_-^N`)OwJ6*J)99^M~|>s#Bb!U*e2nJ@Rnc$A~F&7-G4}4&>QxW})W5TXwedpx%m8
z#ax)jFi*)*D+9TD>C&IrzJBX^Y4pcI+M4ERbnf(j!*+L$tET&*>%yjqRPThTKgur!
zET;SDAqy3H52QcTQH$~k#?k}mXNv4qnx{H`^WGMeTO7Q#4DDC@A@GNHNo@iW=;vIi
zPqwIU^tZzR&9VKhmX6_x}Jz
zPpUu9$c&zIR>>a)1$d{JuR#4C8D`F1x!t*R{}&ElPVe9DzT7K^?Y_#NUWWQSq9ZgP
zrnj?=+AFMG0RRT}*}h0P06`v0L?jpaKK6V#n87@}bQ}&WkWXD_LH%?V)1-cF#m@n9
zD@z8S+4fgZxx3zTYB#8cDNhysNaZ}wFjEijO3e@;11_m*h64xO_vK%t@|{XkD4jU}
zH7#HE8-YLEqY5U{^J4F>#}v1y?vo{-r`RrwW}Y>Lvc}r
z-#ChkrcLUg{1#pqK4^Dv2Z%vm>?9DN*H42PkY|@Me0(9J*I*3Ig1qbubPM$OIPb~7
z-UTtpt3v?(fc%XGQ||N>ng{u%>5P2=7K4Wy5;jq}xGml=08Zt+2ho*c^ey3cE+|)Y
z+DqrhM;~9RAE*h#5Axln%=vXH{XHk-o#1O?JDxCmvHXD%!y7GT{5Wx#!M9a4jC=>@
zTx1r*@I3H?BCKb?hUo%!u5$V_djV7a3ZGp-x&Lye|10e3#
zW9L>Pq4jo)GrVdqGY(zF@EM_dnP8Ikb~>8xJ&a9#CiIL6r%UZx?xn
z0|oiS-%OrBJdA1Ac{4-x9WrUX8R}E0R=Y3C6J9FzsVFEz8d=GqR
z*r#uwr}9H@rKz1Su19EGu$NY+>+{eyD}s9CPE|l62JV4(*3)*P-Q3~e1o^o2z7$uV
zQ4?Q-`%KrIJ#b)xb^U2JoJ`^VOOyQ!`2yvKmdJ~db9tyY()FIk)gbL76o1?OJ50+z
zzS>}k@$b{>1R>lvZkyDc!}<^2C(w2W?YyU>oxF%%xWR>Y`d>G#!*b`u@$ixj_S@K9
zcd=a2bx}N)JLefY;Qq8`+Hot)=jqj7EW-8=yw*#=b{dzh?ZJ8%c%MZgK9(4MDT8_1
zJu;OvA9(zXpDD&!nJMwN(9WHC$DiUplQfuiZ$8j@!n@Alc_7*5{6)-jb&3ZSA?}X(
zQWA!7T6$fsvKvsl54%G!evj`>IfHqzqB5~~eD894P+MLB_zl^^6Obk@4xftnRjq2d
z67Elv`BZku0&f!ccK#jUgZ&qT>Hz*(wtMhh8t^rl3pE>nE=iier4M?Mzh#ONpu9dx
z>vJK{J9F>3umQ*HGdNzj56V|8m|{2#{P|cDdWp;0-S7{u6q&j!YkkR6_3H5
zgUZQaJSRvm|5g_Uc+g_LwX6x`yL+o5st~8}+ShIq0y<6E=^7W5fAT9a-wExtO2xkZ
zz=HEy;-`bhpxoz~d2}<_tv>YAn-_S?`8H2?_CmgXUqjJcS%@cU&S&ePJh@)P_}m2O
zr~Tg`LBP-L=`Mc53t^s4PU1Zqz$RNmCz@XY+~QpM;NVZdsg1QcaX2=y{U4B5A<~Et9I1^6>vSXejoVP7@ht&%p2bM9C|owvJu=H9!SaQ0emyD
z+fpC}^im}HxXRI?FMCGUAZTz)CcwuuH7O+PtWy}9(5~gounQj
zuzlP$sRQutRzB?d+<#qUaD(a1n=ksv!+gKcrmKC#Sth(k67YPqKl-A}
zu9Li|u)3FN)j=+_{{1q=wS%N@A0a6=ouvAPWyGS%@VkmE50`fMzWaK#k$V20jB
zA-r$mZGGO_2GlDnV9ClB*k^34pOpYkoq7M2?H>4UDV~1^ngQpl%vax#4!pHZ**{en
zkG+Q_K99orN>e~@y605*-p{3#S#a<2RR2!?tUmai!&bS8W(&JW>D?sLpAuojlmF5^Oz?gt`(2_R#;u~$vMVuKH~2FtJ_>l!Ot4!6M83rk4)6enF;jPm7o3?$37Su590g8!9?G}@ZXSU
ztvg#_1@h+kBh?-tH{Tf@kMTrcHeUtCSKj(6H}p6AY3;iX?~9ih&%@8o
zR6}{dmF^c9cdry|rtdK0McD^X{`KQ-Kh$G~$^h_yxE%2{2mQ_MTq}w3e08$pLzHhg
z{pBp`pR1}0!~NnVx1`1b;2#RDyM+4ULSA!m9OgbB4=)v9zk4F%hWb^eRTuF0M63sS
zdvV`6^*$;M@0q+}TokZeMy&}hGyvxX2xucO>5WhIVPE>_NBvpIg)dhiK5aWn^NtCE
zO8{B1$)w0F6z4}@;GuflITI)jSo*?*>Mwou-3a~scCDlZ{nuJyPw#`cR4;ZQZ(@$W
zZm<>r-s^j^rg)bA_mFRV@!Mn7MZcN>w{P)_@fv@c{8
z-QQ<)xlmm1yVITG$@h);Qk_LSnuq5FZ%9Ic+k7H!Baq@Bx2I
z6JQM38ye62PAUS>5At{O=Tj=*?nw27?lJE$wu*+){(X#x=>vL$r?gH!s`GCWI
zYR_4f8*XSI53viLH9)(!I=|ELbJ+Zs^0zH8@eaT1xohc$)L;bCjX+6lFANaGw}-U=
z#MvanB9D<8VIyc})rNT#}FBTa>>lseplodt~2g7KhMc}X&C)!rk+_%0d1*r{Pox=;bf|xl9{)?{Rgd
z%V57tRJqpfB?#xfkKfYJfA?`;)ii+q@LJE^3Hqw50&D7Ff3tKgyuJbQmn*$WWYmD~
zY3VIYg7RA*-hO!vw1=<6$`IA3G~m*-FIvCEra|1k6t(R>;F(0WBJaZRF2eGfyEO=K%G$+SCL?Z*9yc=+>U%vfSKUrkvVdWw>K-xPlG>hbQ
z6aBaM|2?qjCaX?mE6)-uEG2w9F`lvnz>3$+35cfyQXMpVNcHdqyZGiZ?-<4gE)lUSRTt{>L`-zajmhG$G_K^vLE}pd)
z`iUUdeCw^{{ba}SSj&XrezL2m?N8s2C;1+sqp{hylk|&v
zWZc@-L5c)RQ@1;HkmHUsW|wU2BtqYAm&D|CkUE=%Dgp+bWR19x@XD5UvMJ%+p$(hc
zNq}&x4o`nOQChw3s@SFu!hX0a>|{X)k(T*&d)Cf&a%kxcit~fZh{brBBC*UiqSh<3#6z`}oI3R^@%`@>()qW+XalRA>=op`z+2i*
z{DSRsdF0xN_QJZTPvcsNz!U9&bL}m}tMPY3r)npWEPv)(Qr<<5ZJN`w2FSh_aBwJ${$#R-+kN__fMVaBvFRPcZdhV`}zCaT7jx?uk$ih?g8N1H@!an
zYR#?0S*>cyi6!misbHw2AmI3{1-9W`LhyUJa~-Pg%7EXpADA0VYb6nChbFPr&gqCPKycxR@5y9MII19QKphdTim
zU(?`Gg?O-d+YOaU(BD48(sdZ{_qMl-9-oni^WMqiV|aeNu=bc!G34u$_FTVk71|fM
zZ~l7yb$s_%WO`*b;5oB#($`=-(*HC{ghRaGk*=zr2{=;ls($+}hzC;FTZJTHo%??H
zyvYUZ3D(Txy#)M%+QXfA4nKC0=eZf&1IYz+`hNibm1ZxG+6a2rTRKB4p?rF2%Gswt
z<>en(V!wL(PrG6s{-X8Vx!gc4^nN@u0R5ZIg6DBxDAX;z8w+}(Wh~t|pmVSC>%WEZ
zzdpGp#2NIrTo)uaK=0|Qrxu%G9AEhar(1&Fq<}
zjI)(zw2vUK{pVWhM4(*TK4?~hymWocFzz>NO*YQPeMOp3dOw#
z_^l`&y3}Dx@ei-wX^N{BS=3TLPi|-Ou1A`|eXwIM9>w6S70)Rj)pefYp;3W56yN`g
z_N8`@SVdBMeE}~i{?n+KOZ{}-p83BG#=1_dEv&5X{_r!j)^YMEdHMvqh
zfBG2$UfC{v}+1CJ)dMvi*3dB_a;1|dM8)^rS@%-
z!zo|zhQYrkN_{ksIKG=1M^Uq{RPS<2HPwsyew&WZvE{>b{)l!wqj@~9^s7{U!_<&Ha;>0k)bG$e|XWt++{~3M%5FOt+m&)jPy9`y+c8csD=i~Y5B=`1rXg_ap
z0iDm=1Kr|~zaLZQf_4`f5*=JeH;!-8M}B!+9wd_C{q!57NW_!!(wpeJzpA0Z23$wu
z8;+!6xr43XW2C2-{#HUgZlhEioWEzw#3d2OF6{qI^PNe*5+Lymxa9t~Xw#&$kBIUK`yxJ>%TuI~`H_M0saMLem$uEe$w=yUHu8FkQ0$emO256aa)@SS}R
zaW6;WrEf8mAB@?*3iU#7DqO|i!_~xUE+e2<_SdOW3-m$)o5=$xFX`WQ9rLs6+Ieo{
zy=d_3n|hNVUat(WmfsKf>m`?cJmzKd>lLpkK>7V8f%F~X@ZQHwS709sYB(hI9B^OK
z(^}npupgEbWEKm0b<4S$Hvw<_O!y<>-o-luk|$4LllWaOnw8Lg$CMit!w{c~To;@S
z0i3D+`_Tm(Xh*7a)|6^v*tb5cKDr3(Ph6n9dNJe)k1W6OV>RH(W1DiHAAtSw&E{V#
zOW+>d_rp=yo3MX{$rond0X>C!p8B_~@cuxhKRi|%-sLSho3`Ts|8f)
zfz8ma+w<4Qo;Jhp;-Viu*{1{Ftylb0Gwg4&!fyX|zN4m*ofSexr`TcA`Kk@JJclG=Z=NxY1_S}gu
zpPt^m`!TbJ+}knBdi2i#Ia4hZus^+zoM>_>p2P+5xpSCj3Q&`nLq+L7`p9>W8EtMN
z1EjwH*w3wIedOQ2&LHja@Sf*}tGC2loG(+$#VrO%@UGGU;n04v{>Mo{>ykdwS9q+;
zP`s5Wt#J}MuG~t}XV;##?Pw=gT-x>gow|vp;Wytq*Wumuf#6_6jaKsL?fs0qfX`MW
zZ!cM7)lT-`Ubso*eHTex-8t*iJjgpOaxcV$hhxfBD|zg7C?-*}mH3`@70gj=C3_Axq#AU!ksLwqbM-zQ}@)C#N
z^SJJNcofbHN)7WQ4cp{RIBw-ivM_X2O
zl7z1K_zk3+Y)NapX(k5oMoRhW6KJ2U&|n-`4d2yb;wpa24l;Aw=y45$E@E)!&w?D7
z&r4rySzaOsd7a;i2PMJ3^nrgX>tVk6^dH-|1?H*LUfBv+SU>F-XLU!xdg>mr)^Yvs+C6Y@+F
z4}pF}fkB2nv@!l!AcMa@`#vva7M{S*>$_BrV=SW0+fcb3a^LZDZ6YI=vCq969
z@HcDOUPs7(8+Rm6iL8OVVPi{_Klr)LV@3E5Scj7)ot+s9{pjGo`!*BWbMnt$ycDS8
z(Bx>0OC!;mTQUAE*Wv$6alz#kOP>HQESo>u3aCNpU=k>`*pzO2LS0gTaP*yRg7pTmU{(-q5uT*ViHG!XM4WTbi10Sf={!$vKr&e8A
zEa?4uxN{ZW%Q$?C_)`Y!eWi;`+H#Qh#l2a8@87Tg?1(%GJfFTx5Wb6e-7}*Gai76z
zjmwB%7B;rQMK|DYC&lX+|DP!Y?nV5ftcT8g+8G9pZfKtO~}jS8S|n&5zPl{5a(>$DGE2s@Evb9iK9I1gH#prwfA)CzHNKX
zjOMrhX0e4({-|M`D(an@!2BI0?@9HYC=X~X13&`zog0t)AZ_jWK>aG)u1N7sze*^@
zlM8icsA9jOgM|&So=(i$A4nah#EYR_(q}~Xd((GkDPGyYJ(J>~q^U`i-_-6tjQVE^
z=0V~O@KA>rm5)pGp?cFA4Jq%O%)A$!bj6A4tshRNa@n4OfdMV(S7d&W=
zfOD#b_a?V4@KbD8c5Z7wcl;?*4esM}!_H-fCTOexIdNclWeHV4w5h7S1kLgm|8%
z@>yv@H;M90d07rPYM{DXyd3UXJ3Jy+-GuwRQ6=%N?qrBx@lj7qO}dEB5bxjdaQ+v0
z&=slx{c_$PfAJRF3&vFN2w>iG^W2D-6rMgJcloA!?;?2T+<0y4BY3}fsp{y)!tIc6
zl&hB2eA!Dh=Z*f@^L~JIDXSzL1bnBMG(C31YuHCc@`_9#&oE`qzpl)WedJ8e43Tb)
zezM)n-+uGvZX*A~Zu}m|Ydp9AeQ*ZX0P*7S^R;-{PtvrlZduLkCtPl3^_K6u$?KC+
zYt!m`NWppC1s5#`2;Y+Qy`F$`-1BV;`q}-&>R8;tUW-1m!7@}Td!(PVJQr`VKip3|
z%MTw`4(KPHJ8lLo%jqX=f(g%eZ0aXX$*b=)FYYIN9l0t0R`-+O(PkZ+9sMLJ%EDgV
zpr6DfsZH_9=_6SYixcd)`ib1WBX;YA`r&tHN(HyH`$@)a{|{k${p7`~=krb1wvwdp
zT~F>WZ6(rTN27#PTFK~y2_ffywUH<8_Z)QmA)n@xKNJUeD7JCmc>ZmzM7e***RNAs
z$;f-huQ#T*k@^Lr+==?_q@(JS<6MJQ(#`)k{DEC7DXsNrNP>EMe_2J?>9-OYh2yEu
z<6Ftw9@DZw`&Q!TU~)UNQLTx?7$;d`kj$`{y5=9mQi
z%1&z~T<*7~-RNp1hZJ?rUYpTIu6lLx70R@cspB__Ki$$nOvS^z4A*y&JF{o(R)u+y
zeB*=FCRsQa2n3X@fpt+=<}=i(*-q9?SGsbCzmpvCU)R*I2F@W1wfQb6!S7iNJ>8eX
zyVY9F40n>*O2!wB_~^B@k^SFxjk*PQ5$EYYpVY#7P~XZnsf4`lR=eoITOV7=;N`{I
zhakT=%2S~7;C%-c2w6E_??B*-HkS|XQ_jx-AcCXIRS+D~9i0z1|iGn_&F-rH}NI#8xIDlPXy{zrfh)vXXLd-Cjgx^xGuE>`hRh6lJy|;XU{yY
zjhKgPp7^fuA?V9^HnL8Fo~(1?na9vx`QPItA0%_3o!SCS$oz=7n;^TL&tN9qt0Z>$Qmk
z-X}#b|8wRC$nOpfd_cUl);x*}+Z}<|+K+)Ab$`^Z2J~vs$}-H?{Rlg{0dZ`m?)3(w
zTbKTlM*ZZe<;T(QHOg#H2gs)$oopz(!5Cisn2vj-}^<5
z-YeS{=TqF|n4A%V`(DO21s$xnF7lB%?uW05?*rrmi_+;m;?+l!RI&dR&jzoce#ncw
z^QiY}aT+}2gueq5G^+}6*FU!s_yri8%a<&9gL(qm#S~vv9w-Z@`ks!Vs8=|3WB~0C
z>t3Yz<;A4k6u;b2KA(v4Iq^s7`!YdU({AKT)j?5@rpM|1qW<9-6o-x0|5~Rk70Qg$?wp?7-Pc5LZt%
zNZq;z_M+Y=JE5M|%4iu4IQO@j7r#Ad4)Mvtab+Ut>6+@@2m`&`2@4-6gI?9sJD%;Z
zU-4fuT7JtE;-8Rcd5$66ckWX0epm`|hsRp88}L@pp01p-M}0)X)L^O+tDh_u5L_Iu
z4|cXK625i=`g2rZaN#PzMRT?(sJHbI)t{ZRB*l
zsNLc6R&vBwyS;f<2a(L{H*DG0O&-tehzyT_c{^D#R5u;;e!abWS-g#?-g*=jS=CMq
z15EwCjdqc$LrYCx8NmA_@y-wTgf^LEk{X`EVb(M2w`-q*29
z>LyAF+g9fvYA50+-AV#<+sTxxk84fx+R4HR%3ht#ouv5TmphmByGgTh=+WBbb|R)=
zd{+SUCu-f}(f!&^YI(Pb?fukAZYtZ2?w0B%@fiq(7z|$X&
z*N;YB?uBaE`!}eIV;N#4MM(0;c
zroiv1HhT-dJPY%=y(}_P8RlQ4qWX?n;G2^*5?Wxr^>E)D+z;z&iC^I2+mN^DUDw@E
z2S>K)u$_CV}5UkAM99`O%=4=3u9|2lPg`OXJ@|zU$-kGvDR{
zzm#1rGZEU~&6=Ec6ZFc4b%K;YzSYRTCJpR#c=#Mq0_yor$rRt+im#3w#JG0t0(T|G
z)h+%`yQ6_7Os*DM4OIJjLL{EgZXfPjgz>g|i>n0g%L-YS^K5{YXk3o711h~%aq>6N
z3(_!$zbped$@HJfG?0Hb?HNSQT@ASsVSk-coCWr=s22X1?_AY|BrU
z5X609j*TDPFK*jN(f4$XCe3g0@7g_mA!~v4wNE<8;QqcKp
z?^0GZRSBY8Of2#dJ?B=Y-9SEm&k}9an=<JS(7egS86p$0p5^d1iWl;ako%}-
zG;qHWa^9vn-3Se)<(ViX-j@ms1>3B!xd%hx@z(
zk_l+9LSiY+LlhQ#IEC%AhBUHJF1NItzRR;O&<((O@T7pbyxNI%W&b3%I`?Aj11y~nmw97ia<}>vA@R>14zd1g;J$Eg9A69Q?dH3`{3_@LJ+4{z34Zr@
z-9llN5ajcBdNft!t%mRFe^qKN__KHG`oIOS?|-V?^6Wk6sp#Kexk0-^53C-Yh4)4e
zw$3m1hW@zTd+WRe^c12)?n*)Ymi{d)papS1_L*AY?k!-~Rj1w=+l6B53yKuf8n;eK5{(pk5AO@Ub0ueG;}BI
zD~IcOe)zA3eYU1e`A{SLot^3Ynp5lh$cx0yvkKjM372ER`ps}|PKtHjq+$=>@6MFy
z-4EfMw&Ti&`p!P$naVz#zNMFZRk1#-0Q=>(;xos^54V%I^5=6OeQzh<{g%WW``Af7
zob!KSF4;{q;|8g}k5A;l0q}A8Tr7c9ZaT3dW7_
zPAIbI>5*J`&$9eL=ES)C&__`^c8=4Q$G>8HCdJQnt~-D87Hv%dF{ANG^|CTszH
zcfpL=89;aa?T#O@Am^$IJm_--y254=Q}gNlWTw;}0r7*d&n*xunGJH(0neou=Rkb{
zZ@9!M+@q+8_YqGo{cV6ZP|kf}1a=!#Gne5Zb2rrJvlp
z@^gwi*yomSQk?P(e>S&ga{KGp*%-E^7bY$+v)RtF`kVexRFv-$Jq%
z=-i5zWkv&C*1vTl%0QQMsMatN=&pYGR5%RjhVSBawm^H7j$ihj0`n=8Gr{s4K%w#q
z;Ub_bRyO|o6zYpjty?z}+U1b?!%hq8nGh3v<|oXn1ucU@{7{d@g9;~HM>-5Sx-~$b
zli~gZ_seIl?vrr8)YLn{kL!p|lGZ}p*916^F5vl!eud#$q~p&vlSDsEtuA3@)JyhV
zyJCoR9tUh^MF5^)L+w8?X_di^3A8BWpqdnXAn|BcI
zmkZy`gbSSDH}*D=?w5|*PO*sRTvA<%>;E0T+F#C>-3bUJ1M5sgq96>y2!gK1^F1!3RJ+Bzl6wko)+cuY|-^J3)
zaGU{ns_sv^KSzEIqI&o3DNXeJBh6tu)lb*NpQ3b4$2917r)zjpeB&*~k439;kk+3p
z(?io0BI_v(8NZ3rx2@Pu`SJR?oa*=A4pbK)pF<|+HY5DCFfthTuN$o%)Gupt$~)@M
zk+eCP`sqJz3Qt42#bL8(JcRpeZuX&kwq#^li73EBFO;bt%$BJ|^jxd=awoNq1nEzu
z{`fB*cb-K1jL#xJY8P?KQUpdpAMNd;{>_iJJ*D=Ei)F(8D962IpQ7^K8{K#+5B-X}
zQit>l8}*}+k1_t2g7Eifa}5;6{c_=~UpsKz8R>etErI-|Pvc6AU>@0e<_Y@3ez~c#
z^byLns?82|cy1Z=OIJ`0aOpT-1C*Orew`Wq6xO4^OIiwXTv%)^%0{T_WUhew|6t^-
zVw?}*OWH22Z9FHenlNtPK}esc`y>;`6DR1~
z4D8<(GUnFUZw`Ys%_!eIziy?4-x2nl_+~n+4|c5khZA7GoOv^R%68CmUXQI4E&)5_
ziqjv?>%x9{(#|ZkvrzV|Jg-!9&0u3
zJ~jpVAvSfY1>Tc4m^bt00$4{EsLmd5Uk~Fxa^rNg1AcxlbpY*_4r=NBXnzw7^|`DM
z^bIkOrnbO3uIiR}9Qpqe@y@{$@>{}eE{ubEpEb>0GzIweSoh829oAoYxuH`H+9z4Y
zDPITsf}^~Hcqz=|(=sbG!GFRmFFNe|S`OGbuJ1i(0P{rY)GLQCP2<34n8
zh^zB@Q@8sZVpAMiKitV7U-zgOJn!L~s>7T@yvs}{R(djAklS8g6WOxcb;eg+|R7~$C4zXRuH`@Z@V`#s+)RQ=wOJ=L9
zhq}vf$-GvwKDL=d#6mMh&Q9WzH?vL{7aoClh+=^Ka
z{#w3GM}S3i2g5UuPl9`R;>X7Qknbg_ynvV^(NK?Z#^XJ!S;XT~?S#gB;M;FF$N}n;
zxh68!4&`DB
zGAvTp)4$ab>;U<8N$CQXV85SLBVHiNB4JfAY{#QLWX_NJSy6j>$=mT=zV{VT?=equ
zgK=)L^J3S?6!33R?^<^J0gKcm3f-t!4R(j^nzN_DxPR{_Ubh;?Ew{g1U}6c2gb(JP
zx&!iFUdn^_bHN{nXSHO<9FU)HDDLn}2KlJhD5>aaS3gO(ebAz)1I}w6)u!m8{8A_E
zJ`3i%p_kyrW{?Y(y}IX8X%6=D
zHva@amvQ_e55puuK0g!Ml#Ayl$9s&Qz`6hdI^Bw3od~;_8CnMCIg`CITW5nE@6nX7
z7%iA5;<5wXX)tdCP8Ik>0G!ty-BAkiS=3ioIWhoxNPzyoYWPl+yfeQA)U#^uQBl-e
zB!UlaEdf13w8lfW1>~E!<++C@0d(&+8JGm~OFbYsW(Lsx5FL_p1%AxBjXp|4`I2zm
zN_|Mz&px#5GSJyBu^5MPNRxqmtA-`<;?*f#AUV60dv;b%u
z-7UBj_!RwTt9$|S|NLI~_#UiFmk%mEMmy?Gf437wfS)K|S9}ig1J~x??}v5OR!?>Z
z>TfMMA=lAfknOW83gxAe$B7E~{jW0vXP5)bmpZG^1nI}>r+*hBj*FJ^yp9m3vhvYy`H?z`5Iqy}&&wW^5$T1Or%-8js5R3F2
z&tF*xGlFM%Bb>4Q$Q8^FTpvvLbANVr_lEHgnJ*335+VIY@E-~=v5-xwMA*KlAOAe2nSN
z&tprmykW}=Z-n`SOR1l8y%%1ngE%r{%c%dak%6)%;_bI8XJNm!tEJF-7gP*TInMMT
z6Bi{SoAG0LVzQpzzd;kE$ot%KtfC|9h$oZ=UggQy+I*JCpMy{DPy&~$tI+f?3s
z9o9XJ<+7)XK;Sm;`_Fwa3!%lEY8rp{H1GKp953yA@*QyeB>fJMD*@bJ+A#Pszyimy
z)IRjlWB)ZAzis;(*Wfs{$&MI8d1&*c19H0oPPZ_iel&KSS=6uHw`rFcj<<%zpCKR%
z*pGAg#8LjxH>rP({P51?(0-jSs4ao~Bpki~f$PBDz4F^49M_92zfn2y*4C#HNMD$^
zx)kNYe*FGPh3t^Yimk0IoRTA{`CuXyz6(r?w`jXGy8|74Y}?lK=Af
znMAN(-($ZHd<6RY)Wzotpj}Tk@eEpE{r*vJSd9KJ$3qmVePN$CJWl`Xa!F|CJ?W5c
z;B(atnWaOpu7*m<#&*KK(fEF!$0j&e*&omE?FZ*V(d)WR-$8kg7Sn47q2D5;^ok{c
z-Z}r>>(_8TB(-pI{^~aH_faaja|h@OTU*wxgY`FSw@>~GSkDh@PMC!2{htu~*sq`;
zZ&CN!`y`{E2<_{CH|`mm?0MbnyH1BoWE>amTpq|FyVN^Ob)&&vbKzUzkJIqGE;mU^
zgzs2qyUcvN*uJX}$*+l)*>0^9gSCTaDU$`R!^#7A{e-~bcb#Bb;
zKgqoua_r&v>VFS7WO4UES|<4Y7;L#6kU9tE$C;;kDxn~^_bzvh3xxGgTOnHG7n?Y2
z_xY{~`>wPlsXq+;qy1G$u>*eh%{{WRKGrb5&sEzs!#R}k@APk(A2`IR%Kb>?Zn*Cz
zejzqkl0*9L`TFS$vB@gy^AbTRY%X#QZZ-u=3J0BGaH|jdUMDe*L}0@@^Of$-$lVOrEKC=SRDJ{0Gl+(4jAME
ze&=bur;8(p#BDTV9h=M{U;MRsIufSV#b(0B
zX&fTuyf8WKGMh{vbN$+lcs5D4)rj;P-$$I4T_kt5^b$|S`+mFQdr8KwJhKfD_n`FL
z6Ja5^2X?&9d!&c-5eb1wH|I
z#qAfnQ^Nbm_#9rH9+cDmMi$#O_kv%wdZ|W4FOd_yx7*VQ;u#2r=e-x~C+oQ8Ddm^@
z$b#|RA~UD-k*QJ=Q8Nd6$=Qs-AFl)ZiNHW|Oftw{cKqc@SMTP=
z&y_yHmOno$lLPW!%jP&$(Oy#Wv-M~l$ZgSS>sgabS!D66-bqsnSmf-jY)+SMFZt5E
z`+)=S>s6?KzsZwD(s$b=&dyYvM}CD8|-h}
z%~!`2_7N?Ot(!ko^pnrV3!*dl`$(0)&qblXAjgVVH4AI>lhnc0O(ij)=S5u+^8op#
zrh1$C+FN~Of3!)X4Y!Zn4~tV8)`0WlkH7Z}!hD;~Cy=-j%Q{JZQ)
zFA-dAm)s-;=f)r21*d@jO2^P&YXW-9QR&C}mREtV-hnAa9espfB15RL3htHh$%dvu
zJ!*W?vZxm?4Rw{S%IPORQa=gLyASql1&JrY;9ukN^vSoWpf}yuY&T2=c~0xXn8;ng
zN8d}af;x!*J$K6Zj9{4W6>0<9CV(7gy5wZ3EX;e)2$PsAu#Qyp9oSk1@}%LSt2y6c
zett+iw7~}EyWKsT=bwQt?0|r9J;=XzGOmUdf?V7FDBx!UtVi`1?!GUBbm%)^sGn|cZ8uXow2hI}Et
zXI|$yDWF?hnO?XX)|W|!YUyYlZJh68bzz`3Iqpa6@9lyIdogHum;4P~f8YeGj
zWS%4L)4DZ(-k|+LPo<;;+cU3ax+2Q6atR_02z#z|&OpBOx7|>~{dGe7rK?EyZ)8ru
zWPqP<=R!aOSfBm6+z|?k4`d--TVj+7et+k+3ZbFE#ymXm`u|Tbv~tdg=ZN3?%xxT4D}>8FmQnnxCa5f
z%qXB^R?5dp9XJjJI@iquEeLPwbW**pQ*jo}SL~QafA26Ge24|^ygij*pd4{FI4dUnhoH4O0MamelSg}
z^8IsIx0{u0Uns+Valc5wBop!n`F@)^LjJB7PVZA-KUlTK{cbv}|8<$c;U%z7!ND+%
zFD5n6+k$pov8rpKcn|o2!IMHR&|6(@7u^pqb>s#M^?t*Kx+(x%@+~&i&T5Uc3*Kj}44OKr8eFpr*_HIyn27D?Dxzs5@yPgW>wevx_@#5(R
zhugrv-_*%=cmvplP8a>$3j54XiL76fp&zvDT^GFuxm~%n@AL)mgDv)U*aESEh+2Zs
z_%HDN3H$6#ykXy3T&~C;1pU#iVsV5G`*F3%#!Z4yE`3&QZWhp)lLl@uw11rJyIBE1
z7ynN-U>)qccfQTgJPzk=p)(f7Xw-nbQYF#Rk-;YBUXwm8^yiRC-VNd(x3I}tzNy?8
zu*=S~(>kFL0sZgvqb?rvhW7~p7@p1
zzFL9avnZcGep4Nsk55WzwFkSFm
z;m_kU+xkiEod~a7y3n7|$L|YeL%bxL3g0Uu{X}-{CVunS
zelm7zb!#usP4nNf|D!RCPuHoxKf~`|FtPQ3`(f~#2>uiEYzfGZ(XIg@z-LRK@K{6O
zzu05;_mwbSx(;@azX1B3PF!$_Yd6Rnjhi%o0iR1f9Vg6&aaKN7<_^xQmWHsWxiD^X
z*1h=I2IIZ&xk%@S8(_bE)Sf>L;x>3~8MF0gdLLQ%&Gg$Il|J%fiS=no$3CJlGrg8X
z!~M^L2hGbj_LDIcC$6g2^^s}w5~gKJLEN|ivD41i`^lN!KkfUM!MU&PYlY{tVV&wL
z+Iwmh*hL3_U;h0R{K^c+_141tahtS9J`DK%`y+lJ4&>i-HRHvP!46RN&^GBk*jsCA
zZpNsCKbylkOQj&-(|z&P=gR?Kuw8vm3(%P;J=VVna<1c6i*cD=SVc
zzV`;^Oa6MNMUv1zXGBy4UO>9l^yaRMP`+u2#A1x!GV{5$oD|6Ib01D#y9mB}Y^YFS
zJ){>4Ew+_}^+%%P=bb1n$dSu-`uoFtZkc$=AP;^YZYAs4B#`5mx<57#208k$IR9=x
z=!b)opN*-2c5JLRYQ^>P!#36QLC`a5oHWJSAw6}a@ML?C&y{;l3MB$=wxr@F#<_U-
zX?p!`fbG#n{ul@S-bamOtzc1~?_i64wOL+PMzVD^8&0;CERE+g$`D0AGF1?LOL{6%vk2
ze*yI;vpfSVA%Du@pHCA2wmR{D!S#7xxsMI5%O`T4xGsYFUK~r?WCwi2Mc%eZgY;K#
z8oKm>&gkyz^~HdXw;iJ?4Ee)%B%;wTp!~bO?gGd^`oy#m{nwmQx1Ydu{pQmaA@nyg
zFN~3#2l=~K?|+T^=A9{x!??d%sT`|7`&fougAcC%1+h0e@m-#13+;770F|?)>~X)W
z)mi~B%7H(k%=LBk0KYsO!)XL)lw!35^D6}_H{t$is?Xkz=Zx~1)h9v$wh2YPMg68`
zL_GxUL|fd#W$^oTB^&c`zbub-Yp4S_?&e-+lk&o
zA*pRi$IV
zG_X6_YtnP^6jt0)#2wo2x*}ikf6{3^+DSXWa0`C71-_K8$ba}EB=`sR8$%)
z1s+j*Z0y`A16V$ty@cl9TX&AiXjW8bcIc~0s07XTUX
zI~KhB5Ef&%LHQydXcpjSdC*+|72z}=(_e!&gSa6?0YO~oE(0xcfGtQDMHJGNP+SM-
z1ivn=r>j~Rt8(cBL70;76HemawX0s+v~gA|PUs9hnR(8S<+aB6_~
zyiq-<%L&Zq_7uqVc*!HCUkW9r)AXxgjm7+o
zYGe@Bu?B1q#N`@b=zbn!>i37)gZVqNnRflnVDN#fFdTqB&lDL2f58I)06ua7a-jcu
z1%4L5Yo(Zea|JI`#9s=1p?v1AOb5n(kh~>1bo>$3N1(XDeq6hUAgt(fV(3kDX?+jw
z9ih1THMkZ5_56$A7e%`D;}|`N-w)h5AzdxQl+rueG2=pG=@1-zfZf=I(VJ3VfcgyO
z6k;z?dOxjfir?Tor~FivG5TrPnKqiA?7`G8WciJbGox%~+^Ik8gpPoFusXAiu>LnH
z1NRX3+UXgC>kwaQ$S>-bRQ!+5SI+`>s^>%t9ijU5_t0c|K9VhOJRa$bp5wrQbEpD&
zx?ZgPbeGb3#NrE~;1}5bpc~;$E-ae>r)X!>`E*yFiT`|Y+e_+?GVM_qjBwcZhc{9F
zTh1<{?OO7|mCn!me=F&ERA=N}x-PsN)(FS@u?tOB$D#e-=m=jh`bjI4)lEVS`y?L#*!md`=_
zZP4Mx*e)aIb2W%h%Ku!9{G3Yek;Q$`YSE$UW00;RzDvsw@c59Y3HOdi&F&{4yz!xuG-xUV%Y?6Tf6p(9LqS|^&o1h(kR-Z5f
zJN|_=<|d+`_e2c{tDh}|{lG=euMGG-^E^M>732~%f2-g4ZsLs@0n@61ZszW9o*Tf<
zIWc0Y54NLZ96x_M&|e>8T6zZL-QKm89w$I9in$T+<^;4iRqJ13BJg9C`OJPA^oPQO
zSJ@ahOg*)EJ=z0`4vS=+2f6B4+qOF!p}v(nHnfF6dVBA2RehkBIP~sT0?_+}4o>t3
z*p%PtdFLI-(QPW)i*rCPF+E(j0_5>GQD?p9f&F$|uSHdI3+%g3tHwSDK7%cCertg|
zIKl5k%tVmGT<$#z$OJvf|n{&vkvZya%+0
z&#C)N8}uK0pn60LpqJ9j!(jm1UD{l&T48@Y80#npa{h%4udDAuyH}N;?KOt;5=T9c
zLwsD`;_pe-){mw=b2fiIzqms^i3-7Xc+lsqFq(s+I)D>
z5XSq%u}Y_)oaxlc%AG);cklihlq0vdOsg#c=)QL0%2xm%mRap|1o_#+sV;l3emAI
zJ3t;6KCkiaF7T6Ie#%dEFzgtChHu|5c
z_dJZA26lra`(;J8pw|p%uegE!nOEn3n1}b)!+s8&$pgGFJ!d)__|S%rKIGr9e7idv>T~sv?!a;WB4_&=VSv+w4}U)Z^iOjCB{=}Jd!LYt_MC@a
zX`;BU47(P!--mMdI{${_diT=NC_oGAOWRwGeuDcCd&O5men8P{DkmPayc}T(>0b<0
z=z4OfVd2IyC|98Bz<(Li?}p#_90K%HPYjsh`m#QGm4Q6Kb>GjtLi>wR(oO~R+r2X)
z7J+tcr^8AX`1_8Zw{JNM@LL)$({P-<3>`UwdPm7q>*Z+gIcid{4&~~D3(e|LuB|pN
zWTD(GB7bo!u3N>Hc_X-P8hETXz;(4_&pmjN8T@zbjA{IYR%08ZLP(#o@op&IceUpj
z-oyONxZCe>zf!ogZw=Plqp?^q4B$h~FE!jp_D0D)LAiZ@&|M3ZtKVdNU5oJUP^JN{
ze{pAXOL09?O=-b{hFLqeZ4C#4baD?;tCCmQedRnl=bU)$&*_)JNr*y}qVY&WPa;p!KXSx1{B-*KDWI$aEH^(|iTT
zF~Hy5hr4L{>+p3Hs_1!8`66eBCdKzJ*hA}|leLE85`%CY0p%<Y)tW)`
zGdBs+@^U`V{ZP+JIp%xP9?%VdpRi;2UPDm82mkSd&@8AYLK4^p`~|WFc>a<%6mA)S
zV-&!nVi>+W3P3yqJSG)B2k;^nuIm8&ti_c7itaE_k7El%mx(_!;GHN|1Kqn(`CbRb
zYe-Lf$)rnxH5v0Y5dcJPF<)vF!euX-Zk)=DA7L_-j4MGI+QngC|O&!UTNau?|47k4(PhEC!D}!{9oBO#SmY4DKl}S
z?|+>xXLu{)2Q{sL(QDeJnQ`Lu0Xr7rW@j+t(D3piT5b|p
z1F(E7ye5eGx$wFm;-^020TismYd=!>a#W8RW6jVXk%nZ*H~-Dh>(A__xc#^-w7lLd
zTRQJ-ZpE>Y&aK@ah7Q;#Om|Q@y^HT7r5~UNY+%2C%$%>-yn*Er=%@H^rS$8=nD&+*
zIZM;?%G_xERi83}Vem&DSP8!b+~1xF>J-2g!t-dpiwqoF0&Zm5NAXF@j9wR7&HT>(
z*BE{!?}6F@e9gF3O#AuSN7RhL{dTevmwf=On+sP0+(~ksp+(
zKh(S@p>*GiNLDVxb`B`5DaP-o+*t+|0?@lY>7GP-FfpLQaeL1Jp;JB|kch`Nstgz{6OuE?-Td`SF9P!4U1_)y2J%r&+WNBxL5`CC*4Tn_z;yF&
zhn1j8$wi
z|8RgUTNIC?TyTK*P5}3Rg@V67k^ZC1K)5jA&(fbtV7VLNJ8$g&k?Wfs5I>;c1tE%{-`8Gyh5i~J*~in0K#n@n=xl2T?HN-g#7zdcC@OZ)
z73hb4Jr+m+c$R;m*?EAwidx5-1Kd3+_&fUD27YB7NPzmcXjNw8Io!d$T^CRu3SA-b
z4E4O~hSnyOzq;zB4m{v0*Yyny{^G&f)t`b|g#zSxfUg?7HV!-smP!R(`n_&bK$
zx%M^ze9^Y51MRdsO2uW7&pS(Ec3}BWzIo4ZUY?FguSYw9_@*Hj-GZlxoz8ue4S;zu*80vxWiTQBj6^vl1GtmK0XmP
z%Fjw0{$wFP8g1ISI1YSg{;5KKi*!1wu-(7+YkbG_4ci1H@O<=NQtEBw^J!~}7xFpw
zl~ZjGz@^ny-dzBTM4(21x!Y6exGHLy?SOG>bkuxRw?I9MU)L(({jnL@>`iYV-Ie$4
zT?WAF=TqzOo^?{8RuuY8)^!Ay;{3W2aU>YWy|P^MUhMamvdPUjKTj^pJdNL#JUR6$
z+Sz5xk9Okt*s3K{h~tb;;%*S~ce`QVK}?^s^Uzsr$Fj%UZXkaet85jpzOxTw-(z~2
zlS~Jt6I-|k=ZpT|n{zOqz4r^1t3~tGZesshIO;t_f0S`|cELshdgH@60Z8w!@xc+-
z9c>r(AN)NdqrJCaejab&6{e3&+-!|-yX)ChTvrm+wtxW**6n-V(U>khq79M_oOAo^
z0!aYYiy3?C5Et7m{tCyXYG{Nj;=ZcO>AGGitPK(o*gGz_g2V*pzS6huV!3asKT{CG
zVHe%sB$tAF4&X;ruYlkP>;L{wun|K1N8Ses9G~NFb==4D(_dYq>v-XxX8L{I{mYt>
zE;c2NuG{@(zvw=aDWy&2cHYu7*Z|@86v?4_!7Sggu+f7bxCV#H$;~yN=sp;gvuGI8
zHIIIz^xtK}V55ig&`fyV2Gak`dPeuRn}_DXh7J4MdND9`gM4`d%z}{LkmyU>w^&V+
z=8qiZr~ButnK9t-4gMes^XNV!>#aiNa^Ipbx=+2iwTsrPw=;{j|0O3K46xuo+vQL3
zCQn8__9*nEbcWI=DW9i8BWZr(S4Iw5H4;I;2d8Q@-A~zr^08_i6odNQZ^E@WfM2ua
zX}}&kN-)&VaB90Hxk;p>#1@R?~XgK3CEF
zv(IZN9CBjh`pQ<6WWgSCoROD&Z!>axR4gO6fB3qb=0Et#$X^rILAOBtXHYGIaym#3
zc*(W56b^(k^3?JOCVy-zunqb5Gnn?N{bu~{eReYPNi>W;tY^6+lV7s|x*u?dON{*A
z^pN?krXiD`XwQ6a{85JP)Hf8kA#TQb23Mzo=>NwK?v0;;{HK>0KBi_d^-NvJq$hVU
z^&M|z;P!e3@=&~ka_mk94@Cg&C`9oU(w&`|?+K#V26#P;R?LTBofN;1VjAH8=?VYI
z_gR<>>BC3{<(`k?*jtEWtq5U|QaV~4;A?3H+@l;#r(}@+AEfFN)O-H_kpCYY?G7l1
z(jb24zjW*seCNO_iuLE=@Q3n=x(wbs0~rMW0&9l;
z>gews0;?U;--Oi$^RK8gc*m%|p*X5XK~P+p?}*a?=pT&gMH>T|^ck?)VEOVex3D=MzJq3@204@n+=ItF#>{0NCnS~7myT~&+3bmR60RFib83g;!9fm%33m({k
ze$vgX9|us)0sOEPLw_E2GsLft>S-%j%sTKqC70Is=;8*t&bF~?VF3X-;BGlxw-2~i
z!A}l$=`0<}r{XvPXb_xl9hIc*OflL@>m7br4F`~57fNK#Aq8}f(Db}zezg8pw{}Xu
zZFN1Z-~SlH*Y#CQyOh!yJ*whlH>KZr5@$BZf6h3O;XdnUrXSafFy%H@G*Et`Qq*XE
zLQi%G(o1Ws9KrosZujX>#8({8p!Zs?dkM;+{P5+G8Y~6ikKzW$`hXAW)Kk7gJkC-*
z&{%2NCM++auoF5I+1A{5Uu
z=&C~eV^O;o!fmDkFf~A~+_<0ib4Dv16G8cOKuU
zu^>kry1W*l87vvtzGUA`cM$*Cce4p$cWu!Ftj~Cx6YclF-RlA{U41s(JBD<-w{Uy|
zaANUl82F%%EHAY{ytXUu1465z1X)a1{#7N2_-fm@k0{6W>a2W+`b&I#jWx>U5emJ6
za{(Im)Gb3f_DFMsB+6fUf!RTLo)wmJ84kd~KcU?x81*2bYbog{&mG>l@*dWoYqmKb
z`JcR7BO2wjA4iQ$5k?y!J6z+-kKskC$Q!PE;w6h)3#^2+)
zQ-uk}P0$ZdDn~v~Nwc_?=gS!4f|@Wue)zLoGgR>)Uk-#eMd_BX!$Rd
zs&kO8L^cz)BCy{-Z>Q(7)p1#i5YNi;*oJcVYvXrA4FJXDDZ`@RIB`sM8>kYDQ)`r{C$uUHC#9O&m)_2~R?4>zLvyy}=gOOgIZl)N>
z=z_i4k%51lmcT*)`qJ^=e5hBJ#W=&u(qQK?yBLXls&3f32J8KpY7>Czvm*jj5uU2L
z;*awo{F3Wk%-3609ESB?hICjJD;xIzze+d73OH5z89xbAv4>k$qAk2>E
zhrRg+`QEjL&hOw`%2d9$no$A^JJ?&sOVD|oln8FNfG-@Be~9q6)O3(oz;2^rL)VR_
zsZy|Uf!(rCyb1XW4=>3_+)>3048LG!Puc}rKAanA7*9tyVl)m6QE)Hm=|Z}`IAxuu
z_MIWoI9kt~W;kYma%;cjQn`P_N5y(WU*f;M;3jPxQ;%blvOopHE@q&0-48-#n)I
z9MxuuZ?|FgtK4vAzwo_ylja{j4Z{WM9nXd80Zx_xw+Q&|`_-)!KP~M{`5Y8rr8&F6qElA
zZe`;4>|vH5+`I~!0WkL&3Mim2h%o6A*nOCO1Ck;Ax;67XzI!Mjf_-QcQ_n^Qn8@I2usUMBJz!PD{1#My0Cw3i>6bS!
zu&j+qcO8|dT!ff(=NtwO!)U>Bvr%jT=z#1(z1H|M0v@K#@L6?>!M~$B6{JrXmH){%
z20wO{ff@}A#3-p)4+Ob|KhUpX^dYQ+S%Pr-Xnp}U7jSA$2Uv>5pqwCf3&2yO<&g`-
z*Nj3LEC%#zM)^5|-2?db(frMb0uGxZh4TuF?BYx~L~u`|+RNEo64rUr`IZ=$_ze=l
z{zm=8>NkAsn*Cy(A|}fkXCAk$To*&nEI|
zIS2Px^Dg|^
z9}0devfsXr+07>3Uo7efyV*f9OJX;jhx>PL{ckw?hJ#$O?9Rx<)o^aDds@n83Gh=<
z>cUsmPUh}&k}{akLy}l2WA@d8zx^~lv0^xX50qauFtq~s;7L9#b?hdW^Gw<`FZPm#
z%qGW$T;qJRG^RSG5g+XPkPCb
z9s3G3oLFSz%NK5HJv}6G_r<=T!5(tAft%wZ&_kwMjSuSc?;+nV@C&{ZW05C^e7prA
z?q*t)!n2w1-ip4O?Y2AqpGni}Sasf`Z=}R*gJcNAuc%&CmRs-JL&|j?eXL68Aw_@B
ziv9J?NC@Xij)QhL(TF>+`c-WY*>F;3VtZ*1
znXRMZyH>TE9ND&5bk?{oa&Mld@ZR1|vgS1>mOrhBoU*&M_wEyTckt#+8}q~P4xpOV
zn|`A%@+PA6`uy=-#Byum7suEh;-45k?V48)(I^x$KakK(cGVM++X-D{Pq^-y=MXoY
z
zUv!gKPoj5~wL;u;-=ds!h`ZkGI94>aw38SW745su=_E-@O~uAd?;=|j5)F6~U8HJQ
zV&f3JLsNQ6_TRFVUF2q6<@A>V@O}MLl^yfD$gZ<()4!#QbnEtPx#-qKPR!bJ)N>=maWCV2jDY+b+m_91Fzq4_Y<6GS
z1Md(%(olK7Oty=-smmrmU)x18Ed3f}q5jy3k$RGEJIR#ncgELlb&|%g#xZ;Ecaq&L
zvLUB(JIRrhifw-$bQ1mixBJ%Xb&=&!C$q0v!+VAOxx)UoUF2H*{PjULUF3xrCwBcQ
z;OnbM)XGip{$kvL=R3{2h?9=z{d@5KPRL8!Gk!baeMb-7i5AgaWRCJ-p}mkUw|HrM
zA@Fhi`=wYt3wT#ivtg+H0=(z5?AdZQ^rx6*LfKo$U#xp1F_v_ZhmY%renGhyqYuW`
z5P$ysab?HlY-rEXY2A()og^porI2TQCmGjwIK&EYp|lPYHR$&;*Xw_RfZyj0dI^4K
zfDaXi_Q`j;$hzK+-wAzPWZcOQ`)1GVhW9<2I@|zeFOHu(^J*8#3^}yUth@{2mG3Q%
zpV3VYY%#H~-qcObcFv2jb?+uWzD$={P}fC{h?=gA+ZjAZC&N1co;8jJu06!5
z<;$h>Nj+qN%Fk#;;8P-HDEac6Ziq`47xzNGhm3r5Q9ggNhdgc!jQ7mwA?v2S++Pjv
zamI+sxQ{e-!@ZZnGrz>4o*Qj0-_0P-?z?>d2AKEDq95v|uIVPB3w2)bHgveCCS_L<=g$wMsD_F$np!eL}Ct0N8`ts($
z3oJ5?|LTsm_#P73I8bK-@pBZN_WGsjvdC(YIje0WSfop2`4qXwEaGd=SI++c;`c;J
zl`d$8cc}WsvNDudB$&JOf|x&x9C`ccZyLl45~|oZqv&uCNo{^tEDZ4mI|r9W#Okof
zc&{MWAE#NwEj_!zF@r^(O6wmec-BqSYL2r`vY?rv2epaLDP0Qrf8>kr
zX3yv)-`u)8ZKgxKpNWZjlIbk6Nwv{xKEyQ|+;ia2<60I`6A3iy1}K!{%BNEa^{Re)
ztKA3dwBNcs9SL|Z;ykZiT#rS#vnG`)K|kjH3K{qb@UV!e^|7ZcB3L25@n95-JhyMY
zzc80Y20m;Tte)IUT&A2_qXzH1n!c>L)y=elHR3?r$b7ERxguHup2!XJ001SlH45_tYal_q`JACAK1#Mvod<
zZ0>sF5MiRmRzJGFk!DDH(grXTwk7eo6l77tt1u!widfuO%*
zdWpQ?kA>ETupfVsEt&7$OT2UHyybzZ&s1o4nC+5q=s*APS9$#!
zy`?9|9EnNtBdf#o)nyGF*C2Pdx17x&j=6Nctd!1cP5z||bk|Aq
zO3ywr&Sv}P2@sDUuTt`;@D_;gkmJFh4fnQw@4w&C3i9BsV2yyS2jITlJF^Y>Z^8el
z^jkopBAb+#`BXo)Vv}Lb80T)d*S0;eRi?uR`~a<2@os0pIiZeP{8~Xak+FWiLS!|Y
zxbE=1cLeVBZCrl1W#Iy_FCNvOTjvdSL4~4$Yf$gh_;Rk}1UB&q%|1A5DcE5rKJ}J3
z2X<|B#UtOPz#e5SzHOcd$h*z04mmNPSH}-u>w@^>?~{0!%f9v#gXQLmtn+YQu}xa+
zC){&0Q9a)@UmNsb1*eLS4s4RUWQ;9mKbyFb+_U>mu!&(s#O-SkXM^R_@;)}7O&(9n
zo>sS(P15%qt_-n(`;Y8Bxm`!ttjET1RJXMlS
z%J{b#7A$3x!Tc_>rmY}fFW?ticM|l0H+wjqv0$fCjGlBL7w&1DZV2K7Ii2@Q`&pS9
z+^1YmZfh8_Nsn04lMWp?zqqQU!`TA%qZ_Y-W1>Lsd@19h3H{P$yTT{78SH`48v%T$?{*_Y$CetL`s4(#3gxkZIk5&4#}z1ZOPJscqO^@5Aq?-T$1VrtNJf&lKR_Y
z#e{hfug?E_LAwTre82eh>7!j7lJa10khC|O4BcCC<#{ff)V$O_DLex0NilDCp2PwF
z?oH>X7;?xj7rrDp=oe|(mKm?z+2m$fpGS2e8{Svk#nuD5^UgKt@$wwfJzm;-`B$)W
zT8$5U@)PXpS$`gF)?<@tzt<)u+=c#Cwz{AR{pT?B>-f+N4vElTJ8Yc7Awee$Zp6EC
zh=}@x#KiL)Vs&C~Y$e3QQ<+t{>$?SqXvinDcSmzb#*Bz*%Pw)qR9XMP(|H`i)#|92
z(a#|lV?tZ*nsG={$zbfEy&R$)+{n>B%^~K0pE*6a4*aDo+OPuR0a^;ng_|tkkP~d<
zk1Tx-DXS>*tZ?U$e}|Js?nZFPlkf}kwK6$mzIfA3`4Tqiv%FCNSm`770|G?lr;tF~s_jFlX+YR2Owca5RG(>C1H?Htm-Wtm3%5r~(#PL6XcfJ1Jp
z%$YC*aS(@R-*LSM?;@0^=bmzd@BHe?=B4VCSTv8wWO+Vr|
zm-H1(xfulYtq|B)+YaTU*YJDX*Wi+GLiv`})=YmqCqbBRviQ{GrlF4^2GpvQ^elFK=rdby8)?~vtr
z>$QqZ|5fnVe0>Q5X-k
zCl2=3zTgnE&CfQKLOjT4&N`co*224}tt)FTL0s)hmwFHO1Rhbd*SzpSBiVKOVX7o3C!|!y_K2ekPn*$s=Kge${E4c;v9FW%667*ULyj
zL;fzbW8{PKj%FTFJhsJ865bVaTU=9qZyS&7?)mam!Z}Lc^
zmPuVzIG3Eipk}$Eh)aG?^Vuf~{WlPCv#EPJk0j)$4BV3Bk&A!#-&*#DOF|p@I01Kf
zM8$db&Q`$zB6}t#MdS&O2#uR{si+g$GeJ4-=3X9YX8-WZ8|0Cc(zs>8$?!f~fZG|(
z5FT;vGRe9Md_J>o`Kj2)Ba5ySh#suvku?c2THQ%J5|d-nY6N_}aGo0ymBk~cHj2I|
zh~<%ux_u2xGI+$@toicWFFeu`P`y6;ACJ`i^G=W9@rb_q9#wy79$E0E#LCVDaMs^1
zli@v-vWR9We`w#JxSrOPFfO^`ys!B3v;pGz@37^aDFY;yEqP2jo=4;(BF?UX{uQ&@
z-Fv&0OYAkWv&IzjNU}>uxYHYG?@?!!$V48oi+!z|?+^X4DZpQQ9*wJW_@{eAKr7Bc6aL2;fpYTpJ-2eU&yKI&CPU*%x_Kb*L6
zB!)w(CPr#yNx=FnW;OKB9Db*`CsRLFaEXIp=;gRy98yy7Yy3pmXV-?`KUo6t>w{dk
zuPXNEl0}*ILOBw
zQ5nKVO|5JQOC5);DfUF?02k@6&VU%gn!GqqSg}9N;S*--%Bg8aWoykqZBg!eRPG=D+`c{&(yszkGbuVEg~|5A&h_eSd1o
zRF3lT|6YH_@4z16`~SH=u`mDYPj~qL|1
z_dTsq{C}m>9ZlCCh5yyBF^Vr8g_fi6zk2?I|Iz)|o=N{px&PX+dNh6RD4Z|~
zD3l$AM@QlFQ5Z7{$Bx27qwof0poSEwVL}agFq+;vnw~L=e^qDlpN!&dqxr{2@sFeU
z%u&3EHn4_B(GIO4x&I4&`zT$|GW<^s**NXN9oUx(lyLy=mu6YP?2_64S79E
zAF9fvt9vr=`d$VW$T9HNXgSaI48DUlyoMBSXYd0u3^ZQHK>cwHv?{?5)sO-@U}{LP
z5|bXeih<{9kkJ|vb%DWebu;*;(R$ziVsM}Ti>d35=lc8p*`pLjBlZ-N8BCI_-5D0b&#N7*DnfysAysz_yv*|-zf#Nym2#ojX_P%)b<-tY_z
zh5m)@rs*bQM=5sB&W4V&i4TVjm7Vv1r7_}#)+1>@^Lz>*+
z0-*l}PP3@q!e7u)Hc>s-Y>3G5Smd2Lps;Wgj0Bq;P5cp$>FZOm7Ml
zfr(&~0X|MgEkH}PS%n&a2j^XVtp=EXVrnnu?^4-%6!Wc>wRlwk?=<{)i0Mul5+hho
z#KInBl)0Xd{X}e3X`%Lrt-ZSm^X0h*1`%bIWNChPbrv1JN7^G+*iMbX{`XjpXF(3l
zpVp+LfchsI+(ju`7{1j6x%N^;B;uO+kLNB!eV36ZSk9vMzziKg%kweXls6
zx$&;ZrGUSAoI2J5E~-0Xg7!BB4bBh(43K%|j^#bSE&7h_7)Bi*KrAnikj3=o;xt(B
z(C)sPUudV6u1yQ(x6KSS!t|B44`@HO$cb!0Stn#+6ONz#0m=D@BQsS)F`YAeo;l)K
ztt(%#ywHlmgNO-jP8ZPLk?BQy5pO5v6kz(gk=$C;-&*~`R07cBM%EIvd)I?-2thDE
z+a4HU{f0fI*H-{$Zv0B;-BEiZcrrF|->XK~jX_B;wJ&*VIbAQGLY~ldZnk%L6Y2%?
zH$jMClSh$rU}M2JovEPxJ+ai0t}EXMnL}ugR>f@an6MwTGt3b4)!DQ?iPjfO(2lF)
zyaI^dfBMmRzI19UEjQ?sY=Zh0m)TL6FT3AqDdJecg;h9@9|c}SV*R7e+Lk!J!%lbT
zx)(4tr~7DjNPy|ZL-lvZFQ9sE
zq?WG3+OFLYDWKh$t#lvh*SFAen$u6v@yMPo3LcLQ|EKtzE8=VS>vaC_-%yoD`HyEJ
z-M3tS(}r-KBENTU$8?MR`HyHhjpO~;4|dA_MJPWXJWSi4Z)Z)>X^`>P=Xbz#K)%r)
z#t+K7Gk$KrMjcImRSDAo`EPD9`nO#ezs$KjhWeiqPo=57Ycd_FUYs1`@AKAyxlqn|
zV1)KJ_|Y=zxAHsAs<;#JL9Ci%@AFQ7gM*c99(eGT(;JtAs{iqOQC;w#put5>ZX@K`Y
z)Mhbw=?_dN$h{-590A{zfO&vZl$d(%g@BnLXVz_^c8h#q{P>Dym}W>navuXTo48$J
z#_@7DQ*Iy3CZ=zkjN8|-8jy>yIY7&1MsFGHGL&(&5H(>{A*z9wM$|54u%C~?f3P|*
zy&c0MU=J=+K$AoU=Z`U%^l{RDYeqJ8Vlc`Oi3+q|%*dZ?Co!MVKlz!FMQ|8EF9DMQ
zX>ma8HOjn`SOc?)^7Ki$d{Q44D@{+Il+R9%|Jq5rZ%pP#P3BunwkH6)1?#Js?ANx*
z@%ta&vzr_ro(0T2RpPP%{rpM(lmfdNWrNA_$6bK3;Uv=LK&H(B(&YdcK4}N;ivO=q
zX&vNKx&^py(oWj#|5u(i3o`8{puyyPzniS@f9+8-A^(5l`9FRbi^~Vn!zb(CIB8GD
zWO>}hm~S+Rw0e;LSFinKIvpNJr^^cPfBpTRy#kYZ|I7bhzv!@oJ{@krE0cPblV~uB
zbooL4hDmvNvY#&|?cF+={=fQ@C)59%oHwt@{6CZF=O^W>ll8DC*TPnI8>+~+=X8NEMrA;7+#EI*F!49f4P1HgvwzZtnflEEt4
zVc-*!^5Sz$x&*pUDCg(EpvUC8zBIWHQcaljJYYqr=LyjX(UeXg@LRgT*o3-WNS8=r
z^7q6t<>J6wqQ2bZx(=W0*W(vB(`<5+ZZwEzbO6~T){oJ@DZ!u{mjcyW+m8bPo8bDak(WRuR>IDo+KFzf*yzrg5)mNNOqOBjEacVQpRzvIgI4Z*26;9#E+Q_dxN
z9hF6XnNwU*!PI+QfN6i+3p)bwvvDRxv73*;5b(1F=+M}N9NS27S{rtZO+r?OQCz`s
z1!fZbN+1ph@I4z`QEZaTCrr`S>kt^sCZBIkqxErJ9-`^O$uF7u&3vf5?Wqezg-zxRTgJs?wLHOaFxM4swbiPqP?MwXVZkcpt>TAeRZ*)rh_6l9ZSEw|<&>aGz0&klS~cpnTPq(3vHb|v*Y
zo<>>XLMyCPHfecsht_{vcni&+yK^^eJvfh5b~hug)Bt)3IBmyVSkVv%gu|$ODnJw>
zA@KgGy0FzDuH2hzjJV>>ZdiH1SD)X6fn<~F6D9wUN4#0mN_nWjdac(e4{a~niTN2>
zAAX@Mwb8R1Wh3rqcafJ{Niq4cwF{{55#~CJm7KJq;HAa;1a;0m_lLXUu{?2
zWw0I@mI)`yy3X87kdNtdCoEJ3?0TmE6nW07x}^WG{0r5?{}3Mu2#Fw{wk+ulM!m1X
zA%e(1c_jI%om>++U0BZP+WQwP0n_G$*cb!;Xg@n2`Kw&nimg~)u_P!}7tq;c>2}nw
zeChf5jXL=As9MtrKrK6amyM8abB6OR)^j>A)tLq9=6t=Sd>~&Rjc}9%6xvw+8T-+|
z>o~^%Fk#xSWoVb_PI(#R|DSh{KE{4MXSJFUz%q^R`ig)R|D|ir2h6E1OIQzR-)&H`
z4)FMnsRLNfhf`@D*4x?goPR&y8_wm@N`M(WKWsC=jt?>C7j6Oh$wL@Lcm^`lXMn0Odl$HtPc>q*$ND{wQ_m1EGihy|Na@
zeuvv!`-0m>PxSq>+*Lg&S~u=3sLx;HNjw(tN@0KUajDUqSikK+|hy0X-t~VsM=Q
zJ%M-2&o`er1GsL|XI=sV4EHnoLB)t4vY)a<0kxxSSh!B)!zH@$oR`VF
z9he`KQw|)$_Rgoyy5SFYjyN=I#`>4?jjhA=Wk1|ryc#h6npcVu;O}Q0=2))u#p{J=
z*N(L>s}awo`qA@@`u8O2M+VmRf5Y@**{|o&jtt9ZJSdmUJxI@cNAmn=`GTJzs@Ps<
z)sP>eOZpxz)Vq1Gtr5#j8PuWsw5czqRvzT)r85SwebdSlXRzNrac$thVca4l_n>~R
z{?XfL@5;w}BXNFp^iG!J{LMZ5#7`S=iAN(R>hG;CrhX;CIgt9NS&R2myE7a2QUA9{
znDK+ZJAR9yes7_z4C1+leyXQaG@JU<=56yS?k_DmfcZ~)q^KXuxSI)qhfM^E8Gm2<
zSdi)kJ}lgfc2;hyo`Lxv9R5Ls1wSxk41tGD{9P~6__4<}$OrZAE#78w$4Q&0&#Ha7`0Pn)p07n%wghdOOq3oYs$aS
z_5?;3Q~y54>j@o?w|$m0fAR4*6bBbuQT#^<8H@wW3f3=+;Scak6I26u_#z{JnPB?&
zbTZCMc*1l*{%Sn71FD)cSQ!k%1hUO(rvLLfARK|b`8A_I(}BS&H>;_g{HvI_&(Xug
z{o6J$yr4gOJ#-WB{vzfa{Py*8s(;(Vnx>bYfUpPo`#YF*aH9`IkaxnYB06B02IQLL
zWh!Vs$XDQY2{8ax4Pqfq3!uPczMnB8Uw_5u@!@iW{Bt%;y2+$|i4Y^ddd|#K+IGg?
z6R>Jf|7bFkuZhzF@{4p>#()#P%)G}>wm(Ch(SM=Jl%F>_5ASi9A3aFll6X|v@iF6<1>l#X~p1QC+*P1Z3X4NQv=x~
zYa=7KW;1w*9$?udY?7}ZkY&=Z?`JUY1cMJO7!+B=;DO0{KEK6=pu8Cx0OxgKCSS{x
zLG4)#)}La~x{=w>Pc0c)V6wjR^dJZK(X~waf3ul=GVz&_byhO%te?CO&;E!V0v^{U
zPEkz)7cQJH5^!MHWZUF@#WH$e2i}#*tV^G9W`5b%Kn?Vk7%=q?1~KXD58Ki7H#-=a
z^&AHr){QOWpK8(VfZUbkk7%PhhDd0){%)Pdx*FGwjR^!5h
zcM^^{6j$x+py;<+5e$d>yz`9xJhz$lQ*0Dy{`&@?fsp^-?*dq{!29$Is9YV+_>FJz
zWmM*vH~|yECJV!QV1R%RbM2yb`m@+H|KNFjDlhMZ@DBO8P10)-m2W(w?H8PBrsYpv
z2O0tCSyB*&0h=ZarXY%D6i~hNJ3cTlu-@BwD1Wyt3lKs%q;@L3hxFktUVw6v)FqmK
z+e!fj4)~<=7s}(Ecc-34d7SkT1`h7seFCUmVNt2H{TKXw6jv5!IADJ6`fW6h?Oy$m
z#wV>E=CESn{`?~RJs#l0KsG#4fuHAC+JN};-mBf1-X|tD6S3Io8V_R4NlR(ucj7uS
z@yH7f%DjUG1nZ_@JtyMwuk-FAI!G4NxSv^c>IlknKPXWiayX&#D9RrbgyG=?{C@ni
z>B#TZxKHz9eYaK|a*-UiXX$V0YJzUrvC>lKFgefrp0OiER)9xroH7A#d-`c0aNY@onD#
zo(DR4Qg2!V-(D9sx1a&)caBbaQwr&-8#FWfKn{+2o$LVlX-0Mln>5+*_qNY&Ow|J3
zwe)EA3&?*SX#Dr_GU)e6>)zNEa4s@)D!pvPCV^|O%T+)^LO-IdUrKhb?_+@t+FR+ijGwZ552`7UcqMZ8)cFR^ZcB0>6@e
z*<}mh_w7|z+6c_cnsxf0Enyv%X0zJm*zkW`xK2CHh4Vqh*DP6Hq5%8ArFY{!u^X>bk*O6c6Z>sI=2D>IW
zp4XRy{NN@R_Yh#|@{0lqs<0k7o2|H^Ty#;6vjEhWsWhy-2INn_jvhY+=r7#s_jM(k
zeDT%KDTRI-e~U=f1MDy;HM<4vL@n6wn+WwiOXN2^?*iwNq^GOjKs~#ZI&S@52>s^M
zT&1!W_SNgnp+knSe>X|C`9Qr6&u_xPmrZOh7hTu@811no4cCAAtx{1tz^T8N8i+tW
z$J^e1ehvBpUMEC*0q@5zD#rEh&e>yu>zLDGYL5b>?;3D^{1@aSRks^z0gaE?Z^3nH
z6cq1;>rriA(Cb%#vFsNdmM|XXI^FH~UURK*)e?7T$Kyr!^SvN{Qaf?~H>{f#N6j9y
zV0;ENTd!-ueBS9AX;=VyH9ub7ISTmF;F1dNixly{@(O_4;|4Twe=N;;za7svGN$)-
z;{JN|IIY7J@PzZ{XLt@wym;@13ZUVlb!%|HJY6??CiX*yN3Hcc*gN3!o&(Q^3h@)i
z@%-So^oTpwV|&RA7&G{vy<(t=4^ngZiqkCJ7|6W+pVQh?~rcnBS*lfZXaVj
zXEvmi^`V}VL(F=_F|`Guh@EG-X*^hV%m@NI^gntFjYG|GOq`38ouK)CgZ)gLQ2UfmJs1*Qz+G0lVaDKDj{SZP(e&RBqV$mzKNSXV1v848PgOA3?|C
zkKJyX-v9gz#bO>9MyThk)($Ff$uZSNd4n>W>IW1-w?Mx2=OIPYqkm|8y&>{vKc+v?
zm!%-059%^YZ*Xaq$_~adWhfWfMdA~Zh}>S@=`PdFmyeGFOD;m<9Y?0u?zIn~`*
z6XXI>#vX+(W?bsGNYeBt6NDm{9+U15r+d_Yt;UQe$0`s(R>fpMHw3`Q$@E9t82KX(
z52VMytRh~S%>RhX3FMM6rrxyKj9h`V71C94`T%z>X0Uy7euq{wGAHbM%zux|6p&@j
z^uuz9!4i1}2atY3e%uyD_I%8Y7nce%FY%N7UwfR-X-o5^aQHxv$CANIHyAx@{2T<*!~3RCl>07EF?RBNFm@l@2ju^(
zWY)=T7L&i>EH(gsa5dw9J~}gg=Z8=VO|NOPqxqzAtPBL(z
z06g|nAq*^xONk*(|1kFq)$^M;PWQjTfesjOh)Zep)IT~L@}fMv@x@*k5a3T!mctB#
z|85n87i91cgL7%SBhk14q2JH`(fV(m{|qC`CNu8Ng_QyE_2eO{=fi$Y{piwfmuS5+
zm*E0{`+V)Q9f*da_o-}nstG0(?prv&QhEQhNGFunzI8Ffb~9T_!9l{k;Y&Z-|1$5B
z4k)K_R?_=&uW5%Vzn7hu
zix}V;tgGNC;ody6HxA{+!TQ=5*Fq!?tD{~JphE}-7vkSx8gv~KXIvLvNyBz}{Le=q
zfBUk=;jIneBE7_ict1a&WosY}cs_N5t|1`L(G@jV&-U#-wHqP67oU!%_ryki1MbKJ
zBsvRc;r;N9?%Cs5Uq-7^LqDc#+)h^lToI@Mt^m#*X?pwc{<of5e%u>im4QfnNtgE+FDUw(l-
ze6Lm-bm`B0Hj#|Ewy6~I4|3i=5w;QTCHPHjIZa@Gt7FcT|6>tForB-S6@ka4#IHW=
z3FWt4n)^T!_C-%1k90MQoYD9xZwz*p2N>@eeh=|PpYHD}h
zvB*M|0MWW`;4M8joz|XWkwOpsd9lY@h>NVr1G_z~kj(+-}!vV0@-8mASq^9QeJCPo)jmwZxBbEH)_T3TA{b}&eREpwY=`%zcadl2Y=-&18(Vik
zAI9lTgQaQ>%+tW`2V4Exi3lG@`?(wKq^=<9zlH~JKd-lXbr-Cc7$M$JZQP0rL7;#jDrD
zI(wyQfB9K17_VSYYp%`>P~WTZm;8M!G7;z`Gx~-_
zx+8pb@}d6{>eGMzT?g}igKNjm1Qz*vf1*<3FpE5{d9f=(heaxkDg^sKv&gv;b<@LU
z5SQh=N^R)HRY(62W8fzS|`x5$ULPvCvP>~#LKL4cVfNvY^}mX*F*
z%?s;N#r*wlCFtLYQMEfwP;SRE5f|Kt0ek@)3II2|9H|?027d2WXucKny_Gu^Mxb6{
zj=xJzfSs=`1I{_Yy72oUz0wi%d-I)}bpeA9H2@KZ`DlM9hJI<96Pv!Xc05#AwFqS1
z+s1H#5B}%d-5VN!_dfZDAgXzXzzaRRhdXtr;QP#`s8bU|9B3{_M9HT@ZkH}rO`D0
zcYJoF{6XGQpYohe^^Id_cWL@FAVR?BMMP=6D!cDeKJZrK7Uc)FU*-AY0?`P2MoRYDE=~J>bvK{@a1>`B>LdrMxa?h{t&ok1gzn^
zL}h6iroHT`Nc6#9J=jQ5|7;LFPtSwHiC;#-eQ8q&97q5i)XQ-{RXpE8*Kt_a
z9N0MUUOW~@?fRi&Ow*elPxZv~i^D~9AL~9!gn@;)C;657hr)>H7?j;5^77CR{p3GM
z^;c@B&&KrDsxep~@D6lOClj%Mv9&MyA0+`6{(~k1~1M}?pq20S~F<<#XwlLdZfp#TxMwjBfm}a1c
z&k*E?UN_&F3ut1u_YU5-U*TT%6ytZW&M~nnNWZ!7#y`AQPp@+~-weFQVgHKd%b@+g
zjXU52Y&c(>%DwLjdZu9)Zty^S*V5g;))4rN;guH$(N5B0Q}7%3%kC(s?E_tsK4+DOG^1h{ptR?vMt;+o`;hsD;ZkD(S@ZM1QX|v>j@1!5fAzCxrNCD!MT3`by2~PiOWG9$rN2C|^2V8!CSEOOFEEo9zqxW6;}uRwbdjA!EXqskd|&>#7`ftppses#q@KI1l`
z?;=tv$J0r8*B3dpaXck=ieIO$D$FO3%}yqJ8@7_ehn)-#Ebb)sy%Sm{(>mb$ltPUk
zZI6gy%CDDVB#-b+5m!`Q{DcVB24>w4W0CZSuA67?2i~t)Ke7S%rlG=F0|ibpubURC(!qnG?=$
zF5#KSZ6J>N1qF1$I^S~IeGu1Adc8&2Jy>6dLq%#%skagSiRk2^m$ZJ$8{v55
zTpIt*Sx?OUwFOmnH;y#Z8iHN@^ZZ+uRG9xeBch|6oz%4?9%sf2Ixnpy=nRm`zp1dLM9gS|4m4AvjL~;
zocJyVcAmbKIbRU^SuW-m@e;cqOyMg8<8o+~f3(5rAVEqXO{rUs?f7S$@
zx(Mf@HtjzOQg9C1dt%nG7@UK)v*tMueu8~qT2!`wj75HK;u$*$`_-RaaT(9U!}il-
zSHL-Em#EF#4zPRF;Cn0jnZGwStT6->7(7hhL0IpE^3m_DJiBZ`G2leZFex}!yG+(gW#jG`kYy|$eTkESak)Eq_!o>@{H55WQqV7r
z(prM|5FVdDEkVCJ>@(qt=L6B4ww36|*Cd(gq91oXDEu7b!&ep4a+FJaELP+BC$rNz
z3eOQCQIZuH&!j5W(D&F3krhA;!2Y3CKk!{h-gQ?Nz6%wUj<{ewdK*+VvHs*E&+@Rm
z;%u7-jet*DUhtU%-gxHsMg}lrPgxi8{@lRkgIIsx6z#>w0KMd|eD?x8KW&;Ij#ueF
zk5O!o)gfYue!FCWNi)iEGJQ0T3$H!~gahvJF56Q+IXdb=@6&QVh=pMKlC3f{4$H?I
zFF|>XV;AKuveWTTap4?ttyLTC6bt@NdAsE%Qx8m+=KnDr(b)7JTp)s9ZuL(@J=fBW
zDi|+^)V$&_{Z|4P9S_kggAdsM?%>DY=(|)*KL^T-f9ZWgY!2l+kNPT}t@S8d$1kTm
zJJlARr9qagt)YDAhU;6(YijC%7C`!HDdz8pTI_aHy>qsDH2qz!3q`N>5Z*!W^^^$<
zMAwL)bX<7W)l&Xq=k81Sk6g4F)eo8RcPi@5^9iKqnG_3#k7bu9(E8h+;g>Ps_qW^C
zpq=wKIa1yrm6RAB$&(9aK>E@L&>Y}_yDAiWyRE5xyd8|)
zrGbooV>D!g-kPAT6i2SW=KvsccpRc~vJ4!jK(@9%Mdh&OHY&4cwosnPW6jjFg*T3-
zJASpL{oE;2L+#C*o+N|iEq`sH`7<`b^n$$e^Hi$$JAM-w2!7)L^F7C^{oyqK&?fj?
z0n)FZ9;f!$G@YdVyr47#F05fbkB-s)KipMJ^`?fsr+PY$rzqOrzDn0yfrAxYZwIFx
zg^t7i^*cvpu{d6OE_|WCel@Q9M=kgl%iyPG44ERz4%ty2SkN-N9%ZTb7dGR5nl{u5
z6A0@$`2$>_!+Wq17hRtl_|0cxJyyTH;XnfS&@JC-|2ZS?QC@o5^YCs=cd1aP_t;t6
zL%Yx~oju}ef$17OQu7dh^GvZsyq0y9mhV!%Rf&4KepMz|e@>&XKCTCIz8?lCuRa%c
z9rb%JZ~B1usCSM9R-&vceNG7XNgm(&96YCAXI(pnJaBjXd1rjbz0iN_H0Ekgd*ZXcU8dcA9g>PLRbzhxgwDzSsoUK1~tw0Q$TKHpPlTy`Q=
zkw{FF)5X-E131fVeLIqy*cBic|aaXIzId)!J&fa?EmNCos~2K+L_7V`+|*N<HVHQGE)
z`dvltOtHKecIFBBBQKE^ZowjBn%*BjzlC$uN3DJXdx%#9HcGh*3dxUwPbs0_9+3JU
z5``0$&1Av#2|f-uk7%!Rshe{j{C=0CbM4x6@?CJ^oj7kE{C!tMAisDWDVARMbAMDD
zsf$|F%wq@dObhIN6N55{LP2BBmhKE_6|jgJ>pl4}
z!Xm%hjxIbl&LRskoro~(pFC2vXg2Uk*V+P;H^3)z>Wwa0!v4&vTU{OX2iCLCE46E|
zKSe)nEZhS6Ccae;UzFi|$d@~z_6_{d&Y30xw&0i5`fe10p9rmwU6BTSSuH9#c^q=bjA+9+Nc=m+{Q`LXl+;K^
zWvJ&xl+0d7=-2s#ve6ni-+bQ|rKk>iF4Yr(qEOGLYAL-HpwAJv)vXz1&vy=feo$}L
z#iOEo05$%aiz)yfvfcI)|spEw=W`MV5bJhI;{l(ATJ;M00>IpfG_Y%2T$~<@u__ONI5qUVrhy|Iogo3QG
zFsSh*=<{UnR>b%rkzBMB&ja35S`DQ@FGpn6DjBf%@zlPm7SKy>E9%+=`R|r1HEsud
z&J*f^?T_7TUx4xVWNFeL#GENN1F;{28_YXUPH;H=3*&9Q!zeofX32oJpHy{NKB
zxpnnj`o2&Vt<#0yWu8r#_Yv)vzpY+}@jSkC349t3^Et{HxwR?U^}nB*3<7l
zQ}(^0{0Sb1D9=3oy@<-weV@aHCA@Q5pQ3gy7cHf7reY@Le=k~?cYozFMsIm}HRZ8s
zqoY1puKV%?y`LTVRYB|X-Zqo+_Ov!vZ?v<&NMHx*wX8IzW_q~@ujFSFKdAd-6nKRb6Job|ECJeCC@c0L0;K-<=`;R@401P;}Jjh
z{yu~ICAf@Z{d~Zs&kma-pENG8=fn16Z%y36^V>>^Z&Yt_!Ad0|kX5b*d_w=!x$>y8
zBH+<}C3+rqS=a9iUab;uAGDQyg0QhZZ-I|*I&ovr9kdlr5c_G
z$i8t=(G2YFX^G4?g8ZPLTi(0^KYDhBBKijuZ4`ZBy?K!96E`=DVFgfxmUu=57##`WGC}dmaJ#
zKaZv5KSaOsF7_<`|_=e|IH3VO3=s(}CIQO&C8gmR(D1KK!#JTi}6
z41cjm*ad@4YoXpzua}1!2AD=%(j;wULJWZ@uFXU)pw`lh=fI_}ba}b+aIDowl;@
z;>%-^@)!38XKrmHCe==Y58*!SKRNYHAE6({R_BR_4fxTe$_D4Qv52aMuK96j|D@TB
z_leysVp{P=EdhA2(3A=nyE$;*%xV65z=uUvMLDNzHR~iIJAN0rzi%Tmjbp2dGY|b`#K>7_fk9dkbCrLw#vQ}HP
zk?yq#Nwr$}q_k7)$9$z!a&~XVjtjX-L@|HiNI^md{Qt`Gea2Fm+9!l0~-V957gR4gMbF>UIrnSO>{Pv2u}o5a*suS-(gD
z&TBH?UuoTJBW^dZAJ_})spx##uG*8Z&)WK|l0i@1{ei|AG4K<~?IQEFn#tvW=6MQb*Wo&H-#R`$u=_Zew{-=}{2=izbiH=c*R
z#rMK@SPdU%IRfwd=5gF+VPOrCl>2KWalDaeZGJbH#RcENp0U}OR|fHd
zoL~v?JGmyyhwovZzDe^Bs)YT+E2!yV4gBTHqNt)Nu&=7_=so5IJ15?k|M?3%&*+n9
zqcQC7LGjd7KG+Y@?!l=$VLwj$oV6f+h((IMYetqrIp_Jw{$F%q95U+I*3gcou|yoc
zrzoVHToKs7B0=|`bVR~>50T;sm4p3Y@#(<$A~;_oZP*S7%9T@yba{qEr-KC*g%
z4aeKQVEa)8S#d`I<4R72djp!(n7d&68=k*cz_{9&xM3FRuiT-|hxNus)yblUe{-e?NS02+3S`Hf^ntp8h
zis)J)NO{z@y}Q<-Y`n3U_AgZUO)AQP+TU3?-U1;rj8VSUwM7qcRJdt2Vxf}O8^nrf
ze<*Lropsp`Wle{l^zXiww#CzN{^FLn82jnC@BL=v1tl6*lwUP8pP=JmYQYnX-nF&=@Sg0yU8h~}ev7Aml`GDJr>q7%xWhe`nUFK`9KU7t{P>;C;~P3be?iaA
zGX5QwlMz+mz;}<(xh(;hzbiky!X9vof@r=x>_f2&CULl4b`ZngxE{?{=6T>bv1irI
zD%>~W{l|)~!~W|%O7hW8w;>+~oBAKFbL}6sPyRuBqKS9=ZUc6FiMxz^tHn`5
z$sbTDFEFkX(&tH@7SRP+OC@mcR^SiKC5rvHA38G%f5ieOCp60+1-c~e>uUvvDuQ&MbKV?hXRiR#Y
zPZme^5%5#%2adP{A4;8bBvb_Qxi8O{;RfXvefy|YwGH^B*ZJK}(9YWi4*h4)pXHvn
zv|oUKbxYumYJ&J6()D930{pq%zQFk9t61bmbEBzmKfL4Z;pag9ve8u5b;}I!!@<|{
z@84jNnU@cqxnt5v_D9cJ?E~lJNdDMF|9x=2w+{HSq?bkhT+JT7bqway6Zp~FZj!7i
zuVaFdyZuuZ#1MgwjpKH7^?Q9PR%oSO2ZZ{dAn+I+BtrBh25a)tY?$;%>jza=%y
z(sIZFA-^f_{4$8aT{$JLiXxH|q!X>Q{UPa)dG@YFHk7Ma~_1ZnSYzHsN;Y>U$`hORntJ-*;wj4(XpQ==O3fo6K<1o@J0x1LviQ
z@%LvNNxt6cuie}bPY#)DN)^F6UugWh9(YkgkJ|Yvy@zC4vz$kALLJ#xWj>bsu8mm4
zcUmm#gm<`0(QcBkF1IcC6Jz0$Ps&}J=YCvPNyIZl_6VGBBH}i8f;D0N2cH>xzwAEj
zC+Wz_`2xA5GkNR%3-b5KX}!7rGn5_^!}(hS%G}zBm7egN1UNrTO|VW#*UTdl)ly0R
z3fW}7xO{
z;1^D(wzp@A!gpI7;U%---fpLrTbwoA!}+N$dYljh-+kPimG{LK-rMrEM5Tq`UVr25
zrDx#XSJApG>oVNKt=OuQJPiI}$ZmP`EvRoR`910c<2O6v!HWgp@A7?I*FA#%EXZ@c
zng+ZuRn_0a6!iQTxVU@=z4hD7pOgRw6-p`p0)JvQ(zX)6n;B~`e;Wq%T5yGS>4AR3
zwJZ(-^?mNR>sAZ*aE~lCZvO!-9HcJFfHjKzeOTh4ntbZl4*}7*7l~FW|#?IzRui
z9s2Y4x19bygZdw9Xr9D(@S2#RH5j)BmD5|X-o@)1EHPdXyO$LyKY-VI2CMzxUjAuybhLr#4rR$Ip%?eFFW@{YH|GAh)H4z%R`L?-Z&V
z>Ic2xp`}wkfPD4K&*wTe^pHLnS54zy@S8bs0aI-`jxG{a*`0a5VhgnHS=zRpc>X;3V
zNBy+EJE;GIaLquyH3CN#Vt?d?%9dc9yVSMH7~^G4YCf$`k(*;2c~@E_>$VBtABlVs
z)cY!|Dv0IpOz+l0{X1`0o1k9%@LS3QmzeFN?}&2_moPEg$hBLD9gU$oQDMavE4=gWF1&zg|@ERC{BdV~=6r?qDt4~|!`
z{*JwvzCK?}265@PCVCInv%!l8_2QH-%c6bPN)8^t{ClVL7b3orhG7A;@%-@@QC=K>
z5DR$57l9;%a4#l*xd!zL?_A!D
zdbegp_ss*`@@lwC1n}3(8B_6osNv}i_=6(gD`nc9xc*k{sVNo)lssZ;hH`{xnm4XT
z$18W7&>oGXxNlm3*D@E}LA}5eVke50qO}&`)rANgRgJd
zy8*uQ9PX0_oN_ze1Nn==da;$q!R`;M$!X*(&DLq+e6SCd9F@ZGT-CNA$vp;esz>-r
zd(c}sM^5J%tlxpdJ5^0^zt(w8i-7#f`$E}x{~UPY*Rl1WC-%6xdjNQhp_849FZdyK
z-y*df$T!qW&-TOm?mGB6g5Jxqq7ZfPle2s66RUuaaQ1(Ws)v03MfIQ1Ue)$SNh^51iv%e*KM~%Oaa%
zeTx6&6cK|tyR)s|W{~7q{kT=iaIa)~K>c81JF)8buIW0|N_M|HFnnccKG9S!7IaB1
zA~U5Hr1bA#k%E!~zmngzlb3GI<+G$)$PoWsm8ASU_CYJpDdV9rm9LmP)@EO##;$b2-tsE=
zo$2)P%H^#jSMO#SE15+U_D@eA{gDU%-%e!x#@#t2$eDOAoc0L*pJLw2p}=Os@$+}T
z=R_O1{A}pInQQWiP5ejpKh8YzZRL9DkCykzQ~%okOxHA!>&q<8xTUm`t-)tZ;2SI=
zW!h@`E53lt9}BLC7^ol%>clTM%y~jw4_NL!q~1b8wv3JXv=kDJIs4+Ic#DYc8L~d_
zK?NC6`*MDcZ$06xoAUkkk!B)y@W;U=;VcqhxN%hb6!8ABt0~@gz|X_}OW>Oc{y*3K
z&RvL?a}00BNU_?;)T1xBWFZbGXoZR`g7;^)wcA(uOow&k9FL3e_f(HJY2T8A^Wf1d*~p{R
z^t&g%LcCln9TMja`{F^^j8FWqzQ5lZUdRFIJ4N)Hp3f5ZiRk$S_K1sp9eoe=?SFXExEsnZ(9^g^{S!;O
z68&H0!}}NWL2o~M!LoH=?>-jXYJ#51)}#oGQ%gHeUfcz^zPQF1c}=y)JHsMCjr*D{
z_&uCeHd%pjZ;nBw1Af;_ctsQO962GG<@5yfE{)qBLI19xx6&Q`H%mO|%o)Ju3(K$c
zLi#De;K3HqtKV!KBnau*nj3=!05v~5e+mJNU$Wz!5unw90D5k2@`lgLKwkMMSrX&r
z_I1xqF&?eh`+2VmV4-eAi#(uLq5}QBFCd2313}=dRslk3eN)eMUQ4-9C`K7
z-x=e?ff>0e7;kQvyzoIif1hHLtAMqq+xKe#I#gUcg8e-6(!maSL)j-=J&aREPUpRD
z0Y2hYGQqf`+q^ge-z94E<|LyYcfWfY>Y1JJr}2`vVFQhq<$}v;yzHzyNaM7?lhZU_
zemfg~2>VyPz(W-A@J~hB-y^mi^!}q@u6z#KZPbc?Net)RT~lTuM&@p&_aVm@m3~B7
zYjxl~?3e2qO;416*@{Xc-roG3-n(#}mj4gstqS337r!+D@bKve
z16ZGG_Z&scKlh^O5$eC;JKiG#_&&1mmos2?$oK2mpS|;a>38=w3EZ(bFPr2I|IPy}
z8c)B2as1i!_(e|ucf6i{745AN5b@dt>v7lLsViu_-0!Hc0peeX*u%k>kUvYU*3$;?
z`CfGi()(`k&>`
zIDnq&$-g)JfFBHK`)=9-_YDuWE2*SFyxjRgYQ7QpqnHY!wA}`HJNDL;7(d}6a25(W?2>s?pvQU$fw@0hTl!HvwV8Hl|>Q*Y&V|aVG;j=$1|1zFZfyS
za%~R03mJX7v}z^P=N6W6tKl(=T*}h_9mmfibu#NWE{6Itd=3b306!3&ICQ^m6Yw8{
z`A2TQgzt}JE!wlX+sKPc>@u9Uq*mTVXQAB(%3Hjjn#28Z$w-_s__5w{VV!7rmoZJU
zKRn|we0NlozKZ8}J2@HATj%?_nG78{tRK#wPdcX#Ia!{sAWx%<5~rSIkzD_`k19ku
z$d8{R2g6gF$o8HoRyTg!Bi$-HL(UgHAbb%V2K?~7yJY
z=Y?DzNxRdqdA2xwr=qnvZGymeO^L_CO~=|vBIn#Fu9{}zb#J`SZ9JdM5?yMq=Uzn?
zhg~_@ItuSX?!SU8s^RZF%`Lm4Ra=Srp9{W1uL_C8QK>@p|EdV<$iOwVnnH4+>UbCz
z#LEbch2<)>73AP{^S|pLUfzz{XT=ZklJ}$ki1?=>GGS}|%4lXWc|6ZUyMJ*73F8ue
zx_@RpNxEgRL~uhhDG&cNB}TEF1m2%IvieCWxpT72{|r|Jsn=5a-0MB(|jw5)#xg_e2b@~rgzh_h?A}4n#8HyBZZGhr1aY2
z&3zSQ-cyGXr;G>0?Eh%G?m(`eHy#-+*_4v_iVzZ&R4${etd^Ccj7X)jvqxLQCK-{D
z?7jEivUf@%EqgS6&)uh=e_o!u=Y8+z?tS+>ue@{6)7XTqg8
zUi#|ibw7i2rT(lnA@Ck)qAK4_7sk^UOQo&Fpx33cx%V3A?T@+_C=Ysn_d0HP2YLqQ
z8(-t!(JG%kFw_oupL|biYl9wj9E~hFr&a#;&X9Gm-^En3wg$#?`kBEgeD2cl`Yf6p
z&~qGDVJ(62T>8m1trhB3A8<}27|Qw8AuP`h`WnF{Tba{8;^ny8$kPpw{-1l&6&x?^{G+$r2Yf>RGXt-y799JVg7?=X2)**e{*h+3vz7{Q
zfqzdv*0W|Fu(SoeAj1c4J)mci`XvqPUF52@#_PL-`s=c>ze>qV=4XL?E8)3}SU?Hq
z6X)=Hh0d=03j7|#CENBF_TwR9y5?Nq2lXHJcw7a&F;S&x%nzu=^x=g7ypM3@t+xW~
z3GwvD@$$Ecy9)Npd%mkb5(b=^Ela?9hwsP5;dq(Y&6*Pp$nxsK9h|=Bg3Ko&jJp!y
zmvHc0_8ZgUcr@KtX#5&bg?c|5p7*a+?{>#}vc5|!Sbo#S!-DlTd@#6=<9>VpusW7E
zIt}(?zqGG1)d|P_m@~q%_`YnDKb5eZ*?uZcIRBx042h4v4)v_q-kQmw7nmO~Zm@w3
z(Xeh_AWHUah|c$u`LsV<*#u6Ps%`m+%QG=eBk^16Stfap;W5}u=J7kNHq~Oi?yDUjd?6(l)_ytEQLjON+w;%s*M4@@@4_xj6
z$I@50Je_!TvQ8mF<>P_td5*C&3iGJnw4?nv|EkEB%UIv*kQ{mc_4`>ZdA}uch3^(#
zZ`#7h(Sh|mD(@0rB4JBoj`d@9hQkjHq2FiSvBvogCtfPp16qE`|BRpWtN8Xy#$eCN
zrZqC25HF>pwtQoOcsWhoR7wN++8VuXv4GzF-7oAwZ(b~(3D4u>KMYw&0q^Ii*>M}M
z+r4viC*QaEoyoU?*WLO)i`)2vzBq$#0Y1ODeAA5haY*m(|2~R&XD0O$d>RVM@BFQ%
z0jPPeeOo5@iA3s>+G8LaO{R=t-qO)%?IHl^D<_=w2(a@9U3vgu??(2!J3v3hR^Y)Q
zh^uB9Gj&%q;CV5Og}K0d`uo;hmlE^ak->KRS8cG45SVvY`90|MNp#QWD8hRQE4f6x
zKO~#SP58<>XzR(3vv6Q1Kp|ulzaJ)%afLSU>~=GG8g#0
zUF{4vAA;UR`X7Q(P|mSdrW-Bb$7jEL$M>IxeS5U^ksM%O#71k=aLB(}@q`ESANZk?
z?*`xv`)V@0&^GX=;RSC8wsfEzUg}7RqE2+8F5h?n;w9C=DhVBj9CSJ5dak5e1$tcJ
zYcn{|j`$OYKJJ+XzcVuY#96fi)eR0FS&}J2R}RjpelaRUT_LS6`C#he#JvL@
z+RXhn^h77x8$88V7TJV?mapq@J%5d8c51gqBmrL(W9rNR|9@mq>mfBSyobFW-6}W_
ze&)bU_an<~=#;w8w+Pt38ln~c;kypRkNZcRW?}!cb3mAmNjSWBOR}uzebS9|=bxJv
zZYV&3IgxHgKk`xKOz`y!`%2Jfd%JN3u2RIebT=X6Y&p_f+iE)m^ZDSlLdS!1i;=4!
zlgod!WoWSr|rp#R8I>WdzC6?_Lw
zs9jH^0wqf+Ue#bLN0ujR`Z{3X?k6b
zKEGu3jh!k(1%t-RpDP=Y08^vle@^vCCfP^Bq^Jh@vn0~muB$@rp6Xh6VPCPO{WGsp
zsTOowNGHppqzOsg-76!ur5+VXPdb(BRih~bzY9~%RmeNuQ%x?s1ts%L{*E4PMuXd$
z&#oV?M&or}JU3{y3k0!;7yD-H6mTx(>gyj6d(6f2_70F|$=Ph(f-L
zz}t;}4zPX^l$M~++KzO04N{8%PtenIO)htb_k(L0EKQlfUvd43;oAy+wbe!-|4Tbk
zsxy3g$O!z2tcLE_E3gmmt(r;HMp!S7KY6hf_>=Yz;d-3zdvZnkYyiZa^ULG;FdhY8
zcT2B@dBN`*fuFkI*Y3|4$#K_#Uv%?3SgQj2#8T;Ej9|R(xpl&^1^nZ^OQQ;{ujnNZ*J%|YjrVEkL$IRE%02aMZ|(KnTVzgV#S
z=A;MyqWe&U)e`v2?u&OMwt6|_D9@0N9I9KtEC~d6yzFb;h-+yTeo*Nd@clgezhJ|or3%`>elzMKb{-ny%Pp{
zx`_g5t5EKhf(vDr0cE#~=6{0p@pi9gIPSk?wHdw&_)K-t>@)bg$_&NlRG{a0>HRdA#|7uAb|?Z)He{FLIB9$3cR!BL$}w&8&p^-brTDhPAgeqydm;b)4|^FJ6m~{ScpX
zo#e5eSxzP4Gg6{JA3%04J0*cJfHjuzs546z2+ObaDEoIms3s8o@;zIK9`~y#vJW=x
z=%^Xi`@->!@So{ics&Ag)L)Hd@A1u`gzj7UZ)Pw7HJ*}Ut`y6=Za}Qs~Dw@)S+qam5
zj_g0`@0+v2>Gb96=diwCei7Mk^Yf#uCzb~vT)2q)gZ>_OvcHB#JdEI-?P}ybO(Lqd8kgJnedjRdaobqw58`zs-%TrfvHpCWv+qX0
z?b$s&*q(Od06L7`hx4~%yuxnbiS^l31Uzs#&s{c<@1CStCwAlfyi=cc;Pln1jnr7L
zTKmpReE+jEH8=4*FtjyTl?KLB`*vxO6Tpuq+1TtB!H>R@;d`UW3_Q)%Wb=8jb8aC!
zcn9dwY}ozV4&*c5tE{*kq&;JQje*~D)g4h*0(o!sSg0jnPWIQ8O3)wd{q^t>=*#4d
zCx?Tdnyw7F#SZzlB8KEU5f6sHeZb@IoAT~wFG0WhWLCsxKtKPKU)at&icbx=0E6|@
zE6)L{WN%H!ymdzz^?oU^zwB)43vBnHKT3}sEgV)@lmv^7_HSC9u=9vBW$G^t4T-M~*@Lc6JzuZiV;vVf@{@ST$iiHFT3n$@wnC
zTsmZSTCx$1%E5ztZAVcI?`yX{
zt44HGbTv|nHE3Vcppvw46B_90x^E-jfvjncdH=Vr6Dbxy5#)s5|8L!BZsRsn1Lp%
zx<1K&9`;#wZTUqrND|f|u;I%&+m2C$)Q2Fph>yzv1P>-kQ`cvxl$j~zEobf^v
z?AH(yj%jH_KD|Lw8eQ$^cJL+7wQK5;@z0Om<@}AvID0?Kx%2S*{8G1oi!|*h?A?Dx
zr=PT=peNt%Y0KB6a#$1@p>9I;mHvZkKexi~%ey-rCEJlgLmfy%O1nG#Ni%A4
z6-W;?Z9=zBmnz?HZ9$V^SFLk*b|C(InoUga;CD`so6k)IbfEa}Dd3cF55zIEwDw|4HDGbo4sO2zPrmH9SwJMq+E
zt(UDxe6C_&OKTU}niRNi*H9<=XkcLaiyq?o+77XqQ|;)8yZqbJ*V~Xe?M}Y)?A>Tf
zgDXeg_AX@o?Thsk@a_4M?7xyhaIVC#-4Sg#k0EM?S@G*k7dkrP=hqJVsc9mPZoje{
z^P1`p=_lIJ-HdB;{@!h;;SdV!*@-%SxcssJ9qSr`X
zY(rA+^nRPmT98}QuNaH|W+WyZy(guy9)6#DaEseW4V=5N5NEi%4mAfji=19;MV(tN
z)UDaxiprF8R@491qdoHUBA4jvQRj<8ipx`th}LzRO6dDG^ugTF?O}NfVtkfyd(plL
zEw%{E+ETZn^`GV|Pu^}vH{3VGZ7XdRfWb1B;vvwt4Gzpm&O<8I0$lzXgsnB;NquQNF|S*r7T!)2nu7
zz`ha9HSE|~RR;U*bDK*&flo@FS14$8g?$`yU(zcb8qw!^)Vn9I6>XBnb_w=7A6*e+fbp!N6YF*WcpT?>>+6x=2R7ub>hHS``y^Bi9tXgq?UEID4R8
ztp~lbCLw?A#$Sd3zz0W8oTvT_Fi38tQS=;-~a2xbA|j
z9eiRD>ivjc)C!&pCzIdj<8WQw_Ha7ExV=fG~HlZb(06?mU6Zp>5{kX9W4_M0yuqe|mYxD^L&cOD)%NY==Ypx;NWFW;5V@
zgyZj!%9`UiUR?P8cr(^NTY30aHI&~oA*MPDn7klZt^n!V9&I4sN$q8-@W$&o((BgJ
z_(1;cIx%}}cN}}wb-cbYDv@yruVdWPthK^%o~EbZ>v_PdZv;-90d!+oe-Pj2S(wT$
z%-_z=WIV^bbLCyt4gnT%<`8a<=SxO6z>&HLv626$i7eUsQ8v64E
zu>GT*EByFAis@|^@bwF6avTjGW+DMP>_kXFaiSzyZ{rE5*zM@9f;cRB%_zsYtFT(3Sz*5uG1Sc5(5Wbpf
zUQOiwI37}uQ7Kp^1U)PA!9n2J!!kr(dPvdVvX`33yFLJb-c?wA#F+94iUX{UK1|T$
zCq;h&R!4EV>qh{<31NyXL=AKr4!hRcDje#h~eiKB)XUEOImuIkcj)4p+s>b=5Nkd
z29~higi;vc+eOSiX83x>yy!gK-v^)GB@IiE7`2wr05$M(?2)RFqgvr*dpPI)Y8PmZ4lNIUN`
zctXwvdBAdPmm%mY9#m~%0+h7;P=MvBza>F;WoEg9
zxSgV(X*1#WvHo^n9qS*9U?%AsE;xPO3Ubpg*zOMaJin?9;|XDY(hf#;Wi6PuCuW#k
z#QfUc=e5iQ@VDF0SNUr&|28+DnqkN__t084iBioqMD(2I5cuXR$dNJCW+R61oWHBUmu}?@|EOM
z87u*(We@Dec82BK=Hm8w7nK4ZzJ~U@_n}l9^tZ12WaSES>QSl{{9f6*;u)MY5APQg
zQ~1>Y|JWlqhy~^mJ2o?nLp!s{q%XXhf%iSpYRyOC+=H%b5oU~F-;aXio*xRG$k)ND
zvl`+M)w!K#pFrHaYNdIuQy$ha7Hl6VL%dr&+?=@<@{Nl6E1kE)@6QYj?_PxU-|iZ!
z+G}th-ZjBu>mXio8EdS*g>{|OnH|(;;P>ldua|7&Ar7{!Ub$Wi?IvI3_%{^nU(9`5z`7v*Tcw$XQx0TF|%HcV*i$v|pCQR~Z&{7=Mp7Mz-F7
z`3Bd+2}yxg^zr?Mw~1F<(3H|a?ZNvU=ksUj{H6iCEyVr(PiXJP
zI6m!U7Fbu(ePmJ!&n@iYomW?2-gQ=U?`)0%jLU7uH)m+VxYT43)B~PH+c+m_2J@(a
zqdE0&p#L8$n=TauJ6#H2Qaw-ZM*LS_G*5PQqXssU`<|NJNOUJ>5_e%2>higFr1er4
z8i>pjj!f=C-vxd!KB(Q0oiS-XTPAL3qIfFn2XP6nruwtg5-a~<*~JizBR7^=Jq0e;qHvhA}D*kv+!g*pf7
zF*U2}X9)QlcU1ZE!}DYRB#=o9a#we4>M-!h!wVHAW6%x)m7DDZUw5Gy{STk!Cp%G0
zd|RD5Eu5D#9vaJep&iK>#*}@aZbzegHqqaz??f)*BTGl%xuxIyI85W$fi4x6R~NwV
zA227Qk-83~{>%4U4U9j{8?jH#fTv}0Ch*(_
zdzPB$;x6z(|4F
zPx{$}oCklw_%Y0!4(h{n|3BIAZ7>hgJmS=~p&hjb-%e6F)q&*Yqm802fWLfb=~Qk5
z?{(IHmM?|%U6_R2z%Dwgbi=RE
zZ;F{|UkbxIsmMy}w{O7P?d-SY4Z?j7J@vRmtpJ{afGt8bN$!u|NZzu%J$^%+{b?Z@w@;OFI2
z<<3ICTkWhkei`ftpL!+w9`uS&MFn;PGM(z(6ax3-S#MZ`&#A8XE%GrD?k9a|{IfRP
zPbIT~qZP`(JrcL!3)s_=>S&4MCrj5wt?zK%)=}O!^q|k(E8~gZa|c~GVIc?QX6`m=
zGVgqv;l0XRAl+{sJ(8UF%jN_ddsdY4$tdzf}nyO*g;=ytbpGSUSPQug4dzl?J9M?gUoS+yF?WH^nb+)FE@f*
zV>pwhaJXZGXwd$y{VgzNh{ZdU;K=^H{!jPsbnIt(LlVH!hEA
z@k}kwHyLm<$N9zv(ux?RrlXT_dfS5v^1fKq^85*$P8HKk`1SaHDSq01i}M-GPm}sH?j0uAE4*bVymT(`7r8EvMT`^M
z6|k%7IMx>kbSM12&h<6n|E?3Agr|kRFA*f=X?!H*=s7$f*UuYLc)yzKFtO9F)Qj-F
zV3h8o@Z*>wb|S9{27vTwdJ66~
zr(kP11Mx
z^a^MljPZE70GJ3Y8DpC*1+%g#$Rl-;q_59bCFt{Inba@c?Eq=FzIU-?T|QwErhjmK
zrQk7w<>bRT(BIYnBQleHDA5-UrnE}~KGp;DIfSGM`ro3oi*sQnDL+SvqW33NdV^
zyUww77*oY8kK^<;-CHhWKM*gdW`NTpMugcgavi!#(rwn?7y0Mo8Uph
zJYTe10b_>j0-5hmd#9LTnL*e0CAP+pJJ(4Zs#iU!}F*G%S|7#U#`vhvWU~)d>JGA
zmb_Lpys*BvA`dtAo6?5+vbF$L{$8(-(^mv%;K9H-IZLVGIBuAYR6oM=ImulIzu@wU
z*dLAG1b$``mVX_e
z%`Cv>B!@F`z|%C29Q3*e{Nei5TaW7@J^xJcb!Nb+sk_2>|Es;>K8Jk}S3eEt{N4q!
z
zSJHaC7LY!WUNsBsqS<*ckOAfiCDPA#+yP!j`;{({3HJG^xQooTfn8FWeBJR6px@V*
zkKX~l7CpIa`al-OH~m4GKXAWa{nMTEFrPBZyDNVY@(qn&*@i$n$TRHll>l7BepodY
z+9yC?U2qoW3#OMho|J(0krR~nSOLCq`FXL664*QbBR_Eg^uKzT)3AYkuIDlywZgn*
z?!C2BC8W1jweOvVbajtpc?9~GFHO>1hxGYV#!1_t9*$p6ALNJkC1Fub8rJ%-Zz@2!
zuoL*?W)uIUufXRv1ky;-1CM4uc{hR`WXo+gZ;itJwO>~DCjmbd`F(iyCEPEk%a-dR
z*jfJR@etlOc%NOTnH%bL>DV)--Ecn*pIfEPV8_LsbAgw@j{g`mB7Z~uyf#If$N^G4
za+qGl`E_IETcLdpz3+`40Q<(XajrXn_!s`M&
z7TdN|!aOuGB3iK@_~dWLYOC+S6NRrGoLvEaIy-;h>OHVy>GIVE1a=CqYoVos`(|$+
z-m?YDV+lX~RUGQ2cU|mOE97SjbZdu0y2+=J)D4h6%k<}^9pE^R^#OfI&o!sAeA
z<@a(u*vE12M+)A@`CqRhKYqW*PxmfB3+&xknq!y_=pDSA`3m&ksP4Lmd81MG#q0}!
z*SGqss{qE1U0f{%T&X}PO9i5{aT;C*drMxim>w|iaJHfG?jAsYu?
z@q;Y1G{bff+TYJ^U5FgqhrclIohJ0ZCjL$XdKhQXPbFNnK<|g^#qBj<*YRAw%`c!G
zE?sgHu7&$~zKm2+gnF$pze@cGuBUmFV2XKTCeL%`D!8ut?+znYK;D9HQwxy))tl!B
zALJitozvjHU7Rk7N$`BOJBtYO_X
zFPkfgHTXGw8qA)K2mO!}I-Ws*{tIz(xE(LAALPOO_?E?;-aUZ97o-kjzp!L5{~-#n
z^4&z*X~3U};bCo1?wgwYY3v8~?WvN>1nhXyLW9?(yk;0yF@HV3PPeuP>VI#YqOS(%
z@7$+}CLuk{V`XD5F6XN(9bV@SQDw+vhkSX(Ge`07Uo{r9rtU!g!Rtp^F#rE{II)8j
z&^d{%)EiK#JXOmE(7J@C9Q&JPS!ehlFx+onf`BYwiNqyVMZozaHXrO?1*!B(aa^G?
z`)>*R(|GZ;In1MDK6HwFg!?gml9s{#DmKh5<1nDWbBA1JjP
zOs58UJ$suFer^^M-xBb=p-4Au7~4@2@7{*}NQ{qi2QG*C0pdFVct1*s*r%|e%e-)T8-%87aCr)I-;Uw68!gis
z#TaNN1%wRV@4U((Jf=vXfW+~F(c>f@FP&#Pjq{@;){uN_(T~K=Y~MM;%jJ|GyutZy
zi>2}yvkmcwAMtyj--j?(>QqQ#%#GYYjqU1AYB+=C`?G(5fr9_gQzX27?*a?aOa1ti
z@PVsUGrMqp!2-~IK<;xV2+nn#2I2#L!-AFMPZ+b1^!U^41Z`!{6CRZy&`!z|sq!Q|
zPK%R<gmXY#*c4)5Pu+k+~jkBzEm=qsX_v
z@{#n8b}$>RAHv7E07|YMB6h#r$3^5tZfGV*mr0}WTjAJ9y|C`srhfShjl0+UE)Wyer5_SKxI8;DHyBaAWYtsrbRcx(zF(-JaJ{`ddOJZU|VHxORh}!z$(d
zO>7B8FI%2sw_r7H0N`QYjuVvdj3WL)cn*FJ4(T86))K!^dqQan`RUeb{fYivMZE@#tp{{Z%~Q=zl(AaSD(z{ixMb^1SsN*r8>e{pq96zj|zh;lY&->a(DOs$~t(A#z
zI)|-j`hS3>$E%}pI_K$PV*lOuI_+4W|5On99jQu(>Td)x-8w-%YT$87Zre8XfGprR
ztMLwaiW=X3zDeNUZBMn&CIgNfh?4Jzxc}E>wz?5;ao~w^DCl2`?7i>=@M1z)z&O;`
z|M^vqs~|7hcroJg_svVDF#_6${l16mOaGwDAQo^(_h8{>KrhE%hp?T@;X^xd{hSU=
zc(VYWEu1-WRHs|hS%pr0$Y)0>-^>9of9veVJZy*j;CgGwKf=8z
zbPq5~wEL7dVA$AVr#|2nMY^Au*YqA!nb-zc{2)Re3piv$LERxBG0s(kSMmOhQLjBXPdX^_6RY4>a~g
zs{q;tr*Po=$;RfzYyo_>;P&Bq
zoxr&OxM;dDQ5G<`{^gft=qJL-)Rsq}-L)#E3k@J$Hdww^M>tX&hz;G-dziBxL<9o=mUoc^PE5Tui^3XFe4%c_dn%P
z3o~3Fe^-0G{{X4xu57{ms9HhQ6W7PeRgRqZ(h}`ojq|y~;U_476EaakIN$8U;|+K|
z_9S$`8p~^ne(UrCO4cqYuLpelSdVTB?Bri9>edDM>tQb)XTY8f<9Bd>C>O6(!t*=7
zgZKII_|AK=h8?d1T>f_9fFNM#vh_Rs9PWGM@!~dz!08Wk&%#11TtBCL9gnY6b34g)
zuu~-~T?*Gf^l)}6ejf2E+^_Zko+{9!#`U(keJTOB*JFzmb^JV6ir;`khxL;uR-(8+
zRAfa3;qoMwo*ltH{@q%F^}8j9nJ~5+eQ(9jgOw|{4(IE#RcGS#
zz0+$++{(#oA$(Il!F2-r^Ukni&vE%R8JvHy{)TK05_d;7g91Wk|e?
zq=N0$kl!8jjKrnKrB)=Kdfx9NarVO~2PyBiaR(3_c#fjBBtL0Bo8&tyhLLhCdt-=f
zUdBu24_ue@ha5PvL*brW8Gnf92d%r|=n5DWDgiZ;p=>2ehH1loM_Z
z%?|RD5z2MXg`xRC7CTSLAIHrHGRGzg24R{6@*gh>hV7zUf1jBmx8lnn-`9|Wh6gBi
zO8=WrP;H>3_W(P>dO82fJ%dG%N%4T8|Ku9nEFio8E7$E`I=LOBU;8KbVDmu^#O4D6
z=_Cl9k|2;hf@c4a7Ml%vxb-m-0K`uV%iI59%fE8}uebkSI=LM5{*T-KrHg3cf5312
zv$qM;VUT5gDVX}N9TNWK7vkoD{AN)K;$NF$8RbxL?%)04$Dt&>^q<{uNH>uUa5n_~
zXBYte|DoAG{m_5q-^bku@^k;C-~Olf`JW7f3;uv7|ID8Dudgihi;vrC%1{
zVFBq?|JqGMl_E<-QOdvlm{M-)zjB7ayI}p&f9>`ciy$)t%f#rv?F+#@KRF269A_r@
z`QYJ?vE#0iKt7^w=_
z2+EmI-bZPfQtG3jNvYp4JdFXp8kOS&;}R|tv`3V3RG;IX1@YAdcX;5Z^c30c7-igY
z&NM+6hWXMTL*lpMH=b3*^4{MC@Caf3=Z>T~#ygo|a6?%44jzDlKtIm`o+$8no&}Pw
z%!jD3oVuNX+(-QVCxUMyMxbHgJw!-9spqkncYxvm?uWB5dGH75mJ5`
zTLQ7m*2a>gTNtqt%n+p5?XI$p+*iby(r+$Y;#I_P(o8BEhBDaAbXSMOwd4u*`=
z2x9%d9trZ^=EYUYI+DG(HmP4xsT+wq`rj{~#Coy9HH{eEPK1EthWX7Fe$sA_uC9P9
zfq1B!OVaxse?w0LzU3T!T@BD>&nh$Kp+fuXVFN16pYGLpodaw(`E&*IhuOgiZ#+M2
zOnRb-vGZpQff2KjbvjPox&9$;jgEnbhY@8>2y
zj4KW=Ph$Q2*z`DT&zOW-&lbQ#+eW6h1Gb(op~25X!fVA1^X4}zS__z0U86T(#OXF(
zqcYeY8PQL67QC
z^@o=U%hpD$0vHpv-@S_U&eEN)z{u!n&x5bO!}F{V%R`Z!5J}-VJyL^(HR%6;+-ESB
z{9)HcfTp)Up27X+R)XsG6M$NF?%%Ne^z0nf*sj@?J=(->Z94Khdeaa;FDy?QhJvey
z`)FPE$8t1n_iHTMw%&P(v0_*Sma1T0S}gB{@li#F1GYm!>cj}fUMW>@J@8&sOotcS
zTf{2ag5{LyU4;L*rKpg7s#PXN@Maz2%Q|mx?XW-Yo)M`}9Zkg!tltBSh60SFchtX^WUFeCH-Ds
zc#9v-pFPD(##Km%6uj66US%_^gwrF~7f3&|_74Kb2K%=%Pq!)p&YfoU#`A;p?TP$&
zp2RUZJ&o;6kFPh!_Du>n%42)i3xpoRQThnBvwfa%3YTZ5$Pj{`%jFk7
zgnt{QMyBBOC+x-$F<^b7Npc=vC+Fa`ie)KTF;|Q=r9p&0PfvJ|=TJ1BM8-w)B`;#<
z>IYs346v>;{*lzHvN2f;%gZmdusziJY2XN9+`9cbf%B!WP~E^d
zw5#tY#;LFKoEXD)rf(8{8ljnXnv=RacjLXGuB;Oqadpe7J@dj@1HLqKLBY=e@3z`zb19=~KV~ldRSpoe-DEW8r?Loe%LCN3l4$T4b(`btPV@R3E@8Ing
zPT!e31SJ|I7i9*PnvB|E+H`Dp2x^2~mgrdf0r>!|VlPGk#ovPjPhs*I)ol#?1hj
z@h=}X{Sg^I6`a2G5AXiFu2X=L9`vt#%YWsQ<_CTCfBGey4v|+v%Fxi^%6DUZwaDwr
zSkQ@gRp@!@j-_X_O(@@OzmCRb`0nxAwBYO4>(D`6gQoGr^+?!o>tvBpJyQ5q=5f=q
z3H|K8dz+^Le&6@Wb5=5?4ynXD#2!D~hJHG(zsP-{1GTL4)vrF&j^wv)7C+_OhF2_%zkF~AExYP{T|L!}qT-J=R?zjJ&52TR
zH#c-4CGkeb58k8bc(|+13R^$=`&Kr;cB~8iJ+)=cLK^(O@ZqgVde}$tV$0EHt(8us
za(3II%v_-97nR6PCDh&~627aa
z8qS#k`-@q9RXcB-Y(kPJr0iwQYS6ut>xBKnCF;Joel{>vSW~JDH9Mrrz7x;Wib+#Rpbe4-m;@A5E
zU&NQA@*@MKT)nx7{paDb7S6QN!nGzwGPaeai{}!}tj~K2AZ${nmJIJeP(xy#KJnI4=pc-4SeewM$0)A{Gyx^`sy@
z-^E96sVPXa{zz%T_Y`!UMPhpEqZIV`z!^!7eW}RyZeneVUK+Z_`+iB}K`Q#m^;y`+
zJq0CIAK7GAkb*MZb8FZQ-XilT1>;P!WEA>9X!hBbRP;DBAf{~&&Vi@DR-7f6j?#Yl
z46%%*p}R@)3zHIQXgxjqZ9C0W)Gw4UGR2XKKB`_xLx<9kc#m=j`kRUr?cS?q^QWRt
z`fK;=^HPw$pjp5fIG0|3dfy$Di4>G>AA7p;NGiI^CF6bicM5W5Kees(K+vp7%C{`K^S4c!rdViLp2jWmK
zo#&C;&tp)112eVVu2)E6wk6QnDFKz;FnD{{Bp%&c|MaEl!PiLjz3|lw^skV4{Fs`!
zOC*Y?xmSN5+Li92MZnG7X~-wdV%}OP9o1e~P&sfY6MfZOdh?ko8wIxJ-_R>bLmH=4
zvbvhn;CDOPr;BLP(L!cOsd;WXQsX_nef!%Cl)1kvW35#ZvdTzkvusa-?{Dn8=M(z|
zi90d$*yg@L+O!Kf7R^bBvF{QKv*IfxCX=ortQUnAWn0Bj_)C<)TOUy4^b+a(;b!S&
zdI8`6u9K{di9`m{=Xiy;zC`PPl~b#~4@4i{ZW7Yl?1?xs^aE#u|0|X5vyig-k%gpe
za(AAP$U$dP*VW#H?>OkbKKVw|rv&yLN~V3@QigV}Yi<&b%Rtu^KkVM`pNSq=Yx`t8
zhVOk2pFEQ1QiufauRWkrT!KXFrL~&gB_WafZr_?q-=aCYUF&
zRc@6&A5qDuay8SxK!05sp8j_?44HimdQx@w6*4_})3oGq659HAYnYT~8rqaxMvoGt
zQNhPu_a9Qrp?#cjea7z2=x9a0M`lql@|9IqdhhfKIiGB-_5y!c(Y?%(oL+|dNA|UT
z*jRxW*d=b=fpd%Y-iS<&ORGlqj1z+DktL}9vEU1z?PZ8B%eJK`tQFsB4YaO3uCnv_I74
zv^lhAUE-_G*USZoLWJZJWY
z5At=QiqiWp!d1J`WpkaIg%&+%Y$Y-zNTnT_yV*rPhwq+gpHn$1c&ZC+I1s?M+|q+m
zR(`GJoa;q1oKkGG@LdwehyUeWhka=kydu4J%-v|joVO&qq!;ZB+<(^jdmlP8mi*_M
za~JYlFYT_`(S^zmbNPMQ)Qtq@uN_5;UC3SZ;_a5UE;#p_*=8?yH!87|Q>t|8Mn;MI
zr%U^~(b#;vD@}7Z>SMj8%bC=TmS22lkv8u^>ji1r-hS;tUpkdvEZ^-#`pH)u``-1U
z;9s9TWj^(yvUY=(BQw3|-L;xh$Cy5}nRb7$T6#a6f1h@Bz-|E9MZ9>$(lCJD^!fb=
z1Ur2NX;0E*_oMUXcFq>UgXq9wa)Yqm5K7rmx}~{e2%W175Ok>NLJvnn7cfr>>^aIS
zJJ5|T{t`NIyr3JM8p{cKbOG$WQ0Y*V(2b}jjb$PZ_aMCoa>%rz2T2QEt(%JKK^&Xw
zxFvh9<8=1MjZ(Ag>dRax+KtPQ!aucFWQ1us$?W
zx6a&HzYmS4>1gk#7J-c(v4ni{QTAm&Y`GS9{*h*VGI3kCU_yChH)wN-i{pZLliKo#1JnfN>B?i%`)`5iB
zqXS5HXs&hOdk4B%cgWHTez(WGa*of%6VBP-tCfG-+>Z89|9A`EmPT{|L3=&Fz_1i)^6jO@TqkeEsZP1kA>-V5jQ`{SRyStF6M|{oo
z&Q5g6t=q26T(2blOx2UDY!g>BH
zpMtJ8x1rrvtu>kInvi`^-KBVmdZem*$3uO(4t>7nZ`vW!fs`h{cv=BJwmC6k&Gw-g
zF&x*@bQ`HhN-GT)-di=G2mEUO=}z!_-G=v7H3l7MyWY8P?_}GMh@kw-fVFwKQ;sC4=oz=TmZ28i*?}W_8jx9b@y_)0TI3ym%=P6L_}=y&pFN(TSFLsFVbdKJ0F$VYqr*ZA88
zl=?d5ZZw?#R`rHPizG^mPZ`W1G`qH}nQs74E*1@g@ps
zZFXd*2@6Ggwmn&VEn15FX7?WdUQ~pdt5d78%nHyOz2b|$ba|*eEBlAiSPrs}Q4M`G
zTY%J#96IEol!uxEre1An%tjXin#MK%W+Kn;A4A1sGm%qW(#;jHGpyp0^!l^usA=Iw
zG3{mG!8aXQUVlnL&pJ$vj&LQRi*fP?E0W%zaM#wzo8zyM8DH)0u;@rsEc3ZnrXdV5
z(+k}=;r#-YzWsT8&s+$~+$#1cJoGto%T>?in|6WUkyf0zR%eB{f8S#mU3EsbLmU^G
zzvsZZ|7P}3@oY4M#0v{GGmw`BoV>9y4W;b8M%(3^h9<^7mY1ewqMg|%18I6Q(D~g7
zcC^hIXw&{BgG4|09u&2)@WM2F=W6*XeGX3&a*5#9EDcIRCA3t>>KKyI^jK#Z_4O3g
z`MB51K^6G5g7XOn=NG8mK2tzqHWKk(Q(tJFCbN$Z@BXz@%(9;O+bhb3X2Ruh$5
zQ=EZ}7;WacDXIR7c>q*J!uH-q6|76x8tJ)}3m#RK(;=ua|#51F4h+wy6{)poq~p
zuTZDgh{fP%(ldo*BqM9pIMh-
ze8O0_lQ{IlBk*{Nc>*fsPTKy3=M74Yc^{uqmxNR}
zVg1iN8Qy!}))=`IjXosr%r2KnN9LQxsbcst(B(av)dMoG(16{eBW1^<(T!m)!K&|X
zQ1Zje>vuM!qp|?Gx?0{;^i(e9^X~srk#=?G5tf2f#4%oaanG(a)V#ZG-p2>Vg?&=`
zD9kT)7B25Cfaf{nT*u2mn~sWNxqRt`Qjut0#Gv3-nD;jRZPaUk?~B!Rol$W}KtE%n
zbFW$_qK#|;k=7P(Q6$axrt1R9sNS2ueCw?=w3Qc8drf7bOXsdMZ`_`ZKFpPUP}>=S
z%*0(kU%Z)!uKTOE*(IkTk8+V;|6}Mp+`0PVIF784U-r(PSqT~6d#;gH_6pe}T7Df>HRxCI>ve{HHDP>VjGP|5A0
ze0$sn@=U$wB8H|*t-21#wWIz)1ygcP~Tc>rmIW*2Jx-1{9rit0#;VY2PxbTox
z;A=gBEC{#xb-Rwx$a9JEPukS~I1JO2}5k#BYAOOsy{)o*!CxsWabRlU#4GBGy-
zHF@!CWm8Ln7+;xG1o0<=p2OeX_nMJ}^sDD&tLSCh#-i;0gA71AW^R5!hCU<@OOG1)|=(2%l@}j+baY%KSvQ%Qr`*BP&XIVtbfDcuFay-)l+5}qfq$e4d5!j_*1`6@Ha-huB46#J&$OVmkC`BW5M4Q&-o
zq?A{+7U+I8p?@+*(}ji!K~DhTUR^Zd4&THHH?A&%rAGhitF|wMHd>d6Yx;D^SufQH
z%Y>DteKEbiS6#*lrw!jEU}eKi3bu|PVkTRgEU6TKS^A#GB2v5U+x?E3Bn0&ss#_-r
zHmgf`)8TW1YrP1w$r?RIEXu%^dQ}ViBEajMUdoN7-DtET%RPejFi1;&*LV)QaE|KJoE#O_zWlAQQk@2&wJp6roy~w*ykGyXIZ+y^aojjb!*&rP
zi?f&p5kkoReV5PyPhRXK!au{cCW3_0qiHiN2!?4Amv50-AWQEz2#^10V>Yc7rG>Z}
zvhkm;D2mR0pQvXt5kA#Y{gaoo6zWlJ*M|2ATe#rePU-jTLOSbXYaB7~eV`4TbZ
zQKt~(>FS-akIsWRsNqmY5tHxikjIp#$dJVU%*mN!puCSmXQLY{9*-$&l*v5&v~=7pJc
z1;1n$%EScQkob2spRiZ6&T~7XJ{W#~uPh`p2D?8qmf_!@jRf1zF2+xEU^LDu#AmIV
zku=h`sqs#&Scc`)MP|mgh*zHAja;MLOO@y7iP~mpW5mrraVC9?23Rrp0ZE5@wfVN#$m0cVEFolIo
zOP-ufB*~F|o;hnBdn-z61&x1TRP%`>de61tBwYbEJhp%=tGkLzd|JXLl$qoeZ>%F4
zwvy?`pbacRoML&zRt9_tu`R>@YephnU%Bn`Oe5z6rX><9Hn5$*6xVdCe+bEnpx}X{
z7+jm~!^mzhLjsl8e{olcf$bFgp11k{mPWQZ#@5XR3vygse2Fi?aI^DWt#0qIku1-b
zH>8E&sr53v!%!OhBB7NL+MokHBT6f|X5Qe(<*=>f7lu%SPK4R-&=SzK+1#Wkb%t6|
z^)FviU4;hCWaS&A(O`O4;ppC%6Od=*>a0G=MaWLtDNH+qz!gbe!RoEMz#)L;yi~gp
z%;9($d1X@rY|*pc6!PN&#uMe4a~Uqcb*eO$ll%cx4}5u8dG0pkBA!n7z2*(CZ15Ay
zu6cs@%(L6tQjZ};aN4z0nl$jEWn$4lGY?!}qej|U&0zW0tMgWG5@AY;Yi<0tKQt*#
zvtpit!22t4n7zLvAnU*UGsoQrn1^)7&k}Be&UeGLrqgM#{;oNgP$>n{KaH!H@N!VA
z&q)yN4Fn971e+|%C-D9&4V}>Md7w!2-uat8H{lu|J()kr3;0_A>EZW?fo>~J+cqXS
zAo$`;o?GQ}xV`Ox==bMCH?>)+R+@59_v8Fos
zX9mE^V}_l5xq6`cdo=4!Yb!7>j+tTT_y(mbXP@@ZEds5|Kibz}H6YqhIhIx%(2FZN
z_W7w7fR$HYZGtxk$i6NAB{|~=9{yQdH4P1ij-t_4w=+ZGYL=1Tm6>v=J+_fqYV-mo
zjj@rPyO#`Y+cPFUXi*Goa+{K<*rr7BRA>?7^9e_Xz!Gox-W
z5e8VA+iP#CMS-Wx!9u1i??CwCfZju`YcOQut+**An?*P4tV
zpD`11A;ba11IZn&s~y6lxOyLuoRJd^E_MZ?B12u9YWHDb4m)%3Pyv+abuBJE(+(3h
z6Z815ynrOqQ)Qweg}}yluJ#RU3~0FRHe0Ih2(z9wKP2Bc53jDD6?*Ki3tktVAv)1<
z3bMZ=@!~O;hI+5{I}#_fz|PIpgZfq@*g0!k#u2Rsza{eW9LS#slzyxKx^j5Iy_mb0
z)hHjhrNkK#FXRB8wt5{(Qv+b-e=e6F)q4oQh%9yG^5`dkjp5@C_BpL%6FffRiz@QDc;#)>~{uQMuB!BbP0em3kdf&2;Y
zwt#b2k#Len{H;@y7>{|Rb|=w1Qgo}XTYyUm_HT<(P*uo4#_|dA!CEq~xxi#YTi=9;
zN-oVr_wHd%(eu8JLe^kqU8j*#Q4Jns8Ou@e3d4|uHu*QneBiY%^$GH*Q}ATyTRVU6
z&xrbJ5ur-19*G;hra8wOi}~1&m%RP#iOldFayYy-LL3UAiS4Pc7!Z=Z@Gz+ydvnjy
zPS8FP$)L@CE!BAs%Z^F@Z9a1w88J~i=Ug(2j177_X4@7Zk<9)5l`$z;X|;s^Fk3Kk
zu;F^Zc<%xBIl#%=Z|E0>f3FpjQz%D1e4X%Y$SOgGNpnjTd>>#RblECx%OVh+sE5ck
zdpg)Rjb`2O_=6cZySKKQzenDG>^u<2PeQPeo$l8~H?X%HUPa@vzL*c`=JFD`8pcD>
z&aHNvfC#ijPJDQ(g1z$nS}>p{jclyW(1h#yBAXNo-o+#uSmP%FCHZUtWYbB>{g1N{
zQpIyvcDRmUlkKmXPb`ID_8QLY*YX{ZdE>MBomra4?~_({W$JMDs_6*`xr=uoRNQZr>Tn?^x
zJBesrDk3|?CYqc=*9;2NH87-g&nuLN47uiP7I?+;kg$DAXok;}2%FimmuGrIf>hV!
zs~(3BkhqKMi+1aR*s>GJLcrmCQyleeZ8br1q^D5DvE}(GY-2K1dys(#Vhy7U2{{QiIgbAK_tN`
zX&_PlfC~9U%6ek!>NtV3=k*uj-1;UNmmeP4E=d)VTe%EXGe;V^)^WtpSnxlTty)HP9eNA#EX14pZpt1c>-IC{=UfK`If+I
zZp`O5+)Ma5r1GtkV!Ej(jo<9aqA6nR#t@@IMT+pWbkUBHgFN9*&@L5ft;iehSwI`azd4&E8
z-CW$-ErbdIg=4ihK=_!pVQ$;g+oYbGY+tqhgFyYcuO?TlsEMO_C<>NT5iS>8?NZjy
zBSdC(yST8;5m?5e47$vJG==<@@_8XzO4wxikRfootZ7hPDXi^dJK@(Xe&i#Q(=>oZ
zZ;J|$V;ZxPcLM!)38rf{N$ycVDl)2^wSw*AzlJ61C#=)r9?y(f)_it9Sp~)whL^Rfyav~cEvf~)z
zlW4`f|2cYd9quCZDe_fM7b=ij?q00^1<9Bb_mFh2#vSa4CQ^38un&>=Z(;m{!~`Ov
z;F0fmvl(;iu15d3l_AoC(RE%c5t!(X{}1WbFPP2yz}UAA-3
ziLN^%q<|DG6t%5My*L3AYzx?jgK2;;N3ddNA_ug+FzE1v`8<&OQMVH%!2)-OL%3<*
zGlH|Mv?JoQoG{8tqo-+00C1Dm2rrN+!B-L;ZJvK5!P8D{Ig%@9A&>npMe8YP;M2T4
zWbGvklOOS_D#~*J;l6~YuBFBxw0t2h?vxSS`_I6vg4_aJ-cS3_LDUsWl%6m8^~Mim
z{7{$4WU!j(6lPaJmTiq5IiUpgil75}#=Yd0%&~_UU
z6Fj=yV2J)Afwx8O6ir1OAkK-+RIi9{!_`=tE`3BB&b?z%jQuDJ3N^gXevIdUL8#dM
z{oj)C@hO8ulJtvELHlOUc9sR`TJWOvpEiUQjD)n-*2}>9O52XfSp%R={>SeJ4h
z{G;`(sbW4KgxcjTmYURoykOyOngATOr5k@5;g10~1GyiWH)_HYAMII^D9b=p#ehq3
zZY!K_oBC+LSPN6$G(Uaqp9ig&Vl}_NOb3!xWq(KPFeozxhNC09!P)H#z1PZnffn~Z
z-CFn?(0cnjv+Ma*_-oeMz~;}ObZsM0CRxClnVIxX@j
zKSMLsl-{w^TTpxU1u^c@0Yh3A$+lU?z?UiblyYnw*8YSH9U5!UVD?Apx2g^>d-i&P
znn53MH4C_1+BylV<1e%-1vY>_9zB^0&i{aok<~}qXFq`Mv!c~_k}r_*UD3^NmaQs4e@*g>Q>y1Ag&o~ektY_`!@je^0xTmmw$rV7*I7(G6X2v
zh=uBMbD@l|w{TSG0Gv)45OtHAfL?y-!zMSF@&D%YLN|5k@yiRGb>S-e@XE>Wh~(QY
z0O|GjtoN3I^Cd*r!;TK8O3&p0Y-@mLiAC|w|JT~Bkc+JbYIKwtK>ASs~;W)rGqe%Ns7iopb?a$Zo=hVqF8{pir#p9(}4=qGK2bNjRK#EE=;yBGFaLcx|fAqugIr(xX%du`p|6v$}xgPMh)a3%dW%nHwBnqm%E}FYp
z_!wsM7YQ3ByMsvEt8upJ&LB@upT4Fx0T9#nF>(y#0gvpO-suuAFd(dB8M+n%|7j_x
z)6i)__LU1X>>RJ)w=4e?$P7v$bL)Ab;N`pE0=Lo$b*@OD-W{qn<7Eh+x_wSln*0b1
zw7>2LPnLtFkF+a!iIp&oP2>5Kw+~^j_P)akuFgMJU_QfqJ*lO`iTV&v;G1qv9L4EzkF1hm#r-ok;@`0O#239UmVaXLLGoT)FT1hz+XDA`L*6675_NmpC90S^x$%V`$39EV
zMw|yKxPn;H6<_W#YQ*}c`Hswro3M0sf2y#n&ymltfY`N?8S-qlI2De5#|oc)75_H%
z0coYPRA-QTk9<&kD&Cmijn%adIb^8_f$#D61(WY?A-Ufl_^1Z$VRzAHmD>S1$QK^`
z6UD$Uq+UPuLA}={(4tjUVP~iYET0Ll4<*vV6!Q+waduj$*7V1Wri%^$*JYU-*Q=1k
zn^F<_ZZ()--VsIc>oDYQ>&)6T`T%)j`mayCHWjlrlKgj0qZavbN6ja_u^LIgn}7;b
zWn&HV7Xu>cB9O+rG)P)@5nan}?4Z3C)AbKlmi*d`9TqH=ir6$GDJR5ic#LZ>D?fBt3}Xh5
zrmh36tsP`3Ro13+dli$t{M{~$Fop<8cIm5(cOmBb-UI%8S(tk6TdgH(HH0ESIjp1G
z68W6eKuP`B0@Gn)31_`(jJ(aCw%QwrMOp>aXIFFYVKu*{^xqfCA)Orj1-w(Lh%(?w
zEgm<+(iTNlkw!OcZ_UgwH#!no3Zd2fe%l`NS`u1nD-}R`fO;vB!QzNJ?G0}uOiY@~zNcSxw0abHqB&L_`tMs4IVgT5ZwSkmRD)LPDjq#9Zz`lY#ydMXq4dHK8x9lJR~*?d~?u;nTeoBQ%z
z{|QCRK?Rp{#4fi$~?Ev!mv_`(Su>n@{
z?1OmB1!LqX`qjy6^B2KNurOE{95r3i&s?|vPL8m)U!Jch62N*jBO5Xz6_8BtoRXiM6gd{Lj>%l9N7#Q5M$OzYp2T{1cvxqltgjAn&^`r
zCHV0i5rW)C3=a=aV`qF1Utiy^$Fwpc(aLKqdRHAYqqjs%!>o`2|BwM1LIBdgVyPNr
zVTO?T44K^^m%%&^4s^X8__5E!vD)=!o`|Vj(L94l1QJs!ZgyJA0Q)M^n^Vt
z@W}?u|JBPv{QeXEbza09`7lCrUp@LJw$h{T=i}jpd|2Wh-U|%FDB@?I+UwjvcF&Ki
zOIcfE*^hM|Gb)8*ir423qvf6;b@jU`*Tm~F=|H7VdFC^i>=J*9tLH0(bc8efY{{+cSI{xX0`AB{eF|8yPU|Bxo-_pb(g_Yewp&q022L?qq|N7{wrtB-`D($m_?{f?2ip$4buE%>ng(-XNBFX
z>uhX*Zgm#PZDxW|2b^4gZ~evC?bI|gH$EU1B_MF<;R0ei;!X)xIG{&seB)I!0=(r&
zej!Wc2a;}QZ^{-rjTIn1fXM0j
zoPckp&_q*6vr^L(d^*!Q`!eGylv}LUIofdqz7x`>wqw_!_i_UqryOqSGH4GTnjT
z%=d_12TRDq9ZG4Hc+ZGU_><3b)Jy-n-!yQk*u
z{lr!{<0@H~=U)bXU7lO(dYupBl@@3!Cq_Vd(9bHjm@n|;JsXNQ9IwC&YuPKP)gZW1
zJo#Fc`!hHa$(o>4BE@@;GS|N-|AB_zPcKmCe1-fat>eNgqad!YmBw^#4}!v9?*+n$
z&@s>AMqPP&bf?C@J2{SvOV6qJbEA=`g3gef;K3~>1&37V
zq;ZVuk;&8}29(j&B`3~`0wp3#y+(I}6c4Yq=boeG$C-Dvg6~DiqEsPzEL+?%=vTgz
zbcP1hxa}IH(usi6_y?8KanlF^-H1C^+1M|DbFY0Q9PB9JCWPxE>Pd8H-GS``S_buF
zzNp_uPqaQxI8)-CuWpK8>-LfujT6MLwii#G8&gB~`Sm5}aZOZ&Lu7?k_72W}&g8%C
z5)-toiC(bQ_a-Xl$%PM?QJ|woCiFuO1W|t4a}lI>j(`>98E;6Q19|%jq1^;VbeZIm
zbwB@kRPu#2TfNv>)YXBJf%_yUI>^_^+x)f>>c1p+;rV#_7)6Ou{5d|aL@~1}zx&Gg
zK~mWDZc}NTQ;{;;wYl;LTcbg1yaGEZRkC%AXjgux)@2kgCCDSiU)fKA(?vy+t!IBTfzrc=!_
zREUY_u}CCA5!1Es8
zcs?D;qO1nW-FK%oQ4I^8;*e`f=tZ%KMc8x(@2K-(;7OIjj~Qs&lm7Y8lZoYuSNJSQ
zyd`ZmJKPS4?n!%$JZ^$a3ZN5{4~5TD>eN5({e)+V);<(ROvAIZ@gk?p+ksCN3_f&f
z0PQVP>7H2y@XxGZ(9g1cFmCXQju`y}oVx=ujjoSEB4){imBljH9;_`Y?(`Zwwq&~D
zczq9QXHhxRdmVsTF3VgC)eW!|J-&}T?t|y3O8V8dyI|(Nw@>iGIUEXmW%$ktX(
lzXaYO*Yj$8lLqV5%kETn*n>BQE_wgPbijj3f%r3jRbVz#-09YT
zb>PI`z}tyZIiMu+*06ZE51gc5)+($LgDWPolY*1_K;ZMkn5#FtA>Q4o64_b`HH90V
zFjA%i&qt5BOgcSaB#p{FCy_vyxkS%x^mY&`^knpoycvW%k6o!9Y2#qI0EKO%VHt30
zZadzvONMrNhO~;Zm!LB%@7HW9WpH)jbeUhrv3@f2Qux9ya6I(RsqHT#1Jy<;4QhvS
z(EW1S;nRfUxze}Wjsthtz@PO(@%>DCAb)7dnrg=j@uvE})GC)jb0#ZM=a4pF|J0Z!
zmnj2nkEn7>RDWady@QgDi;BTA)oc%k@qPGIrR_(awL0{AHJt+w%>cdozQUP9P57`o
zkK`_;92CobmArB{6KYMp*Lrdz8D@;-J>E0+0n(cJj%$BpARkG?&ooMRs7&W?XmgPR
z)DwzzC*!F>Zsbz4pwTwgY;!BPgLWL@+l=7my*7tj-1yHq@QMhG7?A70uWVv9oZnxc
zblygsc)gvvo~~k|Xp>oqcs;^9NmF5WO%`Y_M1M3OtYTN%;jM3lc{K7DTa0OaQ5rx4e8i*-n^*5Z
ztN*6jix1>LN!TTOjanHXRT|sPFDnDxhmCx0b_#-JIR`N&-Z4ZuLUX0FqZcb7w{I>J
zeu9KW2&+wPJw|}3Xqj4RIcDh|Qf{U>f(&x}zN2`fABiH~*QvN%jX8REUyodPig>0^
zs69TJg*@tHdNlNY3w!#v^K6mdXKW>9Gk0008F_s6(!|+YRalGvXuw24GbxtE#NmMfmfG^Loxy%;M*d?EGu^Ibcv
zkgaQ@#LouFlo(wRPE0}`(kR9?hDKszr>)p2=}nL|4O&OPEIEYwaBW#H!wwsNB-7kH
z9))CHdu1#?{RFw?A|!P8zdIO*x1$N0i4meUnNW3B$OVzBcT3@*7KR=3xr|
z*k)K@B4YdDl}%8}18h+~&7L1xVebWJ_;YN!vD=|19!oy!!_=dcsD6Z$Aw7wL9WNgy
zU~uP%X7*wxcKsB6M%{O1OvcUk-zQ85scR2Q)2=nfOn8mnzht|KWHr8Enu&-?E~E5Lr|#*hB2ev72A|8UH@F^R~n499j>lpsS(4>LUj3o+$;(rh6uWr$$1
zlPJqZ4W?~XUt}~sg7tNaS5`H@MG9t{aur)^u?pS`%}aPQLeBYH&faPk!;eB#Kd@5(
zR;jpO_m`8gAo;Zni^_OB87?pPVpYo>_FG+wvRCp9}qgn7c)K{fh0_Gy>l0x!PL#u_wQ?LAZr_r
z*R%GH{dY-OM~?^uCS7?T+M4?Z(YQb(zvw!Ssot%g+xP89F4&Q#%3CdCZkOIx`wZv-
zk^vRm8yx%m%?|xgWeBrWCMNHeUBn6(sQ!?O@B^Kv-tu}EMIeh9Q6p#48PIWVPRost
z4^r_^l7{q(f#q+fPO$n*!mKw{S*fDl&@V5!uDwkfa%|LQOP`d4yI;zxbL7>*p^WZx
zvJPjMq3~sptSJr-`OJOMOqPe@n+sO^Qz&Rtom!am(18oz-_#<&5|Bq|B)O1B0!!ZY
z)hjwiU^G#@Q_WcqR_jd;S}7U>;lY=Vul!|UL?ogY%pU^B49}nDH!=gk9QA9@1I(bE
zXJP-uwlk0x(7Vs*9|Ik3YOR%~JqAd)=ffdhYxu20+2B;REBI_2${DHT0l$R|`d6+5
z!Q$&>QeVYdLI2{?O!ZMFd^p00$T2r)m~w$hu21DhpM~20)}~d`by6fnC!0DG_+p^QsjosWmg*D@OVVwuXi7S
zkjv?RZf;{>=^CqRUC|pbzV`Y^)8#q*+}Fi^2g`wPGMc`{S}lO!G{L6|t~0=}`gFTe
z>^pcVG3;-tTshRLbjkYdQUxjv1W0NxO@q(>kxP8EB*yEQ_*zB|-@(w7iBu2a6fj@n
zviD534a6sMD?a}-2brwzM>{Nkg>)5Jrb^qRK!++sM~eFeR6Qre>U{kx95T##kb7YR
zz@zKar_?CWWoDaL60YBX_@c`>>XB65uY%}yb>cyvgaZDW;%)S@nHXLXJ+e2?$%tR89i@w)E&q^j?<(4rS>
zy$QuiXurpwTlQx1__>wq!=E+vapuKUVpnb}lqNT~bUNPWn8RYcDV!;UM@OlBJ`-z%
zr{}G_HzR1H%d0IZ24mqUI80$B>^b2%Ma^xOjE&Iz(Zlwe*_vnozxGAmcg}dN()oYe
zxD~3@mO(<(V~fgMVRB?1)yAEd{;Rn7+zpr3R*Wu=zm8H8OLC7I+(Wn8I?710uA;*w
zk=)#~dgxD$2~I@C3%z$SYl1*ziy!$%dD4s~<73WKPTrTE;0q479$W~$hOT5&@}9r<
z0DtwJ%iLDI691;xrC9h3I$XLfJ(ewxX8qUg6f`D^I_T%#w<+MjZ(MnL)aoFDH+klqnR=&y&qk-V
zXc1}R4vIx3{)+r)cGV1;Ric0jtpwgYyUvdO__eVkGj56M-a7d;Dn<_-FuGyA9;%D4
zh-d1x+Nt6{TF-L-jo?SmWNaj{AOB~yqke-#Ml!mKn(Bln2ce{rXXS~WY2#^kC3qFJ
zE%3n06=kmFIcP?pm7pt60h(l;^LO}F8tRg7=vUq6fNptb`gzwl;vG#^uGiYm;(z>i
zY&st>;~ST5U1ELv3k=y`pZl%401Uk!3QX^f0CoTP!d8qO4Xt4P7opCG=g4rgyb+|r
zd!CPw3aCwk4a?g1Px}Tz<+Z9vowK6oiwBDl$ag7}+e`B3(c%#pe0z@b(EKl4tyvV=
zB^iX|61$9S7FThSq*aMbc4fR|$zi)53ZZ#432v4NT~NUB-Umj>m1Cbk)_6_U1y7Z5
z4fB_Dz*mT?HTQLmE#dyZ6)&>+3Ne@_X_i
zrX27a7u;s>?}q0V%z_`mYi-amCA)Go
zFc!v)rNs-`B?Bet?Xn&2Xc+G*l#w-P2E_%DS5DNQL5g<0yw&s@sN4DH`#K>Wy1KnJ
zdTNvner6A;CPY1jUN<+to;{N^GUFQa4r^*9d(KMYX>N>1AtcLu{rZ(a7?@)ZzsT{!5FQRJKkGl^3rmVa)i1az0$!Q^`VzO(
z!1rI0n3^>)r2Z!3>teButc~76Syei*8M9Lpl$TUsu}Oe*q3Im<&Oy?Mo1X|=V=A*(
z&D+J&=;`XiY1a@<&Unc$Q5xXG)j7veCMH&gwO08+;C7n8qSnG&IebWyE{6I4c}~k(O>-ut0nFcN408^B{Gd$
z{txr85T&S#jXyFlb0%%o>zq3XqwqOZ!CwQ&Rj=avCr6vGe{(mHP30;iFWtnjFfY*X*v
ziU`|abT#6O4f4Slub+~qB5N%&$fT?2c=|h1OQFB{_~M&y_|
zw&j1cWB14skv4K~?Avs<9vOO
zJKhY$izajQ1(5(9x|8t`E+Off$YxFeDGVfpIGdy5pkPJUA#KDK1U?|LcjFvLY0wmkIq3me)*0rQdlZ0ho&w9Yk8)7ULlJH(s{yY5
z8$ADd9H9d;+|zCr4$Lf1*zaa&firh*-#FZ&gx+u86z-;i
z<(E4ofBeFKNX8FHSX)mDwj{xiJ7GJOoX?=^u@s}lu0kEbgUzss`ND&;f=r@1NsX_3}lNP5@CSu-Dfx1FiYlf_Mn);EhJS_|wZLaF&(n4b`>3FgD5ONv(1hOz2gAoOf#;
zT&jDwE-tH_oauwElX*J{UTYHZ|*DrKf9FePc8>gE=-t+
zAzBaxPc%#$eyzcxmK%~UuT!DA;a?RR=Be@F;s>j%4D`o6JvTfDud$&ftYVWJ{i^se
zWtu+<@S?!1y1DZ@Dkf!g=vcUCECh0NCCn~eloGSa^ADn$oZAlZc#O_nHga~-2T(H(qFyzI8jOgO$E
z8F$&@xh^_FdH>$U5HsAo@_}HYiYw}o8}#60cr?DNczIm5!WQT95SjT(sfqt|OX3!t
z6-5oRZLO=0%uu_3Mi(O5ypP|xU3z~8AK*!)u{o(Kws`LqF)oed%jkudAtYT9`sgFO
zPZ5){+PKBlih(A=Yy8#2v%4u|hPb8QN!3ucTlnaNi{-To-e^(}I+=azEgrMc6kI{n
zimyJ-X8&#$f}c$2U8!XDMW@0o+IZLsaXGr_v0FXuc%(^%-p5xsuB7^$u72w&uAng;
zLHn@;^$iPidUP7Y7k}zHyyF{Ld$TAKU66qe0>Ry2;ykqRFIkdtR0=97
zY^_o|>4+BLK~1LP<~WlWeW!1EIeOQ7gX5;wOVnZK8+%E8AxeFPH6?d7675w<7tkS8wwio#7v|6Z+VRw*0twfzPfJRsZKsYDLQP(_n*gm6>>J*%Ra0uAonj^9Q3}^
zbMiO%*h^DRqoRo_b;Ra46j|YnasT<>DrG`XS(RMK_{fhYY};x?EC9TcX-%EZ^D&yk
zrA#^hKnHEs9{C|kY>yW`ieyK~&CxniHztW;74))_V2aC+=V*d(`oO)v3FtZA0UpX}
zFMOd*E;4Sg88T$0YCY?7MxTk$oFs0kKxL?pi=>Wwd@BC4OkyK^UIf!N;c~F
zw~nX}E=>&hUH$pyidtrL)@O58q=OQdSqq8%*#8R(`CVQYxvGZxyXy1@+5-GfrsSVR
zM{;~%6@hcSLiH!K2n5?t5xo&6|?E{JFQWVu#-&wiVgW~EeZh?cna6`Ca4V;whS+W`#f_cpP
zm(P41gpcfaZJRFgp_>-sRv&^-;WPoH5~#UR_Q>lh9xnNcm24|
z?U!cP%MN&on!_*1vL4h34b)b|K8HX1g12tmc7keynLjxf8ljfyj9>oD4BUeXGqgrE
z;DoZyyMpRuz+OC4tNX7Oj+U$ZOfj&6uLrp4EM-i9`@>JVWRooLRvKrU`7Q$_axdT`
z`XK=-_MMrduZO_X5ksxCw{F0yVfMexPWqsdV9=4QrK&o3hS8+LE$cQoN8$cDO~o6yjSevt)8B!_GhrU|
zJ}!W%IQL}pyae!n-f|;JaSf~2(!m3N&0;>(Iku?^a|ny_r{l8M5p0>$#+CFFFU)ag
z7TliV25$j{fXK7&m?&q}8i`py_EWoZNvHc4VlOo#BO7A?J)#CS3$-QSXp-XP*b7X6
z?cvkHzA`d6Jgo6#>u?#95|?=JW=5$b#a%1S{
zEq7MiFa|Oz8nT66q|H17vHLTNg}SG1#VXBVx?X)b8P{5o!C|Mv=XOsJ$}7P~0_64B
zS(O26RkJB%T^b555;KFqXHVV8I{FdA^F1x|kH;|`V!P#RuNh>FZ*ukD?i{wttQ0dW
zN)MGM
zdr}@0sen-v6*qxiAEjWOo(x=*j&KV+YY2&QB=0OFT>+;!XxsM&Yk=!DhVT3{;lQGh
zC4YX;2R8cY%)Iz{2YkP@xN%qJ2?&V1(*3lm2!8Qjy|5zp1jK2}2k%g)!p^R67T5n7
zI`2m=yD*GrWmGcDNXW`262&?8%$BTdWyWhol*o!?6*98--t&8o87V?UT4qX_At`#l
z|G@Lp^E~I=_w~81i^!7%cT-lg2iKy>6xDlI?r+EfWCmZy&~>Xq`J0gBBhd#yfe=y
z8t#(8;2R7B(ai1On7S+(ZC?ZW``gJ;U91UEg!J#M-F*d=iSnik_XeQ0r*w=r=mhAa
zp~HRu=jdQs<XPM7+21vcyJDG8u$Q_Uj
z4f(p$#;*NzS2fc%XK8M`1f612{(35pJ+UmVh$hs9CD
zf6UBII6~0|cZvc3e5iWZRZ>V@1BZ1#d6~?hhBYMfW8ChkVtDf7#XpXBp!~-Qo_g~J
ze8aLQ6?`+r{&M9NFZ=v(xkr!EC}{-DWPDI7NmGNZCEUrxDtqv3woycZ^c~DQu&8{S
z^9FqD&QZ?n;tewcz80}O>%_D_r^CLiH(ip8QV6z81KUp$sWql5tb$zk
zKYa>6F$iCXS1~&%{Y!@8-OPkyd>$+cCCA@wC882%4&)@5ZTwOD3mNVV(+BcJ$
z&#V}d`tUlCyd}UkGmBz6g%>dK+MniWj()ggxwHI8yBo$Gtkz4>VYur(^)C4ff|>Wf
zQp=}SK{1tAgoWlTsB}p5p$YMRDjY$6%eu}5+iMINkY!my*Bj$h@3R}=5M>PUY%YNg
zJ{z;P&u+s(6%l!;tCe)NQ`?XadFKg|0>zbKy12M3C0**+iohiD#E{d=qj2AWrcj|~%b
zg;QxCSG5a%qm;it&U%}jfE*W|=8MfyVS-MJa*O;7%BQcNyZAy8Z>g&3jHg_{ANNXg
z(mygo<0ZFHZM}Iw)u3X*=tYZ_gx>fRO{+k`u&6ppc|qJUA7lN7oEaB&MGG5a$rC)cq7u`1|jgA!ErjVSQ229U}Wz?3c&V_}uP45?>
z>wI+TlUw&dX{#7#iKrQPA}3xIUT%xH@Xc2Pu@2zBU*fNiDQ2N)`FlbYEUAcN;Dw5&
z{D0uVaQC=zuoy~_Fw%ZwcOA`gw)|=??FAYnGVd;vJ_1L(Je=JWRo{))X1Pjlfqz+}*gU<@|zCX_-q9Q*Nqx3yn!2U?HqyJ9?U}6~1^HUl@
zxf%DnGn@+mZ720G)#(Ssy~|qt;rw`X&`#tJ|O9G1hc21s*xrD<77(4m^_c5BD;=zU_`c@GU|pt$K0hf0INSbWh)
zU!g8wskqkh%p9#)GD9bWLESmdxi1BmE|-9u^d~jZetjt6ME3~-iyv5hnS17vY$j08zc*r2m4+%u
zivuoc=K-aXb9_^k%_wj2t5vDaOYoHT;=82SOyEeByQU|biK?WSe6ISG0{U=ry6<)`
z&^6j-I;Qu;dtxlQtZNww#2;O*6H|!<|7j<9wG1Sp|D3nk*-UCtlwgZk!;4=)^t$_a
zcl2ZM(BI5z^hOmR9p~?-m+L}5Q%_I%^}h#i7q9&ZGdK#Hn|2?KHBEq?C|$iV=SiUR
zDWUCr%?~uq9_Eb3DBwo7^{bL8Y1k(hpgzpI0E!$N6GD{-5WSPd<~mImn9GVbcA&2S
zhd%wGc0bDxcL-E$CnINof}!P9io+1_VoTrGzVsCtwpx|dT2Vl@eBaDgMjq&EKN(`x
z_!AsrJ}&*O{Ts46H{eVf_!$%yM5!j(vB0df9GQZ45lrcP@aIz#J(OvBiLLdSa9l6B
zZGI~UbhR$dGvZ^#3$3p2Ggfu6{?XCbbxY^r-rq*f(|Z@OyR5^AYOxX=oSt_%8MBT0
zV^1n(ir>T9ITzxV$b@0{BugUIO<`>P?2ZG8+IjdT%xLt)ye95tx@R7`?Ty)He2GnO
zBJkK)Kf_MDG&X-Y7ES4+44Z@>Np_W6;0KSiUwpE(#bMm~e1VeydMy5-4xUrT7vy(m
z;KQr1WR>Fg7b$;y*XEP8SFt^`qrH5EOjiL`UyowElW-0fH(b^HpkxB4x7Kv!dqeS2
z(?JFfk{D>Bd+7)rT_7Gz@H9U4pBO)e;J2;9?Awj7
zOny6X&M^l*c1iS_`Vj#aRYIB0d`3{if^?z4keF9bSxVcK+y-?iirkMpErUgInmfA9
z>5$IkWlD(&f{yApU($$n!tp)@H#=g!_50^qrSsqG;Svw|zRp1@yxgWFxj%$pm10z{
ztR9B>QIQ#YCpw{TiCM_Wfo2$=Me*Gxrxr3PjTtlA_&`c&k71cuC(LUXcf0hc8Gcg3
zQNAx@07v^HtInR5hOeoPT~X7Cf;$X%D5xYwu{)oW-5t|wxG%yyZu~lNuX~59tfh(%
zKFnR4JM$(3a(8qzzbp=dqc(=eZ;?CU82YDG45L~wY}!bu$w(Kr1=CKlTa`hYAF=^r
z8mW*tFtdLdb;N~o67Si)jd3!8FnRKj9Sr9~_xFNpq5jrX4v$g+oc3v0v3jly(JtXq1&Nl3&FUT1xdTK55c~XT&a_8lsWML*<%W
z)ZT26DuBnEeMbQHd-NThB$z|V1#c-1YkMfIV$*b5_7e6B@jA*IDUL_pj|%V&U4ojN
zpR%rNQ(-bT`Qf9aJ4ok2vc}nikDy%dYns;{MtR47lQaG${x1dkKNtv7VP@AtuH&q8
zNNu_-j8SD0*p{=<1#l0bi8J|+bbU(E-pkR@$y3yLYoOpf$1^f4%r_q*ojDJ}-pp>9
z`WB&?w=C-#v+q!FmHK8&un_DB8Xk|*IfM_)7Z*EBHjq+1v51Cg10;WTv^qvQgB(lC
zSJ#A;Ad4E$mQ@)OtTVrGWcwH$&R61UY4~;kDw8HIo#31Wl0qX4*V$g6=*H4mlC*Nr
zDD&UHzT1J|GL2qTnNkc|FeHRe?8SlVds-%4-cL}-N5PoiO;v!??d})*KOt!T)U|a<
zaToNSvhELURuow8S`#Zi`~m$m%lM@1RSc{Ie#T$iP6y{%Z(pVQF9YQ)WvJ>`C4t=!
zubvzBE+H9`s@?pd*JwMFFQl8k38)S%kkkn0pvj65w00^5#WFQg&g{Jcc3mFeAR?FjD*T{8j^sAS}@6}C^PW(r+Eh=TtU7^wx7om
zbb*h_$oZWCRs?^2Bl!{&1JsOyn>e&ok)lY&h_i$SGF^LZIXxv&!`I5_
zXP=*f5yK1jN!)!v&gf%3adCUJJzl?MrC<#FcEz*)lF0zq5BD&yXdHT!{AS4SS2*Z8
z$FP1(P#Fc2J8*x9)I!4_Brh3rn*-lICW@BP7No#^^7=Bf2(WEoI(p~#Enw7}lm!jF
zk(AJz=hDk2fc57Kw@^nCP&57Ej#O!hLf164=ggTc0*5+%dza=ZuzLFKdJVfhpsf7*
z&qIqIFlyAQf1Kk$2YEcfuXOpqs5H-aRo6@C_eXhww6iL}uiZiY8#fQIFu(G_{UIqj
z@n|<<6w-mbpKm#WrBu*~fMBOzu0cRPc%>(ZT@uiD6+FLD#{+(SCA(1&FM|qX&o&`d
zW5DcoA^&Sm1URuaTeX#S0jwy#nAU!HpwK(Qyh(cDJQ8$qy3iBm062c$is|L@1k^0b
zNrfG?KaPOCn+Sw{Q$poY
zc}Cx~D8W|7RF+7V7YaY^=uk~v2b#HpYHRF+!T8DWfk8SiG{3^x{Vl)=6gV?E|K&&k
z`}SM*b=PB&^P6wON-VM9%GeMC&W%SDT0dm-#nQlvP0pa{i>3;WzX#DlA2asK183|JeJHD-T#ylvrx}zXwy_Sxa}t
zf1rpLBPEL}CZs-#v@T=V_m|X(+-j#>3lclGgCc)fYOPV$9J0#YWDT;CV#!du>R3
zWy3{1+0G3W#xQ!kmo;O>7WdB`+*Z0S1xubKAMhWM#h
zp-=DOGRN1#>s$GBneTfz!PD}a5C9})PvnWDB(X&lVr_-6h?ksnXL
zOt?1IX9ok~<%5rVdtq_a@z3Q}_u&`S3=Zznc<5$ur0dq?W$b)zjx1i=1Q*%QJk8fM
zhW5wqSKYqqge5M~nNoT+Vp_ex$KM8xaC@h2w+HaV$C(Fp*sOz;LGvljY0!j(BxoC5wz^__6M0F_r@IqV
zUR@Xe{-7Oi8jL;scfA`k_R5l^+^@v~S23I1b_X^k)B9U1)sE?DXMwz2J1*JY*@bol
zyccihtufz;PcXzOWRthxPu6j!uirFb?XS`Lwhlm3IQ;gp(4K_iB
zFWa(Y*ARTbzwGaQ2Ej9WBT_ld2%gRMJn$pl<91d&u6_8^0q5&V&chS}w4Y<4(Ya25
zE~*^f^V#k2@O|3qjJ`G~S?~7EOsoy=Zk8ziAm-=p6~0!_ifeO$j}Yll}ty)et`a0@m)L%5tS
z+I2s(3CfEU8s-&O!bIDfVw<6n@Ht^?qsYJ?(~bV{j<&vm+4bCaD?fL@#pfSb=MP2%+izv8`}o^a-EVYofk9$RM6J
z8V>{gJoy$kq%oCkh!#y!0~EMnKDNZr48^*y->SG!0}rP@Pn&)F1nN}3J8OL14|eO;
zc;;xB!&lQacgC(=gCahQ3#O5>Sca06&aRmU6Oz{z`XF%+*VeMXeB1^1zq(4*_4+D2
z)BP(|q8h-&*rP1n0?K$hxKE^5kO8WsDHn7N+{T5X%%rXs*7%S}+Ks3jBPb9>;vUJ*
zi~Y(+`y=cWaoEXy6W1jroEOMgoSW@|(}z=eM`D-I@m~hL7LT>y-y(je(H$9VtNzR1
zOWPk-(WU(MU(m>@nT>#&%8niA?H
zO1!5Qr^NkH3AxwUeglgmFT93twSv1c;RTtDXW(+b`el(tKByl@e_w8CA8mf-QPd3l
zgqnIo;)c`rfTq@}uKyopd}OzFOhsiKF_U`>sP!|z0`9)w-eVM)=F}}ukTQnG%eV9p}}9RL;6*VBw>2bU%@L;XEEbu
z#EZzbf2hSyMO@1F2&9LQ@CT!xAHb{g<$7GA&`y+FLz!%
zbS)6YXOG_g7rX=}z85K$vD6^4g+aTU^VR6J*oi}NX$0`rQ1bhbE-g;0@BT(fmjIq6
zRZIV*ngkE7I-R}qXau>kl%>{qR04@*8`|1=GOTTwrYN$wjuajaR-8_p0;vWnX*N5O+0|=+TerRc>L!r+i
z_G&_i`B5)^KjTv~1V5h!Oqu-h1*>vn@wbjkgCByC-Kkjx~3
zsLGQoS2{gGroWFJ4HgHZEH}@xQwbw0ip+a)w}e4tw~j;0`7>z4BmTkcrUNK`NPmao
zg(#Sjl=;o*&yQFL%&h?v+(2dS3EP(j3BX5KzBcnm9yooJ)m~MM2Fx{ASo78r(cVW7
z^201Sz_%@Tx3IJgsnMpq76@)Y3yU6F>YPzPl1%v7ymBlGc8!vXNKOUk&YU))9jQlO
zhD8^j?B)aVHeSAt2jN8hKY#TaTL|j6d48D5CK~K?1sU^46{5vMvZscGp96tKJsYeO
z3LdW87qtBhMA}JB++`oa!K}07w-YcQg>#$UMk~GG-np622{HcQdODYzj)p(FbC&g;
z;eI%v_KM0nvRH&16&)D3Mmm8&YppB$>1V+FFx&fY-V-1@s!qdcX@R)GKa##q{R=1!
zjvRV?P6&Gc;^3~mKL{GK)TP6&bpX44N2AD~H>mY5?T3l&AK+UhVOZIo2bSFE9(kuR
z2{bbes_8_#5ckvxI+3s@pe1WwL2Eh$R%EK3-0Yc%x=&wHjN&Me^QZ|q71E4S+n9&?
z3iH6$TxPlf?Hk~7I1NGjHuLF^nvBf?V6DMIG{EdzZ@&kgJif|e%TKE0F>f)
zKE9e=23l+QoWaOhxE0j3l`bZVX_Y_Re=I5mg@QM(o2=8oSamA9%c*{N$78DG)jB_H
zQCgkiV-ds}$^%SFbmw5JUu59G8CmQZK@*-W<%Cx%MdiX(x#4jg(OVIU-1x#Om2I1W
z2&@^Fl~T!4#e~6o`Z2yr_*wCzqSyNra49A#S<#aYKh=As$7{t2z1}>(|JL&YuI|@A
zXucwW5u^_6v(s
zH*r&}uvdD2Ej}a^QB_2!!ETl(!?;c};cd-~U-B;Yu+%D?vU%bjR+Q_rG&U*2&lrV;
z%?upaH^S7YLpE)n1
z29D3%zH7Je1m602{@#L?H;ljJCGpj?7Ba8YeMl{>hdnb|#Re(WFwC1w`{Q03{LswI
zW7~WWYJQa4*i0yd`P6TgEUuSAFcQJ-8c)>WX3h}A71LlUYp$i(BB{M#Yw}5ze|L#hTnrt{)&gr`#C_Z>g=;!dTKbo
zWGaOA@M*YH7ic6fA`0uo+4g8+eW3yKD;4L%SKxN<%ZC{fN?6vP67JKTf-1ACks`A~
zFhuRgR%R#-bo~5#l+9WkdoMP2Zz*>IbDs*8Q7sB6lP0;m!a;^>r>f~?H$
zg;JwQ;nrBU($Kvr6!_GA*8r%NJ)J0@eXihEiq3Pe*tR#8pbS6QsFRv(zFi0
zZ-^w=x=WUH8t7hf2MVwcC8?BkNl>&Q*+;wuhmMfrbEH$@1=Zh?Gi%y|O2)U~U!r(y
z*GmE-{m%cm|7QpKRTBGmjBW|lUkUP7ZC^msl1AK%U0vX--}w`s>~<(jte-P3uN56E
zSeh2R>jsjA4?NaHrqP{Tfu-!ST#!#+v*#jP048QiV&&ZG(UhjtTgQnkBIj`{?0vv{
z#FqAO|F&%@m?&&)?O4tMwY@mq#kB#oPN%ANzHLBKf#t%qIu!sb|3{+u!Wz+Bo-fa9qm7
zf@`TA*{@gX@q+u{+PZm-=KI^AELZM|XYy^-`04NE9~KTkxc=Xr4wGhda1yK>M0p`0
zZyVvAkUL=Dd3o2$A77+M${IAH%?|_;|5M(((1KiyyW}n+TV$T9BdGr44j^=qd#=9l
zM_oUgZ#I8c29q)$raf4N!52&aEUsmKls9!UYa*B*tWixFcDjj##S|AA{4;QPMv2foD|3=dGym;GGgXlMaA1k5{FSN
z(YCmCl@1tJHeEbzB7*7?3YK-n%s?lPm!RxD9+dKA85EN9pl9E5-#F^afi;zB_B1Pd
zB*o|X^GdH5;ENcj4vrE*Jf0>)^#MGH+)uZNp$`I%){IBJ*Im#9Ve`9xd;P#DZTu?=
zvfJQdaE1xVf4-=G+3Vjj@zTYv662QwC$*PkqK~lCNt~!&OW+v^5(;`
z;5yW@FFt$e3|QF_XST=yBUbi1_u`*vQ~vGKEG
zEdCt+u360HpsxXEA39qdes2yZm6oW6gxN8j6dkFUxg<1=jd#273t-NIu@rf7o5!a_tZp%xxu^0Fue>6wX5HUK
zBH>=wlSy=7`@mBh`&=Dd*LgPRJ;i^x+2v6o=U+SQb7klwvnCt1RSNt2yGI|p4!zZn
zhzo#;LHAQGc&0*!AGd^d6Oy4~*u~S$#69}6o$|p(K3D9}Gg+)L9}a_gteNziN})>u
zNruW&9^_gn(0_T+4gNY`sCzo!2eUZ*vD9sefTucZNKTel!mU-w&EC8UXhCyRE#O2D
zw8!fwmM3Ci;ZGiIA(#N&W)3xB#xPj*@-tQGBX1}nAnEf_Jro~SH)NaE5XUT8>@KI9
zonhVW=ZkAhc5uR#%kmbPE@ZCwXK%`!k6i=GM$ynyT&o~0UsLe}tFA_SiyOz_p>?xk
z=>~VOeb$>S{*h{2IcIk0326oHX%rdl{ZWi(TQ~ONqtdXRe!!Onp-6mPJFsiuXC1!!
zpyDtys>2y`6n9K5t1w5M)oVOa^38lF5Yk-k`!QA2G~3t1vrQ!T=i?a^qiB
zutVlw$%>;@cQHpt*h{uy2dqoS!z7WW46QJ0*2Sksao=`DRDHYQyq&I_4o0S7>dM8jVicCde)3x$`t0*
zs(hLu@@Y?b0{UdMWHCuPcob^E2T#7G;CvVG30>`f&gVc()HwI%Zx!v4b-=XOQj$l^
zdHMHVqBXmX8!9$6ZV%+YM-SZnNV4g6klL;8W88ylAm(>(ZQ0UwgvwI=?azqebIX-|
z=bR@|Ky9`A80i6s0Y+9-M1AA+uHe;ivj$}Q;3kXIbw%8-n3QoTg9&OEY)h*1(?d2d
zj@HtFqxfI@vXkfT99Vr9^j+Tn5+*s=?@rG<1GSaL@3Y=K3Qd$R+T=>mVz(i=1WnHk
zkm1fGVm6Y87S)0*f7p$oHsRyn*lZ-hj~7OZ4X6MMy+0-HV4=RjRq4M0=zT(e71fQPzpd*eU+yIV+9l&2qriUf
z`?#wO`OY8UFtKbyudxaKwuq736dFfvA!{ps?+U?P&-!Fzz5y^=^204dts33;KiiV&
zXN%Y#PDHNX^99zU^J2&MJdjUE!HO8!5p=d_a%JqG2AO?6oyqy?26&sRYx}qJKj8JA
z?^L6W7wT+!OLg&^0l2IrZ5Gv64S3Z|tkiB3dr*>OCgQxkP@_&>)t$%gp!}xo2X=aW
zB+%e}mL}%~FgJWt5XPN`ih5297k`RCY0ItuB#u4+WdF4@1+e&|iw}?5KK%Luh+HU|
ztcb}%WKd0-%_|b+@LAV2t9WEFG#Yy$grbLF(sSDg!Gg#PzN#+{G{u
zy*{bV>`~8v5|6)~eyPEVgs8YLXR;A{DD1*8Y3xxDm(?Oxup5BBUTi(wcY+jEIfh?8
z+HeX~-1+hOH=6)r%qBf_U0E5Z&Hon={nHv;R!)w-00|0~g8xc+@6w<{$r3)X(yYKd
z;MVhRQ5O-E>S%LwvO<}gQ_=M7w}7X#kV|)KF!5s%jiqW!LBo7+>@S#ffnU)(nIZiz
z(5FjPF(9
zs?h-Bc(n0~r~7w7C$f=b%gP_6
zVAHqokQqcsyzX6rgw8u;EIgNDoU#Ts%y*7NoMXW8`AyR`@nl$(z(}3-Z3tYI@e)Yk
zn?b82s`8Yn%fPSiQ$+WNU1ZFBEd45zFlN6;jZQjOgERXS+-?I=KqElXx@VyWdCs|R
zt?2#+3>v}XE3c`rO%Bb|Co*j)+x6))53?R%%A~a8eK3Sr$FF?W6(0i=efb8vpBUhH
z?)s*sMLBr;q@3RLS||F=uCOB{ISkT{vx`gee@9<^4#f?sAH$ao%!L=2Pvg^aLy6Oz
z=V7pWv7@<@D)uXWcT2R!8rHOQOC9*!fJZGtQgZUnVD5?1Z2FgHA!&X3FOqUmY%!x0
z(YPs#FKkQ*X!hKO(}f*>M*D~}-C(LHt?p^;y8N-nU|Iq)#v97?vY6mMf8tNPv$BE|
zPm-j>_etPQN%3ESyOj7|yQK0{JxaJ9(UnAfNEC)VdRnZ0!X8&gcYNr-#ez#Oe_yWh
zBhFWQ+TDj8&Jqb_3x~uX0ABd}{XYA=4W!wrxq4T~6#DCD=odQO#&VUzhi{iWf-&AT
zPhE0L;I+I`)L}>LfwzY_7j`|M`kL>k-<%>=F)lIuJ#3AGxD9vuh&^-L6>Z{+L^RGmNFxf03R`fPo72OWb;$kkwMQIE<$a`tju_`{66~>=ah>a*>BM&(%$Xl5wPWgvS7o-&%~;dDPn*l_Ii5AjGt-ys!yP0M;tzJa@OvfeOH!Uh
z{yEe|qScLnsatZI@5=V!rb}@K!&5!DH|^=G81^oF%HVs#M+J;IOk|E%NOWM4KWcTf
zdA&Ga%%(f&OE+%f%*MW-iTcl>GgNi72yco1bpF{)AZMXtVUOV}z0?&LAl1^oLfcJkns*XRt23M~oePTJ@
zfjb{wni{2S!rLl?$=hz(c&Vnkz2T+~3}QP|Y4^zo60xlQ<6SM-#tUB0&ac6T`3+QC
zv>)M9)$wTo`d-kP`E2Qw>kVT6p*)`k>wWwPHK)uxHpa_BGdEWiOrUxM-M_utm+({X
zypK}fWU=ekg&RF}!T5`(*(&1+bJ%&knuFtsG0ZNHo}nGo$JERg3GYe-vDOumSqDpd
zm}2|2sxsIGQm!Mek6fdJf1b(I!1A5T&ZRtP^H1e@rECt
zX3LAb$SMu5P|$1gc-Dcmq3d+Mg2cJS5mLu((ZJvoxe^+pe~=egk=oOVK)ogFwCRV;
zxJ%NXXDx3N#7LKr&zX@zMfN97e!jABt1s*w!Sn=0GQYhUs-K~;8j$$(^6OZ59LbZx$a;b3!92O+k
zzU;dJ(%$LO`47B6vDS0DSBmpci!o|{H}wKAhkd%k=Ryx{z(#;3|0?)2cHK)eYX$|_
zT&caKJPb&Gx#uQdY6h=O`zWfq$)L4v(hg2PAN@Ql-SqI-0FmDq{vA)<
zj)LL?#`pDqqa|0X_mj`&(BysRt&5jl0K2lVlzNRCWdGUpL;K5WBvIep!cV@9Y)o|b
z=*V9Hq26L=^1&LQtMX`S!KxlvsCYko+f@VNp7N;p^$vqbjoH-W*F%vrm3V3*VT;)PB<=|gdbFg-G9r`zUy3C0=2xNA1wUs*h0oAYHcMsb}
zqRXRY6jhQvDD{L?Psw5d3SMww(2OT?4F2bwiq8ilPKNjBv8^{c?{>QUL{BM_;&SL>
zEKUL6atQx*hsc5Qv&UbA&pD%OI(b&44sIZaYJ}5%r~%AhLMQnuB0y6N`OjyK0q6qP
zcIUj92QW6#ypjKEO=11e9#x}y7vTwg!P&`Sa>V8`8YmNf60F_REi9c>L@v+pd%n{u
zh^(P8CH%r1K~y$%_>uNH!RTz%FwK_(1;Z@v0T1p|=+$($(C$+ML_eTr+QA~1pAP?IpVX%)It=}9~h+8RrT8;Y&
zx0(tgS?PBO>k6H3bpCJxUcog5hPoz&=}rY^X}uAG*44gW5}l(85|WZPhyM{2o=2Z>
zn^l)YfuzaDQ;s^I&cTC^ce6Y|*uzrRL}HF4OA{>bc%KFCM%b1sQ-=bzC+6vDng*zL
zhmi20#T5LriO>EqV~+|sL=q@Of&js{j)YS`54^f&W%W~90?p^m*e`|gqM$Xql9BI9
zz-|zYh9A0#h&)_J__QMkUZn2yY!*TxFVz0b5dCUFM24$hQ5uO0T;VO2vIf4{%BR1^
zM55*SvQMdJq|v=}NBu<(31T0hg@pebDRi;%=z@PCkvG27+@{AJ1{Cjv)Jk;bfV5u$
zz41$(sQ2jDno8Cv@SI7S9Ps*y9$PUung)^K>&3VI!>^_ScA9vR-@zrQM&|Q5@6`sd
zb-;er%ya-%6c0zq#qOf$J7ra}dw%G$XDyBTZX!_Fmd=_WDL{AU%hHZS4S-7e$b;Ko
zDnU)a*j5R;gA&jC<@f690=%FqzG@vn-0Mlbh&5_J_MBck?j*#X5F?;!XW|d+Kbjv?
zyz7G=x6S2;AN2v(f+d$rlb(Vq)~wg@yzkM@%M<4y&H6nMif`%
z-2SUys0N)T4s87XGlt~f&sj4nZlQXWEyK&=f58nk4@%!EV$QIxQZCDs6PM&s9aBH)Sf`
z5B`n<{revDcYVH~w=Bnb7hfHP=@EhP!cH&IoItXf!u&MaIyH2X(qb4Gb?cLNP9ns(
zw0uYE+8)tsFtckCaTW$tyvUPKGJ~HrKJ7o9aK@K5b)`OYd&47!w}vh|+{U51bRkcp
z^0E8N39b549{fAUs6%J=2sHCJ_cBT40{$8A`Tp&?2V6{Sl(0{`hkq^Knyx)5fHfzo
zSB}KV!_%Ki>fM60@ViarKPG2vh&;l%0`k+2P))l!{ZctEye?=FLO~;iHShOxrrlM7
zvkNprKk2M6&%JEFi`!oTVY);y>8vPZzj9*yF~>3dD^65unesT++v{p)ysHcASgUv#
z9PDsF26a^zSs}J;q)iP!R|ACv+?rf8sYNJKHp>M0f
zoc#?yTsps`U8#kkrcl7S@Yyb?neA)(#Ti3US-}vA_Vdti<-N3VkRPsdwYS!%&4hQ9
z^X%RhVkmAjrE9I&3Af%DXLjp3;zk+$Cl|JE!&rA|_nfFW%vTVn5F_~vnmVgfn3}Z^
z4e3J`GQt-h)PHtyFhfwI3yH;Dom4@#ueN%b0=?$N=Wiph$AsX0A6>^@<
zb;CO^4&`1s-VM1JJCBv?^up89R7dDyx{2J-iSAhz0#vfP|KEAaPPp*tHPyj#H(dQn
zMiP3X8`@CV`8RlXL;h#dbeg@LFw;wjRxr8?){0$>pOx=~e2zq_jL7pwC8ZErCJ|IQ
zdcKv3c<&q-Gn^5lLa^qs_B%yO1bGr7zvn0tAXs#yI^gbr;;&q|DI!GiK@00sy0LmV
zo>JHBXx0h;n=G`GZ)t<#l;&Z~?H#bZH|21_(;iHJZsm)FN-tibElX2x?#1P+BFnPz
zy;#tpGgGsq7mJ3(=~z7M!eXQj{S@Ne_!4RHz1W>@jQ&elXrAlF;$0Crd^X+K0Pk`>
z7i!0Q|75oU-*jMo$2rn88H7bo57Ap4#W~l_p1!lHjX@@c@e_Vq#kIU=bEU3dj$4}@5E;r(o=ckKjOC-t>xA@bX}d5W5>}hJV{AAetf?ZPxd|PbYE-3v)ff>&ll_Qjq`fXAp^$$3@h`a-7)q*BgCnB
zw+)9pWSKg*RE|SB=$(hoWMU_s+$BMy7JUCa*GTmHCLE|>nTonfFc0I8X3neO*y)-`
zSQ3$Y`M%q
1A|T?$;p={G8gIpL?72Hz)Rwo4=l8oajfo*z%f#M7HF@_k`BzfcP%
z;i$U)hDck#cp-`z&l#?^81@;=*j9LnII5XAzq-xsjW&I(${=
zq7{Bk61v;r5WiUt{FMcS@|*qzE0Ujrrpk`Ot%hZ$*kk;#>9ym>h^zegs*h*&iuF5C
zo_Ly-^U(nq5A4$@cu9)mA5G261qor&s9f4Fk%HKL$6779{4iA1=RbSR=qcc$e{9G5
z`Vjt*w3nK0CV^`>W7Zv*_R+WMy7)Zizkp)SpKm>P6@56ommR9th{*I)kNj`jc&I)ITX$0#w?D{n?!{|`uS^7iIf>l<+A?muPD6Mg7tI_p}-
zRuXb4e@r)~NQxcq%N-{)_5xw2M!6<z+Ka*afIG`go5RhC}48~GpidH%9A-R>IHN9LK;Aq;*9}pD`IP%E$Gf_^v5#h$bkm{3XbxAgV4|8Lm-3bYdWUbfMN3T*-0Ees{$HMKpENO*UOb
zhrg+>e{~ZF&Q;H|UCKRyF{jpGfr=`6+@i?+FK$=CF7fft{}h!#t=~nRoBOwcP)3@_
zzbkqutMqGwj+zKEx_sa8zs7UmjOSH54sT)fJ>Nj+=J{ZPSa#^7%ga%MZKzbRjlw&H
z+cyp*sMfy|977b2qeTHE?3+27IGso^(7sYJF%_>+!<}(oX)~YjA~dYqm2pymDcZut
zQfHU&#^W*5n(GsSvDJ7pmq(F;%#F~l9EVCm_f|eHscnaX(XK@1%Yb_Q)pJ5%^Y`}Dsk_o>z)|Y)l$jDZ
zdyPRjW$hLa)?cIWRoqlyqsG1&1KWhHkHet>s$}3fTaSD51rg-@gO+6bgeq_=fApBA
zXMvzf^)J_0{Uzb##>&~klZy(}>-nA3F+!S4iTyEsoy#h(h@
z4)4qVr|7)nss8>rj;#15RH$qrqlifOoI__k-~Ih{|GW3x^Z0xoXT0C9=QE38n-qHaa4vF*9SrW|ci8zw5tIiy?O&K!
zBD$D4hiGRvNqHT=@i*4=V3@{d_`Qn&>XVJX$K9ZSI=)7jgeMw;%nN_JYiV53zJXWc
z{iHY8T#=taqFji6vtn0nUpGq3-ShE}h#?vgOX*uVWdz!`+y}DM>`+DEfB$y=d;-o%
z*;QqPEq!8%0d4S^emycQ+ZvUAyFo`a6phYyl9|#zawz2RKDy)CXTj{>ab5j5
z6ZAAFV%+kwJIGIW*+EnN82ri>n7aK3p(n4Ng@+H;f+ac1fr+d{`?4Ku)TQRp$C-Ur
zqFKC9j_aKW=nMuYHBujDG3FxYN%jxxdbQxpfT%!6`3R~$*=;8R=g>mElGC*#hM@P(
ziFiiGyC`6Mf9iD^eK2BV7%i-bKzHu@JI*)8QP6cCLuJw(iOM$KNIww?glbb0O`_sZ
zm~J%H17>41*A~?;$vpvPMozAL-r)jDgYU@)PJNc4ia6ZMe-qZtpPS%=ykfY4rnO=4Mj?eH!>!tK!{N
z&b|0BZ_@2Dw`p|lk#yZ((*+Q|KGCBwwvIBmKR>&Cn;N!U9hY$&k%ilLZ<^>__JGDz
zXFB(bd*Nl@zp}gS0-=VnoW`Jj3U+ul$(HOBfjjp;&$qc{1b=&dKg}~?jBkPDkN0=j
z!{r_IESx8AVE&iqjd$(}#mU;+!RVVR4*7L~$!6zi$h-T8=7C5Z-0!5P{>jAwb}}Ts
z_b$Bx#rnAD*wQ7Sm&YONn`zQ`&mW$YbGCADZ#dt}%f~J7dr~~GSW6NnD^-((@Fj^DBXYIXYdIUYEC=X
z>)fw~utqd+UfLR#o8KF78h6M2T+Tx>Ep|B5sLtT++gDJQU3%lyN)~M4d58Pdlqr2y
zZ_=bo+#%{d)#Ci%JPuWCj8~I?gr|b9M++8u;lVR5XC(ZsF~?z726LVpSgUaB`x>S1
zxHGxdb!5vJ%KS^Rtv&38d5d}(&1}ozOpABpVA*VKq<6EW?|1aZ!+6V^D)B<`NzW(FJa?%
zYmQ%?PvM`{0e!khA(;1Tdaa`TYg|E>vMdGP;(v+VSKc-!!7b)b2fF8@pa5T0LY$BR
z-gF(?sq-=#hlK|}ym28B>#gKXZ(7}fXc{d|~5GppJ6i6L}M++x}pY7P0~Rg(Tz=HqIG
zItV_rt6mAp%a!?&%+8R9KjjfqW%m^
z&3t`%UetwBZ)uANRrY}EQwC-n-_vo_Wz0WWU;iq(RS@Mg-R^7E=T@PLDGuZ8Owd^=MuyLd?pgWKz;9eM1bc`tLA
z88bWFcCfc~Gcm;?LzjdW8xLc+IOERcOF+@Icl@H9)>v=vS1;w*%lMz4$?)Xi72q%w
zUi{|VDF_ZbT7(obz=#)zK3%!$1^<&2j^1`S2G{>t&3uud@WlhE`HIWVz}cbnvqz80
z!9u^o;OE)XP{m660v)S5KBf1uj4Pf7FWjeR%Ze9;9N7%+|3tZP>{;fA=Go`5_}#$p
zJ15w1)Zu%DbaVocPrY?{_t!nRX(<0e<1jTYcmO->-Bsa**DN)Rv7+#Wm_>1D-ad*S
zNHXjM{Te7)su1E{V1U`h-2b9E=77)}As;bP2sfEe=ki^pg6}T#)t9cXgHWmbVA*wA
z{PWR&cZZba;3d6y1|drRa#|u^lX-mxF~mjp1ifX1yma!lE%db5e42Rtag+};-lkz2
zyvl~PXdmclbT5NV%Y2Ckuj`Rk)6++3T$8ATKY1i#lo|_Ee(1l{PVxCTE+;&^Sq^?m
zmNiMOe?nGbJ8P+md%(T0M`yOKGD9A2jU_S7eniLC9i33|3HAE^Ei`)D3uHTT_)C?_
zP~PU?S<_2Qc>VESPM%+5C?Mufi@?DFV5aE$Pe-B+ow-rmL|s^eRDC}wybWZ+Q;qjR
z-v6FND-+4@qY}P?tvhqfb%M=E$z;FZU#n6SnDaNFEAtO>9vULzxQjr-SAO1S%Y|Un
zhj?w&o{xM5$E5rpC4xI`2IqWfzJkAf(l505%fO$?ex~KG8K~N&=+)ui7_jRm8=Dr#
zeK0w;sM+e@fT*qxcc*S~CZ$bY@s7~N2F>tRGNgb_c
zZy*c6&|5y+gYkB#?6kCr8ryx8+eH)KxIYXe@_APG$mF1B)ltQ6_HO}iX$Q~ooo*^c^yd!cxt&4#wuAWcFEWGM5fk?i0}VK};d|P?o(p-|F&5F*=_58;
z_MJ+1m;vj7hkHJj{Rb{TjF{oQ%7n0RM$xldYs5a0wS}f*I$%XKz6y)&213HT`u#t5
zp*CPkH}QxT@aOM8>s$1NG#hj@RG~YKGAraP!k-ACF2Glid+$GR+FPM;;?Ncu`f@xb
zuZ!XzochI|SmaMUHqXAn>KQxSZuC+J2gho2soK8^zI;t
z-&PI5<~Bq#J!e>+nFINvI)}-x)r0sE5#GHjSWGC*O5Ai;G$iYfcuy>t$dk_BlSMB+
z)Fo^?({e%=?vj_;Yed@CGRZmWthS|`^Tg3V3lh#|0t9=Uy5yL+8d>t+pq5wnZ6e};
zkBZO7B;tz1;gvnyr^#DN@+GXrVkASNfEvS!20`Ux6U^e{LEbh$r1YZBo}h`$Hp(zr
zBvNDOmgmNn$i1Bo$Bx%+6H!v!{*K4@q3oAo-9@L5qPFu?L`S(j*%-=sv?d^vbgF)O
ze*bJ4fhWi2ytD#IRkgIu{42Xq?B-9GGXcfq`c5jbc$YF_T#Ge)z_N~HxwebXVDc03
zRp4BiYWXw)5%q|FNj4$Xx6YAemQT{*!*SY46@-~k;BJ||Zt~knB=b^uh@8#+LgefS
zBP|Ryh00!~5bxJiH#C~wlGooZ-`|zlMYLx-uj-37ky*NdGV(t1XkofbFWgBLLvQpfno`!&OJ%wPR*hY3e?>H~xffa`+wDVr((Je0hzqJdwowPlXBP(k-``ThSwA
zb9Fw_;R?9K#pNxLKS+=$;r{8#O)~i9xq*9scY(Rh(pzaR;>bkVi>4WBpfYi(nnMP^
zNZm&9zTTH>M8f=x%;F_R6wAidY@Q_wxXoRqn9j?9YDOPlx}_h)zYf~hMlTk~^wOQ%
z$5d%RwzY4C-Lwdb984rH$qRttKeM`6>N?2Fa41d?h(v>jkGqylCV@D~O?kLK4~f3z
zs@bDAikfZ?(Mp(Vqp}K5H6EwEfY;cunh?=IA%Z1pt7jvCx>~hJkZKvqu^MZ>>$3x7
z2){Tl?s6VHjEa3ooV7rnAyOiwcmdf_!~vP(BC5f544uVGkcUE
z&rIEzZ=wM}X}i_+VS5%32|H{SCOL_0cHAsUFk3)Aj^~zAq~3yq8@0Euk7S|CC%gB?
z`4xcMw&{tLu?4`q#b5a2r*BAVSU`G>j?$m1W)@t-oeowkO3x%e%SGhEkAW2HbV@(W
zMepe+!ys72O|Xf$2|P#aEWTMLqLm{Pd$@^g(5hp}WRu;9*!;%fN&ZO`FWA#NpGpH+
zrmOw6)px+Mlqjn!95nDw^MMVhcQRb1#0A-`t?a#{s9-S##y-&G`Xvn0?NS#`D68NA`x$2puw6+BR@FN0UFo=?A%a1|aX
zp{0&{Z4Hlnu>9TjZvkkk7Ujq6R)lok^FPYw74fX%u2a4x0IPe-GldDNz~DP^#!rTJ
z;{V)v?#~WLVxLg9v(uDE*^Vbg<>=2fMEYFz=o>kWWdlqnXodM;JYlB=tvAr24964!
zB#W2wCEsy8SA;iguT(u|1h8a(QM~nrCpJ&BSW%z#$M62ED9JTFz;i_zv8@@NxS#X=
z!nOOZ`0-gsipudaY`9lzk(+q|4yMnyJ`^{=XER5cdhwby)y^!
z{`(OD$74D$D-wiL#ppxDJo5A43vKwU$a>?`?=$dSmG72O?rD5HPi#Ij{T!kk55JIq;YWjMTeiNjq~4hpc)|9*IAJCr8zP
zWoO7@NsIed|NLF?-s0Omt)f
zaUDAYskJ842MbAP9JkY?U
zqJU?TeZnQj&SOh$qtJ;V4ID3SStw6~W%g;Cbhba1Hx9^XG^L7H(xnrWv6
zW*u3`uw^n`63Gcm+B2iJ6Ar<#Pxp>2Bp-v^&dZl+%xG_wBEXqU?Mm{-uQsF)$Y~JI4
zN86AaQ{~mJehHL?aL$~Ou3ADcZ5Vpvk3@TWUQHm&{k
z!7nr(3)Z<4h`s)Kq3?%($jVx#ZAyY3M)`Ez>Zv(|9ZIcC4&^Nacem7)cE}rdJUFi}MJtXo>cuW(BE{P=(AEYj#hxJvt0t5vvvayY
z`4#e~?ak)Cw-XzeafpAl{E1{nYi0}F`@sjl9!cgst?04r+V5Jgaun$}AR_NHg1)}}
z#9*6L2X_2pdBzH}K>JOaQ!?l+GDzC(rgEfgd7r0)f6@t07G_?d
zDZ>oBvM(C!Y@9q~rr`mCuYN&>^9^X1mmi0!UJgq9vk+9H5(nJ(>cncudn5CACeNQBhd~0&(8G}BVqL4F1N$C?;y+@`Q>l&U36hcCjG*NN5G9Mf&R&{BBZ0|
zX|#V|KJf9(jO#JCr|@H(1uq19qd$-D#tTnV@|boT3tZ>Y!Rt_+r8P1fY~||C(#yJ|
zHN#^^j)htPId&yY4ns3=>=Dh@6zK>O9@4OEM|hzB_^&&yGOGeQvC(Zqk^sLqveRei
z^-ys5{`S5Je>D7^$>+Pf9eA3z4Gu8KQR)`9vW0nP&;vu;3xPFfz?g<}qn2$ra8ip2
zrM-9#s0Gzsa~G%F&j#@P2{@n)#5^K0-Bv}=Nu96z@TV})c;U00=PebG5c1Macj5w~
zgyEISB{e{I#^cQUAC99CJIQPJ3XH+&_rtwbA!pDVU#q7t(pm|8=zvjj=w4tb>i1}P
zj31>FbxV4m4Uqi;Y4d{PGkHPZ>b(x!3nDpR^c;)bjjEfY-yV{s0xa=o_DeTt0nRr)
z`E!F@;L{=vyCn}B8pGcLKGx8JhpbQPg_3@d&yRAc9oasKF8{E4y}BfZGHkEbo$UJ$
zz{w+gEaHpgFfBm8D7-k;U|ECuP-;b_%edn3$@=qE3cA|EtCJ=
zt^P@_PnaDjiW5RFdUtONZ`YG1Q6E=_j}H@PDm_@79`}=C6AE1`Ce;)lrS)UIenwDy
zeAa1q$T*QZ)Hpq8&`mz_w|)_HyN-yXTl&D~o<&@)ZM-CwwMe!#GEszMS1s-g#qonc!Gee$KjO)A3$`Ja5xH;B0l5OwrAqRt?-md|a_V!S7L36Gv7T
zeE7Cp({IQbHq$IkY*Gc=_OzK1g$r@*dxF<$y6yIvY2^>ppo}4TL`y|P<*W+a61E=@
zhqMFUEKNTq&nZY0NpB<(mbaZXk1HUeYLZq<{wR9z?UCne(s(A7?xskIc!vhXEu=3b8ddp
zxRprsmFabymdzqHgHt!&b@mdKA%{AtvZu(^e=&jQr;j39{@+LY1T=y7+Rpd9hfR>`
zLF(SkA~$e$%;_(6;uEm4(@z
zHuxOWq|HUsMlA47$Eg?7qp^d2riRx!!B>@4p66d>P_Jdoy2h*l(h{blAEN(B^4|>4
zSbakUd~#TC=hmVA#G3dRB$XX`T-*)~k-OT`g
zMX+gB^BqNUze%HnawX7tJ*R9O#sg0sce@IyRFvWh61{f51=SV<*Gm{l6sWrOH7e{U
zn&B)IQrO
zkaHQtUCWS-?Q%j~zf)?X>}~&*0l)f-`|E3?
zL3gY3JM)rBbYnFPzMdWezx%&klt`fP4t>I|zj(a~{@iSh%h}Hl1Bes3r@xc{7xAVG
z+S;{f%~M{(?sgpb&ezI+Hg*&>+K5fQYaIbvZW$Bj7*c@3>Hqe=BLY#fZc*y#j5k2S
z-Tuq2sW|k<=_VKb_hA&X?9bSl{Svi6lPPiDXFxgQt`OVNG?d3X!(n7r4w_yv0DG5y
zKpT2t+Guwg$TMoX0;pf3OFc(o-6viGl^%v(%iw0DEZX+=?u$?8sjik3Jz#~$J+8IO
zRB*$aT}qL=MNi;7*Y%iKJsHR^^63F@?}yZ)SGUOFHl)c8>{7q{MmYY^%6iBu$dePd_HX)w{Dg3`%sf3#
zot~rjzR?cae|@pNIo|_b2X{}EnpC4ougn9Q>f;icteZgf
zP<9grD7zW?1mraSN*
zNm)H6_jiiHyE|4GH{MA?9bO~nTxWVbZGpQ?wghpkO!oGBr9C)=N@o|7@g|yY<6jRn
zXT(QE9yWugY&hilj6wCA8PuCq%TaH&2A=JEy&gR|hdwUScKRKhNBln2(I@W?fh$@s
zFB?*L5QF2nY=fHJz(x12F+ayVsLReVZM?DsUMxJ(3=#MUn99|}$j*MGZnHA4x$84J
znB4lJi}gE-(=t0MC_I5w$C91+rGBA3VBwfbyFV(t#6K5#{41!O3HUdz&4o2|@;;7o
zazUR@Non$Jdtu|LzJoKkN6&99KN~}>?1%*4yuiSr44NruM>6ddfqx`PQ
zD0YTPWKb_KO6`e|_8aHy2Zz4^tLJ=~i9bis9Z?_mug^ZBUs1*yVuc;R>dCKNOdlJ8
zU}#RuKfVD}JsG8N#k~#uTHYxVY|;a)(jPn}PZXi7BfHl$J|OhpjjBqR{0&qW!lAtJ
z_dvtq+L1&`9yd~!OpuqZL}dv=k~#6YpySjhr2CN?D}9=KS|ikg`ahEsTFv!wBVDWzVMz
ziVh$%=)S|zq&wirT6Rk?^BZ8`#H3{8l?(i};7P9zN`LpDer_ech^l_-KiJs)5QLR;
z__`{lfk){tQ~KFFK~$pW*3IEtNV?%>UGniOAjdj#P}ocZB$QGo7eDku340V0e;c`?
zumk#17k}!5D29vnz0JI6=K?2l<-9stk@xv?{hAuc2w?K^uD1uSaF{;|Zn={r2AwH$uD(_O#V&1QKyh*D&x_1+54FlBHqqr$LCn?FX}-F=B;RzEwwo5jkw!75G;`4QPnM`;&PSwYql^v2n^7ePdgUUFv%{{eM=x+&e)6(eH1Ji8s@WBs$u15!m)DhFNY1u+DTJueq
zdiM?DQqnHxVbQmQ!smiwwS%9Dhnm4G;>yK@Z}C;0it-floMJgcCu1aGn%`7*d-)dO
zBu|_1Bt3dFl?B(83xnV{|dzx}aU3VdEzV>*Xl#e2<-q`Pb
zXdOf7{b`17Z?2Hcm7SxX8k9-#H$oK&Iadjr_$sF58>Ix@UqiGK<4KBL`ur^Arvu6M
z{?{~EyH4yl*QoM#&lSqK!TK{Yt5E(sW$@?P?2Iz!vPJr0?^?~ao1MSW&O?OsmygEH
zLvlpkp~S{S&+p1P`eu%;E9Esuy#=&b6B?CSRIg4=n9eB+*k^u?UKb1DjOst=+mc8D;ybH%)75{*DIhVfK!%na;dEeuHJC
zCMRpZU7^XA@>YE66jj1`qAy@vIH$pYoTB3xa|YH#{=moF;P1Yq=%*i9T8Bc&Q^V{E
zyWC}o>nANDr~QwUn%QI56;c6l!WDn8zBQ(cU
zT$bDlcxWxGE=`_O=^QJfoS#&)1IKy1D{Jh^>XZB2mumiQy0iqCN)s`9oOAVqy5xjq
z^#i%^OcLZoN9#X+OH7rN>^&7!My^_3UYknkBO3M<+`J@FPN)H9m
zTg8(a91Bm9DRm$7A6vTGz?i%}z6Y{4wsH#U+A2PSvrg
zW1D_-^nU>7(qvzIA&wUR`4vXga3bZqa?~P1K?E>gcxj>0MLv$=IY)R65iuLn
zotKZVkgc9$q~i`jByDq%kxI0H0~T1H__-#u<9XE%Fr4mrU<&31q6A+DRs9frvj^5sW*`P@{&+
zp$R4rApH4H#(mK^pi^feIioF%JWk!jox8+=_@e#$qfy7vyc&r;9*Y8vy?Qwl%U6Nj
zuc=?$<#R-f+uileo&TZh?5vhTr3ZmM{k~Tjelke0__%+{Z6lP@JoaeIs)A7SvsX_T
z*&s%o&l%pfXF(o**896OkAU#`
zZ|U0wX=L`ho54Hy5K>%jESmOC15s}^wuLUGpq&)XbFW7dxaPf@((6=@?wwhB(pTS$
zoQ6_Py|KCsEc%sbMS0bMrOFnQslFTN$f$U=)M*VC_}CunjAo(sM?W2_R4yUv;yq(a
zdkjF==%F{~_FqQo-y1Flv3Y}p6Ah7HCZ2&9!#m8`2aG`RgPq~VG`i?d%C3r{c|D*$
zcyubAvR*$(p~>8O@EE;T6FO$Ed>RGnW|Nyqy5OA|G#nogN1a2@`M#`s0uzp|dxpXR
z;JAYPXG`mLq#MH5*c6lpyrdZRdcOMw`b#{YwS4@AN|Swl;?{3yj)Nx9jl&@{pa)UOs~3zH1_8OZ4Lly)kw92@g8dv-4l?k_
zj#WLL1{w>M6nm;(0o^)I{b5IcWJUdzdG}#GFm#$Zm3B1+oTxhDG|Ur=Zi&vYs{FnN
zCTMS@N6-X;V;2+(BaSDbV3t+h+7TB-h-@_1pN#@mp$7UIb?Hbjtk?Na^dN=1hG%Xi
z{{k5aGRZ3~RPd_jX#LqI+dy;EYmc_WI#4SzuMxEELG{{wXOl-;(W`+MuH7SfCy+Rr*WdDkPs
z;yY)4SzJU8H)$Elu4V%b$;thfRI9;pzW4j;rz${;x=D-beo)@
zHs%jH#ZkYyt}u&6sZAMo1oQxAB_6wb*INOdhG)tr%jc-z$?ksEf=(26n?C0!dpEMF
z{8}g!QIGDcw96N;R3Kaw`!2HmAvo#I|GO4efWTwBI#UI!!A(2c9zE6)z+Y#+>Pwpq
zf|_KiRe0RdynM~%`G*;3;u-zll!i=XVs_9cQ!o@%+@(ov2=)X2S@b4(Z<(X&QK>Q;
z<9IZ%V!tr*DH#dr@fR>BOr!J1P44ZISwa#W($=@lf1~12wx-W&Mc|CVv1=cEKLg2c
zo;w&0bfH_ewW2p;yHPg{vwsvz5;_{l_|eU!3?yD=G$g;j0`21rUfEPtKxMBF?RT15
z&~tUoaW%IPd}1=L7_aw6VhvX=Nn}4m=f2wEH*}e(n|@l0x;!71E&CK6eVT>lKb%aI
zBk!W(WUi-Tyul!bn}OSX{yB;e+$EVnoq)`ZS1yH0M}jm1_tDz-Vc?3He9%&gE26II
zJt=XBvcLc1Hq0xX0;c3u)ZVH0AXCTNEyZ+cC@6o>T|v7XaFQYuXMHS{zHth8^CqJH4gE79er3n6@9Bxi(x3A*{Q+-K={RtB$G2$EK_d`#
zEc!CKpk47+$>BbdlYDWWKISo~HCQ+=d+9t_RV%ak@!=85^1U7=uWpH0Jk(G06}W(x
zcW*zf`|gDBh~Y%{QW(nrxN-l>V^;vh?QH~q`hct;iCy6>C%~_Z9JHcCRUkXocPWH1
z4cJQD@pN?bM0=lJ8`+=d3Mk1^#)=!)!3LKj-|>Jr&`N7~eUwHOToMp?d0O5A-RIOg
zU?^k&PF%ZuRekXkD!(04@;K-L`Z%!K8!2RiE-ugLw_iC8dJ9yRC11#)#3PSf`viGW
zRnT)98*evo@%a0xW6hT+=cGIBk)qP5a%?*Y>j(huN>$eESBHT0j|iF_?}X6j_I;z@
zcrDO^&$|evepZyMb)eCObszA;YBWT(45E{ioZ|_10IlDS8cv!i**^}JE97J^yyfGUd{L_6N><72(gj=0KC_N?z>1j0xCdqg}pNbwUQ^##lR
zWI6x43;COSP*u^Uj#tVQA|)*OKcI`D#+{XxMw<+*qQ?)12NR
zG@sa|YP^^k{+TEb+M&7E@f#sutNe4)w2V}|$>Ms{B9b`LICp`XEr)QkDA#UL_(U|z
z=c-BYXA|#lob!E8346!xXXf#F^@NaZR+z7fupsx?9n}pNcuR_##0ljjZkY{?I~kikmj0l|16?3uK76w&d!gwXSDWgIlf+;a
zjiR2Z4N?8Y{Ft>F<@eWej-_6FOpYt6uvB19^
z1lT9Fm+?|CnY5#GXca4vAUSE&z{!mGDF1Cca?P5oyLwA@smhZiW`#?)hMdT!!*(X;
zsLv6W6mts)mn?0^RLD-LTt?D&|bpt)kQ@yRtUmYEI$Zh@Th<__CW)~eZhS1XctboR`p&x8R6r<78~qSim~^A^n-g_Z?Mzi^6pbCKk-#My}jNSKj9w6
zk9Nd|5=?z#kHBria=efjuWq3H64!3a4R5r6!RlV$S4Tbiai{eDVd+0L*yd*D@Ku*`
zycIIR(%6%Mt=?B047=F`yWbXd7HjI@WBmc#mJ2rUUUS^vl1rJmfo7U#*H8fd*?#*;
z67@Iuvze8)HliD{Jo<3aaJCRa)$jkRMiZcELSU#@gE!>Z)zUT}*9Gl8`XY2Lx54*f
zhOYp*L18R7xrn?axuyLq^Iq^~(Z2cO$j)K2Hh4lrdw2`kcM&#N4
zn$j-l5ozj^BHIq#n?pi&c(p@~hJblRn_`&yo^#8wy#WsX7~tl=(*gA@I-{#4+oA4S
z#T72=b|@?)WJ?r(#2ysY4^L$;rv3cD1O94VMWlPbZ(j?%@|0TA{y)k%
zm9h=wSz4e|=7D|RR_mb7{oS%B;t*VOnsB|_hhUxRha_p$Hu(1OtHg?^h!@Y9kaprH0_ikRd}qegtp*
zQng~UBH?Rx6-MPK5@tM#Pw0M*;N$Q|pBz3RI3P;HaMO?CyBXQG7kNVAt>A~Z1#}Q}
z6ga--MHxpjyFv>U<@*ICy;b+!M=(4%_3q1J4Bsi1z0^%4;bA||^atBDuzR2PzHI{v
zPrkTRmTv+K719onYAWf4+QA^NQ>d7X;TKQ~MH0LOXVn_7qPIJQPb6I(wAD
zpBE`uDYQb+*B|OtcOuy9Qu6HDM+!$C-{P^bt%lFlwQ2osA$aY<$s^_nK`Hv)K+#!>
zzi0eY6H{3YJd&c*HXKt4?Q4{D1w#=$^-r)_(H2351JkNwl=&EEIDfN3sv34(aCSd9
zRSqv??dH2ljbZOA#$B&ZV>sryIQygw!3zm0!g`$5P{sY;nY-sJ;6C3Y-1MLZw(oU_
zUmC1}LfYz8%MBG!hrv=nFSrc)94YADL0<|ZKAw+rh^mHeO%_G^0hMsVMZaj{W*OXg
zUfX^5Knd)VDL)bTs|bz>^PJ)usD$0(=Y3OT&4o~Ypp(i+J_GLX
z+E`QE5d(d=@ygvxUT`7Nz>0gS2xkAIVG+{EhllmS&Y3gm@W&yAA(fwz@S)fN2l`Gs
z_(*`ED_AfeZXS9`=eIitzBC><=KuW_!
zNsla8n)ow+((ENF!cocj3pc??V;%>UXKYLmJw6aqhmmj!!Hu9-rFy7@LQ=
zctop;c`NbUy@D0%iwN&l>14Gib0-Eu)u{Nt#^CtX_$zrS5yAh{&Y>)8=QTRxl7oQaRx8OqBcZVBxw_@sf
zMt76hW_)PW#9EBI8J_^DzfTP}<0A*DN(|1n;f@!p6~^Z}@Lt>NzI}!5*a7cl=Gkn+
zaa!D?59it_{dfD+idH&sf69L=b$32sw?C^wJ7~M`ve~+x%GD3p`|0a9w0}GB()7=U
zORSyv`QDI0Bf4(f`AV;#yQv$?tVwb<7j@$Uh-!UHrx82-w`=;@i$<*QwI}oZWFvm!
zEU`XJ;R6WptZ$@cG~ioIyc)jD%~+K9y|MRZGtO>Ha|F)Kcq-4tRwld&BO)>RaaAMs
z0LN0emfA2o|5n^{#dcg_rtpEbuMIB-oG!PqZN;avM%J&(wqVY~Vgmmt<0RyHtLwoB
zyvR{@mxdy^<<3^O`Zm{r#Wijj)S0*AW8bvZ7)!dbXJ7pp3x{qTC?;vjcCQQXc(}J!
z_E0BY__j+zPUs!pe{$r^sn-pd)_|7QD4-5IMf}t9maE2xJVjKqdWvy2vu}~bs{%@Y
zU~9pd%0?{YAz6Vb&l}sto;er(nkF6BH(y1
zUyEgR0%AD+khq^w_rtsh<+*v15E^y89mCKP(-$-?SSCC-|Fvubj(a>cG09kmr+=A$
zvy3XhyCS)`M+UQTtgcddWNRE=tWh()KlT7dN8Of|OizF*#<3sml8f-$`=KI*X}LI|
z`}k9kM8wbY&5?wke)XTqh|`Am_~ge$Ha?_
z`wW!%4M?M^yjX*;YCd5VGAYDfQ#%G_1LN_<$d$gBN6J`1KRWK>vjiAdv%b*lLwYDf_edAY~&{B_j2IH4Q+VtDfVD6m{^|elgaN=L*m8Gh3
zczdOrU#+wXhI-S#q^)@h*Xx#FikQ5Dy+IxqPn8!!x2Y35D$i6>yvp?5kuPfCj+)?J
z>Fd7Gzc$|9?_VO6>GwLk?{@)Y(j1{Vuv7^fEk8@d4pMk-IWHu?uPfrJY$xu9#5g!d
z3|RkoT>xKx@|U@wSOu?#rW_un!_Y3n!y)Q%HH@w>*vZOU0}pd~on_U=ur^}oQN!ta
zD6^NAdycUI_Kuxq8}6!t35(TLE3uSyzW3Vv9_?CK1)d9NnASsQ%Ylbp-x{DpUNCj{
zvl{3ckjm#igWv=g^Jo6}I{57C#r1^w1~|PL$L8YJ1Wg;w7hKj!c=6R;AM@+A(5z%&
zeB2W{g?hemAOx*-6#m=qFYPL}J4eYn6{*_AuvTa+
z@{{MDS}W8cV#PcsTcGB`k$iQ{HmFk46w|KU0ng}q_5JzV0XaRAZhjH&fZ3J;$=_|7
zp_jvZ=j)(n=Bn+Lr??c=35Qk
zUZ{gPLU}pXCAF}CYc*ohsuo_!7itoE+Xx+bucr0gZiKJeGltFyH^Mjl>V7K=^|0+z
z34_{19qb$b%eHd473%KMVqksI3M=mWx_?$~ft{D5mByVL;Zc)#7yYA@`H7*X-l5O|
zS071f_e8bBAt!~(WAtq>%G51qQ?40m4U`5?i8VsATMBvAMHDaNqu{#Em}+>2brNyR
zm%u{lwu6ZxMUW@`tm5~o0_a3Hz{wtm;mFshbr?Fm=8D+LH&>sT43^S0W6;LTGn7l
zg)$31Kip|-;MI9c!6@()-{_DDJSJArmfNpZW^moys3WTmsvstj6#)12F#Lm9O8@Qa7Or)1HR
zXv{s``*UdeC-a~6+NaXgAHVu7QSqAQuxioWDVGe|>%JvCnLjdUCsfUDEwRg?u{=)h
z>Hm;S`>=GMqmy$g&3hkZY<61d9)waVS_P2c{Ia(R#g+#EZW8M
z4T231vuW+re#P0CH?((R`ENIF%cs5TyTA2Ra{*1Z;4rN|5!R`#P*4NrA5LE{e~d4e
z7AM3Zzd$UXmabws`2l`U(riQdtop4ingo0~?*+d@xqNqJ$xo47+SitAzct*TAHCe+
zokwUi+C)^b(7GI2{O0>IIr6!*kodS2LOunwGl^b#lt?k{yl-c2p-nNZ>UgF~NN_H#
zsG9X%x)6=F-OTaYXFmAdN`v8xoMr{I2j}|VPp>YbHQy7Tdr?kcn%RX5s((Hd(kw&ej{BGw(E3~Vq|@~aXhUb2(hP1E(snx%YNJm@!TJnY
z|J8d^M5842ZhZVwK$Fmjo~qwdMAP^-d3fHZh-UL>uW1zg-`mLW_V{ubXMLkuzI;&u
zO+}e2-j5IZt#4`=jt62Ssm;*McqF{eV>h~0Tl
zt^>t1&h<9+De1+u=CJEs-fw9%I*WDi`~Zy>JrI&P|CvS$)woo-zq5#T_wnY~2V5nz
z1410ZJYVx@mp*PTvOicr^C)^6aRPpa>cd+Li|qPh+RKfZHhXTD(CAUsY1i0GX!R2F
zN=o`AG``D9Su;%UXm>Sjgtg1w(pYx+XNM0K&|Dvnzn(c)OndPwyk$J8gtq7L>bI{r
zOKDNEns`?8rQ480G)$31`S`
zwxoWKHNkd9xfcTm7|3fw>dhGaxiaJ}jCvo2oMc7xR4oX;X2|@Eaq1cJ^-H8)z>#3P
z1HnEk#~Lm0*M^!DF5To%S;DCBeQM1oKP?u5%?QY)Ozl@fAP+28R8Y
z7LT^_oJ8O0Khm$ODv^WC2ud@3!~Lk-@531P
zj+FD&6I|y-`sr)x!u3TwX&Ci3k#;xAzF>K$ERn5Gk#W>{NxL*>BFnN6Sx2}H>xVYi
zV*L8C9-~!RBgW1x9T){RlX8BsAuJm-eaC2B^$TO+^|u&1hi5VJ@w~vuvC$3VBTq7)
z12Z46e05?mZ2dm!=&&f3PkIHxR_~({w+v!=$-*j(scqdD6Zk~1z66UP)=MfNr(hV&}kE@Hi}60sOH=H7tI=%Xsm41#Ouqi)V8>_@2Nl$47ws_QNs
z`94Z*t8xXxaEpDoUGCVNFOoUB-EsS74na7Iebkv)b2#FCl+~@lTL^s`*x)GlQ7aGq
z#Pcx_wZwK!Ep&g3%Go=wv)BN(FWS2rA)E7}MpWL)ec2!3j<`uRRPHkVeb^e1PjP{^
z6`=jH1l)dgTY&?rA1T~`?Yvcx8+QR@^`Kh{c7TWZcfCgaoW0VAF(C5JcBJpS;-N9>
zZ_8l^ON2>OwI9-R?JPNg>YE~~B2l?5i@;}8e%t0;t}S4l>TLm3pZBHo399dA&fJU2
zJNTb`N9D^+c0WS)9aUem9@Y1x6zCv3nid==M{+Fdw#u`BEY~yJkzR@8y1jUuzPPh4
zNDgR!eF9MTW7%%hpZ(E|Cqn_-DfW{{uP*sU6TutxYhWwjP6uQ^kOpTCiyG=5&~`V&&9A9ELDKV(Nlv2PKQ=N|}MMEzxM`D2Gr
zXrgGq5^(9DqdMxRx?)og!X(?P(x`s{^D1YA=}SEF0|2F${rYhmFmn3y^i{xyoaoC)
zzrRlKKcpvD@{JYA9ZH&?Q2%E|@8IXyx@!*a9~~dI5>%ecGLVL_HA!a(;c)|Atp8WT
zDiO(EFFxYu`7vL@4INjOXi_!g>vj|0;Lg92!6E*i&az
z&(t)A_wkYFEmKIA+Ub1->CNbLFSr6Yxz6D#l6{BQK1Ayy`t^!5(o3j%SBfxq1D}(U4?7_g!g)Gu0U-k_M;|HR{w&|?EEeVXi(FNF!riDEiwSRYFEWa9vQF(sE2dvMPdJRbh$1r
zmueZq*si|}kF({v;S;1s1;pe1@vD9k`_FrWyYad`wxwY`F?ll18;ZmaB^45Hwmc$u
z>=;4IS~9PKuOyCK`9b1{WfG}>yoz8TV_xEoq<+~vDYp?N7%)e$foB-oTXUJjn~EZ$
zAD~a_3my~szD*joQ-os`#%CNPUU@|mJAOtHyZL2EyHrN}hRNdoXBfQW8zU||Zzttp
z4E~_a&>La2Q?)1cJ~V<6jJR%ng~&~e{*EyC#aIuczbb+was+dI2(I-e$Z@6_+tZpw
z?9S#R6
zTZaC925(!;n3obm2Iqk7He~R!k)6c;9Sq)Oe~#2!IT5_Vkl`@lcK`4nKmLcX330s?
z1HlPkxrl-Au(1q>9^(y$oe-?B3_%eiIA@Ii`W0lX9~@d-{*O=pHV`84ky<8H~)WjLLn~yXa7Gw{*+N3&OiwMSnuCH
z{`cJf^(Vow_Yy;Iih%|UtYFkvGO&zM{_p+}!jLa9>Rr#1@%S0#Ee!m3&nRZB(+bA(
zjbya5X6UsuaD-9L#Mt*VM!R`N{lDk@fl(g7MaJ!6?7JyOdl|<5vp7WR3w{%S5XHC`
z=rZ=B%Vko}r$I23vCr2XA+ihncro_Z48HADNaXeqg0-&*ev~C>@S5P#J~G}Cb|R~`
z5IL})$Wb4N{Kl7@2adJvcpRS9#2@|eB0T*1NpkM9&k#R*CG`{5Q)T$i04~yA_Y>)F
zgo*I-ze@>!?cGlLt9wqyt>h&7PsNBIWMle>`{Uk8_um2SpKVO;L
z4+0&Dy+-+GaC_gaY&h;X?X|};tzkcIx4}jY%VvjXv0V3wtkc`}Oe{xPF2nVnzkXx6
zbj5#oT;0^q7>6z%$8o;yuNsbnuKR1@NcB;Qdz9b^_EEK5f54Rhe^4yNJfg*rzYxjW
z#vb5!uNxFwh2--Y0l^3@w!FdPaC63Eba*&R#@U^W+iRXvG{^1#3S)j{`kXvpiMmU8
zelx4L!qx`fQ(IP#u#RU39{<#>QQQyhSUk2j5fdKFqJg6`gVs&!m~IF+=5g%x)hLTJVhGc`MzH9YpeG7v>~;z=5+O?nsU+
z%1J}~rH()G6yjeG{zML?3I2RwsORvH*J2zk0*`
z75Md@TdVg1nrkIUy@YZ>(bvhRT3|kNmF_Z7&b&fKrQ;hsr&0-XSEwIae@!3|a4;%D
z#|O~x@cD&^-$zC$)S&)N8`>t&I_Cae?~Lpj)>|fd2JrVq?gz-;l#$yWs6Ow33n#jt
zjZ2sJs{{TVNnl2L)eF@f(K>ZpIQPLB@OGNhaoaT?^Q`d}Vp?ETA%v61O*$hnO@)-2@Cs_6jfsjA~GrK<#qXnb)Iv
z91*#60O7bs6h4P;p>sjVKIWite2%7`F1(KH$^5PoiOLg=WycWq!|zZdblV>X)Czd^
zt4;n0H++$`w~y))FZyT(X#d}m9cX;!lg{uz0Ac*qsiUa=P==Hc
zLf*n3@6moOn$i8G1E_mQHy)jn($xvK&^}M&o=ZXYs`>AJfb@4#*Rfy8D;L6k&%${F
z_BY@3bg&-_ZIQx$iCK8#E2Lj79EJUAWmqY$SN3be`o~)Ld_namBNqWx=%d!RisACv
zW=E`lOK&;xPyg91NA-VC?GHofWABacXW1gFu^pGZq;R{s#HY=uo?FZUegxqGV$6V`;G2;D6tS7US#IGOIBu;<0PB427!H^1qPmGCPpdOhQ&5GD9ww%<5
z4wAU-+BbmvJv&VJfY>&o@03RLi+&LOa^4`U=Q~TvC5p+t)6lsAmj}0!IK^uxgyrIq
z2#kV@W-&S+C-=|^y&5csE+_rmWAJ%d1}}TS$bVjD+>BBL&&9T!jPKA)^fe1ey&hvdJ%b5<30+0zsnAOFk~oMS
zTPkTE&fuw244!zA!7J4nJT!`d{R}%HnT6+X$td6QkB2hk?~MC`K7()GV3Zp%@E_l<
z(o*CnXVxHpzE?8s!Sk7fn2v*|d|EBMEMU5ISPxL-Li`v1qTof)z{!@gRE49P5P
z|G&Ka5=K27K3v|z7&nw5k2CNx14S76(Tw}=7DhQFopAfD4EZO6m(OT3#$)hpkqbl)
zX0)HtBeK(4f?15`4bF(z$!NEbq4(LDl+!N}EN6`C%INpLA1UwuOxDelv94L>r2H!5
zIa0@n99>1uS;$r*Q^5r14--se+|!39$UZWTBjsDl2v$dvef8U$@cQ8V5p4I!LQ?+P
zo!CMBAm>uXh3K8%N%m)e06Di-e~29s9Yk*ZP5hCQ4bjVwCG`d_q~4Uh36HZaoa9?%
zPcFcIMr4GHvn_;-Zxl?%KfZ~an}E$+xLxj!X^hez(=f&>ugC4xC_yZ1IZznQ3rIeR
z#f#Wow4Tg6K-UJ>hw>?5e)+IzA#OiGBYKZA^Kkv2qpxuJud;PGztj;*{D;ocH@N<$
zC&_nR9<9gmAi^jMuCS0t7plRskAuw#qn_)llF(lR6`>_3m^;i+*n0qJF>T&Y2L_|P3*
z#4GN0`-6)DUUn!G>*<$8e@1c?`>;R4q?v;fh>tAEj<|vB(C4@Ri^@g)*65-Bqs_Yd
zP&w1%@Q^UTnRU;{5a0X5M>~r2wuu{lL%4FNXankp^LN?agMj}{95Y7zrf=Ti>}kNg
z@84#le%hJRf)KCYVz+7H5TM8%+YN{xhkWy2jP4J4BC~DrfKuF7o3{b_y?HO04m?gY
z%kCoLsga-l?6?cKjXBP3C7|o;o1=)|Wz|(*PX&Hkv!k9L@&0XGeXo}REbMIF!+3E=*MIZ@dB^+EZS^cwYS7yRu>3OmgI}Zi4%cU}o*n
zAgC{4|Bo{tkY6C<)fCKY`^=+tj$p69W#6_+=y&U)>RJQnclB1OB?ce|*c?}>c?b6+
zi>EqSMQ}gpo?uLIqu)Y^_Iqw{5PROAXer-wctjz)I6
z6nj|O9q#t=OR&QFE0Z4ieCO&8E{_B;{`g0c^Q&)
zbkEk?<rczvT@2++Opm5i0Gq~dGa-G+W=nf?-;v_@6NLOgN>~d}
z55Nt(aQ>t7lRWm<^P!&>AiZ>XZf#`m`i#{tk)B_>a5BQw;r#^&&A9{S&^^aQ!H*5e
z$%VhZB6RaUisOr9b3FFX=_}+fpnH(fI1hfF=a|GHWeexM+8yg{z7jNm~-Cke+bR|8iZw>HZO;W$-&Q$cz(QN5E0mr6v3p!&PVMsff1&)?yD
zn_)KL4JOm%-jZre$K&;^BK-gH0O9$g$yV4-GeyELT8hbiDsN;G>*ukOdy($8}{f@nU;o#>jjaRcN1S!Xt>S70W1TA9p$eksXY?@S@(
z{nrQ@E+<&{l<;ozTC%>|FOYV!Du8d0t+{_)y~=t`F3o#(sH9EDQU^>DX1czGvzSwnHy6
z9-s5=23N2@d$dyv^LA}JAw0fH@++**DoegM5N6`6K;xCOEXBNB$!HY&7o`G8+z-wE
zAolaKDrc}=oBQsnB7Li)`S|;Wpdty(!%J2qU_Ttl?uq^C0TzL(bvvY?|b#b4bo_(pib}bTx9U@X8h9
zjog$nDj(|jgyXXO&iS<{&RB2J-Hqa^-F&DUs@E#c$8n3b#oG$WGHljH2n)R)HK8~e
zm!qwZr(-}9vI>FUfVaEL*rGR
zUOR~NnxDvZAsmU`B}NC{7JWNL7}+sW(u?D#!cjG~3m{K_t3B!gxH5Xl88m({Q$q>j
zH_PsrZ$Ve+0_0l81fv-fn)7?0cDzjrI}d8dosA0B~Sh4cNZ
zvaSG+y}X%?X}*tIe11lAAK1ZBoRc5|{6E)Su6N)>%O6iFY@RNFJhXR{
zZ!g>%&YaQ{85)QCWFK=mb3X9;C!cn?q(a`>xnTWHSeIjVg^gE%zb-6R(&^K#KQ
zu(vf2
zf{x#bjj&!*y}geE;CU?kk?McsqLBGm8!7*OEv_anEwF#
z!tzPk*8Tm|pUulc%k$`zMV$W{k;O1iOY4;b{%~InFciLZ2F6L!x<#L+Q&C0gLTerA
z)Uf2T1>Mj-PQX?wU~eDwa?`fC_NiXVW_S7WZD7xZos$VXVE^2#dsl2D?3V)}RZLKR
zefYQ3IeJ1wx?p|-90M6)&`yUd$v0Bo
ze*%4H!zatS;oiA=`LnyQE_XZCudW5XpViJP)feD+X@LuNQq5Ynfy%haj^Za`*xGy;JJ-!XUxVy#QDqm+9&cj8o
zC{3_aY|-+scVS&4m%RO|0QT?cwRTVhoGsYcgYpcCZWTG+Fn*HEyu(V+i#3-EYKL;$
zVzov3fUmDFP%;7ZjTav^hIMzfwYAv>sL@s?Cj;f@@^s6R07EDCJ0O1{^rD=z=m+>s
zy@#3Txx36T@2~)U;fS*?4p3j@cU(RQu;p9s-c7L2PZvG&4fziK$9rpX1C%cf?2}*t
z46szXzYOeg^h=v^1o_({A7x~3iH=qx@`t~C7KtJMxl$_oJ@PyLaT~v*`~Q`aw!o$d~J*Kthp3(
z7Wu8%Fr9SdKfOMT#3Mh#eo$u_iU)O%R%oDg$rJV!+68vzH11U_gmrvT$X$FL$~~^^
z%nkwMV96Vx0R7H$KS2F!%U_*FcGFXC;yle=?oYu;{%ISANIq{-9*%Hj7+(yE1BrP{CXrmb*ozb40wca{XnjOYQ*r!r_86!`ep>sR
zl_QD^a_7?JQN7241zO0SGV$bbB>!GHGav|f`by46Q$W4C9&5Bd8be(?Xnky7xFjOG
zSmMZf7Tp+gPKQ!MH%pRE6tn=l-cAZ~M?nNd%grBW3
zeT~bLern);Yqw2d*}^*(V^J&l&Uj-%IhIY1yv6(|jAaJ*d#P+1msf>FVU#r@-wzJB
z_~QFq%vpcTTUH(P!0R(eaS>N3+vr~L-Nvw?u4fnl@h%`9>QO52FPRkl{iVg
zupqfdwS|+saIq254_mbd_ctcG7VGOWz6Ti&BJZR0Y{+}9r250yZr8pOSU4d*(Qe=FdPJ)FQglErvA@)B|C*MaZsFV6G#=WSL!SC}g5xt*`_hBB~MDO~2
z(vKk@sgFn^^VeTP_T$h-GQTNRGT$c+gztY4A@lh8!Vu4|ZkX_S{=a0M`CgKKM3o6o
ze-mbj_2O+v{jdzV-zGdF`{1RD4W7S6=r_#oZ={esLFE}Tu8lE4qaF9K9&=#;?(b|Q
zxwprvko^8ZE^A!h_~{(pH}~CDFdvtgBI9@exJlX@k#)XsoEht_xjps-?SHqW_PF0E
zeon0S-_6yy-mA(O?#Uz{F
zM{;?D7|x4}PP8T<+03He1>x2sXY^3sOv3T=Z6x2m^q~{s8My)ngi8tokD_xkwKC}^
z^5gtTjo1#)<^IN~K1g^1=T|=5I?aj3+f$Wv6UB|sIh~l-M_4?B&&^-T@6pNo^yMQAK{Z4vS>L%eG6i5lb+TdMp}d^)xN0|(0IS=F6&L*x4o3Mr#;
zhASuWeB|AJ-ADYnzT5Bzil6`8I--F5a%=fgc(n@iRJPiI{BGV-M>ph$m$_KHc?oev
zw5cN%-3P+`H}6Dwt{uZWc#vOzS@~lC@!uvXO=X_juYiPC4jBTUR%8oYFQqW2mHwCG?)%(_xYLt58&>l+6GU7mvZgwc#iUjcZyZ~
zj{?5U4m6zxTw!VbW-;)bGx2hhiGY)*T!-&~ote7yh2jto$7^{`A(>aA?&>Ls|F#7S
z!>xe7c*@l1KZN!X8|)2*0UxM&xuA8(c557e3H<$I;vp6R$afo;I=SgWc}(o6ydCso
zW}{x|4t!?kohuKLz^a;r4DE_!#{U+qOTzhveQ*&3R??QDduKUO2jo4!_6Kc;DvAAoWF(
z*Nft(Q)=x69xZUs5Z`$J$HN@3v=ByFhLr;kFA|AU|RrOFaug1SbmJL&j_qtR(
z+A&OZ-BnzD`|==F&41#-u0Oq0caH4!i|=|V)|6Ya96Z1epY*MVA4j2#O+DU3!*dhf
z|AbvTzn6M7D}2xEaxbMWMbj|4)k{S-#I4w_1?#$H5C3xD{kxZ}WtA%Kp`J}oIrInh
zP)B)nq&4MxDCT3ggu_=r|JH)59(VRq4cW{_Wz*f1l&$KSZOz@(Dq+V1hRog6Xk1pX
z&8coG_sQ{5Vf7yBq5QX-mn6HXPy287M(A}@r-RszFVX0tOja#%USZWm%^vPkJEzi3
zHM68_>KX&TAR+B~2<(-9es|CY_JN;`jm$IHN4+AAN(%6Nl#I_b#J&aoc&0;~a|HIy
zvK3PY;W-AB$J{yr`)b+M3o$JqZkqT5V=
zpx>1?vKNnlzK$iQ4tj4TJN&vO3C^XWZCIfw$O4bDMBm53xll}^o5H+UvpF1;;2cOA
zXD;%D^KG(WB%=Y=M`fs>lIbIFtbIvtFM%
z`d$#m*E)5P`2-;Mp&iN2fb?@L6NpcGR?)hVziTr7{GtHxPaT)=F~DB^;Kj(F25a~2
zLiYuFV~{Q4^R6u0ee1za<#f-BqadezUaxonu-mh~9p(RTUu#=}{8`+m&*Bu=>nk=Y
zhwl4X=MKs|g7WCJAt?dCfU2tr@krkDayb+DnTV{-9&cgY(zyif)j^-*n5^M`;0kC)8=bt3oT?=V$wC@pKhL`isFUFVfhVj0R^Nl
zx4Hp3-_>Y9_l`0CfhrVNQYtnEBmE^$z89=BIEn4A-lL28r@V~=&a?S^(8s(a-(6)0*^~O@4(9o(Gi})J8zX}_zN_@u
z?6@r9~K;&yNPzNQ6eu>W7h@xr#1+
zFC9PN{7Fc=itNo`vBi8xf@MAK*E~K1udfR0
z9;_E>LHH!SnD9QUdBQ`R%ZQzKqu4P|+kS$4M=tZ1=rw&K{99sG8MbTZelM&iUsH(X
z%Ke0Iwknc)!QnP0TyIu!3_rKscgX!=Uq_;@V3CjV8Gm?oGm&-&5@QZ!ae*r)|*VM1Rl725;_>yzPdQCh{*kI7C~Kzp?w}!H)c2)6Sg=
zp@3gJy)UDF=BAmj9})=cU`KjVEE`ggf0)~RnkNuYrPs6q^IrD37s$Ul+{hb5{oLKT
zTha*dvWYp{JMeE=o@4(ZJ42a_G|)LVUc<(M+R;6k@b}!6CO%Vjpm&^Eyc)IVdYfs1
z#xv)pokRXRT{*fV25@FoLCao1mfcRO$Pa6pZW}{+-}paCcM|~DEWU9P`O)_7A=yt5
zKg1uuIE(CR4OZ6L323Xn>C6CNU~EXX2#i+}ntSd%#E-=x2`BPF_V#Numjb_BE^euP
z0_1tV;FRBh+Wo>|S`hC!VoPn&cLqAw?GCwvUYZ}zL=MEYJiDJ>xxlw}3{EUNF$MO<
zj{WBcjA=;d>)
z?4WnGw@o|%_>JIxD$f(*2ED1@IR@-xQQW7T4RP|o^}Kg$Yx=0Qo7+;0pnt2Uvw{buX~@E=$e|6!&fNZmZW$!uVNUuSQM*
z?>`>BZFlVerL~CpFSEhN-dt1=&#@#p-N_hsG8Xkid8w#
zcZ2R2wXorxL2T71wd2eg=5#y2!K_#Bw~kWbA^uZfc3%qcgV*yQuwVB}G%t
zcTu0xpJlnK0uSX&7)cxMqL^Q26`JqwrgUDeI(=)XlS&l6xH84ElUfjNyeoXVlN!m-
zTPdu$aE^4ot+~ACDCl&LoQ2XSv4(i-8g-N;pI;c~8ex`m)x+vDr>36$!
zc2Pez_y6r|=%7M3-%~W}ZKGnse@lG3-bP)FUh}nUwv&2Uz2kDjmrg2t;qAYr>K&AH
zX{h3^kTy#A$W&zJ`8G=O_f{nj!EUPRq?JT`b`MqkZR7O@xYxLGtUDJA_lOz61EzzY
z;du&jtsnW(Nxl0t!PDZ{O}#(brP6N#yyomq&CLCDN@HP@PXq8;p@Wm5qP!i{fZyc@
z`G-3xYk|i@!$m!ms(yZt^fd6P5AG7rVL!Efe8$Y04)*Hjj8>JjQ#mbyM1k=q$NuAoMt+yZINClHmUYfL!U1_aK8xt
z%dXe!zkJx=LvE`S^Wg7R#%q`ChI>~>_}b&rz{9!nv^E>Vdy|l%7upqY@A_V7{rwi`
zRe1!^S>YUbuME|b1pcXaYsI^5aK55reO|gBhWvh4lgk0%_n*9vdk?{TnU{4nJk{V2
zZ0}$HxC!)CqHozG{{a80XHpX{OsA%zQ|F}NoY!~`Sere8z*As7xGciqGlcTjPkc?9Q9L+sGemeXtp8`}Diy@nFFKV(
zqP#g}oA}!Y&`gn=55=J`PtMsOzo*gEH1Z4N!4GE7Q2yY|>Rl@kuii5~PrNuOlP0d?d~xgc2AtRV(ZCvr%9)>f<{&$7y1bf1a#@qtU4$Ncc4MBfN=pU*
zJwscSh5)*UEF9>-d7}@VprV-)fA;DntmNY7yx
z|2(SSX4SDD&Hv;!iFelkBNVvg5nqqlb
z-@^;@BR>NX%zr+O(J_x>F381p=58h5LFDZte<#lF@0^9&$I&$~Kl!bF9=|`|*
zeBCu%&MTycd0`}TDwAU{rDyF_lY@d@35ZJt}vYMOunLs=dC7Ph3EHc
z{|YQ8o?D9BZ?4>p_eDkRB$i8`Ji~gk#m}%_$q9YDJ{}7k@b|RSZ(8yC9h`cO^O%}X
z{Bd6I@RenFAM|mXoJRBBo3H!`VY%#U+%L0jF8)rBdopk@Dqp$sv^3fmLYJ8E`yb=&
zvqh*p)0l?uW6pPLc(LBau@ST%PD;7q`%mt%taeoXxZfPxI(ku!1l^h_|t0OHwSxNQJ#`zt1
zXN%mW5&nSp|E_aH^#|-+zoGVm#gDflJ~V?cIu+@D|bLjts4B+vjC=R
zXg-?G`9Fg7TgfuX%o*mlUck$`zlESLu|`om@DJSYkEy%&!}$K&
zC0DY6-`8;LePRpoTX(TS?R(HKUlUMGccxS33X7gz2Rq(cDXHl~|1kn84|OX+f3?Zi
zg}@){J{tCGzoo5bu0XjoEO*_XqC`&3Kq0UUk|h&8*r$rzVq6$Y)j1DN_?4
zGu6R9>Ym|#c{#{q4h6qngYt_KzDgHOb>Y4LM*f;+h=WE8B?}cHk9qa*K*cWLxr<*|
z7oz-;bX0n(!znuT-8uc{9nD_KD6@SuIu7!hQEigm;gAQl(wmtEUh@55(5fE6QR@1|
zsdGQ0$Eemox~$>@I_3RqO`!h{xTj5*PODYYsh@0rmR(piMuoX_E^v+=rEI1?!LOjf
zx>qNTujuZj&JTYd++#jSUGlVEBVIE`S?}!mTR%2TJyKoc{C2pPatR9&^r}HpDrJzVz|<;Rml%i?1jg~?EeG5{3$ZicA}ShX^`wOw|0!G7hUh>Bsof*
zaFGw4xH>{j(&R0birXmtZTgoscC=CHk6s;)AM2#@PQP-WRPLpyw9uQ@4)A@jcE41@
zU>g<1wzgau_}P}BuaB9|byCh@;Rzwnd#JPGt=5Jt@E$U>+g4(%jbi^DLXSV*Mx~|b
zCyh9CP+^)TAK8yXelzy&=keYi>Whi>+J;5#)B)d{N}Q2x)L3h=vyoXlH7Ye}^PmFw
zXW_vx$Ftp3-=j6}FLJa~w^?;WGrxmDWcBT)lMG
z)&uf)1K+t_c<7W$X#6&-3CJteOE1fY{a}92@}vRmJIS3^>>Yo5DcffD!$;x1!fU(H
zl=DA274o%7xgPWcUmUKFg1n&qo5!Qvg*}w^y&KmSsl#*m$>JOd`{|C_qxD`N;okS7
zU~V%!FPonEhCO24RFux!4Zrw%slo7)&x_gV@ST2JaW{;wJijAwI0wF$Nd0K{Q>%-5
zuXn(WeR~hp6O+410QNH*&z)jQh)y}I*?2erc*_$tt}Dl2-%H*M?+^g~qEhk7A`{L}
z9^b(@HaK5p-WFoYaK0946zkN0onf{IpB#Yw>?kVi^&86h2J#n~z<#|yv-T4D``-7r
zx}yvrkIB`su1pkohnUfKsZ)?2O^_+`{0!&UE$;U*^xeGG$s0z?V&U9ay1p_2yIego
z&dj%g!qzYgM|1e&|_Vw{{wxOA)j9<
zhx}Wvj&cvmqaKwM2oC}@`|V-J4fT(v<7lyfRmCr6k=`nYky`Y9f`(&UY!cwGz}}DA
zU}xZ-R&iI*lfI?4=QWg@4uNmm0
zH41WR+w1}KUBuPw-#9zJ~4f~Da$Qu>u
z=Sc6+xy=+Rf20-PhT_T>IfFLDAHExz3S|Jkh`z*(>Xp)4?;^kc`tU13#5XUu$74Om
zL<#RYkR=+foxcINx{B=@%DaATDeORcmFLS0aXgb=Oa*{ElV|$j4xr%qOU7t?!w0hX
z_km)$w&K5gm(y90gT^qCUv3$NgK*1gbY9>p_<4`Vo!uPR0J5RqtKh1O$(6ZsvbPKW$B)Q&?mB@Na4emW_J
zFk1Zrwkx=r1%EeR|G?@h(tGi85ign-ec|^%Nbm3BXgr=-@TNgjpJ^=Uitwz*9OfBC
zC&=GfX{or_DMEKU+RV-WEllR^ly^^7*zXF5fxV`4v
zH8_vAKll#jRle`m;{KAqCutzPD;_eKFG=MIVP1C1WEkgDd-$uc{Yetvy74-sW#f5$
zX1f!2{l>`o-&1-@ta
zFy9PMk;Ttz-PlUZhidzeu0wh|wySZXbyJLwzJ%J>y*+@x#}v&^#`(bry~#7ETqtJ4
zWyD9OY6L=%{788ie-AT%;pcU!(f66raloy93$)OE&Pl`!|9$wk
z%NJ!49}-C{>PPnE$(jsk0Iqv~{oV)ATkNSbiuN^k^d-z|*~a=GqW%2raWMYJJ5FI&IZ)pWHeIoKH#20Y5e!=^8ux8cffB66gh?<`^}nqO7;PkoVD2b
z5%9rM_q8`+yu!?wdZk9-Pg}+-?2w%`HMP&c4+sg_+>8Z3Dv`zd*9-jSbi((BkQsPh
z>4*K(K=09ifA0H(-pX0q=N_Q9(plQ_9{6GInm=!LOo4y6HY+3m^<1qNTvR}RleTN#
zHRwme$K`wg*p)N$xGy9F?rozzTx+5Iu~x0o3!v}1J5eYKu(K|G{v7a=BU4RxZ-5^j
zG5%ah13P_GH}JK=ICG{I@5`s`^OZ~C?6DD+OykT-%E!_86MN9A6K|bI~
zb;1%>;60nyPrhW^PNxlRZW$*~BmNfqbCAiWQao_aXl2RJf?k
z_EDxbdunu);tug=79RRR
zStwRoeS0RVW7OxSik1u@Z<|aUI
z{@SgOhkIyta~&s7J5?6gxRLE=8|A?~D#6j&M)~d!7ieaO{Mz2d9SZJUl-bVHY*no7
z)TKDK!Ye
zG#^VlW%jq}uMe~zmf@0<&T6O3yVZov54BU9kYg`(7m+dfHCi=@CBJ
zMdjaT4k+8(LwRh9Q@ZmG;`nt@Pjvx0wY>jn75e+eB8Rx$1*)CYpXF-~ZJ&d@qVk8g
zn}lGWu}8=i?x9m@>Aw%nv(l+g5|ag|zO_?-hWMU_Gj&p5NB%DQXwyTbo}0?s4d+48
zmPP(Le6JvG>7$)B)lS7RrF(|3bWlnKelAgkT~uRYPvobrUP{-*PX2WT><{Cu(x3c0
zC^^HC`eTbbsIQglC5~?HqEx?b*=4HLOVy2!r8ZjwPqcnOF61`r$j+#v^@th>`Hq3UqvY%+O()17{i*l#SOEDt?zE0v<;k$$zj
ze>dVmcSa>ou!EjSeARtp@IzvOn|Q=Q?|~KV{u-!PD7*UT3g|0j!oRqI@j_Gz@7aT1
z_htE<5U5Ypb~oP(`e*){9#I0l`E&8(C|rGNsFeEgAbP)EUAr^~>0fe)DgFR@XL>>c
zKZ88u!*&bh>oOdF7$P1k7(C{V@&k__z5j#s@7zj}M0Wod@i+jX-FNkVgx$s+HxRxq
z?%#m)P82=hL~-fI_g6l+o=*tBXG`6efO*p$+X!70_fobo{X)EAHb@f3W1HL)IG&5_
zy?YosqSMuL?{JTL)T=7vz_S#U5
z|E}g+cOupwzwC?OCvJQwa{%#z_N~hwAiZ-(-rzjaqIi-&qk9(Nyp-5Yo<~T}IrQ`i#nb>(0t^e9`qj{t^v2jN!I-l%*OH9#xv)6v}
zMEfp#2z2oEoLM79_}!IdwUNc_gd9;x8b}V3#(=`20u5_z}&YN%Kr!iH=}c*
zqg~mC?r}-#kNMAo-U7Mip_jm8(vJ_fZ%6HgPv;RWY;SWF&_Sqh7^miRAFT7g$9{lX7
z(G_?5AP-|_Z*i9e-Ve&Mn;Doven_)*cgralXR-2eRc-J`@Oy1_HQ-0bl1{48roo@S
zRMj`RO{ca=1T9tpzwY<=q^mmgC%#I2unF{J6~5KHVWLwec4`w-@OK|3Yt&A?0{^RO
zZ0fok^uAoIHaP%#SKbD!{sZw!c-I;(l)gK#guh|FwLtMrD+s1zCUh4bIDs9Nm=+)CN2@gZu+HGyWl1^p66uBe-
z@w9#~x7yR7e(KFb``bDEBb0g1XurXgA5_bLbV&=$OY}_9O?^J_L(kWh-wErd>a!`+
zVJ6@$3*%Exs{WvED64KU_UxdF2a7937j;l3?kX`m*LP8NXP@_PJKjsZu|I8Sd>PiE
zyH>%WzJpqGz0O-&rGuI{($kgN-AQ$Xxos=y?4kb78Sh)X3H;MO<-R`IPU?YmfB%D~
z4ysk^+PT1zPRjF~LMX2cyk8TsWAP8_r9zBk4E_B&DfTd9`%$edokeIQQeS)52cL
zf9Zo`uTA&_3;AeH&~AQY+y?d
ze23kI*HmqYLvDA?d^jO*VbWimvIO*0Mu+_zpnSC1sI36bS>nz^=_v2>i#M42
zYZ=^Uj_RL4`~R+OmajaZq1^!MdeD2F+%jSS{io2(m!BwxzqfHpQ%?iA_IdKRjc^W9
z``J6$KyU3fIeIgUzv0f@q0>*&(GNak2r^9A*5C~^Y-&h{dy=@P_SEFx+gP`qe6J!^#Yl4oxnhz2~svB?09
zqy9(!3!3N14*mB?&vB?R59z7x>$(#Q*xGcr7R9fOT*tFWuix_EFw)brKbuD_Aq
z>BGx#-lS=GH_mIsv0labkir7blSnU3TI(tr-)YSxj^`G7E_htaN%FgFHIdRduQYBS
z?uq(KAJ~U^g#Nj3oS(_Q`Wxq=9-mu*^)h$9!M_&}Wmmls#mm=b-7^2<>dND>dcHqt
zQBp!7l_ip7Ns^@+DpA>rsO&9dDI!z~Ws68smJlDi7F(92LbjqTQI@hj$o4!)gpl8v
zxxWAW{(8S==G>VxbMMS~&vMV;atvnV$@#&9EDu*=`y1KY7mV#?-B*t*ST6X@KhQ&+w1$|1fZyAyA`gFX=}DJes6-F$2_s!H*b0^iur?>
zmoeiC;LInFXWIeCn(8Zc08^JOO2YQiU()*EA?Qc>(z^w%0l%7Sq~LR$Zn_bV@ORnr
z-@AQvfO;wihgBi{r=apM-uJPvUNp)D@imRdbIt(1oT%pi0Ql+2Yb#NZyGp;k2A31)
zku)U@DCQ%11L(ITSsJHDDxRpp{eV<#%*>hxt(K1YNo*lpbM`?!(`U_8d0uxzI>MhkBnb5@DmE{?M0d
z_bR*KzB7eDwYhHeFfrqOsCO@VuRoi4t%rtME|r|4_jDn3o!V&iZ(WE}C-$Q0u^u$m
zQO#3O*NdFLR2TURcI73^hrv*ydWdf1=g
zcVs!R6UM(4Z(n`|x#NN*c_)iZCd*o3ka1_SJ9`lFK*x}qz
z+w}B($S>;+gX;vydB-9>E1O>Al$84OA=v%Qe{nQl_&yIEbH0%W-|G_Q-4R|e|CauK
zYW5MVH*H&+d2gZ=_JhMZy#6kR|B9=OGCRQDv#gYixdh*t-V4#dwO}t(GQ_oSLqGi!
z6ZOa&`U~Gw%5zruF6TGA;faKJ!?V5_M(aQGXgLq-O;hk)dyTGDRz6!!_+gb~q
zK<<5p_XQJ>8#mi_ekp`IR(RJ{f?S5_G}9uGV`krQ1OE>)+CBHQ0@gP;G&EBKV7+m2
zo2|=ju=}pgVP)Gv&SMw*8*%8*O8Ww>e4t+Jv)?X%gM9eAm|ILizG2apB1e$B`SFYY
zJCNIHUp0&EH;SqkSB8E!YxJiC+e=p8s5f|A+I?`-vIxK@8!zgt0t{&jI)Lq}o^@?K
zw!3eBiFf01`ABBx0W9aI%G}BXXqFJY9D$sl{<;@t5QcLq^_CI7khvN^3rq*NoZ&LJND-Lhx+_W9b-7(F1fbG5`
z`#Q27Kgeh0iS0hJ3^u{}Td7LD$K~z_NaV$O1}mEg;q?1~Tgd*9j%i0M);E;b&>qWq
zm!2c`(n^6&?4@7jA7U>T={6YR@)&1W!8Ii8AO5B0@2r>Zkn`_NHcQC)jpBRpR#@(*
zp`Rc|S2oID%BRa^Wd9@Z-BJBQ)wCRiUb79PXx^Ip>RHs)t(f0AlKaJtVTaxTd1okk0m
z7uBnb##mle6O7yKjDPuhC%|W4IltogmHbO%F}mD7O5_{u^2z?r@%+Hc_&uJVqt5;0
z^azpjKWz722VwbfeLiyU3HO=NljOY_{}+A?;`t-q+akbsw}x)C1OD`yPfvk<=kaWJL~?$VBBia_MZ!2<}QgzJimW!`tY_h;EpHih5JBXd$uszA=qy?vcvkrao7iD
z=-Bo*%tJp*8P?bZ_Ojk#cimqYUpM{9v1b9h&-rgQ!~y0zdmY8Hm|*;nkl%7_zk46*
zMTVJqkgu}+#Z()}cb$CN@Aojyymb}&Z2{w`Sg++PrU{J7zweiGiAM8V)42;))6#|}j&`QbSLOGk_aTS3?l!)`
zz{Bia!)fsUoXO?w>JCQ3=&C#oWhXFDzDuG@9PE=cb(^!*!+F!m8|5CX7SN9Vxm#_c
z2awU4FTGaRM^VlR!vn9OT?DVS+ls*W&p62wDGBQaO*!s||7k#d-@DyOmhMOA{6Q*ku{$m!tH9
zkyJYT5C7S#b&?K#Tz0(iyiG@&m5cuF5@sOrH*8$jdg;hQmOk}uijK0jR-b)yk&Y^s
z8b3|3qa&e$WasgF!|21X;<8e?VKlVoLF>uo4D=)*6Bb12Xs7k(bHCb$5c5)_w^CaN
zkZRqx7x%U@P<7nD&43I7g8u~I@9W_%$L+_TD
zF|YcMhJJqxX7LWCp}>?81JiICs(kV{`IkElebn)4buytL`t$da_f=`Ahi}CCav2S6
zV)j?~`Iv@^70hBP>;N_2uFK}8q4+aTo4Rw0giXXKe!D
z1%~3`r8MOH7%eIUzS2?j;jsCVZY1gcfn)jlZggkEc#!f#8nUZ%3X_8JH14z1IONe#
zf}eN)_g^#=ZU5|>YF8Js6$;jW*wlseRdlS;$sXiqwUni(u@@B|ITmv{j)ru`cbM?O
zI>%_cS^Uk-G^CzAoOKn}0eqUa-jkIEyQs8SLvI}oeeE(E?egwI(IPoJwBfwAaL%nq
z8d8N5!Oenri7#4u)sL8hP|ecs}J2ZX&U$j{KzmdLSbzMoNND7$bG07c=_N)
z#Q`{fzuNq~TmbNBv8`uZ4#W35Nl#v~aXYL>DBq1gG6wUb{Ksw&fj664g?Q8eUvivj
zQEr8Gh$7ErhZ3NF%um~8#lv?jy``6T3-p7lzMZ*ZFfOdFV#*tWabnS)2e1hR=lq|%
zV@ZN_yz3FP?FmrM#VrbqO5puNlh?051pKi2IuG0GKJ=N-LN*=zh|ngw)~EtcYg>Ji
zjRW#wYF7JR3H>LbFT_V0_)CU!=eQ`06OLW_>E1B@wEnP4y9VQe*g4)`c${p`jAU+x
zarIE^uCO#19~zQlOeP>csQKy8JeG5wDaZRJBUhjBDgOk1Z)e@h3Hdz>lPh6{ak=8?
zksU=K&pCVOItR!dRW)yGgY=b0yoUb5^SYWQHhhr2E%uK<2jKBPYKoG8jN5^arXZb7
zAnKAa;5GgdN4y^L;;GTFH{jjmt=gC`<*>T4;s2RKQoqQn4bx(u^+_aYk}?IWXyL?&nlh9_V}A-
z%6?Wr-bQQkA0F>o5j|`#(noq)aXt@x5+|`;zWsW7Wgy^8lWHrrGicrPd;I*bO)2EQ
zv;?I>A8aoVIf-4zcCd=$*?atarS^UD|Iw;8^%7j)LX~b}S7)7M{$idd_@z-0=X2@M
zEb(u0a;ob%w!^}-Z*~~pi%v*mG&@q!g`d~UH2V*S6ZmroPg?mnqaB9>FBJU7yo^Kl
zNBbrxDj{>0HhojEc7w5r19Nd~Jic-TZT?{{?F^
zY6y>8@qE4n%kdj6x`nZ3{mm;FQ~nDguIXa2IhNlXJi7sxSNPx+Iq!V-X5bN=
z9$u~=ioY+ZiLd#%AIPlkNWtY7eOV*36R_h{iF!Kp&&`36Rx_|JeYYm#=1myK-ot
zbHRZ3fX++?>eoO%oZD)91@O*;vO(fluM5Xlb>aT)xa4810boO&LRAwESDz^U3)pe~
zG#l2ptG6i{=X>bMi5ARje}28VFBLFBZHoRD^iE9-tlta#aCh-?*)zE?Zp?dd4)VZ$
zOxWXSj~g&AbZihAO6o!F4~m($^};x0bZ?6YjNkqV!BtT(K8x&}vR?#tV(6FD=wJ9B
zf7G@&(E{2*Z!l)D0gTTgHKHg6c_2RaEcuqUPaBv2~D2>bC
z2kh5s%l^snr@cs5lBw3rVGyOhxFAsy!9b5v#UgsAhLBTJ(f7EfVYD%DVy}}mK4o8@R7cb6CZAgIY+zib8Tw9-5sR&9MKokfO=&KN{1
z{FZng0=pZ2X-{T5yhpP^ZRtJG9vr`Uaw}zFJbakiKVbuQ^7a9~xqS>26=GJHyqJ!3
z>o%6t6^9UK_FxkykcLk
zqHNCkkhfA*ocdDqj~kCI+j~|kn@9mYibw6y3Sv(?ECrM$Y74aQ3n6HMSgv;$4S2%U5yJa
z&`jw;#!epjKV5pz;Zli|sVzOot|%~@sj?f1q_|j!-Uj&xUmhoj^`f58lx)AD9z;vH
z<@?pO2mB`4JWH4axkHNS4?DV%jfg_Lws|l5d_j2QZQx(yDkw4fUJq(~zn^hjwFmzD
zR%Jfp2Ir~*6Hi=b(9mGl?Vhn^ps(T?Q
z6F1F#)$2xO0&mhxE(32o)+1b&Nkf;uY2Vci=|Kd#Z0hep`|*~1*e%rGjby9UE9*IXkyulWysb0LR~pL?o(6ss
zeQr~coo^4?a9^Th;%g7$&nn*Pxf<3TlGNx13|RL@^tCL&6P@_>-7kMbL*+@Ic3n>D
zMzvp7mW9E8Z@ERm-VJbmMWE!>t<~_K*xtjHmp^-V!}<6Y%gdELDF2b?67A=}yN=6-
zv_d|E>$T7Bhw-EDLSq5#Q7_W?x$Vb?JFp(3Y(MJu4|pBBc#;Q<54(M~Wb*)@<2z-%
zO^FWY_}|#u3d{lj`Y66<4Ep;rn=bn+z~d&T#CK*vzc;ZGSIdKOggdWEn|T1nbw7`T
zrZ5g{=&-taw-kmvpqB6Tk@bC_8ns1wHC^0
zyWD$X6yonPk2PaERJoNAgzccZE9;lvz`y@^m|w^1Z2fOJZgd0w_jK!{t$>r^=P#TC
zbgbYgF9u}!Ic+x#=s);*2HS-Z-lG$^yn)~){&-(2pC0_-AK=V3JO9&wo4r{+5I)JS
z+Ikhjc7Od|VY}?x@nA|B@JQTG1B^TTEdS#4xq@y@~DgnX%^Em=B&jVXB1r25oMf74zA{Dsz_D9!o})e#Uy4wXP-Sj7|OKx^caW
z^7(Ro0h`u6I)nM;y|Gd;jGsfxiC?_^d`>+$e3N6FA1?o?N!})Gzg4yVau+_XUTGGPL*N
zbn&FJ35><;Gi2W@N`pxr>zT4xKmo@;R(gyic{4|{yk0eq#Xh)sQaN=4^j1Lwxax-$%Rn)t1tLZdtiQ*$kz!^625A*j`BxQ
z95qF?gC=!PiGV*Rk$d=ykMLX9AO=bQ?n?PNb`7QcO+=izI=ka&kogP_k3OIA%5Sb4wW8423%b&ZplAI^6y4X;G>zUivej1}Q
z<0a8^DDN(5mq^>Yq@8H`^`ZDV{l>)mSWfdF8%Y=3&P?7nm)^ZXoc=qQovZ@~I(3ot
z;zg+^$n(`d0*So(vmInUFe3Ane6JiCa%6qCV8(6{?yos>OS&+xo_S5@$2?eOyW=iB
z!10}T60v{L-^K39xL(8i*OPHho96u%%ik9?+O7d;Tc|TE4j9!X?2Gjb*k(`4=Ul8t
z*6ZD(b}hy7>&N2qP61jCr|Dw7GuzLS^YK6KY}Ci$N#!lSalhvo@{7mgNzketWWD-v
zxA|_I{x@g580VkhVEF~>Z6xAdi`Q3Er;lY
zP;gO9UJG91>joW5~Lx7u{z%y6*A?m?u~leVPS58!EPNK7e}ZRcF7vHUsv#O(m>5uo8fc$yzn1GWdw_iw
zk-yuQ3B1OfW4;ESUq2Y>e-h>kt*?T*jD`Eq)fdekHz8inb@?9UyKsJAGE#ztZvg3z
zA3o|2`_1cLt!Y}x(~lIMUt4A2*@tA4(sylF>O*0VFLWOOeysQ0f5pV70TdY|(V$F+
zb6?HhUS8nuM>f+|-e>dske0~CrhgCmkyvH(V!n)iWT5K$`{4C{r1Kw7iNNB1WMZ^@
zb@6N1_tfw6lt}}d8Fpt{qXcNf`em7SCcfHID=6jsJf@K%fVBclG
z!S=L5@S~w_RZ0W@X)ThK`esnRo_GA?&ByvsDG%$1s1TSxYRPWTgzs-t({iu&>%C~v
zr>;AG9_+sq$Cm!H$d8v7WdEa`N
zkN-J?7*5~eKEM7eg?JmmL`-uVk
z@N{8@1N2Mt8&AasA)F)^bb7}x=qDS*>K?(m>1e!n@iE}FQoco#LwY@Eydmp@*RvjE
zxIS)~ADlxNAH1O>4Ea>tISSbjezteFADUSKyTNN
z!*~5)919+}e{g9n^i%uE9dzi=4U<=n(V#!4#ib2+0FRgTY|8%wJX*ZD=pv
zcF~6skX{+X?T-2JW;b!J=aBCyBj$G%kbbuLA3ydFG!q;90IuriZlC7RrbocLvGeY@EI7)dvpTub=qh~+K9Tm5z_kZ@(gUXuTFD`^rQKk7=-ojwbZ
zaGxG^FNgPu3KCv%<21>?Ig6Eqf2vXUcLX%YlIPwQQoR0MH-%2F6u;klmg3*R29zJW
z?uC004|!2?e;4k3`M7Xz%GnokG)vc@T3Wb^nNOEOkG5@WdFY|1J-4
z!mB?n-1E7Tzlr42v4yIqN&~g-Z|FzeudVimO6TUH=4aN2sQH$|Nvi!yOtzBtd=p6d
zdtN3&%~PJ%Q}d#f#nk?R^=ryMUe3b3qWlM_dOc*J+Sw$fi|8SBpDNEm=N-{AZIJ@`
zu0Gs)K)&B~LMlJ-_xLC`X&3(P)QzZli_sG5{Kp=?Vx7gK20B}??O54%TpWg0P&K6aky<2XmnPZcH#
zh&-1bHP4DIE+Oqa^PT$c{x!yvdDo)gzjJt83A8C%h5P&B3{zb^4jo0KStkG$M_bAI
z{mN~cq@VnEU6;(auexn{itE`CC|Zceiw(+aNxjUwg~@oa@`MkO+x_qj=?^t?%H+P}
z565ni`c)-`kaIU?*$reoVATB=f$OO?W?hW+xn(Ow^ndA2`73nRP$&J9-cGdd$)zbJYwI{ojX=N
zhJ8`BuJKvm2e(ERt!M?_dB7z;sRVdzsH0Q*x=G+`o5GK;gmunEyvx$Dovb;@)g%q|
zsJDz7Q?m8(I|2Zh%jVZRP`eJe&XL4M4@k`w>m$TDX@vj&vs;n3Ld4D7Mi%{y-%
zz9~bLd%Smmq5?s^2cvH20hL8wp>ew^6zYueT4f@)UNLJ
z`2D@ecf9!n^p*Y5eO?3f4BQ!MQ4g51o!=L~$GHbTHkJS$bicBA8|WQdLmR>C*H>PA
z5BLoFOD)se69?&HPtdtvkUlabvm2kAx^_9^%5LD#N0kBtpG*PY^v$`nV+{D}t6SH9
z0nh#+9PauH<}p1Y_w3_9-^*R;YRjRWk6L#7#6bT1PK|3XL%r5#@P9fE@|H)UB6%Qw
z`*&tvd@d>G*mR#5gg^Njmb*i_vg)1@YoLDDrw={H{?T?{^3=-#{IfdTIR^A$sL3m>
zgYeNWK08{We!g?DmAw$&&&4+?3+Z0VX71HMJ+5#G58(5ub;AygmVmOfWF>!?7tj3t
zEBhJdVR=6H%<(xE?~TJoTcLdgvK(8mUo+cf7j~M#yrK2I=RXn1r`&ImumgNQ5(8X%
zg*Y71wowG~bF*X*#^3o^w$Fh(P%dX?e=@$OG}ztOxDIfK)5dMLK)y)x#0NaCR0l1M
z;s+GvYtq2u-I#Pr?l-_8tpcMEz{pB3BV|ZuPP@Mvk7N29#@o3;pU)6Q
zU4Z?!IvYYD{%(*&>>qeir>0Fcp*zw2Ha}!{A*u@TfZZ}-=@DUaW&(A5J
z!tXbK=E&GdK;;+zoyFgo^3C@6hk)_4aLH)E^<@!Em{)J>+!}<()e!fS_wfF#TO`wk
zKtRXX@GN6MneU=Ac-*>lLbeQ#=NUblbFn>It8(g}1R(Rso&9)xo!M{A?gZF$^7b8U
z-%#(tX^eT=mYp+D-@fBrKk>R#rI4@~9w+B7`VfCoXEjS^@i@&X`(Os^`KM2N4{nF*
zl6$Fm9RFJUjuY$0_gs|t&r*M&{Su!eIriO&_*LK*t1-g*O}*kHez&5=S9M@}(33Q^
z67zO?+;ef9o+x2(6}Pw8zGz3B?sco+62>?50P-KDnS9x5TwbV&JXuF9aJ93+=@&P(
zOkq6Ipecmw{YF)`mgx2Q`6nE|ep3Gg#>u@iVmN=txr!CozHx=epTS|L-d&e*`tMKu
z^%%F_>?O|~c>m=;9M03%xQ4O&?q#x0oDh>p_{L?8Ny59sS47$pf2oc2D&Do6@QxO7
z2U5<4qV?pz*qWRVgvWSUULia~%jq=9N9NycQf@DII6VC(35lU})I(5&;s~?m<=$=H1pSmyHKktx`OrBqLmO5uV6i4Z`S%TufM|mk;d&GkB
zzwFvXJ?||>@nrrZlpOzVDjd6TkNd3!elO%g@p%h53hS3rIIu9UfR{qmV*vve;#qj9
zc=!lNI($q7T^BHE;XcV7x2Sl}h4|P799zKc3;8Wun2&@m#KUF}$qx>t5QLw+3C=F)
zKYE9%x7$3$@3RW1c~0qhYQC^}F@-5Jl-$<_>iOT@RJc)&dX6QDl8X+c(m!ibX!e9c
zy$MR5&Y<%DZbroiF3f{0^r-Ojw-jDTu-onluq!jJvHCSs4yn-rQ0Yyz0#<8l#m*A
z9!so_`VKW8Q1`pmT2SA^>4kc24W;CzbU%@N2eMxiy!wNx|I3B_D&d8DT=gue?~K`p
z>OWam{Yk%i7!^$Vog1fHDjq+2#9Br${xpjqJU!_WmG48_E+Q`=Y(Vs$QV1pG3D^b_
z`Qri9{LriW9EsmHmqXeK8akBqec`T+H>$^-aA_C)w
zi_|@t2cE4ZeyvWa#FKVwi?t@>TlU&NAF&5f%Q4l&O!E}&ZwFZbQaPk
z^R7%w`C%+~d0z_;F8^pAOAEGdBYANic;0g0CGSDZ?`a%%r|~+P^q;vj{9l>n)j_h~
zW;DHY3AUdHjDHim*L;i39oO?H(>7O}uVbAy7jB24BkPoK_-@j1;?LCS)oxOr)${%#
zY`-rG{3QMtSe<5wUvn0J^}ASa(|HpeY^P%87q7+Zb<5wW#F2J(K1}w*_Qz}^_A7|L
zi3hiHxZSZ9yuOvcsd6j+zvjY|l}h-$=+=sgmF~dnBu`dVNdQmE{(5Lu6ZorEK}jgq
zUpQ9c7CvVnlq}e}1W-ZGMFI0;6;%tb8}NLFTDlL;cc^fm5L@b>{~0`l#yG&P1$l{Q05D8-U>)
zvjuoP@d+b?oHLE6^>f4F{Q1j9-N5&vRRYX}K=0GMBMy>K&Wn6Dk&jTH^_BnSa{;fa
zaygNt26`qbw|>Y19F^qN^94LBA^#TFKm6j(unx$tCuv@k1JLiqr;9>>GRvQT!Sh*9
z#J4vR(DYxN-6Oy~x+0-?-lwd$>=xGVg2=miysjF${G>eQeY3Ak4&eDUOZ1U*mJ4=v>nFyP{QhOk
zvD5fHzOvUz&;VrLC=#U(DE45kX)hq#kLzpk^Lnnj`!GKojz{N$0sWJ{1Y*5a2UVZn
z1^jj`nVh3I>pw^ShiMpK8nu9UZf2WJW`K^n95ok%o+gf&HxC0^ZTw8z3h{Ae^)IC0
zz3dqH`wjcOV%{I1D-ZF~k^XY)00kuTnZ7~%z8&Zx)~lny)s(c;H|;N)5MFtYNfPUQ
z{?A+lUhkcA`Yw$1Im&iK9Dhek+1pNHeOIcbp9}|deXsHe$M5d$uElyuFnmq$d6VTm
zzS20n;_IQim{&(NX37`<-i}zh1oPARS(zIyfb_fjH@Jd6cj&aW*xz(iLDK3ifE#_M
zmn%YgqxhJw0N@?V)fTaUiB68a?*aFu7+=GDT@yxZ=Fj{??D;7rCN+)pA>-+jdT{MxFZkM%1k-Sr*6
zZ`b~{r*Zw<=QnuZ=QSExHsb!GbpQHSoX(>b|KJkf4<5A>xP7;l30R#2Oel)B#raAE
z{N07+|F~tZ$K#IL#HSoQo_rmwIgitWEb4(;!2dZ{W!N!tsSlI?rUn-WAgm|<`V%@h
ze!_~A?30(DjV9-OYrGBw;`r|2VGpcF&h#Z=9ImK3#)9=pa9eW|hxaoTko|t|tc@%1
z_;<9nn%oa|^*cAQZ>!rCiQP%AS~7^`12m#7@$&*^cL%XNT3vh&$M?y*ka0Z5I&>3G
z*Vfd&h|}5L%8_xqa95KTmXkggo%)iLaHM9+Ncp6AM+tb~`R(Wi)=@>{)!$jNV~&Slt}2axr6>3uDvJ={)E{8(|j
z4v|}N*Och_!Fw^W%X11lh}@KwJ+a3%lhpcpMnw_PV=Ry2tr9EUNxIK>VInUTSW3bV
zuKJVu>ut{<`TGk}`I#-I?E8Ur)cWdCZ#Tl*8O@v5KJZ-`vN_McQauNQb>9aaC+Y}ESp#(&iLcdJ^8_c2~l=|;6wK5tJ^
za<7%B`NRfaDm*+%$q6m2gFjo?AGpdy#g_|H&(|;T*E7ed@^>DkaN|J=!!}a9Hu(e<
z-nYQZ<@~6y_a3Ue5@#w5i?F0VN_QzdYe69_4wLxr3wU7x;}_-&E(_tJ1%!{3$iqiN
zuwemJ7xppW<0bL%(GdLKb3nRC_`m|fOC#a`kN?xBbwM5u`H|<3F6aZDm4xBrCkT^og4-A5UoPnLFo=5pjSKTA`*dplU_3{C?=o+x
z@B7BWyilx{%EvI3dM+i3nn$?tQ}QhVR5~~pBjrt`QmFfcl8;)LZ!|88Cv5rDUA=&l8cb
z!xic~y-@g>W^RC^4iQuD4~C#iAZ;9KhdnDs$d$v7Kz
zDwFKHR2SKe;c>fK_XgPyt`X)Udj4}7AmgcT&mS_L8|CgK<$kX9?7{N8bEtjD>_lA>
zFWE);iS=rw><3GF7Rf)~n9^%tP=M5{ou87|a!x1t>s=Wp>owmSsrl9tSxWDt_o((?
zrj$eU(Ok<$%F#Z2_ye{dld-&6*j{re_kYIWBYZ)c*q$d0{cOhT_{9xMFcbhkY)vBL
zbXA)l(cAm@FqsFk#&a*n&%X&#O2l$v=n1iZEY7hBIQ;fz*#O3;k(q%w|GTArB;DHf
zGqG1G@zl9g*X&Y1ET=B@RvIJcYZ|dfj;r!WJM)SAk@*yV<;5R3-T3qmQlHm0CVqIn
zkma*k8S{ymI`&#TpRzssZ}J%6xk9c9%vY8^PA7irUh?#l|K|m}mP%mx`}*RBm=BpD
zn-t7HV`R_4Ck%QYW7>~#-&y`)oL*?{xeW7xuiAWe7_ARPSzK!?7yX*aSy{h?nAZ73Cy#<
z7bl#{1iTdXrRo`=#G>eJn8z;tmGcG59oz9h0k4yW2r#e5JnQf8`l73VWzrJdn9sRz
z8n9zM^WVxi;rH{1X_XFsueQ9KSunDGG&k(u@~glMc)GyG|MFW
zn}=StGVuFQ{FJpw2QWrhR0HeT@-uuGzX$8!SL7b>d_T`j{60fZ8;;@k);oW78Rmb*
z!#grC|9wV#z=g|8zS~`m^_oy*Bl>pvuTaL}&q7(Ja6KAU>?P;Cn@xX_eZy-8OUb$~
z<98l6em-Hj;%}_KU+}$Hj7N9A{fFz7y(**&W5yLjE?vNjZ(gP2^Cka&EMAZGUz-1P
zInIBPal~o7?)<(urIE~6!E-FuSN6d2M*RG0`MQ<39lcaq&tdsUUS;A3X{X{{a!=S;
zh4UXApXwK5jms~)RYk|orLia~VpRXL={5fDqSCo|a9C^cM$$hT?%R<0{D^%sId5vW
zbbcj%e(JN=T3oN|ySxd$Y2bK@;}ZqGy~6W_B9mvQaQn>L+aASwsx=>&#`zy;zL$&R
zAIG$EVSJ`>c^8h~%B*gV%jKmTHR60XKXDJhSTwpg6UX=Ns2j!kJ6yYIisS#}$B}jn
z*7$i5$A@S9B>eO~vLNfmZmU^!ar_$Z=roK&&g^7dv076@#ue6c^Q1rJr!1+(={z5T
z$hdJahzlqH^cO1@Til)jk4$B7yD>*?y@BITn(G#vjv_GJae
zv!_jny}BzVOUB(PI|G7=%uh)@wMN!wU^x%9N+M^h=|aZq!UxYuelgO?ojBe8o-G;I
z?R$fW+{=Rb*EoK$Z!;<1d56Pu9Ik%6mbA}b#ml4|O}X1-{EnqF6HHneP0(<$8cENH
z&LiQqm4yTaLMS_!7f9VVy=8m{iGOKM@rH)jITH3ue?@T5-Tg$*XU=IP%qPS~^tW3{
z@%q_WYQ2(W=`Ny|#K16-SG1>icV_E<YCZjTEe{E+tfuT-St_+odhs=7r^3vs
zb-%hq>R!YN7Rvrt?*2;U(@5E|#(k7MK2S-`3&NID>82@EdcxWT9=(sUcWlq8u=oUZ
z4yWiQ^?dCDpWBv3*;B@Ps=Vw{inqrUQ#|3FA(hW=MJl~so0@l6C{TRj-Cim`0Wm6{
zY8^^m?-}(xhYOXyc45BoVj(?cA>U9@>ba&B6#8jW&qrRNFyJSeh9b9@b74A@@uzO*?A+o^NU|XEDd&0sxKK!If
z!Z#N9E_8AdR$SoMAKa+$-i7DESd(;cIz|u%b%ImI6z*Bz!9xq_s~6;h7x0@wX!wdG
zI{MXm;)S@vF#7Y6sqWfA22y#hd%3B70R2`>j(F7BhvwMEUlyEWpnHP_Y&2#%s>xJW
zL7a5tRxTc(jfTLV__x86&j*l}-S(IF-GjK?tbl``w{O|W>?lb
zaG!G7QW*uf59EHlJ}XZN_zSxw5LCe1kLK(~*t5V7$1jdw{aOa?XkgxQzHBe}bNF>z
zc=ww=q;~1TM%|!3^zPi4rkd;kVzKEzymDVNNd%(Hc
zn}*pJKZe4&>0y`j+)te-F%G$fcy^=l_Gy92v|hwyd&Jx&81~V{%@6awfbrT$atZA!
z+@p7W$lhPO2VGyD=TX!ldbk?=a*3`q5A5hgkQmD{?J@8_`S^ljbfCg-cwJHchJJz>q3^7-mS=HqM^T|$-%d3x=g4VQ;O-ehoqZ8cg
z9$xY2LP#|8YDQug%1kO8E(ZT%DkkT@TWNnqcPgGseOup#29k8nv&VFyF9*}r)`CB;
z#N~B4W!zneZu(vE7x)Q1Td*m<>Qfg|)p4EFG3-QH{JwTwq3!5w&AY-?cJ0VqMl5oy
zxC=?8YOyV@=t2xVkI`Kbov7I_#!-l~1HEJ#xKa=I!zjM^(6uJK3t5ks+?u}Kg_v*t
zS@ET$6J3?#5jDvIf9S8h+&6*WbbqV%zx(fXp{OeRh!)!}v_{~Ja!_U`5*e5rY|igM
z*6}+LFHZ+D3ZK%7cY=Io4u#z|=|Zw+f8N>uxD$m)tdn`x(1A45LJiAQI$)pMu=J}-
zJK`M=NV${Kj#$}0e@cA^_s}HD`3!#tJh)HU<6k?v({Wk*zw&FsIM2d
z{&#Wk=l)^w)_cM0I#7*zsN(X%c6333%eT3y9euy1=ElX>f$q;5@~H@Pp#EL6;Wxp*
zd$Z}g!aR`<@Yi~+s2|D^R8C|z+}D8=ndF`?mF_?%%cp72mq7UjdnXP{LOZP&kt~zz
zK%1-%zHo$gQ%SdcC+69K)|GP)Zrcv|s;n6*6YD^${FLOhp}fo1W~El(PhL~5YVQ*8
z8=qz9&T7uofnGWJ7kmJHw?=%mI=Q9;m99mHYi@wQ`_R$cRJac^?$fp`6TuEdACcd&
zcQyE*Z)!aA5!$83!tlJ*+YY4H8}@jccqdXdGr05{^ez-cod=e8prcxE!cEshz9La?
zyy`m;|I``2?;sZ_`X~3UR|ojP-&7i2*nx}<-kmQ7fAvF5-mU+scA|HUr`*%_ccO=*
zXDi-313f#>#jT(1K#_0gW^E9!o9YzccD)nb(l*O}9M*~Mub46sn(091b@oT1B|4Fu
z*9=Siq3W^V{20{cOjchlT&uk?;h(~F4KYj(0-?UWW1phC2P%R
z8*c4FCv|K-tc3olIBr&2{;~@RTdwrm1bV*93Tu1X*@>P$4Y;Tv-Gx5wKib0#erPO~
z-Q3Cq{{A8s)j6d#ccL`@`}>&YI#JuexuxauUFh>;`uzgCE;L>G<;mH|E_9Qn@NL&R
z7&m@I-SOH9xQ=5e^i&u6AiL@?+tn^4ntkDD{lhLacv$Y@(WEX^a?UVl82s*dr*F1i
zx}S#1*!t`FlzfSe(EQMep@b}NmwJO(vhLqcG7U@QTU&57KA2XnQ
z>;E`eb)no{gCi4v;s5HBO^tJD;73Tp?)BnpFK8%S-A3~Deb84a%jzhML##XdQs3^T
zp~a-PL
z@hp$jNWkmY)K5gfJp^*RzyDkW|3Vz{Kkt2j|Gr(eZv0~P9pvU)wHv`d(e=0%H@jRK
z5_~H5>B;~NY4JL~aAE02Dgni5`)6tBGS{`La`->IEi7?%T`~CY*~z~$56Y8yUA+Br
z3k?~**rl@O2R!$IJ8pS84GqzQFYNgU`IoM`wdWi7Cwh4D5(DhR$TJJac0j@E>e62D
z1Ju7!|Ja%cD1UsBi4pv7zEXJ!i#nA1+|4G+U#T0F{L<2I=mvYDE57Y-Dh=&u@M-VJ
zrlH`}Uyo)d!1vX^)UjIv-RR(|K;0Yc-RN+ih@B*q+r4|!U-o$LLzGzUX%6*dWqz_I
zaGr)H7MVL{^wE%-M@@cF4YW&Cf8py};9u$8?kjQ7u9|$NNk5<+rd!*(Z-C#YJ8AVz
zZ=n5dunPKV+tN^mLtET2a~d)};qXb>m40=*YKDusGH+Q(MG
z3Vs7htgnpk=tdbE^?MW;-AJ9Mz1aod`!%yLZ%rCmp`LjK@ecQWH
z!q{K-PK_RuKNo?fJQzVoGKZ7;fLn>%{Z
z2G(1azqMn9bEe*5{App}uRvv3hdraH2l3nt-@ysq<>=ARRkSMMf30BKg!HwrFWh^u
zWYK1jJF-ll$EXMS?wTJJsD*GMe@?JgFDiPolwUj&?z5wJ9yn>*hu)|Cw7dfUna$|_
zKyB{*=!LhG=L0l=7~BkF;r|9uN@YjdKKQTgX`r9_lKXvdPm~@@l4?Ja8ZFXK9qUJe
zH5G@9&kUft_Au!=I0qYiRj9sL9Q?hB@H}X#>qC1CPB!fJ=tmZQE1y|_{P0rowmfdw
zk2e|g=dgr*-IilVOyHiJA_r#6^0$3xU8HA!6nj64TeS81H_-E{!s3@=4DizxU%B1&
zG0d|o{nbKc`cTF82P((Gudc?NN3rO;0rdL9hF9Hn14uTQ>5UTj+prO;j(H|Ci1yA}
zs4{_n4jp6rr7o8S(2zvS`?xTuU*qJO7xe?Ew?4n>+WY{z|944d)#*Vr7cKZEmeG%{
z@7c-1vvB}bsW!5mIz51VU5}VIL=K?4IX331!vjc4V)e(hT>a=;w9aszRzG^W>_EVa
z%6|0tT4yH9!2u*6o+s|*I)G-Ml;r_!m3jyBq{cw-lRrznz{Yba!
zVaPsdXuqTZk8;mJ)O(Hd&c3=qq-|E|de3+W`RcRrCe;n0rR%mm%>=)6b8%X&oo5CS
z|1zFqD_#$xHFLsMJTUL*;`Q0)cyb6?qR$x$;D_Wk!@5qPWdPk9VtKw3{L&?zsW?^A
zKZw{CsX5Pb4k5+7cZmx3hLHEP#hh8yWP0hx
zqMF@9D9=G+;NyV-#2$8RV7KQ0`qtZId3DPm@YM~zZ+;J=e_7Irrp!a=>Z#edtdJqp
zZt~_2cN866)otSx-bhEkmsgo=QKh3-(;MrK*3i+#Agi-?lZQ~R4=opdU5t*j_@n%)o#-g^d(^q5d_%~l
zxnsi};UUy=@=&%x4(PF3z~#?5Iw}u6ps_NJj`V69#&3V7qm#GBOd@eC
zzvV#CQ#!i0q+;7683syM?0&`6K}Q-5X<9`+bj0@Q>UIA03?wM(Ybo=Qjy^hflLrw`DMbP(_m}~}rHP(E;>dt`w!J+tuZLSRTwPT5h%OyIRNLH^}
z^O%kbr?0NEou#9^t8Pa)wHe@_a`6Vg^9(e-mHBdT4jq|)*%;ebN=FIj_d5J!WuRow
zaoy*d3}jv5eZJ)|1L+DT$Qr`^>v7znd)@{yP;uZ$+4d|3s(*6xugbb%6epAD&2@Yj
z5of#Jq9Y%X6c8dvEL42F+-9WgH;Uuq2+Q#<`qnPg1Pxa}`
zO{0i)Iip5jdK3-)F6gQV9YK6=`F`~mjiA;Y^z00cQFP-%S<&__qiEd=UX?oG(f@1c
zTtk|GqByQjH1bsux}2GY;EEoGwnZstA}t>&mNuuD`I=_aDWV~gB{*l9S%PU+vnAyN
zHATagDZ!H0Tk(&$K7a^>YZ$BA=Fs0J~ukbvI$u8XN9Ue2yg1F
z?LHY1au1j@Sl1Ec?YhRe0fZEaFAi`*7*2d)=C>kv6X%f3NE$z*U7MnHK5v%v3!)MH
zrE(d!8KM2lDgM0*grVE+uSpET7Pe(WK8+7rJ(0#0Q(xT1gFP+u95OVIRQBf(if0ms
zLTeFp>ukziP5t5USVXf3`#*wDs!y=wl8%yK0(wLWFhg{;CUY1U#p;4Wr$(
zKjVz-Ykv@gt2do&ClEH&h8T4cLaZ)4LF|Lz>tG#$J_J4}_jZ5HM95P{FiR<~Rb)YhS#8`Pg}-I6EN7zv2$O!*8cttT+|)kqKl-zN@HO@l@-?fcO%M!Bwb^4yF-
z%6V0@C5nkuI_D)zAxFXpyfnC(%2rpvH(+Ep-@FWR^|nI#(N6X~~ph
zR`7IA)ta{gIZ=b+MK(}Ia2*gOkrN8b7L8qa+O8>hM6;?GAY#sVg8h0n6(IlB}C)EFu;H1;he(&

diff --git a/previews/PR1624/tutorials/out/solution_000050.vtu b/previews/PR1624/tutorials/out/solution_000050.vtu
index ebc13ff0492eb45d426ec93e7a27919ae8fd5e4f..3f7ff613157ce780d8bf4829b0155b1659502814 100644
GIT binary patch
delta 12189
zcmV;OFJjQDl>wub0kEY50yZ?04FoKc(*rpHF_R4hWC1p_as=2Ge}4=B000000C=1g
zS9v^^SroqZLXstE3e_Z)CDKe8b#I$(lr~$4?8#Eb_I)T(SqdpBsjOKNLNR0wEwW3g
ziAWL|>qLor=k(Y8-S2zfd(J)Qd7ksUCsoV>jCL@5%}6P}y2Y^b
zPli{0&WqR3XV_bDf3merhvCAqne5BE7_MED)UyNL^9j0i=NpE7ue8hBZ46KT(){bm
ztqc#oY-_jz>-L~`=@Z8&HoFs+Eb2ya=$YNg$l{HN?96P)=HH6Qh3xQ(JY7T-
zu6WlO>_g>9fr?kV;16=DE)rw=)p&
zU$xvpCKHjxvM2m^Cn2(5CM2#u8uC)M>gaxmNc@;l@l*Ie+^_Vpa|$A{bjfC)bU4@g
z(y&23B6)j_GEq4qm)@rdov%ZrCAD@`t{xGif5;men;PK0!`#rM81h(i`xRl
zijshXw7)iM5r%gL_-`*wVK`3qg<2Zmtv*>U*`bc%&Z0t%&;;PYMeTe4S~L7Eoa^DS
zl3};#TPn{-DZb{kXY7}oG9*%}LY-}2WR$v!JY
zjuyN*NmW;pqvTuo^?!@uv|&ykZuWX@Dhe}y?BYi%XG$0HDtvX*glEI?#_f2708ODz;<
z^X}7(4260q{;KEYNbzXPTOIwsC~lNabr#E}*uK&C<)k3!f~b#9e;~tu_o;qf!HbBp
zS@QML-H5!D2JfO=GVC%z?WwA%MBj^59fE8KemND6HfRWtOWgErRfsGIYIF{o2PBpK))CGd2xw4
z;Ou&^{uRJqt(~z`?iNJ${HFExR4^jkN3RTb7eU>d@8;ZtbESSXfBx571v*o+Z9v_I
zVdW8HKON{prc6bAsyD@-G`ua2L7V`|gJL>x{*7~o)(paY(z4DCZW&?t^^F#-FQ8|w
zhWvcL-$10@Z+{H}-Y-rGv;O@#BGTG=rZIOAQM;ii8V&e<@=Ic+=P(xs{)89
z4?3*6`0L7ChKJ^s4+Vk07`-oCcqoD5!SlCx5H(rE5m=d7#gat
zMTDm;R$3f***P;F5DNIto6Gho0i0a(za5na^-gyW+zBtsUQm^AkBg8BC)nddD>gFxZ^H>@1v?MEz+5+G9zS1ss
z2mP_lJ2W-}I-%J;mqQ?r+JKL94xm3*jc0fEiz5;e(Uh;62J@Wjpn7vXJiq43;Ng0T
zr#0n9f1|+He6_f}WE1pTvuXRpUWhBl%ky6s_?QhRn!ip0?o4)BMb(-ztis{z&}(N{
zI8EE$1axl0&VLS+{=ApRh~*ua@6L`&i-V>a-Z-S|_tcYNH8*X^?pe@N;fW&i6pGz$
zI{JPApCeUKKa>wTwdn$PP~Zy1M@5%K&3=S?e+s1tWf?R4G0HaO+iT!ebk_4hs2>lX
zbg^Fo#3@NESFi)Toxht_vJ8<=z3<|0z+5O7Tc+Ise(25o&Q>$fQH98u(*kf0r+;;Q
zhd^hw9E!JaVkx$|J!|4M0`&_$Z9fHdlUvfbKLz>}7Rsx>0&wQ4;MrCVd@;269`xB4
zfAm)1qL~Ev6l!HIEL6pCSSHKrQ)ZZdmHRfY67Z32bL$cKMa%OGEDKyIp8RC=P#kn#
z_U0cSZNZP($gDJ^;7?|X?X3I3ALwn09GnMuoBA5<5e{>Cw$OHF`Yia81G<77!2Njo
zz2Hl241Y@$=vCB!xn>qqz(D_6r(D)ae*>=D3Z{HgU_E%#RiJjqAoUJI`JKzMARBw?G`Lv?B-NY8bYwyiye^
z%dqj11-dJs-zFAZk#O*x6*li*F96@WF^M;L3E;cwiN~n66e8}V=y{quB3=gZf4hvp
z_atQb2zx>t$7n_M=`utPPb@L64uSqyOsci2M=^CJ=;A(x-!B*w@0kyJ_+xnWxHHVp*&Qqn`qBE_;Lm92&nA`p
zFU_FCZ-1#D*aH3}_U$!Sx#ge-e@9Xe`FMff?{&Eq40AsEQ1{gqsMGdL@m3q)&x~xu
z>k-iZ87qe^@$lVHW0P(&BM5bAe9QOrIN&(cu;(b$E#cP7qs4F^m;KM?>%;wek_XL{
zV7`n6Ppm#k8IDgmHBhd~Fj`mtIv(a~mBoh48}Sq?S-NmDfImUW$3j#xe<>dDT2-3_
z-!V^OHk{~zy2ZIaNdLnh{O&c)>r
zd;e>($|XcTEHvF-1voRSAd(V325}A%?{Nb2w>{X=
zN;1-*+`3XdZ4^bwDl(EXD>uULbAErlU*~+!cb)GU-$VsJLGh_!#-{ckx#fH#jE9q<
zV}|@A45^{GwXM7(jN}ZhOb*WoLn5(QS$_*u%PoGN$34Q}E1K>Y(ZFCg`2Flb
zS4*&k*~KLVrfXm5_y)VjGO?WfpuT$iXz|Q2BXZWLumKF_+`6I&y;8q;S@`rYBY$GT
zY=t-WNe^~R7sUMBQC=Pu@XWjM`bsd$=});PP{Wham<9&2x0!o@qGO{?E5JAGr@e(0|M%yV9-0-G$os;gDSm_AAi9v
z_WJfTMaac9P6hP7&-J?N=w1HKmpA`}afg9ms|Api2i`mDfey|B#*3jp>Yw_Y&Qm|0
ztS$}?94yyW2Dy%~=lYS~<$rX=%)s+97NtetDD%W!iW}D9JJGb>)U72)pFywp*Nqe7
zV9?UbxpXeIli?mJkf*Bp`}+__`xBCmdV|qgv0VGXr&c*@Z-609dfQ9D^S7pUuYmph
z_&!rw@0o}1(nio?Ac&9NbNwPGJ1a2H>H4#J@J8zb<9slq-)8$I#DC?8OFWbOX}gl!
z-5mQ@9FI@F4td5;lP9>pw{i!kCcCuaHmF6gb2IQJ;NkLpVb?*%VyK544}o;W$*Pz=2PnklOZ?r;3(
zfi+n9NM|19WA}NU|9|Q*zb+&{>kcUWFg)H9`X0|(^J`c?deK0f2lUarT50L<2aoMx
zokCF7>0a4Z=ocjI{Nn=HFUWb54XUNd_EFw`x9LM&KXmbdMv#;OMO%8J(E7O#5ZbFxHRie2_usSXaiwZ-0vTbrX>spg8My4`1ql
z@tEVA_U^+xf2+KA42r9+cPr{)XRA|`x}4&Cp%A#vZa}?yTHJtY13#7gr*kFZs&EUGn3TmdMpq
z&z7J497uT))_)^h4}C;*wZvJ-r~DfAeuAeXE@X;Qo==;M#el0CPS~(v*T9q9(+1Ad
z%`lRu`e~yT7L9X^Cn$#VkRSa_GP9t!RaQ~@3i(8*su0~<-X}#|2|QFTm@=2*xJ}C}
zp6ZEabh;z>fcbNp^6m3TnAO{NMrY
z*O~k#e<`ggy+ZXtN@2$usz+VAcXC=_ADo)QP4U~9^;0#4>W^IVcr)dtq{u(26i-|!
zA1Q9-{yOuqf%2eBts@oO=4_ox>;JY?HfKSeBj~}QJH%bZ!uK@dj`Qp8836^
zkl)W-KKL7Q+PBRcxScY+FTYsTj^uD**UDj2gjMK00ls~(+Y7bLC
zdzsO+n);x(Sfb@Z(CB{84Au4P89doU#W2<^ozpe$O?)2aeLKrHcLMJpR9mw6Pm0fh
zD6=2XR~ShTQ+*Z{^lfPe-2{%t(|z-|JzP%fIQq0W5_ery?V@`AfgxB)aUM9hvr!Iw
zlz*LBd>Hdygz#R>p?tmaoL?CFLC3N}T1W2X(lwN)YEN`>sa{%J?6sBw?XKk{O+#PK
zdDcmFEZD-vp883^&N|N}7&i=ct)p|!9gbutK)=)?>cbGoOqn0}805(n75h%{IL78&
zrGJ2WGv5I9mBYrNx|sKPFN;Oxga3y=mrato9?UQe{mV(!O8(F6@{?X%*5dAm2OVb+Y9zRlSuNi=mw)r#E57h50
z%y>loH|Jk@s*^Gv;k6Rbm*nfeBR{+ISq0ELX@tkk)<7ytH@Uuy
z?8nMi84)dGyV{xb&q-USLGPdS_ZosfB-|xizyVX1C*{u}U%x^-@{49^
zT{q_U54BdaA#Vy_Y!Zig_Z;``5QhDbY1wQy>wj+gx5Qax
zSww=)Z7QcHK{11o0m{duZ|cG?={?Tzy{GqWO{lz>PV*%iV|lTzLfq$h0^k=RlkpPh
z7u>9H*9Y(Jn5ycf{^vGFgZ_!v^ydt?;(XpuYnLpPqqV4u)7H&SS$HF!>2Qu!B{s
zMgN3O_NHwV*EObJZiBm({HBUP^Y4pRWYRlln@!xHe@c+p
ztLjSJ?`pJkp()6D6MyXa4IG@qI5HpO8`G7_sjk+Znz$WJ_0y>!Vw}#Wa8dLZ`0U7&
zV-;XZXuuao($~c2Q$J`*w|N&wbvA6Wo$`EI`C-Q>xK1M2pi5_SyWe8`BT+
zQ>AGy2dX!ncA*0F-)Ii1G@^JI3Ui(hJNBs$V(;L0mGlir%94S&@N00000004NL9hZ4L)=Lz|vs_DvNV$p#DX&P9ttiPJC3u@Wd=M=~Au&~o{18&&D
za`S4yBl=x`|3bfBgKpC*(9&g`(*&v)$b0*MvVqrj9>VzQsr0Y<_|8QYzql`m<7~K6
z;PY>eVK!6w%zwX_cT6L{>ZM7J;RK(5lrj403*EOJz_>Td8G9GUO<0^gtN
zIu*7J{#x>pR%1l^jmjAkNU;
z@#6WkPDgv2Q;3&+|GaKH+7q`~T|S1qswlO57lZxc_DuPEFvrktJQGaWqvy2{oP4{2
zk4emk4X1JIR^0h|1iXKxF_nq8)Y`(AoHPX(HHH{C`CBC-U-pp?PU1?1og6nR-z3x^RpZ
zIP4UCbrI$1mJxFoDAx4JhvqS^Y8EO7iuN3S^8)b~6%B<`{Hg_|LCX*?=fiV{1EAF4
zl!O)9>x<4==b}A^H?-gt=vrT!LwR5It!jfOxLx+X1m(TGBtpX+jO1YrDuJb+y_IOc
zwtpuJs>~qHN1ha}+o;cN5z3h=u;YULx+$-NoX#v-@85OrW4UNt={(KZ;4)Jgl@IXO
z{kT!;OY?Zd)h$l>auOK{1r;|+#IB+|tTOqc2ztnV5eY)wTu~J^J_-BG)2|^EXI@Qv
zxfEzR_Ss$4j78Vx$J`v*%JyIDeb=KS@JD6g+9(A;t@?^S|Tw2gd!-^X<0>
z4|wDU_t5u+Weas-`xV*tHX~29Dq8)t9uq#-CCXq9qog$kT>6^7h4%ZOI1S@1;EZqW
z7MkxvvB`NnVAA^j(}EN?BhZhJ^uX++R6pYVZC(8R9rE<#xmChV_zSwlLj&P&Yk&J@
zE{(X(EZM}rKv!d#^TF`5QYU}qVcnu9lv6LjKA!2P_WwCq8+}fLzMnL@fbvk^8Kt>~
z&c#bVIUbBx*JH`9g}s9D(v9@$O~M`h)u3JIZDp$GzI5Ss%Ig8flQI*UpF(@@3UCX%
zi%s={w#1ZFwWK*
zT=V|x-eus3`I1P|5rJ*?fm8?j@#()*=$yJ2$*u%DtNEg+KD}2(`YM3844T@@(eL)c
zxhEK~kL|Ilp?UiBERQw=*$&|z`)J>dW%VsVwHs@U`_TV^qQxKfU|Fe)4S(&=mw1sS
zG%j~%hQwAdS^UXgv=36ouLtOS&$mAIj@F|p&8(t+?EZW)I#-)pHk3z$_69sj+2E;~
zq;wllsq$#&1;h;u4Sudl>tsr1QN5__&^M)dG!Kk#q&zLSzAIoa?DK3nPpXS0SvK7Z
zz<*sN&SazgY;;-4O;9Uru74c)XP-8`r3oG!-4ZJds*K!|r}a;1DOQj!_ooeAqWtjW
z#@^!xr;nE}QU}v?(h^9ohqgacTS{?0d5`Qv|77R+Vjoc_UjEtjacJHz&Z|8lbwl0orqiZ?M#|WLvz6&~ebk6KNBpBj!zgpPs
z_<;E>*UjrBUH2CbF{AUcw0UI!#b=uAcsQ5lvF&%FGPszL&7}GsFpKrNOL^%uINk{U
z(mcJL#@E>fbqRqt*?-?>kHEh*^qJon;y?R*-<9%Wt6lWf0RD<6Rab}Mzjnw_DhIaX
z_~1{fuX0|gYFgjKQw4F--z^$%O@5+1bkBtYl&4*CQy!mT$FS8m(Y~A%+Ng9MakN)8
z8BrZ>SD)s!g+Fb;#DNX}tMWy!SHc$VYSMa1_h(UsAkBX=>VGbW@-X~cdxG*)G~4$9
z>G#pcUs}e|kMqNYQwEBPc?$ND-@)c4<>BFS8%=jG`$2+l8tGd{eCK+)PlD#~QoAcx
zi-jX7xHRD#<@cZ85)9H)(O&t^T2P0V(Lj1#E@^8=wxO0n%zM(q4@R1_-WMS@Uc6wm
zQxk*gV0DzY{C_a%L+*~XqyxL>{8iNpc5l2FNqJuGX)8mz!gwpG`}i
z7wM1p*;ko#j(@d9x*fu}A99L4>&ey@ub^?3hi)&X^ESA4TAu26>(wlU0`N`BOf0=8
zR$o0AJPhUu9ri8<-#)yNWyML((T9^3V-V3DHX4T~qlxK#q
z#u>_kTm92gWANQ2Mc+-N=jIm+slHSb!cU5k4!P_;!oq%aTYJuK0Ndv}_&tWdvhj5t
ztzUCOM}x*~;(5qL=OHKl3v(ej(7NVYAf1;%@gVYtrW*0n`;h5o#dd^F|7Ob6J%GN)
z>iubK0)M+y#`lU3+6x88KU4mvJATZeItX!TJWTmKmcbSxUEw6l3Qd6AJMDK%z~5Lc
zBS!C*6+4UOE({o^~sG%`!;TJN`IK{_u=atb-Mq#|KK{a9q*f{MdHi=
zaAj8R5cEKs+~z%t+FU=&i0onRIWt<{upeH)LH>ARVY}Q&~W|
z8l=7~cPaYKe2CDm!~UK(-fSKRW=iQ+SbxJmS$5{Z9;$bz%>VDRp{XOs!qC38<&*>6
zdn{pTb}i}Jo#N)(V26QiXcFl1{Wp6jXxixHL+>@6>M6%W#GB@mbYBcAWJ++Gkls3+
ztF(drPJI9Q8uUL|-<8de`s!K16Y(7F`~m+|C4f~dm;K%7$LDikRtM>&W!s#`w14lH
zyQ5@~=V$ICs@-(oZ~peE8T6XvWxN)Cy`A5(=sv7o9+u(?`=Hk1wTbw@DCCH)q4Rd*
zRjZ3U{E{0an>T{oVf!zNf<^8uyC~Y_&OpXQjCQiW~CCQ!#A#d1IV%~$$^
z$6)^loz>$^VRwWG}X-E*60$W{M0*y|?^_2piQm%q^vQ`HOHfF@JfU
zptz<+L?hsDs#5%sMw*)a^Iy$0_8!%YFl?l8&eTZxU;~Y}8aEB<$(q-tCZ*g$^S8g-8G-9Y1rZD-6RTN+pAJeuXSh{m|tn$KO7X*}`YzA3nkLh+BM
z*ZxkTu&jIeD<6S{?roKXe|%X4tJWPdna*Nbd=Nb{z#znVo@2u|27wb}_2-Ocaqo|=
zZo3+b=J`L}OZ8YhEWVy_QGbiY@5Z+Smqr*Y)v+44wvfTfkdAdbO&Of>D!LP7F9D;o
zYTb=j9EOib$z6!yFxaVNYkP>p}ehj8+lw>^G>RslYd}yQf*mykpvk%
z@-lJ_?w$2FNf=22hltnK%PN9$K@+GvyjE4irA
zdCLT_)w5<81qc}VwtwJc*--)Co9&)2i4*YrP41EhX9XzteyCQAC-Jgvu9O7}aCZLS
zpGw}BFy>jBp#msVYiFh10yY_Xr*9r+Wp}A-2l>v+Ty=ftK>@lxr78Pu
z1sqi=F<ri
zuEOWsX*9Z>EZ?qerogv&*bTp?u&PpTZ-yd`H<4Fbd4Iy4`GxI6wq$*A>aIS9Dby#a
z&OYW&p>6k&5&KJwJh^#p{as?Xg|$vRtVSXGbf>&rDg~7f2hzPt?LGXnf2c<+>#
z^=u1|pMh#q<;b}#!UKy+7ZJ{=j`qLLiBW$lK}Uh`t$Kpl*;FSAz9a3241Q72%J4~3
z>ZYO7F3JlZVo;u19dJaSL+$`I|FN9}yI&YvK7U`$L%Z9RRvs_l*w|K=jwTAHtp+7e
z=27rJbWS;#{I_bYuZ1l+$H>#13rqi@pm)&Gw`M(!!PVf$#uK%l+EOryG1!+$xED0$yYCbd
z|9{yr)4wJslDuy#|8ev*(ZQY&udFu=M6Lsz;vx=CX$8KW@e;_yqzC8Z@UTc&5cQSl
zTJuxshxPy8Z*;zI2GK+RUA=qrBPr}0u9UX^Dn?~*kp4PS-)C!ATpXWF!FVXa?({tx
z?c(aOD+te~Tk7_?=W&ov(hgrUg~$4%!GBp>GzCN`>Ni`TBk}!erZ098@M~;X+(x3i
z%J~v5r;0++(TGRYgy%OO+)DhH>|@{DAzF5xLhOH5JkG5rzEwNZQ+W@Itlq?IGc5_a
z6^p0sJIBLtOW@q-KmnfLcD*!>5^!o$c|~e9kG@W=TWYr%Tzd?EjwgID^#
zfT%bgVFA7m8>0nmW`>{1+X#qWYOe51kYH|VbAyHe0p4I9=UB7MO1J!tz(>W~>)`jX@=)$R16aYRqO
zfmh@Hb)XR`D$>bJLTqc*fw6
zBw@jcY#N!8@rGH|6wLki#7`x2&B^?Z*g8f`cDvpzb5)xL8Y0iuc6SsI3-&{nnu!{>!)|iF(@`X
zbJdOTG_avPH;njLNRyZOx)dG-^C(#v;zyTF7Q9Q;AUZUDZM={8g8j?SYquVup|P_d
zrv5aAe+RN1tH@lJY_0Z*COlE0u2`KQe2KimJgg_@vG-NYeSf~3fvn)>9ZKqy`?r#a
zCpzoH9DQ}tFOAb&4XY>#+*Q0P&5vuBObytk#cmp6wBSx;-3ehFCrE^9+F54>68_^6ibb9+d(xY9y^ije)MWF?Q2
ztdh(UGG|$|v!3e2KPzNs-pz`pG5vvmOU8EM4_hDAB|oFYrl!FmL!W_DBB;h^*4_v^?k!-v1C5-w0x_fC%|&B);2{9P*34;Ij4-&lc35B-^4-#KK
zwCN?Q=8~b%iR6AUiftQzIgEl*#{Gq>zSHpB`cuk^Gc)GTRB8{PHpKVbL(QwjzWE+>)9;>pKg%;+3(QP
ffBlaHzOqwR#1kH-l>w@i0kEY50yi>~4FoKc(*rpHIFk(oWC1v{as=2Ge}@bJ000000C=1g
zS9v^ETNmD@G)OWe4QO&H6sb#BtzF4gdWU6T3>~DR4?cYA{d-hs;t><~xb55w3`x(9a8{)s*BFLE&WQ?97%e<}w>)k>#l#&KZMMB{!w
z;DC1oz6)k@pgHEV>BlV&4qH{U>xFR;viMWG^92qroV?>TJ%EE}#ixb2(;Q@T+6)FVcdwRu2r!rm@xJE#gM({&`;X8*
z4(8j3`dn$^;HLj8pV2H1jOK|v>zGUEz7Konjtnx0caBei1_lS*f>-L6Gw4wc5KVi?
zpjN+YZoy*)(QnrD`qnVmytJWhWg~+EhnB9}xG!=^e>&P#!XQxJm|yP}gYeD${&{u`
z;`;v4U$T$T+?^}E_X42@jdfR;suQ{}PFhlPHKF3uj)fGiC3I_AO@`n`LPIhxm1XM@
zDj#w3so_pSt)r9M`gRiQnU5&D*wVJMkEXx#bF>bFw~
zy|`?#y>uGpBXjK3ws=BqqywXR!U)w-KG4$sfY8_xgv8&`jSwI
z#}lJ+xUV@k*<6eJ_m_*r6kZT|(^6x665p3h-F~p|IiVs03v%o5_k;dzPp9zzX1vSW
zdH6s6l5UG><%EW3^m=+?ei=fB8{XX~^!X;w(Bgc&=WmD7&(AQwPr~N~3kiLA#BA)<
ze-lDq#FdK+;kjXTy;*N?AE6kx!myaoHzQ%eq3MKbe37X1zlMG2%3k)?i%>bs)xSRQ
z2+i7Dn>k@dXy<8zc6Vz+XLN2-v^a}>GLb*4YeMK|@eXfC56t%;gJC0lZlmhoc@+Ec
zX9iEP(~i*AztT-U=wSWC1RdPv3H`WCf5D5JWbkaS7kgjLAj^D(a%cgAzwcPhlB;8&
zd0VJZL4wd#$tN^7P<-xtYIAWSgIJlDtCJCLjR~>|_D>nSFUV65ibecg(Z2V^nnA-Q
zA6M6<4D7;gsZc$j=vqL|eaMu)Z==Wwj#S_j5<6pADfyud@cuV7wIq6$%CMs1Mgt?S7-qyb65V
zCW_~FWt*&Di#ijNo(4UrAA(!Pf3r>a7@V=2nETR+!N~sc1?N66(DEqS7>RgSJMmVc
z>j0r9xo^)nB0kNFvd%ZxaPV%2x6spI4${pg7PX;%$V=@%l!Ez-NqA@T@IJx2cM5;&
z;Gny_Bjn6B$^K}(9-Zu*Wd)!ODgG*Xa)y|YCT_lordR$c)
z=sNR+>&4p$?H0e-;()ja|2k*)jZ8u#g=~hb5r;d2TpreApA^bgKdQOL!H+%vs%JWK
z5UH(rz8Q6~U0LL%ejW$ke@WR#b$Qe|v;F+RcNvHsH~o*v5n8oD@oBaLp?;M@vjmX0
zIpszbxrl3bi|anF$d@N-*J`>E5577Uk9nA9$gxv~OHu!oG@Pz_ALC$+&GV*c)Nctd
zcOR<8K>vDxK?LIOxvi;VmKNr>e#5&n0fe4-_}XY&0il-lcQfzdf4Q>1Yrl0qA~fd7
zrd~B025W~*y>zh;Y0_meNhdk@rhd}$FvdA2v0GFZ&ksMp=Z`*|Pm0!A0S!Y8T5dFK
z7(hL1FcK8_^9J(LYu6Kkyq}$T(fZFSJYQSi?CKpt*WOSN2}688UKm&IHps!EyG6o2
zXE=~fH5hp7#DSf!e}I8b0S8%ea)EP<8E{KdWiqD|THcayxXclCusO!eHj7ZPf}O{t
zF>i;mQzZ(>cLV-)>+4V#qs1fSdyz-9HFs|;VX!+@ELbZLaqjKy#_;^3YWF7oEa#wM
zP+{h$!wfF8E3{uky!i8qJA3>HRg7q^X(%H!^^wvxXVg1Kf9VYVvN+`TfygF7L&U#^
z!R*Fb23}@8K73Og@FxTxh{XKOs)T2W;M_giCR4lWE%tlTMRQRM&M&cIadB-z&&T$S
zeU3q$^%vF@#&ehJ#VE^QoRxBy4EnqYoxVdnzzFBonil2lB#A*))a&exc#c}WhAu_i
zN8gb@FOBy|e|#jhQcaXWdVJg1>K+DCFPc;@p`NwZOk8fvBh++GqFoHur8MxgM?*59
zR)cQE)&^J?)p_5SW-(w>i~If2U-aCcnYSmFgZ>MPD{dkF!;UvK`{Z*lrq|-M;3xxd
zGBQ$AB9yNrQc4VY`R>=KZxG@;Zz{v12yqg1>UL;0f7UzM-cLdr^PU`@mLiF|)-;q+
z;*IOUjXCy_pRn%z$$76aPLFq_ff&}OP`|~h7UN|vw;VFUy7`FlJyu3MEl-c)c*wVX
zueA$ZP=9#Ydq#etPAIoeWm3$e(zkoc9`z^MbaHEt7@+~dbvdfZ=!<;pRd24sKK~KX
zx38Lme?AS_;ZXE76PCBaWGsg-n
z5;<_a>EJnlK1Z&sx<3bX%ILCBpHKt`ejiiIM1Ls1X=pyYF^h2#P-ZYw{jw*#*^%KH-
zocN;W*^fHA(Y{d2JCXy7+mn0UhtLNF`Pq$Q-Q?%j?n=Zyg$D7fEkT?ek#}n`AD$&=qrW^eM#3Vy5sT1_5a-y-OMMTjsLKy$Jna)6^=0e$n{C?ESe%IcWUP
z_<oI{Jg&TF58
z{SN;l0Pf+u-CohLqYdYO?knS4vuX*g`(T|OhrDrJdv@!&%Y-VM8^knV9Oq;!lc*;Q
z&Xq?z3X;KjGk><;6709FrO&KO=sR<4KDN$A->VwWA21*B-So_LSX&bP7A>ese|91C
zxM7TrG5Vh9bdQ;C7{@+LL2aUhP~O;l(~3YsODw-Cg{Tsm_Hs}>C62)=*?s}Z2@dAI
zH7&6~KU{4SpjLr8U%$9qqXc;r^R$Ru*~y@L_J~-=4D?062Um<9BD8;U3yZ>jysR?(
zI}H1?RV8Pj9(A~XVRf$-`jeaQf1;1dE=J#Bm9)phoqr2TX96~<07kQ2k;CGDveIaS*1P(5buQ~ev?}8KGV{@7j=b{h1$X#0o
zNut-M`Jq1;xc|LSB^-5kD300YM~JhXk0fUv#yIQg$x({)H$Tx&CtsY(XZ>jEY-oL(|nKNh3IdkUBoFo-LQJD!I
z)2jPtL8Z_z^TBx7cjFnuOqrqB)t!RF%%mJfE=ORPseh1Es>T497nHuaz(33sE}85Z
z;TvYQ&yl*a3$*TEaDEQB(!Mk9~VwrBh4kjt18ywCIb+raNSbcm7V4CikoES7p>oy=gzY*Eb59T5~z2a{Vzo2o&!(;v!hppGZAB@6UV-)iXr
z%6%VcTMBlir*wtD-^0Cg`9(p-U7NyV;3lJ<(B9uXruLXncqZ)DN+w0!gI>)xHhJh7&3yB_$Qujt%bu@U317<0=cx_sfh0PxnX}J-AhT~V$1g!cjgA$%z~`JedBBhx;cxO
z&V~NOo{XoopYG8VEgA4*Xr+M~I24vX(|->%P&SaZ0OJ-~T`d85tP{5>Z;}sHvB{rg
z#k`|WpkKeD?8GRzU-4oA?Mv@ugoirhA2pvo4IqyWBqSg81`ji0`1XUPHul3?|!d^0dz?}RwdiW}~fVSKqA-YeqIZoScz#^w>Pny7cowG~}!HiFK4u9v6
zmmgf>Srku)<@|1zSpQ~pe9AS7qu=awH}H{6gWhZK4~J#J%P}5f{VjAEnB&JuC%eQO
zQ3vLM+6mh$C?DCAJKE?R6>)Jc;>gQ0#~0n}gPxVlxkLG#Y$PWh0AAj7a+?atCnb%g
zLC#B-k`5Tya_hb=Sou&-g5G1_X@7zLHDdnzkixtwQ0YNLyeITyo(-1&z(0EqH`W9C
zJ%Ws^Y{Wxk+abMTkl}QvVhi;1lHIk>fLu||tMinHSxN)+ZYS(|msr5hP@})^K~T({
zZ!gt}h}|3V=7a0BDr1KcpS9T@hsD9k>pydPF;8Ci`EefnhxXpjp?xqa1b=3jBY#;^
za$L%@Mz;t>dyEG~Zwv^;JU^QYtxV`&_O&i;f}NdSNyZ|`TzS8GWB3yg*;1wiS;~SD
zLg!y!V!3%A^dix?8Y-+HLz_zuWw1=GrB;s5WiEBKN?Zpx*x
z+tA;r?TA_hyW5L0KktQ~nSbFQ#G9Z`W!K4O(*5|g82$o}N1n-@4VF!seLoHwHlMId
zr+O)n@}UcyX^>;G7;Lg*gtM`ap9xDM1Yq~h&n!2O>ZzK#>Nm)-y&7V4Zf&1b8C5W{
zQZ#iYsM*D^h{yO-9d@=O_=5FolHS|!aF-|Df9HjBmf_&hF4h!}>VJW=n1=+|zR%W}
z5BhsXE+Ry70p~mY
z`-9$(>lR%e^|Qx0ZL6pcdP*l*%Y&x(K1@+vugnoRPt*vHo2GsK8T%^!0_Eo^p_ya2
z|AP%z=KewX$vtB66Z*R*^LbRCWkh{Dxa&HjauzOT+r!|X_x`#JqU=4qxnX=7JoO~fc~vs=N>yyh_lG5hvGUj
zMpPf;vp)|6R8!oYhPbX27rC|pid)sm*9=9l+_xp}FPiUhP>lW?e_qXMrgMy>Zu@5)
zy_0Q)Q`8R_Rq{CxG47dvcQMt;xgHS>ve1_m8nsfK-GyxY>7LXgVyEjNU$Fn0+yK^B
z-q}rdnSaId4s&T;*|m=;F8*~Z|E75-B31L`C=S2lo2i~DpY^uxgudV0qn6@6Vz~Sj
z)lEhBxE%7o=e&vxy_3#Owz@|^ug-z3t7zSE)n4kSb~o2@CU8#zt5nRWu3XAhFrsz6
zJ2f<{=zo&ZR6+Kml`Bn%mN9+ZEbx5Rnq_qV1%Hp;1QE+w!qwE5b{jCiI>4VxZ^7^=
z>emyMPq7=R-mwcG)DeqAHZP)mIv)6zNc}OVWDoTPwqN>eEz~!ESvR>3u*W>klitrk
zU%%q*l$SbJ8~ZW8aj3H{9kO}ET(elrt8on2DGqy{dBt=;9=m2AkfF`QhXierR~*GpntyaCZRlkW6>W@F>EX}2NnUh~g`CI#x_y2d>uzw(4
zmj2aTsVz;K6we*+jp>~oUC^VZhH=r_a~~)!iP;x-(0eEqIo*GW>fdnx73$Avv%JEo
zkLX+K{6+uwma%S8szXa-on5F8#>@F8QXTnuB^5
zSd5j?KgCb_MO`({cZNN`*c`n6Du38>5F8X{9+t$oRkms+)zue?V>j5;C!C5RM`<0E
zv$OxDdU?3^*iA4k%>Roc7*ii#Nd4ehwq0v1cs6{zo8Ebk+Jl}EP+uqiGsWTNSclUT
z*0H&}i{(fCL3Ps09^=g3?O`I2h1-LwO{gCYg*!{aE;I41bSvVn-W{COO@H@MwnomF
z`cQFY_a|CkuiIrI&Q;#%B3lpnSEq0_{j(0X%PiA@pS(Lt8pr5;c&7$ZKQ@U!vd0tl
zw_3VhxPwv`C*`Q$=5Z&bM&RE*WtHhetj7z`FrAQvLjBm}!zOGK#V)QdvK%P;x)d{pSVrs;*A(A#u#B$l3-L6D+}gX&SY;l|=*Z^{Q{!_ZMv?P!8-HxUfL)`O
zvA;)*s-t!mo&%Lb>w1Hs_dhf_^atiAx_&9T4~j9YE;3-JKz+YlIpl_+W-jm95u+3Q
zE+-AySw=1VGK(s~Nd4}6Gnn_mu*a+nv~ZgoX#mx(%li0&(g*+95rOrUvuR`c_|A1D
z|CnLeaWh)WGx!U#gTmQ=v48KddTy1RpCd*;xb}yeV4fb|W4k`At248z8wao6PxY$@
z8EJcXzk?yN8R>ek?-_GcToaTJzNU~5{ozg1KCiI8LzX+t2J)oj{jhhSReQUd2lVTX
zEpcwa_kVSt3Em2Qg-C(C7S}grw5(Q9gx?p
zbd0Zro#Bef{3Xz9Id8K)1ADoTFX*;o{M=US8z&G~m6c83g(0^*oGb39{Tta&W`LJ=
z>+M+v{(irP%K@C15gkJ74A+#79RnZTtWR-(eTDin%V!~LEG>GU1$mym&Z*;&z1@Yz
zvmtZGMCj*&;rsIo2Y?$!Bc^|Yp4zmW-Gl~4o2zqJ2A+<
zbGpn8$fCt}E0o|z4D-HTDD=ET6?p;BKT)1}BL%X3GSmD6`hK&T0LqI-a3uQX
zxDdu?atgcpAn$2?C_~>X4EGz9fc~l3gM1!vhe_L=1n_HqjepP-;=&o4QEQCX=Ox-?VLXB}==xi5TW!@T*J0yhW!lCup|6)PFwZ
zL(i{w8|dMc9ryuyeSYa&UC04>b{`uNrzc8Uz2uJ}m-`9@Fj2X%ISQ0*<8GpSKaNp1
z`4eRKulkerTPOT;5eJyOp*NnF;+1~Tp9^d^|0>oCdv!MX-#ZYe2d}N;?vfupB0&eC
zuWy~Okbi((FNSo&B{0ZD>Ovs&k5YclT!Y_&rxa3>ARo=}SEaZ&8bl_l)A>F>x|HHj
z+!e059@Kr~FT;WLs(K7*1F9G0Hy+flUdEN`uLF&P9x6~hcc<~UQ(PmJYl=+Co?QFK
zHQ)wjH?nZ&SwXMvQN+Vt>>nLf-J9)`9$2WY^96MtNTHn>&;0
z;9!iM4b^@B*xuEow^$KIePq$rZ3n3i3}e$~l(0_CGf#Rg#ig7poa)n0Bg{_@EHG?n
zFUGv;WeZO!Lq5IRx{~(k`#~|%9Ar9%cRvtOxpVBd@r{-(Lw&GNjQ|zyrSN0VLDfqO&g0NK|4c^
z^H;!AmFLrJLGk;?yOLlxFevb~3drZUIFsr{$4TFe_R-ZhxrySmv|#6fy^xcbGTu}d
zt1@kSmV*DfiJrZJ@iUP{g?B;q;Ds`zXMc}3d!Pvhjs6+U56ZJ1$&&v<>*Y(RFOR2w
z%c1yiWko;Y27jI`mQ*AAI;nBgufJ@6rMe2_8Sr84!~9g&CBk1(Ctmxn)XG5KzUPs0
z8TGGBj{{4nPZaN8LH06a#C_;I>Ri6tJcIG7krIu?l!wmVA}`2e*EddR(|su6+<&?b
z%&SZer8p417zRT)altm$gx|0B`APO%2ynyzDLZX_f$|k
zx(rX&gHxLE+i87`T|hS{vv2yZn?xmav^up
zV8>9S!I3@B8?@d~KglcMrTx!@SBy{`zJLAFo~HPu
zE%19v{oCioaML8k`-j^|5hy6^&HEAhDGoJ?wwj)FE}q8urD9yZGq!63$O;hT
zq;Y4qCQE0^|EjnNif`R7QDy3<;vZ$Zsz6;%tJW{
zTr|BW)NUmP4uL6rM}3OHmS?w94Z%yL+R=0$|B4Wbng>3$pPeSZ7Ju&jm_+^BvbIMo{;N&vsMESl9M9P3JY0$$c31{>Hn0EZAjtSE5+of8V~j}OESRtB>Og%Y>HAtaL9$cYF)m2+>(oLhpCKQ3
zvfo0k(;fAsKI`uoca!>>>`5yRU+|v+4+qNInCTgQy5C0&Mt{`k{+~ae?d*2EZ;nWc
zI2-^~GqYCkf`O?oNAKdhF&amzp26-GqvSR0n7=WS)12}W5_Pea?t@mZx2oh{iG1v9
zSx~s__GmigDTc*G@5b!s)h*Frz#k=i`{>E0{mmta;wf3h^c@gdz8hIT>csULp-#oR%4R`23V@3r;iv(5>yH^sHsb2%uUAz(M=xdDg9C%U|2bMA14)T
z-k5Kj-DD|G8y7cjqW`Dh!2GI*#iFHG`GSzuqPC3fkX!F7(wR)?Yg
zbh9^b9P)oDq0a@A(I*&xt_%PG00000c$^KFc|2C>8^#;iDzX$w6CqMkw4v=HWT_}*
zl5Axw6-J3v_`16GGIPvMs|FWU|m;0
z_b?|x*ZtO9z0VTd=Xl%7>h_5d>ZMhoHd
zW!BPrFzb{LT*Dy7u`_ABBZIRuZccNX$6(ZS-Mb#@3`#y?)Vmc9_K+1G`Zx%2_x?x3Ah_#UjS9B0pj!k6eSLtFAnh;!XSr
z#gj==w0EjGI_{D}KDIomElrB7PX=jIZ%c8kH1x#HZYl2UnX;Np;(^M>Py=%wbp3;+y=R_{hYZCCG->zwPi-50@o}mS00#eNn
z|JFh4J|q5)r*uT)lHJIeS1dz8bs#8tXs6FBhkCoD!vS*X>A)a;weLEp|uyISDjml@D=g!g+bmd5_&J8dlITs%LLQ6dM}uCQ>ae(UQpRq>5e8#ss(dP(ME&n(
z`(i~4!9eQ<=fTGW_K%Fhb5t2*C!BwOCs6LlmcJixq-*7&>&6%)s64JYeZMzB+qMBS
z?thdEinG@DbV;x__TAXsS_FqvI+YYp6ZHHWmF?%jV7{K3tm@B
zutACdjoHdMTX|@AG_NNr46JK|r@aa$@Y(zO&jY^@1j+dh_edm=HymG|rA|42kom=Q
z*+GIW9t*9$En|SDaEm>2Su_~wPnTQ9VPQ{ZpVm4la;|)w5*kA9)vYM^&N=}aTeY+m
z>A8&J!YXR!5!7mq3@MgM@bplIfePhYwX*e*({2RL!|!*Q{Et96Cm>7hBZHamJBs24
zSahCl2#q(EqPCCBzU9ng$9;2u+q=sJO#JA@sE-kmrSQ(9y_q1&uAjd>i(uuhqv|{8
zdyjh(xXh8BW4Nl|q{SZuT030>8&@;fY`bpHiC`9f!Hv#Qxg4^8s4d>ZOEG<^f7%~v
z0y<{8S4zJM2w&rM`2*FlOm|}RGuo%;=417%90{f*nM7Wu+>4a|FK{A%t^Z)Z)#wSw
z2=X`8_3ley;Icj1FZU^nF0VeR>O3j@vPuIx(|M>TW$!2`5U?;~PU1JJYxNH`Z&p*k
z^^=mo9I6NAy3s$g69_g8K9aHjCP7nogz+lc-)AcppBa-$U^$TCobnHYHc5lRV#+g3
zTf;BjMN-Hf*N=0WD8P1q-;UgMx*`%(ja%%GiqHve{B^#E2#G>$+FGi+2eWx;K|Mjm
zzCAY^D9=mFt{yu;_wjCN?^t+(;NZB$w+f##IQV3$kNS2FY2C;2to3-*t5!~mJWBhx
zE^OxBFcDtgw?42+6minC?%wGJ0pC0Iu4-LlQFKb2ahUSEEZs7Hq=)Lz-E{9%S5bl&
z{&v+nPZ3mi-MjXvmqAoi|8PnjhcJU(i>4gsky!CIp{bYp+0viX(fi0`>%U&3Ft56Rpzq_5X`BU%=F%_n
zuT7}s64eVsm}(YZs?yqjYgvN%3GZxMY2AQ~{;Ymo
z0yTSIuYWf%_^whB6z9dEEoqZbyqiaMU#{v!6%huh4WkDrSBoxh)V=>uz}teN6U>kl
zE1bMi7qzg^h~A;IPrTU(;?3XTGs*{NCx>te*d4wot0t3z1LE`}lHcP*hwSAJ{@#UDB2lU>~tWTLYit0&r>xHxfE(}h{
z$&PrpkcE%8UDa*}I+qo|1GykB~ix*N%^LLGZm(1a~zDvUDCDwVd0EGm@M(o_Jk5_=N|T-}SBpUO6>bqG{vpD@FC
zGWE;LNnWP)gvTZgzehIvrEvRP|M8$4hpW!{Rl0vOxIJCD+$M{l*w8XHYgmGU#KK}d
zdY%Ib^4WfyDCftdk1wI~q4CUx)KX6YGs|CIV(moe26*
zokN_zt-hbX6wPuz_Od-ZT0*+)O=%veS>Sr}3Eih?K(6x9G7%HRyrJqPH1BX~vI}U=
zTJ)atnM~(rh1}HZxqBICmW8zD1QFEmDPEO9&^<5sFM;O&>-WmKN*>U;-#5X3as3${
z=WBnD_WdcK<@uMv+EfwBzP2xPj6|&Of8uybLUYHM*t%R@DSSScCz~&4VYRZZ=oaO0
zMy6%RBg%L8j_zZ}=sv55noMfrDaS4^-+PsQN7y_Mi*kL)VZoUz;|xl8bT-H9XiX4d
z)8Uz3KzWdy>u`BlkASP0zA-z0sowrlAT6eUa_~%Y`4mX~GtVQV{3_+a@dq~?y6I}T
zU?65J{k|B@wT(F!OE5a8c<$2w7_9L9F~W}C+t`WbYL{q#UaD4UauYtDT

diff --git a/previews/PR1624/tutorials/out/solution_000060.h5 b/previews/PR1624/tutorials/out/solution_000060.h5
index 548c53b38c3ef5404c004137b64e8a58619a1637..b926298391eac59c17fdd14b5489b5d4cca78ddc 100644
GIT binary patch
literal 205696
zcmeFZd0fv=^gr54$Py~qT2N_G5fw2lT1X-(Dn*M*MGB#)NFmZDk~N`HSwe_XiZ&^Q
z7SXD`_q$Ch<(@OIx9|7+``+*4-aqc|asRpZ^T+2nXU?2CGq0I5&p9*i&KR!MpFDB)
zL@xYKK!A&nYa0FMzyJN^{O7-bmx})J|H^lA{=v0HBqQ{{ic{|EHn
ze7!u5G4ju@AeS-~@N$jg666x#;@<1v;_k&Y4g6wU)A|or{ud7gxwJU-CW0c=bKHM$
z*8!J(d%d`3OylC33x)rdlI}it4^KxgFBdyEN4A!fn
zIQ?~Z^m4X6aKyur_IElI{_F47y)8?Swzjpb#X5u&>aB+9!5|SYQ@p)7yajE`=DD|EIv^}k$l+PNj2&Ys#!lN8hR+mYc#V$??@8k+o9NAF_?dJcWRqK;
z82*+7qp#=9@OuuS!z}W*jNyM{ga$v|AaWpWCFn4VEOkPMSY%=bqZcO1l(#1`d?4uVfx8EFOJGBBtQ)E;)%>SY`Mwk#hLu1<%+#_v^*>bXD*Ap
z;KCWlChN3)C@*KVm-0`ucEJo{5v4>c%Kum|M$7vjGvzMbz0d%Qyx&p`Q=CPt`^#zh
zS~W4M|Iwj>@~-c%QN8uNwJP{sNQftV|PSp32%7q4g6pvSC12#Cv!u_5bE8CFmfVSTEF~
zevb4#g@I%f{SP-gP*3Rj3fLssMDuVjEFCs!?-%Vxxlecn%qTXo^%cK~{4)N9O2}uQ
z7}Q4UqLDrcX|#!CK9-kyRKW~mlWWtgVX4CSw~MYq`Pq*pZAb^YrofD2lZG7{@S)ix
zsKvHv4CE`TcfbT;lhr-E@L^%RpLotje!aS~43@KJ8w_#-Z`^r{n-?gF%)5`}GJdnV
zhoRiBwd4#B@Lkh1MEQVZDM!g#=<@2$;FJ*QGwm-4a^cT`s
zu0~ZTzZk}9K|1H~)(-<9Pc=BW1^GoEy!od9T^n|^pC73F!HI{kJn%Qq+d81f`3FQ7
zf}A};?#cu#FW6w00rH>Emn32RNq28*BRyLcRPJ}f&H7&
zw89+qb1n=(WML7IRENaH_&yl_;zI_
z+7XOh1A&f3w5Nu_xr9xwHjbm~yj1nG1oBHx`gGJt&h(Q2{_}S?Kfr(92
z{lA<*evcb0L*Q2yWzhLA7H2@$@t*q2U05D{sM8P2Pb+U&iM()<2^~+qVc|t6zv8?L
z0uP(;$n~#6Ke>PX6v6&jL`;jq{x51@?2PqPdzy_2D7m>ZeflOD0YkanbP`bM~VC$j0Li%O;l|a?^RP66ZwYj;BZ{IK(2GYAvb1
zieY7R{7!7GqjoEs5TQVv8Uiy=f3g`6pxZ-19ps*)@{}hSzCms}2c{FqSHP-8zA&8G
zFRng(LFFnc%sv(OgxSCDUV(0b{$*IrsBi7e^e0Qpm&%V^W$ezo!swTaF#BIn+g$4B
z;`05pUUrEF)z5d8puF5mW`8m@X8h#F?G4)L!jl`cvz&t?B+)&PXUQ`32ShF8SHkQ-
zx(U7#>KjaDs3ELo$UWaPyrK~v
zyV)eMn;C~75rz)#W5#3sEQa1kb+EqytAlaHtVX*q`6#W%>QJ8!GstmvV0ktU1IQ`-0v;wemBXg^zsTov{G7y*Q}_n;
zq&fOH>(I^}PQP6^`d>KXMp>}`fTJG`(Gm66;IM;SoTI-CryaeJgM>zGT;PQcX=14J=$&0c{45wU@GY*~@u0Sum
zlWG40pds{Q52t)T$9@cFzC?dx%8NMlnmF_tn~^)w<2#!S<}r4i6&T7x2Z%-X!LEkw
zE#&Bxg)!^p8SGvt*WlprqSXvv*oF?UNWDKp@2+9wo(W8QqbnJHcOO%prGa0PO>_nr
zKK%ewuA0H{eZklNRE@(E9
z6g*D-Ngew^>9h=7VQg|+ld&J(kwWDz@;HM5ml`qlB?Osx6mdC&%Fi@BrS%?Y;>N@#
zNm@53pY&xBEmxFg;xEsJ&r}}oF9`<^7CE=bm-4AyDKKy_@A*q9KNN8mW)O?)eYO||
z2FCG*F{Afm6Sdds7ejgNlfS9nz_&TzIN;%UZfYl34&WN_?kc9x5f(91@1cHO|L9Nm
zu}7P4!oabKO4n0}fDpg;euRSmow*A-zXx7~5SN(|`3g_n|A##lsAV&NGA
z(hrE@{gx2kwK%CmO8Uy}{g&T*qJ;!@Uq-c4xlp36+*df*RU
z)^3Rf`n=wv2-}|%`LlTvlqa?(n-l`SYD!LUBhV;$nfPrWmp=FEnhsDqGNB?8_tPx`Va?pL~U1%@X{S{xfshQ{Yt+)=%2b4eMpOf$1gSSFQYcJ9LOe
znlEN882JO^aQ^boQjl-Z+cDz_(1t>_-4#GPCltbhV-pdH>^wXN>|FBaR~b
zJK1G7)FZ*q%!3b}?+1D^R!y4=8G-&)H_Xzv)Ghq*e?OJ8<|pudj}?~FuzpKtK(98&AUUWzy$h2$4s!|xwca{oHc
zH{*AT{S$%KtPZU}JCjq!UB`K9v|9BJelO`8EP}5cfsBWcLHNu%9kH
z-73hN7Mj$cpPx4MZ9*Kb$SuhI2IxJCGrQU^;xY8oM)Kg(FB$7k1P9LH>`c+32Td>Btl8
z_w}yDs;K|+>;w;74-S(gHIY^>%b@4$ms8ZlQUCGd6nFG9+S|Vy>(z@qg#Zb8BI8qr
zI3C#+l0i5RB_%4-P%rZ5dPt$czP0xhJ*QY~(tetP^54M+@VmMOI91I6dTjBb?YIs`
zI@bTeam&&vT7-5R?=;FFm2f#m?c
ztMI)5@mM7Ho+^#EUz2|X;yS%FqzlNLMV3Dt`-pn}(o(soXRN_@9_j3tg49p_j0x2K
z#wuI7Km5psunGEH#SC4xZ9Cm33If%s{FW}Jcwv7?Eui{KZXrNrk?>2iC~b9u<2LZk
z20@hn`U2br{V^2;s%%mqtV{c4yW=En|IMrgRPMd{9j#Y)xrNrdJr)HJjzu)(PtkFz
zyX;Q$FZXz!Q2wC4JKZ;~_QNoM-Rx>0K$j^ocx7t|!!J^X?twg*?<=*dmB;Mk1@;U+
zcm}6yY$x4?(Z916fe)MbTt}n{``c>9p6>`#Z_;x(27`P}3T$^kd4@Mo{dXmpSY?r4
z!pwd@F>EH4S38(ffAdc#Q9k;hHl?>uJ5c@RVmLNKd*>D~e*W45Vz3iV6U;19!NFrU
zix~Nj0!A;a9;!h-342Dq0nGq^3}z)#IbOyOcSrbqV9)O{BUjgD^b;s>hy7$R)BX(w
zMxLkz+dk-9L-q#k@Pxr<0se3fo-81*;wD0A^j-v72=A2_FY=Uwd
z6oWh&hY@&K#FSc4AkHS$6wpH+2-QH39>;)Q#?J!!_BaiJHx*~-V$OL8?yIO=6=4^Y
z-xOr}X}ON6w+*Kq$n7}i#*2U&(B6)%jK4Lz8GQ@v4(R=v%g~NPjGjNTAg9%#9@awM
zfkQuW{DRG$+NIME$^+0W=r6|Ug8H0(;A}y8HmAIqa<&MmE`@@+(Q_30pc9KjvE0W>eF`WT^y~Xr3
zP$!N*e4KIN{m$4CgRC6&^Esopos<7}HemQan&=RlO!&;K6Y=Q`@6gSRv+OB`ue!pl
zpM64KsJ(!-O#MT5nReztmJ{tb9cSXehDJQm!91DD$fqMb2L4EqJ{LUD*AFyLF6Aw5n@@#Vwr6a@4
zd{Lf-CuKGDgTs(H%D-e=*gsc><{a#
zX}M4_q2=0B7(D*`(oPrv7TGR3gW_e!u9dXh>pLbIAU^eNru9Svm(sXryhV`eMO|ga
z&1BYHT3&E{B@8(1hrxGX0N5mA^C<|tke_4CfDW_Cfz=0LfM9?1yg+fkW&0$Gw+*jc
z+mCX&fB`u0Kwey%~f23q_B4NR_u3!Axe6ht>}A)ZXN9hhd3*&pi7Mf
zw4kC8HYgUkt@s5t2sqc*k3FRJqueNt@0)W<5b!t5eG@0l8BZBzA*UN!Ybeer4&3(fG$q2jnV-5
zaQ@a-Z14EP`e~Tg&)QaIbXi41>XJDvppJstLtq{+VH$EYIywOZJ>o}GfLc{
z+^@;>yAsefhNgz)fL{%cRa~D1{MWKsg8PB?EIA|k3iOTaM%|YH9v9KQp^3OzOkGu*
z7wk@Ss`};+{Jam=r`&(TJ&V&fbqmn%KAPXCAj&4=vWjP!KzVkP+Db#v&-(4w*#!7F
zMb&6a?l9cT4J5^60Itrct$uhL^wjSxZ@}|V=*=CshoO8$ZBxo$;3xQ;6gJ^ulU)lY
zr&vP$Sx%L$&fxFtfq7{ie1MM=r6%vDIOq=k>)foq(P5%iLwup#Q%v2VJp%euxxt
z{g^Nn#(jV{WE_l7R8mf&2KdvMvP>KC^9sAjbD=;tJ@dPQ=bwIUDal*V-msYFSOn0n
z&B15pf&AoExpC8hYPgpO;e6OQC2&j@=&S)pv2URFQgQCOe(nhee-w5bx?TqQ
zc1BH*4CtLcZlAUrXvwzV0K8vPTB`o?G>k`!XIc=>!-$QiC(M9xU9j`aC26R)!o{z0
zA^4dS6ui?2{5Fd}SW*jAv@PDd7|Q1@(q450=0(3mLUuINA3WG2kLzpEg88}V=jXdd
z^qyu_!2OdbKecObKt9l-L%Yx5z0gmCKl5*f5u>YSPIy##-mRG5Mumz4BlJD_Kd#PVa@IP>Pd5#H~tX!_KMd4}On
zKSD8Xoa}r;<9&C4a5%<^M*~5{=;!_&Q+jc|dl|}cp*+UT8#X95F>$Gfj2^6mV)>ou
zXN8#ZH;kLA4XsKjzbxW<5%pE1j;13Wy6PT_@x;VMOANkhgPZYheR`M0~ilY0&yomV@SRV5?O9l0-bPB27oyB^z9q9_ULX>fE2_qD@lq9X8`7{d;1fUQH
zt(bFKFDFl$4Y#Ezw?{Yu{Y@=jOnG-R24~HQ0(b!ORmSjnKu;}{!D)#mxm1476cfvk
zcRAikX}TnHu3UYX;dKSvsod{oE7hMCpGkS4rA+zQNo6Wms`=H1cD%nM;D@}GT_L5{
zKd002aY#Zy3isH$O!;2UeaHe=FbDPC>O(P3g(EJj!Yn{~jB`#f
z_hm8@157utTTd^5AntMQ-L60s!1{J4n10V6VsLqi
z03(kwWX9p>1Ezdv3oK{Qx9ogDsmu?i{T_N?gS=%u!@twOjzL_IR;BH(He=2OmFpQl
zG=2iSh4PqTX1zVP~{filAKRf1o&$`6S7st8GyOY5M6}0~636Lg+ex)?B
zDgWIPmJQgQYK4#K;2w`RiSpI2@ZtjYJ^7`S?~$2D%lrHD=sHaQ2U
zR?)cjJLt^^)IXN^^*zR~rmbQ$j(Wuqm}sy+oF1a~zB?+sMZI6i7Jew7xxWIAtSs^^
zMF~Cto1A#GuNwJjDqMTaejdEYF?UwhGHd_NG;#pYH2)zhe3G@ND`HAgguW
z9Nf=DZ!YmwHJwaBoe)`iKASdx1EjO5u%!3dmJc
z;tP8KpL7{D$sk@({~b@Te*!m~gs`3G0dpTBu06bNfo2-eAM3JiWB-;+|DB1nFtor8
z@yO+|5zlhq7x;WPZ2%fp8}g?B=)0JyJd#*%d%A2O@VxuQJ=UoRpQhWE(x!)jPxHh^i3FN*tMt^+)Z`lhg
zPrA~_B3JwuEvpA!IkRc2GvKs~G3Sq+0Q*iixz_bV{W}(KkE-zlo_#;Gh15mx_q1b+N(!_a
z_NYd3a24dALi&W7;M^c;c2;%%MtCpgkudXWJNv*iSw
zkH&I(Um3x^7JK<>&|z4|nWu-1<>0(L-MgnJhefpRs(HWM1n=EWY+rbMJd2Q5HVs4I
zXPu1cIh>!|JPS6)!MVkCsh=d?OR**@?H$2$-Px>z-oWb=i?_Uhb7ixxvBPPYm#)|3
z-6q0!w7B!CdNc503T5l<%E4~)`p-G{VE#>bUy^nozVDKjsF+q3Nr-WJcL(&o9O0cj
zO9$}g;3L;9Qt+KsI!`5cb(6;jG)2t`;3{2BSG@?>=Tf#TORMW9>*G%DdHtcAh>0y(
zUg+CR1g0x;8>Vyz3w{`;Vtje1q337=luOcNFXDUs
zv0L14q5X<2!kw-#@7)s|6P1A0&mVD?hw@i><@I=;a%k#`Z3n$33%~Z$&_B)8koA{9
z{`NzV(`L|5zG%gb@gpxT{Z7YOxW9OGBIRrg%=cWy>Pc`uJ-Sg>_bBKkm|T672lINs
zp?7HUa5p)7VR7KhNYLA`#YG78JZ_q`;XG-!t^KGE<7yWQ*Gpj6kn3a@o&&DEcPzv_
ziKFs3YmA>V^GDS%?uBgsosW5eKaM+RV4OOCT5>hY8|V8>#XQB(J+o4J|8g$STNC8A
z$33m50)3!;uMXp--tG6Z@E+;@ns@iG-r$)3mq_r_%>GKJA@Dvu8L`;E$P}(5yx(*0
zUCo8@ZlR`a#V=Tw`)un1@qY6|R-?>opuBc3{(OM`PRwbEx)1s*4I}2^Ib?f)k{ZU{
z7<;$V7+2>6pQigtVE6*q*x~$o(kT=DoVG0}0rh_EAEo)U)V4E|aX%@r`C?1QV{$vq
zQ{`-Z_#FM`bG!SZ^Uw+WzF6=K74<>mg9NRN|}$w_t7fc27e-HiI?N5r|57b9=ttxEBgQ9yMo>MJb_r}@ojF1YrC@>OlmDgLRqNuhXY
zk>xn5KhwJr03*z6V?uFJ%{hPppr7%Tc@N4)R+SL+y@x`he=i?w(BT
zZ+b2&iFPu}Q|UQshvNs@54J!h9mkA`QdI9tr;rfptuy*-fz&eFgy!q4+#@O8+Nxg$
zH#UH;>@ndE-~9}89#FMmaOr7FFaz=%q2LbCv`6P@zamPZJHX3rXK?D0@-EuXp&2mD
zAeV8uOKI><8!Att4|^cr;16y>x$2egRPLkpl8#IB=e^Y5mPTWMG0^fl#{XS68ULDd;28(V@AWh1
z5Rv&zxkej<@9*$2@>jnlX}eu1jC^Dk(=UfG7;dN^tRdB;iey65W#x}W5&O?nLj8W`|csF7qqYj9D;Yj@)xLl^o-~X)&<;!%RDQ@H}34#GQz#u|cg@7bb;`}m&Z&Ymf7AOk$9$G-*XH?jN07-xRC99)R|
zp@4&JBX;wh3Z?W|8Jg|oUBrIsi
z^UiX*iuIS=@#9Yb`TQq>fw(Uj>A$V|0n}wb`x53|6-^SdYk=1XUF(GVQKj}7`x&qw
zRtTNgf%{srs0jp8z|VEVJJ4@2mnpIsmnBkjXT-yPHMMWeB0ML!pVDa01-+%lIk}kk
z6|mp#UIY8l3!jM-5D#CkF}iyb%41ds{8MOH3un75Y`f3;u}(C=?vy8Hz^7n9y%
zhkpEI-K;4Bx+2R&$Oh`CIz0|Yymz)d(6en4ypxw%A8`iCk3E@vTn*~qaGW1~5vWy$
zxMCgb>nc`Td+|PboKIU5|IZLbG?1g+$nMSi9ov7U4?;J_QshrT@lJQg~&dvfE<
zE@DyDI^qwwjNc|$#0vEKj1pCMtOY#iwBFs}b~lmNG}3Yy?Ix)YxR3NeyGuNd#;yQ=
z6|DEn69e2+IY+4F#OEFopF4-|;)`DLWpc=VGJufYL6Xm8+J?bx~8UgE2fFzjXo_at?nt9!lsN$ulx
zb-fS!h=}Be2wjanl2DeH_xnK~5gRqQzWQx9S@B78-29i_5O^LRA?cYgaqAFyMxpWeVS)K6}
zTRO;Tr}KlKSnXubbPW?99{9aa_Ty6vF2e6&LbpH3fO*&9RNI#h^UEYY^24=XfE(7>
z`7LzmCR-dVTffeLJeEm;OeEy%|A_03YQsIuYqzOFK`_q_!
zUP+#xJ7M0OE(j=^3+wQ_v%b$v!2h;mTN>YXkWEMY*(-i^5QR&{i2;+l$=~W&dwiE=
z+%#9IY6;|nCVby$Ij)09OkEgUJHLY{_MdWn3H?5@QsB+zGw@C?t>*o+d2rrv*U>11
zb?`!cyz5e!&psMC_EW(BlL9j~-3NPPGLx!wK^`LR-^vC(xvAe~S;KiqG@j>m5a`9}
ztUOHP*^?2^S8#qlC_eYX+*xe$ak|@DVK_e*y_C`rf^+iWt9&=IVg8n^Gm_VWb)mTV
zjmUXzxOY_!|Bx~R@=UMoi;RD=$XB6sIR}`(&i4X`1z~0w3n(PRT?(XUJte7Q{FjVzpZavDZz%5wEPjjjBZ8G6p$T|N?I-m|a=t3PXJ&u3rMQ!;djtJ_myl~4{oR&&
z!446gE#3MbJuU$v;XRUo3kWcuy(6tTp
z{!DwNhV#UC4;LUjcxO^3KLhP{k5_{SUN8=u6aj#;NY6*OR)=yIi5!YI48)kcW%1n`
zl&?I0zZCTX#3#}B1xm?tDL&yYt){prcF>gKDaYcui_wlj{Zas$kk6XU{LV+5Tbk;N
zCgssQr^axsKkDgtIni-wmq@0#Yt`&>0F3ZEtJ6ia-TW*$`tHJSQ??|^d#6=Woc8N6
zK9GSt)x)I#_yFg?Yh9$toech4IgQC<@7W_n<=Hn9sQ$Wq28UL=fjgl8a}16xU}xT6
z3(BvzHK(}h?p}Gy%WuF$Jlwah-c0$x2e&D|GW$KvuWpZgM)?x)UEU$=X&@SIzCMsU|NEIdCk+PKN*XR
zXg{O557E4GoSZTphrlE=s`piDEwx*8Ux4CEDS^y|SpUtY4*DG||FS6_-FfB(&2t~i
zw4&$AVf}+2QSZE}8w?oChn7!b$e-BR05>p@XH02#MgDE@tA&Us4eu1u_NxE>q2EpJ
z?9*`6Yii8PMZHCbq@UqDtLlbln?R)_ety7mvz}2UvfVgvpd_+V`2f9~PRgA>
zxhWs=%muwm4})F(`R78Vz>iakGkh?f2bl6H#lwE2s$>_j^)9?C(c;SV20Nnas^eoo
ze|YCxy=c(a(aN1P9`>)OzBcy*Qf#vL_G39$*zfiz#U|DO&Ky}@r8TLEMaJ`$RlivZ
z^UT$(aoi#Bi)T!!81^|w8`eDmzzd0fA4P@$M+mrU@%pWRchEfTvR|LTJ*j<9&mGuD
z&x#9G7=Rt6xUtE{)ZkszWpSgFd);K(=J(rrh2cF$)yasa&ye5t7UG>E3G*Yl#{60(
zGNCthiE9xc>%S%J7-u96FGwo8oR}PTsHThS>4)+n4@?^>3o^Dd{
zBlJB_eK*9E%Jlit$l%C-4@#NXk<@}AH>@-q5Da@xy&!oT^~Tc4umT3tQJ4N}TT`9x_TrcMb!-9EjZIXxTyZ>y~RS
z>T4(O|Jax6SauSt1Eh7GN++=o9J+l^r;}9P)lUx>>mZ&PZ9+30+sRX#&2QxLJIG{#
z^cJ0<4zlP+9-+x0+@DJeB%dS>#zr5Rt
z*^_AnhKt%seY%>Ngj^@-KGCBoGXUQ^m$a3^d8hV3>hbrTZDc}B{@vU8t>oicO@m@9
z7>A90BL^>ZkuWKnENRF`hq>z5{WNVOS6=sboDym!E+Qa11aUEP_^H)R(5sL)%)J2l^OYyl
zk`Y&4zTqM#4(*(M?GQ8<^o6gO#(f5Aw9IWsQybi)a%FC`fc{w-T_r3ShY)eC&wM~-
zN*~Ny3VtYP=ej6>{_MiMFLj`&^QUJQ;^0qT(>Aq%o_(f@oHLa7o2ScQUSG=D``$zXp8)^^93b+V;*l>~
z1TLc9j~~tbh`*nignq$&S3}UN75C>4BIkrr|J)Ztc!A6&(OipP%b
zT1?*yty$_t@$0>7aD5DVzt+Q_Cjh<1-X}CK=l(vM=8=ERTuSAUEx##_bBJ!B`pw%8
z$zuELSEeOczeIKUXQW?7>S$gr=H6_Iw>}9UmPCK#*Ofy;80;^;zZLZi@;6fa<30L=
zj#K13wIleQRopiWp`PU86XvLQ?QA){Z~tKIvk2wU#sSn1c>yp39u_J`u9
z=zLy^Paj6?q<%h?V{mGBAimIt_xKHuDE^DtWkK;~mZ&AocP{;OoZ{TS+*c_bZjGXL
zCZ_+U-$ilYE5)7u5ARSs@5$#*>$_I-&~{czkJ9^+=8SbT&)DY1{7&TxiKcjXjlCUx
z_rT?Gi|S3NK27a>{Ub(kr|ss~(LIfZmmSqmAv4|2`Iq>*$&5
z+xghPt9QC?psIMpvU(!eKFog
z&K#+m)B(KUQO)1OfV2Ky@UTGt&Yw-$iub>hNvd`y#G8*#1|ML)#M#2(;(OR1I^XF$
z$$)a@&;(U{*D%>J{@HSfvy*B)du2^pIACdzU<`+If(_
z@|v+}%p2^F9J=vtI^6Rwe3zZ{3G4)qcpLS|LY$GmFnN?0;!4iTemB@(y6qCD_bY^^!oqpAyw%)8cyS3
z--~*)*yV3GG51~n(r!Z^DYMPp-<&f*o>fE_Hht|So01o*b6*4e`-D5XJ{tVh6SY37
z-VeXKZB!1PJwUt{&-cBtxsQYkh8($gtDC4_RF+;puZM^$1Yg>AYyjR7zsr_&=_m1<
z&bIgN1RQD`uJ_jk?me5k4n#Kf5eeh2XVZlTi1_5FwRz$FWZP%oBxkz;Qg3qf!_+ms
z@OQbaS&s4pMEL!gM4Kc1Bs@T??9bkQV#@!j^Ke{0yr&3FwW(?)$${=mzO}ayQSGl^
zB96Bb!v-Z^Pn{03N9d-<>pz`jV@Fwf%+6MbpRab>47ZT&TX~*X8?_Str~998i2z)u
z6ja|1xUp%elA4NXD;a2PJejq+mF(T{dfM-y7NU`A_>kKg@YHF^W${Jrgx^-PJXNQa
z9NX4XJ?z*@PIoL>Y|Rh2DR%0YE#9r<+L89zhqkwo9g0V$_bas$jeNT+({{9yS)&q3
zpEX-aPk>5`8vJ#u@}^finoC;Y?-zNj+1}YnRt7u&oom%iV#L<)N5FiY9-4S873RZQ
z)h|j0z&CN3-zrcAT=*#Bm}XiRdA#&Q(&Ie1zp;$HI|_NpC9YXN<=|YlqAEdFx4fMg
z=^nijwW^cUzNlVzcchzSgm@Jml!A3(eP$p_0`fT_$5a%Kwvkigzc{AcYA16-#2#$F
z+eKcUR3|Uty!FkfcabsV5k&{@cMP|;63NG0Gx}TFNKx_Avf=zra(}9sbB-(Q_v`%1
z7r;CHEY`rmI5=lDJ=|)m3-~bX;4ICD;P-d?)<+qzo_3_IX63>968JuSZ3UbUBX;=N
z9Pfaft1lEhJfMjqj;LL?ZQ+_!C-pdon$#I1G
zPt)$#g#DdrJ(VA4}sl
z*W>&12ALxlo&pX2aoc&^k?D{|P;4f5@N
z??yF1U%9>hF5)`9c6HIMK%4VQBk*2lb~ayGAbh{F@s=AeqkZxzlnr)m%0-vG0==AJ
z75D`q+^?-27ES>i+u|P`%?)wojAmet9O(HaWqihWhQfM1bT{RMf)U)CXbpVlrU>45vi583@R598pj
z67?I#8<(8*j?80{(fxP|Qn4jo0|J;uF<&{S@
z#fLK2bc}Id96I8;0Qb!*t#wjJD;2xo4-)`hzy8!0`;l&tEQxv>I$ObD-1${5Aw##EGT~IsR^O@B=gct%1iTf
z(>&UXoGla=RLnm|sh))NM4GSbZ2v%UmTd&L3buP-+KXRkx2ll&J7{8p(iEqC
zny`)HqA8{Jhf(ix^%%8xQP6b;^7H34yvBMj=C+w2|Fc~n9sRvKQH|!u(u!wMT=sl+
zO#*_r+ttM=YMu
z6c78dnrp@BJHW4ry3vREK8cp!6u-|LU4s!R4v3Ie@`m~-NrK=0#&
zuFBQG+i-Qol>nXaDrw{c#6ebU$TB6+|NDzap#|cLh>%BpH}FLp&sCu
zuY4#U{HWfVbml6=51BnzPTq%jEin2^Qsx%mxonTFG_Vs?Zz77nquF`k
zRKq90Nq^T?7_No;sq>X@GPY{KKC^4$$ROAYQYs3_1v`q#nW_ZN?an`UOSj4Ze%>NF
zA=nho`|+!fXn~(9_b>Xmf!!Y=;`{MEo|J&=>Z70+)b5(T7~=3YE3PBi^>Ch!tm*WF
zIIN%OoGS+LylwQ&6b*H_XKRa^pf1=;reD_nofSq~7;jYERLaKCEbI`6I-+^5N~-*$<;gnaG_`-3)O{qT3ughU6}{Y0*-
zfF;=g`_-aHyXysD|LRa^vGIiWp%+ub)ge#S>E)&QzPp!LTV~1~ggo9G&Eo}ERp1_L
za*kHn%3gBjuFa*i)nF%K+X30?UeX`5wm{dlmvm*9PJLY3PZUyj^UX=^C$ptI42#_Q
z$(hGO`zj*)$(M6;UGG@*lh1`!BUgU)ll0On7L8E;dHL(x&-e6`alcp(SIhPjtIu6R
zu8aCfcIpeBvyj&t)sdN*Y0ys|B|Yvi-PBJ!hnLCcP3$KRB3D_6{_Z2wf1O-E{I`!d
zYzwg0=;|Y1YmQ84RqiJ(i>JE@F6}2>2M_Fd0P>Y_6NV<$^pViz!KZrL`p84~t@k@c
z`pMJy9{DzzexkZ);R#*9FFQY0E*rnSm3&Hk+WumBE780zcKZ$BnBa{+$lA#Mi
zcN3Pi65-3=2Ld;=5|z@)%g1kPCGDOcXHQxLILY$fsJm4wQ4rtoxd!kOf8+KIlAf((
zfmep*zKgBoqT9Kv)sd~_(t@@Ol{2mI_uLkbtIKI6v9BU@3|d;rYk$5GeNn(uPf8|+
zu7~_u@wk}gO>Ja~j9crB@oi+PLw#0~L>qZAqb}j@&Nkv@9#mp*v5mMYCg@b(Y$H!K
zYWmi8w35aRLPF1*+K5lrp)$wPPSWA{II-CS@(z!6=Xb$8vlM=7FXP%utXDRE{s8Tr
zwv`aadzYV)5)!KED
zy?L`E{0qCu*vr9u;j?X|qWH}2!+%;y;_cMhV<*~(LXgIk&ElQJx5!ZbLnhp>D}C2p
zmfJ@7Ua4>Q3ITj|U97`2-$h2YDBo5CJa28jWW6Zhn8TgH
zAufQ2F7>BzKL`9`Ar0XN)>&!iJ<$i%@P2KmoqNS1xF6A8%B2DFmV-;;-C%vrnr*bB
z0pfzI&g;ox8Qt)%{MHtcFL2%)|H83(HoU*NdH84%tVhR-=Vt5ye70}VnS|*HUEt3{
zHS4i%^3Gye^B#B)?aa@fX#(f5_VeC?IdIOoUMQLUXIU2sY`r)BOISCFjCG3q0dcBa
zpw6lY-d`;0d2mex_^q;JFcNUiYuRsSzCwJNwX2
zOQw~o^8sE7QfLbw1uCxJ66y$fI7=tR0~z2iFPY>z9@@WmtY;M4+cL2@WDocuJwYJV
z3*u9WkHkYm7>AMOV#QlfzHEMo$0@LT>qE{xRj}9oTSOc0Kd!5k9{U9T49@4bMf~Ek
zWCp==b#Us-Jj74eD{}3g0{>=6>oMNTINW*k2=i()?+n0;J;2AuGUO1KIW{jBAt3)T
z!}6If@Q<9u_%NU4m}~d&JIKF;d(KDw&yONL?uR%!F!J@RDadWsxbkU(+~u~>Ud+SE
zDldJ9=kQbEwMuNzOFodX`x*FK8~NuJ=HZq()J9_-;-s}e0KV@DS<=godANbO+XXSd
zR^$-hhWqS^u1TwLA2reQq5JpoR0*12=~k7?M7>MmcJQzf@N%dP%^z`p_liP&7ZF`a
zyw?&^NrxXG!u~VQ7k=pkd8v*xx{vQ4VrOE#057RuI1W3bD?+eer?`iEP(Go{`!?E_
z;Sozfyz$UT{V(F!%6pd^ah%#3-yK2vd-e1kh+|_my&A!KUb+Vr-}dgn(a={@$O
zX}HB#qh#!W2YZap0x>8jI$6=h}qV+hAnNkLc_?=da-)wLY
z=wqJGWr(-NV*An&=XA^x%ZbSKobX2b@2RpRwJZo2k?ba+rJzJy2|0U
z3pd22p%qfIqJZZ$Tr%nidaL&D()kYa)&JS!Yreq$JaE3<9L{fsvZ_8spr_qxtbxC?
zt&7~1y9e|pi6qZ-gLUb^(>2im&PA3ji|63@xm27yaRl^q=c|qwfZnP;ZtEV<6Fz_M
z3F@Urt+|i$PD>|K#RsT@+{`k}zkhEYdZh<(A?IV=LJ`oba^PkML!PF6*QCjJp?siS
zM(zrnN5`g&537N5SZ;5&Wi4%CJI9qLcdJp||dV!t+5H^|RSPHGjN3ijmVd8g^X
zdAz`)V#hpa$E&Z)R1|1wsL|)}tZs6y#))?t_}8eJ_pBb`yPo{o?^EFS6&hJZISZj3
zuR_VBzt_8oh=yR!L^!Vp$%wx#T?Fx}V->Mm-b+f&ypBpho<`F5QD)h$UUK8(^?YGP
z`2BxgPf6+yz&W!dlgj;iiPQK)$)6$rb8_G1uWBK^q%A-5esE_uDOr^}|5z=Y3tO|f
z_Iv~UGt=&MQE?xsT(aMQ;^;j6@9E|3#G~VT>^qZoq8`B~B_`HR&I&B8jhBFT
zU>jD>OgHHyMD4d_=9PBh^J0d-+Ou{NWH)HA{A4>RpH|m!X>JD*>UB1K#@$K&&J1ml
zDunyM@v{C2>~@mzS}tnK6Ua05t2Ww8b&yAOSKe53b`aV5uP-DkbrC*|yXiXcUhiOe
zaQ9r8pX`Qxvpt0%Z^_ryg%0F#u54C$`yTdpvUG;Gd0Pi*O830;leddZtm{u6
zzZCWhVTrS`&~8Ki>YqmeH}%ZD5_=5tmBvjk=B$JHyyI%{$10fj?;Dlbe$b!fNagdG
z$2iZscV`RmJ|D0AodtRqnvi2i|l`PIs4JDXv>-H%xQlpFSK?YQ~dQ~@vS
z{4{iE5cJvtqz}Kudw8P-vzp-?uyNdzm|^gjC)BcN4d}gGcEElO_%&Z|kKa|;FHf!R
zU+@C_VAoWNtcCh{hbp6+!Oxfy9pmRf<85-5n?ir~T}{n52EBxn+OIT0kEbB`5S~X(
z%LB9%Ku`Q#Q@JJ3vi6h>KR`cp`rRH$DA(KfuuL4NZ}R%O7${$`;>>R^;7=T3ZTtdy
z2dr1od#uE@f`61j?r34ebq(|mec8$V0r=%dG|lnv2mJbI`w{n5E8#ze>L3@tb+sPP
zzrR;z(7#{pr8wG$`-avZljXQy3jLV1ALILZx5*N?FMV!EYP<-%Uq`(-?qk{V(Q9!3
zcz$ac{kwqo&d%M8a#PjcJ3T;;@B6`%xc}74q)f%QsqWJL=oj!$o)^HMID+$InXjTV
z(CL@F+wgtMGNV!f#4p}lja>M;XKTTQT&5K@IJip
zRR+d?t-{6h_cmu&UgSdm7XFQ(d6S?Za0mF5a#y;~{vS+Tc{o+y*VjanM$wEE8Hy+>
zw9QjW8YD`Y8fZXQK&KV(UGaTgUmO5YtQT
z%`Hyi@I&Q;?;Jj`YbwVEU;NJefTvu))vr${+PQqH3E9hlQ!6^C{Uz6T29dqoX?AQs
z*~=WQ31Nq+{8q(4y8qoaySeA;uADQHx>T)&*`~41n1b3y=)#|-=>cD|6kej
zPAhzueZ`m6kAS_bd%P=>{FHTz+Ear-^A{A(rMw(VbTdu_L9S6^Y}t9>|E#BmO9&6|
zsL`?v1RY}caNH=!wO(kcDT18kAl*7C$f<3=8bkiYoLSX%)SjHKx~DPZmd2bJP3!GW
zBYIqSLM|pJ!|*obwygW`!xwTc*EM~jN$%{*Ew=c+tr}I2{DNGWM2gRFjFyRCH1XX|(?F)rQOqO*7s
zem8B%pZ4dN=UlhTy!H*ejo2^O_n3zmXv|3egS;u%RrG9Vf9${A;rTBjI$5=fu7?tS
zx9jnz>QS)g*>Udk|G|#4Jb8Cj(Cm|@CwlL8GWXXN>sCXL{P5?2E|`BzF3_`;2R>+S
z{t+LCaktjNyW%$D+Z}nz_P^eDvNvM}pS!GsakkOsN5mf?JK|=>58d9y96l!YCT9rQ
zoSU1kmm=OGK%&sM(E|JC?rlnXE!hRX_2D05mUZE~v=J-I5;DhrRu-~Noh&S2=IGb9
z_>P>rE^mXK{-*OYt#)n~b7_^x@{Sj>vl&k7S2lODSf86^>#={=BH{RlYc)dF|M-$8
z*-Bk(T;ZI=?YctdKjPn&O}{(Y&~Zn@_QN0lIe+BpAF%T~p81ws8`{N2n;(4VHBHDG
zVn6AwPx-DqcpOL8StC-d0qklz<(!(I+)5)ZzExSyzigMrf1
zcx;B2-0+1T+u4U{V>eX6t_s~x@BQTccp$B+R0-_Q%8UE2<5S;E3Q2Su-k5nFC47YqmtY@MVMpV>a}
zS*2RYW2cWdYtNsGIFq!|>u({xD1U>l#TpA9J9Dql+t!WG8dqk%lY_mq>i0@A8TN9|
z%(13v@Duf&y=eMAJhx5hLC^NXUQRL_F<@c?;*rjr)(OP=l#W{ejs|tCzb;-I@EG>2
zW6I(aLqYrQvwxr4!ehmrn`}Rf>|kqW%2X5@!OvVPdvFfybYaD;p=;5e(ah_!%4GTM
z+vK%FYCJpGw+#ocH0{ND0zWOy4fFYJxBQg6@IEd!JF0Yh37<(lwy&O$+rdst|8n>wH|L#Up=>ynHF{@3>2Ey*<4Y_XH1$#Nh{J5?meGmEyiYEA;
zPORNy9);_?_U$V(g+Eau)G};8
z-h$k{f7|#fkTdD@U$G8ymdaecisE@|}_87njq#`y6;@
z)klv>u5Qf4p+6y4bKP|};etem8ShDMo4>-uOW=z%<>!;1x$VXD!9vKrTH?bE
z{|biP^R|HYyY4(moXy#S!jpMHpV!i?1IezQQA}A#>mx4%zx5~ojov9yzPZw)
zr?|X#m3o$3f7LJel0@=bX5INn_D`ty+n#>kVq%3m>7!~|^PcQ~^Rn`N#5)ElPp5rm
zC9)5LNzQp{ICno5fjQi|%XY=Q#q|DOJB44__=}LXTCl%T^oGi`%L3mkSly)rbO`|
z`GfdZYC#w7FFrH_^$IN?Hp!sf-_hd@tMHww-@ki>^sMRK=}!F+^;e^n*899BOP3NZ
zc1(Gcatm?=?}vY)_h`?{ID67J_pj!z50E$9bU$G|_y*q9le8|A|DRTtdm
zAx>h_?~(uwj1OIRi?;e>JXyV@R_OMplleCOWT7$@j@05v)@gx*qoW+OK=_=b+skIZXvJ7@b6FX-OrR{ib36*Ig4?4paYE*|IM>~2Ru}S0JEja-C}i3-
zbsi(^p}#&v%tZc*w~l)b<^!)p4xKow3hNf`
zr&A^>KEnR@3thV%io2Np*6L)>6+)KRG+aMd0da=wqG#tK4)Z`!({vAi?0e51aZ_R@
z;)WY`hOZs`ja@U5o?eXS{KjS2BPC_v`hRbk>3;Zm*7jR+d$&BikLT0i0YyZRuA>w)!f7u2K*}UD^CqhmM*$>mANX2I8@3g|S
zAy>$B20mW#J_qsSGP5)0&h4_1b8k|xN^?8A8XG=j)h-@$UU8``@IL$-(=-2EddXwQwKjZAUe9B;
zMOkg;z+t5puMVZT@Y&cyI}F20JJ|7~U%Wmb&bsgEYRRBMJoaK+@`QU>&ro^r+xPzc
z4(4oXbItr~C%d6nQ8*|J_%74sU#UKijoP$c@`V?liHjLo+edY>SrXEhM(%GIqC4^(rVzBCzm5nyt>S1Z#Th(CvU6r6T9QbUQ%Bg>~z#kXRDr37U_Fs+~)b=YF{^6j;
z;XKSI0>{s~y#%=a@tAQ}wb9P7u#Kwbz(2}TCT$BaJ{kXvOvu4H#FR_ZW}9PN(2&=j
zMDfy@Z}M=&2j&lAL&wNr99E3|y)F>rPuKf(IXf{f{R#fClJY~2GZlP@!T6KAK-iRv
z^-$NCexuG{9q3$ST*4C6dn69nk0CcYXiC7>5q8dZIoV`b4ep4!;ArUG=N`d4jJPe)kml
zYpz6TEm8&5DSM}{jPh!e^ffg9i&ELudmim+>=@Qh3BO;hV`#7ua<||2j3>PGJikvD
z;g*Y9C$)1x|LL799tt^?je|B4{<7Y+OGtR=$e#FXkH8n^g$I(qLH?)N5+#%mf3icC
z_C=4eihV}w8Q%-y(`a6tl(lXf*^|h^-{Vh%ng^=QCOZ_O>(crL^-d?Ahmc)~O*}J~
z=Cj%ouA8Dz{$-6u2wgubZn_24*V?sl^Y*kQ{c6b0zKESYECcl1x*uZne8%+5RU!XM
zjqz@AvhS8DrdP-g=U(|dp8PHS59;?4Ub^Ub=q>pRu63%)kzF=h`;p6gzbzwT1=+#(
zLvL|$sgs)>j*#35fgOMd;ydn6ZzXvh*RjZO2Yl%oW=r;4>gV*6RIWKGGMMDXn5`d6
zyiYckFTnr3eVwC1|BW2p>Fd06IF&DR-|~p$Om8M~akcG>=Wsad@j?kMKBQw-7Po%E
zThYgeR+M+b^DrgS1MX;6=Hv>7BD@yu
zWP65+|M8Vaf4Vl@p0cq`!7G+AWawKZlA2+aNPZW
zFs+Ev*B5d3B|CXGhY!88j&l6WKWn*o5^oiayCk>aJGP0T-3OX4
zxV#FxH*MtNhMhlJmy
z_N{U}7uA4|kGvT|^N)Muu7#4GU&fE&`fu3AoUNpfR=@XF^d7Yz)OkViDP5mU@~J#T
z;+YQlzgY2iX=+!vL5K{Fh!ZhNUqJJaeaA+hJPz8sJT-%)%q
z-g+a-=U;o~L-U~W)7JOMKioIj+v6;7+T9KX8G4STFD$$1{?>2Y&`kc&fh(hf%pfm3
zC7a(3`l#lBC9Ss>&+nU)4Y{!Q)kPNI$8Gv5U55I;X>X_XVB9_9elqVY+IP`y9`*+N
zQ!a0IRrd(P`}Op_-S7+Wi)>H%;@929D$g#;NX0lgE+XR6AB=z7p6M9Sx{LeqoUVp&
z%$t_47%E(eIM*;=dmq%x9?EV$2KkHjkAp%m?i+ov@U()zC_rCg_H^L64BLo*{V<*y
zZT%>-2>rI?;~~LBjQd$Na&}WOf9F3w@N4{3*dx!=FKL`^;@xo=fOf+(V}C~ir+#{t
z8K(@s>(~9irn=B`evz_@N*7BlQLQ*WqKnlH>1cHW4$_%+*4Q#k$Wry?9sRQeY~h&n
z5sK>tY}0x}y-yZhtUbbB{3Y;J=7Ngla;C^b@O^vd>8Cngb3N~nqbBNrgITQAHdXL<(ndpb}VN6l|5#r_bU!Kct5HR(wbORd?Av;$Q9KyqX
zw~d$M_g$zFuvvdRu9nmb*pD5fy04xT!2kYsLMif{$xnUSfASLG!)>cxDa;nKCAa%8
zx%F7U#Ip=@8^Q%_L-nGYjT;3lu4VYBMGQDk>2O|$3~=Fked*zE1Z;%;m8pS;&`x~K
zlX;L+VLd^cmkU{l;JCwjSs^=K*7$QOdu#{O_cAzB)~ACNYV!6?(8qqSzju}R
zLpzzir+jnfqYjo>Z>v@5-NCAFw5}MYinxub57w0+|5%q~->>hifwv}RMLqo8!Tt@J
z-gYYt`5#Uz-_$VfV2<(LoA;jNGe7;p*AKAXRix3lq5=LcpwsHp&#>=Kc6E~SAs(y9
zoPTUQ^vu|w>?FCL&yIdQJo4UJKHDqt?c>Tk9xKyY{%o)>k2R*%P5fQQXAd_D1Hzj5
z%)9Do-h;V(<}Wwy+^7y7lTVCa@Bnx^d;Pb<@l!jPRKdsJi#qr$+BN@f8#l2n+v;h0$BAa~!hG3l`
zr!Aluc`1}f+eoipz+)318GPF1gMEba(+>^V-^r#V{2uzz5B8d^n;#j%WAoFb9B;?+
zn5zD)-U<30Z0xV2M4AT&e6VfV8H{UxPik3vz#pfzjxSF68unaTcX>0`sl6uTP2Ys^
zwr;IK8pfT&gPV`N!~F-C
zmiDE2ZC2do!B;@%e)iq@9qmUsy9WuO4yylDZ(a$O2=86KU-2yf^!}-(CDr(y`qxb$r7c6Q{E
zvug~X=Y|P=)@%mnUR0YI9*w~6m;O}E5SOTxs_2V6hEQ9C5VgDVNsTnWH+x`XAh%wih*Ct
zLkWMcDQ*`M4(S&;X9TtL!(6eP{9-fRvUSP-nCtrbcr9r8mmMmEJN?C0aQ5BY!={<=
zaU}m^HreZf*|Tu48lKNPGjp=<1&>#{k$rqpb`OcG5Fg_3_8P^NE43e+PPp^iLkSLF
zn@%p_;+7k}AJ8VbjCLb2!U5k7hQB2|btqyP7hfDKKX)^=dtRj&M}=eE!O~htey5cR
zF7F5zylXIiM)egLTNe^;SQ}~qfQWqp2YHF)4?C@W>p9g=cq+%?$}WdHcd35rh;8ar
ze(b!{clv$X<*mBJchBkoutc0kfhV{B%53j>4hJq?{j-&mpBTdJXB)U<0JlGE(#tpw
zXU)4i^&?%^_eV8{!|M7PbGR{OuqwxoT(FUAXRT`A0o2aq$R@7*WauDHZlD%m8?K+#
zlg81M1(b*l`5yKF_JI~&=;q3O--&a)Za228L4NMEcy52(K1IN4@YVB0@=F%L?FfF?
zb%Yh+`kLXE9PKVeG0N9@qIr~0Qw8h>|NJ6gC1|p{0ava!EsUe`KGGbXOtEv~{G)w)
zqE3_ChfPoC(epUJ(t(qEGjjn)UyZuT;rPrBZw}Y~Gzw24`M9bBoSuq8gjwNt!vdN(
zxgVAm+J<&9a{5%&V)_F4;36gNeb_$hOB<&ryl_-LLy&sM!;{3?R)`n)%
z`D2pWc$|kFpoyxG~I=O;!NV
zBktJ1vS)?lH*DP6Pb$*FjTg+~9o}AWr94l-<_aN8h8V!Pt%T
zJ$a|@cdA8wxy$&gi@kwBod&(=)Q|hrZmlx)f-LyFJ
z`@8y%-$?s=-e%1Aq;dM@(^ud3pxh~6cRJmF)h*4|E8u_E#;jEY^$~X$Qhd79E&Dc_
z|6ZRfu)2V9bl-n|9w>fOzr=StV!ukuS_XcH0wE;9rL;!K)<%tnc|4!HC5I
zHYC(kE_H^G@tV$L#g7!S!lw2^mOcVDYQv;$dsYb8&2#LrRW`vufjIc(l<$b(drHc9mM!7aFbhn32^|)=Yz|ELtb1kPTnm89KWSp
z*8=pYcY(gm9oVf0(XBBa@OyMs4*RVKykq=sxQwxYjh?BSJ8rOm`Ss0OasLYDQ!^hv
zo`>K4mZ!1!w=v#>spD1hcMF)c+QxujI{{OQ8!B1Ti?~?Br;fIlaZX6<_Pag1F@Jtj
z_w8*Q_TK~u{bpW;Kj+TpqOaGm|EXSQW?>AzfQ#
z?y|ecpbTX|Z
zbG2asoov_CUEivqXX&+MwFmGIuBx0nEIOLjMdP}Qq!3pa=C>;3X(!W}PkL{Z8oF^ZoqdBTMl626GmEhd=S<
zyW{mgZX>@|l>UYI2;c^{pL-%0=HUY(V+Uvg{~Z7RFAn&x;hBp;9`-A2i&=Z5@j1q)
zR*k~YUBJgqg9koX*vVuD1q;7S?qtb2_QCG?9qjw%xS}U+9n4N{;1M;%v3q|SXOIuS
z>O%V=eP1IF+J*~_qo0O!Fo&B7Nx5q}*khm4>Z5Y_?)M6tcUB|5aG27mt=^q%K|CIC^#Y
zyNSxs|NRs_i8-BY`x(KuW7w~%rPjWC*TxR^TC#TEnAIIDrDy57fulRw`mGZR7FuE-
zwtMzl#V3eI8u{gKMK!F>L7o%h*4`0l6g8ulU=agp(*bCYMlFLKK!Kv^5}{P@a0V)huv&YLA#O2J=n
zbM}>aHRuo7vpGJF@Go4F`&hXH@9k54TNMFtyUtyYP&14_!tt}cwV;o2m!0q>-lwRK
z!+pk~oxy=G)lXx*oc#9jBVUZCO0VP;jBuTF)pYN0j8jK0u1`uweZIu@UW(W0xzY8d
z660z?hH0-K^nGiY?lT7C?$#p`>-=zix{)xI@=p$t{WO9_TkGzO6p!#(x*%9;04gA%7y-`|Yu*ZboQde2r_fG5Ca&
zx60E{e>LCw8Lbl@luuTU1^wn_oTrTX`+lpW$bR{E|CXV+Udb8>&wZ%3E9>1)xWR1q
z+5KO@&+tvuBiw#%RL=&=%dzW7_c`(tUt3_BHxg9*)E7Rj!}`q6mL?o=BqJk92)UW&
z@=8m>VlO39Lyg0}Mbc1c#Qu@7q+ogw799y3PIVHLutDV_g
z+;ryB8Qea`AeW{!WWOz^8OiQ7_BT#^KV>`BA37DQM|j8Z
zNZdraUcXI_%iDG?`jZ*e*AG~lH>j0<`
z&o!f+!>e@xzxR?Jt87=7Q@zV^i$_FN?2&dG<(IcOb2v?R<`QG#n|4|0Q2luSI?nH0
z8zVY5`d*sTJ1T$ubtQ*0WHe2xh|h889Zc7ij;Z2sa_E+W98UeQb}lzB2)<&<;b4{F
znD#(^;UjE+gnaWhYzGA$d{&hI+C+2gcdGBM`oP_vtj9S{&QMBp&XAGToi|kfXj414
zpK_nUbZ$N~bMZj#dHCpl;pjPJ2Pe5cZ$$LiC=o8UU0KVOmp&Bv#UuUEJn5|^k~^{t
z>VWr;7tI5LwC;2AH$H)&Jt4WrQLkbw;(a1S@;$#Xok960+#Sh%MsE^5rXZ64gk7z~
zpI8NW3p$9zA^+mNNUn6Wi0`~3;#DL>^O!ka&p3JeucCRytg07i2zYXL4tL+@j#qH?
zk>e3AiE_))leu`?)~|pQH3AII#YEQ9xUP@eHFPp%#CM~xZgkUX%Tl{73MoQ`Tpm0
zF_DB_sM*ZzYYps>;pC(eu}uc$ytb0cK
z{Ol3sd$Ds9opaUSCz$K^jW4Hj<4nY7F%6RQ)(bzX3VL+q_%9Tfl;7dwME)CasLm&Nhu*HV?FP&f;`OluuSr@}T@dnN1VC
zY2Ib*sk4au69&CgTu9FaOt5Mh%5AeRW^&h`ALvW{cJ%g#JQ@!hFC}}C{HlhWU9{fU
z`cQD1?khUhA&jWl-?f=EZjXFw@QC(pyv+?iK-A1Cr@9^X*0%3b4R_ezNvGYn_5&Vh
z34bq7=l=y<$_rjJ8T;x_?Np@rn4H$|w12SM?M;Q1HL!Ps3+^_x0v8PWGR&t9cK>&5
zmc0t(TDFg!Lh)m^ecmO}`h`lL`vZL-uN=5RC4}tNqOQs_sDD>mA5QjdQ2Dmf?cjGO
z4W4-bG<3#i*PD=^;ZkTZ8TIpb>~)DmdFv(DWXc0@$fWMdAhctzrJc76^OJ)3cGVEX
z1GcE-9Sfa-IP#J22M;@qakV-z<)k;x^M3uZ#uE1a(}FV3T=@OQIx5c-!oR6wJ!xPM
zaOcgpCy!SEZ+6l~Ks2i);N(>hWP{o~LO^7THhOByp#(hT_M
zne31?7SMCtr73&Tfv2VgrV8~Dr#~!hOcCK2pE$M3DwGcz>lu}fe#`qlLe2p4jtiHG
zEkb!-qx6X`yeAT+XTM2;jXqnyESoIXHq
zIqvs+Xyoj8=ouw5HaZ3Onbp}{tc89G{oz;g5ACZc4ddIO9sRXleG1GFFI4vOnuiYj
zEq_k>)WTmW->|OfDdr7tq}^Ym{pr0|qq0FSm`wT}itBUyCL1irbJd=*
zVqQ1l#KVyqSE1*8_oN_qJP$j$WXTgKuP&QAO$Yb8&?!e<2hXRf$nBL0^fdQe+tq{e
zD@*R$CPJ@ErP1!x|AU>|pD3bT-Ovc@mH3_BXW#P2fVTSIt|LFl%AH4xQgJ`Vt1h2(
z29;gqpjeLg;@hpI55mgvon|f19l(Ph{{HY|Yk;qV^yR|D;U6C>7vYo%d4mapr}Uki
zlW@?B2HtMhK4ZQH@5?yX9>+WA&lvTIh0@Toq2$f(GSD4$_KUaR_jfoBXlTOw!c#k&
zS%rSBy0hW=2OY44h<=3MX|VpTb_2gtJa&%39{SxtqhGe5
zC#HT>B^)@lo)>-)?Ha1@eYO)+K5JDt?Wak8Bi3;Q`d(WwqI?YGD%;dTCLvy1>xH4S
zH@=I?cWbHx5g(E}IkQ+F^$!eQCD@@nJO1(MH_*qtNMYa#v}ZhLfGX+vHtTKf3HXa<
z_xLIufc&hUC_N|CH>Zc5+XH^H(~8FtpxQ-h<7-h~IQ*^HP4KgQeX1*q6%VD&R%%1R=9@@JQ>{_@V
zv~=q5^i3!~YOM2*{OdaX{hp3S`ONyHt7LCpxaaJkc|%TC*L;fWb55NkMR@g_6t9@#
zN56&a*ew8kC3Sz#6qJ8_yUWBB^orLG&TlNM+mc7)|JjX~4-yXk@pNj*UhtYOE&_Y-
zni`HxG)}`8;zzhTeL!qA*~5)eife5^zqIQllHY#KyJvUk_s;~h9wq!ArmcCK#($;3
z!zNIC$I=lnu+@MDSu1zlJTxHxc!+G
zQvI_CcgrkP>!SLyXF-yr$NiTrM@UcK<9}W0ckFjvnnQMNezxO&lB+&{&zE?)?r&qs
zPKo~v44Ty%F_-%67Ewd}4X)puG0mVlm_*30>6
zD^o(@kjM8YBQc6-L$*l?(P4L2!7+~UCF385r|+wb;jUZib(xE^sMmYO;dY&Kwj7=}
z`sc{uBZ~=&98UXe?9Ji+GSdSbZr^b7%4>{Ug1^9`2L7zsBcZ~Gn+8?_V%OYs!
z?I=u(Q2*JWj^jtn0}I|KXB)={hnsVJv=q7x@`)0-eb7C!=q||JA_7_qv`gg+X}SSE
z9$9IK4wC|0MLEZUxBYKE;NK>a*FjV;$!R_oQSR7q$SOlBm3Q2sH#eh-YWZ>8}-H&OUSDvI6NBpn;Bf0BW
z%qc^|_|6x{bMttIl(gr>duiS0<`+7f@cBZ1vyUha-)2AD4ybRS7H^2A>${WowNv|^BNub}%*zqYqgMT>
zK1K3M?HiB)6ZYs(|5-FYu{b#$GjrhChif_i$=IN~Tz?yOS#o}ns?QpDAn>DkY#&GK
z(tX#Z|MrIbp_X2GKk)8{&yL*%d)tvNv3fCZPtl{!2C|z8w-?%xebKyH@{R15eaR_n
zTK6u0C%-ruu`f?)K=S7|m*OFc%
zx2mpM3Vbv$Ww4<=>Wid!&Kkg>{K4w1ATp+MrC+I|A8JaFyI3Jirr8>4Ec3iU4QmMPtPlf
zYsLUqi?5thFN^yx^tRGy16??w);$>IBaWEPB>Z=-|MGxypwoiZzFz}co>rbkxM7pD
z2^UYQEg7-e743I+bf@kC_4LwSNq)nJ&lj#E{g${!eI&Xfg!h#Cf7EG>@!6o$9@OA~
z3+Ns0mbxEwNWwUU7Os4pXy&I;E
z2!Wn&)SOi)?#}DibiEt6zAJG2duPZCUwzk&106jwp@H64|CQHg)BB|`c@!Ktusb`A
zlIVMjeKkvozN5Sq4R@*k?f*LbpzklhdrcJKp63_EFFwR|&QoeuQ2(o|i$ACJPMH>d
z{B_WDh4qK&Jyi`DxPbh%vcIFo$$~C(j{4aMy`)|rO?(L&xNSsVI*-ZfL3UgZsMlAe
zu0T-9H5L<-Ais3^1v(Y
zYqL;Zc*L(T8ua+8D8)R`*-qtg#h|xuCGm1WTNhtlM{yqf4bj$wi(mI@4;csgcdyYN
z!ol;5ayE?xm3$R?OA)l-LzE5m$B$IkO!{7n>vBhCLQmhJQ)I}`@nPuCY1`57O9nZ6
zp9KFtYuXrZ$cy``z4;72$YkEjPTcRDKWbYpf;UY&&`aY)<<^%b
z^fBBvX3jUz82EE({Onoa1*n34^8d$;*X{M=2h#6{j93s$`c~GMU?Ug&=+1w_iHc9L
z=Eh$OoiCGUJSsL^zL$PC&uu-IcXyB56V4y+)7b)tCB~06X=?O6NYB3~M&p_wf6#5x
z-*AkvD)E(9=W>1{>x_9^ea$>O?%cz-nSD6_pxNsI7Sv8^UT2)YPc-VtL{1N>&(4pi{_M1|oSkSrUMf!YkuJ+P
zJmIGy9ZB`RJ0v;#t6?&mt3U3wt%b@}ul9pu7vqtgqAu|Y+Woou7l$9hFysB36~o!f
z^ByH8RBx0Z8FVXc|lM!};
z>!uBKmQAYV_~)J1xb|YkqFYeEM1~yc*zaU0
z!tDu;Py@U}1>Clvat|>b1D$+cgxmO74WaACJ|Mu3_>wW_I9iyof}^iOUU2mfpW?Bg
z-fN%;FRh*9%JB_PMAv6b`NqjvmWll3(hVZqTvkxQl}r5*;kz!J!b$C9o`u@@-R6a&
z>(`sp1OfSqo{OH-4+;xHy(PKNQ9cQ1iRkDmJYLX715y3cF{0~hB1HA`{%enZovuGf
z#gM0{BjO3A5^sUgpD12ZuAH6%)ze*5Id>c+_{{`gLFqqHZW@95+W+`9|H(&?I{#0O
zs|L?aCm?_8KY2pY)IRkF$^W08)Voxk`Cqy9e{_M8s9g4@NPf1ah;O9Fi|ctGgd9b1
z5&p1l_1`>;eG=7QwGfq$0a~MackTxMihH~&wc$MiaYZ=}}*aukh3^u~YvwvAZu
zU+D3G{yZy^$06q2by#HN%4`1XFC=N@_z#$UlYGC;KR7v$-Jx$loPYXr!f%Ce6rOtpN!0$+t^esEE_%Ow{(C>FY1#{U
zJ?_CHZ!x`&;0^zh;;=82{vN-VH&@io&x4|R=d~i9htu3iF7ChU2TqRW_)!>r={gNv
ze~zz~4nsrmXQ&S4{0w2sF)hG#lg`mZ7WrMCYH)l53&ah=pE2PX6vcjWM|z{+&)d?*
z)n6G;KZJNRso5MKd~!L*AM6qJpH;I+U(36qdF60N4X%DimuTH@lCkLi&HY7m>J?Ex
zpH;@P3EFKPxQ?T#JY!6}fY)@7a(Zms^q2FKsQ0HxB(W=sG7~Jye|Qw?h*nxpHsAVs6~mU3Nx|
z@T6^`^EYe=zkU(nFdY~11j0KRlMKsA&*3$L_fWfLN@tf4U-fjG3-Nu_
zlDCt6?tK@;;X?78qsK@O2a_Bts{fn2Hk9O7^v;bXJL|2rh#T*9)y1$Ni1B#!X?)`F
zuZ{LSOgw*@{baJkc?$OU5`bIx$zZ5KT${mAZhcsKKyD22ZyqapP`z2|=t+b_mq~xE
zBL8LkFW+yKpo=G2{HAp#AB}!EI1GN*U&V_EKTHgmwwm;PB!8@q@JryiuSm2DT#^%T
zl5pqz<%;Y{cVl@ol?VDH2N8~AE$2pk
zK>hHoc2=uFYt!}h7lV%L7aF-1bnP*_p)`MR5^f0}0cupC_MLvGtu>?{`D1OghbhxO
zyJMS=&nJC*V-(z|zXIOMXVZPlZt0gxcyLqg@n?jWlw14e)Ab4E8a?!UzMgy6K=`oy
zOm{!R`)QY!Z>IYVN&b3`^xWH9^os5;{k?%3@!A6d`cV6`vezx9`#Q5Vpne@_T*;1A
zv7iPUt&(Y8c{L$tI?dm_PPS$#gZ|mz?$nC%=YA(v6TV)0Sn~_@A3x#SFVbt1!`gim
zSDm5#s+#yI4qeUkoOV6ASV{WII<07?``dThI+yh9anIjHdX5uFq!WI9TKZxz>D4@*
z9V9({bcc^7z2{dZadEe^vx+bTL$6z|9R9IQZ{*@3CfMY{)FAFU;K?Z!P^m)(JEwsf
zG#~p&^HHaV=kur^1E25Z;zZIi6^p4qO)QEp5nh((kL#f8HyS@mr+H1gY0L(qEv`L}
zNYB<7rbGQ35Giwq^c~%*!Qn&`Z#hoSHv=4|k)EfV&IJ)&d2L%c(Ntashev;GO5(n+
zn#K`I^g9Fg8rl&(SX&1~f;?9pLI5nh2QQBl5IrszWKQqH3}eTm6hCuiNKXRwV_d=&
zIADPbDl0SS{l56>{CrwR)oG8y2#@u<536p{_p)A&D27}YQACWDGN!)J<@NUmaY{xssfUw&Of^rx*W3<~DU
zS{aVP*iEPM=1ccr5%FEA3t;gv4hKDzpmN*3IOZPwiSAc0RPf_VE-fS~
zb_v`5P`~f#e9<^;wU>L(w)*udP`w&^wTEb*xQ8%Y*oX046^0x09);%|ZsyI;OC?@=
z;z!PoXa%3-^fQ~~#QDp67m0A_lfo=%?sqo!W(_BO!Q}^TgY`R_Og~L`k+W8;8E4?nNpGNAU
zel6WL=o*^d;JWM6L^QNhMD6hPP(95esXl|$0Tq6*-Q!h?eR%;@CE70q*zeiZJWpB3
ztW_sP^7ac^>#ZHDlkJ3S1aBEWoZW37{(fE}t-(~?{&V5!}<
z_f{ogUEXd_^UJ5$_mqBEA^tA>-0X7yAwvZ$wl#FxN!Jc`r((v#Jz>cE@b=r3K6da|
z?z=W=q!QLu`$yh5JYB&0N1V4DsM^dDhUdK*{glU^1}&^jalFu&XkiFn(q(pKY;S7F6c%XjmhATNTo9~+s$L%yXDHPvN2W*&Fw?CG+0%iXQ)?POPzO+D=_
zG^_85jMvBqIv`^a?n
zyk+YepVphL)#+>uKmwog@`31|h890}wZ&rYT?zVQe^}wj|%Rky!20x{AA@YfPop#%0iM-}-
zzimlf{kx6*lxh~cSJ}p1Z=WLYFm*vc8Y^taVvY+UDBu70(smuzYacT
zgZpc}YPP_wm3`XSbmh>}R<=EKhv|2NHg-X5(Z-+<$oogMbs4oXqfa@-{)<|1&d;tx
z(Ks(>)lZe@J8-_??wcdOSEJoGmU4$jYqhe{jHZ$6ac<5|hcm^c$n!qs&*C1JTZh9X@&o8bE@z=`Z0A%ud;3{Gc=rMb^cfj
zTYF{N!_IvzEO5-A?>>iGn115jhF|urY=N7-h6>KjQ8+ASTI|%y)=%EidZrZl=x>O%p%c!p8Wrft9~oSn_d`PZH2GxNvi!
z`Ic5T`Ef|1wpber`&|-vRH==HoEWqIG0qjb>ohC=$h`|G*NF#_+^*G+Gm
zUyf>LD|~$1Hs26O+=$gacm~
zu6>F3tgHI^z!FOTX(W0bY(#H1vO*;C7*T!zL@zV}3!1r&K
zyS*854oyQ|sz6R^P0u^!>NfV~YFnSCA?-|FZ-}f~Ks!@gZx?<49nSA8jMce=b2L-;
zPg)>77Wn~HN(Q#fY-f|cOG#|m1-UHU-*2kgndwt?r8^iOT9%*q+U<_Kzaji%JhwKs
z`WSCTY+M^NlT6-L@CV_a1!&8aJ+=7bTt{&v!9J8pW$Le@Y|B(KW${~FAX7g;5
zu3Jn&x2)i?pw;({?JIH4%<_xy%GUGG-<&A5K(B8>UWxKv@;Uk|{S6FvApPjirH%V{eGsV?&)Y>;A-;t$Nl~xj;*{r%4=Cg{=BDE!p+tI&k=6D9J
z#D4VOs*j7NFyvu6h2atUUiXpqxHpZ@DrV``7PRx2`lsAb7d)3@t#UW5hdidcC9cLB
z`H%XTh^d{v#bYP<%aZ@Xe%;K>vvtP4cjbql)ptQp<+R|U38!KA|2B`Da+l9!|LmN*
zrGwAfDy~2IfqVh|$EjU9jyz8qN7gOx!t?2G=CZNy5}(}`+wo=9D?U43xg+a=5P6?7
z*d=8>==13A%pXhnENE%xH=!T;x&6RC>mz*R@6xEXxz1;+j2w>)`oLowHYGUOeSzIi
z_tVNBfPTx58P}+ZJW(=}-^DoLJ^Y@&L=Sp)U0Cn633->M>(4lqc9F++>8M|9MgGdF
zi#ppk$?@5qAF8^Shjp;G5&FLE3z2W>hC=)(
zi#xDi>4*91k`C7Mcf*<(*C;UX!>B{!V4|LD`PIM~#
z#*e4JHCWog^pbDCY>(++CyojK9gswRRhN=DdE{e{&)j-RwxyGKjj(wZUd3kxE94c&
zb%O*f@o0n<4|zO_{`zV3(?q|7zke6vfqgpddP}9TpD^*?d*zfdTg^L#ntB0A2P=pKY$=f+Smxj6U*bL*B=ArFY!
zwf-p&Qw6L~f?i5TCH9wwU78>FR=_U##58^wBxFNREwJ6?B4BAv_K#mj3fQT5bHfp+
zUt@j1{o)8AlRa9q^7d#U`xxz!a!eQH_miHlg0XvZ984|xqz@%g6EPi%Pz&hf51zVmAm}6slpJwQJZ(`uP
zM>%+J(=yGgzQbR3YGc5vnF2O<({`_)o&uZ$@MWdKWFfn1q%URKD_~PKIf@6|%nDS+7no@(H&+`B;T>1|*BSPRcn8n48=N#r|p7A8NhllV-huIUA+i
z+=liK*~XYP_YpGTqpEwew;?~^ZtIY}X#ZM~;;oDqLUwvw^yWXv2l!(BsNDf~ksoJk
z;6o$a*Rl8h533-z#(MgXrw?%cfcrujYnd+Qu3!=?IjM_HQ1I9lgLz|1bcmEnijd9D
zZrSUTD`eB|6nU1tK|4ylrDb1*>~O;BuMc98N0DzEv+10Wh3+>q=zau!WMn>XZxu4>
zarR{jVqL6vnIf--_hSYSU0{$$SSr=YU=+KvcydvABdqnqYLX?6TG_E
zU;S0{_nhcr&#E_MyGC@eZi#VuceZyi&H3+D%%Z!ParL@wN%>uD@A`z2*_B;PQrlS9
zx2TIvc0V{!;NQhsdsObO4eMh4?tFRDlLP&HC4Kl$y4a<|_kMkT+J&>Vf7l#Y2YT-6
z+a-Qo*q7>-z9Amxo9>+Gezp+(le4F9*K^!&)8gz<=xKUpc*?l}UChhrgN?B{@@GB{
zsPbKj=bv&zN6E2^?TFbWrgpE3DP?vS-1>y;)_gxY4(DclE-Id|a@aTaY4-M%N_m_|
zD)!oW%WFK3$0h$`?@i;WjJp3}Q6ZHn6v-5YN*bi79g-mwks=CdG?gJ0We81@1}PEA
zkSSzHQe@2R97%;rqM|_*((qqvU+4b4_ebzH(K!ZvSCs)p!CLK`!IR+^!Zd*sDDgWf|uPfM4~+#y7Kvk$;EyB
z%N^i%$%3OZ*YwB@6IHDQ!p#CZPq}KTcIb~!;Vt}NUlsc_{m_JAVj3dQ&^c$A%=sR<
zQOtITd{JHXObNb!bBE~VQQLJx@V%`%-7ZIlh-J0T0h#b2Vv?fZsxJw@7dG$1xgYYw
zWS)EXy7f}S@OhU%?F!4G9nbm}Y6Bksb;eBwe$PzVuh^=X=NVY2F`*2}nU@H=BFp$)GbhRMT!-NhB~J7x&^k2d6xoYii5fM5$O!inA8=0LNB3Ww!{ra
D`c}h4dhNa&rQyTG`{&B^U;M+QeR-YS
zHYm@-xACO)!C|7~wB`A=?qTvI#4wK!}oCJ$~l
z4gIl&ato6qJ0btsOEb4PbPbd3J|*qSpNC1Jj^dQ}EyLspxtZj0e3<+$tz0OXI!vk)
zlOo*mhl%I(bxY^p9VXn*`=pQGABKDBve7~4=X3S>*NY{fJpHWACW6BxEt)wB+5F;TQ4*D%>i*>#y4-@C)y@^Wp!$f;+Tf5lzA+q<0Tn^vM
zAyV>0^~g6}n2$G$>KxUE;rp{>A3jPQBLALPWQm^{B2Aa=%$`8J^PI<{VFBy$`W;Dzz`8U*DA9Oe$T~by?E88%u0vI@x?k)jys2lX0Em3G>Dt6PuyC{I|qr@l-f48AMk$L$|#~2|^PD
zd>jt#b8vKV_S~`L13Y}Be-pTP;kgU^(SHy)_IP~`e>WKMe~<_Cs1cSWT=a?*YFsA3
zwd^0gH|2uw=;5P!Zf1WnJIyb%Lkst(u-Yc`~T)Y+#G+woB!*NspWs~&!w^r94G$&>(6+5>=7=m
z|EK+lefgjM^nm~Wf774;dw=0)N|?X@UBAquk)!`F{I>`G+XMgYf&ccv|K0=IOYx6Q
zR&bC4$p1l2PWmbiuH@i<^vgK12?v*Ra5V@2Q?3R_R_5R$4(f4Gm4p8&U!EiXT!Q~$
zlYJa{ItTZ2@CL1rO=LN;76+9$c#4Cj)KNB>$B_*=_#gi+a^#O{_#Za8&XKt|7|6k&
z>i8cvImxM4iX-phl<&)t6=;XCiRl`~K8d6MwwRIM(+**ivz+?wKf#8w$q?-bHawqZ
z>|{9YkKxD(mQ1=hZ3vsB&;eqTaoV9UjvRX#>M)y}+sxQcn$KW2?I1R3;@BzYGxf$6
zGw8RGv3o7YV4^sKC#RwVY@#=xv76Mvq)QxU^jl(X)-I{C*Muq$dlagKSbb2Ro8YzH2i1&k+eKPdChfkz|vKMn+D2gENy&Viw~7
zvf*a}aptf|x#$&&CxWn{Y@$9*mtx5;oat<`c?;9d^mnu1KWvic975BrQm|v$MCOPN
z#m`@u^1j?ir1jQ~v{UR`;Y;;G3vfoWNyC9I=qNThev20}z&K9HqITjLCusSO8p>4v
zmrn>ffK7axJHa6~nV8f=^RFE|PxXp=s?>hLi5FB}6})r-+C@xEpmt~NPch{j+(_;6
z1x2AF+3>rVf>RJ%mSW^!lgd~96g_%JsULm;qtw3hn)gA}x0SA>>Goo{(%IzOsUcdf
zs<;&_K{lE7ab?94z{U$!Dzv=KIWV$p;#su{28>PKhJ{{3z3F9%fW?6BPj#RGHn~#r
z13C`omOQwHU{+8+Zuq<3tV~7&fDmPA^LGj`O9I{%j9j)!+JUu(%sYmYvSGAP%om`
zyl4er?hL6K%pY8`WgpgW_&6~S^%51ix7PrE*mWZZ>nq-9raT)k#LaFY=6mEaFzN{?
z=^vu87Vy35y1daPY*IFJ(NU~#_}kW-=Mg*V4`Vx20!D1mA9b@^1pV|k(1V2man!xA
z1NB#4)>>k@Le4G%%K>k^9=nU_T~0?}uYq!2-nvK8w9FLy>)80U_2}n>F?Dma^X0m@
z5mEU0X*QldBP)!4q_rJ`NCWf7i4OuA`0+(#GotG^M+Gcb
zAnpL2Pck*F*D>F5F=Hdt*E>J8Lv-m|b&rl?-GvgG@8onb%(wGK+f%gP=NrvJY?{5z
zANynX@zovJ-lY~=M=;%^K*0dhn5nddeF9$Zuvh~-8k-AclCbf{^GV*At8
zq(37zn!j+w`uM%Y={i)i*H%RN+aNz}$MTpU*qPYm%!QFq>eu>Sx(;h{Q)Ez{elcbt
zV$+WZYG-WX5`q3GRu$8AxqvlH$GOFBCbi?AFrY)$*+DtpDYRUnr&H`!zRDkZMd`FCyYblUDFUHwOBuY6brb~ZMN40uT6$AWy^
zXxPLrC5gt7H4|Z&Ku*1)Pv?)kjT*II>(of?UE_99y|Hi{MVF>sbI{&DIskS`HktDJ
z`Etazh4!@EVaryEVsD?>VtP};O&Sjuf7?Rwjg%CPx0~0-Q@bdsSlAI^9-Qd~huHAF
z+7{A?W_2;Newkna?YRz{g5Ps*)aQ2
zp5w%{Czld9Hn}sCNf(AuLj6-K26z8v%KZeZ0n>knFzNdL7Hki)ZZHeq@Z76*8j!?#Y0F|x;I266SEewb5E8YS4U|1M|Rqg%+}xD#Xd7-kLH
zPngZ%m&J^q3x}Bg-iX^i=x4(!LKKDF67dtq-%FSbGVN|a>J}i5Hm2X<_(Q7!85TWF
zf5*vxj3fKOY(;xbPI@;7vpM=Gj_i!n4D3#G+EvbJCpX7#E2n=JCSwMOb2u!3havhP
z+HmA2uv$?TLIB*sL2(Xl;z+#tsB?=c7MIr{(f3j{a#W0MRH
zT5#|x2O)^VADe99_&+*@amrK2kC1|WSO$NmRIUzEe>M1*~0J0mM2eS~xah;E1%
zX~(e10borid;P|W*yIvt9uHYDb`k;%Zbq61`W-8=p|GDmVC0(Xj6SZCk$-eCvZw)L
zH*Xs@5c2sl`sijxzk!dz^dTmWCEdpkfH;lx7w|zOr+foDgjIUUvY%
zZSKtcPjh1Qvqxc@g7mfz*g!T}ZO+8gsW$plPQ8E~z$WJL6QE$&M>+9laGZ$)9VJXT
zPo6V&O*+`jR3CH5ipu@67Sv9uz7QfR
zo6K)v`nxXg5w-KZ40H<8XPz0Q@<354I1GGehBb5`o3M0WK*zy+D!57Ywkmfi4#tW@
zhe6!m_!|a>O?WbjVL;eqoukvBJ}d+FkS
zUKd;o9mgh0-bORfFW*nsd6BQ2TVHW#529TAN3?Sg_SlbnYuAKgdCc!_bVwReEF$-&G!m|p*H#ae9Nyrh1sb%1vrMhV-EoVLZ|x=S?w|Nd`Y|{_Sb|3h6VGlSO~21OM2SI(ZhPzyEx3Rutfn
zQ%oYZTfp^WH%kr1`&RA*Lr8zDa;t@CvdR4bqdv>U5QmR$4Lt_sCVQAFBfl3~a^6Q4
zu<7E()!44h&Y_Yx9^)E(?RdWNHsl(V1T<~Cb44DoR(Vt$)9)#<3o-qD9&gxWz#gv&
zu)u*ojGfc60aQHj@!M9w^Mb>f*e;=&$A~BR^DD#hMJdROh0B-m0p^Zt?-2o1U7iPr?TuNo}VU}kL*W3
z|CK8|Q~*2?P+)}fxw4x7tv=xI`fI*e?rXK%MrbGe-1a1%gMueq*^BKLODZgt0F)|9
zzlrC+kB@8jKY;$e_$kQyD#-4?AIV_5(?t`kPe8l6vPExp0T$P;Jdfq}oSb_W*U6+|
z0ea4TcP;t=*8f3A47OZYH_oeO;d!(#YC;y;{rwp@9or|uB?CJWoQED?azT84)qOsm
z(}r}u`Yi+$*V;P=*HQeg#VgUi^`z8n%%9gBQeg!6=-UaY4S<*BS-?agK8fAX$9(4o
zRj*^cuOxr*<%6Fx8v;Z0Ku(w0bJ`kEB{6vf`_Eu=mMHeS(21LMXn!F%H4@Ql(+Ney
z$K2gCu9?l=M&sD7x6A2ybWh+yQPh94v$a6{w^n;TVm7OY#_PGJr$3^+ApgoF#6a7L
zQxRts&4G;#;^VVe*r;G#)|x6{yB@2pY{hnoh^(P;xM0CNKGgpUS!sq?`7bCH=Wq6c
zqrY&Tg-^Ej!hBqwpXs^!#_~YV4s8EZopjj
z^(jq*$}%ruc)%`*b8h42@cDe~W=I#oVgX078bDbO28)CkEaQv|TryET3swo*sc?90
z-Z@6@z-l4=8^?Y!Jt^f5zbidC3Jlpul-1VKGInZ;U>v
zjnVH|#mMH^e6Zui<^$g3_}vP#4&~t~4ASE-$gsF*dIHCxu{nS}H<)yPWoBGv
z;Btoab2vPJot*xAmCM*=sKPRYbXkZth_03lhCXNX7tOKZa6Zyz=G`sM^N3tvO_(10
zld&JH!UjUzTFK1kpRG*#p=|6BHqk!M^!N3D%)F9&aEzwcOk(DZ>?3CU4yO(^f2qud_T}hNKe?v)Tg@8gW7E$VAfwZ=YC=yUUy^uSYe7M<*}i#
zpAm*XU+&hSvXCawP)J`LewCub*KOc9oU84hP+4tz87=?(xhg6LHGY7B0RHgN3}!mq
zt28qA*#}f{BEkMB$oTzSHU}CG{8h9L0tnpW{y9bUQ@%3oZ)t|>K1dHwZKrnn@|WS@
z0{o^K2NrnJOBTiAFNbM;c>#-H0s}9UDFerWe_i*X`3t%CQ~fCgxK4%iS*2WbU)r}?
z^bf|-=(G-6PQvZe&~Vr%dgY-b;oi4u7zP0NO}?!t?kkJ_?zKj_=)4JS*D0>$6c>a}
z$;b3ymPR|4cl2k+PLvgRtZjKoBa;GDr=Y!SXQQ!KiD;;Iti7E8nbE9ax2l{zl
zM1BkE6$DRpqQ16a4s0ZFk8$-qKknBh_8(W_{x^;55UfDhH}BNI(u8{4ouw85-fw)p
z33(uYRS@M1I@OP0fZ3#det65Zkb@1bTLxKE{C7VTZb^zzY<
zev=-!c>{mjwDuYRa{E`?ivf@H+!nL}%PIH#poskb(o(@<vY}ntm6xCwGqt{o{68&wS3x0Q
z`n$B=a7g#`O6K+eUj6CsxDxvD<>a-x9KiU0=F%ph@3{H1svFQWTI+DsO1OWqS2@%S
z_nn$i9{Yk=3B
zKEJ0znN60*dKcz`{cg{vdV#k_uV54X!$(WUAit|$
zX%^2S;Pa*%BW{6y=f&ZVJ3xPC=D-63$oKGUmX{;o@ntFd)mFp&xYZ$5upI6ozb-TB
zQ2}}S${Zfhf4JnnUYw^$iZ|?UCYOdbK>c+q{X#}%VBE{6`K$%~#C`o5bD%xG
zRyJoVq5axts%>l_o%`~Sl%wdM!)vQUfGR#s`(hz|X++a839#?$|LZ*q+NCpay#(85
zwXgf}Q}E}io0owApqtWShZ4ZqEjnL^q1<)TOqV1;zRU>?b9~?&#M@h#Fbet7T3?KG
zKzk1sw3Nz2zsYv%Y?T1}lbTIm8kOLjysObZ9{jah_q*=9Ans|Y&SJA1oG8|J-hP^aBcBc#S2j1`iZXv{{Xi1q+PIq`ies?g#86s!7#}i
z&w*jyrrZwz?egjd@f_K6snf_6FfnMXOAN64+mQ|aU?1M{=(YpM?)$p$3qmu$Ida`w4-1xfRixh#~z5BCtw7?&B#MQ!epf_&bd`n0l&H-PihUh?lq?FIE*$OCE
zITnl0v0nJ{KK#3YO?;+)oA?y+Z$Fpsjpv3#%EvtMe0;)6*&f&3NZfCGCBV5WXL<~S
z-6M4u|JQ))mp+<`=lPWFHc9$`zgqvTNBQabBsY8Tb8yUECJ$u3edk$e;Q4piw}BTZ*pr0o(SnHzQoVzhjHQ+d_JlN
z^EzO=$G9}u?e>i-`VD&fyygYMfWpZp+vfsmJ)2&J@ooK&K@p5EJ1>f|F&@3IWsM@b
z)b-v%%v`%!1LMa)V~Z5ZP8x4u1B3ClkpV^o&maGCU%__eYnk<9zYMsq;>UDT!;770
ze{suNU}7*{@{51qxc-X$qJ!tAm243k?1#03SwIZg#Nzk*I~W&*xUC{F{+u?QV~PG+
zH><)c25@fedF_h&xrybgFkWhsdoQv4)<=h7gNC@=jTdonj@$hRhy<*w?O*ABa_(pE
zHk5M&8tFdc_@|%xtq>7M_d&JHEkG!MSKf`K@jE&{j>d1Lf_7k5Ft4=!0+WJqPWA-`
z3gbHU+e7rjEy3a-)*q_=Q5n-$MN}Q9`t2|OVS3TZ&_;}}`DdK}puNps;{_N`Wo)fB
zqHI?rI0-TQ*aNz6*I!|Hlv`#E-DfpE!SxmBS3jOit!Y`(`43`U%|+Xx?H&amz|1
zsPMe`2n++D=4?h^!!1JnoV!DorYo;ErD#asw1MYP@U#)_ZO)lcKLvL%=R4W$!Zh7W
z6NVG?W=0Yet#0B20azDj%c%W5>9O*Rn{_t#I+P8)t3}D|(W9%$^A5puJRG>>xUf~5820)RCFr0u6w;9=QdN7>_tK`dQ
zdelOu{7EJZ-*Zl2{0i-6{9I%SW|03qb_bvjtVYCr96tVZfU$FPW77Y1FtX=Ds0QrD
z?O}NU9=yWn8}T&}kT>6D=WK#m}Ysfhqk|
zu3l(K>vfm69>?`+@4FKg8tgw^^3Y*$e^?c|2r)D$89D~`)44IX5JeY{(KzryA$cRp
zX@|GM#DaY%$2T4^)WyOF_lZZwfz~K@%`W*KeD;_U+
zWCr`}g6S{EV1My5c&&$
zFgEd_4%RcvbH!!k$sb?JK70p$RcADeL;#M&&v0r2Y`nI3_94(uy;oVq2YCIr#*Igi
zE-(5@>@o0^%QGh?;XN&Tz10e=Z$LtH1)e*&;vUV#^Fe0So}}x*b9OuX%y$A=UDxI+
z-j|D-D&-?@;o9R@YzjDfEy?^X*eNb~rnMHzUpinT;|S@m?beXZfQoFN*LNV@x%70T
z2kVl^0gqn(Iwx%>f&N|Z
zbnh?d+sZ_9enY;p*qx?B&cI)VlSUh2V7~6_^xM7%*7t7z@|@pLewo~k)#t!oC^p#D
znMG#3(iuK+wSmmv%_md=y#26SRMrJucs}X4EXBqp@#@>oBG9k1)JJmz;k@B%^Ka8Z7U`^h+&NsmEPAi{V*Gevj9kYD?zLk7wTIo3l#`nBb$HYWgIKPRjJvN4M
z(n)!L@ICbFMd!e^|KMEHB{M~zXMjbFLO%a4hjT#6h1W~3AA)=M%zeh@Ixt^y_PqYC
z0OK1l#r8P#w{x|WVDDFm|32$d9#3y2E^S9tZ3kc;DHNQ3l>^TMOSXnS$ZRF$ddKcn
zB|ak+LI$HTL(j;zPsS^wKD3hlVy*Z#xmJ?WDHr_sQ!5FZQ#$AN?dQaG|2n6^`gXEg
zXi29Eq$}Syi_?_`o~s@d8NC6{ooC-Yc7XMgab-?IAhb*AMpQ~5^!I2`_FM%xmvO%}
zObY}*zbp$I%!2deV4&;$L6~1U8UvRn!#vGRb9E5|`RBz;xAv4lTs!_^$23DW{QkE_
zhak*LJ0(AdF)=vzygK$5=ie
z*L0ml_Md|vDErJJ)(UU^IrQ5_dM#6gDfAOcmXIM`HQ~c*m
zfOfmS>ruY|c8h<`e!mOsJ~kaLJq_#B{ZyV93-p#tvRHm#zxQIim^;}0TalZb9L6GV
z&l#GyPJy^AUdysKf_v}OeW5vE_c`Xg&{F6>ZNbeZ&dDvrh;PQ>(NGqN=(yl8U<~&T
zmD~dnu#T;=ByQQkdC^&M!4iD_c51nzt`L+j*`U8>A>f;_v+yPZ%zqW1n}{)K5w-z<
z9*ev)QC9I0O2s%nIwQdw?fLQ-CaR!p)~q)X{Pqi1J9{1Eo$D<2$O7^v|IWv_Dk-6T
zTnFvEx42gdkbBCvA9&u-wS17c8qnjYRCGJ|8EeG08vz-9
zR7(@jl}AXd8}36BWOb|T0eO3efv~{*C`qEvbqw3T&~wsVR!jO2IAK3JS4{cb
zq%tYWx8!cAQG2zYV@{~=Ei9quz%s|vK;YrNN=KjatnG_0(emw&1oz=MHYlE-fqK^}
z4Ir?vU-`w-d&#skd1|+7#G(i7&W$|-MhxSepzDctgFEtoxWc?mzRHC-7FPi|{A
z#nX-w6s4mWp1e8;=pWb>B+jCKdRc-9^1;7W6ctOM8$eD@VC+I3_R)4tI=P;M=&Od4L%y}>P_*t4R7R~5+jG;QP+qnd~7jW-c
zm?nV1&iLXG%##cAseEE4Q;xPdOgBhZu)+roaPJ(bO|e(@2Su$;>=@8Hi%@i{nMQGG
z>1*0P>m8Czx;1kS=ySjaW)P=l0(}J3uMel~n72EGmN$FfUy7GLv8ev7Exbkyc9CZU
zD8IGbbCC8|+}Vk=KjMmxQG4zC0u<*i`=*EX!H1alcghCm(RuWJ0&|{Us^g=B`tybs
zI}r={&8VL(KEAY`a6UU~|0R3^wfo3-Fb(I?(ysIL9OCUAMC-8>Xry)mYV+v)UcYz2
z1+<%55v7FX*1BAx^SQ)$E}RH~7iQm}=lk0~GqfzfH&FH(s?XDVK1$({kk1ISpxoYW>YEZBUXOeh4Z{oz5fvM&yogJc%laT$g~C;
zH^e0l(73Sq?^R9YzY3?FT=8Bya>|D7Xs05xe=^3I=8@pvSWa!_I(UKv_tu?a<(mN~
zJLM1^9$SpnWSINQV)_Lp(9Yvst}?`kX<
zMgPvm{!y3%sJwMU_!~gIy45|#fRE$v!y9FAPgoaJBnH@)aM6D|U}ELDH|qe6BQ7Nt
z!Tx=7`2m~f;OB~dM}cF<2f{&)cV!J*1AdF*
ze_RfBP9CqdZh&7iFIC>k1YXI1N`I9f$Vr-UeaPb@<_xB!gZ%r=*du;G!3hhGrp|=(
zMd8V?r@+Gvl3jx8!R~@xxwA0jCr=l;KZAQ~b6Xep25#^-g3m(`_=fDIQ^$^iAENiT
zl6qi2e77yqXK*&$w|MX2H$M#bc{r=Y~bb`djt(Z;MMU#QAWP&~u(@5*ioM;4rs=`0eIeTK=Pz{E62)-|-0Uy=~tcU*)YL_r@=H
zxI{l8FAhjZ^w~C(+1^+Gja+^X-;XHv@$|}S65HdXTd==?#E+ddmOlTKT+Is2che{%
zGh%9<9CE5B%VSRu*4`;5R}Ke#{A^Z0PVnXw1{1UL-M(#*}OLL+0Qs&xosSA1Z38*foy`~9
zTu-(?xiU9vFU)VP7xz8}vk1SJ#d>kOCX)E|*VXqEpOP3)VXq;Hy8tXplpVmTBSijC!oMI7)*bP1vz$fc6
zuAX-WKE6tAV|Y98$-{(Ol_1}4fC?m
zlgB~qdp&t`V?^QyoYxdKt+tn*#3J!Lrwiv^hH)0~Rg^vs`@K+L{kHaM;(sGf_;hMD
zIbc{O;9=H84)=CCH3Bd6-RLqnu?_Ccm-QKM3#cGlzAd}Q!(Bn7+FV6b=GGIV>xwRh
zHZ0P2!E&-#62vuqeGf&*cmB|OC0uuV>)kF|!@PZ96eQ~c>+gQGY{4X0&+PF-I|9K@
zx4XOKA)G^O0$pdYTI)+p4I(%Z-5r;LoweLCIOF^
zE|uKA>_O7YxL
zns}iS<8z$W!d24&O+?vSwgJkR*2iCg_|~**tfLd`6CWMkjq5k5PwGz_$jc3jS$Iwf
zxS8&{5ZWJL;m=zM_WCck{cM2rBllP~v4FqAzHV~>`_1Pvg90`}99wqMEe7P&%ACS`xSodP);jLgl
zP5%qq4eV2{t6xFB9D9nyBA=YNVAf0I=hCO=5RC6N_Lft00q3)AdN%@2vV97#<^#`Y
zjr)P^b|~DQhWtNbH29+o_%*M9chzi=@7%awU
V@PQY&&+NR=b`AHn5dW6_Az*j)
zf?U2Ypu57DGVU*j1}D<}%l-8HFTx;yex)h14Ez*+JyP)m>?ZpTJ(>dPuO!tz?*#n&
za(~n;Ky{r90a$LMb(|#Pj_)i-+~;KESd_P1RbL5&1>(L}?I`;BvSCaB_10l+HJDyD
zi#G!01-TZJ5rFcq^ja=I@<$sE-{_9z;CTn6w!EcUwfLV$w__5rr7x;tFE_>BvAXf06?M(BO+zR!zjeI*C?tVjJTX@*a>y=LyK-$lYP1k&s22Nhr(
zu1%-o;v@pJ4bmkQ_foy{dM4lHEr)6P@>Y0^2Ko!1;TjPz%Pxxg;UI($QsDiD*Bw;m
zyK`zQ%HflnXuipFrD^$V=RTw7pMrhd`l!#gQ1nJ@N&P|d>Fze9{kbeyjP^^M|Hloe
zmpkKUfmofpjrP0I6EFvP9q)3Q&w0-u+Ryt>ZlULMahG~JUJd>8sb8}^;W!BPj)&!G
zfAFUt2LcTDEG{0@&r3Fcgi&8NdAl{DNbY)ikDC}RKZgA4=4LtCZomCs+A*Jvaaar5
z?aF^d-}8R-mUR)&(RZ!=hEOjd%I-lFHB;?JyVB85K$M}r%WcNU1M4$h(fTJSD|Dhh
zBVUN1eRxCbbDW=3{+--y4tRau?j@LRE|rskc4LMz^tq*W!e`27EgUs|A+J*g!iDRf
zy*tVu=gn#F8ePn{so?ZxoZtFcd&982#f5xlvHmRQV>-wuv)4|yoeMbEZVwydjk#IF
z<>?TQ_-0+2kN#@Bm0GL=SfVL2AJ6e4Lb?a{0ct<>Bq3l|F(pw9?=kEYDkWzE4(v8M
zYJ+-z%UJZc?#+YF2M|BGCLI$;9&Oia^%U=SC-wx%Z3ldGZ5}HM;>gIY@xWoQo9d^e
zi}5L`Iw=z4n)lPj6|JDpj5J-Z2KA}kuM~L)`_9Wu&4FnkPYg4=ejD_D=N}es277%C
z|4)xW|8LXx8!ZrDufKV*NeJw&y-v2km-hL=I@P`?yb1C}-0#iB_aXQP%#Xi=
zeB-YNm9!zQ%SUm`Y=Qk(ZOW4M_8@;6^ih4N0_DRO!Y9DKmKr)&{IeC@7er5ToDV$6
z=4Xok8sN1ZpBw!PfG>2dnUF5G6V`cAl>diwEHZ_CpH*elLTby7ygxP#;08!%2xP3&6~|_#~zT3^&KhMBMFZeBqn-pHfte=-t=8o+1Ez=
z&0T#~9IPP+8{h2QWf(%_t9a!nMO2VEir?N8&uJwC-zV#;Mn59~21@dtlS|3Orw+$z
z2QHH#``hns3Y3wo?BaF1gqny&WSE$)@k7!X?_;JPoKI#Xe(TYkUPW>Z{{724T1Ni-
z6K?Ob%Y)xHw2v;|dY9yQ3auHKP(;ocr_Y$z^n|Q;o7SK`xsv2qJoNs3=sLWQRd-UP
z`!3mf!+>@7-F=e%Tfwa`y@Vu3a8+-Xd_v~0@z6Y-bDxy^?{qhp%_jXLHx}3h+#~-I
z3v&|Ui^7`@hx`jqqr<%8Yc@KFDHDzJ2p(c{)o)FBR^z0
zSU?mfyc-F1DJN5ujug+#enRZ~ZR*1^%g9sJIhn(UACM16Y%Sw{$w`zTl$)L%eRR`zP
zkX1G}f2=cZfqR+3c|P#Gq`&I%nvtp!!W#WB99(pdh=uOxsYog%>(4&kZ0^!PHgySh
zUGQvy--p}r^i9lt68OaaSi49%Db@9m>DZr7-tZbt4-Kv+;(15>K22;S!PgrW-q*_`
z8s(8GH+Ak1cgZheTpzQEy7tX)q01{sgi-j-DJpej{?Ol78pL{05-Vp;lc+DCev+6pyk6gMi-5%Dj#|4c$)&juq@5#7)
zgY#nG)#J@wYAs~<(egrFNqFv`CAiM`GQ^=-QG-vR+?L&%TmwSzoUM?@PBXlL82j0o
z>F#JD7GvVq=AQ(9k>1-;1kZbhZoV1|nE~;7xAgSXo?5cC%SxztWi#RPc+`yd#`9O-
zaZx%8@B3Uk_A3J7r0X(m?$oXDoU+(ecnQQAcX`w4_`cXH`#|FihHvdzDIG0s$
zJNi7FQ~h~0#^iOOJ(sg$@$(Em>X(l!UIORfB5C){U^gvnWQH3*@RBv}P47c}uRHIL
zpM&<9ZTfwv3F2VTzQ8Zu5I>zvE(ZUe1oy@=FEy1!;rT^s%fWWAo7d**SO@(TT3B+e
z2jcE6o`$DKV4PK)f+md#!u>$b=kiAIYlW78`yQ~DxRf-{8Or@!Th%NI^}Al1u3QEF
z9Blm&KOO2%GZ~Py12lBFDuesZt{*FWyTQ(7;mAWjkR5+?%p3&!sUdwW826i}Nkk$)
zwqJKdGYX_c)ijMGv#DMzpulDyMn
z3I0tx7O&6%b}c@y9FX5OO>JFw1McMuwk`Fvfqr}T@JG`cu$zBj{bDL<4Q(88BB`cjdYN;yhM=CHyhlKLi}OTn~(cs{aDuA|+(u)K#>fYnQO3E+KAfb#aXg@7GaUAAbq?DR7pKQ}(-JdG)U?
zJhA-WA>+}Ab}ok$5jTXyQ@-_Zd>y?f+{qtE-*3vf_COW&lm0kTzNMghhw`yFC!kA^
z?&rMR0n4q{k^YSB^4{>f3o+GBG6?l|&lXNcjGJ|75Ap*id6J9#W%~SE@B%X2`;@7+
zqh8RroccA{H8>B?Gh_5ml?`yH1I+$Re8qN^%|1=}u;4F)
z1*q>{sz~##_syJz^4o(E^d7VMiY_sV_sP~T8bn!Ze@O`Pvk&^Z(~!>@WHr@e
z{b`a(ZxHqQE{Y?M^sBzN4Ebx^M@<2oZ%(t;E2Dl_?MZr0Nw(Ky?laAoDC7JR`o2mR
z^F<5hu0ua9V=C66+`I9sE#`~3X*+=B>Ai`*jq|PV#9MeF1mdIRpJcS_aTS<>`92<=
zMCX(7a-preK@Q>%*oOHsPw4C5`a8E=Efv@G*QMHfcL5$BQg1vC>*{TLr!0x!DUKq53rC$bA^>eBndVo3-Mk_ZTYRr?r?;*e&gbhADtCDSgWk7J7yvve`=Klb
z>_(S2@jHY3OFdo&&%3U67rW6vgFL&|Z-A%$JJecXU9U9Sw-R})gW;{y|6rY5|M{y6
zI#Et5&r^nks*0$wWS`66r`TEec8E_um
z6#YRF>i&uGcCJh3BVVMa=i`L%z`sUPC7#9;>OGI>34eyjR$up>;a&;-%RS
z5$MBkJauy)=yPvn%<&
zS6C!F$i7Ta_8IwXaD0Ll#Cd&VsTM(h7?&hNZCQ99Av`*B>Sx%0WFGSb>g;bP+*e*j
z*#Z9?Oj)``-3j(BBXjG7WENQ&RC43a*;Z2dXN~QpSI@}xZ(CN+uxTalZ>~}2sfK&P
zpqCm_(4Str+O4Iprh*ND>#Iu^gj~_XL5B(pLCPgQ6yS|HL
z@xwoc<8R8z4V_EPqiN5`kt1Gzu5$qoo|SWH_nks=H&5N}TH`fRU>JSdZ*eK9eQtYf
zS4=zE{^^og1aCdzxupO8Yn2!E7y&dMeaf?V$HbiYk}di}ZiEn0}0*qN!@iysjE
znMxw$c|7TUb7bCqhkO#cGp_!~$t#4v^y>X|-%|4ErGx3LfgI9*Wu1rUNGiGZGoUs6
zK^Ez4Y^%EKlSaDum+dy3mq)(2EV*-R>KziiS7^QOhir25U~ikGb_KDVc-e74wT}F{
z_E9alF`IZ5Wi8*Vnnvcx%&rQWo<-jL*v9+Gy^O40(KP?^%o-wAIpx9P4cSEMdXfE;
z-^G`($WDx$0XWE^@cS&rpnF{gG
zBkG?5EP4BC$*R3A#c3MNWNzBj+hs6cYuw(SJ{u0t544R>{{fz2A+R_2*u)xghNsO?
zsdx`p-rPX4W{B8H^1%AmD~DwX>nr{j
z>%f70_`SY{OSOke$lTYZ8`wp)#9?w=L*hIZ*}Kr-!m~3F&v^2CTi`qv>gFzP0_#L#
znQ1cesVK{%uj7HgOuRq+^LJ=R-^+PDBgPQds&Li}OP
zx*6jOc6S4J6}LgVItr)ixqw~tSt)5Puv^k
ztcR`OelTQT#w6gerw^>?*b4Q}E(*VJ8p@OH(z~n<_JXq`IzpkGowI%B;C;ij^89x?
zfPWmbLXf|x7sBgHfF-w93(JFD0t9!p}zGa>j8jb5K)xS~!`TCo4bNX<6ZhUR+#pkwn`FQuBKdWzt5}Xf9
zy&p>B^WU=VYvpnMT=4IFPGFrJHl8iN8RVWZFZ+pruQbQun~b1eLo}@4zoqizwF1k}P|9V^0DLkxz0VD>{n@GQx?nfCLr^vm)}NzK&M?mZT^~Gju>I26>a*)X
z|0Qeho4+9IO}wF34Do86mFSt@AZMJqaJmBYqmnYQI1d-Tnc^S?d_=(c<~J)SPgc=#
z(KkqMU8DPX4%lhN1=?Y}_}F9MsstGOP-)V7cy4Q)bKznz_<1F9JVFHci}&t7%J}|P
zM(e?LZQwNpk4_AC6+ql{bbeM1`L4^&e9#5;IOJEc6;$A!_@u_pGKh~l)vkecA7C9s
z?3r+jK>V`jeKQ-L>n6+XZCMR*`Iw5H>GKW1tB0o4Y_w&Oa(4-lpYVLEMOR_)ER^H7
zG@bVt#HGVhF?C6qEHb@&kXIevbK9Fab8wXe+ynTJ&vk$Wrf_D$mHouOpf-MkdAj@cJ0v>t
zere*HY!bWc#j*WMGs)FiKhGXmoK1FY`1edPD}!u#!~N!*NeRDBHG=~i8UAgvH
z=`P`UyDWH5TNcr}DtBRxVG%i5W49~sRW4b2%i+f??QAk3dCByrfp>{`@3x?Us7&(I
zXUom%$~?0A#)4C0&iBZs-_n_K+o8TTi{}Xg>Eu9#NcZyn>Ey5dse*U&Gsvrmt$BxL
z0Dp0h>P-2PLCn0bS}jw(OH%VYtoqO9k#m-xR;(z=B!k6fPEl*`k~c?HA3uDbMGA6D
z48Oh2ASV?scZ=%ZCEc^l#upvQA`?}+y&JPK$z1778HL)rME#k2oz?FwBGU08F
zNN1A7lLaUD=4TP*U&=c~VzS923vKn@_#EQXJoiInP(GR8DQ!RBxr8_lPpXtFs3l)k
z$HfRNV3C1X@02+aaNbZZo@{s_o7_8Pwe7H10nu4i?3ewail|H4vi5PckV2z3`VAv2
zvdxgYpKnVxQTxIroyDI8zkiUm)Q!KCc(e_;Dc!3lJ3pPt`n;fp6kE0EFFJIO{C#MU
zZennkq<8s=oB0%w(w;dfyOvgyru0`)a@`GNfNduAz%7>?+SbI6^Chao{HU^S9tkcD
zw{^KuLA(UE-`bN@N0z@|G4mwwc1h_=r|a&+z9*iQo(z1bLBZnU2Dpd&xyta97x0&g
ziaaZ`zbx`}xSd^c5AMZP>I?7dL3|T3`6)de-fx{3c(xtZKf7vK=gJSwWT-Z4zW5-E
zG^;9}w~7Qlbg`@Fl{my5!Gzljg<-z8Yejl`Hjr;MTpyC*{ndfQ7sn3zLfraDZW?TZ
z`yQ*_@SUO%mwzAMGt0h?Ty(tm`Lu5{k@#?McnEk;rQ%43nh*4^=pN}#;D^G*!W84wSJf
zA+*C|lR|o*CcL-w`=s4IDEHYW$?RWHpL@FHt6K0Ua*6-AFvROGe#ZoGpV{H8{aX@H
zZj-p@V<>-b!@_$wSIdw-3y}R@WWa`x?+qBIi$nUGl?leuQ0|LE%a7vz
zl;~BYItuB1Pb?Bg2Eq@KTGFYs)
z5A0&Dszp}-k4^4R7{&AU2ifPklcAj7=Ut7{z%EbkA{+Ub%h>&OxZm3DtcGs}g#Fg}
z&r#e*x3eE7;J%$XiT5b(!@0{QPe2~<>{@0M?(5myD*lMQyjzvf?y%JHe&n|)e4bqE
z0Jl0wyWzh5L&Lxe_xJJ0;d_ttDSg_OicU2P0BX)Du*82)CyBfyv
z?iWM)IdyCA+8KBcs4TEMTnf-Vt}MkHkhe&>elys;P2Mkue2K^O(k!%VGKevd0MvSL
zaWo3ha?V^aW3V&&d;JO4SIW*Q!27~Y2KF1!F1j-3^ld<=8M8M?f}Q!Cv)Xu$;8%IH
z8+l9AKF4=xx1{PkR|epC)DFR6K-I@0Z$|-ZE4hLdVf>zmX8G)b{3E+NyZix%5>#(U
z!o7&lok5`oV5ef7q*x2{`LdZottPCO%ELX4Y>;0a-QQq119;&P|MY`kcOq!Vk=>Bq
zCTw+91J-5p<}vms;5#B+HM|F=!u%_Ws>A;CZtt9E4R!~6J!~VvZpP_pyE5^fXZx{f
z+rdt4TdVN^_;u`8=+7Ns_fAfCc^=p;SbMu96!=V2&0eQ^uuJ};{XiS|i?h32Y8%9r
z)_UJ>Ga;TjC0O{?UV`_Ha$kEbn*z^4s&_=55C`5lh1bn&J@A{ZiM!pQ+}FMvMSdm0
zduT#B++uIxT$yjTT1*n++_9RlOiR3Xds?6k`Su+;^nYl&?zo=5H!c~KQlV5zBnk}@
zDe7owNqf*vDyyVrgh&w@Xrh!fY41gQ?=(=9NF|lFkbci|Ki}V9@7HtAbI*Owx#v93
z`;2>U&N%ElN}hdsA%eeyJhi$|=>`1CD!QBVw=v9veeL!xg8$l3RB*`~)|VP;>62Qr
zJIF$6d+t5(|IM#U-?*5=ez(X#X5vB{iS?A)vVo(AZxY
z%?>g}XHsUAQb}@lGDT%weo2mtRUA62Uq(#(=Zj9mdkzP)C0`DewG+p!S@{BA%gN>0
zj19friG)7WM}|4Gl<>-g=f{h7kXwHrHPyquObm1BnN|9wyE90>W!|cY`9e64^5D?f$PV(SD8QRv
z4c70%`Z6ADFCpD0WRJ%vWDb5nPAw@Xuib8!UkR)w
zV_)eRwv?rlh>lD~Zlx^p;joxz>ZNQF>!5G#$52YnMvp1U-mfAm;(mLRw`9Wm`pYWa
znR19=_wqyI*K*12U1sfqQYEBIG4$wUas^Q-b&UMIFN;jEZWZ#lpG!LTNByKP%_F9-
z&4rlPmXKX*Mi_PZD@aYQIB~M9Cj*P}hNf;UL_M*DULE@1kLH&qFD}A7!ktOq9Jw<>LfzesneJ`DWivQp<8z=QAq>2MIOsvs6gZS@R-b)<4d;2yp=t%PywZI|>w
zn9nkV&wg}+`Jq9jjMsq*!g}&&aOK5XveD*M#^ZoiA}q0dHva+i*ULQ#9v*NmpzeGr
zBgD(x(x96eYaxz(+H0u~^{N;#!E{@y&;U3?nm5{@jTw&RtHgx%cQvPGp-YD+p*PT3!q|pVkD-^tNUmm
z-{%ra&NH5R~dW(AdxU7Y6USFaRbr<6$|GGvhURrET
zq2guV#49RZx-)&cgz-|OUfC7(iccrXU|e0aDM>`!ZzQ)Jp&;(z;Q^Br}#}?AXc%rC1ZK39^MSZ`!
IP)xrMLd>+PA|9EmX9f+w>I`1>=184MU!u1YHKcNb?2e3_>GEat6!dAx1tZe
zRb^lO>N8*5d)V8~OOKbOqVV|+hBQHK}myw;ym
zRqBgNw3qvJ_!Q&48}tUkiSx?LnCcA5Jv`Bizv&J|J#fa`)c0Qn%NgjUa%Vr4wlY-Z
zkLh%)aw@9*s&Od)CMPV=O$u^x5da~5_LpDEFhRfm;9p6I6!HJ%ImHE!O7zX241Z<0
z&-muN9K!pu}-gcRGle0Q_ZOw_Y_2I)}Hs_
z=T`T*pW(fa%2p<3X45<3dmhT{6@1Wz7kAOg?UW>;@}IBmlc_AhTj&Pu?G~lrM*+`A
zM&T>?aJ4T_o4FLQ?KHR&${-E)SZsMEx+(_tjyago-CV_8Brn`D;%9-+Jdi@M_i?B%FTXOnu=u6O9hwWVhagOh0avz$FPPOs23zOl*6Y1#PD
zS_L~N-79>c;C{>F-gmgo8`r#{?AQ3EIJKnkUjz7m%o{h9^NItkMPSF;S8fn&If75a~tn~|Zd5vjgKYimU(mb;&)U`gMJeG=f$OE9bcZAKnn(jO8xELp;o%)PHGXa
z&_o5c2WQK>Q7Nnab{n48<31bAKZT;&N&y8O$Cx;|)HD
z|1sZ6RJeNQ`Nzi3QG{uSMyz@lIxYJkcIftha8B8OXGWqh+*;My{O{BP8XGhpV7E4p
zdgkWUtTC^z;C+Qz>(Yr_T
zJH+0hy*rH9pH36|)9JJPF*)9hmsQ@GX1QF3n>}Fq!w}nu3i(x#7H^N?5xuROny$Xa
z7W_ThuZLLWt=-0F
z{ov57-ZcI`ANV5$&G>lJ1{0;veQ4epx)}$lT(tW9>?Uuk5}eI!MQ-+56RN@{_s~tf
z30+;jPrBuugX^>Q%xV9)kI%(S_*O23pp|FPlCbDUxaMt7&dByUJmGw8@0cjD|3_>I
zGSQWRvNOM@dwjqW^}UpQ-GrkA*ZaY6f-%1gMHcO2Qn??YdftUnJcbE43;H>v{Pt$P
zCw}3$kLKc!yk2Kit*jxb@!n;eRl(w$SoSXTTe7=|y=Me^e)W0vM$8v|GppIINOv4(
z+$weMg^L-QZT+-mV}U7Z`u6M4ojpf!rAVbRdFhMzw~q&%O;-)k7)ddQU4K>Z`m%)=
z_d|{GhP(bU=C2}f2bGAqGa>4z%%8&k%NDBm&4g@yRa6gWP@8=l7IGM!SG)H6v35NE
z*jm`!pv?_k%Q}-j_}dgM$36eEkUE0jKdEO&=Vgw52$_ky#%qCUy$vGX=v$*kBOH<|
z8+PN&Q9i)j=n`^y#Oy&(q7oW+|HIgwXNOVq0bz9W7j^tjs+8#4tl!x7>rbO%IC#+Z
zclh?9dOkdHxAG5}Qqs{SaGP-WVn7y6%qjvYBXcD(!*{I(R)Zg1*
zr%O&8SKM>Xes>8sYUlNZpYQG)Og$!?KSy~ddS>skppe!d>`Gx={lTIZ>xxo$blcuxVOV#W^r3NN19N&r`K|$M2B}Vn@v1^}7e>(I
zm$^{a|5j7k4}C-=j1U)eqaW1@m|(UcLlBP#
zGW1sB`09mZBtD-fuhYj5fiHd>B1PI^Z)|QHfL(`?lWDTQ%qDgtik=6i3rY`Q{5N?U
zhENA=m2YMItAZwS$AIr$B(ngvgIxg+`=gBgOseqB(6Gf$Yza_^ePD@9xGRPZ*h*l1
z#L-pb_1{QOx7{D+d?tsW{Wo?ctN0^67mjjk-`Rtm>z|HRyC{dfU$l?3pOQ!R`6!;N
zzAuaIdOZE`%H07{kKq*EO>g4B;_r{sXU8kG25-5vd@^{e_5Q$Kw!&Hw49OM!G2!))
z)GRvhotoCH#pBRs);Lq5_1pgQw^h9kEuBAM38voTq=5zmeGF8zikRjBo20_!2NgRvDephHhw;#g*jAa
zmJP%n!yHabj{HvXK!(wNH|f{$Sdm7Ejk;(vrd(gklfJ9-cgh3}=cwZxD~h;vJPl`X?n3w=QIv&x;k~3Ppoe`G1iF8em*6X&16>??l+)
zgI5v5!spwcac#nOs_ZWaPn;t;(~tPXlx#t^2OC=c(*al#bK0dza(#HV8=NB8$Rxoh#aoBC@TpsTaN1BA5(cd~5g((^;
zhiP0%MshLnA#0;jOjr8Ra`WaEtj+9~=BNlC3KmQfqFu!BFk8VE*0^n`7`+;%7_5OG
z^mBa665)=oc!3Jvd!G^GaLZ3u@AhIkYW+g_(ya)~ezV98jAH17lHA$>m(BQ4!Heix
z<~D5XO=PCflU{_QiFY(?*BEASKEtHwSTFX%p7ZUrwk$eu)WbGy!xNlET|rCd^DoxyVuOj31d!eF)QsCl|UNO<>uBP0wp~
ztRX{^I$X|oWl@kcsIt=Kjn5o?Y;MtU3{}1GbDva*CmyzTJR|B;JSy0A&UD$0IR7Ei
z$J=~d9sOE=i&1~+AkMU5QiNX00G}akA8mSZ9^dXS!?=XzE|EI%(19NYoO-z^rhMoSSMNRauKuZq^3|ts=P+6M
z1Kk!yDf{X&J1HX#UyZw?(0mUeO_>%akZMIkOQ$n7md{ozWr;C@?{s)M^dasd&m#Bud+rTQb#E)GUZmVimw7$n*`d&NTojQiw{^qpWo?L{Mc}JI@YR*8#`0JlO
z7b?Q72jvtc49jp{jckv>Ur+GU>>@jgJ^gW(AODrc<`era`&`j~PydkEmmwj_b0r^j
z{;7*@X3IvGxAAR1zRemx+~}(&*A;{RJBoa~VN1M!O}^cAp*vuuv}sF;A)Z`W$N
z@clZv?PX~aV^A!r>y&k|CG-xuLran`V`ngmrTfW`eDy(h8^7D`(>{iORNz^Bb@w?g
z6X;@z$u^)xx~o11b6%m2@sXx#S6-t7ZKTkjOHc4+;tR$uzHGeosIzn6@dR|2i%`I8
zrZUuidq8CDXa(xEDcl4T%SK7}6~TL%Skyl!d2gK;ajvVeJ%{)EY+UeBOY#n@T>NWQ
zYvL`_1YAUigR%5P7;gJv<3PqYKU7~TTD;UH9xXj^^=Qwb1ayA?f`a`vUsUek-r(V}
zv#89FX5J+CQGBM+<3m@qCBCw0qHC+k2{eAYnw4eAOZ3wXPX5!PZ&CicKCb`7ThVBS
z#RIwb6Y*)ir@6?rRJ8I^Y|i#)6{z*Qs|6YZMQF$w(d^IBVfb0I@l&~+A$Zgn>dzAq
zg}!cC5m1iH!H2h;$7CdOaWmoRkVTCfxVsYTRO9#w^yu4X=RNICK&$1ST$4IuiiU@ZCCE*lMYShhyupdM`rXPTQTy3<`GJRb^(lUgHC7x4~|l@7=KD$Oi{BU&~^y@APTBj@{(1s;L7${r=YF+F#n}
zk&*s^7#@AJ)k$UMtX=@#%%xIenP`TNc-$(!<&5Im)xWoYW#z)9$4(?Q|2v4j`e~wl
zJ?|9peQVBLW_S=)4as0!l@!KZ*yDb3ACo}!KRKMfmd1rTTJHCI`}7C0;x^3N&?<-D
zC|}B(8(2Wn9M$#B7(QT?{QXTij~VgV!0$SLWq0B=!xym6wcNx$L`DwSNIzEQk|Am@
z{u~Q4a<)}x_=oIiE$vL}*GIWkn-5yIa-+9j-6Hlboy4lR)ZYhP{)-r;{JyfMXc_Cj
z^7@P40TVQVzw>F$zFAC6@KmU3I5+;~)0oP$SL|qwNlk5g-7=>8^vmx*bgU?sasZcM
z2@`Ryj^gPPaZSi7Pa|9Ok3PgAPM)j!>;%R-{O@4#5HntRAoNl4ks$=*)%dwK)QkkQ
zTN1xhwqoDxw)>90!x4w3#lS-$8_+Lj4j445Ct`j|9b2>iHDV{?Nk$?5I5Kg1*T?y7
zkFdjgiwb|~(cvp!_*pTlG3=+p=R%#FZbUnjCE>+A5+-ta#w+Ja26kd`TzI=TJ?^_2
zQxUuL5p(w)s4XJm`vM)0`@g4avE=eKAw#ZgjC*hW^qC#+vE6##ny1~uFue^nNpr)w
zNbA;Gqjvp!Skoxmvmf?%5teUr74qRNNS#eo|IwT*WH=%*;M&b7?65Xt&mH|>B(0}2
zw$fe^+soz4ut}i;yNCyWymvMg^AO?b8MqmW_)N-KTxxK`JVGkm2D7a&v1{g=FU<1BWasXE4`q`>Za*9l`qbcss3o0MZN5EAIvB=!XWBLsarlHB
zekJIIv3Z^DVNupY3@Ymb-SS?^;tq8CoCS|61}N?PqrJ9k_{9Q@U+J0lKa
zTawG;x&?T!y8}LEBLz}O_q`aC&}>CaF}v=0`G)hzbnv-TzgiDqkI~G&>!fC_Pr}x6
z0TY{%NiP2V)hv=2TMOgksH{C0hk}qfzr{9;ajTW?>qHi0%_$|YCwiXb@w(=$4d)!`
z2}j!xTVX!z%@zJ$9)Au*#FcsaUfmFBi_`b2#snKJo?TcqExd%aIr;Co~!6XLk(ab_5s8cIO
zqBWi2Lub`{8pL;Y#rMbeauWM0WPA$yo*+;S57u8YjPbnz<3l+P+B^sWHl7ooUTC$$
z19sa@q@{>`A5OTiul;C%(hutI38$AslIEQ}-IaTAZ|j0*`Nm10E?V8A5M@c-Dn4EZ`8596UQ2og{Xa!mSG`PzY)KwreP|yr
zb1GA~Yf=DOo#($?La%^9ji~}5;aqrk>u0E?6$;Zwvb^3}i~yIS#;&tB!a9hZG($q&Ty9-w>n))8;BU!sX4IP2yRx|UC_0t0MF`~0%U*X0L`VV
znSF!TfyV}^hcb@MVEca2M+wpp=!d@ZG_$J!n%lN!s)=QQ=T{&86_;=aXSuyjn+Z1q
zdDrocVK!|bUpVQ5n^!AnY%vqP4V%FQor0+x;(OB*{)?gQC2io}+mO>Fg;sE-N&3V7
zigsYNwPeG^@m4TdnTBw2G=n+U81Jg(A&|V~-t2Y9A&`|2LN{sI4K8gc^Ir*T141<%
zhs?TLz;M#E4QEa}r~o-;JJfr?{OQRfOYeKYg|0K%cl+A_9e3)eOHwNsTq^5&(X|nDJ(5)AEXRQbH>bC-HxA4uAJn4&66c$rGR!CZC_a8(
zF50&V#CX5=)&7bDJ5R}$r#={HR~X(R>W+cg-^xydY7L;X_;k2eWg`%BIP~NhAiitO
zF7}^#i35gfv0E9xf662A&pP(58Qd1NIGiw=MYUfnvd%7b~0V
z!I7_RfaAH9~)Zo)vi{efLKrE##`SpC)j
zb3Hge1~0JoL!hl*^q&(^KmRAY{e8bM;{Q0CmtMFKNxn_|F8d+GY({Rn3#GpiS*QueTNA@O}CwJKN^RFv$EyHJ~ucBimeiJo5X>yRaSQ@
zab5Z=Cwa%mao}nh2yXS`U_ogl$e{=aTisKa>F?p7LQKExcR2=9GpCRlAsmRZRnPBJ
z#X)KbfAUs4k`{et5&G$E}bBOSGzrOmxgLUy7lW{(gD?AoZ*Iz
zhAj?yio2g|t0aLXOp%3?*iVn8p~jG{qa5Ta9~X#|F9R=s+I+khf`JN+GYy{&>VV8#
zZz1bkC6MT{@O;x;2Dn;nk-Zz90q5xvSUyhzJL57>hDX%`-HV}LUcRUT45IFb+izC@
zj!P!QPTZxy`kK{=GcRhvCx+?=ox)Y%woC9@UrI61W=jML+gUOr
zjH|)Xj1!TEget(&@{^#uiBExn%&F%2d)dIQS;9v5Vmv70C9!1)mjS-nHZs!27K0@S8cwOf+IG@tFOFEfa{Id&nkEygZdZGwluxY
z0mm*?x}OqC29+i1r(=lomH#X{%D`z}xui$W$eyzdwKF#N`CA(KwZ{`Ro~R
zFBd0$e3%Z6_7!@%WE8`1eEL83@K?g4j2RYBZbSoB!G(>>p$2dd>h7j-I|rKiIUN(W
zD1n!1uew(Xmcu(KVG8DhN#JL{GdOQ#1>~ej7OPkDp!AhJ91qzl;P+c`S6`#GFcm#w
zA$%kSy!-VyL}q3$RDaVxA{LbkdCvA|kGNGp30u~{4=?Iq5+=OIkE;rnE=qNrx>Ezk
ze6B0MSLqE>~or(UH;wzW%HuiMNC?um(c{b#%c@f)_Smye`5=L%^YayYTOJ9a-BbY
z>1>0qOs?$@u5E`&wR6|Kr8^*t=5l;7A)_0v?fs0lLx%p^Fk|NqICHi>2w+|CAOrv4
z9kCvG)vze^#larY*}6pwh5N4YbtXHltQw24|Dh83_+V_C?0k{eN<~&qFuUQW-|L6ALfN%|D
zC{pjAH>`kLt3P*CTDQZ1$wwQQT3VrugWK8z=O%b%ALbV_1tGHexKXSm2~LISYrPZi
zgzD9r>Q4FX@SJkSdsDL(7@XVnW9Cr<{F;AX9O^=t?&e
z#S$PcPxoGT=oT3K(EQm-G6Mv0DE|#^E`{k;FNElFp1^3%eZiJtsc_FU^P0(G
zE-RD*ww}6_u*e#)e5ka?R2Kub*=-Im!1
zQa^}V#-%j^Z*eYbvDij1>h(&3?pp(}dS{w_UWf!*dB1IWr_cn9)+92l)0%+qZ5xFf
zT}>eJ8zbZIyN!VLWDDa4k46x=WlI=5)C_nowjM7%&;q{l&j;-bXaP5R4@*zEw1BN|
zo;RsHZ2^S%6YzJ9zVb5>oPEa>wf&ESF1fg>VHBr8uz-6fCBC1#iN)Fx4(GYF`U-gb>{N2?A
zT0gmjPJC_v2j3DucXC01CJx(2ZE6IzNzIDiU;`-ZS1+GhZUp}Z
zm5iIXTY<^QlitP1X7D6rWppQTE~D}+*WTr53_Qto3%fR20~G$T%85TH1#zFTOcUoQ~Xr$R)!g|Qk$l-%C?=HWBolr!pD*--*ssl5KC`lTL3@c+wg
zE+(FP{$o4hd#xH6rTq;bCn9de;e%sF2_>LdX7A~`Lrs9x_^u{d{eX+OzL834Csp4y0$%A#k2Ze@BFo1jo3GY@$#O%J+KfuT
z!zQyksb$^6@RCu1H9-^v*=0$ob6z`m$I4#
zv#Pv!7^Ew~%kMWtdWmz5Vl9u2xNk@TRkcw%f&4Krps%z~e`f|v{p==b&z1;hxb_d+
z)XawAkNo&v2S0^w;n)T`v2w^@B=WrLOEt8R`f|ZgJpmr9l6s~Pl>?t9JG@Ori{PBf
z-?u~VWl*R__ru?+a(Hy_YU6%^c=*LYd^|-o1wOcFV95S34|=lwe$u(W1b+1n?6VVl
z2ET}XZM6!^hPVGMcY^smXmgx>tl#k|Ob~pL!NXY!W0iUthXzaGYG18NoPIjY+_<{=
zZ*Ve1_E}Z2e9DCHc&CnSlq-Vr?14K3W6Gd~JeD~ZQ3a(YW$ee^*1{B@sEYK_DtJ~*
zv;zNB11Cz8zB*(Q`zF#w%S`O5fD>*Ex%4$vuvg*9j~4oBXyB;%+QhUDjyfrA(q5^B
zHx2Jx{B`Xa{5oNqBWhU&kM%!`R?4b?e{RfY>2I%w$h=awKuHa>-}Thx(2r8c6dC`~
zQ>Y9k9c164{-+!+?2qNq=B$DuQuJR%HdMnV{lQB=m&>5c*v;lW#D0{~Cw5)zRj+}b
z;UCyM+^Zp%xYWjvlhyD)#i%Xpuof~edT4)Nz~J*K&YMSyNpQKe(DG##37Rw>YmjfL
zhlyGiXFSa?couoYjcAeJq-Kd?))Wc;d)lTRaSlR_tXFrDmX>Px-qFxp@O}-fnaRIB$b-QMw0^qoJqfzOQaRR=ZB
zi+Av!gizp)#h>@>4bXYphPIcM4KSs3%ef?G5~SzK3e~<`3l-GjH&rZ?pnyV#riwfc
zjYJ0ii`?D-{ZGvL?*3U1?LO@Ow50=sQgfTX&4}QTesjljZ#4{VW-7@pE3JcG%nIHX
zO&DBM{I)?dq#ll4yE2!ILHI*HSgKFI9$sX56}^1A9$p}^$?WLDq4=!8X!;)#wCirV
zouo{H(dT5}eNwH1cdMQr?jOM5ly`Tt`<{9jw0B4HZ|Mf8l69f|!pT~AoPUqgce(~R
zT~soAX_`cw`*!0_Xf%Ym&Czty=jx%xp<Y+=cyeDHVg!gUVo)Q+W
zhuN!Q&vixXp*)k4qvhdx=*{BOv(O9S;>!x@V0s8oIT|GI%fjJK_ZDtT<$9Q>R<*Hb
z8wNc}Z+?9bFsPiRaA8^mgEtRdG2U_(gE|i{UChmeu~!|{$7W$!0M|%Jb@GYlyU!PPHBWE
zWu*B&7B)hg6#=uno_e_bQx~r;Q7#7cvvnh`pHZPr>ZeDIFnN*dUPxdgv>K(GnKFS;
z_}VEwnbZbY+31-?0;BFlZ$99qp^y{y8n^a_1EqFUwS39azv$+de`;KY4UphQ<_{%
z-y{V{xAR5_5W9oMDt_a@AdW<($d%MueW2J>9qX&G_JRI
zy`FVK7ixd3`?cN#8ss@xMz<$8OiX?}3f?9|c9IeJs|A}&(NPfPcv^}}_Ve)>LY
z{;|tc0-C-bt-gAiUrW8ze4QgyieIJD*P6zzMmq))ejWjs(el%o8S)$4WD#Uhze2t+2u*Czow8{k6CLgCBISMpTbf#
zYWm0*s^3QL)PCU3`9jIRuhmVdSM60kMK-=kN0Bw6pHXDoLW&{>J7^HM;uLtvT1iQN
z-gJV(5Nv>wpS<>!7+O=n^=)=FnGSq*lys|GP83$U@1ihNY6mfOOo8RLH^dN|0&Z4M
zD4bwVr|MO;Q|g(!y^$DFQ^2cI@Hv^v6Y4c&#+QB~hU^sJdT36`@8&WfhTs$ko-V#i
zW|zuYirf|Tw4N;28p&mn$>Hx$@k`w8oT?ik|E)NfOHMyHz)egUM88)cN638UZeBso
z@3zo9=uS{1a)=lrQvfXG^pj~XE_98|{mc67WW8y#FMG&*d8&>WvV=XBnpiRqB*=Xs
z)7@IFk4*9EbPY1I9v@61v)I?=Bbl%%gBU_nz_UGpU85b&yxf)
zIr555kr}e_@M#Z%md`n-$a;o*mAK+`F$GP5>#27ild1Eqp3<+sf~rl(ve~mx8!~MYe`S!V5FGe~Oy2x5
zVk(^i$M(6fk?lzF-QG%OmZIwia{98J`yiQf?Y{@e91Xa$M9!bF_|i$1vq~rTQ_?k#
zs**XqGs}uhUJvWPrwJ~mr^%cp7%Crh*O_3K=*0&v1P^%4rI4v-*G{?53jg~_xo)-(
z3n}-_SlBq_evD=Ok8+(wADtwo(kWn~?nm+4J+Rv%
z(n4l>)qTqS*UREhx!;nj&r|wIlJk@iIe#&bdcRc|k93gbeHmfY{xG{rsmD`Z(vj?^
zhCx9xnISz#Uy}76DW_8GMloDFNtQ{hCn){|tlp!L-)9r09etDWly<~K{-n@eos;6H
z*NYX3-_hPa6g|d*x0L+CwsHz3g)=Dqu{I(`$=~cr9C!GYZU#S
zH&YbOwo&J!EJo^lD1%bx*={+El0R)jowts=QS(LRsq<35Ej3+b_7_EO`|u!zRh`s%
zl&zbZ-~59r{|KVC=R7|frCoEp)OwAsQ|ohnLcJe+BCIL>=Tku~cix3s?)nSrJUKf|
z?Z-|1R6pe8sO1;0P}%87JgIzphRS|@s=YqidYDC9FEwc#
z52Whxmr@A^sQP91sj|pHDpA@x*%$te(jV>y)bu@lROBP-I$@ip@m@FtVE6sj5Z9OL@Pf9v58B*9oYfmt3T|Q6aEn2_)
zqWLjRqti~RpC{9(Zfc553|b0_n}Wji
zcyo&8=OitixaleR>+u+|SWskQlA^GQrawcIAJ7<0V{I5!Z*5(V#+x*MUejb^Q2l>?
ztk<_rVh~aC*W*57a8cxS|JK{F&h_|qUH-qA|37xtK(Pc+TXb$^;^>FfQr
z-kvo7mmQtRW^eB8p-p>Tucxo;uXEl0dVTA9>*d$|S-)=U
z@_PH$%dMw7)9NP{{r}I8b-U~RFGtH?x4Uj{yECHwzfXc_^Gh1dt|={FgeE)DWdAs7eV=LZJR5ENRM2E=FKT{51eI+xyRT^Z
z{4{w3O_rsNyU(=o>PX8^rj7SKH2cSC`lht`Y>+0;(fpFXL9MT(i%MTeWzQ}u?Qtq^
zeWZ?mcMWPB8hVJDUUrg7twbsd1$xNsXT^_EXoN(rK#wqgSYMI@b{8db)kyNZg83pvjXOugmVI>W$1&
z?J0!LQ}q1kdntN!kEzcg`l{4;dx)uzlE15*x<1me`%?7&#m-XFf1K^6@W2{Xb}VV8
z$Y`Psh4u4Q6u!B-MB)4{>iT)kp1KbB&Rn3_i=Wn@aKZW!MNdPBnIf|u(WTHeqLz|A
zlM_N&w{ID%P~!+W4N83Tk5z&gQd8hu@dzc3k#3HllKyo5QpKa(EMXxl}kRp$8GsThh
zc4Bmt_@zkgGcmx@+(D8Gsl&96rTcL8rmt(1Dw;vQDnJyLm)n^Euo|_2hiVl|OD|+41+SPnQTL7jp-Y
z<@l`&#MDZ(qupZ@IbMIwc=HT79#^)P$aN%Wbxvu9oNlM4lR~zCb>-ln0D?;|hewYP
ztXevpqeU=t8tRhE)p!MnT_iZt#p-jJ;EVfb_mkuO%1+6R_5?pGCM=NcG(FJ4$aX`%
zi;t7_H;qTdpCK4w8T^#&Un7`)M7Fotzk-jfzo)oxp6n;S_gmvNa=yP#D%pOyW?&yV
zJ&XU=xpM?7KVE5%BB-|4GE<45zxK9Y{}FUxu#q<>=;|GnT}|+q8aJ|?;DzV_`>a<}
zK($3ru91-c%8QkdpF=`gG>H&t3KZ$L5}^a}T;=A~Nbc9kZ@a|FvT5BY52e4xy!AW@
z`PkWPQA&HvKG2idCM~#)Ot-D`dENxej0z~vOObo--6yxV>Za;ha=&y+Fn=a9ZEw1;
zJV9g4Q^E%b7HE|xlk3~A$Z79GkbTQ_268>Vn+qQVk>zXG>+=a(_vuDu5De~T8O$c=
z_!eFvx4%k2af;k;9{b$N$^Hts!*67JT~`_6$uvH~OnKfkG1v7b`?*h2#D{!8JpLP_
zPtLa-t6d{AEUuk_oZc{-7fkkZG?(3uoL-wMN`w$ZKR9d3kY!(sw?060fR==QZm_V(BBU-wr+@vfcm3)s@C$`E+4x
zk}M%vlFCxDWlJhXQHrv(*+M9kP$`r(DvB)G3PmU>LI_cmH7ZJ?&CX+ALsEIq%>Deo
zykEM0Gjr~lbI&t#&UMbrt*9XXm$=MvnG>i5j4yGE>|f=>OeB7|(*8~r*RT6j--s~@
z{;VU$!HgMj#qc{U;sTcsze{Hn1MbgXc+C?$f9*%5opHShn<0slWb8V~xO=TQsxCl%
z?eM8HSnn4f9qaM?2Gp6o#S#PnR$d
zdpF)j*~2oS64Jg+fy#p5bg;4QO;bCe%*054?
ze&Son_6>t*CB#AFY*Yg$xaf?9;4*1@1gV&^{4uO
zLXSsQ+frpkPYR1$DHK^sAzJ{2{dC?DYeALIO;D(isI5>GF+~K2CJrk{#1e^p#hoED`;C3JZ;`g3N4Pe}_K48=XgSgi+78)B6QyNc8%T$HL+zW%%(4#?kfskI%wpA@%>oUH{3yp!WyH
zi?qW(1Yt1~tfpfGogYfl<@xFI;q>>GJ=z|A$TVdT{cj|Nq|qzxe~BM(SngXh7HdzxVmCpa1%A
zqrcBVcj`HLdcRTh=l_#SqSx2c_W_OmoH@Pz6TO^3=jZ1PsQw)2?XT$V`tp-=K
z)A_U^y`Gu=d@7yqx6||HmGrnf)`oiSFnztT(fdiJ$GcyADZA49l#;8V=l$6VsIplF
zwcqTUsr9*5nyNRArSO;rh12xB!p#-bJ}t?h>ctOGXiuMqbMsXBq$!07#+3ZnWon$M
zL8@$*LGgB4I)!7Rl-&|6qw+iQX;gj4W{RIjT%_{RxuG58_q
zLBu}FeWvV%h%Y7IbByZ$;7+PPotF=Z|Ix^x?BLTX%8r^^QSn3ji9*s}lTH<}`#OKA
z{%23qNWBNwbAnE{>dAN|D`kkkT>X0|sedg$O|W&6dhUD7D>7bU|B`s@hni!4kmuyB
z>qxy_ttT1hzVm(JpL8GZBmSz!=q>5DVX-}F57uWU@*B<1ljkdp+K8VI|D8?zWzr}o
zXt-``1+
zzhP{hYrcxHV7|5sV}*mt7t9Oh#kP~@m(5HQJpzx{68|clF;-=jGVB}yr3R@ZY
ze9Ih|v_5qBF+Y)08}l5%<+6$VG|U(DtnRnq@*m~kJZDbN2WZZFs!T!MLZWf)x927D8!7HAF_cY$vW=6~m1#WrJp
z*U?}-X$2U1U|JXRG?~doPN{%irJoPsyuP&=^8?K1RG(?TxCHqA=jMI5UyHm+J_$gD
ztTe)pevX@#_(J)1iTMf4yXRMx4i5t#H*;Md9tq|97KyJB;4-6MAuj>5E6X=XK>Gk4l)K%T?(@*L*R
ztEHz89RbwQtE!g-oNp9e7YFUa&3Okk0Tu477{T-CQzvM#32=P#VRkR5kC*Uw?fL@m
zo5*oT8PL8mSHK9`^?giOjDPf@krSOfhAn-_;O@gSelY&8v@4nepx=Sqb6Vo?e8S+-
zgKwdpIpat&$4B6w_x`L*9PUFw-=z#ghahg64J_maxdz^D8&1G&2Ew~Ak2Vham2v~n
zA!hrNIzUb%?g%Yt-@FjIstL;5Pet?MIKivaNdnJTRhI~_2H@HO#WQC>&xGJCmo+f%
z_|_6hIj9ers<6ZJ>(S2=5(Id5a{!ZNJDg*=wYoiRgZN`jT3r(KZzr0)K?%?|GQU;_
z&|uLBx5+5P3&W?`=4rr((*|~$Kz-4wTn-}kwu*7-}gjERRIA^Q6Z!pFbS^pzg?zG&yI~YZCeqYA>
z&&Z=C7t3FrJLZ7*|AJ-5HM~#FHI#rVz&c$t^EV9e>HN)sV8HI}-=E<2!Z2NJET780
zlI&|;9*gQ;(DR)SqZgisA)`a@Zvx(bSN{a}FK!yAjqQU3|Krv8`Ad53B#ynDpg{P2
zNc<{dFIVx561!Tzjx!pMpV2Nv@*1y}t|WF;xkrQ8&nAs}Vn5GbzCrM^|1YBN?)%-O
z-0ORs^dmF09w-K^`?gv`Jgy;kPak=&4nGoy@yd@8yWFSB3X})d&$eeIzKj*>IEc%I
zjd8@j@wE(*d1JL&bPm_^*?1B=pZ;Qo=)?8;6~V&Clz-b9PS1mJHWRshnUo*Ob))iR
z#eG!%$&8n(XDxFk`qYN+CH;u(i6JO>cMEwBf0i&(_Of&&`tG_rN!}|>z<|`J25%+k
zo^zheW2Rl{C9LOVt?Oj|>Q5adexlUUo8*i3?q(tL$umRw$=N-We^#aQiSeTpPY7(I
zc!`VKKhnS6YRdl}eL=}vR8jE&hax4vuKpB}J2X~D@W(c49y+xsKJm|pk_-KjLG&DZ
zL7nSA`&L2ZX|_~+B)jPiDQn8HlJ`uimm=~)pDalI{OjdpUZiNhNjb%V(vzDWKUvb_
zlQc7m2W_CoC;I&q4?;y0meBcL`%y}+!GuDF;}j2auBY@}#z@s)NTJZkk3x>hax#8u
z8O4KEub}3mbT8Gv*&rn+9Z2!O8)CIY{s1f0|EWSsj^PINdrM<{&bL18*w
zj*Y$^Y?e~;J@k3YqVwLRk(4};81kMU=;%ks9dx~d=)BsMUJr-tL|&O*uA{&I0s1-N
zf4u&`{67!9y@bxoQ|WQs6g?hRq`zkjy*+@Q_s(vo#<8c5XGxcHrO&6~9jd(}j{04O
z=<%llUH&FrpE^xSu05SvcUmW?a`|xz=jh|9@lxf4OVoVrze<%myeKr^L!sjVs=XwF
zLh0>Pz0`6Fr=+O*PDW~-q(o5r#c2_xr(ZcGr%jIog3PG>_S{{N?2{u>)cX6kk6L#M
zj#T^Ll3_CMU+8+wZl>O6!HT}GRjBd4hEaKp`5o_w{Mun^UmML)Jl|b~%Ez>XKOpU;
zYbiTe$@h?ylO?%`9u7KG-rFm^l9YE?Ws-e;ysMbV<;7FuQ=bF=lCsF#0pcGzZBuHoee4OO
z^5VjLrKEkz(1zgfnM`6g-$_?I#&W!N(Txar
z^_3=|we4~m=80CX;$zQ2`y^w)j3Mx{vfqiP6oHq1)@OiJ
z1Mqgv&fv9x2a7+MuK~UmQ9N+=3SjJRElU;PpLgams*9n$_mkk7Ou%nGqFR{0T`FRA
z2?AbzUwi$c?{l@BXPymwD-`@;VRaz
z@cgFFnzzOWfbVxP#qOzv-;1Vu;(Q6LQ-%1lX^pMrjKe8P43`_YmZF(rqIezf@GeTKw)
zG;|>C$GuCy|35XW-c5n`xldEiuf)xzp*nU>IMxR9IQJr?m|PuG9r&3PQrTOTpu4+4(C1jwpg049;&{&^zGfEoi?>YU$#W2YTxmY2W)1fyGn0Nig85x^WAB-*c)o24%yt=d_%KON4o4BMVFLmX)Ihv$MejKdz9xxyX#zcO-tZ8zwt
z@Y};z2gX^Pc54jplLyMKwRk^-^v3pLJF+8eO_3DfaT|cj>qSN
z!**3eeQH)02acbUesZ&5yS2YM@Y`AV-DBMZ<>X*p_=cbA%>nr{{J+~iK>HEJ^li5R
z`x8S2a2$J6LmH?QtoyS3A&lie>Pi0O;?Qt3E^nPbTZZkA&j&W*e}4M0>EZfRv$ipe
zcijCJFmiu=`VZS5wwLUkxLj3qFbmr`k4%YK{GN;={3mfe$Mf^aI4<(2t9yat$)Vt-
z%W?hInSs-IoDh!aGCY45HzY1}0D7NG8^rcPu=X<#wsW$rA6ammc!cBg8EntgrQ1oo
z;-tTbps)fv@mEvp^oW1k?zWlW&I|LzzlcaLCix$+Tc~?3&f9Q*xn`n5WcUj#_evBwt_gHGr*cXgWpeo!B`Bx@7rlnNagpT;WEv8
z7~e~^c;VmAN43fZ%bV=jMf{rmU^nqgNwN7Pex2ryAbKvC2$Okk@}qb_=UwW)8Kd_G
z;@=m=<`F+Cp-9C|C;N>_y%VcH(XaEzVp9HBeu3n-9Q0dBze#sakoH0yKhhqzOpug+
z?rR|N&`@e#4c527zJ~AzHAi_O*MI0CiI3zikM-jBTwmu-=Bu>1i1@YCl2kJ9Vjf+j
zpKyuw=_K*<
zmGnGnu{RZW*=cnVM@;f|M6LRd?c<<>7{jzLPl+B9P1p4e|pj5BAJ6!`zu=t
zZ=Rv@yETCne;uR8U#lyq_R0Mej>VS{eM00Z-nqn#lJ`7E_1_jrmCZ#c{)oJ*h`bjQ
z#WUMiQ9Sjy9wi@4k0ZX(-=l@j+tumk5859n{Y~in-;U1PAJcjHK4nU;r*2e#uJnAm
zS{PN&vyDO%HcGG7Rw~b}luOlLr1Sg~5vn}9p89=%c{dQfr-iBKD(LxtNBVuv@Kcmt
z94GRKTqiR%et8ciZ=z1M`!`WMzTRG){C=@lDLHcuYW{pwsrlg29w7ZN+x#L}L_ZhY
zZAkHA?h5Mnsr4)*?UB!G$hr#%r`F91y*^Su@_~|<>h&Szzd_w(e|WhklCql*AJHQy
z`#LEvY@_VVjbd41S7pX2Uf(-)o3y(w(WCU%`%1=JI5tS`)3C%*_i;po0*IU%$0Z^s
zCq|uzzm;Jp`?_UIE_u#G*Nd#@PCZ3Zj~HzT?|x#q=_^_9e8yzoW@SDg@=4Adq`#zw
zLuB78DEtt@azC6Bi2XF_U?KNkzIBxlJCe?ueiOIX{5<4{?Z6=08lq?Kp9o?Pa~`%5
z9`6$sK*Ms#bR4c|!aayITQWbaD?P}*E7S{1!tGJ@*0R`MuASeq52K9U^8{SK^y5S<
z#(Cy5#IM*JZw$rb8648_#s07aR~P#DJ;s0<;C`_;=+FDHYYpr
zSIT}05aI$q3
zhsNO?sK2Pc^r$r8;$z3G-vT=4-d(l_FtMKD%pKrW2kah7;{17~__{UzfbZ2sU*LYV
zUgj-+4E_9Rxl=X@{71~I-5m3$lP5olKL9-aUhl*xw7)b^PjrBBfAQ2x<^Z1>;L2gx
z2<=Om{x*Jr_MxK!`z@i}lPf0zBEkQQd5Oxwd)!T&
z4LAe*^UzGqHYRw_*yG`uN(`X)y;{zGIf!q9eH>;=A)d-Di{A@;o43nPK?mZBCNG`0
zOk+LB{nF7v_o5!8zii;xrgy!lJETeID9mHZeb050tTbf863>OEyU`G@QT%qzZq(qM
z`0A}d4~o}xxH%vRJlpQtm$wWw^rvpQhX+qLl3c#KP~&13$~mjW-n-C+(pQx6hCB8m
zk4E;X{`>G9#NpT5^C)D3>XBN#6jxRo%sldH>nCdEM6#Z>kLm2)V+09f*~ZzXtPjY0H{-
znNYqGF>oC7c0-AiA?qP-N)`AjhvQikV_w?;>-MVC_@4lfd)!>x^%ro3hW;Sdb6H%S
z6`o(mv->%){ZoiMmWu6|dvFR|S%>&VaN+{y*?$6EE3lnP;B2nO^5@Fe`D3}Bud5GZ
z>|BtEUJhur%(@;wx9+O@M|=)cq}aoZ&lmNyOrx;f>sJ13Hv#WCw~}wq3Md~DDl)-&
zhI;uV4r~t--;Ycj1@!${c@@Y1`NOswS)lB}by5}ELz63953yZ3!{=Xl2ynCU_EIR%AJ>Li?89=RNxsMN
z^J&%7S;l~QU&KVPp1MXoXYf1-9E`kz%kQ2u?Zk3FBK(M-^$~bQ^wHn`hxpgcM$|pp
z%Tr9m&)zgmC;79S5Ow0Gr&Xj${h+2B@$X+gQQy0npWaR0=fw0|BLCY#i1f=PtVHDE
zQesH|pZS>~bqo8<*n`v?Gma5|9d;;$`0o`Z{N#KiX@ww(!xogk_u=<*JHLnA!=ouU
zk$hU@a3kr@+Jnm9TYJAG{bg*Yc(u4E#j~{RC?54s-i7F0!A0?_9YGXt@7qrCzV_=B
zZ@zHXlk{_AjEWv&Q=yP5i~u*7o<
zc^?BtCZccGxx+*cw_V9ZPT~P|Z<=i*#SdkFP~!ydq{>|bl>Yr(R6Hf?N7YNy&pGN<
zDSleVMDa<_Ich%U<0-z%$Vbh?atW%ui=TS`j|1{#d@=fcuy7ic7jFMdooA)YULft7
z&Q!iK$?rWWD{4~t;@hRuq2pG)T-TsL%3da%O?Z7&HdvAR
zdW%D3-FaTqCHYpjKL<#j_;cLZpIENqx(;b));K`ca|8DcA{Trp{v&SZ$V?~akmBbJ
z$oN0JTFLM1Ug1jiq5MoC$MqBPa
zP~V`t)?Vy&AIe$by0#3;!qGCOkVkKNaz7Q?
z+qf?4m;-N^b2Mqq2Y=aMcJU%>1;hyqCgH<|G(=-iYAyH(e#Mb*)tf%>J0C|+NI`zM
zB#5uW1@g(CwfCB(&H&%*$vA2Ye!1(qLzNt0cyI0U5a7$zxkty$S!l?#HtS|E@VFCi
zcs7bde$8h=EjJRz=Tpt8S~3qjAl#*nPY2dd<#g1=HK$iBFihR`C0
z8O0!oD+;^bXt>ePnV7|2kM0~tOOyQ{$eyC1Voy<<=bkjgtWy%vQre4NN=t=GE*(YA
zlF}m%N#iJAe!L|;lZGr+#2?5Wg!R%gmG4o|hxVC1o4-^%j&3)9`r7k-6sZrFYUvg9
zBERw1OFlLBBCXXtMLmfls6k*tqMCC8<*s2kBBVBox{j?!^27bet8tRW^Zp2m8Hf;I
zkeonMD_4K3-9L^_uUMTT^?e+v8l=UKZyZOL4(<;S^chFe(l=A)4vwQz$z@;S7f+zb
zfYkH5frz6)(%(dESYrwiE{`owzL_o6KMzn#@_aKA?~e@>>d
z3mNB+o%&AeMCZIFTUB~GQB_jz(U!G6NS-BR-H)5S$WVYoqh(_kG8@f2=*iZJc(pd%
z=a=h1bstv+xIXGcE<0)RcEDfVyajz!13J+%g_#zbWe4)!#qf@|z6~jx?H3p4=s<44
z5#0+-9msB=$<~my6GeG8S29hsqX$1CzHQ2CLuoH3e(u}Vj`)usK753;1MQ#FZ2SoE
z2$#ipftEH6l~~DaT>|U$BzMZic!*~jJgiTy1s>lTAo=-YU@vmHn{8mch=x9mZ+QEK
z19_=OX~U~KsU1ky=enWhi!Kzzf8}D^X5gI}8VzqZ&``~WvXUy`hf-gKvdaFoqx-&x
z#K-S-A{Dp0&#syFq5%7@U6QcQ_*APF+e7~j2JRB(5I^wjusOig3;F0#*O)D^UR5?6
zW5;>-Er^?cktG|G;CK|T0L@c9yrPN8(j5-e6yzu>!UK@YIH
zgfaFq;Ev^oZ{axPRAZIFEzmP{_tS1{Kb_0!Yp(-t@j|uuzM2s8Ho`|o1ID)ofIT_0
zY}rYB&^K<3m$m@y?$a*uczlMcKYRHBmAAFJPJ({ds(&^;h4(y^eSaN}8+fuR`Wpcg
zgC25?z<5K?5_=CqIaR`OC${H};>a5Nkt^@*$@dKN?QXo-U%XWL5QO8BF*8;r>>rXV
z(tB`yDoZ*q_D^aB6D=4s&J=9HeD_zkz9zQ28?x`GU_Y~>c7S|u6xh0C1@6CkF3$$*
zdHswzKh8%iOnyI)`F-xC()ZZU4CL#>(La1w-tGVBHRvgkx~II$jX!%@x{->EEgV?Amte-J;vpx#RS_sQ8+gii>1G0I`NGWD&54@l%Y5Pq6&
zp-K3Fh?pMn>)rO#dRWhz!XN3POdQ9c7bHl3%?_^rGK;)i}sgU=Z*?fn{U7Y$wu+PW%4(>l}x{LH1
zq!vebk=Q)-J^c~`8S>o5g^XA%*PpSL@bH$aqJ-!Cb2lOVPc0uJ`MaB!8p*shT)s=@
z!$Yxy^v759n)LHf(VXz4eZ>@?ED@yi<)eRBQEIn`=vC%R@zC4jFGyMVFU4cmON@{*
zAJYbc=Sn`3{x_r>`*1+l(USH`C62Dn#|XR4^C!S-mqvp;e*W&5o!Ngqv3YFeR#d{Xm~T={d@<-jN<;vwzt6*f3Q32?~#7KB*_b4
zIX+WI2h2O?k5=|zKDW&Ih$fb^$~ppn@(1Ge_^6w>-;l5Vu2?RvM92r*H_3tNPCWjr
zRL;eCKb5>&tBaqj-l7+YQJ~8==pL>=7IOrb`?qyH#d0@C7Sga@*_VT#VLP&qEvDcZ
z;2o_k`|vrN_G!jCY&V9DrI+G(Sonfp1?FpL^x2Rj*u9n#jn{a57cUbQyzg#*+C7Ht
z+&3nBah%6IDBzKVpU)}&44;<3e8*`0#CnV9H(ted=9tTALkW<3wWxgJ3)orD6AvF^
zyL^xF>>gc=j>kgqb0LTKU+V!BFPuxh1^J!8nVy%JCwek1isb|B8(2^r1)OZ#@$w_s
zu>;?^mc0eP(lqF~8PDU|L$dwC(Eov$mK*)R7Zpa^9!NnRuz81>#6$w{VxN%#R(MW4
zG4JaUEa&RK2KzTRuRUQ`Yx)q2%srEfN$_`Cv}{gdp2-w2U6cv(+3&M*)1W`~Sq5f1
z;IGOmk{9-X9}^kVE%XLGHK(S8Ucvh@nfrX%1OBP7c;7}v;1LcwjLqe;G;}1!MDrQ=
z({rEI!q!7x@LpM7NG>
zWgJ~^$gW$z2hR5cRJ~sE(2%PLBd@0h#0w1(1F;T6XuJsiyX$?E=;Or3_^1aoRK9Z2
zQU`cn|E^sr0)>6(`++yx>)wnZ75e}$&%_BNv**G$0RtMEddg6h1^SiHELFEgLH;)3
zMD=#}2~@k>`1Y$K<49^c>seG&JG!uPhSM~n6DcNk2i4m3B8$Xr&O?>N*?fZ;H^+
zw{=aUT^HKXdf7>di{hP#TX=|j;e8KM-Xdi(%1A?tmbkaq%F|F3KXZVSNIOc7t-QVT
zKnK!tivQpq(T(a{rqnL@fWP-@IFl(#LzM+QY^lH7Q0Uv2jFWrXQFNjP8<$cSGUu8<
zx-YyJsR>ScTrP#UkGu0I`(PVlxNB33<$8p*CNF;rABS(SA||yy(EQOuu%Rzx;+-+Zf=HmbIo1uJHR?Txn902Oi{l
zpzYsNh+oc!HtyXt0{f0b@T4Zpmz%}I@s+TSo$OauYeJl}*HN_J4)UEt%?I8#S;KnP
z&u@&}OhZ;{AP<4O8}4MxVl4!2K!IdO}Oj`$i3@NI4chGT!gGF@co?F
zgEl(W@Z8p3AMj{}
z3_0IviMjMG1>`D^t~8d0_VAnWu2%t9DoM^_9=YxBwnEH{_6k_O5C?oHpQ1Gf^7=9N
zR|i5_&X(aIw$u4_Z~mDAzW=(yIUn>5KbIxV3gz>2Voz{C&X&7)u>9?Q`;P|zk6f}C
z#(XT5>+xG`pHD`{OXGgNvA6eP6lq^RhmrH$n>V<XfkCegnXq#0%#-#Kr
za=xZw=T?R7drD5CGwwHqL2Ey5fBf)j2*!D{ha?|7Unk>;?J*K?C->nap0n=8{$}6z
zgih=)#I&o*aoIIg+{Xa0p}3cPS9{RtL>AT~W>Yd;p@lp}b%qI+`?e~Q@Hge`SiMvs*&(A
zAIU9*5A{v*RAGICbu7s_YwoiF(yrz-M4o3Kj3#nZ&ihF|-RV@?BP^eK?l|FDE8lD;
zd1VD>f0EC;^W@o8+@2=EL3oZ
zYMzHa2E2PD+84{6DC#DBXzMlZZ+N`y-#;GUIHD=H#Viic)Jh8`=PQ_LqC%?}U`Y(+2aQe5|KKRRGYMf7Q*OfE!~C&s%~W
zxh63>*#!KhfyE#-8tQ4LhE7U=UT0)-8UPz5m!BB{J1p8$U$p`9aH;l=y!)YEv|1oH
z2yjv&c=JlYP}adteD0)gN~2u^KJ)mOc3(Zz>pMsw7Z`_o_qoD1*e+$Pt9SzSze@Y=
zDpUba*}8aT7?i)oJvw|E#;N!l9J2#>&Z>UaZ=&Gu+K#8*yO9R=qUo!iJn)yXL8+HF
zAZ{;i47r^M{>WiyB1RPC_1?X8;_IcMb32j8B>1PGdwT|?!S94$zp}lc66{o%X(2B^
z4cP|El}9Y@M_Vm8)%3tG#_2Yz2g7?Yo%9~t2YzqPRHi-!{KX+>roy)$deMqE`B|3G
zu3G4??GE_FS{0(`=ARDe=kLNHy?SmOAyC%@HLs|_c
z=f+Ww?dc(Q=!fySOt#+kCU^hy2OVN@Q=tWoaey*|o+lE%JdAhLSRT~mKt|@zHZ#(+o
zBJxvFqZ1i?ljJ@d*oDqWd>Z`7)sBY5^?H>h+tIb(pWM{D+E7WUd4bsLcBFbyePf_s
z2V$1)oVz00j_y`HY*FQGM`^3H*G@9EBZCoC$?4LL-lm?ItK{oIOFK*C?*Y#`Q#7l-
zbR!KZ8ZQ=n3F}6QZG*AK3%Iu{_Jhj>)|=(lqf0^;dQj%D(VOqEZbz&PF8}7EA&boY
ztf57)9`3KS+6wDcja6tiAM%Og<)v(gfJf?Ie%Hziy!Y0sN80w#uYPt`#$+kvmv?$C
zaog8{B(zn>i)_1)toURIFDJh5xn+gF6b*@dUcq>B9{7UJ`Ad8Lwxc_<@LymDcA|S%
z`^&a%?nVEW&F#9zO+&e_oLJTZk1bcYs=jXw^2}G3oy!3J@_npS%?fyqwcdrHDcE1C
zas3;&!}_ItJA88`)O)yGOIr%--P-+>Aq%t@)ZGf#hkg1|Z~5vqAcy3VBzUtSZ+>aT
zVPRPR7Q^!DGqA6jhj}_nV4wEL=B99g{I29PC$GSMD1I+7G6DOHm!T({2imotJh1oz
z^0iB8$NivOsD3KU6?p9S?G1+?L%iZU7q#CC=4)o`V`ez)!vjt0w%vqrL$_UhItF$k
zNh$h~80h6`s4d`o6hdv
zjo-`N?)_`b#}fXSi{t()g2mb~kE`38X5SC-R?(EjI8F{vy!4nDC;&I7Z^<>;0(jGv2gqBzZ;dm*JnWKCYAJh(C@Biy?WO
zlqb|ZHSzUx0l2^Vf_o(2#CFes@P>6t=9T!lxRAeuUwzo37>nO?RCgf|%h%>e8{qam
zmrqP!^pILk`Yl-GIfmcQJEI~L`&HiBLrSecxJFbZ5XKX6rC0=LQSvIfnx;SOz!i(E)
z^A8Z-xaNZ&2bNR5F25bieShIshxtkDXIcxEPyJg)<~8*CPLc;x@69#Aa)JL!2XK7j
zq~r*HauC+1&O>h8&%q@kDY*ZGL5yVFx64`I_(EA|eI)sArNQaKdMxLDllptKdv}|E
z!~HGuJ+KVt(@V4hna%@BwC>Au2V7x#YYgvevu8De9|0xn+a4-_9atxK{w6;6)1Bxj
z!t24~O}IJUkDr4SBk=yrE=+ut3aI@2d-_$tyIogVu%4wG4u|9T%zT)+WEwEA{ogx$
zzijy+Q`mJVPrg58h4a$fBYV~GdbefgavlWykY`?R1adJvcK2^U*=%=w2#&u5)}1D{m#s}2YxNp
zIG<}V)Z2mXSKkcWtRRR7#@DdedZa+J>k
z{}LHs+;@X(5S6cQll@={yyuP0Uxmqb^g^l2>hrlyWX6`iyz1qK4)p%S`p-Q&oycjqzD{C&54zR(Ip_qOXL;;Ss;LG3wRn}(;ioknXfe`i
z(%jmK4(13J8wvLyulDve3Q3T6qZN!
z3jE{E;5>3r@K)Dvyq-j&Kl*f{*AG@k1b}{7Onp0|GrQ3k>-@7J9mwzfalW}4_~#3T
z0G{hBAn(#*_(%-=FU>bFVieZXPlrdz2>j(%oA?)BVSjW7_%0h%prQJ;C5`e>U-;sV
zP#o~eipPo>(~w7b8@oNL3D%jofSLXn$FdzJR?_%fu
zO>8uDeU*uyFXW|mNH{eI!20mLsV=%3c*rcv8ngBA+|kHepK!j@E8A}@8Oq&fasqIk
z_-9PdAud3(ay_OlSeKU9DrwFj_fomgSR3l&8Ff0k0pI$_m9hhVbBE=nEvjZ{U>xB`^l>HwD(nVtY}}
z(7zJyOY6!ima6Kq|;
z?F%cHT*LCZE36LSypY)HZAu%Vyk#`yC$`6XzWR3J`?;@IRV!mV_NUG6JNB1jr!>et
zV~-6|>e!C&Vwfj+pU4Zd#BX{W_#upW!eQk@oESeeDQ>}Xo6^*Xe;AvdU54#@l$zZI
zY;U=FcAdj=-=|!xu^juUaN-|!s{DaZkima!@QK9sQ$;Rp2bNRj%p>=5#4bepVfiyA
zTnDh+i=?frSl`)bJ~b?NIwGFL`Lxk@r-)pewI!Al*cCziC3^jc_{%X(>Rjne+8pth
zj~!Jwu)Le+0pkA_&$kgCB_KaV@;&F8<4AskFR7CJ9mPE1mE;^r_p=d^E1L=-=Q4)(
zSgUbAo4q;6{h@g2FT|gFY#1ctE)Ays?u_^bPBQMbrTN6~^3N_MJYr#hx(}6=&O`V_
zP}OGQ_q8TIx!`{Dqyr!0`J1<}T8H0@InO%~*C!M-eZhW(Iq}U5=Bsx_!eX$T;jUFW
zSZ<5xXEQ8UvoNZF*OL=RfivdEkJ^u&!Ey_Xf#e?2=Fm3s`?@mnD`L672ZhG30d8Jg
zRfOd}M6-+I{=W3j1mONEnWPzs+{RIUJT7ZT0^y;W4R^@z{%gB4Ssw=(-jck_=BgH9
z?8o_~hB9sd^4f+)<8|i4@x~7S-fTnNX+wa_jqdtPus#iVO=2RTT)RZ3vLVusG5x)1XcX|9W_Wuo5oSuDvSJ!$Q@PSJfQ1G}){fVnV?YgER15buYZxl!crbL1SZJ!A#tY$i5y
zJZ_Ya@jHC)*7d59NF(^eTURSt7@&Ny?((lKV3%e!(}j0}zpM~%tj>jT16HrAvIM!)
z?4Go9AZIbmSa}!X5tFS(r;0#MtmFIs07Egvi-!+m^Bl}_8z@(6Gx*!*NBY&+w#7hpDTj{e_};0y4V>YRZV8^B+pr1*Ak@Rv^n
z3gdJ52a%+Uw6lf5I1+JuYcQ-rLvLbsmOh93+IzJHN79`k&vb(Er%2HtvSn&uR&f|d
z{(Ksb<>7oGRzJU@<~QUuE(r|yy8>^uFXCz2IfUjWUicdYjiWD>F?`B!uYGn~dO##6
z4ZXWt?2pd}4n=%@9Qb??eQ{(1os=ZWngmsDmaPlJ0c8oNy8_W^I}kJBmq06cQ*L({qo&|lmIj+82p89@a^|$VMS=ST`-6-E)E9Pah2wZ4m>xqc&D2*tA&Zvg7=WC0x^)}&U*bb@yDj^H=QCY8_!`P-nG0Sk
zK`x;zyBvQ9k~GI(j_rx4eq)#!?5A*dTgzlpF;tnf1tl~tvq{a
zDF6NX=m+-G>N+boVYv;1C-Q6ojksB-%iWZ>JB;O6
zU%w%7=xnhEi93C-q_|)?&%RSn@%Y8!Q>(F@bivvn+>dL+W^y0lqSAVje?Ovn`WBWC
zeUe*=<8#L?V`My$h7y9M`j%o$$w4vtfw*KP|>Vm3y)O4G*r^
ziPz0w5+@nw-82WuPi#6yBfQ1Z<>eFn+@$L{XY9xQwmOmg0_y_xz2nqvbE1Fhx~&he
zoQ2mU{E18WuDNnG(KqF42swA*Uu8zls}6a&x?{P!YvjE!Ht(9=kJlwbO_37*eIr8S
z&f`3s>ym_fcpv$({bkyYG2n+aZdcIGOTp(A!ntK6uja~ilfMMYrDpch5wJh=~*S(6JpVIS#r*LdP{jMwXzyfc7!
zho!Rr*e3AP#V;$anZiBFZ#(&H$MN$|!vAG~+=Cj)d#+#~m1G6ww!uA#9i3hBcTyql
zF5odT2Du91H?gIFdBsn5KZE|+i$_27g1kz;V>T<;(|h5WM-`#{Wxa;fWxykgP&~FL
zvESDYnt*&mI)6YQ*eSlBUjJ|$anXro@kT)Ve^q{%w-}09uF?WM)YdUx*$KShvVQf*
z?HQQoW1jPsaPLy)wD)#P@K=JYJ0A>d4#MBzzPNrE^7qP#E;e^6;djy4uq%}dzJD@_
z;hl?x_4Mk~HI)}Mk0pb_U*yTRuIW4Ul=`p}Ah*A?x~rCeFflbZ!iCua0`by4bdx!3*9ieuBZt
z9Q+P*3pm3Gwnop2!W;%sY{j3Q
z2mD6h|GPK%8`L|9&Q6|VSp)ZqKWRkh^#GsH)*s6`1@}xUp7*{l)b2#5hdE=Hh<3ug
z(nj;FHE?brXP(zS+=-M%Sc?ST_LaMXX%(^~O$w?(li?
z%yc0c-b=ZQReMlt&k6R{hAyPmD63-vaz{2S4LqjS1%I#F^xD+M9yGUof+Hce7rj4F
zSov>XCz^2Rv8-I*iAt+JJ%1tBgJQp~c6ti?mx6Wgb$`I=5nB)HX7i{w{M?Ir%@#i70AI{~`LRD8__Cg#knC2-bLC{&CaA*w
zXZFL7XPlk`-yAFm;|5;z>x%l?jj(QKKg-Q30PpdUL+7+0zae#R`PGkb4pN}G^ZEzi
zFK24mk6XZaY!eUVV}Zvs@11*OCPRa;aX1gbzPdg4P$Rej+O35{Y9N2owi7)_27cZV
zotnM^>Rrl%+FW2BZY2MB@dN7Jd=~!Udm!s*s%KN+`God8ZFZ%QH~G!ygX0s9iDj3n
zVV{_YC~UNX{oA+Lat*dicF!4D7sI~st}$X+2jlFMv~b3G$;-C_F8%^J<@M6lWnd?#
z652BzK+g(0!&(;DPs4|3UuFTlE}U-v2lAmcV_U=ktG#SXDgc*Q&dun+^ZFuz%D16x
z=+I?$A265CIfET=EP&f+1?c%9WU7-7$|vi8)nL12vwls94xq9@&Nb|3gg;JQ#Q9eD
z*GVcrq3qw#e}6yVO5LopIA6dg5?O%#LEuc>C~3!%&_+svU2$dSC#72CN3
z55`EoU~)yXeGllp#kKdnKGYvo@b&cu^fYAc#r%SeKg=Ba!@TDkN~S%U&Ka#PDow)yco0g-vUF=&M;U__D8iM#aDXc%D3jv&eTl+gHzT$8tIU
zBuTv7x2ozpE>{&DAph@KkkU_bj#4khmWt~?)cmFR3%4fm>x!;bD{#Gu%opOP-^xmn
zd`FPJAjwbIeyJsSjJ`do_E_$)$#6SH^N&vm4@h#1BzV}Oi{vX_?56&H!*UmH!V6r_
z1rT0wd?&?+7IrNreDJ|v>hCIE_ZTPbx{s*xxu7vpW@e|(M{M6w-@RQ{r@q76xrh2Y
z^gYH!MDF!FipROOQQx}>(Et8|ZYOmwIK7VgduETUsPE&nwJ5nwlGOKZT~Av{KUy7B
zzn{#hzlY4Hb%gN5&#kKoA99dLBKefV(w$ED{Vz@Y8^PC`;;+e5S45Xf+;1ZFmHlzMum9AS{WW7lG7G+{Rqq>;Mx=k3-K3Rv^n-xj^Y30w)i*Uc(
z({g0qW*^xRet7WvLDGNIm05DX?fxSFe5|*()H3E^K!-Q1FNFb}-s?-VfgMpStjRbC
zcBy1vuQ%>lFPFG?!`
z79V?)hUIj3TL)mdTALspyiaGdGjC$Mmp|ZXq7Hca$U+_i*vaRaI;#sHZvCZlj{PP0
zmlJ-Q&N;(5d%e@~w~M;r+~MxMSNR}!_G-qp(_jZR7Bof0!F~ml)NUyS`|f(;ZQ>l<
zBM8}Gc=sac>7DdG6W{k`6O%q34)V@P%L`KBJyYLBeu@SEc8t3xw+q@mu0)>3_n4(P
zA0!-u^7I;;GIsDIe@ze8je$Pp%DP*y{$9CvRPlVPZhvxOE8uyXOWJK9FZG&4W_wYy@spEB;C$p+$8l#~p;45g>-&4-GRUul{bqd+erT_7;kl|g
zuxA5X%2FZ!v*_`lYQfhLq4dl1>h7p)hEyulU0mMi{{Uw9k*?Ls2tWjL6AHCu&)f4al%x=9%HWAWx}
zg7XALpM=B=$j`ekk-usW{93zZO#2|@tJrvw#hPye->>>qr(iaWUR6Fz7SS0-nxV#4
zTGK=5>=BdO&iugh{657CpB+NYGh#KjpO2vB8Yg(Jg8a-N>%QT=LrCN3iECeVhEPg;
z$yHCKS&H=Orb#R%tUrhkGZ
zX#Gf7d#ufD9ORNZRC_}Pkk{*fA9vMuBcr9UTU+A0(MKo4!;jZ@qo1L801xcCJp}j(uG1L#I~M_=|N$U
z^EbMUd(gr&yFcCUyOHo6)5FPT-Dne!;NSP=-6;F-IX%1H9>l=ABgo)k4{B7}R1~b#
zgS@BSKYp48`6({;#PY+v$lpye=zn!xc{r6%*siEZT4ar^p(sKNp^+9X%9bc8vJ@>s
zvbB*=SwcieC8;P%v?!u}ZOWRml*2*x<5*Ib@;x*2_Q&^K*Z0r4ubFvf=9zitow?_E
z=RBh-)rzDST)kS~(1MDMW@PaE*Mfd*ZSh=D+=^-rr?u<^d!V{#OZ%4^HhRQuJ?D25
z&Z`^~GAUWkM#HQd^7F)7kkWn))xdMDsAcfV;rBC}kvLz?L!0ev)XHlkQK8t3a#BMd
z_5`$|@BSYP-ot*RUubuxoNY52`gvymj>pYNz*F`?OL8lEp{p#O1LJbvLyz`h_gLl2go1wSO?+zBi4b-rnRAaZZo^yzXbjy
z_@uNM>}|spVb(e+VAtqdvLPAlYaP))9t6Pqd*suSo@z+<9C7l}h5ggDduzvKV4u}0
z+F2U_`^xKTk0uY|?{yph2pdEA$`|og4&*yN;g=E4_n~M(g#nynOBX#V)B^EdmJK`>
z@P3_(-L#Aa^JUq*8`+ED{h2&tU&3UV56u2fh)n^x@U-Yp1(3cbd)co=AYY){m%Izw
zC86DOW;)o#)}P9lhvOx+M4l{z`M`C#+hSp?Z^Fs8z3|?4JKp2ifj*kjn*(1#IpY>h
zlDK@0L!tl81}wMmW~+dF(v+b(4ak4==KN4R-+UGFf0+vSbXL(ET>i{AIqRN5{*lcQ
zT6lkL^-;?s7O?eHv49ESL$1v;JP(g|m>1^^c7&}sMdpZI%1e}B(yF~IJ$2q(<%+&L$o;`AFU!6m{37Ju8DdYH
zFxDfD(ypvrCZ}=S(Z-VtYVNp)(a~I<`l6(SA&4foPcm5%~Xz>6q;bCQw
zh9sS}P?hj3OEYSB%b!gDk5-nEpGNeWO8>v5x|QDFv_gXVD|o)uitwt6A{oMG=eAUm
za&=7xiQZSj7L#-d2YMeyLkZ=lT4j`1{?exCZ9w^Ih?)YCAJV7sqq8YS%%^;m%b{|r
za+Eh~G54b^X8epDWc-m;`BV7=(|jDjGi+O{!R7qPo#dz
z#zSejZxpDX4s8)CXSDGTDR)g>3>gpa7PphNWZ+@*H7xt
z;J-ld!xB3pA9v_FDc>h*;0-Qs;rvFDe&=EoLAUBCA~*Gz7m-svk#z^l3GnlT<8fbf
z{nBn6zKn#`FdBAVBK~q>gwBLxyC{3;j46)aXCPBd-e1e5aK$C;OJe+z@VP7N)o1xb
z0PUU}OTcm(DIV7_1`2S8F=h#V;KA?f!o7#d|827t&m(%KJQgGRz1p&#w2z&sWDVco`809oOvT+;?&jPZnSc{J?@!Q!b!K4O%yYOOR{o44{wfsGwgzE7{`AXa
zUlYK^+WBi118U`q9>?)oO1WS0yq%#x>m}~D-rmpTd;ojtoCIwMPs}nooet}GW=vwe
z3fO(Vo?ED$3HDFfpJ7|A4ZuEmqFlj$KKxINhfg{Q+I7p)qAL-2@+7x`Zv6kg?F;k*
z*8{&^SvP-u8rUhP&H3@C0`&a&_|?U_aH0RYl)s>^XJV8^CXB1nRPV%yc2wMDwnp8v12qlpa_s}2A#D&VzX8e>
z*==520`(n^y6$%?74|o3$#K}vZ_f2YNe`iYL52K(mjf@?@vSRR13qD^y3k=C@T}Nn
zTFs|mA2i$I-Nhh~JDd1-pbmJCTd1~61K7p9R1LzbvD_Phv=I2t9Bppuoi5mg>NIA*
zlr!r@!o$m_zZUO6lVg6{yD7J$JByC0Zpdy!bKhA!{cPERO3QnK(hWOMf5xx5^D5d=
zo@URg3`zL^phB(Sw;Zt3jl^EWcJNKdGH-{#{^XT-^_Mu<-^eUH>Udf(46I@_!7d)M@ZhXna9;M!w#vt6
zThVX6xTKf7ttj!}QHzpCCvD9_+J&un%pdRb&wG7f$wUdc%W
zsDPdVO7$)Ff8iX_(T&=MU|;-g9C&=>Hhj087VW2X6L?2ST*EQYQ&n3auN~4gi)7kP
z!#G+TiYRLfz
zkA>DOSwFQESv}z#u+3@#f71(1-bb{c)X$~DRqx>c4Wb8&6{5jjn(Tb>R$?ovnj{gK
zd%P9xk=`orH`Iz2mAEB$1267XI=0SZ68wL&Za9Mji&#yVE-BADNo7BpUFzyCAvR$}6uD23>IKUhMv=a?-@7h(x*Lx_xwTA>d8Lv{w7}e_S6a9
zf4Sat>n^|rW%)8^As<`2D%2h3u~qw7S4Khq%N-i;R)b!1mRNe^!8yvxPf}9YkD!N%
z<%H*uPjJigs1DH2GP`8X4Zs*3@1TpIr_Abp{tDQ=Zf@19<@RII!o;Y%>OT~2;8a-_2}uSRZoTTlm;FoJApp-E{*H(
zIrdd86aFNE{F)bvvwHz`(`8~gfM?z(^prsQCF6w0fe_aJXg>QKU`XC{86&`y-+L$D
zg7mbqXGgG|_~l8rc5F|bt@&Kd3NYuYlNZ)6UeJ9awmapfSl`3AD%)oswx_JJ-lKte
zeg5rMOU#?wg7U=w13kA@>IAkx_@eOr7kJ&9V7v!m1_ypT~$`XZ-=>pEY#9x8;2dJX+4z6$Jjk|()y{O@AUXJ@y
zFi$xf&!47=DXTHBU;nC{>{IIAX_0ebb6#E`{^<(iJO;3y7ahZiKf2EcMUwFPRIB#A
z2CswGmx9T6+9`h4_p!byX}yQA9=qf39>w(xRgETMIiH<>BXGWxaSxtjy&Xq
z$hZ^#`dhxF6CO}w5OW#l%Q;$0?96dDP2F++x5;nfaQy4tfrR&<2`QC0ocKj>H%>1T
z?I-oM*W4rgNQ)=qB~I6wBSY5l4ZWr0JhzPWVP70y*T0_detY*x#EvMWu!-1%b2f>t
z!+QRiYEJBVL$OYuu>7Midhb}@G)=;rdAkz`KkBl+LCSgZsE_2=l-4ACN9hRl-}>=P
z;76QaqGph!k8d*~yu))#A(8vFfZiw9s_>lbANUkpNP4|IP2Z=bMZ(S2`b2)yF3JbR
z`o#!uKXBWgr1O2D{PIJbHIX}IRZr}VDcWZVUp*k+LinJ1uq{bX{6_tl2~S^1(&fXp
zkaEsXr@XU+m)fcGo;)M*e^cX0{@8lzx1z>^^5>{26GZ
zTWEMHlitkKZ^K|NQ|>gz@ATLK+OPRmv>$lh(0xO^B(=-0G^Y1H?aQa-J}995F*T1q
z*ORI=|I<|3-yS(sZlNi~8m2!!OH=uE%)Oz@HEFtQG{x4{6e}73(l)ZRzkBb{_!l}1
zGI~fe_sKk)N53QRWM`3n$g)f)sP4U+Jn!VsW@LQ4J$L6a9`7Pu^ts+vp#BuA7F{Ot
zqwe%R+}Y>pxPKwpMB=l1X#1KT&~aq+`Voo0F1D1s@0%IF-ECu3zp2drvf?!j$8PH(
z_20IqdVbQSwY9ER9zWXtYyw`Q*KDC%P
zr$lZb^GVM6NV0!P$tWWBfcsi4S({B>j*gx-}c8FS6Wq9plV<
zT>^N1fAF8SEw=B-`Sn&}y$hDHE%5g{KV^@#YJh_iBguCgUcHD&JYVMbU!H{LIg<^C
zl<_{}ck2t{KlQYSlNT;mf8*w2jN3jGlm0*jchs@GgZfxL#w^7%tvLOq=4|qvM4f9=
zjl;hlHmKq9uZQbdV!7eJ>`OR)c=aVBuhMpld>4#KNeaMnPaGwC@jhywP)IZWE^$iO
z{@nAPCgOEj
z!1&Nm7RbkSuT#bAU%$(V9(*n&a{p>JUYBmlZ|TJKlvLQB$KOqE+}}<7s#NY0=iu~X
zHm_G=y&U4EsO$k0_2w$#b?t7WuPfGbvT7#rr!8UfMGx<@9fTC8lk-?3eC?D|eo
z8{h$+6P#=XfnSY@CuDFy-pJANzq5efBEOa%0Qtd3Hm2A=VBsIPST4jTofx?04*N0x
z+Cfdco`N^OSNOZpr}O5sszDx2|LYM9@q6oHMDYEJ6DEHs#rig>1sE9v&W${UDh%R1
zoNBmJpoYR;R5oWUo1t3AZ(X!c|;yi#5Ddq*3-3$&-gB&1?rW|?YS!k;j&hrylW6Xr8N5|_8*e6TJ1n2pi!8icp>C}KJQBkZcpcp
zt;$spZ}ne94CbAy`v<1g0~T+&ef=@SCzZ^V#{SxD9v@Z5d{=vPKq?LJ&Lz{mm=`O5
znrkrwzAN-bbUj-F~=jjQmtT+V
z{)_hqrg~-xIKPjn(f5zQr@r8(w(5^{0LLcNQhTiYvO?IE)OO-mjwX+=8>3&=beKAN)|T&Lq0ut
zW)Enq{K^~8htp;@-pBJc$GQ0iw#&GM-8-=zP*v6g6`leDPl-z^zu;t^-qY#d=Gtk-xIO=?5GM@i_{jfNS@$0v}ez^R0b)Ba;y`|FE3WxWc
zsv-M?nSYfJ;qsEVL=n4UUP?FlU&W)FGV5^ur?+^o;Pm~ORb)N$lh-Z9;X*#cUW_9@
zJC0x!GtedNlkj^;__b%86|oa~hMU2a*ziA|%}!MfX
zlqFbh=gB98N6+2iLhSnr>o=2i{*c!cQtm)<1u4gD{~V&nzLLXaefQU_AnWX74w-+kNlXSTt7D<1%llnX3Zv95W_ST{4SdYs)sNb}&eqTsC$KT76b^h=ndfuWk
zypGuOPpWqizJ01*
zA?cN-)IQz$i1MWs2YE<&Hi0>$d>@@Cl0VUk^7!C5E{WIt`-$M*5o&KPEIC5*dHxtB
z@l$Qe2!4yFJW8Fp5Au)sbRu{6WD!ZfS5Nt-w>-74XPHtyzOUgiNnccvOEC2m<#p-D
zs6AesMdj|MXOQ%{>nU%G_(995%B6DS3uyV{O7xtIZvjpJagXw2t#Zo$CY+{dSV7AP
z`%Cj*XYLo&*-F#@FuZ-v5Uu~~2O4i3LGk8BiWj0Nx-tCThvD@s2J0C7b&1NsU?S}R
z5=n5F!57TFVI{L4fI~_o9Ud}4_?SbG%^)~IAz>JF1mPth2oIYekS&5ij9xI9N%#SS
z@Nh^N20g)l-$jD%(r_EY?+f#2cy2fCM`JGK-zCic2`;Q
zUcC}BkLj%XkJt%J%#$wSdEh`Ez30ZMs)Oh|kd;kPN@N9*A6St=_9bnzX}QW{bbq-b
zneHzoMd*8)_4^W$7d%b-?cfiZUUZ4RS2ss#JHNZUq5ABh=>gwpI!_4Q&mI0m@8zt0
zVL|jC`IQ-i^*l8%9fmOQxQcro7_*a(wPDN8&r4E1&t-WsAC4YA
zeTvA>CWNuOnb=wPcf3KisJ`j?8rTqw%K9jI9zIBO7@8(7fkALxNC|+G)B|KpX70UZ8Or(
z;qdBPVxKYEO=gjGPG9TcBdo7b^_xu02XpO!-)pkynRFV$A>j>#zt5{ZQR@d`
zm;EcHv7Y?1UPxnJHz(=pV_eSO{jU1BoYuJEaopY}ftqNnXPe4M_#lK+PsQ(-16XS>|lKwv7&ry+`r?FjnTNjHf>pNh{pj+JuijJ*VT?E
z=WcK7)U3zPt>Q$XFy^PP?oExs`c5)kUxxJ@6ESVX#D)1Xyh2;pOEQ{83(-cQ=3i}f-Q
zjM#t8hm5WD_J_ci=HuB?qBeh1HfP~3&<$zAlY
z9qap|TASDh1nQK@|G7UX2uosl6A`;ctY?J5=>>S){HS&y{H>AgNyr|4&gfr3E(-5ZiI#z7>%vy9!SOY#^ND?3t$j6_mrbu(
zk@YPv_XXkmU1#g6v7WC!H;{SorQjgpzn=YO#BP7VQHs=i-9x7Ymv`QIDmkZ|GAKjp
zTjH>StV8cc5=c9qJGRqxMQI^PAAU`FK&=3^`%PFWM)WBExQ6J-zeA3!A33ur|27+P
zCF!hqEmCgD<`lB-i1jZaa<^V8lJv`o)UWh5(Ge2fekXwF?IiS&=usGyO4gf6v%E+-
z<8%KIxp_xTNc!vQVRL7l~iX*fUFZ({)_?
z7(LHpx`RW~^IfRDJz|jZfVC!+fAGAc{KNOwOCslKNAs&?&~z6$${!w7(tU*R{f9(u
z&jxCr-h7Sn>BeBnn{OSa`->WP%D0C;Qhp^;P2~pUX}@zcXnq4HnlAbu)kC$A)|-8T
z;_xJz&qSG~uVC!YkIiYiA7ke}_l)XU_>0E>X6~Iovx$af7=9Z0i0Wk-M)g>+o2I)l
z{7yZMhEsBx^m{a1&+xj9el)z5;iIW$G<;?U#Wsc~uVr{E9A+c^^@+KUk;B|S@6V*e
z;34T643Ct(#+1jDpTOilz@&qbmB_
zkAOk>ztZ95{ai)N>~D|56BX!
z7bXXS9~cCu5hQ$wK_De0{I8w=G2u!EK-mif!_zAvYb&~_@n86o*LG5a$GMCIQ;qPX@VlfRD6V^h~M
z>9;6a2vgiNPEq?T-A65aNW%q+6yusGeq;Xk5_gk^hf6$(o_Uw(_eqyFF*3h)He4X<
zOyA@mgm(|9%p!7q#Z+&1ZY9y%CRU%!>nQIQyjrkcww03j#7?^J$~{KyBy0WYJ}R&+
zkH`me>HJ!EobHn>cG2^Hi=t?`i?;ic`JZ)|_8VUy3g6MNPJ&xd{{}zPe
z`p2?N&uU~#ga+Oc3JIqw_XVLut`8)>nO5A&y6Pgxn5
z$9(@IqJ`VDS-T|_=L@-3TZF^E66&3CxR_lpio+hS!`9*auL56@_8m;Uatx=-t7-kl
zsF)vf1>?bUPY+<6mA#*=)2ruue8%B28Qma^d#W1<&l1>MNXj|Gc}e;wQZ0*|5BLKeZxt$<^v2D2)
zIiFl!6GV7df!FLMSYDbhg4{Rq-lmNBm+QDHB!c4~h7FEm%$;1n6qk4ORk9MsCkqva
zaQqeFH)KCvo^5A?^)%D-mcsjr7=RBz>}@7Ia>1QPn2+gY
z^6kLoeN;41$Iro)V;hE1RWpa^dCOyq0ZtbWN`t8jo@dCEG@MU+EH4}*x8|)dj{g|0
zYLC&VLzwW&yTb0oKau&nXFGBJ3A2ibp1l8+!4wD2Q&yLZBh|uTGOj$Y(*KJoS8lq7
zzh_Pzy}0xQV0ZY1T+F+TA3x~9^$8C*k>_~I;+O~4v*mVh3GP?EO9x>}1o;PHM34To
zmBi1VxzMZYIN$5!3FKUg(SoDL5$!iyp_Lnp*+etLiM)WN_~oL>5V
z71<{&lXE5Whs}F>e*L&ZG||`T+XT`-{^oN@dgLu?XVUU1B=O5d{R!T_vV^36yc9t4
zb2rTwA%gZv6765iHC*M|~5x_(;Pz9#Xf=2AW1d(-nV^S9A?Kh%@zH$jw!
zGo$55`Bj>f_fE*6>+u>d%FD}RX?)`2r=;8+O4L55>_gM5nrXTOqU+0b4f_2-RsRK%
zPuxiLTe*a`yX7oJ6WwBxo}Ee8&!IZHo(HU@<*b&a;qBAtI{dn>ATL-_!K89Cvch8`)Lb}2j_4^
z@5mI2p{Hs3)GIV>x|3p#4@GH)_q1-HVKs*LH#59_9+M7EWJ$SAj9oX3;q72lAn~#c
zkNUyjnot^_!{h^}V?-{KNpEAqtXLXfvX7z=!^^%gauXRj1tvX;sShScQXVWm1YwaQ
zc#T2$h)2Tz_&f|=5{8de1g)8LUncyIC;yYX%f!PWJ|cgE3B%@^gkdru2!oTLq>R$x
zyu%!{d#_5&6g>`nk2I<-{?U%^KAe*=a2d||eMxE2N`Z5J+tlR6;s1jjA3CQ*9^|0=
zOA3!nNa#Ys*;`yb*n+?CHEX>NpMdj^%Tf&==XN3gDJ>$V)t%^`YKNXZ{J+r8^^Vz|
z*B$6T!H1dd>)Vla*wD^rmYwK(!e6nJh3$w{XwP%ap&j|v-uyTttsUuCRem0t-+>;k
z?=idw|G(q8%PMVBZ$)>T)QT&lI?zDa$j{ReZAjwG-tO<=?Z`)QLCdCu@8Hk2z29=Q
z8U4Jvm{%gW9eq>E^qN%LhMLFrL|-jyM>f1>N!M!Y&|-nJ8jXo;6nE+2>yDprz9LjE
za;;n&dXZYBTw&IZ3cCDi{;0Cil1J^xd;|Osa8H}8(u-Eas#Gq&^AzT3>4q7d@o=uf
zSvAm9uNiI3^xQFjQVWu~@?1G=s0H;1O?q@n68K$Og2X;EI6tqpMK^a(GupO)$z%a;
zGukKb5g^mpf(#<{FFPv1xt^oD1#TI_cgD}NYC^zom)e_tvpIEa6l-zSN{^q7Mifsu
z+U#yZq0YCBlHW8UzpBCw#rDl8dSU#*MICH3ciF6a|GBVH-J+a(n_|Jwmi8iUo=_7q
zU^QrZ|6n7dSr0Ex)@P#$N9zn)4cKVe*-53U;D@8e1XP=Vq
zyhhaDXwn>$)r7(&pK0gsY(i$&XB=`VZbY>e=Txq(tVU5+`zp?Z-|&-_8=s{OgTIT<
z3DRZmO=$GW(x-Ylji@y^%68e$M%2O+exs_Lg^n3YT~4~%fH*%Y{oM^%=z+nN1P8b`
zukF9k{yy;Q5x#Zb?fn{!DB#j+v8CWAJ<8bVoqrO@-&oN%n%RI-CtW7ZgAJ%LTiMTxpM^4bKS>|VZA8N93l5sEZbba47J2IGEI5zD
z)@=CJfG!l4e77G1`6Oeh_X&+iPVAkp{lZ3Mmcjb5XdMehr9U%T34YkG$y}Q5-O_+w
zf3Z8j1AbEZf<&fHvtl7$@eQUOZyJ!);Ub+q84c)QR?1)5hYje_f{NBn*Bg+Tq-pt|
z(+x-ekibz-l5HpSrRPNo_tm+5Xv+Ab1yye
zX9Jo(bU$qY_?2I~MLz2hJg3B}Hvxx4KtEq|&Pr1j8ur@d60gcaoaqiTt5!j|PQgdt
zE@Gk8Wjdjop+0Rb*Xkn&S?K!GgENF7e5Uu=g&Qi+&Wo*S(}$sdE82x>9a%_x(iu_t
zxsB-70%@L`QY>`AYVC%3x-2AcH~p<$1Pl3p4fx@l!9wou2lVZ;q2I5rJXgDnh31Hb
z%{}DGLRLLSUrlC!UwmVkBY(XbQOELiey@`ok+q*g5uXjTKfyQW$R`$Zm78P~pxcNR
zrWegj-_?lbn%$NR^ld~J1IJEu`LNKUZA~XDxhzz4?tn|KWg{{?fA5q{U?ZBX(606;
zqY-UMEiO3>da^@btdCm?_hc^WGMt~%h`7l1`5%!c^d#To%nHLMR8YR%O5#5BhksQm
zg7@-(`4b7}tVWa{$QE~$YeK{7iOy$Dn@}3Bf_hnEBl1%5+m!mT5e?0ltI#vni28Oa
z^Qvhyp@^=1*A9T((f-q)B#t+t%jxUh?u~8)zrzl>V_zEK{#VNh_d)Klb?zPKZA~cZ
z%{raQ_Z!jvBPN@rp}g(>$mB486FRtkn{M{TCNwR8kN@8BCZum5vMU4p18VPco;J}I
z{M~#@eW4%HgxaNE>~w^Cn*NF$N*-GZ^T)@?HfO6Q#P(1#pL((h9oz2k=II-_FY`d+
z;FQlWPhEMsCH=;rqisCV@Yn3BH#_6m;3hKUQp!?0Pop
zP(R8IQj0HX
zLSY@nKb8+Pq4`0lQ}XUNp&-jOHxJY_!T*xBPkhg6LQ6Pj&t-jULf2N$&)L}z{t_=7
zK2&SWM&+z4p7EA!RMjwRW%MdGvYDqoR4oAhh=!jfCH;c-pB42On*;ua*w0IbmDtGH
z`>#u@70mnXI}J1S*eFm~huP@ltaP~_88H8+mik|X|IzFH8Tq=r
zs|l^!c!Lwh&qfz=U8~+&u#xfc^BHE@Y-DG+(EM=)8y(6_ZF|IRLS@~P@-)YxAEv&w
z&D3Ti{d(Km^1k4QY0qn6={syBD)ziOavB?*4{`E+3jUr_cJ01eB+f=HLMAUNt=Xt4
zqnw3&*vQx9kGL%8Wo&QY<_!I{;^vrVALx50E*I_}WyAlR3Y=pN+323xDdmd~*(kCw
zp+tO$jgIcz$J=b!jNaUQRgn?fi~=+|TwB1;0+`HR%ofq&`G&B1SZ;Q#m~sr_@;RPpr^?CAdUax*ywgK%Tuud`rEs9o8i)C6t=L*{GCHHTC1L)nYV$Bs=s{W
zwR3`YeVw&@5$Id0RxhT(2lWbC^*d-cqekb7rNW@+V5UllgHQ{a5;B*g3w{ZH@yW&Y
zjI^MxyF8^Simm8)*tH{{-ZsNNe|ztG;1Q;p>PNnwZ$Vt6S2NtlTF`St@kdV=w4#v>
z`gvOun~_O^riEaAGy0{_z3S7J7PL!Mcu!|}3tB31uIRF8E7}>79L)heWu-4RkHs{j
z$vn+VPyT5}^W#sfc;nQ9Tpue=xq77q-IY+<`&GXg8Q2=!EO3H;5H~$v3;JGXw>j{E
zf2#?#Uo{3STF{2!$RZ87RC`F^2W&y#rg-&WyvaU+N^{JN;X86JyHN-Jx0(KW$*~J3z)!q(?;P1l?MU+O!M?JsooMR}f9*ox(KjZk
zPp$UvM7BBO0hQoCw*I$(O&g~ZX)lcNe#Y)VbtOkfXA^+MCIn>H44pnYz|IafkYTRqCz!KGKFFY^}%V?QKWSc~xGD
zL+wcUz|(8PdpeM|9xq>URtFj?S-Znwyb}q8`pms=*o8#Zq*}$nKV9?njCr5GccU|b
z=YAa1;h^YDnPs~>I?-l>_lFlO?Lw{T{ef44yO8Z0j$=YR$hoIX+@RQlgf_+2Nyl~~
zHhZ4FRCyitS>v{mZlvb2+a=Ts{NZI5?+}>RgB%=-#Q9j=
zNJsvTv+I#=WYHeJzqhs#gAI_}ShKe)>)NFS>Tad5=`?hI`%UX~US>
zw2zPvSsya|*^OLw-Bokob|b-)iS^37-N@g1X6Nn|-LU>2HGY`bjci0Xhk1p1(32y3
z*w6O#ARcrwZ6vP?E!GjZ-ZY~dDJhFy+6nzu@{V7ua8VBu?7P_-J?cTH=iZeqw+BCi&mR@l6oX&FmQSe}T0KbO%d_d1ZF^9t)<%n$
zpnr_b?vNh1&s^#7QXg+G4&vW;b<|F{2g&mNckqK)5Bg8d^bgy$2i3b3j0zs?K~JU3
z-^+ORpc?ZJ6ZyCFz`fhQdo-b(Q3siUUkN?Pbm~yiF!bw?!%Mpkc@8R7|M2rU=-CrG
z8urbp2i1N_J9<_h>fL2jDk%#0rq76dK2fI!?ciDST;M7P)!3xkIlhN{?>=mM{(*xk
z{^)x?f5<`m>m~N;rgG5J_?V6o8xGPHQ_U%GhkK$1GJ1<)e>*1cuFR9nL0_JpzqqK1
zg9^ffcoX3{#%oJjJ43zy`IhYp$?HL@cFc4-qzU&uTnPW$=fOcuLP^U$N^#K1k_P_l
zV;r?
z&O>q>bXxv*dpPKE9`(>ca#ruGtJtbDRanPIcjcXr8a?phg{pj2qJ;=`T
zipIMy93*aTH*oPU2N~&WEv7-?M8@Ooy
zian(bGr1^6^}xnM{k=%yrN|VWrCc;0?eKLihjDg7B;p4=-qy9X01K0!0AP^s{{j*_@Q3u@`Qw^
zy@+pwH?rBk7iHc|=)CZz7si+L)flM%hiv}+tUtZz^@pOjV|%%1)~>hR`{ACi($LMP
zzFKh6wcHP~%cu6CKbKVw&i=?n-P`-MJPf$#k!Hx@t#EIch(~g;q9+%bofV5359A`<
zpp|i!u5e%2ZC}xS@?5kdYgyZS9f*I`ajgvQ`zV_iwd3ajE_z>A5?r@~ixkh9HpWhd
zd&njR7#q&vBAfjDzg}is)V5`tzi$*5b$qIP`Sdy$`D(7IS|$kcR-K!-%;2Kb;F6an
zaNoz|vkmi4Z091^=HrhgK(AuM=Hs!sT$GmOGxY@IYvHQ<&UnQ|Iics_=7V0OdwwLv
z%z=Zp#N7G$c^wyBUmz;?<`noXj}?47KE_4u8?Q~8qyqP_Imq*0)agS~El)qqfc{Wh
zd6FHrlZ#G#UeYFO$wgm(t*CO`%tc1Gr>~A!!bRbV^17pNpPKZ2{#lo;W@G#g80@!f4JFYAMIPwhw9}l2dnk_
z&}z@NdAf=m)PCUI{Xp8h5dY2x#lJ=($8@eH&N!I<+mp+9^1o3R?6oG
zZYOZj+Z!SgNp@UxuGlK97vlFMZqaBu1NF4UcLv8oe>&*4+PJ{;8xVIDgMKJ^w6HiC
z^s&-7uj<~+Md{jNw&s3N{?F%&AHA;aYW)HQE-J{b3yODxd8+Jp_aE%(}Ec(HyY
zcf1#^i@wqRs1xST^_c@_e7j)2jpUDqb-mGjG(EkAXTppqT

diff --git a/previews/PR1624/tutorials/out/solution_000040.vtu b/previews/PR1624/tutorials/out/solution_000040.vtu
index 65d773730788cdb4c32034740a5b6735a06e2610..22e49cf3fc352cee5b96e42dfaf1c17f0c405060 100644
GIT binary patch
delta 12169
zcmV;4FLuy^l>vv90kEb60Wy;f1T2%&133XQlMMu90X4I71lSgTa|{3g00000c$^hi
zcRZEfA3sSnL}bR-Y)M))h|VeDqDUGP%1YUau7rjw6jGVFH>+!9WThkxl8lBzA)~0I
zNGVj#^ZWe%ITSoiXXc2do7Ps?8k3OYC1}Blm3@Jt~bTPyN9CB
zcu;)5Yi#8bH;Na=9A-@&rub>|?~WiCBRmIV%3LY7Q>`g~%YjjEB2PEPk->X>8{*Vh
z4Cc0Rc(s}gRufZO-yz4~p{Z!N6p7-2;l11jkR^h5BqV|7~?vMdZ&T!N$Bfh)A8y)3cgjuy&~b+=>qj
zj$Cp>z4Hx!gDv7^%Zpw!n4ju0qSwOUB8}Hp&zl(hBtB-A_(KL)o89iv&u8#wgUHtr
z=NUZ9v*lXZUIssl$Bl8v5t&(kc}i(v#r4sMh*lKqKaE19?dvPm@Joo4OGQ=t!2CbI9nT$85oxly!HDq@
zIj~=J!%#LNOA@uJv*9^J>i!kPMdZYi3*V(LAYzF}q{1E}QZ-~ZyQ&J2hdOCr3vMB@
zT+~v3*7i0ceta=awQ5Anq~?0))grQ<_r3CcJt8lXf<($25fS<^oLt_7$o#)K0&`$g
z%-ks9)qqI1z`4XRSogz~9b5_PdX*y|8NmFz4(-{)FmI2GhJL{Bp)nis2jKVWH}lf0
z;Mqx8dpNWXkyk{ba{qlqRx9#i|ATYvHC*w3!lMik|N0?r5j-0<1U9F@x{X}db0Ou3
zEI4Mm^TPvJx6Cm3Nev?3bOsIzK1AeP*tqGp3Pd)4X>2`_gGjpnV_nNQL{gJwx#b~<
zjM;@s+It|Prh7}%-U*RqBciKJ+yFq9PG#ZQ7D$9E-?*
zj_i%4MG?TKo07#reu(&%HuyyBL4+=3H+E`4onn$+RT7B232b&4`2l>kIl_6_$>6Cs
zb`fJk41VCk+QXMcM4{r|yW`Ip?86iH7KQbSQj^9?iirHWt7EPp1^1k<8yNz8ot3h<
zJ*?Z_!pxXR1N<(Cv9qBS+0{}99fy>J3Q+M%RW!cJqb#@Qf=39!^i!J{ibE#Mz+c
zQ84IX`P7(V54^YS6*xTb+{S30$p7!@Y5`651ANw>cXLlW4?Y|esHELXaia1qg(bEO
zPPyoBuzncu7W?zhVKX9szo#irdvO+#(ivCT@g>0D)#i6jf-b+FZWiZ00bUB^9h?XC
zh!{x)sRJKck3P}62JaWFQREyvMRB_N%1O_5cpkDbje`4htre}!Mxe{T@~fRK5m|Jm
zKUgdtbRaaRKM{0USF+h60?xf_A3xwK_={^snQIs9Kk7cN>I=Mon~b$8(gmE#3jG73
z{lMQh-3uydfV?n1QKxFpV5zY2{IHpbT&c|dC43Ngk@J+3ejSm%iKQLFPZ60C?Q3p;
z`~1L4GU>UBh(%3qs0f_HDEI7al^%*a*t4@^fsZnKmP}ZAQOwe=7%+PXxv9|c%Nx$`
z5@R&CPaKgghAOFl_q_rCfQAtDQt(k3ZzdYfZEO`+U<&8HY-DbH+8+^_;S}w+pkH6-
zmclRw@{$SU6`TP*sPV02p^mH12jik1QQX-(wb|d3!PkX*Q6%_zvHsqjpUxnHXA8MY
z0xrGUHHxbM_uI}Ledrqa>Szv6j)llt|73FF9)m?=A`^OlL9cI=>bkT}Q_Qm*HxTfl
zIC{obA2-N5k5_A?K3!&T>UJl?iyI(sqHWh2CIUWH;cv|W7gqSij3dCCPqC8|hl3!0
zh_^+fBqD)_yAN5LgKw~oYJCKJf`29s^!ZY}lJ~78Jcwe+jHCL6;43o&%?4*cKK3bG
zYuajyNEc~;{?0A}KU|c>l>uCy>~>Kb1YdoV_h8dj$hEVZ4cE=BVz6l<4b}WcvD5sG
zMN{At-HknMYk)ucVGB{u4T{TE&v5GOAXlrkB~OC>MQS?A3_+hLVMy>C+@rORJZCNN
z`-#b;$zI?|z>GAhcEEexxm9+{;M^tL2|Yge8f!a$SLFu#fJ~
z&mu0Mi|Qgrk8R*TCw?lFeFy)EdcywZ4}8+U*8ZITjKQ+TGgrF6{z)HA9*WpNj>opH
zsfTl9L7jF7ZC4)+1FEAJ3HE
z>aGTT$uu)L^9%;R-#qAE54q80^|$yE_^I1}v%H3nfZwMW|I)-QkS`{Y@}-bR1y`F7
zRRHfZJUqP!j1UlJpQ(4(LA9Uv=+cWDZ=$tR_`eh;XcU;ua2K6^8>)S4y4fuWEbbb0&ieq-!
z98rNjv#~v{E&}>iTR9gu=~v*7XO`-JC;$(xUr(p@aNjzyPhI{&|6{gwYEZrz`ZA3l
zVk)5b!JXeX%Ykmf=AWH^M;-dRdhdi_9`ubu5rsy;C%>lDY7Y3VxrdBvjT6NeA1rS^
zb%ElN<-a>rVP0M)ROcArw1{1rbo>P9(l%Yj5&Dr4pNfMzuush4=`mMepF^pCFV~d>
zATr!C%SQp$C+={xcL`(gZ))AB`vH8VP+h4Z7;-sfTVXfkNT#er_WS~hjohUtT0lo-
z-<>xp1K#akC%20GAhPN3CZT`e>vp@p1UNxo_3L5Al1JcYIj8nJ<@zF`ZM%M>FZjUe
zkbb>NCkC_AD|`#VuV&0Ej5h>-{|ex6I*kJ<&X5z1TMzn4QkRfUhJIpC>T?SL=%e-Y
zJ(op-PtIPwpwk}e@w%^hRO{oi0pG#YYu`ulMz2Zlm_}K4)5ose+2zYOOGl--*CxwRstKo
zV|4JpUVY#}*Y)`hKj3?ZJRaQ}xQD@sr}E32!Iw1|?#@Hd4`0a&v@-_Ztncyrwi%i%*X%^#(f#;`nu;<_MYD*#XU*_>Ia{?6(?{ffn443kSV?vkn=Myn8ZtDHyMMq$v@8lfJD!y$I|t
zh-+O<``Pfqi3k3TZeG*jfPc4LH%%VkdyIHW%m%PFYHDyE?5ZjxI#+=YgakwCpm$BY
z{ni=WbWN)?e;)LWd{9Z}T>fLoJq{c_%WWqqymmN)Cz^bUb={>vV#154BnmxS{Gy$f7AVI^l?5C23ZlEeDpspm#o+b
zutwHKPYinZaHX&0KiuZirxyJGpVrC4F_7;ZF<@taI;Z*)7J+IKp95yF?z!D;lOAM$
zrG$%2u&b&&+!{2kaI>a!TMC)>cfd~SLzx=g_Xqy^nP$i($A1SrL;molW{VU<-m=#I
zY${kFfBP`+2x#sfv@V=oGX{LD{D
z=(%H@GJipSvX(U=f;_C=`HZ;}a{AOR;R6(ZJ!6A5oNHWUW>g9CMxBUM`p+@;8`5Rq
zRLYQk3nyH
z`(S7jgtd|v_#sVYxDDZW98zU1;UbbM#
z9@woC_gB<_{~H1Dul56Wkj+ph%gJERHok^ds;$`zpJ
zf5M^j6feKIuQhAH6Mn7&^j)vTu)_f~Fmi6xf&bmv0l{hTzjC#?rXTe61s3%dkcDlF
zOB}&Y%ebYN!B+?QMyW2USnuADpTK+le0E@d*zPn-`tRlmX^PY5;%!b^pj&L-q64`9
zQc0!72IxNu2P%J=K|duFnlJ}Ft8lS0f8|5V6>pB)sFRb4rJHtw6Nx1?6qn3|b9F+n
z>)Ix8q!#;}JrMlp9qi99@Z+F!3tMq(jKfaw=q&?^hlQty-Z#kd`!dFNfJGe{dAG6d
zppo{^J5FHJU%OL(UzHYNz~`Q5X`Nfc|UoCLwXiH*E(T^r>Ea6f+d)zMb__
zo>6{^mb;pP9N|K>jbL++EsN^#tFgEBLGXfE&AbBWIdxpe?t#N0ik(!4YfY1TXdnBp
z^_Go7e{yuB^9wkypQG;&D7y0Le|_4I^>>4;)y;O)_Z><{xhSvinx>?^rhRrsyrFv?-g=N-1~k!6
z-B0)E{?)g$lJ0H3uYn-+f@~uq>Y(P*n;Of(*w9bKbdDZ_nTJ_mVQ-8w{Wrp|gi9Vg
zDsjP$&ZDz;yDt6jgx<$^f2y1NlYA-Dus7TF^hFMM#cQKp5B2waMKPKuW|yz<5xsrO
zbSUo{4>UcdxcR-gq1{UT%~AXpeg7S{a83g}d5HTG)%%9i=KE>?20~NsXCYtYsSPFn
z;-0qxpMYM%)*O8x*NJr9N$~NA%|?okCug53^^^HYO$8JWL;lXEe^iHekNvfh@-@ET
z+nOzOKBae#^C|B-E225U@0C{>)PH4X!&lL|hTN@bl?9I
zV=Y(*@*C~3rG0n_f3l>gzlYh6By&T)b^3$_^{2yk44o+7YX9E!wg$92tW-|>
zpf`$UA5;c)xFT~hXfE?I#3}zddot&YDbJdXe$m`6U1Za`gzhiBZj|z`Q0?OTePCv~
zy9C8gOlVOUf6X5sg=611D8ID+h|_$wH*UVbq-noy0LH9(Jco0ub$LZJ?kk`F0IPl+lqD?h37=D7&2hUPo>bK;|CR1M>Np9+*
zIX=%ZIF<5QD(WcJ&y!W#qUfF0e%FPq4*O^yvAav}e+~66xnIDZg)=u49p8Cu_{{Cz&n%9A{aSW{MuIwE!
zp+0{me>cwxdeyx@vgzFt5mKeG7j$NK;H7)maM#vR1K+oY+o$=^dYg$L%Hu=#FTAFD
zGqvBCL;bV#)Nf`O{2llyc*zzN>}s%4r}>l5(M9jhkCO`$tHA-Q$(j*RQsTMhdvKX&
z+C6%w-u69yO%8Ez9~RZWgmt#oFY+lLjonhUf6^gGWQyIQyzL&A2|G)5SY#qq2wKTV
ziv0shTe36cK-t~ioOJ#;!xQ85K5%y(T1D?d!)ZZ#WB3Tgz3@o76ZhJOE
zc{=71rScXO%`$P3hMlQ|`MD<)XIU!|dN2O0?fHP&XmYG=Je=+bC?eAp%7X_;0=qB?v@;CA6=|*nKf77hPkKyk_
zLRjlCnDCa7$)x!yknc$KQk@kv;S2p94js)@#CNMja{+U##!|_VY-dX>5yktIg`G+SA~V
zD-Yzeu)d`{D1!23X~`E!U-009kknz={~r#+@8bag00000004ju00000004juldnAn
zlgu&>f7aO-#}Psc6*|IvEW6bRGGh2BVj4_9;J!4g1_seH$
zZ|4|ez7a|^?3o{BI^{lA-~kt@d3G7jjWRR*f25Z(!EtlOcr$okWNeul`p+=>zC8g`
z-bh^>f}cST*TXyDr*zDL@}B^oX0+IMfR*DF
ztye)$#fqd!%%6>W@QVTFkEGgV!cWeOtxMC)DAPG(fp#g^81tuLr%2K4C^La+!g(I+
ze=8ia;a-dJ{?+VuPEe>W^yX4<+>JZT6Rf}F-P8=$&Mv$?25Kt#=-vf4z2OPn04|$;
zCi?@-_@cox2<{0e;L`{Hb>-iD8U7<4Cg`id4tHdpa0QQOR>Z1679i5uQvHpk#c`i|i2#Drq-f4zm!@~zm9n7u|z5$v3y84Gt%;#Y`^J^2^4
zu-*q=x{w(CJ9te+FIxfB{xjN74}4<2@gl9WrH`Sv5_Hp%zgvy(+ozwT#|M^68Z>pm
zzeip6aXObmWk3HCjB|XcSVG?~9aYk{742E&JB=N{<%Wl9#L>QQ^XvIt;Nl;>f5{X_
zXO8nf=h5%VS-grD_H@%P=ilMC-af=45Oz_{(diJ_3lk>zuftwC-NS+TU>S~g4i1KK_
zaor!Zu6={`Bg~7Qu?dU;UTm`r<{vnANq-&X
z@AJGCdm7&mFhYLi)fH=mL2KDTm2DJ{n+mLE^xF-5;yVReXoYh2g6w;i=6vY4GX3!9
zd)Pu{eWCpzGlrv`_FtXLe~P62nbn)xHh}R){yR^izjSCUYa{Fvm&>o{z&@y}+bRq8
z6$cNT1{-^Id7p!k6LL+IA1nGI^=Z8HL}{-J7+5b@pbPG>JMd>9#vQ2lnmr5hJ?!55
z7ICso7Y{gy{sJz^*A_VEr~$5xp6FMy63}UbeT+M%w+re+A<^&uq!HqIeVq
z@5+Ks7!QnSHNhUOOA^?Lc_)1`nk8U&MW`B!VE%=XhY4n<$
z(Q>NC37#^ltJf-mkxS99c|Oxa9~9=78K*kyvWkm23`WTv_DVy0k5O}B0obOhc8=;j
zBjR4{7ufd)D~dOPe>b^%pueePx`*oC)BCd;ov*K4{~bD?&tuIk^0e;+
z&kc@XbBzBfI%mxZExnuQzxBb0O}eqiO8g$xlW>E0AoZ`baS7BQJy5;0Wh=Nu;8+H=
zi^%kSdxv>m^E3K)KvC9$sZOd}hu_9&{ie>TZxojyVT)Tof6y+~`CUQ_c0~6(IjXyS
z4!LZUu=`xOW%88>m!9ej9c9nY5E4N
zbP9@~{Ule%`;-3NDDT`<1#0z0C{kRnxfnD|fayuee|)2$?7kE>%~!mpY((Fue^oiNtf=A^cmHo|1H}lp2i7=7YI{a*e@J*(sxMDJ0;P23%N2+5GPc3d(-z`^c52l
zC%u_rf4$E~T*B&{SX8&y%;kKQFixSwM27B{$q%2nNFOsAbYdt^18!;keUI|BpON?f
zb?nEcrl>BG7Ul6^+#(Ni2AkIDP>orH_U_bm9hCnida+AgzybGkCJPKonS4+2+QrDx
zSwZ=eRm32@PB<}@NV;dS+NN{>TyR1%S%~iCe=a2<%CEb%!=al=kGkFLj?les=;1;7
zHf9|9jOtNLYAO2nA9591PWNY;
zSq|mPT$g6P|LC2`7&jZYR&(fH3x*gt7~ryJ@LwC{6*+_B~04H=zpF?+1{ke3=2bKedYIKKxCUKUz=kt#%(n!EL0E0-}3J
zcar^vyhJFk_#6c#!OcN2tLPkEr9(s4e}l1uX@&HT=s99xKzgA&9iKz_a%Nyq{S`QQ
z)ht()^rP1xNV>BB$%9NfCtsat
z3-I;UEs8hrebOCR9HR3C*3e;IKk=S5YnH)vYCq@*3e+{#&EkM{qHN?>*`99##1i+JET6
zp>EnY=Oh2?vh)sc7!4rZdm7`^=>X=*)Ckf0$|z|K552FtzE(Y@`>r;pd5Uzn@45?%
z;uC+SM%M`x7(QJ^w$Rk>+oV@fe@umkmAHRWxFuSdAgkXlj_M$C&8s5H-YcwE!GgDGUFDI$=V0b=?Bz?O%W?BFbPhkhymvf;eeHR+
zYx@oOZ;8{+q&l40Q7{`!?BIgkH0uJoH&ve>6yWEM_b6
zp6XdRtUPuv>7ri$1A0eQSu%AfuO`p1(n)8Pw!aYapgb+K4sV6uD_sK%d}04>y-14Q
z{k>tw=II{!wBw-2HK@5}JNFd0gtg-ry-Vctk4ZcRzqbA}M*H>9
z%!!i3ycI^(_Zl#+bYD+Ke;VB{mS((vkuF-U38(xt(Q~mOeKve~Fvkp>8}NEV`TW;fbL%jOL9bS_@MXFx;{Frp~>V5x@
z;NDcrRh3w$l^vR7jl8Qfs8m}CZuTr+Z%%oW{iulY+-rTxMausPSU3H2s*ex_s?+~bR@&pssrWAUHn)%FOEnr>qA1&&5Tzt(wKdwtC?TbkvPP(oGBlH|_nh~<=bR{8
z{HWja*ZF+j&w0;%Klgn--|N!vpSUnHfI@?;Wa6zP3c)1-MO6tD)J7fHWC4X09wn?O
zf`5WwtRT8IjDqW&87{{|D743{lvfX;kYw&?SoQ~n#hZW6h}uUXEBw8|7at14debg>
z`%-9CzjgOAD9miJ4RO+>Q2uv?-<=s0GB&o2s7|F&6+4!nI3a|RMCfDnQ6V&QOcGvw
z6(WKzs~zqaV#}HqSNqRG^jlqi{E`zQRDa@&M5$N^r&HC7v_(Qp8(+g)E)(J7(Cx!#
zx@hPcJ(+0Bp)p|msQ&sP8b$R9Q=V+4@j6rQYmhySC9w&YrWw*$SX|}#bRLa)$>jo5
zDHr!V>)w3dSLm@%1qjRJ+#|AFvZcBI52_%@zhf@&)ZTVi{yre}C%L
z*vOzV`Ez5A83S9D*uM_zGx)dko2l6n2Ij>uO4493NB!zAYo{{kO_i=a(hiL->ZjK9wy1;TNFHa5^uh#WFJjP+g*A5wtn;bS;sf5^6
zb4WTd@?5=@LqgCrzPO&lG1b3X`!YFH-`IWNf(3{3%hG~FgLtHlziwJ?!+(Pj%t%Pp
z=HcdceU@km4{MWA$M3d0{>WLlcAhtnMW%t#7Dst(4d~zfIFY=6K4gi`86I*=>qg8^
z^Vs(>T)r@YM@vqH^i)5d?B60tKZS>G=LMgtR30(^>35z?;!%}0{pu9*zU00rU$=8S
z(srNsT78O#<3jnr(}HTlnHx%Xt{dEs62<
z;_(j4Zp!A7(5SR4uI82wMcqRTrnuQHd#-X+`LY|!LjGU+R9e}BMYt+DH_4oeOK
z>P~9^F7i9?sn}$~A+9oRZCVT2=dOikBf;X$i>4FO7sa^Rm&*hWGVs-2-w<|$!K_;y
z5;tubm>s$`;_c7i?jOI;_;Z#RuiICiwD(PLgm+T30dw0{pKf7ov-LWXnbi~bulzQ{i)yG-td^`#Zu2I;Y#Di
z_N+M9AsTK*`R>Jj6#i|RZ6Wi6f>LnYu)8yj{^+Nwj#e~cmEV>74^h~e$D@HbX+(RyzC0XgURR4}V>F5`IkebJ?#MNWn^J
z+CLYhD9lNuC+>d|;%?R>^M~aBRVGXKJtp^Xv2tru$O;-i1c}no>LMI};43w@h=FU(
z@|;{tG4A@inY?#oVJz80?J?&tnYGEwf0V=aWs_T{%Td_1KH;P~`R>e@T{joXP*~KJ
z>zqe;_QAHs{C~wF8a;1{>oR0Tu=NN$o}$FyLyN>>ks2y*wV=@=Bk>g0e^FC)bC^aT%<4Jl|J*Z+CoC54XXF44D$
z#Uo?tkI%CA8D|o-dDw4T_`7h^~hM
z{a#v<`Nc%(ZZ{;{tJb^wf{b$=y!Xxj({r!iljq1htf@?gyKAO_lXV^u}(#SpIgNwa#Y7P>0acJ^iU8(3m=Dlp%U(MwEQvZa$FxQ~K8pe;O5#AmS3kdg`
zE5gqPsb}3Y8RVVrkg@!Qg{ki+yK6@{n9BO-6@P5zap9r#wRLBCxYQY+mR!I?LGc=6
zJ(t70`cJ{_mnb}`$y6L7zTq%GvM`0{sJ`juO50irWm^P_9|&j7|95A@s!9fN4LT9R
zAQs&RlomwQaX6akv)xO;L%%uqXm9}ee!=WtYtuL=Ur$Vr&|xvSGS=Of%;AdFiP<2!
z{eP+}ZFG59h{}nEGv$R8HjMZb^lFH(uK4Q?>N5ka1FChcWi0ZpJ?+^#lgGNyJ!?A&
zx4fELKZ!MY9B{ZDb+VmBqhwjdbwx2wc>jIBa}AB)@d&Zpd*q1*l
z#_BEG->n`Oq4eG+XU$)U&vmM;v?Oz{+&%oqb>g#IPc|@ZofN)F1ti3#i}22Q^4!=R
zG4vkPEZTXT!=6|N>wPgi6fcR*Qe1e1g-$h=nP8!(?V`d>GMN9YP`jD%L8rh~r+-3{
z#^+NX_4Fo%cv>Nrl~$!N(idj2)Q*Oop=;VbQwB+g+p|)4uy}f)uFCB@hxZ27MHytD
zilX1<8WOG^Rt}f6CB7FaQD-kXDuVt;pCBz|8rkQ@ZB$4;QH*QfG}1>nVi`Vg=NN?n
zwOu1e)I{)@srJup;&ZPAr8B|_Uw`E)JP&!3J*szmrfeqq)(APT`JCwF%6GdwDKhUD
z-xm)rCVjix#Uxw!Q;@HpT#!zlbnf1oYb1{-8JQhwBXweh8~L5&UW`@Kwqc}Rb&ldw
zG06=(!dr4K6aG}F#h4t_W^hXB{+2XSe}$%=g7bP73fk6<5~Pm~i~5(AgnvK0O2O{8
zCmv3$?Fu(x3h9u?j@BDp|>!RZM;Y7#Wc
zLTskgtfrvZ`pc^Zp%5iE+uyWLqi}3Re0Oygsb9T9lzoXttBb%qLf)FKI&Yz_uhyL(^wv3M7{`g!9BgLQ%bZ4l`$mg*WbY@nUw(J#7Js6=L*GYf6@?#=(6C-f
z}S~bmHR5v|JLl&g`p%jH`N%tUq^KPX9%CCq#{D=sC%$&KZBO{
z=1Uu&vRL$Ugnvh?B9HsYW4|?$d)n?i81r^EkAVYu9>;1}+{^#`$m|;f@z{`0gBGb{
z|IjbKh(dsz313S5a9+(*Cp85MqgF?rw?a=jp?RrW2}OJh#a
z@($uBGJEarFg-$S$U48??FfbOm1ov;iD^vB&>u@D=gBzZx2}R?aW&$;8AJBd>)z{J
zMtJ#Evexh;(OX6rmA&a-F$O06mDTG-h4UaI~8A2

delta 12138
zcmV-wFO|@Tl>vg40kEb60Wgyd1T2%&133XPlMMu90W`C61lSgTaSQ+e00000c$^iN
zc|28J7sq$Tk|-g{W2{FiNi-1cEkl=z&>SK}C=wY$sw*T?4@&OMkZa12DWy_|B$t$m
zLZJ*vMNyRO^RD-={W
zg*BlRoky|#kSA+@eJ;gXol+dB9C+5%?LU}BF_$ePJO=+?EO2Yn`E-gK!W}gu(kSLz
z{V9=xu}%Kbq*ffoy$Y-t@d%2|MM}~8lN3L2MH`UsIvK$xyGSYv`3
zi^04$Hn&cb!RnG~>pJ8a>}M!L+r=6D)3MTKM3BLD+hX>VPEq`lqu>7MJH^vUN50MO
zhyC`}#m{?9@xA`vRrQZ3UMMti>K@#0?~T7EE~7xV*aC9uY2=xg)zBks-}bC$uaO@tA6RGj5KEQCmPN
z-HM2$ZM)+Kct2vdS?#zjBDU8TOJ1-+M7%gLIc^JoB2V@8N8C0b^4E&n$WcZ_t@uw}
zY#t)>CK8obTt%ee0=u{&7UowJ8$P{&NY|H_s*#rvxhHd>$_L&ve%o6PrXo_i_a-CB
zLuAK(iS@(Th%CFJQVclVSvHSNCY1B*mQg9oQ
zjS|*>a&~tR3FAp>s#PIkAv4Fzpbimj?r7zQ21Ht}hKQH*;od*KC6_lLB0ZHOFdN1t
zH#SHeZA7G7;M|pQm^bDY5LyZI29%>~jN$#K4qdTt@ZK924gY}OBjeWR_rvd{Z|0`i
z!n2*S?zgk`h`c3ImHQteLKM02|G+u+nXY_)?p20JNW(Cv2%gtA1~;d`Jj`)B7gmml
z;&BU`kB{I!%S=O`)FLvd*Y7A)jYwd`gvHkiMAm%fzYfYlB+dV^zI6g31<7)p@-Re@
z!&zwuFGQB=-_~?+Mnr2wVwJfEoYz@9zJ(3@2aGRwI*JH8^~I$N(TMPyHf3#&M?^P&
zdqZha6e3ea(#0Wuh`5zD`b6zQL}*C>zf%YHmy|wwUkZ_);AW?h9}NC??_u_fP6q#e
zV;?m>%-|wd)-IkLBI*?n-koS+urGI>w*<^rkomn?NfG{kPtQ_829X~(^`pZ;ugtZ7
z+apRDocrKVd5{N#*`HmK-YsYFKvQpje2(TRvw3DAw36W$z8V7+Y{7i_N8YXm4&h3wW@9Tr71^
z!j)pd;|demfiVA!la?89DBnNy@y|aL8+)2r2OnUt$-ORvr_bS>wnF?g6-4yLbAxy8
z2K|)eY6n1Er4}XRmlPs0e_D_?1?yT?n)-DdqWH{2o?tZWf5<`OdoOS-FDz=!t#OL|
zqZfWH+R5Of`R}&d)H67Ki<}F8QUYB{OQk#SL1bmS#=nM;KNU-n)}{i;Grg@^2)eg_
z%~^f|cy()Exat(}>D;HvA-QOZ2Mw29e%nOxpOf6@&cGv6C8d&pGzKr;dbRn+97LR>
z7Y?+6uUE3NRt}s(Bx&Z4f@JWiXRONFPVmXNQ87CKbe*a9_xo#zo6pvN_rE0%Q+#YV
z^HSDDidzKCll&m=WxUavr`s6ZMAJM5ARg*jY#qH7i1=q5Nhk#!3a9Iy4Ccaq#8NvB~W+=~ieg=^&;cEejC7@qz%eyDRmxJNW^EgiskrT*soD27n
zG?NL@06p-1pBP++^=H;9vK>!RoS~uh`$#**Lc@D4Eyk1sY>d9LvP?lU}Km@DI)a?yqQ=yx2|nMfd!m9+{|)wxIggtTZ-;m@NbYy
zYheUK@fRkTTX33Uw3ufr2lvQq8A`ZNLvef0w2}W2s3)R5C>nTPZn)cK@H8SOVj`Z>
z5SMqlwTi1C?!_)0z34h{_M1zO{t7Y>T{@aTs(n7XN$FJp}SX
zysh}sh@5hNe`t>-tk3e*8G!isj$P^RJqGod`?WPPgkq%(U&BJ+id68HAz`SG?-i~$
znb;xHL7GQ@1BwtiwuR`V@}e8
z2J@QVJ)n~y;WU|ci1&tbtL&G-xf?l?20Y-}rfnR5mD4aTmAD#D{2?yQw>%19pN(Um
z#9hG`wM7TLwgP{G#uUm%fj==%0>1i#o(!+IxA2}ZSbp@vm1m
zz&W^iMNTQep9?ER?&pF(!;bHf&F`al!Jxp)Sy!O%@jN-j0Uf!<1Uh5j$F2nFeVx!J
z2~vN5B=HL3=*Lgvt%d)!*KrFA7`&!Vdz}T0YU0DkpmgrGc0l%sEzRvpM0zNO+RAJy=O1(<258sD#+*oAd
zsm9=EGq%WEW-z$TXvnhx>c&Ugsp3n(Q_p9Ad5r@QzrHyC(kq(~ku{HAR0{Pd?^^Sr
z3ebI~*O8+H>Vn`1OXv@9PeyWCKo;;jdH0SFZot_|qvKlNE5RpyUFKWV8EmjWa$E2%
z2B&r0`}hmu_EuW)z(&YZ8m}~30XS9h`(d)+0qA#9xScjo&z4;F3eb%N&Tn8>aO=v7ekb7J{m}D$IiLs4
zrR%E^_(bcLva(A)=-gSZdzLTs1-wOWpBDp1xTvuW?*ByD&~BL+gSU-t%oMyvaqN!0
zhgG1@eA}K-9|e7@yS%H1>`RLOIlWYWPXTn0nVC)9xSE))}53t!O?K
zLosK?uMSmszi7c(z2gukwSdyACxQ@JVVAz(AoL?cJ{697u+K$z!MOimAD`5J7i&ub
zp?_~w(p1m4l?YJCR%*aCs%o4^Nc_N0i$K}0^}3;a>|33*XDmgKM$`V?!P
zZ&D)o!~BeQDd^0hfghd*<3P8J@e`=$PCBe5Vdmf$i@8FgaGwri**$rG5I278$k}+v
z&+ND%@kr<^1qB-`*QX$2?~2eju-#@Q1;BTV
zIsV&Y2s-G@obU7lzIWC?{=PeS7lS#c^2?in%j+19%^~QA6SIQtHv>2CcKdzRhB|2T
z-f)d;1n^n(bEY5oR5>yd+kOzfQ`+}046VjBjH$P!4d*a}xy-6322>tU=Gnb)n6WYVk0=d}
zVTM=n2ba?eh8cg;pZIqauw#6@;6^>$FoTUH<1;bOV$3alp_cQT#RxO!OuIA3VjQ;)
znC+frF>dG|?%jg0iZ2{&l6&LcT!}`tAuFGU#uik8=a2vS1
zKGb0cs1mSs%@Q!e(%ID(^QM^#FYb87xbGV;@Pn*^D=izL
zZ{GIWnHTnr?xkvR!2Ut|d&W;Nep4bbavRtbJ~O-k^QzuTcC7~=3JV2(fZjFsL4zx3
zl&bOOI(UEj{F4g0=Z1eq4qXAqqj}6^VDHA2*-Q?sJO9@muFW9Rk1MSn`^B)==39gJ
zOmh>b!2lD-vS_S(5bQB=0Nf`KZ#V_Lmfh1jUC5_=uU}!q`s6I}SH$oiWgK6?|EWYC
zDF^+;-kw$hIqPeL12JE!h^e#&O#6IhPabGekWqg^@2@t<`Ah_y3F{J|{R|FfTnYm}
z%Gl_LLw_Vx;TzcxvFYt?#{NMq(!bNJ%CRbtWJ)zD93ox2d4Ntu;taoe^)@y
z8|j?h4>@l$Yf2Px_}20z(-LyZjJ?QFFj~h@zZLhI6rCGWfNZH1mPGsf$$ocT890-*
zxq|!`Y_uzu{JX!JSNH_TKOnN97rftD7f648ybz>grVTqwPNd%c1ifOOlEYc(-5+=b
zHPXBD`a6$+KA#s)EQek-{7VF#`})Wd^-<`1ax?yVOL@USy>Bt4`ue
zj8}=3yP1NVp~5xwU|XLpi}G;X(A(Mzj5e)aPyqeBrrSg|_%l$xi}Fz3B)*T%aVT}5
zY>eV`Y_#hu$n%3^@E5prZE=4NooCmP
zb1U|2HKjUmRr(T>-XZR9zMINaH&jBtmy-P&hIS;_ZZa9sM)_A#IYEE<`6RCGC)LsY
zeX6arZZ4F~w-VzFweBB1bnlH#c2xHr-;8llTvwVTCclT?vn%W)z3a$sFLr5APdCYv
z-qG%xZ&wAq>q1|BA?QWeMnzRY4XJx-tHH}by(M&ye*L+p8DQZ+q$2Gbeyo&B77UZT
z=1%w7c3_V-?H8)^`8t2)P5!h%;wpa@{oc!kT@vdeISn4FvPveszi{{lRKaTUPh&9I`xbR$x_B2>BYNJQ~@!=d)rg}0Q+gL#U&|BP9OnHC!@a)pH6tAfT-#6`|
z`zf?LEu=c%^)`Z&@~h%DgX-_<-=XViU0v?(`hTbo%CaIz?zHY%N%e9-HF5#HU)7T6
zF^ZetlOp=mXX^*-bT1+=EBg9`s4i6V6by<&FF9>nPjT1AFn6Q-kH6Y=fZqF0aTfO$
zkk4ShEuF(fh_!!`>ib!T(Rd!nc^A%^Q+@j5;ZA3Yx2jd!8>-3QCltylo)qr?&Y?JB
zglBxpU&|Ee|@wirz#iV#|3qDp({V1lYN@^+9Apu8u
z@*h*Fx{UhTw{FkjdypH)^PDNpctY(Asjpk}*iKMB-T!~F<Aj4OZt&vQ*hQ32o=;=$()~HKtktQ$MMrtEQ5wH!kt>7Ul|_#oQ{oTGcx?lJv0)^vYw=%@N?HJ`6r2Yv9tueDSchcw!x
z+9|*NJo4yXd)4gOZiD;>G<@8^&J{iiR3`?j7(=3<_N_|uXIRh1?QRxG_RkxJih$Qx
zhb!-*FDxv3_sk#kPX4=v;;vXsL4f+KgQ~YI)fxB2KQc97CpAFk3IprjR}2gpgB}-i
z^Q?cMS2^%cCjDE&11r@I(Eoznk&oU%{jsf+8s&YcL$VM3KWwH>Qyd@9x%Qs&&A~x0
zi|Xg=sBvZp>^c4`bi5yo!w#ay8A-G3+h4>%T
z|18)U8^D$OygBLq7j~YTr2oNTw~_VqFEpPOaxjE_e)gJnW7uoeJ8L`2>~XJ0Oy^KLu)OJ|;CAXik5<~}Q$Bx4
z+#8~Nc=5bx^cpx@CS#F~_5BF}@92Dg=6{u=Jigdy)=KuqdW-dVD2`1sPCSRbhM16+
zpY#uIV5Bpt{s`tfQNA=}oSyQ9-kC#7Jqi9bFmEa#zly0vo>@hCtlhtpkNy!p9euoX
z&R7-UEX*%-8rcy7S#ZBZ*%7LTiF|(*%@h|leg<{ar#7_g2mpUc*KVV_ru61RAH8qQ
zGS$qhSntQ2buxo2cs%_O{a;=Fq%6s=_ls@AGomaw*deE000000Duhu000000DzOPJqMG`
zG7f*&TNB5Ty~6D#ktCGtlu}Z^pIu6pTPQ?!LPR2x?CF-2#(6hi|IOjIYa)E-X?k^2~vlwaK5(0k=aN0~|vJq?%p7=u*V2%FBLF#9NyvuA`kfycL0#HfM0;}5O*1N00rFWgOx%f537
z93K|D=?Uht4UN(|0ZWijt@m5p66Hi=E;!5ZwQo^1pvvIPVB3;$)oa2QC}5->?mAu&uLd
z!n}w%Yu|8ioucw!J6I_1Y1{`I8HtZwfPb&$D|jp*%j)d;ychH8E8c&m3_~7I_6gYq
z-iTI@qkJ8hH$dy}_u3m6~fX-}Yph?gp@>JFnFi`VDnQ
z$H=dOnq2jA&|Y##aVI!^TQ;&0z85=(Fbx#&3U`$k+_|WM+956S9}KcAF%vyC`7rEcm2X
zo4XqfpOR{z{E+BBtxNW*Q+InE!H`c2KpQ4-8b6PxkmX!
z(>ZHSY3SU>__GniKI$8PS&BZSdJ?V^^`-GYtsQ;U!EEKU?OGrY|H)L+3$E$^_5t&}
z7v^-!z*UipX1YL6yVaBQ{jRQ=ZxolW%gyinp!Icq7t?=$9P#Rd6xCe;o793h*qq#A
zqX=G{Uq3|O@nwlVq4>O3a~)g)j*MNdrMzj-yC`}OA}oROfZM{Q?mfzsGA-uTA+oR^~~v;?!u~Z>ooc(K)Ky^|4jk==;M(^*edM
zfA5DreT08;jerpM63|Vd^UYUKqKlRL6P!rqo2BnC&feKgbrUI2QZ?eq7+Th1NG>5J-8S(Jr-*)*GkQB->B*+9R2An#
z{V%_~YwLY*TYsn=#kJH?zitZ5PQ1j+0%i7J?xTPCa@Q3MX?+Q1?tf*lg`;KwWofmqKf-!rI3MijWSX5^J4c;+0^@Z-c%4^(ZR2OTmUv;Lw)HK#_
zSc~!DKI=HLTO3lboZ`~^+HMc6Lwdm>k-jIyksd~5Dev{9^%i=q6cQ!BbCDnO5f{H2
zhe&^_+bd>LUh=S8U2MFD?w7A4pE;;Mrq(f|C{I1^X#D3P=s2K~N8fp8J25jueIs#k
z9vAGm+{{$^z`0K4sKwB~PukE)`F}?zM!*s5c1>YKQXjfJ{gL9OtCGX~llp2#p$hfu
z=&+dt>U$>A)^`U%j<8jigrL9oQeKGitMY&GaF8b5tFK&aj?=jqxVce(V;Kd#`2RgU
z{378w^wpt7;YD-~&+|+vz5yJm`857esN*ulp=kN@Yj-hz(9cO|8Dyz4B3w_?fXgJce$x1M*B_@T^l=x&8}(Kgyew=)F~W#6WN-6yyPZ{(Hie;v@0FfBk|)LA
zjj4V%iEY_4PkHWG(?Ri8;wXGfd3AH?;=FqJmH1Y#Ituct@P4FzY^p4Cnf8Bz=hG|N
z_xDDh8dI!vi-QBPl-T
z;;XeCK>p#g_eowpv$vG`RXl$~_E{zFpEypjHU=0qU=vGqaAw`x-}FBGY~j&O>quq<
z99s#x=7vmerE~f5fSukeac=_zOTiw6G2d=5b2uh5jqa`3g*iG0PS!{JM>^FZr*Xl20NBUB^f2vvdUM~b4rGg`M?Bl8uLWJVrabgpQ8<4@c`SBN_#@S`
zP;hz7KDu{x2AVXA=NqO}%J~D;+tdibqAvem?TUH@$@SFc~aFG|}yH<D~V+
z_~ZiJBem5brqbZt8NJSE*rm@67h8cc>-0Eh=o}(<{i1h?bpA=PW^l6Y!36EseM?UG
zD$L_Eta?}nyVCtVsmXuPzpyanz6zeUSQkS1$<%SQrv9w{#yQ6n{4(hAp6q#7DhykK
zF}~k-<$&GuzOk+tZ;ViPxDR<&=2Zvv!z_!Pd@hJrkEq={>Q85tv&Ngj$JhSrGYVQK
zS*)$ZcUt>`60MMTkMt{5SAbjH%azRN{?BeIq&#<3y8I92{})&n-4v>iKv|=LdC2>k
z&UI$Mzo|hzedcSawd0{*TgVNj3(5p8g-P?~s
zquTY`XF);fwOy#Qh(hVDa{s)^6fUi49a0@jp*D6TH~zaA>QZ43)rQ5;$})+2`9+LF
zbm{ZKJ~6h~ymoW=BnD@7_F*kAMzGX>XQ>j2815&Y&7a4JA^Xinu+(ADJ8)x9;zt@5
zMvuO?X3-cneo%8|H;r30abq5Bpiz~f_a(@IhA1{JeS#qkwW3O|vY9lL5+27a9YCA;}BH*srT?x31l2Hra4+jFmg6+Mp_RK
z<@|~ZSB~+R%3n=d;K5^3e+;kg%Hw_D<8j-(c;ue1?zw-EhsBq6*%{Y(cv`81+CJlv
za&+jaS~HL2pb0`z4Uc20H=BDhc+_6qy6cQNk0hPc;IJS8$=_ZzEV31U&=)f~E@hs8
z=yg|qVHOGqG#PgKVJ9FYYtHhS-U1Yt21c3h6X3qHZ|lQ&0soy2U8tETKvlbX$n2y5
zuMZLO1#tr2W|hl~^%wA)jd{@GBmsIIXM8GCNS*hKJB}X{(4RW#{2224*}jo}>rM$c
zzV-ArixUEL=E&bl4Hgi8cF_K4jkAE4vFDS1Y6`eMzcA>TvVi!1BeZ^5Bw*3Rh0%W7
z1Wb4l>ik4qz-WGUiO5rcBQ$Pzkb92p3@9{o7jQvmSSemhfc?X!80B|7Zu&|x@^|yt
zlo1+hAj3n43pSdTL-Y@AAATRhq0Re1PDM6{RSP#Zt`hOE8}$=^x8395VC=S~-IB*1
zDlesP3+ad8rPyG?V@t*1<*BbZeAzrF`2aXHy=XWp^Opp{00g!Z!t2)8=-kFEqFg{D8e#z-)n
zS(zgn&q6P!JyoNBo58}6gm-@|WpK?U>_y*I8ein^m7XPYVUljpRxp>w)HgPnSKVk_
z-k5pVZGgrWqg;<7e=^Sv)68XmQcwv#Jm}#<<407Ps*@FsROL610|qGU$XXZQ@fQWV
zrl9=O4ish{8db@jLgu{Hsc-F&80S>w9Bl{0NKs0DU^Iz;f`O+lya+$0`n&!i3MBKU
zG~w?v<0z=b)8Fs*icygH!0bNx|3edP--l!lHI$nhLJeq4IT$Y!rN-diJ-=}y^I4pz
zT9lP-DM3rXI+M3fM0e>fYP%Va!A$pU0mD38bw)Q#nn=NIW!!N!a_;2XkJskNQkd~E
z+a-tatkSN3%Iw8_8ZC`Q)feO#*m(vXN>XCc_1bY&k-h}|b#-wa4jkl^G-jup@{oxy
zl3q2)WAnf<`vZXjJhQjkw0{-D=y<<#H!H@(wAuHS4T(;N7EkT3rBMIW_2S4i8g9$h
zyGGw-FiGyJ-B1UME!~gCX-IPjwiVqCk>?SSa7xsF*~Nnn>u|Z_C!nBCaaf1=x-8JY
z){^KK6{)+?kZ|vX-t9-%Z796+?3`ivm2leFxn%5D8gCRH^fYR*h}_$%urp8sY~Oat
zmH4noe7a6Kn@7+1-zx&$1o-d$=4V09{VAg=zn}O&luZej3DLQT0e=;adrqW#dIP>O71hxTVsQ1(-NRg?Jwo
z@anaEia~^cDyya-<|B{i0TvaCjud9=Ot{%Z&i~mK_QGrig+9Y0-%<&0_lNI{*rvfi
zQ9mWwBZEcW$#z-G1sn|gdhIXm<#N#2MUsikIEXZEwOt};m-OP#dn0W2LI))T2{d#woY@uIEX{n
zE~VL#)npzsd^T>25wNf+dtdNQ0bBE@t$v=$L+eU>+I~$A@`kYk=OqP%s~;P&rezZe>oUK0Nz3O@c%
z>uanTWY6+!j$J0f>CTip)HEJxqGaWNf>i?kG5(Fda9DsrzE{6H;dWBLxo_^U1P1Fj
zzOnem;K3bt7tv}OnjO=ZSQ6bEw+{Y!h3whw$LrYE4hqBLcE-h~F=%xeJvEXifyKS5
z`I`^%h>CTz_Khazrb|qdTm__tjWw42&cS(}s|r8LLi0(%ye7hjh52rp<RaQWa9G*_9%l~dnrRY*ROJKW|z)FVcPWki48K?>CLEkk>!
zGw_@;{qGxO&o#!BOpYLYl`r>y+U-sHIJ?s;X)W=$cIau*Q{t2BKkReH5xpCKXbozS
zd;8i)CzuCN7+*6wJB@s@*}qp^B6&pF$aHrrsS_z)BWxyf(f5qD3n%p|vJ}fCBscsX
z@jB});ZObaXp@k6EE1LOu1_WP4MchhE-N{x&$F(VBKLS{UQ=sH_#>!)u^O^X?
zF34sqjWX%ANhND3WIRq@3c|yo3fWT!IWZSm=WGi*Uqz93Nmt(GCe|`w80|?u7isux%aBtb
z{yE>Fa__Sm1>f4Fz<%<7t_4a)y`IX%2Nz}BpXE@f`kAF|eVWE3-{P-Pj~L8pNGR+c
zCqcGHXoK+z4hN5#yB#9>4f}+Byx$?glI?H8Tf$k~FSxT;dNu=YQk>6pDH``eZO2qu
zP!KgQcv&YF&pHUz$-x?CzA>s3^2F{2UWvBvxj>3hn&+8?f+Wbd_qFZNMdMqvO_Gpv`;xO#uu
z^K#-Z=U=5f$fJ=glQKWm&LnS3jB0wLS9X4kY)5yi)Uu
zVK8f{!jW~PKa}4~k7eZk%P*cR2qU@qQI-DN6~xzBp+b(53WJ7Wk6^n#7UgfvwCl?_
zC`{e&*{mpkpg3XV*LpIig3H(FmT3aU1m}1jtm1Gp_tOK@e_1q)4EWT|WuV}2-~UJ<
zh2JNd2qk16YE_jvPgkJ8TkUiofMRM!v#E3ob34d}^G_&T;;_+ul
gW5`F6*IYGY0&jnz5pgz5QcJk@KPLw2^OI^y9X75>p#T5?

diff --git a/previews/PR1624/tutorials/out/solution_000050.h5 b/previews/PR1624/tutorials/out/solution_000050.h5
index 1f7390d04038c0360ec4ea608a5afa0e509d4695..d3221341f8954b182aa4b11af01860fd8ea8f5ea 100644
GIT binary patch
literal 205696
zcmeFZc|6xo^gl`z6+$YJvP2S*M2i|(%2pwjP$El2N?H_AvRAZ_D1;&{l2jCxvPB}v
zQWBM&_uE!l-E-#k_WgeE_kQoaf85{W{&Vl=kI(6xIWu!+UNh%B=gho?8S1YQoFp-c
z3;*Ni=i=oOrvDuK-zew50688i8so>xcX0l}&7q+jnm&&C2RD~E7nX^0*?;}=;ZQ0
zTEI)Yr2qTRdVCza?GD;|{m+K~muml|zsY0uoZOu}oc4JC&vpNBx)k7AO8>>f#Tl=M
z)Bas{-adPs=y*k;aBRGr_U!leIRPD2re@$*BIsD_|^Yu
zL|AV;jv45HN1waW#xtfkk8bpDP$e9fthJ|GByT-i@8(yUs{J457v06!$PO<0U+2jftN#1{nK;WLKj}aIlm6oH-#K(lkEze*?2`Ya
zBOJXy9Qxn&#_Est^WV$I_%XdJCS(1N=kQFpU=tVmDcNK$M^3AavPcW3eg;P#Kn-P)
zX&k6+AMTR*1eNMSJr=NCCI~vrljPWC#p&OkPB4qeDKX`KiVnxMWICg-o`VBt6W{aiLBJyBjf`IL7VMZsc-tA?
z@+d>SlNotR5W^pdVfaK_oM1MwamAIvB5rp0L0IItC9Yf+IlLYBIh#DchdYNwa$0ae
zY|@m46AI&6AxPu-{jX{=_mpXM8CV)*Qryis7Z9L5QkDN@T<*JbZw46Mi2OY9W
zPrDJ-3sk|C%_gd8Utl6xL^no=(yN89gOLBy^_AAM|N0odEQ=Hw-<7bH#NgXfmo?Fy*{C(#An#06ft|)8%8{*HwPJrPo5?6HWKGOIgE^2RG
z8Se%xSN}TUE6T&2UJ4+8L3bP=1U8x4Q3uF}P3(PMbz-^IA%myLS37CMBUKWd+JW}n
z+4R#9%THC=Z@_w5-#>_;9pRVOP5S{f=Su24EWerm+#l)BL%dcKKu;}J@oyXOt-9wI
zV|%H0n_h?lm8_hfih6FrMl#rr3Xz;mDxJGvL9)qy
z!3fx?Y_hWD3v4(xaqrfp-)mLF!xPw^a`K}ZwD+~`2?1=+HkIcg(w8`O)IkK6Nfh@L;fC*LA=KW|7OW5faD`B&Je*=Q;ZZP<~P8
zC;gszgEmi)KV%dZj#R=`pRS8I!$b%fY!dKDfR6XDs||P*Hjz5emWnh=t?n7>*RPmZ
zjWkzvmo&=nol&Cm7Gz9rj067l%luZ9+l8mK;(Uqzy%T`?$(IkoMrRYNS?_84+s}wn
zdjq_~==xSsb*KBpS;Gk&4~uXsmxBXglQ@4tx<4HAgXn%ZnI}&5Eqq|xfu6+H$S$N-
zx~pip)TzZ(|3)T26sVutEFzEe#QZn2k$R1D0M8HmOX&kd3h@6)W>jDAsV-e_WVy+1
z)H5l&wi4;cRF_w%SAA!}N2DMw8lM#nYNelpe1YqX~G{yusi)oW~F?38(rftW=y8-M_H
zo&m+sPBH}oEW(q@*t3}iYKZeE7(F$ZP1Jjk&hV3XFd(x?j|ao2tYOMcVb!2q!jR$H
z|1ohS6+b`7Us*HdFL4@yS6#}`a1S@yet0h)J=x@T3kFOUiH=~}5tD-90H0C8=>Oov
zZyNqVo{nlzZw?j%zx*jfrEoYPr^5jGdK^AbbDR#4izqVG5_TEN*)WS#9|@EP(hk6%
z=rDYXDnldRF#6fB>rhVuKMT-UHX6nviZ?iXCZpHSneP}qMh;Gm>W$eYi(;r>k7B4#
z;SSLI96Mq2Qh6SSf|H>7S2_Gu3RIyUM?R55&vVMRb9g;i9oUZ{XI!QnK9Zw%gEK!f
z?HN67Y6y!IBYXgR+Box>%i-yCL3?8qoDS8)uZ{Y2m_Uzlc>39Zr^AAJV|vsK;9WR&
z!ltG5$M84q4zzo$eJ{@Zt>(zb*1=f&A{_Zz+X`b)bfL4z07rhEvmSTS1;!>$
z96y!JslVnUQ_l@yD%d6Yf#E$l`?`{TU^clX&6E$nX3E=S7+&fy!&lOc#3DxS3?By3
z0NYEqVfbaI8GimdGz{=fIvNJ?1l=w0THVaP*ZYktfKARW#euR&1w=!nt1X!N!8;i~
z){v>kA74$&lk%AMN;rNt$&AtGdBONW^#EKcEK)!xf=xu{FzpC0U})wA2)EGhbaMqt
zyM8eB92PL`96E>-$R@4^IDssZuVVx|f<->YqC;hoH8&W$k6dEf`BW=R->7YY
zt;r(q46ngJ0N0$%0ha(cGOQC05Nsmew5T5W!#tj}pKtz-l-kzI(f*?@yr%W<$m{}-
z%_3!Q4Z#Jlh;$Y+Z#VtRsQ#%H=z>^eX-9<+wi~Z;3$`wsJmj51+bJD>P3=9VWepR>
zCdYVlVT-ZJ+K<8DLLg2x)zfxT6xKk)Y;wH*20bV2l2GG9{PgeroSUfke2_U$SSU)+
zan=`mQXD(D`u$-ZsJC>ddI{=fjZgE#cAlI)n1=N?PcC1C)TgrlH{!GEO^Yw0Uf%;A
zh*WH{>cFrimJ9SmnPYj+_{dzO9iMgvV!4Upb2gSI#GMmHxyRz?yMF@T{p0sA8>nPf
zO57iyQS&-v(GIdbGP@u6gQ2?=kXJcU|5FlZ+ctyB5s=@G6A(juY#3Ic_Zg@}L0F&$
zli2Yaf?tl#Q*){)k>(n`ND%;p!d>kVFo|Q_vH;<69M|B(@qo16R(8L9f0!4
z&8N-zfY+Zp|Fy_1_IV{DOSn
zet9X>@7TPW`j<76p7tYOx~;Yo=|sW}AsO;#6S5#QvWTheD**Vg4wIaP(9Ze8OAe#m
zUcwssIDc>1UI1+2+-!Su59-IxkQ~N!s^BH3(E-#sE@=?=)pD7Ka1{wSmUr7x)Ngtn
z2uPVlynSv{zoOKWO#Q)jQ+?_m*RRGv1?R6@tEgW(cXS;@Zuq`05s0%$*92I8P;Txx
zp7Lv?7C}U0L;fP1`lol*n`t|)6_;swL{SXYAGdbt7Oa2lOv)#ux3$aYJXDP6Q9Fi+
zGp*NlB6bS4cU(l-9QC@LGwA%@9ufj%z$Q=k^iq57Pg_LWU7{yLsmcXJR`8v&fCW9P
zVQ3ckG{*y!hQ(rFXOX*?n7Cy0fzfN;$WZ%Tm@s3L%PXOKpcVh18_?f~VF0K}FypsB
z@ZbT5O&)b3FaSI)2W}MjZWnOdKr1I8AYc(vq(%LIW4aaf`*tF-w7hECW=hv<`BS~<
zP$teFGXyczpNZ)ypf2FGv7ITZ3@=~D=moYi<60TO$dwZqy=%SB^VzS)m=t
zM~r@P3Zs9CF!bz~yR`j#rw#O0V|O6$=aehKsz-TwBO~XIVCZBH
z9*gJbzw2h?bDuNcMe#OMZ>bw2|22={b0~lXzb?ak&xSfie=7%kRU+c@=%AxnYvF42Px>>JK}1#`;FIPLD^tQ#|k
zBG^A>HIV+sZ4LEXIQ5TwW_Zo*%zmlLWB3YO?x4T!Bg0PybcOZS>ofc117ypve5Eo&
zC#y5%eUq^xHrYHMH#Cg
zq%Fv-NTBIZx??W$IuMn<%fFO6zOP
zV)W+izzJrNBS&4S{Qjk6D1f}6COQfhnR}s$(oRu4Xu*BJ;-6qRizrprz>0=E%1A%u
zJBJ-;d1mTFs=qo-kJ433q+!62pA%7~{TFP^q{Qt9g4=qnIm;fFK@X(A8
zv|dPR4h#7mwhQ^O{+&H(v|ObWFB~9`Tl5i5Vz93|KG6D2;dXEUV-YhMOr!yR{VfcE
zg-xPOM;{?>I3y{u33)!L*cGUE`Pm@23OH|lZwSHi7|lzMP+n^(NAbo_E6k(-9yno3
z`?s>+yc~Jm<-MB`|40~d!^sWKnSNq)9965{va!BDzXcG!P`stbj^wy)*a@PT;
ze|ob#iU8lN#`0ePdg}hM|GofTm%R8iLHiHn75&Z}=x4(@B4;GH70Sbs`7SF1t$lj!
zUJBGJJ-9(H2xx7%@YX#bFX%b*F9WEyOw!xAAP>_yy*Ux6&-ihocm6`0Ipfk{0d(ON
zn{*?PYu4#5vITnNblW{4=+FGpscR9yzfm3c(R>`-t7T=>{02LxY;0_{{mCNF>N}Nh
z^TN5qr!TK$Jew>RDLY~e?R6@x9GEh~A`QV>yo&^6Q&mpHO1NuM(i%*{*C;pmLAD3r7Gwmm61z=aTn7)^16Gc9)y&n350laMmUt
z2Pl_uQuW1gHN4BMng%rVMjG!En6IvYU_&{u*C65d)&`J=Y?`Eq=Ntip-p2)puL)+DVH6<&cGwSlY4+l-^;1(2KkXPnS<(`fJfzCT3LUCeYEtI&i8(p
zSD{}q>Y#UHu@GMrv~$majK_PD7n{SMy#c$VE&bvVAJ@C;q@i43g}5uO`+NtZDm<6f
z|5KN2hjNckwTlaZUYk=v?}Z-A#7x9>bgabZH?C`*)f=OB0e!pbanuT+E;hpPxITw!
zre4kmI!=9!;4HAiYhvtrd$8BwcRnBiSiix`rh0)})^I|UW*Q)b3
z^f#%`C==*QU9p1GQ0_lR-Fz05t9vHjtyS+u5@N(f+dTMaoFe
znuJ;6`BvU=^+P=Gti2kcfaj?&zPaace_Y3!-=$}{-^e;zkI}NnJ&P($)
zaeD*LT2^*F4)pFA*Sj|Y6}TaA9rc}yRGP5=&HAkOI3KKbV*t!762uin^B+$;_fmiT
zZE8{~>Q!autV4Rg#Tf!5;P4YB)Gx*GNm0MKHLKSI?Mxkcz6s0oZdB4Zu(a+ijW6n!
z{nS6toSsVa6Z?|lslVD)7`F@ceeE_;y(jm*0Sdu=(YFE-oCn3n8%yw*#smXZkJ(@G!qKFVMK;;0$2}_?|F12>5W`_gPKx
zLulN&@koMxw{ANvIxoEHgKN+Z+1StKNWXEnG-Ca#$^qK`9MM34fN+mXfj7i~Kg_u$
z7~cJ%{O)@oKqu{B@cqiQWt2CjfFAI<8Ui9XPrXKf#U}GkGx+4ae?5(>E2Izs!a1^;
zIWI|HFQfXcqYN&QjcBFg%Qieu^-WThsD4oAJW3s>Eui|l?N?35b}W1$3AMS8RaQfbz3|
zCQ#mjEb#X^`qEZd0rKDaOt~5duU2#L^9h^|C{Ofa><%em>VKZi$oCpDd?LpVmEAD?
zpzrF%jQ2b}00BETyrTyb$m5_o)T6@ydK}e3KaVp`
zU3I4X1ne@Dw{ZHY$Il1+JUp&Lz4fXLzZT5{`8eD*K-1|#hfQX4#&3Lsk&oS8Q^
zQ4DU6J;#(^sAbA$K$aEl-IdMY@Rkgwot0Oaev;gn^(q<1oWI9KGVA@>KIR;fV#e%K
zxmC=0JEJ*I(yzq9ERPZ}vla8R*{S
zUQe)m(hqZ3!Ei6^z4SWr!Ob0)F;BPn`C@Q*kjGVRjzoFit&21+`+wNuh;fG8(JDgR
zTzB2O7|Z9{eUeB0#+=9Zkr$gQPun}cMnnkX{?>=NS5V&kX{iS4MMjiOL*8X!LLrvV
z&Oa-NIN<2;8kz@uKh6Cy<_Gsh9^uCLEhTZU6Y=z!#Oad}ADwVfT7Yp?fB3N|+VS&_
zj4$&45<^Q7SJz4&*eVZl;gVm1SpU!X1>l+>pTBtK2GqNgaajVXRsO3;AE4SM3eV7v
z>_c|+98|c8_bTS4vv2uK=5MQNV`G@u&SZxI_#o)ZMS<2cHUtKsmBKA9+
z`x92op9tl=2ahNBO}X$E3f4GXT$*EcNOu2fG%VR*vhAf%8C)sf4X4#O0@#EQ$a>8U`f)wuSaT
z1q?Y=3BWjt!shJ+yqxv#deA@3Apc2X)T!!c+iWxp(zvkC(|u=
z2H|&JQ@`snyuWIm*?n6V-UklNRkp8(^*qva@qrDT7iW6vNJ_wYmo3eE5$sv&+;-%?n_ea4{|&RPFXimV%k`(W=(k4n&=HYIwIDAbo;w)@Ww(0e`p;G#KaSmdeY
zuIC1xP(S6FYJDEu7p>76&H(-SC9f@{`N98qyi#mk2=?8Uw27t>pP&1LJJeo53pqg-pm00
z?u#92mN>2st&dBAuILb4gZFR;))imF`@K1PbmID8Tu*Cs44(kK`u0<6Iml)6whYe(
zeQ9pb*XS=2TRe}>hyK4jSU5Tx#Xq
zesMJeK6T%%-{=>*f8L_!Pn*gB#kU~OeYda~`{nw+Mh5LEf9y%~G?Q~))nmU)R=3f3
zSoKL`BI;>+6#ApQZP6M9)UQ3M04ZJY&pvg{NNwkT_>OieUZ|LZR&GqFu?MGW%uK-pM$DtfWX;gd{>>^1K>r?{X{X(qhkIbj>(y8JC01n6-%Bxabu%Z4!&H<}m
zxBWtS7Jn3#^U5O7Wy8D1XH@?*7bct`@9c7op64uMZ&Tc|>25WZ^YbQBz07qiI__yJ
zUMr&hv+&eK;P2?F#igv|#!{9=4WwKW+`L29SqMLd$(%^&%g#
zmf@q9Fm_}gW$1SMiQ%N1+HW;9$Q>GQ9R6GY|7C;4?z|RbdQ%f4fJA>P<>w%9kYr
z9EEb}CCoY~P&r5Wb!VA#<)Z`^8U7HO7nF~>3;mN3?~@QpVQY-y&W?nDepe>
zEnQD9^LVMB$x%K@*CGG
zP@5>+h_c9rV$17T-Y~p21i*s
z>pbG;8u{hp%Ye^4w$LB#Jm7C`=cam$LqIwI>wpqfra&oKyP2za01T*)jE&Q
zW1efT{@Le<|8s9J@lnP0s@QIj#~;Z#h5oz8GbI>tc=kK7G<@$8$+vkn=4%}@n}m^S
z$E0v!UcEV>b~VPQ>6<>7%KsH|GF7r
z{@>up)O+`!y^Bxud61qVS^4)hu)_<%8`T3by6NP|IOQ@{jf^uyy
zH932rjpclIr!+{D%U_ZxzhSwaxdVPkIjKwdH|1$cC(jF
zYGM)hKI#6$cCenBr$q7tj<@(EBV7yjOnFo-&I0{f#R;dj1ONGS&DRMB0iV6$J0!FQ
z&I@aP#Iph49IAPraS!z7=Lb6p0`8o7K*L*R0o2nNe{lEHejEjAG
zFX-+Q6H$kL*T|)uZBHK{|3u}ibbVr0)m+#|+xi-^PWF?Rhqn3T
zo*p7El5bC#bGM(Y-LtS>(;MzR?nbNH{OKjHb++nUdNN3&%~_h?qX$XfgzAt{qki(d
zu<*>YSr9*5;x|@)>mw!pSu?%0U|!os;;q5Xr4wTUrhxrD(`$}Pf9fOOGd?^^xX?!i
z76$4I-|Qm?v=2{Q4C7t*w;=x(oKxP2#tJnx^^%_(W(HWM_mZ%8JGUI(+DjzOmh%*;
z_7NV_i4lIhaNi**6}Cj6mk3>(G<>J8hw%T~F-1hEhxE=XNt)@=LtePq@LX~3C87SU
zD&l{;N%Mx6%31b3B=_&a3|q@?Vt35El(==1md49>JvR4{j6jK`{O4(F4vDaUI7r(UYH{OktbZO#g-v^K!`;-yi^
z4}cFAUcX{qTOX0G8@HSn=6j~|8GSDiz@tmr27BP%#rj=esQ4hvlXIM-%7R{URP?R7
z+KWEY_1HXf<{bErX1gNYVcy&wN0LSxS)@F|T+nDy4_TSsk+<BO7#DL}9&FyndYoxltlL<;8qMSjXoaH)YlVb<|vXY983X;ODW*df=BGuDF&C
z>#2Fi_w9KgH@i|YMIGe#+%Bo3UbvrGI^I|8dw5yvG_)Hy<&jS5vRLGQk^n=sympw#0|0OB(NH*|QrMtGS
z2AaqBJhTerMvos(d^cipTl~qaVxnT14L`
zhNp$dUibmHvu)X5OORV+I=;typ2-`t(vJaews~z6`q#yZsqScRn_%rL^!K{_r)j($
zxcjI8{du?g>_sSFV=Dg`{chfgeb+Ir#KsKmN4ev$%@Nf5ZmmM`o_d*Mma2D%tNN`bC0{)wr*ICq;*j+vw=V8N#f0Gc$OReB`NBwWL
zqmlS~q9;DS9vF8FPS)MWd#0~1{Q&^LIZQL=Eb3?M5nYb_%;Htl{@MGFP
zY)BVDz2Dp`0a?MjMR^;F^Gk1yqk0cS_$dxp7;Q@Rcd2x&MLUHH>)WxOs+I%2532tb
z*^T`9YWx8o*jZ@*3H2UY-G(310Umw!hu$X#|2a(Wh4ObCt3&-9VHf(|wD>{Q7UX}h
z9yK7f%3np}d!*Z5itE4LX6|oTUq&hJn578Q5B+rUS5Q1N{}(h1^}G+jvpbMG9D!pW
z@Il)d{3x}M$$yGm!521gUmGn;<9+}28me!5^d{A}(zc@QhHEhSTX(Bpbev+ne`wzE
zU_U0fAwSn8L8-*n-&B7j>{B7i?RuGe61GGVZPz2<0iEA+&0AD2_yU7N?P{SI?VQ83
zx6BY=5%9jdnS5_qDswNP`;F1#kp=Uh{TzRoW}u<=8>s#kF?^5&zt@|@;LY|!Fnz$U
z(FPa;?U|-9zmrns+^hJ$P^5CLQ75W5az%yO*>^dTj_-R0EFY-1?=W-D`K|3o^&F4J
zQaz>mdO9BW7Uo_gbIC1Q{_CSYt#@Bqkj|q+=2b{YL*CiWl*%7%{6O{2?*rHea^*A3
zIWWp%I$Xd(KH&lL9>giNk>bVUJilpwT7B}yfDpQ$KKrFjPfi)*&pZ+>u!|M^`f%v#T(k5aHSObztEs$_~{R{Xk
z@E&{F!4-N~{({fD5$Od116A}(tfQiL&_782`4f!qnJ<*L!%uu5&zTbU75(cG`9H_7
z9UIrBPM8N!{czog0Nr`}+6!!FqmG3r&QsC+wN_Z~`@-n?_}+M^dY`{7>bq?`hxLd3
z59y;_$7Y;^3tEVOf~~(KQ2t`-P#*BBWBK5T9o%0&T+5H~>AJDC2gY5`V6)Nz;L9X3
zWzE1ZU-Zqti0vP3ux@(^)a%XzNxYAL&A(Tp6ynh#<;i*3fd9r@G#r_5CWbuA_
zx@JhqQz#c*5Nd_@{R^v~zpaG$y(n6>{~gFX?tU~q1A3k8Q>^8{KXsIxy$s^rtW!~v
zE`spxA*H$k?YIA^KZ0>`*=9Sn9?;JdzVFct@vVMU@_jwXUn)djQCsN8T-O
z$b-CYM5^n{iBs9+{AQAQAMlKf#L>eU=X=Snr_KV-fKxVvq;mN|ywwSlJ(UZ2v~QC9
zlSM<|ynEqggQjagNqTm6Z?zJe*rt5lDY6jqC6jLU+*l0xnB^CPKjsXOw~IWZOs4!G
zem1i|dTnEqB}e|1t|*55)Ty`GL(u==3$f2Ho(_@lDYrw;$PE*dA?X?Z<$Wa9Vd4pu
zO?^anCoh+5${>kfthx2p&>&H$Q#=0VR6l8G?aFghZtID!L*%)9#rxvFgG6Je|En`B7ho3w?L=wZKYF0iUB<(-%&W?^4BCFyO
zeN^5Jk%6AJ!pVP!i2c3t6la?uvi->Elgl-Rh}U19!CB%xL_581_t(QcM6xgO#6pK|
zaxQLiZHP-ZQC%5yA!thv@%YT!9rmi5Y&g7QosxbJ`TN*7lc&9ltTJ({^xfJ;q|2w?
z{xjG`B!^yJPTJBWi?PzPq!#a6Blj}LZEZpN?
zTW)dqJKRIesz0FtcE1(;`Oh7zoX3)as|?QQqY
z0#2I7JrnOSw)}`nf7J~4kGF51cYu7io>OS2BdqT^f9iSIfG5M2X>Lmce4q98bf*~T
zZ=85-YCP;y5zWfRFQ9++qwM=9;9uOo?aQdvgmaY<>t`GIhw$#gaBj_1q>EGJ{a;*XH$?HJh;?tT(&cJ{F^%5RT!n@$5
zdXq6vzu+Qw6~+=s?&JyG$#oD00i8J$+Xiv`kLUo08ro_$3T(diF~=e
z6u&(E9=99u_a3E*HfaB;h1Q-JKi%HeTtfX`u~8bwFOG|vi{(3dSj$j<*23hosJGfe
zmgcR0UY`b;S%@Q2rolKbH!|JA5U+2XbPFEr!hPLODIR*dbUnpCE^5IPR}K97O6?RaWb&>?xnMrDV`XuL;;c`L^C<6a7eVomaa=0J
z_gmY2XuFj3~}_zWj6>%mHolkzyJmN21W*%LC)loa9#xzmAqTv)eKOuf%Z?U~YrCI$*
zmqaH}xy-8d^!%{JuPGkof|~QrBYj}IM2~ngpVJyqdv6|C(+4hV7NfslUQ~
zS{H?YBT(Q41!kH^@A3ey6(?j0JY!f7sVc&kZ9gleO)v@WR
zxSyw2Hx%POQn@+tK9;)$j)RGWd*=WTC@?5wLmy!_cc{~-Fql9+*y4dAy*r+%1+cv3O>Y4&TN9_sy_i$JgY#{1}h
zP_DME)8-ZUy+dj4hwwdV(gx!o)N5fC+TicuZZ3d+tJbqUt%B*
zZ9Tf?;(fq23VL;QLoH2d={(A&|pz2IR7;O#Fi-{NH;|6Jy`B5FV4XC0BF
zke8jl==0M((C_Py>)S%x;62u@*=r&Pc^#N^AHZS)%=iZPn{
z_KCwJLh#lr32S(VZuU&#&L$RFWh%s7>kT+LV4-TsmSJ+U+*@z3Vu-Ajx4Oaw`6iL+
z>MaQqVShK4EUd`wC)aBK?Z5PAn0%Y=Q8Jn{2*1O*)AF1P-;=Db-89-q)PDMnMt>Y6
zIyupL9m2z;M@*l^Wi&`uzu139VKU@x%Ir>^n+1P|z&}k>*J7Al4h?eL4ft_Z)OpST?nj#S*^{R&MAd
z2Gh05TetU+B(~N(A>KaX9HYs5e{~1pOPIdFdO`VEg$PIBFR_4%a3-6V-8X780`*e|74
zCl1Vl_2T~_&lk=M?A|Lm23?RJ=yq1uKY?=r94u8=7Q3;(rL|B))&K
z2j^(_kgN4iL@Y=jiQ?PoV=)un(Ma-!+=c$loVl*}H9=fHW9m_2(M{?LhjJA3d&yjF
zxu7Cg&l;vrw5#Vr+_-f7u_)LlX=E4Cg6n>;dW$uz)6%K+m2$9uZk1&HxdQv??z~A$
z+<~{w*;(ZPcFxp_-`oLsZ~CqH^+l++LvMFK?B9LqO?Ae=uh^k`vK`jdTyag?X5fP@
z$DcR^`tRpGc)t_+Q7C!!FckOQQ5A(Va4ur2%<}ICyB-QxT_}e2Jj?wrJtrPMabimb
z_=85hlS%fF|BmJ=v5NZy{z0ksa{%J3&_lL(FW{}Wdr1_GBh=0{KNtGbJosj@7SM6G
zf)mj%d93QUM*kOpe
zNCLehmyYV-dwTH`J(@LuXUk8C*Q@|J`_Vc1Js^K+8k2AW`0}GSPvg6Y(7jzv=q1f9df?MkjBA0H
zBIhBWp&hgYzvrTDw@i^2-M!{^O)i7Gb}0=U9efUi58RQ5yFDMmV3wr@MuF@=*Wg2@P)SFZr54
zGur=rP6o~YoRM2k0=h-+tWoB=n>kiYo+mHwXh;-Jh&l9^Ik*CN~e7
zM^iFc-H7!b7<*IsV;>)?*Xv|JdCvpPyKlFt!Bnq4<}Q_Ac>~V_q29LLZ|L~gMTcm6
z#^RS_P`=Vnm*S-Z{yTpl|7EWa#Y0lRR?s||V%ui8fd_mdAWz42wb_K?CjI2M6hA(i
z@rcf+t%+ea>MfF4@(Sss>8UhNoSGwX2YHLIDi<8@d)~?A^gDONl%qVxM;rhp#E~Uu
zDUMTqSFMSBYxeJdxV{>XS>>SK=6-7ce1KOJylH!OX@U<>uJl7p8S%)|a;qBTy}w>d
zLOtQ1-bMJm+g@xb#Qp*XPwL`&@)1c|f#oBU2hx%9O>^Kwy`6^x>v3PIOZ_Rvd)4?!
z^P;id*(bZA5Wn0UKd}?b*U0L%;63Um{!jKJ;I{;M6+U1)v46Kuz;)x8;SIl3gnZ$K
zgB7SZ@*;V4+Hcw-=9Vk_S4gdfHU%XNuGY
z(N-uw+@0_N%Ws^sqWDa^WKcyE{9x~y3r2YE61{NV6Yo)y9#6l3_eJ(E)Rbj_zxZv+
z?|3+;2Ah|)^n#tD>kif>Lpk50g$Z+k&s?-oV-EPWdC*f$n$bjsjm4AXQ}q`mV3+
zYI5ozE(g!w7y|qa5rawLP;L|FdISBI;H3}E#(-lw9C&rzpxr4aX8G2Fop)8U6l{P`
z?6_bw4*b`_m}ymbKHTy;U*$dlJhgkhivZx4de6iB1kU$2G&ih0199$TTGoq(`HHe;@IE*%bH;aMXu4n>PG_dq27HhQEU0J}*-J`LIM5_>W5EXU|Q0$v>@UeR6RA
z4^pjMrUK(iJvKNY0PY13+>oC782Y{6lYHmUzX773TV7YF3inEtGX8hq{UVuKz3}HY
zz+pb+-0}GXWWce8CH-ocOg$SRW(xT3@%}E$xM%R5pmVO!0?0Fr_y^{el?;;8S0=mJ
zLH>%^9Qt*`s*j9D7KSZ`_ohZ?#=HIH8YYu%5}s^(40*0Hkv$J*50Q=8CTDKIeWc+Q
z%iOuG{lwu*{sG>N!{kfuGePyJAu{gJ_t>1_AtJa(e)b%LLGpN}lhk#dVWM$x(DC5m
zAtDkKR%{gv`8X+Y{U5nQ#8SUN@v-d?xirX~{08jsjukg5So4(=8o(+;@9Sa^m?jiElprUK5;1J0SsOfE7F+_~CCKo17rkPIM6Q(mt4M=*g^L2uZryr>>z^1dX9DX
zJBYflan_~Y4)QZ>$noy9PO|LbRgV_gPO{_a$f;x2-DGv-*^Nur_Y#euH!GLGy2!e@
zaOYMzI4`OD_iMwx_RzP$ZTNe8hk_Wj}u
z`9;aU;o`|f-Q>gd7b;v60GD1rqB0Zk%AU%zJR#1V#5XR?ulaok>D0eHGeoAFJpLwc
ztGvID1ljxy6@WaPdzpi%Ea1G8>9d4iLVm6Al12M{z(b$ZVw;j-pHJ{KE--?9rk8x<
z+B?V>hE(c&<%M&CgUydA1%NC5T@{&G0Q>LBk0Yj;U?(}Gx4Q`Pa3jHwChURz!JqA4
z=L7pr{F3jsH_+Z(w#S;Zu-?X#Vu(m}!l`m>Lx{6c)Pyh!Uw3FsFq_`PrgK59m7
zBH}BLxbi^6d96w*yU<@hn78;O#wj0<=yMofmi`KRQwV(3e789C``)$r7cuVLwG_O8
z@#NXV0YBsg-}O4+JN#{2E-DQ`Z{CH>9>{qVt?nY7B3%Ct@z&_FYxdWHf1%{L6YrC{
ztcF5AfZQ_SaWUep%sw|ru)w>|JmJ5$fjTW&LC@Q2Th`TJzK(m64L!$4lpB3PJewa_
zE`$2!X7&DPx8a-OcS!wQAJThW3vIl(h4u5pIvee%NU>JI_%=FPPvg0qupGTt9@cnG
zans2x{v?chf1T#8!g_F0G^6@Xei_K~^xdcTh_kMJrT1gMXDi2`er{3mDbzdcmMw_X
zSxf#S;x6;0r{r;-?_Jzhh{oC81(dt|I^c)>@AEuH?-vyeB`FRq)0|0hSg7tBns;n>bjwD)p9&Z0yE)&V
z4{3ir>Huq@ovHSGiAZhkwthlARuTR|FSyTqGK0SRe>bW=O)#Co;KX6hC*Q?^=SM%|m*fZF^fcnRAbO>YWYy?N6y_@+0tT$py(Q%nOyQ(7%lDZ64lLUykqK
z%dMv!Py@YmTZ1Ff(B3JReM_H0dChTh3+*`Jm(QgEdThSmXC{MQ-`@TIn&A9++kUYn
zzBlPFv(M}R+G!S)fqAJaNlPnyNB?>MFL-eQ?=)JaTwI}D^#woKpTTd)?V6Pl3;yfG
zL4*App!{u`VM>7ppsO6{RtzSp0)KY-sbe)_AJAM!96nT`Dfa7@*Oa;;~8AB62S
zpQ}NC36JL+D1hG*Ub0d98{m?Xa(QwUa9T`wnM1{9xF-tZ3g-jAs{Zt5(sgL>dU-}V
z>P7hU%#eq3|0RFr@9}1^Z(C1C-2lB^htx%*Ku=Mjf1)PntzUlEwgKYR6?>7K_v-*>
z`)jVgZ2aQ|;A^nX-Tr?5FuFz`bSO-Chwn2guj)d;U6+2>4Txdx5sl
z5NTMyeWSGNAX#=LC&C%>3Zm2Q?NBj=zZ0y%7xVia{N0h%gTFk#4-!>@slwIM2T3mb
z&+kJp{ykdK9?b@j-~Svta-CYPtDM`??skzcJg$7RKws!$
zvcG;^7pb}FA+n>ihlC8?d}}P-M>6ipSZ=-wINoxTs`E6MZwnrWaKny!?Xcf0?>T<=f_?Q~e*?J&xZjd5u;DfABW1y+
zxD#-kI6$4(mGj`7wQ0pn}&6+%PVe`7K_7eu(d-
z(vNTOcM5h~pBN?!_5Q^MI5&Wv>cno{1klqc%UiJ<^hETcACy48E5G-P+Z^CaQ;oSN
zLjM-c+bS=Ep7{L)e^7rOZ;qoA&^V(+lW$P&n_wJ{@7oS&d7Z=ap@jLnY43o~lc~3H
z27dngAqAX=Ro@KxyMZP=eI|x~7vj`|oJ!31m;6|sjDCHf#*f}>EV|}%82$e)zF-r7*S|l^t7+@j$##O%Q3@
z^u2Z%2W(lY^gW%cNoyhgUAx6Riso29u_#a$@y*7KS)r(>adgsnj2HW6H}1xFDUbHP
z7ex7*zlFDPKC%W<Im1(>@J+8J1
z`Hqa`G~UZ6wf{uEy~^bpQXy{jiP-M3iJNKOM6i4seP1;9q#?x%jTKLBpkDOle``^1
z!XY;r&o>7r(|%R9F~8$@@UDpBmsxRzl{mgtw`?hnaH@vqKEVI6@1yvj#KoPySMqt)
zLh}$CGZzP8yQ4x+^01%E2@)naudyvIM{!+c=QyNboSGz4orZeeyz^X9uf&EY8}-hN
z6SPD*64+O^l8bSeL+B5rkLiYL;04P=VqAS{B^PADDJ1W
z#gS9+?~M5HIw-lre!n+AbKxZ@-#yO#8?KX;lXzzMigSbps|vk>Ia}kEc1QL%iI$>F>hd;4hCIF1fxF;w9VcU_&G5
z#dZ`7?FK!$IhXTgLC>pQ#U@)E{PefqLQf6h9yn@oVl(J@l$*9)0zKu?)pLvScaJ%2
z2VThME41)R&sqy{d7s!E1F&ba%lAj~A?}AxUB9CO^t=o+l5N4SzRQp(%7phus|tpk
zePBGhHmF@w2fddC$2X{e|2F!ddq)BMew9mVh$Y<1=xwR`qzt^~1iN-Ah?kv(0rmln
z@Gjo={JLXp@Q#O9R4;hkAbECF{8|gd4eu|KZ8@vpU0k2gwUy)giQP)?J>0B8((s^s
zLiXbUa{J8vdpXeW!YYSU9Zm3`&jx4yoz+kFzH9!G_j!<95n6WisnY-vX|&5vhk1$>
zotx;l2K=_i?d3ba!ud->TklNEAbI$t<$1UJ0Qso0qk0dVns|x|I{k|cfn1-
zsXnKBwR;CiV)eRz%%e&@`~
z`}aK8=lcHsd9LSq{`&Z%(>Zfy=Dg;f<;)0CuP+>!u@L-cW{KZhxwpQPRBm;()-UTMQlC4<^7eLf=Jx2ZbCMTteI>L`e%gJj_1Iq
zDf?_C+VfT&C{}d_KH=F5KCYM!_eUcSab_NZ^}=`TLTS*WzV@|WoCW304}I;_4)y^5
zVO5D|S9*xq%#7BN418~3+sJt6pHkaAxGSOG93g^c641|D%^&h>wSXR@-|^-o*yq?L
z=`nerM+*G7_3Z%Y8`946X9&Rg(7WC9mpy^?a-nIU&U6?@$3Csn6NdG2-)H|x=+C2h
z)7niD=s^;TM7rTT#rf9BmcuZF{?!?;p-s-`dw$}vm*>45Wy!JogYD}cYH?am8a
zFRPjogmJyB$<2a5W3ZpFlUYF5H68AKJYRX=q+p1A>r
z_ikY*;D4%mAMOFXsP%I%^1XI>dIZ;NACJOP?{g`KQMaNg$sZf%M7?9DMIXK}rh@Dvq7
zx>`dlT`$K3ZoZ57W`8|xod4729Q2b0+$~7@HqyVHtl5U^s?3I!F^E@}#Oy}C4NBsE
z==Zr#A&{;k9KPO}$agNJBn|nF-P~{-`S#0f>PEfe=iCj{U-#GWgCd-#g%#I>gA3Tr
z*X$5RKBW~=swihy#Dl$4=GXQ__xsRTI
z#GCoja?@5+(esaUs+Q@PUSxBg^4TP-({);ZzZ=Cbfi(&1yPP~g%a^k2p!DHGUn%T!
z05t~E*Q6b$@;vJ>qx<`J|0J`KF7nZvu9wXRSNP+4`B$vR1@%MGeMPkWq)*S1M!u$0
zS$ZxY=(|P>`9vSKbfJEh8)^`XcBR3FMbr=G`-u{It|jz0nc7D-6%JE>{NwyEjRRl)
zLndzA+oQ8O@w+7c^-=%k;1+xt^Edg$`s9gvTvNLb)%Tt@g;9MdTrc|z^4lr)-9SF&
zqHc&_2<;m^Nf6h|1qQDlV!vB7*+^Lv((BhIo8qmT^Fs6!3ARQ3uuuk#&bedt8IG@0o{1hjpx{(uE%_uMcenP{swHf
zgqJz}sNWdPw>yXR4BrfyI}_%Ih`kbHB4E8NZmE*o5Br?A%M6R+zz$hzq&}*d8@
z)r#%#Jr1Si+FN0r7ArX4*$Dd`y{9r>y-?1I+-DiKmgrYQ;2l5EpEcw%r$9TLx3Mpt
z1@lZ>=A=w5m`D4+#8yO9fxYQ(@vtlKO{=vS#(5!J>G~10%e<^KYB2=Di{4~+};Yrr43bWb`j+8FGSQb9*=rGQ=K?`=+u9JK4y4?mmla7oD@
zu950|E=jft5z>WolIFddeE;O({NMQFI3qupzm7@%J#WhKPP5T%Z3Hbe<
zJI>7zhI9Gx1LrOe!MbqLbibke5!mOoKe+203G%ozR-VB_bEbgbBETf0ql-IqEeV&ixI&5|!cnX7R}zZ6zF%UnhAgteit;
z{;GSmt$;<&G+Da~l(We8_c5OWxhxWTuH^6vz_0zepIsisB0hE;P3;gC5x2Gw8NAIR
za{1%uC;PG>u6|MfWIGnQ`7Tku%78`wrM`=PSI8o6u5Q-Sl5_7(k1
zH1Nk^MRXmP2mSfw-@f>VpdT&HDe)+7?;*O2ubwMxg?+Dh_w41+PZwFQYm$WVzUFT3gluu-^r#Ju4jqKNo!)
zzp+8maGv{q`;BzazeLow+)smf;boKcZj2kU+i&%|LO9PcZQK(+6YO}yrSJ1KKuo3GaV0`sU9=kpR_!j!Cy0;JH$#!UZGaK^X8aYwC
z66ghcuNj8`Uv5_YKRh>?`pNI(Y4FGpN
z-_KhHatgQ}UlIuE@86cD#=*Q~oE4LdcGSYfr~GmNzq8i&+GR*rbF)rGJ7`hArX9+C
z_2ZvgIKM?ovu~lkBxCnX8ug%%CwHb=0&HxbKfea(lkB;Ebe-wmQ;K?l&Lo~`o*{|{n+dW0Rh2pBw~RGu8$4eMt)3xtofr1@sNIR2iyw0NT^)IBiPw1G6^dX_0b8wyPO)t2Kcu>#A8<@W8tvEP@
zz`n6CiP4XA8>!yXaKM4W;8l!&$na`esz)v7pGxf_ci8$=uX*caM)^X#FH(F;>shKt
z6?t!K#qte*F!$f;i$N{H`Wrx<0(yqyLy8|FX%wD51=r?*?ivf~5%4Wv%;dWR-sDIx
z?+d4NzW`9rApL6~sDVJw9xF!iB_XFNo+IH)?K>4pRg@kf^pW~M$G5AzME*(hyxw5`
zoS)(+5n5hZO#Pfk_KQ(H$XqOq`tPisV7e63uWI{9V!M5~_K22iFW67@xUc0w)UFeE
zMXMOgO<$;+h4B33M9Md};0&$5HbIx-C%2rX{HY!zrO3DbR0^fn2SrnTGk`t*(c$o@k*YG3;+z*54o?3+Vm%4&(9s
zFLALawfB6EIzijj;Hnk%L*k2+g1~jqKV|pRxTceZgQ=dV`(+Y6&vJvvURdt^I%9t1
z%kXyih4nvrayH^Mz@NIEvQdz(+NQPb0?Z31^^bH;fqDGFqC`VwIOh!w6yA3l`2HT;
ztZIgQCHu-v0YA>;{l)vQA*XmHT)6+DxNxBe>d9;GaXXbi9c7te=GA`&?OkKiZuhi2t38dg_~2+bo<9t6c);p`Q5hRtyBRfctD5
z^%3;TNlUFdKO6A
z+skee!v4hn0`;;aE0J25zn^u7m5hgZb%)Q6FZQrb*qqsrG;S8yolWb;q=KAoF2VL=
zq1=gnk!i)SZoK;%lI#TMDiOM!y5<^iuAV19^&`|D6|j$e1lnzW?3e3OFb`{dJup-Q
z=R*he-X|_?_(i@3KUa-mHKWd6)wo1vbB1yRW}
zExBaLmVdt&LA-+BE(wQ!$FPZU)B
zIv`&$9yIRq!dX|~ocOTMfxNTe&n@>(tl%%3M5qkb%!hTi
zv22^3C$x`W#57I3?|-pXsY`APj33D#+dqT;n%s1!Wp*uxxMr7~t8n0uOr_|V=cPGB
z@`u?-RWF;&4mo)5{u4H7sxrNuAjKhJZ`(I5S;ry61yyH`0_+NFoBO<$O|sXx3JE5&
z$+^dAPh>zJ+~N^+P2x0%$SJ=l*W}|6yW!e%2i~yB2c<)jM-Q;c%2VXGav%O>~YerFX!yoU5~H^XC{!7nl3#P(T0Kl)4dsQon1Q>UMQ
zWFNvN4I`XWOCPY|J<8%8tHySd4eHx9chq%}z<1gAH6C=4{VB!$`VjX(?%~0GSt?y5
zxh&O?OS*|lQq&to>29)>7sm1_=_0RUE^Fm@cahUQ3EX6-F7m-WSYdKL#8uEQc@}-G
zoBXts&{l(do5VL-7}a(W3$9A%+)#+Ku%p!Y<{Iu8`4c4XEXnH@pXQ
zU-J@&MKo9n(JQ8Ok92
zC+3~5Jqq@*3v1`7r?QAhZL^kjR2O+`UaWK~u8V|d58odEe$Bei&%Cs{$j&Kic-OPR
zKl%1+)uInQWb-QHj1S|wNxZe{1F?Tyv*6sWc&|cQ7g5|V
zVEGBg>yLeot#RPLvYa#aq#o>B`WIzt+
zt28{CGmXPr}!|3H5_<*xj|l#1QtckMCcR^#uBbNdul3
z2Yb`|@%hd$zt~N=mVq53R6NKc&q+Efte@jLr+P%kCwkHt*^YRvwt
zOMvf{Q>cpObmfBvCq>*o-_H%}31(t-3R1AX%2Xy|NYDYcwl)!ku)c}RIF8z#lzr;z?Y237dYG-%}8Pz6bcNpRO1lB9$dtens?FPx&X`itEtTw(qxGwXS2%g3CpBG!{x~-M*EDiON
z=c?7|$Y&~<|RyqO5?q_)Nf(?x7jsr^oLI!024`5uNn(gXV>>G|L0<==7;|M~V1O;=jD
z>Nnyi_*}@~y6B;1-X+l;7z-
zsEt6cc@J#BmtqKN9l$bwhOa{$P{5n?nDlj3j2;{XUNo4m!h=cg+|1w!;~9KlKB&)-
zuK>jY$PZdP^7me4aQRF|54Lw=>Pzf|>z2UhU}Qk`$%dm}KtMPjG{KjDU|(|in-W5g
zi_ok9Z)WYMWrj*Wvmf
z&<(-+AIp96wu<6Yo`BgBaNha*RGuP1I4%XeU^-(T-EtW}2&za0At=a~@e!IXrJaqwO?PnXewp%k)m6`?`-h5;T2`$b9%yQ|F6O(+71Dayl8)ojxV77FH2Bn1l}Q}5uj=`fhGBe!+3{uyR)D`r8#5mF<1I$9
z@tH7hj}P~9T@UNUx))j}pF_IWR@ZkzkeAQ;LAno8}J@*i?#iI9`G;byBV?<;PDIJr{RA3UH3dKW9XrrN&Q*4uUITh=s?^ksJdpEMn0(O!_^$AqU!(aTU(uvh
zAI3I9oS>b@HCgL`E>YYs-SdXp3+lc_=6UbgzhJGXO0-KMWhwryN68yjm
z`1GvaFI)-hZpf1)r898e<}k`};YY1mqvfTosL!uhU7(NbOU+ia5SIZ3{O#U^L3e|!}RaU8U^WO9Ci{c7$b
zWrv#;aPIz5zq<$Is^9-f)9nuQ&#ad6*cw>>9o8Bg0lnd3b!3$|_!&hnne)O6{Ch52
zm#x|a`<^j2RVF*ZkA8-}`(w}2jI{XWcp8t8LQv775U@%*0u4ts4D
z=u4GVDMqj0_sAg$X7Ax%`6RFI*k%}a8)ejdApbjKmzNfBK0SS-H}@Cp_fl*)DTiHQ
z{~Fo+Vk*Q@Qz-V`unyvuUrhe-JzOJza4UMZoPcm6<(SpCfxO>x~V-TNU&zU*-nxJpkDJon7IV;C+lQsqU
zoy3EF=tF<0YzY_&fc@Q~wv&SU!9RVq!t-t($k}>VHA4dWS!US`2;cc
z9&+l<%?rird*EJY)F$~a-Nf``{PyHoJ>-Z_TCek+9&)tz^~M9s!A~q)=zZQyXh-c^
zasCFdt{Yvf`S%*e?Nn}SC5)d-VvE&-Kwf$A)bL2qf2E_AFMa`bfPxn<-+qGq;`BL@3*V*#y`&H0$D@{9p)_9hZb^f09
zw6L3uo3rsvzZ&cpE#>Fcz&Lg>Ye;v5-?@DGqcOGNJ*2Diz!WQ4u*1a^Mx6z}vWrVC
z4L`tmDRvw?Bn9mw7$aO?0Q8Wri~d}N{Es3Yw#|Wh?=6UNmIl53{ie%?i=kexo$scv
z0ea)Ui{jES|4ene?{wbyWPuzF=hj=+>!0$8IbNP@k=<$!9WUulAJz7ISU%(66
zVQ0lk?GNxBTg8mxaK1dTC3#mr>@y^jZY@WD(ff}-MBD!Xztgsj>!Sc)O3V(c0u=BU
zzvKi^Vt&#?VPQDH
zZwkd<7X$wxvngLtKMwS^(vgAkwC{ZKh=+8d(dy~~^rM!A!cu^Dd-BDh-%LPoQF8&5
zZ+vn{XA`8?EvS3-5a9Wdgcfy3*LEMj3Fqbg%a6@l4CR%iH>lWv9Nt+~4)H*rwx_ps
zCGg#RP-L77_?Bi<(MgbAa<4cF{Q@$qnspXH`hdd2D)e9D+B%%ac|BNOau%*1>pM?K
zD?obfn!)$DZkBp?fa5ld)5>)p&_4DtZ_!B=fbBMQKQL}RTX@2nK7b_=E%vxxUKqC+
zUX%kpLMeA;6~M<19C&&H=CMAzotQquFJU9DhoyYFJ8>PH_B_M)3P3f@;Lo_fF_!MS
zh<2i~Z8v4H{@tY=nW(?J51yg%EJMn#?nb?Sv;G7J)PI*hxYdp8>Yaz3ZCLI}zxmk{
z0QS_oiDJ5_`O|f%2h3~~Jca3EosnSJ1V5ne=@$lpUNHYp7|NUIaMup$4fiam-8kaN
zn+1H)_9DGS^RDV0X>@(F4U^f7e);z?Nv=Kaeo2jlUV>L;Db60
ze2dKn`4@Y@X90aJ8=4EC9-QVQy%csch%*n>!#&|xhMt}VVgY?DqQIw)>N-G-R)+K=
zwG94S8=3>?R?vz_hfO4f1!!&uy5Q(OBRP?w`yl|nZ=eNZIy?qV@jEpPehIWlOrIag
z>>H$AzIsT0UW#m?wr71^u(Ov8(P6F7Vu=(AvpK|`_MIJ-(;Nw
z+h^dP7!yqSTg_4^E_ey9rvm+bIkP`qaHfH#dmm@^S>F!)pyjnqb*KD(XH=mfzz%ep
zZ#9;0asN;p;(w!;(09FuRUX%%Uy_}b4BcED{_ZAccc7*~Dew?>N
zy5RZ3mT+_bKJO?so_p5bjq$bL^!2`IfZ)V;LEAJP8N7)A?YVQ5p1L@VDsn
zpMdbZ71Mq~FVyLGC4Bdwq))aQ
zq8^hh{zOh4)*-K>T-fNre&WRE0;F$-V0zGtbuX~Iys>MhBHr*TcnjhO;@0d!`_OXB
zWNPo2u>XJxewXR%Ta}pp{*^%v%40b7*dkm9mvz6qb^_>Lb)kt~h%fzpJqqyem!^2m
zgY{8%Z{MpN(C?S5vsk|u(tqhZShX9{WyXCxgYlIEC0CyBgzvPLSgidQ_}9!hl8WmK
z-+e>=(0gkyDdRclotZnqLIC>BrYu*~BM$xM-%kJ!x^t){2;lei_w>X;uTeS7efb{F
zKZ1_A=$e8ZzsN?oYzpjq-1vI_UdV>^yKmd8LXh`V*@{aCK`&XW9qzIL_CHA+A<1&!
zJMdxaWlOMgTHLv|3%}z)SF#y(IE1idODUMPJ%l;>GsAae=mc51J~@El}o^h!6p
z-)L(wYZ>mVX6g(+j`#@k;~YuHWvSr*C3519P72u9`s?}>z<&EZH2KKGI>={WzNG-<
z)PEo5s|os`WI)rwiJ*sFNc=JJCG3-UF*CN8LVNrz$7ch5Nk
z?W#(57cbliP&_G69sRlY55DWO1Af8RnJZC`JjiV-$paX@VuL{uz`)ObzD}S&Hy?gJ
zB^31jk?)@6HX!#+>8;|Ppl7Gs4&3sD{icz0&{EXPO)f~flma|rCNf4z{i~lzVRL4^pvi>XkYMBvez09=>k#m
z^5}oc&W@~^26hA0<28?LU|(}*?^H!^;ES4MF0upUje4YKmI3&!Mfr+skXLlAb!aN&
zKcjYX9zV3ZqN!H;d`KUf@Y6O0%Ih!oYsP-A8~jvl5H=7Ou|4Mi5Bu9X8@8@Ns
zeb~W*E$RvQm73e!1jtwMSv&G8;QY(idTK!WEDe4*2#0bSN}rZOddlu(6`ZFALUspk
z0%&r;@iC4=i7tJsLdYj=cOv&1;Kwpuzxe~dFZ=B#oL_d$S2LUoP{@+6C<*x5VrR;t
z-|p!TKkuPkJKku+2K;?BcTLhkdsLfXdLs5SIoFiaxbLv@D9}NB&+hd5t5L5`8JzGH
z_1g0EP%u=&J9;uDR%rL>8*(1Rc~dQ1(FEt!_K}0|A~URuH9P*H{+cLJZ=MZsEXOYp
z@4MQJdG{34D~}9)#`P*<@VXI}`&Wal7zFUO#5r|bNA8S#NkzSVV{G;Y)T@78w=qSy
zZ<8oIferhd$)Da}JN!A6*N^RUU3Q!;meco1?j_<+;|7OuKJ9hTI)HfJB>u-ZuYQ)h
z>w@^Ay%N(9I!3PAg5}1w|8~Rlt6isVpUlYT(9QF{G@j6T^GF~=(@8ffa>kqQmjRozWrDu
zSd_t@vSddFrW?(Ku(5z!-WW^uZ-uNs)K4T=>ky4U($!iFiWTVlA+2=%HrX_b>OFY^
zGDVo*JINCi57;-IK#dUm^m}fC0tf4x|7*H_Ur(651nE1H;;CJcEC{CSdEX~fnts}<
zgyz%!{BtexiF^pBdck3}Flv`CIR(!Y0sZS)c(w=1UF~X4^$e2}-n5*a>`gSkMCA?&
zf7we?z8-ZrjsgDA7<)>O4mP9E&4lsKjkH-x@%#}NT7JqaBZ^lKz;OiRi!f*O`26YM
z1_HQjTRQ`lqGMWe4SQrX*%DSJo^1V;v6WScsQd6W}Crr50vxr83TW;
zoI%sO*E9BS>C@2sP)@u&Q{Uqw&RDS_3
zL-iWyk?$DX_Z9;+5I{QJPC)uBWCQpSs{{J-rwqhh64Fr%M+lvqruUBGFj-MLEgtfr
z*AM2$-4^nn8ZFOubU#u*nvSO?m=2wh@&}LND{&hE`431lu;@L*r)kc7r^pa~5a@63
z8TdGm;ZwnG0rXB(+aW*8iNQ}v;}?Q|2X1Eoj%#GnH%l`3fr|{Th;B}hPl?0ehe!88
zkJ=dg-{`(>k3W+Sy^@f>dK6;xLM;FHSBAbrpTX_9jD9bm!2Hf}#%h$$U3?AX1UtSL
zK1>PmN(PwUwIPG)Cl_*10Pz3iexvpDZt|ggOLLih&Dz;aKS`*@h6R7^WTqbupIJ=v
z*$goCE02DcGrTT{<(>bI2T-uy9B8C)$>=_60cgRP9ylAwkbZ8E;cskW#?R@a%s8Fy
zvzd;&FCuYl$@yD2_CBMAhBb^QwS
zJ6{%pX%O;p9bwx6^t_vSwEm+znD=njj(MevbkDNw^j@PgahihjtN-gM^j!m0lOO3A
z@Ac-zIM|Saf98!jFz`TMN$!Il1NuyQjw3>WiHh|7#iiFwVcy|SJ|moEyN9+LZ;n5mAK@T@rib49QH%NcF1_f(bw%j07~SW!2(G4Z+ST23
z-*v(<4mO5h$MiR&_7v@wF^0IlOwH$RQJq8eCii3=&*3VOqb}Q
z-K{QAy&mOM)`@}+0rUQ{o6vz_AM8D2EwesLEvtTGI$xL)UB7-c
z+;hTpwA*FvGE-0ef8v6Fz&DIc5JJ{?-e}l
zEH4hw@nFvE@Pi9=tuo-#|3TaMu6kp
z&*mXC5)^XBICye*JOt$c{&}*)9Q7GZ`Beef?($bBOQ7GxE3Wr;yf?8YF=zzab-}xs
zDEu9b?`L$^036(J-G+8rj?Zlcl(Q^9%^vfohh^kre=3=h6AB{QSQtYN`ttc
zHTY;i{V0lmzlA9n=FtsTmZM+r`}5a!aVrrHPAH@RsP`b
zV$fqf*97oi8>?>My^)ZhIo>e<-<;lRwg=$G1%t4-j1bnE#^90H*
z8a$DW`m_ACp;su+_aA2I*bn#&&woOBPvvHR!td7JwB#4k_wX&0!h0`AzWjN@YFN-s~?|veKEk;2X(nPzAFCMT|oPKNZSWjZ1)G(J!)}$KKZTh
zkM&mTkQ-=cPx-p(C-#rD*{m$=XM_D}z9`?LE;%p=K|iw;xP8#q&;-DKrQ`Yq@l7jiim+eGN=NQPT=cFso!2+EYJzDql
zQj2`te^2SW&3Zaa*OBRAnv;-jG&dd=K=3o+rqg}F*ZwhJ*o6IeKbz{wsRK=P9em%h
zcmV0n#*LKUF)k1mdg%Wx@VpJso2Tc~_3g$^1-cHleH8=M3(lPyZqanLx6J!*M`kha
zh#UwQN7u)B|H2HAFHBE?t}~4#OQ>F+dozfxQ*XZ5({gK0!*T(9X~xgM01I~JMsQmP
z-1Ru49}ku={u3Fi{HUCzzSn7bZ5yM9oE!?J^$i&@_zDR(D(6;+4uj?K?mwjP{$>Z|UmOcRFXXq|17Q;Yiq_zRh_G+9GNk3)`SFRS
zs}xjF*mIcC>z(SrJPPvjc`*7CJDSnk74I^7yF&}82aulL#^_(hX3#89?rz)`;ddiD
z;FgQ9u4mE4o#|JU`
z|6z372i)D1$?xv~*&+YHafWXSJ~s;K+M|B+Yc4Z_gDM8Hp;u$RUq2b#8v)dtgX$~L9dH+
zGbrKzl^|GnJ*>Vd;B_=EL(w0;Zd^~iSxe-@x?fVPCV-Do>N5H3oOLA4y{&h&%8
z4u;(ZaL30C+=c+~gUAl$L{b6Z9?EF`NBFY>T^<3z>!bYkqqr*-5cb8R^3d%Rlt-%t
zh`k!~VedsaGAd7}3{$T}5ks#v#Dc&cUdZ(KB59_7%fRf3^k`ZD#4pida5Z$Jg7Q{a
zGyK%u4Dc&3+aZ4h%r=Frw-vU}5;sbwYsB8D}##G5nM9I0X31W-|P<=D@Z9@YP?FX?gYB4RoHJ68{k<5YU(Q71DV-
zKgrz=>x<~prgDa)F!!R@w@B0P+*xiz%bB&X5V|zj0TQrh!|$Vr8(-*u`koXYtYp>?
zXZI#r{yZoZ`DE8J-(_%**+*Tv-9hQOJ>%&5qEv$e8P3gfnBV;rHY?;iG5s5rr^Ib3
zO&`{f@JIS^jPYMwum5^{yNtN}zB}|@%a!C$3aCFsa_7-=wgTe{_(frT`&UBc{%ROR
z_kpKno9TO=yXGB$28HqQppLG8`xMH%5dWd!O4r?mvrAxx1HC`%6HUJ?Gzc9R&Moc>
zQM|-@=o8{eszUAvtsJJ%b^VW`D{bfNhZ+CAaj{ucp6GbSztYF*WI65^s!s>!p+0f_
ztIQ?bPmM{LXNLQuuet|UVtV*8U4rFa@2%X9`owYdhj73R=LdhKBT*j;&B*1TU)$o`
zBOo5N{gVP=oOC*>BIVJjF4)XuBs5slbQ;zp77f)*#(F
zV($-x-DWtnVP7NGya91Z$MmlVuen{8Lwebm@`;Ft_VU_Lj~i2*REm1G&6&R&Q7;$p
z;|j_EJikVxgz19$bdh@#^z=C
zdyE)5o8vz15o-}xfWZE0nhsM4_|F{dk-_r5z56nNzt`QwXSxVkR|o0)tmSFv`mvnj
zx;9|ag7K_+8XON`p6&^v^8PR#qH;zY-A?!G(|+5PV|l(#vSBE%z420N_x_w#C~U_Uo3;^3!dh~
z5)1n^J+3A~OTFD-nu6c!g6SB9x*v2vkpugXt05g9Lp8?Kez5J{h&J+H2vFUEaNLV6
zFw#M;%FPs>m}USI0sLO>*Wg8HI0wwL-hlP(65k(&axM*xG{SNZm=SNJuMj+|jPUo}
zS}zP5a3^>ZB6Hm
zs5ZubCDEMw4drwH_=?&!EV&!rB7fWmJ5l7fR9Q70;j^BheF&#pFQW5{WD;IX1b^|2
z2v`8|d}t$`SM+_esop&&b88LI)4xO&8M&>3^g+FS1Wi)(<{_!
zDNeRAdaH9VvtE#Z2%5e+2)YaKui`-Q0COa#Q2J$~dWzrG*iGehpHN2es%iCv%V*#F?PIL&U!Tch$XYmrOZaf3;ejunRQBT
z0GbVOp%5w*GxaFowg>Q4*nCjmDI3P0rQ{BiUO07cCP6c$qQT-o`jTH9|XW;T`2EwHo
zT8=CHn#k`3y$@kMj1q(rqv>dcMVxLw0KUO&Q2zvM7Jy!(`Du1YUx{R(%V9Ra$*4RR
zuz3K#JDR>3QNUqQq)_p|li_Dt9MTiG)-njtQ9}#QcxH9BV;-^`b7^Bmaqy~
zQUHDg3Zn7H(&4^Sb^0*taTnR``RHvI#IX`xEj%+?3jBh9irEWHfH+N8LS}xJhWMpR
z^H+7jJ;ozHzL>@9casH!T?!Ep58~PZ0$-@xgsb0V}NuZKR8gNJ<2pab}U{0dK;L
zAikHJ`t~V*@ZG)<;&pAfgRoqc@>u6ONp$yuMTa1s#HWtgD-=sR$@mw>R^QV*NuzH;
zvCP5mqJ$)v0eIX}WW$@kki;}yp?k*dDKJCzK6
zf?vb2ZR=Ne62Iws=7IA%iHQ0qvCFC*?BRvcY_P
z?zA!OFkgIZTHp1P)Th`tK91`o!<*$~)Y3XhSxx+x`Nts6sEBKN&nn=1n{!rbEW}HX
zkhe*^(@CcMSUjY7s*@c1yDmcHVF#J=T6k0jJBVU;ovu!OJK^eY
z&ntkq>stMuwhC{4l3rhxX-_#nNn+5QeUqoR6L0cl#!snsa%St%U+hKg5C^y4)w_t`JrWwS>k0s$*bdV!!GhC+1C^zwIkstsalaJ5t9Cs#I639D3tV*
zgbTZVUJY^Ld*5s;3fY*JPHo!uRIUm-7{<7s*O;O@@Av`
zmhI%M{_b~4rtM_+1&;s^yLRw5UbolwW;+?t|F_)*=#G7}FK2@sQI3;LzKZgZ7+(6#2!6o+27u(5})q&mOG4O7YLe!1u
z?sl@4yJd@rLzC%)25W^}K`pUA>;Th(KJwJsTF1eVruteO2MPN1a4{T>d$AkTdBd
zpKf#s^k;69TEpB<(pgkL<)&XJ#G&gu;r$%S6&F=5hW9uvBP)}C)pU>^5$A_JQ#;9H
zQ;Vb)i%xQ2=Jij75Vs~~e9!vFMjd3LR)oXAx(>qmkoh!kHcfPP9*T%K@l(H6-QX)^Wl*1a!b0%$tF9~Z}Ypz
zU;8yG26a$=0e@>7+-LuDX>sT0Ru-9RamYPoVi#E&SRI)EjRo%o-E*&nI7I&E^*WZg
zL-`X91Y|)xCEMGE9A#*SWY@x4E0EJRC0JFBkD|E~HjvSY)!)PG~T$+IG9`;!nCz#;GY
z^f5+VB+2w1x7QcqNc*47DuFmiLj|Tb0$hk|x#2|JXNdE($hozpT&jy4KG?FU+Omt}
zHsslL+=BK{A8$P)w~K7EZe4AV%Obl>uemx_u!yfmL&^w$7g3ob_f*+RKyQ*j(F5R5!uBeSL|PDa-OnCOt#b5OVEDl=K_{>sKfpK#5qoX_`6Boscm=F
zjk}4G_?F`tc@PiUFxcTC#JLvv_0oCg1&H@0BRiJY-9-ld@=SD@I%_+~>MqANp<$oG<38#~6WMVw(NUfe}UU
zr&Y07yiXY3$1USm8?o&nCej8MpWn0A6_2Hgb!u;a*$X%r8F&wt<}~&NTP^b9h&Gap9T06G4wX^fkYau!*bMw!M4dUSDYcZ)uqY
z;J2{7esQKd?E9zP91MYSA9y^Qy%Y4?O!*tK3$;PcmK~>*FN5Ek_*UVu5I4oTu`pH1
z4ekl6otT<&3;eks1-~l3)k8`w^Nf<(pnrJ12o4DV|GU$FepJ9cw}r2Bbf;+nf5$u{
zNhdbZT$R#M1@Sndm9H<<^kowj^NHi0-eHsKOJ8cjUqf8Bql@y=P1tb0d_-^GHa4lW
zP0Qgs!G>|qb$@*s^zMwUny*sW#Nv26zwlT#=nH+TXGyb(Lgm`N6m2%i^As?CyOm8|
zYT0G}f%2DJQBiQe3wDe2{X_fH!GBcF-g6r0>6=y8tnO1`lRrzY-acju@9?bJpZabY
zoNHWrs@ZG}=fS2A-+0`B^OkSp&rOAP`SfyA>XuqKj|?#0zc`moY|Ff!ukM8VnMcCs
zPF3WPpA&^2*h730y?Ax?t2!Jqu=3;MxO6rN?s}NHxRp)b4pzM0HjhI#`_+ZKRE2l8
zR}WTt@8poLTc++Dgndh+QiNn3@c;As=QL%IO^%qvkElXG34)IFOJ6asTA!jVw
zzsHtB96w9DM91|U5|VN6_w5@TGJA22&DLNJc`KL}sQ~eVwUjn`M)z>Y4WpjEv6dVX
zV1BF7#Fax{eLQ8HaUSC7opn0od51%E5*>~HKs>;~tHQC~^C4eap5x(_9CB~jCjS{n
zVEa>5sAr-yWwR7IE$u!9fsTzO=EpKy||78M1t4%zMVICJD#qwq85dV12@u#OZoPhW`>aov$dqew+X1Nw~IAl%Y(UTt2
zxy06Gj9}$Fc=urYFU2w2xn!M>V4c?~E-?rf4|((h;vQOfMx!CP#(;2*p?JPhI+Zc>wS+yV4LXY;GkE?i>N-v8>42N&$mUkh2Mxn$hL
z{XcI(es*rl#$)PSqI6CCQ{EOXi5ausb6h-^%=9|J5-NiDL5e?4*~9N?Qem(m9^};z
zef;TeI)~ix4s2g3!6gnG1fp%ExWvkA=Q|U3F4=QKa_7NIT*ANc#h0}ZpYnvS+sxW>
zpxfR*x_$$XI9pzN6Ag0iuka>eZd{VBbK$I7D3|;iy1H3186c6d+0f1=u
zA#tyRul5=9h`Y({vP}?I`(=gcL&Nbrvfc5l@pDxkam`rfGAzm?BPSBoPBd`I_QK@Z
zW1&7#c&#gJ0*~yS7jyDAyvG%CFSh@-FpnIbeZ)H-??bYxBrSh4jj6dhq^g
z=>3`VU_6=6Yrpmp{BS0gNAzT!r3*Sxj*5dq
zpLnFYTq#iy_-vo!}eQNtbiHJrmw*(=V{l7M;N(d=aLDUeIsmx_+@!f%nxe8W(lNgm8(k^EB1~U+JhI4k0%4*
zc;t7~zSJj+c;v>U@`SoUc%Nm_XT1gR-9u3t`(D52k?UIJN&P-Nvie;2-ctBIPoJ#F
z)VjhYa&e2Z5^}i2-NCVU0*sqOllNJc{^OFtF-Fr)LOiis+jIn{4nchQ*^jb>cwFMC
z738(+DwmX9c>j$QLcFfd=+*-;PxO+@eM7T(gd=|0FGB#{uUiy0k9P>>n|rTU>i*^s
z`%j%akCR=aMOm4xgj~`A?(6Y+>gJ6`OpHlhmGU)#e$3Qx!Opzsg)I<+d5pVKuCmqFV9{{3$Y{BH~VZwvfy3;aK~fX-6c|36%1l>VQ*oBjv>
zUwNvdbPC{mghug|{{#Lf=j>5B|0t9hg=^@kW?yg
z1xX&I*N@Ww!@pBOD@giidedmS`~TpdF^Z?q53C^annbimr(i`
z(Onb{x-#X|WiouTW;1-7S{Xd@Aok=6@>D>VLe0fYdedn@f&P05{ty);rIO*_GG`ve
z;|3TxUA!2+sq3);{x7bsJRIxhdnY84B`QLTEmA3kWSNLWvQr{k5{gnPC0j~VWNDL1
zi;y)-p_F}1mgljh2q~$wXtnk`XCCi${jTr(&;6d6Gc#wlbMBe>JYo`wCR~Y3_)Zx+
z$|MzHtbEJ1WmF#hf-9IwrfdJCC^cs;RLmqYvtNRtOcGne5Kk=>z{SP09GFK=VvYqER($5QrU+nyY%}yM7z#%T2H(igO=-dbEJCTXKsdx
zgK@Wd0RzP(p7#DQa7=RSus_S56jvvd&9)E0N&~;J*H94gv;G8hlt~1JnzRx5JUHGd
z0Wy{9ixI=vd2eI7%L4)v&Lr$g8ZDS!cwPY}f=SkoKDAQ;^nKZzzZUQ=GpuJd;FWW&
z-IzaWSaBTlKgS9ETMlTRuULxdgW4irv7Rj(vzMcs6Y%g4;_D7eYR^u+XmiZxwtV>k
z@!>l*n!i@1ijH5ON4`0>vuT~@dn~swqJZXaUK=Wp`g5%>iy$txjd-gJxVu{?3NcZC
zc-|^N|K(vdSnj6ZJzjM{_DX)3I3|%-zu+bg=$&mCr3jcivYep{D7Is!*$TjQ8n%lC
z0F~0u@4@o#>$!emJ32B)1`zMh^9Gm6Bn2~kC|=0kI*N8$8$4~qe7B~#n=rrm!zS8~
zksbfGq5RzEcs-7vdoV);am=OhET%geX&EEFeI+!ES#q2d(9(K3mUis(<<|%MUqsU!d#ouAmj|->ef(bRQk?e?iOHygot4
zW4CrQge+LsDK}gYpUDK%d0*hIDTi|B=@f8;;18@;f5v%omCHdi5-Q3j>ya
zHo16_`k$s~X=-n0lO5HU9D;2J_1p`@2oL@i{+T|+c&p{qZ*?^8p!#)2&^<`+o-n1j
zL+dxSbB=NZ)j#?EJ@tE=D<05x3Y#0KUhXMPiUDr`p!|0N)d7mVX32L0SaR%s417$Y
zm%x(agjo5nVRd1;<0vb=X9vqp{&80PxVngyKLWFcdfLq_Oj`rf3G$48up9veD!@EI
zk43C{mc@XXAS*FVsNDyc`=~r^F>CyI5->1Bd_Ke)#|#WFAfJQTL{y%N+i6%0$eGw2
z;A+_2h)Fsu8JswkRi^CI!!QZ?p^sQ7&&5J@oF+(rhTQ^enc^2t*(~|c2n%arS75%r
zB}-;=n8FR%5|gx+v1Grg{PJ9u{@-C%`g9xy(7T1nfKCX2I7)~!rmz%d6=m)z8Jr50
z3#a6yDfuE!2k32?vir(Ze$-U{)~WVRO^r|1RKMJ(#;if@*g;O$}4$#k>>Sq)+4CZ^P{;gB?+?&$>ul~(b@_*}vszW_H{->S)
z^#5!3Kl?eS^wg&CzjQjxP|ttzfAh0tN{=o_$fwf-2+kk=1O7GD&(~A-TK$iH&Q$v7
z)Vz63BPdi
zoyXFfKz9iGGpEXrA$8qHEWJ=$7P?KX>q}Gn
zAd^lItn>G1FvPp5eQ2x8(!Yjo1Jd75&3F7gmR{{uR{C-9mY8omwXXT5`jy;?0|D!7
zHY@+%IIG-IAC|n8E-WVb%?aTc^wi$4{B>L@tDKuDOTTj_R>&l~L~#I_AOyR37;n0o{*H_PH8
zt#4&tI243+HyAb_u~VIo^6%zjZZLqriyxR%ySP7fo!7j}AWP3S+vS?vr2g;@&EWse)i(6kiqik?(
zf|j5BmF0$V!#xpd_vo{L_p%`8aW&6H{$*dVN($>~KVij=vSoak4DvCF&#uDA@3&k(
zT8%tsoxj~DtmlcUz(2%f(V4=OPv2vNq27!gp;G<
zd$XxN;H{qIdB|Un3v7c>!6ZWW7vI+ae6!BY5%qT~+JEg+ViJoPH6JDb$KTckY=!i{
z-^N<7o``{s<_t(TkY8|lCdihnxV9|jrFRkMse;3
z+*>%QssO01_}DfL?6_3$lzThKJDvnH!vR;gyb8Yo={sY5UKIk0^6k2R1@j+0J*K!8
z%Ki9v5%Y_y$85-ICfUk)cZLz9SEZ{8%7ESs(Jh@T0ZS$ahLH!a3A=a%=apB?Ocl>H
z6*q1Tt_N(6s&ml-3_Zplfc>fDI`JIsWRGfChW!qCW%&)q^?8iy3u(YxBKqI4pK-rM
zI#BP7P@^H{N1Ts{ECqX1^#rwWK0fcwZ*l>gY5OD<+pn55iNJA^`^??P2`KTcBD)#T
zXUAf#17LUO!RM*HP`*8J#*Ut+uJ^itO^g%K*xo<&BR$yO(QHW{f3S1osZ4dOfB4!R
zFC$Ff`L99;kng+6txbR#MJ{l&0_S#pX>f3GzD%nqN4z~~PR}zT@l5JR)_Iffn4Xi@
zeg^H>+^Y%H3&!)Y4L$D-d+34ZhxO#gy9V3q7;HFiu7qO0)6d3&BZhwD9r8eZORcP6w0D_Oe*n&Jh1&08TsIrGc>1UT
zW-NF&iF&K=k5Iodk`hAw)0ubH(tXVZGWiSD6TD#nU;
zYVX0lgWjK3WK3{wosYyF?kg);`E{X?3^KnIE8iZ~L6+8Jqu3E4K~d`sGAkxI_!$E{
z#JO)sL}2|XPf|N|o*bj{srRh->Z0RFWxY$^XnQ(}Uev$;{@OvuBmX8050sBQ^qyiw
zlNrSyQ}+Hq^C15=Rs$#qs{qkCh9%oFp*oNa#38%@_8frW1=Oo%={=?d6y#ncG%)^A
ztp2%EA_x8%rw`I=ZCJQGwv5_&P7j73(!&N>aeu|;WmGP(LShEzSg!!8*T!Vs!?oYH
zpy``DU>YDlI#-n9(_R3OwP98fcSx~t#s(J7NAn>)mL3eik9)J^D^vXzn#w-{(GT@B
zDM1B(i^~z@MoU(@^pt)UE<;E+2JeSh;J~u?krhi{D4CUCXV1dZiLi`8uiJkeMcV_c
zbT09!`d6{abHVOFy)GPPKqVY5z(`aBgkVdt8lp1FtyAlV(n63SDAM#LQ~F-8dQopT
zP6yx|1V9{B#MRg=;C0+4fW1@wGMhS2g0rIO3xIuKz9SFIPVp(b%#r3m`tdC+EQ4K(
zGGnTK+I^5I?FIbrJn`TCJ*^(nX>mYdEDpGR3g6-Ig4{DDS4`DgFeL+Nf&YM);P3*D
zPt6aqF_iPB#wT$K7uvAW=`=yU3T`XFpSYX=v$wLaBa?-GvMh|9;_I$c{Qcp6R(fF&
z3+qi-$cx(*>N!4DPh&qC4ExCl4S@Na!^+=44@4jfvg8jZSt$O3wV$7uvShBQ`T{nw
zyc@->&S^!caG6X)-+
z)}`ju{l}R&R=v?1SbFLHta(>EYEA30bz#Z-U*f`Gl5N{q{waA0ZX`J0FKvT@a6f;7
zkIG-8^e7I8E~XfvrbY3vWC9q(BnqMXsQm01Ec}!);Ux6?YOp@T&0Rsd)Z@w$d&touY
zevpe6m9s)UpaI}hvlVm^cL%l6`b{4<((=Q5*=hbK&njBZ=FggGsCT%ofa)>V9DsoV
zUee1=`Fnrr9s!j1y4F$t9KAY43}qR)Semc6aTyF8-0udCQ64|a|ML{eR_cFY;D9&E
zoS=3kRHV@Mll|K%?lW+-#e9LAPBf1BrWC@$03NHty62AUSw-u+`)BkN#Zkhoc0$1(q$mmqjh=#R+=H}d;mB8iko`K2AC_tbh@H1D9E*-Woz$hT}A
z$~%xpJrI;nK-qhr(j096L@5)tD%^{e-}6M?!L(blTMzJ5|3BEe;J;SM&qIA(3)S7I
zzi$6kb(D|fXUusAyvobJln48Bl1ml{2KWh?!FtT!t6m2y81$E|+7JbJDujCr^7k`@
zOCI@s%$JYXk-t=Wy>|Qt^$6Jh{Mds1IL0|1E@Lzu}MaHRu-!keF
z_1y&i|Giyi5!92f^o5-XcvemAawF*H{5(H?8|oWY*WiqV^y%^`_xixjtlT
zcW+idhj=giV!Z;`wW!pyB_!}7e9n{Bb%};V!&PwPPyP1NdLK>
zYv(G!n$K~9_rT6yx2$b(eiqN0G5Q+plGI6^_ZVb7xqH(G0M8eOp1!-1Ngki;9pr>^
zTVfY}V2ApoWD9=lg4~ibea{)d7?E({pX-p<|C(L^{jA#))U6BX;?iFp5A7^Sw|ao*
zCkeentv8-uwA`OQ?gq@Z5&wnjJ;uCoC$8gS
zTNBllkbX`qe$79S-Cr0;RRXR)VO;(d($)66?adXKD=@?qUvoD39-hVj`a(sptY%%@DVzNZA}jn3+mJ^?sz{kJUc
zi=?V&67qmG*GE-we`Lt^IpF!`$%jfu++RU1x<9agnwr+_1n|35=~X=EbU!|~3HOW1
zDKmcThobn7C%?enwuw{o@qBp6`C|Z{A4VAx&REa31tp$Xze+97%@cqzw~KT>g8q|1
zg*HKun{FG$;<;$Xw+eGSU(EacNk9tbfoI8i?`2@`*2V44C7}P~*|tryAwOu_OCbxu
zs2tB)tbdKw!UDv8bw^+@Oj6EKMg1(-nhXeJ;BW5aEI>Qw&$SppS@zB~A;bmV0o0#<
zsSuDydCYZJ1KL@5^hqYxms78L0MpBEA(Mmnt^vn6Kr5dEu~^?=;yHEH8!InLM|2eL
zpz-3LM+q=iCRwSRm1YO1t~zrKo-$Ko*G4qV3nDzTALv)vGPE
zU$-k~(0*n*eWvxL@ol5=Oh5gVFY1*<6T%;qYvYGu06bS8PRB!i)Sb$CGTu=Aq+n+%-??g_hWa}W
z!1Mveoq%ou`VN&)d>uPR+zMjU`yqv@J3%pgY?&7*kB0)`o630zKqB9+h_Ade7BKXAc0D%*%|q_|p-#peur
zp;^%5o#K~b!~N7wW(2Fdk-(+c`;iY%yJ)b}40eQt%)bgu+airgja
z6t@Sm>{P^K9F$AP<0xP#ngO^3y9p?1$HI--EL4V7faywT24J`W3uFCQ^SEdZt6#|>
ztn_DsESxb_zt~PBxDYpbSp6`r1v5eZCBm{NAL%5>_8zSJR!4fUgZoU_O_=V8%NbBd
zi8Y>+FuN$rVlto$0$|kCd7^M9Oa6+(1L>DxRw+*9|HS14>FMWL_1+O;$puJTA$=82
zAE3P?3ujKv?}!ddp4h^|ceqSJZ^(kxk8dBR`Xk4}4y2!uU%H(ozpP}9SB^YuUM^4Z
zfAgt#O8#Y5{81s|Gt7Gr5@;4NN`+#x@1?&_bO|bIKg0TVMUk2MLrdUlv0VbKfF`3Fj
zS$Hr2{&fuKEZ~ysiz(Lq#DRo$9Bfaq-!X}nE9bgJWvkD_RA#q4LFE}Sc%p>(BxghS
zt3lvzx{k*`7t#EjyZ33n+9{k^@B=rNY9e}lcmx9r_usnOG<{vvX{z_7BY^JzpKI!1
zz+ry$YoUV>FMjQ#Je)hg3kC#uuZA4VFx=m!41$47GHnz;RDynJ7}5I9rgFf71%GGU
z1tSY_=F3wU7>M_0{HWg5V?ET5&btvy>lMl$qW192{?dwi?}C740~YAwK??4jLi(v}
zy(7y3Wto=YE!eKk?Nm0Dt26^?|5LPl!STVm=ZliY{GqdtLJ?iI?0~BR*sqp>A22_(
zc&`S^-Sg8;ksp8bbe@Jh>4*HZ=P3VJ=Uk5Y3bIDvD1oOL?YoJxV$=4O7}qRJTGpc8
zvj@AcVY+q4y$ICfl}@w3^uIMLcVPThS*0a_{Jp>1{UWCCsBxq5wQv66*O=ZAKJ6~@
z-lma)Yj|IrXgGf^>YY<6d4TPBH_Qyg`%$0I#=e$-nvwqHct4-;@60+Wz*DLd`r3f(
zomSVe9-(NdFIyqLTm6)th4nvs%ydT{V8H%+&N4vr;n)VO@7odmfwzESv3(!$p8S5n
zc5}4d%A-9~QZ!g-LcUw6#q7~mCyoAP{sZi?^Q&^~1|;XVc6Uu%5$j!!`P(#emL
z+X2l#ianhM>1X}Z`vcYCJ#ASn-x%?%`>g7THmB7`HXt=my@pa8+w(3GnUYm($wZfVXc>c2bdndduxf
zo*e}}PRS816Djb6bHfcEEQWsEo@w?D+R@p`y`R82!e%A+%}_{p3z)530X&^wL{Ka7
zHTVmog6(3OP`}UVkt5PDZc7CA&Vc%sghl)jPJ;X8B3HHpGH{PKYmun80m$vQW*f;t
zJSf*ppDqUbVpROwv~5gsRXN1v@(x%BnX)hEE`WCqDbwFckT2$UZuaf1z-t!IQy4dCd0u0ro2}-e%m<*9sN_&+C@mYy{_=Z5uEB;ev7XuyVfo
z2gW(czr^dQJ)9T!+HC%A4d)u;)vc0R;73|7@adKVAGkjh&VHzc7;cfftaiPH#1-ZC
z-YJH6i>zWZYn+E`-dX?N!g=u1De)`M+emNTZi{_!ZRCVPceURJ;EA@&IZeYD!B)(<&YM{+MO&*cYs$k}A&FxY?gmWa@Cm|vZV88ua~?$$fU&0P-XKBIlJ{iefu
zs?i<)?qvk$7g4Wvy&eWhZOU8A_m)8__o;LiLHTJjq;_zFUFpe@&wXz*NTy4_-^C*g
z5)~6FyHlM(#13!3#W1*e{A2#TohxNv>Vws}^pxEt2e_z48EZx)W#0z%4
z-S)fO3~<1&V-W8f_+Mw8nS<+7%>DQ^kau5K8NqWvnM-AWB;a!){^!S_Uq>shKM93-
z(_Nv;iTpgc&c-kh@YGPEKl+`m%H~BpurB}Io~9rV{k!RJGXDjXliqX4{1W8b4Mhm#
z0nT5v{nn@>+*fqI{$dCEPFIHvK0&>F&a;mPgPo6?&w4t-x@fRmy&U=do8#QiGyuni|8P+-wgGp?ejP#
z2lz|w_1HQ<-StvR=nvr^@uBwaGFe0O!^)LoP(CmY84YZ;cy-k)v1oki^lEjbKK6ym9H}?41C&;_jgIR!m>_|kxU&!!Np2KH*jHd6*
zZl~$$M~+gNYZdE!a_EfGa@7Ab{FL&M^UpkK{?b=)eF*7YZY5N{_645%KsNe02qy#J
zBVY3Ac>OWBPW4u=^r8JW3K#(93VbH{1RbyFc6jav+1gb^9r5$LIy&x}#(cEi12M3C
zAl+siG!JQGO|_AMV(FkpE8I%eyaR!
z&F^B3&nZ_4n%-_FPW77n#HoDG%Zy@|F$cxAuE!MDCVZm&HpCteJ`m@ov-+7nn{_@&
zxKTjU4|DR;`kZgG&JVovzEk;#U@6T%beKWcg+jz~x-Q<|6NL`L`;Ck)R0#g4P=)F@
zCv2qY371dMee(2&4Nbpr$CK)9h}%N*HO+3*b$)Q>%-vZ2*M~PW-~Yn|<3B;FYGEZv%+61U3%vzxsG;*DE^{nr^;zgD0jxihD-)@nGsL7+CPb
z&wo?@;HMgU1?AHDf;s4iI{zJ``k_3Fg)n{HUU6f@Py9(~i2TD!-sp#F9U^FZ31-iz
zKMc675`_5~&$m*)={YcMCi<2j<772B=XvL3S?shtQypuE_8EFVAts
z{N(<1)NeZXl|DngJSW4K80Rt$=lNj%l3^WczXkIEy+==&ZUc@C`crdf#Zo}2IUDEv
z0Y7?or9c+8U-QQPCAO2YQQr&8-?-xaX%3*}*|@4RfLnyZT<|=g8nD$0;~PiD7fTsH
zLDSk~=@1in8lXpj@RBH_i?1o;$-mFcai+
z!(VOCE)$D;X?QP||7K;xC&*v!)iv+{@Yum!i2{%=CGYhFi^2=sfb|&jVuOi9!(fXBSTjL;kBlpQHapv3Ee!%CQ
z_UmhbkIigX@BRX^&?yBE)QdN5U)Bx$;NWoE-EiRd34RH?g-wAko9+*+gM0Ec;{_7t
z%b*?Kh;L$WK6q{E^STfCf0pWz$8S;@ByK+NDS0%t7|1+#Vwj1sjQwGof
zyvrb4TC7i1YBm!^H8V~E{P)9KUEPboSL;&s**gru&-C76TRV?Iv=;n4*9H80$k2n`
z9@3LvZBX$7UcPpx{Oq@&Uy)(=V0t*{>usJkvxz~h<^AH-!{EF)>ulQ4AcLIC@tJ9t
z2Hy$1$=YuKygI)|cg+&7=fog=-{YNo8APzLNHp9HzFUo|xchGj_(NkhU$vBKvhkW+
z>eXdsWZ#=GXJ6fBGGJ^PU(MM@hL_(pk)2jYOuMW*dy{gBdEM*^<2WAn@GI1l!LNT1DULf=sWPC
ziu5+mowedzC3&*mPV{4X4S7+ZJYeqt=dj%~K70fI>-o4lJr34O-k?|30$4}qgUi}F
zfT!IoN>+}a4!mr}GC1?YcaqmTV%uPyb1H2p)ra-7+3NBATv%V1g@>mdQfek4*HtSI
zw=syuD~U&LK5$Q{o5OPy=Kr$Ul8e`2-b;4PyYA0;PLv9Q=1C1Sk;UBk%Ox*DyBj#)
zP2U0gB7UrU-49rQS6-bTW~(9Hi@R+~+#1MU_Rj&+^x%CWYP)4d9{7p9;fq(zhk85Q
z!`4CjLG0V)8^G>US@8q7Z(IAjIpSgeO@4a!9Q_CLWSodFtn>V;^LqTCAMs{x%rDqi
zA?ttXU4{IjOPt+Iz&6hLpT)q=1h$03v!I{$Z`SC)hH=nq6cKwu9?yz2pN!!|1usd|ci8}N%fse#CX#<{*P+E)rdrP!x4*IDCotVdz7nwP$0Bt`F7S#i~B*t(U12$WUyp;yLcJMYZ7Kn>N`iHT8wc~_`6Yx`Bo7_+@f9Ids
zh>ja`;Q|rj`ah{e)LS>-bp@U`{w&&m6VrE8{}V$T%heym@;i^04C6i0q{m-2lvjV%
z|AFXaIv$34`z5>Jhkww&qb#2NeMdFrEh$YEln*_&!4F5^J*5R`0pKkG1oGn_
zkA!+`AQC_at8+&Jr^?M1$1!TFDCHwEx=%lcZP`O
zqV|*rW>4_dpuA=i(;D|v;_gnEK={3hP?r|s9NT$xeas(&^skqbxgVpSNWbd`
zFKsYC@Bbts&;6Np+4C^?xjz|$N6}9lG8#xx2Q<|VYDGPp3lG~Z043`e3>^Ty!ymkh
z=Yo9V-?1R%pZ|It>3xZ4$)NoZw5Jd^tr-2(-{Bz>|IJaDo&ny%?^%(RPx`fU=BnbB=g+`8SP>;8`##b$fctDTK0Oo)>m
zh3?BD;oeH`;aRSD&^s|E%C!#oVwaYnu!TAB&KG^Q!O)+8ZnINCGLYVOy2lRUx!XHc
z?Mbj-@wZ+*<4ZF!9loDf4ZKfB%P>TK3&fi`PF)|UUx}-7tKC)xIsG>%S`qFie-z6a
z)xiC&&BI04yF8&^HfqB{z@yg5RtR2%c-z(`~OY8r@{o?M28;_;rkY$7QFNW9NBf52;zW;P-
zC6eqGn?I-EvLj+tPWj
zN3Z9Qz4Jc0iM@J4d_?`WDo=kxu1nq4{eH8O9Q+cHmNWQ({Hx;XyykU}Y;fF`oImXu
z{QbD%@ifLGaw6@QKU4ZV8S2^}YkfV1_`b95Ty2z13L?!-^1XA&`GevD;X8}SqbEn7
z_NF`_FDlQ?Gj>iQ_2WOz_@B%mlD)t0#I@#-0cEwphj9;xS3&%f>E6x6FK&a#uv!BV
zU>rPD@ve%TDRnDXxl~CcP8#msA6rBA*BE?s)d2ok@GySYi|3>~-8mXOSo~t&!oX9&_
zwis$QkzYR!IjH3^NJ+cL{En{-5*qlY`}in>WEl6A2*Liz<<*!F1U@-EFm+uY@Jaje
zjT^OKe;T(R;EWz;5Yxk{FRsDNTN_z647CYk%dm$SYnsm+IXgL-cD)HK*1tCF0eoM1oY1Rq&38~ldy)h~Q9LhxO~
z`?CQhu)lXM5j5T%$sm`ER(054f_H(Ezjgyo;Ga5Pw?Cf=anj&DuL0O+^T;yB8vKAy
zY-Agr!`CQC+_(k)CN4iq!5-wyEd9ljYvA1byt}y{-ZwO=FZo^u`+fHa*X{?sdmlN&
zx_B$v<*j0Q&bYNL+A-bhhWk)AfO;
zf5#uF){k3B3NKbE)@os7L$72YWl{S6lyc-YPiXbc;Rxup0ENFD`QyfqF7N
zj@YPzzRKSq`v#D!q^wW)K)tI{XA64(E-jA|L7pmdu<|v=slyh(cVfJdkE)r2{{P`H
zzbW$5hVk9G7;hDG&$nRwVQBg3VSIWfZ1vU%@XhLb)#IRVYtmJUeA;WR^eMcTSSO~>
zxg79|$t)*1ILACX<5`RM=Dp&l8()L|*;!q3yFebD>~zNSz~WYDS6S9-X|VVC
z${UsVz25nCZ0+Wlf8iW_LJs$Z*F*PU`zMY1#W4QzAA2y4I9eR#kNx1UvuQ;+{e=Br
zjJF?u2yDjmH(e)y@BlC6$lrl-Brp8^SHRiP?(O(avekRZXS84Bl0r4cb3R3dlbD}p
z$^H;A#Pq;E?0@cofSbtIB?FGz;kXDU2Nhy{It)!WjNh-s4oF)9nnx-4W4%i2ZFI5x
zETh+7P~SzW?<=oUlnv;Gg|SEnwp^kP?+&BFBGW%7(|(2b1<`RyUUY|+lNTzZ_RJzp5vW(~yp6V7WaUNm
zR*Qe567FdOn*wb}Jb7OUFE>sVJbxp|=^xCnfrBa$x&TSFMxq{1#?8D-iV(
zrkQLI0DOCD;Mg3%U%cTfmjOE7T~mzhWjOCTfcwhgsl;V=!1rrKfXs=#{`=VY>
zWfq=GWlc{=;Qsg9F71Ro)5yzCdI#+5_kmYEa9^x0>CQpksr&QuMsYyniA*W1XGdRN
zjW+l-MSh?8i$Q*IMR{iq;BuSY5aG&p#FH?V(BQzuPfeFvk>%`g!df24RPbky1y%;7lHp@+?)3e`Xf27al;Dm
z-_ehB9onJX+6`d=T%ccNbxEC!Fi6?SbM-R73&oNJKIy`{vY`3wfK2!(cGyB2=e#<7#LTP_9N!7}*9)b9>
zhZ*sEJ-ovgAFtq<4sq*s)5hgHa~Nd9T*2QN+nUL*3jd!Q;XdrM0cRhM?-Qk4_G>NS
z-glFAPW4XUm+Vqf>|n=Qksr&VI^e!CAG5j
zHZqIx^ViV(W)eR0N<%_4oTE3T)(XLWYx{w7TWe!-$om_Y%G}3d$WlA2pX?8t3DbJj
zb;Z~ga`LHER>JCLa_qFu@2785h-rI^R3CpbQ4IfmG0UTwbbY^AWuwwY1PTQWdF^Y6
zf?Sg(Lpz$Ja}G77T`D2U10T|@8d`{+|GDMI`CEzY!7n9mxQfVXy&j1RsZnG;+kh@#
zMltD7RB^Rf+(}Y@7dIAdZ6;gd&!610E|**lzBc`aVG3CmK8JV3=X>P$uit*36Vr$}
zTYq@jl6124dSTCX_Hy|5+ovybTh~NFhVuUW%VrR^vR
zPzkYYEg3fCXdt5e{rQpY&E%d&tK}`u3}U%`Yhw{l3Q7HV(d(sJ0TIkDmJN)oA+CAM
z^!dV#WU=CP)6B+9lC?(CKKNBCY1ABWj$4~eQbek~M@`E}N>%6Rx&9hra+*YESZ9&i
zfQ!;xo*AU@d5L`BnoJ_-(6B0G@iQX2C-(DMQb{a7xM@#ZV37V}OV_WC1OK6`ZL9+8
z;K=m!8BtttZ&=%GBnjV(Y^b=NvEzC(S*tZ+e;C%&*CZbPvS7F$Xggz=1bTik7T3=#
zfbYw@ytb*RG?3Z5**{)f0r%TQe~ul_gzpx3-j%h&Iy=68G^P{o1q~(5-&BlO6SJjZ
z%bmYA60b(yyA1*Gy^mkbnR$DGfB03*wgTSAb2H`oyu1q1Q8ubr+G*$c8DVi
z9Q(5hA-*=}lx8gizEzeXR0#ae&e?YVd)TKpWP5i#h4uQ^etxJ0@RwJgtum*Wy(B{NX48mT%!EH5^8;RWd
z`kMxnzc!866582UW%wSy7n&u#N7?T=1K!D>SjWJ6Z%_N?vlRBD{qYrvOW=I@d2^Fk
z4tzf}$v%?LF~K0>j@~=QfXCKP=qDyX`!eAhF7)EPXcL!OA@Ibz{(R~hU{C$}V`~?H
z9ZrjmEqnp(ytn#z*BbgGJDcscEns`tv7WzRmk=>*4Ff$JW}RCl=oJTtX&6BMYLV5I
zhX6nEE&Q?t^5^o*IuZ!_SL4{dn?cXZ;YBn0(@pg!7u*L-zbL+OKJ?44TgP7j^wwnv
z>`I4pjYzf?r$PT<++wv9kgxW$1-ylOQTdUSAKl=u)DLbi&cpHQuQkSZ8hLg(`umKk
zmyeH_g8tqGML8Hx9vh9)@3V`ldEr4Dc-(`%^mo3}P28DYAm=wZN@09DDIBecafb-X
zMdJI^o0KiV$a4y0JLU1+DcWe}FU-%jiWfrO>HIqD1@cXw$wDAJa6iAaFZ>7KwcN}r
z+X4T`++KkGzwFf)X-z=Jr=;)LzE5ZP9bdq9uia-404{nH=Y;KZ{S#e=aaA%SPyqGC
zB)4&4y`OI+4Pkt2>of|+_%%EylJXY=cPGj>oICi-S3vr``+g@-?|XzvGU`3x*+aiW
zIAb)Q-t+k@ijCp@&%yrq9T+dubt;EY@5Ouf4lFnR*_85tReEgM*v~qP5i`UKf88rk
z@8wtcxjUd_?au<#|GOvqEZ*ZxmotMeo`ENxT0*~PIQq1e@)mO+Uiuxqj3w*uB=mHy
zqWtBP*e=R*zS(5Zd(C!bW%|8h%R3l;s3&5kG1aSd?V&Qi^*1WNJk;oi`Lje4>3v~f
zAnW&7vXhf?nBH{l(+0#C@$6y5*l_{MqvAQ7^iWp(_LTN7Ft9rXWg&TI299@+hp0Zv
z`+sbq_j>VrdxTJaslwNX7(KN1F`}y74=cpSSGUl6VR1QqI?i|MYb9|!9vxigguGzq
zU_V@df*;t_O2;Gd)n&?~c(T?yq24dw_YZOXXl_jIM7`K2VIi3A>X0>#vgDtIOA*U2
zpTCav-SQK;fb#C1MEJlJ;{F%IeYierBVWRoyWp=354z#~)~o4D=>o=z07u{AuQ*5XZGj@zz04nPhMJV=X{(%@4#|j
zzH+En+Rr1@6a2UI{OafGP%pC~KD^!+&gpJ?3C_TWEN3Zw5`z4pk3Of|pj^7|g}mgQ
z;MXjQM5CddNA^1n+o3-zgk5BN@El%p{%SSEhks_CizC3Fr+Xf{{cb&jtm%ID)(F0*
zThZ&LR|foLyW#z7+I&#|w$Aju@eJY;yD{#(ej8a*`NVcJoRdFGCvp4igLCnVfP*W$
z;a#)gt=DruIR9I0zxTASgP3RiP&)<8@=WKs?2Yhl=IEnicj6TAvUzTi=@8edQbTue
zo0pQm0SX-JB*9M})2aA=9PT#`pFEcl-$I0QRcD<%Q%*93`Sfr4-zqM&4(aq+%OI|$
zZ!?81LO(LI{KbYDq*ZNTVsm{VG2-BM2z5;(8>Bk4*Wd3TY5L#4{Y`2nO`CqJ-j2&5
z_4dw#MGupShwd)zb0*1TPn{pn#?S3!G~4KPl2Q#(RFpke@Z}D9cPHT3>YzL#oTFML
z%ZMPFb6l${mOLg8_RopX$WAArKTmuLEKen?98YG*l5F@6aA$a3OeQgNafnP3zDMT0
zGCjtTl}+9qPyQ4P_djz=R8zy9Gs&fjmHLe#4@hcA$!eR)9P)&($W^^MmF!S|
z^Ql*o)p@`4tJ5>c?2K$l%1<9TfCou)L`ggXztgQ8T@1Wvvb*e^i)IPY
zNOgIpd#9SL<-Py%(Yt0M_Wrf!nl^|FUIo5lur5`2L=r=H=;aAzBkV)#a4>~T(FC(o(8)Gbh#|G|HSK0&b`Sp1*`Id11d~)#M
z!HvLU>R+#2HVgR6FptHNeBdvqBT}|Fz&@@V_4GG@`@lmoh9>O53md*Yu{j64BB&;N
z;Ut{1?ngO%-VeM;+}cBV>mN8TW{Y2X>D(Z5imxE$6++
zV*-Bo<(pBDyTC(l-f?vWzmUe$+mk5*_ga-hQ9^LuY6?=DiT7|krEU{<0)ZDAi*NjD
z1-vk6iK?^!oI`zW@1BNt-+xV(3K!uXZp~V`c6>LGtdb150rmBMw;FSR@iUY-^HvP}
z-PxlJhNaM-m2Nl7?*lJXsc>^P1USD0k|1%KjZmcHTv$Xlf)
zhR;L2M<#YOX@dUmvc>Ne(BJ8CU@rseUHT-C=OCnypZR3E4=~r>
z+5-Lcm&uJ>7*A)HeWKqBJv;S>s}t&fHvM3bA*4so^W2DWYwe&Ce3%LA&dMqk)L%4iw;A<9?|!8`a9xB0eMgLmshWd&
zzTB$xzGpn9p&I3wWP5AG(whbJUP^CK(gKuQAGl2-O1jRW-=~!vSKE#9tw&XqXBira
zN~6pv_K6?+GpxYFj{V!Xb=O`@@3-Wj`5rod={?wOwr3oucPjm+6x#PN>99BI9pRT*
zhpjZm4+x)Ff#^1x7UT#p6ODb8q*QONlr6~G>GySH9*7cGPRqZAmXF$wYC
z|H#l2JXiSz_WZpJc+&O#haI2?|L~zD-oqq~e&*DI{puh0u_OZW6TUBYISa_^dTg;F
z=yfjH_NO0sjHLUer3b+e$;CBvWvEM)0GYWO?6;)edJ~shecj1susg{e9Ml#;T~@$
z{`%(0x1l|rV2$NFq+pyync)GZa310anrjB{dUtzmc|4n%NoH%_T)h9XTv2(_67<$-
z7|GU|!+A7xeAnS#203t*1n>ONOt>y_+r0t5bX;kni4y4jv5PhGw1oLMl@}b>&LFo#
zhTYHpY$o5*E@=9|d*=4-`AK_^!Mm_Xz|EKQpq%EWksiT9($p~fSX$&A`1{Fk|3zF0{`wS^L~Ouu}IRgWQ_Org3MYg~(oclGM=EKu&m8=jU+dk|idqLy~igi1E3A7tSkk
zh{<37WFOs3vdp1Z;l<-JlAIuy@nlUCi5U0GZBJs58K2Y=OMd5&mhSnBCOxvqLyPv5
z;aR2Rhe_(89{&cS9g`h&V6vGsya-be)y^df(%eT5v*(bju|am6X3t2|r&E)Bp3T@shxy<*K^g^
z%NqQd4vCt{54=6hSRoPOrG??l2*PM4q1p39mK(!9vXS{M0q@V;vHMyN&W3&O@gO_}
z_^YhN={FA?n}~MWtLPK(ZWm%}rCk*ce3v2P%LDV=$-ijGSOD%5_UU{Pif$l7-ZAGl
zOEZY-{DsjEBEcV;$uCWUde+*^%X^uPbVyE4#o|&+|XR605u|c{OXISD_*zbG#Up|e6
z^|{0HnwB-*yR^-aSOWWd!h7t-JBaUhe0urzfS#PNDEkw*Cmn3BdoBh0t~snj^D*pK
zQq!~P3h}m0$P=A&0v-}?TZeYvNR^5J(%R3USF9UCB8cJH>33`pS{Eob^f9w>z
z4j+K?&k8Q1(z{?!@4D%C?ZHnxe3|gd3G`MJ-FAwr-$=&4)wA4C7XZC{z2
zFr+8DUk*ME_|!i7K{uqkABk^i1U*5HPk}s;t`|{snj6sb_xHgwfYrT@{hI)*YFpy*
zJX+Uku7UpCLgw-!jF)2De;Hyt+FTv#;RJa0!N*29z@wQepD_LNSX3(d^F5q7OHelC
zQO*eke0Y0JCdQHQzqcGQPV81WlZ^4^*}c(&sCP~{MHl^de^UQ`j0-+zI(#v{Pio6r
zBX5|`^wGpP)i*ER{RUt)o4{s_J4=EOM=k?g8q<0Y^-f(pbszQ2Onhm))ah5E@p78i
zE*dY*H{GT2l3o2Ijh8)7V~=3}vUXXDAV#@a)BZjh{`(!}qao+1-6r4b0#UXX{6X(O
zQZ5A0`;gtirJpeU(Yq@6i~jH~a%hbw%5zUm(st)ASgL|@nw-fX;-a3kWQ@BUt8P#p
z!8Hf|3?s;%fjLVM_gej>dTSVVyhca6-ZtBdye5({Eebkr8
zJNyUhJFl;zeF;$eR=`KB&vIe?a?HQ7Dc}j}f1AB@To^EEHj^9Ye?~>eHSEvuuGbo9
zFJDJ@G|tQGJ@P^#fbj{fzoh`f9AZ;3UMA)iSfagtf$kD}U_GYII#}ioac?1W{uyJy
zJ+Yk1uOWX%%t9AS!1tTAxB7zK*1g*Devn@pFtFbR_D`%tQ}`dq|DGZK6yJ|Vl>Jge
zK<|}#!kE7?0IaB7qlEivuLQ5fKF~W5`Fs+;XO+v-i2ecC+{JC^
z0r}e#M!%m2^q6a?We0j!`ZFaD0zX)NTWggW@C{x5FB~NhFCBxv?bii=#FjYpXqzSQ
z*6lak??Jr$@%UF~_pfHMe3bL0wFHk#tc|RY#3k|rrcd0^syPFak#;f37pSkO+9T$Ue3qFogf%@he{9G^{
z_(9|0MS4%o;T||@e?{kO26?ogIZL*qnH(Knt%v88<#R-W!l2zgncK5VjDdH3+8|*C
ze$3KAI^l0#GYNa{@NN$LzEEJ-l`Q?S7IJMt1iw{p1JPY>HJ;9%OE|r1wKc9l+}X4+
zTPKJ?68p3+_6fHVpXqC&o~FDYhsU|ypM1>*Kau-4w6vIf-+H~2aT0#7D7bXf&1J2`
zL~VV7_Lq9HruWkfW62!yGUU&q2FWt=R4rJ4X7+4pKVmR*bdc@4Zbb`r;av|938c&u1}8$ufr1F4gC4!
zac)TQHmM>qw)(AxHh&GtQJz^4Z3yoMi-m%?{DTz;=-)E1KV+tn4q97SY_Ri@w-
zry6otRA(Ub8`zuj!N@MMgv4+bKFV2NL^@6<%H|(>N_cx7Z}l~)BMueW{Jz0WL~UKr
z9jk|>r18gxSK+-yMB--6B89YKa@eTtyJq<2K)g)&VAcH^;^p$vqgul-{^Q*v0(^b&dxjQnF?rCdJ$Joz8{BIJ
z&ewT6$PRp#&F$ZP=LW)DVBM;|0LJ_F>HlNu%Hy$mzP}|AMV3T~5K@*9En@6hifmCK
zB`G8=Buhn95|K|SMI!sYQ>?q*GxPL)y?%eapEKv&xo4I;=RI@hnR~JF
zDLjuJM#r{7KkM<@`qm%v+q8rG_iIhq$F;JYR}~6z
zaX@X<$r9qFK%w?5j+e*dR~hiSUQ<*5;!AjsRMWv*c>T*~ACc1g6)
z8dSheWnxQoC)jPaGp3;f`DLY^Ml;xL(7pXoAMCz7aCXM$v^F}aX1-#8=gg-?x)Ppe
z-=a9D0}%huI1Xy&fZb}2%a#xDJlB`9IJ7{yUQ1QaKZAVkb<@&q0(&32=Q>A$M;Oa=
zT_}X-r%iUFLO-6ycvk&BXNgNyuap3
z*^n0=A7w<|*Wi7NFVc==VY{2xB7B{|E>B~RZ70}m+3l8&?QTD|xrWzw+sf3_9)fJc
z)4|RL`Vy|;h_`^=(|b3R~+1i<*7sr#I$M3u1xPR-jk|K`N)zcXo*v}3%ch~>|{Z9T}JdP(h
zIjQW}U*qO+_`?~*yWd{qeZcu7Hc1>O%On}dd|5nL_&b)F)l(;Memb^=Bz}u<#*_CL
znY;SPJU;Bi?+R?U(cao-FQ7xzda|E)o#qsIuX8nhiOkc*Ts;)A-4_k@7>wB!tz_Sc
z=nN~t|BhcJxMOyL@QKB9uzd^c*3&SP_bV;$E)cz7+9qk7p3aw`fbpS=0^tD`YK>%n
zt=95q@}B8fo~Jr4XD_RlG>(@{>n!X<0I!9W!GRyJPP{A?g3GVa(C38fP5Fi(H$UJT
zltKRQl@TdDh4WW^cP|p#S0^M@W7#E*{|+wag0N%==23wa(>++f{Ug;mtj}j;C-1-3
znpgkC>E_Beui^IxCVO8rV>wjH58Xw;^7>T=l!g1o3hvu~L%`^wIYUnco8XJ&4+K1MHsCpWaan{Z{1h
zgoH5ge#4QyHhA6cj;N(Q<~K=Pn+6f!6~oc0V6gWHD(&$Fq#x?#SBLagv9d3icdo9e
zMk@fm=U(8~1w73aSDgXlL9vZx8-6b%_vmXY<}H2dUxS4Kj(Je$cu}?fLa-M
z{G`D?wrKlfd5EjKJtOrjbRllnF5QWT`E=lARd=d)&8TWuO+p#0BMhrH{GI^2YoE0~
zzg2_xSXNnPiQv~+y%NsarDnv!p&4`w%AX+BrikCOoIXOUyZ~`q(f5kJG~`>mrNT?+
zKJ5Q;QMK1wWHdpf8^-ZWYZvCVM({pIa&rwA__sA#LiH8s
zrJcICPQr6PCg*&U9(Y6F!=7a0t-wdZhM1M1~wyYN$o`(6nu_=5o$n=RG>~{(f8K
z1K>$#PnLRnLi|^(W(rqtLx0a_)*gLZhc>g;QRRk5qujYiBg+Y}?&3RQSqS65M^5*3
zaekPmEUhR+&w_tJYj%#$HK8a&g8?fIn77@r^W-#z|0C|GT}*`Y3#ml^*GKF^ZcaWy+*>Ej7Nj)yQ`9d+8@`?+|Dz8SoJ=U`kD`jx5vitD)=5iF+
zbZy<6ZDok7<;gW_y%Hp>^4s7UM=|O*XBxkxQiSYvJja&bRU$DFy_}c46-dkA*u34F
zQWXAM&xXCY3t6qA4HDd2Gq7Lk1p=~Pv;?Wxto(uU1
z|7-TSwvE0H4S1Ps8B=UVjBy&*YI;i0*!MhnPwqn0yzHUWFHwSWh50}3f%TG%h%>7j
zfyWjo(6^V4ftl{-YavGy1aOuIB;Z2{*MoI1BH<^QkeNXuT2osdrJC5I3yP
zT2HCA{Afli?ssRAC8P`OXc@f#>nXha32&Hyw%(e=TQwXK3Z>h
z+13E-4rSZQ8GK<}==O2-D&PY?pqYF<0(f6Qv$;$rl-D!Uxg`Off0-C@TRl#g|L*&$
za2WVY_3piW``k1EFYpY}
zVxF?A@cuzJxPFZK4jtxkQ;H4EYJjD$etyJp(q2ro7xNFJ{>Og@AG1AB!Dlz%q?0^EQmqI?X;9+w?#D#s1sf*Z70;nJ73-;`f_9
zB9F2D#{Qcs7>n9V37?_5Uq$%Koe@JRoGzpChVZCdg+mC-ed3h;yE)Er#NMudGqF46
z7fJTPrhVxpe8HA}g7BZAa(F!g*KvNhM^GnGnb;pSqkJ#$w!mSMPEE6$;Jg^nF|bcG
zh3y`I%_sGU{lgxheIW1hqS#X*CL(W>t|1r~dxY>SKc6+7xPI1eJ;;acS=!d##~9;Y
zSdZ&B-dgT2#sDa&v_#@bCscVc{bGm6xM^x|p4Q&%nNp%NAydrfOV@tM9RTd$F4D(*
z{)x<=P|V-#n|K5;-L4kt$f?>ZgggrDEQO19TH-@}>bKVu%p6YvX81_V2Yv-_}p
z@xgB|nE*E>cywcbzBNVDVKluzOZ?&-DY%2}WrsZoZ_N`d7{dAeVAScvcJp6Fc(HuI
zdN(b$yZ)gOzH9@0|AhBb{Qs(>Ke;s5!1FY7Xp&X~_>lu$WW_9uqe0@evkt8Ao-}*p
zzf<7nyqNqQUa&i;uJ!B&$R3Z~xpDcKq`l@p!?<^$PG=QAuVpR?iFkZ)oT*GN0ehC0
zaUH;tXNK%yVE<_omjZr2qQSN8J@!)~
zCEaNYV2-qA_HjUJ9&@w?Fu*lTdk^^E^W1YL9{g5W$6Ta;5w72(?(75mWd+^chE>X-
zeyN5Yw@Gy()8YO)g|)CAb;Nlk;0W|L-nk85VO=~Z&DDABU<>LB+3}RWt__9g9Clp~
z{<`s5gww$L^w
z>X;qjDx?f?shf3OhBaJYJrj@5WpJfUl6wwzxjvfticr41sXNDoApb8xUl?ihfDb$E
z@H=MJj&22*Sj0%xqU8IIyXnBzYIi!&M6{Yu7p&h;
zRrDB01eGIAqnAtFY87aY?l#*+n>v(qb@pJratk_O(juY5(1xxbO3(H@3;TILQz_fe
zR-iliy-HoeHRynSL9qwK*@r3GmI92>4jMkZkDtNwa?jzJp_P|Frz6=;xX+jS!
zG0{f9Y(~iznUNn|rQI0*j-JZbFD>!ttn_25uMG`{s_=Mh}38~&bebjnW
z2eQArLOsUQjpDdXG&j|Gy+4`&*or){{br8;D$bTv`^J}IZ*)*7;aOQTjz_8_<
z?oNnTRjD7^HJXu$^RRJ>H|&G{Z^!JWY~Z`|sUHd!+t44yUDgh;9_g|>
zP~f)EggzXn*7q)IK<}OwIVJYgBWV-q4aT+AC?-5JxP7n!vAQLE-XdCs3U-#NJ1#V$
zQ}rErom(5>|B%cad8^fE{%*Q+3O)3{^Wnm2W3_0@@wJwYpPLZB<@@Z
z+n)*J1JCzd%@DZ%rm2l@O<;Ttqf-c9zaQq~F>(*-V7$|_y42VU^VcQiHXXmsFwagK
ze%J}+`rBUrXcY8SFSeZf5BT72-^T4Dkl%=J#P=ZJwfovcPSSxtQDI(jCtiQLQf3l-0x*JZ?_TUjBRU_;1M=Y3
z{6{$cO4H1%;&^deCy5!`uXirD$NRP~)r#rjeK*JBKNqP&dboKW5EtOrr?Px7-(w9a
z-GKLD@AqJWKfc5IMgp}1Uf)nvLkF>6Z|=Xc$8pYVt?YyupxmxEYDR$FW~%$}eQX+z
z3Ss`XRo^@a^UkNj@;s*icMdz+;rndgxO5Z8xt`F;c+3Z9k~^w!`gQkA?;U`5yaK;t
zJ}I&^-b(o7XDhN__R;Qj?{VC|{dZyp`|(nLG#~5#?u%}~aozif7~zYuJi%mLX(dKg
z82dla-oFFiN2Isc0$(rnK0X55pS^UG@SnKnG&Gq1zG%F04C_PB$Z%slw^)!h#=!hS
zvj2yd%V!Y#ecgGw1lxVp?<9KVPxm%syUpzs-Wc>&UI@zrA6f`+LmJ*EXz&_#@iv1`n
zC0%ZaVmAo?62tmpI-t{lLE>lO3Cj14
zbPIpL4WYm3|EBOZqf@VNy;3JYa{(T`MDgD)?j&RkytJN?*bfa-&i~R=QYGo7k-;SY
z`gPxlyuTN=U4y;c{qtmfRb_(%(eHDiC)lQcj^Oz};spDGJ>dsDzf0GM%q>Yt|H2H*
ztdPFeEs9{+wl*@4D5PqU!F~o!)Jx#@+j*;|6!+iyeN&`AAG}jth38MrEGK_sySVNB
zgl`*nNs{>nS0u{E_PiY?5_g`P2VPn{V}o0C*?5@1Ue70)$tdkK1K6Rk$QICq*#vqs(diE
z(H!u@`oxZnfG1i6O0e9v!U*v3h{7~eH;!uMypJxYzA*Sn@a9W?;cj`;1v&*MEt71*#3
z`R^V-iv14QHb~B$X|Ut1<_4MRU=x|Y@c$8P!TE4pHDAH?AkVVehI#wc6VdyJ0pA|r
z7(EN)?N&PlZwHuvtN0B&Ou#s9BzEsg74U@Gy-tJbhVUNaeY)FK7}wHl7B%oV99@{Z
z2Jicp+-{Tq3dU{54a*UU5Pvp|o1S)ud=E#SH&X_DK%H(g2=Vp)ww+~t!ix7tvf>y4(
zLEJR`bug+|1>%x_{0MA1K(Cx#l7B#YbdG{g0#6%~Sn!TEhxK1~-&pHQaGyzS6}@#3
zFCS$WIAy`Q&bZNAZ&mpJ`klF-Ein)W8JqG8tDxRKKH+D54*oZJXD^XyHPEjV!Py=(`nR
z9%Oz*`ezjAGyJ9-hn?|z14gAP;QyhwhFw`$+EC=7YbVES+R@aeK!a)N4tRglbkI(}
z1NqkoY%ha($DzY)?Jk&mw1i
z4io%O@f<^c`zqX5u2*hrP!K#v-p=~6CD1QCUZsA~g?OI8sXGJp%|&BBm&gY546diC
zCD3l`?xp(RdDl0k{j;^Fpx(a=ha~I4b@_B>t$=53+*2cK4)Z9N#bfsPe_lG*7dgHk
z>po%d-^819jV+>
ziSn&%M+2$mVg{-&vSPZgnvD{bd?>sQY0l0h%^+J7WJD>U^m8
z7FyFSz&m+F-i3C<^W>j(b&41G++5tS`g7oC%EgTPOu?_9!LZ%AP!1~Y*Uv3LAL6rR
zS`gX~6?7!if$ZP9@A3ff$*@BM!C#>sjtnuSGDf!}r3v@7#+$UfB
z*&OBpMw2!@G|kBJ>;Y!y>Nd0_C?ci=?e@pFXd}&&7WltZ`lfvR|HmgRieWUc|E{;9
z?79!kUj;0g_W&<_6{eVTs0#KGW=y}Rg6Hq{Hmf;n;AzT^&hb~kpET9|ww40$ocMY%
z4#M-2(&txp7Us=A&T`O|z;oglaGZ5NJO|ba?GgCA!cx^x$B$|-Px&aVVgcg?i@NQN
zwNM@xn!;wkt<7jyc;HG6V>7y(cw1Y~paosHD^?w20rPx4=b+l_z#nV|k_%ydw_&sN
z$v-DskgAMpeKWLcw7?qy%M;LVc5fGPh39?!LjP?I@N35(1HX6j73X-U0QX~;5r<@aIC<{{6Dx~(1(f!N2tf!sL>Cv4_m3dE6|7gDQ2<`;{WMztj&r3
z0sh?gr%l7-_(Sk}?(!))D`m)6`chc~ANY|hUFTp8>0<5S
zO;_Q%gW9z&F9E;u{ML*Iec6wXo7aL~sNrpECE#^wHm@kqFHCtj^;OG(=OfQB-ah7rWw+LtKFn{McUdH;1J2I3l*9nmQ2*VH`TFWy
zWei^5(w?bl!hBmd;%*d<4_*yMqXUrd?UJTSyzce!-c%UYyV~4M#PN`ydQS&%-cIwbsP-cvkMe70k;MPAoRy_!DRR!Wi$nb4lf6Jp{7M6|*xqekq23
zhfgX(`~9cYg5&eZjPvjjKq_TXGaPRf=`!On|27b4E)iQ5@5yCpx0fsD$%dM}kVm}RL(x|ZhqTG{KeEqp_o?xsG&``LA<;bbm
z%Q!!}qmfYruen$YdQ=0nr3(HZLHU(q9FRC%@#QE{sovOt8&-U&%SpU~=
zpBly|u~Er5-7}yPK12ci=l&@ToX%PV%dCLh{Ff*^_;@U;FLMIu*Uw+0@P6B81H{iZd2hn^98XyjJ?*#qgx_W4
zBoMwUI!I0UY5ggA!h0F<-xwj^CvTn;y^cFfdqCC*og^~zA78>7ePdJzCK&q=eNiNZ
zw@WJa61!4a$QG`PwHB>C&W(}49lITZA;!4AWCkPqDn@TKrI
zf(O@9(vSSBx9Sy02EC*)g~wYwpwyr9YbVlP$4axHp`cynn@D^AG4Nc6<;r!+WWK!5
z^fxJw;s7G{k1V91;UKPN{w3)u;;zsVFn^f7M*OVt=_U6YcG4oU-KTU?&L9@6!`Mzj
zic)VUGNy<==}9E9-<^QJWC}bfQHAtR-Ru;S?)QX;pqDSDTzo4QB)_wkngjzn>`d_U
zmdbDMkDrst>CGc}eA(7B#B~60K#l#a0pN40^P$*I+w9Rbe4nv&CkBiLR1<1AUBl|5
z6&?o~&g?&h(>;eIIWT@Tv?H=)^9`7xLVsHN`5x!*H(Lh{3GwN9hbl(rKkzwgz~mEU
zWPT>`fHfTZvt{=?G9PQZ8$rs~csU&!4Ep2cA~=v8;&7fo14fA-V`F$6-dt?c%n4{9
zFB^OY@V!q6`3}HDX4`u+h#T_t{^7V>);^qOIDdum#h>{7TmyH(&&z;CQBoJL15Tbg
z{sQM)+IRRcwp)1kG!@GUwr#<99@QWm+K9*H--}DXaC$xcIbDp`;{Nl(_K(bCx8ZS9
zCM2eu3y{Z=8!8^&$6Q`W#{Q?NR)%4J-8N+U;`!XVKWln%e)Wr+zB&LulYK1ukQw6g
z!1Cu#Ht^fks!3-Nc!zk`x{Lq=h-azG>rO(P9KOc+9nX)pRd*cE0+eNPr-}uhR{uS>
z^B(XA+q@{>YDgD2LUolD&^#z)7vBGRNut+E8pm}jT{vJ7*4Mp*HE?{*O2`=226}f`ItZ{u=p0vM%uL~OsDZ+lUpYV#O1@x_@-Jb=1nZLAK
zp1Tb5zil^$+<}*!%u-EV|EC2B^N+tO2ftM74v3feLO%-H6LJ&y+OkWQ;R8kBrwQrx
zf8l;p#T>upVLoN0{$1As^sTN|Y54r!+oN4Kr2tihb(W)`ewOdP_rd#dzm{AVmV)|8
zyCRs5_too2x27Eif1O9v+ULQZ-a+g(JNV~x`C4rw%v&y;X8&6X>1S8eOMgK6L5=mA
z2<&$cSXMbe`Wg?xhAo(f?C4s17~YqtMWiV@o`iPwe)p^!_~a(D(*YyE=jh6OMwx*}
zYnXi(-~^dnbyLwG++Uyh>uLh<(_b<5uK0YGnBKiir@_xg>gqbYZ;*ZE=eKQ8F1Gb1
zB0J%J;{PeWsRuvgd7~ek1wZIJo_$z>@+JK`cT*XVWoXCp66kl$6;*IU{ajmCOX>yx
zdj6x6R|HI!tDC~SY&Vyt=2p0#x8_6MH?Ws(XKldi0!aa?JIY`lT76BRvj_O3N!Ov9
zlfV-#D<4(=0e(7q;Q@L8e#r70vM7O{{64F?YvI2B&cBSgAitE|!EFERCS&
z`&%OS3ewxJ-U-M5W85?K`LQz~?WRlQyq;Xzlu`T~o|^n`r5gOZba+>4AfQLblie=>
zHTWxD9|M0q=he@r1EzK;#6^NV=kwmHm^VsS=oMN3_U~s`(E|K-;MnL#z~e~rx(u{4
zo1wPvdf3lUSCHvFr~&2mFyOHQe|E2vp=JgCy5K)py8~pmn^PTfQ2*Oyas!m%K3YO-
zbjP6o$vUb?)5CMNtv!7rB<76)nG1AfaNYVR-_EiF(l5q;ng{(|eKaTldX*c2j<`LC-llmT27IpVa8(`a
zpLCaA0%W{eRrCgUIE!6RcnHkDemRxe;PaZ(szz07fCv8;{w0yh?xDHG<$RE)a9w
z3^|)X&n4NdjsO38>uBSc8|dRt6>P-(U&`rN8-6}@T(4Am1CCm_>Rktv`xf&Nk2lME
z4_?E(RzamwSP}3%-%lnrKyHahZ#=%<^>Qr2aiw%HZ2^y`H$Bh8C+J`uyf{nx8{Dt!
zjQ=4#UR8W@fdB&hS^i}%)|+KJYTsPDG2QYr|
zEFpY_|HUd0Y1rpdQEi0N``)jT{Tce@s)V=u%d-Gj3hiRYvh#yxjbP~N}Ehgab{Jf&@
zxQ}f3g#q+GJx!9H$7n(@@+%v`*VhdQ#(Oalzor@~@~?3Ll5XD&Zo~B&_&689mkgMo
z!+cSO!lR?M0d2zhr%`xvKNF}y&OXgT@I(_-E66XF005KzeLtn=-#kDa(`iV@wU4o6
zI|U~nQ7~}&A7YvedVJb8MnNb&MsDmL;M!OUnnUlv@(k1-M%Zjb@a#YPPaG5(vtX?M
zfdNqOpR5h+8OxiMDYza3$mI5b*8k+_fBEBD$9kDsO8I}|VuO6_Bfc^8Z$Y4$fF?@N
zO#ppw0RU>%Q0nRBM~eJ4h$3e+K(`0|?|=92)}rVY7$|rxpMv*bR~NQ>gS!PF*S~r+
z`DZ`gM#-=9CUz9+=P1Qa_+LGw{`0@ifs#(QOtBvyOC1eC+CD*!$6VLK3@
z82k$i<1Y?My|I&tA3VQRxFR5)9=%L(-@o@de?L&{#+504hgIVWfN^viO#cC;!r<6{
zz>pkUG7q1=&`idK(E(91u1$|t`{Hru*^d#J*nxkW4}*iyE)GmYwv>av$ALWY_Ak76
zg7u$Wr-`5Y!o^5EEX&{;1YU0&N9J21m(IWg27IIshB3gI%`nUYy4sBpc|zkec@E2#
z8^}EB_<_$P{pdM*f?ww;?Wrr|Me^@F%}=h+j5|x#Q{1+cB0sf9w?106BM*
z^Lnqb0i-{(`o)6pccLrY1fx})2pK;E)m_@~_`i#_g{-%*U!%0E(u3#Fb6~%Y8|7Sa
z{USP2o?>xXGi=X9Pnln4u5MMpcJXD5WSmU0jWx#dp``S)*dOioLFnMXH(Dv@nOh`cOB9Q0(`Huy12H{yLnA
zeS`T_b5tJ;{7^3%LQ%LK!&^2eV*Bxua1YcAkdBfpeFSHe>sP
zFJ$9!JZn){w}Jp4x13(40UkFd`r~;A$UI4ODy_g%rq6v@83q1*vNE+i3GniP1E+f+
z?uQDVX|4raAAIlN3$Pc^eQ_cHuyyU@9Q=L95qDl^JCN7r_ORgmJ)SFMu>hvOm+`{o
zT`aCQdJCu&QB}eWcw*jy#Bn3BRbE^^b>&A7*#P(L)!@VRcy3M9Ra}4N9GdV!1=!C)
z&qmJOOo{i;(gT^M?_q@yp#6e$w?C9)f^K{!AN-S3V{TQ4a_n7gETabenmc;11?=*A
zkN-Fc`kgCBS8XtwX0>4+cCu!I#u4;?_7(5G4_GZba1`?^x_I8UlYrrzEpwRHC`D}>
z*$ODSS5g^csD!!1M!+knAv-Xy*|{Paiv49^$}7R`n${s3jd|7<#Hc4KxF8
zzWrhcuJ_>T`+X6B#jKIG#euT7}I~~7ruX}
zx6f_duV%-!Dlv~_E;bW-n&r8h5OMDtwAqbo95c(Y-q2dF8{g+wud^uD$LI@g$Mdl@*=yiKc~CD`H|ul(p3S^`cpV^`zQi;J
zep;q6b(@0BqTB{w5{Bn%^LjVjAI95bwKxG|eXWGZ^F4Nt9j^m$DAme|U^`9GHrx&}
zjHmf;Z87Gj~ERTUiYIAFRuDc=l!3Q+%)!{t9y6qkbABWtf+q!{4nNd+sg#B!O|4b975A_#3z-Xt&
z{sG&ce>k`iV@%KfCJDfMyGp9CzDoT=CQe_qqbG6em~bQEo3{jxf5$xhT(wvb&j0B$
z_bRr}t>PqccW(Nb6~1ng`KLd|&(veDusu^r1`tFzukfK5Sr0!Q%bbek;gCP3SZ~5N
z;e%xwKGjK_-$efV<5)hv_AwbRncZA!ae9T#&NVpwwGb_dtM4bmNqkf8^M*hS{Qs2|
zS%=SPsR8Bz?d~T_;%QZdC5bcFgI!3R-F-&NmgxC1fSJH}7~(?W^@UTHi2g`g5wW*B
zSX7VoZGj0u2w;7v=r@UP4ksz`X@jghiI?p+>xds?D}f{~y^Oaf@if%0o5WeoU+_6K
zxW37^1p*j=K8T}yCn%O+CJEDg&0iE)#5k}een%>wr=Y=Z5~fcGPJ
zlYDOMM4Q>WNqex7UA-+U$QgD{S|7gxr
z?51JXi}m_Pq4@xnW%LMM_(b71x7Jeh@Ns04{%VkN-M3v(ePAyS>;vNvu0B9sLcrlY
z2c}6NSKXyxun^_?Kvs%ek1qrLU2_V;;ju(k+fI=g7%A8T>q~q4XNB|JCVl3lX
z6-Jnx6a1f@EEfNtJ;?@S+?qID>KFydYZH(gF&zd}_NSoxzj{ddSDp-9J)nodoS+l1
z9xQ9*P!P{Dv5aS_7~lQ#d)>eJ!+G2dAw3uaAbuJ!+WxbD_Ah@w+>IbT|6lqQEP_1x
zPbOv{bN;h$w89R7K7i7`Yyb6+h165{v2dNbCh*s(@X^7r(oTMK@Ae%>0><;Tx+~{!13x{X&8kVCnzKq~FV9>Vq
z0>v9w`=0!xEXvG9K>+a6NR
zM4AAS|9}6!Tje#{0R@Nntnr7~tEzt@`McVs62At9Y)QI?B0E7bV~XF|9`xkCK|*){
zg!z!d85JBS*V`7~#F!tYrWAp*%e@#J8gZWL_Wm_FU|A4AZ
zn1?nVF1(BR2>YGe?#BU}3k1k_A!@nW?&9}R;Z5RdIIc@eoh1Cqu1~#44&;R4_9e`N
zI9YeIV*ZxX(R3W=&l2efPX_P-Z&Oji>zIDZVzl_6V~o?KKL7oS%jM`WT8~k<99_ie
z_b;70gRghq*PxHFf9FMCj3HeO>o7i-*dc)NhU3K}*k0u40Q~U^_AAUh?BfEYk$f`A
z14tcxhZeWTt^KcVVcwj4yK@fns-1gknQ{8@hZgeKpMYI#d6=(BKYKQg@wnRnbR^g>
zpu2uE&i~P@TP;Qh)BsO1v~z@F5B8TY@CfN|v;OiMvApiseqoHkd>(e#uKTl22}T|!
z1wMQ|&q>?&Snj;p2^|T>w`gNnSc7~dJ<0##WqvywW8N&vz0nZ&pF3flTQvavE=WvZ
z|CgwB%CTQU(cg|?oRQ^tiv9gjlXn-(zcvSx_l&=CErYSFyKf~L=kKHZC=6quf#xxc
zN(P6?dtaB}3>WN&#rUp4j4k#DRx$n^H{{3u{+y6(z;ZsDFyTLMVvdk~s`6472XH>G
zjUJN!nJIW!lJclzeB{OUmyEcXaXm*ZrcdB`SGwmba?S%w{Acq2D*FpODp>!Or??f{
zKlr>C79OFzv|3?UZeKrj1h>z_HF588f4kr~N!Ep#W)@koUpAJZh8P)lmA=J()N?kG
za=qVZpNg;N^n6R!qZbWh4`cmUz#{4QbKlsXV7*o97V=zaf0J3ndXejchjBWW^c?AD
z!h%m>V1xHpW#?dq1@ltv*LU&!K)-9Z5T37uI0}+}($O4mgZ&%&uA_qe6&Fl9fb&bc
zaGwezmHf^_*w5=5>0khddd%Pc9Jk9@y*J_C=aQbK;B=#=vk)<09<_HLc~ARC{>BoP
zBhQK5!uaL+E5e`U_Ir_baAZp$&xME6UE-&1x)1~gSQlj-BIVj{kO&bR;(xK8GR7J0
zggw~5M|>vVu|Jw4*)R|Se^*)3zyils7+4pjrhF$)&E1^X6;fg50PXX=3kGOd4|)5Yl(TJp5g9M%
z-L4Q`eOCrP?+AJZsW{T!`U(m}k52C(_hBl6Z9kx&+p(QsULG_%$obRNLhU{feRL4W*u
z0@A-LJtb(l6*E70pYw7T!SsU^*&Rj=oPNCp9$&z$83=cPa=jE@{eu>&4P;IjjWL#g
z0QUe7?xyG;$
zjW@-gRC!81n{o9)dY(KqKVUlin-e1wA#ih_-}xxijgfr7Qi
zDEJDZ2~Lm2VH_|NdNoE>==B&icTfSV7}X<#;taaS+}MVBbJ
zUX`LRg6N9%1K54QSNOJoHMqF|AK~HvuE7A9hN}TE@t^)Cu5OTtJAjK=1cb#Og6+Z-
z4E~qD&A>gVRbIZBf?AAcxzp@(;G{oH@A
z0~zz#X`dcyM_MM#`@T6s+_H`lV@d5qk$Pd>(o)~hcgMbg^o4E|GyPC6JE|S|IbEn6
zgZ094HIu*-Rb431OG(N@?;A>WcWnFB)Qv2qEF^=MJJ6O$dWCfAHgs1^VYk@24m2fx
z@nC<(Fk%(&OA;OEM^06S+#d`&P#Mk5PdnCjqUGPCygSF+kj&3SgBZV0h^Lv&kBg%R
zc^z;+cKvHR%2{XmyEd}}o&UU5{)#26gN0j!`u}Z18Yk>upNwon_o{X=U&(Dpl9y&B
zq8OmQT}9+9JDQO1#O~ajN%(%8#R2E@Wi805=ZKCDdm9>DaXe8C=a4O>jSMecXht7S
ztePr(tw;KDi-!02Rv_PR%a)1ZrErebja>7HQj`=Lbz|Wstn1yAz9)L34!MLPb$PQ2
zWE5RsB-dI3=N(@is@YM5{(C6NU!2s2&>4*#+u*#51nCUddu0Xa#oT{6{GgA|vu^s1
zKA>Cl4K}vW&X1*YUmd>c^YLr!7S0)u*XUfkb*=a7VswVVGI5zH4{b4ARe2GT`ccU=
z$H~>J^y3Osnedo+0TTb6rmsEy9;ur{??}!dr)T+55rXQ!iuD)$RRe!K)u(-+=GUp^_qWwa5C7#84ipdBNlDgHln~Yx#W*
zE=j0w`=^ro;AfHL%DBVhLgc&f*^qg29vc2UWpOVo9d#ay9eckm6@3Vm@4l~?hq5$2
z1(o&ZB3AqB4+R5qP~@e)ia&SKQCJJ@?g2RG{p!f(EiwKXNat4ABde>qXv?=NgY*8`
zhgYC+(KVwa}v>weT|QXoRU!-V@ia7X9{}wp99T?v=p?oB6!AQDg_-WQ4HJrC`y_ngK4ZROGc;GW+pd3bGQtv05Yo=c0=*e2)B*hB#REk7!7zpq1&{V15cxDi9hz&zpiQN`y;~j-;R+
zk&}1dD5SzU;^G@(R#K409{mF?M^eyKUC@WXC&}or>+Cf9(-f4^ojP+`HU-(ny?WGk
zGX)i==YRbEIvL$JFu$z5F$tNstFBl2m4YTKM{}aWQ;-M!qjq~ZkN$^S4Xyp3M097C
z)siPE9xXSlzHd8}id2|Rn#l5`qPUizO*&pF=$_aot1v$JK2dT_Ys2qYl(HfB&+C&Z
zNa(6mLpD2{Tl;xvr|e)VvJZ9Hxt}=|#+|=1Z(~zXtVg~?ZCx^Is%cOCDv*q}-UtzV
zADx1jHdO5nZcamUbE?}Is?*R)2je4?Q%Q*3;KsA>K8fh#$d*Fq?s&u(?%*mSk$~dj
zdD9j>5|Li$wDHk12`F4}la`liJZj>(^qsdi2HCDAynY-Qg;IJKhpq@lqF0)UKiAxh
zL+|3YoSw9bMV|UK*ERP=qZsG)x=Rd^=s`zSX3pK$C{JBtd*yNp>f(}K?iNl%OjPuo
zgV2tbe>b(K@6SXmiEJsqsj^XlP}(8pLOB0Ec#b!!J`GjQ7a6RsNk=z%+KZPnuFFU2MWRB->0Y9Cs=d_=UZLo)
zSc|@bPb8W>pWyWQVInHIg*NQz%6Dp!4T!Hl&WN#0EQ)v}t0PVCCtv#9+J
z9cs8hyAWE4G#{Ut6tpRZ?}UiDa~&^5{fj$upPQ$l_&9D}ap?@Sis-bqF65x=6V$^=
z`GqLVf_@)wdlCA$$1mhVU?hsvI`Lq)Y#a(mzR0*SHUr%#PPxy@`VQ5NWV?NK`GB;T
z=b9PQ%h6X$d+qH#mFQTbv}tWfEsFODw`04~fYkRM-g3pd5#1aSk4U5~N25FJ*#+-a
zq74JJ0raon|K_(`9>k3|pa7|pQ?`4X(299C>fkCx_VvHw!}eF82TzZdRcO|t{Yuw3
zU)*d&*_AT2`V~#^U8}G=irL_&Y~{?G;Zme<<4%B+ZVh^1Hy?H5cOxpbt>ui@hV#SD
zy(rQ*{eb$<*6ul@UV=LNR@~3HS0nMqf3DZCHz5(7gF}1bn_=HjqwVauMkMup;1AbW
z6N;mmzNq;T{-?gXO7Rn%4{%Fii0_#v#Ml2GC5ekR!S||!Gw2UBBh>@z-fss!;E`5b
zy}G#_Y3j`$ZQ+LV8@%5Zah5e9^#e2cSLU0Mh|;j*qCgvZ_S#n>TDt@OKf3m8%f(Jq
zCi+7*M5`GUqC9Pf^cJKzckv6OK|7ie=sP;!(1|Y7+{jIHc5
zu2&<{wdq8YLk3%t$2(DZ-u?}L?{%T4ma$FktzBqT_1fTp;VyJ}EOhI>nJy%s`rg|m
zsvAAy8d|TL-hMf3#oj1
z|8*&R0M)&sefP$G0Bvtmu*(bRN7>gc!w*UIA=8vyuN)M6(UBYe<_6R7KQk52dnb8d
zeeY!OYO5#k>|~|ol=@~Q+S6hEH5ksdKTe$~2!8HJnYhvM9o}Dr-pF`*8ukY@hSx88
zw*WtU=IWyZ|Np8|L4VmhQ21RL7rIWUj|;nR(89SH>AIH^w1GF%xr`U=yaDsMP1o*x
zqv=L=9A_BTdUm0d!cIqllbvY%b<;5f*NwyTXuMAss7pj}!J;x
zt@r6UN(JY!t(oyW0_W-f(LZd{=-7-7$QeryeXmD*JoYQT7=!almNRG5ENhUK7pvuc
zH&{R95a^17b1g0#Z4Ejh-vs9ri`2zs*P@o0UfQ32)u`FW-$?UY8+x5uE$p8P^S!yQ
zr>io}D4x+lY1OD69n}^3lyScX@yEMd-L%?{81ILqHN$x|zg>&I2#T~Ie?CRaDL6mH
z;_Hb=^A$CyV%NK{GriS_@5%jX6F4u9{grV;t$m8e7{
z>D+IxGE~Zxm%GKd2r;ZN5lT%fL_c^+?oK4tpfsGd!U$XB|_$T&jd(UN~m(*&`
zrm;oHtTc98KXVZ}@6WzfVzvM=&zD~1JMkV(E}38aP%ADQjZdgJ;k2l4u!8u~Drj##}fopS6*Mp^m|o48&ipsZ4N
zhdG&S^s_W!G4OT<>PN1J1FI6ya3Rf?v-cxV$36zS?G`VPb;o-P(S09L?&UJY=Ai<_
zS+bG+NBld~ZYIk*A(4l+ez40bozFpk4!__!F!2sW#4a#cYvrNZTffSzE3(nP!uu%^
zG+Bt3LG_$&Y$j?Db9%C!Aq&-c%_wUcq@$Mfso$a}lTn}5%x;(dL}ah+Aj8O-h#H^j
zz5bM*fSlwnN>u-dMr|970!LRxFI2jtGQt)5`ZJItH_X$CUdJ2n*9pN>Q(
zSWTXA!gtCyP~9*JN<`t0mg_FjCnJqq{}CF86vQ|vdnUsm1%2;ko2Y*H5-ANI(6O3)
zjSjBLsu)CtqaS{qe4~o5Q0&+;$0PFyR6e8ZDaLshrIw8IMZA56#FV|X0A&+!4AY-@P0C@hvbFcn#0dI{At5aL#PeXM%z6s~eQ_%Tt
zyXcqY>Bujsdv+M=efXDOIMtbS#CY44`b4Bk5EMGL9;`EW0fQ*||ONmySq4
zA^lmx$`|92
zXBS;-b_%*y$9GpJ
zBNt876-<|c%+d0t2h#~?L*XR%ymb;3?SJ+n4%Ht=DLXDCAm;v@6W5KB(6#X$d|f&zDD%^hq__g`Uf=bzRhu&q
zW6NtvZs;e;bsJ>DY$H+BAs-_%-8ZOMOyu6XQwb>7XvE;X3(Pb7v;vo4p2gDs3{Ip=
zMTX~2xLn9eMRhYpYcE-(A>B{wM_D?P(ep$O_Q%4h=#Ws%?VVle==_0a&+}m3vh(hL
zK|LD((@7p6ZYESzd`pZWm{(sa2AZ<8Av1pXOvV
z{Pa|)@pYL0*w3pz8_YtRz9`OeZOKM68k@e=C4?g}AD3nkADBnBeyw*2NkiH$b2Oz7
z($H3|AE7QXS*X^|?7tG{6qI-VPr1fWI�yWH{TChJM_R!B11)KZ7arV_1M>{Y
z8X-~4f4U9)8
zFUu?ZM9Jvjt^Z@_I{dkM!zePcDrA#cX7={I?==b$Lj3H#ipWk;C=w!s%1B1pA!U8<
zd(FxWDVf%am8LBZ1J!jR$
z)cTT$YAThDgr*uoJd#@UiA&Hi5rCo2X#=8~d1@inNKeD~$&c@3!chcwf2_RpxlpZI^@UE0tw)4s3^M|p3E@uZuM
zX*Ie;xMj}3*??Mb{Wz}>`x%WOJD+gps6J4gCN1cpKk7>xrUtHDI@)K-HN-AH#8F#f
zLfGoBR@8`kGe^Gih(GFYlgfWvM=*WJdW%2f6=6lQ=tIS&&jfFVYg-KqRfLtG%QIKR
zItj0(llvaVJR#nHUv+QbQ6GUq+iF9hpoT!|1gdsMa|txj?vCF%KM}6p4wg}*en9+@
zJg+CM6-#-yuF0c
zfP1PE!863m8$V0Els*!2&jiyv%Wfgsl?qrmF<8EA|jR!!JH(%l(K$
z9TCy%%%2HO9YwAR-rou8fw8$ot_;X6&1xBH^#x)@z#Q$vC-_*v+
z^%R2Rdw8%~X36Xu`O{T_!A2n6QGkWc5gf6mq=a;=wJ23s_r2?t*6CIpjo03P)oCA4dG|^E~OkFjDNb
zT$>?)U?~S$Wr8Z^NZ${Z3lnsgu&aKPYR5ilAP&pISoD$caKgsX!f}}*=OuU&`MCoy
z!7ArGlOxMybsIuM?L}+EeMPP{P1YHEX&hdo7;qOcTqjKz8HvN{dBu1c$i1=9x$wQ#
z=T1ofE3KMoK3D9Qp+yAda0|(he>}+5U5!P{sb_^e5<|Aj3M&(QH8FC*AC6%pSCQVd
zzf_;Y?qIPTu<6C4SVWVN^|&>w6XGm&XXerC3s}5?L<6~%3f8U98l>bOiI9|04QUs@
zK;C+dzgr|EVv*m-`t8ia5GK3Y{B;SxBU7&X!7Kvl*vHBES4LPLrmkoEk8&so`?Qg-
z^4U5DTf8%`nNggH1phib{5RZ+UFV!9^+X#GolQFDTX&i;qj>dmE3&l+XGYf9Vx>Ba
zDU2iM?$abBB|Gbp4X
zr2#C-HHSNZz7a9FaHw0`g=0~7EZ2WO{)F5-!CF-=Foapy+<0TJIg31?-Ly?U`3qaY
zGefs1W{|+w*~VLzzp;fS!8c)Zt4Oy7m+7`A38asZ5mfA3!XBLUV|^XBg6!%}&5UFm
zV3Ny6O355Huw!GWzr}tQ2JmxA9iKhKUbs4x-%DOWioX_Z@S81RemxuA_bvWl=fs~p
zI+mvcbC7F?0)sQiX!F;h$d7ZF)!;y;hx!uYtW_NoE<1s>>@_o`TFL+}yM`}wt_?^9
zuZ>mL-=9cR_u_BiN6T2(x$hJqmWN0}@rVJlktqD-WN5P&zyf(~4?ea;kOEl((NE%C
zFLw1!Y@os+HVyWxTt>YXDzc97_-rIvIye>P5xzrqc;VJ=M~9N-7X_XXRPcYQ!Z8o3{<+!H98$-I4&DHTMlPkHO>
zWrIQk+S&R|Ggzi3VQ~CG0^~YM%-M;&4SP-so{!srz@n=7bd`!dU{O&~{1O@nI1197
zIxl;G&#u2lg{M+syvuZ(i(~;vWvcs}{I&>qs^iaYeZB)SX{|lhXp&$J_DpzZE*sn*
znsfS{bR90bSWnX|7Qy05?k__(V_;&-S6LC$EYOrMx_+kcHO$$ee_k~F3c9?ztdzG}
z2>LEFWZu$w1w;+AALV70!F#E8dpQ&@z~s-En(y2_VBh+8=P399gr3QLzyF~LJT2DD
zIM+4^`%VwIH#FAkt9
z`0mtd`+Z1FT-FX@2!q})Cup5IilAs@=qET#N@jA~)mvUgvPwlIz`~`5_WJuL>y%Y}BS2+2${e=&=i*mobS%y@loc;WhKj2w1
zn({cQH}F~Iii(6r0X*@!T6?SI9{gyRHE~hX7`~*O7{l5fKvoP=GIsY5c5D+ifBJ0&
zBZf$rZwJ}HHsg-={frPOy|=$lTj2=7@mPV)+gAWn6#0Ryks@eUVR@9c8wH}Sy_JdHhgZgwEo>}Z;XQAx0oIIu(OYSjLk6qoULtaNaKv$8^m%9)V18k|
ztzP2>ZgxALzj7rE^6#_Zcz6z^e7mxY_;H%0-e_;**I-MWO
z{5I5J&yNH`0R}g-W
z+3%6`NCN+%i9T~M6FP_(l$Y#0g3-?3Nkp+*pnS$?qubO0RK~X2p3jzoOhJhj9;8G}
z8C64a{;eS0Z^&&Fs?^tgKuQx?3H9RdODqr}E8T5&0zRkSvhLzs5zV!6r7`CTaSZOjkh17Q_
z8{-lxu%~XtQXpLh%Caf^)+nF=mibA`erM{Dk#wn3?g=|sT1@@H?S3mjJUeH~ET;h}
zcjLoH{+@$<3Rh0gT;~U=Y+gya3iPlY>(^<&@fG3Pp&S4F`#sWeLB3U>ItEium(qDq
z<%7H#?C4oDF+jSw-H05+1K9Y_K)!B~BFybT@69$Ik1#38XjF>?V2|V;Yl|NDLQViS
z=d_?nWJ*paA?NQa#O$5|fA)!|SXlahS9(_NAkrjEDg$N_Shaija#_L}Mj|g
zaI{p{>t?TS-y0ktbO^~CI30^==gjF2K5xUGiW_)ac+$%KGN9!G{=cB)}6KZA|OMx_uE
z^yQ9)L=bXGZBQ`XLJN}#%<|4oK8vK~v@a_%pF`>gnoX2t07iU1I!}@rj_GPOWo~de
zB46dsHqwh~W2Z_&e8a=H2rH>cC*PdbLvERy)vcTf!8%fyuDa*n#`0ffa&2z-Aakip
z*zqlA%sMwqGLTXiTgj{@v!{BFtZtJ$p>)f`N~AJuy?A2~nk}31oNYxUbLdO;Gm$v#
zdhbBc$22WOKC#gtPy7rr_)Ok4u#N#M4pFH()lZH5Uf>kbsTd+2_uTAcFS&>XO*5bW
z*-C-v{daGy@xOh-<$a09p-Uu~(LM{U-xwL9rO16|FU1gvc)>M7XMGMku53OZ7&S?}
z*J)Cb7EOt?2@TJbmC<7){eDM|WE>b3C&k^qUPI&p=}B6$NO4RlRA<5DI5iR?!f|->
zp#b6<|2kY{=?P|Y%iA_L+z~OCt=UeYI%#NEG?4P6ngFFt5HI}IFo?0
zH%@#`G5E%#q@D1v_>X#rT@fL*cu-T2o(XFU;kXlb`xD`I4MoQqLoZ?E46;`2+fGzB
z^Wr86_e60H4ttNI_I=F^$_edKAhdZGfsTm;`xZ-
z-&N%NhW|>u2N}YljzW7jMvP)YlBmW;i>#<8URON(k4V|qKQjof359N+A6_Vy6TcY}>eo5n5eP~uGx;?a37R~{|4uv@Cu9$3e%`5GAa;w@
zuAfLPCZzZ++-2vBAT}pvM<}Sa6D|`|HcYK5iS^Bz8-Dea*ubUCVNKUy1DD)+tA!>+VC_S0|{lX$q&ZVkAFaChAI{R8k
zi%kGyNRxQ@l1v@pX_Mij$4(&m8Y~l;-`0rKR-Fnu(NhFbEI#e63Kt^zyYENxgfY^z
z;k2*Oz=r)zA7N#?K!LbC$UVL^v_aI^U=h0@A%OXtqwnZ>9kJe*w3HroJjhv@H~hLj
zWY}p>1E<}v0pf60--Fg~N=W-|?etWbKSFuDMLvPk1o=D6(Icm?gta-+<-8i_M%MWD
zeQ57HB1U+HRJCk2lHmVeWlN1FBFCN;OpEmskkflyw{I(9oa4(QhF!kM_t`S(oV#gQ
zky&JHp}z+b^+VSfFLK3b>v%J6$9W?snI7c@bKJ-3ul^=&n#e|^gje}Tf8D}*;xnh%
z`TUX6l~1g__II$_G)0camN=wfJ1Zn9p#VE+I_M&$9EETx&{LKt-^W(1L!VP^#vo>L
zQye^RpJManOA<=RH6-KK^Xi*7KOi#=taINeDzTM$Dy`1;8pJo*m@CDp8~e|^uHY%v
z4APS$umdwqXiRgda4X{&f!dRYo&rVV9AOtS9bJRj_5$}7=
z**||LV>OLkt$uo5Sk^xwxw-#3k)*D6<1WKd*{Kag{nhCc+ScE&WRkiGQROa7-T1ZH?@A_+)6FC0JR%Qr|N0=GFB~Ff
z3lG`Ms#dX%yglWj`ag)v?Uzr}mUa*&<7z#iBn{ox9q2sE$$=}y2ZXNn806s*3t>s5
z19wP2`CFxPLarXp3<_mgu+P|F!W*hBK9eKE5sBNoD|CS(Vz2~Cl
zuA~BINb_>n_9ei78qY4Q+lxU-nu!7D87WXVQqFYhh7b(d@Xy~*pKeBP^xTAu4VWY2&-8>5$E_Gd{XI-xDkg#
z3uz8+Yv&k{+isYsXs-*|;q|BMT!r93tZ$2?s0oJJKXc!qtAg|Y30hjGWW%XJ=jHja
zXW;Es{td~S82qfx^g%AU4ID(ZQ`;1DfHOJ61GisSgQaS%Cj%Nykb7)GRbKc9sQ&tR
zmPD-$&QO2PWhwm)pKQO+uHXL-9t1xit?GCS;hMdJ()==Tau2*(=1GNz{9rdZbMFv-
zSya?~JMtPZo|e6s6#o@wyB)7FXPgC5QL0~TNBf{1srokOiB%|VTE%T(-wH{URqOlM
zhQax+zOL7uBk)Ou#*mKDAE-la>_!sZ3QmxWQJm241lYHzwX^sb{Mkw~`i8XzoPdQM
zdaj4SXKpUyPSzrbI&S?@ljR%an)pd@^J{{ws>tT{cS_VnH#vl>>JGUg2%CT;5on*LnrwQICgPAIQ@hH52Z{T=JZ|!
zbfcn7>q*DZUvGL?`$lL`WwCOOS!r5SVPodn%lT4>F_HfZ<86i#0!_A*Gg&ZM8$r}EVMsrKh!Lhr#9S(-9vS+;h)S#lYO+C8O{r>}-#
zhqIeCE)$UMjzuta2oZdY=23MW`vR=Kc~Q2%p+Re!CLI%_sL)?V&eo4^j6fx$Yo-$;
zbFk_4_Iq)P0k9hNgMGN}B_K7|@Oow&1xMVBeV*hzfr@oBy%_~wpw^MEpe@81n39vQ
zJY-1#zpk4r_W#HR_Qb`vUJrc%jTq$(zUjO0QLd{PmaPT`c!1a}$BWX{FQ6z&`pR}&5ulY4x}E#%E!-k)-ar!{
zLIwlv@fC_la24$$&>a2%h0SJ{8v^G5xtHNPlQZRjQs;Aur(Xlq$8FvTX81#X<|!wq
zo9^K1iX3@h|7pk{)^)i+R1)42;Wu+--~>sQsqr^;$w21w(c?cpk%ERwSKsqYCZI+9
zYv%M$5B5?paJ#7JG<3Yd{xI$iBajtfT^84Nf{A16WgupLdaTd^PE4Js!nrO>@Iq29t}lo{ZNXc2H^1x_M_5sRhfiMeBJfmjn>*z&fcF2!4Urc~;
zXQ9=i9Xs)v+FhXhGa@J0pLuD?0a0xKtnT=n0X)(g{wnxo9(nQHDs}p4Eq0Pr=~6%q
zhQ0Q13OVe5jg+=HED*$5;1=KA@`~mKOt)aBkA1Kcq0UjXQ6z0cLfW$g-5A=jlW9U$
zHA82?CHg@MZ_PCX|4(Z#+DcE?nv#a1>6|%j_GN`jzf$RtdK1Vk)F=+Q@#?&{p
zmN(Fk?UW3s=6Y0Pco`Gl41FnbQ0F4o|1%3);(2<{a6bZ*itN|}WIM<=(y75uAF8mg
z75?IDD-GDO;n#Trz72>0U#+0i%}R`)byf*1vVeoH`2yMEo5;U0$O^a?F|t1U(kZ85
zB$4uy)0x>%h@`M`x21l@(Ptj9?VhNCJmjXgF!^DDh|d)DlNOj`Ku_?ESb#BN6r9_&
z+Z~H&_y3H3-x`1^?RI+7Q!5}t2UaYPI@FOT?ac{a|C(ZwVjS`yf#QHljEnc>3Xa>nnoz=WjZEF6ejaJ8I3%-Z*Fnj
zG#Jy(`|#8iSHpI`l3Dg9)*z!wcchB?nlbSQ2i$zi&yl;+C03-ySCP=gc?SWzOl&>t
z`Ae>FD6--w#lB&%Sn|be&5c6$hnAOXQKP+_uH4+sv1^3#zpX}t
zH}{Cg@?_%POHm=Au}hD^BSDNMmeVw#QVCIe@W$YZ_X_clYRB^Yzg@yRv#ykoWisq?
zK9KwB#E!US{?2N*VM4s5e5Jqs`$&jB!S-pfbB(B*@2vBx_JHuO?04&r(NmZRt;E*2
zxHbmIPE6NCup*x_vaXYl9mhiY7`$(AFd#E+{PF&63s-r7novU^JrsboKIyz1wJ1;%b#
zmYvAJByNsVi9X{;a5{;zu4mN|lLVm!p46+Dm~)tujnEy$8qFekqMw00i{rN3knlrN
z(%Q_{o_JvSlp4D_0lvteCo?;9k>MCuOg_&WeGkMvX0GaXj5TJP6VP;4`5rbQ8!2^6
zF$oF0k|}2<`2i~kQDAJdoWOMQnJ>E~l_JWO3m5Y=i!p_#seH$uS0Ra6`?LQI_h2tb
z6O-OmEFre9)-@t%-(jR27dxK3sX#i7M>I!|1b8>z2YJro-N>M6#j+TFgts&0w7wNA
zL5{l*ME9<4%o9
zQLZXQC{Z5eih(xl{odCb`EjSg`n%YR8hxwSjeAQhz0Ipgxg|Mu;g?-(gT_{=r}PxK
zq2eW!WJdz5h$d!UzFUaH?I4EZ%HJ>rm+^0Z$G%}i!xOGAx!D2r`wHEU$;|KvyMxF%HO5+Y$j0`j}#
z1PZQ?23!Ydu+Q=)%b{K2oz5hW@7@?H!9!P7NX#l=*hu}rn~T!7wH!Y+iXCy^r~p4
z_8AOMv6JC=ng-7Cuv_J7zl0<*>>Tpib=rdh7ZD@X?tnqG)meT9@spMF^Jxs^qzg!D_uA^VYZizp`IJ=6{7l-_*!)
ze>-3hNV*A2erH3G=^wBvAg%d<;4tVDy{z?eeFvTn#c7xlNzh+IQ->L1jA+{Y;}e2Z
zL*Pdtx!6I$Z)oc`_dbA~91TBN{N$JBGMv?tSaEATgDYL=o?tot10440wl=Vmp!FUZ
zp(|%7@!{gKJG9=scsSenmNS<`Q9;Q3OV*kh&tI0*LR~pgsyN&KO2brfJwYCuixvu~
ztyzLrPN^~K|EJm{CWR88vCwFD^5H=nbo;lQ|MK9T7e5mA1*LG*_^lD!1txTazb5m#
z8Wk$_h+TJS{TNO&;4*2=DS#V{p6L89Rt~-6-W&~tWl*Kfo7v9^wD|MYFWC3+Q}_*|
zGpXk;63{&?mOHUc5GPCR`&N0Nf=lMnyB|fOqswomjZ_#k(Qy5ne{!P@@QX7u64`oJ
zaq4|rQ%#4nI85KXB$%gx5>~SKB)({&%;CM~O_XoqiOTI)kQXMXMwXL$9yl^0mp5<%WIX*3g=4SIJ2EFG&Z+^Yz
zlYHF+4c57ZWa@gOaqnJ7e@-z!%1^)0dxnw(-LNKE;LDXm2cp?y|Epm-(g{Rs9WFDV
zSiO(?WMMz-w|9TUk1fKFjwXhqH(OzR{1iv#Cnmh|oSc@2)jWL3u{j@1Btzku*L#P}
z<1iiBRvRz?@0q1g8xtz_4aW)I}hvGcKXI!Ei_
z)Q_ZI_ftV_SMAa^#?Ii5-W#c1uO#vB(x=_@!(YOYkpQxdm`U)0mq$0Gu?6&b9Xo$L
zk_ekG6^=lKyRbj5?TZ}!GPF1;(BKp?4&UZ>ycKb80jVRO8Z`B4KxSUX7%?*ky5uut
z3DxX^7N>>q(zAVld(L3lFNDF?|Q3=wDL|S?|Dkm5FwzhkM|k
zz0S;OWEs@mBzJn5(Fui?hO0;VKSBC3q-AhX8jo4%eqDgEdj?F|m8F@E-giW~rZ~%>oLs
zFR25J17V`-&8iC#;&AoNYraz#3_zABBW$?X2KTC;757#bK%8t&E#Y?>u)JPWbzIRK
zl0O(fE+TaYwvU9@XJ+=okxa^2w!B_AB|kUBR22swg`Y2+_^%M~u`0%y9!rKVbYF{Q
z{M3adJ43X8CRBi3LAJi20vB*1o(&N@+AZ7$CjV)4Pyk164-v`Na`2E%(@yTC8k`W+
z#Uo|eK`SlUG|L1dIA&n(^Xo1j?5lG%aj4P-CrDS92vnBIX;$qp%M|)fHYIx~fN)ZGZ#Iu*_I_^bhs0Cx%wVz-E6~#B;>k=Cw;Nu>Cew=
zv=gwt^k_X=iEQK!_1S33zDDf95!(%_7ci;f^wsv}rx+i3cwM#5Gvpd&vye=84)#oi
zVvW6~7HMU*D%A2DL+tX&yHAz8K@4K2lFIJnVq0r|d+p7Ih)cthq3*3p?BT2i8B(`J_j=Dm7boE+{WbYcPYLU-$i!*>~5FEk;7FY
zU*@}C900zxO?|h;2Za7QUu*9(RW_nM+q^yq^V0o*=^i5KdL|r!
zxBIbpZ@|k%K8qcJkgK9+1b{=;m-B(F!q8HUth+S#3~+luzpX9K53T(0Yf3|+U`zJc
z!QXTVsI_Qj{8Gvf#v4@dy!#~u^&9E>IR2i8qJIRDW*JRTl0`x6?sA4K)JnO#t#NQg
zE*R&}RDdk^hp)^op`g}0>=Iw-C3yO^U|zQ9iZ#naES$%2Iwll8uqX=>x?u`t7wdioM4Q6Gnb6tS`_2G~A7^n6
zkAXMd22mV*c>)&hwx2zTSVOi?*NS7y-GH2^jX{N#Hype*%l_$22<(oz%v~nl1d;=o
z?OR1MpyuJT49mY6An}&TLK4RFP_JMq&PG$@XUQkr{`ZPP@e?K1DRdXc=f7?N0Ww9S30|O!xHTNpU(AhTI{zG-@nkPS<`VNm`%WoE8-1v-S)8^Ul;;6XFx;AR7hE5k|&aa4G#j7MQ
zYZ1!%tYEqPY)U;}QZc>-|EzUTTUBE-`PRpab#yIqBfh{_A
zr#d-4T?aRrC+|5dbjJq+WV{`oTtjy~EmT;{0#Hn8AinIL6}t5Fej!P&F^9oK3~lA)+WvsEuOnMjopmEw<#1KvV@&0E
z{e!^_$*az2EaCb#m#iZiHvMQ?!P^#P3l_4~%0SSjSN-Sg$yrd-!S9CgHrA*bnaQ`_
z@yqBWJ-gB!cWvDK*BBGMkuI+L)RRNO6~Spwts8KMCgGQypR$uLB_8z`jcq{SHvY|(
zex*3a7$u)~i=#=jK#%v{rAR3+z%3(htNoRHiSIAGjFJnuh1Ys!BnNxOqBjkPE_W%W
zphxM1Wd>{&xMAe1@VrV6UfWC^cslqYD&^g?O^A7gX5G#D#C9~7U^m)1vt7l4lFd}V
z4$zTDgD!{Anw}Ct%RxzAe+wrbz@ztP>$)(WZW7H{z@da=Bqh7wzFx%BoqA`t9}A#D
z{N0zi-YKEuKXekrXF1SS#V7Om4HhWvmSuyZtv(u2lF)Z7__gxz6~ixLco*Z}Pk9emaO%<6KLv0N
z=(eyr=Sa?g{MD54@6Fq~~AAsZ(hC3#gw;+%r@^XIcf!0cM!IKtNIJV~9
zdw5b6udDJ`s&71p9=g%>KCAf*e;Lwf&nx}{>erhdMgbT6K>G%5cZnIEeKkzAuayNi
z0PQo6#^>Ob>f{rWJ*z;OM!{mnU>sELUUNII+6di9PSU!R*FeUyQ1&ZaHDIVlvsY{_
z8d?}joMX7#1~k(9RLGC`Gd>CJ?=xaAVSAUALMlfQOf&Fp@lGfL^goI}743Y6WsQ3~
zBI@7ZN=ol3?Jykd5ALq#WxN9RWo?GJ;uxIW4fb?>wFxM*uZxU1O#q4>B-N~80t!&w
zG-&7T1Z~Cp1?4Y#Ks@CTmRP$6?8}p78gRDqO@P4~b>W#P&C2Z~VY?sb8f??EuQmCZ%scO|G(Ze4e+
zOabIqB5YU)(U7pK0q86Jz+H^+Flx^msJ~g@^}pE+lNAa4{TZ=v+w1*bdbebdpQsZe
zE*yQ-r;fEN44J~qDeY(9F?|Jg>?$KUQ`Io-dKl)S_ZZ&B4{>vgOh6v9Cy*DH1|NFw
zep5Vi0HS}FlHEH-1lq^NqAw1#!COH_OBFngpvzrQBDVB4>@5lB;J*+IAAP&*Z}|8k
zAhcRDyNoG-0B_rg>P98_Q*5*@{(}sl*!`~6sp|neeq4R#xNHNJcZz;Jd`|}?9*O1A
z-IoLz_{j{5yQ*;dfo4+0DPs_pP<|!sf)|h-e7bV@Lkq@?dF`0;&SJKIlL@DZ7XaZ<
z_syvzemdm3UB%srU^uTt5Z4m0J3&?MNZ7mEi%C
z%{oHHHMg{5`4Ox6J4i)ss0OP(Klt()nZj6iWhFh%l7OZ&XI4w|w=rrT>K=BkKZrq)
z@nJyBMSw2`e(UF9h7XLUmD8nnu(qxZgRFZbVA|u{knaEq)P34kjVqi52M*7#F@&pu
zwD{@n+bnGG?a$qG`O2fd&1KJxHh~fRb8d|JnHLE4EYMgm5i06%|8nU*uE-7
zm`TH}a`jT3mvSKTZ(vo_)FdW1Z7(wUt`8ept7f;J;43pdDpJQ}J
zC%&@^ZXhfd{8VGtdXV)m9ZLgaMC{(x#x6>=a)i=kr$h8w6!Ogc&9^=^3Ru*bq+{*Z
zhM9W0dzQDhAc6dO6yvi*OflkQtqFAs5~g4_J-KiS9DnsZi*sWEOHeZxI}(0iu|XTA
z>>rwuu&Q&Tow1*>9aU#Xl~N@T@M_EQ-?epQINy5u!n+NOvoWT^Ep8d9*{!^w@7;qX
z_#!0^tUgE!Wzw%UM<47)LDRSWCmsle(zDRU=t5pK8C9?
zG#umqZdpa=7LVAm{WjhmuEb)oI~ksxgNWuUsuH=s&k&JiE0QpR4R+h&Rsf@XD5fk{
zuvyPph2+r`FA35wAXN{;vSby?k<+4%<1bvwuy6MCp}q-~NNCl3F#5e^d@5dU0;sSZot^Xvvz{uh)T`N#9eG6raPS2gEvS6o-mLW4L>=Jj3XS)sudBNEm#%@+0js;KO&gg
zYCoB~ht#QfId~F=u{fF)fuf8(1S8byPaf`IpF6+0Qz%e?&z$0cQz!Xh)N1^-o?aQi
zGNZ3??iB^74NPtiJJL^*l1mnarS2o;)>jPqzjHwE(BI*iX{Nx&)Qhw)feZ}i?X}Im
znL~05;c=FU?RW`nc@l+r3cF3m0wTG6%24YOv-+0_;clHzQ*2gZEFSu#g+A
zkbx;YUS0Jkh#VPJ$ss?&`Q?;1vwSK*Sb{TB=)rDL-U_`&L{%qa8ItuP5S$zBlCtzVg
zhu?Y4C6GfY=hAC<47ap@WbcMeV7l~2UR7WPJ~wC5iqBgF
zl&9Rr9iv3hIN|$r#0?_+WksH9oHzr{$vh2^;o5^<^SbeBH_oCWseLlh)8jBDi7g){
z(4e1jGthjD4lfVZ!MM*eqL1r_Ey-!vQC~&XX~{fwT(zMf^_n;z`k-=q<$91H4ri_k
z#wALi+`m?R3u#nwhf~8hvc}EvT#q_LT;vz5@?ofVN#sTY<^|ky*2M6O_%^M5`&raS
zV2IGXVS*Z;lGKa%PKNiaJ!A-sT7le+AsVmtIdF2t_YeM?us{pPV@{{(2jIqAl|NE3
zN%X~kd`^Y4m+;WFq88Uh3)HGm_<)S%CjQ~>6??;y`?$?9GhEZFA}wIb;%dieoF7~39`$X
zGFJ@o*#Agwg&>}IuwtLWp|&6TLX@KH;ASm8#v--EP2Gg|^?Fa*cZT8`N+mDLe+Ht$
zTx*Z7Gv(smlc6oY`&)1sDlPviHynR9lh`#T@(ec*DR`yY`v$!uI?f)zjNylNDN`du
z^|+8dgjmpn#@qO(4h&o9L*Dl(8L?z~V-mg$Jp&|abhcQ&aXu=m)iSkGm`dSCMOq)NMesMS09sbuK-aewb#5D*1?&clPjIJns{!(y4@{|2^Sx{E1ms;1$|vd
z5uf1ufQ8eDw#^~+Cl4OEF$IqRuCz}b={^s8(s@q05%0bjY#piq2L#5(ab
zPN_0$VUz8S-jCI6?8`%sa@5a@7PX^zOL+FI^pr4qU-f#9s}n!UbE-W5`!7e-TF*vU
z_lpA>a)LSGR-HN?>#sqnb4CzPQLac<_5rI|y#<{yBX$LwN6
z%xhSdr`R6rd~{uMZ|<1cdPwRUu>O900
z`_aS_+TCgQxGQS{-noao2>;0n$@$P!z8xmmp%PQL_2)bg4ipjU^t=mwTv3lFp4Z{)
zzwG08T@1jM-~*BG$qF!LaV+kktTa3>&ly}w%m7?|8b61Qa5jz^7ap=&8~A&uZOBE&
z6p(woF4a`N0RQsPa@>mv2R+h{9pkm508P)|&1W?Ba4f6ro^_QxU>yBDT^3*h*zC{$
z3wZk;z9lM&cs;!V%9kVqKB)PDrVhO465~y1Y|LP4?C%0xSXUNYXU~J&M9z8+@CT!d
z&zVEkCb4%3p-Q7lQwUG9aHPl80qj$%=}c%BAM{Bd&wDb#16-_fZohxIfT6ufam?D?
zSj#yJDV?@8M3wJbf?kXv3@i;ZJ*RyhDoW9e(kU{7(Ow$=(n1QDBWc^tzB7-FCyI(R
zj~f7bi;~dqwvy0tayNyfVI12Lq2oyWbcn1~mcDpVa2#q*HQJo2_5d7wji;{DY98@;
z&DzKKq#&=akd59HFNolhz2)FC;l
z5%{m%CrVCX+!e^D63-^2Al`TUpL7z^+PG>Uv-KWps&uKjum2M{)$UYwoR$S}!}W6B
z&TgdoZev7J>Iimk(WFm4XaadH{G|CY-8A;2VuwtVlM(8oQ=~%|z97Cy&xGt5S24r<
z9;H>MJ*4+aQGka!E4(@OU(rK<64-y|udZ!dBT{=YGyjL-AB=i!Em9zV7qMAw+xOSy
zg+mrPiM~!MfUSWokcx2|Av~ra7TK;KsdlPycqJJ$+{(Tf*>t4$cyr$S3Y`l`_1b@R
z-CGbyUN*UzHY@-`$pNi3wf%N*?+r_NUb?!vO(0RJ16v`D&U2}Qpwt#4mfc7Y{{dB
zB=CFuEP0ID3;2-qtxoIE8?c+>JBE1qXV!0F~B3C$DP;Fong
ztr2@6T=-WtJT;dCgN|o&1d)-WBH6AJIfnJ%@*D8Ha;XYN)>F}+bo&5e_FIOJDR%;e
zuzS^uGAn@aKSk#q&ei*eakI$C%qGf6W>&=W*iwp+nKGl0U1nwxp~xOt*(-Z}p2wDo
z$SRR!RP>GPnV;X^=ef>x{yEoqpZmVwuRDc__p(JRAU1UUkIt_f?f6$FoVoBB%zgiS
zBxqt6Ey-o1Y`TsC1*J1FAAdgsms8_)PP}>x4i+Bt>E9bfyEPUO)C0c9
zWVrJULG;rVV(3q#Wa9210K@8azZkw?fbM2GL7Pf6kW=5G!ON2#=PzB;zU#;fiH|GS
zoD^5Vdg{NulXfm)rCserA*ObuwI2POF`gBg->Z`fJt>U6Gq3P4uqwb~gACtue_6qA
zCBnQ4o*Zz%Tf<%NIpSdP$qA;df!OG#Z6AyvBu=}H0{D{O9JnPL%DUvn4OqT8a(
z^yU#>l;s)@ejbB~io(v)`HJBqVm%I$IhIgxr#;e5`!3c}TaR&?w}G7>Za1)6=E9&R
zuS)ABO}w0;-*s0)1D`D8M;RvSc!+#$G1evkzPpg^k|I-sAIh)AdgU5ntif{DrOOM8
z=IU!x?uEf-ltE-eS&2`W>mUDF(t-D`y?fC%=Z_PjW=$@Z+=lDtL?=EMyTjEVEpn#W
zt(Z-2oJw1?7F%B)s9Nz#!+P851h?jVDEa+yP0DgJmP)@Q)pvN#G1q!a1GVChqi(tN
z`iaotV~|?foyRzu{M6GX!#14fmnrbjrxjm!-Djs=>%^Xr%%pUv9IKh-{!KmBf=z-4
zK0iCWo~5mIwZx_kmupg})qn25#G~gXd!st>(ZKmC+s!sC8F$p^69+OYWqRa-9u#uLjfV@}MS_}%pJnByn9
z;o6h(!Q%H_FxX7xqfxM15QK~xia-*IB}|E
zCo--X9=5M4cX#!|DiKbqBM*C^>2=Z<-Baz5ue}znv0+$1Srp+gN`N;?D@v@Ddtv(h
zIAQ*iJ3JAB|hNJn-Y!{xY3x9Wc&2)3ibzcP89tA!ViwQhtbWWh7U*RG{CtKrDd1}S}6Uz|Y*zuEnX07LJz
zPJORyhUcip-h5oHfLrlHiJ>2#LS0)mhi8|p@Y(bmv34T}286x5T9u4otzhhWfJ!wi
z={G%2)0YHI!GnZSH7{83>f;d0BWp-vFn3MO#|%a)KRV_vtc-&vZ~Sq5NsZ06l<7qu
z$-?dmGKKgUODrAXtG<$84*OPPmo8CB!zSLSfV6BCT*W%V6&%0;^N+t;O_dG6NBi#P
zeZ6}FGZtR8y{Kjc;i@Zr!2PqBuWa(y4-Pe)x8@iqKyQmb5J=qKT=u|kQd3QywDZEv
zu^3w*sSEMd_V_?oc^ul(`f23LefW0oqZfyjH6A%t`chZ(CQcUh^>!(0f|HCM42u*eoEUU>w$KF&#MK4yI*?HK^i+D_r3sRd!Y08
zQw~48KlD`1o%jG5NT5c~nF*9D^eFYsG2B;^SyRUu1_*U-nTjG=XGX)e)dX=m-YHLswMj2Y96e-;o(V}|xrw1qAv>nL%=A(K&U3ROL!&hoxL
zj`-?&W%ypzg1?EP9US7}=vC=}qf$i?63N0-7X$ae`@w?SJLR>Yqo{u6YfL9{*L-Sf
z)7tur>bjWfU57?~#h;u%-v-oP
zn(MaY(PP`k|8690bfTiZRZT&!36LC~?|u2{N3=Wp_Qe^MEa32+skFVB1hHx{B7R
z5=+D#Q`5&2X#;5fY9FUm%0M%f93r<^l90ac{=G5J)6s<^Nz(~V;cYhd-G^Bj)%r*yu3E%B#A~pjgpZrGfYFI7MajH+@d07oY
z-Q8$p`3~8)=QM1)w=4nMOPyNw-AAC@iccX}c>ui#pR{`BkOxlHj#M#lJ^`_5$6xYg
z$D)xh(?27pqkv{wHnJ77KJTCq`e3Md^bagNC1Ll21OGjyHU0MqN4KUK~oDDI~A
z-mg2iQT>rXLDsKQKxeB$VC5$tU>W77SLQv18W>gE?yepO-pwTCq*w2t={R6{AZUdq
zduv@qIhDbWmy&|kj67)l`+qg7o);L>O
z6|hf?xdbFBpn{8R907-W^!gQJNF>P`&@KA&R>>Qql$bY*zm8l1cKuSP%khxXRppjb
zfVW3UZfmSvMP!-Kq$gb1QM0ea=vyLTpK}s@OO%gf;5!dSLLO!hg-t6FMgK5v@>wOQ
zUiS@s?7gp~E_H%Dyo(hTI^m2V5jnutL&|yb#;VdUwve?uD;tEf{$8H;mc$@1_7y)v
z8!vJ+8LEGD`!dLG9-Nk;AOf$Zh-Wvhek5Qgdx`E~g5EeD!;`
zv63>_kEeNfUGH$uki&`h&ObwNo{}6pHL9X!de@b#pbOwXOfx;~X#g5UjEStGAA;%q
z4>q&OR>16?!t+x~hqO;2L8E3@J0R_+Vjj5QiI@c#I87YlkagVV%T9V1kRuoM3>
zzdl<#K02Sn{{!8~*3Ed-qEB_>qVO{$T32v`#vvYzbx(Mt9y`pPRgP_0T`2_5GtQ)x
zd3Pa`7gie#^#P#%b3cinQaYeLrMpd5l#H5hbL@?%=YT!Rc|D%odi1NSV)*jmULZL~
z*=@VLbWl2LVY&#@Q7)l%*UGC1So1UX7TxMae|jW;cTJB1W)f6e?iCJR4Y^)2N_+%t
z(`$PLY@VZ7&L*k`@7HM9fty8pdlTFq{CRynA_3@9)sMHal>-MBnUn&)Hk8zDbtkXo
z3+U|y!S0UK@S((23x^LQfK)epyWL<6bmHAnsr=uFurKf_`4c(h=c4r-=#hp}HIE8a
zBPZa*&D>lI<})<*9Kntw*5V?lmoIz((aaU3qY#ag__NpZ-8mU@yDmX@4<5F
zPG%3&Pegf@oIP5L9G?8p^1d(d6eRvnn)CG4HK1Z9GW8^M0mYMXU;aDr4X`^2to{DS
z0D*tqAPGlu+g1h}Zhe;NFtl|A+sh7#
zQg6w?4MRy!
zW_1~7kb(KVA%ZP&7x9YVtJBqg+_AOX@%9xc3LQ@TcxJ^cg9mRs4GSz$g`qK5Jxiv|
za7W-L{g1GF2Kj2?EHxq5PZP$l!93%kncjWONr8g->1-gM?IQg)
zgCg|2>ua0jBZ+mM5?y)GX8@1w+#3Iq5sc|YE$22}oAc}3?%5dm=2f)NXckEB`j7$J4Sqgh`1Q;IydeS^&==qdHk-%RSZXR803vvTA?j$wL8&M4^O|Ulf0h&
z8cx!DAe35oz|TKLFQ!J>;r}KL8H}k+umsWj%bv1&aOT7#&g-1gP+BxsWZ_~sJgyb~
zJJ((mpKW=#vTtC4hkr|}b=X~ks!tPsh8FQdGE&{%J40#En#-;v;guiM{9MX>g4_-Z
zo29G$7}kb90ar%e>s^68^_spr7R7K{LEp??D+x9^ODVAq+hPr(ch{xd4DdH12a8q`
zEBI6QuABA4LoT`GQ|)k-JSZz}X|2$%3d5=?9y}NH#P4%|r4HZIhdH~Fw{?qUA*-Gb
zhrtmE93B~?y5w^dtN0yRZ}Vb+QQRy?FDnpXy5WbYcJLH*d>Tf>`d0{x7bX$~o|rE<9J2Fo&-y{kTwI_517jbeew~tF5|4~)MqqvBFrz=R~Y_yw_|Og
zLxk10Qsxy5{vs3i%ykjQzo2!EsV0p08_M&E;3yYXf$E+l+SG+i@Dx+#7eg9494KHk
zbo|#pAh9l5J;yN(UPtc4G_&@ig|EP_B&h`GN%p$WKkxzT$0!u2lq1pcTlM2Nx1&ME
zA-J@|Hvxrmstl*qmV=48k`0%wKvYdPTvsW12iebdyqro82R%=FUU-mApqzi%gZwT9
zV3si@Yh*PQi1<_KN4cb-=ns!?O_s)iLmM%jyWU!=
zg>V+i*x$JPmm?9iAJO9+TzU`o&83Tsy`~ZSnCM2uZUHD}HPibrodaHwXY_i#twZG7
z?Jtz7<3Lnjqfpi(6Yxh|sw=tWI=cH;!S!SF6)?9{*&5WvhDa)(HLXNOg5K8)wZ)o;
zx+bv;Q;CEY`XtF{KB%b)JZT*D^X^Ha1CK=MkCQ22)~ig$)zJfNu1ct17qdpxqhI$W
z6b-#lQwS%(qBC+@qJ7TGS86s{%4u3
zWOjS;+wkNnVVkpDPEUvvlvKPLtN
zLM|X?A%_Kv+iGClU9Kfs{1os%|A}+(H3>?pn2(|}paVq*rRUGesUh{mT;H_^exQ<|
za?OKJ60C(1LNDZ=0@pJ9zpRGKqBBnX3O1Ss;1{bu&F8E~z?-7z%db2+U=~yp%N(<>
zl+@JClfWy3EWUT-{`b%ZxLI?2x%>MbI4u~kqwvfHl*J#^*j{%+uB*T9p3m?E65LTi
z&tE=4j*Oi-$DbyndGnC{EBtm~`V7aan}RBOms_K-wn7c;F0&Bv)Vd;V6JMI*jr&%o5ec#;8i6e{o_nRk%vCq%cGT>3ib2`FNJHEnC$4kAOfD$cTYqe@HfSD9kJ
z(08Ua|8zxOC=qs)_0;`bKrMKDeb3+vNL?!B`OW?t*)Q0v4k(ksOJ}3-z?K9I3{oth
z31x&a3icnerRZ?J0{hX|bWCvf(paL!{coTxaY_D|w>F-XulReyg%VmFuT#0SO^3;8
zs6SR{@xs00j{h$DsbN+sTG<9aKD>RDId_5bC!z(3`fmzYpq)|4d-B6P-Daszq?3{|
zbd1_PV<&$To->WDBGY1q37%O*Yss^S`h#)2b{0F{da{%5m0|$7qO{V3hb-~*BR3Nv
zLn&xIEH`AmEr%noe(U^Ptpek#(j(q>tKoz=s^yoVL6||$)ko{G6plFlmHAn(1kMml
zRmkZwhYQTEt`fD6aK&G-^w{rDp@!HvEASG)bmu-ul{Q;J8~>Z$S+`v=rR+cU#}`6j
zX?iphBg?bH++~Jz|G{OPAj{1n?r(@U#*EzGJvM;ra=)HRyV>Eu7@pHJ*0ng+&4nVb
zNFVRedoyN8xM0Q1pF`VO!EmSc^?+u5DIU(>Xta0~419UC
zb+HbcJv+<%g}DJwHRL@$Qdoz1iMRLNmWpuO0{s@%i8>q}aEPamuEAk5&$XwH)ZucS
z_EPi0Mm%_;Ev>H|!NMkjraB9T?8nP0q!?P^%hi|K9U1lTj?~XN??WG0!QG^{zY@Xc
zrNKWJR1oA?&Txr2>;*~R4T$t`ZH5obXvMAG)&?XGGMgc?<=ef~-2@ri9R$vb
z9rC`v>k7`AwZK#P%yt$rO>mu@L^2`00oEmve)dqPgV(5-s<=~YU`L2?z+bi3P?X~M
z3f;Y2sF-l(RaoK`$jV!7aw^0XbEsb-zi4o%|NfW~&OPLlyYOgIQx=xNJAUd70wIqf
zn(LTIzjhxF=nxIKTG->ueKg~iQ!P-@SQq?Ju7`@wZa9J}7aqUFprfpt_PESDoEcK%QfeFtwVnGeRorSo=seMe-lT}p9a`12YVQT}pA
zh_N1O{NW%Sl&geyx<2*gjwV1qPWXYw*6Xl;-XUrHk}2G6bq)MyVF4e1Os1ubIFCn@
zsl0h>PvJ+l1x5`}7OuVzGhv~V+raw?g%gBUdyf!R#oRL-8ki|m|Cx@Ia-iQ2vd1h}~7e96T!=;*a4L_rQ
zDxfQ?2{%$LE>_t|K*??iELf}n&-{~S^5ZxPg^cAM(2E_#rY|Z-+Ss;1&JBUx^LC9O
z>BPy{`%HWg3w~d(p5=#jF9(~}3a>t=Te{fB1kb(Y
zwyuz`0W5j$U#i>>bG2{gET7~{!nDUzYJ^8ZSo?9<)!pU;bT_W-@0-{Zu$Y&(%&Ecm6xG3r`x*W#{WP;rc<`V
zP$@_@ezSAZwG$Oib5X|^H-U1}1T#~CR1~LnPk;W4H}X~K*mLp!0UjBg-YH~cHRbkL
zgC%AdW=L^ItjIX~+%l?QBlqrq_6aCk4af;ewIF>i|KecPPGpf6?sP4s3?*!b`t(P8
z0Or2Tq!^alKwL3V+EUvNVKGLsR1G6AMBN`s*;9Z7n2yy}Y9$@chcapVH|~SNN0h8L
zt{5TTK7G2^*R9a-E?e3g(>9Q!JQNl<6%6(`hPk6u!jVNLW57gM09dx4be1<)1s1*g
zM(J72;B;_x4E5_u@Wj|BsGRya`pGA;GHmJ$sMV%ibmoIVL`;|cm9h>rvsp{T;a&hL
zTGD>Uak_(!r*TZ829Hn*MH){-X)xe=bMIck;oN55dnRBthyhql{3xH8mqkjS>>EGb
zwE~s&!Ti%z_W=iYgpa4IEExJe@nmX~4jo%6ADubHj99wHKU7Iy1AhHXrheZ%f#a(G
z2rGj$i1|Gr>8B))*!`~Ex?m;&JUC)V-(8SEonAEJ8Lcj$`L?8xZI3t@XFalVf@_+A4S0vri$+kDFTSefX1efnh&Voo`{Mu
zk^qD!y}>Vx6~UOgR=llB1h{F=aB=VVbJTiM*4_O?7T~&?a_K~A6PoP3S9Mn-`q<2ydkJ}ygmUz*neZ~!D^mw
zIgQ>BDtRu&vsm^UTx>lv%V~ZJl1y0#kC=>shPND6b2@G4bM4!%*_k>}5vu&`M#FpH
zS0E!g6u<;qV@Cp~^asJUWt*w^ntIgo__mF`HG;W$3GA@UsL(U
zwF|tQV*N^(jRY*Z^4+JW+R+_vna*9`-=Mk7$jM`60azO>FTa=)f<=>Do99GCG4)>c
z`biNfSe{$E9z!?=Q(BCqbiKUrA5k)Ck4*s>^1R}EIprDrQr6`5JEC(?=~n#ZBa3oa
z{+XBJF>yOQ%=CrH^%gJuw%p5EdW9E{IF_znxg`QI6e@>#>X>lvJ=qC96)dm0(>m^X
z1gdTmKYo%xhhNXThp9i~fm+GNjN++sc|l`Y;6d%6rQ5T`wzIr
zf>`j(WHL$nz{)|Un`gh-P6IsZs#p+=&EtI
z|IQvcTP1!OckdL(>@{p{I;9e7YYI8%GMtQS?qSpX>HAf0Ut^CWy|!d;EAjs3v6Aga
zG??s@Tvzf{Yv}lgDI(#05cbPZ{IAEX7{6hV8hPoIhyN(HCEO)IZ0k5m;qlZA#%1Ey
zAG3V$T>4U4>)A{k#5a<+&l87F%FM4SmFQuq-p4|UB*rj=YVH1(@f~be%ag^H5{J2M
z_L%0x!tv67f|Kp95BodX0`}z(zrVg`tr>dLD&a!ilNaY^6CmZuH<~B4+#%y>k9H>G
z*U-53xSncJHFTFxmkCI$fGh?r){l3SVa@l57mw5I=_!H}H1{J(ZZ@Ose^
zO`h07sFi(&pH3wiP9%4Q(#}7ELk*T!zYE1fH_Eqc`;93u{N9DJ&ra!wbJ4wjmiDpG
zBa7aky($n+c&ZFf=X*ob#JU)QX(YV*d#kox;2A7ET}eAd7787mN1OCBJ>cHBRO*lK
z?r>k0so&h>2F|tuwvh%pI5K*h??IzBJUMg5UUGvT-b(QqYp4u{F+R2XmoL!azsiPB
zPlOxd@ix}o#$+v+wbs0A87l${vNj99FFpSM{q~~_Y@nTZ5A~UHee79r1M>~bLe)8o
zt%ySog4_I}+GuDH99KPiVBl#3S^pRWhFsOaA?H^pP95cj#r$?Z1V4yE0-4`nUX&;7
z-k?m+BE1e9l>0m>Bvf#w)IEj(R!%5T6nT1oQW&acypJmmJ_e(`Tx5>i5XTul%P!dp
zt>F7g*UoWma;P`zaL*i`#13a^X5;bbin!mkQjk#6x8D)`UZszr_h%I8eU
zb{OuXDjPQ2%ozj)WZ%V-TcR{HRP)}>OwAD4EtwdW=}=EeE`Lbe~8h)m%3<_G~|-_!UtA2cD`
z#81R2uU-Kha_{a2G6tlM1s0L}0f@~wa+bWyA5eJ6I~pD4*Sc9;AGIikg0(Z*3d7^R
z;J*j0qdjrH$N{DN+A_5P)1NJGBpTEsBBtY9jFGPBn?~Cgk-z|8_PF}sV8;_Rg2MD)
z=LG;F!3s9A4M_BKl(|ZWCHi+RP)%Yn030a%ldpJlAC-JPJK^wI6_nqh7CmHB0$S}1
z5sH5T=+?c|gvk^EkVYE9^X>~ddg%W_C-1xwkgP8hI~F4Vt`QNavoQ;j-DrC`v`Pvj
z`CjpTqEZ5-+jIFhYIX=}##PRX6qC_?XF2|#5T?CM5R(@dIxoO*&t2?M(MOZ5K~;dcTf>*(w-l
z5^*!sGiD>VxNN0kw5`CreFcT3x`LnG>wg3x*
zQ}qG5_P0YnJ3m95&vY`9g=#>0S|WMgToqC%&-yP~(i#lC(e+z66ADzv2;ub)lTqK2
z&f1ha0{rwro6>)C(Di@>evr@&bPcr^OIEUxV?HtGm7jSi?H}Kk(xJ$h>px)pZDt0<
zoS-jkajpZM$^!-=`lX;hx?}E(STEXU`*nS{Z5eEJIJ*{}`-9p%BhNgfZA4M|7`IO3KkZ^nj+
z{}l0+2$R60v)Zk`PpC1s&L4WMevZR@iT%;oVrd*RsCkp?Gr&U>Pv)+!&Z1j7p;>)}
z6nMgs^p;i}7bJJ6NT_O(!LKgGUg&bWhP}0}^`34z^)0b~0OdFc3h>J72I
zHWW6cdA0rRHg3G}McZrcIc~L^sAN?=1dRUqFwPTs1V4zl5PC)6GOQkd-_&&fGB&Da
zG2%GshBE^rBx_ZyaA|k`I9Cw|{`7I9N)7Ac?5TSTHc!2wY~ljbhDQ?gY~iMI35bWU
zpV^j}o;QQnhQkG)ir&T4W*{MFDh!T&$_&}UMR4b;5QE=P4lL6?4c}Wi!sgSr=A-jG
z@S?);B+IHtkmu6sM~2EWc>ZK;@|tBSTz`5tbeY)?T8W9*8ID9jvQtmoU&Ci`;J-Ja
zv5X;bEtIP$F4`S_*F0ewp%IK%)}$wjABp4JLKYQy^$w8Zyph5VvlV3g<6N>xdj&4t
z`KHlzA{Wb!iMKTLrQrTvV&l1z1k5NiUHQK6pL3
zYse&B=dic&-rvz$CQMVDo&Lo`44-;(tUtY+0e)XtYrj>%0T=4;g64{cI4Rgaj0cv(|Xo*e$edP&2y1d9x-ft!OB=ddkmd!QX_hIun$yL$BD51
zUqtZrnC`ezgNDK}vFK2^5~lfjvlSUSy9Az}9zuOBEMMPSmjVyHb=^;|
z!+^Qt*Ng{dWoXixz~b|)1Wc8Z38gL$0X9xjlJls9Lc;v$zokyZ6`e!eo6z*xWj>tei2K%@bT_2`}m{YcoIbmJednKBm+;B8j>?wdQ|bJdXY~KNC>!lLs;u&$hW9-$kYx52lx37waO`DS-7Tmfl|JUsp|=mz@K
zM)}kJ<^w=0uA?fX=?`=u=a6<}BI29!7>HGF1GoMQuYAq*7Exu9UI=M9q
z0QRXg*O&b1$j395O~kPlY}KFNe0yyeoGw=rP#n60v_0Lq3%2io0x7+#oNl+#F4fsY
zvR`p%{XlR7wsnKf%B|BYW8u{ZGe@+O_gA{LE><@E?vw4#b=
z@jGGB_UPw_8tcD^mfrZLpC*NoTuDfmgc!
zS@rugp+Wx*-=FRs;NoLN85_ZVl<~f^lr88Zm`g3+Xm4kLd$0PpH=2tPEk5((1
zdreDk!qN|{b56a|=2=8LFT&Jq`DyS$0^5q#8W$eslRHC8Ap`lNW1Vw#)$y)TBWwB7
zn^1>jhax@C9v&C?wltj0hY#*@wpaHE!N^~w5;rSFai85S5#E;bhk1pLb*VI8=sf%W
z{&{{%SUUUNvV?{k`<_$QXKt5(PtB!b4#*6#YRhC2#fCY&P(}<#g}j)Ke>#el
zhNCPU6pq5%VG+5`lOnKbvMb#Am^F?PEIm;7V8tYVKOH1Ea>IV&;<#dSA-t7CadP=P
zfGOI(z0t9^AeH#Lj_~#Ca3Fs#Qs$a3Ud{FAG)#_xda5Nqi^>aOO5B2{iFzg65k9z5
zTj>l}9K9v4Xe#3^`|;a?l(%qG_~eme8r87er0@D=mRjiaO0#;@!4_ux65(5BdvXyU~oe{8uo=04J9h@971?`XHBYbYKoh2rj_vv+P)itU5g(M$S(3jNN
zuT(@$u(M2(T;k@;>!SJ-l(SUMjo#8k*+Z
z_0oS+51)ids};o%pfc@xn8V#xX!6IoUFA^kKl;Gmk9(vEDi;@+bGtRe|JrN>Zt}Il
z;AoU<>DmeptJv~_48et0R-R5T%y~V!H|Z-x^0!B1Lg(>R%mU%
z!lhcm!7`sqAeE}ur$?==&||dEok9{rO};|!k$@iP>rvvN@wXF8Xo>&5`lKCqDX@-N
zo^HX3x81)yiL1qJ|6XotxK?2?NzRS!%x)}3NNJO`>%<-Z{*UW##r@N_l66#?F;(MT
zYZ-@nd`maa(%rEe*Q>R&T$Sv?#_aoScedIvr}We6W*3Y_9(@SoawOnADh18q^Ih1R
zf8>(wrw&Z;efjb0VU9X^^TbPAC5&?{>yP+Iv|xsLb()CW!{^hXG2@tT$MSsdm!-e8
z;)XH#^T{*_GaqE%5o&0_;{L*yA9PeWHP->=8NNr`W3I8oO4o|)oJu2$-
z5|3r*T`_UH40%O3Z})%ofSD@C&QP{BV3X*nuV?Zq@tHkqbVnx!>)I;GXX(1a4tCwF
z%TD%icSimrHA^TqEHXaY8Eb%JUDUbBFBn4d>d)1Uhy25_`MH%RbLVlJ_ISqEa(_H8
zGx28O#9`hemzVuwya6=4dFAzw9$kEQh@57w@C;VmwTx)GVGX6!v7w;<9mpZk6ik~b
zhe^5|Zs!#!8r%QK2H?b%A9@jnXzNj{@37~6;Q-$*cE9)
z0_|A?8n`^q!w56(i7ulPIA)vj3R+%HFpOdzZyn?1uYTkfoq5@r0T}~l5c?7ol^9n$6xShH1)^bu3iN5
ztY)pta*&nta**WbUU2Y!fi%^D9_sNejA!uw0m|+lY=+anp&^UOmsu+PAli$>_4tK)
zz*kQedZV2b=HKVNpBnTDeOsgaF(K1~QuE{|m}%aCGTUuBN8sSrTc3uGSBtj@ow1h0**
z7JW3YMw`EC2dUaB0oQkzD<`}9K|fP`XPI>{GB`spXkmPg%sqe1j!wjYf!k@kSt~B+
zr}_u(wI?MY`%$GqPk9w`nW*@j&*TTrn9$H{*?0jDPpoQq>oKZQZnUtK>_BErzb~Ge
z%R|z#ic>Q&&w!Wy*anG=Kl&rj!98}%9c8We*3-5Xp-mgoT+jSOpndB_w~MblXbyi)
z#XjwTKzsD8iLA2C&@wwxLWa(6Y|5lLv49FGJ(`~b%1HaLeXs_qp|}5W2e^lRCp?>a+D}MxgTa5>xG_HHxtK
z^;1{J59qU8w97r21IYERxH(%$pe6x<7^Hq0J->2P@%OL_u+40)&iLz$)`~M73NG4$
z`bSq(eL95Eg=>W?HMRo4k;-8;SVc&&o2yT6XMsU;#w*NYt!UYj@BzmS0%6OY
z>mQs~(Xc8h&8b_Y_`au+dPj63_;$e9cMcRH(?85Hh21s4tWs)OPvQ+C(|5A^J+y_0
zAHC@@QFwr|Z!P%i^2Y(Oq11|-+`u2*%u26qLC*~OzoXME+w|BK6(>^C5l1eQVA`-=Me6!y1c^a^e=IOr4T?43~
zo3=RreGPrc6Q`oc-UD^9rjDKcZ82JDirwI7E3
zLHfNxH(46@K@L0L=X>33ct@#$z5N0c=Io+nVeo&9JfeT+zgrsw{|J9t`JCnveQ@ZW
zLJu{h+iX@ZZ-0+AHH>e`a*d-qHqQ^t)B6C;uvfaLDgkwy{jg>8*g=jXvjM5KLhx0`
zKQj$y6WC_^FvRei1E!8)a~kD$Kh*18s~_q5;`pE+UVQ$!`1#-1BQ}$#aqHWx@%z2h
z&?Ixc!G%r^|7*%@5gBxbsWRVUO+tb&$tQM+kH5~~qNogV#XCyS^!8R6T+zl~k2sX%Wk_5iPab1Zjoa;ty+KXgzN
z8?G=d3Jvy>Y33GavDh_r{BDpMKc^3HQZTv#x4!87s^zu94nJxBJY;=^`z*ORdL%2M
ze0W}f05^t)?zyw)atP3^^2wxg%zM}uVDRCckvkMp)-f3L@WMZoUg(%=V`%&1b+YtS
z8?^OocAjv>kWQ;fNbbB0?DuUc=Q{MUblBK+sOi&TQQfOI%daq;{_|PZT&5N3b~6|D
zr`Y1)dKHqdd%lo!kjQmDCK_|Mz8ET2%7n!~6TOx#8(>x{%PZo+%aDItV8%<=1K+k5PsdUzNp&jmkhh2e=?a=S{gr|#BJERG-B@27q
z3X`0Th$x@7!Ja7J50U2&=f~f!vR>C)q183tnYvB{4L76|+GgtEfB%HU1CJuORq226
zi8+F2cr25TDifd#$-IZmnHD%CJN<3+f-26UsxFHhsfP3tth?1Fhxduuf*;=8
z8*8a)fh@&7`f8~ic=_vsOpID59_l42{9E6NhgWH)D4%s=qe^Lw%A!tuyZoJ<+S4}d
z^@)~aPplm~aA~CTY_#L_QWIm%@9kL1l5Q&TRy&TD4SBwKnD1Fpo{uL}2Xtg>V*)Z$Go%ibEDI*c#8C{eyvjSsA+xgCFZV&<#I
zMeb>JV5JiW)=ke_ap=yyNYg_uh4nWH%Mh~$?7*IPapPG#PPF8dZlZ3(^A}thj_$YO
z3*>(Hk1RFe%~WojK3|PTT1eGzGGIJt>LTUtgmKKHI{|_Zn=nzdSZv&n63o6o8YFx=
z9V6MSe_huauvvvro%WYHEEz*SXx?6kBbol%1sI3ny40`d>9x$c#R2}r^!s*p5xkt
zz2U7xKWmQhV)|;9DX`04uzWr
zwm7!J271%+Sg~23yqEvsoLWcU#pVEiWT%RiiK^iwXOF%%V*^}L7SsCtk~tjbBWeBA
z=m`HL-~Hu2Du+3LIlgQgQ^O{${A){V-^ckgi2p>=mJ;<39GnZvqS$yXO(byV`agaX
z`hB$UPCpSd?E%G~w9-b|e~|5tMm>#oEfOQ2%e?k=3baMhbDX
zPv>J>(ZX*%qGIz2K=Z-mUZ?nR$l2SQJ{>!U*sTnIOw|sc?ozJlCcQ%re*CwZ0qr+P
zH7e=P6Y8UQT{&M|#A5=9o(%rl-7o~Ui(dCBFn6H{-9C>uOL1t&P9(GolHjG6EabfP
zouE|H-u!1;J5Z(dRh}hoLvjioG^wLC;5~Rt8Wvgy?6m05NmUMjrqP>ont$@p+E&H8
zMOg<_EGp(Je>)fawp%?-YxWH&8BSe(9p{b2?(Hq#{+^9EWXVcyyiErV+RHUA%+ZLz
zb>GPj3?i~g&pMumm57Bi@$L5dA>UM}*gf-J6zcCzK3Ce~h%9AwSE;+ALGd|R+EYaZ
z;J@EW*T}en(4wJZHFx$g&{(YQWd6t>oWjp(K3*h%GYKXStgVuPh2hpXC=Ekn&cTt?
zE1sb9GuP<-d?WB@(Q;WXSuD?1nhFZk;|>ROABo
zDB3QusH!77`b}EJhg(WTKO~AmpbD67x^y*o*B2Da40f#QUq!ZUmQI`+B1ptyuk=~X
zIq>FO`8#@-v#3e(86`JG5W!JZX{)1Skf1!L%cv*!S;=93_Gj_NBEemGoZtPgAo87x
z6ld6cPRO;P_0=AYRWjae-23u7m(X5nB3VW~rlj*)*Bo1H5e!fBTnjc&AOvJ~2CF#c
zEBWrqrD)og5nR|k+WvEHQTivB!9Eu_qtrNeMz}xJoxq$EuJKVTfiR@^UG*1Two=7K
z29n+n&4f>aeJ%@uiv-FKOuq!5Mkx7dFeVzNfq@erHtmYYB5%i8#_s;f5H9%`sN~)YUOSoa)x_GSW9U)Qi!h{=`
zQ)0I#I@NZ2>QE9_pj|U0Lj{H
zU!`6Ja6@V{(qv{$sjQayEnM6sFbN4K9x92Z-IH^mLX
zQJ+s@*&D9t_UVC>uWS6l+Z>f&DZQMC>fu?_z41Y0IBG-hm~V)_?otl7L|q5}W9Yme
zx%$F5PG)8zv=EU{8I_ECjx8Z&M3H1hwiHTMBBQJ?g))-8_qylUGqWfoBb5~ig*3jO
zKj8DZ_x^OA^E}V{^$K7rYxwDEg9db7ieCHu1nAo;?_E3o0=@oG_`G&b9=-op#d2$1
z1uUR%4@J-DpwzUOq@s)TU@B2ZzxhHU2%6g!um77D=*$PdwcbS_>EnT>2_xEIw7n@=
zu)qS@{Cbp4Hyw+HoKzkhe{~8S`7ZRRqCyL#)LMi*OEE?Q-htk6ww@sQxiQDP#BgvX
zLiJV#iIDc@sj7kx4Ipmt$R6wA8YFnsCd_K@cl6vGN$`H-gCY*De`&u2f_8%m9kFxy
zNHp1*?;dXhcsnYSy+pZ}N}8{F>KCmbi{G!jhMEn5a4<9$OYlaYCpx^7_zb|QTej-X
zVh9Yr>HSUl}GP?ax%%Hd--k#!F
zxM1_~)*vgs|IV{ia7ha)>6|lG6;X!koQ@1FQcRe`%~h*;dl%Fks}nq{c>t$j8(9s8
zldvbETB55^2^QHj8Sm^FLH84ljgI$Lpo>So{F1Nx44x129k%$&i)V6V&il0Qz{gTk
z4!i$13v@4z`w3`FqbB>E+dQ)q;NmRk`YFBxmY-jh?E1-x9hj-f_O2x~e7;-E(03J_
z*!I%a;P{2|Hd}R5pHss#Yo&*;j>*6sp2NQT@7#gcJWgbH9`nHwduCS7IX-|AEBn1Q
zue`u{S${d3LmuPZ!f!Z^*%`s1C=cnR=}VM3mvPNOT3gs)&~YcCp?MgL5Tu
zreqzxt;Z>-)FU|_tZs&F11)kH#>8Qm&R$lkUsCXw5~EhM>oL6Ic+M`+R}%VWy`1~u
zLJz~5bZL}%DRa_OTkqK3#o+HH4r-gMmC*W_%kVaJGraKpzEqV(C5+X6;B3QE1HT2$
ze(}`~!jkK=g%A1M;IZT8g=qqnu&vKzY*Lnlm$zk{8hq=Z?zvFej|vvB#KZAJ?2ISA
z-P5pjd&CC+N?QoHke?2(e_~kM{Fwtc<%=u?q7|{MZLe)}xhFKR6na0da2}^E`$c4*
z48^_9N*8p>eXz4B?ZlXu1!mBh=e&E`3G?Qe?~yR~ghMr({-+l%!DIa5<8Ft2u>4h<
zB~zy=82;qAvSe^6>?v6;x;K;$sh@v2a$_kKX0K|$Q}&93T@iOaEU(qV&U3WfQj9f_
zS837Cet#JhG_pwgDwYjz?0ogRX)cx07jyK?v&eeEsv(w@p2J1*E^9eI-3fmhLbhIdZXdxf*oyBA&;=m@M(RS
z)330J{}mZ?n2#q+o4(SvCqwhmfSO<5W1t>o)m>P~0OR?^%f_j(IR4tzQa8OMytbe1
zt&I6?D0qLra^h<@3gh9)A3CAnki|v?1i7bDsVC$N~m`mNnR_DZ)B~7u`xdDfKP~Rn>HD1vUw;IvQzG
zg$vlr-aL6&j(bjjzEkL#kGT^hLask9##?D|P7DlX`0;G@iQ@7C{NL?G&gsew%)N)B
z@s~gn&OA!@D@*eQezUd@i%hShB&AOPsfeP&2G*
zHO45UX9I0c8V(MwI>TFc#Wb0Utl<}tAC|H0*PuTBT&lPT;X*Ti-Ayjz&}WL=w`xVP
zl%2^zVSfTLU6=^@!efDJk_IXpD14d!PQO-%=qI48vwZB*R)LA0`!R1K1LSs0Hrao}
z2eQvUyz!J$KZBsp%#-35vGT^F7TIbwIOO~6&xqJ5XzwRbudk{KhxcP$jolh}R;n-l
zc*G9;NxZoKPr?z%mR>R5CB}2r14^jF8*BrcpR2Pq8z&E~F
zwz!PW-Fa&omCFcaA9+>|($iuqZvHuPmLEUVb$X(1$AJSlIb*8_et`S82Iyk*n$QtC
zF>x!Nd6a&J)AP?1H5MK?^!a-C7NDv;?(b_;1!xPOmS%4ZAyqR61Lm?p;9ZTwrfitu
zv`F4~i0&wAq76SVR5^s!G7VNE^WK0He!p`wiApq=oR#^|oC!O1e>xI3^BINK@#n7Z
z9S3s^%DRT)oyhwW!ys*bJvy=yxr_Y)6Rx!6915AAM=!@59c-VD0E3X?_J4=k(1$+r
ziGH&RBv&Uk{DG27KlSiiavEPLXjrT>QCKelZ{!~{C3F;_i?58-yF!wHUF6OXsvRRB
z(&1S{G+!m4MSXoNBbkWvAa;#<7zbi~l5f7?@(0%By{j^T&FIEwFR#Gi3UsvC^%~
+-P>=P(_JT-2zXb`I0l*L?$9ilZ
zriqWYF#dnM&2oYV@X+$CDIY+es!{<}=wG}#zkDP6VzYBf(b5O4}3Y(J;S^{4k4NJ_NX(Mql%kIvAANh1fz2n4|
zPLkU{@~j*DWXO=nS|<}&eaeaT$O$3@@=4({PK$sL40;P}TD;2S(O604UQ
z!u+Hv^Wu@Upx<4oo&2GVtjlZYPm}E+b}!lme?IOY6B28;R71RM6TS5!YG(;K?fJNF
zR3ejn>T2P$`d3ISJ-Ih^z3w0(2}$wsaBf^VKDtZQshG^U8_%$O&myjqyq9097Z9$N
z?gqykC_*wGyLjJ<0l&C8UjKXJqS#91?7*%%RR*NV*epEcXR<5U*8x;x+Tz
z$np<)#_6u5q}kUhc+MHm`^rT`=>3d%_lE!a`y>+E+&6f{wT+fmXm;jeg;|lTyo~MfCK~k
zYofW1;Y8lPB2qQdAAXg;gjjNWNyaKxkh4-(Qw5F}k@II~RF<%ol7PN#L61+rA!bdF
zn!LwLNMAeevNL6sWN?2|f4*J`F}1%m-Fdl;Y%P`Z>O4?P>Yq3D)VINPt@R1T2dl~N
zH6^<{AYLx-Vw=x~cqt?Jaqa=cON*R7#U~G7{(JUllMTd6-B|kH;S3cdPIFu-MZK21
zQe9en`X-!jQ5++;{9&J!^+LqOo8T7=pB8DTRuHb-k)ow-wZt@rTVF1%m7EJM9aDmM
z=6wIXO7&IXu}|bm1!5}6i(8SBcF*d`_{IXqH~XQVSlluyiv-@?`;W`Jw1@)1tl=uOVJ~4?niv
z0^{ZQuHZ&_h?f>R!s7UT$u|ef1Jl4OqMKi4ErDN`Ww++T3K&0Od{gCO3QlJ#oG#wf`WVyL-l3-vYg6^_9!!FTg&BN73vv;K!ML
zUdz~l|FB0%Pilc)f}F!=b{Nk^4t~Z{FfP8`+Qb^i0latu?S8qQdoh
zO+WYxx8~K(HsB{j#4_}UVEo(f{G@>M&}o1D1&o*P8uw&kyv$czVr~lSAvYV|XRUyE
zZdU7YjtTTu4Ak}rfu8Q-@}6s;m$63A^$_U2dHQ23C+K}AkPtOZl-4EZ}Bmskt~Zsm08#PT)-^d7^w
zFSU^0vlZ|NUC8lmkpEYO=_T%OWV@D#=mMr***1XX1@^PP!1*-4hy;%^=wn3oifCr5j#PQm-cV=2qDTnDBOYR
z=lT4pa~Z1{-QS}uGGn(C=kZ&m_xm7^*uBqG2h&ZrZQ6iX_4kz~Vp_u|4a84#cO+2X
z+M$WsH`#u*h}uV4bmb$aiS?&eoIa#HG)VZC%yw=qI$@LnTwnMsK@EHcg
zf~?IHui)o)YQewzE
z7(6^fv0hDY-m9QoFe0mu?dsXlar7TN2fNuiJxm`vm)(l#S(_%S5F?Ayw;>ORGc|9;
z_Or^H+GY;;O>#-p6Nr~>AC{?(!1yw`wRK@NjF(XLvIFlSUe40-*6as6$?;di<)9Zl
z_I&M1&^s>qP_-KFH?C9l$YTw>-!`H0Ef~t%u`=M5A>bv3>%Gq)pJj`D#W@&?4;Jg6*!K&KHclcg;XifwCs{qbp
z^~&0x01OOuX#W9rJga^{5FL5A`}ANR#Jt(s|A{{MBXjp_t!p8E${(9M(hKqW%Z9`1
zdpjV$)t)`E8rB1@tPb;mbyCrbjO&?U+;NWBk3_$0Bg!T5%`;5x@Lt)`I;#t%`#8@Ld4?UzeG7!oG%^She#k_&H+%6%9dXueD=G*Q!A~EgtzE
zq7Qc1dhBGYoPzVeal`KgV16dRa$~m-w42wzH1kT}Epl`eTC+|ho$7fsjL+}x7&$wtG@J#v+YEc
zZCxdcToo~E;A4^Z%q2O=qBYx&R*|K11NZ)Pw~}uaeX$0U?PSfzq>4WOW+K{=wcFJt
zAI>d4(;lm`%#JE)Hg)c
zeR)1%c|&HTxGg^KD>|e^rsw;&(e{Cv==YGLoA{JHfpYVy*
z_o7RPb*%Sc%@3tSFvIfOCb3F#(Zp#+(zcph9#JgcGFd}3_uiXN+fzo2maOkP{kxP5
z|5*Jj%czPR&hoLX4yqEOI;kF&Vsh>h4iEcO2&F@oJlFHCfkY
zkaxo!&NGwbuTC1}Bj+Y;?k~K((t@8IIX{u+?)O6urOOr
z#v^>4%S%d$3IF$rGV2=RoV4?VLqj9!mf(BHSX52aUcSD|@}h$5j2K>fQ@@I23+&63
zHidfG%zR7HX(iK~9&O8Os)@t*+Rn1&Rb++BzEf#6wIt?~|53x37P6e!i+td|C!ZbxfB8pmJJBOf3yJd
zZgB6Y2S4Z|e`HP
z`OHAu8{|uCwHUAA{snP(l7aL6^>-53v;ou9vMn&a2Qt~XVL7#X7Ej~6y-jYO8RMTb
zWpflVrwCB`&7Md7F#llyelP;oM;clvUbH%a5R@Ox_^(B5
zzDqBImg)u$Ju+oJ}KbG@@N*BGdNXh3Ck5ccA9qc?r
z$pw!6Dp;;YoD!9uY6iz>LFQGgrZCwSen$uLZgKbRh<({@W>|l|`4&c$*YK^^LwjuF
z7~jd(QNf#Q=Zk9ruO%rR;B9{Fk1eKC2I
zJ*Q2ff-p|fxYi+xIx^KDZf1PIiT3|Bn3%`!lKm+^RE+)Tu27PME1EjFWZrCM%ZsgR0KK@&GlQT
z^HvpSUT9wc`C-G4C7pmGI^Cr>&-p1o;*53_?-jPi`d=B~EkipFitLj!0A%Q$UWMuD
znMu)DuPZvPcd=YGM>|CvN5Abmf8aToLJ^_2roh{hCA2?`0*_Oe*e4$V^SyJ;eBoMp
zFrFK1Gj7MhdZ}*amyh6QzA-l*SqpxJ>-5bmRj^xWo$t0Zz|P_)T@|+>|Fpe3o}W80
z^rqJs@Gn!;zj+wPY8eMwgyH(2zc!^D8(_Wkz{!ChV8_X+Da4a*fpPfN~ul|GKRN`^*i_^aVH5fS=17
zT)P49XMGj&%Xx9Ion%~lXX4)8LfSv=E6L{1BaMQ8`Zb%uf4=+Nn-6^G;yvZPB8Oof
zwUD&@NeTFa+&ej%n;XfHXqwa{%xAyksbn0|2A-|%q~5;@{K1_-xs^&#Kc?2TX0T3E
zL_guc3G33PM|OWV0{?e@=4{g#_|@i4!E{B~&ly|uE6y#djSMIGS{8zxyM3OPZG`nD
zrShq=VesRT^qS2Zp#9m8EDH{eXoL4HjJs`R%Se3XKMS5GWn}(`*N$=3DiXGC?d`3a
z)ucxy`riFNb!15L&!-uWTJq&#%zIV!3L>}aN#N(93Ub*X|KB6CIjBVYG$F|LF@hOo#W@bmjNDK5is=d1YAw@I3{=%6e_F$1soolr<`E5Ak5Kggf23
zk$lj8(9Wh&N1U8$GA5L2Nl1Wg#@Ai-#9L-CTvxK0MEmTfGhEk7@*1AMh>~a{46z5l
zoQ|p^B@QkBSmPSVeZ5z`AGS3U!_~XTW?_A0O?qr{8Jx>7-Y8R_rO`k}|
z#S>M(BwL6;9OsV>VQplHPW1y<5bRel%fF_3TMzq%I|_c-HWHW4X)jJV2l2Xi#i;Nd
z;P*{!DGUQJuZuP|UoNtZ%p
z%R1rzqG5vZy&Z(_-;YPq{V=asDWo{}yOo5tn!3ly!Frl5-=GU;7un&-XY^LIld#;_
zXs%?`POP3ZTnRhZM)tdwd++vaB^5=Ft!It8h=vWh!(!J7U!|YXYkJyFIG!8nln=mp
z-wi*v^!$W!m7VPaWxI&343lcUdMBC4R+whE+fKwDoG2-(X(Q%?iKCIQj(7KwsbJQs
zE>fWpFjBptlN^63plImWPChRluUsAj?=dEJja0?A68ncOZ&rB0dR4<6#hruAL@}qK
zV~<$_DUI{#cUG(=>-G-Z;o4D6OuAS(EhFJO4tl*AZns*`Q=x!!EoTs
zI{n}KI;+Wu<{g2!7p=s)aq$^5KfK4sD8{gDRTD9KJ-X@VUwHnV^noUE^`!GjjQ7o!
zR?@rrK`32!3!#%=$4(NO$ODB>f6m)ClMiBFFYk$NA+uR#E41PJcey!FuKliRA+0*0
zWf^oW#PslSy4y_fJq)gH$9I#h}ey3JalEtf$&yx@@?H4E-m11^^qPDZt%QC*5^;%gz`P@C;pVT64@n=
zLagA28itOjSQx{;o_N9^9`J)+8UYd#Fn{b+b{`vj-$uGm1}o9ofPa6`Sv~;yF&Ejt
z-QkD(N~PEd!*}Jn?s&K9TS0uTd2zzf494uZI@R7Ykgw)EXW@}cDXe|zDj(4
z2=d&a3rD`=ddaEnpI-|@Ib~C?b=NGT|gmK
z^CNbE-mm@oBmRJcr}|%(WZ&9`CUWMy!B`CSQHQ{ki1O
z%l623^uHdcMZS1C-r*ziohp$p!{-6JoO=?L0T~+CJi~HCM#MiL4^T~_GeCZNYR&W@
z@=5j3-|*`|I8Pe4eg*OtyXxmfSgy07^jqWu);caexKAaq?Yb(i^OetrQ0Ji6-fj7b
zd`Z~yXfX1J^Wyo3&<@x2r-$)A^=V4fK3u8ud^P5ue#be4dZYF$pj?CLCK$8@p;<{uW%r1;^%`b<;Qvs!VOI^QE03$KrX%>42&T%
z;PbyO*h_s^h5z5rX57aNzsIJ&8E{_nb7Cd*bF&LcLa297=4BQ>-`cYun6dmOEn|1=
zANw{wQ^fwZ_sO3i%;%Dupw92Z#~{j3KHXbF*{?IO+a1?m9~&8|;q^Ts`HYD3NBNu)
zO;0U-i~6R!)gwg#om6}1*n#M_fgUq+3;))LVu|o5LFkBa(scWvlacW|C)(qEE-8YNK90GlgYvnQo
zkc-pL{4k`S@~fXxyKr$D_$_T?wkc)sn_OY%
z*PMjwJvYT%i-7eE&P{9w@jc;=5u0u2p!|D$aznF#qw-^`v|-*J={_-ve#K4bd}jn;
z)gjSk-=JO}dCmvnKD6`b#%NdYpI2vce_R6npw3@`g^<2F`;)0LpkWfr+Dy<_srd6t
z3+yl#P96~k{(Jen@W}1pR#KL{Siv3v&pS*pEZP_5AL_cD{L
z&pViJyt?6_+U5-9t4GcY^uT+ey7w<7b3%Nt?r3iV{@s^ptcUyE=_~fs^DhOz^y$7(
z3=6ck#jAyRnD1mizThPV>pRiI4p(16|1c0dEE@;c>1}w((cIcbG*0oDOO4c#=zW99
z55B?sTUL*3Mdn)w3mlPl^lF3eSKU#!=4vM&t-1?S?=+LnO#L25jwZ5$t~;M0q8YxU
zq4|7`A-r#;I_kR@)}h9%e14HS@KZ*2#k!MW{pWIn>#nV>M1Dq7kMndZIWwOBSqk<6
z<(!jhb)j7z_;@Pn`9Z%lpH}rh1?LR|d>o#^IoF8SQ8%OtJLB0`R?E0Z$LXEAaiW
z$J+}o%)`0m>zOt5oTd;L&OT&wxew#(_JJ$4(2pz+`^MO<0l&M`z~Un8$8&G|ne5CB
z>jFjETjt?8sjc(9bOr2OZ?I~KVSg7fSNSs)zSu<~HZ#=iaOooV2c;M6{&f=T9FJ9x
zeL7+NYROB-x=!NzOkla(uTDZgos**bq?4R_@gY8VM<>zSuP~I+)=nDUam9GBc91sX
zPj<`pbr9i%X6MI0+DQq|{?-*??Zm52kEsdz@6|>bzm-|Q52GK!F>l~Y&%M@+hQNM9
zq=QF+Bg9Gf=%IHoPFmy+zL$dj6v3}}Bv2jJ83jbGg@GT5U$8f(gK;Xw8TW*Txr
zq&@3g+evC-EFK3Mwv!Ym*;J#cC~|?9U6I*^cBu?
z8036z6>lTg_H=E}zX;#GOBodF_H8E$2Rej8INC|)+|wTipg&mm{VE&QXd|ho*Jx)-
zw~;A^GOlv4bI&%h%vkUrD^nTN%EzJqSfs9NhVKKK?S8bF%me;D?-IW%JdfiZDY11M
zf#($Z&vyf+y%xRL5Bn$CRquAH>3}>OVc=HMO4{t7tbEekLQ?&s#llv$5bNe%`?tXF
zm3~L}rj53e=rb8xa*4@=U_gl99_TW
zAG{xO`oNioyLH=$xS*ljtKl}1w0R-XQy16Mq(3R@L%d)(JZHcLakcm1qByj(;pC?L
zKuBNkaS}6vaUEGvsSph3QHJ|g?7VpdcxVN8R}bU|7g{XDfj@Y(iQ8!wc(#CQy-^G3
zRhcVPeuwYwW&Rxun}PRv8T#C=vz!LMJz-*x-}Q++nRr{87yKN<7p4{Pp7Q7kmS#I`
zI1e@4=*|KD){#Redr%PkQK;eUZ{YFj&S(ASEMWbNSx~YF{I_S+o>|t@yaeJ
zcST#QAn31GhP8IU)3H(^BhX&&?iR&iT;mN1Pr>u-S;2g&9Uw~|f5jIE`M)>2Eh7K*
z3v@E_hx~=gzKIx++dmxU(*WczOK`z?!5{T&Q#T<$)-lZF0i=iMMX$zv)=7iP1qQ$s
z`KwPdLjKf#c9EALM@>21KL%(}VA_Yg-pyj^V>yt0wd=z0yzmA3;PO>aUwet$SC2!w
zOxpty+=px-&*@D;j;%`Ac?RQk0DLCtoQL^_9$>@anmzg4er$U#aK^%(8K&2q3uIh
z{;o$2S23S%H`g%A5+PeD{+}kr%SV;d)kg8qrB_WIP`>>;Kp!#A0X|a;xOgXv;;XZ{
zoIt$cc{tL(1NY#>v27^Z=*>{qca`j>>iJbDi{jI^?BP>*KZD%+lpc2}JH@wKMieMM
z{mdhlx^8u;EXCt9bRZe@*&~_CR19IRuT7Cqq>Z5!Eo<3sj7uvjmgCm5J
z)!^+-OdklNmGc;8c_<%tqv=oNL$yJ^2CaqYW=lhrBAWh|Dq8vWYr$N|PlVGGh^m$}
zJt=q^D35f~@cV0+rhzRTR_tlDet(dzABlO4pSvhXTOg
z?`WZ9#+@rEnX#Yt-@mQbVEO?0R$WIZeBipALW_MLDYRV2MVpUQ3Q)3LIot;HcZ@4i
zn9+s?g1_0}Lg9?>VJhAJ92^UQ^oa8`yJxk5zJr|BzJt;$=+yH>S=d;GdTzfMEvWsF
zfu1Zb>~Cd5J`9LA*oxrBu#Oq<9Bv5d2Ir~%^zI7>v@G=hl`s2IuW{>Zs$J;bmr&?5
zcadrb{=RT(9@ffun)>eI(kqcveUx6{0|fsn5JSz2!qRB_B#JB7QsZR#voWfFO;l7+
z{j!U(h0>dQbcRaL`|%7K492;jfHTg&7he}oi32u#cya^xn`~`*ir4@P43#tSJHwyt
zXD{G*n|b579LGtGh4n)m|LP;>IpZIL3eu(|KwZDHD(?eJ23!wb)l7$xP
zGmFd6qrd3w`U?|Zn1}dJP~V;56IsYY`3kGxQ!IaTb@*rWYbw?j5twecHkS=iBYFuz
zl$|@i1l!GTuZlnFcW^$}jRx#_XDjjyaLf5kbedpyVEEsmZt%04m5-Va!@3l2y0RPY
zdx%?P^q~K%%gNn?e*2zVqMJ47Pg_Ze&4BDMl5`C1{&nZhIb65kd&#y8{j!GF6H~k{
zFj?AEAM{^c^!SH<_O6YAH_o%RxGOQDe(ep%WGr`o&eyXIfGat~rqTZmz4~?@?SB^a
z-4XQ-S%clz0bXN_w83?v@r8-ebztW{8<)KAFpqNUx*CxHyv4%$OyQzD^pm^ct)UI@
zzJhYPvLo=a^b*JD0JtucKJ!C3#IY+6Ci}jC{SOi>9)q!}#VidZ45bkrn
z?BK2i(0^jtqYLvKa0e+_UN$a^+N*mEqvnsd(JdOSS
zQ~p=M@IJ^djRUoS%Q_yOyaewJ?tBt-@-@_JUkk^ubYr?HTR&sONCUGI6e1HZ``Q(9fQ7vf~R+Oo~SvxK?d
z^Xq~=^P|kB9>9;1*$(96`+@35EcVaBe!UFu@GoPKwIZ#>>!7{kv_DxchxYyu``z0K
z>1#lo
z3Ys)>g?d^^d%5q1c5x63kIfNv^Kp2S?t!wQX{yo
z%;%xOa*#ulf4kv$dnoed0wdU4=eDOd8SLfX_rsPSWSz@-S@=EjC#(}wZvj6A$pzuL
zY)>nd_v`s!-D)K3LL&5+-v)dOOQ9dhJYdms(}VV&Y}50A`i^v}I^p|^ucQ}~PC+|a
zSrvgaz;n7
z4E3;A8R-s#dR|!!QsD+W(-HzQa=;#@DDM;8kk2sk>nff@3wE~+TMgLnE^p-nYDX%ou!OlfFJ0ctW0wH1@(x?%rZU*cE5Ib;#~~+
zs)e15IR3vFeXhxX{HlW)vyUM^@FJUcFqC8V#>Mv#%mZ?E@(Z4T`xvGdI^elgV-5e%
zX|PW+W{L3{(4SEdIlC45HwUx!*-uceb9?prTuAT!+;j*1$L+!M`(vOyIkApa%iw;8
zn0$|3fb^FOzx{Dv`j}PGUpL5qa^v(f^!xpDT64W%j|FeLDAva(y{OX(FyCCMn-fs>
zlJ|@NV8e!h-J^gihi*PZ|90l^p>x{+KdLGiV7|wVxH?=%Iy>H2CJ*;<5HlD?Ke@7U
zuem0ms(``6IY=+4wDLiJXUv|RiT0Up?O16K@*!VA{xyJJ_xVoZ`e0ttkq|8JUsqi8
zM!=lzU}GM@`d(N<%xNi9)c=s$`A5}ikg82p8cqD%wu{!nhwGY($6g?$tlGOf2WtJgXbZ)h-gwB7lwi
z-Rw~ov-|}M+2Dr?u9u*^#UzpkWgq1Qek^y@!$+I(`VtpgAj}xY3Ux8xn#7nOMn{<6
z!u)s_wKUAH@H<1DrwubwGsFCg?C&`d4Ub)(MSD*hAF)8az|BglFrD-4Dpy2zzK?U5
z?&n$7fboAFS5Z64hvRvu`~2zVqIg%7)H~`N#=<^(ithxgE9v9)(pG&G->ljTpG^Zf
zddi66-E~Dgnkd^1{h+S9x0aR4*BI*p!VCQMoEpW?9GjOeMfuT^wC#u;>jWvjcI-TD
z9o&@u2zA~3tmzlD%V(^RDzD|12@pz{{~udI@v!5OGZc@D`zuTFv&xq#R5|;`vZ(7*
z&vj6EIpi$G<4U4w-%U>HfzQc7ISGtp~{K*_>GdsPSU<>%~}b|(x6wXeVxj8
zaHFSC?+1ABT)6OnCvtfV}?(0ARyAnjDr+L*|>b
z`NcTWWXK=qrXl%nJ|N~#(|e&xE4Ko(K@TwSXlO41j|ub}d9gw;{_*oWfKy9p>8sd5
z1o<5FW<<3t=vIK9cC>j{V*#xkt|VH1@PG5OckHz552XQZ2K{Z9FTw@j&qOq+ei^$N
zj(dUJGhIpb>)pvWbWpy?*$4%~^V|;GSb*|Aoz(R`ONXd_yfv|JE2eu*UZc{vkJ08m
zRd$dJ>HplnQ|PGb4F=#j>}8ZZ7=DOC_eh%E%6H-Ohme0E&6P^GE`wuBAdjo!gM#(l
zio+BZO5p>7_TLah*&DE`^A*bN9{$iWunwYil1i5&1J>v#gijmH;`kQ0S;>d_Ax{rc
z<9Kjva5c(44aU&1f!D|!*o^*S)u}tw^^dv1eS&;MP>33DUv^2aN14;OpBj$}2bMt(
zfjGBjIc3i?{b=gCYav5WF=*Fg^UyJXZ=`5Z?YwdePN*TS1j$n64sgP>1>_Dj4{O9Q
zrKdK?@5@bC+tA*!^dTYi=Yz`~EO5S2SbCuG9mJ9I$z~3yAMCZ2`tImb+b0Ln?w0ab
zjaXk^11FI+fD9)3mr>7P_39^xca~h%M?Z2SPqYv9nzo8Y;{3w=UD`CJ&s-!_dpNm0
ziAR0ux)&K}$13TlXQF_9>m|-(I=?d$6va=M>M+T%ADu)_+IB
zQNKDQDhTy_TpyNS12pIrl)`wg!DRLx?e@-ze1N!Zmu4x(Tlbc!I(ESPgZp2M0^h6U
zF1>IAc*u@7Uf%h@i{c|BZaTs~kJ*=TrUBsHJeNKwCqup5jr=V`0rihN=hXwRaI#4F
zXAh`*)#H9B@R}_@BZVBm&Z8UR{W+j~UuXRfHGsbhKIaZXdZ54~EJ7>9@SIqT{{U+G-@
z>`{>WhGNEz!Omf`)%(YRS2eS457C8ua_L7_64c9zKGg^R@6Fr0b>9uJzj)cx@&!Nx
zgX86o0f)F2U1gvgJ@d_n@SN}01*zISfCHhKI(S~TW8JV6@}-8AanF&@B<^^a=MLza
zUSL-YSXt1!0n@Lv*g9dqto{~ci2Zr3o81AFmnzLLAPN=D=KFCWS3~2g%>AW8xOECRf2SA(Cl{}{b
zdBu}%tp(JMXbnJn+2%Iu^8+4wGs1@bNbST+!VAdAH&un>cS}8!As6CWUNfxc$0ja2
zqKNoO6l1}5%
zxB}$M+VVXXh3CQ5`}9jE$n5ra^jHDqv?5-r0A9Ky^v@7*(&YCsCO~ocS!fGfe|eLy
z%orfYw2UQ=bFKYWC%6Gmd0lK32K+HI9b^KSk{2>}63|BEem&NsJJakmKKJ+94LZ2q
ztIzd0TnUgX-=rV;uTx3*FZ@4Dz^CNj7s0OSbK<+@pgyY1!po80j~TT#)Pdgq>VIxl
zfB~Ev_$?tlVDYL4`mu%6EAyT~`iYfo!ZbT9ffQMhh0%&t&}qAysuk&I5P*_qlh3hmSJi#Zi
z{qGOyb?yT64&5%R2)O&r)p_*u(_yVi=x^1BUENX7{fwg<`X%eD?u%HTfP-oWu|97X
zgZuINgx9=(espT!1zLzo7R0owSR*GdBpao@BP1$10_kUHnWcPDNo`o$IW!!vl@
zn$)=_EYCG^go;}XFQg{1oROiEPx1b%`qK{KeR9~e&tm%gw-teiUXkOUF@I5eITg?8
zrEPX%xhs>_J7Iq2(knkO|LFJnV<->a{jP-Q8*Ubd={Xl9G7%+PZ);)tfePgVh`kNF
zf$_n5!j3FCv~T~yDr)^q>it2AAMY!0If413LA*df!H=!DDvsscZQJ+&+?q1N*;7
z<8_-XwW+vSbhej@|H+&OsB%6yq)_+w4WjYye-@=w{sFNa)cs|-Zd3WkSpMVdQvFnZ
zZS_sc&Z0CL4=K{wN!h8!HA&?!IeCf7XJ(=0?|KeqfE|B2&QRt4Vtz--tJ9b$eD)9M
z4&-Zg9Hr1yGnUe;^V?6MXT)!+T$g@Y`!UnOV+MVtF-a=F%nq6ZbnYnr@rmL|KX(RecFMw;A%$H+iW!HS0KkJI$C&(hjutsE_V
zGi3nydwdL_pKd}!MWm@9KmD(r_Tl3N*$OWQ%%Lg<{uo#$%EJGpxBpjO6+Q+?Z^g$C
z_z74zV(cD$3a#5WP#9@Td)~MIvoly%8cG7cQE5O5eSH2izFef3DKz;jK%zQuVYoP9T`a+Y%G-^BdMep4RA?c4e}
z5fxV)u0(%)H?^)GW$964Y95nB&!vd+aFE?P#9h0+Y!EvJL^4pH&w}9McZBlfzhfdd5otU5&5=`~SJ*`s*-H`N{z+j3w?!TL}7YF)&wLbUYGf)juj_wJ7`f$MmTAR)*2N7syYHl;`U3y0fp!-t7atq#MD8JmK5Y^;}B;KXN`U
z#eA!vS!J}t_&Qe%+FkHL;4Gs1o2~&tKpk%Jm#BBY#lQfOry#Kw_4UU)1*m#j{ydKH
z{e{&Xc(5=}QdkEb2G(WWCM^-^HAcZe=pO>H)IQlS-f4GCZ;R}Njt~9m)KhSvz#n8k
zQT3!VTy_-o1tsU6vjE=BJ-P8BU@fmR^}lDil}tBNfhXlxU)q7Z;r9x^O~_aF2A-nB
zcy8XBReTuGMyVqk`AgNU=DB5npVe1hM!w@)nM!@HZRd-@RMfX^FloSYMe-%RQ>zu4)Te&0n1=$17DN51w$Qpw9lj-$_`J@C7|BFiS<-`
zA6@y66U+agxg-teG5hI%XxRcX9NBMl18_JYI|X@dZ;ka7@?z$b`8ulscg}0CM18vG
z2M6%Ec^2QNzC&Z7sE~p@=-lXAn2^IdWa_hPc;ERuR*GnsmR2-u3(#3zQ7mfCd3uL-aSI>JS1a4U4LekFk+3U$}_xw+@+a%Cct6l
zcd#G@^O-#PS89MvJ*#eEyW7-$rbnJ``uKS@_7iKxzgMvT_;BB@!gAjjJ5xMZ_uv;L
zd_Gf}W-Vy%>TR1~qaNDxns_PJOWQ5i81;?y{TeadMZpS)8_ch2%fLaydVu|1C2WT+
z_68m(^A+Y(-;p?R=hbV}e|5MPCW%pq)ECDv-;~5bgaZ4^^eF!sqV@Ydmbc0CM;oRG2j?sx
zMv5Jw{PpSkc~pNoZb7Gv=|Txt?GQH)4^!72-*uGo`>UU`Q_rnfDUGS&7$%TD
z@%ASKM7XZzA!WzT-g{K~HO{Y8Ir`$&RGj$TF$)X^`qd8s>b|wks#Lpu^D&|H6f$Y?
zbE&8^)!x08?3CW~5KHQLMbnK_aVg;`gcT@9Q*alRo+}YT)z9YTEvg(oH>d{WZ(ZI&
zp=E$Gb)Rm%2UPy~lFd~5M7k-3JNC0s$o#XMvQx$pstwndj_6SsX0(<Epo#>Ux{qNCbh;p!)*sgIOXX+fo3)v7)0C_J=^Y2iXaK
z=LpDP4`u)wE}>!WS_s!5f4xkzTN)n&$fb(_fSv#GaZ@|!9w4(Q(e8U59}CF7cC>Q*
zM``63iqPyoxgDAj^51aN^6mcPd5(A)q^Dk@rJGF9WF-a~{ys>6%EEdD()j~P9*ys3+9JsPJZ8rVr(5B@&#O&R`|Vj1eG%H2uwn+HR_G!dyB2IKu6-pQHyLSVgqdo=`UBq}p
zu-#pz4t~u
zO`U|fF8+z*5PbLGcvD=CQ#(1+b-i+zRVz8Px;Z9V9R9l;dSP()hH~N;;%S>vSWX_U
z9$!|rq>J26IKTbn#|Far>H^!$x=NC5>*`#iUQS#NX9q5uEG9V@Gs6ZdKM;`?hD)Wl
z%7JG`7|D5N5bd#XRtC#rl68M<-KBqT$SFo9k64>lV&@}v#eGj|aeAP;`-bHYiv>5n
z`lj)*gg7;3N4a|DlVrZuFI6GWi|vFA@&r!Q6tf&(qEYUfNA3tsnfbkcO`6mz#on9c
zkSULbPy21kN!!PEl(-g{qEPsGl!^4%`UnNrIVzVRROi1Q;6;5
ze^I&uY2OD5+u
ztu;1xClNZO12GZiN#xhD#t(D{Q;C24#S>$$Dd0~p3SW&$CiL_R8D}+;NZo?z_|(8F
z@~lDM@{m1zcRn$;Ya@KeK6!2N=r@C8^2_?e7?0yi5`Ai%JBB`y2%ZTzQKynZti|mQ
z1{9_cvj~n0T<4R>ccULO%oh{LvJD*J8`a~<7r$+PE*(!Hifgvpru(H3kA#%urlMD5
zDaT;#&dGQZVZL;tSRtPHlW%J{&Lk24>e_p|CX>kI(Z7tjvMFTkD(8+y_+EWNm`(iH
z_6*{F?_*%V$5&+i6G>_NNAUgqa?Qdd!DNzb-pgA3DVg*xuBmD~l1c<3eEYq+6Nrt_
zR7aCU0{Jo^khQEeo)}I=e_gGR0RPXm+iobEK(wVa!d_*^5e-g-S~z+{4!W5%uM%|M?Xu!Xv)2H7&c_
zxdTb&zd`nEOEXA$!`mIThck$|pin}^)eJIqd$WJj=S;H0&5VA{rfd@UexZwx$R%r4hBNeLtPM(@7A2hT=|cHAweq*G$Y|;BQ{MW~q@i<2NL%1Jab=9!rZ@VOXgM;lIz9>`mw9Sg
zH1~%Q{|VuPqFIj!Q^5h9t($zwavAC7>Z9&N+;bb#EQeF^wWPNkH=kcGo)vzoXSq3>
zF#oyAFk_iRhLvI~=6~grPm6wO>%xkOvPb@u%+FF1^cx7+{!DVb<$1loToy5F&x|oD
z$|YV}y#LfBUz4Y|JftqWm5`)6p>~3GugI-MyQA4RlZmEkU*vdX22p+{G-uh7OA>h}
zHiO5Y4W3U7v{*66}05Cbl?(h)Y;;{(5+-
z*jTo6`qwdY_@6@f1%WGf2!GY=njDV^(($EK@~KHIk=*~7(Pgfbyj*wg#dF0nV&NC{
zw$uu~Qxw$n@Oo4wdE0B>&(B;js$E@07odwTx{K+ExEH_L|8uCfgBioMQv0~D(mTvLl=ThQeXt+vWE|#RO?O3fS^@@Z(
zr!zain?-aJ_LfC=7LvKU-=>dt~$rMik#kgAzvoD=6?*GcOaEt9L5omy^2a!h)QNc-gE3tA)}C4
zMzUq}Q$_=sWfVg8%pT!A*G@*6*)p<{nGqGg`|thZzURL8z4tud@A*ET2XY3wIIwGV
z!*P?i;^&1uFym_Pvp?>=uqx*Mmoq6Xu+P9Ta`wnan3EAEAfYL=+yn2vynoG_=r@|G95^bb(*PeI
zAy50yx*3uThTd*D-vI{;82!C|_dvJB=Xx?@eel$lfp)P1k(2DI`_A=WFH}klw1}T2
z_7OV!H`0iFfQlDd4=ozMLiQzUi#$J~Pb8xB>2-%bnA^=9e9f8YA956NK10j{rb~H1
z5ob9B?Wj%5NDq4<50-Moo0r%>*x1W&A@UlIGAd~LQMgd+Nclc+3~pX3akXy!3aKfNhYhn2!QNM5{*TRvV7}1f2ei^d@V_|t
zd5-9h`o!RNCfjO==tEEm`uS-H{(UDUcSCp>vSzF4(wrNHKax$nZjuheD^YhIO1BQf
zPAPKdWkaHFI!~rlb!-%_PR!Z9Q67a?&9KOG@O9kJS&uM&;-=kJbs7Pd=h>P@ue2MHwlA$dKO;M
zO+sGHT=5#GZ;%!*jZ&Tb2Dc{1;yF~OphVixv3q(`(Bngu%8bAiB(s^4j(Ri%-#@Y2
zLjpsvbe*C?Rq-q2JgPr1Mf9B|k*Es}4G{Zk_X-cJiR-*Oe$H$_av0K{y%MUaH3Vb!
zj@5wHK`7JDF&i5<0HtZ?D{haBLiiif^S&B|@)x48xVRDMW95QN;clXE2V
z-{4O8uwd@bNhql0(`FDb0ayKo9tYTuL$Zp6_rm+czQmige{0pmIlH=sB+(jTeShTs
zLtY|pMKiVML%Sp~Kd4S}*@KwBeWF%l@p!@j{76n;8)rKJXA{43SVR%$HsjAY?57cZ
zUT->|E6sm}krj_p-q(DEg(C_aRXkszK$BL5**1|=V)Ok?!yd6e{wRK>o0s@`N2C%X
z!$+Xa-G}LZw!<)zUYFc_hWLCabJm_G{yyt^8C;LJjl&72>9NAFF(}o_{Ly@L1ipCE
zT7Ko~F!X-~-Dv5C;Oe$`4l8k;Tp?HP3o-V>^|V%0oB2*`?+f&bGL~`g?u1rT@r^ty?U3{w3;z}bo=e+-xBpcbr4mzbmSl`?w@)q8Qcdq@=KXciGLw-MI$YJrMlqk
zlytHP0
z=H|67$j0_!EjOecns%}?&g?fq&OP=nOP6}6?mtvJ`?eZ(jZvJoyWR+IGg{@IIM)EL
zvfs=%4?fQSY##ajEWS>
z6t5%BCyWbBMpi?PX0vJ!SO!;?bvYK5i{SjlwBl4eAI9T|q_emh<~}OCB4|?qoo$E?
zeb+)*`scFLFkL!4OoTOSor&-p1;^jl^R;l!ddJyw<^z=b7;o+nP)VHgO*m1`S^+7}
z|Cqb(PRw0;V5{?AX)Ub#He;kwOyolb{Z8+vs)X!O*5U6D%3+e|#D?sN3b_1O^nUXk+XE_`_p-856vkVH6sh^#qErY9X&KNuBmci|BPuq$g6LsMGjjrWnV&0a*
z!yo%Eb6|btlb1KEbKqW9%EjDox$s1h;RBArSm@24S0gKt2o)ubWNYlx;N0M!9s`FA
zn9XN7qp_3;`Co^y9~`%YykFuy6K4b9^&I<=hPz>qHqzwTtV1-^bgA$U6^w<}F7wji
zyj4&*|7fbknR2*N&2jpHUlC;1^x@aYDS&$*w6bR<4<7Io%icIu2LH-v#f7O9L3L%)
zBQ?1NuuW{vx|68)D{7EHJ$%$Oyf2?xq!3=4&`;OVK)_fGncnR9e3
z40_;5uNU(gYOt&VCWE)o_Ge{J%S;-a4W2r6CNKle*wP77pU#I)3AP&~#Cj>j@Y7?X
z6YnAaT#s%#*?V|F<(2es%6oV^@}wk*Ch=FNUU7R36Qa|07|
z;UewtVxh@AC`JDDjdmk3r^F21e}2CRPIf6Wsdkk>c8Y7k=F-{FPxg*L%eiuhqH2P;
zGl=!4@f(x(tEDh|^zUM0NIuN17cCJcD}k5c{ok7tWw737x5DH|8Kl9F(r&od|wUi$Sa{|_-|mbQ;hF@L%U5;x(F=88f%
z*dEk9_W3;=)*qq?NGX9s=bvYWg%?1sd_xf`Vm_JJkn~0dXE8JrT-@BeRs?w(WS)mj
z7Qmib3Wdm}YRFse-U9+l;im#>7roIEIA!(UkeEmfjIR`PZ}+c-tz30Wo6E#HqAx3R
zET9M)==-#!Zk0mWC(+IK%s&wQrTxi19UtJ^r<5y@JP#VwoX_!hEP|F&B6~L*%OIsk
zUEo#9D(G{q`|rW&577E5&yv}A2^5yo=s-71;n62<(E?Z*r1Jwy(>7!bD)B!8)
zMkZ>D%i$tnhdmkAL0Nk7tJrV@6i&>$;P~wme7}~%R!rnue2U9{9jaOdBZfQ=elS!)
zk)!v~HO)r&)mv48Ddrkk_Nb>s@jrE^S587Oirg{s%h`LR000YVvB*&o0Ql
z>~xu!KJmUD^|eH2HU-G#`h`Bdpqww~sJ!~0+{`CAmeXa&r=M=rw~cA-anXI03(@(5
z{nGeQ-|SH6Ja@ZYu989`-aWaeo~f|OgabQ<{p{`j{;X|A?g{?I#=>W){`#kw-0=5<
za(kv^GPAC0^Xl$7s8io9qe?n*7HHynX|=v>gPexhy`_G7Yx;W!^G?0L)W{YgoCC9HzPSFklm?Fs
zr1+r+s4z#J(+O)=crc2OH(2;O46we5_g*=_MX~!7qHh%y1n{ptuNzvFe8k_|2Dcq|(#h))#(fG-
zHOLw;;-_u}N(anwVZjgGj()wzj~@};^!}`ODED-e@KI{b4~y9iyy|7GhI>?9HCqq>
zSd6AtzP7v~erQsDFfV3=Ei2I5tnfwSLJp4!t!F(k+6HBvFg_RjA?@6?_0yhMw`&{m
zeS-1UIS7W@X;NaHD0c-r*_r1~fx;Bh=
z>$3BMhF6$oufyebeivLo+wPN5ggefAcdpepF(3P9xE5=UwcjF$Pw&Vm|Bk;)DzH+7;&#~#$q3yA(NE~on
zTcsrp#``E)ADxrvz*4-uma8nPFspTShofQH`0#0BT;9nf%q>tlxm>>s+Y%D|tza>Z
zDUM0keC2M%HgpH>nI*U21@V$?E-!nrHs$ySwi^@6ef3ynpab`5
z8{0fi8isKQio|O}3nlNrSuK5k64=;qePZf6i{6-?pU&J-1xNnV7w+w{1(O^W%lBfG0Y_HG
z@n@YfNb3sI?F?27JnBFBBH9Tn)IS*XYYEXIRwV$
z9zF-5+31tZ>PF#g91xjc{6r#H0QAWw#uI-up`*R+&dyVfC?KNBIL+D*&~>)`C$y0b
z6j%eQ3W$HN5qhRrxxG4&S!wTFF>3%R3zyx)y8OYZ_cP8Nb~WhW4&zDko@U^mVSBWh
zya`Odm5ZI#sYJUU$YYkZE0C7`wwkoeXE2xQyO#F49C3`OGpzS7A}(ew?dyW0NY&4$
zYntjEs19q=8hg18JanW4WFCY*X^i*R@CxP?5S#g0}4#@vN6AriQM0sE{ll1MANZ?
zQ5UEu(12EV!o+4XqNjAs``n!Y?g^WP<|bDHhLno=QTaOHv_ol@>iioR)O^3D(0T;Q
zWgDAbSlI)GADJ`q_Id&HhQ)8L!cx#iF6u5~pA0UkU8NCsN(D@;M=$y>JVrkQQ)~ZT
z2|%;oWx`k4UBObL;DJ(X0BU;JS28*p095fwV7
z$)idvwrPOj<6Dt*lN=QH;4WneO$rJSH{(|GvP7ijv!Tk(YsNpOrYl5vL(z+8b
zUS1fC8d&wXY5MLUNn!mux32qud~#wkl3x=@FwYO0dg-G0h?1m6${0}5$X3xQcoYO1
z=%`ff`=X_2C;tyd(x5!zzqJn+ZUcVS&sS|Pg8A(ic1U(T)4fo#f0w#3NB9wmjj}mA->i0o>NEp39t@J6|)IwDq?UpJk$8*qt8uy{-RFaQ4ZRnEsYsf8D=(5GUO2mc8@uB1W%h1@A(Crj`nJ+G3V1axCidiZrl%Llw<
zBmMDJsRYdS*BkLu<#a%uPMa=7U>Kj5uFm?l-G_JoRB;v`YQb(hCy(mb!9z-y#@f%#;T4Og4y!~ou)FHqH1Tu_xE`y~G|L4?97=|-ce!82
z?|B^IrkSzB!pKu1DHv0*;3(et%=G`Td#?&eUryh}{;l-+(DUEI6=)a2p5fN`m4Umh
z>Lf+Dsg){3}dqU)noECzcBt9at~0zxKDyo|k(ES=Sujo>>$tmDe{D&=o$2}adZsh4*FLFZ*u&?f{-&~@
z+?=eGpY|!E^Nmq^t&MoWRT^N%krxe2I
z^6+D)oI&@6<5S~(^(2J}h84w&ayRcKvQ?U0!qSCFSEuHG)N@^JUlX@ol>1rnhgLIg
zwmwzc^2M_n5-cv@)0v@M2|Vn}fI&RZFF6N)0j5vh>-Cp>gAMNr?8=qi{Ak&)MUB7x
z_NGfA|pCQji<)acZj}MnO{IIoc+MW{Le0Jx-zh^k-+LzS5H$iyv@&BYq{XMY@
zA`{Csy&kxE!{y&`5Bza|5%V2*UmisLNR{MlSlKm%>)({wv<2cwHSBd>xvcj
zE5Y`Pgdf@LRpW}y>_GNu7bbpAbA0;UFutZ$B}hmvuQan$`>0yA2>tr*Y%O?4A%6$-XaJ*H@#ITA>{HI9JY~@CJ
zSyJhOq{H|NEmLx@*S#3wgc8|Kbsv728Ouu-oWOnFoqxpYK>>W&Qmx2czhfFDWHw|X
zL-=>Al1$<>j8E1oX17rk{&*fWWP7ISc
zo%i9Iqo)9JBMr&ag#sWwSxe_ai6GMH0@6wP=YRzMZRohSI>6KfJ4F0hQSJ?yXy$fS
zkop!hA4KyYC5{%guc0DfE$pAFC)fnY1Z3Vq$1R@{zt`o-|m-Iw+@}
zDIon}J)JvKcLC*|_#MYI;{G+6A*!~h1ERl94qw_ZKw(K&Ui_zJ4D15~c2Jr3)Vnx$FTD-!2Mhb$pBd`+B=-hgluzEao+-t+OM(
zy3(Dhvlz&taxN@dB2NiE)c_BsoZnw)VX)E6jdUojQhc#jHUy5%w73UuwsKc>Gs6QHMQ;IVF63VJJ6n?gR6
z4!kz8)%_FUh#aR)A2AsK`&$C{O55j=6N60Wl<^cARLWa6Io5#wXtVxJZx{iEl~2KDOb%ErzXI~DA`*w`08Rf?E717F}0KhTBvam$I6160G{*3;|G2{+xC
zXK_nbIFhKI8_1~p*V@jAD+?6G6z68e3xiLC6OrScIMk|?-9Z(w4!l&T!
z9gN%jPN$z?Axy9siAA{65&p6C`kin64M_7JEq~Jc1tR|P(hazsCA_R~5#YH&2T5N`
zroz)_;oIvWS4SJ!AP7hUBc#ZS(|e0nm-
z%_kZ;!9oh3)R8=O;NJoLzPv?4KSscV%9o}UQyqwAO7}+J$2E}FEFf1x{|;o2pkofl
zzoBo(sejr?jv$*2O{ZYNUgT-C&cI2^NU)*O{1teEilAt|%2A!RiR|ZJF_m{VgT2Q7
zu#AvVz$4oHN0NpQX1=*VYw}eHMjXnR#j8)Y1kxmISc8wL-oVM
zub?F9)GKJV>WpZXv@;i|;yDAx=
z?b5fTe{8LwSyJeq_=RB5AkV@rKl=wXlh%IBW&MTZtRh4jmV05!eQ&TzUI^Z$9-6YU`1w^kl4aDXE^fB}G!87|LAS_?Q
zQhD18{MYgQogS$@5~`ZNT{Nr$I88K6H@_r
zK9LbKIgOFJQ_ORU7zK2)?W9d++J8X9*)L~!KM{GQ8SU(Hx>KY}adA$W+
z{L6#N>vk4e`l&(stclwYYYFzdXZoQ<^bQ{Tr{Lq47IcdMeXd(SBp{{cEQLKv3E>Ud{bu=-I%zKXfMp0SDKC
z(C*8tV0m;ch;_gO$uY*P#mg!IvivOEn<;USu0r`aq-da;k^8KksIUocw|X
z&I`pv(AMEGB**QUk5%Ea)PjyB(~HiA{hVcwnpJCfGXco&bkU5
zMLn`UFD)N4f&?v}_+@zt)Q$fvKU}WHElsm{9#v&y=3~K(`t}C6_JY?9TW^0nIW9|*
z)+!Y1e8|#js8fP_2JLBoBrCvw`Ce4$uZ+f62Hq={Qo7-fe9q3>P6y!&M>}b(db+WG
zPVO%bPfD?OPcB`luS&y9{?si5D@9=~ZxiGzYy7crwsc(&#x>mCAWHH&*IP`^?wFs(
zt1?W}uXTt?zXZ2nq&5xxoQfr#=zZt&gbuy-3tD@Yu#QW`Xh~6UPGU=IaaY!pI`O-F
znKuihYw&B2qxT*%MPom(^{)oEgs~QW-%*`Am}czeY5S9#a$Rg8)H)@y_;EEa&(lo`*yWacap}I!7*ao@v3vIc&dD$`
z6d|dFng29)4>352fd`82G`hO@5612!(HnW#u@{glkRu$s^P5-lo67^dASG;_qWKOc
zLDp|`NBN*_|ZiSLIBp?)zg&JlJvS@2k!%qPSQm8N;5k
zKGtl$w-EaL23Fjo^W|)r3hsZ;__}XYZ+)DX5Gvlhgn`qFnx}iraP*O6TIqK@{@syt
zuHv}?)({&hoycvAn{Q4l1?akB{!gy5K68x1sb3|1)CqURKNyph@?H
zO8Xy7;I&}fFrD2d@!EB);xrh!jT$hDL&`UL3T~JbwXN#2AthYaBJd==x()v7;n3(e
zx_6lR>EbEEK?dGgA2&V5=7o7BvHA$_J;E~|SmbYJMPNb)=9W34rT9&T0;nEeitQJ4
zQW>%2;H$1IGJZRm*h;DS>a1ZgzWe;8K(8{$h^j4E&2nj<_S2Lt9(m;L^c+0?Q!h4@OxZ_zAJd@
z_YmeosY-jtga_%KP)20@Uoh4y%rqLQUvN1com~mVVT|LbWTvytCf@X1aQ3t$E7&Fi
z>$aMj@lpeo=W)q>*rRBh5<>G7?zYjHRC(&8vleGre|trST_+so+ZUQG!KB!jN&1Tdo33$D}js;7Yw|6-KN
z$YrFM+8?>iAOQ>(o(*4e<3|z!;aLk~ktAbqYWV4mEl
zm7iw^c3qn^-;H9VG_n>Unz9o3%AEf*%V>7#K;>C-6weEF99`P(N@
zT1(~yC5lTR_seIWE5v@e$7XI=
zz#LF4F3+5BF$Cv}r}F{B8Wwt-Od^T)I^d>$2&dX?5DRWiKnhQRz?gS?9CtKO7g(LL
zjK)w3TI~3^tpv<|s*33en;}ZRE~-nU-r!(K!&d0zJ5;VJ_C({tU0}u$@af;b`{?#h
zlovtq0PLHcHLSQBfRdYe`$FX6fS|%3X*xE0^lLnGfi>R~Sbjdc()h+71$(bY_PGax
z_kYw>9)w#UO8cU>tS6k1kR$ci9xYSUpi_C3-|8kfL&rbB6yXCTsdN82ZWf}{Llvv;
zjvS!%Z$aq4ygdNENk^fHJlei{QDC8~
z;<#U?D|)o-ne6qY9GptP$iIO}u<7O`v?-Me{yZOl74rQlitZeftxkRbo-r1ewKR49e8Cagit%2C;efY;5ZivL0_^{*rQ{Uv1lsL1|INPLM}yaxRufwM!NS*X$K%V2!Nr3U
zmpgZgz|2d`Wq$k*de7}C>b1r}SpPhxT6=m3{LVfzoq}&7n!}#O>s#NDV7a!^6MYV7
zudHlS{*f7u4*pe>`H_ttsxYrE^DQB#+r4#*WX#ahHC}PAg@s_1elY@TCm}-lIkEy(PZA7YTw`BUTR`SUZgrv$uEO6hfO0fFPrBMGRNrGs&b4<A15$d42kT5BLte%1VZoxJ)xy(f$85c2FX^UQ-=4i!h^?#
zqgpDb;2$Tcw)x9k@N^i;Kk>^3mdB_uevLMQ=&9t<
zm#Y!|AQL$uXZOvwC63=f;pNC})3036h3MgF4r9oC9;
z@e_m?mGW^3$_oUlu$aLI@2?Ved79${XHLL(5xYCaQW8*Haz2*HiXYC3ZIP>!a})gf
z{+Zi-&>~DI-G67hu1{$D*%FQ@1fcV4TOEPRYVhhc|5BEJe}Edy{^L@)fI{zEWsi`r
zpcC?ZOpk7U01r#~uDuTJ1yeJPUW&5&;G52u82;nraQ6Y{_a9yeNEBMP1ewhs_5W@X
zM9CTu^(VWQqKE7Rd)dIC;Q(qvf$=x~TL$ZZQ#DiY3^oVS4n=kU``3UTN8gbUa1|t8
zzvkd}_&Sn*aeK_ZnTfE`T>kanvyb4#^2aoGnH5miYG10csScZBUge))lZA8NE7s{`
zSP872Wk;RJSPAcIsa6G=m(a>?eC(EO4>-nSSv|8{2DB)?k%-O2Bact*TD*>yh@(5J
z<-Zt5;OVMxUiC^HdO}dwIl}TD=*dxTE!C%j2ATA!Q^C)Pb)u%8-YHL{6N@$H_O+q}
zRBltN(}G^?(eBEUCm@Duj){ikD8Q+FtRaZiADkFIcPCk81$=?Nb%$+3z}s_T05^;y
zp2<2mFHvS82Cr_>v+Lf(^CtEMU5R<*_VJ4%sp~8Vv^2TPZZ-iv@8IIecnjJSIld6*
z`VsLgHajm0`T}W#DZ4d(XF!su>$^XC3vq=XiBP=91M)RpjoiK)0A;WwJZ$5GSlB*`
zzctbZ2Dgju3rU%Pi)h%f`+qu73x(nHr>^e+^Q)t;oGzpQ>Q)U=F^yzY^I-NL=QT(0?KwlXw7U@?X*hRt
zCWr&=W$KWd>YhQb>{45gByxbDH#vf|g2&JvdGf2lJvrii6_KMAmj)sxsPN{~O`O8_
z%$n#)21F5kCT8<28}L27+mifD9F_HP?Fi+lB0Jv8&XyKu!J1R6$?@S|n1?l~*=x?T
z$WUdTLWI;8jD^*wjXc!{nfJtFfU^b)k+RX()RFa;Eq|cy6>1mt
zAP5+0^R94GSfS=~fpq>c&v7%lHhEniQ;@>=?v=+6R`mI8bk#Bczj)fIdsVOc=P>z{
zmT@Y(K78e=t3cbScHHWq+Q*uK6DX@MThafX!G3QS`W(p2;TIm0Xqr$?V7r}@Om#KY
z_(}Et8M8P66me4H!j}iLSO8y{Ns#gaR(dOXHBWjTf2lY*e_eSL`@X^3Xl&00G=JY8
zi?8EGgZ8(kOS4w7u$3FkH@v2>DcKWWn!fDee+X{%tLC>sT?vt}ug#AFt9D0~rZ|v^
z?o(sCFK58atf8iVAuBR=NWRj$R*ffg*+rk2sKpfd&zU?TeSsHTEZcW#zK2V`&`uY8
z7lVD%5Pr9-*o{|T9FXDpUXPc!2{)3+zr!rkyGlzC*&)>-eT<^5W%d{Ky`S>4>^FzRf1hdve*F6B|)9|7C
zE*gRrm~$G5EWg8h6?hcqIg4=ox?TO3!8j~hO-JG1TnOIx>ev@|`bf+|AgP)nCJ(ni
zRS>{VVibzp*gR~0Eta!e`jqp
zCUeh;eEMMp#uNCao~xn_uScdU92J81BHdzUzO(@$!sqORS6leUr^
zukaki!nfKKH6Eo*396yMC_q`%r5LsyEh5Q
zZw%nnlN9}tD|85RzR)@wa~#w@y_<0I!FlwmsU(ceOHoK2O_
z?<<0ThCPigvzJjg*(tWZUvgl6GIv*X(-@_zWJf%meF`q|?%Jn3)djEhej{TiYOp%}
z?FZvIc{C}w6T>EG1U@B63i*(ygPVK;1-)+-0oq^vDfG|)ya9h}TU6}OW*7ta`D`R3kt=fMQX$ZtaQOP?V>)8B
zHjS%UZ$x4(-0IW6N`PmBGLQ4$3KUfJBjCk!9T?M>cm15oacP9w1
zyzbT>O%Q-l_l%n(vX2tXMMi(KUuS_NQts+GJ=_GN^q10qGS5N}&H>du4-LZO+(&+~
zP7;JE{vpT3!Yfee%foNy%~c3L_nVycFB`-1fW!7PS&9TMu3wj9&d|Y!eg&7FupJ`|
z-mcwNSC@xd?*>y97PX+$T3~7IV|&7fxCjZsEOE#dPXDu3_#{;P$$8E14IQMB2sq;t
z=tLNm?o-iwt^qF;U6~0K;UiGF$T_opWhHdgl1}N9>Jz>%_G&-X^MP#kdJ)x1_u-!(
zE5lzXtO!Qo9*o695=6c^)6;Kz*3bmK(0TsrIpOu+?qlxcr33@sANro#*I{jh_oyU^
zF+ueBC6^(X5`nh1|(vOO<3S`u0!$&4>W+Y#zE-@8(_JcKhU
zvf}x3W^nZ`%Yj^~JwZi^jtPkjfM0T7M6LJ&AeoZhv5oJt@Z`?dI+PSm5PMNAoo^IL
zD0sYZ)uqA_GO*vt<2W5nunyUuS)s`%P;=`!P446pI$_@+e3wnwo)91ODJ>z4A8AOT
zC$A>>bT##TNT?x5=kc;Uv??J8wKkS#aTgNWN4b>`_ZkUrFN9cS*47a&?efTSI#&{^
zDi7-JQI!!c#f3YsJ}o4?lGC}@p=w8%l@kv6VCYVW?gl@i*z5`9OiI}w=QIdUg%9Dw
zUS7gg;TK~kRwD^Cvm)4ZQ3T<7KY5&obr4~x?3en>XnTS|&em=&4HHZm{$w&BnoS^!
zjLJXH|Bi5=RJ(IwA%t+(H+#|4*c0+hQGcf)3xcyN>LNS~g#>fgrsKCp(+IAIJ)bYx
z+7U$Qf2DD4#lydMmY(uuzJv0$27a~{B?P|@N#{S|S%k^vxi*FCPvLYiZZ;7~^a)#5
zWu3j52fJ;xYIx4^!^d$0``u%jgh%t}t?z&y;Ry?$mGwtPLc7oAq-7cgt?^9$NE${M
zce?$tDzgE+Ug0EkFrq*h`-TyA8T1I}y|IQ6S&XokZl9*S9RSB~hft&pg}_g{PWcxE
zoF_nW$9In_6V5#KQai?-4h88v1MC@M;Zgci
z5z^F8pv&H+Id8{2NL8UWoc=za=qHx%;T_6{|FtbW=sA%N8^%MJN7)kKFU8+6LNox9
zJ7#$6zqkZD6sbD;dB_MXk=ANR@d)90TlTz3^)8xab!441k|$_~G~X2)IuGAjc}%^N
zI}cg!-EeS~BqKcjy-_H3`xmMmFvP-9|BHD@jX
z+RvMw5wH)0SJX-WBvKg>43Ccmca1Z`eA7q1owzc5^dzO$`0;PFn4H^P?A?d>`rt*r
z;whkF`15Mme+0yMO%sF*_MxTtD!w$K&u=2B;iYvrrLCN3}h_
zJWiMffab*bzz~;qV0M4>zvP1k#2>|XU==P5wz_8gB)HQ+*qd;b?}}+Cv~cBZ|6&YM
znYEd7Vd({R3C$lF8aMftQ5p4-eY0EM@D#BrpRS%G=!qH`1p?aY
zo!f8@S>HbLap^)dIFp=|uXpAdn8`~$xp>S0JvXr(e&DMJ{BF|A`0kj4&nMpfBaCIC
zAOMxdR
z1CWK(Pi;?OZ_pd|_eM;>HMD|Btj)R9p!aeixi!*7p!>1i$$z0TsE0N@3}XsLA7>Q~
z8A?Nem4MIR%+f_9CAJqY_$2}C%EZXnRgIC~v0!_?xhZ*pq
zdLWocb_K=gzMEF$U`9gXmzhJYIlzs@)HBPOXOI(9PXEOrhz$8%Q`LsG0AnXR^YZj5
zN+-fJ-kw5SMF~&$
zcF&_r53Nm?#iNPeGcSO8M&QB*pNA;H5YfsdGLArPaND6M_4^M2B-Icbo_{F~T%Aqyh>{MeQw*24LfFLh&=^UhJv0lM{RlBkkGR1OT<%*184D{c~84yCuZ==?AKk6tIS{|-WnrE
zjKAOtr*9v5!6%4vChHHRm1)7ZA37l#vkll;zgN^|B{R5Zh~;lqhdFH4BO(8_rY!PE
z6^{8ga2}Q988c)*qyu9lP-K$u3;&OLX-i9ZAOA1*{R&N|9#|w5`S#$uBASQ|AQheb
zi?Nii2kZ&cf+IgN#H1|QQGPbJ!wug=>}0&b^vb8F*ji-#p!?mIcwI2`8#aP2U-v3ZId^Vz^-mDa&L4N4W~&cZ`4@MCr@0B^
zoxPRVu-Ao)-;fT9797NUDSme>D{taoGNpZdGig9?(z)bBvwjR1_4MDr(~FyOl{K4N
z3}7l3xH=us4_t>$#`6^SIY6>^fpg;V7&e`J_vx187|tABQmFCjD>j(7y{)i4jPcwn
zIbX(f9@v}{DKP%b2%M^1ZDe_<(Z5r-|C&oMfz_S;maGU)RCcD~eW2|jCg&u-)7C}>
zDo)av;jM>wtIt1e=l91zMn{CYdxZ$P&Y8u3J?j?eW#Y5XetjBnsJ?4$$98b$Jv2#E3ZAC%Pe!Q@o%kl~{t2C5i3S0A%6e
zk=sFi4bXW$>5W}e1apxOqx7i&Vy?pJ9n@683GAf-2nOhrSt}
z09(OUj-tp5DCt+v$bPm*EUKilrMnK`VDOb7TDy;owGv07jeP+YKN8JN8H;`gw$>ii
zO+xg62P;R(y#QAjv-clQTVy?OZfv#Q0o>SkKb^lCjbbxj9!;!o1lk`F-@7zFAXXQ{
zqOxuYVhrzCB?<&0!!Pe+O1{1W+r{J~sA>REOL}|@?9BydFe%Tjvq{K}ZskRuRy^&j;f!F+~D7{ouygW5m)o9-LT08SV1_Q*_?{Sbl#TuaGFRGqMVyWR`u7nborQ
zUXeY@3Ym#MwkR?aky&QWvB@eUk?dV6BqQs4fB%8|as6;V&ULQ$`}KUj@{qbq3k3yR
z6_BmSi2m631vTx&Un3svMdAm^H)qm9(9uk}=cgZ4f+dZ(>veI>$fKe&fjwgg(5_sv
zRVg?EIh$Aoa<@CdnW4XO8G_X)wT9+CM^7HudOYmvuQd-?wDbQ?Gt7V$Kx!3HGXcVf
zW92E`r%>Y}dB=yHB|uC>HDzYMhSt84smlp4;b-HwZlH)cP@L%!c8hlfZ6!~HRi*3z
z77|H@?hslm%*##q4+!8VTN$)@55A)#FFQjbT=&694@2?A9CB=B``1v&`79iv`eHF4
za}icIQuI60`~{@DRtZn89>sS&Dr2I0iQ#kw>sN6~T4-*UqrlUpjw>E%eG%BAg`j6j
zxzL^&yHT9oTy*1uhPin$vyYVVD|)Iv;6y@L*@V^
z3YMNoYSqM*syb^5JC@L&#wmlgMI4tXuhW%I12}AdIY#WJGX6PydBLnx7Y=cjvIiNN
z!n*WDYNMe|@SjfXz7L}qHezngIkF%PcPRsvI>-1RY0qEHpPGJn+Bhe`NR$!BSE^@`
z1&HG9QkR1@XC3I7pUIk^8cH_otWW4gnHwq$Ntg<>u2+C
zV3DcT*SaVGCs{sJ<#>&;)eEE6NCAxR=MSGfxMzn~PrCmaxcC_Ir^J6eXJ3oicc>U>
z7n`uDe4&)ICB}>L*NIlo#Y4RX>7c;Hm-y+&&9>#j7JN|9@_;6&5q}*`Xb+ug!zW9o
z17E*?28YS>jA~_@u*he{?laZ(*xF$>I5@Kvrx?}Wy+GfMB|c=V5G&`yp!8uM`s0O=
zTJ93blfgpBpDw=Qcqbdi{1g9Z@*o)|$ugNs6FA_JPdeU0&njRmpZkBw1aDBYQ`X7e
z8?WKdS$&&Cl^i%^WTK%48{wbCW#w-)jgZ8M(d^pydMN(a^Y}dl0{#HNPrBVGfOu&Po6
zm=5^;jyWp86X;b!HN1*VKK1m)G*Ch6-{CEF<-D+T&VlyiR||YvajCFiS{p8i+~G5%
z6ob3v=c`Ki>0^jxJiEgT&5bCYNbz&a1^u2_CC#A$BlvPO6HA??J>RdXbX!|V4+
zoyV_X=bzJx_t!kJbcvhZdYKod_H^GOuhD{%?}bgSCa6QvS&C-m*SE18n~=_k6bU}2
zD5LMne1OdPp3W4l%z|oBVLtMcgng*>OVC1sKlkgmb(CkC=rpdo
zgJRwlvT%fzn)#zawLWEig?F{kr~poj7Oeur@={gQ-?G!y#wR%{kb*sM}&w*S;O#6GXi#Z6ocn&Zs^;}liA?6@qHm8pK;TQC{)`_J(|d@_N4ZSd7~my9B|;i}^|Cu#tND|Pi!YcT4#vhjw%
z@1m3v(gnsnB6vRH(dgFiuZE(Iutnt$%k
zY98u3S21o97y?eGnCIO}(?hRLlpK4>cMl~;XL1tgA`rCaA^xCIA2F<%W{I!FBBSdn
zEct&U!Hngnd5=PGbPhM{Ryq#=o`2^kD0>?Lna}jM6!u(XzCdK$932WO?}Ta0c>f0y
zr)aifoajJmfzIaFM>2@^_QI%!FEz+{W4QmCs8hD+;u(c6MyJr{fRpWAYVKfJ`eEaT
z{%fG{pAh4nNO43U_7&?c@&kNJW+Lp?aZqqUFP(eW3lT4>6WL4M210l5aE-Kbfr0s`
zSQA$vG&fwpwEvGE{GFqJ!*?|paq$r|`tVwzl`Bh8LOiO#+;qA1N{<*?;kC{dDZhDiLri4^)0>aW1Br3#pB?|;3?rGR(``WT6W#$^M4XJq>)
z4>cIRu+}a(`bkzKUnYRO@`0
zZIGE2HQM%s0{puCi^}s!kL;J`hd(`ZK5c+2RdyXa`C!5n{vyunJ&`{%6+bQ&5xw?q2>^w?6Zo1T-t_;L6t9mC%
zD5%&hwZ}2d4c!Ie&
z)iY6)vFQ8rH{A)#(SW2eN84BOKDfsp=|j-G0JcJvkC~6}AfA!y(uOA-!0Ah;VqZTR
z08>E@imzCofcZXlj-}veWOO%~`#?Dn6qyKVS_wQsz0G$Q^@y^;6VGH8iI<7s>61s%
zRS`+3cKzE4jlP%Q!OuN+hQLmwX)5F#ce@vS(N^10xRU{nCBNMpkZ}N)>z|VJe{)0o
zRxTsk1MgA#nL}1$jV&PXb<%*!xDe#uYkzI>=szI)*9kNSy+&tQro%G%J3;OjutwB>
z1QriSF3$WM1r|IscTP17A;Yh?+ER{=f~eiOhl6ePAp25nPRJiJyqHH4&U>#Hyg9y=
zBF^87CS(6F>2OSgr1e-npA<5zm^1QJv7ZdD(z6t{!B3!JOU8sPcLG??RoYeBZXo?1
zT9o8(PQxDuX{8ZlJaAu{mGrFZ7MS`V_=@rN0;>33>l=D564hR>9PCeKfxMetOW!9s
zuy2x5vPB>fR2*Fou~Rut*!wTuiB+Y7H%3I~xcwNg!wt?Q!LgfIsaLV^7sD!2-Lpwg
z2--)2#6e5#j-1dz?NRk8k2dCbKQ-Y~E)R|COGNvq&f^U&mE!IJQAjraM&X2W_E3
z4wKxrv)y$(s2|))uJ{OhgG#(AZ
z4t6fQvB|@U*;!6DY_0gs$91-9$9f$31!(5kg}`Gcsr)@WVxW*n(~&CrYFwuN_I=yA
zMx6ZhUzybNCj3vek67YrJUlD16~EoyfFHXb3lGz4#72(p-HPNbSS=@Ye=)TahZuK=
z*lyI~dkz+_Yg!v|g^(I*eB6lbt`s$9+PC0{%g$FW%(h|TjN^Vsb6RlCcZOT}3XPbT
z0MI>CZo=H}_ddUxYQgOGOb^wqTk*XAT;eD7_xL}S#2pvr27EN7Swn{Kdp|bM45!^~
z!Fi>Qn-7@V@EWztV{4&CTorI@Bi|okzrp*>fm8^`2%cGf+1iY=zZ^`lh_ztO?f*vA
zsk>nDj^CqTg7>KG;;$>uH(Mdi?PDY-_z1o{!7G$E`4QZ^Z+&(6d>_HL%{ehw(+yLk
z!gef}+aY_z1X;+lW~i|HY>fX`BV-}Zq%BwLg=-@NMfCgK(0EoE?D%!S)>&~~r
zQwvg3qteZ=|FW0z@u6NAIIG4yY19QTPzE?Y=%A
z?RBL*_$9@d#$fLr!Iw0*_v4}`?xJT`A~{CzGDT=tZCKUAEE0o}M4n>US^Y|$aqmB<
zzvpZ1rwrhi(1}apD$UU7U3|%CB!;F;AMysG-oo@_^PLflxiEO}KPUU`IC#Y+$=SrW
z38GOc*LhlkkFWXKbjNBftTGAE_M#|-8IK=Eab3@WvwjcNnQ9tg?|mClhb;{GZeQVK
zJk|i4H+z2W`&B?iGA2Etj(li1eNQ!r*Be)hCp@v{aK|y`t7mK~wP0BLMRw^U@(`GK
zUg^Ak6K9H*hGv;6!J|~2xBJXIG3TDsjlp1fd`?|xRU^&-67zi=;Z!<>DO|1vs8+>8
z*Ke}T3{Q>Vq{c=;Bc%(T%uezY(Yy)AL{0f6e;W{d*S3Cq{yFgOSE0+vF0pXdi1aO_
zx5E+n)U(vdhS)Ode8Os}7W80?JDNF>55xEWrtu%7K=#f-O2=q5xW&QJDXeOSFYkO)
z842LW1=
zGn<@j74Vn-IugUvJ3#5)Unjl^2FNJ&CA*V~76(nn%9weYz_El$m)ow4usYUnusBZ?
z&*!~4KS#m|t@2;#PZZHY9rG^Mw^#I_*CW~Ti}}}K9jA}$68&Yo>O3jXZAO6~1Q%w1
zeSRL&=(;BRUlG8s=n4lzq$S`Vh5}N3GCC-y+Cph&$U^8?D3@{`9tDwHFPmtSsIl$8
z`smHpAArL4v1o1iXP{~8LlY?f1vSqo5Ms96WOS1@J|G>z!@OXW)3Rq)O6x6-4ow
zBtMEj0)L9_=-p--K*RTsrf!~UM>i`;4`pu*1G2LR3yV7a!0||teP}SDgQ@01weYbQ
zsPLJgxY>sY@RlrNwB@%C5Qy+UtI9usWKBXTyswo3zvfoT&qtHMKLzEgp8`+O#XyOa
zQ}!`{yhH6@OYt|P`(5AVfn*0tTvz415K{^+$K>77yB~*S)5BCw8m1u=?#uTzB?m!l
zOM%aNZ#N=e%eT5tn0IHNkgq@eZSx}p%<)f?FGR^L0_l`r||+h&q)ncSuN1-)vQbNpVh#*YfiKN_vFEAdtWLU*)+66
zlJQoG-4|H54Qg46X#t)&KCX-~6O=hKAwJJ=6I8g3-@JUM7~N|bT&n)+kG#vt?m9$!
zgAW3HE?L%BAl^
z4NRll1v2@$L4p(KiS(sgKx%Mp`_?!gc$69CZeJyeeq??o{o5%CT!b6n1^Qh_tX5yH
zDgP8gJb4)@Y8s+|lE<3Ai$fY|YjkOb_J5F
z=DD4yJLvtP7e|Sl7x<-@uU_sIhvYuVFm4t-1_$Rl(+6(c1(NPB1#Y=HAksgyG{?m}
z!2;(-%Xo7nk_|k~Ee-AgDc^IGk>}lkg~_VQ9W6J+Np$nEa9RL(&0fQIg*Y7*R6hSS
zpZo@Vl8Tc5J?RDdm>ETN&MG0VXp7Hp=tPjRHEP%lc!1VzKHv{EeSj#i=8t-*C$JV@
zpHOJ@0DC_ZMwcz4ki61yuDy{$pc};KSe`lz(slkU**6vgnmf;4op|yDJZvCZJH7N8
zRTQZ)tzGT_hl;&f$4(xCz0VFugffePU3>1BU5GDA=)zXwPuu{*EzO|vZ|_m`H}+$s
z+uLAMmy6t%t`@kKp0AQmPXK-gtJ>kBbtp*uu5jS1DUfd5CEQxF2S`)>NUn~nb2~$&y2glf{u)M9j_27XD
z{`*5-^kv3haR2Ca?cH175#?>UtuDDvbYH^$u?p;CyH|1gJF4sNm?&YyaSrRdQm3(EJPGLR;e=b&
zb1$c-l?msxa<|D#O=xgjw|7&E9QR9GyzwAD2gj4GJXfT6@UtW`huBR?{5tJ!HhF~$
zq4$PV*O3$8GcDfNV!z5j^B-5I^A|~R`LWN>a+6&!SIe6ZPVJ9iDs$gc(8+}3iaG2m
zRp!83VTn@JH~dhls`$*!0~_2X`O}8&b|9p1&%3jA<2YvQghRtx&f2iW!?-IPyMA`USlH4*rvH%zJcxd&%4F}3+swkQxlV?_6O=3yv6>z@
zc&_jF(lI5hr8(ZY;B5iFbJNcUWcuNnl%VqE<7rS(&FR4}wF>OI*Z=A6`vCaqx8k2I
zzX(XBGBCD(hL9IzV~kc(G7lBYvtLyAGK)7>PAv
zH=R2n3rYTPLMuSqmBGFDQN(`dvq>21eNdF%14vGaLuo@SgTbJgKN!E1c9VuMqY
zsTsphSv%%ZwRo=+4cb@L;XM{%`;y4_Si-a$R_irFvu9=uHxw~E;_EQ5@Z~+UY!fYJ
zepvzAdL8EncwWIx&PRtA2wwTB^tYM}f*2OEjSV#wA~?+yX!P;;JGkR7py#Ym44FE0
z-}3Py_@Bo8h9B7oe*b6ok(=N%?qm8b{r-9#yhsusbW`yy94_rjPCSAk4Np<@r{@ij
zoOzNr&{$x
zy%-X=-wuANlm*|IvJiZ^_OSiV5N(%|4W2Up(ZucV?Yv0gUY?p9GIXA>;2RhUjo-tZP{Lqt)^v
zoNqibe!Wrwjy6iHwbRxilA{yFSBY8iN!w3?#+S%oYWNL`T{$*5s5@zO##aqKF4xk1
zJxY)N6?IIk+~dU8R=vK}2XVp6+Duo;`0wDTgj*)>+?1iZkNx4RGP&Fde-1i&v&
zoP7Htfdc2JkK7Dz^~IvyEMk=>t#GhozUQx(y6{9=PPl$Nq9&Cgq@pG-YQ5Rn>`MM(Vdb
zSr0-!{7msBxE>EE4Av+UE5`vFTk=8)F+cp$)Bn0?feij4-VQ$(afk%8$cIjYCBXS1
zwp&wq9z^!I{kuAQh>FRm6ev5mU`XAaY)7XhUrTQ%HXND-rMMyh5yvTTPFOaHp#bSej2p=G0l)ne1gJ^#Af&(7NJ_ZQOd75`M|C%
zDU>qj5&Eg>m1hw#59C>o4j4WEgnk>h+^sKsfo7&Iew>mj2Z{S(NiyW6h>Xg7a+RJ4
zs@onUCKKir?&WGfZEgEep?DVUj%yR}3S28pG|dJxTN3|Znr%&T({}xq?
zcC24njY0Km-y@@Yf&o_FJ1!9Ajx3ZZZ$>njBByu4_Wcu)AX_)T`c6b1dgm_V*t+Ne
zgtE%Ah|(>=P2Fu#vLojE
zO%r^Ij*#Ac^9krO#8Lga(gb45A6oQ0%RzDHc*0-$KLoFzu)kF@?!0EJt3nWpCqk*GeiT+#z^6zi91p_3#E+67Ni9)4m$i8n&7g>kRS
z&S@R7;wYS4Q)*=u1r>Q3$9%}*T#&=tD
z;|3^W^sDOc6dAx;%ai3WZieW~Hh!2p*aH7(?cn50FC=O`wmNt$7CleR+&Gac0bI;(
zD!cqPM_C`k?@Q9(1$V_-n%0j+q9m5Zdc`dpj23-?_OqbDDWb-
zzX}!}f`h|X?NX7!$YVNJxrwO`jomurmzP`v^Q9Gh58Vh8?Q?v=^YR2g1M8P^hW=7?
z#3H@;()D)0cp^&BwQ&msyubR5%iw-;t}vg1wxQsY|1uX;Z{W!@Qu_^V*xlbW*!YZmc=}khw?3m&
znA$NuneZLi&X#05z=VxNuJVXn{0A!TTI}>*JVc>8&+@`|NT7Jw^!3o-llZOT^%53C
z1jw3O-?Hgo4^e5#@d~xLnPERO*Q$<{~iw1_G||tXy+ThcP?0e_5!W
zeo_Rkuu6S(5>UXn#3EtupBW5{BFXO~)r7Izx$y!9q>$vCw0_gqOZZBNCP$z16}Wbu
z#?Rmegu!7SM92OF;ew?{uDMLi@BwpQK8?IAPHTQ@Q}X#XG?viXabzFOIAXg+OC>hjFf32C(k&HwmZ9
zI`9ugO{Y_xFSegP_Ps^<8NrMG{`pbAw^(-SKS_AZ0tZqH$WlD>z>J=BlwL)Fkax`F
zbLO2i?5xG&dv>D%pHK8kGAnV!#gfMJl!ux)b=*69`|>@wEvPlBlUaq0o$2U*HrM0V
zbDSw3hPClXtEWngVi@dCt0HDqNyq7pD%
z!*k{WV-Lym@acW__?W?RticvdJ@%>sw{>>CA+vph3o{jjCI%a^78P4eY<(Tx!e>lGU%&LJ!l^BHZ+L7~W23gNMDnB6xa0j2x9>&~&gA8IdT*~B
ze>Lw|`un*8A9}Gq*Gzwhtv_YPIz(6FAEVJ#c(>6IBWA0hOf!#E#rIN!>y*+
zi|S$so;$R5_PvAP)uRRlRqhD-m-YMo$;0qV_19tQy9nm1`lkv0eh)hqkDl;nZ-D&2
zwdwfC-@^;{?(Kd1P!H$aLtE==>LEkkx%}+AweZzj_VGs{HSk%oe5*`T6{LwI`tF$W
z7MfMjxN8zT-8Y^@aHfB%f(&G`jAFa*pgm_#n^$5b>{xz_d)><6nugoLdzCWyQdz>|
z(ckyboJi%EL|HZb`oiVJaQ!RTSnGaP_*4k|(pPbJ{+BbBF?_i5_$lFjc7){3xy$uX
zfUM$MPGTv1w&)ccWS0nexObPrZrVWBAL_M!U+Uq-OdYSPg!j;yKVoNF@*S-Ex-=>$
zkOxb9&-5*NCO|{mqS=?Nm5|0@JR!od3XW_X(Q}P^2NU8b0~G0t;P+!>TdZPP@W4p3
zfFZ67()QZVa7vZK!|8@%W&284?`z{+&8c`jy33kI|_dn)mu6bap
zA>qi=+8S_3`{~nlVpX`xesw~NS_7vnT#Z%~vV*2F)BM$S?)dS$w+Fo{H*uNC%2wfb
zeMs}NtRq}_2iy|-)L_Jr2zSlQV}!oiL$>i%^`4s!*yv#NZ_#U2$j77`aAHUYrrhCM
z93b2`g(Z0(bdSUkdNdQ(uBRH~b%!g40ZH0ejH=UdF#a}dIdPWtT3t5eS&G}aI1~?0
zdnxXu19_;F_d??7gf7mXh}jeKA^2K!?5--lyp5MCGeU#c#Brt5FQaFzT##x$muLCi
zQQVuy^@W>*25%2YR&I_F;Y;4$k1TiOux66)So_unxYtT4Jz_);yC=v!?0?eW=zXfg
zrbT^Ns<7s=5pWzbRC0=*RuID44ze6^V{A}=-0D)Q3kBS*6pW#1(}p6iwL_N6RG`8d
zWB;C>0A7vCwf1>>6!V3In!D(qgOkBeoXRD*VW!u?Dn%KDxqZV@3^YGcZRQCHbyp@V
zcyiw~oP!*myXh|WB%T(t<|vX9GtGl%lD{fiKaPTow6G8bpAj?`Vy!00*a5hW)K5?d
zGGeD3+J*vvTeNPGN|^qSvsZo2}Q99GPqEDM3T*6-N#BO;52ncErl5&r@y&%
zBhhmZ$>dpxzPvpRHeQXKGkHCPj17hs_bGVb$_J(IW9K))BO99d!FOBe;l<5^>v@ZS
zqRq|Fdgc>gun(U)qe2Z!Dr+}OT8Ln3ux08))@3Bnz#f>un-40@zftcCOn{&>FIJOH
zKLK%(e(7=9Dlj`-Pw0<(ff7#(iFJF21C~Li7A4(>z}H0AU(Tr)#aGcfTNhP=%GXgI
z7kMMWsYhDEKTjs2l1@2IQ~eMS8MEOn@p%jh_k=~3y=q2jb%ruu^NYdIC1)LykMSr?
zqxDdw9
z0jg&s7HSm{xk1yv;(=Ml_d+D57ExO
z!o2FE?m#NKt_P|pgV7z&8ES1EWXJj0hB(^<)cw$AN*eV>mLU$PuR0EtN50r+s89fk
z7rBc4nQtOBn}U4THeK*WiBCfFKM&MXX0X@$Jswa?XmMP*Cka}FNSqz3|vVtI=W^;D!N9vafwQv?i&{cbBou7Tx=JI{2z^FU%K
zaEKOoj3|6FOdnnj2F;f)$j%md0MScF6pUKuz+RZ3$J+cVFAbP$b4khDGOn=*QMZ8;RtXc~VY_pby&QGd-&
zig$t7<>L(sW!u2*L(E73%Qkx7+s>qHzYl&`#~Eao^dY51RjY_mNi4mudnJ{AgK$s$
zueG844+5>emEGPupp==T_=Vvg^k~mo@7*a8T$eboYg@1azBPzy%D(%K=v6Y!LsS2N
z+R=goHCAFwbvn;xFNztz4mfx1R`n#fai;My%g1RnF=P2!lKCfr*F5soTka@!-bqva
z6LuP#Gq`-6b^HRjpBDBf9i2r=6}N-ws#ZZ
zKDk6&cNWrcGd*wkzz0#!yOrOY)_Bb9%E(9P1;@{wXWV4w#LYY69tF!xklWuxm-!U~
zypZwKDRRREN46?oa+~*n+rtGL%00^{GVg_n{aF?`Oy4vFvY7EG-Rw&)q=3E5i{AUj
zxj>6lv!UY0ePErYO6|lA2FSzJ_u!2K3l<+Uh-$05h~>hDX=f^|q4L5DDYNY{RQ_ec
zvT%wO{-iEv^dvuzg%Ve;&L)Xr*2}lZh@>2#iMw$gEuS6SjOG~bt?`BTFU7v@&&|a;
zffH-B$rU(0k47d)uoO4UJ%1vfZ4Axs8%5K%cte|*Z}vTzdDwuT^lFGr2{!Obnr^-O
z5_8*ndqkU>!x?|mQ`uf3?2-TxOCfop@V8Q2^(8I3g?8|dW@a)CDvCkP!FsGVY
zQ}nS89$h{|afAurbJZLxp@ioQgEuxb6duF7>ECV5w(KCO`(#M^8)Yn+MehCTxCHi#
zPrG_;I07DeAUl?xK=`XsKd8jc1a9>YqS$I=Ekhzn7BFSmv^BoOs)OHC45T>
z>Ye?b_d}Ex_mBRb-#MKiNR<@ejq>d5q|Ft82C`
zsRAt%YqJ^yg+TU>T*~LyI>=m|%}BB28X^-U#*5$3gr^5q*)Tv)nR<<5r*o0D
zK~a;D(K`Y^Ke<9TSXEUBHH
zm`w{>3cq`wR;B?gE(~2G)oDV5G2Y+waz;R|i^QOa%nbS=~emEq@wMkpn$+4?nINe2fuKZj)%WCW_!EO1rT#bJm2s!R5X`
zWk>M6FQi{>sWVnVtts>79=5w1(J%Fq8
z>xm;@%Mi%~B{tZh$9Mjc^%!MSW3zsV8eS_tn6JXd8b&OFeM)Xev0944_%}hF1e4{!$mgx
zvJ7E^%tt%!mg9K9>TRZU!6U*Py$RbE1mZ)AlT)8_wK4VJ+JFVK6xiS7CadQGkJXZ?i@_kdQ+hH
zX!TlQd?s9()cPKB>
zyp_V>b+a7TrqyUDyT)nGxflVZ_kS;3@qCJJpJpgnTlx<_8+yW}yKasH{Zo^#XgEN0
zFu&lMYzqr-hX4>!iEZ_yr4#acU}!UnSO7{6*t{}CC3#mg>cFgd-&
z9etPb?d(c$NB#P@_rkBRhJf=+A&p!-IxJr2*qec?C$HzOb;IHMU>;N(_XQy
zci-SPd|T5nsDO||o+Hyy$-yaa4(Vr_BVat+x%0=C9>5m7_D|@x5o{Cpd~-Kg3JbQj
z|LiRk#y_SKzMgfoh5jYq(`i2HLPNRc#f%5%#x74n>{3`GXwK4rI=;livx6x_MzeGyNj
z9JuKY?*YzFCc9(Zq)@XB
z21L78gIUR8ra=L1S@tYgGP!eY-An~C6^n3;@{8b;b`wQ6o^oP#@1x?m^PDijcAKlT
zDGOYqmW;UiLlWMl+vRf?CCmkWcy@D1(Bc$nQp%aVbC7+(n@M7j2wVJnluVlb1!?k~
zUwZEO3*3_s40j>RK@-Q>l-Gk>(Cruh3ME-pfmk*RUB_M5jwmFph&4jcfYi5x)^{(f#;>qs73Q^jOcO@+QD^Vg7b3
zX(uAH5us+Xnn4B&QgrS)q}a&CJG;wt1XQ=#Upk>Z2r4!9lYDq*(4QUU)WPScusOA8
zlU61f#zYxsrjIOwaB_{GZ*&{LcV!pipv)Om5x}dcpWh8Owig!stMfoJhu6yjmS>1f
z+1y*5ItqCF*-Mo34*p{Reh%~
z7@(|{z1Jaq$i+Uxjpa@zDsYK=o?MXz-rX?!mm2;Ajg=n!+~e^=^`l%Nk;MzhF_S)L
z?_LGcp_+-atE&Vqx$?PGtOY2&?vvqqSqy49vYZw~$dg{HbMPFqdy0%Kb*CLUD?lO@
zbK@7DLbS)QAy_FLjBM{lmp0tE4qEOtvxw~|0`jgc+PE=uq!%Am;$j&I4o97Cn2sc(
zSB*z4$aJoNBbu}qNlHvWNA{$OviTh}{nCM%J?bH-KkIsNIx-t*tGVp{Y$`n7kKn&8+0xfGGpLyr|u11-a0n58vgU#08V=xhei
zN`P_Og(s+p_`#VN&KKz46J9mZ$Vzay6?5n0pU;T@{CQF<;wofj=(VeUDHS{&;Nvc2Bhd<#13ew?nXZ9+|!MqXKP3++v|T{w0n3{M3PRDd-_nkKp_=H6!+F#B>mTntAL%);(l(yk4!#4xs
z91G)Y?)n0g%zHhCf{&3ZR$|;wNCnsTRjJ5IYY<1dG}F1NX7pxqePb-J7m0oUwN5wq
z7i5Nn)YV-h!D`9_`UCgbq5SdPq@R<5FfqLG=Tz$|dj0n1;Em!9;FO?#sYrbbDS~%d
zzG=H?FoU6(h2Rt8IbrWcm0O5541O!mSx%xvI#ms_-A!O9K13I~%!->99X?BPoP&05
zmKEOg;~~L_?%5=9?&_v6~`gcT&Is=l+POo2tX6lHKK4+c^UXxTtXFl(t
z=F`7Ii^jR2+qpv1!8Id%Jn__hWiNjyX_6T~Ykmc~Jg<1*!X}N`0?aqwDJesaG}vo;
zjll8DN`CFBwZ|rZ(Z;f~8!oke{4+Ai3R=g{_&*ZQg0kD1G&+RdJ9-Z`@ht07SQ9bFS-ooYp
z|Ku~_HE_POKV^@i0iK>3F^!gPgxzk}{pE+M;1v>6vJQze)@74~*gA
ze0sH#Z4EQt0_g)WPtT%B7B$I(YxcLb3ETfpaTb;139Dg45-BeXe2+
zFsw)GS!U&XIKgzZtDCL?cBOk>)pKoxH{NylG3g-qo^DpAfTRuUytDo9GJ6Yd>$r1u
zfu<3Af2*TorK-ceTq3t6R;%#AHOnLXD{Yv;n21EKss&TsZ1S-i$5>$gFx!!$4u8+L
zv$nfhjeSR6eE%KSfmL}8I6{)z@W_92@?mn#nDsI9ZSn?$ts;MtGJmPZHUl!__xsy1
z|A~i_dd_W_stFo@$!*4kjhyMul8ty(ac=U6I>MWIbi)pLZP@zH{7S8FD<07er5vbl
z!oEcBG&KRn`xVc9qKg_Zw_x(O*5xW}0fx&Bl*+O6>Cdu)j&Cp%Nzc(WwF0~d|{D=iMHRr6jwo5N6#rd@@4}zSZ+3`A>@pT
z`|o}}w9dr^+*{|tavY|QSML-GYryr#S;cs4-s8?!Z(pqu_E&En?ztO$OvB$k?D5Tc
z2jM;8V#Qpx9PH($-+N;@318|bV<9WE#T~JzI!%{GF$4avR*nKf>wxa|!zl
zZCCHw(Uf2uYc{pRFlGV8i!&}SkLyA)19IvZb6tG1@Hso%C=Wio))-AXDGo>e>(;va
zMjZEc?HrA(pu-@d)R?rF4YCU?bU79fynweF3|jk^0I&bQ%9ksu7_fH<#+I&tFX#0F
zMo75fA640U_?ro5=%-es+}6X=@iRFA+X{Hzh>hje1tIv@?cH(L?0F=rcHWgSiUZFs
zZ=P2_r;XKSmIyu6l$eF{ew+P~E#RZQs5I-?k76eNUiw0|h&*ce
zrh5F(V}oS<^5ZZ1(X;96!_}YH0fqe|ispqyq(RkbyTzP?=!N9P?{`*%fhcVqLxPV{
zL@*hX;+s~@4uk4CNUr*ZxPk&=AFG*
zKLpZ_$1`_-YXa2^CXX_cXV9LyWYJPi85*_Yq$<2p4)`oduh~lEA-8~`H2REK)UO{M
zAQZoh{=L}qh+C=$zm5EQ9l1)t`N0}W;ge;kN;%%VWcv{yJ>g8E^R*PDn@S^9)D715
zuWPm^WufzICFYJTVaROp>iUpq22e86(MlVe0%~HMO9RIF;2EXJ`RFT!NbQe?YyI#G
zaOmvn_9WXIq%N#y1Pasu1|WM#DpQLT`{qqqO#Hwx#ClXB@FA$)Ni;o@8->hhK1s5S
zF{8Y_z0Syti(rjhUwi(KETZ^Xk%3;?0pFaLlchhskPJ`t(x(?x$Yv@=qK@$rkR?gs
zHpst%{!`-?kZI5a`#}?@jVi1_Agd_#P3fHmTj#=4`iI<%SkzHJseY5Asjm~rQnVqhuf=mE%4hl`gVq%H)2Q#DXoeP153@{PGZ|E
z#GTLd_lZdxvbX=Ur826HtnKb@e|%vK96#`it4z3}E63G5|2d?9WuEbejVXmFJ4VQh
zChZoY6{%x$eWVTyIsyUtnhNU0D>VnxZs=wn$!OD15%5S(xlC(u3p{uHds+IrE-E?V
z+<)u7IiMfCv%5%PhvM1FA8*_)LgzzcU0W5JQANg3%1+HRSmOU-be`o8^7J`Eyh2I`
z2_qHLW?Lq>(;M=AKd2qK+mDs)6ZM1tQFPw_SiOH7H?mbWsU&-sP2AUomSki_lB8sW
z%p#R7(Xuj9NoFB?pX;*8US(xw`y^!-`QG1u;C`GR?sLxjysy{m`P4n6`YW(w7&R(<
zj$ao30S=teUmxRTgB)L*-JX$sb_S>RMKAj2pf;;9%%V^aE>o44@DzpR|
z)dwn#ql}dJY1B03uXGl;YKpG@JbMJE9V=oJQ&onVaM!u>y9k`2ANf~X)&RlnW
zW2uG#mc9!Mv<%Y3VO&#+f1U$;+pt{q^J{Oo-sIhpKJJfi1U;q~dmRl$=Lg@8+1J2>
z+jNh1y3AnspA!xW8Ae$Bs(tO_Iy>koNO?1jHUL*sO{qNB&x8|SRaudg49q#Q>Bur<
zg!2R2u6r_B!_+rUmvkm>;EtV?l+(SBF@xj9tB?ETvCo^ALpr>=P?>*AK_}P(FLfWf
zJ+tir!HX6#&MP+1z{4w;wNewS?vp9-SLBBi_RbfpCnVsr)y#*>3AR{}GvSc6pgYvE
zklg;*Qw^1j><>?Em%$fN75^ko5)c=Tb?f~i_oGhb^~x@lLAIF=Mv(d*stL9(k+og0
zaV7I`b2f(OW+ujl1qrC-T^wGvR0~h8U-<+&{c+k-_Pz#_SCF!HRODhYf~lIUZu~LT
z@Q{<2R^W0q3=SN6d)nC(Dv`}jyEBn^;6ACaav&XEQ)PQmnu6g$!r6*xq!N-2;~(7B
zy4XNP0q=WZsv?bXoGS2rdrwF(mQ
zweIK0d0izce%3JJ@hYx_
z0^(D=f-#klu8MrbT~P)#26HbDkJRH88REf2NdvAQ*yL`$(TJsw$2jClHj>ZDik8AU
z$o)4lQJ*049A~z{*&^Unha>OFXDtfV;~V_EHpX=IIFo;P&VHm0Po2-{e?OX;F;>LzYkcBlqZ9(w-h^{aTEFKbV+)LX_g^?KZyS>E$?j
z<-I6ZQU!k4v1@+pTLo5BTzci9U4Xx@2CMyYEW*JLA_p2^2^Nf)eJAjqoFAIGAMPz(
zhNX1f^bSNc;3EBO-+%Gs{^5rS`JmMrY_(v&vP_S0;b&%z>w{Ie=dhK>U|k)a^O;^8
z@2$nYPIOYd=HwpX8#(E7{}BGj+_iP6ioj#X=PC~ARAYC0GwQ1+G0u)1%VR%}anp1=
zFV(RsJa8bdQ+d4<5C1Vg6TDSH{+=25Pk^HmD;|4HVrHzsGQvC-Y3v2q#UNJs=!Hxy
zz*rQ=w^NE^;$#9&P?qB!&fNWmVMW-5{)D-zVmdZ8PZ6b+jmPt;IWJCkR$N2J`*qcp0zT_i6`fgb;9;@p7?)`hyR+p4NUqnQE~X`Y22hKQYu83fFH}bd`Q;0
zhd3V;uL`11$>2uFLDBqdF>nV$lcRX)FJ*~$>?bJ&6$i7RR5BHzK_N7YSC6>wi
z!gedqlfkaA<2!508>j}A>bK0Cm(Jpgo+|AdK8o|U~b$OeT0Wvy9G6xh(!yAbg
ztp0ou!jg3_cir^1P)sQFuJuyG34hg%+-o(Vp~2i2n}h4%Uhs(dEm}Uz^5N>{crq6p
zPY^%3q;(p;P!@gvH%As`%7iIjm*vHt^YUBb9>2i&^B7ewPg
zJho37yYj(r+J)uv>y(f;>@>wIy=AnJ!R>XRxgXeccl_4Fy&&!e+s(+oOi)X_|7zO!
zS9E}pNM!KrK{jbV=l3Uff$C3hntZgYkZ;Kik*?l#BrI@tWZZNLdGS{nanH7Z!w=+^
z8d9?nkuCgGIopkM+iGfJ`c-2V!=lf#+9M>T-i9UY%UnjG}WXZ`y>=0d>Y
z>~YSzWFD~Bd_As}kOv$G3?k0sLS!KCGpAyh1$4xX;uh$9(Ez;$)AoE9u#8=6KXSGj
zeOcFIi1fRGUJT33E(FB@m-X55nYJi|A4#u^xip~d+7%U>-??b$&h~|WPh$W#cgzf1
z(|vTwgYrVogfH?9{HZ=ST8vJWQ&ZAV#sdz4;E3WWTi~a6{t4$lFLVXo&Ab$M8z@}%
zV-1qR&_ZKq|xc*UC+P=x(zbjhwkBs1am&CB8%gkr@m$^_4GB3E%fabmrkG
z!SmhMlYtIkT3@I3B)QLa?R7rSe4H85JHqYQdF>C;L~TT`bbB9|n%=`p&B}!;qJM{e
zaybUOe#2il1He(T3o_YIJO&*s^4)%_+EADrY-+&qN1b=Pklk^ew=w<_nx
z-y(Yj3l6w&{`*P|sdBgprt=HgFc&fok5|9ds?LgSCpf5Ijxhov)Y(c1>dyLEDwFx*3YOm5Wg
z?z~6w!g~#wh2>E&=^j-Mp#~UwpUD2cq1dlw-j02(rH;
zq4X+91aRKF)6%dZhfZF7YsNi$4qR>A_?F#n2eOIW-S6QvU?HdWmDS&a2-iPxqgfXR
z>D61@+#Y98zEC|h`Dl&kq`U?5IO))jaZNR$7G5w@A?P1oD2B>l5`Dh%IpE90-zu?m
z64?Ar+}M&!L3>5E;?6jK1oI)(XI1_yM&sH)N;yJnLDci*8~m@~grZ%SW*U1y88gPUZjp^x>AJrb+O>fXGlnepHMBsLr=RtDz-`oF7h&GtbRRtb
z=y)ki;t}enG>Gg>u0u7drO_G3t|KlLq4_i|chKUN_%7bYANeHxt-c-o5>z^yG6`QQ
z1Z{oq_Pje~3b?q|dMG(=qdn!zO^KHt0?U$#E1jxuP{#?qTbIwhLf5<2=ym#*L2iZ5
z`wqWVbfT-{OefzKaGI{O)!(xKzG>UpHCm|Qj}hs_d2IH;B43IO0a(=T70aUL@I6s=QS@U1qJk=ftZk47ioHYc~UyeYvu)7vU;Cy
ze-DF~B7tiYQXR;!tCI6b*(_L=x|Ze^N(po6*xpK!&u8A%FtX4A%K8l&NYP3`m&D`og3ZUVwV)~I>y{Hx`$$)Wzl9<8S-W`hT)Yx~
zVOvXWF|-Nt`wx{@$WJ1!C#?yEL#xQv3x-7o%0cIJ;}m}bCH!uIB*8pOk6)iGyZLCA
z9=bHXH*BY3!UZ;OF8BK|htvc>b36=us`)es5sR#NZ(iNKh>NM|BxGycTeg;D}CvIFK?RR
zgSQeHEesqW@h7B0^=<(6EB1YvfI_fXB(+Czya1+nDzjqQtpYO>N~0#q^f124!jFBu
zq3a4b@0)1@M_u)I#P1s6)!G#C;rBkl^(vpi+CJ6U=jZ_K-`D5=Z=?^;w;^11|h2;N6-obaz$gfIG
zH^}A5Uac+S2fH+j<8N>TVlQPQ@sbl^IHbm<%Rj{!>(?odTkDoX6=@N=t&DQ$urB-Fy@bKZV^uk
zG#8lIs%cJxlm~+JE{CT>fxkjxJm*s2c_!N%qK2;_Ykp!X9WcXbtfu*#%l9Cy_@Jug
z+zaUWa8Yb$F%o_$W%;Gw_8jWilcZ(&sUZBhHc3T=Df|
zt{z%Bd%XMpb#CdAyLh@FiK^G*9;WK1J~Nnai+i-ois*A~;Sb_Tc$0#fykC%U^ZPFmzb}>7w)$3#e4gYt>b=3wx=l7;5BuW*r!|pP2V3}I
zA%$z+!2!N>VQnm8b;NDz58H}^tRa0AqkfB`8~M4l9+y3S5{8XX1f}yBVS3``#py6t
zxa-#0VSmXDKKM0~d!a=fE~vk+;qJSH5C365BUXPNYxH#dq!voT<*ZjsTC#lb_y+nL
z-|0&B6>YwmTGhe1_g|x&#;aJXdRqJ8KM(w){mD^o0ZVB7^Q7CIiw3Y_cw>;dLkGvd
zGulhlCxlB{FB_kFb_+5cReaok(*Rl=D7!@Gs)dvPoDIkp=Y%gGr+qocCkG2AKwNsH
zF$QjKEV&#P@YTJ4H9?&qY->X&Rxcce$J<4wv4bE!Wn&*VpQVp?V#IY6j~<7I37L#G
zCuDhbno#pz=Z*=A
zFP;bCCptfV7G}Z1$0@hD{Mew|_LJV{mIy6=zo;B_YYrLFvx-k>{{YXcO`e_2VuX7_
zRg5$;$^Ig~yb~^crhLa51I@o8*P>H90K5OCYPG}PfycQOXO(zzKIwQz(A^jA
zAWuP>E6hN~)<$q{{}0Cng=+MlLKPFB$pakWq;#s~IFRnH^Y>xsZ$RgF
zpHaT32mXcr_poR_5u6EEvENV%M5!Mh9Uh&%3H;cj9IpA@0$B?ES}b28fRpy#
zd5YK=^kYb2&-GJR0iSKDx&)^MkX)S3?zM15Ui)IsIgEHBvFPu7Hm%Xa}yv^04_NK%=|jfNJIa+E8J2S|yeKGDSMUbk$Bp2{LElO9>(7UT}wa(VVo|7VJgR(mQAzZ4yW_&v?#t
zXrhbo7ifNUP@EOnVvl
z(PqrTtE5{ohDoWb6zIi5d-bXy53o?0XVc6+4orGO8$YqNlD26-x`n+SA}qr7nK$>%
zk*sF*|3+@nq8oJ!I(zpqp(~-CZUT=h2+N}SeLo(ykdB9@OiPsx5I%eiTJ0f!Ngh#0
z9^2b25`XNfnZMdJ5@`W%e?5O(L;A_}OEG!6nz$Nbe7MEq8!@SUN+s_ZA5dvKy!Sx$
z4G~L*U^*K12v()5=(2?
zfP(b_+lpygv;n{W{_pEL>DTyom@1?Ra@*1(HuD{j#OoYQ-GfPJBKz3eZINRpj;t<$Ig
z@irxaanM{;!#`YSgcwQ_HT=T8QG}N(uli6p(5L$JB{41)O&nY=9NgE8Hlxl;a2-1e
zq8+9v=BeDkvbua)r)m(2dh$Sf;z$Y*Y|F79x?BKw8#B$Kd;|c)6SFs9?`8Di(aI!;
zvnM#Dn56cUoF{6x`0`4;r2v>DT#43;WP*o-kE5(O7VKbn`Kul!51b55>=*Xn$F@=@
zE+*-p!XM7j`tG9{2R@rvknYS^RG&QA~cR2w4z@y!S@99tILyh@j0QP@o(fFO9}X@eVBRz9j^5TY`xoH
zXFoA=#z78#Had>FgcT2emf<@?Pls*p$b>JD
zG~m0=@;_d2`=HpdeH_92$^P_Tq|MDxUKnb0S3kT?9B2P6xUJBmi3?M8yzbh#!HfvH
zcGr+FIH}ehUsDx~ufBN{)BIl)b`xM^eCO?jF9nKeemHM{Z;q@ipR2Wo6!NmC+ahft
z{cx=Fgpem*AI)-$qr8XDP_O)L+jAVRSHXP&ca8CxW7J|Jx6eU(dg!!mc^%TbUUv}P
zxd0aq>6ol^E5bF&u7Bzbr{O!xfPFdv2KYjx{t0yK4Bp~wJQ(1A3cg@$>XeoTWS*;4
zvT8ghIUrKymz0rOiGn(|(%{iOml*?|4zb@RtLK&LXnWvp#FU9aY&#f2m
z8+Ej08e=rHy+M^Qbutoy%cts@yzb%TjtuL{hI^Q_S#RbyYllOQG_kAMk>KX{|D6W+=YN|)I-$Ac!D#}MCDJlJ`6I_#fgzk^hNE3ArdTC}t?W0M(=Z&|DI6v#j};^m
z9F~-i!E^j-`T@5UU`LH#_gl?Kn0~`daBKT9tgw$5v_Iqs_lOzhQLww=S;ta_)4SgI
z>ofX|^*Iyx)0f`<2frx{oznA@G_qcmsK2yYr%DhaM%F@XJda7q5&7aBZUn?ZMeB~OV{kIA?y$Q>YSEx7Ou8FA74Bm1WBvA
z$3-ONu%(c=llCQmC2~rPcj*-15F_pWBbVi%=g;M}z+eWbBfhEk?5h}7)hph&RVjvF
z*PdwZH}S?&CpXlDf_!n6)O${2FB{yKTKwKL?>3~pvy^gs^$yI`Dsp}|!GphUh=tu)
zQh+QoRJodS)==$*)YMOJ8^|l*@I7P83{p+R=4R`ig^G{u4}80-3&oe8r@Kfiz*q9a
zW1-VpSo^nO;9KVNxcf&#)j~YL@dIV?^C>5>z`VcF*UP8K^R?AZqJaP&Y4CO&>lK9&
zeN~=JULsh#R;OzNiQ$Gh*IbbSA-H2*XZ?;`CHJ#?#qSxd2s@{;U9ZUBfv<-HzBL}b
z1Gf&D%=w-Y!bhX4w%em^Fhyy9q(QY2cJRQGaRQg%UGs!A>Hbx~<8-e0SA;P>+Oh8M
zlBb004wjaLTt5cQeN0rFuj^ogU#a^RL(gK1Z(%E6G`XN^e4sSj4|!O5^5*+vFZrPy
z8^Jq%ND$r*t1Du%B*9bz^WFJY5vb{rZkZw>h^?5qWBzgo;7jVtigBFEa6Chq)l7}N
zSCOx+mNb{akBCR%Ep~fxe78-HzcU+zrpMG{lT2`YtO@_s90j;YCUwRBCk@>NuJA5V
zieXdzIphD7{-6wI-CD`xvQR9l8UNXr3YHdb
zQhqxw04KLYH(oua$2ZRoE6KHNBTf@@qXxw_;Lr8Ny6Q_FAi!7ZFh(KVt;4FgCj1k%
zEIbTni`oQIx0zZ`-JV84es_2MLPwC@#={RWUYzi{fmE^3Fe5y0(>+Y{#SDrPu&MG}
z$VJ&d7h_*8PXo2*E+MTyf1$Qxr-MDi*&vJb-71m&J?N_Um5fMo9%MsYX-+7A5pbT=
z3tz3Hh4qDLrzLEE0I3+8mmwF&&>Cc_4gl>%)6fgdU_m=TV5s13AJ0hbNxK0q=ly
zi3I0g;5XO&4)*-x@
z+(-L+^V_!^ZIQ{;ZK3)vAHnrAr(80&>#<{Qge9YpWRcW;az$Et&bNHT{5-YTW$p^?tE7k+s+ljY1<{}_
zqc=D6=l+tMP71{rCoL0uRBZS*De4Ip{sOlyW
zD;H0UEJ^&DU4+XeEt0z6s$rSeMZ#(==}oj`0EzXH#OidwbJBe_%~hMdtVD;FUqQ+`
zCCP_FG$i(d5a|_7>Xs~}4e{y7@H^KGFH(1{)}fgBm8$F90*_9aGY}fsG2i6DN>%0n
zrRAOz7f3=Wu1Zz|_X#=Pv$ghjIY`5jK7!^=M@g&nu9k)$rHI7CkSp^MBn7qnr*>T`
zj3g;wS1SkZkrGF4CQ@x>6UWYmP*gS`k}`F0)vo&j5x5Vt{QmTZB-p!q(IqL36rpVu
zGH(8fm}+>X&s<$j0>T;1R+?XkY%ynAhUsY{rK(R}@4XvQ_t*)Fs|S&|gM576P)89r
z?+ou-St}-Kjb<^qoy{ZmYkk-BFS8`n^#x9}a6w{SpnxGq!GXlkWA~`hBZat}+;GN*
zwt=+&t$`cIi$IcW)4}un38BQ$KCyiErDr5Xf$p~bq!?n71nZI-1_^i0=67RmtVpYX
z^Pz1uI|v3dVYwkk(RJYLtk5!Oprc*-m}}9G)Z_ky<+N+TPC%bkjfc@*pQpO-J-Ab
zApoY;&Ntd);=4P4;lVma<8xD|E1GFQV4>?;_$VC!=WZ5yfmF|oW4
zeCpinQ%m2V{)ctzy$7$MZ=Jy~NK_C-#rIgmEuTj2+;LtnubKn$TjmDOJh}J&v|#!I
zUp#W%8VX!fr$b@;ivGEB90#sm{MQ}%wNRI%4!fZ0b-=SKox^Qy30ygRA1Rz+1L4Yi
zPS*b&MXZVIqUWmQfo-7v@BUFUbd|1K?bJRC@Jy(NPGE-#G044Rvng06XuS!+I~Ust
zfy6l;lF&&|B;*lH!RLd7q)2+7q~0S5zZuS@1Jz(zCgM3+@=fOZf0gXgoCn6yTz;mM
z)G%{Oa?35@9lG=~tl#flDmcS2CsO8>gVL?;*=AEDgQ*cxq4CTYaQyAT5YBct5WKLq
zDO4Pa)`p^wG(Cw0W>Hx}CsVS}&(s~d)rNZXN|tZfPwhNVdb_BrQR)WH-iqlP{TYHj
z-zr!RR7(ZH0V65n+GQa1+o`JB2tV}p{)sOwY)+up32jX0sv-8#M_Yrk$>mW9I|kx8F?>^Uif;d2Mk=wK;UZlPXlP@{sw^
zz)f*{YOw89`MnE}*Zl_1L!FcOW8mJyGJnnR(`C_Dno=9cw>&rc&!`A)>bc|?-XI8T
zgzxE{(U63~Gtwf)N__ZQAkVE=NB)ArJC{{-c&VW04cornC!E-DdntVHtA9X+RAl*c
zh#&U4jjMPyt|Id*O2ZeP^q>MeQq7l6dU$U4nU+!!J?0u~ZVELqC
zL0gs*zcTApf1+{r<;)?R=%dB>X-xq?`(SjmC==jO|0jK19IDu&lb!ao
zqyk=ec=I(a<58G#Q+EC|nfoVlDd{JR8V{VVQ&Iyg8{l&|ZAE!yE7%?H{2UU(f%7%u
zQUxQ){DN!!Z$~}Gfwr=rr?3op9<;_6g4h_~B$d4Q+>Z%xI7CciQ-&JP^@rEck^5F*
zeSL?#B3Q5}&+^}dH%xev@zxu?8h$t`bezGVMhJE`O_St5u7f9F{(RS@FfKf#|7?~+
z5PwcL>DVKrfE9-wjvf9di9Oo#K6QTGgU=Qi-)r*YhHvhS1h5$kzzdr#uQIQF1|;6f
zmAwx;Q0{h{czI(MIOzLEV&LHt2;+Ev@uj=c!|P69x9DE~@kmWWwxrjmk2`)HwB#pK5aPAZnK#5TI9H2GjvNtelM-
z;QiQxdjs1?@c#Mop2hS0cp-t)p&~{OiVBvr(!T((C~Qft#$E{y%N(;h_F4eGE^Rxt
z-(M6u^@|A3D@s7QO`DJ6i{u=l>M@F+y%hMy%6~<=krROHgD}+|t#J?<>c?kGw+DOI
z);V4rWyDdee|TK5ZSTsUp)8z-bJ}4MNi?4gD@+d8+p#9
zyNEM{7V72FTkAMe!knz`evUiTxW!kZa*x+E=(u3F?Vz&*BpfZJ!)g%X_`@z4e`o<|
zyN^x8)8}2J~|k<
zhEl(OPG;M62lsd%B#%Cy2A@)#iY#?{K=%9drPe)s&}ql#+ZYQAOoK)_Kg8z|XFaIKQzb&Re0>@ertg|>GgCv4I;yDM@CtB2kf+m%<}2!tyOJ6$P!B#pYwDGOk3fp5
zX;bdt52V>vp6c_pA2na}Oq47s1CKj>Z_u>+qma`}e?0ZF(R$I-)nD){xcvT=c%N<$
z;P{eFrArZu#18-T`TjZ*<-S_KQ<;+j3~G7~*M#(fq2@CeBOX?wWRh{i%Z4oEWO}Rc
z!AJ@S^OYN=<$r{R)Uym6_RfRi$SLPGYhU0<92veFUx_RO&JgjI`M{0pQ}~c^2>4(#
zoVq#l5s>o$X$g#ZD9*__BGcgvGT6DMdh|dzNC|9vbju?YZ4l}PicNV)`x&R`cPV$E
z0IT?G=lsDE!TBrgst)qERGk|!_R+a0mJAg)H_2Ex-UsU)_x%1p}
zM?i0#`%QxUfgWp+TfEK^g=S=7;DfYhh}mO*;+OBPASc--O`H8VI0w0zC>R~l*gUUl
z|2=!KJ)mcyxEYJ=*6v<@PqGDjx_k?r`z`}x3;&BxR$P&jaGa9qg{LTwpGG)m(HXrR
zJ$&u3p%vgi`|>K4lQ!D>a=(a{B1DyIywbz+`XD$;y;$SIRS-6N%|t#_359$rQ>Y><
za7{8IUp+Xk1f-9}b*zk-f>na{PqvdGQk|rKIaRPh>gY3{&g2p%`?Ra3B%P&@%Liij
zhmi}?Tf6+*?g%^hOG8UXcq;%86V@*%K*{
zU)kd??spl%+rz(|J0nzq(RksXZa711OB~K-ezr}N4F-n_wQP_SX?i%)?#>XbIJHlx
zRTf1oMQ)19|0R}~8*88TZ;}=#>So_BED<%w{1^LfRFk~ljBIhG90iMEmhY3uzMBO$
zX*tQ-Rf6kFUFY|N0aDU&ijGTsjp?
zBazOAdPVRJk$zQ-O^0Rd2kjQ$-0!;9lgw5A@nxZ8!r&~0jWW?sde5AbgQGe~
z4qc@Mt6sUp$Ytx`YyZ6<*%s*CGmm~vqIUP5SnmHu;+Rn(1REO&>)uAG+fsk42oW9n
z!+&0q9IxbI9aVTq&O;o;>Www^ZOI%DcA5XGh=JttmJ&H)Z_BhwH^l=|
z*Qi?tjBTxA^pc$3%=uom|G?c%{D_(~>Gyme(@8nv-M8wByT%eEl?jFvRpLw4;4%K?
zp|OwZqT$}AKl6L4R*vqgo2_6b+17>eYmTT9pIfv4wVfGMH{6_G@meaZGP<5;?mgwL
z&ig<*u6yrN)t~xc)k7bZNG#{hpBwKOsESc^f3!z(qzWgi9gN$VQxCY5b5=uxhAX*drVCNyhzz_o8h$nT~e>f!&as{p9#KwqH=o;
z&Jbr_Ke5hcH6xjsYKcCuxIqMR$;wn52qEQD#TQ4Lg_DK??-ki43lYWEJGx0aj3guB
zZ#|RiOr*B<$;Q+a3nJZ2xs)R|h{WlqXp(bIl*AnH)H8!qjHrLzPpxo7io}-nO0gqM
zmXt!bE84;xLfrX7TeTpjM2PP**)&qnB(0!82P&y^#D1vn^R)01DP?drW!W``6!2o?
zme`9LVnx6rxp;se%@?cRtUK38n9}~`kr)3)_`VyMO@N_f^n^3V4Y&hrXcQSrYegRt&wyb=AY~9SS$_G#p&QSL)n__j+kIc(>S(2sHW&|&2Hq$lx|qg%`K7-RH_$UF#n^)=QKHzE
zGo1@)T07)qs|z7opQ_vX@wytrV7f*2Kv8tT_j}X$%
zbrf!TJ3%T}a*(Og76JtcfnyulW=Nwq+4Fn<71T7V7B9Lj3tB0gi%32)=#~h@jAnoo
zXd4_lb!h*2;835SgYG+{ON9I3R;@-<8nH3?%D5T0Zw{Z27imS?4|zh_ljlHdOGrmx
z`){&O0kg<6C!v>DhYtM*V}UFs`+`Jb8ge@8vKY;k3cjya`1MEhf~W`E|a);>`W==*)*G
z^!R^Pncv@m>z_o9>`Y{W$HQ&rN(abZ1U^!j!L4h^w2<2Dk<1a`cR$v8EHwe`fv*n7
zE1UvnoBm6l8U6*By^5cx#;%|Z_r?zIhQBDW{IG%s+zai#=C_->P(pDoy6LQ#29P|N
zv3>RDSMY^&V*GypG@{rm!MjW5pBdVF{|IaP1GEn=NQp=$D5K|C|a-IG0_pS}a>+c&JNHbkL2PV&<&2Zq45wM*tfwh3_E>m@lL
zFa{_FZ`|stPY1l|t3NaVMz_6-uAZjuLynoQ7Qy4aD1w#t+2f<#Xk?~60RcF(XS;G(Zo~RsTSWL?Pc#0i6$cYDN6*A*4i<>feN|Pv7
zV()YR#cjY>;LjrCN6t@54nAX6o}FW8({q?Eq>
z12M`TsDSh1z+=x3sSCT`LCw>c+AsUdLA;%K{GQ+as8KGlBe1d`sk(aEDQuCi&)(qn
zqsw1Vuizh|Z*Lb8$-LUnar_bDNHvu+n=1i*bu$OgE>(laA20g5|E&X=X-+4Z#~VOF
zbmG<6i=V(m^vlae`<|ftZuSC{J_VqjBW@*dxD>EG`J8FR8-a9ehaca)T!6#^4P#B{
zJ|K@D=TS`iBFa6|c}2Zu1A($T6rxyV!&DQCq{%pj1
zlj5B?IhPXNQreN<=MG%LEjs4z`JpTRQZH8uvXN7c6k&8c0(@s=uG)X-J@9H83()>Z
z=5-6Src2+|AY+H##r$;~GPzYIp|1ZHjjK<8%_j564IGv$
zf_R%zLNo8N_?9%3cc0ne?d?4JzrALBM2nC
ze4X{{jy9mHPGFO}{s7&!ntjb?>J17*(kmR2V!`IK>!_TPoHKFGR;y^gj!F$)wyIpT
z23_Y>edd%N1JgHlkNgYB9La|z^=+<>sLAGMcW1^GF!NqdgV*8=id`BxG2#9WH>Oauh{k|-4+E&r3!|;4nVQjd81;q6agi%J)zfa2u63S_)cFwhv*|e726u+
z6Q^wmn#XRoprq8OfnoM6z-|u=YF?_NtY~f3XUziO`@sBt6FFhPBu)+kJ-UFz9ccM%
z3r?XN3y*b&9w}haNvBhpy-GwL=kzlRJBk!J&q#jhJcrEQ-R{hPaT*9Qi3Ll!1`z>z
z@0*p}k0N62rBeFpQQ+|Xy;6;~9eQ&^?OnIEI^cV}7bHnYfSSQ4_0l_1NOaaq`q=)X
zz`S$u*OddjV3+-s>eqG|AZ#SBtXfWka;B`#%JvYsOy
zqd7Y7lr}>|H_`iWarKe15(^(YM~Q;j>b+w@G?d`tSo{YWi8bO$d-pHX|E5S)R7Uw7
z9Bm|Arbx$}BMRW&mIHD>7{Gw=TurX+8WDN^UYol6plS%^}?e(yrEw7su&bLS>cGHc=~%wDZJ=
zOH}j)X`k_dqHkyON%v!GlX^v3h(=wBhxUbur1V=A2~HA$qym8pJ01q_iNm?PP2Ry(
zq-m2Md`})W5h7u_+U0Lvl5Q=Y?EAs%P5R3gcOv6YJR#sB*YCwyLtGy7J02MEj=1%7
zcKQCn2oio2C#{PoxO9a-C7JQeOn&dQC8U3J#(_U_
zq-dye+wQg%Nn~-V_1tI(N!W=hepKF{c!byeJ-%6C)FpG2hY?5RfmQkubWaim`%T~+yE6!En=NZL+#;%s)U$V&j
zz2E@uiu)6GNUEEB-fJBg`LYI^jPH2@rpmC})jUTE(HvYdv!Yxs@EzOIh!l;B_2Z{u
zY_v0uW%zRJu6oq35?pFw@+Fk?3a?tcb{G3PfVt}nwIa)gu}1H6Y7QEV2bTjJd>l(K
z<%3J-bXd|bxA}9aLGuFm!FkPnKj%k$>yw|rTAvrb+qpqg)}DHWihnbvng1!mI!OPVI^lr(
zkMkLBmGGW4Xdh9{2QHm}fVgZcIU{j~6Efh$B;
z{6V#5*e_G?X~3f%zWp6wvesS$z5Nsl(`4G<`FHJ~%$8fAYyNm;O+zz8HL2wae;c6q
zpFNF26?HJ{34YZK-OSS4a#S}}s)^Vl6z
z>P7^WlV88q=BtNFp7DVt`)c55C$q|hYX~}51f^{LMv%E|r~kY)h9T}0POR4(;IlIc
z-TSy}VfnYSuU(EI$a`-3i{&2zp1!Wa$xi;g=pUQ<2Ur^5lh&md3Wsaq!-sbQ0?7Pz
z@#;8n1#Ox
z4m^GQqLaG{sxgdLzv(K48ADI3i)KooR59yM(>eqVM+zD)`XShM(0S3@p6p2vHF;Z@
zTngh31$|3aFNMq$Mb9dG5G=fM>upCLf*lP+s_$4O+!s5l$v0gJZ6_ivRoY4*qe0L*
z6Hh73e&DL9!B_%M>2@_I|0{&ncM^|igyusb>Ga1UQh6}1$x1adtpxHNbnu`Sj9oF7<%%!L=_4bK)-|!{-zt*
zFnI`&|D{=Q>&@}vx#kkct0%@-Y*tL}r}N9M-OGo@9)n+eBC_B^bG_=x(NuUgX0TrO
zYzZ`dRv9y-Tm;`<_I^(P;}iTnN6UJ9G!t@G^85=ro(A_Xya1)Dc~IXzq%X?mBa{wz
zdmFX&76y7h6-%$Y3;o(;AF(Vw#63anch$6V;nZ|EU64!`6ymJyOAR7go=N!e3EDhiP;Dk4jWtl2`gc&Np}{Exx9DJ^lQtDeNf{DJ;oXJd#D
z1!Ihf%kJ=d(pKAhH-!@yUDm3zkE9VAn@=PuPUR9CJ9>BiQGH8fQ^S__`}h(2?L%Md
z+eQ=XUvo7GAIc!kn#}$k_s%EQZ~xP3RZvLC%vmhl7<^2aX2;ZpYKtE;0d+i7JmJ^pE7LwoPR1vkS1}YU*>InWj
z8o`cR-V>|1AG+T0ses?t){t|ts3P9F?J&9DR7)_d57qyp*g%{JGI+X^r-^vOp%umx
zSVc5x=x_K7zhAD(+2p^4Qcq|+H|w$DZzBHoCrxm_YbNY_C{sSm)r3^=2rK(+4RLeT
zwc>XFI)c}{{ZDXLEn)qDw`+?x^&gq_e8ilv$JUCdjhB&?5`?sB%U~HPMF3t5ueh-S_;-R5o#Ge
zQ~sJwL@LL~$)ab?MAy)>Z-VPu2_7*XeG5t}aY*LRRqyFmqH6APjo)B1Azdl(L96Hk
zap<76(4y~0!b5%K!Rew;#5rMcGx&R{#6!AgGNZq1h^B;`qjVCrM8H48A^tP9giz$y
z60!UmVkS)Zq*Q-3Ve{vl_v80y3m}MJ*p%gh0&=EZ7d~R=fC`k-cdxnd$>}NOl>6EGwmW?th^@}-&5|J
z@z)X*SN7evW-16Hx-AKZj0i$q{l0o_M>An#+OlB6)69%nbg%`tzz6H^OH_
z2!Xe`=4V@OgY{aE_T)f6MV)mU=etS?x1ftpl$Uad5ZTgyTJaG?gADicCiuUF5tBvy
z-hPo_Md)>mpz9{EV;9@Ec(D9*rReFm9B`nkaY*K3
z5qRVEoH?!X9Wd0;I-XZ^8+6)gZBFBj0Xr^BnK(a$-=h>gqusNq1h}pgr?MU*K*H=V
zt&|PB!O5D-u17aN2U*@}hCUHFVEK&urns;Yknu6y?4w{gP}?T@(mkjMgoJf@jOdhr
z);-h9*K*3i24&tuKUFHhqnCT6A9+-Q%7p1_mUOB!+vOm
zi@6F=t%o&vjjF)P>^LQK@EzDNw-hz}n*erkYL3CKl|V&Qg3@=T8vGRqa^CW!2CVKZ
z?Xds^xaupDn9f}Rh9c|2rM^@FZ?#O`hxcp20nak)jS+RA`={Hfe)xZW7b{q#FVIy2
z-lgjVpGpn*CZ1^+z*rBuzwPq~S^pkPx!kTVPpt&KrXg#}Fvss}FTwI2I-hsJT(80(3N~vv1l`0k#DlJ0>t&4Ok9*42%-1
z0Vzac#9~l2U^_D`U~{huyi$`@5r0_;?ALxW&oX)s{_2DUn1|Mb3pP``{zBf~WukL8
zqOS&whyG-^d9E6Wr?Y$=-`xcGc65H!FoFK}iws4Vya$`QG8y++)qw|Ey%+U<)PP$e
zSsBYu32-7wu>O9^J5bZ|sx#*{A0YT-Vm4$rtO=Koh6Q=WZ$Zf7>^b*)Fdo
zfan{|tJ~%Z0i9@V+REizkaH$$Ratm8a1FWrOE>}kzD&Wk<_++F$JgrW9_fJp`_mH8~rsrx@~PR%ZZC&E+*;)#5;{
z(zkWT;_m~cOBbaXFPall`dcG^PvrvTTv@R;(H!tDuK7-IZ~`!UHh7nB(F+_sQN50#
z;ykhW>NK5*O9AMOS*0x}m0@8P8Y6
z5g)fNF^egs5DpguLK``9h_$mKCh`?l#PMyy?;QAph`FD>n%m>!h^`W0Yt6%H1TTet
zP(&r0Sk65Y5VCAaOt8+LKV#-Y%oHBhcP)z|1jVlKC(0%fn?C&Ve=(j$xH@@J_C12X
zk7?$3U`s{-VQ?hoj*a6BVz_at>l0%Vam%p6_ndPIv5+fR9mjm1NZRG}>gxKZMErN(
z?Y!p_h-6*)rV<)~gvy6uP7%f|;^%(du<7P3Lie9#ZpP;vBKy(pip_0##GQKQ2WzPFa)sFpLf+%nKYh&{!a85}!3tjQNNmu-a7=2E-dY9)B9P0U7c;BWI9g&UO(!DZ>sO(W|=`&eGE<;ag
z`FsxH?{?t6`u$u&W+%6ac27Pbkg$(!Z*T$8ne-qr`}JGGym;vTnZhEXI
z<4ZUiJDo>Fcst5Ue0fW359+&8e7%5ZAMMok%Pt~V8SeEq*g+nvEzvp$|0gVKB79((
z=PhySaG0KCaRG7HTQ-{^w1^mCiOzLfE+QrbcDPUc$RlJoZc~!{nosndfA=Prvyd3S
zXP~v`Q86L3QYg%3SV9ckmSg{6lS{a-x0X&*E{3;st9Vi
z{?AJmHAI7+2IqxSaNN^5^HstGF_j@%&U>kXm`icge|WEoFk)}C+y1DAI2Re4Ikcgi
z7~4N}b?54GBJV^}v(ck+B4_MqRmJmaqV-tQ${zn(qL|*`fgf`fv3+?%=-M+iL^A*A
z8)ADkQM~swzl~ZQVKW`J+VD&xvB9(d@={j~VI(5DX-i%;Avkn$ldWzYF=v$2`$D
z@C>UViuBxiT7Bz@TJbNO#)-|u$uq<
zqG1jV!)e%ljkKe;CozIn-bq6?TDcUhJvUna+i2xJG`ll2eWfMY&QY2i;6Tb-Xz~tP
zf4{28@}#RI9=J~8>nkL#xR7}NIEnVO{^Dr;f1=3?wEnquNV~)vBvxCIs7aI4E|Kz?
zHzdZZkSO+;#2<1b#(W`hb>LV0`S;u+<&%me&S#QHH$v+9hug7T9w5t8?vfbth{V{8
zF07}vcVes(CCfiY)AU>aXmFjmHP@^KfiU4hw8tj}-mgj@CW-VVp5=iK5VOsGB%<=15brMt^~V03#pt?|3)%HKH5^4)
z)vw-y%5_VZ-4VL$Jfa}n7W{NK((_;VF>4DkS3WHcm4~SRoh*8;m#Wx_HZ?|MYekyO6PrM4zr|DM1
zb%@-`+9Nk1>Mb}p8bHkdbA1`L01QYVmJU3I}9mGikuWFY)uaoq})^L#D#
zL74np%G(^`&-t2*sGnxW)(Ss}`~e?hZ$bRBI4+L#{1>RVv?10xOi=P6DrKvU--F1c
zZoF+bM6UV6OsGBK>K#W>JCod-AE5ayyFMd>{Bhq9tAOxKfDMy1#6iUgFXT^WTR;ku
zJNAfQMtY;gp>l+3wwbrk{Mp2K0Px;G1)YxalCIC@As`u1Uh|Q8|*Uk}3Mw@a{&W|GM-l}TAHF(?(DUp8yq<>YKg=L`^hl&C!rxapwjnfXbiwO&F71q
zP`!s{t71?+Qx~R4G!GsO7pg2ER;@KHz68<1%3TTRZ5m70IzbHl9*g(I*rvQ-)PBwS
z?eK`B0J*>=c(hW$^nveqA3AnE_=V(M*Dv95@tvuUK(c4)JRV25Y8HWH>#3Y02m@60
za6H(zMt1_qjjTKI{@tnLg!f~VQQlkRhi0q@-d6%!#qfL%o35inGKSHNZ)rVSsU4VFmYf%S3W<*aZV@cFpf(r?x(RIp`{I222G4O
zUR=;(hZKPVj(JVsc(mp(8Ry&HX=D2|;KE&J;$EWf;orOr=W@Uog2^c+NRTo?dIOH}ccuO&X3nBHmVzDp3HVrVRF%aF
z&xfi?aJ#{02#h?JNWWUS$n%M9`3o*jx+;y^mDM228U5ODxhOwr@4iBw7rMM*xSX?f
z8Y6utd45@}C(oH~&R$&3+Ch%TF4F+Z(Me>xlkueeBX9Ejt5oE|pYujeLZ`dMgBp1V(PVi%3i+;~Qw&w)G1@)bi8bzDi52_*55D(R1H
zDk)!%A+eptZ!XgK$XZ9z&UGEx?mte_PFI#Jmpn+y>}w{l--fjFf4_GJmY-{p^P%yg
z2g~L(esq$?mte5Q<^D7t)kxD5-embb+IZb)_F6Q15o=OkZA#+n2PCG__?ro>9eF3R
zT#z;|jsm1S6hh*mizHUk_}Ett($8^ja=b&qq+XUb->+%o7oqvXNyEpq{@|j+=^1GAlTPFNTD1M*b(yq(NYfvpwXaEAFB%%8UY^$8JsRp7ljXO{NR0YT?vn?#
zq?~hv#HWcQ{wpOh;2*hP_|i#vWHX8C-${I6Kz^RjrpP$7ew-{9quJ#TkusktX(ux<
zfamwO87cD`kbVq_l5)aba$m;1Ao+6oJh|?vfQ&PH#rkpmKUYa!&ib9yJ3Eu{YwZkzLZH^?zZ+5#>i4K-UYyy
zRZu+qbn86E+qNYbTZbqZ4bsT=HP8EDdDn9?zerfajNASGfXo+8tPRKd%Kazsd80wK
z!snFW^foLjtI%P+NO38y=g8oL%X?;*lTrWeg>AU}>IFZ1u6=U3jr9?$EXi}GTLPa;
zmol8Oow=4cZtsB|3$~k>X~T9xq8t^dy$kF;ST3);i0etk?#1UE=q<(kKU2X7pJPdE
zJIauKxZz{Wd#g3vs*xNRO~vQvL4GVYw>kxE=R4YjaB=V1
zn+TV<9e5BwI#!T%AIY6-cEHd;0bd3F#UkwXeARR;qhG8
z+HD18lPjxZ5wBNcfAk*J8?)Xga28_EVG|$J-`?wO;i#RG54~=Pr*`^r_8CL;(_c|V
zJT8VVe-hOz5vrX*?V4?ze{TwLQJrP!0mQDo7Q@dWMy1<3BL3Ga_O{0mB8ywZvlxhW
z3&nY7Ao}+cgIQlG@c)Y57hi_5N@mVP1VlET66<(~=4t26^`TzLWzX>ph-N$V+K?aV
zoIJ~B5U1Ld&Lu-EZpoWK;V=_G2XO*Kz1D&VY-X~0
zpkA%bPN53=H6-LSSJwsSN0y;c5b9UJo*uql3fQ=LBNqdd7xVJcmfK+-;1Kuk>KDkH
z*~4NKU^(TJO#|Zh;?9CRN)Tt0t_h-fXZ`cy4$1@Mn@2JL#1&v#fLqxU&o3b*)SgdlGy425Hps(QlqtZ%u}~S|(1P|b;>X9w_s*jAb3ICW9`Wbuxg)V&5RVD{
zGC_H#m;RvM6^JheS&h)=l~k&ljq-z+TauFT3gh=J+X&Con;;bwW`0nzqR!Xx_xNzn}1g
zc)>@Y${S+ilPqgwSNd2u9n}l_dI{$#-P>$!xuI;~uqG}9qHpzEt1yU)F1njgK3IQG
zBp1z_-w(&tsNKw>munIJ03A5K2CdtXiTpEipTcoWgg>qW`6+hZI24tOm2%_w^R@c}
zj(2JaM=+jtq0dL{tk8YJaVmYX>>0ARkXwXTk#L;L8*#pKFS7trE|~ur%dJA=JEeW9
z3+W|S)FH)#`><36$J@Ox-pQIk`Klk^8Z^%RAHSSM_AOz5-H@GXQlB!y@qu|9*ENQx
z@OiNAge*QEoMy@UtG5h4u{>OJ0G|u8E+_FgSN}YZQFO;Etp7o~KRe4n-ZMpimcZvx
zSiULN+YN|f`ZXUFn@959p63`HC(hDX(fFQRn&_-0qQI-w?D0mp}g`
zh5bF@v5!2z&l%(LQwspwyRBox@`pUq&*Tb{2OT>@@}~2zNYtS5Ge^2UT<^m^(!Q&e
zw7>Y0w99`-V&65=kDYBKZ>sx5j&C`ew7)7s+GSR-;{MGe$#T<7lD{}sk~}Q$FUjj-
z=g53HU3ERS*ZoH3+2#5qkGg0=@-InR9&^HiZ2vkfA26Zi8@p({`br$B57H&kpT>{p
zX{bfx(Y`*U{?uy{skD5kie|?{%OADf$Z>8ippB!LwC7MIQPYS-=dI-D9^6jan@5u6
zS;-{E4U_c?XuLm|=I?*`(;Zs5fiG#_MC0E^G=G$7{C$Hnsb{B+172F<&&`FV_jV-Z
z1sX5!r}+tkBev6^@qPyy-!~5+%S&>|&(WP$|K?q?e3K4|IV`09&nNP8OQ)@$cZ#Im
zHJijlIub=h$$4-%N6G@UyuI`}DVJ!H^U-ch+V$=t%Wu%`|F`9n`qxx)JvMG3_sQsE
zQsyus{rKfVt|yQ6M!>=gco`#(IvEWFDlf?}6(%%x=VSk5Wy=cF~OF
zya~M!!R>rv-i+h5_8nn-&!o;x=|=6ToTA6^Ha+DO=DG9we=zTt(xKvfBGCSpD6+Hp
z`Vpqwke_c7!u;R=9%nt0A5{A4BYb4Ma08*|N+Qh^IxM&1I2StW)PU=+^7)9c
z-n|Ux1?!mqbRs!oP2xj@8q1;!2;Y60j7A9J-*h0nQcZUk;Ri`ue9rD;yK@T3SGQ=l
zqw|zfE8>r2MW2L?=sbKjU}KGBdWHjde(El!3n1Qe<&X1EWUuJ`zz>~&Pm0ek;d|zt
zJTizkP}R))P`Q>#vp;I5+T9<<>tVeot
z-fve>zaq@pn8%+z!P1KOBVYORaa8Vc;cpnSyL$d{Gpd*Hx=9-G7!F}+3gT6Vbk4aU
zKUVg7mLk7?sc-2+^`!rnoJD+NzK;DE7sT%)21&?|a3_}4NWWOaeFWM6G&$Ox2=P|1
zj2(cy>(1fkP{f-g)(Z_~Lli8W>{f+%=;7PlrI3d`6mA~whjPQxqgccXB&BYBLHGJ*
zc5^d5P(QWF&0-PCTjF&h`+F$BYGyeO4;U`M{RF)v*5{JKNwd8>rxqZu0~a)XN4p9r*|CeLw7da|ZI`&Q`8t`#T`NV%cl4
z8`^7>OA6;hyCs{Gw|rrLpZDGNx(f9Y`$e|d!Fnz>x7INCQ@~mFP>*_;4}iAliVmn>
z5p#HVd<5ptZ&pQtL6|RuSw~mH_Urf*q))?od4Ge)pUgns#Zoti`1xjvt&2_x{ciIX
zp-+eYePD?(4uCi$ZhEZ^`akZjZZiO7ZD+Bw@qN%A)}ZRm&|j%51Lu2TUc=V2_Dd`r
zAJYy=`-jkOcdYr1C5To>hbXMDT?Q$!7cZfIN18tU;ezeCNo&xzLsTB=f8k%Js(W{?MObyQ>o@|4IFGT#&G|?9s~~a(@inAC>=jq6
zdJOT5L0&gn4+`y@@(3u$tyOJRhq&w%;Is(+Jdi87YX$n_f8zA6wNMTc%&UF}^(Fz?
zyt1&~R1?3vHk2s_6GuJZe9-r9Q9$dkGpeKc9vpvLgtT5LtRGWT=$H@vbYo6XT00D%
zzj&oI?}4B1z*cv~qp)3$1+SCT&v4x{Pr0sv^{wZ-2F)Pyc?*K45W~L1^#W1V-1jKj
zcfO3~hh-r8sa{b;d82SnXS)@|ScRYA_aN?+p5BA@?Skda|N@lU!hF3NDxwi
zs3I1kb{?W(sqr>vhy${}6EvWojfSF6(fCG0&c^SA^?+Q-xIUEGPb5?vgZPeewCXe5
zFH(gTF`n@AGTXGllmpt$gLOR85HGgGRG_%AN9*BgQ-~dRGb52-M>(U~QJi5`JC5W1
z?9|K&G@j?$-*u3m{;l6r(0<=n$WVadP^D@Ed?f*{XT=xu$WEwZzYFs7Oq;9&ikl((
zGd3f=uea(+qz?|!w@0{;rFZ58lAr8oMf3OG?1(#xpJKo779)E#rzAs!KR9MKqCB*5
z!CDxNZ_}!>N@Iv;#}{XiKj8wO9g+S~MGL&5h4&P3`S||DXU#D7zoGIlJ_p&3xZ`t7
zT}X2j*{yi?W4rNJ&+xgI`|}~rcNhc6{AG2?Hv?pMQ8&IF;fY5awFoEwUBvmFMl3x(
zANihc^+0}pw~rb{^P%&@0o!p^T&80CV`}*Rh-;grB$^j%t^g@izEf59HnJDEw=W6R
z*M8YDiOM~`I^%Qj3Cmr4Uay(Hi_anbZ6uFaH$d_QiKuB@uHajV?d3MfV7bpw9OL=C
zeYn2CbR0fUZQ80Zs_kEk=OfMR57sXVUBKg5cb(r7_5V1LJq2OwJUM?CrnB)m%=GvI
zp1&UpcDQ}3{Dw=&j!Wtt&cC$oeZ=`+;MVQ`C@}kjDQ#Y0uR*@R4@Te@HZ2PtHe_
z3@Llio?o1zCuNuG%h?c_Yulf1k~m*l;>%*lHQr@GzPZo_JFKL47N_U-%0dw1(K^|*dx
zP(5A`V~?}2eB}N##>c10Jb?
z?%$l`y1k=BzULEhlf1{L?|g^r4J55iLhG|4HWAm~x!{XYb4v);Un%9q`*AEU9>=e~
zthHEwnS+Y;?_cR+zrDtnuwUuKPzP#ntFACE|2mzA%L~l5;`ng`{;Uwv-*Q}qM=2`1!T{Gj8u@7X!W@@?j3g
z_ffWCSukaV_mlHSKOpp-(|C-~eB+Lv2$gTK{z7puTs^l1$)AiiU>?uKeTs~~M(rc0
zy!uhqPsCSST1|~U2$vu4$*_fq6zu0Y4Y4?th+Ahy28D^1}eSPkmZ;_W&B748L>_$|Kavvo;}KYjnEYxgDPW2?MZh5T3VS$}gE8V^06#8
zg~nwQLwp=L1UBoC@?YVg#!FBvobb#kK6}Tv#y7)Sp
z3T_EF`SZcgHL7QP`*UNs&rd{6ZWM)iMzNq$bUGFI6<_n6R;7ZFS$?b3Dwr=#w!M$t
z2iI>;(z}L%5g^|gnr{sGz0InN&irs49{TqCZ7$4Pqzgw`lHfW%n#Sm81wWVQZ9BeR
zf_}ESH(V8gJi4LPX6PUj~gi`lIH?g1~%n}>Ry{P;Hl9Az#$PdAK!9#500Om3K8o~v41
zvV*+z^qQ#pk`ZwH5+%~_^$4g}d1ZCPXaxNFBWZmpU<6DXo)p&{9sv{n)p5=+|GdQe
z!A}SJJEfe+f#QR>F^|_C==a(>?x*Pa3zbf%`3LOZ{iSVU7u;{N_kUiFfjGYs{jeJ9
zB~8}eM){PA0b{W80DPY4_}l0ng#!N%?)1rM*iJ8_kPrdw4Q8Z^9zeT;mj%wfwuJGy
z^L77nB^=-12s5AW@Ok898y#1MHNDO7$x(4ZE@AERtZUxbFD60ax_b-yG#1VZeQ@DI0u*vW8p6urln;%K`T
zQj^_N>cv}lF#N?tcT4wkozX>WpS
z9DM%wBp1aILdr#J8!WdorJF7k3I=#pDRNoQ<9gzR7hj+B2&)4gigf7zmi!YZz
z^JD*wYbVMVrvcK`tV~b%8xX^8}32#x^$C$Y7FARA%P-8
zh)PDEkD~lKItQhBSYDv^hcA|SAita)4AT&5zt?U=sAcMa@Apn$^BF>RVpeYW
z{X>zT^&v=qb6;c~!k&H8l?a2G*GHlA`ZMb+<^{QI(#xp4&D`N0!mnw5I8QTK>!yT0
z&-7E~waC8B*A?FrZe=CJQ9lvy4Avw2>Fl16ClD(|9@V4yV3*ERLiT+W)eUIgYtO3S
zJcnsIKNi{dJ+{U7o5S|Xn1B5^8;bc)c+*47m*@mXao+M_xD4~1+zljebMv*pJj&%b
z?R|(h5|}?}p1fLv+7(K5!tIuJ9>qLr=I8>(n<76t(dT&gm_N>=oDMtVaTGY`;(0vc
z#fSa)d3mx9*};GQWkKy_O}q6Wbndptd~0jo>nJ399Aw9SnBOPwfkzw3Jf>QZ=
z%?YMX{fUTaMszCI<38`n?^Qca5(r#^q(-
zbTCi7y{!oQ8+&>Qw;#+vzCSp^Duw-8baBM@v7NzUcpkZqlk=CA{SMo2)|SQn>6VfF
zepLm@*E@yDdwCln(m$4iCAeQ18#2Gx{gUMOLzhT<{c~h~^23`S_v_3?mItJe<9a+t
z^6_6roLJvjrGZgOgq)wxyd?kMGe+ifou0F}zkPCNF^`UMB6+`3C7DM!h>>>mH;S;G
zyXG1^9?q(8EWdEk#qF1^tj01i_5sgR1EVFrzo)j7{w&Io`A2$SDYoMol)>|{D*MN4
z^!bWuT*2jBvaA>ja$aJ6P(3GJ2X?_zcpZnuxa0Mf%;Spt8M>f={kGuD!{vc{L-0P-
zbG5+vxjRl%iaq3d9w7Nff@2^f926M
zv_3~{1u(x4Z@GuZRWBNj$1PzS;)?8EIlX*>_UG-!R@`29OGrN|cM%xDc|?xQ@iL?@
zzk9wE#k*^S0QPg|BA!Y-aAxZB7XTKiU%GXd4?$NwZspvM(>k6aGO*|_1g}u
z;P-yHwJ(LC_<3mSSUb80vu?c5j^gE+yU)r|J3$NF@hDF36qZ^4acbsH~U_=NsqYdEocM
zUKskR)o|PCIIPd8dbSnKTjP)Y&SLO>dbGTqFRd8f!xbA&qr61U;dp}_#GPAAhs@x4
zZ*$0r{df!fzJ@G|2>lX_jHPtYk)
zL920Z%sk|0M@$`##>0MhS-+YefIPY3>%#_}{Zw%Ncis99$h$suZR+5JJcl{|vc@0C
z=Tl!^f2Ai4d81qO(+!?5|9Bcb{Cx8;`10$Ttsty_PwV~2=Xf~JRW{P|HdJ6?8E>5C
zGy+N--kqu10>`Z+-sK7VWfJGVXS@ublgMOLPh5uQsJi>nn^9xn`LK4XpU(sk+q3OM
zHoV`Knr=Rq1KZnW*1N12Kn0^W_fwVMPXPJ1i*DPK#z8E&Q?X71^8St#^CB&n54x~(
zKmIfUuGJjhS#o?5G;bdtw2A!=o-t+UHLQW_!No)5W&Rk>p}gnx2xxrB?s_d`1jJtEXWuB_6l
zfB%NTP3y?rZ|FvV&IgUqT*DFYThiHbb>}dMxCAWx
zffA2geY1N*pvjC%)aVX@xrqz9>B||T1%@3ufcrlTFu^J
zxbDP+jqLj1`Wj#U+Rq5{hTvM;K3}yF;GXNU^m1kdbSHGEu7W(-fX&VHC(P?)r~Q?F
zK;GFD$$w*c-7q-5WUxQgZx|f?c1GZ*9DKg$=o!74iwYj>_+H_%h6*@^ru-}34T8#+
z85O6AL11DhUMN0246Yi3Q{N%qJx1S?;(UY(gilR}Fft7SS?S~vHGx53cq96^JX~+p
zzY-G|;ks$Lq3qARg$ncMZQ6(7K4=TNxiJo|=VcDN0Driz4uJl3+n}r|`0sHJM05L(
zS|-pgq1j8n0gkKSnYsLJIL=F6=TcBSV{q6hy$AL$Q>h+Q$V3I7hMIW1q5UJOw~`0k
zM}vGLCuZULy(dx3J_YSB@Xc89nNYz^*=NVTtfqqfF*b+FDqy}b+rxMW?i;t!%?$To
zJUG{z@tqxxC+*uI$(Vgq5F`4CB2*9aF!hN}PRN7Lg?B&X8Hai77N_|U=x4#73-8f$
zzB^k2716$DZ+jhq;+>e%-c5lJqr?Xop2K)FFfRNZ#p}C8{<7r|b;{h+4nn+?I{6pz
zQtt3%Mk&~y!rL2_tzriQ~V8OdJa=x
zXE+buTSHY_Vf+%8T$}P3+7|{ssXaIgA_7Pb3*=(&d8&t3E=@A%MTzKrslpk1|hko|>2?l{ko
z&FrjN4duG1R0eea($&1skNSO_R$PMkxk{t_sT}C%=Hl+&Jy6z}SiYwZaa_0m>_do$
z-2CQDAO;F~;O7j=<~8`<>zht$2kIv)#2oX0s#QEsk=*0&q=5Y7m}3k;^1D|R*q@P9
z<#8nY9GAtsQ#3g#~xYB@1K$(MhN`GnqO3CyR~ujSf+
z`tR3f!Ov~ZJS4y8GwdKj@|i91m>-#&aug!F(|T!m9$r53jYIA4d3hi6rYMi=xL>ax
zhD=mmvcdQd!qR)Wn5Sv-9mD>)e^4Y)xr8lw|6UqP^6O+V^1biA!*kd#_z!~^m37Jc
z=y73^&-d|>{NCY_H|AMUFK*(ziB*mKerEfs^SGXv@qMgUJLQb!n{t0%qJB4OJ;UP{92w4$IpjO
z_{ih?WFkWr=aWv(5xBf1bPB&8y}06l>m?>Pen;(pK9**JFm2{?AZmwAP?Zz$r++&N
zDo}Zur`;$*rvQ0vga^0g>__X%@RY@KOMZ8J}@k0^QCQjss&bJSuD6S;%Rm&p17|Xc@B=1g&{P+vTi=ki3IRBK0PUT1Y
zL;l-JKH^stC5Nl0Aa>?nJX`_oxF<^^&_42beXz?5q9VKg$&(Q8eDT)_gP65{|M5TY
zJo)!;EE|pI_Hpw!w;^8LSs#JoMuKI^(s!&ic0q20K}FU1Q`F4v^IK-kwaot^^m
zXVju&1ME*o_Q}@(7&oP&+O-*Ax%p4M?YR&wF0PJ8_Y!mgDaGSZ&vfj)BQxZ20inAT
z#$mm_#XW{cVBE?$*SKH?<->Y!e?~(4w`#dYSK)o9%-Y-T60m~j4Icy?-U_56^o#u>yJkG-Ck>`*%
z6?E=&pZSFD!=#+FKf!*i
z3oIvw33yKK{k&dVkqWr~3N>&xQ-Ne-*R&tJulyXW*c}hg-@fe+RW8H+i6BdxVtBup
z`(n&fH3IT`v8Qr6@IKbNW=m^U5xn;cYkomJO$E2ZHcqPF8Uc)E2kdXCP=UfTjrhm{
z7*80kFk5e@g4d$u3*YH2i~`d>95&H(VGU(^ln8wyha7A$L)wq>#2ZRb-8iZ
z5jcO+YWMysPJ_`urr$Q(%z+D@1uy-f-zWBbW=lvO0sWn`w&gBkpx!zp+$41tSUfp@
z@1NBaSm9Ws(cC=(6lwiQ3%u9yJA^1IJeYfgc~A(=vM5mVrC?0qHEiW%Ul-OVfc
zdkXwcU1hP}VG``M<=_*1G6{^;mmaJhm;q;2IzpQ+&47C*tulT=(_mjY^QJ2ODNwpL
zd?5GKBuI!p$DLI+0|Gxc4^&;90Yw|_gmSv3LE^8d`296gfGv>sX(8-4pq2Cvz&Z%r02Hrh^rKXYMD&5;3cZ(+;B7M>w+&e40F%%>4xP-)|Er5dig--!~sYX<;F
zj`+ri#XfMdXYx0%?Etv0C3V5WX&A^x4Lo&$_t8={P3xZB>IW~Z37
z{a|(Ilh)}qL%_q2*?={47z|M)3Y!@ELDxe3g04+3*sRmJUEH=01ce?cy`VA(nq1?}
z8^(r!&va#N&>gOB%NzpOgy`7`lXbr}RkSZ^+Z5$KKtU
zhIv~@)9ZTl97wWT^wnzU|ApTM_L4BZ_+IZ`qf3l{bnc$HrA8PxPrDWya#BIqlJ~8b
zu-&!8*5WNSFkj6gx}`jaf#~98jh40%aMSn*Wgpy6xuU_R>lWejom$NjKFF8z_i-O_
zgL#WXvt)|6!Uz}&lMP&ier&nKl;jTW4NpZzp?m*V7p{)|n1b&;c{H8yg7GIp%7iNt
z#+U73@>|$ooKs-seVhw1nJvh`62=v(zW4z>h%!;%zX-v&H1hnWW)VcKt;cU;eqY<#4;kI?Ery3w%JenOE0^5DApL$sl+RGezIQJRq
zqY``=(7itMkIlyNP+v77o{jkWon2>MJ%D)Qu0cLJ-+rdu6`z529{XB25};f-d19X`
z#II{SwGscl7r&kd@#PEiDz%o-PlJEA2K%5)Pd~Xi2x7M2#h?=qHICi?i^gZZDs;Um
zjGwYHqBHi;UfFOj2Lo)sQLWbPD6;oqos>j&%Ma0Xa
zdul7Tig;d#!4z9Qlus}+r(Z>R$M3Ox{Qi*2iat9!hZ&8F4kPR;=f}J^PPa1?owq%8
zC%&R`B36GDI=@#chBT2JEBY+%K17pyzs%9OYgwrsi}?N7Yd;2%zQ@9@8=d236TE^l
z5R=VUY|%XY&AKIs_yypf--P_AcPI`*viTwYP%elOCtpsYd8%Cg`T8crkEiMHqWdiy
ziP2Z6z2|LbFdvH(lc1u0*1y!o{Po|4Js6|c^x}I}C!-Kkq?bPsgLzR;IKcPZlKSnK
z#|f+{!}bp9_3IE{f4N)Y0kRKxxhW9gqsuY)UT%%+sD*M&Wrnm#>kEcvh;NQ!28ndvtLlveSC#IfKxA-6#CLd%5o0*pJj1cFg+@
zhw)+_D7*P?I>76|s3JSI?x!qBPBTccN9|?Qa4H~vM&wJKLFH@isZ61I23ulAklZ(X7e5c;
zkl$pa3uW#%E)}Sqs%_gpAU=7~;4OU?RE_EV<6h&5EDdYi+fHLaNqAgd~E6rlzlA%ngbCkyyQ9tF*b3m
z9mN~){ch1Rigvv&A_{3$6c5oU#UyZra05b*o}@AseD3T2~p
zf4^Rab}yHIF`@TC`>_Gsu+Sqsa(6YM{^z+}f7{@1fcQ
z5B;N8q5`g{;+$*H-;}2fQyP$WGTmjl>D);?m+|YT>yy+-O}XWibpM7?}0mg!0``BhO9g@cnbf
z@uf7_AKx*B^aS`Ex_!DU=uPklxFEbO*t#6vf4kfYpoe}nEGC#)?S|iye!1CnPu>Vn
zur;4%79R(%=UA*p;Jw<(YCWO>KHuG!*%+=_HVNzwYwqZIGz)stWvn<3PXhQOm)Yn1
zM}Ty5^4wkcT-zq4@PKmpG)Rpvdb0E545)RsInyUT1(f6^#RDTofNc+@`Y+da5GlMN
zi%(+~=-Y{EH!;ls=Qmx>2`6TNw0hIYj7}=3KmT|`#L^U~cYUh*v3&-_%EYd`Jv;*r
z9_EuTcbf&gCsU#Y3&z1p?CsIyu^Av4{e-@1_YAmcE*bZXZ3Zy3MIX`5oB{q5hsW9&
z`$3$+y~8IwuwKj%%kZreIz-{6t~w2P{twHEa-qyv#47(Ju$eT?=dvc*UvkcdM*2Pkw~)
zg6k;J0rRwL+(#FVz)6^yrKTp5G=Gqz6D?lsIq(>IM}?SXbD7+T(?K%NsJ
zPyNXX<5bV9RNWOg|4vQ*5!w)4Z=F=#2~k>C>sT$clZ+gDr3C%$kV?6Y;#>Ecknsnr2f@m(|8;E#L)(bLTF@@}~2Gzz|`dogn91>W|HB!wx{YWATzvOt-^dh_D|b|M=fP`DrUPG9rc1o`91kV5($(Q)$MJCxfLT5$;Lb(|AS`HR}eyOifxzcHif;^N4B
zEU&)1O&#m2+VPF%AK`RJ$9%5!wlrR6%_&29mZy*>jc;{C2~qxY^jHz)2^GoVl&39_
zNTKo6%BsefSl|AIS16w{eaX{+^<7)(dk#_c*=v0)U+cS?e*d&?8878cnmhJTxvhIN
zXn*{-X`~h9Os#W9uwTWw_^F=d5eD=*e96o$#PWl~y_3kd*=6$*u;02e&Md_I(vxlm
zc)!E0&y+BoqP!>#`AVGnZMCC-YB`$pJckOY@L-f{G)`QI*rye>4WFl-goQHR*Wy*w
z9_;@YnLd*^519YWsuG{yo13#wV?TH4Z=RY5IM1Jl?yqsYuy-*&kMV5_?>Yl+p0j?|
zK|m`DAwgXCklT3l8qS-nqraTI26$V2go^;qR8`Q)0jyB1xS0j^MJ6=rCyoQdmY|my
z-_
z;4j`Ty%e#ZNV3VY0S`&IVRO)bl-QQ672i%pEE-hu;1_b
z?(85I@Fjtn>rAJC|BEZA?keSmc)Z~E9l2`VM7O?`D_e5_^OWNE)BKlV=-#!lg
z#@1MR!te)af1A1g2l+|N*jFos{0H`)J12$%>R0;9I++;?{rSO2al)pb9OyZ?IRf~{
zB{#GAFW|m4?iW8lWI>!^hj$2nc_&dhEW2z#Wq?#Ey?Pg^*h>mde(z2sz}FmYXHJ_9
z`>@;7*n7`)lffwAj9okHEocPX+tQ%ap>faru6d_0_s#
z(x;zTSaPnee%Vj7svh+Wwl|XGrKKJvtD49*wa87SeT{_cX1Hn=@R;kF8POxaUyR+m
zPG0M8B-tSq52A&d$TK$M7h)@$NZNg!Q0>o+g>WY)YBBGn;HMDvxSz0(!oZ(C2S41xHW{o7aaqC(9?-oQh}b9*z)d!oOu
zb!aB}>XK78&o>j@qo;2Ut#2l~90T%&w3>-e*Rriq-pz#Vp{>Z9m(7GbCg}Z%>}F#7
zx~aw#_={!D(HXXr%|uM*a+p^pi=0@Q%#*eU{P_dfYB%5why6l6%-q*R?q6>oJfPN0
z#ty`XaFn(Z&dpiwfe@cvnvbpu-ecpY+jUt@WZvd$mJ9Wo2*+NtR)a$=ka83EQVqtJ_Gd(6Gf$h<{Hi_NW%gbdXY2
zf&0jZ{ySdZEdy~lLbPw$!Z=I4usOs0HN<%e?;PQf?jS1^TrM`YgZ*X6ST{l(s(I44
zUmon0=qAgk7}!rR>~_W_`Z?ICIC)d``4C@^lh#rB1olJu#L*fU&oi~3bQ@i3BXu4c
zZC4-;Jx^fPv+b}SdVHZtj=%tXHznD3Zzbr#y(Ic5w^$oVZQXGp#lD@$vj4YN1MJin
z&g_u+6W__x>D<>Zf&G#YbNGG-cuvQ$d$YxXzudB1H>(%y&0dv0oj)Ki-1A7}9@s7J
z=7)_5P_CyHETs$kGYq9C7o3CZ7k|!X!+EObpWSx?KyI7u&&}R2eiIIzu}*>OX9le|
zwFSGh{J{Kv2e7B}roY#=0r`Y|<@J?$upTU^wQ@P=yEotLwJXS}ywXj?`{nLU-&X^A
zd>Q%a*#z^aNMHH*RN#lXCC67LLwSpM`MvXCKaWMmpNfWhS?ZURuR^}dqYOoyXRC;{
z2_g@8?8Lqv_X)YS-*QF2k@LCmQVqzB@>~>+h4jj=*NuXU_%r
zoMJb8dy4CGRo`aQ^?Bhs+fwWY*Ao%J$n#HYZy3}9+;nKpq9%}2Hh7?d{l7@V(i7z@
zELTlo{6wkk-a=)-5P`ZU$o~!Pj=+)9(EsP%g%Rtg$^=~j-8sY$D*|59(tC{a8P-+H
z|F9n8N)7tHT&pn6!umW$?w76wypkBpHW%=_SyBiOU|IQ_H$MS+;yROR;QHemmqiu=
zI#zdmp9k}}FMCbfg`xa%`RwWyFiti_`Y$*F2(CHDA_FfUqOn)duE$gQ(UO4|o=Ia=nydwm)ivEjiVnsV^2f9bEglK
ziY$fiDV8~T5-%7Rd1FOeV4nHb{HND6i1Vk;;ZN;>_GKGp318*tAf56a*GN}8d2nd@
zmE~a1I}-g0E5Qy&pL<;T65?N99K}1rU?0@|ksUA7W7^54mZ(>bpx@Na=&rqB-xqg{
z34a&{o){PX+63}F2Yb)o_JZ#b#?P*qxgX*xkz9%3w;U%xcDAO1@iv*pg=9-`#Qg?#z2vT6|1K-eKH{cT
z+Hd391N-&Q4;*GS6Z?cRy*07TM9@CIyYF5z+54+B=YCi#NuFnSPGor-*>E_X{S8|S
zseF7=x%O8x`StkLNKQdB36w}+>p9d)?yYvMn=D`vWsdSw3$$9u`K{5By=E=&f4xqd
z-K$&3lLYyUQxdI2a#-cJ-8L4H=O3^Mc+x^%Zb|G`%5EX@W+%c;Z@_w8x{q+=;#Tss
zu;9M+NGnl%eBRcys)a=I$!RphzPD+ks-H*Rwh+^LPq#U+Za8B0DtjFZ)&u2!6|scc
z$h$quoY%vCy-t_!*B*dhP7vUVG@XR`LD`}conXKJlt=_Ue$662XLB?CIopZing$_T
z@b8Td{m%k9J4n%V{&E4}%b)s=R7`AO5top3RZn|Z-wU64rqUATB^R4qvgUM<7Rgxa
zMtCnovn&s1{c0t?_lyTs=D@zctX)3#lJNcsEeW55IL`hz0^YpS;e9o}Id!bLmE1B4
z6X@$@k-&L4ZP33Pmen+g?4y~
zq=wnTxICeAS`XtGPuK6O+yvt)aOjojRgl~5@OA7FjB}AWR{KjJ|IE7Tw3U!v)91YI
z;3$-*@O(%u1;6O)J-(m{;vKi{<
zYR#{(Dv(?2#P#ej=wX*2oHY#X?|jhHs0-hp%}^<9I1b}eXWk8;Zjjr=_RX^k4N&`AK7tP5Axrm<{J{i!i#f?8#cR+q`
ztm#9X#~I4(9~A^tt@C(-`;1@PO7QLlIi(Xuq0b>dC9&a9BA{ZWR0tL_?Y5f3@G$xtXZe6IHoyzsw};UWU7X9U
zi0cb$MtJDCfcFCh==s?~6G96cp&b|dc5TM>y1~dzRmej!R^%VYb%fS04>#g`UadE|
z?JLN!{oM0lKcrcL9v_jv?C*J-igH!!8&;xU|EZv~4&|mN*)$`sVQqSM9Ou>T;pX$K
z0NHl;E2014-lj(VWkbO?>MwP51*pHA{E$S~XFQ9(((@ND30@XKe>v-eCykq&uIqKf
z_T(jRd4uh|bc**9>SxMVErD{ElgDGQJmxdg(+a{qLt0H&|^}Fz}jh-`;Q^lN<
zIeV?=MU;Ee{oEPl)MP!*qMyzjb!@@>w9lu@5WPmFPa>XmUA7(NJdH%oBj$X6aUT0U
zb5&<{iqAjf*ucuZ*
z<4*LKem5PZFiyKizKajX?<|>MrCZ=HwWq{lox9S%=L2lLl*<#N?&P+~kND|iJ
zQw4Kxe(i<*h9)i_K7m|!eZ(R)kgLAwav&1q_$^xxJ_EUlCl6kw!aO0oTjSVeklU4R
z|85t^MVcLq`v`JU-f!8@!F?T~uU5D~eL|uV8QVb4%p>5{c94s$b$L4n^xiMhkkSC-
z=0W{KvF)JWwmRSQPT&{&gMzm!g1@|(l)h>v`0qf~;(2ypNAnjS%l7kxeaAbwm%W8F
zeE$)-FAvuD?!CF|HU|Bu{kLg3)E8g#q(;ga;`Q*a?t3sl;rsIC(gK)=L>6v8*7z5$
zpJZ2^CNZ&8I@4a_qI
zL%+1vg5Uq%Q^#rrJIT2+xUT^CLE5E7-%kQRa2cQ5pVrz=xJIt|41Z`R$->1!F)AJ8
zXyOycLwvm?(P`HE>bZTSvT*MQ#lRkNWXYqLf|7P}tR(p6D~k@Y+B2@*X-zNTe&}>!
zD6^M%Y}#;Q*^?f!O@PBD5cZp-loi8S7#$=nB=UQMU@y5Dc&9n-axanUems=o)Wpp@nC^7Jx}%*~f4}ZwlL7HE
z=fw}4PWF)Z6}!EkNc9qf>oN1Zo_CW{F{iT+VE;@<24CKj!4~q>FLjH=yB2b=*Z;#y
z=~nV>(dGk!;4jOSj%e0Sw2-Ivd-tAM(@Gw0spS-FYawxBtDVk+U!8gNyqZ#dE8%Jz
zy~(E9N&*57d_8p+)`5Pqb!`LxJaj(Qbo+`{64hbwUv66~nJd-LZx8!{IA*$UVS(S?
zaw;`$yL>CzW1+kB$@*4uR${k)&jJ?tR%_~}`KFbqj+d+|5o3{&(3^V-#ahWTX|-D!
z8m*-DwzBmi*zbN-MLl-~_{&)lN#?QOFS}gW4aC7;ddA&f^B?%jQ;Ex#`GUW6HD`(X
z*S3;*Dhguxus-#Jk88gDEU*JwaXQ@4pK(9bV`Cs5ajJ3p@4UuV^3jCvrN#t{u!jd7
z;unYaiR1l@9`MtgF(;2}Lw!dyvVLfCvq-}q$w;F;ZRFEM-TSGBfw#S##ji21gX~gi
zUb^=WjLSNYo=xg3qI8gbY(NA2e}UGP^6#*}eSGCuWFypfI;%EQ2I>onU-R)T$SFHnp4|d+{6@c4
zErW5h_VFf}Bd{JLv#C)>2IL>E3-j^@x$=<4XC5%F^~O?X=fXHnyLQn-3gQ%rx)(d2
z!+f}U((5A&?8ZBR#Lc_F4jkTZ(}3&72^rJbw7_00u9S0jraW8^a!tiig4j>9^x8a7PDJ!!
z7s}ll_mM__l_XHC+5vK&Y42?gLYgJJL<{|Xr(8%d$|asTcnbYb=(%x~-ykPC?C}o$
ze4nqRCHjxbS0C@ATo7-3Hu}rZ%wD>GmZ2%jhCItm<00K2!gj)g?(cOf>QBUaQ@^jjk9ezO
zA>CKet3Qpdql^prQ+eOlac{BQb8i&gAKjnSz>VpqIiYV5JG@)zKA0qDMY@k=Nk2E$
zSLL>m4wjF+vo%A1*}l$@9s6DGa-9vfw_|?f(yf4hR3qNwcMd-c6J@boOS;B2us;ni
z{w+KYsO9(Ikv`zB#QK^XkmK%ejKg+!E&k$<^&8ynufqP~-K_Eo<$u|IHZ%j2<4P38
za{E9wN0gUHaf(8@4I#JhV0uLM$Xb6uDJ5%Vl=D4xfUf&o|5Usj%dZXU|3-aJ|2Q0q
z^**?JsTFyGO2sNUEf`l1Jaubvzm`nNDkozYk5ZYO_MgG8DthI~;r@*qJKDcXf!t`o
z?KBrik94IS%7^O*<`^H72R|J)Kdl&f^3yrm1iW1*Jktg0UAXE+nIrJ#
zz2cTPu0wu}=;Q;GcRuvc+86NhnH9B<0mJ0gVl_aHXLFV~zHepDgNDdsw;Nvbzzegu9}-uj6J_94M9IbQY;u6N1Q9)oiO#*O9`
zuK>BF#kH+tV1H&QnLgbM*Q@Vfm;53D>*q59Vk~Wd-w7T#jnB==SV>|E{JO<<73Z-+
z;O*B{UHZU|yZk59*Kn5&)=NS)qUwJFUub8~Q>=E$j3tfRQ@1Pr$u?X;4oK
z@k;EmHu(?sJMZVQ`jx~YVRqRocws-acti8T)z+=#G|PKZA9z^%+aI-0bXa81naJ4Q
z$8BVZh6;aqR4WlKO~0Qs7veGn^J4G7dWLD(%QxHN+Q_5n8=E|Vm+lvdn)-IRmB`OB
zIoh$MmHfP1w9?9$MO1`>-!D4ZPA;dK^N$|_d&3brsj-7aETy_$-U@3Yxj)kSbA7=c
zrD`X00be=Lee|U~@Rz_JEFv0I|82)u8`=0Kt*#^u#@Uj;BjLcC{tm_PpM!CI
zs_aM6>E7lVgPnNC-Rvd%qLEt_M0aJ<0WeUh*J#M
zrF(PKe!l=dv+Ae1-fXZ>Jt@tL`F3fKkf32T8zz+Yr`RmY=$v+(<>918Yi
zs!GnU9rSp2ZSIvWus4sidtDyDyyxh^_J`wOSGxr_o>-O*@fT?|^Leli6p|`8Ul8iC
zzH-f72kOz8`Mzcd>}lGj?Ozn29*h2?QvFb`SNU2=JWuDBD1Wvm)RU)Cb$J8Gcl4D_
z>jZt+SpU^u0p2z5{h|4wmOV68V?JqKtd!UqqS{PbvX&7wuey{9?J-^`kEVU)_4xHXZbMRcqjV2-31m?guRZ
zrR5{zt^j^~vwSEK%K12M%9KM|G-Y$-4nWVxm$T4cYsQPL#C3;hGN&ccztu%gZo=y~
zdnl`6xt(m@81k3p8yf@A&vGZrM{Ux({Wtwmk{+ZQ@rIQM6Z~`!p0)&rtqT_SK8xoO0$|
zGLAhHRGx&E(724k1at10?xXuuE-9Wlmrd)`C{436uhIR^7w5mEX-=PQbU(1pZvmRt
zJY-GzneuKus`t#liN{cn!`cya-%M+4SSGeJZn)kSc@*KDT7vx-*QLjf`I77QiJ<(c
zM?&W$Yn%z504LWqdA%!Io6-9Z~U=KYkc{{T0ho-?*;0`HYNkbn9dR5*?0MXVxm>`%R?8TmLKt*+@e;<$KoW7%n(hs1l^
z6;^_tbNIy{y@Gk(X$0I9v9z1!o7Rvk}DYrf3({K%X`f)EwajfVJ
z0RH4~Yq-AFshiwCS3kWS)@veIGv_)%yF|{+|E44PgK*d$vE42P=bi~~<6AnbhrD|B
zfkT|Ho2-gVoMrkK;ser!MV=eL9;Uk?j-%sob|$#JBg@ZqnF{fZj$S_
zm#S*_4XJ?p4&_Si;O-=?{JTV!j=*P+|i*+$OxKRdKG5%!;?
zt-mFr3+H5rtQQZP-vvB-q=RDubT8M7rBL
zPqRp7;_Tns4OwJeuigB~vA=4}627K!0qnPP0p+6#`wz^-2I9$~jx5;tXHy>L&>vJW^?gy
zicc^0=yi1VgYHJGdfJD|UB*QzxQ?ZiWQ^ngPk#A{xK
zKNN;}+O-79ywza0@|LP@!np7D6K=uNc7h!oSLI$g2L6oKvB(DMsdx6V-3#$!wWco-
zmcS?V7I+NcxqZ6h`}lW5JkHxu=}!posMzfFu??U<*Z5TLPOwwE=NNgdgzpXe_0trm
zfj#Na|MDRb;&uCUcXM`v-R*euD9!-T>WSC9xnOs+!fn|!;rawF&w>+R$KTFiFE4@i
zn2`r6R$!;%q-9Jv3cx>1@3uFGc72reeJ>5y8#G*XkO8~9maWDo9^~#^HU0Gw%Cm*T
zjeS6mz^?Zk0npBc-PLJ)AQ!3~x)9gN&qg#rS@LH2RUEZk$2X~D@LJz|E?~5AN_GMf0`BgYmHwT
z1}=b?(PtEY}MvrvGJ_s=q=1e>_k8Dc!&P
zFyov4J>ZbMCEvGzCN51qb-pAs)3H_ce>qql5ET7_Mt3sTi(t<6x@YD_+qUa8D
zE?U+?=6tfqb^27!WvYv!eBF0?j)Lij;7V-Yqu-{6h!0oK9meOZeD5lin;*#hU&yHK
z8s+f?J73WFq{O(rKFZrG9i@5(8X8eP&0C*M<=lCX({|q2B0|5rTTy9B>uZW(qk5~Y
zSE2Gdg{J7d!Pvj}0oJ!lis=Vk?;MmbTwsTY~mS&W;~+zt-;u8@AweX9Ns>qx_Xy+dQuW+IW5Vv=Y#LhDyi>!2auI
z>(FilH~yU$^99yVm)X9@@w!rInjiA&do2qW?gcbEZ8eJ*#{VtV0OLYP>zgl`LOnM$
zmoCP6qfos11{_}lF*ly#xHivTw|WBbROPcseD7zZRXX8#DyoiE#yG&j`Tb7#{7b9+
zD%k-;tVZbfoznZP=zJvW>5)k6xA{6Dt8o4>gU5aF7+}(NS08n-f5yMh-uDCl;F7iG
zBz|8UX>)Vq9`G+8)84-;03KX5A;E10`f_J~TImh)hMlvueIS0&QX6UcwiM#Ei)F8f
zWx~42$Nn>@=fu27f7Ji<#{naZ`-fgNKHmoQs~NbG*N`8w)A1wz=k1E;vY8)Ik8FdK
zMS#gB#=QZMf4Ma3#a^)A1?DV2f5@-Bt}=yjf{^$X3&lW>C##e11c9DvZz{iegWqTp
z?cs9(y;8QWdg}pk%3Wd8!-vITKDR2Js0w$J>)mEszAxz|lIGi&+JRqGv5{T+9Q@n?
z(F;pVykLG&5`F38LHO?M%ClOXpPg_#=sfdG;NfYbC1-^pZc}3S^wOTcz^^%_giAu;
zT()YtF9t_ozSehj^DfY1^GeC_`=BqM{gp}wh#Mc&%qua4IB-Qe+hs2aIFy5f7$qtgq}9&BvR#7)$F`o&^Ke(<>SjWJ8c6vRxaU4AVG5^Up&L3pydK2ANSvZe)G~6{ptb=^DTdw6{
z0sAm3Cx1^}0Y7`k*-&&0=5Hql|KkzuBv0Q6i=AEANpehHG+lvnh;swyEvtS6>wy+W
zFRa@O-?-jt;I}oegoh3*Xt@q2Euw+cUod+S4TH8{=jJ=g8R{a)1regw548
z)97#q;dba2#B)AoB#chif^`n<0%4E&n}M%P?+K2J27Y#Z`I10=_-@X;y*~)_bpP78
zj}7K)PQsJ*+Q)GpRgie}k9NZCH#MM<0O#Zi_f1o(>mYXV4Ic$AcL1La>CXVZlQ)gE
zXqpC$!v%#gOGn#DQ%&OPC+|AQvG*e`s-7LBZePf&<7{(
z+DPfuX}a5Xwh^h{!(;lt+ll$P2Mr5fwUc|*?T?3cx0AMcqee&e!oK#NrOoxe?c|Eb
z&zhiP5MP*fHSg$dh;yB3)YAYw{l`b`TM+QikctF>Tp0Jk=T>$+g}A1{!k*U;!5?h;
zyU!EsiSOa*o43Pzxnj>w{%m-!k|*3dIQN4-*+Cmpr?rt&GNn7N^R^Mk*jMtA!Z6SI
zdweLW9`uN~k(>bYn@dk;EPScnMmQti#27%F*Ldac(Q6VA50O&Z6!Qq|ReMM&@AY<4
zecAPm&R6L7tKmJFH^7f@MT(fixLtL_O|%^DD^V1(am_oh$02?mo8dk1KjX4ja|-;$
zd)U_xcF=FYE?61H{rJY(gB2qX=j~VZFMv4lxbtzR{q695Onlx@In3v6brVNyz+Sq1
zHGDavsGV%7RjT^{&r`u7YIZ%mH%+B)(zAiD&q+9D`X9s#6iS;*4#T`Sevy3rGKeoU
zG<^GQ^&R3Mmkh6&0xy@T3EBAq?21*g+L@{ec;1Rr3l^}zp6>2H;0yJPX*&wyITRg7
z?REEq{oNYe^5hxt_25zAuwl@bGyHhZODKOKX8#)Pdt}$_$Esj|KbPo-w1U28{~Ofu
zfi%DI{a{_V-fm6wAw2h8d2h2k#!Y-JmgVk*`MbztUhYw_gQ6_qpAujPuarCx@P%~0
zZo2+jkUPH7j{6Cu*F4-dI0MS1BW!=41ADHsc8|z2D4*`K+3P2y?E}9ad=B@C?(p#5
z2kE#4ZyLs-eAwSlDg)9ZLqR1NC%a;Jw5%HH^VHTA!*k=+q6RXPA>Z?}*QO(Y4j<3=
zX#tvVIWH*%$c)_0V;4anX5Ee6*$RcCLiz&NbO=W%^AkkfHmF^T&IU3TxR
zLw~)dZ{aue=j;N9ccCBJZR!@c0m@%#Ts6b{RYvv{pq@60b0)Ao2Rt@PBL8;p+}3g$
zFll$b5c+S)x5hG+YkLb)h*^9h8lSkDK^^D)C;(~gZ)QSaTIVVdh8U1^Khh+uw
z4KDA?MHp}Ew0nxteB@P6NTPSSG+CRW&gMP3qMILPoH
zpS$$G2_-B0j-vdhq8$t`+|gBq^@^@7^u+7Eb}n$h@(!+Wx-b6b=cF%~zHzML4eD_s
zeX~2Jx3DksKopwSM){_7!v@OJa;h(F!g^-*N^~KQ=rstZapw!Z%zvXjc{c>%b!*Kw
z(*5ma;o`oSUMyNuj_B@_NaL=X8_emvB};i9jlag1no}NJd$E+Rdmo%uLiw$y=@-iT
zx>K3+uP4HX60x4ATi?+3vDP!^DG#1!{xif8arZiw`_$zpAin#qMCUVI?^u*KN4_1RC;cB*Oz()50t#}v6{
z=To^As()wV_>kteP{a9x`)h4|1%+eD|D$|&V1GMyA!SsRgO(5?RV~BE^24u^_u6R9k_P!L}NB!Nw-8S
z?n`
zrO_TXnb(CWU~o6gqv8&p;C6%kfG)cqRX>MwnGHvF#+A8%ejfGj3L(yTb?`*oFz~fJ
ztE_#?fJcngdMq&b330V7c>j5j2juJfr#g;;yzt{1lk@PtBsocc(0>N=?iKe-4gyb;
zId)$&!@PsoEB7fbIom-3tl1i?j=}l35Bf*gqjBAA=7B=so#|7>7IOLE-*@dT{|WWx
z8SUTF3w-NOaP`XHz~e6O5wgVj`kmIc(sl5>o=$V$5&-kVgqIsvvcr5%$za!JJ$RlC
zs;Y~hgFKJnLp%OX82`Vg`Ab1Pj_(Z2#M%I5Gs9+pz5+8oWOGBD;LO>eyAjYX-g3hO
z_#evL&!-aE;JI~g-}}ehs)P84vA=QK-a&rv42kuFesY8ieQ4e0OG*F{EeOr0GV$`|`oDbQ$*lqRUbD0-@dv
zdH<8ra37vwUzI4(`^&1e7Yd+%G`@$q;=by`M+!~wyQ(jC1_#vPy@-6DwQCOa^M|x3
z1@{kNhZfukRQUtrW7Y0eK@cDJzv{bA2(Hs>&(Of{yiAOAT6}sykFP~_YvFw{TvKzg
z6XpYV#dq8Lf?j3a%LT?^yj5NbJ-r9?$m}uus0ZWFn*VwFR_ND0Pn`!Ipr_)BFIvHn
zKUwtci6)@di)-7J;6D08<8$!4OC>GUi5}3i(_wl}I$(G1UqJ!L*Wa();0pCBxJjR|
z2HbD;@Wm~dhy7|2(20X{c%1Kx{)mP4cK4Mz+$?~8ckMd*4fdsOdc%=_xC`R6YHDlR
zppNIS2EN+*_^M)p?
zYuRlf-Eo(-`x0Q{Cc?H7o>SZ&p^RG8&p_!0{;$%~LYJijXdqPGeiyE@cX~e~56X?@
zo}}Zx`a6c9gBPK^-@$tNBhV*w)fe;2@cy3JW}j0A&rPi4wOl8loSkg#
zkmrziZ-sn(XPKhQS-@-E0qYmQ^)V;Dso*-Y&E4h=^?)yC9Fup1`~7+Ef2ahs2{=3J{BiU*UJ4V6IB#g3s-9s0>7=n0CHDY7wiP-Z1pKwlaJ?zuDjmht
zH-MVbr?>nD>|9u3K;vG`1$Kuay@0I$;0hR%2Sgw0yUyz#hUfF~Jic`a<=&^($RV#D
zef8JM2QbI~w&x5ef4onp;|`?XIwzjR`wRSdT90wWn3J1Vp#AUQ>Z>#dbPOJ>GXgyM
zQi31-%<_Un`aj?s1((av{)=6|8D$2jnULItyx)r7I}!Z@%Pz;o1~694q!-i1A~$Kg
zEg>Vy9q-c=Ww;7?`zY6C`k!CbP>jwA&*?^toS$pU%%rKM}{V|fl|
zaUr5rQ^Gea&lRX$i1*pwePsaq)!@^$47}fS!3bIObA4CKKjZZutSvWVddhPLJqLI5
zX%6b&L%TRBkH}?P5RdgrlyX_&^{V9ol(*kp^sX4MdwezaD`H&tCRdazzK~{uX`v10
z?;r+c*Eb;A6!k|T4#zdo_AV#qw$sUw3t;QRPO;%Ys%{!=f9x5%{65&YL%%7HT
ztc#|2ou8lbx+nHoG%Y%+PU|mzA5PPbp**zwVSENn*LE_ze7U$jO~+n5M$5NeD5Ghe
zbIkeEDPhciPgZy{^1-?c>Re>7MV6Uw6rE>yGHe#6*K40=aGfE8694~S`-}!!uK(}9
z_1ypdOAjAQ)BL7?^-r8+(mVgv2SIjPF7yv~{i~n*-+${3|K*$6Gji?!u4nr3baA=BQQ+nMrXRt);zW^n#EGmk3EX5_QpGWouo41VTf^jOQq)Z6dF
zyf1Aln02QAt}_4oaed0X$I(UBZ}I(IZx=?tJ4=*xr+SWXGXL4hnZ{1jPgC|$Ufs3W
zjE;+Ow?3LL%$7rulyK61s2r1{=_Tnh6r&QDdEC@Eb58jk3+8?CRAT07BQ4LVoTXhk
z#n9`_dQNeBE=}jYWZuJ(*GxaF>jcpEzTh4^oew!>d-S3HC0^y1=(w+4*MRmU!m*Ez
zr^}C`=sd@dZzi3`Z{Tvfg7S|uBIx{L?XKVSK6kZ~Y5TYb#^^lCx$r;<%H?NTQ9I+S
zE=b$g`7V^!%kOc8%HPy;ru)Yt$M4hh%U{g+lDo5()?2_QO6|Y~2d2L>hDzyr(NZB_
zx-VXB>Yy{)`P{0JXq@L5$7x6g01od9SAPrm;K-h>*sf-&_4Ge5e8rgoXnzmx`e}*!
z^_B+FbNbzdYT7X^{BwgkVu;HsdY(b<$+v2F{mKO<^gQi5A&gxxFJk6V6~|K!p!}#x
zQa|=vS#R4CT-TA{o*j(&8^#}3B7STSr~U0{%EUz;+pG^n`Ne;a_SXUi@$gVbuv(@uapHLpmczsozWRD*u3=MtAi8mmPi1mIK~2mgB|!
zp+5p5>r;U5Ree0yX$fd`NV=pJ^#8lXtqGr_#nRB#crMAan~5(@LA}?{{FU?pUMRyG
zn(PJbuj{k>bpX;X1KpAxfNJ|Sm*RI#&xKVVrUSlLxnZ{k@KMuQKI9FRH75P|9F&vh
z2$ey7u~#-9!1c8s?rQcJH$3*@wKwXQ)*yf8GobO=dN1sMiPBw_qJWbVy6O!eZzG>+
zFaW3{uU9=2@*U(`rZ9dSQf<_L`&9azR`mKqzTS9#;t=%jDIulr!XPhnb*N$h^pw6b
zjrAPz)%Dl*Ed#j+ZB4dPaGlfIN$FdFjz2jlk66)qqaXXJQj9|s-(RO1j0
z-G6UnpC6!03|C($;EE#m^!=cx%UnOxZGcxDvdeG6dvd)0;@?BSr&Hpyp5yl{;T~UM
z{Q!8^rQpJ9K>4?V8FC=EVkGPU52TBAJie}j`uI(Ex?;bkk3BW^0%X|^ieo>D^hG)1
z`4`WZ`qfB6`mNWy-KzkX&k2iO0_fI%eG$el+7xE&jsUb1*RPBL6tQxA6%IHSWfU9@
z`0mu~<<9|y%>QUz1$-V)%y7K&svjJA2YL?H@_og2a@th=#P{QS-XDqCkPc}Q>BjG5
zR|E=O$GCy*wI2V&ppV5tCFzBbZjfH}vJOx>Ds@X9U`y%+E#&h{L|)#<`zZWf_4g^D
zq3^qRjEhVZ+HmXx%-!=a>Iq=n4dMT=-4W+16|?{k)oSzLbG(v~x1W
zqlP;H^FwrxqFvxyUbN*h;Jo~mt6c!C|9ow>0X;u>OG|J9cCwjB=|g_Q`Rr-PueCoH
zW}{tjQJZi436PITMD9ozB!&s5c4gK3!#Q+>P-ORtQ6U!E|X^%wJ(9`^VT@nz3U
z$|DzjkyFL`>TWEo!~C-SN|dioeB`Bm!u9?!8n0CTw3hOT8(}}TqFlITAB{`${?elH
zINch(2+ZGF)k^>05}Wn62ix2GnU#b2Cjw8>czS%yaq2I3Ro)xJ@?kc{f5!Z=p#J&2
zeiilaYN-j-Z`Y?r(D{SBw|^nlvw-s+<orCg#+eesv<(%7^JW;OPU=G7i{H1As
zE#D?YudCe2OY5J0w~>}#P~SrJ@ITI6zlQ4?|5)(9}X?a{ygr*{l7
z|BV@YAVb@~GBtq8`~G}N>)FZ0@G_r^%zwq*oA1-|kWCEl)W6B_y4`=6czJ;i!-L)A
z9@Fb{PcS?*W{lyjmi!EljmTtp>g`cxpZ;1N7L`xdW6HIcF?{x~7K5z$%>3eRJX61?
z4|DyoOHBFi3?>boLhFIe<`g&mn?Jzdq3JCZ48rChns)q$$NtIZonZ1;{X=LhQ{O*~
z{D+tS$-~DHwEX{a|G$6sKe_+cFL#=egO`rV!^=Ps1_ed4e+Y+m)AX)?^N?Z*CcWqr
zgSWLA4BcHyqW8Saj|K1O8ai&}%
zjKQLR?`z*lMn2qusb7-ClqU```6o9s{q$h8ep=q+286V@TnSFvy;)=9<+tFfbkH-6%_HRGHjQb{WMvuvKW_`!Y{Rx#%
zJ|s@{EciA>?bJKXsv6Y0_C%yP+F3u`T
z&mHAg?b%14-;%Uf6y4O*={n~=w^P&}D`&Tip}b#^Ce^d1{26VJL_!3W&l(?kg5?JH
zx5Ll^zi{yr)zkWXAeA@2-a+LmK#0zgnqAVvWWjFuG?IZET+6LPt0x)>Y+Bc
zX&1_$me{V0@^5Y}P{;fEN38I|a{j!!--sF|E-<7ZzWCs6HOg=M>NbUbQYp!e`nw(d
z{dE3xIlhbTqvUtIz60gs*cZE_d_t`F_;tV(?E&9SfO*c&395h(?dlS{0bh+Q^G1GV
z8FTm`@`v2bO6O2t`Il!ykiQHWzg>p>;>6l7{b7JIUGbU7vp-l$(Q}0t=SY0V{#m%Z
zBFhT!XGGO3M039z^!s+hgEr~t|BwIm*pKL#oa>77K%;ke#jzcWXR0)!zAl-X8K(hn
z&b{7+a>kGAFJix(+Gp#9yodA4p+`>v_icLnNfXfgdhZm*b9cV}^9$!yk98Jo{{px|
z#_Q`%D6bgIEeHnW;+LoYt7tbePd9|L@vpx61Awp99B1J3)cZ5O5bs}0#Q2dP6|Za)
zaR>BxIY*z)D|fU=qn>R#vl8+7oJxBgiu|xD>2o@wb)bS2wpSxPa}nxU6=`IK{c%j*
zzy;fL=Ip*V*p9PH)VL63_p^uMxdw~F-U|E$9M~(Zi}T71FC&3RfF7Bxd>Md0jJD55
zJp*Pt)AR2ZuyfPzg51}Bc#3*0I2E7f0a$j#`xvHgkL7sbb84I=e-QPxS-Ac_wol@f
z)Co&KA&py#sHbBj%Lw`FZ^gu1)JOZZQ9kOaJu>kf^;b}H8p8hPTpR|j4fvRh3gtg#
z8PPRZZxi=c`hUSO_s7GTkiL)?k@OKzqg(Y%E}*>WT50UZ*CLj5pNLBNlJ(f1cBO~s
zdIBy!${$4gsUY?c@_na{6H5@w4^?=eo=K^qUFLubjtw}Xz6}PmC{MhmKbz|5c-Ln>
zzCU*~Wgj6Xv6t+`{9+uGkD~{LXGx2`WYsHt~bsT?tdr^<#?f*H^_kQEI
z_gKyy@Rgod;Qu^$5XWuZ(RzB0bcMRO0$!ImattU0yk91*^qlT*vHsifdWno@bp3mL
zl^>1!+rCoRh|gDb>+2eO~ldaU>2!3kTlzkU{e_Sk+>AYmnX$?>oh_B=>{Di2aEJN+^_1DH!E~2Hj8ry4@n6v@Q^KYJP
zLo`=m2TB9&35ch>xh^f(0Mm=&Z&QEAZzhm~Y4(`)G+v&O^`5pzr(;6}=D*QsrE)H@
zZz*pV$~jK${;yDTiYH$iQ@fw>(2@F=uC_*+udy+N*0aJVn(C=|%p?ixtA58v{q6Sr
z3)F9fJ^M?`tDBhrjMuTByMuBwXM0n*Su-psj_X~eyxOxpmR^_mMU3VG!z;q-nRx9*wik^4oJ?9vh~X7;4H!Oj```YaeGmJnoNztEHxhK2
za${|VKh;e#ysEp5;eE!14BxxCn#mvf%b?sBhIdK*VcO&2%djI%x`2Siu?|7`c|Bq*7R4Sv4
z%4peq6QT4@QAR{I84Y)X)NN%`q(njy8PSwIA{7xWq^=^_*Tgkk15sK1-uLgm|2mKB
z@j2J~ywCZp^Z9(v>-oz0cYB=jzomT0zvbos=CzW?{{8Kn^KX0jlJWohZ6EnxdE(#F
z{lEF}b(@%VWDn#$cvlYYAhICV)Aq>iAmmv#%@dc2BR^t5@%vvjELe;x7QcwT%hJDv
z1uYa=Q2$uWzG8+6{ZbD$NA)s6>uW;4PZ1M(XtQekWlV?;4$G@+V1m)-ia0kuCLAN|
zS2I;-KtaXb&=@xch?M$kYLNku%*CDAsFiS2)
zT~y~OwGG|1caKSLTZ~wAe&%$b~qoo6=FHQlmQA9y*)GO^#u73iarNtTqy9BxQs5+
zPlio{PmJ!@P(g{1p)Y{`BzN1)j#dOwKgC
ziDVGGTPN?^g`BYDUj|(StoNk4e(dQ0bcN5@#v9}R!|r5=52H_D<>@s8zp)*=-G#Q;
z{^CjgRqIG(cUopm
zeRs&va@xGe^*k9$U>ahuUen8ph%fi?__!Y9aAuSUDz&$CY`
zy!IJ#(F+tv0YgN{4_Oql{5}jKa#czzG495jr&fR1OoWd^7p@AR4`N6u&8Y{E>!(sL
zJsZ(4=VQ^0mq(_DA+fET5-&hRKgIpLOV3HrxWw($2RRayco-@vE+>M&=F#WR`-Wj>
z){PxrSU)9ZU&Y);B4kdyBZ>(UVZYp&mj?Po5HG#rm4rTs!4@-LC$}T_{qr_q8#yA7
z)U%nPqC}8iRVMn_ng|<)49fx~Ng%4E_pEpi2?Rx1`a#&Q(%6=6EsX10I~Fj3-1!ie
zRhWPV2{!XyY?Cu1!6uv6gcczjud2{X_mqi19`}kCEhfTXX@7%_3JD_pBhI}?{=DpJ
zWsyGSFwFm1lOnJN{e$ZL_39rHq0Rn~$#zxbxkvn5R}Um`BV9Dj&?bS$QG<1@)+8ux
z|C7JED!d0O#->}
zXN4vt63nhR(Dcia1QQ+2f>-c36RiZ_@iY<`-={5p_=w}WUaf6rf&|ai%N#tBpRe33
zHqn2I1ga6|yMH#}IPAE8t92C_w8aXwCD7l}I`qkhutUhvSB_=)Mv>rk?|DB?KMDAl
zVas>p`82Ja6v3~7^>nnsxRC@2F{|MP7q45crKxwE$RJWSLE6|#2A!a>j1=Vfl}i-K
zITNw}QR^4mMagjV%2api2|V8}-8rWFmW+O7DK>6c?}}fQswR&F?V5oxyEl-5-RxoN
z9Y%)UTl@-wy?9=+{yHZwggl%6wOi;qWH6UEVMN-J!F}z@rL}nc&zIy+#IVR5+sTmStR7cnMur110XS7AL$mE=)!ld=elQ`Zy*pT8ku8z__>Jl^=rw3tzJ=B_xnb-0&rKAI6gv_rEKtLxJQs
zua)+ouKJG{ho<|20w#f7!RE*@e8~6XIr|p{v~+6UnaWb(f}YjP2=aW+eUywaLmf4A
zbj^QFpu(o;$*paLnApeRXRgsj$?!@M|QMtmpt0Jl^(E
zjj(h!{kXzTlnU0hPedJ1_r6W}WZ$-I3Y;7II;n+u+A4W$GP^+qRo>$pD&@#U+A&zH
z*@yc2Zxz!Qkr!~MoL^lM^GxnHE&7f5<;e2)*&?^d;Iz!Ot|lrtJmX|Z4^qKN?oz6P
z8Wq?f6czLRRQNG4SpAQleVEtrF#z{(LpZ(YZ2=YP?50CZx+!3wE=fomrGSJ#X+-f4
z1p?9o8~@=xg>H&7lCnYGfR?kF*=-6`pZa9C`aT8t1WhawkvrvN>^fb`q(DIxpP?ci
ze|*p9gcWCT93mAOF5rFt#Of&94&)Lz2!?q?twlY)YLm=%#}U|JvPs^RNCgeUVxL^}
zw-r>UAAgxhg9|i;*rz{fP-7CKHmExSJmVwWEq8Ih40b*$=%&J;{Y>9!LmC8--RV8|
zG5#_wk8eXX*#CFlK7HIDGQ!)KDM3_-Ec)^^s1^ANxu4tok!SFxvZA}imo0>)vlntXNf
zxwHEC$O681Fej*Gx}rX>@!Yk6tnDMP`HR+BALP2``90do5y9`mv60u>it+ot_Wb?`
z)bSiDRsX!4LW55Ji(3v!pfB^K)~knajljxx0&aSm_`d0iU6n>SzUNrd)hM|W{eG*I
zj_U-{V0JV?_9Tl2E~~dnt%@3fu#*XeSKf@khV*0oaj$XSw6jn~>kkzid*Ew)Bn`55
z4>`aR^xvDCy!vP_`qH?0#f+k_Zb`|z<;-~0GX~fiG#k)CH_gX45a);2FXtS$-_C&C
z(ww*P$esH`+bJ_-I|g5*yE4gCzzA1~Zy*UbP|clXq^1v)rn7;oFC%z*UT{65EM
z27D5>%PlTvfJx%RI{Iw}&^x!Ro;pPb$I5MH78!I9u+J`xM!lh%W^SF8KGs(#nh)B|
zfHhGD)xm|xUsbUDn$tZ7^(`{TtS#tJQBuMA8|!ng{zh8
z80@JR?%w
z2KADD3P=BUjR6lOSIjKcXF`jLK7rH2glQv1wRGg8m=sv72*CZV{ps`cDI7<2L#g@3
z*9@pP*lx*h&4fAbRPeSyI
zW?~Jw)Y>o(?CydarOU=aXmHX!z#jR7W_5=;
zx3NITD0;mP`ZSu|G)VnE&jMHNv5fUt&Isx1b|HW8=6Su7k}M|Js+t-;JjFtuR{mBO
zHVb^BjYV(DkApQ$X5X1ufI`pnE!A{6IXpqGDONJ?iM;>y2nb(KJV;7pgkgf?;mV)vq)n-UGE4;Y6pn_kB5e}f#tC!3$PFO0*@y`OViw~Rwn
z%i5W42juAx-fe%7gWQ~NEar!vabUbr+*B>ZhP*(R1|`hXcyOPbi{m&9{h2wluVx&|
zU!2kzP+}v;t!G%`EE^=9NByqOV!SH%>(wM|$Kl|~Vs$purqYIzy^!zxQ6$>{^Q(y&
z{yW|txriSYD+6qhv*;K@kbKC7iHOS1AR!Js((Txx26+y(lUex
zqOr+rIO-$uwP=_PD;ATy?qa=Pf|5P$bLd;uCqQ26#0HD6K8^Gx=tn=1bHMSo1-8Sk
zwJpNDyr2Ckk1^(eRc%xq=^h6L@&eB)r*mLcBU5;-GzWAil~dR4LM~-?!0oJD4&=O<
zn)Uz00eM-0@Y2T|@bx}r`bP%)Rkr#D1?xpxWL6%{JNnUm3~;*%fDA
zJ}Trwxu_{O9NVXTGhZvq&xJJz^DD5zEb$aO8r!a$8N>1uobwyL63rl?&pDe~KkPbKp((hPuNP?AQ0J+Lt*TQI;fkHy(5=sLlyr<-xiV{?lXv7u2oC
z%C=bZ;OUU}?F45YB&QvfrDgM=>PFtvD>6JVl&%-0yYN8gcKj~&7d#m8Zj`P};6cI<
zNm~_R9z5|My~>v7!Q@neoD2N2eqcb
zJ5}HFz%ixz(~}e)*cU!pppz?l8jq8=^Y;#xg<(Hn`Tj{i9vph-^)gI}2aLoy
z>pZM?>-xt0m1!I(^eOpm?1TQfiua3%I8Gr0ni)5j@*r34Td;x#7fv4!*1q|b17oLc
z`mL*ZpkB56ROU7Gq3djQ4qC~BHA5$Yr1zqaoY0r>1EqLAwcRfp!1HRDv~)w7A_szx
zCHlQRk9@M%mp;8eiPx9uqC6?QkE9*-hbK(j-nc7;t;~VzpPQrlcd;Qw&*7Y1{WzEv
zcpZP>H4dil>$PL?`f%OKb7AN=8`tk|x^cM$uS*|uu5HEpsrxMr_d^3rxFwaLFO$lF
ztJW8USTo4So$N4Nj{MjZdeb+B=S(LJmoTREQiZ6qV!^
zq(rHlT^BA&2cEU&b9F!WzW4pw`;X^&J^$?e`{TP>Yu3!Hna|9c_gXWTU#yww^6}%Q
zj^n}qjUCIw&m%$q`R_lY-2d&J$wx*1_!gCQ0z{$U))1t+aHzuYaI>P>`3akNdI^ANRmz|5P?N&oXd>j`n~-W0ky@V__wKUMor{fYdm=kDX~>+To)-`D-a?NXFSkN%5~
zhdW+h_n@ttf_M76)A3G(!hhpk>lYLp7_uceB+%W>3bbI9JpZZR|6Je0%ZI5i^q=Zu
z%W7K}((a(%zkUUHBypaDy?otyCdiZj_&nMoJldn+h5mK_Xj8UhQSo2$ul~RJwr|3((N;D4VdW32kG|7Ffu7Wqy8!6N_ozg)hXOGmhr
zsn6!_5;pm#_lsNpkN>ZFgWPh)r?AOC{=e#pQURMZbM?P-DdR-h
zTs>yy*yInFzs;pgMHczW<^T2j&yHrUJf1edCUdxUbaLgwTwaAs_tSy0$YXB*jJIKv
zXI%Z$-0~^IC}5EzT)vB2--F99;Pxj)8)g#@ok$j0LI=zyL)`jv1h68Tyne^D=frKN
zliUAMF3-3LHVOXB=nZh&nL;OoMV@eZRzFidOOa`JD!1M16BvH;Yo>e?^`~s|Y6ipG
z(sPhaf@&E4fjpzXbqB+H?m~lEM2uVh9X&MI9SxoXX@n6P%p%LY&>$8$c8{s&A;pwi
zUuJkaJPAQRy9qmFlgA+pA99!}&o*KB+A|CtbBjgW`^~|X%p$p&jNGc8(Z8NKK;<)k
zaALdTbxbOH_K5zSm!TH9Gjd@3mK2pb^AM7
z-q_2g^|$#xp!|=u4e$Y26(vMMUVz>QhAEy+r;Z(g|W$9+co+~OKL3*P`+bRO$SoG36o*P!Tt+Zg%8ap
zd#cWS9RazEzXdE1Hpx^x1Rs_~6zV5WMSh8Z)Lbn8*>QD<7x-B#Z}9Q~Eo<3$8OvvE
z3TFR;@@1LP5{}s?475AuugN`-OXW?tjP={J%8ilU`WWqr7=!ia(A;TI%Fw`v<6v
z^WJA@kGh<4H}c;;o5PN05s|;E={(PuIM#~w+s-ZH$N5aZe}6X4UsT)+SoqL>|3y2b
z%?`Kj2?5pKIdK#AJ0{`fa-5&o^*3FSx;~V>gzX2ctUX%=)Yj2`<_(}Xdz;K&0!>N`
z^~3R%%y*MSeWSx^@o2|h6Q7evpGGdH-}~gaR{FhVBA(OrR3hUhi~VO#Q$B)vLE4+Q
zAYB-nOxL;FO^4}NZdd$c7SgoY4rNFs;~(upny)`tj{19+@YDK=+jMvQ1%5_ZWY=>ec;wu+fzx%5)ASvX(MB1FN1
z!@Pt}d5`S~R%wHyW05N%i4d2-IvgWH_xbi42j!73sJlnU6&J<>M<|PYEEEF=&LY{l
z9^jbRq)OvhH1dTm>U4iAL|&uwA-dLy?qi>mr+cux+y8DP>ep2*q5G<%V`&`9g_e%e
zd2qiJ0*(j1-=<0(v~#t`=_xp#pKiyFSEWh2hF+bX?5?5>(Ha4_Ihtv*;p9cWxD>{^YOidFr3G7mrZ+
zM%n#TUvTbjN_9M3sGT8~p3(71?fOaQE7Nrm>4!0$V~);
z;JaC_qV_ttHqiN7>d->l6}!_)>B2wI4CudE0|e;M0Z@l}pLQ{RK4%9bzbpvL3FJ~O
zjQ+DD%y}`>s+h{_7BlD6^#{!P6(0k`0{tuN7=1^jNXmOkhf^Ax$+Uax3Zwt9mN}=z
zPt2xv9up3t_QVP3QGL69dCCu!LowJnc^PA83LbCJ&$LM(26;UmkHGt)c|b)J7`pp9
zL(Rckqg;-QU&O?jdh0?NS~r8C15+3(4XYOQOX?Y_Mu85S*q&y_-*|TxeI?&Eb2rGa#a`kzzTj0+k0ZMSXq27=R!>=h~XdoA7UBhJzdZOHRie5ANqlL_P
zg}8dT7%4+LCoVGjI>-W#yAtUK6hk|7nt}g^S8KTTz+pl4z6SD|2NJjTwaemuj6P#Y!c3GH=kR+lUsir9T1y{Dq;bfek@ZyO@z_^y@@G5A&(!7O}2EnP|8uoox&y?uP}CS(infIDuo*w
z;+D%1)Skxqj2}s+`dWc-9K##
z2E#sitcviqop}Q%6eK;Q`jz!i=yrH~J$6;7OEb<^&
z1tx|~^mr_(yu8bfw&!epit<|=hN<3{&T(Kk#E;HGw4GTRgw_+fY!3rr5tFklYS*jN
zkLfwKXyzH1I5tt8{Qw3AakKby2*lVVM@Rv-AdB2RHk0z=E`O-rs7Nu@TV&u4E|g7b
zD|BJP0I!}q1z#R;ON16IFcz6#8AA2fs}<7mt&d0oSH>o$*WOcH-r6`7wl`l&UPTrmhz|kE6YWnQ9hS9n8pP&
z?99Q{ut>4#HX3K}pF9_X<&z}(G?7ZAT&_WC=rWz+{sYtc@{m9F_?Ff%)GO;!QEUb(
zvq|3z<*UR;LlJk^AKBW6c;=p(jSRN0D)Yj>9q6Bz>Ti&bFxksSeXkoru84Q^Lf_2k
z1Np`6TVscS{#y724qO(o_~Cm7^>Xqhr*ME*OVm&h1wPeEUv3Ie&&l4kSU-Qp^+i2U
zUh*XBn-uWIV{0aT1Aa&&c*Gc}@0zY@KY{8B8zoJK@#O{cyr
zfpX!3-g#)plu0Hgh=Ub1rEBCsZ{Jy`i71zgKCV*+bR;X@7jdbH@}9NW-n)F21)qU`
z{pC4d8c^9I)nC2>-MmbdEdu4azch5810QvMeMu`&>;2!)ID$OxYD2OKP~-OPSt%fo
zP!e8&aYaYhp1Zd}ZoS#VbS>~trn-8q2Rmm^^SFU{Jx(%Mq!1?rZewu)Bz^YCArq%$hlKUx1
zdMfE?XY9a*lepf>
zzr1N12UI7OU5)K{%y^fL>(oATQ#O9DZ4)`h7#}DsDO1AlW4QC)4%8E$WtWQM3^{mC
z8~Mkn@77|xCx1+X2XS~)ucq}wpmha{)^!0rny+t)^VE9UaX-?te%i+ZfQEdN-;DgR
znm5b$qF&a_Ir%`BKM}G+I|C&%(~ugCq*tT;w>z%=MtaKd7(fo#KaYdT(ay)tZuC2S
zuHw^1efx}3f7}lemF)&dqk{E*p}hO5=xo#vpIz^Rc0N6+)QjzWo~sjq^xZX|g*YF}
zUZ_OkI-L2{>@Mmx?2;Xz`Vf-UaO%#V
z{Xi?j1C_8{g~4usK(LQS_fvn`cXI*tA3KENsULHGuwn+9ph@%P~y
z>G_~GDoV%GISCOUo4AKr(R1Q+EVyCdJ&q7cueL$@1lr%WoI~jxhv}3GM_STzM4{;b
z<^S>{Qe=}UAqZIE`&v2Camjy5r2QW|gg_tiWzJ7&z2F6iPPsk(!n8yDpf07L%gArPP$koCZNA$S}DZ>@Oj!88G{x}L!Y
z!4O@e{uJ=uD8C-c*sFUHktT~|u3_2}=ws@YKVjmd$qE?20A3AY&R4~^m{^5#NSB!p
zLqAE{Uf=@{YOmwZE0j;YV??QrlRMR~ZkS5>!fA}1Icq=+?TAsJ3;8lG9uwu_v%CsM
zPp1W{fxR7D82x$J4Dg%Ke4xhCj2-Ei4utk@Ut{EIri^~%9fnR8WcpvP#_(fI8T|4L
z({j+xsbu&pP`~sf^Zn=8U_b_W(%lF+S){0snMWr)zCkWh#I%#ejRzf24D#zZji3*k
zm{JFbrcpk`9cLS|z{_H{Ku_Xl0sRTM9D#)CTCcxVL0AlF3L1$3SSGtL5Yrk?(6
zMxM~X@NwLDQ(_Hcuc#NJZ-B#r`lFELL+Y@D(cgHIPNNaP1N1&WrLV
zrX9HfM$e1}PLQwT=Kl|_VA{F47!88+ryp8+>@!N2IA?+Hd=1<9wRz7F-JEAbb1D3zY%2da9++t;RLXW{)j52!iqnsJfMfM
zV?-AR1ouE2C)4uinGYyG+lg7PDxa8jJYdQAyA5-5Y5DYH47F@y){Cs11g-CU5hG+4
zIj6me@+;QhOBKlH1>;Q0dlGa
zm|zyMyq5z33!Ajpq@^Ptv{&gq(z5W=wA@%wpSGL7znS_u%VmWSI6*#re-rEw*jGbJ
zG%lDv5C#(n_jF^1tFiu*q7$&AA>Vb^8&)2RY(HH`@%^s#))mNmjxAKe^034SsmL!p
zcWMgK)HMY)NT;9mQ^E3$hB-Ho_j_SL=TAYuK_Qx&9^H8f_
zwG45?lHOrGq>k%;dt&*xq(*JjH{2VRjC@N!k1gV?!khA&SwLOQg(+SsoVS?TRUzgA
zI|1-QaxDw>`$xWtBCd1t(S!^btcMRjgODEiIUY9n-T
zD^Rsx(XtqK+<(l!Ljm{+(YA5#fnRRj_iz=^#R2Dx^g%8r7&aN(k(n+!G9LKRiehtr
z;4Pe%elrDr`|25!76GpQVK(52@xtY7%i#+^zdM}D@rCj~+Umd6fu^;OcD(`oI_0Zj
z62=h|u84X11GOJ7cj-Rp|80|&QUN?(zUQT@709wKI^tE(o=q@LtD!Tl0YgpjSh@K4JW0SJ)kr3H=RvROVsacA`~$8jkadpKIAfpcQt(3Ai3o
zvz-g)0^R6-Tje9@&EBHD`vB0atT&sIz@EWd%NtJvJ-_kRkvX7uBw;AZ1E{gZu;VAt
zyRMSl6$A6}RsQ^O?9WQrXO$$(YkvJ_;WEW(Qg
z7U9gY+K>LED}F!_=P9P3>l^CDxU@$hHPxH~$OHCm$*3E~q1_WBx3q!YvA_FA*+A{C
znkZuXuipEcm!P~c-xMy=;ruNVr02*M{Wt&ya1R`iLhqqux#w&^qmnDiiw_Sj
zK>Y|G7S%I8Ka=*usyqa#Vu(MrNA+=D1P4z7B82;v8jUA7e;Nx;!%Ym_&#GtxvV#1}
z$}D;gw!O-x=WtazOasWJ6&d{Skn@Df<2v_Kxy$}`s&96X!7FX*m}mw4y0ujA_tY))
zJl(e?fKvN(6?(1}#lWyY@1E@gO2>KZEIyv$1Gsxfp0dW|K$X9
z3+?U&tB}r{#^Amy>dd@6h-czP8*c_bR&a6lnR|O_e_=7RD4k}60TAGe3Cy@Q#WHqn
zgs+EkM{fKOhRuMT#rS!E?$TuJi5g<`3}6+YemFNy2-wT;ev`2wn14V`NWV{G>bqWJ
zcw_Fo6uyMz2JNyjtOvbrcol(ssoY7q
zhw4dq8Ni1Jzt8ETIJ;@=2I>ba+|=p5FZOr{69aL?^#9n+aGB}UQ<(G2J!-x
zSI;B$vG1jMt&3}PVIjjk5xW3ZAe-!7o=M-;=xmOJw}9Yh9X|MCd3wYv+TN*P2l}q&
zP=*D4=O9#Bu7~Bi$HW`aPN}mq-(fqpOHJrIfYJpfQ}H~z=DcqY;-Q?YS_W8twYhUC
zQoD7M`>|f{lgg=h&XmlROT%-gcF%VR?7*+wjDZUk@B`xe+>nB5HvTKDJgB^Y6sV
zAO`d7`K?1=k(bzyp@A<|KmM
z$D%&?eH;sM#rH4UlmTC_`7x(r>unbKaZETa0qoTomV0#?`VGqS7Zh8}Cbz@+Wo-Jv
z-`rI0`eOw+$xNU<{4=~unsFq+;U3^Yf3xnmGw_bbrNrrnRUgsVyN0vlc^~=wy8PzM
zt$id*#Kezv1M)!G2PVrsfV`MTK~PX)KN+6*db?~}ANgY;$~NQcBWeme*EW?xKCG$6
zr7scoiC4qTGceA8*2Bh4Ft0oF;x}!X0PFQs>W1rJzl!}%p$f3qc|bg7!BUuyJsJ=1
zg1se`=|u+>*u=N-_Pr;dSNi*=E#~!i&$!8HJPY?7>3bT_2E+TA>A!MsF96)#zS3Ie
z2jsJ?@AvS&fOm76E`L@*|LY7>H}Sgw&YL^0SWt{j77%^Da2hFG1wgS~GBKGKKw)4%{u@R`*O@4Lm%{3(O}yJq;!
z^@VkRp}zEiI?&^OH&j)ieA}&#evGHu(>*)eK(C-j>gz!mUt&eZ=aV20mEGrW3;HXL
z=oDaF94+wr06Pxy)U95N_P0Si7->{64&u|HdqwiQK`*aDiJ!aC`lDXC`;^P13+YJ_&rYFyGwnj>f!#HH}rLu+ScN4PP520VJ`g@$-
zH_z;Qw;jj1(q~x(+Ii1rHDtu$K4@z`&8PW`OVYefa{v8JsCVSD_-Q;}236mep#3)6
zJm4UQ`;wnLtFSzy?rs>)&nLHEGcjItN{Glq`+4He{=oc_^2+p!XirhsVGX3`AD;8a
zd7Z1!o4f>guctQrc)z<}LN5m6zxkIB%i=jWxZj-Snap{60J%ZG%MQ`=y>KjCdqH`E
zO)3B%i2M0woiguT+oFE*iaFab&R+E7
z%~+g|2!E)nxV+fF-=&y`D_-n;s{%C=SZrdWYhl)2o^H6v|Zs
zRA0dO8vsVQ|I{E97a1HhrL=UxRjNO0=mL!cS<=fW4w|*h1PuHD*db)c>9g%)LXayHMa6Yd%PVv^};+n50
zPbCl*g8cI?CLR#=WN_(TM`#A*+Y`VXp!@$MP&?ILz;J-Kb!2cVrw+nZXlLp~m}ZcN
z?9HL{(ZP*WzNnJHzd=z9{^UFNiOL5w?$CMJeb0~D`$tfo%6~pkq4i_mDpTse_5_Wi
zZrr&D4Z(T~W9IL`($CbMDJ{(WxWg#Xj$#bQffjJz8|GhN?E940LgjBcOdN8bdw*rr
z$>94xJdE76M}hWxEuWEV%P`|=IReuS?N0H;ix)VzkLXdVsmII@=NE(j%|#e|-;eP5dOrTrKYfQwK+K4%`~6I^%Dd3xKh
zna+dIF1#Ru{RP<&)C-Gmro5&6HOliJ{sacW`--6>RBooVfYM>xZ?ru9+XA`{jMpyj
zK)sP8v2?w;99~QNDSBW-?Kw4$kK*?L=_`uZkBNx`UB^!AT&W#XZXTrkhumShA4$>_
znx`q4BTDP7H-Pjow9m41hY5l6?Ajp+U;yu&A@m;oWqdJR$AMnEX?yy)bE$nNt!`6%
zpBndME7s#bmrC1vSaF4}<2R1`3bCD8JX52wJmwYeINaYx9lY=X!7nJ6Qha~<`%-G(
zZKK&6ILVrXfyvM`kAtldrEjtURtCPjQXp3
zE8qnu#P9Lx^gOwFtV;#oE!)jdo0i2MK`R9TRIl0ADR6tcB;1<-sgPNjMMhud3aP=UPFm
z!eh*fIeohtjCQwg6zjx#`~9V2vf#Y3ioUfFB*SwMM%)=Lo^;1~J&Zm|aXT2|}R7&wpFZ{IYd9sc9<#=Qi(
zc&ccfGt?iel(P6R;Ju)ySBpDE*kmoQh<+TDx7pur(S`bFB4*tU4N
zKYb5zkNTV$_|D;DLyFE_DBpA>%IYADuRvhM;&Q;PBh~A5$3cF$$T41c7U<2c;>(-?
zxTE&S-l9Cf1KTDken|$Ld`5u(cbo>~Azx;Negb?}dfE1zG_;pH+%WbCoX1PA9rbZi
zfw*YJ@tU|C7I`ZjYh4WOH5&?fEqw{+oci_oC4d8+L}ja#0FQl2_ExT&)I*Ns+dL2K
z0N!J)Q8M7mCC4fThfU$V)4uN;E6??k=kGeyR5`t5+q?Yr^T7@k1+ib!U~lj|{(X6X
zdu}zdUiQ81BW(_|I^W*xCo4;*hCc-R@2uw?nN|t--Fw4F2M35N#FrPif9xmQ@`7W_
z*7TFio14jXA9y#rDPzfx0LWYCj=#94rJqFCY_xqRJ3y+{=PG8K_mTXq#+9rey`(KN
z?V@~FAKWt%^Ll>{soYyEG^eDWR68yCQ&HMaOhnX2#kRsbFUc{Zo`D?VT(N({P|*N+
zadw*4h~@z4Qs}nnDI6dIpT8XLd)!Or{=RtP$=zO}+cdB0L_jYY+%RaEaJ`qD~@D{l~?hdRw%jnrb#jw7-b%v&#AAX
zWUsLl>>~T}?N3!t?;={+6;rBU+@31ZysKm39iEu^^y|~$952vX_!9QP@q$DbO<2z_
zLvxmkf&EEV2liZn_QbEsd|L|gi*v?GazL+oUTK{(#6#Z>iR?K5dZ!e+52Ih$lUkoq
z0`c>D!RL8$($H?qk31=epDWLOK{EWz^BpA=MOAlRk!Tu2)H|SK{4RP4!8k;QG
zcY7>Pf2;>R!?y7{VPI$Gv`pS*(0)|ehr{Uac6q)0g87T38fq$tzwKNT{8Z3Rg*CGf
zmmUq{HJ%6ZvQ5{1R03b5&tr=I-twYKLph9}Ek9+-7SNw|qoG<8^jPQjOvU$t7pqb-
z@cq}&@Td2x5J%?OU+y{#yoXHg^|4T|xqspEYoHfUo5_dx@a%!pOTNJOo;MN}pbYb1
za@PdVaw
z@vD04=~PnDP}P4I8eh?_hzI27j&*Yj_v5hRq`P|=Tv`g$pAj)
zN}BS0;J;)gR^UCzgZh=WcrKiLs&9b#rlIi4=h*IMslmla6UTg`=jXStk11Y=yl?Um
zRj70nlU-pN+R^xwr}Z+zaG9(NAcela-md
z7kRs2Z#oaDA(tuc(($bYzzFN3=>_fg(VtoLJ;RdRCltp_>#V0ZP52N#kO5o}qXmEu
z?!m?|xX&I|4e}0B{nVZxkrOCyS9+1^zixo(gz~u_U=GjHN1Hl?1h0;o786sMjVQ
zNz3;nx={U3U6NF9t-d(D4{J45ruR56i&xY65gs^3%PrC-QhP4MnbL9IdUBiQl`mDQ
z(0SOR1V8fwyDujh(RR1Fi&A{~uuMP+?z=VN6`&bVzyC6S%|KmmE-4!h)Kf_l
ze1g18Pb?_FJ%?Qjjb9D#nZ3gDJ5O!lA_)B41!?%j0puqR2%hK#nxL`=0vqs$Cb2Yr
z6wvmk_UkRMqVZ7Vf+s2*&}*K^ytCA}mJ)~UDei8GL29Teei{A936-_4u)QylW~Z^e
zpqebW!GX9|>mH5A^`=^%L;sOHGke(~pbN6EPoQ>%E@fkWpkn>-WyGBq(~d93ceM$n
z{y8gvzUWM!j_;j|l@?6F{lD|!zFmkH!xa}_!2CqnzGLTSz&WvhN!@eoKV;jYLG(W+
z#+knOemlTjR15to-@_SYLZG*U_d^Em%NLPp`iN&cPb?UN{&rI8yqYb*D@oRDV}ai^
znDI9&3HW@?au399Z^wl{N52zfX}jzI`0Jm^Uj+)Fyb}uPZpUtxp@}!%Co=?!IgY#}+$($7m@ZG+PNvSu0-=9mtU_-9FJRv7~%CTiq!8{Z(gRJ-M#=H~(8nb}$)NUlNAH0|U8vS{_`j>%OV!Y@`?F)M{bk`%kf9$3O5)#cZs^M(ch
z-}Npr*f&6aNSEjGgm**S_rQ5)LO1CPye~g(XE*UNKJxv_`Yw{+kg$I3Kqr~5aG)k-
zeK)D*NE`WRbdya>R!*ui=_V}c(dPv+UF3vi70(9uPO>Ga?cLU=UF6TP(9^mHx(NGk
zhk0NQlygLM#Vb0=*Y?$$VyZgG+0HITAB1dAYxG+$jFu5VAPV#W@
z{LT}~9Yj$%H1?}h7vVYZ=g-TJog_~;apjR$o#fR#6$#T*9pnky|E_vPJ2~xIxH(m%
zi<}lN$*cnX>0vpkER7!GVdztQ`Y8Ajm`v$@b&>~T?fle~
zyNUMaP=
zjLwF9bmmNd!n3-ASWlUKaNmS>!WUQmN0+~Yya>H=Qe2{&EL*hDMYgbqY`8I{VzjoM
zm{+H;pLMm7WtYR|MTWPNsToxfolCn&dOjZz-?%_{}Gux})H0wbg+g2oCM1DT}?uiv0eirtr|A&qpN5FpNl|HsD
zu&)n=MlX2*>-hb~B`%?Cc;6Z{EqXi5Ly1G8n?^6(qnaEp-Pj5KeZev1oU=W|D^<}q
zdZ?H5FU&65GzH$(<#)}T1NR5VgfAVs1Nb`Sw5{ndyPJ&5(_OV{BjEZ1t%1W}hrNO!
zUnJ=7sg4~!40iCZX}MYk`;-)iH;hAEePwX%eDHfbVIoI$!5`v&ovWf&Rx}
zBc~S7Gvl#YG6CAn`)Lu_0eXXT?WS*q@_pic`k2@MJ#W$F*I-??WKk2(S
zPyG?Zt!{7X%*#N}s74lkhz<7#J>NqbK>mDnH6%14e`CJl2A;n&zxD|aL%Y%|=1;*q
z#Nv9DLd5--@@4DsTq@mTa{Kz+_$Fpsd2N}Tk$>D=3S;Aa2~F;e!hwL!qVgWIvm|@A-xy*
zndarv-pNqBwS)hJ0@@>Vu>lgoaK28+UW|2Oe0WRm+iO0BQ(TvCx{umXF|?ZEzVrccip#FW4^uoD{tRD)!#m-9o2i|lK}A$=
zv{i=cwZ!F8{FgLq2F06K!q?M$=elILR)>C9FG{C$cs)FagLXCu^waOc>&M`4jWD>@
z0ln5IezbngQ%ocSUQ|)CM){}X%W0nR{;~0N-sYC3P&~ZS(S^Qy*x=Ag?^Oog9FYuH%12Bhoa-U=1A^_IwuFeRg
za|a%)VZF^0mr|U}_f|g?_44|6=3#lw*@eM~KMMlh*`VH;mr1n02FFCYK1-K7_oDpG
z&|NQ#vov+(z;B{XZlQbT!wg2IcR7WmfMtwpT%|QD-{5L
zC;;A1l1D0-$Bawwzv2jVs$1ecT<2K{d@Z;xr#v^`hIlbX?yyiD@b|Xa2h9Wec%kP}
z^a~Nc-_dw|$cQx-{e|bx6s-)9i{!i~8bA*(Sg4ErpZAnGiu-88x)enmpR9m+AnLh4
zer1ye@~Pf2OTPfW-`3Pe0{Ba>Qqp7(e!WyFJjca_TY0PmUPy>1NT
zF%P)u#$o)trq8?j9mp%f4cfX9-$
zw877M=bP4K11{_--k32R{Mozft&&<`r%3bF)J;%cc3Uje8u(Ye1Nvv7y$zl%r3(Qk
z#-0tA_X9gWyIQ$lfO@WBS@%~#Ugz*l&LbslSTFOwjJ|>M(5Sh&F&5f?DHYY(4d>fo
zjS17cp&k8^q~3dT!CyXib!_5;_mUbbrCs5CsZK1f-}edjLuUQ{w{Tu{IIsKN40*74
zV{wa+z#h_qP@?sLJ
zEcgNEmY4XwOICv4AjouBdSmO)5IeN$=384rNX{|KQ~N?(A>ZwrY(QRM(>8a^5fiz
z^?*0;q^>-BMQV`9ZCHM=H-STB-TftYxeSskzpA2|9r{UN^`Se$vj)i=pCdm9c5}%2
z#GV9I9}el$@j5JVl|u}Rk`0f%Zzs`K*8VM>ZDi^FYr78bX(vfLk|#%)bP)}Wp|1CT
zx`}AP)O{r`?PTu!pj+#HwUH5B)rqz%+aV7s+OhDkfAoy?QKa&IT>k1Jm65NIb$
ze;r$EyrZ3nwq{tbS=T{swQs&SMWdZ?3|Q;_I=7STZ0Uuqi`$8{!t%%^q@A2`7PwVk
z)lN=@<$ntA?j~)&eClMKdWpF6e50eVUITpDcdo*ESn0RU8uR}LM85=dYj=~86M?y=
zH+o2GY{c0-c<i=&(xpWn5>zOqR>49<4Vz7Wt^~e`KyL=e$)5fpP*@EZAn(Y0L@q8-UV;B1q=0mpLZ}uCI
zt4f$`MVvT(!IM{P;4ejA--G8*`y6@Gt3ZuDq;g9^>Ae;}}@5TItYviv^#4m?Sn_JP2@otN)@LXJI
z9Xu2Hb7AI+NW*xh^`qWIzw+JKkMs0QdXFEm$duw7FTZ*t)LZ5*-zO@Lk#yu
zJ1w_idozUe@{xX=Bo~c(g`b7L;CuXyA5zaFfAY}I2S~kErk=t20%JFBK%PByyKX1&
z-&#&w#BraohC(fsDtg|jJ6i&|z+agm?$
zwlLIF(dE$g4%qq8-ytOxe7cYJG^}_-an<+Xo43(kGZ*-I0??94$`qF+80KC=y)zGP
zpFw%`gxO)Jw>H~uG4c`D0@q`Ci<(3f@`*eu0DJ)lrPv<9^V#9hW%_-jO6Oliy>|w?
zbx_az2xkG#U&i`rZ2Voany&PGl)s-jiRLTiR;yZ|zOIMMXZ$`n`%Zac`JyFVw~=y^
z?KUCaS>hA62g_&5w~C?uZ~qm~&jXz#cnuzCv&hcHg|+xSxg#F^J6FrJEPoLN5~_`FSyn=lWxd!hRqT!%7+6A$A$Uh?d3
zQX}x^1y`Gx1HXHk`@XTDXRGvw{ym7@GD`0-9y3pV7Kk{gCryAI0esT)r%C4lH!S`f
zuq+#R{qNs7nD0}mnSLL~b^fuVWjpNS>*uG{`$Bt$ek+<{K(F%jbQ>Ee&))05!4UX%
z?^+qO)98Jlz1N`zc%TRm?
z$c;a%EGP#%Uc4WxdmjA9qmZZT(!pOBTF>%Weje~?a@wP-&<@Xvi+h#8e@+@m-1Y|W
z#7;@AM~mQoivMxJk2MNFY|3G{8Z$8a^4DHx#C_icdal7f%1DiOIC)jCC2E6Uy
zJ)?N;N@K8dY@3sSC$#(UZ)g8~I42egOpDzEdVXS-SKAp?#cYdTbHEQv
zy5C!z4*vW=<Ro=ll&VI
z=r`_l^1=Iniyqzz)ZZW2N6d0B1wVp3)enn;RF$57GRg0N{j7CxuEbiKyj=kIU1CA}
zHy8Djv%Us578`O%_S{y{p@x2<5^oT}zqg-EDdM~?dch&NACvm#U*ixvCu3I+Zw?9G
zrmOP&6yTwv-K~SGIVA3uz|(yL9P(I5@~i814*3@FGD>kPhd2oHZAet-kl$`n^Xz7F
zNY-|ZoGX(#gsmDHow$NS7Cbts_RyL`_CCL>lP&^zH1RCGr92#RWcuFX+am*{VC;5{
z_}&2$>bL4!AGEXR!r;Em+8oFu9xyxqV}RVew4?NEBjo+k%j9^wfOoi&lQV@w3OgTd
zUZKn($1I8H8ufN^<3#PUQO9<|iVQEPSlCWPZff#B1RP`5?^d(mbvtp9S24b!)lTFK
zvrfiZwv(d9pFRSh|F$CP%mHCJ)xqS$t#;^JW4@Ti>(mlm{`wY8ILHk*V@q&mpwQ$gwp79Au?b@r21
zn-22(z=uy0ggVHLpsUiim+wy`-FXW?IA6~XMAFw
z_mUk75&d)GI>@w)xHA7gpjYxNaQD6ra$ZX(*KtNS;kzj2@TIhe$e(z9dgJ2`(qyCV
z>VLR{%(`6=?i=1g;y8&Ct1rlTF-rb0f-Vsh&Xz(vo$ACK|^d^Ue~z4EL}J)|ik-OYZamq@8w
zBznVpXr-rz0<7VFHZ@W%^>H7ZC(UD~Ptoonf=W6{Z;$qp{bI_QAHh#;PoG!&0^VOt
zx0!u%D)5p|XK_*h=Tsb@a{V3nm!GGW8AXEMtJK`@EX|c9A-@>;T>X_I@GEkAYe&H^ogW*d%nx{_
zKzmo*U!ZsPT?llCJX~j2M#EjO*T1Y~yAbq0!QM#-afp#&T$nf5VdgfnECl=$sV`PE
zgL#lyx?Cy)%0ug##BkdJdfIp
z`3>XF6n^uBIlg$vCey{h!2dAQ`yrxfsB%TBGn2=f&U
zf?JC5yw27tcExu6YKBUYUi&?55a&T&ZS*jX%U^@9ALUIY>n>pXvF3{OeeIkTZ~h>T
zU2brz73XP@LCa300<*t3BaW?;c=8+Tt<3PE_{Bx;Ce4#MNLE}y`%h$vC1bhpDRYVo
z7s%Gp-+%Ot9-!}bOVsygq29o9w+hq?P)!~}DyPsv$Csj|E{1#?&!SA!^JmM_{xam|
zo1uK1#UYyKIXJM2=Ak5Il4zc}?`V3YjlV1A)k)Zh?bO;yg(4ox_3OBc_+F`LR0nay18GA^FQ86h
zR|{}HT~~U_puXpB<5JW+@$5q?=1bXUC+J23t?({3M4W#m_DmP*ZQXETC+dl`YlJ-k
z-r#JR=ux1{T;F9R0DasdzHJ4RkIg?yf7jhIETOg&l1^5L;Q_<^_LGRqSC7gqhr>R}W!+QbBZz*|iJp=LRuT?x3zd{_D?p(g<
zH~7KbqY1-BFrKRQeZ?`57Yco}4gfc-ySJ$^=Vt;gF*Nual?lHmcshOVEWkJBcLsN@
z1KjhYXo0Z>j896)A=(|{{d+pDH;skh-ge}7f-tmada84QA;jZ`3rej6V2lI<&uw1mMv9-$RJAHlzGS|M088Ghffu`pnQ1oZRUQRr<;I{fZnd`XZv
z#OtMxPwu@1d7~q2E5qo8{Ukt3!=>4igZ^LV?&f}C^XSi=_W5wGL^f7@+yHi*>PXIu
z>?aEGGuk)mafnS)rH^fVKN%lnN$q~>J+--)@XX%-OI;Cuk7C#}aK0Z_e`5%F
zjH2(6eP_;fk;TRSxp^k=t}W_;lV(Q;+3}IZ+1CK)d3`~ZQ*Hm3u!
z|JY3iW)!~Jy|9z;nC8xVV-4r{@@4+gGM&Ug`o*o&@?G$|=QUSLth>o%?I*Xcq;?Xr
z>iS-CtCNh(nl8F18t&}^vhT*qb&>Z4H8+cRyNU8b>qfEXon-CWH=Q@wog}_{Udh^`
zPO`}Kk-wc{7x^e*?Xb4Hi|h>#)u~j6^S{mFvQK7X@9Zo;|!9XDf)Z8SHDH
zO}C67kF)8M-sn>}-{sCE>p6CGk=ag-&BJ^>#BPqgh>#}W2ge=Zr=j15(#b!=5udw#
z(~5+A<;ju{m&{;2s}(OPeh=$@%j>Jcdtp2=3*(|NkI^6YthpWdxwqSYNQ0iR$@N=$
zxSvhjy3}F69nTmZ!oTC7`Z6yo81lD?Ws%z~Al|t5QQH0z#2<&h79IZ!dZjIUZ(*KA
zux87&k;(A)D*48Ae24wU(pLP<1LyYgOod==zzemrIA?x<-cKjfqa`r@%0Y*5Z4d{v
zEja#m80?*=zv0Ug&||ko9L2cE;cJ>}8l0DL1Cu7-0XufzeJO1U^+zQWPJ9D9&kj7d
zC_#P#$zMObHfVs^kr)AFGjsqRr+_+q)whPyaalut#5L66;i0e;8{@5jbgjcNdm^0)e(26`X%
z9oKsWeBNRk(~qF%T9{Ia=jwfp16KHV;dt*ge8l+IV!N2*4WJuSr8#(ROqAXRe_<2+
zlkqoy^zYecr^xdHU%f@J`8eu%yqkgNnAfGsC3t=;&Tb!r{(bI&S@1v!;(VKD>->R#
zF@EPhJb#i$_~8G$2J(G-_Fu#Ia2{7n;fLk$u2Wf0X)Dl#$1mD2t}dT#D2(F}gg@ws
zzgrm|c^`u3MTEX}peIn%`rtWuPNn|wq`zCNJu1Bp^?uA~*o)^>p`|80FSBe4>3M0d
z)O`*8x>4pC`a7C?H|$mMy!72?4nLd+e>j*#^CRX0LHAHkt^F_6+n1wB|DMY9lch9o
zV)X~if%20*hR>0HZ&?L@NeJE>N@;tbeJB41QCA*UhwF{AucoF18VhYh}f
zH^rAXqMVqQV>Z${%LYvxPu0~CG;CiYFg6hN<-^E6Gt`$)?s{KCefe_JerlW&9o{$K
z|9ZOLP6XwwUd4W{0=~j6Q7{YW*`esuh%4Letry|*I|sxBqH9A2ACJ0IptnY27Mb8
z-4)b&V~iZR{i5|M%nybql%M_pyWx7{aG4X#OM8XQRCwUNc82Aji^0xeXFIZd9C)Yl
zk8Gd84zcsSd6gaP9b%FphV%FD+v-W5&yA6b{9}(K;J!LNmm_&WpKrI){=>-)^Vc<-
zYfC_%i*0)@kn&=T{FRtGZVUbR^Nw-T3hWQRsHiQg03WFD$vYYW^X_=iy#0UwjFB~_
zoO|s%#)yo;@$HSOFdr{I_h#)MCXx8PTk_QD3DOhkd3aS0lW17;cs4=2h3IDUtxFAI
zo-pd2W6Cu_#^>_#I_v?xdM@qKks>C!`L%qLU_aP*@r%nBn1Y|}-e>1IprNar=fZf=)CagpZ)}hjL--?6Z!(dqH`%{?-a%`KFm4POcj4L~@
zv4wR9asPIGeu$HJJbUfnnS3U>TfRYb75@auY_c!!R$>wlrsr0j+F`P=bi*l~(P82x
zP}Z3GXqY&KWV~G>GeY`)+~eH<`tn`-+S(k5`&p=3kj}3Fap$vF4EA&m6VWZ7{#Jrs
zol<6sPxg(F!oxCm{;G|Tp%u$rGp>)2nb;-g3PC?hS$@r1I5tAo#f;vw
z4<8}DKQqf;f!@|N;Ss4<1pPeo?Z&(fP%e1>p|CWayn1{qinDx#WGUoySgfWKz3+!?
z??ZgiNedNaD}xd8di1sZekVGy^Je=h4El0BwW9DI=*v~_8Zeb#E6Xu%J(EI9u`(IDUios|1y
zu@E}P7;%C7nlYDsI}7WJHcs)5
zV9(nhD#~?(JU)F5X{K!!bfOr%aaiHOC^1`GXv=;N=5N3D!e;m%3qzDO3INAGSTY_oK^<8B@@T$+(FKyNoslnv^T$ZU$-~3=a6at_zB4Q>W}_j1#-1=tK)E=IOomc
zhc@utY#Di>_8s(o&ewzco`Alt6`9Y@jo;zN;>2_CU5D;u`;rgyOtxB*-DCJZ&ewJ8
z>4Dx}#D7_#48FsC=CPE2wv;1V^8?7|8kQ?yzMcCSU~!94>^}oN`zP21
z<-P^lsG`1zTjY8S<$|U1gHHiHkTXz>>l?-o0_!lYKsw9f3%2)m#OD#}jqVj|Ln47L
z+Yqt<%e{L#_MqOXR!SQ{UZuB_3-#iqMOKu)RL&)oJ)Ne~OzF$}YfrVHp0wNBbqMJ^
zsVmgFN4OTGGX~!9JisZqepN{uwO{pG!M7FY_it<4Ma7pit-}N$opa&256DHD!1FhlvlFRoO7x}ZIjwd@%{~r(T2u6NYt%?%v|JAD9AAoYHaj@J2{KWo~)VhlRbFV4B
z|H%hlKw>3WFYsu-SOz@z!l5YCzuUH7H1&b^(5sx+81ojm7JWTj1NzcjzL_^159uQl%al6<==KeXm>^A
zle$LuPLnLt%dww6QAZEqIGoq7{)+2;*K`K?R|0jYJOX@=$zRP$;LZMauWko@Su@dbcmu@Eec)G&+z9i-(4(Kv_QQN~|Jb*b
zH=u{a^r~;p2YdLVTdMC6*b8qqR@`)l>rbhbuD>e-_L1rnWhU55+g-9Wu0i?gjgfzj
zg1$cD>vD5Fv@6Ox_pThwyKK@q2>K=EPcBnShv&LK*JgPn%x8JImf7zBFWlg(f6&?v`8S&H_tcD!
zksZEv`cI&K>}Hkqj|3*EkNBR|y>WtUn8+_WXvQSbmuAji)q?e9?~tSYcCc^7C|++`
zpGlm0{W6=3C&=*&JL9&zXAs`As-33*ubBBP*SZbs7BP*HZM$pX`_=!HnfH8x)SWvU
z>R`hp@|Nq9$5z63a;kCoImBVw7}T$y;|sWj`I7zR1}5o?oA|4+aFVQ;>7HW3|2%%e
zOmh?XpNA_fjN3wBp6*^NtltEF)cbQa*TenYmPKoJbu!7)BT7vX!%X7-`@rAak4%Uw
zc1bOW2HMbkPV*#_jP4M%jl2WvDh>0l(YwK}mzK)*fctqh(CT@tf0#uPltF2w{tVw>EuuHjmj7yI?0VV!pX))C-yNGR}PI{!WWdq8#+I%@;xYujQA&sU9JU>H;c~
zAtR*aWBK)I=$BuS0>gg~ooxKhC2kM#Ar7y9fAa>sqP*P0M)$-Bd3=4|*QHNJNWOh)
z_9_)R@$^wXRk|GFlDR{k-q0T*`i~Z*^*YkYw{Ppa9zs8?UTk>!9GZq!4Dr-v6@zIBUssJ1jgp@?OuvtjA)(0of{`P{R9$w-
z1I-b#BhKJ`<576t`SsiP#f%YaBepf-!Jy~_;4bv0uIb;>Zl&
z{iyeq(eF?1PA0HfJyFtJ9XbCK3a9+xzyKno(F64JuveSo&p*tW))VX#n4;
z+>Jx41puFwth&4X8{m(3OHI4qf_Y!DF?6JYCHy~eO%##)X
z{jn94_C?$kY5z|}2KqC8QvA#_kjpEnnCAlghy5Q?(O+{>%TZAtX!*o_EjcJR<}7kV
z`;XV{_e>77XBE)XBn9^$cCV0g2f6wU(-Mf69`#j1;&a$v_WYS*A<&#{(^-5V$8Uey
z4e{5;y!;8oLo3CE9@GMFS8(aXG|(+2(H7EBzU|lsQH+ba*dx=3>lxk|wztqO-aD_<
zAN9#~dC!&UK&M2rHlrSD+8;F73iT1G^f1&bXSaXRKtIO&s|OFqK)FRfZx~*G-|C0~
z)~|O;-imn0@X``E@(1wTm0vt)f7>f>WXJ#K-!{`#=wHcku;N7hUL?P(2=#3x+fpiz
zNRF(>EaIj9)Syr3FF1BzP6YMxk{kWz=x?~M-ED(-p|pUC51F35!-aC)J;FLj_n*B=;h(8KfM3vl>e)!D
zT@y=5O8)+#V-&7RQDNnsP!54~0w9++BuB|9Z3edpw3E6wpNe;uecMje3-hW{<>{fD
zDPFVn71fTIHY^K)Jp0X}Y^t5>$L=8A5^6!=n{S_HU*k9%+~TQ6xl0L?a3jb!^tHzs
z
zBj00>2=Hg}xA{rKcluGO@f^N~29oW4_#TvaE_j9Q%kO-y-wynT)X+b7fewVq#$Z1`
zj+RpET5s?E^vC`^6>9{xX%*(TRYvM+~D&xWe44zH}~ow$oK0e
z9?b)s7Q}eE8v8AI<63eXP_?fItMEH^zT&fEE42UdXNMCVsB^C0CtPpkc>7km0OX9;
zESfO@-tSPv?;5DD{g*HG59ZwmgA%)%pnWHVv$Ye&tvD
znoBUR9tn$aj)ZxilUBT35&T77J^bm}faj_vYVU|pxG&?vP8hf4+fQ3Ag?ZmL?ASJS
zu;1%1*O(5hf;b?tN%|FR@Qs~IqVZ`tN3QwBKoukFV-^WZ+x<>4o2YA`Ojuj^gG
z&(aghE1@DdK|I}>_2H;F9|Kl=sexXBnhcrWxK;+-UDrBmYhA~O4l>+ZI
zE+*NiRAHJ5euONpHjj8)&=B)RQ51daivITWyGe#`
z;r&ROp3Q2m1^xZc$MF_&-5KOb
z=~CG^!X%;%{4E|r&|gUv_qVMKqW31xU*!UWL~#rpV8D1}ou9wx=3XXon7#Gm&Jrg1
zV>nr&2y$lr*K|k}1NM`9sr$<_$Yp~d{~5r6qEjn!cgi!#x8gey@4FbJ@0_T|W{3~k
z=dWS0g%|Loh~l$Y9*B>)v`jq@;&|OlPs$8Gqmyj=Zu6|0baLq8_?s@+zt*o|w(_Ov
zD6wo@Pt%6H0`VVBS94F(NxJWc$)I<1VxOV<&a8w^gc@x7U0#lowDXx4;{l&|mghLM
z+mDhXoQjGgkdL7?W6s-e?4xAM4G%*}%`sxX^Lxh--xzuC_?7!{?I;O)Hh)LA8^j0i
zD$o;>2R_ALr!H`eR5ox+^z0ZT%kJgLR{tI)4>dMMUXL6lzaC2}F47$(-WT(a-n|a}
z;r^R9g6mzj`5(jNcm{Y{
zciYGZ+0~;&I=eb)+bEqBKjKXoy#P47E+W*~VwCXzTF`Y9@>kF~yB)0bVZE$pUqjNO
zQL;c?dpun{7%s9Yc?euOamJ-XLwjuNlUvrppmV7{;$Px=9P%4wGE
z76;%?`@Yl-nnI(boo}70Dda^H{5T=10(kRPK;`%nz?ExbeyuVUgnVgFSNVqh93%Cs
zGEY5#c}ODi$k21}^WNmX%JUND&+ihdE*se(ugoVt|4+r>r>UzsWd?R4?}p+(=(l%M
zVclN-uCug7?ebZ?#T<2BMn!9cq{9{vhH>Rh-yv8NF
zL&+Kacp35{xUbF5lU>IX)~SO-*D&2-{`l1(aifDGF{W?PO+MzyK*Iv-K8R~N_omQisjrn!Ra|`s-Dbd%`6ZrbiQ`>PpEvlkh@DuP_
z(BwyW#%du>P>NLeGy#uIIz1i0`d1Lt`-L3ke$C$1+kr|$$ziWFShBv
zs~Y+r$NfyMbOEihy|oH)XT62>FzVsY)-(NxkL`kQ=V3f)!HmW`tRGnJppW|9=f0v7
z;+#&;_mH>>*7*l}i!rXe=mdu%;?BI-8`P$56
z^oM;mU`@r7y5v*EnD;mQTF${(=46rB0{(QB86*3g6LZ4^jKCCNAYr
zIPkJ44-MtjO#`X@Z13ae&0stI<%tx|ioGD$iTp+`7M?8pt3%<&&(iV~F7umW`TKSr
zn!|zZ*eCZ>Qv2hAR|2dCUfYBv|JMZEj=;~{hp-}$KPX{H(YcV71Iuk)pm`{7l?T`jyy9(u
zl|ZHT0_=qHlal8s>V1%h!joZLdny0u<*fVZC>IoW5sst>+;tt&UV(he@?8|o%VXs;
zn~n6QaIJD>%MFyTI1gzHf&O03!qb^o2I0o=ozn5(L|6yTTTJbXyCSufYG?B;3kcqGvLmQTq`kU8)?ho+cas
zCoDnyZ<1U;zBh~Bs>8|UfP-aieoF#%{Bx6vC*xMRMEN`WjMxpZz28}hjwrt(X9t|z
z2>#TZZ~R!Ep&$u=H0%dSc|-Mk1>X`Xj$O%^EeGFUm97Y?op*W?l%11(r}s6=8*IC%
ziPxVo8G^(YkeA@XcIx?uWDZf|A>KAcy&nS72dTUdM;&Lcqx^&1j|wa7R?QV_+4u@M2j0j`5^CuUwHrdTmMk;a=X=QsCZ;I73We^q?9{RNnRl{!r`
z^k5w(qWH7@^>2^|E@5>f;_Cc;AH4Ct*OQDC(Jy?#gYE1yXg`ss&Jg1o3TcYkp-|r9
zaCaY$+j=j##}9zNw24%#0D7Bu*BmCu-yIMxLHjrCj9D_qOLSBQb6W#nbzdq%2Ku={
zYj~Xy`bCPo4g5jBxG(v<>zp+BDVK-&8p1l*m%FZI>;g=3j8lSX=CfL_$T
zSQaV?`=NeC8?U>b3w9XKAN3pBux{U((ET)&LAKu6ST6}U+;QuL_B_zDZl`vXxr3j0
zz3izX-ExRiX?w?441Uco?RSka&f-v4mXipK$N0-f>)Ajr%G%^Ld&7EO``L82aKQVD
zY{Q8S&`;GvLQ0tp~f2p{FjI$s}i&l~qlfGD&7o
zWzGse1|iPXBe(P*Z?B?|jQe~B2~%1uJ7&fps~3B6-cn`~Yvq07=N2GFX
zj`m)6+=f9?PMN0K7QlP%!;vbkjDCmX-S(Rxf5ezHQ#=9i>^9kvGA{=C=9*G#4EEGX
zo&KJy%+tM{IE73`hk
z*{5iD9P-=5rs;_ggCt7?$>;(7U1olQ>lxVH9>-k$4&u6h^yNNvz!x9d
zm-6T`NWSUrXf9r
zP~cVwe(mC^`t$Sl!TMyGb66GRTf4@mp(pG&Mzj(gPPgj9JhH2xup0b>%r|H;j=#Lks*?LOa12-w|P=QnXG0^R-F
zB@ytS&d!+n$FN^vUH6*{z3<@v{Q7{zoPqUy^99dpcaD*SItDvUW{m7-e<;88$tc-7
zyDcTxX_P#=r6{l(;@FQ|y*u(4{Hj-@E|0#2JZNLFzscJ$$Oku&D^z*_><)nzgG3S7
z4<=-G_jKDB>7bo)_jem3@BUu%9rqX`Jv?{(F6E7pZq>(H4d)@=#V{?*+Y{oe)pkGI
z3$#Rg;IdH(>sD}Z?E(b-Y2~u&mW0HJF7TrdDCIO42~MF^Md*Gp-YsRE?jq!
zE>aK;^VHi@#etbn-*Ah`9tbpZm+!nLm{(t@-<0=({(ey3Tq+9luEdC{vo~CSdS>1y
z%s*+7cFgA$(1m*2&Dy|zzn`ogaSnJsjV}UUKrSF$(^nk$jSu<)UIN{ELhhsn&_!%h
z0d-Ix`E1iV)c3Xz{_W9$_G<;sck2S5s<*D`7SubPiSNaA!sK(yq~o!E=cf`ms8^J^
z&4c<&O%@_-p}n&A$M%OnefSTBP{a*~BZ9)mfnP2otM(PB(YtCp%*%1PenmR^iRJoB
zN-^H{X|)F(*I^I7KF*Ce!s^ezqfG3d21CLEc#W=0iKwr=mO89NJ3#JW^c&Peb8lYp
zJqXl#DQyql&wBnBevD&#Sko+p&vT(g`7!EyMUm{QsK+yBs}G?57CXLp6Y>N3*7iuN
z#ppWp2fF*M5XSx=*}wk@;)=Khi`Qd)&vK(D*dF)(ccWO}
zY-LTY&!_CLr}j7UJFXAFdTW~qb!@lf?5-7fy}2$YmA9>OOV2i}r~Tul;^rf}4pVrq
zy=ehIwtrYC@DSEF7jAlr*PHvY@^Nia9ii}Q6cGtQxl7^BZ?Jyr?X9&)6<X7=e~J7+?*Rk2>Z?i=NI{`nKk8z@f`-}@QMrzec5{Btx3@&Wn3f_M-q#Jy~4
zqHyvRvnUFuMm*D?>;k@>+bA5|Sr29p$RB8e?T;Xzo&aJ0K%0&NJObKKBlZL9`#y9~
z&v(V`HYIm$+@2b5?(a`qvA*gPjoMGS$!0yvPp-Ovs=u;soTB%mfPmc578d=rgoTT9
z?7FD(+*>Ta__!}L5BJ@n#F8trg6;sXafM|E6t1bF8
z0M~Vz@1*JlJKsY?kY}E~hg8U*Wl*1;%Cav^*GW*k-F;<>+OA;9
z{Z?m`4P_hEhdwoGdOhPhI>F!DAlv#5L;d!>I-
z&$WBrUohMte#Lnawg2|;`)gq6fj$dv{)_nh!}4ItPG}Quq~4#r$F0ibi^7E6u0v-3%6p4q~DN!fpD
z`uTrXgM8o};Z`?jXXpH!Zj5t&rYbyn1?Jyp(D=2U%CbkqJP5C?};h)Gcrir5BJlTJul0|
z^*KwkumBu4t(e8dI3G^Pt9zpSnfUeYxZbBxx-lJ}DTz}eJzys&5&6hU;Jv!w2=&6+q=#TJ}ItP$9Z-99M
z^!D5}0p{(X?`GTfo6rCkaEj+F9tXXjmto_z3goW*TRq$d_xn}$@;$C!)c(#ZIR^6O
z`8@q+VSH4>Q=eu)J?A{-WYljm^4akLz^`;MDU1O6ef>1oV~|&hs??By`r_xb{pX?F
zXqVzO%mYxp>F>dX(2n_*QO7f2pERoo%AA9Ez#{=kU8j`+_m>~6<4A`%m&(Oak#5kB
z!|&gygT7b3)}ZzT{C>l96W>YjZ>DF;yG{b`OcfJmR0H1RPketpu^QHyb@oZ?LiVe``dJOjX4_D41gmeas*
zSs7LJ656%?GDq_T;H@)r)o1WL8j%{odbC#!xDy;+!v9sEm#DJ>;|l{$KTm@Hc{jV?
z2DDq#?zss|fqeAe>^mO-ztWb}->?II{N31GYj|G1-5l7r3rczpnvs
zgjIGJ=F7O{EGk+Ebg_biAnrq?Z#rmg4gc#S1yhqL7`O0e>WvsrU?_1()fxT|?uYCC
zB|~|ARiK0_Jg-i5{{}qY@x!P57h9p9E5;Ynr=k4o^Wh!I(61eC?`GYrX*d<4%0nPTm1|;YSM&?}K(WDZe`w3wWEes=xC9yf3mB96mgQ
z@l?&()!35~5XYM+P1K)MJr(<4o)jyv<1p*bh
zm%eKu*!8V8GY>CAf4ze)RV@X%o|5%!(h#q``*xIw8+;de)}k8JjsnKaWQ^@4g29qJjpU`4J$$_au*&Ez}?6=zkat
zyyYqFu5&;S6uYT+LAmC(KH(DJdw2%F)&
z+IkW3_{7#<)?q*uN*XVtAEr&WDliYI^qQD0>d;=@An&_SY*&!uz6X>`xJ3L!|N3A?
zOM@ts9~9GggnDESqxKlu4e|Y1y12fqc=hEX#H%wmqAM|eRI?x`kO5RW`lp{Pl*e$D
z>goZlueE8y`PV;W;bWZt<7MteAr2nT3Qjo-ywJ$)F*D$QO-2u*9P`HY-o9H|$MIK>YvFJ@zTi|8|=zS7Lm}znjf67%x(%@#h}u
zneAP1l)w4IK&&X@?u${8L%1IQIp`w*8u*^&yienOXC7BnqaUu+O*a+!FG>3}(O&V7
zze1gF8OQAg3K9ID7cGuqeNE@@0i^Smm48CJm4sMk0$_J^XJR(_=y@!W6rR>96Q!;z`<+L{S;UkzQn>w7my{@fOfbYZK!shq_O<}C6!J8FhmKP1XIs3W_;{r#F_eEm7I@qLCK8ga#89m>W9l4P=1Pz4+Q(^R55@2+B%?@VFIRhhE*nCmP#ej#|>K;Bme2NwJmR!gb-Ecs?f@m?|d)O$X){3{*n
z+r%L(2k6%Igp%v*k*E05jgYP$`uB+E4JH5hD44~-k3Afq>W{5p*$;~z!SWki|KZ(s
ziXJ?|vg4*S@xftU*+yVD5likIN8Mo+rb457*sbjq_p}
z>%91lNgt{8pMi*3%8pvUR{t%|^DmRIi*8=H>u~*muB&U_qjh|`#B;O)H=n(=N+9mzYeE*d`CM+
zYJKN!LnY$y3)$N6kAr_Uys#Df=X0r@iq~79Wk>B#=z4E#iFUxYBzY06m&^YckL|xS
z@o_@FyrGM_K2!Ma5VoJN3DgCYzu|vHUaPWZBsEFW8Qf
ziWB8Oag1%C=Gp$&b11*a1?3L-Ltx$(xh;(A(w96pO`{!k&gihTH}G3O3>9PE?XnP!
z1?X2StCOLj-V~M1-iG?c+bZ=h>aV@QRs`3*t1s5Er$V{s?dnkM=b1d~eOTYqCg_Ix
zuEtU-3H9Y+n+)c9py_P2l>cmDc;*>gPkN9ro{RFvf~}`fPrsHFo`dlUa_Vd=sQ04m
z-+LTqQ&+wVs0Z6Tj;uia96)NEuwV1sw$Lm9A8qw6m@tR>hEG4w;Q1``26<_Vq5lSt
zZ}4~kZdp6s(ueqC`RqbhZFoMm_tHly;rb6QSdzP&M
z`YrQEtR7r99%v_qI7sV{OW`e`vWZ2c70UTIg;h}hZ>x{1!hH*Vc~7jO`I4LWpMvry#ziWK
z{}hkSfs?(!FO=!_`2f&Ek$-X!H*9p8D}w!4ZPtFs5$@j_HkJ|$G;qXaKl%+5$Bw#T
z{{)inbtAn<6ME8bhnuL6~La`{<0$ZOmvQYZ#`
zEwg7e`U@CWe>O)!zwh?P%3*wkvx2iw81(bs=P0>tz^g{y-1rEtuYY;2-45i_Uk0gN
z2Fk^l?ZD>`G`^^e@7Jlfrr^MVb<|t=srbEVnR3bCcch}<_X@|K@BH51`2E@O8pR^+
z`LJ_k?rXS?Bhz&+j{oQarG{jn+I6cF9|CpD6~BP*Der|i6ZF@bnB5c?20GgQ@E_)<
z@7XHi(+o5{_`dLBz_Tmk46>$y_HxddIDve_tK}HS=;r-)A=(A}dwS=51-h+9seTtw
z!G}cxxF7e5XW=I`kU^NLSV1d9j~EXPq(b
z(>LqlEY1^}tw;6IUr=}CR}$iRNOSFr_c^;TISap+ID?l{=oitxuDuKW@L?W%rj|pw
z$;MhoeD1!}s;3cOM+`i(#P?l;bA3O47w^Rbd2qhk!+4yAe)F*0@6^7goy-6|#KE6$
zxt60}=DK<_H`+Cky|zzHfG&MLaUA{Rb~dLC5w|8hIn06c!h#eHD_kEJ5IF`1nuEQp
zn{9{o!Bf@IN_^it(_7Ete6reiYcuv|%2;p{?)R~|ahi(1ep7bN68n?0ueS|x`7z!;
zIIs}>AnRk}vA_Km``2OpSEuf1q%BK!QuD7cM~yVjN2*b2XYsz5wydG@?w-+arTp#)QRm;NMv1axKQap6mty~gw(g+LDXe+4o$?zMNe8iE{qY_{KJ3Sim_MNK
zVBN@OAwVUFPnPimKnMEw{DB)-pOQW;i{)JH9H5}VZ_$_mfD-oCofHDV0d~Wp!YZUk
z2PCO}WJlZ7V*Q0nODR1u#aqLP_1hB;Qh1_rh~qrgzx9Fb_E6r_zKN>;-g^Q(#NZd3
z;s?Y6@56F2HRP-3&ZFv|+rOc3wGu}R00dYUx>~vwDeuAW6#ulCwMQ^vv{E_#nCJ^fav0t#TgwIT%o68H)3=u&!7VCNNz
zPO(VcWa6mnkCm>W>K9J2@N(5C8#t`s
zas@2BQU=-dkUt|hmulxqP!ZLR=6~1UPAQ`B!m|IJ%cB3Bs*f!zpvv7xV3?u(%HJ&f
zrn9n(;+HlTQ|*;6W#OphOVE)HcJ&q(ZvVX(x&i#Pj|ddOuT%%-7*P3a7H%uWW})3v
z)hyh8SN=9t@04pp(ay6?RDE0{{4H=@$pZLWfG*N4&t3Q$1ESx-oa1pK-GhnSpe0|6IpT739>i41>8tx|xRx47i5AgQ_
zy|IN=-z&ztzWp4le*J&#VJ)^BiNzofhiy|l1*3sqr@|`7RREMjs|o4BPA$vOX5j%XL<{ID3Fhk<_&(86q1Ic1JOz1n~BU;i7gbJPRDJR-o7zXMsl
zv7M=C*87oG#p?GGIHd;5vs!OZ?Y5a?Ly#Anu*$6=I|%aQ<}7<4zxp#JkEc`Ob&LKR
zpI5#tev*UrerejF0PH`)&j;v5G*6(N18-Tp6Re_O`Or_6yp|@b-Xe-sUwMedFG^&!
z6ONxRTz{|=1t8ui3ET?6ztjP@2k;#N5!8J0hRIF&86H@%o=^KX8nO`YC;-zv$d%@v
zfCq$g;2Ue<#_&CbqG1YgnQAnue%lt79mU_}zVm9dWN
zsQSWXtaZOqBi8d=@nlh>r>uS~Qv$dT?eNUtgTX%0c`I76KNfshR6hbmzf*pao9zel
zQ15Ujo#DdzyWT2PedRqFXb9%#H!9S;>^3zQCLF+xVp(rce!kQ*s@*uZ5sEsy;7TCG
zBb_m#=Hnpk7)iXZoV^}
zIH_?vdht9}zB;s+n)kbN{MI6#e?QmrGukchcin16{BYqKy9MH~%Z2t=5bs#-PI`^~
z42gFL#(wNptT0F3G+$~z@+84C5cTuCtI;qsK)m#xqe<8gq2IOJk(aL|
z*k94E)-Q-(2F2SU(Jsu(U$>_r?zENLz=8VQT&q_H+wXb3SQK$f;4{(rI6f!-yfwo(
z-=y!q>#<(eW_%7lpVJz*6ugePNyQn*^Q4YUG1lv7)>3w`%+qU6A_XCZ6W~PpI6$$r?I0V$SuZkb-hyFal^MXLN
zjHRdXKF?>)u%SP8z2A^D4Jf--kTLc*>E3oH9Iq4eD+=(r3krfBARc^KG2MuG>3~Wd
zm48>FD{cz^AEiK^9>j;=*c0a>-d~!(SCE%TwWjtbgRAO-XSQj>1c95d@XC*
zGlt_gsblsV`!&H{5Q1^lleX*Hk$3!PG>HFG?73I(@p+>?Ru16v)u)I(zMc7(&h7?2my;4p=M>eqay;vu57
zE5Km`xG<~l<|?3tmfB|&fTow5cA`C+zo@PR$1%-Ea6i63$|(mca6G#_hw~6GgZFwA
zub(U^%S5{-%IBUdQohG;YO$ZOA#XU2O^{{${SA`dK%%4?I-h7k4eB
z@=htIoP2}47xxf_M;|?mrM@p(^-5{Hk4~iSE~GIL!*B3;b;8*w-2ML9%`$vGU%SF>
z_&!)nFS>~FGp{(iui!ZPY2<(d7RE#9Jas{rL4by^+4u+rWYM33grPN{#E=aLTVV(|WZ6`(wPRYax~|Ecmw>
z+Y#eEl7@777bj&$j_frF#QN#U7RsO99&$O1x}QaU3|{{_xLOD4uKcvi_#HB0Hih8-
z%PVwC60h6C-Es%zS5!Oz1q16U`xtE4&wDSM9^(77Khp{@2;}c`0oV?k1%KgfI
zsDS*5u9Bk^a2ap4-d!DKQM=<2sQXdBZ2yuEm@cBUb
zmReqFzCN#74T=i>*OfcVkq-FYq~`DN9Aj!8-u^h4de6L?f3C#(_e<)7k(SxK0mTLJ
zXZ`Y^xZ!<%&_Lnlbn(pQP-QPNm6pkS}eUe&~vp5<#>B6DY*kJdK6vVkPnI<{4v{HsOv7)tfb2M#aMnn
z$E`k8J37tipap%934agt;{nGlieA!X;euOR5$S>dTNjoypuW1Fh08~ZSbjt`W0+Q<
zyyOlGf7@+k;T+uz7B0}PFrw*ydY~Df9Syxa`
zaNoxZ;O~d>Fnt#7%+RIyKmYlKd-z$n!m63o-Zf7Ia1hUQ8Q=j>@e~+lpdUZN<3l^z
z7_JXIW&=Ummda{(>thz)XzpXx?>Y|RAb$b0CsO|VtbY3ahs$~^S=Zk%WYr%%#Nvna
zSPzpIj2{Mp?sIfr~!JCMAClFPDZm5=GOXeACm
zTwj2X3;lY(fJN!>_9EZ#kX5gQ-36Z79s$%5%^SGxo&t-8jj^a1d_5?KRwUMEV|Re&
z$zOf=xr#|1Uv_=>{569-{rV*#TZKuSb2b^Djf8v{SA3tpf_;OQOa8rno5UpjGA+tl
z227GT_lSENJCp3$Y8`Ze!65Sgu5@ny${=HH&Nl>Wfa@-LB3c5*@X^r!5)eWooo`ijvUdoHroukC#v8Nd5bs-4;YF8{hCBq52a`1w8D!@D>6!y_10?$QC;p2K
zaK2ssv3l9_kU#0!+Q~Nx;8$RGw!5>GL3UUKd0q7TMtHA$y3V6PBQF{@_!`1`dv)%t
z2?xY;FB6RBNq{(P$M;_(dDjk-rSGzy7Y2@yu2S9TeI{c>%{pEC0>tZf5{7H~BbXPY
zzZ|V;fIQSw0>6e;M&Vqm={4??qePZ|nVZH9`pWB7aslK`NmLXR2w6!dmmB}aWE+f<
z;u*aic92(LM}U9J<+o$Rhh}3`H#kOa6>-S41k#CX$)<;q&ql~0W$~iQOvo=G+7ZTQ
zH9{7g`}BnQYMA)hq(wCGjS}&Zx?AsDAn!!_U#D2uhZmaQP#yyLztWBNmM#1{OzeHO
zSw_4XBo+^H0&lR>Nbvl0NzvbMj&z~-*UHbsB#%~eq8svS9#34(IlHc#9Qmd#b!}BY
z8TW2^HNZ7Y+`qZy7TXV#m=>R#`=1Y!&BQsS!nlX@%#Z6Td^|vmqqn={Lw?ZJwP$+w
zXb+RZGgX(l>q%Z9syA|l~@zK5$Ne|Qg*jVS}h7vWr*9*^1db8s%;Yxzss
zoij8-li}P}2l>Ps{N{#hWzvYUhW?345dZD*nPCab&*aVuSYid5@nIhjVj6tq0~k
zw;dvyhsE0Oz8WH;s+!6p95iB;ZrJ(PaEN$rUbrPrdx*@-+Fn!aJVaikZSyPY7$R({
zUQ5<;)5sjLj&W@V$TNRkbmvm&Z}Iw6B|peNZ$2w~l5+u#7Rj{it?RUg8jml-*xTA0Hy^l3o&b|3dk~80GYHgG3}VzE?PVkk}R)eQ5eU
zNEU5zjpl@YhHbm>#P}HWtD`{7fDQ82@_(++lBSXJ(-nVO>}llt&Kr%*2{bZ${m<^l
zDKwH}!28jNi$)IJ(fb&^mPV#NcO>20M4(4G9Yh$
zw0TktyjL4N>AcNLh6x{oKPEhQh&1Yad)`?&M1EARDYk}we%i2AVy!KWtSVi8`2$cf
z!vNdF8#Hq2-Vu?*%`_70ePDJ(7tRmTwOy_Q?|+M-&w^9%UEH_-etUH#~*ni@Dq(`?hO4Bn-~#JuWS6>Y;X@sCV%^*%EUd84-~G=CT-t{h^B1JMi5OyG$cTmfc9o|3f2NRh5%_m0~pT$?4r~*Lf#7)n(A@T37-_ic|$+v)>TeeNRN=B#M6RTqejT8xQ{*_-$sbT
zplZ=^Lki@sAL_J}#Lh<|E{UN!8gGl_RA1*XG-XJaqEwzEtg1$OFVw97uBb
zF-%tKyWjW??TO6|qCNE+Azu>i>hFQ}4zJQQcnaqOIF(12F5EIqCgdF+Zi?2x}G
zTl9RmH0-BNl#E$D%tZ$~Ghyv^RXRzX&gA*)KquzO=Vh`^Lmnbtqom^Vbh7ySJzh2F
z=V6Wm(W9H_#G7+h7l#dctOnwIS)JC
zBR|4^aBHZS|kpC#4BV<)fDeNx~|8}w;@-cNCTKwD!_PrlcZPGdb{nU)w=O&j<
zC!e*N7pc5}yh*ux6WvGYgugm|cpfjD6Cva884r1$jtYxaOh6td-D_jKm3dIl!Ch_l
zkxn?y>g=gv(#c1kULP@aIuUpkAT?nL`KFwIJN@*glb(ZPXG|{9$sQwnw98UdfPRV`HWl`Pyh~a4
zU88gEj*x-BZ|~C}f8`~^uW5cFbP{rT$LmM@qonY5+W5fEQSzasKWz!*Z5om~WZb1T
zM$9sOou0$~sAZv|ZIe@UBF)vSdRcjt=;&8HHOn3)63^M3ZH32(f_8`c4~sGK)ijk^
z`2fy2vok;DGE9ea8FG3yn2(Z!*2*89Do4rH$KvbbATNP>t>~dfc#r11OmDp&4*#RU
zM}rS-kUy*ajOjzj6A-AH>9XB=lF@pDWD5J$K@+6&@h5Ynhd!J6k
zyIR6Oz&O?`+rJ<586)d|Xn*-yF-G!C)*IPa!+u(8{csJyk;c7WInw}VJysJJdSp39
zyvIJ7zDyp2eZcEujn;wR^r-1a6^L&XxkcY94*PgsG`XhCV;du#28ymJ%3~zNU8`?@
z$rw>wck|RnGqA(vs_jWh1Aoe3@d+(U$k&jS5i%GzN`8db)dg^kktd^u{(O*+eWUgW
z=RCko#s-6lu^;FpVk)dp6>yQf$%2LLupdivOC#5hnNcz-)U>M=#(QTPQ`-^tYw|bo
zi97%t`SG~I==S07GS|BD3@xL#YGU5~a?_!e;VIqnVC5|RuOYnM-snlebF{!fuL&J6O)RxNAUUIr;S
zPjmnC68vRF_r%=VVVyVHtaxxCgG@*D`qNi4h&$)))0}{_#jibl>v9nMG8dL}&71)o
zzC7)Y%wve>)`(l{JPzx~e~cz0*cjxqnKz9e)?M@C7Sx%EgWqzCZ-eMw@E>1vm-CK-
zbveTfrxlkWUY0v8&cla6JmcM57N;@D!8x7{I*@m7C|~L&`NSaJZne4Z7cj|#RuVmL
zKZ9`T`mh_EhrA)Kg~!aGK6<2e#z7GBt8|7ItQ2LEc;ip67J_|Y;1RrA&VoU1FSYZk
z&18`CGhe+m#u#K!{&h{|AcN#x46!`Uf$a;e+%3%@+;^TQWWadk2Wx2XJ%sbW&p*g2
ze#am*)5VVi;D2wvC!23p2>u7Hx$~WVz`Bgp*PRWj3^F>kHdn`mK^AzgT=@j@w)~PU
zX%U}=_vc>fpakU03*CLru?fyG6932drq`QEwh4&7ST4sT+P)IwE0)6fkPq)2xdHhD
z2NjzR3kEZZ>+ic$>&`Gqy8fzfYoNdSkzT3w(+t2*EA<|MU3M#I$~VS`Nw)eqT$zIN
z1FS~E*)!`IM6oRAdc_3TTa|MjZH4oc!uGf`e)}`YPbtOcaWPC%9rP>rxGmUU^<4**
z(-}mkFN^W*8-tiS#D*M%_W2sv9c1P(iQH6bYkDA)9Ldj^+jO2uE)D*0ac*W3)ptor
zYax#@2cPwZqc51G$Jp<lMDiB>GP;oauf5`2lb7dIq!s4Zl$y$7Dj@i#35k@l2wn^lDh^Hj^Bk
z_ik@NEt9ZYU2S_0c?(SjOBz2wKEesTvJ*zo&-`$ic_pxK*?(_B-z1F7!`n7nvko)K
z{6Owj-ee}Zm>1-GzlKR(h%Q(ZU&$op#o^T@Jxp?c{+|0I?2ret(`oF;4c84!E-*aRuD>$|0YYJyx`)XsnEJCh7-SKpxzd3_9O
z0`q*;Cdju{wM$*>Cy1Tbe4pJXA?kFu?3xR|nB@0awx^uZ6XfgwUqZXt1hJHy6Wvem#r{;AipvC9Sno163*%rX>d?G9VFL16zGXjMHbE8yc_v1^gZp$TWfnp|%>wIR
zk3&9e5_Vyg-i`@ku_#MiYX1cBso%i1-)4eT>2J^udow|j+nozcJ10m)-Dy+54#=O$
zdvTpR|0FT8jF$fNf9$<^IF?`ZH)@E46iSjLQ)FnMQie^Xgh+^@OsO;}Ny<<($rK40
znk7Uivq;F8@u8$N5GqoJqCuUt_WgX{>%GqP{&lW%&R;)&^l9zA_TFnxYkl^<@8@~o
z_qkTS+37sDZG@<~+D_l{X@vB~)@!ifcew`Jlw_8|?`|m?yxFH>H%d;--6--8+85xL
z@h^sY1XSiPisXax5zeXB@VjJ3xz0!h$&3=8Q`0!2nqfWu*!@JL9r~yH>@0lK3cTNy
zC#A^=zw@PVxzt-~l<>R`IP3@WsA~Vbm%9`5;)cyN^L5XqeN=QPOM
ze!5Ft7$HC7%hZB1Mo1Gs{|Nshn0Nl{lFptHa!k}ovqWf=sN86?d%0zl>{=#pD>rDA
z2+q%`{L4K`oVPqC-hQK`Uh&ItJJjd(Bf?g%Yn0^vYP|K&W0Xu@%ePuHd6e{?4P5fz
z&nP)LB0Sr(6@G8cS1v*a=4IxZ8uu&kd$q;~9bc@2avd-Fn{9?hiO8)F()4Zb
zX#({xGdWuI;qEB8GPF6%6MomN`9i08%Dqvtoa^HUgS=7Vx9Zw(sKh9Ftrewt7Jp~1
ze6zLvtWn}Qq;%QAYLrCYYJpfbN?J5`j%~_-^4To{(g9GeMz(qm^s8W_?hcF3BV?1X
z@-$Y}2wC5;*?o59C{gy7aW4oNC5y{=R{ethU0136)DP>xQtwW6?5h#-A&2FfRtEEs
z(`V^bFiHYmyq`Y~{XTiPjVn{OVEkOncnED;B2pWDIldD$?LUlAge-wEG;pt?rsG8;$EijN
z9C{4=uYY+t@VyTX4$kSE0-lcUu8vOoU5?l<{Rj^q=^rP@WO(iZfAk*&wmx2;!(Rd;
z{x9T)e9RG+CLHvN6k5~};L!L-?@c-2J9_wNJ{K1kt$>eRA2(-dEK}pV3R*Gy9}#vf
znD<%ahKnGFG|l0rH9=hm9S^v0h(agv9+|0ugToV^#nOvis{8-uKU{2kIXLjX_J99l
zTKPYZr@C}K+lc@F#`DGw91#wV|EJ@LWBK2B9)kb>e>0x{=kZzspDAJf{?GXuhZ@-V
z{{#Os0{=4t|1$#rGXnqT5ztb{KNeZe#>H%0$wtD)W&bysivL@=4eb2?%4xIZ|LR@E
zmgU*Fl#Q#|sLVzUHp;PaoGP-&UbZa6#)E9UO$}oaS+@LNfBtLdJUiW#R?H%bY+}%15T2=2S*L>%!LmIv@VSBJIA6T)u+AKz9H0
z-(tsEq;chXiq#d2tZ#%9#3Jh)8Cjy#h{|W*FzuXHLIYXETM{dR`Osj>do7zqjC$`*3BmNS8cLAd5sSK|@$%R4j_(_yg=Hi#-3WPw@obLW+;on0_wMm<|77
zk$aqHX}V4Y4lIkzbY}F=tK&qnNWhFcwB23W9Tb&^j?wh+eF_vgD>|VAEb=ai6AWOH
zrvneEUWu45t>3ChndWaR;DZ5Rkx54?VT-Ydhe;PLe>L|a%`bkokm~b^yr6Qj^6
z%;t?W{kJ%?E~0FH(|T`6${Aw*FEgHsrGPi2Gho12gmoheRw|2J;uD4mW03`)I%6^a
z(*BzPivdNd4vwHKY5xTVjz!3IU#e#jB{PKSbAy6kAwChkIur9>Z`Ij@sO6MKMJ#vQzps-EMZROYck!bkZ2$9K=XlItdjIxO
z%)fW^*RUGkVHp=4Wx&C+k6}PrL~X`^q$XfV%u55z7wZyMQ^E8bl~G#&_phB|je2oM
zo=9Uo{I+A0uwG%iMD-Db=nHynzR
zfJb{%Rj}L>z0jE+fc6GbvU-56Tc75PFJ+PY!F{pVosDkV3cZ%ygyWSa
zB4L1b&RjTSGwKcO9;A3uTH6-ox~JLS(e5hQT`RGFR{kw|T7Y4!KmoK*ant|4V
z#MjiKewEy(7DUs*PY_CAeP$N)VSg7{gdfNB{C3}
z+r4hpP3(uNQo#)DKhMwW5K>q~!@1H4+dHnjk?vcLjXDqtS!7u-Cu~4i_q7)wGO>uI
z(uH$~dLvbIAF76z%tN_mB1{#r#E8)g%jyJ&!6GZGGw8lto$#8@a{aee5^?Wmh&AjIgR<>@=w!o;Wg>9Tm|y9&}BQYf9pPLfdgZaf1+#Y
zI+C?7C`SEEaV2nwuzqSwsQ<8u^re2IM>2`}n}^mgO^`2^st*)}MVc?617(p)lLo4<
z@yVIy^K)OPxFgn13iEeodZ{7WN4}(f>a@iUTJPFim<~vcH$IPh=VT*MW^fwcV+PDcnZEb_4s+$CU))qILY;{YIY
z&Vc3tH~TU1WLm*pDobjE`+@X@V@L#G9mqkmfc-n5TEL6({j~k*$6*>kUZuz6Z^&li
zoNx1Tny#StkJ^#tcZTYlOg5nDvjRbca!X5L`T)Id@zZsY9s8Zi9wLrZ-}-?UmC015
zK9f02`>pSo@+Iu}d_n@$pq?L;0J6y1E$sflXk)r0JKqmlLD|fT!SdNmz1JF<`mAA<
zV16YBgR)~;bunv`$Y&Jkl$U($O~PW{`ogBHp=Yv-6|b=>(S<)Z@+WS2eqzTx`AV
z?D3g-7Yl%&#AyL^XV3pmwww>U71L)T0D{w^_@4}$kjeu%?T~*N0dN)@|I7cczanh=
zp0E*|4b=zdM^ToIci6azjd5)JuYVia_6xD?&0B>EEFwr34Der^9*{4v$Ke(w(y;$A
z8Pe<7B&L3Y2dvFlU7?F;2fB^eB&hOvk>x&cVvK@U_cG6vok<%2sJ
zTucXmMU;>}Li#%PItt##$d`dNVY=D`HpC+0>~$=$g~@-9ZVA#SU10R=HL;@*--;M{
z4jmX4ajj%z&Q3;OOP|r>u*MFu;QN1>{KQ5ke~4YL$^_%bY>IFIfH#QaM1tRoWc0Jl
z7~FVTho;-`;J`rKtpq@HW7fZ;6O%7K866CZBs|3qvPg~t<4;$tXZ#xczC2C85Xl7<
zvxpIU9^xi&g|LWQE>q947mS{o-A~Y9k&bjGeJq3-r=^w*&u{L+9R~Z}_9HDH!(z&<
zvDTxq$ObWr4St^}#(8Q}4E3;~n6Q^Af9Fawl|Q~|fd*h+&p)K`L}HN1KV69{2jZrI
z6>Tp>-HPfp`DQ|g!LMfH2^{>U)+34;B5$eQznCE^7tg)|21A@xH-!O&II-n5?N{~b
z1e(7tIgg@C1c3pA{)+L#4g|k%0`v$_%sE}tT^BUgEchn240stk8;CxSJ;{?;%S}+1Ii*htCOi6Do$LP7~fwWGXf6>
zJp0moSn(_}@0}X0_ko}e>Az=W{icZ_9d|9UNUW}*B-Cb<&%)D|}%h*aRGr8?n+3F!lkaKM4MxpA2V
z=2wN+*C21!&AYV>SkxU8!)tdVIS&A=L+pbzSXT*
zFN5W^UwKL)in@-UG6y{T_>dizyXRh&is?PZ-}SJ6f6kmSUkj-EYmY1eye@TA8F?I!
z$n+c7FAoduA>^^gc<;%I11@=@{rEfZ?$sMTg3LfZ*WVwGkPWF12DFz&ouU4j2LQ?ceh34B($UEdhQ&Tk)6D
z9Dos9WH?R&3U%zB*a#@HV;^r6*iow?I*#XyLQ#&{Nnl5Ci15R&kZyeYl*z9pEHY!h
z@$%V_e!yzk?kGS9y;Wa_Aiq+0vs&vS@c(AP=7x|i5*}Bsr3rCmf%J0g#c&=n4E6Gc
zde{BwbV7cgb#<$bG~j2cv!d89YrXPtIv>Gtf3W@XB`aP?0P0oQKa~LtsSHuYbWKNV
zeN0dJ=3&kY=rS%KiR(a0?iA-X!1*$Iqc(u8*81CUy!AG1I(Y=_dFi)
zDZU?3QF7z*m4K3VSI(&drXSug7xz)@vXz>s|LpJSxmcb%IRC;1z=+s~e8zwaHZR?e
zAe47+n>G?UGKO;vBH^c00$uH6TyUec^2lC}f!+s08@kHFi3VJC}r79qLcr
zR1<;t`~C%>641{;#*Z65cc6Z3z|!6h?Uc`XErR)G7rels!npM>qvs`;zbVvTC(elK
z$MjcYgFJ}hKPvKfgK4^4*GU_!)o0}p=
zDJ{B+>o+{3cw`aahVrSNSgxeDo1VXtdM49*3(?l?!0f=jiD}e-_7|p6dz9BgvydL~
z%$DL3z6k2CxqpRFy_dv;rt`GJHUK@|C!!Ry|LmgYryn2B(|A$Wjm!$pyM0WY+R_tB
zW%J+HjWB&(!ZRA5zE;BZ49LIN@1i)#Z3@LL-Er+Wezz*V(0CwX;Y|HGN2?+A|2+2L
z6q}wh^_RhJzUbDdxZl{ae7rPviZo8rh~SBfzza&%m>&m%#D
zdEOyS$5BShjh1&$lcVWdjx+X7-vr5^_jn!vpq4v>``G86gXv5^$LWCs_O}D0zr&N!
zQ^{m-<{4)EcYeSF3HXB*O#NL88T|-cevsaBTPVyoc?Q!1s1LUoh=RZ-5HoR@0q0<~fWz1fpy?q7KTT&a52p#z
z7hv}Q+hCU<{-Xx~@GBvVJc`XiIs{valh!i0be!pDnK0A-=p82g!%s%8WS^IWg0*Qm
z|3p}(P){CvTn|5I^i{MN{Y}_?$QQ)!1KwiWU5m68`8o55Wsy`VH(3
z;PV7VZ>u6RFIU$w>4i8wke|RFzZhyT@RfzI4Iy0vw>_Y(C4(t={R?vQChR!y1}$dY
zag{Ua5$t>U*dL7kr7G+o+_!5n>$#rAq^D=%fWZA#7&G4ff0=b9km^m-^?8_e^R0xL
zzXRG#dnY(eX#UI5CTI}+tCKO61&5jan6*`x%EA6jdlO#{Q9Z+7%>HYB$ka0rq9c~)
z3=yVSaBMxr6r)v){8FFdcZl++mvk_OqJrZ#FdXLR*Apt=j>8iq^kZcum8}~;!9>70
zwQduvbchQIT2#(H)<*M}{AKLUY`za2hkbFY5F7~bO6PMl|B;Rl#j#B0eyPL#71a}q
zKWl(`vrTbfgFhN?p(tnMMeFBFR)Pf#=fZ_0U^w`pb0?@>jX`^8{_h8tFi{Xc{!OBB
zNp$t>3G}0NJ&fIXpD#kkA>R2(!9W7v()kV(0RFIEPZZ-yT)CSy$}&w>v|n{%DioV$
z4P|4x6QAl!tWUL6!vW<_k`fAt-DV-Qecuhz)V{9Kp;M@LI5Cgjzwa$@g%uBc#Z3by
z7RGh;!ZFN$ari3?Jn)F4y_*ortV3bPgC7tarsXmpu1-T)=%Sb@#<4J~hi+)+TUO2%
zEN|q-(TVxT_7DC=nftB5B;-4Hs$Z?c_*dw27aTM2h~Miep9pQ&;X@u|IB;tK`NQ3@
zc2%rbr@jWBw84F~sU)2zZ{5ZkIRkhk=e>Dm*sgcSr^(3g0|oy!AV2b%mp>bMp7!TTFKOT{ZjH9y!mPj_Aa^+0!i6WqUy;eav!I@~<&LKhft)ECYxfHB-|9JeBLA;a+R%^(s5;U4
zI~3A?p1Sh-0Pt$g!LVB7gKd`tral0C_1F4^G30+vmdfY?jI+HZ7{3zkcSlEmv!MPD
zda^v>eRL&u^1Ph)issk(6AJR^L#
zPzlhXHcGU&h4AX1Tz?<*U4`~11OTrzdgc3bAJlL9F{r~}1@PeZm>GYd{FW7i-+0u3
zhb5h}x&`@5o^|iFhx~Ops>w!B&Oq+Y>79T%rNFm!VBVUWC3BX69dSy}KP-fMF#c2p
zPRM^hciMSg%B4p0yO%cSd=|wB>LPspK>n+X&~Xk9yCFR9Pfw)rRfk
ztu4gt+VLq{pud;*`ab;W0sOE?rDQv_|IAZ~zhEA$*8-=jt0Dhr?z7Q((4Xp$Cuco_
z{$C%gvfT>l;l82w+@bvp|3xzo0X9}_a|ceoj|rqC@q}^^@#ExGVX)%
z5V$$ANe|}DCRCw45A=i-T(z~q9-&$8x@#f7X65PUn+;gR?pyVPAQ+Deljc1Q0K9)<
z(g{3|HXE*O;(>V+f7SV^2Fg#&Ja!oO{S)=@oo0X;g;Ixc-?^>U*r5REs{h1&6!g?t
z@^N@>nep=AFWkRZZtl8``$62zWiy`dwJc2Jc7UA*?&nr!gRIG~J>@UxuXS*}FAMSt
z-F<98!mfKV=DCa+SU-~c5KYzvk=p;yQ$dzs!fqBX|7$ECl
zUHe+`eUb#dT$4oM3CK_2QhO*2c&X>UC7#>9a2|jgG~lld{P4s8;x
zbOfpi{@UZ#Vbs5tWlQ6m>RVes%pnl`)f4AM;tpVX}liE$BSJUSIfy%7L`~;`9yJ2813Iz!_zc0Mca)cF^%!dON=g?U(QB0Kx;$lky(Zeizz}1M`IE
zsGq9oxx@C&fkc%5nN-3P0@&~2hCmF#ZyQOkL)<8LOdRp%TNr-87J5+(`(Z7FYmg5d
zhiL$G5MuHhXNgifeLSUU`dsym6x&T0`9UAUJH;)Hsh!!E|4@CFi!jxzQH1G)`a11r
zQGB)M4$T)6Ev5QpyWdl}RYQp8e=)D3_}>2ZbS$@jFCQI`ed~|Xe#}<+LFGLD^Hfjo
zFAN*#Nqa&yfE8FhU`+!6U}+d5UsTnj@)`DdO7IcW&x6kxz3wXv2EXE?dfYY4cx-sa
z^lvF2qhFlD=#5#IP(8;SSWeK6X*eDL;NFD`rV~&hnUVi(2%zi0<$Ebj&(mee
z9_}A*GIqt#2Q6?fV+qNS{vC${&_t2J^K3r;e1PeX>p>=+8(#wfJquN61~7aF!?(9Z
zG5H_a@42kZVC#D@`s)W_dx8E?87xo0i{VW97CerFynQy)-$zbp5cpp~M$aaSssFDt
z)9-ARO*H>NdNRed-ORoiox+?eXU|_r(|z3-{;iao2@Sx#F&{HNd*JCf%1J+g#sYqt
zKA+;c4>45lCXu3Yl^ny5C!!hoA+Iq_Uv^2DuA>*GTc~{D10KM@pTR8%qW<4OifO*_
zbRS+{#_Utiul2Nkm#u=(afnOD)akm?=T3nG2HbzPJAh%}|L=*={8x4+RDXLUP88gC
zijUF#{kt8L&P+hbJWhZS~*%T@0rh3{&&EdwrhOU
z={N4zxfgc9#)9^5N>Sb}<`|-e@($@(7#LVrZtAxX`6>sgA9z=hVurH6`4(7Mz`thi
zx`8<3lrlJ4u>WDAHOd!XsMC7ggTGS0YxY&j3)7{!>}F%TMP*&oZ@Njop#HYay$?1z
zyl-)qXFB5Z(LK~3s^#q-M18U6lCGHU)y>I;c=-49W=#J)`4+wRb1iX$9R=}rKn4~%
z#I*(eaWucnE*j&(!7qxM=ucnpW$nfMmD^8$z__)2Pj5TM8HbBwUKsB*SmBjFA-?GP
zuU?MwS%*A1EEiLmB#z~Fwzc$PIceQ_#Wb$n@-9R@_fa_y?7w8f5qJOr@g=9P1@XjE
z`xXgj8a|exko#uG|>0i+uI1?~1aaXP|V5pGQVrri-Nk+b&
zXt5!>8|(@RDQmq57-MpLbt9lnb?jny$bWfd>uo;3RnsmUdJO4)y$k1-15eQs&$Gce
zb91lQ3T(INfUm43V8(cN=|aGqp!E27;5n;)iZ65qS^9a%jk$ome5*Z>x8!QQiZBIy
zJ?bdm4SFZG%7p1*`zL?DcZ769ZvG1vfETZnT}i?8v>u0pz^AN~dcUm&exda)g1Zg!
zQ{Ba%n?t@!=EXyNkRMn5Wk6pK;@jh1{++YcfE
zyS*pmgU5mI$Uh7nZ=B5{%Ax`r%791v7H{o8Oo0FO9b5MU@*f&W_j4=5d~@9UI_LuD
zjDK+oI`ObRg7-z5?grklQSE*9pB8fa_JPC=A#lzKbLv-jX(4u!Vdr1R)|0}8pSBgV
zTFB7c+5n?95O2TAbw;_teE0q1yQJDc+NWxB%z0T!f~23M-)zUngh5~RoWcbR^E(n5y+23AUl
z08h)awRZA{^NF!r^3}sF#HB_zd2v}YNsm<$YU6_Q7w>u5xFCG*Qs(Q?mIfkd73u61
z)j-_O%Q?&BHdt?>J+9gdgEYRO|>r<%H?dh)(TRbS~_3*p$i
z+}r|z(yKl08u)rWi2Lz6?k$g;>twEY-Y0{Ej22f^9j;5ozm
z2a8t-fOS5&rNYK=11OcBVRWySfV}BMkWibyteC8
zE2-XgdT#B!R`SgIHBZgwR&sf+%D>0AUl46SPp4(i+KFabBDdc_E4gc$8LA2PD249&
z{LC15W^_{BJ=iZN*4DQ@fOT8y$FC3${acZtbzd3yeS(r(jvLrH^>6tnZ#YlxQ#$jv
z6xP?1zMFZYunvFR<8YY{@=Gfr*`-fl99HjXTVx1)z2}?QYFL*JS4<^&#o)cf@gOVo
zBZk2x5*z2X60U5H9XA8vdBJ|s=-oS8NWii@YFNKwN0PeYVLh9vu1)WPcFHIA
zkADG7pL)Vx8c=mo_oo4{v!e0$x=XO1`BDyT{sQ}HduXsX2lV^tD|ajGcjxUCpBqw-9t1t}lv!H$&bN@FjOqXC%-UgVqQ
zQ;p9Zg~^8%)1m%7ecM&=Jg_J7H|6thU*Elke*KGl+0#>i_2cEAsGom76oY=871(tY
z^<#~^@1uWPHs(8*8|+rl`{0J_RcY;bD2~I?(mOZMuZEaRI;Iaes44sNmI{lUT;dsc
z4CLY^(x*}Z3mcyIq=G#``-jW1ze@zk!+mg`jhbDp91s1oY;C%u4(rBuW$0~P7|&+m
z7KgQ6G{zN~JMJE6XVHVWIJCbhY3)wDBY$oKTNk5#@7{59d(6bzOTg)tBz>amIX^
z)dln%*sU211Rng<26%lEkoWf`T7R*2Q6J9Z_pS>wFx~2xIuKZx2frKe1P1PRC+1PT
z?64ajQ1AZW05D>RyWXpgV1DbjJRq)cu9S`AK)m(iCADWxY9__OL(DnnSC}T17oGz8
z2YS9H!|x
zpWw9tkP9}(M
zG=1TVMW&R22QSQhsTqE9c7j@|J_9qir*EdQuNO4r2UgLlb~|N
zCgvP)^C(OglpEZ)i{d?#aN3Wc4`*q857gm$81gM8Su{U3dL8Yr;ai>sn0_VVFdeT+
z8{t?FvaF;J)&CpDPqBZ}XUbbAb@x))<+37ON6rg_={Z8B#FL)CeaGA#uslz?Ikj{7
z;Bnf{pK61v1{OmwVQ_z*P&E_lQU6F5%?CUG+iLi=*h^ovSie)s$S
zcYv7*ce{-M@3xgM!S@B$Nhu}G1iY7cF~AN`SvqbTo+sW6b@=B&{LZY^Ond=$CJ%jl
z9{}l1ladqgUYvW6L<0K72;D>bIe;%Va2xA}fxNG^p~o6=5Y?i(lOS9Cy~6bl;*|Vl*5h=LKg{yo#Ea>{J|(f@aDMCb(5(g@eyC^cZ5_t5
z)D=HPpuCTbKtTiCTmKRl4Xy`X>AN8H`V`W$=ENtlP_
z5-(+rL0p)zPR1Ab{ZdugKzHB)nO!%lQXpQ&UpM(%3iVoRDg55`s|9}VZBkJY+~@mC
z*eGiQzdF5sY4u6q9W%`xe_oJ-dQV;BId!y!z{gY0y_*E{>Tu%Ae7HY;y7fRt74+Lb
zZtdNG+!nIlux4k?5Io-=P#R2!cJA%SG%^SNo8mjg;8ju!{7%Z=pmp~^f1YNpW`8?*
z*W0jh<#y_KvSsZ`M|0iK@jl&mXQNjQW9+>c)_BQNG=eb%&WA_3PT
zSG$~lLBa=5biUK8BBn_n$R+1o(qC}ULVQNh!A!RM(f$>Sl;R&#F<)Cx`cKUwW_U8T^j?yR@xvu9ETA-Ja^-
zLUc3L)sk%+Np^kELtma6;!2+F-Q`k6BpXu?Zafch-62gTOR|}`2+!)W=BgDP1}6D7Bdw?-GhIL!|Bh+_4?7!Xr%*uISKX2~JI(Q!R>bPzymcThA;aJMi5ZHGL
zriVvKpyyYpuj0B*iSlH)geu=->1fE
z6<(8UB^jE%LABmsw|$?gB`@4h-|cVq)q>|vbM0*J0FU>xQ%)0m2Y%w%^Pur^@WThf
zK0guxe;R(NOkNu1doI7+E@Ox@{#+6GoP~duaAG&uGZZkn7X5R`P9fc#r(%G4l9J4`_kZ^L1*i9EA+1g
zV{%fDwu1laOE4^i^>O|D7LIzD_a1(^(rg%y<3;;3Y(QV5@JRhZ(0BQ`JPP@;Qe@#(
z*v7dU@cQH1AItzK3&OM=Yrm!D?}n6Df2664HZ?)fVi*QWO-MV$pb!N~GVFTg;R%?mKTD8+E@!1$G(
zIDbGGkh{iquLjuJc48+yPzL{}?w`*O=|8Xb4>ME!-HHoSQw>P(tLoItA->x&F={^)4lK!Ussek1I`~?fYM3-ZQRQb%^qpjq`6(-qSFP
zx!1Jx?d(SVJ61aM-csdbDCM8i)&)}j#1jHE2l6jy$2>;dz1m*{(X67K-gjP_GEWO-
zF_Y7j*CgdxE1=y6e|`?bbRP2?QxJ=0l~R7RVt*m!af{{G?89_F^}UoQ&0F}L-sheQ
zSEKE*TK2BP{G~z+pM3q2xvv(x4aX2j*KJaw*rOp#=S83=g_dhm-$T=nX*1;vrW~W`
zKgQrO8k8$2yGZ%u%_Gs&juc*ekOJ$(;Wd@lZ49+R`MyB|EjL&_m)38vqZJy4=iDM(
z28gpdbdMqmUmc<4PMtKOwXGj{RO|bTO*EK
z-2^8Wc;9vZgK>KPGc~3C&bYv$@9j;@z1fWU-r?Wi0vvcy_f_~L4xA%iwvS-FqIqug
zbCuCydHtw&=xA5I$E0T$YO7Dt00*&;qP_zm>T%acvfKXnM5xqyDHOF(%;E95)+
zd0bY5_D0TTC3QjEiA@*oL_4SOJupp#eCf_b
zk;NM+h}R#u$4p9n;J)#PStRmkRj*y!%|LHKtyj=nDA&HBu~i-X`l&U#K4LE*4jOH8
za059^Bxz6HLb%td^BTVeaqSH6U10&>DbDYF#IF^%ke0aQW(_GA_k`M@2f!EnPZWDD
zae#AbmWf(q2=Kt|8oX3%Oc)
z=K0jUaDLGD*!)ePmApF;wM!BB^at*OnGMfd$d5O?tEQ&6l9(svNr%(ny@tk9)nlZU
zbkqg6eNDRMHP3oLp@IpAFUnM
zY$YDy5qVQi6cP{fQ}ZnD-h3?UulLN-teLEd$+vmE=Q+8!+3Vn=!_{QH`K}dTe9n?v
z9ybl*qbtas8!E5lrCZ5~r#qcKMz@jLiQXR3DWxR-*W1Ulm7++T%^{bqqEE=lh^G3r
zVGZQ#pQlTCk3AxSWZv6ihaB>!aLH~j`AU)MWn|Nuv$Nk!bl!((xlo<
zv_mWI2LaD>u&v)KI<1PV@*eg|0X}Je+V7`LP7`SqoAXM3w4NL(UHzwZR~Vy
z^`}I1kGkYa;Mp@X0_Ji`R+Bd$bt|qGRT8O0xfY?*Rq%Pz&QtUG%ZZ`jt@E`oUZK?z
z0aNWO$ia;6(Fe-q#Cq44rAPhBfye%tGZt4u)Hm}FJ|4~`*UpBFD7ci9SdXlG6`5sZ
z$LsX_=QE#>W{EK~5AQ+}A}?}XZ|nhS5V$?rEVP20*s4)8=jEhE75PR^*$yqj}%-@mZ~O;Uux7@7&Vh(e_=@ncwV9z
zS!6M-wwR0?zSxpq@Q~bjdc&svPAT!)B*N*vzn;Wp3ChPFX(kQpPN`-lMFSVrH9W_xWQci5;Jj
zd3{+0nR8Lr&1PX8SyZC2DXAO$p1*LLTp2uXzu~2@0QSGz@}@BYe06GnNc4Q*ld5H+
z#wURX&r9Tucw7hm-E_ILGwk0;QN=x00^sK#P2&9u=fzSrR`laV%|z@aSEjZEoR|6S
zJWVdc{rX%>@mi?Y!qP?V^K`g}_j`G6O=vwSQ4D&x&c2x>@NTo2e+JHnm-W0#;9S!3
zbGiPR8Neq!(p7nSYRRRARW4O4n@HktPgp4Un=e{!jVi%#eigEO8wP%I=F;;E?%IIA
zhhGszzmps~gM%B=KV=v#Oap%xA9iioV(@d)`Z~=aa8A7+HzQwmHQbk^KDk#2=gh8&
zq793|Z$?Z`ziI(`$xB$C`=@|EQx7xEhxSxbj!A|>|0H5g+Bbq9e7(MS-LKCjEvlQ7RV;o{53r@`|wk?n>J
zV3&KjRsU|#?>u#4o(t4#>EJqh2DCp|D56>k_HYff#0f$BNB100X|6j
zUW=c*xouF)i}9>GY!|<+GWh>XmwB8(Unjh$-xBOI54Mo02fe5^pY6zReOpR-65w8b
zfy{T4Z7^=99jjmJg5LRZ4XvqAPjO*P*A>tc{*huO2(s_Up?c&2C#G#rLwj;nUcnbz
zKpdPJ3B(GXFDAX3j(q<3oU2iY^A9WKVw}8F`|S{-B@cfb>P|A%$e>7AXjy(Cgpu!HkKT{*F5&kJvc-2IWsI
z4MTcQxLaXQ2ikehszMp_Z@hP+d@Efsnewr4Ugm!EYThPCtk+}yoG+-~UGnPGs_5OU%t45qA8$<0lHWWblnmT{;&HfuW$a@LGQ1_)lUtgy<2(lVj0e#r}b12KXj3z1rIofewp?Taih(tS;!;hFMVEu
z{8hd1!XMPzXQ!fs`HkAs=sBe{L~%0e9b5KF4%b(S{JAwKuXfg1i|yYkdA}Cr?45%2
z^FGPvyx{|iaNqgCAQ#u0n`f4#u|!ZxsH>m`*7Zkk6k*p3-G){iMvRVuHW%Xtw5x_a0U!1xK
z$h{``<5SS<=g#nqe-3^_M6uQTCCC@|=)X|`^xJLv(0vA!TVEHh0CN2`&W+-rmt|Vv
zz7G6j*ENr>`{4Hie6EaMo&@&Ixu%#0^(W2w8HoPHXk=779PTCE3l!S2;ohV;Iw9BL
zM+=d^Ro@{Aew9z)hTloBYpz_KY5~-@=;)`pX|rJ*ziSJyfRDy4e{y6A__fZCy_FZ>
z`DsVWn>|yY+=s4bX?wvRUwR+upZEwqmst7h#cd(W$
zf_u%>lCs5z!&*tzhU)H+@iro&bmZ-=9j)Z^tu2qNp0>byrDP}`{Qs4;x^WSXt>hQq
z%3oTD$^ZX0TBBhz$(=6&^lOjabmIGZjVNjU%Pca`{FPR{RXe1H31
z8~LO0aCY`TxCiK3)fM?5kAz^P&^iKMw#kEh@>;;b^6
zZ{xot@^Q-JXG6i6r0A@`WxLEolBMw`jZdPP+>_p>{gzco+(+&mSf6x*xH>)*`+O>g
z7~Nby*xnaKzG`2eu;wo%P8lZyht(dCbQXtdXHGKttT2ab{ltCJ&Z8oB#w~>mR%vt;
zDrS@NmN^2y1yac4gQuSD2LAHzkl%=^b_Lm$wD`wLl{zARDD}{2V~f3TE2%?8Bx*ZF8D8#i8_U+0re5>0u0T>EMX_t*@_{N+vL
zQKV5=HmujTt*ie=ghE^tk30MacuLioH@7DyRg+!oA5SxBttXp)oj-X5=K0&kk%r^I
zBl$ls@>ysC>&)t)mCB7W;HfzRhh|ihjRyUodheNkQ%UgDZVZtiX!^>HsInX2<+
zXBO6yB(I21w-&UJoywj!p9I14vG+1YFW@{@y(iny81~7t24$s1u>SIwesH=8{3T+3
zMDH;4qa}OW9xh`zm);T8D-?kHyFi2U!LYxVn;pu@)`fF=RKTLuu&;yqpNg|;fXAGR
z?OZMg_l&B1df(xE*4w-A`+V3R->SZSngaD|o=7%TmTD!g3!a9FoCLp~M6|B}-@0V<
zX-Ww2&*qYm#5+ddf0npt)jWacKG%PKkcV^I`#WI=dB6|IL|!pff_u@u;;*Ka8UkNZ
zk}l>4zopn@`;dSgJX@#jwgSH-F5!2a6XvB@?yfVw=WflNf5!{-Hg7fZdk+2LIvH)a
zAM`fajL2w#-a@0-ji#VCP+xQ40r;J@GdLx-gZ_*%!KuFWa`@JtEJ=
z)m$>5cPp`)zK@Z!P@s1sq*vtq(ZJ_nSrM<|kzaJy*oUtJJB5aw-aZHYE7{LZ9Dwvk
z4-6?2>F{I#`8G<4)2Vi_qm$&O@EMoWHvOJ3IN)DJ^e@w?7Yv>
zSy>HwAHwh4L>_B)>5$Y8z?c_PEAd{`!0@ai-n-O46%fMsn7rGp8sm|>b4nq`^_~7&
z^nUQ)YwP8xx2PC~oYtu(foC|lQ2ye*qKWdCFHY|%KU!D#p1vos;KkT5>W5xG
zMeoh(-&Gqm6sh_P5F_LLC+BeYTV*t_wE0g#2sgY4>evfb%64cOl<;
z`R+&@?yuFGw#y=4pDQu%InEDP#?eLZ0hby|!8ag*-+Udr0oTK-c&+L9{CD4z3!1io
z8%(aY>A^m^=;gQB0%T|6raK8RbIfk{F~G_wIiGj19?w-Z)V;&~y|LL9>laT?kDmc}
zOH+Nh8z9fe(iVNtoARsoF}}Z5-RpW4*MH956Fag09B&%_)j|Fz_gfqPf}HSZUtk&d
zEBl2(wi6&5=PGWkfc)6ADd%t9TIPaY+~lV@
zUSN00TiXgbz$_hJ&&BZE_VTLjwtiseCNVkF8NgqJ*6_~N0RB=e8j`I5yyoVV-r>u+
zFyD{t93Dft#WlRk`kmHxaFu#+dD{~0k<1cV^5mkcE
zL$oh=X8?X#J4hnrGx+V_Pdhky;dz*Fp=Dt=Jm0d>SuYy~^*l*!cJ&9pG=40W$0NOk
zkUMvOG{SpsVCPBsH@kHDEyb|hJ^
z72Z$Fi>|9|AtDENFY^Pvz(8@cBkHZ>n$&|W0(Yy3s-DbETk!YXej$OqS}kO|U8=gA
zaVr_R$s3mK+(b6-`t8(faDkZR)(B?U`{FNmR``%caw?FK!DC><215aSItsfA(ezJd;hdEF_SnBrYsV^n@+}p
z2hT0`$skJ`y(is?OM~~_!z_JG7L85R1tB$FII&O7k)Tn5qoy>^MceLDHdDOGaWA)9!&FI*7d@sMmk
zZJs=0nNAKyj9DM~bC1j%^BTz2xJO2v`;D5E(}?3Lp0^)FGs&-3AM4D)G$ImgE~%xG
zL4Mc%ta=laO`g?;p3*EyC#Ej;9fo=tL~?9RZFA3kQg*i4M6WxI=w@yGdcz=t_(T@C
z4td@uF&D~D%+5$B{F|Mf>T5E{)rW|a{)054-j`r=sv?~j
zdjFOe8NE+}N2Z$9n`Dw1gB^j2d$P!A&)el!o~Mz=X2pwU&P^x1y0!z^+4tf5OCnS<
z<1)#H@76kkH?zq6l(z@p1HXJNIU}ZbXECw6m!IgITT9ybb&Um7TFCBSGpqS8!agsW
z@|PnV_-a^if8LQ?a<4;Bx%*=!xxHdy*n^{)oO2U2DjtLPVy(I3J*_iIM9#>-^eOkr
z9@p3X2jM<8W7zb#*~4cf#xseJ6Yec%xo}1nxIZLgVsEaI^%+Ea%uqtlGnb^jSwG!R
zy^1^u68+)QRZrSv<@i(F9*}~)LYt+pW|87QA_s@{v*G(C{4WM1RFK@HLtpam)R7q5
zM`Z`$d9Lnx?XsGDIG4p&kW}D9XI9rA+5r2<^>@}658y9Wr2G5Izbz!C(DHu%L)hoD
zhH{J6z&%a;g?LFJc%S}jXWI+d|Fa%XzajsziFAmE91w!{RV|dB-Q5xa@htJ@!cH-`
z-`Q<=Us)LZMyV
z$Grq60YB_to{_r->d9G=+v^N=SS{bwx)J!`F5N3{cFzF2BG&O4LO-rpUg!I$3FXyp
znYco|{ZX&menESD@6%t^q94x*92W+Eot9ZYmByK$@xLWNzCZXY;xW|kyWzu>vmmc~
zJ-HF@;ez(@ecB9iT+;62$X8ra3KGR2eNEWgO>?2%`Ll*jV0_x$WVUb|(p`r9kD7vf
zqC7ZZH~4#rG>+*%pkEK?=aed;ewx)-1L&1bYEgd+STQh`+6Q`Ob4t7`fX80-+QqR1
z^x6+iS#D2H1=QGO^n-ufiq8F
z9Nv8=m;-r$-Ofji7}wukUgV4TVYUwC1Tm`*^-CT39K8u+_
zOY{JRE~oBM032d<3xp<$-qJeEO>k`!k$f
zGLAU?jqwXCw>c(xE%Ffc<=Zr|pFdNEvoOEr0T&gpnWr<%1G)m)JqOZ+l}(o*8)MPC-PB+zF%1OEv*#O_vV|@?}g$l
zGjPXoF@iOEuKhi`LwgNgw&WYEBd2HMEc^T?2y*OE#0G2oJ
z_nrgT`h3HSqktbmjGtP9UgY2Q1<04K<>||y-oUv9@Xf_=zL_b%ItK95v#D-IpqDX!
zMlH7Y=}_-uyf0ioQ+fsJeF|9qCK2#%$=}lwpy!l!Wdoig8oZYtMBZ{y;XxznjjU3)
zPXk=z+W&h5a4`JQi=Tk5YO*16Fn_U`C*xe8{31<`VjsXmALZvwhI^4i0|6F=pm)vQ
zeRd73XX(!yomRnq`Q8y>-URY?6B6Qc1
zvd9+FyP&@GD?jjyW#3x-#bAG#@7Qh&&kq_8zwp`t^=>F!EiC}|HCDPg<+<-+{~TQP
zZh-`R9?F!rKi3l0V~o>~0+d^2AS=zpq#=b;B&DJ~w0CK0OOkdX4Wi_C&g=dC{quf4
z&b{Y#U+2#AyyJD>6(w_-Oy*F&?jySgVEh`^`ggTy6^w6DoEA#i?Zl4tQ^*4t{~sQ@
z?{Li$>iw&Nco3WuyS<%9=?3hR_V=5Pjn1}{?T;otZt#F{H6Zp(_@QUXl;TRy^-{
zUP{VE@7HKawUc%CPIgM&A?aklh0OVN&)$$Ls-BO1NhXm!c=Axq(+Z+nG_ETH
z{&IBQeMDY1okZw<+`POY9oEeqrnDt937y3o11q|6VsAGRS{+(T2KuzuB$vJ+hAI)g
zHx6WyO}BQX$9ZLuI6FQD%T=Z1c(BHXQ$bba%*7~EB9cJ@^m-LuGi4LWgTID<`{j@<
zc1?@nJtc(gh!gi@N(H&#Z7O_p%uuUfb_dM+jMZVOSrP$iOn5Bho4LGf#sV4Z`NO>`~z%N?%P;cwVr
zezM{u`!UFe(ysRnEHEDtWPR~b4g96oxg+;$!GCJxFE`o33jSue{@O6~k1Ota{zc%2
zHCac`hJ&9PZ(dWN!+XwNcz2E-#uKZYJD(C@ydd@7
zxp)pO^?CQ7&ycSp2MtnjKK4~LSrW(DJ#RK!;<`%|lTAq>Qd;JU&zG|;5yM`^AIx;!0LzfF&Hj2Ev3W`dTEE=t02Ge
zV%)da0B}5@I2!#@O?Qt5+HtH%$7ng|#jGDwUjVePPdRK1;e+AayM+Li%Ay6(uP>tuz3GVlE%cX<8GR{#$tLSa`AZfZIm%zM
z^nQRZc;UOV!;SF8GPI|H3kv8@57ikaAubt>Z$>}x(Qi5#^SiKQe+>0D|JrMW@qbg6
zZX%9%eUw4H?Orw1y(@b+@loqQUwUEr2*S_t(?3HzH5HPCdip!=-pBVYS{$LiZ!#$T
zO??-W?4wzadM&>hHlhE{{WwW|XBwpP#T5Nz`p$71^p}jo@m=UI#TZj9(O<@UDSOWV
zT8ux1tD9iHtKgE4yk|p@p*$C0)_n(#*MQxZ+ia=<)81FMYXI(BeeDFU-{p&h!-bhp
zpC-2FGXZt$s$Sm))EKF(L4W_$|DYJQOEGoNMQ*@59!u+E0Z+`l6C47RYS5_1^4U2Y
zau5B#`k$=_QE$hO9jdngdm{HRX+!(+h)-EF2KB{szee2(+BN^Wo&1!))T!6@1An>z
zb2#W79!_E`1wF&4F_u>7@7A9T6|L7pIsPzh!gUn4UGz*3LGK*fzgi2>`%wJ4#~J+j
z;0dYS4|38!F
zVm=T4@>$V`tHIzeO@sEcErP#v+ZS=v2K=RmGt2syu%EZ}Pq#arIY_63gr(W$-1v}D?w@2<1=^!e(
zW$EKEPBwp@jJ*W@oaNSErUS5_wOZRwMT`!3W?sh2kkoc!@~}sy1o%di+}74DM!=(z
z;$}9nLHp)drL*h
ze88@wP2hNR$9(?{e(zBR+eA>mR{xOpTu+0gIiRTmP_xF^O
zi`t2{S!DgAz7~?6W46C^RS9v`Ew>tL$Rtc9NLA;`)o+R+82r
zc;=*6DcLMku_p9LF7bahC?Ni+gm^nLi^Mp$6SLYudSMl)Cr>f%rVZufsxGebRAA=zmY?1KHmM3QxQ304r+#h5S6
zQ<=p0x-z{9e;x^68xy^=HjnV|PH(y2Urfp+%ECe?;oOe?!`2L^9P-slRp;D>LbBof
zXNwPe-jbnIy3oA>CFFbmXRS!J3SyZ0u~@+5EzvmnWOOeJ*sE<0L6t~-U?Y4I+D`z_rVGWHb{9S4q*r>&(V!uZ``A;oGE(>5YC
z4E}P`X=5WJ_{&`W1i^dYFKYrH`P>J8$`_O~yQz4{0Tb+*7`>MMHYarIS?jVcoLe?r|qbKW%^fjS}#ew|_0M3PXQM
za%)_`@4vrYbJ~aBmyBJ$!np{%q9;7ea5ePnM7y((m7#q%*=~%!jQS=h5{OkEFyv@0%#mxCVNL
zWiGrufOU)INBBYSoGy#y9nj-`P+?aIxOgpp-W%dO#XWk|fM>*4hM0MQo|y^Lwr+^;
zjC*#+A5gt5%=r@NS%2f_TL!q>Oz;%?%LnU*D1UibC!rJlWvomSFZxTaBSUle-GOFi
z*8`l_Ecv{VFop2u@4k7$fQw1LZ=(O#?)f;e2IAXeSpc}wVPvlmj|_}b%YfINloZ~W9qq?&`)0UWiCey`j{(({_?!HD?eh=cTQ@bhVZ`y4ozy-^$DPvMsQ)LZ
zhDy&t`K>(0*VqqzNoKXFv_*^PSbUdq(`e{5`f
zBFZUhaUAl10wR8uI8Ww0n!OwSZ}SVwb?7gB7JBdE{QR6=Qzr7F$EHhg6FAI|44rKG
z01r%EWj+bGH*$VF5cFn+&DJBYcw2Y1674wg297fUR)w32V>y3{@K}YsU_)}o4&)tY
zl}sLBzVz;SYoZ**Bx8oBnU_Ma9
z5UKSQ!bL{y6;6ZR)qSg6Rsaq~(tl2b@M-gX`|@EtSgExzF$eOx(P|DX$5YF=USqjA
zI`h$g1v?%Viu>YyV4W+jS!u&Q$hc3)fw#xtJh9ZAUK|^=iwPf_*iW#&Yretu!Jc+9
zko-{V&^jou48ek@uueK8GJf+1jc%ZhG%3eIz$zMmCvtCSq`Q!ov`(MAM>-^`jj!hH#Sa!Ep@
ziuAqwVE*ZP4$;%R(In>8*v>P=ySQ#M3x_V$+y}kj}$xX
zeLL{3ihS^W#xVV$l`Ny*CHWlYl{Y==SALv={>7NAYZ6*RlD`(Fj234R)@X4?^LsEq
z3J*VM3*T?$pTBS22jAiS+SvPcWNiz%qBBvi`?P>~zuxt>eY}8(GHRRhcNLNTt8HwV
zEJehyGyCN|AxET6y#FR57`|wR}z~rG!+>ACp*hp@jVHn;0^>TuO+8
z`BcFAaU=6C2hERpGmBS}Jckn0GZk0cUJchqpJ21QSxJ6eiXtX!z<=wRoPBnqoK(0!b>KML
zK+<#)Z=1!{6CvFT{U@Z6oL?&w
zv(}-Wh|&cISVhzjece?*n})049=M)EGX5(R#c&{>4BgNl)t6VEt82;`aFo
zRg{#HSq7G(##1%q<>dlWLyOqWROy{K2}9MYrJ8mjyaYvjLYpO{kci`=NE2;?3miqw}HMPX9-6ikbvu0xW<}HVhVJi{;yqjSo
zV>R*leU8%+_{+}s#&+-T!*`KcnVUSKV4dNsfyN;8Cmty-*}91uqT-kIHFmm@OnY`)
z#X&!OW3R%m5e)N&Yom7qz>o9U2JA63ZXn`?3Om2TzQ)ah({~Lo!1%p+t*DefoIfr8
zkt@pr=S<&L-MJ;yLU!g|{VN9dqL}=Ue`5x`>y5UcY0D<)r`}H%ZUJ_kDKSyeXoY*%
z&g+;1-!gI9nXwso{l7A;4cre
z)JpCGe<5_eeeYr1ThI4y?mUbq1LqQdgu(N(%$X&ZA>8HdRu(x}FL`y{R-{aD9feBw-g{&OWe@DuezFHb@Kvx1^F
zv;c7x7ehoU@0sW~BXRF#lZ+K5Z^+mvavkqMNeVnRARIxLxm(=Ez>dJtAm9Wzu
zA$Y!)*Ub$1$o=apY@$FfklDnuAN0EKpX@|`@a~n;QXAk-7rPGJ*Vy`>Mko5ii^)p2
zhXJLv$D4OTI(L~`*K7cpOSnN9;{&uVG8_kt96Ohm4CC#r?WOOnAdAUfFdKz*+;nx<
z>w&y+%O
zTB9ul7gqsaoVrqkd?}Y}uOIGT1m4!q!1;bvoaF|6z)j~LoIwA6$=%Wu)7gDgz!B%|
zwtM%kL;p8&YnHliiMZ$Io;HsJiiWhz1
zRi(~{+@K3%!|)xP74oR>=*CB-6To?#;z7FJaJ&}6r)TjG`oM3C2TR{npK
zJR*OX;xFDiAz9Fi-TXll(O6HN3P1QukP7E6siu&p8UBtA;%jCvh$D{r_#Vgn-}$|c
z8Rcx}IeoNeQRLMeuAjJcCRAd1uDTIpiQ$FoZ0xZdBDHo{VL5y*{;?U?GXnE25F@~K
z=U>Ags6hIMPi)2XPW^s(9??>M!XNclgl#ZD{FNVl0>f*XW;Y>TrjMZbUf{$<@{b(Y)99&9C+>l
z_@U+2#&qaM>CSAQxd6{uUM<3Q;$wSfh8a+XOJU9dut!MHV-lWUcfXOD9r~B*ISaVC
z56+2Pk8el(^;rl$T*P|<>`$EsP$-HeL8;mP8FQngVgWaiBS*gze_vL15
zR0P8K6B&Cp9MWY^yv}V3XcRtfzX1I>%{+LcI6S|-&-~7E9^hwB-q=im9f{jsIS2r%
zICoO>l>;G6i{a3(xf*+Z?+0{X_ISt(@fnfI#$kY`6@CWd`e$sn8ZVwFmUsJ7Dz_i@
zch3DRU`_|#b;?~w3hr%^JTP@cN(s&ls`dTQX>BGAAJ%e4)8~>yCF)`M&A|I_9
z0JPKcD@Ml^VLyX)tCbO~OEajaUBr88g$AC4)5CoBdK3L7eb~>jU2=={h=qH}I`VHF
zfbqYHYsv6Cl&{s3oHblv_nENhcjdr`Ra5$RX+3Tw?mHYbRAkEtOQ0N&-m^0D;A0|x
zf=U%pD>k%TW>ihWS8mjt8?S?N!J_?pPSp}8#)ziIqZK5lx}C*)pn|9$p8RjfzK$p?
z1f(mSsV2Q^f`={7wi6CFzO@7C%|!pBlhz~CDmb5VY9PF?mRR+_U$JuN9VzdueDsYC
z`qhfV_Wj&#gfr=q)W^wYlFKBr$>?2y$ck(rwmq9ae57k7>_eII7aqbng7JsO
zRnB0)s9wn?n+Br7bN7+3b{$zAu5jBB)<0G+lj+pnSxfun1!t($lE_9=1H*0A#FFWAr$-Ew`-8(7=2zi+tB^{5_0$Hk
zZ~Brk@8eqHpK!=0q@$YTZ+AD#GIWwvnf#U&8fQvuBsZfEYMwjOL_{JP*K5W#laGqq;*R9CkT5I1
z#igod`2S0wNGln8?VLpR)Q|)%hq00TYGQpV<5kYK2C}vAy9E1e3wfrZ`|32T
zQ^{J>2kcC%C*P;pj*mn)kY%O$YYseu@7Gku6WH}&obx%NBk2j}<-#KLnXWX$I2%8_
z0`A=?zicbU1mn=%jDx
zrAF3<2ktOG_2=8Nc^<}@`U?(v*F<3-#p-*KH|(qN%@;oyg>lyAXZ+6#fLif?m-)ju
z7QQ6+(3>5`DMQw@J1}nB-dvvV3D0kAdUcos{MwK2jBobJ
z=tc80ooI*mSTMXu+zsjX95#A_>jsL7{5x>J?79AJTl8CB*QV*U!FvNGTr?Fy{&2DS
zLIcFFVH$`;o?FV4JcR2di5*JrVvx?-d%eqh*>USI$b0v<
zp6G!5f4_A!?lZ{so3C=I@mT9?7Y8H&w~!++n-b?*#YMs
zN7x(%i>x!^Vpn4EdH;8Y6fZRyTA{KzfbY>(9s#{d%`cb
z0;u<^dm2#XR^sYM#ouRtM$Hdm`j?o$Df>UFgm%8|C#MMN?GYe3SifVY0ju#n%AUWj
zVEcH`zswrjTUgT?YW@=~WK5mk7poj9L;cgs#HsK4+=?T-aQ$^hN0c^|@4v}ZI{!uO
zc0>75*IR1+lb-<==s{Mk=emgDJU6STb6V>+CQ#pHirEZ4#c-96eNcg5x9p&B8Q5u(
zD->D=a_1v|6Wk~Gb(`(N9LSYY_dgzhaqJJ5IlVIRdy-@5lO@X74Zx2%tWpL*QBb6qm5l|)P)=8A;zlyHvpq`|lq-@)gi4dW(*
z<+!gkJbzP@r|bdvr*qm-sZlUa@&#wJpMmtZd7lix^3WL5`KAx^_UY@r&v5(`+NUNL
z2e>0MrDhQF6)>(*gX`aW|6>#Nfbp~831=|gE9T>6`Xv{_+Yb4um;;8E+%icA{g6B9
zi+W%O@7IqJQm~G(qtXBBR2bZk%a~iD^BC5pM>1`LZ-XCvY;a6B?;SZemkzA5mTW6C
zqgw=jC_P>CVeKs3``cT)=i^P7M}Bj+X)I`k{g6L-r&l%+t~lS-IhL?)!_ks!0Phv2
zGctG&^NpL+$GZF6U|!-ZtuX&_;Jo-b2j)LJ9B&Ts)5CZw
zV{3MQEv%dTm$B^+%y<5|Yuwxe>pR)PyZ607KPQqP9}mw}_4F5xwX~AY-dXZD$Lh!@
zP1l5^p$0-P(%AgzcQa8B?c!m$)=C=6j(^C8`+R#Gr!LxBOc-!u`LmR8R9f
zZX%9PElt-M0Y8`vtGVeht8R`)h`)w{O|pLXMtXUAxYq1@60F
zn&5_gz|WubKgq!Uw9a+?6&iP--gh*#-?fMKYun40aI2LZ4czN!>jmp*jaH}Wi{bvP
zeXm!XSq<|%udJeuTTq^_OV)%0fM3|&uICQ(n{eIJ-!fC-9<}go*GvoB$k)ry!f(lU
zkTs*jlFa;YZ||A>2*InJgmJ?w>)nQMe$Qt-`KELmiR{~cWSwn0;dPGU^wsKsdkHqx
z^<3#7Q`Z)6XA86vTS*U-dN?P?*ULW_4D+lQ<=;ss!rI7(G5Law_BOJ6wfOihn74gj
z6nFQ9^=IuT7Y$cVL%yBY`LS5w`B&Cpw;*U&P5i4vp&jLhGsm4`g?*Cjw>7@7A1@`w
z*XhRv>n#g1^Gi@qayA2|c~?4#YJ5_u>p&-Ikjn6X{I8QFhp5$UcZd7iin_lWE_IN)
z-^-F?Z*~xyq`l9)>pF;&Pu6k8`3@55zcc-4R0p|!ahfxBTL%eo{3-FKwT+a_-0SdT
zf$szUY8kJ9eTP`Cs~;oa{^D$|!2R@*ZNzeB@o+V?-zddm1%^y$XXj(p9#+7*fuWf1
zWCW}a<$P_{bJ_>(sV~%Ke}oRSPkx`@TY(=Pz0Veh_ZN%@Mjl-R
zet(E%Bs8C`gLLz2-MeViM$VrqJjw$7+wP+nTMh8=f|mgS`*(DZVxf#PceIl1Cl?h8FSHO%(`fq-x7&!w2Tg&njcp{oVybIDw1>UlpAJpvwvyki
zig(lGTS;)dc62G&DRee_`EwXQQl1>lDgO@h39f$Lra!GjV8*&AjT?AUwzQlU)Q8n*
zy;B1}^rPirnI8a?c)j+0f_>u5FE{?EAAxnzcF%nGk`{8RhVfSv+*2nvH*fuh5$5HQ
z6@A}<-*+`LnU_trki4dymMRZgh^5@jobpmL>GxXcIUoV^tcx){QD$HlXVkDQoYNmo
zdXlxM1o!{hW~oabZ6&-n%I_wP!M!cfAq-cJ!uh26?lzTUus+(+b;Wo+ly9V6pEQ(n
zv}x$zZ|>%1~&IXPgwz1ux
zPsqTzmzmhbGhoMa^S+PAfX_#1u4VTIJ0=_kO20rn%dYWrj{)CZ88yf63~R1s8FE4T
z2i~4q*bm`m&1dZ1!}EKYQf}k9nT9LJhc-if3HHWs8-j8v9LaWn4|W?!CVm%!@l4Et
zE6Hak@b4$_9-^S1Aw@rn>sgk|iS`7P*S&fFX7p<o|UXOt-a@eJha=k^NYzVY{Ktyghg@MQC=uejdraW>8L9)x>U-B^$N
ztf}kQ{9OfT5?ZFa8sf`bo%xeMrdN}=X98&T-MAlly?@_Om=ef)A8_f~f_=&CY`M&k
zUtfXwYo-u>_ViNQTJ6m2+Qn4UY__+6fTwp`-4aLkcX@7IQx54dt{JI9>8tn(Sz&zWJ<&RhKaikLoukw0
ztvHT)p?{?gqTagK>{PrQ+-Z3hQU6oOF~kA}S883`
zIVO|htBR{Rfp|l|H7=oeVoj6r7L+}Y%u&xTbxBkC{O}`#;?w%~BdPny!q(I7*G*$$
zrTDg7sVc>%W3E1@p8Hm)Nbz}Q{2OjaFS-bV0i*OxD0EVXWI#SYLW^gBRehAtq|xRB
zexGRb2Eq|Rg`2~v1&nWqq@`1K2FwRp)03uelnL1e`5=@QVyrCH#5fyrel#EhC@s{C4n
zLa6XftmTjZ@W1`dl>GAwBPIWGc~7065P?>X@!`z}DfGF%g2F$8y%d@rw>QuOb`8EnuuG8#Z%W;5`KkICx^uj7`UPgV(
z$B|V1rYp@;^FLmbw_MoXSW?Z_BC?59z>8u0-WE;qe@_`tsy$gerGt`%b1yO_1E}Xw
zTuqhBug`BObl>JjmBU@p5NaN__-Jblwo9AE2r556=JY9h9(To1^P(G%MX7y~L?$Jw
zpVV;vq}rF%bSc#?dEDMpdLO;4sPL1g??Hh9|9=`_f%EUez~d*gmn@cx)j
z>j8bMo7?f;P0M0l7wm7&7CkGmpR~x_2*&=eG0i>N<8i38r=8z+gvt@<6p?*fBZOZ$H`sxIR*N$nbz51I0=x=)0@fpWi1(Nw6&a*t9m~LYu*VA`^uXNL>
zdtllpo6ex!1_9;gP(QQzuMaz5TyfobTqjZr$#r4_J2wiete=2+6if5%R|&vd9*)&^
z&8fgV&Txq(?p-V4`cKKs1$Y_bkfFt0c#h>9tlfbhn^?7D@(bwOSJpZ50)Jpz<(zZ`
z{E^!|u5Bamo<2RgSzP~d5bW+h1^zx-?n1-S)C2}+Oe%W8JrdkmMyRejCfjC#b_5`0DxztxsZtkl&YUt=++%lDEz5
zDZp3wd6N&tgS~2hcw$@NIWD=8qiw(krjFlQ(u8;(<>255uybihL+cilTgvr0SvS!4
zx31LP3FVqLcKU5BJZIbSe0v;(hh^(uGg}4gu6NyU9)S7Kc!_TC!HqCp?t3ax2>fR~
zul|O|z?(Wu58aV~avyYGMYO=arpAAzFMX67=Sjbq~$KetlS@=b#zLr}Q|T>Y%(MS1tR#
z0?IqCMEoJHr2Y_b3sU_YY%5BH>}^T
zKE)_#tPA!!xGH-gK)&`X
zBhMkf(xU4&vx1$+80(Mgf*r%s&fR#xc@8UbD&3^&lBt|
zz3gy!HP|DdxOaXGo+}NBzyAm9klVm7v=#7czeN)#JjWB2lz0d9SiipYnumPoJY&+`
z4|b0@ABZT1c&SJwHhYNQI?SAdy!&&K!RSMXw^aRnEgaGbtDJjFU>+dO_h7d*yys{p
zN4PSy!#p3K##yk>>2{YgE9e)uNq)j{;mPhJ_I;44FWK7RsxsvF0IF+UQnL*1@`
z>NV>>a3J1&Gb0QrJ))MSbuu2w!#E@j8xoDYqV`qJ5txm#sMia>Xrq0an0;0-aO1J{YrJ
z{sE?^a@6w$KcE+jq8WbQz;1TK8T~{$UtRi_Y46Xlu8Q1F~RoDN-=Wl%f-i-0F%lB3y+7CY#u?G~`
zweup>e`oe$a|Wg}xPP4|
z%3cg#T+t7|-DYtHWwF#{^XR|QzWJ1(+>;&6jk0_jhajfgJpEJ%pTG0+B()y8p8M@l
zjAz{S!W{9`e=b)s{>m@4*ZBOmEC(J8ujM^>98rO%lLJwss_ZA)yG=^_1nLbA5@rnF
z5yRqv_$I6S7ly|k%zB4>VN-8*8_E~;d8zlfZQDfgE|tJe>Kq2s180iwbQG!|!{F81KVp*Cv
zBCC!l#n;Moa;bRr)d#8PqSQ>jpk19N1yp*C{pLU@VI1Y6qj;F)+c}EIE&r}a@w1Al
zR4SbZUoxrZ7oCtmga3W*MDe)3;dfNLY#Z(WXhe4PQt{K>H>h;Vx`rv4emCu2>;Cfh
zl-z9KPsM+6TTbDTZ#4dEwmg-REB=FR6L^nS0xVYpmJ!GXVBVN9g^bT28IXTp0sxfg
zrpZU37E!M7qs=eAB25N)nu~_2|Ca8XFo)
zqdLg%5CARyha93b*%8fy@J~1m0Hph${;&VZ6B4v=hyTr6+W7P+Jz1pfpr31}NMXAi
zR6EGte2Y{)+-w37WbWH&Ao!&jNecD8Xj2$;GK|8*Y_xoh|DfRyX{aX9597xS!Twnf
z?YWIG+eJBbJpkY^v}VMCblN;h&W1Mcn#-l7Q~Zh+-}-Z9tq4jHkj`oSRPKIC0LNeh&NhAnC7sYPV7*IIiz;W&r5{u|AJW`UrK@y$nkrBJ
z6u(ob_rW9S00_U2{g0A0_u4Q+e9Jq?-4(Str9%=Lg{FAy}7X5u;@3J1WduLMJJw%24#sc)WFVt;o
zHvpbFqjzW$_@4F&q4Sr3hY0Fz`kM>9Xx8@y$9dT2F*j;!83f+#C|gjQ0{Ob}C+g$_
zz}BNl?)bkp3XMKOX91rJwmo_Pyk=P&_v*7?=jZijqBcVMZ=y%P)&L%M`xY|-;kBL}
zuG64D@c23d@)*sWbAEz=c5Ck&;kutj^@<&FAP;?%jc5Xtj2d(i1*{a*vv33)ytv4X
z^KZU3<4XMgu!y6X0r|&t7IUEm*m-T44Eq%Ds@dyr*Bpg-FV*s>SCFqGnOy7f|GoDv
zZ?y~n`#W3{d;bFdt~IHD2q>!m`3$ZHYDh7u;5py;vYc1D0Sg8nYvXy@Dfu_5$d~eR
zUdAJzS&-b4?+X|ABHDgX}8CAnfg86h-%M)_Z3jSrP0QKJf!oC%Kj68
zO-vOgIjj;U_F|86;~vHom}4-pB)AqRf^Fb1C;RkosGP*cA?l0*Msvq
z?M&Z-UB5Sl%+SS-l^cc#y|dhZx7(HGzn&ZNLOespFSK=X@B;gJAglr
zEoVCmcq#N6{2?riAD5D@u>o!mjR`?}yQAM6699~5bKijNh^O0Bi5KvM%>F9uzeTfpJN#rHpt&HCtEBf05i@AN(aOHHBw@pcY(d7-+g6cAzaf`(Q!BMT270(
zWO=|&ql0$CpqJwi@Ld(+Bezz>OF?}YY_S>W0C}wXsXhx}@s^onO~A=@{r`*rJ!cz?
zm;f`IW}BPgd50gr_OuhHd{g*(umTv<*o^8&k=bB2VKAU_OEKfdAo
z#%WZiybknquJpNE170aS%WDPUwgqQ>a2!j}-;(_dtSSs2*N|3e#ysi
zq{i!DKbHTg8~-+8{o68#UPHTGF8&d`2zYh-WIXcO1g8lD3_rt}{6rkEwRpM~`@7X)
z?E}~!M*ej+WBtA1imAnRrc)cAg!ypUKAevE^LGxS*9Q9?A9Q;RfoytTDUcJ8S$f_Y
z&!0N%9(UCNw3ruG#PR6V;nV@NCvErD2y6$dUge%d-d?S0x&iq)KXZ-;*83?g2VQI+
z?-Q$La9q;V&pM6e?{REj2lDdkSuBcbfG1_v{K0X)|JXS7KNHK#Ong!A;{FSsI4<>8
zw$Wj{cgL~)m>)AKvjKeG{P(T`jK87z6bJ_J3yFkb6F^Oa<#ovKx8BjUlLCyttwo)4
z@!#CYh~r}U7w@0=oc_@tjhJ4LZy)8i)@@gt#&m*Ht)ub%D?_d;<9p(2w45+p-JLlU
z@z23YSgD2lE@z=Q+P5Q`!36at@9m@b@s#w1lNg^;$VK^?YID~e7@s!I
zvw(PW^~s|czpwiT5C&+6Q_vj&2b4D$A)0e1QGWmS_XLXHEM(F6<^kbM%8y?;T}P+OeqF}oJ(bTWRoeP$yJ-+GFqkJLpP|a_rCI@1{=?BFlwWV}r17gqhv|XQ
z!G7ljOUnL5$w6v<@MN1lHp(g})5rQUz(4vl}mv@E6KyX3b~?=K{Q
zRQyv$8c)iSg=>M}dF#5%l%2ULD=56RQ-ZR?o|DEejBLHA@a4?3c+)sA1D?A%g<^p4%vG>G0vKsX<8|IxEFfPffnouy5d;9tvpP&+!Y`mvAQy^3
zwg64!;WdCfhBRc0gXRTt*VXM5Ui{xYVa(!xdd^V15Z{N{0(3<+z-oLgV44ICi(Y_P
zApb-Fe1`xi@jv;s98Gq@>_GU;|H5I>oRXMwD-kI(0g*+c{Qu84|4;VR#uvai`I#2qc9JGL|DwrUhiN=noS!B)3(!!@`hW4J
zG<|a?TDfpj8G!E_|7UkURs*CT_rH0R+9XXs=zrz32dfw2O(=t5y%eWGAp1qpaPR-Z
z8~>MH9aaOxmtge+rUDD6o*y%yaLp_ag(-@(`u;HpBIq3#Q-GA9f1HKk0k9040pQ#;
zt)1vg>r!&3n=6G+-_Y8}H>#jue)PY3{;LPw9i9`oh7|~W_5`h6q!-fi#o+dwoS&117sI{c8*fs2yE|$9Bg2i>Zv!>RCv`rWWP}cmihnmM^vWTdm%)G^v9r4noM>dEI6W`Kt
z{Eqzr*Q5fDhJ;h7z%vF-8`d31GC2|dX$ivs2>nQHfU?ujpq8@l4l67tLHJguf&+*m
z2E0cR*WAP&1L=lNQYgw_L$#014*FC%964+OEd+Q#=#CiVC%cs-pbNvk2j}0eJ%Gvy
zCsrX(DF02*K?f-An_GhMZF~AP(2l<24$shT%?3*t7~$Mr>GuIqz@d|&NvP*+bHosl
z(Im7M^{Y4ziJ;!ro+LB$?`yxSLB&G9736`I2blEN3UOs#1Plp~?_WpV(4R8YF8X4)
z2Y)j(d|2N$dIke1@PyR^R6bKGigi(cVJ!OPTEH070m%!18mm4I;(MG@(%jR4C*AFs
zQ9xesFom86`3l?ibNkpN#UF-S&%jI?
z+QZ^x1EynP-=2-}@pWsR(av>UeYerC^FMwniuv4G@+}bgglv!jwNBHS@6^
z1GM+s{+vYA`wuR!LRmCcl>^gvncw>g=P^J3O&iz(J{i|J5dg?_+BO+^ZFWTa4Dw>X
zWg3Uq0nWI}F`_>Aese{vx4j9omS~@u#_Ck$bx&8^5<@#RFXc!?Nhv~3c#(N+?E0VsDh2_2e_B|zxS53c}hjwajoCi+p?BT%tC{>pYqFndGB?9>+cc!f)JK*i5)=(_h(;NyS9DsUf4b+hrF0a2TigxC&
zb4*5Fc)(-vIm+Jc{&D#J`*YJ)$j>4&PQ!o%^%HbP2H&$Zyzf4icO$?1CPXGl4XXdO
z6mQv#cJ6U-r}8JH6fcPR{2mpZ=jt{
z+!Hp)Ya4FLWMVq(k>8^*d}LK^7TUx5(bWjc^UOhi7?_}+@7lTv%m19_zT?9E?BRdzBI6KkjvHM6@gKxR0o_Ro)Qe=^m{VLtOXfB-PK;
zZco)Qfh@+QK8odHGSc))8_+*o@fw!<@g332k*6!(yI76wq$<_O7u(O*uUS=?u8q72
z#e;e3#}8oryh*>%jP?p|-vAr+kX|lpDduZO(`z%-FR%}X1u!Z^AloKV_ek(7B!Z`beyP_{gW>cK#y`()`-broHOb#GoPE@Y8b5PI=3qq?_Gw!t
zS~0wLL+M|{i_c-(9B^GgHr1ZiAEwtpS^wsBJH(iGL)3GT``{WSkXzZ=sCrw@m_dzW
z|9$_Ff$0tey{33}g~l?VI?xXcIH+`=eC?zB-{M|1Rn+6lD87Tpsw51I1MCQvr~INO
z!+RE#g*eZ^N-E6H_AFB4_~~$Va7Zvdh>ZY6f_7Anga+1QhfW(IhW>sE1OWQ4wh(2{
zqSZF)y~a(@O(2|U={q<|m@hkpP3W`Sio2;U)4P5Fr%CO?6}z`S{k=0Epb
z(W1)ju%9`l*Pli6pNv0UXysk9fwE7n%8IHl%dT&fU&?x2M}^-#y^{*RX-GRyvFzzp
zDxJ$-kPUc`Gjlr)!`!I%6db%q#s4W0qQbk^!FY8}JF|HrSrZqVZIU^PPc7GN!i4%=v$0kwegOyKVnb(Sik
z;K!-)_)8`6F@L{1cSSvsU7`F(tEiqNuare_*_f0r%d_SIO~Y`7e9f(
zca3D?*8JHeyC%YuFtK7Qd@p!sr~Q^oJhjC5iUARAXeO)NE!XYufqTeRx{EBiU_H0o
zzw=;qE4j3(g+H*ig}n2WE&7z-O%7E2do116NuIljI97yq5U+_93%|7C|Ec%0uuX63
zfcqrmf5lFGAzhDZPD`8jk}N4T8|I!aQZ5seJ3avSm_Jk!T(PT@cwD~WyF9u
zmsYv9k-g*U^NHkG?vpnEw+67BBa5*rkHbsTe*dQE?HDsz3KR9
zA^Ey$kog&RHWA-5zW?R%a-w?0{iR=f8GIjb?Ae*SrDUXEP&)5=K4I6@Z)}&(AWItV
z%YOb>0)A@F?@eY2nfK-nN#-jhYMkFw(z)M~QjXSgb)#$&^!WZjZ)Y+5Pr&U@CoPJI
zvun?XXI_OQA+#iaPgFjcx7Cpy^2s57Q_N
zibA%#W)O|X>l(hp|Jo!ddb0ik^bopy+FyLGv6-py;3?
zW;&~0dS;Qf8=Zs3M(N~BS0Vr582Epm^b2f}%TtLA&yb13xm5DsXN_1(`BakF*ddX2
zDV1ok41EpzltR1$nhx9bydt_v*P2r-UXjQIjeu!|G&0cLd1cfc?qx4>i~k*;LRu{k
zh0p80A|-orUw3^;Ch^X5MZ9OzNR;Q6EGf9hes)dXlaod%gqfI=11?F#lWWkZjs7L^
z`FZ&eon|Uo5vSU5BR`eg%G6Q6boLdIY=1A+;F(B51y^M~(SAX=`{fyo%~OfWo%oNx
z@1~OQ(0@C^ijs-g9*%4KXI_xObIZr))n1SrS9bLY*}o!|E3Cd~{CGt=WhAHb_NNlM
zxD#7XM8p3U88tsy5KkwH#{!m2KO~cjCZ{zzLSK=p(6j05H>VKS!O_&kKDZy>N_eZ8
zF5I(kZLBoglSp)2O1a+cN+eew6;~Z^enFnOUM{FqNhBS&1>L^xOC*6o;Vu1{32@Jj
z)Q!8L&&eq7vScCdc=%u3b`Ya?!z0R#94eRQX+#?-amFdRWgyB
zTPTp3kx3-gzNwX>?_LnA5aSP__g|7_vh3fu+L8$iPYT}`qd4;8Bj2hPugr
zuhHb^vkQm3?nS~qSlxUvdm~8;%c~WwsrSkKDc-ePSZ|Z76UO#(dwgMkQ=3_p-?jM2
zYL23)=l;dso`U9|gtExJP(kNu>uln3dA8AbF^{A_i2b?mNip#$RT}s%ZVzid_
zW{^LTCy(z|$|PTdM;?C8&mlQgpNklF7m|r>du@6=ONf!}c}2InWO8ozk*=UCDWos$
zsOwa0IuX?WI+@d+LpmRKN2CqrlfU$dGWWD2iF37OzWb$zq|S5PHYg#UY`7rum~A?h
z6c}vgIaiuagwFjNFNwA<9&PwqbK{UDnRf6_Vf48{l4^D(H~T&&dEpU4wIk2HrD0j*~x>-D`G2&RZ-K7@tSHP5+Jcqb`kCY${!lnohTtS
zTXbi1gUg7C*xjhBvlV1)UTrGbu!@Ao#=w99|5MAy6WrkcmgqJv+n5RW8w(4byJJ#S
zMh*pW8ueLJ5aGJmYx4!^aDU?#&Mk|XM4{s1Q|X30(zvJfw8r;hqU~~5%I0$^sTWfg
zG}!l?$QMeyu7v-UXsNu(eT*lQ@Oth|Z|f)^I>z4$w@8(caXov_KDeJS=uv#eKP$LD
z@mMxpsCxsMEz@RBhyN2WtkMXk+uK5F`Nn3QPgRj^n-_}iX4R0RqRUUU>oky8pRR3s
z_Ntl0i=1)LhyOw4pIsJE{J4TRxF23tv%H$@4s0oI3V268-IP|_(%Ve3PR10*z&!;_
zx%#djHkA?SraZT+Jr(5jVe?=u>3Y(8qUZl8I`4R@-amjRl!h%MM6!|*l8o~_c1wj2
zkzHm+kx|J?qC{jT8QFV}^IUsxLiS!o_WGjV{r!9YxcA)iIbzf%c4Ijw@@O=@6r@Km3wKr_tVBRjvj-3g;;qXK^uC0D3qHkC(LHC|5
zktcBKWSpR7AJko(5Dzx$hlLX>)qRIu@aepjv^E#9&YQ1ynDmtBWm||-Z;0=Qu~`kl
z|1}K2<~{BB!H+~AnA%(^c0lacNqmVXOYDOICTH*76!--VUX(nkb{`~iFWT3ZC%U1^
z$$yx*M=!K>=RWxFML!IQ6|plD9E7YV^^d1`4#CgcCxsu^^}$Ct^Tm1QeuxRke7AW-
z=v)$=b7Bxq-nn}_Txk%N_B{x#zy_faC(G`p>mcMc
z_!p|BGz3XTItx2Zhv7ZDJ0Gl>Mqp$hUnZ;i2%NSrbW?vl1m|9!u�*hVk-|WZJVM
zF!|7WzHwy~K2(~t3bP)A?`Q@{e-W>9;)?C_%TE2!=ha~m&p|)ry7y9eW2hgx)+KZe
zzwU>oPWN5LiJxCATIBobJ^;mUX;~bl_ytp5i%EFQ3_yWs&s+As190KC`XCh&JXPUHe+zK#a`B7bPQe*^V!r?8HX<2@}3Lo<4}*PA<&0!
z9NyFyo*#AZhhA^~QT*ZUhbhLBg~1BM`0Hg$07XwPbnpLSsQ!~!zY+?JpL#n0Pyf36
z^23z@NTQJYmg06lalgd&Szg^w*tNNdM}8s3X{yP8eSQtX7mu!xdWQ@`1xIJAm@mJG
zzGa-e6V21(A80>>P`)UpMX+|N1;$k3Q&IrU`OA<u1=UqoJYL`0Su@p(M8milO-4{}>82bT~3gvzweZS8i$&?)n8wUqx5
zjAqf1m>c{BL)My}_<;d!zX!bHu_
zV=YAP;KFR;D3R~>Ye(CwiOAKK^G{BRu|F
zhl}lKFR`C(-4ROUi>T6wQr->igiqqg-||cldG2yy_S=@tP&tiajbq>^T)do+*#BNwH_dTIU;ZG
z%}M7?X3;v>bk~ZRCb$ZE#wm423lZSOQvsG1iY>4zb)-}5O(V?BO=M5!sfEHRyUW%@
zF5szPIjbCEJiPSd$|3E&R(Ng8--_M85r!CR*4*Ezh7(=KvNoO1VtGhx=YLzcx)Vq8hf)Hj|~
z3{`V33T9ao;|MLXHiy?4kgd44)R-+EhJ9r*%L|Kwu_WBDn+avGL5Xpo)29^X3TawG
z>p~dmVZ%GHnGYXS%Vhd05IqJ*r86_;%iy~xyfN)}3A~5#1@M4EI4d8zM=M+ayVL5L
z>4|=Ck;T;rcVe9FdQPYRgmykW#uz}5`kD<#CtBZxKTL;>uG2dgHPRvNzNBvEVj7XV
z`0|w>JSIMvyze#nFWvM(RMBa{d4xRDAJo8`fuoUAvvZ>1B@
z{lzY`E$A0`F#2RY=PqRs3?!S{
zN`RqodFerw5PcDR9W;75fFmEiE%tGzeVGZ#tfAV;igf4_ob&k2oSU>-eWqL0KGJD)Fi)2cHFH`fi0{c@S
zt%(UuqirgjiLZ
zn+W@rzqaP(hQhJ&|JL6&MnIPlUh_TcuTZOG`#Q%!927s3)+(o;K`Ue0G=U=em53&Q32K7P`g_KhVyA+PlTq3qPjA
z=xyp-nmL(JUfILBOM&PyTf8_hB9#W&pPeQ-3{QoNCVjeMZ;5`@s^bfD%$YEQhpzRQ
zPzJowA~1IkCBYamY0dG0AJA$^c&uV772e!id{eEM23zhEEWpkJqk~2kNImtIRu>j}r4WRzG7;tJECmnxZ{BoY7vm#$c4V&!Z+1o(%^SagJXQxav;k!oeMb{xFnY}OQE@8zS7<1Qus|HdrOd*w~qIitsQqD
z-v8S(k1eH8;=p|08f>IsWo$&?7
ztsV?1gK<<+e7u=B4B?WCT<@-ge=fz!8M@WOYmWvWO4-&yp@9*{gtKKZud_m&=ROW^
z{ddKsFrgaO83yu%ovw$!NgY&Dw;N%LXUhfJxKe1hxOan+v;w-jNd`VHs)ljd%L=j&
zo8a3v@fE+9&2VVwb@Ea|BhJMTJg>$wmdXA!mii!*UL#xd
z<7GiXY>6xnalblM>0O0a;l2Ui`cGL>%`0>E?a8vfH`!KFq|0Sfx5o|F?c!vgW<8@d
zyjdaJpmHWE`RX6p(I%JgfgVv66;|3b4|lUG&kpYSP?z5rJ4c*9kQGp-8I*q1S&>I~aodf#
zSGH1}>7>-ynry*>=I8KWCF~NZ)Em~%v$7V{1noAZvWge|=c|kQ24xp0t6IXGdMnt2
zqUZa5@ZdDW|@a@JYHbRB<-+#xBrLI?S@0!m<
z%duhVg;K(78mI7x49jx=BnnLGpJdX!{CR9W`BXNc;67Hvv|m6^A&Pm_haTke@Zthr
z2Jh`J3SptIGjr5`i{d;|*=K^(pWv%lHJj1Wee7*Oge0eu8Xow$=S>o(h4osBFx1(Y
z;-4>ZFHAXJ#=~Dl8qY7#;1M|n|DM!wU=D$=y?5X9;0G(e*>jYSV0lzKLgIaHSWw}`
zGY>9b$30^GHh%oYu!h~w8@Dbh;2xik{b&)^#zxBxTXPwLaEqJ#TBoQTvB#Gq%t9&b
zaFLmpO3{>#SoY4*QwmGoc!=S&ePvx5_OtHKo1+Z6*d2o;w$=YsFf6luu+~cl|4_fK
zHm7ESAEimYyc=1M@i(WJFJ^>brf=5^8%AyMH>Y(kkVZbkTk5r!xZbB>gR;paZrMYa
zv!%-!i?cWJu@n`o{pAz9a=ORH$NCLM6?OQ}iR&|7MMgn->_ZzKW}x~^>*7cJScW2n
z1M53Xx3=%soNgfgMw`ciR?izhCF$6H&XD-G<2~OD
zg(-^Luy>(cfYfArHlz_X21@g~S=W>vzPdK~*1l6kX*B)a5wD!1Q4om%DyF(y>|%>__>LnIEu&ce6#e
z?0XMzQ-4h(y`QrPNBFVD$@E@M6U#nZzLZcAENq#c%FNq9LSCQ=p>0g3$U2XlU9kmfGVc0ne(r!09p2{uN11y
zKyTgWSQY~%Fv!Cmdh3l8(i&ecPpFjzGH%8x68G*P%_C8)OzsY7aBGhLQIRHanP;uK
zKcj{y95!C>rRf3JuvhQQf4@cMk3Np;ipQZfUJi|2d3&JY%^x=b+>!K{@uv#$PoUKP
z!vljBFO;Y=(3e@8j*51p2zaqE`b)5CtIc->_ckt3=V`d0sMDDieCqEJSy{-XSAEf-
z{Z(h3`hYrUr0!TXl66Bx$4pu480FA=5=GvsrC1=<%GPSV`UREhms~A+=>?({jwGzH
zIifAa8{B1bDL_7R+=08I6#dB&+#jUu_^Vn^ZtR18ZD8-@)wZb?fPPnDIKs4m{a%ttwz~h*~x^-DkNxj
zgx#OV4WzM=XVWkL05dk=v|3ID>U6okfX%9by_bEnL6()E`r}oD7wqm}CW1!hqIn5&
z`PuZ>ska8K&1l|U-K_$45(%8RZ6O-lz5YI1rvQ!VYgzun+Ckw&7keVuu*DSJ5xJHB
TO06xPB81vlMR4JUU;(N>k{s#0U^it=`nMH`!L@F
zNW6Siho6Xm{bgnLd5mEyGfo&8z3|B^^s~PN*(Y6V(A8}Bz
z0QM$Lm|;|=e_M(11oOc`+)tefGFz~J@6A^+#QOVRR04qnKVNM>k5NTmMRGmhT<2gp
z?vI&!XE%^|z!L{2m9Q`Ai|E06>sA<$cutGXAn`h+=pM1pa`yoer+qF!GeOU^>IfN^
z#8boM+^#FDZQ`nv@jTeC_f}c=0N!qcYCY!|V3rPtG18nB&S#P|Q=G
zl*f|MM)dcs8^!0HE7ehRQPESuO;SX4ow&L67&V*x-5u|B4`fL?gG09pMhJGXtJ+CRFNl6CJ>
z@>fe37SL<20(SxU>>2JDC?|um^CY73nOI?3L%Pj#yfFg*&m&0Mv))3K$XdZvx#m{X
zemmhXwNKudc!22r4B{esDUo{!ntZ3`o%k8*`d+IGWFFm#(Ixrx2v~MdPtL-=i4@uRaX9@X^S9!D54ryPz-glI=L<9x
z@|kKwxG-)jwtA`9ma6aO*nV+IQjc;H1%
zAB_L59m&S)I=)8K8q1sPbz!3hJou<2@ry2&_T5;1=QFhz`)4MdxusYhG%@AJnDwRK
z8RG?&{jjlz^HLQiE$rV6qIY*;Ia<-p2U6ABXxr(
z#$mM$w{U*Ln-Ww?9|W9`+-8Mk1NIXSaX)EIyLaNawOLpD@ovCvJ8r$edKxk@PcW}uVtBX?$K#b8
z>)Y`26xk{~3L=0PER7h$Ag+5>$M3{E|0BPit1+O^m!oE)fXvso6~2S~y9_G#EP=Nf
z^>(M7gLGprwb39zS%-i5UXWj~+R2Xv@;kpb2gO0WH&<3?*M)qef0
zi0yWG66miGa_^}I-VmA9TS>$9CuOY0@pM$#Yho{8;a;=1xWAlEpUpJ{o;3EL#HIoC
zKEOvaGa=o6RC#VO;GlJJ#vn54`>Dmp4vMQ2%6Yz{4EOOZKj5-wJVCrm9YA
zJ@6%Kt}Am-*dPu)X;fSb`VZX_PahYD`wgd$IC-l9?Z4pNb)HgBB?^Ci8o+n}U+Od2&^y}41{-waL%li|?>!92g`E||)
zYk+T-cD-BO
zlK1z>WYTEpvSYz}w*A0|G*cv`_W;jy*}?qBwgs^|ZMyh;e*^lrt;V?UY8&!Y&ut#M
z){f$h1>d^B(jj@!ct!O+}V0ky}57sb!wdnQ)-u%&_k-l50b!jKdFQ#w-_}gP+PDbIt*|YUOOBBh1%zieC~}
z!g@O`+~RNn!TGH{+wNg@=%)lGgV}K4+kah(-}KQ?)&3CoKeaRza%)xEhDjQF;X{-8
z8QYE|Y;?EZVQfdH6{lzq9@7xJb-24fy$!VsB}H_Nv>~t9_NbyD;1eS{!G$mnuDv{}
z$zn`HY~}x!;_JHV^{-A80}QV_dj_?q$mG>!uUW6jwcp%6JjXdnJ-yIwp@m2>+D+*h#
z1SmFra@%^a^F^BQYwV|vJ>
zx}QM#9{Nez*dOlLRl_w8?J#$pdnX0+f@!HowL0v7CzrlH(*gNCxt5ZLVO~7o`k25j&#W59NO7ylWA+xaAYKo#%XuHG1*FT!@3
zvL)hn&EI?zZw}bB@PHYVBgs>S?LW=q9*)nc
zevVu0!sXwsnOu$SP<+2M1ovlcC9C-a*mi=
zuHe6hepq2UbJM~IUy9!}yBDVy4m{q7vG3&Jq0|)P9f4$Gx1uF}KnwQSd5MH~
z`sq{ml}zNth|FeQNbJ>V5hr^72#5r>mYzqV`d#futYA{@x({Cic6K=mqv<6W({~gz0juulU)R@Wj8-
z&}@)p1y7Od{T>6Y1$kRMbT8oV^Az8o5&&8Ya?Thlu_JJrvO9nWnhoh)Q@aVCEr)F^
z$T#oqAXq)}gCGmHIhCISG!*hp#&!^Fi91c~^*+T)|5alR
z88@d+m@c4q#$lM~U74{U*LlXVLPx-J2p8WT#W5`(
z#Io`m0~iS4%?5+ZFu(n(b`u_fqa(jg^)Q^TWFN5v<4CXk04_)Hu_)ZZf&J%i8f@gi
z{^25W|Cs5*E@hmbqg*MEdA-rWFSVxuOTNJ47l4(^nto!OWZGtp+vfydf>Z@$6I>^u
z19;eb&JeHrz%JJ%*gwp?5a7jVaIio_5HLKe*%|whR@ry4I9{yJ?l8plte%RQ!EwbT
zCPfCXzvIg!9O!^Ielh6d=M{RaE9Y?fY>ApL=Hp36{oK+4ImULgJO+%|-q&sjxWTx`
z4WG|(ziDo|2C}}Ur}tNoyS4PHXmC#IuJ%Ho8`2FUB7U&~8jQ?}J_SsW%GFf?KWy`v
z?O_J!Jsafwn+WM5D_JA$05^|h|F{q3eD$mk(g3~@kTS+$1U$GwYLF)i;#6SlzKTnb
zzyIsU9S=c&u5ixx24gKqJ+TsEH?9**FF}Lwf^l7!Zb_noA`-eaGpM>-Ol6x7I
zzj|qC{kOX;EKk}|`0vpHC&L!BS1xSmT2?z6|MGpS*H9ZexGd@iGH6FP4Mcg4t*0Sn
zy=J>F_K#3=V#QjI!8SB$8%rw%ewt4ga#hoALxPH_cC_+3^sqK^edoqjloZb99d;e!
z*n<)2MD2F8XaCIES+gcIU3d8Bsi+#1_n~l6|JgRQcxb8bvcH`uZsyv&Vd`t-(`4Mj
zmQ#)-*SbRTQ+-)dY$GW$Vy$i7&YskL6vjd^#h!df9
zG<2nw(ep)YE1LITZ8k04iuxqlj9Ove@K5whGVL$y(+BT_iC=3&*LiL2-|uWjQsE*(
z(q2u-N-mh4XL~!6HK=vf`%FXj7f*{-yl6#&xx3`8E;pi|S7lzihSbCFPjo&FPq(8D
zjvD`5Tiejby|%{9Ssdf6MFy3K;Nym5e;?kJ-T~CD?Gn;^;pMS8hXpl$f0u|?&k%akzT#{6}l*D
zJeG5|7B$7cQ#Ln-@fr@xe5Tljw))iC$4=1@Ye6p`*Y8^NEzUbd(76h+ZK>%PoTx=s
zjjN+2W?E1aXGqm2AsYJOWB6bR+(%eW8-85?eDXJ2h$rSNHuC}J;$eMC-xdC}3f5~O
zyt&}88}NVQ9;dil!83;5ib
zEBQsRUcB@UIBp03aOlV-fjt^<|0C3*D}9cJqA#q4y^Z89o}!jSztKTD#Zj4cNtE
zwb`@<>}|aIVQc`}Q8d|*I|ubjNZ!!t2jqF$*O3e9rq)kpw16ku1+&*$gPh<0urv)&
za^taLysvNLnbqdj1-}<_(y9^Kb@<2W<{sejXJWaRO@V(nZYP+x0&tgK_~sk1UzS$e
zme2+`(^kbW2zrY|xK`l%Q6CR$YB@tcj_TF(d|L(g;dWbQC_($PmT^b%gJ1IX^lM88
zzJKAU&AppYUW{qp0sNfa1O3|CePGWK`G`2ouiNE6jBf&LEv`I*<(%KAl1_r&Z1IB&
z93V@1bVg#ns)_91ZUo$2(mIag;4J5H(InAulZNPle&Lc@2*Ov|S631Dl^!gMWKVud-+hL4~h^xivTO)+n
zF)x2Da|RCRp?@?6$n$4K??Ps=-U`VrHSBMT!er{kICi?}@bUtZ0)zrO1|
zF2i<&vzf2K{T{b2Il_a&3)GfjJ5&Z2S73Xk-78jNJauChh$6JxQ)M5y6t8K9C
zG@d~E>!reR!h>B-y8zLI`#o)?kFj2ik(L?mH!05xzBoN)>0WM(uTstuJ92yR0XvM(
zx4VQ-KB=xIvf?o<(vI!6Ye>7=_TmR+z%R85lk^_0Z-lqGJpW7hY5hwxvcH-thwT#R
zRYl|zy)~CU5k6SX`H<|hf+KZxaelk<2U2cMd#QA1z7Ua2$~c{2d~1u2v?v=D4Z{ZeKF{ds5HQgNq41w#7dg9$-Gfrk-&`e4W|P_F*=ri<|6gQ+Q;Jl
z9lPS}i{-HWR{A(Uso9;J2YCPG5ygCQfAy>$9@kHnTsm0qV3d0X)=O?(N}dOFT9x37
z^$u{KPr-b#v!?z$&i6gzHHR^~{&O&%ZxMz|zT$NwTe8-7FW}w2bL9C-UHkk!xcvC;
zqx`u1s&pR(t9^}Ky-K6mrQ*AJ6!zNaus0S)o_(tfFUZEe;tN(EpmL(1m>NVe?PP1^}DZe@(w;HHp%;9
zp9;vd$nNq@z+e4rx0eA*D~zvO3E0Kd-FXL4m{G*!9QePtKRp6-z`x~ba4lO6=_|Fm
z$nz{2OAH2ZKMcKg_^=ZE#}}#d%Fm!){oU&E{*XQ@UZ#iFd(!zmuQ8w9)U0xTJNTWe
zw-2?Z120@buhndY{KmHpmvx|?3#q{uR3ZNjpGB}B`1Ob4p2pW$=}6!Pcak?;Z~V*c
z_C?T>P>`}%4t~z|hy(1zEhu&)siTkilovi8g|HIYqiXlU-QQ1D+GieD9E{PYmScg@f2
zq8q@!kM_NIewv0>y+2yJ&8{7B|4CO)1b(ae`Jm8N2bd?jJqz;TeC+691AD6Ftt?3XRgd)bP9D>wMYIdmZFuajLY
zA}vU>TKuley$JMT{piylokeKdp3P}zcXgnwy#XJF`rFY5p?iLg=Zg_9`@r7lb$8H{
ztGDHnxGK;Dx83RDCtXPGL)?~Yr`pgrccJ}q=d#f6pv&^bRVk=D`b0rfOfF*VdQkSU
zBM0>f9C-6^GzZm`Jj|W7DM$JNWAUl6RjBVP-=j74btp#j-ZiDIPm$Et)y(GSbI@|Z
z;_mg#7O~AV7oF+NLG3DQYsVaO(fN($QIGnIP%{7VC!+t#
zQB1a1z_P4LwD|b9*5SXo$no?!=ApIE(Y(?V#i@PyNPw@@$3(Xnz1sC6@(5!&@{v>#
z5IB^9WV{-Vr)ED#-MU`qo=oH*({DdT+a?MSZ+Xq3
zkH+S4{=${$gy6q(o9)Zd@uQCoD&D_D5wn4}Sr*lyt&2xFez;d5*MPH2_YT)0U#Syu
z+7m5E`n236_gdH|m0s*WEnSN~{s=E&e_f80ery_cAFM>{!cBONuY8R(Z}U$Lz_~HQ
zjg>}^=BrWbM~BXT@frm+KDuiDui^P2!G3Ow3dFx;@a~ES
z&(RCT=wC-4rK30`=hXVR0!3@H>h_t`qc6f5+3f5kD5&oFj+Od3=*+b7@*&+M)DR!@
zW4m1mVsW^}&cI-c1`-0yNe(UlN_@vy1R}bZ2U&#Ms
z`PcieTTw1&RYB)s@E3H2k8hs={~WQ&Jjf2`e!OjQ_P|psUPyHM7D+>2x9=@XfpRTY
zF8aIlG~7qJxQ{*iK@CdY`k^G0yAiE?o=V`g4*u4C19KYQm%ruhm4E#U&h;Y_zU=`XyWgaP(**X*
z)*6DUs=$|T-oN&ZoR4SR?^FSQsd|&A=_uISyYfJc75Le?(B6$3!C#u&RQTHguie6Z
zc3Sd0_^F7V+}}|D-bw4IkFYQ2ntU7a7WQ);{W~WtK|l53*=3_~a6aIr@@)pLJITK1
z*BI=txf8DnGl8C|O8aX$*f%Kemh;5t)(URRuQ!0beBPsi_24(3t$JUH&qe5R9{R3;
z-&EO|93b5_zUrhi=xpnKdV)(+4YoErA?hI-d}F5Pz&?5mTq+J(26Y4f{YfPDJj
zh`blrxw8AsDPf2!2~U^?c0u|o&ShP$pm&9Jbv^#QNQwK-32f)XPDL5aClx|(yI@RR
zazYF9$s_xw$n#0R5?R^LfV^V4sU<#FW@-4Ti1j>f3UuQ59^|M%_%7G5<}%EyBUj(h
z$MN^T?jBn;z%XIGLVQjX8EzJZdB93hr|bJcul{yx{9C}{M9cFgknV0q3&!zZY;494
z-v@~79O?Cie1kZ+l>s=kM(y(fNWVJ%l|1)3mErb#4afuSe?MX#;rMth;alBi-tL&6
zY#XCf=W=Si5?Freeb57=*pSj3w!?1`&sr=WyG=c36dKG<_+HRFS2?zSNk<4j<}1VX
zhEK45=o)r-00Pd(-x(4f%zkkPykP|5YHvpm*7GxcuoCyjpt|g5%y0H>jFrc7Va7jw
zA;1mouZ(cHJ@y-iae9=vkQdhD&2=8boYVNNCbD+%US%vFHPQ0H
zc9hu^68;z_T1(pHv_h2d*W{9i>dG9u?_uk;>Myz+kYe*Jjoqsv0*Lta1
z57v9W>j=pY^-F{YjbOk1+2je<+np~##=$pCa1Z8_*EW}CV;(0ZyK@Ma>!fjY9qy0V
z6Bf?+y4qj!Nmx%psH#f`@JRMMajbW2`8hY-e=Ax(TjKl;hC?6mb+)@iemMe`axve)
z*RRN6V#njrv`2d%o~Ot2Yv7j%un*jqyyX&LOP#|CT#nblZB5udwS$SIAEZl8e8qZA
z&Gtv|Icr*HTpE^H|CXxY_3&W3l0BYpQawAZGXcj7Md8L9tOG_JHY33J<45xGIjT&Q
z${60yGHg@jt_RfYQ{ae%`Kxq%vlL!OozgMp_UFIg
zydh>oK?CqtHR*d=Mu26$k>WAHKU;W>lp0R~Z=dz_d9$J&B`d2xE`W2BGJCUa%D`KV
zc{6$?M}UWJvEXio^RedpV%C!)T_}YmCQ=OQmA#wSb;BLTi&wGSuZ{-%Ln!Ey
zo4A#`5vv_Tg>Nw4Km0WGuz>w7zrNpr6dDSW{9bAm)sE&As
z_mE<%&HdNNF+O38uP>@>)mDaQQ5{`KvRXIo@cA}$q+wsGxj+^Q&326ck&Pa2lr`3iPeSo3k8c{6z`0V1
z`z?OX5~TM0M)RC;Ec)J3xu?4P1v>G>ZgGQ87-IPTs9wDN1@dGFlaiB3LBjjCv`WgP
zq4(Vh*^z9SNcV56!dKNSG%w!#l}j%TiOVX+%)d=R3J>3P^73b+{^zvEb+i{EgmE_;|58Rei>y_Xz882p}vnA-H%oE=m&NV0~OE{uu<8zch
zcp%y!AroyKWaFZjJVQ1THjY=O3J}*ae(T3hmFSSs%X&W5G?ZDrarYvLEVS(Q`K#t~
z$%x~EsW5Y5Hi`}XkuP|-2q`6;?^!aPj-LD(-^uHgh2oEv_-$cLM_kOlaZAsE-RrJ}
z_WcHZC61AVvP|S0w55MQA`_W;=cPV8lY!8P_K^DTH1zKMO}F$@87Q<>dWb2$29=d)
zOe-P7N19O&hcl?Z5hlHK{>N!y{(n#QH^*<1#dkvDc_}N{-YHs
zi)}8Qy$F7;dR8kW9eCM9>%khsm#A#T4n-F8N@ViMR?p7-H9D61Qt)mD4cX~mnCbz&
z9Tq3vxU7F3P=d^4@vtXkl^7-9#{{b@r6_s?ybehPl>eP@jttpAmt^7p(4zO(A5
zplm$w7nd*|lU(2hE(QrfW}Ofp7w7bj?}PR9yNEG&CEQ2Ylkrs!^h{jW-ZzH*V%3!X
zTG_7e}rzE(5ubjUr;L8i=jkZe~TU+6N!@!A|(6Tb{SM~b>
z*8#sv?mU~#2mGuw{5$s<;AI-Sq@04P+tA8LvBo8Cz?;0@-Zk-s`^Z)&E3W>dA@5&K
z5yOJ8AFuj*xCHEUuYBH%&n26O_Jk{d|Kh!?#2Wy9BQR*x9-lYLF?{$Q2T^x$p}cQ!F=WA
z_~<~8V9in`n4dRgQXTU-Gh($ebiImoYCrl&Y=3K=&{;JB;S
ze$xiW<%v?OWmukQ+p`8gFY)1J=sFx%b)!DV<2YP)Ly!F4wai7K0jEcAtoOptA1JK6
zGl-u%7aY|i@5fS?bB@OS!f0mv6qno2`bh!DQTjRkqqzR`*PA489QO9Gtik2nEGV_Z
z@%^6U`@7gq$EHtrWdN@%)gbpN2ddtiVSR!2J8+N({l=|C_|k`+>j;10OXuH>%P|?N
zBkw<&s=OhNYkPCa_-2fa)H6&i30z^#ks2mZ4_y{73so0bN|rO*3I%%(p*Uoj!=q%Qm!B
z!oejxSK_o<2$z#vtfYgn;@?#TJT9W9$p##N%PQ{O!sUCdyZa8Ow{I#7#r@jrGkh42
z$L-O}M-KtUp8e*5`_o*_dKb21Xy(#s%!kVL{-yZHl(=n*WZ3OvkFV1!2dMAZkI20kBQ+NHVeQ@8!C;6E5El8L2ITerVy(_Vn
zy$15zOrGfCb^FEWu|qEK3u~6AN&I4&pHOyKX=U6*+1Y%cvt-ilLMZ!
z`|P5GXK;PqzbmKeK~FB*AyN~{nXlj4WefLv<67fdUVuOJvJKle0{-f_*rK*?SSt^#JjGXe2q39P|DwZw*eLptY-KKfdgqk`eti-6O?oE!dp9XR5Uj_kw7Zl23bL1}|p(P8n)NKe>y-^1`s
zRNZxmBV}tQs8uB95z_OEC>Cdz5{y@mM;%H}3w5Bc|y{A@lN6Q`t;3XL8}9C1)1W
z6l>k{^LQ5eL#xQ#%$^)6qAlEGB2QEcC|HJm$P$Hk!Lw#B?0`skFy(pmBd5D%K8N
zC*fC(Vm@s6b`$u^3HPnt$_zAQed={%c1JpzTIqX{FE0ym2PDLA>J(+U28d1KJ^pBlBu;(^RK+iWA!6iQ}K;e+QtoIlT*-iH0_moGmA
zp!^F4cJXsy=j4bVeFFH4jWp{Ue6F-ajxFsv_@x+Hu#zYEabt}Z(@1F7F)59bAn=zO
zLOkC8gQe?^2kMLBiAqCuDm%&^p^|WAWUml1l0-wHWTk|RmSjX`6v{4}geZGvMk%RC
zMF~aech2+s>-#zPo_F8f=X=k+@4ir&=T1!Ah5j8xR;+~i%rH;Jr5DbR?|Wd7?F2lp_LkFAIam*t=T}7O
zz&dKkK=I;zn1{b9R5xx40={$qY5GoB=l+U0+${t1mubEM{QleZtr6F``*M!UpndWb(bjbVoS5vtDF)##4>%w02fg;BOlI6iZdN^-
zkNfOnNi|beAV>cSImijTkm2>Vkt(Of9_g6B}Pc(dha0mF2E%a(xNu}aOy-yxs-^mlAH3U(#r$;J)V5PqveV9pqx
z7kRx(=?}=mH^s~_pG_BN_aT4Dj2791cDGe(oDS`6$m+rliYE{{o6{g0dD{r0oj=+C
zp&IpvL=6RzmrZ?n3md$!Z*#N9cmUF^zP;iR?pOIJ>*Qd4Tw+z35Cvpva*;s&*1fiK
zhfwe2oeTKAXWkk1cq0feX3*3|z02JyQ^?OU(}V@@K)khdp-mScTjJLOeF!H#x8Z|%
zuxG2ipYsAv)<<|Ff0=PU8IC;TU86r6;>XWF2QdFK`syR-S2=Of{s8I?75saR@hW`B
zQc!*ynSUA6i{7Ze3;F5(aWlC11o}xKrv{dbZ{r@yueH_T9_2^esXlLS2;ng$3|y$!
ztSab@dVB2ZGBF*&(aJDP-<4N+8Twha_$l+DUYXp@7}Wa`&gY7HoE-C-D1Uwz!DIoL
zdu|&Y@)!0~=Wk*?Z7~&q4O!SPTffmT7BDKyUp)a(%bL2c@6Rl=UcMN4uBjP
zbRrk|t{wmM56p+>C%Ns{0lyjVw?w{>n|6TT9Auwbk4O%{f`Zsc%*WaryE**;tz}3M
zo~M>L{Xq%&yku-M+aq|6=GeDx8yGKxPh;8qfxqlL+LivC0rIJ@boD2u%X}(;8T3zl
z&j~&bdhDr#Zl_>;%3QHM(#Qty3B5lHKTaWuq{JTk_s8X(2f0A6!#GEI9q^s@xayA`
zz*ipB#%zs%_cI=Pd{+nZQ)%lWedIz*$pUwll
z#=&>DR)F5!Mv0q8K~F#K!h><(Q(Ce~-*sWVwORG@-QU(Q52ebooriU$#_7b{GH<&`
z;ndCM-)TK0N^0$0N1CD*7T|o=z5%aZ2bVIf>iOebR$>1NCyo
zMsIl++1F<*%P~)~Jk#
zZnbVbsWPIe3{sZABByN1F
zrBmq{QMmt8)o67AF|a@R^|3@TS?Md^eWbmF?B<)=?Bwu-DE^eZc0jy_98aM=CVsk*
z^i0$0oCqr+<|i%UUf7h8&?M+ct3zHX~s*<>~OGB4;jtn-5G|NFba@>eqv4&OkcEb719W^Tt(?!B}T7Uec{|1Bq*1_*n7}(2*RdYShp}$L11*!f5d+GhgH+Bf@
zP+If0(1}#&@6~Y$lOXHSh?~a1xLUV!PgA=h%m-3c1>XU!==LyHm%usnfxcgp<}lyu
z?Km^X0sZ}+4W0EaScerm{~Unv6Le>AUv)9;|EO#X{dfZW<%!`Y7r{RpG<54Wo+J3t
zQt)0;73`(xodY_Ba30NV|4fw|tV?+WHgj!)d7Woxb|#Gf{B5N?p`aJ+vq5JQ)Kj6w
zw^im4{%xCy%mGN}kf=nFCD_YIK0!&e*A0^g2Yi7qxUb?j(13Zw;M$077NFS0`W4Zj
z$5{Do0qx$jQ{L1hthcNldUU)8*~@{oI}y^o>X@*i67-IHUufuqaP#(eKc8tr{iUCD
z?}GW~S;PE1ALtqVUILYUFRqo3I4J-U$|TFoFQ%Xg%5LJ|NSn9S>nE>
zz}>$$Yylmu2LvM_pL93(tmpt_-DMVS0PiU*(19DFV0}Cp5{UNFRm62I+G!u}h?h?Q
zbCROjke6gA^V;J1m-qUIUZ5SAY@hmv=S!ki{FBTDz0Y1(U*Y=RyGsHNUO{_|9Bvd
z=dUmxJ#Gp(+B9ge8PM*=BVM$Z3rzg6Xdk0W_~D{v@JDRlpo{WP-{)tOIytv
z@OrwLKI&bGxo{ly9D=g$qn^XjB!cp|N|>$*U|7Zlb=2#e=Ua#2w2$vn&u#d9=@5oL
zk9oBV?T_%X)dFZQ>*7K$qrJ4QI+ly|w(Q}#s*8XQE5Dw_?=2G(V;N9Soygd;0baX#
zSQy()>z)2rH^3g&swfxGOFopKgnVN3E!_rYz!UWSPfY>&9N8QW0cH#Qdx8B&{*8Vy
z*85i}noMj*wVR`Q8i4H+or2zg8*W@KSPlL5P6_kDty0vaeQsfF8Lw?Sg^fbVixq0
zW3KLWg7Kp)*>KAn((63lDH{*zMTUqC{sFxoJNQPngWc~Ncy>D&xj&#M*W1gC
zJW-sLlfE7Dm-pw=#~jdOdE4^Jd>h=uF*Eh1)dbr4)92@V!CuBaiTZ5_-;1ivN{{$@H`=anKiY}lx&-CW)p=I6V&hbYd;l1JlO&-9GIU-EL@L?^uWii8nE
z9oTdE9)&cncG$;t6E=zW=p~;@=1{;?$Ld&^$NDeM
zcY5{0y_5&4rsldy)PCM*+p!kH^-kCH#*0$8?=;IMH?WihM~1H!DFA!NAK>HO)lHnX
zH1(h36&bMP13ii9_(dgH;Z*D
z*vn!@rbDe@FCFGpMrR^m-mvJkGZ&~KBO;e>Xmr&Pn$C@@N(bTl-qqh%dIAq@w2LbV
z35WjVXg=&{Q9~Xj$47;TJtqhKw4IFLoYUn!75&$K!nuBliW^Z;z~iU1&(`gEMzk_S
zWe(qYL5^&niaco2L7Mx;?Nnhr{yX_~@mf5jm%|a`0r^ah^l{riT~BT~$26)9bPz@E
z>qfrN&*Zka`YA>O+RUuuD}?cKL!l?f4ei8}C)>bYX7Zj6d?*e326yY9`hmTi*t++8
zPCkrBsd2gNv%q(A8XwcZy83RWxzkk`*H2Q;zEK0cD{Oy(4#ByRjf!#Zt}rjYJCqy*
z>3zSp|JE)@Z)e1{!=JzpVEnD-{9q;6;Ws~bSK7ch?UU6p0lTl_mi6c(jN_?WGd8TC
zw~T%`t)~daug$#PC4D$gBhfTnjCTJ9y{HtFH*o5zfF=Py@$#?o*T7!JpF2E>_R@85
zX8mu-*GA#@wsCTh|MJLmw3nsIyMMcZo@QHCFE{Xu6Gw)>PJ!Nx(B&(E5WjryuD|Yp
z)(fJ&4Pd9kc4leJLV7`wE#@n*K4wK|JRx3h=ceDM0mt-ZgIr+zUS7V-GzsLjHY>S8
zKrg>zaaA(Nek?8$(=d+}1smBJLH*FZxvPrv_!rj4{}^GO<*u%7y90V*(cf3%eD|?;
z!v#5zLuQKE%3xlry_Qvi_OfGegA304|2(!?puN<0Tk#O*Q7xLQbJd`yp2*t32jP9D
z>Ls|Y;}Ks*!vz?By`0Gk^osT7d|rT_4QJ71^luNIkMk-3)F9n4
z6`&WsvKtOIgWs9yha1|p`2?;sv@3ySQn_eX)YizJtpvT>qG^UoAam_vyn^dTZU$QT
z(E;xFAJay=c!ytaJ=#nAavjQke){Ec3&YjS`yw!1LDhql{cK8*yoKv$*@30=8y=}E
z6LB3M7ZW&aA0pCg$7CAdq)1jg
zmiKEA-FeKXNZl)GML=)+Wlosx+$D8c@D58#>?CwI(4{R^$z7=-rw{v%81JHg
z=F6}R_gc^^HJhp(-Uoh7nsd^Xus)Y#_51Dv>vEG-o;}I{xdRdG(!FHF@q^cTCgA<_&iNcLk1+S2a#etJ
zf%c~b9=rT5l67F8zqcja3s{}|wGHN*bG!Ec)ds%Of5e%^M;p%nZDxN}3G>nSNE>;s
z$xcG&SXgKP^8rUTXX0>+Eg*NS|ccLs*2Yx*w1+I&|2h0N)NfdH3t4h
z-9g5`J}uj=Q3ZVUC*wV%N}_mnc`bK(If+}ecDO*I7+Ke!*_Hq1~-g|Wy3x>@AaK(z15`Guh4L3cP-gISZe4g)kV5LnrtkF
zexQ)!%i9j~aH-(xD=u&z;mgIDgc$Y;vQ6rHh^ulrdHbc{AWz8?a_Hop^zfrqT4
zIn)y2=jUB#OP-Tg2gG*s(lrnnk9*3?9yXACNrnyFY&GPHoJ1V|#ahCCbYnBahZiIx
zpU3z2n|k7?+rpmlwuu})b$aASNj;J3IJ|m2%L|e&8a5L?RZEUC^yIiYJR`SsW7=Zl
zYRKJfViv;Fjijse__oemPkB3@Jdp?jr{yA
z(-UgfLc+6G_H-vUlGdOci|)vJGVbcQkJP>(bxB#LUEutMsQa<>%&c~D_>NquqH+sS
zo)q+HxZ6nb*<0)Pdo_@k=5JV2o_CUqMb3+tgFDEbSS{!CR;{Fc#ahAPkS5}`tWF@{
zZ39s@`e6|0Sxv;B-wuk(dq$dLJ%xOLzogzRS4-5bCL97U3tIVWNn7mvU%{kjBqNUh
z@&l_nvYOVt`ejZHAqUGX$~V@K29^kyYXx=WUXX2T{oHe+lw)hIF!Y=R?D;ELD)|i7
z$qelUw%|`;Do(xC(LhYD_K04rY9N7J{_hfwz90{oD)Yx6UGq)B>wOX$3D>;7L+$c5
z;=ImcEap@r@bI~nbkARqC*``UKUBUTO{Z=$-EV3kt?B#>x)vS8D_F?>GwjRp(l_0D
zFj-G7=(k9wyn9n4vj@2gCk
zPuv*fhJ)?&8h~|7eg#z!&b*^b2Khz6HG4_qF99JX$B;J}WYadF-3*GIj&B64@K{Uv
zOB(IktKIGpEnkuh;lnVnCs~ix&~86dWZ7~CaKC3?xi%ne-=A{ytIVWxn4q0s
z#S|Eh?^hRx>j?ol19hnHwhR)cR-^n;@pmlZ#o%c8brJANCR*w|Zu9EvyD;2JKab)?
z=H5tGlT0!|bwtQtv5UT}JV?Ax$Of`Mj*pvZ$y1
zZ|`=*A2Z985Z`#O)JOdJ;3RdvdFZp|4wMCM1FXJc!|mC$pB}ICv1IwiRXi*DjVQFP8K^L!P!V$dCu$Z}2H-4dR8*{?`%5lS;dh
zZ=IOuqP`DZm*GP3!O^w46PQljPt8!&qbbs%&U5GRH(nD5d4s9zVLboA!186!^=&Qz~m`G5pi7gF+AGlTzNzBhKOD%N_j@_{;k<|ab>=ok2sg-Wk`)Jj7?|Q)Uj%$>rbl$Y
zhcWQP7dCSBQNZWy-JN1#d<)+lo2-F+JhgVREQj)IAHGAs9pA8JIu^K+tDCsrZ$4uVcIc{Ma^K3iGV-cTB}VfK__=1scG&~3lMh}N
zQ_TkcQ3jKPHf>;MHkOC}VF!MHr{?P$nD@7FpHiTKb>>wE|IifRi(l;1Jt|>;``yZ;
zO1x^_q~M>%`yIfKUEO4g@!q!UtX{iLz08LX_sh1|2H=-gx55{j$fIh5-CZ@Eq$n}lIBv3$2)h|<>$81M#uDs$vwd5LKJUIP
z#g$FqPx;t5@uizwf1^ytvK8uQ@I=F}*k)4Gd6r$&x{-*VyYCm++)hNI6b{61f^$uW
z*CsY-bdyulIWzrmPl3?+w!9O+8i>9zP3^2hI~jS*D1Pa86DfPAA(oTUOq8;|TXrk8
z6BBo*@>_A8q&50Pl`PbM&Yop~;k8Y~mwxx7g5_|oLPfEh%LwiV@H`}=H4OV86`G=6
za1LAMyUIA^er$_LN(UKRxiI|Ueiw0O&J?n20>4J|`0}aoR$}vn
zX-fHU8!21<+iDVc#1IXOrUEalmySuF`fA-xMq@Zwi$1rIlvO1iiXU3Z>zAsPH*7me
zv<|CQGuY?Yw`*5lb%*mpl9@Jq$NET6^TpJ9*pKVm-j}oKNH?s{rk)PKmEFS^K6Kv%
zetmwu)ab?1K5~03k*V!zFS)S#weF|hZZes@`^ql3=lZ9HbF2eMD+{b&OkHoSG
z2v5%Tl62X@m*V0*@PC&-boXK15}jQ)mIn7aC~XhT6sqeZk7|v(t-E>&m*s@Y97_**
zX)7@N&;j^``-t&QpDvPmUe2>JtdC^uW!h1ByO(4;{dTY%>n6>IzV3B}acyw%^o#4Z
zU1arJiEow#aE`Zq{7`U62Py6yq&0JGC6gYUffepeq@~WG??H0|d9J=cZqU37?gf7E
zxA+_QkMxKeT}m6NPPa;mJJC#BR+@OP5Nae79J_}z9HIPmN^ZZRJBj1G-WtJ@cJgf1
z!bYt}%_Oa8Y;HcZiPT$oI0b5Sk$}4|{!PSplG>_T@zyi#B>P^9O?P@Lv}ga!SN}Fb
z({y=*#ojLR+B}#`!VTKpT=O(de>;hCy1F?qr=6%pvvpWr>LAmr)A^H*Hj?!RS1W7*
zKAklVtB9xYeO*bvFiUR>i5n9b3bpAVf_=B*2g)1C&x^;|)*Y!Qej9C`{LyG64Ihio
z<@dFdG{yI;wKF;i>+|m%1%*vSB*-AGr@D#U^zq2Lbh4cca_m^hH|-)JzdPnbfoIgq
z-fK?}Y9sfDT)LeFI>;kwm*l;wFmAnu*;gNe{hrye{n1BZ-PnHnjQHseqVr*-Y9P3i
zOmBO$K7$F?X)d!f29CgEZP?G61K(;+4IBN#-9bM6`*rT19QY5JluK2Czv|q+x494I
zud|v${%2rbxz;pW((BntgdHsR-l2u}G)*|Fx1B=(AH%eK2ER{J+nT)T6X>-S(v|%JS!~M*2QNsUwVP)a&u?d#y4CeU
zyx{VStC3$7POmS;^V7?QPuMhr-lIE)1IQ=m*J(UPUb^8z4RbHVyIbeoNBdg9ZnSU>
z-ak+u(}MQupikS)Ifzd!*uQ}G^@x49j1%B3pI(XE5dS^;bOiE&o8c8LOn^)arjodC
zW9GI3e(48(l{`7mYmJeeQr-vQ!=FdfkRK(*vnq%H+H$hDBEQ(d#qNl_
z=I@%*2a&(7FMGm~PnLVjGGY8f^?2&}#r%&cSg$iql8_;&
zm(y_VE?}0)8hONDbHkKBv*$Ie+b5@}A=TfiH*#tY&-QlRMKXN=fyR{1PbVg-!6
z{Ur7ppi0Q^6sjNk+W*9Lj-PRr59$l$J}X52d(ho7X9)UnKw9W8OxJa_T>#>9xv)K$
z?^tuLdGvddm63)jfE#uupT+Q*Wj7pA?+EW&DxdcfU_QCQ;gvbC
z1NO)B-ln3zwQW;L_jS;lbK9nd{_js|PS!Xt{8P}o#1Hd0&ED&aSk57b6-l^`l$^-4
z#C=-BNM7s5pg*4SYK9Nyot%q{@CPEW$6Yys*zYBN>{dk{yH4onMlZmB{oN69fU7>I
z9>wuq!7Q*9*Ey#?rosh=z^nPaqL9ZobbmBpfc$&=7GIA9etaa?Y2pR&@X5!A==FeC
zOOJ}THp6#4CHuSd6Zdox_jwPZ2=j4&$L-mbu&$Q1{S%M-8*f9p-pRrG{-W)?qEHWf
zzqY3hC3-TXck$M-ui+P?=G#Uo$
zfn-gMxq#9xV!HNKM|@Z}5xJ*W^J)gZGv7jc;;&)@No1TZnFf2hqJXVL?~|@4O5ANTt+sqcS7dzbmVhXfC!QZqH!hvrHt5Cc0}t@O}13sCv6NYO|qlyC7bcMUVx
z1&2LNQeZ!wk{fFCp*|Z@)GJ=XeNEQ111rxq1ONOdqM_Z@O6U&Wr28k<2I~ZYY8GX<
zSG7u=`rmqlm;LwjsBYrVzV*Pjs%GM_)^li?Q7fsJ5>!fl(m?`3rBm-^cM)dhTV*N{
zU`Jn-n_PtN0e940^IjyKq_OY~&yJ%VB-nv(V^(@6xq5!Lw3%fWd0Ug^mC_33xEMEc
zCjsW6)XbL9lrD1Z*FvlAR41&rLs*=(y2!Cn(IG9k&sf~|;`|F=IM=H3-o`W+_9bF@
zSntCArD3zTO$2w>VYy8G4
z4)gZ0n>YXN?k4}@fADT&>LG1=_cd>@>>**i3DY+PyU2wa^7!XLxc^LzkKsM6uZ*01
zSVoG$51+W*zU~M3d&F!qwc(!RZUevOP`G#b=3MMpDdhjGpMV-CrlPohQhr8MV+C_RDY-LqKO!9BkEQU8`~-ge<@yya!#=oo
zps7dLqmS^GM8~aU=_AtvKRfr9^b!YmCGUKz9&$o4{+hPe-VdYt$nFYJ^P_NXkToJ;
zv+Zyr=HZShToD5Eaoy-MVddrdF-_H}(1Cv7j8YiziY8_-RBrFL2T?C&Cy
z?|&%1o#`Z=1BZ9YZ0jN7O&f;8vbxCw&7{r6!Ek>V+vZ(~aKB2K$)bclJ={Y*GPX6b
z8T@r$S@x6!!MJv4vQ}Jx{gecHf&DOEEiS7&B|qyT>k~OeGOfU0$Gy_V8uH_NZZODT
z8Q7JPEgH#dfHx)EZrTC!0fTGm3Q+}EhX|A|mpB98ZR{^l(s>8>{iU*!fhX-mK*M7!
z3+}fLu3T6tNdw>8YAjO^l7)LQ%Jx2EKifeXSAH(R`(fPIYBmUcg8R?@8nmC<1oy$X
z80W3%>m=T;4lO0paIf=Ur!~9p_Y)IsmTsWCsa
z9~XCK?+=0ZNYtFD#q)GF{6*Y%02d5)9L$0Fc5Zh0#|(&%Z~t}?@7GsJ550}w4~orn
z4CX=n$J$MddjRil)3}D;U2;5%YDGWUqWRm3R0wyDeXWN3gQ9_#PJM#sPXDktbqng_
zZ+TYkFzA0Xee!w%;#p*@dXcxkC!QkjrS-3z-7tkhjx}$aErKw#X8Aiu>LZ
z3PsiTK#wNHwE^w?CzXHOQ16tZ#5&|nnKC@$nC?Oj!?+G$>Yf)B=+_pK*er?sBVQQJw9=iYb!BLDKsonJtF-A7J7~|c#4{XKuyCvdw
z9P^=mrdAZ=9|$L0#`xTxPZTc`;}_4u_^6b}3mD&h1g?z+6e?Qjih43U!)p<{4J_7U
zxRmh3Z^YnW2_p<2JM)|3^^A9HD8BkN&)o~dcd_;yqMpm%CW83+SP;b***^PI{%!8f
zhAkNWoq3j8SKQ{T^AZ7~p;Iai|T`2%SF7T2g-iFpkJpXbJh2oi?sC+th;t$@T
zeRvzGb;jK0(7h!WWBC=bgv!4G-C{_8qW>y?}L
z9*AQ6tn&M9h$S9Z1`u!d{G-+zFOOb_hK2P<^Bt<)Z?9YW{|fJ|Gu#+1e)7>f3h4u>
zay#qGQT5QPudy4$74N|{&44q^&X4f@v25q5@XH&Z_#i(o{zBFJrlCJnJzJQyxnuap
zgf@3Pj}ygSovR8s<{EGf=P4p`_!y3>wBx4T>M;NGIc8?zxLr4V%n;W(Ywg}2!*!Hd
z=`i&_>+a&Nqsu{_SqPfO@fx#@=ON~!Xbz6Y0;cHw=*4x^*_-Xuy?3jfwKrnCR_CfN
zjOU(xpM-M6_6%xY>d()^DgL$spvJzW~Y_ZF(1B;wngY~(4LlD
z9E5x_8FockDB+x4{>zJn9BsPv`tDD{T^1#{3Ceo1kQLiVZF&4
z{rQM`E3;F@!~v&cgca~TuP0LVFrTOEezM{Ga^-vNFwRE-*V(IZ{W|k3kP*u@Q6W}I
z5AUZl-g2;IE!+zuFBqo?c1e_jXA`X^%-hMc0>!|WZ~QLnehhqChO7dcBs&ye_gM}5wRgMNqd@O6XZ99$NIx!0oVgP0PWId$7a>T$
z_3{yAO*qd|WxYpq7Q!9944KhxRGszQtpxhDhy4ulVI5T!QX!lV_PyCa{8b`^S59j>
zdP4b^-&`}33;L=db1fP09(SJ|A8SDWf#QovmAWp%r>L}iCJ_8QhIh}@Hp2SHUPjM!
zpbP$+8MmE8z`n1|CJ($1p}5RNpA*WhDfvya%nb7TZvE3FS6J^`w|vNj_kP|g+i(Z!
z;n;11h2_KG{ebHvm7{69l~VB9oSodD?1+pV}0c*fr$+Wt3Ce>%tCuq!})Na-|d
z+1-Hq9uA1{m)`+=TxOaM=VR&az89GSy#?W{hrgWQ`3I|#=Rr2hKR3+?{!1@T=g`1)
z;Qz|%S$+@rN{ffLMhW;M&j#C#zlHT5z2&x99;o-WAMz%a!2eR|q{4B09r5~b=R4Hj
zPuK2AM#yIYm%Zv4c+UBClNtwpw8e-%Ctyo7Ct
zBb^5L$7t4qm;mIz!yweo5W)kDe>^ui4(GA|{gbo34*m4k>EUr$F9~|p3El94_MN%r
zrDqF#E+F)1f(ZCe3{;)JK>uh+t6dBQ|L28kKUI-u$tRuOU_KM&>tIrZ_&g5YBWnn^hrWK6S@DY%mA`29QjcMm@fC(NFLI=JhQtl
z8Q$~fa*0I^eW6pcfaJ
z+6V3Mz=v6^1M5-5RZIi1T=Am7t!rf+8LH^gD
zTO7sl&bGi}Cxf#DETrT3
zWvn3@!lNGPRMNw9a>{(!IPu8|`UE-b#X>ldW|CprUuV9f^U8y->y34`yTDw>L}e-
z^v5#&FbP8YDt&hU7qq+l%O6(YJT0}|R~qBhf``AN9TB@EnS$TLw5FwMBM*0#-~R{g
zQ>|L`cC?c#T-vvxUj1gNKpao<^plIo>$z+{1!KOWRMku|Ucu$nIXd)`SX?~CuVhswDBgZQn=f&yIk2lDuM0mbj_m0FZg=E;3Qr9a5|fDz@ySLXx}*BGZz
zd{jpWK05|Ej&JGz9E{$qz_5XzS&UHpIhhTr4P=vZl~n$&1qlKXhk9@Er1JT#J&Ov@
zkc8{)AiPyAlhXHQ>!jp?)IN%TUvOPIM?+KHOvPV%N<*P^-7mY5HIAdf)|26W!o7_??)>v
zPe5LAv5_k0j!%(#D6js1f2n2+jv0ZRA|$pPOyQ!qT+NG%x0J=q0K#RgUp)*Du@%6LP0z?%-e+QhW+wCp5Q09rc
zsEW8E;4d^Z*k#^MYP>CS9W%!CKjyxn#;H}%N^1Olx1FN)mr6?Dx*dqW&D(GV^Rp?b
zfm%mZ=o?V^op3Ct#-F-R2vu))JH2gCzyAD~66W{g*==?>pDv$%MaKu|`dQTk*9GB6
zOn-=i{amMc4*qZoyy|CK)cdW-7}H_N5_r?~P@rlJ1h67e$B|5q{g4C|w7dBhyf-;ULRk1<4D`q*LVG4F6p|X2*1ca~Q$`u{}dwucE
zs$XruXSQ+F(IrCow33>c0pL(`EH~~;@B5UV`w`?#p1jUlpzrnY)K*%MWs6p>M|&lb
z^!+`q>x55@sN(s_(vyR7c+Q^b{3MS)gn$0r_Z01j)p>h@?R1_c!NLL1C2N%)+Qsm!
z$Uw|*#|;?;l*jnWuj4wKY)RdO^)H*R92f_vEb?MUJ@B6`x7FzufzO(&gsb=i?{wVE
zyMZ0fIdzURMec-h+`JSlegyIv{;%QKX2?fT&98F&{_keDtoIEFpA`Dcy#t<$pU$<#
z{rlhJqdd!@9c#ZIKZfNVcvP1Y2z($lN9cMI#D~-~eV7E)`H?D63dq*O!|x7yl{sc#
z<^WF-cCBFucM^V7*av;6{Il#0(f3lE|95`(*=2D#KGtHpnOp?m&92B6~0#m<9gWU)x2yM=>678ueb`Bl&0lb2x!k!
z-!2I6U&yHu+X47l+2i*q(68w9JEsLWr@5B_^Lw!6$}jY1jM#6ac+dWp;LA!N#|x*d
z!G6S6rkv^rc*8+O^B7>>DX(+b?`FFL*pa_w@12mCh4kHjUY9~#fKq
z{`(-j9Lj4kc}wXXV9)%$WE{sa`^3^Pe`e+kfl+{WmB-z2-ZNV0?|lx>D@f_D$8sjG
zuUL)qhg0+S^sf*ub?p-^4aiDfZ}#K(V%{|?ECA>ebLu1~puVqIVEkAG4<0r+N7
z>;UG&A?#GI61@M$T3v&O&~N@Ijj<1c{4j0X!zn0FC41QWlTbeLcu3v~$|D^(ylX4O
z8{R)?qX+M0Xe!-<NoSun1E{ElAgf6oH4u$SA-dO#yTxi;j>aYDxQ*se`3JJQ1dt5vlf
zk^nuU8w6zlO=a}WvA@(DvwMj1Nu$%{Aq&8xEV*!^0_H!f_5>k7S@JOx<4=ZPVn*I-
z{r<4+8bIhJFn5(OHNY16RO%i<)o_^=
zIFIdNQz*rGPJ$z8bQ$2r4eJgOK+Az(I(*-&By$Ix55Z}_9QFKi!pGDBnR&ZrFy7lS
zP6pp&`uQnkA4l$-`hj}P42_;Rzu#3CX~%e@IT<6AUB5MFBPt$HqyFRK8?T#>@hcMI
zzhU^+E*d{XDZgLsSU>57Et)9bD~O@a$M8;8PoX@LHSq-FJ37`;_OGcd%K^h_rsAmo
zCF>?7)nYx}@uI(qdLwxB~;k+n+_k2zH)2*2T>``VhFsAqn
z@3814e158>W-W%#CZ|*U=CP*}#YavRTZ&`2O?|yPVnC$-e#HD!HB`L&=TwUK9Qi{}
z*>PpMb%!zj`B`bo58L2iPTAG_bBPq6deLJ_@uH0)Lq({k$KFcugL8?k6mL%!ysm`d
z*{nDLEqnstv+bzb?h6SHtT;
zW-5Z}2NYx0qOejIvH|k#>LpBI_G83Dpo|Cgu!f4&xr58==`DP;N|v34*V4jWPE1-%dD00h95
z|Kr8~5m#Lpj=c&|VA&GyFZus@EjSH?;RC0a)nD06RwcN%WH5=>RM^
z*e6j!;bvTxfP4jdAtED;9>gSAEhDat!571R(Es(8ip>P_SI#92zrBQ7l1u3(!L9gPg#tr8!{2xj!#I}maX?`y-zg$X0t*hH
zEV^ZwnpfM4Td8^CP^BTI*H&8#8H9G|GJ=G`{*+n300DncV?Oo%6@J-N`30^1Q1T7^
z%Tzhp3{Fsfg8b!Ju~5$+oG85DzO)`+ndwZW7d|fq6#{;fnx*yF@Nviv$P*O|6uMO{
z^^=9tI!gcWt90nu;5R;J@)&W|Rb_p&13dToRME}|2fX=$cJraokRGN}`(SBZw71!r
z%ICX2+F=aOb4{cCJJUzlsPKPLV$j2Y=NrpIg8?tgfn)&NW3?s_4?p`vl}oKClbVkf
zFKwab$CMSl)Vw1t5}bqiY31Ecl_z}*8>JuJ@RlkEPuVo|NT~N;I@EK1=S(@!-kq*q
zJd3=|Ny)Gtc~CJc`$J*C-7YO64*=;+il&h_1@?Nr!{;SeSoLCgi}u#%=&Ybw`_5-jmr6tu4P+K#_
zANP&?yx$!+1(`E&lujA=)c4x>^^zdldN`IF1Kx@jro-p+SXkN7UuoXLOWgy!|L-~&
zK~NuM)hn=mJ|6$bj(o%P+#~8f;enyGI(#78o#3PHos&{YK8X1p@ZYirdDycI&wW@A
z71F79aD95%mxua4fnfQQ9<^Diz2+#^uRQOp
z5b}IMxs`{oUBB0D^b!TMjv0Q6`Lmc$iNy6V9CMpj2K*bl03IWFZ*+1I#!n?jQha5~
zBftuIOEQZGj0E5vDsuO+9CazSzp>n>He6*v{fFaXlz(x{#|d-f$@;Q6S8%_G-hF%Y
zIH3Gr+1+#y9`EjX1^Jv^fMYqP_w&}4%h-?B8L^A-06IKi5y1TJvCTV({chdi8u*|M
z+KGPn66SB$uS5%DKmc;!wWFU+SIPsm2(iS>3@DdfNOq!`21!=>RM
zG4L`|BbYFNN5ti8bLxn2+yoX})HT`SC5`WmFFPZcn(qs9qSME*Lgx7o7j
z4fuS0ry<2-YhsrJfr9mYM%OsLr!Z@=-vsdAPQf^=U+vpZnz8s9gjXX}=)LA|x2F8vs;#63yb
zjnYA*Ul=~uslb5gvbT818;AkU>&GyD
zVw)B=Pd_-iGkms#DTK6+t>|(V1PfUA3h5JTt0P-dVX#PJr)0(MUIjma_q3h
zcqh{mZp7>1!jzqU&l*Yb|Kw&oAP@eHS9dle)=HXF?+Xfs=@MkOHd-njiJBD@w)urq
z>33Q_q)>mPh6-nSzLBzDJMLstxTZ~pD(9}`wN!jM{M7~1)BbsaLfb<>sc_TyOO*ZF
zl6aqzHL_i(cs3KrF6f)(L-PT)WiIUtu-?n3^f}hjQt9`FHB!&HG;?
zN~dueXaR(mFz%;t*O#T|5@D9daJxY$HX!r0CH+U5&|Dz`OW{wx9weuuY%PoK6guUN$;EMlFUTAgo&74h#$k{4PYA+znw@)9Z`0C
zAWa9~F&n_ugm|}O8mZDyI?8SOxQ`K=+hg`#HdmCfuva*
zk!UE}`l_S@&Sy{W35EX+yV2gzdW$D*63nBtCvL
z#lN2jPkbGJ=s84YR(w12UUig6tvD31@4;&_eWFF4s}RmDbJ8i%HV=@9`wsRj@8NsI
zd-_htb`Owue!5qGE%cIBuc^e}HGSli*{>?Q2P^P29Zx_G)5DUzz$o`8Dae
z6Z==nxt5fx1TmF(W)Oif=f!O+IO0qDT
zc}Hh|3E^5JDSCChmdFjfPwyFjMraqoQM2k9kD>zb3G6S3nb3as&Eg4*Ye^yiz
z(+QvHuv++Ucdvb)5#0afq2wK;%vDWZEgBv!%YH)WqSg!9x*yK#eL}Vbg-pjvR*=ecxZjhvf{1;KEP1B~_sECZjFx(o6Tvy51%>TNC~PBbYZ8=Ff=r^D+wuY@P$
ze5Yyj)pvy?FFw06k~xoDIv~Ex+%S)neyyD*SM!Kn&aJGI9+hOP=ozbj48=rP&4W(p
zEZp->_f|};DVL;FzROh(fP3IC7+uMDSV69oJmKB?wV2EVb*@tlg?sF!*F-X_=8)*)
z)1^P}<&wdCas7GbGE%vqspWjCm;^3&qMP}WPj+v1DQkiM+!MCly=wG7iq1QptM8BF`pPbQl%z#Q$|xh;bBt^@}P_uV_U^UE|h%z7^*Vlz}O#EJ@*B(VYI7?fvj5=G~|e{77WRQIwC@2
zEsA;2yr;4_LMjJ7Xyct2xswGC+u9XY2xY;UD~B%Rk$Cc6d&^nDt+}u%2;343$c9@p
z)g2j|>990rnxkbe4Ze?d2vI$o1GQM+&-IGskf0cUO_fOgA@gZJD`oyPm}wg|`(J7b
z6igeFQoBR)_~_X*VMRI6s~zvnRmg;b`E4iv4WyFR0~wvhwv*vJ)t*rP(=3?8G4N!2
zJPUpfAFmq@&Ve=A)6brIlKe}HK^Kl2kvsthdmg$qWx|@qqesR~vqqrv!Ip7
zhTvK6Z0MeQSep8MF1+_i*{V-34JOy=+PGXVLDQ4>*R&MVAp85KyG>Xc
zEIcO4d6ho}D&HO*ESF1$WHvri_hJ%Z{t*g%%k&eJeRnCIr!fIm>ir9*%1?&j;mto@
zJWYguWVOdc$m1Y&aQ+MOh_@tP&=1Lr7Ox>ybLG=4=>pgmKYGGfumIMq`iw1h6u_&B
z%}ve3B<&h1atE}ipRVTphRN)3e
z_z{MmVewqEcn^h&8*;6KqoC+{HSs8=ccgtS+v;rf*U&p
zyQD?}i5|x{!WMCXM7@)_u9C!OD1YoAcaF~&`0IKcWvE*@>{51h>OWcq_uBilw-&15
z@b7aUL~M&-p;UtF#7Hp==auZ$`dtdmf0fUv-$HPjkJ3(nrW#rwlOwNj$c9y(p%hXi
z&(r!H>7*MMis0c=sjE&crBH?^Bgc!Y5|Z299=+=H4!W)!iEFBS51p0<XcSW
zHT*F4$!+d@Eo?3My*%Pr5AV>`hQD7Uc?JSMp3CB^gtt4SV$WvdaG{$qf1|C6)K`k6
zl0C15)yJ>87+lsK$A*rL`|QU&WC`+uqK=#o6zChYe$k)!>c+sRDW;w*sWPNFbvhsy)R1qo{~dX_J>Rkm9uQ#d
zTe9zg!uP|=v`D;wVNJ4E!K|cw{1Ix=OPeIl1f}9U)pQrhM=?+Kg!CRhW?NPLj=7KY
ze74%yc-aYsuhPB~N+Vr=$Yk-+LO1*+l6`DYqYrwXygThesuSiul+zWic0xi&xv1a~
z$rq=;vF|6<3;X-yQ^n?gLXGP2zkVVE&^oF7iZN;bl-)4S^Hl2NnK>L^h+-^$_?2#wA>3tRV=-3)%HTquc^k={=Ja)
zfx;6uhCX;}REg#Hy*`qkS!cYnt`FLa2A2OR?}IKb@Tf{e9~`PHOsRbG6Sl}L>~ZY;
zgsQF-G}H0@B(8{n+To#oC~-FQnnQm-eD~I}*^)HQ)R#-Ak4+E2hce7|wX1`W+FQLg
zB6|oXGQ4u@SRI1upSsVOT^@w^g%1ojj}5`*d!uFc9>cI^jJ)?p_6Yo<8rZUSaumwg
zPG4;a>V+D`A^CB1eelL}Xz%!Q`HP1d(tSZjB&2r|1b(Y)B*~=`;EfM^*_`eYrSydoS5trQXLmU
z$q_1B-UAg%e0iM@bwj9#2Gpdx;6SvZvJu%&m=X`va~t{~JKO8In|9N2qc
z-R5t-ip$+lNqKgJ$94c(2(6uK<{N;srNha*4E-?pMD3rycluz;@H!huOfM|kV7lp$
zJq%y{40z6FIt)EltfTqVhhRl~{HL#R1MtKr{dM<(eyG0~TC46q3Kw@wsf%<+;eefl
zfYg+44^4*YMu;BWdY&T>`FB(owA^CyggHu)BkkoMXMZimA|@a*FUG#D5p@q{nzowVwQsK)MUm?~+0
z3wufxoalu|)%|1t^CRWU(utaY_Z=`l#zxE>euv3k^Y8Ceeue+A$7Mece1+1h#_o4W
z=RxlZ$ht6*d?-#Yn@n4GzQe3@CM6v^UtzZ3KJ%^U7I@>x&sF;ST~M@iq>GMpj^Np5
zD9z+-JG|R)I-Zv78+@ST`;2$71zM$Sd~qBg)#Fs(A6Ag^u-uExHc2ZTFu?GZQ{{^`
zXzMve?fUL3JbC8?!v+~?z8szaEgjvkQ{czu!Ag=3CZ@LD$?!XLW!2RvFKC5w|GBm&
zskOkuh*Rr+BtG8Lcj_lgR2pFOnG%g)5@#mmKfu}btQt~W?mi{J)dHj4FR9fCH^W^V
zeQkWv0Lf+)--f-dft}ty^p8pW-Mm3nmg5aA@WqtiezblwT#UQ!@}$2W{uX^8-=0kJ
z-YnQ<%MByQvg|IiRPq(JxkV|T;Aw`668OsfD|N8SU?oug1Oer5C}uiHmBV|U_o8RN
zx5AX~w2K&gfM`U{EUfY-ul$8*#RayOATMBoQCF*4INSxnt&hE@^k}rT#{p-3Wf=1{OPs&aO)Xe?5
zr9|>+p0T9(_I?mS_w}cBfvXj;KXuc5xUv)yWOOlsz9rCfPjmkG#1{zTuUrceFNgOH
ze@e@-mcYmH$Ftt-7eE8BGkUr|8@^8oZI&Q;pOyok(-hX1K-cE3gg&QYxa`#Y#fK{w
zQammsj#8(<@sKx0;Bf*}Jz?2+-ma8%ZghC^;(Q65hE78iT%Vz>^r)q%WQnNOM%=0h35)V5KO13%=1U9`;4f{l%~M|A$ngmM<#
zm6z%>NS>O5a=Dus@K1e^d0W;y7~rBYYke{XYM+priaQ(!_k|_(Z=`*M_A+)oSat$b
zvg8{JP|zjh%`1kEEf~VkGn&uooW0;#i%f+Y_hFNFR|hT3CmKHrPONOk0rWvX>F
z%&e`{a}oFmx3V^SRjU$UJVg+_kAD&r|Fp>MYMTk?>C4%@N%QiX5PiW{HIipc>b9Z4
zd^XgficpC=l?^p4x1f!CChROO^SGsv1-Zo*g~Lx}L7V5^?o;O3kj?8EU)^LbB%2R3
z&-tAW)f3NB`k@TSCde35aygU47c*AmoXCPZUM2XA+d1$`zpCBUPsz}i-MDmED-AYf
zrZ@eK&44M+Jp99X88Gud`S5FH=`i-unhs@0B4nq1Kp6ZW{Vp#j-hY*;@H_qIwsMOM
z=wAQNtE3_mQav;2KA@cjb)C3U7he{_2cP}p*WB~qQuYQzw@WHKHF`6eb1f6Tz1|X9
zu95?)%!3oIO)0*l`!tTl
zT8V|!Z$xoy>OBj{hfSxy%2U?m!`IqDKeCt7p=t3^>xLtlkimNeKMO$aUQDFC}@U?3y->%AtOfIkZVDgp7SBKCoZ~&=V7U
z(fhX$vJFD7gCgZ5{^zsT!JL&)M-pSnlqrDz7il%0E0OwW>d}9SSTQMQDmqq7;;ffX
z6dJBCRgm@t&hlt#3VozwJRAnRWlx|pFX
z_;9pulyWK$N)?_E%M~kx*~@NZ6xpAl$W`+2T7go?`3!%5fx7~llBa4>R1q**D_1_|
zcMaTr_fElkzYf01tZ{$VTMtDJB;L48LFy}niQG_%Dk$ASex_W3#8aL4ZB+7KBTNpY
z+@;NGf%Ru%k1xHcgrn3AM_2%YZoAPInb
ztbM+%)m;G%8|K=gh-!#cyz3aPY=-1r=YFRiBIT;NdPjXp>pF^p0cuf_B{2JMsq%xW
zYM5@(<=wT@2>&~9?gaxWpRx4$sN^%h8rg(h&+gbRD_I}D-if4=7TKXM<&{N$A}gQk
zFEzTJAj52mXEC(#-9Fhl@%7OYCpIb*HEW-|
zRw|RF<$wRDRryP0*u~7m#^i6Zvy^f1V$Z)-*7LM@r;IMk_87PY{r9{~R*8Ox#)iCD
zwk08xBk>hpdHe5&%w6wsSuH_y`i2WTR#vU4{@=oGWg&&J!hyCu*<3b`I%b!HxSr>i
z6khG4m=n`iGGAsXoaf{5{)ux`n0TzSi9phNP5%lfh0xObj}I)msr
zjGg%&OKG(V&VT8f=Ob=iJg_coN~uX57ntZ2DXY4Kp{=5T8zJmC{dC*sR(or__ghc9
z;amXDZ_sGAKe{Pft}e!PVP(Ej@@w(EV|$#~hE3@j{Z$oQFLbZ(RYMFm5r0emE{!EV
zyc~1FE6@zH{2KKu|FISR>%r2$p$bP#KOn=0PUQnmM4TR;nzq9N>~oKJ``hAU+JoJC
zG!B>*=ZN|>MK7Ggy30>MItJ6Fh@g{nyoCpFpiYsrhuCZ8^Ob9F-0_aPA;Rg~uP`AU
ziY4;p1Z;ngl6SQ5>}c`(We@ofn)W}N*CEGuws^*(>wNqxPu+jLH{E$
zSk}v-&J3~l_~Q$ca~!@QX{OC02AvxRY456)!_1b~PnJ
zj0O1MEUph9kB8v6caDdz5mgd?R!oOA<4X!wwbZO)i@n0Hi4>T<%n8Q1lS1+tSHEFy
z5{C=u(z-A%mjFlGx7pbIi)gu3(Gr}Y{qv)?fmerjZSi8f63v?htubQoxqmt
zy9yd4ms2)dr%ewI`I4WzJHd!(1Onfr-=GIU9sjLe*_yzT3S~Cy4oRW0wqy~VAt{h>
zhzU85Lp1sL$hC?B1@P}g?4PK!_fV4HaF+%RA4uMz`gx^Q2wiJ&{QNIW60Cm>uKT#7
zgBCfm27dp0hAwEwM-LqoLHQ@I1WVKlf|Um^*F|bY(2ewaP8y~
zZp#c&Fip2=$228~OtT(uj!K;avtM~Pgyxk&z%Z?LTADpF<~dyuaZL=g1un&yT@Xe^
zQ6GN|y#io=`X)onHy0Fcj!V6wk_5XG!n`$tk5D*fR3szj0KVOZ^BJ4l)hQ;c%3=`oU6aQ@;luCQ3ku6H;+w3ytLoZnA0opjV#{(
zYo`Tb-;m+YdF2Fdrs^r=jYc7zR(6*V)Fo7*XW6%
zZd5}`GU(nsJ|4}Lh(e&!E0ORLbTh;B%K
zKXrfiBN|ubu{RLv06dv;)~d=aK=AA}u|AOkw7w|E8So+zct*sH({(q4uBOai!w5M`l-j1ONPMA1>?d?Lq7nUz#Mf^@&fwWbF5$6q5&;M#F2;InFhO_%;sAB
zb0A`8jqIN50;;0@o4BU2g-&tOm}hD+!qZKeXRbywK-fIO^?WxU{L$}~7Z+>-b>H~R
z>Gx}ZwSei1^FKquskh?uQ_g0$=9{T@V=V<9ra&1g!lOVapJ7J3$K`Lg3qIM_%pno~bD
z174_V`xsd)12;{vT$bOzK$2x%pVHF~urhJ(Oq^;C*v3N@FE}v5v-Xj)$<$QvWerQ<
z(K8!hXysrR`>j@R`6^z*8PbaKmTtY7yJvw?YLA$2{`dqu3!-jR=p+E&ts{m$5{4+@
zV??%VTOjhGcorPa7>PV0$!o^Nvd{w7FU&vei(cH4_cWu71Z{Lz!(f;{xMkwae^WFZ
z1YS&H3Y*LVQc|*1r7>^O=Id79a|Ma$u-8F}5Zn!U+MVj@_D=wma*>7
zHUDNHbaQ{(g*_Na%jMF_Ja$FqJK8pMT^>MzHLNHi!vc(Q8XxBNuts{j`=uJcUV`Gc
zOdj@U9;1-jZvC4*k5SNtHxat8Zi9Tk`0wYI?*OW$+arN5y^*8>3Ys|(3NC%x`tM4y
zGibW;KBKqr3E~W5FW`Rr0IlSzjZ4U7BR8|kIx3Wj^6vE1jQ;lusC34fY0HKK*XaZF
zyI$A8_89Z*J+;XEq*LcN_C_=`thl&&x^g7;8Cz|?IW*OOF>5waV^uip|^B3tRgj!_OP
zY8?5D+s+N_2-$waO(s^keO73|-W0)}w@`;UlUEq-iuT})ePePGB{i6N$v>@E9+!}0
zi$+S_M;;_l|6_`e*uv0Dy^9vVe_%Tbugk{N`f=@KS6}{m1CaNXsp8)QF~sY*@WI85
z2VKVRX?GTK0@uZ>vj(4saMs!bKiytF!t=!+Rwms|!SWf&C;m`B$C_%6$!yy{!ZkZW
z>isBO@T48Z3|B5354(D%#_Cu$exKW_px}!&CVlC{iH1Y2SgQcrc9fVd?nSR~p`buOW<{FnRz~O?+t-lP6#{c=y>2!LqDn#ooz6`^czJ*0{oa7!@Iw*PYuU}lET4>tJsF=+Eiey
zK3*$h_A7M44$r#vU?6$%D#kdFL($AjkKY`<^>O;UI{x6C?#_ireweEsEPa;Yj9DuE
zS157U3ZIRC{l+p)9lLV6LaaUs8k=;5qyO?%$g&@N2cA_lHZou+{Bi%K<%C
z+@J9ZOMtL0cHT^_Ni4Ymf85l?V>tB%zhpw+`0G?6Rwk`n!V@2WD|~D9);=AEJFl9(
zT6VdHkxwavn&_Uyw0d?EVg`=m4>lN0Xl}4!ENsb%L`E$0LXRxcj7Kt%V7clLa
zd}DdrP@F#8L6^146=O~i-n=j{BLR9Lc7LRL>iB1_#~n-KKb48oQ3aW)^(SN
z>3|8lw|eAl>S3w1z>IUq0KZs&Ot1f(CAMv_DpU8$1JCOlOfTut#q%g6($d#_k2Z=3`#kKAUCBMD^fcT9PpdUAyX*S`w^R4&-F*HEyYlEsX0cE-
zt_7~g^1FS(6zmi^hK$NEL+5>ykLsy1xxrHaim2ik@G!A7u4l#vM>+d07kg~92u1m
zx=Zh9Qk!}KbWHX5kNp-$_qD&!SkFjM)Ub%D5^Oz*^eUGQB)eHjiRbJ`6
zeJQAuYhIZ;#0o@AG(U^~eG^IRYgjXXw+82K=9G<}NPQV3p^=E<+Z#tMFPiMPwcdeDro8+1DPibcH`R^f2Ro5$vgFrPGzld0sG2U9
z&w+u65vJ;uc;IpkuH&~`(CPlO7kHlxftyF4tIOzCqGnDBLEh|5H0m_`*H8F4T3@l0
zkUH26e(T=+RB>S#Sr^_M(#Y8kEv~
zc%tnW8c(z#N=*-w_@NQ)ANYDuU-sdk^cre{DNEdz=vo3Uhg{mUlBu9B;`EcDWG99)
z+9DKNS%^5C$0{BUfcQ@C6G9E+Ae5UY{Il|DA|%9VqkVe^*g9NG$@)(c#$NvIW)va}
z(~r_WZ25kgm{(d)yYvDhIvF^sX{0pZ<=m?|pATGto^Rg#ix=h~R<|GbNH@^J+Y=mZ
zboW;QXA*DGqI@Ipe3f>i?MW9H4=Rv+#@2xj-q)M?DL_jI^GAFKIvwn40CP8({(|@4C~BwK
zCc&4O@$VFNB9MRe`i&>cXW{N`WgGbX5Y(kpQ@wII4{dPz?3wq?BGKew-uAzYgiF25
zEH4)z+zek?QJ$A0w$3u!IM|#ath3a@&BvIDC=q7ucs?Ad#=6r-%*>;d@@&((H+6_y
zIAXxtqX!&0eM?<0rXBc;+?SjEU5iHkEghNpnF#J@HGg{DHHiXaTYp+PbfZU~p1wR5
zg1|YeNbk#AgTP15*4Eg>$HxeNU`-BQ&UeU664m|
zPYfp~l5)q(U1-6J_A>?w1G;Lm#9&&s>>&Y!7HvaB)_<)&a(f?mlJ138Xk)
zbJzH;C&Dfr>o;bo19f^u47X|MVQk*Gsb~Z}Y`F13L-z6rYSz@hF3P=%LhlLW;p&x0
z*e)jd)qFbOydH8j;)f5oCja6Fd7CS`*BDTBuRfoU@M{jie%+U%d4bt=T{+Z@!
z##QVMuE%cs?<{&$JVYPD$4nakhsN~Z8<_Wh5+nHEbD-{Q3+r69JV?;!db;?L9H^Y9
zdK}lzf{t+n%&ewNVO0Xdb8)=akjBSbqZ(}*D6TPlT*Zn87}uo>RrYd#Cx%j4B160Q
z)sF;`$NUIT)bso!dujptf~e_~q!m%SrXzQj>@ifF8CZUZS`0iTD4FF9ywRh*=-CEo
zPXJu`mln_saHh`W%9A`xlzH&>ixtbWU_1SJYOOvwxL9l@JjXbWSG_j6a|I0IThwo8
zlXv>C!M3on)$@rMcN=F9>ta7BV@r}0_=>uI$U;}
z9X#UeA0XasU`kQU<@sipfQRFbJ41~M$X$+A;40xpY^mki
zIaldX=7}BgixFo4l>qsq`A{M5N9Je2cCZ#NOfgt4Oi09pE@2PT`7Loj>F7P-7YTT5
zCg;|v_1>?sL4)$2={W4pI-9#TlP}h!@TQP$>KEQCKj(MG
zKLMXPFj(1nwiwHurkx=ob8xThFpj4c{+O7ops{xFCYCWIT-LNYhS##OCsjp#z=b|e
zlliCP7}Mf_foVxK4nGP@`KLa5-3!++j?JxaMf|qd?(MfV
ziUof7YmEb&GSg<*c+oN*yU&ijzI}*7V8;}{PG!{+H~s;meXY{+9)x1McPzfyi#g+$
zvwvp4NHD@quG@Ccy|l)Xv-at@*&FfBr&hD`Okw!*x|ly6wy&|RH+uLQDjKHmHA71{%6N58*{D)Q7AHm-D
z8<=W(V7O4vyY@$os#xg3(C(3ww@K^gPv--&tniuV@>Wt!5}3W;fc1vNDa^9#5Z}=Y
z68Qc58a=AJX4uWzYH?xK5N!5Y>wXKb0$#AmF{&MN6gNEC5_2X@9CK{A-hL_C3=b&K
zK3MJg7O&f+y==HGjU6_r>763&C-@tkd;TMU57%BJ7Z<*-hYKj#FAtxO!qkEonCC3bwiesrxy?O
zj_SGfbre&3Jx>CoFXN}io=;HwP@s%=T$GpJ^9`1K^xA`%ZocFE~6Q
zDO52xj1z92-_$rG@GmoLA~e+9nBiYCK`p_5c)qvM2|=k#h>HD;S)%YbmM(Okqe*!j
z|0UwITTEKr8>PEw^wW|NiGDqV&-cp!3KP6$vuGP@z4rLjh0~P4A}4R->XD;pub|g-)>0d0Xj7qrzW4;(-rk
zKwQD8rC1$#974v&DCEc1%5-VGGQ%*zI4xl^7pm?)XbxD-68Y(=Bx1
z-lszCq7ZP=>X*tgnGRIY;ujh1v&r?2}|BcK6)|c!Z
zGK*9Vt;&L1B29?zs{O;e1C|S5NuJ97JF$y_@OYvHA7K&>J3{xYuiak1Mc)JIFsu3KR3O7KCx1-C0Vb({&6_c>e)rjKDzl?v}(oi|)JBAm^y1`>tmlsI}^g9yDn*ONP#
zFcBp)?)~3eY2vSwDBXb|jF@(Br8Q+ZN8~weMbW%IOr*y=P2h9CN@#xVUP$&-C8~U9
z>^D;{688(Eh6@j{5JZr28kqQttXSV=8kbyy4}dxcRgy8R>4_hqJgN)1`Y+*U-$=nm
zssuZ8$Ol*dj84iIa*)`A1e>ph1(6&%>Ct6kMJ%3474OwlA&ye&(J8)>AzrF%TgV#+
z6N+~)3HT@k6KikcB%D}%iSQ#w>Vqd733G10SSvLHBBTB8HHUc?;^pt7KT1vM2#J(B
zxAQ3{h`huC$36ivg7#_rPZpb9Kw2TbRz>Qc?E2J|yYyI~7`q1ONd5!*>Q5T|Vmn4`
zvb1k1o#rIgPhEL0HYWt%U3p%p6URcLmQd#kKI4auJnEnBwemydd%^8x;;hi^(%G$d
zp(L#9iQenxxr)S(X{)0kXWz-EtQTL66N%;m2*M4`)@*b=r|qELQB
z#9NWF22{V)`EiMdo~Zghe1!h`6mmQGXy%)YJW=aVKp8+Q1(|eTj9pI?hgLpZ)Y-B0
z1jDmihA;oH5Iozt@vnFh8nTg84ruxe+zy0nmd<2>q5ui9yYnwmw)Y{mGVUl;Jh~V-
z_}@74)i?H|<>^3qL6^Ox{31YAL~;0vz*}IO$onR_;Xh={?Z{xKwg~2P)l~uaH^8^m
z_=D1;7|p8=B~4a_pxCJqCB|Auu)`wBp6xz{&bTKJUJh;oic`@EXJ=Z0%D~_DV4J6a
z={V>dd|iM<{X{>u_plJ$87GFuZ4VI2apI>w%~yl{9S)t#*M9kj1Qle*~HoAldaB0I)ap-0RBsbwrDDm?9b=ktLRy)=4zj>B?xXE$+jDNh6Z$R
z2XUB7p@Q_b^XDb3!BL@Pc4k&x^yq`8Ek&IkcxX=7>yaJ={7-aU9!j-9C@oR#;J;gl
ztI1;3O5!dEz2RFy_i`9;spk|3%cTTWrBmZ&w#2-u&;1}9D;s!PJf%6RU!inXRDYKjg>zUm`gn8FQoWa};6p^IXGVk4q4+-?)I`BFu}m-83jwGUAd7F@-=
zvPGZA%QoZ96Hm5wZk$K``7BNX=H%ck%dh5~fm!TK%2fNCPs?~@rR&;^&KxH1W_pn{
z;hxvp4_rk-(Uqxyxi6zQ<3G*DcA^vCC-aEmeE1nNvdgjabBe{?DSK2ag%fe3>X0hx
zC#%?iURB^~R1@YakJp`#ufXprEZrgHrm&eOYFSK1;n>^C#eyYvGPEYgcK&Qy5$@ic
zuhgSZgPn?Xj$Spcz|}Aw>$vd8*qo|R!_8mhVCIM8?xlaf@pAD}i{|TY7=CanNbYkr
z_U}qt%gNt$_*UGE!6&CnVC;qGWJY-h_J%@{GI4VQcL*sbOEFo&Ou{4{1*+HJ(l15p
zzA_qO7aTcPHj~}3YMTiY2LI>yKRs_9j-3c>%G{RTSGF5t=!xLi{%3@J6P`;GdiWTR
z^U+z*o_K-@^*@@+TH(Noxb8`d6*c2cA8%&UM+RZKrIZN-WM=di9_3Oen_Tc%iy3pa2_Lew*b=>T5HZ
zCPJ*zKVw~7&s@;a5?n*7vCi{IA=W}pCdp#ofJyP~kH!QIV+P3!m&wwHF>yzsn`C^0
z_-+5SlOI;cF$G4!NT#2Qcxi~}kxSwn=%22DjM_jH{$l;ON$X%fc3$q0ib#A8&TsuU
zll;#F{?*O&fy%>E;6rsaMSj>grgHyXds|aGE;Y?|nfR|3JG?8ydDeCv`^VB_@VfRF
zKCJ)PI=cKYFcP(@8JeI%qNjGjNC*>nJ#t<WY5$Ca#;*FD&N`iOooYXja}kAFFJT
z<(KmP@6;E;#=QYIxd{=pP`5X?6DAAX#Jf18noN|Ipb$?6END$T0A~4f~0u>jqo!~#OM)r
z{95L7@OROSLCz!q^(%5#$w=l*KoFLO$3t=VMbD;4Pt6
zX4C!!DVpU4(8%W?t@S+<>V^zZMa$bax?g}OrBCmh^;CiNIsWCiihiUx7C6i&GlJ~d
zJeICJOh$s&sJ3(-hogIvA7ZHKzks%%`uh}hedvp(=G}yV8pOYPEj4yC3|ui|loQi0
z0%J2@M}_k0&|9;U0WT)|fYZb+n=;Se;MvUZnBLoOfZYT=5sY)~sgxhBx^+~4`PIQ47SnTAQli3rM-AZ`4a}44hMp
z>pM8Kh}QZ{k3YutfKKuEC%kS{L@a&oh*QcT$h}YR(DZ>9&fDCMIH~aidHor@KRvt!
zY%kUSyL*a?_-|SNvtuJG%<1zI9#Omq^X|6mQB;$Wd?~l3i>@9dey&Y9X?Y%kqAu5B
zU$`D3ehUj3(Z!sE@$rHOe){zxgZ!!==L7bDcdUFpgZh6&*2zO0?@K9Q-l5&KqZ;Rl
z_DSu_t-NQ6r%b9opBm1D$7T4;-!)N0Sxzt
z9LB-)ci4;xk;)$nN~_ew!wn5H((f59ldLNQA>n^mH{KDLBz^k)6?J%W-$yUN
zM3-nZvNAQAeiyPAw7uYm=f{9`~kf1b58tiuq
zCWhP`#PZgkLpK#McN%jK_&p*=BFrX&klNm_|Ir^$)xjw;zpW0X-uddZBh
zzX2&6BN=hg_fG+#{F9uw5zivt?u?u`&X_^;vfiOI`Cds_bCxXpv#cO~$z7xh7Azrd
zoj*618D2oR!g1l1r&)wDe#J&bs*2cq@>bvKNEz`_Nm^KG;0tltR`PtH-Dl#tu0zR4
zZyv#$o$SxZA3|^|+kHQ@8boN^Nn-o^Ff
z%>66o>3c#V-@}PnBbX?!+7~-}>jB~H|I=@y0
z9W+s{eF^%u(7cIH*C3Mo-byUH=o8UTJ^or;!iZwuD?c)>i^0L&{>!iO?!&yBx<{wZ
z-iGaR-<8>oq=^QVvDIH4*NC~6an-`rAQ(186L+>F3=XoGeOgzFfEjg%Mv$jJq|9_p
zFMI3^PZG`sA1sogC4O2<**y(D@K8B+q&*FCEL4;^b$^2DTg!9j`=cRS-keo$QWoU#
z9+%E~p93ib
z*qju43EhvLtgJi0Pe}6o;dAX&A&y@Yzh3-P60Y8!p7vr&hUu>9yop}XuqMB{mFN0x
z_-5zL25*TO!8&>7q#(Tv`o!@t^#I5YOQC%b$Lrz9_mFk9(uQaY@80
zRG|ihpWtBBIzvGyoV~}Of1v}Nztg0@`JxlObd!wT_Z>v{FZI1SCEf+TCPi+Ef8>Me
z(m~3XQzlWjsptuTz-OfAwRWWc(_diD?zv{7<_bzzKDH{82;nDze?Ba9JdpFVU`(;c
zF@nk8N_@)w@
z%bY0$^Gh6O9FjhRuA5b&%-L;dTWdwk(EB4$-nLfkzmNz@cYV2-PUoU~&Yt{j3xgo1
zh2w^SL~+Lan_jeutR5Q3vfZ-&XW@
zz)BIqcFP4vR1v26k-{a&$n|QcyDSqVX9`?e(F6<
z^VPPn@B345Y*8s}|F}I0t%*DKj=uyYaJQ#;#HE7ECmfQi_e?=x%4Ds&n=iWHeIVhe
zf%DVbpo8kX
zAW!!Q`ghALO{5?h6#lAGNwvxWaf2Os-+9Z?4F%som$B2J@wiw)_(LObHMmIkDw7Dh
zH!E1ZEK82ghHRYtA4BKij^!7JakCWJh05L|TVINvJrc6_
zIaWwyRYnmKvPV)fe(yiU#!W=%d4#Ks6dmROvW%
z;on0IU~jm4$>*X3N^~x)6*?-0qD&1&INwHt9@~g1b_NTAUoBmpjoc71RLQyBv$+By
zCbx=i==@fU*-gE_u#t=kzfuNy-SGkZwB#PAUfuwfS=tj-RsW&wbh6I}QKlg8M*R$9
zD@IBazLpXVf#@ZlNrU{QaBxojfO&Y^ACVKA+&J^?CYbFSA8f9Z1R+ASCpW(RY5r%Y
zbl&RHYIDX#qf-gr))Yls1;(<1CYm$1U*D|Z7eM4wPfLsModyS%#usMXhMR>%w8A-Q
z=b8_=;Za-b4Ml2szLB5VyeM4drY5nV2zp9Rvfwfa-7rN5%A8>H@4NA
zfsvq`av3%ywA$gi+eOa}j7DglYwU@D!r%L0-qp2=ZzkexYL&~HYxy^e))x{LQ)iZI
zSpI%c9COrda!lDL_}m{PcHHVvJid3eEPZvjndPsx#8LMNMZ31{sZ_C@=ByK+Xnl&e
z6~8aI*$44{RupLad*+&OSM$S)7#EZ6E=38)=MvonUyVsk)06(SvXmEWn2uu034wbH=($JRO+!+TXa>lw%Gb{l}mgwcX
zrGEsU7v@yHo_G#y-|ch#&4@rMpL?#-yM=+&FJJrx9SZ;|UY>g$R1S6r`L!HL%RzmT
zw_AERIps1D+l7V-
zlp*~hu^U2UlK_hH(!BSaL?2Ifelz$n2PWmVnYcZEp^elj@%R}=>`VXP9xeAr@S$3;
zelB+!dCOFqKd9LPySo`H9B)qH>q6RIJ8!u0+1fiL0sjus%dL3TXB~TBuE%ZG|NSA-
zuM99zuiZwkS7>f!*b76ssJ7>EvI8jd$guyQ`Xca6E=%qXI*MVFOqW(F!6yPASAQCn
z!;hjDMxM}+!-b;m>K_l7@uU)6?|BIUxYl+xtL~T*Rx60Q>A
zeu)~+pQx&oVmgkW@(OI$3M=D~GevJK#&5xSmFUipDk;oYqF~XUsSHnePY3okT*F&V
zy84VvW{|eRtt;Rm$Ky|q9+Z0;!%(pFf<8kB
zt8sU)HeQ$nV#&`8URqtnv{!sWk_P^RE(v)?FY-Ka@wF?wQnzjKkwfvg#V!_1JD1I+
z7HbqX#WvlMNk+$MQI{p%h>I4_?_EevdJ0T@qtROEW%35||#(=7;T~g!DfD
zC2&!w?vB%ZE62GuQ`N&CT5x{neThf04VaO$+N|JwB#auhXK}uniqX=FWa^t%tXxBU
zg*%`HuiNeyG2d;+B+NWC{rsu;)ua5BPE(Bk+xP)lAGG3kvC5~9Znt7PoA-L;U%RlK
zn8)iXt!1*7k!iBuSOLOsdISM{sy(4lJ6yql;M`uPZ(aSkFV>n!y1R7W2iX1-dM^R@>%
zM3gZzDR)BL@M^g&vJK{zPw)O*X@Mpxd@{d!a$vz!bUyptBsk<})U|op4nC)~t(9QfC@y*4oQECAsl?LG(^bpeiXb&CaV$zsZ(V>#)vs!CWlI-B4>;V)v?hn
zB(+e#S~pyqwgO)C`+bGfH4S$C-U(JJIs-MVsDuyFJh9cPKF!5N1fO&r2@7OwAn?}@
zc5Mjvh(c2{wy3mdcz)}id}p2#HuKdup>w7cUb|0~DDKh>Sx++SSaiRE!E0VBBG;4Q
zix>53A=iRo#T47VQUioKg5z3bTS|oE!!ayOlpn93s68ZS*+WeoHT(M++;BK2pL=4|
z46h$6VZUmm4eK7AueePv4tG;0_5Cg7aht(bx+Lof2&rH3Qw!b4WQMd3E_LQugVw&L
zVM`C%94<2~kMm*6Q@vxPed_o(*XXq1sv=Ik!#v3)=84~Q{^Ix=b_}0iE!fRzH-I?g
z$Oy%t0*4`*TVGR-iC|O#zq7(N3`Uu_;1d;UR+qUkR~{ni0(-N;$-?~Nd%+sU&&o&~;scr7(dju%SDT1mR}
z5n;LWey&gG7m>G@bKrgwF>LYCY2M$sisNH;&i(Ty#TOp@$sJ;0ha}1G%a66uVU@LC
z_gRUy(Hv#g_`^08Z1Ar)c8Enp?m4;n1=|~S(fBt{UHjitrB?4-S^RjYwi)gfr<#-YaP{(`-D<cf;h%5#K%PEOK{pD7>^+V3p7d7DX>kI;{*LXj1TpcexV;Vndh6
z=b_-mr#a~(;p^z~i^Xo&kZn+Nj`7QH`h0Z#!I#Pt-reYqzQ6WZM?V-klu%V9X2&!7
zHsyDMHV~g&IOIRS0-h;d9P3d2iby?X?^nrSM8o`!(L;wC3zVdtvS?wzp}%i!4-GGY
zJ*qFiTAT*}Zw|FziR&VA`qy%|C#n(!*qE=pOXUF0JGEl7zwd)L_ntgp10HCPhitF-
zuO&$T^OlKUB>{=WI4mkWe}$NyBcjhb!GJ5OL8Y)<8}wwiMo8xepKS2A-6lDZeBUex3ixvpjp?ix4aj*ffi%L9gU#VKqEu9*kdyo-MByd
zUE+&6V5O#GxZ|M&PJTRD@m`DPPrvNV1KfFA&_ZSR?irsLUcR*PcKUNZ5opW|qj{b8tNE2m@oZ4W
zf#R&9`{Q#zInlW2TCDf*b)c*e-fj1j7KqxB&_3h-+59j%?9!_BMzc&>!1mpAZopUM
zLbKCj0OI?{ntEbcf%fCoxXtj*qBfG>0Iw)FMxQPIG(iP9CUBd<1
zZX4g+loBL3P*UIW2A%^&_NUW_^`udJ(Rs@lawQ~6FIeLoqXu|#&-b1ux`@IGHs1cN
z1K{hS+wb^$X6USWsM}y~AUNA@#a6|j2U>nVtxOkFMok;Jr^kIHz^43b%1ur)FdDgV
zU)8`DkVDppj1g<#n|_b(gO3gBCZmY3dg}rf@5NX1Tn%y*(xhsRRkz}Umfaex1(dxeXdl|?~(K-=lI>2EbuKL
z+=62qf~-=BeeDA`q;B28IaSewMv6muS{TrW
zyC++cpB(}*dukUBivEGlXiwW&)DCpL(s_^Pzd^&PG(YXmW`N=aHiND7cm_M_n{LgPt$1z1wfv07|FySLrEe@R`rwCdK2$F^3f6Px<&3v_#kV
za@S)9aF|c#-^y4+ml%`0XzlAz%BeU_MJi#qq2x>%)jbP1lpHOL1+vhQujhn>Le>$o
zurU-n&JF4QdrjlL%Z(YEpO;^(J_NqsFBCXJa!mJ!Pg_rp2F@Ze8|M#Hm_@n0%6eY|
z^RWe;u4Eeo>w?C~I*I#eEwSZTu`@k16%XHf8770T|2UsSmLY}3=*>JE*W@v^Y}vw>
zQ6)GU8&}D;avA$Rt?Tt{)PcPQ+WyxqA3_K37kg&L?08xsOzy)bI~;lD{;91{0e`sW
z^4G`15tb*N6pzpF$2YyiXDXr
zYlPD;<1!iCP4=Rr+)e@x%NP&D=Sjg4?Ogk-wceOMhr-DGV$*sh!qk#yc^sw|0LP_oGcrB>qjDI
z@?I=tQq3TqZD_@p8s|>G`s0PWm_k~Z8U3+l!nMM=nOMk>C|l#{SAh%YiHfho`f6A0^xwSa0m}|3f6=UOm2??F!gx_Ci{k!se
zJHDvMvNJH$iR)ax7&LM=;9)DSB2MBa%+#nimrVFSnj7C8y;t3ibI(%dn0UO!F`e|E
z##Vaazp{a4E%qMREbydRj}<8f{wDd#)13;{606_?N8t?(I#G+PD^?V
zN1X;&qh|PC
zjSA8azkv^G9+8h{BtZ&cxkL|=K&TZ^mSG*#29NQ2UskYcfyE-?eFt}|Al-i1BC%aI
z9BqL}RX-MX##6HrsdvB(jfBk3#a5V+U;Bl^svf=p`sS}GGNnx
zp$@@jDZ1_Y=2{r*mM#(o%3<6@Z2i1b5Y$QbVHWx3fXzmyCSS_m!lxd2Wl3E(hj;GD
zsXpzx0^hVBH+3S3gLfT*>Sy&VVAX*lH|>ZE77pg`t*yKc|6RDRSlMI-AL+X`pP9>s
zFXB{JPdti+{`C3bA`h%_R_ZDLAc0%BBmCr1)3oa_X4fjjX^wE;spo1HT+f2`wA~xV
z?>%AV8OOZP9A_+FX11^7YK*xkd028q${{}X@h1Vn2(5^2(M>%Mfphz(e+ze6;*&o2
zyDAC|@iS+JltMZ!xU_2SS$0VUU%v2?bN?hKJ~f%v{79c4vOM32jCMYbdH?VqZ(_58
zJezcFo3i|{xl(sY-9#3bADXvvUgL({eF;eWJsWgjVl_XlpbIOkzR+mM>A}gE27>~9
zKAfWXf=%KmA8rzvq?}FSf%2aJw&$t;!+}43bdMOQ!w&OjT~~E3K)+k^D~c7;xM+-x
zdc~Fme;o&7{YV1daOSUP_Y8VzJa$G^R3#Xfm*>G2qGw!QX#}
zez9~3v~Y1eG-m#TTB6kFUt2Om@z1#(^md(y^$Q0rllD5g8?O9w>B|~;a@He{Op6#d
zSCuJW*4hM>pUw!x^APgSqw6J7T!b
z9@Ks(2knJReqkG41?2H2$z%t$NPeBw`ki$x%G2{{RJ~FKf^UkV!JQ-|QSDcyqWudz
z5>n!>`8o_<1ktK|v3-kfpAqW&=u-}Ee9f5Rznl+@OO}a(k_Ldq+8nPEZ8Mmh3iBd|
z`6%v>F1YR=3Z}}qt%na22syhwr;WM<5SKUY?{zvD_$Ks5bLrS1Mmo19vy+#Bx9`?5
z|Cg76zSwMj=d(v@fWtT=!s$3VlFa$A(B7Wa@d^8%pu^u&7%xrog@
zS5(3_0mY3z{aXp=pm@A9)!z(9hwpm>bCr{?Hr8U~+kIX{^Vf00@Ef`IWRCa#>W|^d@@vh7vga
z{-Q^2<_+Y}>*e+PKns~XSyVn@avr38=e#}G4pD*^tWix@2WK7mJUoqTki@_lY3(RK
zVDNE9`{ce9i2R!@$r*AAxm1hQn~{?NuL;?xg&ke=D@6UvVZ?Lb@mFfDW5x>L#}%bH
z%-7Hp$1lxk)wjUHMQc+Q{aAFNpJk23s0H-f&e(~Yx&zreyU&|Xx}m&lb55aq`
zJI4FOnW&ofIps)0HP|SMa4*pD0phijA(ttyqm?Y*Q4?nk;DF@ktwWz8T78Ps6WW7-
z#zThb$#x_Vr#x+i^mGuVZBd+a>pk>VttridxDYT3@|3!#41jwgkx#>vUx0{8L1O|_
z1)b*HPX8?a2ng(q*KF~;2FY6?F4IRlP=Q`^^!1k=;LW^OL0womN}v5Ic~Fo7#ySp^
zy5c{Ab{Y?#;@`7qPyW<9om>K^UCha
zIP$%DaR^L4@RV>Qdxt#p^49w^-vZu@+ckhKr@&ayk)~F{68L6w#NY8=X56B2
z`MVMCAJ7%p@}N4K49`jy#Kou3LZW}uHgoPA__J~U>5$LT_=N7aS0@9G!jIHKzp`U#
zF|{1?uZ>ME*v8C|z+$h8)u-xCxCH4yfs2e6gA;bqu$~r6e>E4hPY@fAIVFmj)epPn
zmv~_>h3rp%TRZsa|p7}#(#AA4g{`w
z`@L(+Tricouvx883yRJamUPz`5v~A2iaEgWNQM{@ywW=^>)h$kMP!|@h*HZ24+GCYdreYzp
z0Z=o&hL!J(7p~za-3{VV#ElyIY4e5Bn2bp%GZUNQ3A#o$l^0%+`0oX-e=D`v)f*fA
zdkNr8>$$!j!$2%r{s2;#g+l7Mmhs4y6#PrhWK|V+VDwYb$GA2SzL4Y)TYDCWncufc
zyLn)oVf)(b%pZisM3?)1aWvyADPa{S(?ej8zzurlq(n>~eSyKMwH1%2y*$#f(tu|l
z=!m#;w_r9C_Ey8=4ltL%r={fG6l62tk8&L{A(+vcj^z9fc
zOCw&E=nXjbtPTsCZVu}hR^g=RT=Lux?U0s>!jvPS1yU!+%adM0@TXP4;MulXI9WDm
zO@5;sp0^?@37l+!H_n#~!50W>fO~M@Eru5UFMpQGH9!?-`4f{1l~C9Fh1}Eg2E=BR$-j
z-vF+9!Y$BO`n&yQV+51W44X7oBAC^wlrKw=nY8nXkZjJ~7jyEg|sV-6X2-k+xXGq>@$NE
z@fqQ@To^7=EWF=iX@bLB0Vzc^g0GYp1fYjE;hy?sq-}5JbvdO
z9h0KAk?$&!o6YCDB0LFd%U0H_ocn>yX6lNrbQOZaUVKZAkAQ>Iua_&<`C#U!)sH}e
zA5F~mu~1d?90-kZ{*tUYgp8Kts06sGLDF3ngCsiyZvBfyPi#nVLW8YiknRSkK=;|C
zyfB!R-1$hQH;6v0WRUZRy$97g0&foERF54r*_caf~ZQ7sfxnfC)yW3V(pk65z{FUW;yI(&6ZhJ$r
zeS3l6%&$U+yyb4B^|Rf!^;EG_Lx3>n@43|c$x8snwnbEdn!xO+J@v%=+dKIxn
zSnFT3dxa8|6n^~YR0Fv0MajEly+#6;whO1pe33`6#y4x*^FW98qxrRnVUBd`X-8X+CJ6sMwJAju#@EHNWT#w1Hdq
z)=tQEYc%Vy&Au7r29yG~t!5Nr(G#K-GSWbT|9>Fdn?+p%q_6F9MMqsg4b)fjvE-bV$VTtF_NPJ$U6EW~rc@#x2FK*6${2S)
zqU@iIqT}@=;F+a?@@njN^nOinipgjT>;z@4REaWTW3`*~W{?}N|C1*cjT}JDj)`wO
zHRpf?=O;rl<4q(?Nxf{7w+JL&9pC$4B?&dZ7Xg^T39I8D*78l
zRgY)c;mJ(t_ot<~@XbNe$46fO1)n-h{4d=l#`^r{?!3knkVi2sW=fa{XCAQh$T45S
z_5mrhbZMy&`S_^OCHG$_``}Bq`X)qZlA|@i&dS^
zVkJ0m^8{y~m=-R1-?Y|kb_=ou4;UUNbB8%+ZU{9Qmx136{jHjPM=)7(FORJw8xA3P
zQ@Hfo3KomOjAMDhn6uO>ZEE>9xSB?VBg6sjLQ4`s`8u#aJMx@Mo&(->6Zg5J;)S>E
z4M;QVBykbD
zUm_kIcvP5t*oNJ>r4@>aJuo5r-t~3W4a>QTPL92K48zG4u~JhtK0oea^Y>gc&I-EA
z&Xr<-OH6jtgY+N3+@I(6E|EUR8TUBV=>!oY8y8$&gvELUfQ2WEupZ`7nxe^S`
zzlWR6KX{F~){GRqCd#pbh1JQ~l~R0`zE@d4$rm;pJy%m|;*NF9JOwQM3$cpK^ffJ}
zO8m;n%y#f|Gd^)9?(1?+J^pTx;rF$*9@{&aHWgac;M60A
z_OmN3_>b1ph^wykczpD|mZfSP?$en)TCH4<|3;lM*0HR`ui|ojYYD#38-r&1oxiJa
z>0HJ692qxx9gc7X%|%f0V+&`^t7g@fflgav$E+Zi1)XGrsj~
zl|wtTp58@^bXdBep5AN_4lSJ*3%F|$yqfp@!WQ9u#9}`6v0bYs^mJCGTxrUMG)a>6
z@7m(wr=GBYUZZ+Qp}O0zXV3_x^H=Q-r%43fltZVw
zYiFhjJYsI*U}@?*)sT~J|jVjb@{J;O;lXJya
zOrKtdgc{5be*+WX7Ms!iKUsGnMTd9xGj@9J%7W{tsu4X)SAvCt56_5Iy4kwq2*2jrG
z;4fl{o21ruI51d{RCPiZ_iM-XZ8sOfd+AyA6&#sR#htWHbL#bD2|Go*Ik>ZEp<%c0r>}0rq?@*(k$r4VF
z+h_G{@WVWdD@z+R7YOrz32(wJE=Xy?_V*t+1EY+!$zHYHfUVu5zN72c;km<8tamB+
zabWT4Z>KO0+)Q#zDnXVTR$S&UE4prgbIv??ZhTh_n*1;~BXyUAeVO#D-j44@*M#|D@mf_i*FaK9_7%Zk)tU?@hLUo}|UT
ziRuY_YgBmbqDuTq{{k>_EpCdQE(e_wCONN2zN4R2EN{0tS>esvJ7Nb`B)ExFSapeU
z9vyY29U|6P0>uq~!(B-~qYXy;!3Syl5UaHFcDA2@X?`Jj4#Rthh+4I;!saV@NLe&S
zcXS>Yrxw4Uw_ilAo6&sCznP%a&0t#d+F_)=n-uJowTh%{9$$?STnC>iZ_I6dWX7tN
zzpttMCW0Mrie5`>FM?CAV}ArJtsq9hP=x@wcR)8fQtbo$fi&+2kGp6G0y!Ro^V7S<
zAY9{usW`!FJpQ#YQrpN33<_L-s7y44gnr-VpD5`=Dj{ZHJ3g0#B^Hh)hOo!T;D@g0
z%CU459;(&5!}}9RWo3~nv0_xnIm5m4ijV`mr6S^?Uk=zmp1s>A;(^v(gsd8peu2Hl
zp1G61hCr&>);~F?x5(tkVqQjlIdHvox>Zu`qtpfA-(I(J3HYuj9d`MMArPT|b;U$b8$AuKTKZ$G
z13XfOeH~WZP|H4dYfkz8KWXU2ad{VKV!P
z@Du{wpOk_LqBVf#y7iIdi_eidr}D(rN2%ylvyDce>o5>0O^f=bHw=P?ki5z58t~@!
zWWYPORMc(hHtz4^gN~4$s<*0s59k-w>pYLX1v4Le7oVT41_~?Ym!=r=(3=3s>y8mw
zKrEWU>Q-tFco4M3v^-s9>Xma4<(KO`!h$&<`$kRbgp@bPspe2V
zqV*od#0(0n6Cx3T?8;e=rf5yNhiA6V6bZ@jQ0P3(iEN6`VAe+=SCWft}
z@Dgh+-R?(DOy4y;MDmEBr!a>|8HG)OsG_N2!H;vOcZOoxCFBqIoc?OWmY5mipD)FI
zcuvAgdkM9(qJ!X|tcmw=$0uZFw>o<^aT$;jP0hGW(P9VPgotGeE~r{iCeQAhilkVH
zo-H%JN1i70%HO?~KZ-VYYDdY=lQ6#oF8|iAJ&6~|$cl`r72v{$kv+w2HGCV|`E72nt_7%$G*P=?SR3=v-8^b{ID=jx@KGfEHFOd%(Ypx%^
za+8pY77;&R_ErKP^Vy}f5LJglgEzUd>rG*q-mg@XSA_GB#3t{G4Hpb5psS}6ki$Ks
zi9I?lS75{RYtHUhHjsqHU)17+ALO^sy`Nwi1{9mlzBrjtjO}g}##{?7$IrZ9wU_?R
z!?*Qsm4+mH!Qp~3%VnuMSdy_Z<$(^i<>=v9~RL=tRRKo9c+0
z>}-f@*Ggf*9}3(4<`SsTzJBT1kqUUGlSe4y+8f9qZnYKmxe$)!s=bu?Sp>y`ZRRZ3
z3Sd9|FEciRz7dsEMts|_1jdYJana=$K-jr%?roa_sl1$Ss}k}X8{Z;}_b9XBK*QYS
znTcFjO6XvVc%BVoqz$c9YBe!k9C_$mj}44CA>#VH_A#U+f4008^8#iDobi0wkO0X`
zZ$|dWMq?@FIX!y05NuRw!fp2JHnjO-nIYQu2o5mK1r9_+Lanwf`^uI=D0{`H>w#A$
zJbvoBrBsbSbl33fm7B1{>GpW`{iU1u-%8i5-${fXJ6lh;&o!B__!Lpi*4JQoT<}Yy
z6oG5-Y}JP`zS$V_Q0>Tnc$f{P5w-lS%rv+}&*fI_^>i?e01@MS2^Sk|Z?WTw>!AorDo6r(qn&i_=y{H~6&ex8x=N%rl#PgFx;
zt&FSwSszz;^qiLI-x5)%y-H!z>j`nIvyMiMhY3~|d?WZ(>m0l|P^7Ya^gNXAy0M*n
zQy7mL&xjcHpTzo?uNRH0nc-1Qzf+QqT=?0!I#9uJ5#F;cBc8d-jfdGOGnZ94anMR-
zyd&dHY=67ySqfqA_F`6~ap$W9JminpjiWe@CFY0vw^casd#d`FxWUVK^^AtA=mE{fI#iM
zKpoDFna5&tZ~7gApP?o0wm-Ijr+<6u*{w~KAQt*LUcLu7F0n`nEpp+BfBtNa+u5jP
z*!-7*%P+9q-To$7WD#jsMK%)|)}eW#pIxKxm~dr(;k_;DdE|Bvo}MS1LuJbn=~H84
zNbCAbt5myglxFF(Ql@zdJ3RESRU=TK~2JZ5x8he5#6Rd0wSkuV=|LQz}mIy
zqhl1+=-49%y6-z)Xv*wc;9klrKy)l{+#?|o!9sYSmDDi+)PfRol@H`(q@}5*5Fw#<+SL>-EPicb|
zo$4#VV{E16*1%nq@+$CFgK!Smu79oneO?m;lCg{TmmUY~!ryxZDfp1zgssq~lo2Yr
ze^|K|nFEY=QbulAN`SG=R6feT63Bm{4_}j{0qkB8iP-^$ph%o9>fl-+;uw{5UTY=;
z=_kg2{313%CN;}sqRWOrkMsCe&b$jiTDogb*eXy*MOMO^8vtihcFtPw-9r7(^>~|f
zt$|`aP2Z-bBhV=c^o%R0Mce`Ta@TtD!Hcu~PYr2OP~H4ysDMcVxKr!LW?P*P)Ll%l
z*7pHqM)!kaZ!H_>zSWMme4dIv$wb7o@g;*uwM7C%=~8Nw*24-{lKAsaBz|vah2hGDTpr_SdQ{aT5CHO5vsE(2wY4d|9`Y
z{sKAjO8=d^B_OEP(XsQv5K<91d{_B)5&RvwOEfk?jZ37$K9ka%gi%e0u_WaK=v&ej
znv5|-223ygipwU!%^wmCXI-dp{^|9Y`p@ie)cSLCc7P=se?GQ%YpV$a$g`_B6%L?u
z_xKJL`BjkA-o;O`zk`BR6$a}vW&qWlJ0#eDc?peX!`rAtz$1I{%b4!F7+QHT-sax
zrv8otMpTkmu48FDFJ^eew@C$_%P8mn;(m+J>p95rPn^I_NxVugZpQ(i52yB1O`gYH
z_qieht<~YHos&j~x#sZF%L(gR&UHlYYW78AVir-c-{CR;wg&D;e{njeB!z3ukEssE
z%D@-ApWY?bkz#tqryb2{GMM5rNv)K*I<&qPbd&o)563;5qNl5I#PRn#mp#VM;>=Hp
zRhQyqvA556UHU&)VS(u%P0|rf{HXf)A!V;0R({5q$J=xh`mybObNcIw)vkM1oLUdZ
zSC7xAYrEaT3ZDY(1@C%by}+$>Txkehl0Pm@7TzN8ek>kt#L3}bbyk@)H$7p#hP&t!
z=}@Spnx{!)O$xt-M`v$TxWakoWx|*i2km}nEgscPf~S+r_wR+|K5S@@xKpLj?NwQK&4e#zI(Ey&IOzdwE9X*~6DfeV
zqW|N2N9dX5RH(53vtI_yB14I_6&~Zb!3#f%w?pu7u|(+1nj`cmdL%1uLC7a2{+4+=
z90}QsZy7VbOUF?MOJ8IK6Y)vWYB8eQ4>9LH?{~T-Ggy=UqRgYn5nAiS{K{}H#qM%M
z{x6OeV)ebD&zDuwaMYXkxApzv@JM^i9*fsg+-2IRKSNrBrEDkeG|81?uUwjsxh@6x
z*=h8k`cn>0p_M%MP%H}@Wl1xXgx6u?V2PJbR+V_#>%WQ{QLnN1>yU0MnIg<^LGAmm
z{X876HOU&&8wusN&R6u-iBdAG&<)lb`PZnt#nw93E5S{7b${5oZOkjeN3)F=0Qu5R~1?>31B;
zR|veN{k;TS2)t9#m&ADBh|gZNw+b}ieEP58xD=#3z7eFn!h(yXEd{;xzoQQhsibDg
zv`~|NIl;@_2u|5qbN{EPMEH*QI0guLm4J88&k%nnCgd_Cy`(vLpy!))tqFNcSmxUz
zDmE_-bu$T@)^D=duh7nBeV`H-93n{RM5s$|;~3Y4DYau(uZQkFdL$
z+l&5s2#rM~F~A=)NmnEPeFu}h_5VqY{Xm0Zgw+%0XE15RNzp0X3S_b~2gAGh;Raot
zu}}ODpgLQ0hV;unq#kBoOR$d6OR=cPx-y;MhQE7evBez5I)8
z)Yn|jG;V>zHiF3hE*2T5so8RSccLSyzrIg=sspi)=?xAwK03BK@Q5+^B%@+W9o{nJGIP9Jy|6ZA82>?5+05cbG3jRPml
z`Cl@Z7J$s1arHsRJX9jE%WEvqj@XcrVT;NrQn)J1w6$9TuAsl=b{5H?$9F7?m*f@F
znChbJwN(XO`W+t_t~et7@khmfa$JGx_4V(Aq>hN>aNRCWs1f;JlGkGzKM&SLHe4O3
zCBf6elYcDet&olPi6@07u^_>pN4QP05OsQwL>~GQIC{r+f0D-k2N*j~aXxo*M4k)A
z5zj@!!O~xQC8?TJkO9+A4E0ch2+8yQ4MwDhjP>VZGe>10==g0&zrqW>@PAhjzCqx-
zIr0a4ck=^Lc4juyX;rlLdC^LdQ4>rN-FhsOqlZ$a8;--t1XO?9Mp%%u7KsYBntf6w
z@Rq1gzYFcCM&%J<@!5uDD8b@&*w+eTc;~yc4DV11@=ccCu@=ik7hNVlkc=QecgFHl
z@5~IENHEp;CsPG_bce$a>Yf8$b)I0ejMqq-V`lI(XFYhIt?U~4z6bsHBqNJUdJEAS
zu!J9{x(-?%?Oup+hyw?7c199;g{V#=_KUg40C0$QHjg#CimuS>D^*s-q70HRzgPF3
zgSRK=Xl|FKpmj)l?rg3nsE$$YEgWA2(Pbyj`lbE@5$AqttB0+jca}i>i+U@#>%#Cw
zhLjnWWw^^GYRTa@d^;^YaMi%k8^rcDLNNFVhbYOgB203Q
z)jS(Fg{Cj-xuCfXAkh(Pym}9VecQV5XGU|47YjICZsEZ&oZv*IN$vgsdVz`Vm@9V`tmelWTd`z1(_Khp9#p$?=!I7bdZ^wG5ImhV${4rCZX|bkL{fa=7d%wt
zhuyGDwsBs_sUT?YSnu6V==nQO?EZs-wG=v#rK<%AmP3mf@;k;JA-J;rIDf%68+d&z
z^J~uDE6Butkws3b1TOZD%=M5}Ky_c+W9=Vo@H%6UG|h4_o{1_*(7croD|p#c%Im7(
z%!eHql28fteQK5)EA3#1S1UsiuQMKOh>JLza~~R2Uy!h5tAjDl%QT^jbujI5#~?3-
zJ)Alg;CV#B8(yOryHKe91QU9VFKF%(dMrzZt}@g#!QDTpJX7Ee{5{Jg_0)lI8#X}yDf3=pLNB#sxgv$Zl~!mYkbfrVBOx|7m^Bycvk!ai&+|D2BK)#0w&
z^`R5YRoLf7ahDW>GivJaqvVSJ%KTKRYCR_oHm`H;0axbjMc{B79zpmC|
zg{&85_wvdy!=L+`)~iMMP_arc&kf-`+l)d*{bo$QGoCK+rVeimA=;jSQcQcIwo%eD
z2T$I77?fGrj7#Xh1e%#P{EV??@HQ^)F@bEi)zv!oC7`~}f9_yd8hddc
zT3^&=f}f6-Et#=#z<|S3PkTwMA#FIN-sV3RIIy95-J$^E7NXCOmH(&cJOiol`Z#V&
zA(TpzmHoG}GVVDPGK$EEWM?Fm8A7(KvSpJ^Av@f2>`};y?6OiKL`J2@^Xk4mZ|*(6
z`#az7=Oh0g6ehKImki;BnWt~p6&<|OY}=Ins{}h4EV8YIW%j&BCPM&D6kWGV(m#wt%?3pt2<@Q0%kbC9yL&*@WCNb{
zpF|(4K5LdGP+_Nc{njBCC$W(7wVt7%Z(wA8l0}Vi7OaMGY{f^cAr76NvJ49XxU@k0
z(kY*P^j75Yy=sy_VDF1~Qm*hfByw+7P*-mhS)J){sXK8NS8AN6bD=z#AHzB~={hXi@=iJ0pk-kIpXWIbKS_He_gk+_mz)#vn#7X5^>qbZ
zFaC=Vt9*s9beNs7eis;hv`KQN@fvs$Q6I%r?1t>MpHrO<2n0zVZ2X94643$o%0{rS
z8_k7A$3&T0qU(XW)|3J+Ksj9PMKF5+y1@3HjfFZ3yt#Fy|CsVC@FihdW$KeCHeCbFk)OPt-`zmy!#_)L!)kNBeu6}9+*Y7;7M@Tj&xv`)f^$4Yj=R5=f!F`jI(6)l4dxje58iO~
zgDJ7mQ?(5OF#EGWR?*S(_^B*Uc*!z^<5>1))u;icVm&g^7;8#Ynb}0gC&!Sc;I=sVVJW!O;Wps)QXgk7kDl_5q{7^{_0m!!n4q^GCr^74D`vVL
zIieiE38g8W+$6qg!yih2q+h9%LwRG4__8C6xH>?)ares^NF-xd))ke&;)XI?ejOTk
zwwtHiy}=gR)%khX(s;x97%n>1!WcYM_WtXakXmReDDzG3ay`UCqQz_V*WjS>-Z<5c
zJ?uJXa;f{78)kC+-JR{00$Y=+5*+s%plOeRE#ZFq;f?cZZIbr|(0?{_DHtr{gh)
z=5FLxwmvpq8=|7PYyx}Nf<@gTt+2-}Ty-PS9;(EO-?^(F1skuW{>eR#APaq`vz$;R
zJU!FdOZpK*+dZ@KfrI|gS?3P63^ok84!e_0I5t50@7MNjm120T)JZ)hssZYK_5&$j
z5#;1PB5;zU8S*5ac;QeM4gaNdR4hEHf)X|^1{KOpa7&Fv7x5r?`nJ!@EZb%nW&D|A
zr!)xj)Fm-)A|H&LQ+7R5>fmj)D-K2?P0&SxdF0b!1j(Fr9*i+YLZhpDNA0Da;!tpB
zcBwK8a&-s)Hap1aI%mFKO+HP83y!iwC4v~9?>He?u-*i-@*OGWdRw4&iSWYrLv7Hx
zhSuS^P#Zk2R~biD+YCt>qr)Uun_-^9!?HT@7HD~fJd*QS3!LEhuS@A`hW{omcGVu7
zb2BR>O|hg2o_^Y=6V%uQE8@&%o>?Az_CF*;(M{0x^~wlkZzI&yythLqj9|k_9qXP<
z1b4QtTx+65Fl&1q^A!^z-<};qi^V~X_lfIeef>t5J3=o#7t{#<3RDGmlpf@R^Bw+kr!=>thQ9TXC@8
z{UZtt2YiZ>w8Dw%Hf)^W+V2zAhD|?|dYw0E!#cdEA>b&&`gb&(pLS!+nSZZTX}Aen
zFg3Ka_cvk7rl!@s@%;(H<2o!FHu~{>d_4}XrwO~D*nl%c&fA-6
zG~j^U#hQx;`6LeMSF!JftFawbgVJO1TKo`>x?Q2G!^{nFCSD$OSWv6FbtA0n95_^T>OqQ-x`mI^f9c-UA{QAhFR2-fP4YurI%Evpbs
zXr8g6YsB7788hi}b@=Zk?JH!oUw;bJ%7!Ms}y7Ip-4>Ml7W5eE(cBpTVQS3Arp
zHb+A=e&nroOfp;tr#@vk_h`lHApb7#q%!0>=D1TR{O@4iOWCf|fAdZU%6u-MJh@>D
zPd(dtzBna_mv?xtEK^*<@&vN!<05iU153pDWH`dtB$+iIi*^tG${(loEmXy}&1n^r
zN9EvtGaKBi7J+Anb!yks1u^xUFP9zG!KhEiz<+hQQ&fq3@PvfWcXtmVY|Tuy
z(mi8=AAj3QTfC%%g;}#Ao!#VMB%#6v|6#*FEHY~9i+ON1Et_w5ix$4)o$l%}EDj6Y
zee(c`Y%e37~0=Lum
z3ms>txgDuKLBl!fmyP
zsDBJ8Opk-=YaQoh_!!{%IQG*oIbNd+${+bVrpA!S@9EUn(nFxW!@_RGw*-;T{kK7y
z@fSsgtA%%FE}^>4Kr#8_gWxiA*U)%xCHixO>|+}YJ*sXJ{b!QcKF5vSJ|q-uyYPbV}d20YBCtHzuUj3a10~C)4jLT4YR<1pZf&q
z?OlN9vb2eolRu(AP>61hx&Z2I%A~}xI&k}b@!0O&7_iFRW)v$Jgzl`L3rLG}15KhA
zBp<08gDZ~xL2WE#sNCWAOj7bXLF@0`8fA4Vw0d~EIDAtY{1)7qbzC(cxoN*_(R$X=Sx9ug0n|+|n(tS_xJz}hB)-*}5
zV+<#|nk;~X{eLLS4D=J9i%lOYQlBO4PKO>afR~8Mi(iZBymtsjEpv~yRfK`CUqt7-
zNID=g9=frkt%(GJ6BKIBS_0i)ehq!m&k%ed`kjFz2gq6}HgzrU5Ive(URv)cqENpI
ztF4CXK;c)ohHb1H$~^PlK34f5dQU07(ip;r#;F!?R|7wI-5a}n@39OjYl<28yJiMl
ztHYm2D0?E>J$2)^|D8rV`?li^w@(0vCo<9OPRwX|^jypGQw3DuEBAqI+6{O`RGK{Z
z767CUvD879Vki}w@CmibfOiCsr&ro8p%PwBiH$Ztggz|*xG5&^Nhc+K&Yg7GJdq1?d;d!3@XJPT
zwxd>L!UjS2_p@)5wC7Q5#AnAS9&!-1qVfD2-@&
z=$ONd6*ZXb1t@he${A{8urP*~%*g*hC2MRLZ8LoOnG2L1rptUA6Al~t?I+}%=y1qA
zRiW+>PI&ij$6VxJE<7D!ay8$Y+kg)*^p*I^^C&_grocQY
z4JBC9jtHJ7PvOF`qcaTgqzApMz5HhqbAUHMt$IaY8biJQbn!Fip>DHCVQ_l{wCNgu
zBzPnqmx(R^_~eug!tV`EeQDcUY|W~hO4Q1DrR>bpz@i)@!K{x;LCWQZ}Dp7`1krbLfvPjvI;iF&5$sk1A*q^aPDxoiV_pLzFH?6HzN?4P
zA-XDdUgOqmvk*`YI=WVD&inVeOZN{QM?>RGTJC>HsnCE;s@Spp2<*qQ+A&-sSa|?mE~Gf
z)<9E{$GpF(UPGO*D_Y|61#l|XM~=%b8$P&tyKzab3~Jmb*glpnhlxu&Z+9wQ!Thn4
z^i6;B;ppQH20FzTa5RH&h*m2Rip#~R$#rK!t{VcHG%h(XIPIn96N+qTdaaJKT09Ms
z4U?&Ih@?RYBDnTxDF_b4E*;#<&$0J$<@--OO0m(A8x7SSl{g8Sx)N-XVdIeVmT*@v
zg8BS9v@p0Xo*yayM0a!}yLQCNZ7?|mTT}%0mHF{k(PE$=qXcx3S0Xsj6VCV$wc(>trMz
zGt}GC(T%{J3VB1WC%q2#eN94Q_NG{a`F7h2{aPH`hd|+aEw(HEsA8i~hj)myJ)v(Z
z@TvD5>VlOeIC@q~p8Qh{R&Qf+4T!D9m#4lN27oI3$&ZBTzpw&qYnh$M3sUjZ6Jft=
zZA-Cl1BI*p*%G|?=-m6R;dCsi5)$69;f2>$+aK-AN))>O7?T
z*YLyovl!&C`5exZWQcvrLi;zlbg*46)o;oOVQ9d8dR*YyS@=||^t0Dz0X&@brL$j^
zANPLxo)Y%h2>+}!nE0!}0pEg?{D#LQpm)lj55s50@x28zX}^i`T<@A4+!#SxjVo|>nuW<&G=1G;7MN{MQPlB#@)J5R&=)YnoZ}%a_=N+wf*@v(S
z8M&c#DJjfvRL%dYDFGiIzRZ-G%z#xXB_m&4qQZ3kL)v{GDInF@b7QxwQD`Fa9rHgW
zQS9tn|N8`vhl*{CI>#^7`)G4nT(Y`B%pTCg#bcm51^Sa6P
z88Fbni@UZgy%Vd5c14z6_|_hh7$`3X_c$=M(X4=d;U}<8`9wEP`3LYDZfxf`vV{CR
zZQnK?@cmZ{b_$*eQ^QjtobU5qYyb;m>R<}zX%t)Go`12v9_U+%Z_R1<0#X$JNO)lY
zy;mPOS3&%aXoXx^N;aFpCf@Y?X;*QO`w#4j*XlsImH)!-pWi_m8&zewWdmT!==<3A
zdK0MJ&{{csrWM7D8&u7?mIHg00yBctA51?`TS@Gg#W5zc~pDBv(fZ@&gDGRK%ICZg4PR(%8s#|u*wB`$BoraG4%ts
z_xs#p^0`RhD~pMaTP#8^1}w%5<3W$M!!Vs;2of-zjn$)nhxT2+>`)X%B90xpo=*D&
z(A}yW>X{RPZuhOL^<`WI=aRWIgA(h2le0*9$K^br)f4t(Ej9{yc~MePuQ`E;VddJ4
zXb(_zJ*{!~LldCtN0lHc8#tWFdV78RDf+yxZXxOV2z+UxjLWNh0A{cGSAOs+0InP9
z-Si!XNX+nC{;SSAXz-Bom0cej&@}qmgTzi8Sp@L>%!SdYh}#*8{&Pn4Q|}_DZkmGV
zr*oQ`r0Pg+^&%tf?l=&4YYdk;JhRNC6oz6W~W1<4j}YakN4$d?xz&w#MJj8|I|
z5bUW=s0FRNfkpC>Yx44@C}}&rPpbPQ`a!E3RwyeC4D)I4S)Y_ZC&sf?DCJs+ssvW)
zGLs2{1l+
zH?&**8p4T|pB)MmPNoxL&g^HVe{UvUI;>hVIMPbAI9ri4Si}NsLM(5*V?Bog1x}by
z=!${d6*BSEqe{rXv2N)~nKK#_Je$;GzCrA~Z&TnWxlYhyYdlfp`iHo@{-ai}-AH$lix!qT5!NFIL#3AW{}jQNY9r&L-Gqkmx{`4Y$NNT~%vE^9(d
zF7+l+=ZejH)AVEDWgUya65<7e#oXKWL-&!gU!C
z3yYu6HiH7wpF2PFu7d2xAI=0@XrV6#?&m+4-U5YD1Glz+1|XrG1n
zW5ky#&Zr_>f_D9C0XS7(;PvE6F?z{lI72<149<9-FMC-!0~QngHo2B)@c12}Laord
zXc3HkCE3jYIlJW&Pyj
zoUjJ7$F-}zR+k~t=qSI?ynf(dG+^1+F+|vRd-G5--4$(R9b;^5^af#f*qNi~0s#5+
z@nJFcLDbyicKmjNGP>EsmGa;qXVbyk!|*RZ9@$I=A#2$ZAR*khn00*+)Ex8#I0|A4c`D_lv_f7bVtB5
zA2&_0ALGcFhi-%4HiLchg^&L3
zIN)M8dU*|+N6=J>k1=dL3`)9}2&&!^!HRD;bniGxL(83>U)Eg8IMz}2iQb|N{%!t;
zgN@q(HcpWZWW=z+6QUYToB6!>0%r`%Mz-pswHCixw||U5B7cbTJ-8}
zx`}Y&GmnN+{B0O#(LH-B$r<)D_HqQL-p5?-C9|w7{@B=F={|?UeQZ8(<8@Z!b^L`Z
z`(8$}6gF9OJEV8o2L5uGy)^q#3W~8-kc3KEUu{HH-C}em6Mmmn
zh?AFo4wuBO);@~)L2fnh(Vp4^Ykah2y?G`Szl#!Oa-TiuQHxAH*7EX)yKE*l3!b(R
zsP##7TA4?nfj(
zg16-FMR;P>TG?aK!&-Pid3DJ=@7NR{{WM;v@(5dG{%qU&ix+by9K2vv>nIU
zZK230-G_&lwP9JhHbq;JK2*{3>-DN>fg(G9RuJ
zB-_5RYJq9olKf3KZ(vKAymO8h&%-!Sko+E~sQwGS@DgMYU(I
zg=>@u;G(4Cz1As1sJ)2XvU{8q4n&VtGzo7YL0N^gz}*FO>|VKH2fqY1%H=jMHZjl2!35OB%}S>2~J39
z4@E0{MZqhB(F)uP$i+2G&HP;_;FD83EZmX~o^+Sr%y`j)uB)WLe(NtN_nLt%Z|Vei
z=lqGfPy02HAgAlN?bwXwNEZ7=E6ahoR+!w%m4E1OKdHbGrG5af38wW-b%K^Ewre^3
zoyf=Mtx1~SE3lm>vONL{0Q;u$YG}t>u&+hG!~QJ;Nmi2IJgpalJQc^OJbvW>Y44r>
zj*YiKAVpIt&odj%8|=Fj65^2ZeZ>Jent0%FjhFp!r4Oq6Od;9n+kpm7A5<>`f>BKH
zl{oF!@xX|p;Q<;CLs^Em1UaU3k+55lgQ{R12)n}jNSnFAr|4l8>Vm0b@x=;+Ln=iTanLR^#YhEm;!CgSBFz^p1Jq9vdO49FC
z3xOKPp*1sid!RkbHx`iUjMz%u2EJ|Bf-bJeC%#4J(eUxRb{j^a=w^wa@i8HH9
zS$Ne1+!e`Yohc`vcS`)e3M}gA-ZQ?WsZkBEz~dSeYJMLu{p89p6IMpVD%MYfOPuI9
zZ@2nmGgTm|Zx(Hk<_~031g$IwjFCmB+}o^o0KJu~ZQRLH2Dij^HsiPz(Xlmt)7D*I
zFnLb=BJD*JwCVIALa#v_opby2g7P!~iP;aA58BFTqGH_XlG+v2!Q)dt{YV{=%;oy`
zQS$-j!WwCtnm@!pikE|xxlg0;TM`v9I%=rW_iXHhp&&@?-#caD{)@n9U!Z#K4KPBmoA=?^lXXsRQYgA@(Mgq-f-Zv=ZmoA$070@!GNw6%ePkDG>cLPo$>hycb%d
zf-=|N9s=I%h*I$4GDQgkXts{uxVl7&xMNSO9eqDX3~3>xlaf6kY(|Ds8TdUV^5X9b
zQK=~e7V9-Ul3q{DkoYg>0!u%UrQXR|_BtgYsXt**G})fG#_k)nK;lG5cgz>3Z}BB^
zJT4TP53L~vHr6VUKeqSW8&SSunru)j}kr%B5CW?aL<(-P
z=#7NxrV?^>r6Y9r3yBlLT-jOD)r5L-*V6%a`iYGc!HlGMh1lUg*X1oAN?a_Scc2;2#
zQIuUamDs&Gf{ER`5H*p%L_KQZFftt)a9B-rrOJd1yeK$VsW5RJCB46VDoc8gNM4`Q
zyzz+yFc$NE&I_SMe9si33r=u?_a3cN$0sDfd|=l*ioi`ms&Y?G#E%`K+A}U$$?p{4
zZTtHmmnAl|2SRREJmWYJRB7z4e4HX)04CWyT3-nQk2UzNzuYDs66tn$_U1H5nQY{_
zDgB!`7=E9*&_okyD(b9k9eAKJf9*MhAN2c8wO&%Ynq>eRyBpD`UiO1%9*$@aDusi-
zL+|i~1Yy+Z`Q%*7BQPM;TqGPR_-z+=gQCgiJj~538VE@W<+voTNut|mV42IMYf3mGQOQ1Pm2)QjX
z;}!}thl4k_
zi*xsRKOnu7IRDJ7k04DxAabwzAaBQiZz!ya9NS$^`Y&uM7FZ6CD2(hr0fF?B;budx
zP({?+QHqff5Zm`KJ*!|CD5dNVP&`jSf7%{(-SvqBOB~Jx>d`Ssb4I+6IN-J5dQ3h3Rx{;L^S#yrdMquK%~Wv_l@B<-P-xp^8H-jq;M_SEx)(UW+a(adlC
z8XtbI@!xt;NjtL5UKi-g83H+c0nc_1d*IgVt^;}=n~8PlQnaM@
zC{M#X0~JWVdCboq2!3%iCM4N!qJuijs{p4Zz}s&0UrN9zSZ#FLHFf-eWab$7x&^0z
z(^5O-OST>qIL@DO{T($H8;uShIJuL=V0{OaUonBVbDdlX^sPLQRTc#
z*1>+o>|7@1+_<@m_;@~X4m^ok@6i7z4WCEvzAkN$#4ndMe(vpG!af%g|MDwu1Fu!l
zReA|#9HeUe?aSmgqS@HwUnB@X!RMa5=W99OYMW4!;=@YtVf@LF(I<@Xopa44`X~}?
z(pIH{7LZKLShz&1v%~xIYnGDV}?4BszGmJmQiG2=XV8T
zR@@@`R@+BK7n`uqEO{(&9ptsL_|EI=!R%)BOA{qV@a=m)X_09|`0w|1AuAVUtaUBz
zUT}jdwtr`Q3g^nXbcQd&Jt7~_O@iW4I#WvOAtVNOyBSxy{~
z@W|}Tc>$dH&V=IyD?eUseUrg^zaQN#lsRvb!~}mj+G)tAaY3HN!Wm6wPT1>u*`Vbj
z8)V7Q9bOGPjkgr7Z=gXoti4Y?u5_FhH_T9yQrORf=)BG^GQYNgJ<(WGdcqJ^8d)}6
zu+oJZhbnJf0+KK))937<`voj&{*CeT0Z+5-tWx^x;8jqi*X7v4eFz%KrwweM{RWc1
zap4SBYefMXRY8f3K>Mbym-l4|5727#%
z6SP?IWOemx4Mt3~%kjM@{1=(jP)6`Kc7uD)Rc8Eo+sOTN_-fkDAN0?UPsOEW1URlR
z9B3=aU^4IJHZB2DsJ_Ma*WH`|`+Fm~5~Nh2E!n8p5Em1q&O{9#zDr<*jw^YfR}d$I
zxj2dyA;_nFQ%IeG3Chq|?RRLg<9@eM7kf&2d~wC2IA?ktNTx3uafBYh)d|$%t=3G~
z=+NPaqqH3$YW!CPVSyAzT-=h>i`)Q{9Whf+b!t#tD836?B(RCuj_zc_`(g+
zyX&a_SG#mEKIp$wu>Za7K?Ym(wgi^s4tOm-MzOKHztQXKArbVVV<_I$<PGoERcvjr$?Sc1#l+8P$7?qF+soXqUj(%({
z?9_dn2f1GI%g-1-fvY9Wmn?XT(0XS1j!?x5H0igkE9UbWn2_cBDXPySn(MDg@A1q4
z4H271Ye{)PU5Cktd%Foe(m$_lUQ`84x1*>N`1*jOuiLBwu?d({JoH#hDnjLrtB&H1
zw?U`^i4*&u3@|tweE#LTTu>P+WomT02sIk=b{-z@wzmc|-Oq28^;ZL#fgsz|zQ_Z=
z#OIuP^GHM$FYkWst_MhHDXHzXa0cQ}M4Nt9Hh||&-K@)48UF_s8
z!1Quam6?Gr5K9|vm1Tbcq{Nq8qjjxuAcHU2mAJgpd&0555VJtiu>F#J0LQ+
z`IdvAfkr1YDQS`!z?GKWXf6XKAen_%^CtiQKCts(G}9c3HF$OP51d0%vd7+zheBX?
zY{fj`76A=0-c_T2;tS{vh0Y|sFhw!9$?{LTOQY6!&K4b&i(vdU-j(!FM5I)NA5Sx`
zps7!4=X4IMBd`0^UgW8Kfc#)zd%bRxSjV+vsd4!%Ds+DQPAXCf^?0iVkIxGNdyoAl
zUC9+f{U@>NP!TF5_^@{PUGyOk&YV@<1TKN}r!D%W-E`>WSM4N$%M?iHE`82Tbqc_f
zsC6dk<~C9Gfn!&N!zDz0?mv?0R8|y!Jvr)&DFZN+PyC?EOo}9L=Chq~7$?fbBp5Lq
zQ$b4RE2BPg4#d8ubUa;ggAi0A
z@GVholu+ybj=IBik|@QKD8|OuMKE3U8+11~3M%OB#f-Qn2x?PbKQI%AiS5pW#SEiv
zLI<@ksa#eTLCwY|G4TW`QhD@Bmt^G=(fW<}MuAxmVaV6@Jd1Zd@tdVAt1H#hA|gDVh2tzZ7Y)@
z_@Lr0hvOPV4jRLW(RwGs*6#0jQ*We*qO&4dvHSZ9Yy|mPOV^VG$~A|ZW5GN`TCy)G
z4b-ZH7VeI7=A`z7^bZ4cN9HRU6F>9~CM4`N{+Gx6?D=zgLU`(UcB7FbF=bW!TIp*+
z;;`&+D4W1QBj=%QC$H*n3VDC8Yx*1^BidT;KOOudKyZ7|OZdquOcak(ns^}jULpH|
zGi{OVKx1V7pHYtUYYKJ0Uc#UhE}}}^;m+3=t`S`Gj)g>sx)HKW!&7LE#t|b!Sku66l?+{iV7td9exDmxg<(H{z{0V_aZZbuEd`^^j
zr2RXG784&mFf6LPCP$3ik9}2g{sy7Yvh|ChybW>rqQwhg>tI5{-^UNkcG3w_StdGP
z&VFnRT9R$&PF5iX(Oe9rlQtk^Kb+aBp7$ia@Ocv|!Q)RnPd}vc-CKpgf3qvuhMb33
ztFNE5pCL-*$o$7}=$;FqV#oQ<$u!gYx$7kfo(>mfMB+I`BLcAMk+BcmaJyG^f5ukP|syp_BP?$
zc6ae|UmLNgzAH9v{v%9GyskrUOMpIyj6
zfb$dPT}cik&iR9MGVMQ9bv#k3&wqiqdv!2k-9+C%&K;H11jwN^pj~l5cM_eIWBVdg7?p5
z`owozk*d_#o`%98@De*l>-=a4G!|{Vp=|$$%GnRvpx!?TTuHc9TkbgO8zC|eue
z^D74YHCf7>?mghuP!B7=Q!57d`IG#5c-sNPQ6>GVk##hAAj?=6KZ;o#2I8|#kHIfp
zA6jd@b^)`@lx}L`NldjOU8#L=pC>rqnQIMPLSiYG`1NT!kmcoCBBksOSlfO^645#Z
zMjt4CDtP?@5PvJ!%l#e({Imo2+p6B9*hQY>9lv@|F+ozsCtwV~b3>s5pUY5NXO!Q#
za|cq&Nl*K^)dNzYSfQ);J&>X|A2dWZf_^q{DOssE10{bqL3@uB^!E?t^2_r*2&_+&
z?!W5=dvBCqOdXko{nt>B|;<)UYWIawy!Tt9GdnOC^??@FSfA|l%n(I0)MlPUv
z`ya1q;y#0_BYV??=r2f}@{t}dIXSK-uMx;*puib)#~Lnnlj6#=bT0=7M}di-%xkms
zFTlF0mE-N-FMzxI)g3%P4*qHr&rPNbg6n}lW*8rSKp&HTl=;O@pn&Rq-65qeWHv=M
zk(^fp(pV(yM>tx*nRD4Bs}K6Y>73Bv$>KBs$KT%Nh#5hYy$$sLzD%LTQ(EN0Q4?r}
z`;KAj!+G@5B<=R23$KvdiETJ3(}?Ii*jBa)p$gM+#>H8w1A-S-De{l&A|3>%@b(xBbfZi;G?8^v$aj
zwx3%NVoW-^q1=I3{l?5#>td0#vis3?yJisMenR&KO$+$Tf9rQ2u?N)Tq;|%O^@9kH
zk}Aot_h9o_jq}OXazJ&zbLy>q4w^B|my-EQ2IpdG$|FvWp>uEbuQs~xBX6=>ryr^4
zpjVG)<-ea@0pUNsY1!gf^wT%MAaA+!`tE
z`}@uz2e9!YLgQjq63FT6wwNqTKpQUwgiMuuk=!Hwdp37mfDo_V^h=Fv$f?8Lmc6_h
zT>V>Q+OYZn{C#`v*7g1{;0zN$O#lFIWU$rUj*bK{|?I}sGvh3!fdEo667`0z0-~11J9I9Ozu6u
zg@h{pgA2)unt5~9=?QH#u<
z5&$*v4<5hce
zR)GpM|5R3wW+X+`=g7@@vd0L8ug)>Fvs^--R89?dtsh1zSm?^s-y;V(a;lNIQd0Ch
z=-F9n)G@Q`s8Ajjy`l1dkHzsW2|~Z5K-5X~b@&9r?Mdy*N?oe&OG%JvdZAfAIK7
z2mYgy>Rud)vG$sZ#5tBadllTZ)h`Gu+fq6S)U^pLVr*6tfT>kuSr
zomq}|GdXT0(9K~TH#64ucca+o6ERiTr5^Xcc_&U0Rg3@LRJRQ+OU2+j#l4G6AMw{-
z5|+uq34FTi(YGzrCQLisbb*|`1|$CcA@nilzz@OVayB{%#`QG39(h-Zw=Ou1aySR!
z^EK1wB4HUW`265|ztS_j=zA}H>-Q9ti@PP>Vlf02nOSp0g%3Q*^E_%7LzAJE$0tQw
z$ru>>wZ(OsWfVp=m2Ljk7=Vt#hkno665;Mbk5}ZUDrmGv4L_XBhbi1byrg>X;fU(L
zslBvL$bE@bI#jC(S|)vSwO^=$E4oUKBDF>EIj>n8cV-{_b$_RvVZR^V{bT59)7J-$
zhP&BlBi_T>K;F7fH1DCeD!mDOGzwS46phvY$JLd`WA%LhRG(6)gve4!5hYue3X?5a
zBH4rMVXaEo?Ehj*bL{?Ce{rM68M
zU~|7JIVIZ+q`R|f2a+4XSe3E`pIin}k@e+h|Cj_3H!;ZJuLOTL`b70A=xqS`O>7ON
zCJo?Uk9y07EA2#e@3TMlT>V6*f8B|R*HwUZO#1k9p+=yuSrfhINhA2E%FJ!8*b2&n
zby_n^8o>(t*?YMU>Oj7ktk3PhYT$oT`SxMPN_gKl+Wi1;Gq8JOu-n+M5pWWPyHrH$
z!Lx%!T5Xv%;NCeSdyB*>@HupD#>=7!{Jr%?eHq&8DZ0FB6ju*sKW*6i>3c0WVk^oN
z>Q)2#zqt$*@iu~Pw(;OEHjTjG*wD0ua|7^lP?MkYuLGIU*1;d6YJtBsS6+oeEzmd@
zrp9(XG`ZTOv43wDjUtbg;W62yLF;I@R{KY3-7{-vV!
z1K4(IUJ(AB0Wg%kFR*nh0qh;DE7FctgQI7zt^8Uo2iHnIi+h`vg1W}wyH7$&!I#J8
z1J6YXpuCd#TyU@k{QB~Ffc{q{U_W*DZ*CC*Cav3^$MBQ_*9QK@!My}9yuR-(Lwzmy
z)A=!fVYmvo@SRbO1ZCh^a6`=N!4E)a_-e-yrBYyh=)(T0>RO;icuvY_R|6Wy1lgYQ
zGO#wS^COhA6kI#Y)_8}v6fnzu(tZGc&s^5TyHVgoDX7U&eoeo-7&Q7uy8fW#0n3DI
z3{&uXHV114-fGGb;A2aR%#qYm_&?%r2KkqZK!b-C*WdU&ux4~tz+^&rwgLYB
zIsaN%*E;yU`krbZX^)dd;Lf_{b2*`TpuJn3h1iz|yd&eR_UtVMx)$cp9}m(_M1O}|HBxu
z%mwi(?*7-V<$$futD)DvWdg1ndsw&MP6rhAc8A}&sUWyy!TNK0E@+sMtCi%=0RwZ1
zt54iA!1a$uw|!$y0i2<=+rqRHKwkHZC(W*0P;)|2G3HPB`{7xr=
zwa2~x;`?8LHxFKnT$IlNQ$nZiT-~1qa?3Wz7d?0fcr@=yx*8+_2CJIpJ34Woc2A$*
z?Ui>xb)DeUSB@0W-xAZ&<`ek+Cca=4fSUnl_Gr-%Q;8H%qeE3rg
z5pI7>Sz0rlU}|kTeW4+j$jb9`ex_VN&|7_A9ViVYddp|){jR?zc-W2ziv(s6^)XvL
z;N1cuS?a0JeWfBIi^0>;?@u&g7g)fvRX2%n4an1VTFxf)TEEL2JX=h#uICbE`&&vV
z_UdW7Nxmj#qGR?5Gp7+=9hQD4FBT9}!^{Qm;r~IfkHAk4ELIS{Ov;Ih1&M@uSeuRJ
zL^@%6pPnC-781(&|BZ^@tRR$Le>V`}sU}3XzLYG~FCc1uYlif5783#6e%7CYzq?j&
zxGc|bu!_hY3)XnfTtj@0s-~a0P(&0Mbm;3@d?418OHO^CEhoU{&mQ6K)r4OB7VWyL
zwZw_&f6a_br9`c_cCzJxazbF&g=;+CRRlNP4wnNob;SBA(YnXg^~CxfQSREc3Zhg$
z#)`#&{6UHyDNOte4B@S0s|9kA)
zKo~K6*{1cRnW)>PecgJng-G*gjoc<*Ll{dPmDt6LI1CP-~V6P}~jj$d!+4_dQCSpWwb@2XB6Ty?h
zD&8c~L@11W%&0DDAikAv-=pqYPsqLeWAhZW6E%TZC)1C%6Fr^r4hybrL@e8D{Wg~t
z;(#Ij>rb&wgm%ER_o!4S@kA})KhL!e!l%jTgYil`@zF-NTh*+MIBBpci+EPPblbA#uZD~5&>G8m2L2UQXCx)
z4za*`H{P_~*G^YN_@4TF<#T)!u}=QXMcecSBKCor*{fIZJ6At!d8KYu5S7Pws@9$^
zBbGH*BDhmq;O_@rvUheg5$eCV{r(Qt5gAk0T6f1*5%&-D_{phN5Iem#OZ^A`*W{y5
z=GG;mZq?w^zdk-j9vYD*U-zNXXmz+~0RCm-tuhw1?$U68yhr0Xo*5jzo#y
z(AF%^0FYvRs6bJzjOehnxgEP!MBH^WioBito*1ccZ|yrBL~uPed?~oW84TX`lRKbM
zLBwrv-OTc&lsK@I?L6@+i?FNrh;DrmLF{btKH8}(2Hvk-$w=#Z0eXu#uRU6Q1A_ay
z3~V*Bz=e&9(OXX!0Hr-LZrd{oL9-ugpy{nJV41NgMdJ7yz;J7rATKsEdeLwAMf8(RR(f3qQfsU7Xj6C?M0`mKY)!dEhcw(mH~rac3z*6
za=<26_VPdODqxz^blS?L7}RY1<1YgBjM98w(r+jSeZvjh&#Nmy%k6C7s8AYp&=lJJ0
z9$ZzxTQlR_UuaK#gPpkhu`0mlANXI<jS>`srs`6_|WgX!A0z
z34n$y4ciM%pvlp6HT6;hc=h(d>wT|lLG@LhOx1lgfcc!i*;{u4_zlfjt2UH?^vgYE
zblVDmj=Y?chf+2;ak$_q&CU#vzG;tl5uAq-$If*p=9U7r`%^n5UKE0IW!l4EXmdcZ
zsg8XAs|>*V_{#W>{tB>g&5icyRro*JNtbu$Eti02apyIM9_IteM|L!Mp*bL4B%oHh
zstUxEZEtY7S_$kGg}H(f%D~a~4Vj#KN&w9R9m?C&g`n`q{iIx$8eqYsJTyL01#J0z
z&s7@3|0geebC1h_0NIbj+L)P3z!Krt^SA6hprH|DKi={VG&;|(iE_mQar%Y$B*Q1b
z!jff|`Fab&%;ttv!^I4sSe6wonRFBS4P09<=?+w6eLRbY4HSkLz5V(>|U?kjOS9|*rlD|rmRKUwN-MyD4Q
z1Pa2P6^(jqfW(!uCxyGO5`CM39$Q@YC2ZH`#{3dr5vH1L_P55932pWj%Vp~%!Z(eH
zd2NFvG0k3^{cO#XP`z%o)EE*&WNDe!O0G{K=oB+eL-O7d3ez2Sfw4BAfvu1IU=b<
zZGKJA-(_|VS573_-yL_8XfX%ey!0X>Do(_{_!0Vnz;NPD%CjEZr}0D&OWt6)UlP&M
z(P7?sE|JhI?G=Ag@{YK$%d=-ckxAU*ap|o0%qC9M{UeyRAB@Z762ui|p=)oGiA!|%C+S<>5|1MXvK_PD!(T`h
z9QV1JMeyDK{I_QXV#A;L#FPYrf1{fGD*XNG$4&0<^HIco^LFwWzHvAI4!?7FgkT9tGL<$#C};
z5^{|elo#+kOYll_waCjM)DGP)mRrap)=Rf1CH*TPLhgsEomgK)%}uJP}sM3;#hZNnt|
z&d_Vl&Z?88#9QB(z^Bac|7(FpgX7jhqBm&U^sGk-v6xn7u3T3_=m{T8XtyaPIuhc4
zuuYT^mhvr|okEL=AG;P>b)FRwaoeT8xM-IUuPwLMADk{FbXNa6{K}||(Eo0FqENVq
zxZD)u7{XmlxP9+v=05p>pcri3u9#R#yooM2_k)!n_#|Cf{aNyf*bxI`cwvpWBW`=n
z`A#VzHL8E&j%Fz_W;*gaN2?6}9{O~&vr`e#X_RPj=idinaaq4eQ>~mBRNA6ckzYwn
z`LhhN`z?iq`T@5+d=_M<@_&Siv9$whN+_i`fclf*#T=PE*4C@cPHRW)&HQ`W9G
z(PhNX-mDaDZ#W)7C3BAM6~rZn$&m9`s|m)zvaf$OKtB1{BdB*fL5ywL9472vMjYLB
z!Sm(23PSho878riDnebSfg(v;OEkLwe(#V)5J_Cpw(>FMgqE$-ZQ9-nBJ)~Tc;HAi
zv6G8v4jZT?=n4vQXich#Ch_^p{R=h3(~XtaO0}zrj*J=IjHMbPeT{9f&AOI&C%tKL
zoWF*MebxPDskDYraa%rrW3HAs^5@t`r?pz*dTPAhVo@FOsq_(T&bk`HEn{ZmUF$l+
zUDk4cTSy~uHf)?d^Jz1&PFJwU#G#q6?MZWRU#TIS-c0Qi@2n?yUKE6=$+r;qT}8Tj
zW;=+>jQ0fTInZc;UeYHf)vuq3X#CAgEqtKTHk8a{ukOCK7;}DT^
zy+7}I|2TM0#SAKDQ!$u|`Tq}#sPbDX)==>s6-%g?OT}_3CQ~t$ivMf(fB8?T>D#IJ
zm5TSMxakIIZ>=qf%eP3pL(T7@q7xPOQuP<8@)4>G|Lzs{yMc=3)O5+)WcmqedLtG4
zsn}~n=FeUsF_&t`47GeXHDAS^%ujnmqQzYjUs{vsLe-!1CS`kb5}#O*Sa*{|VQT%C
zsO=x;A=B9&kmzGXqLe3z8{|oB{7NE?7Ktf7BpyCPVkJRhI1PycktF&YB~h@B#9{Y7
zJWjE9dohl<4Pd-uL#7LTBIV*kJvhDHa|UA~Em@x76N!fsN&O3Br2fhBuUPLxAt@*R
zN9xI4AkqB#AkLRq`hsz!hipI1jm#H~B+EIjwBvj$mNJa*mB@CUjFR<#;_blcHDSLn
zYJ9o}r}8-HI!4MYm902EHRLl!!BK$o+3v_-xs>@M#_%4p-T7zTSmtUb?aNMlgyqey
zs~BH0=V0W^*n#VbpghBJwli}9YPb1S6qc2k58?cjZTGOOnP7$GqESz{WX8dE4?8Tg
zSnS04iCGm`zQC@H^GjOXG14uqD@6L2hK%7-9tR9gr{K~Y2a9R}a0!ot>9C?cgneR_
zSbtwc39iSn{uGvzmgwNp8wUeLg#^O3&lhohXUcEk`~ylB*zV^w-Wb1a(ThfUd}-OZ
z{>gq{0yOEI_EUPua9HA5FVd9FmDCX-8>ApORNX^2>iR
zenPU9Ta^86h@~%mZIS$6qo5)xce#IT_7=qV{msQUAu9C^9=!^&)@|Gy)z`}-7LW9@
zDR*zWLUixFx}Xcud(5*_5#rometV>6GBNNP>B*lAUh{z0dd>CxJM}eUXJ=@q`%%=CHWe}G}oJ#ogiM@
zpn3w?!PpeUiPE2|9RGvt{g){7GytM)NzxE%-|k2CR}F{@FZA~5LFCw)_csh;mgZxO
zg~Q4J2K&ar<_{l?enFgjO@9X2`N*McJ9<9Br46qTs-Ie1A7c6k1u-LpLr$xV_jr3n8C%JU&~d}@#NPjmw3L({-WrimOi{#Tf@v
zC0tnu0}Isgekaqrg7xYgBwA3qrK&C-5827}xV|T~3b@=*+Ey%gu3Ei7c7~@vaz*;D
zf26EK|&{I>TpAgU8{N1i9Zww@>5o`+bzO*YYOWKHod?e9dJb&%^Qr600pqd=pPS
z7njI(`V7eQEr3KGDlgJ)BJH$FAB1!j8<3_9fBJ
zjKmTuZ_BkQSFG(A^XWhPuhP~
zn(U`a5~-JQoE%pTDvz9^@d8}|s~`1wR#3;yoobhb9BJoXIZ}T!b$$@eq%3AiVlTD+04n~E?*pn_
zPJJ$L@!{t?OhsoZdQt64qoM~jJ%)-)RD4VAzb=(uzo6n5^DJxR8p+28%HBt^~BcFHFWl~mh
zB~e?SMA>)}Yck34-I`CzArd5x<&ao*ki@!RlCS<3L&}HNk(i)Bmbd00dHazcpYZq=
zu9EqYE2JF@OQc*LLdrbfC$U~Jbsxy~BI`NSPOi({b>zG?HYN4{Yaq*0I7zwMmt3!(
z^15(6bn*2Vzdt0?m5-A3b$XKZH4Bg6e4RBTyskMU$$f)wbPA_uClDALpP#__my9B@
z?0${h-!A`Y#
zzx<~7V)@6+GhCnX{a1Lu<%~$c@^UN>&OdUy2Fv{kE?BM(Q--cD7&sn{KT0MDi=SD;rUMt;&~BBwyUpZiU*@@7CRF
z1krTYeYoX7KjT9eFUsE$VDk#uZMvC3+!A8uYw+m_MA?R$1WI=n+HoB5ZLZj+M~I)s
zcB{G}K1xT?<}igARbZuq^aPxV4R;`BO*lQj4DmyOz9%Y|y2rTdCB#eDttp5vZ64Nc
zLOhObUG|Op5S7|r79gG{d3C=37{t%(&z2xNSW9+)@`OmY$Dz>;;%kAigMEXHPP|$4?sRFlKNvG#It{5{p4W&;@k0u$Pe25{n)@Cn0|kZ|1@Uu;?=$2VP0Z?7-G%7%_YQuZnIoYKl~7*L+@m-JacYxQdK1L2bOIWW
zA)3Wmu_JvcK@oT4FM0knYW@fN;Z$3*75O===5D{$VLJC4`sIrdAKbSnWQBJ6(vB7Z
zm``)1&<^<*?~kc2pniU>u=#?X;~m#sy8-NH^7G)@CD;!WH*saluj8Qd&C=GNFui|G
z)14h+43ms7nin}c&t7+g$e;3{2F;tWgFNksFY}z<(u(?(9?#s4=F1EH2S4p0o=keB
z@(^M(`?e#foZ#cgZD?NW%9nQ{e=D1DNKy^r*Xx%ul_0V|9eI!DWxypq8+VAxp993u
zyv(0dwMEZyj-xjz9O9L}m?R;H8|*tAk3#I+^^_$UqI2G$1rxOQ(p)ZkJ;Ya^e7g%^
zI+NW2-Dx-ubU9=8XdbZuw`H`R
zTfXXg3+eA4J_tjsaC}W#-4Vv$&n`m#)8Y5ms*t{8oCy=M*SF)}F*H98(9^6(QdZF31je;ru}+XT0w}{6+dBo^SrE
zMD_BI&l&f(3%I^U)j>Rt>gH*`
zQ9XGtHY?KKSySl5GD{J8j_(O+!*c9W
zJjO~_l25-EBK7~=BjbH*Dm2&*h0IeJr&Fu2KAR3%-^c7FENid6&qj6xa(%+%v&crq
z4-cPugY&E1$Z?8YBzcKs*BR_5e)|-R>n;A6gz;~|3v&GZNjs15lK#Lfm5<$|@;4O*
zlKjv4lge*xbIE=dw2=1JQvD=aJ+fQ~H7;!Mkn{(OsO|qv
zAoc(IL+S}q`FX$}vOh<^ko-}T>NoP1l6rrsap-8OU!X|k<#JTJ`>DMB7xlULQqwu9
z{ne-Ht5E$Wt!2`VacVl8)Oh?hQ}GkE9rtVGxD`?Tg&L~9EY)9Jsv^tzX^`XbSCf>d
zWJxTdj$_R>vOPaDGQPb`^-~HANj=&$63@hv^=J-|&-==CvYeeFX@?FV(VdU#21DPvWE;Hc>;)kNaf(%oMUcwKWo1>&SIxYO)E>pAGq>AKvXs@^D%%
zGOn-?!ie)z=TkBAx{-c}at8U_3Z`0cI|sDMd6@OG2G7q0Q&Nxj9_dFk-yrJ^>`BJ@
zJn#?e(7NU6CHwOq4{85Zb?STN8P$J8wC@Kv?te5zZf>CHIrm`BvON?|qrL9+mz+ynhDN
z?!f!t!&_dM=LUQ1!TZhm=^kvC6r)l%s;{v`5%YN7Jr#I=?#?-c_fyL4y%_mUapO34
zzMyS7(pN}Q#O>KTIOB5$+%k1V=_|ROIG%R$zcqaB*v!&Gmm0=x&kZ{xWaC-+hVa|+
zon~~7{FYV2ajfQ?kDVwzBlE^Hgat}IT?q4;h5w>+2k_feAlYbhKK5I_X69nML^x;a
zQ2Lh3JO3cw(f08L?ysCw(<_u7mzD4XVK&PSe9j&}C!dAdEw+9fh|=rLWL~3l?5AA2
z2rBQF{4yHxgS0>C21pOs4-X=}wYG!j5r)(Jra@S8m2n#JjthsyW-uQY@a{q7heE%7
zMfIt36}KR}BVQOBqI!I~jMESge)%;o1f^#btu-LM7ELW#R38xPev0^5_?B=n6i27$
z>6k)xnA2ALLiJj^=6^=0Cf7QI;;@xADa$C%oNWL5Fw(oho!y1_o3F{572+G_4({^E
z?@0G}y#?`n{oi!#=se%7OuG*8F;B*7mnew0q~Bi)gs6S`6&{~W`i3qQ|McQfXbR-TbW1<@Y9Lnr4Y@Md33;#A+`z_-6d-ys
z`|37&3Yg7by6FzfH5mR2wS@KE^Lh1K^dsz#^4Itaus@-EiW{_HKZcbv>d|tMM%y=B~*G<|LSBJ;JNLcEBEzsVz!7rA!LvY=v8*b5u`N2JRKtMa>
zWwLh6o~>|RrxmM+%#H(>5{ay`Z;-#KoZy$rh5b!A@*=Db#%l%YZ_AEAKHm9d_|mU&
zAiS6$%oPLqS8ZujS~ScLTG9!Zr2sDduPn{d6tFE+wfZRR$A9bCUN=Hs|K>mh=!14Q
zt
zuwTzYeHbd>c=Uvt96Ar}&~6;=mOV=WDd79SCp8KP%y(iB=cNFaxRb%+e__3YErC98
zoaQ<+4j~?&r+p!EC$#5{yufiUh)jxRM&DsNU#QkYSv2l9ww&jN=^|Bkk6J=IBuY=o
z3&39UAI?Ug8o{y#n%9&qAKjlFD8|=tc1IxQQ
zi^=A|e)P~urpiNmySJ~G?uFxHXWVr9CoC6K5M9;@{le_=XT347pC7-TeG?AL`E4AW
zw};PZO)4T_5!%1+CB^&{eD3u}PQ1{B<(k%Hnh?*Qlgu8w0OiPL+jGw#<~FH(M1He`
zQP$53&`uR6-`rp*ufOJFkNk%xf?NHO|Ip2Oc01C$!o2BM8LTHoO+Jwp>RU>lTpNsuqy6F)7+do*Y`H-vu+BYN`rJi{~
zJMH>u-1MMcx5Y;~Nr-=H02`8BqKk-Z}m_#!d%;4>_52w
zTcJby)AX#@4DyfSXeZr~-@<30x`Os0o;}idzg4U5fgw$3Pv$tU2E@P*ttqH~Uo~9H
z(7r^mbi#hk>g;Fiw`gylxrgeNE&O9@2(j<9-CyK435^%y_qGRH9?GEo(nKlj#}Q~}
zN@81X9+XEKT_jMwRaeT@uR^3t@=`*6T1w^oGHPG`gAqO#lb9zkzi%!TB~bdXxDq~}
zD!v`T{>_q^_CJ(vmNSOWxwmtV@p%%YW`gy_jyUz8d}U5Q{GRQ=tH5R?pNtK_sF^Sl
zhUA?q3v0;E<<>fGgvUJ`CeXN)Xt(sE{F-!Y4b;z(T{hjQfB)5{Gov_8W2+60zr8-C
zONaEs<#&BSdVb9kxP3<1419jc)^%fk!Q{$_&;ORXAbj2r)Un`xXw2#2^PMr~7}g(?
zd5F*B(5Q=8e`}B#KCdIfl<|4Z;&dMCPco42w*$KV;c`4|4pylB?XBLgk==4DC-Hf#
z*Kds5(eb{Gd55Ozmvq$6-hP(zNWXGxwk__bXnY&OUE@zMkGRJui0u>*v&U#!w2E=O
zljM=<79_7T=RAY$WVlEAH(UNA{Sk1J^dI0i5pcPYm@OEaY8bFSQ)v;F1CEmU+HQMr
zzkQ1&aa^x?E*L-8Gc7lYE+-IXwQQgVS^N(qVq&$9NWx
zORGaYPCuMR`gdkoB=6KVA^B-Z49RZ|4GXcJXf??fn~h1oYfBo*TbH=Vatq$P*v`N2
z$aJ&cq&}w!8P7g2Po_6Y-o^DTG3>_rE=#08)NfAm%KeQLoPUS;Eaq|Mo#g$G9|yUy
z{>*3^&VRa}v@@RSubiXu`~Ei49>Se$PqvjrQR@4Hu~0HUUYTn5Q4)8qBl&C))nB=9y+0OMSvfm3Hq@CV}$nHYelxX`MW{Ty`cI
zkN@66`m){d-7@G`c1DM!19^14S0T?kv)caDBbgXQm=??
z=Zj!CmJMj#@w{4^*p-OpUA8Yd-`Sgf;q`56IF9R)*tG|*Q@R`X@w~jaO%s>*TPEjs
zQcfY36IMC!{G>Dd(}e2VK%a!`DfT~z+flZ7hUdSO;;mYg?{4xAuivl9LKy34e`7R{
zxqw2B1NN?)#=C5c!;^?cv@0TsUS>k?aXI#Vk#@`>Dsz`5{cp8h=|G9bxjLhe?f7pkh_|gHo}m4H
z>Pc5R;=7x-$W<9bG`Q^g4aGxS+cbWlxcT_DOnr0@^7H+-`8zITzwjhD9gD
z!|!QrTdIXJ9pmb@^$>Sggn9Hp>}5W$^aHlrHgHZJ@pt=uN9Pg$+?4y<*%PLRw6blv
z2l-Y;*yW%>$ft$W(x=ZuUbScX-ibC?zEFqVWdQQ9a=Y`TzL4Lq@(W6eKt2}Ys((mE
zoC4l+Z5T|2d?lB!gq{NNtG3|*CdlXb4KF;n$3OwRVyj7^Gte)rqkF{#^Dk)@xNL{{
zwkOWpuD|Cq#^5<5OUoxK1Ux?jJc^{qy
z&vk+At)W>IFmi27PZjc7G47q}OOWrSGHet-_zK4JPn_fX@^KQZI{0sN{x$`Cyz9Kv
z-@tLy8eq~cg7N9MmY4NhDd4E8pV09B8L%Vj!?CJw-@s~w=z34s-uvw0MK3?--vxKm
zZLfs$ab!VFR)qrAEDQ9-L@A&W^f@(!Kptsyjy50#uGg^uGrlJjaBp?W+ml;LqHTT$updax%p!aua|S2zBf~GX)yX
zt&0vT{|0ilKWeh%p9Esn;hXghC*XSvvz)oz6JWBrVS_`%6xg${+pbXL8{m!da;I;Y
z0Er#$`e(x@z>lfA(SKbOQ2f60xVsz$sF&={c054=;bxPGcKaz{F7@+M-gOkf^IU5G
zJRC1(P2q~n8Va~mSFnpQh61vy3)Xc53a}p(;5f4g?>{+dTtj~8qt9{gPjgIwO^&zg
zni?oT_Tl$kj!MwqS!gX7g8tFoH%e+wa6F=9szkY`C?GY>gmdKq1q4|P|GfyuGj^@%
zFzI4@Ne6zWUhyrt<)-TwgkO}`a9pNhigzi(fQqZ6FJ0AXUxU%S?c;#6{0dY2>=@X0`mugx>==-{xodY-${1LCpcQ*zbPR+ne7b4V
zF$TB?zVMz+9Rnv?5+%~r#=v*RjXS+X$AG5t`p@EmhDbPjA-sTMs8k!V->F970e
zMl0_FaNqH6f&ZTc?Ug#mLVq9HVgEwdKmz8A1c)odz+DbCqFLoWZ
z5pIBb`Ad&Z_CYLE>oB+h+dJBq9*q1LBioj%Xuth!QNfDN4|2-NWbRk*Ia~FAV{4btsT{V&q1&36k{;jXITSEP^zafix%84gW=u!P216(zb
zo=Bv58#<3$4JsXxy$*M8(;)rFN5ApZUur-}U0+rxdgQNGP%
z%O>>v7r%>Sp>g3;yg!fXRU2c)Jn3X(8RkQ@aq*acFtBi8I}UG`zjfZcQK!$sTw_m>T}e6g8jtsl5YHd>eju_n6G8Le~NjTS)1r4WamoKQ6r?!
zYubeQkx+eB3zBm;q`X9!W6I-*=5Ld+wFk0aUiXp!O25vy4*v82KDP%}c>I_y6=Qy{
zSg{e;KlDcg^H24jEG#$vuEJ=bT!`(BxUd(?_8ugkPfc0E^4`j9jJ}KCaDPUV=rJ~A
zlYUQ!#4^sWc`JzPKlXGA^RPqamvQ??f05&mq$!TaEk>G*hmQ(9!2aTx*Njt3p=_l^7Bk#-8L@wigOq71c^TN=r
z0=HW&MDl-edD8DV%vgleH=P#5`qL>3Sbn+UiE&ZlG`7cHi@cAsRG@~(VN23`oG%s5
zfzjO66h9xPdl&KiN)45bL(f~?tpbk^55oe^Pv@`3^L3v-9Zt`eCl-+2X3+?2r*0b^
zj$?RfE#i4KBBaxY^2_trw9&kI;1!O`8?qJP=iBb-g3GmR7sT@~OEDMgTZ%^E_BYaS
zj&Zu9_4mC-io6f3nbXeYASk;_WVc^MCe2
zR2RN8%nb7lw6zY$!t-VCvO?^3xPQ+mZ5~GBxP1Jx2#Ukm&De!RL*xq3kL9H?!$w%1HD9(X8}f!n)1Jf0^YFZ>|Kf4ltBH0TVYd6b&?4TL;?KV}Ab&4X7V##v%8ephf+e6!;>AacQ=wZCo}
zY%BBTXY`!}hTKO6%n!lm%kETXQ$GQcHMG`V{hI}+ug83TSULqBmtEiR`@j^qCpt)x
z-t!G?@)`e`Og9fUw9FRl+d2zUAB{gLl9~n=l3%^}X*LU-A0-ecr{};-Ylqf4`+4x-
z$D3@wnHgYs+Bzd(bs7|oj&0KDodO59e3;RU9|ICQ@0M{zjDhmXk($C2kbkl>KJyVA
z1I^wEx;~Jn?GoVFvj}-;eOh)3xG@I*XUq9k!n!eVtJ_m{U%@CiFfsULN7pFW*0H~4
zTlFY7vP>}z%NhmrVivlQhNIw7celJ!*9a)|GKg%<7yH^|B6({BWZI2=5O_8M0xE`MD%S$uUyzTTXLtPu=TYsH(v7uG
zqacDHa-t55fi&|=!Z)6K`@IM3KI(YDjzWg^_-!c(rjyH{f5FmDKe0}R9l&eaHBX*C0%r4s<*3b`$5uZ`p3;o>Q
z<@YzkpnuBi?4%?I*JpOJ%-Tb^9^4C=T%6!KP7F6M3SNV8aqIE5r;v~D_{MZt9P-uk
z+F72rVfp+t1>c*nT!O+JC={lER+Eb<5j1eV3tqYG0O$Xd*oPl}a2=l8S@KNK74@BjwTiilnzrIdiZ;yd8YdzDE_Yi-6aq(w_nCC7vA_@Hz9ub$_;jmwE
zaymzRp#J4_->tWx%)Z=p2<hU_?3dlV%s)S?=zYl})tq5FagGdsu7e!ksFX9m5W
zER?k6KRn?!pnC_}d&kbN{|HRaqSZNw&Q0Z@$73%b-g>`ZzxVd_3{L&K1Ju^hQbxhgXhk_s6cX7o;SYNdAL$>KhhggD#YjU
z;#)?+#zC)>?mtB!M~^T+4U9kM?Fpx%#*8MGGe~o
z8=HoC)^q#!n9qz0zWqiRbPLGk_dl2bycW=bJn)S~r
zzNaRc>W%BQah1SvzbAr5k;tyMVmS?H9As3wKOy|ckQ#|FDMJMN2VibrH`0swFb}UJ
z!SO$NeiWg8pkM;R>UGVSFI}dPe9~8rMEV7jm@>^TV6@FEJ{x
z%j13;ZX(drs$N5k5r7`wM-@ttFPSybKHyr~P=7Y0G-{E?5ZEoTG5tn0F
zKQ!6}zgMFZ6{jFOnORfuIKDq)gU8|R%bmDfVRRYhk6rpl+0eKp`Kx33GSfcHC-3S!
z#4s<+=-ye0Rn&IDzeY&54+w0l)y%Uvt
z@5SJT(CxOb0c!7VQ1Np#e@-PWVgK#k$}!yTn%_3eJC)+|j-K6u#i6TbHx
zZRUmjHZ7Bb*slqceu?AMo6lXq&-d{=*>SWV$Q*jRjM|?(P!fvf!!ny0H%d4D`g%V~
zXPZ*Li|z?OH|bo5>X&-L#Eh`Nni7K6^|FS#whY8=rC$?z;l83Eq%)8B>|#ieBeL_5
zmVYGTclu#+MnVv!Ki>91d`_V|Jss75>imzo#}Mb3eP4=0jCA+X^nm+=v}ne^LbzYh
zWt={M^8dzLx1;z;`|XkYs9l;t??+!DZUdK@55jhR^kcuGbsjPI)DZ0-w{$<(#6i^c
zTJ%Nxuc*9!kty6S_nN7d@xb)^yqvtYu$?@*57MnLUy{4#loo6+`Ri`_{vU8XFh#Rf
zK)o=Cyo)<)Ze+mT!RP+nDWfOKZp{
zy^o};Bto8dHuQr}r~^Eo-X~73gY^(Ddh1X;IAU~*Pc!7N6JO@UE_oR*TWz$U2>S|$cFbo3^~@^(q<^&>&ROj@hKRula!j?+Kl2P
z%0>q8-0OWtcU&3XU+}&CF6`e?_?$i-lF@+oHa1WG^KUOizY{IwUWDhM1Vff?6zrc0
zPrg;b7`efA~*!9=LNT^L^r`
zfT~*Ql@l;N*O)%pkcnTZaPr|J@0>#0GhH(9VlJ
z`V-)_Bu~Bk%kSXP@&y^io;kp&VxM$28^#OmZ`w%@jDb*9o9C|U=YZ&C{Q~>Jc`!fe
z+q!UY7Dx>)-sn^u1Ey72Kkw$;541v2slkIZ%u@b
zfXAmN1cIi9!BU?$t&7Ppkm543dI0lJ-RTJymK*_TZNiVAsE&Y-LV9W*HN#-~b6})}
zbDJpDBswH;XN=eyfO*?
z9*6s8`M%fd^@f4pwL!7BXNCaBD4}XTJp{O-eH7&{je_D%9=gkK$H1PS9@UXz!{9TW
zq0z;_LGZ-7s8hmi5d84^%v&`*447leZ!f!yf^Vid+_E0f-Z7s~0hY1I*e>P0`~={J7Xe{PoA8MOEKI;0r4G5CxnaPy(5O=yXKImy;7CfcrS1a^lH`q
zeNqPZ6_d}&o_i@kpMLbXGW6G&&D!VP2>8CL%y9Cq@hDKsva!Xgq}n?3TRY2
zN{Rao*K5psKT)`@pV3r|Pl}HK_01nv3<5_%=P6^>dyk-hkm5z??WTb7@2T3wtML6)
z@1OhLLBl}gReW~TuG$=AM?kCh=l6W!D1^s5pAB^t}pdHeg`<~y0ez#?+
z?1gFAKJ6_%KX(`xkN_-CQ2bhNFaHv{m(=;>?TzD5zio-;96HDLiuGF}{>k3aRGAFz
zd_Co}Z9mlWbQ!&i_;HMEvzIEw$wDS&bf0kK;w=AV*ssHSnOF3nemhg6z#vSob=?tV
z3^840Xd618jQKZh}T;d9j(gmM$SRKXwIb=WM`kUdOJD~O#yp6;*%A;Wee!s&)XA2
zkIvD$Waa1n5WmL@(4zCS`-Wo<((~Y__@eXwg@qRW-PXm9dx3&bK53O!fXd7MIM$2q
z=~%Q>Jwkqf%N<+%e%gK}cNFdSJ~P_S5gdEoPWMHBk29
zcww`Yq5`tlbp9NEPp{x6hvRFpx0Wz}mD$9DdD3~e>jKDrj%95%ghyGH%us!KbCtLr
zp*#cZ4@@w8e}m#&U2k-Qk>0k$%7;(r~(sp8)2WaxX`*zagVO*^cZpX1Rgy%{~gG!MxLs`8(D-
zxq$`8A0I0#{6%`24wv9hcHupRKV-c7r`Zg)Pt*1>=FMuG*hi3F(-#G77x(HeW~4vI
zr{{;#lQU0Z9^GhWg5N(J{*gY0=7Z}htvu2bi+Xhy=_RxWVE;nckrm$qQ%NzWAl|k2
zRxW-&k#qHZHX1L!ti%myyaQtXY(?ct=Z$+bAubO@|3I?aQ)&EN5UaEZCFBSbXPx0?_K6UFlp>>==x%>*v!>CKX9;n^8
zwoUiZJp|U|;(%ii1#Hx44@2x+Tit-x>4mVIcc?v%wlz)k{w0vhBXuv5S80A0LX_c7
z&(VT7mvZ=wCqxgEZgJ$tD%q^a@k0y?``Clx7oJxga}iHnPBMJ+3ZmLlLM|snnT`9U
z(S6U{ope{k;C`f%uyhXj&+8=)o$`e^IWQ-N;-bTiXRlO1y~~$n9nrpaTd?+QGE8UP
z`taCUm~M60%Q6D)A1qvZzw5$!PVRHo=!3j-#rx<5eV8xZAo!mU#N_)@r=+3YfzRqv
z0Wi){J^%N%*)&`~c|X`L!Fr`^7d0-zdT(CX^veV430$_EshWWCyX_pibl^VwR?m^I
z9`48ftD_6gVBDqqLA>T`sb`&mPNI?h^z?h(K0ur%Cb!iV?&H0^s$5rg3~kzBeLR64zUzWe?)#o2NYTh`
zv?X&Cof2C+x=cHU{3^M2sYCr2G%iM(LtOKtrbz$Cu30o>@hDIa;xB&X6cJu$#t@r}
zc`KDX@ND@njr3b4P}|kr?YcHUk@#I!l|+vb^h8TL-0JigGP%kY(q1!NoTqp3n=85>e=@Y|Ga&vYKM*90?IKoicZ`+3+G3?xjD3D0r79;
z&yqPXkNOo)PtqvQqi0Vy$DHw>N1<1TESdA>kf>n
zK8GWzlcygQFXb~0o$W*a6ix{G-5EeEHp(GJBZG*Rvha_5a~~SNx9P`G&3<%3`N%N4
zcON>~nGl$Hu^*)iS|mi~44}LFwh2a2^rJsgww$E~{YZQN=&6eFJ|sNf6Kz)7hunT`
z`=@xPA1!acYr4wckAxoZpQ)$oM^8Av@vQCdN3R)d?LV~kq31grCyXWg(E;vNDhX)k
zrKwG=?877!ZeIMF9{7u2Shh$sj2qQMPfG`2z04*r_K!8d3u8O}^jS!fkb9`=sa%Nn
zHqbC0=!ZD_NYTo-bpA24ex&(`bj1K-+iY51**b_qYgk|8TmioIJ*T=P0pcN98GB!X
zT)L&+HO8EN^!{mNeEy>W;u>0IF+Vq0LxQq+*Sn)HFdk}98ta^#?JhkgL7RnL$4vXKIR?&mM
zm-NN)ybbUIPA6ymzyrVwou!)vAg(96pJJ{H;*sko$`fYc`#jOvIQbsrze-lIKU5oo
zbCS4P8d=AXj7ErGC-6waR8d1GsQ>Hsa=QC4-f4Euzu_-{c;RjBQWc3YG=E@bAr$!C
zQ6bjhcU7PV;y;bnfF7uG4X2{t2KszQXP+CyS?_&v9@!24AoHWUSL7<79xY)x4&bpa
zL5z&X(9Xs-1Q`Ze^rBCLNQONXF!f)eo@N-
zc!k$YrKKClrO~p=W4%Ad|NQICSNS;@<8}r;_4-AKZy~g^`r+J%T_EptSJXBe
z@^vs=gRh`{DuIF9jbFt4E(>ZA0Q$;EnK
z>)&}kY}Zma3#_nxoiqRWB^t0q-03;KPjZR69?lQgg?zBRWFs9W_{-((CIs*KVXK~j
z?SU7)+Z?v5_g0>{V)^B@=fwW{wuF`qoWHEE|3@E?K6<$Tzn|C5DWN#-6V8;|hV5?2
z9nKDHZ&ABQ9`2`?=a1*v12)!VXW;r$ymg566$ky@U9tS{k#N2kzzSxmZ8+U2YtxIb
z+flNW*w3>P5mSZhOX{Q~c#ENcJAwCD$b3cjP}FzauJcj>M0~19#h8e%9eW>1piu!Y
z!E5Nc9Emtw*Co9KEPql-kyvNWy0Ezcr(LXGMPsb3&L#MdN^5xs?xz~F6e7M^LCPTR
zn{kwf;K53RWjy$K`elw1e9PeXlg(IO*}Cj5u4i%DjJO|#QQ;8I|B^1=iFuP`_hlMf
z{l@O6mv;wDBNoz=Uy
z03)UujBt9csW2M%Q=n$F4sO@a*>l9X0e^bqSj-_^K>2qP_v;~+LUGK~vvp7L`vTHE
z^Q6)SoVIwJGy&r>abEO3ey`2#-aWX!S}|`9>^B&3EJi2+N{%{>VIEeg#m$O&KL=k*
zOaLHV!lvhnfLjLbsS;p3QrvSe2nF04!0Z+Y`Ck644=^vvL)X3r16G|qLh%Jqcuh;z
z623c4vw&#)owTMlQDM|;7gwqT6kF|3I1b-&+nmlI9momDQ~ecz{Qa$Jhc%#ns&->v
zerPvy;qqWRq^A^4F|5G4NdCY5)A+tB2M$FkLcR@MuwNC_bNcpK?T?_JEWYJ6D}r2h
zxsFjAl*dY^)?xg*j9^p1-I9
zaWD>3-Bb&%d30HFPlfZ!9O}8_WE#RWk5sxpJ@zhzIGS^GpT_GslpNA?^t;ia)N9-Kw~6%DarB6G+^BQsCTe-3G%3DkRJ3iU?TG4qGdAx&=2rbB%F
zXd8u>@^y}W^r-fMQywqGiA+NZ9UJ@5wNRGIi{kw#qm`;vrpA*??*=_se;OY{~YUam-A2QM_y)oX7@bnM*{&PoN{9z_s>J~9K!&r{8!Ya
zvC)t29<>bU-8+DU!Y1?|YYiak_-C|_uMD8;H{(~?6muTEm!0lZ92&Tx1i2k03So*A!yV5h@}LvGC@GXRJJoPgmgk^elFq{sQYq1%@3)tx4$7DBbDVf?-s?>{S2ZI|Knh*!Oc-gdC4B~s9
z!m}Z2z@G-hc_-gtWKfD7KgH+{d$xf}G}3-L6&``z!RN`M|?%X6}j0bb~Q|NV`0;E_VoGd82ZKYBa@`dy&j{>iQRG$8+vqRZ&$
zFW^i033n6VxylA7ZaIR!G$VR7Fte9p|TthL5%(64W^EKXE^1N~hdR>lqU
z{DNs6=1<}N^!f(0_2NZm3v}G&kcZtYX
z0W45ok)1I72m;=J4p%mwkI9QVYK!d|C&iZt+|DqSJDk{F>imgg!aPQ)KC27cVO5)!
zL7abhrPnkJ@YMJb!k((fKe&nQaYAn$9El9^iX;nS|5?Gb3oo{_&eQG%IR5q+qLr5b
zb7R+u^H@%gvK__p@%+>&6o5IspY3r!q`s-8#PLaG{WM{0-(AX1|G;*9RD;ua7a(I{
zLx&1rc<$B~%pd=_?me9wXymRsZ0Gsfv}bH0CZ
z`l8bGV;px$3t}hM(Vf2#z>M=5otv*?xv#h434T=Yef}%v10y#%i23xSMXgO(j>g0+
z3CrD6|F;{b{e;sjYygkO=MCZKM%uX-aKDcW(F9}ulKT4t4}SjoiYP72Q~Yk$s$%}W
zVj6V@^K^LwYYE)XIhE{9nD1AmAL)w+Y)s*h$L;L8G5HwBg?c!0KH>KpRPWu1>$y}u
zm4Vl9uZWYnaXd&jexlq5P?allH|BHSw}xy!2~5Nt`r-qVYvf*LLK-$6kMfy
zhubls%($oxSZBs+n+I|~g|a2WAdT`rm?=Oyusm?61;%THQBwM07?%&Gn41{EFS5*a
zU!DWGz=Owk;eFe2jIO~u3kles2eQS^#-n0JrD2)5*lZ8JTpj;}{p7l7CH`(T2
zSpXa!6h6NX3McX$WAh&mgx+(|0M-!Hrz(5SKGVP0$A7o%|?g;t}{Bq>9(x~_#2aUmhVqkCwxX14qYl1UM!&H}wdFLeLiPpB`U
zGA8~n%!3N_yj^0v1M#Z_mXv2lMiATnu@5`I&t@oSDRKgycEh6}#UAE8q$hqvpS%iu
z>g9xD)`=0c?ATHv1?%_+lsn$%fL>EBg;_+B3zB?se+HF3
z*ki2*?KqT?EfZ@F`s}UMT+p5gq*$@jCYf^wB&6T>+yMLD
zE>U(r@9zLAz4uwW3+(>(r|wMKvIo)ZqX0^Au$O-oc|8S32atSV*m$4ZAd2$vX`gKy
zKnLPfwY_2nk;pSgPwrPk=+8#=RSmO2lq2pdy8i)OpVHDWe`FAS9W40%Gh+~4j1e2k
zcN&6qIp_JS8wX*(_)!DOi9xhUY~k#7}1T3~+A@BKyq&
zmA*TM&`}mT4=$(n-(H0{g))<@#DU@=#B08FPLf1I`v+*5k8O+~Uy6Q9Z{PufmzH)Zf&AsX
z6vta`Lr8V;@zYn1B*bUu9;9$*1YKvyRpbDBH5}qvI}PLK^P}8QPP-xWNY{13N{EC$
z+PMaC%8nqX$C>I^fk)V+PqG$3oUzdTVE-?e5B;tx#rK|(gyKwAmX5+YxGbNFY>!tk
zuT3TP(E;q_V)B>FJs|&_G)D8U6V4SY^j;`~a}+o@2HF|HuG)A?_^-qLeOrsu4Y$KQ
zzTvp61APCVf4g;MIEL&EA@7$f)vn>ozAlj0Boe#dN8~f)^gMK*gNu%}+_(I8E
zi##0nHSI_Zbb#;7?!+#RpS-~TQ}@r@0=_bmbg@nz?D|s<6Kz`fo-c}Y?qdXgA%47-
zDV7I#(sh*G{TK8-lmGdLAXj10R!#%oyW#rz6-kf_TVv3tfp(3)+5Bi6zR$9cSy%D7
zZg(vIR8oVUxSMtIpb^x|Na1%!9iEHr?5}Hhy&6?W)L;f+5Skjq>(Lh0^PXTm8y^~J
zdVCo6Kd7(ve1&w*j>$0rklW(rtR@X|^=a;wHIRS!u@l7`h=Z-(i&CnAe9vH&eiGdO
zo{wuLJ{RwvacwV)W_ndgm$xZF^d?L3xme*D1K9`N3?hO_N}!rOCO#Q|3+
zMrkoW$X@e~xQ21BDsgV1`EcLaSx6f++hs`sz84i-$M#Mq?dnm?i&CpAiE{x26qpEm
zx#Q29f4JQPa`egAUT*Al-i7UD{uw`QY%h&Z!h%u2-wZUwdZHsjU*6!fi}6G)zOFXx
zx-PZ@dpJW7&KKfz7{d8mZI*j*T0yQ?5vMnOT(UY2`16u3F>mG)Ey{`SbCTLL2+I|5
z?BT)b-J7z%VEacmzeNP+H=Q###`&);^n`Hw$(J`od`SJ#rwh3M;^^t6v3zC2$P@-$4lr9?$=98vus9yZLA@l%^=rv
zk7fn)2^rhMY2075;`dJBbmpdOzi|6gZ?`J`!g9^8J5EBnkukm67jR^EH67+jdmc~B
z9RcJKm*0xl8<`h3*x`Au;0sK*@%o{+QP=O_=P`Y&a})CoM=uQ_Jg%%ZU&;Onn3SG&
zelO%t>#Gii!gyvHw;FAQ@3FzT`i?8u%Zeu&`qVH!k2St7p#nJ<$#-Yof}8_oUp`*9
zZkkas(F?r0nDRvwEy$sa^4D!po@Jc5u?gf=uhr6?2Dy)iUTc2<`xkLMT_49u&CbOC
z!#qT0!m9w=>9&tG9rhsCp<^5P6wjn^z2WCPeE*s4`|HB7z2sBy(g1t;`l_AE3tyP8{uc2_%YFnUCqD@*2ff4WxKrC7
z{EOQZ=bPSxoHePu;?yOW*I7EIdE~+f%BFlSo&|o_4hh$~2H*!T&t#SfgJ0Ao)mr=c
z0z6;GQGtRhBS_$~ZP0$0_iqy@qmruyp0jPTJ_qXi%t;@>84NsX$?jc-4(ulss|;p@
zcAHL!Pn`#TAdoi6-U#a#dii-x+EamNCrW?&4eNo_*hdY#E#bW^C!AM=eW4K!xu-6&
z%_3b*+SVrT88lcOn;922gCZ@~)!&|n_tdvXs?K5pJxbmz_Hbh27~H
zw7bLQ(xaSdv|*)MWH>mD6xEXhKWrL8%XaLSDY=GFPJ_{5;juyF7D^h`o*9I5Lz~W)
zn}EGMW2QzSJA?+8)Ap4e8A5G-tuJ+XU_DT;gU-D>E!rrRg>b^IGt{0S2
zNaPHmYv+H7Mqe00w7=%;n7N0Lpz^fA27l}1j0DGDK^hSXQ*h|T;=l-RDUA2!2lo5ja
z_Gyl%-De#^S6cfxDhgqIHwaZH!1+fvKl3n$o`UuBe>5pBvyqS@OPD@`=m;t{JCxO2
z2L9uQtilEGt3RCGMXv_y)^4RTi^RNy>T-D$ntnXQl+!>_{_A)NYzoq9djMG7B
zpLFP6|U>C9%~6^8p#1_S*>hF|c>`7q(-m3`Y(h
z13Bl6HR|0U*H%NVpAPwsb%B}KKU!PM>&Eud{s5^P+snmZGY72y2N^q$Vg3CtrD{JV
zT-USXv#=bb1)TRE#rnQucrFCb=b*^i1K1utUGDyc^|@5Ww@B>&nf6LKVS6^1gVQShuUByT?H9T{{2UQGQ;2mK
zuimRf;CxZJSYqD7ug-+nXLd5AULVgxh__xN@@?mi5bH^F_KXwr3UjPu4!GWiPY=pG
z0F9Cl5OO&a>BKzB{o{KaoR=->aiDxS5nIzx|g@J6Gzse-A%^grJBQ*xcCss*LF}PmpwY0ss-=_Nc(wxU**n+@nL=^dAxxu1aL`7vmDnuc<>V(842x=i6H)?k}#>aj(O>&
zis{1{fGrBsS1v$1yZ^jgj)k;%S9p*#+=nM<}TZS@~dc${i_+?A5=oIjL;hIm(xc)XKrXPTNHg_2;L4BO3scN49-sJjZ
z+z#L4ph~Vw0F-AL465!1WcqiaXBp6jgDapB>V0Zq^bqe~4}9)G5{<%bY
z@K_TdlDJ#C70{jjiXq-t6YX`F?F#S?Q=@5mg&K&veA4^u1pZLNuwVSSMBph>v}z?M
zN09%Pun#%lU+kW)?I;AhdhRg;&kOLcOzG?AxM2Pw$zIe|_1p+r`xKkJ1mC3-{{f{x
z&>pHSlDsS6&xl-zt^2hP&PCXRGQPm|l8Nr)+)!@hqSgHl*G?nmF~{1j@e%ZynfH?;tlzkjG<8*-
zY6jJ{E{`tvPb07Oz}Z=(Q^xv`?7eTtU}^RhKBHkg=;>{-^30m{+1V
zF?O9q?CW<^()Pf*f$)ppUR?)%9CwXB(+%Ra884Z$YG=^)2f1uilryM0fa#tc$We=K
z%BGbZx2H)$$}{)So>CS;pHsq>vZ?|nLUlJ4^e!)kUEWOj2jx0?@Xig
zJ^S0U4#9r9s#fYymT9D^=C(#5A8;kY7h)
zK@8l_zHs1tEu6bi`J;59G#=I!XZ=%Rze_^VV|NxsPr|z7=+Hq`VG^3(S$;N_orFA_
z2i)A>laOrm%Z|}15;A5h3J}dAp+f1@yG~vt^ptVNBIpbW{Rm+7PsEF{<)cNNBK0n{WL*2@PM}^p^tG>josd%sbHl
zU-)>_&8l?>_T6yretw^Xc#P>tXT8BcZzw+T^DqfbKBJ=d+!#Wc@{@N@R1cvEgTNvQ2^{bULEy*MJiV2{tOa
zy@2)0ry{aVhJcsFv3gJkLA)gLQOuemJqdje?vJdXCZWTrFPG0T51|1s|I#hKL+FU_
zS*sLZcwVZ%3qCL(QX?2~sW4y&1?BO?E$tsdhhMdY%fa(Wi@Vm3r4A#+?0@uF8$8eL
zhZ9-Ai+-^!TzL{Vge2=lPu!J)^~{R)bh~s%kaXmIm7bs>G`TAyUJ3Y1a177WGKjkk
zsoA$t0e_+2$C?=ee0rmS{>kyzLnu(;%4_N-68gg;q0;dS;#B6_f1H7JGVb4Mj4nYO
zr2dJePb>8AjR)RZFQ36aNItih!}q{{-jrA(3q0sY8|}d?s8{@BuW(S{FnX_{e|xri
z1i3BWJRb84&L3*d{aX$^imi$n(b1(3#OxMWUhe3}~os-&n8SqH^$PGKtHzt3U235charZZfb%OYw
zPkgjh3zRqX72gVl`zs`i@y;%J(CcMCZ=DzbJ!NEF
zzvB5G?4n<@Pd(_@=fyLWVo*=iUOQGM(Bm0RHtW|xf5GrwCV`#MKfM}(?U1BH%L`j5
zS1R}2%nS9o@m+$WnPL6VvZKlpT(22Y_uwLw|Gj_N?ls_d@slpwL9U)lj#$SbY(Zz{
z2>BB+@3ruLX$?Eq_-Al^YR@!1wgXlp490?h4<1MN9e{SScqgsUK>F>u{&{Slc30$=
zVSklmwe^KL;L^g!fm=||K?jGkjgT(b+^gxLwW;@l!bNKBiwJi}_Mwh+Ul%pkZsiGv*87S5FV&darfKiuSN
zNW|ywmB?<#{N-8G(H-{y&o72hV>^GHp}Pw6gey%x#Q9y$*PBA|{V8{66a2+Q?hUcd
zwo$J93+69RriOhnCh?y6iSJu~orBn)w|BoWv9Ijcb@KTdpR>PoW4T*Cdo?h|`Rfqp
zC5)Q46a49FQV4m!nQI5J?j%~zmsk%R&Q9JRcaV;as84doMPj{e;>1(|u1Dp50<9Fzch?rAL4TNF4khqUrYpPiE|3QGPXa#tDoT-HCP7=U$}4
zFO7glO9~3$>qD2evE%n3c~Y9-hhJxlyKuhYGlz>9`>WfFh%4Kn_fN2GQkZz`7P&r9Vr(Vq+*H_w31`voId!6sXl7
zjicxKiwBBXVO@pH(x5Z6$9BgV`|ghrkEWNHaDX^?zXPee*B08xy^O9y+@{Y_bl)Y=
z>pM-(QD;MZ+@-r>=QHphb>#amC9DB2){8$xr48%G4!ULrz;i8hGBqyFjNr;wOpqQ;~b@EtR!)UDg5&=#S^g6|b$=wy8M
zK$qSWIzAn^
zuN(w69Hx;E<)~*(>?E4gY%%VfhW9idFl#nFf?hUB&Z>@&qu+b;8QxG$BNcSJ%Hq}}
zI_n|${2|p8TJ)e=vloZ?KlG@L+NIBAYGnogls^_`n#c&E@`?JKU`hLfmI
zmT?bV`UGn3^+tK-Nu^SGVNPTC~-@{KQkR_Y$iAh?huZfe^Wr>9BzSX6~
z_K?ul@LgOCwIn39RCD%9G6}uuexAnX2fS`tdH4#fBl))NYLqDZFq+*|mczVAg1EOx
z{^74Ah@b9V?t2D&G5^fR;H+WfCo0XlaAz3hcKw}cGJs2VYHywj@80XCvhm(v)kOIHHdHmiH
z*uQsjA!B$L9XU{I^8WrX`c$f&Ib<@7JdM^rq=b^t+4!Hf-a*iQYN-!f#DL$0&mf(j
zLrCRdfJS8F5Sq7FMPtBozuSJwdkuV+ajaUo={oRP=~ttJoFv4|I=(jx_TN>1IdD^e
zY8X{88op2MhjSyMUh7+i!@LyTeEZ{H5Qi&rG35astKDYrQgdJ!8J^gi*2q1K9P>p@
zrh(sO|84ue8QM{Dp-k%MUlLNWQfRqc^h6
zomzv#C~1J*b8-Oo-ASY^P`eHxkww|LX^7XY&t~!`$PA;XP3EDSHAWEE@Sthp_aSsy
zgl6y1`62Yy|K|xUUJ{~WTG7>m{+tq^|L8q8f^6x`hxi5|{&=x4LYtd}mfTNLdYm{L9
zOIT`#?T&B5X!+yP?~*4YNY83OrT-n+k$xVHv*1?+os&^+gE-TVt!dFzyGBsu$iGRK
z!>}HAx?1@P@YWqVa%I~QUG~no8dpau)p3Lrvp+f-o;##$Y(q
z`w4i9-|=h9;AehiSWWW-eQ7^wAo2ouFnixQn+nj6arXrqE<#*-O|bg%9nh08_r;EB
zKwNQD>C0~k&__FC+yWm!`8w-K*+k%r%qJVG{Xw5GDaL=8hB(*7XrDI&ps(RP3voWs
zhm5~d1e8He?jrqI1f**>|M21{hxkGB8wqZ(ci|H^_gBGvs=c`5K7m~c`DSRS3wWUC
z0QTK<>fNwD=10<
z(8^M^2HWwYajUl2Ug`A8mg4_4p`MkK29Q6o^)3n9W67@jw{Sb>*>ivELiye2hwSVj
z?cK8R>^NZlviV&c*IQAsmBe;gp-0q+A5hCIbd=1E;g+%DOHfel>V!y|GT+r6`L
zyZ6!pqVVU9m_NNo%oW(~rzHE@<9MNEcgz9IV@zX}N}>SASG=g!ic~HAM*Kd&8p<{
zk571!&p9)yAg8NL$>+2Q_irQQWm3rX1yYjFd0wX@pO^hPSC(kUkqPp7$LpSJ1V7dp
zX(!ITh;byJ4<|JISRJ>&@T*@o?vKSHI)b-P2{`27d~Zi@qCeDnv(Mu6$3RsN{N7Kn
zbrbxWqMFRx5$^$l*Sd6sdr0=S)B2``lnpBm5Cg1E)-4
z{0cb}|1H!1Rh)#!ecI$2;b-W~2{vK*3Hb(pjBOrcoVb3;V6N|qiPo+tE>|ch<{|!fgmZgtUG4h?dT1oWp&cZAB
z`B$nIKV!L6PbL~GK;f$@m1Cgqy2rz*#NqxIJY+XEO~U>5i^fIF46;Ea029gXfx1eO}fJ@;5$2T8)9c
zdgjzT7uX>&6}JtnSA|){UM~QaBujDj!*y)UCMh^x{Nbd{KRhq=>&yB!bzwLcNw$+-
zjSAuo3-_D7g2xf_K;04F-w+oRGD^|(nM5>2WqVp6E`x^nGnZgJs@-X-E(%zu%0;!G
zTkzWiDwfC^G2b?gM6Rb;_d+~(mHysIPvEKQ`E)xHAs%8VR`l3165=63%5taAkD=4N
ziW+yozNOzzw|@h6iFrYONNxq*FO^Zq5VV)7Em8A?J@`4#g3uvI3wb_eya@J4*oAAa
z8^mE0tXE!y?;k@I`}BtuVE!xMx&FD^z&n{4U*$RKO`^zs*;0utlW2+|GIwTa0;T<1
zTkDpcK%-1d*4me+Q1R7!s(fuz=&_2Wf=$>Yir96^>)E9VWbn{9Y#Y}Eata&gmb);9
z;H;|;KHtaDkK+Dx#?zCi>4`yAuHYovIZ=Jqs%rwZ(pY9a^@VuN;ZOlHJDA^Hx&D>s
z7M#a>b6Co|VjKx9=Zc;0g>~`P#ty0fz<%@XUF8pgaXNk2)f(o5`H;xJ0R1s^>+_v1
z_dHk^+jZHd9p;f2KCnIFTp34RGC2=BtjCf2s7C1bM=(yZe_C?E{;uQh7Ky)My%)Eu
z=(~`HadeS7;j@eAIO5`U{l0t`)(w4S7#BD^hT5J-(C+M_wcuwU
zZ?7t?O*@Xnq*sj_%U~VUQvXwv`Em3~oPW@FavU*Tm(%qv97hA!9A8bTj-zJ(IJ?C8
zG1R)nrteDz?3?3z;-t}XN_FrC}^9N)|Rm`&dq8&C2gMo>(zn9LFa{&h=gyRzmO+MDgQwQ>>SCz3*Si8c`b)8FcF
zux|v>nXxd$Jsv?!u~PBT;77?<$B3<+9Yb|RGWAbkUcun;qC=}L*vI1QPUF;|2R}a>
zj)8XmwcHw+`g9CAO1`HUfcaK4ISrW)QNT|$TEqm*fN$Arae0-DA+dFrhKaL!5Wr9TvKj$8tth1w069}aJ?J3h)cg8FlI-Y`!ZMs9gsseH$V(ds_8n|ml=
zUD-FPix=OGpv)G=Ujqk6&}#NhidW0P&%MPBtJjB-fVzZMmkE5IdOAkR!VqusJ(_kL
z{Jo2QXW-?4{up}CX@3~Luas4h3lLY8c}1d<@C5(yknFeft}tJiRi^bg3E~rs-(Pid6tbw<)j*FW+v<{<|raOOFEr${RN&BNF
zg3unXF-vM?cur*t=4m)Du05^BMheagG*DSeGub?Xc0?(xrU{IoMA-sUo;L7*EMvcO
zr@(yD)fc}VAkNzM!(Z=zhp;{~?74$%
z9jw=F7U(ksd8ak~+jzZO$xh|n^$<@kV-IM`xD0xPLYl80^w)=$FZ(>P9%ctCc*pI`mNgYH`rAce_KN+FWr27WUd|7J(*0u#rB3N+T&d+#H-I+L^Ioh{??Hd
zDmejsJ38sXt3J@%c@CV?cz@1W!H9TfXs4S~@)f*ZZl=qH0`rSpKHaLFpr37CuWoLD
z@4@rYuWfi=;qC6A+zs#_A{yf(T0u_6o>llO$VGj&wG9Hj?(XY?te|}QfpbA4*qbNa
zZ+6;2xs>r0rdFsYjv*h7f*g|Et~CPJb4dED;(g)l+Z0SxAZ^fe%M{0XtU{)izXD1}
zeXgK_>#Gty74qLwavG>!gpywZl0f=bs6%%^+oT+eA|Ill7$cOwXC5
z%`1OF2IDRXpG}xosB%;haZ8sJW;*=7M~<8OW4;yBmwXPFhisP6!#ssjz|RpsrvrVa
z98O1H`rCu+SCDlvz-cv~cspFKC^e{$)7qNHiTEuSZHfZsSrJ}4@^Cv;ZZ*f?@|)VX
zu3-!eTO!uomsVC0|0C^Th`5F2&aGYGz~%Xh2j?)##!;+cT+{Io!tLzw(InQR4dpjI
z!1Gw7I{KoeUy#DSXah*gy!EbGBqw{cD^EV@Ljz!O14kEriwQZyw
z=SQA5D8wkU<0Zj+?OqBJ{FNB|JOZXuv|NmDp(x8{P?mi-;q%!%xG&-8(^NfZ5|DSoXCHa3R
z!kJ__brbS`Pb!5NAA&-0=zX};u|NrkI!)5Y+
zQv7Af@{$we|B^H2Y$E)IR>}>+FS@uvP4H^F>H#0zPuJr%6VIg=Nk0GX{pfk3Jt`Z+
zggi+ri0~T~Nx=w;U(?9f=pn{T61miKsv3L^L7q
z$}mK{w;Kn@einNe`CRApRkD2cCp|*mbDeyywDj5j1l&%6{^Vth_J#f?7%R6}so?P%
z7vL$1$NSaLOT>GL{Kk-r%iljZ+lrC?;4fmIM~U}uVtt6%_7*~}AiAFz-<}=;MEYNS
z&j&2O_AEvhV+gl2QGP>vAK@46db&XLLto_Sa$G)X8AE~ZpK@vo(NBrLH;Ct=s@P1F
z=e%2qJo4&H{eV^PmZ+z!rQ7NTEU4rmbP0^SguR(H-*2q1<8{Ul
z2?8SImoWplD^HLK|A(a2LGP_R_!sAP#uVXp8W@b{W2{|`d}$6?XHnGE3`pZ5zJcc*
z?dywIaXjK?oZmI<&%Hc{OmVwPy7H(F0=~JBVUPRwiJ>ed*4J@Y?o6Em%sLl09ssy2
zRmi~%&^wCeDL!Z4#cjxg4eVNS#&)-On6Gu(yrZ(P0_>!rndS_vQ~x3I@^dlllNgmu
za8QHiYS}5SL<#mi>cY3f)L?&i?sH#L2h`x;JC(T(^F|6q%{b0~Y5Du$W3V?Bxl1$`
zpdU&&c*ot~Ij`y8Z!rh|p)=T7j{@4UQu)0EpZ5{@+xjO3v?n+p*?rn?kiHmep#cEE^?TTcc!2C5y%c5@h75H7*2RfI+z#pQ{
zc5VhgGVcAz=rQ28o`38lZbG|%?v5N{S%UTo4>{xbolwHc3q{~P^#$DZ<&Z8QHJ;Q5
zIi{!IVT~B@neE$`FwYvFZV;1(`Rl1YB=Z*_7i8_DfzMSqx>hvV3OwZ_FZbGCnD=hD
z#(Y#Bo-f10Hf?(pocp{zFII|j90k5@I`QD}I1=4ca&0riIAV)_%DS%$*6B(qe%M_K
zex8-#z$M^k-w$w(42X=OYUTDVXB6QayrE_%ikS4D7B0{k&WW
z^cTgxzgHlhT5?|6-3r$A3i++pD*zAtMk$i819+C7HlrbN-pXNTjSA@Bkd?)^UBGJ`
z)SVxkf__I6`_}>?|Eg`vpfEfq$x=We4g6k-R~qKl(4O1(?t2S=1708R)A2ig3>`Rs
zPjdm*1?;3XI=!n9pA#hUl0Oja@sq{zc6e{oQq{cI&+8
zm^tkFR!4T4qxL;L%^R^5iYEjgZ|z~(pgl1evL2|QTKp;{r*OxAPd^Lt21BU
z493;De?PrnK))(*upP#6LCx2xLFF)h#8NeN4?{bJjgCaUhH-YHKCsdYt}F9Y&k2L)
zWAeVT`ybSAnJ8zk2>xRKcGkL8XZWrwx4w*Zf_3p4LIXt^FwQq99zehj8cXm$eskR#M;`4$++_doBy7A4%
z;rN6u{Zn^MK)=S1$-4oCXg`@Y!u4~4ex9cAT<^Sv*SWz&szYo7rcX4yv{!SD(`3?w9}75+FS$5-w*sfwGR0q7hIH#;5rqld#!H)eIHIdq5)*^
zRZ_Eoa+c>SudN|1nG{ojd32?*^1%~;i-&Jk8N&Vh-%#G31v&ca!e5^u|Mj(c8#_Rr
zT|v5cQL
z!eatR`{%Pi=Jozl{Jq%UsGnQV!S>GCYnBz;+4}xJw(5Y4n}n#bo%!dw&j4R<)_uPN
z^Lo2}zW11y`ug%`V18doX=R1&;%$m^ao9g#%P)(>yt@9ABoWUJ>NeoV*I)gh6^i*k
z+mj>Jn3p@fbl8FI>uWw|BJMu8$NfB(tESN>{!7s9v_%=0H>Qs|XOdiMSEgV?~
zaXYx)?jhDE+(|Sb_T4&#xR~N{cb#1Xua$h`Oz>phi+^mfoji2;P9QEv-O2>tc9;H2
z{O2X$td|M4o56IcBUnz{;k^u&6YK9L;^#NArHFHU(^~yhu-v;dGPSte&drvH$MU~Q
z>Bnhj6;6U5*%`?4;9*|KC9P9@XnrB^906wt`qrk-<&(JTyC%yalPcR
zqXeH34R9cMRL#pEUi|z;Zuq{Ds1kb!e&(unkcex~(W(>uq|s7G@VBxjtbeO$^Tb*{Gx%l
z4|ODYzToOv^8W^Fk9ZUHo9=ff%FnFcAyC;yjNq~9``;32rxb63PmbM`ChB*;qDhn|
zSCIb`!+a>6kn_9#Kd&Dq^Z!(z|NVaD!6&aNW$p3jX
zb0()R`;sx&nvA_!WHg*6*Ps0Vf9N-y$ocL*WVAa(#+q(&d-nYQAN!2||7*|te>x$E
zEGJ7t#yTeQ{Y^&6|7GE$DZjB(yHz|_mgP2q;i8N(O2XXzQ50=DuN><)Rq}kHQ
z<6_){JipX?dxaQ3LNten@l>En{twUJrb(in6{GhAc4zt#*UMg4B+`dgr3wFxk?J@b
z{+>8Aw{OMmp<;+Xhrj#M5-}~TpWs+z;yIIQE)n-p-PBC@O|q-rN?87kd=K%zbR(}%
zM7`OcItcmfEnA51asQc}M1DosLE?YA!7)FH^kR-R;YSH^Um*PR)c7L8-zqv0vs1A|Jpoz8n
zFDp~{ak+Km3~^pjG1GoQtUoWmE271EUgBPQff68BW;U@;z$5JLQ_R!moL~Ry1|AbY
zE7GeCynuwVGNK{BG_{)r+jGusTVM78vftGk$N$0JkZ+z_0(*M1;u#%oFHO0SE%ql3
zMjlOJJMFhX^Xda&-~F3(t$=MquU=yN^E-L*3V!b4FxDyD&aVR%5_sNx{Ug%=Zok-<
zlsLj}DLzud{vk`}`#{`psiDtkYye3C+GiL67n4h^4?{cq6n_eS0$!Qo)3j<1?eUul
zZQ}zTtGp<5ya@9nljgeDzz=8)%?EPeI<5vE^(NpSsk;`UZ9x85(Z!q#z?Tj!|301z
z2&Jhso?6kb9bfJYy{s}%=i3E7dG
z1lLu(@s+^-Wb(O#2DSsduv@^NYu{$?tpxe7ts8BVfU6?`
z>mh)YQ6lu?P`?XjVdNxWkD^Bx-nVR*B|oGDX{Ew%F+w2Mv}v(y4p4y4?G*lBIjg%?
zC*G%NDLm840~mW%z%T*I*F`jZv43at=4%2IV9mID(-T0awA*26fM4DmBKB3&UFOYG
zhqT?@lhty7<{BNv-q6m|ycB6s(9b3*Xa0PLwC<9!Dn745TRrM60=|5eCoBj1(==1G
zZTK7%(JM8(E<(BR@b~B#cy0~lr`wGHSq#~ir66C5K0xpoVCA$T9bTub!}>)~8?N_h
za#5Uz>-%dM#poez%uTAt@AJv4-0eevV@gPt1GftD_#=*Wvi5Dt!Na)*c}^-txCP^Vtfd8?$aUV?Sr($(CbS
zZy!=m$t?r4JTOecdZdCadKmLr-B!xd3P7Ll{8JHt>$`Oh=>ujS`z(sF`E~C|5uo0^
zr5VuE*E$&Yi+3>{1NH+NjV^A*JlQ4Ea!du^wC9QXbh&*J08U!}X~|KTkgbSY$I8g!NgDEKfex%Qt7@tZ@C5zZAo<
z9PBR_!Sy{{O-jb?`5wAxhU;mfB`sb664j=0{!t2ZSB!Fh)*3PL>r_pzZO&LAxR$b^-!OCnuT&A9wcUL|q=h{nw)arxLw8^Vre
zT^b?y|K<|C?^teG@dUw>NB54(;&kxlorxF&PM~^>e}qN|zFz85Nyyz{Wg+5+f%__n
zbtKKVUlHZHeO-k8_9$^6c*f7e$7b>M*|etl80*5^2%aF;|A;83yg5PG`2Z#bf{*Nt
zWGCW}uT_kR`<#*2ChUEG?G}Q!{~m}S_)}uDBjMMm-ti!KS&yhS!7r+R8W4O%Fyjd!
z?@)Z0D8K%a%u^nV-X-MT|FtFBIn!iHw4>%HnIDDI#1Z8?2KN%>_e*k!bj_eYQQxRZ
z3Q@o1F9V|7uB4bK|Hqb3pjBcZ!E4&~1QT*Pr+o;yMzaS*`M0W91nLHa5ar_uqC|OG
z2bpJP-Y5Sr{RsUuA*Yb$O}>vDc^!TRa|@B5?MP1d-y`$f0~+B(evkBL0y(3H2>jYi
zj<2T*k>wu$C@1nOqR2ckPKnHSy@JX7*lv+LZ?OJ9uUGUSmw$64V|*eR_x=ALdZ+*8
z;HO4Jy|FH2+$Owaz$RGP3ZT`o*|JM(PZW84^
z|K|glB+}sQ5eS2xz?~P#cujF-{>9jKa=CRJ8QaCl*!B7UbO!ml
z@+q>vrSpG#Dg4RhbN|oZY-uE?nZ?Nccjte9=*s{7cJIHIt~(y9FN{Y*Lq?@VR#r-i
zsEkKORy2$XQBkQVDO6SxQ5j_>Gov!nE*c_bml6%(9mM;3myt;Hd+vRIe|$
z+;i`F&bjCLK0*BQi`smg`oWhg@PE7X{_i&xKYn@6|JRL5^3%5-^5YPtPV)bw6kDc|
z?Oam-og5#nJ;5Y;hw#V8A0B@^{WRi_`ySOXvVPFD46@#qasIl=B!xddrXS<4i{`d8
zlJ#)gfW++;{Be=Jm*1|7c3;VSGtFKSx#zhg+LZJ4TpG#8l>haPpzzyy>^gruTEz3y
zw-ySJ@6*xS!{jpu9ngw<;aOPlkZ`u`84u7WFJ>e
z)+@E;zaKj)`0r!*H9V?}^0Xs2koALh8lNN5Zz`f~$a%2C?FnH=`-xQP5Z~E9R{nLw
zd}(W-l0wugd!a{u*H?`UlKTy{Z%!Ey`Ok;6you{4cB+$pH^r;k$a0M%J>+_;FRYSW
zFQvGyBmG%cu8tx5ZJ@P-To+BhY)5_vPneIsCgf#zUsEFF70(_d+tJ_KOX^#4QHPY1
zT{fTG_Xr%LWf1iXlh?Qs_VKR`o3)5_%czlNATi$=$e&Us@&`v>&J;(q*mJ^)s9#>Q
z*_~+Lyd^2l#5}k3i$o+*znQtqGZ^v2jX_tUp4|Ofa$flJVB=vTt^8YNFHwJ3)16$8
z-P6q_d2AWCx1A))>rX6RN9et9YC@L4rAl&+M-W>R>`1?!eHxBpM82j~jlm*}^N*bUE#aT0WN|5f6?
z^voaGgk9n4wCn}3Zav^BaQPAHN$U^OA>wu1GMjdS;6DkDD=H%F;mbF&Yn>639jzCh
zL-aV}J9`IW(&*c1#6GfQiXK%4k$Pd1yb9v!!-1xR{!?aa?4uC%4THvrb;9hXIs;V_GXt=_z-D7ir^Qb-DWbM;s`r5Lg{{#6;a;%IgYS96UB#K
z5PH6hYi}pk2OkVt(uwP%O;=75`?&8m?oNtD++DXNlCT$te`UNT?!TgNy*D9OR>7HE
zuPdF_(zkY&0P&wmtwd7WcH=*Kt`u$MsJ4{BzFo@lG2tAxdzthjlE*vJFUXI$Z<5dsa({YcNI8KR*WT0a
z8xnHgKB%Z8`aeOk;TR!jW51C+fwu=I$H;N{-kUsMeQBK=`5lV;D^A*}XB&=@^O1;U
z0r}liY-PPCu5%DymP*JK|2`*=z`td}?F1UqYfFjpp6c*}MEZ^7vJ(V;cU#&_;Jot<
z&k2<1w;my$&+&trq
zoGuhk?sHGdnFkW}Gc_wo`M1^tkn?Jr_g&H+b?06s=YutpD@nOGLN}51ya)OF^?UW-
zl6JDO=T&mvbR6m=)2BJ5q}?7ayoAgTs}Ug4SU{beSM&s*wGsD~nyHgZps6GO91F9!
zo}6#Af)dDn6iO|}tzvxsFIz~v-eNkC@(%|7CFKnN8zOOi9Dkqmwd5D_K4N+yq~76O
z{C#)N#fIcOJo-nBEdS`&{p)oX`Q>rVe7lTo&9|opPVo1~
zdu8}`orux@c2yN03#|Ei99qg>HNW7x1sse89K!
zQd0Q#v9KGz{>q2^>vv4?%VW9xd~ZEIid*pUX)gb|e+T&}t;^5v{{R2Mxj25>u#|s4
zqyO#wN*jKDa||CB|8ECtY1IvJ(gu*{(75pa5DU_WYfhY)WP0e}}x|SR@@0Uu@1lu$D;e-!$G~v;Nd{EvQ~m~2_kn`X>0R!7`oaHM
zs>%HI{b2Gl*dcnXA3}e>i_oUw`OfNSyHyw;IZsI?%gqMk23{4nO~$wrc6~0^$H)7D
z(QtmT%oz&Ieo%MuAQR)x31kJ?@4$E;wvP*TSoJ}`)a~8gGI(F78%^t_+d=d9mO|z(
z3aqbVeGc`&^NR7o${$N{Kf=m$hy2BUa2k`zpB?@KL|eU{@0*~4hg#yDOZs>owsqYN
zUsc=(sFrm7P>lP4`|nln^`gVHr(K7?4$-0Gw%sAxmKp%dO)>d@#)mUUf5Z;EpJI(FYJ1jai9kItgW`jZ_~)3
z!GO4=N^&s`c8#B6XkDa%xYF$Qf4`{kH8~|c8vO>1a|g2IBLv5UmD1Xu$z*E
zXyDSmFJA|FwR+tqyP7q+fRXSv==kMs$T2HGmV0cXHcnNZ2jj$G4g8OZuuVN
zXY~cX{(OA(U&D8hTl%q7yPE=a<=T@g~3AW4re+A#$1jkCfU%SA=#t$@gFM8QAi9c=ThNM_<|KWXQ|Lt
zBP}w}PJx|snSswx&!*ck@$slnm9l`!u~sVFZ%y|yLmttaKa5*k6sh2Pow6*for->?
zmYvd-R1j*+m-Z;3LT=!hdbLwjFbz68Zl{Coh_a?grBi{M>Q>>FiTvPQMzdC5#W>WH
z*CVeT#yEp@J1^U6P@#oYq_Xn`6?R)#io_IBA)@(Fb1U9YdvivXoEH^x9`=o2QJ@0b
zVDXYmEDAim#m!tKg5$+sTdZV8Hz*C~mFsACN=
zlG+XH+D12iT}gpL?+_E_77Exo+}w9vt{d)2S8JzZyc&&%qRLHgx?$hWD5sbW6u7h9
z>B}@TjLYcPQj;Lw4JTeaF0ygNcsy4&`&=IH2FC@Z^>ro`Sk*AJbsdb
zJUrqRyB|Lguk6;(#ajm`ASiXEW4<~S
z+@?D(Sg@T6{qJ167GONdu#;a)L+?`Hk9_@{_7Ms&>K8|?S&sa*I;weYwp4iX)T1y5
z$A7uIl7WE;733z)C8qgO;dGg(vObLp6;t;Ng^^b#LR#U;6($9$4l>?7UPOhLb)0%`
z2o-#_3T1Bmq{6%09diR@Xt3(gdHVu+Dr_mX-?J2Xc*Qxb+{|K(v-#Z1)m4B7Nek-=
z-Lz=nD0Vv}!hs5t5$X)-Abf8`6{Z_DP+?o9ZJVkj4OX1C2&h<3gO;vqO+TX;ZHDZf?%pOc#bByD8uU$s>`EeSU
zEGu~_f%1IiUw2OJDZojc25ygx1Dd^Pa7JBGuelug2G4cW(;3L~
zyRX%1On?q|`?^?vb@BV`A(QoY6CK_>qGyJj#<&8OjLbP%$Qx+*&2dvc9SlzWe4vK>
zKWk+tt=`)qUy+^Za921T66D^e=_4Oeu;;{FP4ri9YF(uF_q+T6qNVGt25p^(wrs$T`8poDP;!O0KsL(?M-pr|*LEbOZ*VJaZVI
z%jI(864ZA`pdj}>wuk3oc5dtr9nL;~6|7iIhq8NRu|8ec4XI0*W&K`JQ@_Plw3>uNrSFsw%QVtG@zdzvGiVn<5N*F
zOxK7Gnf8(8cS|r{>LU;Brd)h)JZw(o5q?dO&9dS<(k0!OuJtATH@s1+KGGsMz*hKFUGhRFXydxjY9qb
z=U1{D(2st5bUw2Wc`I{wbp5cw@x|J3Z|VMKIyAm-&e2rs0r@34DX-#sV5CetdZ@k!
z&YI@>67gx}jaA8^(^Q2~YwFc8+QIKhR|7|+lkS(2?yrmb$@^sI5$76ijDkoLea@=1TdQi9i
z4gL?ACB6SJ@+^s-k+4ib{w9I7y!*jkyYvU|DCo(#uMjtGiE72
zRch+NI7AMi?f-hfg?BU22;*%y<$6BeEr|Pnf?GVYo$x;@@4S>d@)=wSyjOhGxd(2@
z%tbfVuyd{U?yGgIUnqvK0A9?hL#0iu>Wgtn!klD;V&xqx==|
z-+kAtU_%a`8|_{j{K?@no-5@DXYTbu`|$pfodM`aw{bzS@xU(JFL-Pvd?pX~8<_s-
zsvD3O<(JA-QGXwJZQ#!OxW69)&9ZL%Ozef1D}VS*??=9g*V}*Oh2cKdw}j%S$SYcQ
z?8!}I-#(ZS5RC5<>4owGChH>i^uipQ_9MmU5B6+WLFGyi?pM5Gjk?isU&k!At9TXi
zdbAGqd)hGIo#+>ihOybdR@D#g%>e>(KYvY^=ok-0~UOn&RXw`{3K)k
znS;nHlGgt_QY~r#ShIg>Ef2u;!}gxcZJ!v>eEIOL!@5kk>%4#U&qyXrOKsb?O>6)b
zq*{lMNHCz(^~lhJFq9Ycuexf80duRzRVO@{aC&t#d$fWH7aqhG?1|}zB<1mb&vx|N
zG!%}HiD1Aiku@U$;!J3C%L#I@X2J?)w8j!ET!);v)FB$s54u@BqvKi(a9ZyG7vEw3
z34KozMxK>^xrWM*>Bxto@ci$X>i|5yvRyk`xPMv*owW
zLO!1vo!P(tWU_$0Z9+t>fCX(EQr*makv~v?^6mv=0L*XfKUy^x`Gx8sJ~>TVBfm|(-oB7y-pFfsE&8=ElLg74x!3xau;EQf)iU>q0eE$DS?)HBuU;Os
zxPDFu3nnjQ-goU{fqcgD{xbq>7#shQa^Nlt76iW7aHEa|qn%IZJ)X-3%bT7#!&@=#
z!rddWa>%2c;Jm)7SBMQ=3UPkYqbzt`@?_jyhYhxxHTT8yF`k0Y@zU?*Y*gse@vp14YBS?w181IoIKLt8-e%DiW?42o5KbT%`SbBwa5=WQCN6Cnhh~S$%7v{
z*--aAgpuxw@;0}{j9pc)ZmY`0yr2RU%Hd&h7H@=+$v)c?>h
zn*+BOj>WbHbD&(DGw9sGfvlN+0?RjYVNxyirK&0i`i)|iHg4fSPKZ$Q=THvVf0StQ
zyUu|+?YT*Dk2v5HA8N97fDK=NEozlQo@*x;BjMMpIiUGd@a@w84n%7jTrp4Qz_Pqs
zZ53^7a2bCO?Lr(7aQ{|f>&k&N^9t(dV~khdJ8Q!XItO`_#XGJbFVw(I|I=w+T(I=9
z;M}q2f>)GMs5!>tDp&0Ld6hZPs%)+zh=Cvjo)ctJr^EXLFL
zEOo`b8~bV4`9p#P7xL>`Z#iOo)|%HP%_jO>XrH;ag!++#=PvGW-m7t8jhA-s5ku5R
zeD!ph|BZF6e5FjF;H>f#@a6SgxwdfEua`wFHUdNzv(
zg++^Ay!PROy^Mf**fTE79}>Fb!{x%c)OCtZEtuxExy6XS*{XvNXHO7TAQ
z%KC|FF1VO%9G@k?1KPftBZ>=nxE|JbzoN&*b!7CL=pZgg1h~27RB$2IetY}|5gw?V
zGe|k7$OF-0%KF9;ivAUJm1MAyVRB@EK7a*tuOlJ*SfkG*$D%|EzeRm704`Tb&2V
z(cSt-V|c)pFc$HS{q)p&4KzeG#Lkq1`G?rq3*7otngHGN`pKyAvU
zWZ@bPq-5+@e!P|g`i?(0((yf_{Eas2)L_Hi9injl01Jw}d)_M`Kbg7D$Jyuh;W*Cg
z$T1(r?}}&L439?qK34^7^u~HK=_9;iLl%rhzDww^;Xq-}nge#lY`9_V|D2kE`**2k
z0w0bJfUbO*Z%8u}D0T1U95p#m)vo;GRTdkrzg14X2P`;x%Rjz3o(UggV)P>`kaui+
o52F^_RXRQSZ+{XSMBQ$G<{)p{<*c=#Lv;+$V9QSKR{0740aOzj?EnA(

literal 205696
zcmeFZX*|~7_cuyNrf8s2WQ{pY{`jdA`vEYCwl|M-994xE2*b7&-ori^3$!Ob~9^$gow
zu)jq9=llvV^Mw8S`Z;_5w;x82(=SHnLmu;`zD<2y!Cmm&Z4|Nr;qx3Plh^#5*t
zWvJT!d4ADd{2N)og8zMYjEfhn$U6~A37GUaq
zVKMwG`T^L)q=MnosrzFShh~N!;Pi8-mf?TX?ad}l=NW#Z1k>KTR}8N|kKwmWX8529
zXb6kUiDcwaqYVEhgW-*y#3q7C46hZ-
zlxwsz)c+7uzBn5vf=#O2nf}_);{%^?k)f(*a0Ri*;>4gekT6AE5Q>shebzG4+L1{;tk}%5CcKgRn@BB8Z{Boe?w6?1xPI((mrj
z@<(mBQrLw1^JmJ-g-cRCCKW#zi{$deQ~rGzPB4quYRsYh*3LxA4}4oi`P=;WVE`x9$Pu7oT_phO&rTZVqe-
zHkqp0N%b2#J!w0i)!ZmAGFbw)CY!7i-32R@P1ZZ<(E4i)V`+KIdORlBr21+(r8jnX
zwqU!POd4tX^A?_F>@vAP?QCDKJBZ~Qi=(MrLVP2oTc&8igt19>uqAvD7BOD$C5`-7
zEypcL>&F&~BTbhOgcS<=%GwbokWD(2wPCBW$W}ht<$r+I@;#yb&P>UBfpV`IqTnK6
zJxFOlM{KgX_t@Yl$Y+|>0%Bv6S-0NM`VGA^b|8P@(7;D5e{?qF!C#;+{a<~+b}r8>
zU4!N4JcZ6;`IxfxDdbI0x!R2b%KNF`2Kgx+KclhU`b713Zs5Z+m%)+BCW{uU?CS@)
z`qUWOUd5^eHtKOdDYM~+^4|EF`fb2_NO(`I20HbaDz*RDJmoIbv(w0*hyDDWIn5B;
zd#-(F70T6b_?<+1essUJM7iMZUqNU`+__1z!yvC!pMPWo=%Jx+JF$FWY(o<2_m0>N
z`~>>zahw$&$kTtO>Y*JH@8|78`99?cqXD3?u{LgESl)0u3j5z*bwaKQ@UXTxux53cikN2>b0LmBNTH1WEK
z^Sf)y#h=*kUHz*k@H-yL&4Go@B1baz(seM^=H
z)d=ZX>(s|c$1e(kGaCG?#jqplm3gg+LaKP*b~jSz`U8it9lqFPI=`&rq+X@#G@ZTe4idAMMw-a^y$eX8Dyo)GPXRVlsXo?YsrneV3XNEFNMF8=<6G|%h`
zFV3gl59PzC$F+H`1=8(rO6mSTtJQN5*X`v~=6X0DLjj{-sBfOsGY!XkspeiQt|u1%
zUFwgniY8Gic9rp)ynBSGpK&+0r+%eulG#!8@EE02bDb;s-hsc3VR2R9xQ4jm?-3vM&x;C)Qfxo_U8?86@
zWDMoE|9A&SJ&R1p8Oz6Z9FF`^K)tM|=ev-vy`56o^O&&IjoTO819@807kQV48p~
zeaxI!kGg+TJJ(#xrt)ashgAP>0HYrL|-3M^u^1U?f`
zH9S5+K7oT56gm2`YRo)zS~Bf!pTg)*(r4N;`^nU+4@3aOCZ9~EQFnaR|vxXs9;W;1*=vY=P($BeV0gwYp*(+cI~oOxd0%;?{M
zQxN6aNPyC51AAaGQofW!H*&_ik25aZMX0|Nn*n`rdX)c%KMgtGCzd0R=g=^Y{_7@2
zKN(gzw)5|MPUW=!j$@a)E7ShHPmJDr&ieR+pA-6x#$^rl>1PMN23g?o>tXq64j;k!
z4zxOyPvMm3a>{pc_(f8quSCX@yXL01QoOU*GzR%CEOuKWp
zGjxp2Py-I$O`-#0k!*yO(BF@13_YvBP_8$O{$Y8BH+{nJ9SUd=iyWpC%O-cvGV)3{
zMz1)PDgS#E4TAIAn2`sbV)V)l8NNf2;cqQPg8&y1hUbf7${n^d>oiA=S--kt4DX`O
zlXyi_x0$}Ohh>qEZs?HNB)4t{Z8zL^3oUn;kp~|faN@Bo
zv^;Xj&7a6GNLB*R2smPAKDFygI1^vADlzA6&tS31gE<`;;dmZy;nQgLri52aPJk>647bBN6A-G+K)lfqcDtqypk<
zyWntMl)s*|GIj#cLG9D(?jR3)d(!F=(DT#2PZR{XwCwNfML-SnKFpa3a{n@Ib_UR=
z2kwQYa0)c%MB+O)`KSw2v^-5*v=gmSI)3&kQp
zm+7=EJPvZ#?&dbM`{h}WJz79xrEL|yfc%YB)|_adyVHgqZ38)LkJDZopd*d3HJ?D<
zwNfy-6{zor^7(}zPc5#iL%U;5zv(>w1Gt?pt#!^Qi)?xHWF|M*nc7-4_aRW{xUKRR
zezJ&g%jMFH-w?Mg&zn6RlVJA(kk@>1-n|*9dPAxyT~Eixy+AwCmge>2x^b`ZhzyG*D$
zj_dGNm4&GHyEaM-%bRW8^znP&a2(6TdfEznHn^^ptqVzOsYI8;XP}nS
zLNq@13!PGq_6?|ouETZ7rRdY834D>Ic_qeqv+7T7oK3souWWILexUTC!I0btA<9IA^eA*#uj>s35b8i#?I&OW;6pRZ6UU<3UKFE~T
zc!}ezFznondVQx}#61PRfd5ZD?$^T(Qc3t-i$a_NuwU=EjR80xTV$JSLV=FlDhu!f
z+EKg34%>Aheo8p*?5|t~IFF-$g0#_2L2Vs3te=t9OyhyWdRo*kZgM#Y2N$f57sm{+
zfB(B~0O44~=%!~iQlI4mG>#ajbB6lqW>JsxsQ2!Q%sHgB&y;zQ`thIshWfh%uEGHY
z_%$cyBF;ylg1IWr&)KnWTT$M(y#@|mHrY|;4G@P#c;=NvHNw5@h3q-a@0c4H;A`;p==I?to)^r&55f>J15b?*%=*H(K*&xH_;`&7=S
zV?p)bRDrqBo<@Q^wR3{pTsWB7l&Tc2{pcuNMsWou3oBNt%3jv$|pj&3Gxt6
z#-2$v!&KhEi5m{{Bd~|>6@f?)zIQBi3;K@h8U1gP2*@G7EzaNor48Wjq1=Kd9$`IA
zfMXEK9fcA2L;m4F6+K6fR{W**M({H8q1nW=uXXhv)fW{+Bnf$%M0WrvEHaCK8?7&{
ziU<<$3*Spx{|pDe4cRdBRKUSwZ#ejkHV^vq%2~9%_BU{>Lwi;2&tarPP{PgVU&gQ&^o`+kMZMGc8I&H;WAdP_ZhrwA^&M+O&p6qs}K5Q<^pMYo=<#S+lV805`3i9VUcBQK_@-|M~;EU#g
zo(c!wjx)syaNasH^}0B5l5-owf9qiAQ$7UL;IBFTKcoQ=#AQZIdk;BrMnE1j9zIRx
zyN{D(+80x1^irNN^gG8+8m>aSr*Qhfjtkfg@OGTTp*jbcdOT
z#s!+0$c(3=nX&6+DYM?U+cN7zd}zUbOA
z_|Pn}$9N;PLpaiy`Ww$nO#a44c_9QOfL99AsQ)zE)=mBM9v?Z%FM6d!soj*NAF!R3
zb;+g6qy41?*U)~%72GH;zogg)6AO6slM~IGDFh#;dH1Fc{XDE^d$@51mQQ`%
zE`;Z>iF-QDV_h`gN%4Efp41xDi0Ix^RL1=h>T8m8lWTw@G78shzg0tmU;jd0r^aWf>_iKx3YjO$0GanQlx3V^z;5#h{xGRHuF(_
z-MJ(h^_P73B9HAAe4bQ}{5GM>{fL9#MBS{$bF}hB+i{eO48%jC4f3Fik98p4N^H3x
zhxk0-Od}TUaMEou#PXl_t!E>he>+|Y@khGz#%r}eSC1Qo4-4PVBDZ%4c)!1eo`^Ss
zq$c%=@{0ss!*dEf42S0;=0AdI;@}l6-xC8t~=Y
zjzJ^~dFbsmPU;|^fBCrFI^cJd75$C`e0fqg*$Ly3O;6tTW&?eBWg{Qr3WwhZ_5Xt2
zTAz&*{Dr8j1GL!9(y-K!r^-&t>)O{FmO
zb7tDH8y!hg;axlK)B?&vTVSWxA^e%k|`agabNeWJ26Y-j#^HDM%zQeuO3zyph-+yrY?7v_~a*wLQ3y`PI
zb8hnjc}?r$@q2(ina1z51j;kEy8V#?e!+0-dpF?a-gm^C0R2^)TDckO?f7)4N*U*&
zs^jc#py5-3R5Dr9~=^KzI6rgobUBRIg17!K#(iYIPb3D{yG!~r1%IG_4;~Dd
z40NCU^6ICcxAyDg*SLREQqL%%o^aXK_zl4C|68*L_lsDf$D+kRkIP4mRD--?hjf1f
z(A8IN+fM*H)m>dPmjNvnQ_04-`TEX}30UvuBinuGH)`j}$_m5x8HqLW>4fi@Thb<@VFcJ%e8G~8$c$MQXP_NnL(qH6Mz6@Yf=a|K|#
z>T?%HVIIS*Wbb8+(+sqf+duxL_	nQhE+$klXZpx|Qb)2|G49f4Pn7
zpD6xH&$;6>U()l$KW;VEU!Z-Eo;!=P-kd~xr&nVl4g8jYKBY>Q5;VUwf7nS8$JhDI
z@e!5}JbShP^&KMPCL?eD`lK1soZ8z9aeNb9CTgI67u$Q0#w&qa&RLYEK=mT*8NA?A
zl0wgGrGmZmyuKE}=tnE#LkbpYYeWPMao|~A8?;l4bqfF&mRiHoWFH!wchX%@P6}M9Qe6K8(OaF1C
zdA$jq+4Q^e)*+yT``%X`R8K4$X%rtF4)dY&DLNT}
zF5Wu>%mDo}azKEN4#2d5{2&FyY|=2Up7J?1&@ITV_CYhCw-m!1;DsIl0h)MFk;+{S
z7VpbDw7)&BQ~|!y{B4r-lI#oLez1Q|
zaDRLhfjs2N9x&|xsJw|FvBzSo|ihcO%jy$e&Aa_(A&pZJ=g+e~1&
zLA@3XOF?flr=1Cw44-q683&nx9m2kR#l34I%Eip{
zTYF}_R&p5lu*jM(#>0c*UCpeci6IKK{8|PRZ^y_03wD0z#1#)6IQ%Gs
z-?j5hXuI)i`Y1Ko0-qcD`Tmlb*M-{gl(%VM?8@i6O!>4=c%ub*lF6$nue)%V@)1Yi
zSp~?K6k;I7A{tAXc4nl%qP+JjGZ;AR`vp0$z*!{pI+ORfbZ-VN4=l%v33&I=uTS^s
z2enIdU9x`r(eeZ9WNAN2;SH1*{A}^-dQrfmzl9r#gEunb&%k7XVEYI^`Dq+TdrKiNSR^TL^8qxy);Yp!>iHXxo!KI4q_mft%qlMZ^XLZ_cG0REic#S{VH
zUqo}I@WA=*u=RB|;_vF$SK;X%ypx#M@EXURb|+YSCeTSLp_TiAdQA7!#=L7;goc6~
z@Xj6C3fF+Q{41#`kNUONCJsRNt$sFt9#EN#I;ZncF8Ep21oOA5Yp?kq1zu{Xd;dJ>
zH_dLy88i7k5lqr{1tHUsV}J!n6KY-R=hGfv{1O2H>V^YK9*x0q
z*}Q9g`p5%|$o++HSmfGVJ~_U6@L%cNdCBVl=e@Knj_2(skHc;s%oyw?z0P?nen5X~
z6nwVMdcY#v&X^}hw*ubw>f7~13(lo`S$7;&;oVzkfZMhwEFz#+?y~GXyc0;PlWf=s
z`$yoU$Ad$F>&pjTSnmeBYA7~*(*^LWvUpci8@yjC9hfAg(n~Tg)Neff3gV-|Zwiaj
z;r&(i)i+{u`pAXhvdUkry=323sbPNkUXq{XxJUg3DY#
zT(*y0UAM4%p{4y}schi;x3ZH|LrGeyG=4
z=O)D>b?Ijp#=^Q$U7c^C0_)P!IeFtlSdYTvCFY(0od0yB$ZI=%|8L60C)pq`x!3jZ
zVLQC@lS?g^1L{+@-Kq?}=QDxNo-MdO%HFBfgPr^fe$G}o3wU_e-Bn)hfXm;uR7WO2
zUj3rsk+twWkM^Bt*t-bw%4K=V-zKrh@lPXVX|H>T=aiIlTKL;clVGBbEezot?wah>w_6N`+CUr4%OF3z)lH$fzCY3Tz~jA?S1Nyo}t8ozn{X>Ot2Wq)Te>Nt)a3qn%S4%ftZdcLM-Z5zOI!n?Cd
z0nZ2b+;?H9=f*4W9?!dc*Kwy&U+`|*d`XZ$-(J^)<6CCCe=GWRm9LY;aookVJS|wy
zbnT@}VjxeR*zp|a|~!XaevATp00{=Iti`D3V4y`|b8j
zL%linD`b!+wQr{RKXVN+8aMb`NYVJO(@2Eg$8lZRLUBs&`VT6oufFVA47Tez`Pgwh
zk5y(3HK4uQ_@|m-``y7e=)K#_>qf6oU%%`4#;+iMyT#oM$Jrz_#}4HWHeM0ObKO5Q
z=r7KT_4*3im~M;LcM@$eX=0q4T%k5A*&xja
z)GqaX0FOZb%5f&Yt=rr~O!Y=G*mQn~-vs(S3?}w7{m4mBc@IAZ
zfRG;)^`&+ii#>pWv&mJFI=ZeByaed_O|=qTgYwb0b+r8}e(ULX$a`u=_w5SlQ?L-&
zWMirujgMD8>ZadK@VGu*$8k!3={i2pbBC_mzebk@aUZ2^;DZT=+jHUkKfSDDgKRG36`G!E3WDBujVEO$kU#G|&oj3P@Q>p7
z^W0Wa*bUV70D-YJKg&cl3X{)JK1Q8?dC0yEF&4np3eqdLqQ^2R;q_H
z#Hsh7_gG8+^mwr66Z^5885`Ei-1ep8XTk5Ha-XlQ23&cG{o4Ev^y|2`aTZTA;O4_b
z`!nSCRXlRYXdlD{+Sb<5b9;!}v<-15KSBJY_Ox;p$Ojk4>bS!+$v?5bWe$hu(Xy8`9=f*hgdzMs=g{YL#(+8a9`2N$&+`Jy5!rO45IJ$@AseJeSFuRB3su%HIujnQp((Qz+
zd%GbXOBt5$>?R8;jU~Rl?k8
zj%q2p?(R;ad^&pc?Y2(x;+xI6O9y(0WcAj=Wj}gIuvBKES{dNp&ztsC1D?FU<2W~g
zbtGK4sJnTfo75X{*?AhmIGx@6Pwa*D`_x6-Pz>?_bzA&y!2Qt)*O-Yb8@ouQ5_}%Q-XcQ=#rgz{@;y
zrEFneh&dFu%!T`wSWTI8uCNcHD?$=+f9bJSj~9YjxVerM_L?S%INQKHR9@P4X5
z#5D9|1Nf)xU)xWC{-3g4PxxSeSM*%{tN#Pe@jCrXeE-(p=_u1|0Kc2c*SlaO3Aj=5
zxqE>G_{l75qt{;{FC*=5RoufOJ$Hr`CxgG(`cPI%4*bju-mv#8pueT(pU$ra{_wDY
zYaY;{4&Y4bNpzf2-}+8=&{p>Za#+h##(n=hWc6Uu3zs
z_e{{ATfJWf$60!Ofj$AcGE`@K8rVJm+0-0A7*{B};g~e&9rNKyP$q
z_+T8&L*CUFOH@HGvdL=dJQ&}D`yb3-f*+mctfq$dWxK;?EUs0T1z<
zXxtkZis#JF7tVFaPaIVez3B|*
zzp#aa8SrlLgxpU+f9@Tq3Im$@!BQay^vZ)OA#o1)F?Um}IM64bcmKfitjMPPCdwn+
z_S@n4`S-fZV`Np7P4KZqjkARcfH}t?|W_
zP8HCg<$fP#QGTP>6^g$ERxhDEYq|@iCCUXdyp1^}0zz=m3X}=AXpK{TzEsjUNqI~|xZgJ$#AKORc
z$mjXHY5sVowgCN}lQj3gM!iTEE_$!;GFY?@^^2!w(t9ciwej@+r=y$|iR*Ch>jY(d
zzcF@VjSuPth{GQwfcl}Mg0$ZqY3t#`!F$p1FJ2-Y=o=Y_dh>1`;KTJ0@g+AN*YUwG
z(rw63(x2q|0caABGsUa31}~gO+!=4FL;XP7z|FZIfj1A(ciai|T~f$F+&{TT3Lp@I
z->Gs%RAC%^Fl?a&R1Kyj%D~p`B?(qXH4oPGYCz&Li-i!uo}-PgEtG
zR}*_y;=TBgQ}cs7Xy>_g=EUo5a8B{e-Zmfd#phz8I%bB0ogVA5LLd)4rQAc68*rkd
z0neJH;D+GeOnjg3!k{n?%WOM(cUH6g+G^h-16)v^7Aa0
zw0!O(QU3RZYp;SGM@?59$sZt}b~!cG@(qv*@$^4S&h*3YG?!k=iRdTZ87af(oCnDK
z^lw+ZvIa|UTW75czTc2)ICbqDd?dEIu3Z71<+
zuPaPb?j&mM8FybP(O6@*h`7c94z(hwHZ*brLIK#m!c%4zeNZnpI6l
z2N4+e(p+&*2MHcuV*mYRJLD+?idy$~lC;31L#lu$zgHOB9sb-&THel>61EcZ!qkX8sL#0spVOwbtBsK_>}r
zk6Wy>yPX^xA84Joxt-|x$k(44>LlA`*XOT^=_aP$qt-QGM~ml|HQ^O)B<07E=cTLL
zNal(YyfiGCJ!rwTnLcF(?T=~>D%cQHFD8=92
z`ut)ix%|%WQmF;x-y%D(*J~&PJ=wc()0R*5TEidJu@o_@Mpn*
zKmxueda}Y!T@d`v*^q;~5TEWyoc`Sz#&!3vzgC#-!a^*wE8Zz*Hf5Kp#n5NO?%}V(7&=r`%pZ@0gr=Ze|f>X
zGNV7DkCPy7klNM1&=CBf!Zp@W1JEzbH|0`+xOtcI&mTQw(Eo-9YxyDWuKV1zA_w}v
z^{ho)0BE}GJ{!zmSG4a<#yCAFF+L0P7S~rFzliv!BJE8r<}*fC_|`82-f5qO2EN0J
zTA@#V
za%^#^hTd)Vu
z$zb6T_#p(`FP$ZSQSW-|Z1`m&+*j;hu@h;RT5)v_$mKi>;IHDqebfwtGkE@bZ>=iD
zIAHyx{q!Bps$KG`QNRyIgjS>dEzQ+|cs?n0_`JmPy{xu|;CY*qqU?_NUpLVDG|sQ*
zCp{gUzrt(x)zID}$EE!cZ#*39x`TMZk$jCrx!N{eiaXC)Mhqga^xS~ruZuh?tI$r<
z54);R?rZ8w@ppN+2+iM#Ou&Z~@Vk!*eX4vwyu(M0v6Q
zUHW?%DUa9P$X|9^N^yUmwZ%r%Q_qZ{_}W7@=?dbK8A7gmQBO<8)*ET7=@og@`#3#a
z9r+Zla}@twS_40a171Ju4#kDeqbDgn@a?oR%5?+ds2!8KzfxTH(Lsdb*HyvAv|g{E
z7#-)q_c~Pmrq7GwuqGpGnsAKyOuD2L^X
z)E#_~#w@)@ai^x?5bbAWO$^29yqWtb9uDbN`hnwd9#}-<1bg8tUy+ZWewcoT-ATjm
zu^m5+El#)|;?Dhmham9#>y}p={NfS{o
z$G{;TdAB;hW5^e|Djvc2a*?g41W_LA^fwvT!;J4j@RAX5dMMoxPS^7j6x3-BJwbhhSWv@`tf%6#1S
zWoBX%Fb)olS|W=3_Vnk>Q@=od|M0XMh$pYSIarAGN{aa(U_XmhzVuiDAE?%1xCZuj
zd}N9s+B4oI{15I=3D&C#nV`3M^1AD3zzZFEv>Wr1C1Wu?*k5*Blzbpikz0FBCxKrW
zI%0R>EVOg*?DQTLz@1-zuW;K6o;=m8ki*vdaT?cqimz{Qz4d=&W*;VHbLjN9j(`9XbgI_5u@NR|v
zo|U+LTn_!Wm2EyMb^-8dc$HWmw4-0orPUAjgF^-rlcS-XpF$%V&2@19c1mSh)?x4q
z7T2qtK(09WQ=t;hF~29%bX4H{`C=h)dkEggMi|td%T(+kK7XH&U4Zt(XS)};H$ooD
zKpw&p0GX3lcAUz
z_;X6v*Ntui4Vmu8pXb
z^h-{5ZX@rv4UfI(Yb8sL+*jUxu#M!0%$>xJXeWWyR?|CwbgNbd
zEwE`LOU(}+{Q`EbR+usRWTcDqtlgOR+z9+b?lv~&(GNU(Kr{fyELxm6b?J`|vMBCZ
z!AqSk@`+?9H@xj3T^af{3n8EW!%lHYxGemR+Qe_jD50IiHHiP%UeZBc#!tyVce0B@9nO4+8=8t4Jtc3Rj
zAC2t0FK%rE{B(ftFT0J%^}iC6(d{I2d6rj=i1d)G8*(WoI&j}zeyx=UaM_a#%j?r%
zKc+Pa-tY&!llpRgC&rD9$23IoUL&HYcy$ltFS~n&r`E#0W?RXbx&eM%YqOVE7xWg7
zOR%^Nw0*C=OqB+lw`r3e-GKcgCCWe174rKgd~t@uu>LKDUcH+L{@~I2R1G`0H+Z&L
zyWM0i++(@7?K=(aHs*bOVg~ywW5MMqc>l07b940$b=dc<7YusFD1@)z5sq4YXS?aWdY^OP;yvj_22^1i?RezC;}g>+u18d99_+
zS>ShmX|P6Iv7w=JJLZRPP3#iFd}EQ)uI+2Uo-Et-7WjMLPnXW=J_32GxFs*<4`t)$
zYhzxt)b+x_ETC3%B1ZjyMz^}}#60dQrO6-hJhX{bjKVysqx_!%?5C(;*&CGKyO`;K
zbTrR_;+N*}D=F^jNnP0033|u#4CuRoit+umsK?HEm4N5*&!>NA{^$A2R^6|_k4BxL
ze`jio^GG|6bMvF;vrzBdR26AFZ$Ibs(f9NtX+4981I8ctHiYdxJ(v#{*x)zKy)tng
zPfgE#gmy;A#8@Fe^B7S?YO1Za4BM4*d?JYLrPw4>{F3z+u3JGaK64YrN2hAFXkJ(7
z&y-r)Zy;2Ia@KWQim&d(@lm}tl?^lxSDhGEf_n3MFY}<@vk(O;A9p;k4&~u*jzIzv
z;wiN$AxPt$;Cc~gzqBT8&oxsGe)tA)*kciz=iUE2?=AA)M-u7$41~U=cC3AS)Hi1u8?QAdWej-Xe#1WU03g{awRR
zs~a~EA8jv{zl851nV_-F%vpbLJV
zP|hmwAp;#K(EaDtcm#MLYaA57nEBc
zx0{6b(pm}fM)>~5S?iMLRq#iym96LTy^hj7Z%14&bwlp2=Rmo7)Wc0furD`N*w*WT
zp5UolA@`v?GA?1e2bAaB8GM!k`+Q$c;hpEeZ#=VxUmW~}>}f}S#8<;sisFcedY;c8
zTLF4qSAP!RyQ>3Au1oyj9o*oaZ*$Pj&zDPoyFnf)^geTewjW
z;^rL|JQh#Ej?j9~Fx-z%ewR+IgMHe(1_oEIU*QdA(f*UoERZ?{K@D0p|1W
zeScRh4RNlbl5T7^oJTq(owtAbLEc?pQ{FI?2fJ@h)qptlTVt+0;;47aTmFm(|K3%6
z+Tm3o+_%X3J}?BmR9UV~GawJ}-K9xa3EmNgXf94X1NVLwdR2;Dklz@~zw4C@;~w4<
z{yK-idqkJ;wR4Q&eyS+;eUn8mnX@X9?Hx8qOdf{cHCzI}1N6wgs5&3=+ppy3tk;0=
zI_=)Z$(sG-&K)tooX>+~=AP~MK106Z`>w(a0)GG6OYX|fe!!bKA@#Cd10>^(e3ZNy
z;GEkDC-+E0JUZrVenbcE!9`rINR>kV>r1qRlhy!9@*Dlp0Qa_+qtl29->C*wX05O=5Ah>HOOqx!
z>T{A9I0ed+Dr#7Q1EgTGQU6u(0TLp8w#m+5fRvbh)T-4SAbU4wTt9rbpUk$Jzj!6&
z-)RYM|c}(Y!NssUJ6Wy~##g8GM$EvM4U?MU=;^lKD
zd5rdxcLyKu;h8=_-c4von!UD-9F@5{Cb*`JoENKcwE{d-XB=ExB-=)MB?AlP0MA_g
zcy4_e;Dd2#?mH%c-h`k0^PODUh~8SSG8?ltva&(g@wQ(EluHiZSx7obXa0#fb>cmM-+~KI8bUlZreW1<+D0t4rVGsVX(QS(
zgMXvlI>@%*iiNHg-K0XMSiJmB56Qk|%5T}%Mrw+a+gBfj@pY|Ee`MQEf-j1QUg_#2
zF6LRKWx?I>ci!w>uAKxsj=c=r>eo&pjwU`)PVV#2qIZ
z4P0Cy5BG4~hH<mR>)-2!_pAGjNJ_6E?P#$|b
zt4kN&FNYW&TF|z=o6I+wX=wSOhb;5v=`e(LKNwY~N|=?aB0`%=^~>_Tj&4X(oObyZP_v&
z-jg&*+D9(|oTagKx6)3)UE%rnFFu9%?6}W{5(7Wvt(tVw8T^ipevSqD
z)qUD;9*2H|^K|A#qle(Hb|y>zb_c(Aq0IO94bYQRpZMk#_(O$lx4#yGU)h-R>Eu6?w4XzJ%RFBzk3l=ud;O|oLI3qj<-lIB^V%n;_i8XM8Nc+_VR#4Q&o8em(Bg?H$do`J@fK~F_5xga0f_jkA?aT)ZLeg^(<
z0Xi+$>>}P5RjR&Rf#+E4@G&?19pSSzcUeln>!1CdVhMau&V_Y&?kz40RYn~0()`W{
zo+r<)?wg12;8QkO{t*VbgI>9Y4#>SjH#^|@H~GQ}J3O~?yS5MDxp8W-0ZR(Te{{-(
zFg)*c6b$Yy0=ZPA;i(+pHKs`wm4cn_;R>&AfS#1qP!aw;9`>!epw*z4wzA+B{_fXz
zn~W#g8L;B31LB{B?;|%OehKObn1uM{;xJnU$DOzOuJ4zh?FRl{@OaK59A#e{UzrwIU`9DBskIS$KX=)_ipm@niJ{w-;!a;K(({
zA)sdh11NrZY+YuKbZV^O0XV_ALwRZ+tXH@!}uN
zoq;GH-nsl3wr8%#wG#R9t35(cFKw5#5%R*y#}*;Yej@uFe}COqXQq#QPe#>F)EhC6
z+k(7V+BKSY`f_j_7vg8`ep@}1=l+SZK|TAYimu4dy4#q5{M6SUO%Pw))N6Z#-*bce
zi%^u$Xx<2aaRAoi(v@d%eW@33;C_yJ1ryKXyXxZ4#;y3BWUN(=A^p5JT&5J{n&Td|
z?3#9KYLAO~WHNpZ!0%KH$1sk)B(J`Jlyb_NwFj+9poQ
zLpy}rwhvYUO$-itaRB6Y8xIttoviUE?D2a&QEwSTJM`~2-NW^)t>k(69LV)oex!eY
z=w|Mnow!~_kG_gGhjrM<8?_4G|0{%we{2N&@@;Cv@%d105I?p9-*ePl-|`LfGZj&D
zJR3ppw>j&;FyvhV)~}1$3*}ZS8aq{h8k)(NdxBm_aOKha;73N!D0fiu5Uv5qk(S%>F1@K-;<@Qsr`5-q=R$e$YqKAl#)Mw-Gd=4)UdOZa3>CNsBKl{Kx
z=1X{~3BfzGy@g`A^64Qdft9Y(mGF-LvGPn=
z@C&u~o9=%#=_it&oBp_#^pp5!3#wA4!#!X{V7)7x2kdjrzxL^|$(Zf+$_sw|q(q{D
zn?I|cESzI#Y&F&g@2=Kg&V+l*`{iF2UV9F4>;2K9XOaD6pP7hw$$@_I(b(R>ctJmr
zaaK$iGKDz$`<~z_!o4Il{avZ!k$xh^pRGBMK)cGP-iyxfCmW7GDzfu{zkgD|b>gu?
zFZm`~eLvN|pG==nR4=C9PaIf{pABdA!{0YuArl4p4V9|%uTvV^Np4nEpSou|IosBH
z!0~biS%36d4GYdoiBq2ol8W2O_ak5CZ<6gGd}|t3q&{vZ2@7xZnnrdItC8o9kDhjt
zu0z=ZeJeW1rm_XEPap0e0V=^`+m>~Z6N8((hc0yx9G4s?*&m%X+g6YnJZo@OZC2K{j5
zDxquNI|!GAl6>{LPSP+n)!`VNm&Bh{yqV?V=nD5$6}jsU>A`!j
z_t(d*`SPrj)Fhr?RQ9ur%uNnt-;ss*XU>NkV{l%7+|Yhu3+QwG+;cq1yOZ#g-qI@v
zT%qv3)#Kak9&$Lf=#dqC@AGd$iqc?RG+P(Vmv-nRTPzPH{hHTBLO!}~lHS-uM0vG7
ziU5u{dwuuvQ?MR;UnNXxw}<=7rU$7`l3nooV+C9YPGrtoT+}oFmt)7bbv#?)kqOOjuf%7sp-zDTDtosX@cgFDVnc23GDc>P}SofjoAK^3{=PpAqf|b$j8c?JMpGfKh?d!~5=taWLlG@SsbrKy({9LCRFX{@ktC#{
zl*YU9epgoL_dKun_xIO%oO{pfzV7S3?!D)E&pr3tB^3*t!N=;3uAq7BbHPwOb*zuN
zcSY1(1b<-S9O<2q%li5Grz809;_dOKpxqa<`Sy^TpYx%p27FTQL`P@HX-m&2ruSB<
zMgFw{ZJBP?cnS4cI(VuQ>4E3>O~=tZoxd+?FzFMQ@2@pTznzPeTJR23E&8i~{QUHn
z-r7y;;cwv~T)dK3!yy5YP
zz9U9OC0TCAKLWiJL$`|YAvZsq!$d^VEY^W@34o){0|sEZ>}CKIE?Y{k-y~q708V`b$SQw
z2kl*;IC~;!srtrYFClkrK=1xK?6b&9yqQdL#kEV;Q5?yFUNJv6(Db{ayVD_;SIK{S
z9(ILNfpOq|$Q8w}TT1Pe(_^xAvgDe?~p6N_xyI5ksBbV!5q`vlaJ4U%HEEo{Y!5cwtHDqw>?xx4Yd6
z&+hDCYn9q(Y{PouYp1tTEcAZ+!#NutKyK5>M|sy0J6W6bhD)0JI+)E2y%+NEANm`w
zWIhnIJaSplXUM%+5NF|r{iUO}JJNqgcfjxV-Q+sxbA=xw%JZ>aiEzs@&WGHyL25@s
zFJs^Ak%X_VV+T9^d(%mE;Di2I`$f#KPP^m#CCnA`d-A^7dBfr7{6DY2kr~LxCr~gUV}W@!86>
zkfUO5-K;0tdztb0Zg%NRabXGc^!6p{6Saa6-(vgbQN{**R-zN;XkgRLQigeMA1m6;
zHpg#ly(-BPqpATbfRR5xon|;={%8|JLX3#%Qlp^>KWO_Oflii=el6#P<6rwy}mA0>fQDTUjAr
zV_X~Lp0r+HIzFn6Ew4+H)Vu<}m{np6Wwn$5wtCJ*GXQ
zjh#8NY>5==D|02Z{$)7iUX>Y3IKe+By7udO=*tI9f7k4RzPxLsFyk5g%lURWhb^Hm
zmA%~89D`mp-@12~+eqM)yEav2x*bew)|F=K3am#Wg0Acbu5=C_@T(AhPCH7)o~<6o
zW3NKI>nbcd*vq~0c^k2>*zSB!vmNDa-!3~c0DkLNj$0Zf+ws`5Ro1~L<8U6}w_M*%
zjDz*+F;C>NPLFwZZ=Ota8%sPqSl8hYkHyM=7xa~Ou;|xg#5V%ZKK`p55d=ApyDdIt
z(3cGkg~M~8FHNK0_ZdQ8=BT+T9>#cjrYJSF2IFa+Vd%oA*zYMStM%3f&h4#Ix_Ak4
zV|d#h_G3I|>OSm$3c2dZXOAYLyyer=o(7=2IX<0JLm_uH{Gz@Y${T#qB|Zak(|l^b
z;ovgsecU}|4EBSYU(D!}fdsOR^P6Hn}Xe=46g#8}_z1%}f
zAt$i$pSBnJ+eNoGjn03MbgMdJ1AX~4=H0D2JU@9LrA
z$%v)k-y9WnUk3WkqUBm5^nTv<{br=+oj&-KUx1wWpK+CIz^64;zSRYN=AJmA1@-*!
z^yxa9r-yFb>)Ql5tBvop*CKyeSr9H-fFJq`wSJm^oksmuB|V+><BYzziURMI?`N5^rT1f8b!F79R{=fWis3v{)*tnQaw2$?=
z?cWvJ2eN*n(nWF-BXsPDcikcyuL^#x>tNZ(;NONsDv|!H-rm$idVjF(kWExx#EK(|
z#NU3^IF{WR21pOcjKZ3q?g;K`)Jd8T5ea-1j0E7TYfKm54zq+Wy2DrH_ezh
znbyw<#T9eu`Aw%+?j`xExYf~wd&8ymovQwVSKBO(V72
zR~cS+5a{LRnOwb=e`wAjdF4~@8cDv`@?;h1_aNJ%8?=wnoU{~y4{=VV@39Phzr};J
zT_iVip4Cix|J_BaPEx;mQ9AC>Hqdd2dzPtyI(nTrn1TM6wdYx|H~Qt^hSP_RVmv)=
zGFm(Z`Ay@WHf#ai6`nm$0dis+rbdl`oQJzZMm)x0;KDmn$#~ClH@z`4A$K6QdDmX>
zSDISJn1ZfPemmw6#*gILx!k#npBsNm+(G)Cgm)tIAwPFQRG$mVU7WX+uYi3Fhm3hs
z33rZr-Q-*Zxy0#ym1HNV(;E?g7QB3ah#U|0yY8_=?nOi1RZqEV1lGy#H4Z+=!Foi^
zMDX$g&b__fdNO0IUkAIju6vCn^nZiz=qvBB?l#x%m3$0)q(bKw>(N-RB*}iWIS__(
z@nQvC0(_TOc4_7})aTX-`=uk1KiJ$yakeJ*xqADJ#IRpHGWclYbj)*dcd}F|UR$7g
zqU>PILqo#~-{4VWAt>VIt$wh#Pr!DlXw^@>CSXg|+@or7e*gF2
zi@nYZu|GRW_Tc#*>`#PU-I>@WV50|G_g2Dkfmb
zNm6UauIOe>H$$^8o#QhzkJQ8WG4Gx{f2-=vd9;fivurlU|BJr2;gNkei?=)ylGiAJ
zU&`OVt06ZnX~XnYSeLFy60b4D{!Xj^>U}DX0=D;nmwiq~H(T|DpX!_{V3jrvRWGj#
z*ypD`U;cvj=*(7_e_X({91qr;>=H2ZPVKHK9kL>iK3NIal2>{c?PUbaB*N&N
z6WVE;yR7>B5&`?Z&?RofN&(AU=`<(BQo!yWXyg@Y3fMZYDKoNhzRpJO?SbdN1nf*g
z#3Ik>0#@IC;*g`WfZYL(cN&8Bwmxw9-&r2J$XDB$dxFPgr!5Gdwhg@GgT|j&H?Lc_
zX!sCi_!}Sd7@$zfW4oKKo(?VNu_K8#$qNd2%(HarsRfZdrl=mZ@`w-gZR{U|?s6SyYkp7pUr1o4R~zvgiD&;;XLM|6t;=Co5voVb20k3
zh{vAZzjx#za7zE)m6PUwYGXDf-tYD$@tCM{NID<#+ryKF71;2YcyF#DoYdP>qQHYDh93JA2_*~p|8Ry}q)m#vjZDU8b#B0{=Z)1D*Hz*(0;<1ui
z!x^Hxcve;@E$R?LdFNo}nEmq~IT>T7$a@s?R(
zI~(3*Ex)>|jjim{j=Kt6wB+4do^47SyHqxN=~&BlrsCh2eq&1qyBy_j<{8<>?v%J+
z+X(!!@X7jpgWtj)F|uNs6~`P2X&OYRFitNW~-Et_&Ra4GgRVl6cqR{;mTHwxw_qulT<
zpV#x#P>(1554{`Vw>^FHOIhq+_VuLPm;jvep}jlB2jjl&li9vfs}2@7SS3c^8|#!A
zO=4ES?Ll=Oxn?vUN}F`ZV!hGa;Naq_0^Fgw@BJItHH|lqTHk=@$5b!Au?6$PlOwdiAQ03Uw-MjeW1-|Vds9sPBLEaVfZV|
zCu2`(O!$R$m`UVgSMpn4lxn|48t)hFix~9)^P}upgA37kAAeN9qTeX@V$jY7F37hy
z86EC}d8xV7!CwS+pqQ5FBUWQx4Zov#Y8}ea^S|kM2J>Oks>_pygBmbvmpiDZQ=f7l
z>6ahp{`=C2^y@($lUJg=zF~TjV?oE6e)W5V=N;~yEEk2GT$fIw1!%BRV-4Y};Y)hP
zlYSgr|2Fz3o0mjOWB0@7J?V{Sem=5vu4SOqvlOzU&)YKFpY&!+
zl*B$-mydeUwT1To$9Tr-6MpH}Zp$G(eD-lFr-$PtJ&(LZe%6mEDkY$rxot-X_k7SX
zttWlF?M+(|;U+gX12?iibuF7_Kv=W`N
z_BMxKRHSEee#P1WIUIi3=Vi~G_g^HC)S+@`EFR7Ibr=-{bLYx>`X_VyB;ox3I6ss>
z+w-I8e*`S|;LeeCD+dmw@+XhF%;Bf&8v?lVVf?Iz+LYcU4i+>l6>hC~&ip>ymcf6J
z8p-*sl)gE|#W{HYP>D<5I!bt-N
zZdcOxu8o;qM*l_Lq1SsHsG`alM|xkzi8&ljJaDyx^8=4KtoVp<)twugyb13(tuN&K
zUq<%lT_*X{v+tQu{f;KQb0fY*yWO5}R*u?U&ac_dfAfEo-}oo4fb#bq2;lr#mhU~i
zg6dOTaDelBQNDbRJO3sIdylRX>+Q-qq>9jXvNt8qS9=2K}quT1Wjw
zbp9Q7r&On5={HuO1WN3?lCvHKyUw=F)-qxw&Yof=O4EvM|{S89hvvYNRxE)0&f
zoThWIT}HB;f0xVDJU!}C>Yd
zHk<-Iq)_lnr4ja(q3Q$MfnTDkT8B6xJvL^l{7=k3E}Q+Y)?t3IN&0@o8}m+0*4S}1;M(XXn~Ujo)_RIMUCJSa8|~@`DQY
zYbt`jkmmJrEBLq@Z(rU3o=g^V6wr9SazE3Z{Ov{-j!~rX9#^xz!W;D-dUx=}IVd+?
z!DLn_)}i4~^`y%t0N?KJ@n|r^KI43AGrKea>rJb8c=c8{Yf(QsEEeT1_xyaK9(ZqA
z$Hx(W4qzWrMv(KW1~|B_dgyZaO;oCC4;zU0RZmbj;{dyl#U}eZp}?I7i*t6>!QK)+
zW9xlo@P3&mZPsI+t$lGdZ!_8@Wayj21HgaJA1M16>*gi7%lk55_fO9+{~UpO4`8)7
z2Ln%D+iWzv8|O+=#j+G5$J0626PGff4?-S3c^?J4P3yS!G5V1LR&5{LQVl%y##OL#
zp{syd9`st8^_0&d+Ea?w{}nKIlle6nI`E4bz47I9#3`_ltQ;FKU%-alFCTGhfPi%l
zv~qgu1wWja-%sO!kIoqHnfUUbfcgB4iQuuTUfw1-4Nk7l+FZ!gAq
zjXkP6KB@@V!1`SK2UlRn8T2B(rIZi9-THyFE-y^DlQdff|E2GnSiA_IolB`Y<+4}6
zt~R^O*@O5B`^74&$9?59y{r6!t7rL4+CId8;blHsYQJ^n+-UG&kN$pJ$7eHKqUNvt
z!e=_{)`OC6KF-BQZrk^Q&x%G=H@a=-v!ITubt9+p+4iHe|D8}5uzl;5L}X~i`hzzBH;WgaDZGSsy&4HN_+zAZoe0L}AQLRHa@xA39sYSoUKq{5d9WJG
z%~=}l%t=K1o2miuNa8oX$$uSep-x}t7Q}OKwEBK6BBGtu7aj3B1wRbCVQ|kH%~V&!`McC1O0E$`
zJQlI#-=^_@+Sn?uS&#IABY#wzEL-#*xX(^?(t+zZmkrws_@r_O~&&3&lU|fn(c^t);eFwX`fuLo
zwX?Z9SBpy@<}ufh*)z5s$3E85;hTc~Lp?VY=;**-g;zNzVv2D)3ySqEFaX~Eyxha=
zsC_#-i5pqg;XJ2=Vd}$@mNr(d_Dr7d(Z){g>ONb4e`;AK8~8m
zV;epx{ZhmKiEy}7_8xI5_NCk}7=>|rH~H??A0yk@!lxx?+g`V`M5oCmEe`PS-z12&
zh286H@1%ltSZ@Y9@D8;hzD2*9wBrZF>qy&l@Qix{?7|5d$NaFL=ad`yTp#m~{I)}H
z3K9Q8blQ5i6__8EhJ4ZtCfqtgyZ;s5cjSEC{h(&p5i*}Fn$zFGMpiyp{}uD`vwu+$
z_Q1IYURu^3uwSP+t7IPlE}Y%j9k(0h9=Nikb2RLW`MamKFT?z(dzL?16!!R&TZ(lL
zVV=BHq1sswyQ1oOFEyIagX9tqsUY8Sr$U50@?Y9aN}{;&bq`G>9$=ojHM?*64&c(h
zu$;~-z*{$r+x014qH4jU&akNrHhdVLr4N^{NIUuDR@
zIrd%sZIttOOU!ffPZ^Z@<-$hLlf!LCQe5-Q#M1`~k-ydI_}_fgCoH_Bl>Cc^Dm;rJ
zd(G+0Wv%0pzv`6zu^pgW4zCw6K>n$q`m3a8_rEw>mkGI##;ud}ecgxaUjl2aH7emuVm865dE2
z(#_RtlG6NJgd4sgS_a{O(&v|L{{!86UtFHb+oJu;k=}RE_TNa#SCtyV`MbQ;m}f|K
zMyKHP5cLe$z}RubjEwCqtM
z-llLB7f;XJd={=e!8xU!k-I2=wQTG#%J*Lp&7DJASSG}i4g2Ca+$nb(+jhwR_P30q
z4u8G4xKoR(a9stS&rgTB5%~>)i#UF$$RrNeetPzV!_5OP0{%g6=m)@7P%BFu=Yo7<
z3Bn10FJjj?YT1coq>I8@N9q2E!a@9s4wxZ9?*`~`=|86cE+hZ;3z!>`o}C2P2;L!O
zGRG%LsdF^6`T!Sa;mD}9Ts=D8*mLPK&gdK<`B4e-h(wM5xK@S3?c!5HI6L5iV>VoS
zutD#+^MTteqzec)*Vq2waC=H_F!y~s7A@m&?TKPPPQJB$0_QJ&YZTxwMF@m(f+aQ6x&6~g
z>r0+d|LrY*@3xBMdjj^qrGEb`&55gTq4)<*K2<8JnB-%31YM)&FWfZa
z+C^dZ_;yM!+pVrb>FIuMoPSkvZq{AmKW9J1!AJNJtx3WU0`5y2sh$S;CtH(rJi)g&
zK5L_Or&#oheLFzC_H8&Oi}XSL$qHomF1OoNq6J>!nSTJSQ^&{`6>9_UC0}dQr}yR?
zPpYNy_R%4GYd!d%C5~T7{!f$Ub@Erz%WmqR@%3O$&vP0_AJaUIX#9T8nEIUdVVmU6
zhrR*T_g6hc^()wL@HTz_&NP)6l1uLT=|}t5J4K~Z13-U=A7539_4}#M9?sUlV;@bQ
z?RLO^VtCZT4>9n=;C<5bYFiHf^I)gZ+4!El83QDbW4?TP_=ZRv=DABDvkP)@PC5GX
znw%)$&VXR{i0+XvduYqbkso$>S;rp8=eHEz3B>b%43nhlQT~Y8o)5|HtMJ7wd^hCJ
z%S2rEz`S}e-EKn>WPMN`*hKYa`T&TtnCF~l_(|0YVu=PsXB9BLUf2zHr(ATNtj
z_<24^P+Tw*_8jv9zs0J%_)IP!J8|be0o&}E)a3$u?#YC0`}f2D?1IFCC6ln;2tEDe
zP~|PGtGmsk^kDydckI)tNa#0nr6dar=q1kr)392U->+%VV1)Wg+FjB1+mH6W_b#&v
z^{vXCsyzUD?d@9QK2i8%DYw>q^Cx?;RFj<*)*X-hepVIE62PypM1R>uK2shtZQJTY
ze72-XDcm@W&;BXAc%n0b&%FQq(%1sICK;D(<|bhBJ&jtQUjdiSTJj-(0pgOK%DOja
z6rcUlk@ktVfFATJlsV7`d)3Z?mj)LASF>|ASCjZm{o28=rv7|pJ?!d{hE&uq|DwPD
zKH%}EPvU>UEUsg~`F3S*-3sVsF~jP~2C!Q-)ql!DTxJcoo8nRP`D~zUVb9n`
zoHKg4+05=Z&O_MA+hlkGhr~VFF=Q~GMaz!Q8Y{+UY?|)9rf=BS+;}XWUBxOcX`=piwcYWCp|4q4ncDmc~f8V^GCpHOjIi70v
zSfhVR?3NgFWexmRBmVvgyV1dt$DgoM-vGa;uWRLU^wIC97zU0*zml){bN*2X>{Onz
zV@(FakK)*W+cO__Fz@uw$M#?yHJN9gC9?qi^}TWYO!zOPUb~W6?GArKkLEOcP0V}d
zUXGclr{^x!dy1Q|J{%%pr>z1z$`HK^)AMj|L*jqwCcu@(Qg7dE@_@f|*L97z_+L7M
zUdeZ0+?vMlT@10Ge0=kQ;C%Q2=4dn}8*N5>xDoSqch2l!x}HD&+!XI%@<;Q7hCau+
zza%y7UpW6~&XgC7d4m0oC2eh$*dP92_TJ_6><(7$GO9Q!ubu5XF@3z)_I9>>koNs3
zHTajD^?m=Nu7jnN1iYKz)WHU?X&14Azx>SAYo|qxg&(89>`k{H?0)^KFMce9AFzJE
zMGV$^Rbd$^ul2A_uU~Ai5xDjHIp>{~SFtW%v}c0<3E<5)*2brg;{1Ltf6>kO*<3$Svv6=@bZ$85qc8c9V|lh(=Ri%4mL#I^U+*}_-&^TSatSx
zu!0}@FOFef=t9C?gP$0$R|B4$eeDWdm;FcY`FZTWEt_|I0PM0$=4Wb61Rgg!APDn-
zeYIuLi6ei$0?!l;4l_l*=KNm@;lSB9TwY(FBn~{8ICb|v;N%kTaXaTWwX^Ss!VYh|
z*}+Nz+U~)
zvCEX=IhdDZ)psCYH}!y>G2~OiPUY)^4*RTAdmi(8X0eVM`O7VJ+{K-fvvp99UW0sP
zv8sz}P+m;e`}fM=b1F1v9mV?LiN4V|@~ine;(HODGj{A7_*EX`=D4iHp;pX~_WfV|
z9ii{!?afaS?%sF1I?x5>>|dz2#~*Sge;-B6!t6+UWM`VE
zsi8#W=Zaa0TO(ik&CAc;NFRN|dC_^$_}vMYCn5hu{p1X?ul><&k
z+Ic_?>GKTT49H%(ZPmpc1$gd*PW!M1v`0wZ(4r*B9oA&pX2?(5QMHTkc3$7&732@x
zwQyJ1669MrkC;OE*fiOC0^yLhkavSWAV0`w|6$U*BNUeAECdbuYxN)L|0usnx2c?4
z<}W9ay|{8l;TFO%?g>YH3D?(l$$65#ujyRM`6I_KJ+DSMLtgI1a>BzO+G_S@fvyOD
zaE9!7@n7ekBYpl-Udo8_Ew_FjO}IcMwMm-RA>nftT&D5|R37ExTR6*HkRu$h!OR)~
zBoJ4t(BGHrcc&j6oI}*X>7pe)zpnZ*XSa+~xS~aJ18nubQhjw#R$Qn2t%v6TNF%QJ
zh4(AzI|Cb@!9a}j0U96GsQl~WeBz0J)VR@(@?DKZIDXB^-JJdQ`(5L`gg<{4P3WV1
zO_)*UgR^oX#Zq4fWpd|Dfc&$WnIPjdAIdSi*7-FJLEQPH6;hE&fL
zC%2VRzW?!#N}}t>o581JXuwcwAdsl`BUq#E;{pctIR)
zo@y$@_w`JOpLJz~@!|fCQRT{8+fvHWsHuR>sK)_Mu&DRaEke4y6mS&jCl`+9()k;N
z>9bmd@-itxJ_yn4Nbc4$n7Khmg$wyb&cb}>E}{HQRpB}$aI$dSQa4VRZ}&yrKM>%{imp+w?$7kB4{E*V0FO=ovu0;yzX8Xfj
zId>w%Ir$ievm6c|^B%tm^5GlM%qZ9QGky#BR#W^g(6r~PxpmAwM`1k&jS{w-`DScC
zKt64BJ|{O@X$aS^UR@KmbM*;?NkP8bZlV0%S;F<)gEBcTUC-tzS_*o^@C&{i^URKW
zNyHb;zh+AOrmbNG)ZgWmGCBFR>o_|!`vRmlDv|uEtfBnogr+u2!!uI`V*&n+;`Tz-h~Ae=Fw~%rvaUn
ztjWn2t&QXIC1KFUzw`|utl3(An6%C8=w|FP_p2%qzwaA}E?T`gm&*DjPT9kjX`uQ4?
zANXR%C>pOX+)qxT@!i`|pB9DmLqBJ&AwLcS)t+;-pJ$R6x#vqZ?GL5vK;n4#R
z0q`HQ@{iKKg7f~J>&PDQa?q%cRG+;!+$yP_nR~^J-corbDw`=?V~*|^DsN=$Ky#XR
z^@e60CHu{YFq@m?Z)^H*!`tnkx7qf;nOH|%DxP|KB-T+@ZMOdy&ZU_KXig_T^(&>P
zcU8~9cco6cL-Cy-hX1g$g8uIENZXVE{ae*?zn~n{RC#=%BJf0jxR!4e-lJ}~+fX0+
zCL&V%U{Ak?(dHIWl
zhE~7@lInKB)NYSw&RI=%E3?br>^6X(n)mtWF38(7c;8)&`c&60pRR)VB5&rKEKq};
zFygk7{-DLM%k7GOx*7Wv7u1Z+95-NHF!|SjZNM2P{xzRIiv8;sH;<%m{D^+_*S*{E
z2Ii}#ZGCI8&U@XP?fDJ>Aw*
zqZHg+^$GuR)nZXQX{^5rUTkv^1&(qx{}5ddd{uikVfuK;P3ZSZ(89V)B2JJE6V!)qt?BA?jrZ8$T?35ck
zY%Ws#U^AP|?qcv0y17Wcwgvn44bwtgH=&-dd}0^d1814+lFZ|w{vV3uY%@^r+;Ia|
z8REO8B^qYZe%VARX}^O=Px2FSufqFZNZzVW1bq}3r1Kr$XVD;AH8=?RaQWuhFW13u
zFFDHOrV9Muv%-wB-@&eNsM{+j1NyAjJ1!aZoHnc1`004qEnHP{6XACdl$bwlKXBK#
zDmL#6`eUOYrhEbXlz-iv{#y~a&aeKJG>x0eb%%#GqF&1{{ijIxAofW{1XKJ&QFr&<
zKIktZufm_Jg1)$(P&@+ki`mr5C1~fZOV9skg}l}9!o}y&&zy%3juyjn8-C0kpn`st
zBwsx*7XN+kp*IIlV!i19=$MP~VXXU`jJ`iPi1F&*+l`24)#wzB7N<)
zsDsUr)2bhrPx!vqYf;}l@S;&i+$_+qB1OfNBq3MQ|3$(T&$XM&E#~|k9%j^C$G%}_
zPvDU)utQAP(r|7xCchhH6dSQ{3dV){!7pF6ZUsdzUKwkSc)TJ=y>bg
z0`NsMU%eUynv^nNPae{>(>3f{@q9#b(XAs$54o1Vrw)`iBr9ej(qH-96HfvkfBAb9
z-OGBhwe|;{OZYHxg&5&z`+4(Y$j^07vZQ1k#)p5FAdl=s4Mt6=(~v(Np1^$Y8#CNv
zD?s}n^v-pMoXz$H7To02@+o)NAW?ASYk(g()hvLHP&adVpk=~?T!Cohma&ZAm7o%Bhei>(#OjlOYBnevl6
z{|qL*_+DG5nehFcy#-6@JzxKv@gn_|BA-!3cKO~#dS1G-0LVX!$cc17#1vgSd@7@ZRqI_nru#fU1Tn=*QZwD#=caHesCLvL@evrDM
zUrzjAzR4o0zt{elWa6d2E$AiMoMFx3+fT)DaFm5T@9_1Pl^m9c+0Tzq@d6%JQ~iENxl`D%vO0NAijRSJGaauR9I0FGF<
z1)Fd>M~eZ4P|!|k&lig37G{Dv$?cYHX)r8jn(bGZG#L0h@!gNt@>
zIPtr|V-838|5vF;<&P85%_HF=6yd7zd29S<;q#>e4N8mu~v^cytn17@O~BLg+g3k
zq^>2n_XEv{{7HxK+rY=3#Bv6__(m*8K?~q*MKmx*D4#ZsDu{JJPp;_a;?{gzkjv$NYQ*qEdeX-Ljt_Vd$ng(F
z9&r8XoP#3A|Fpoi2i{Y+M;JHW^+yTM
zt&7__mXW;ia^e4-II@?Mw+U3>+GUH$|Abhur&%C&h5%|IlN1
zR2cT8FS5o&xA$W?0{ZFK0`9x6{zJG!@UC^nO;paYAY-ol6#YFTsGiG~DO?gD{d`^K
zW?C1mUo^mu=HUzjb5~2GKa1)5L(lgNT|JlRg5NKs$sV#oV@4V6*W8*r{FVf0;KF6N
z$QS!7s!{eNx6nf6D7`OyO1%`xI~mTuO66YBc#9tjf2*^1w^4rKjsWgF!t1nCoLymZ
z*l>BWr>Lzj;rd1A54bNv{`jjndm;`NfVMK4iC=Xl3O{`rLc*8GmE&Z1X|sN#Li;xy?NQeLB>{
zpW8o+-ECY(?KA3#jY&K5)$Gsz(+0mb`|Kz^(1T$Ye$saxF>rrK->rR0;sm|FHu+KD
zd+^({=SO*7%%4=1*Cu`NF+B;292Aweoh^}zoA9K#R;_L`1@u$4}8}`-@gUV
ztxBj3wT9d?{;bU7pr#jh&7<|?tT~fwDITzc`le87k6F_j_NamG)rxta25P>}QhgZm
zRjZzQ5KhW_b^Uw`>RGSr)*cJI-Ee%Q`b@O*zWG-#oe46WCbhhDWN`+(KAgvZ5@|Kz&x>lx6Vt&uWxPQ@+W*^_8l#zO8~
zb=ezh0pYGxudtycpMJi(cqr=mW0>=eegTos-#48xxlQ}d^f`03@1k>Ju77RQtC0RE+@z1*FMYz@Spn%+)~fQNz$0Mh@O7YDHlJKX
zSoU;{%b>?NibqCX=&wu$FO)Xi5`VFJ>P
z$rH&pB0~nkeyvbI>DRfanqT+
ztu)Wv87YxR^R21Gs>}4;_~1Ly^nWamKIZOgI282Aj2qW2*3j(u{+w$~G@tn`veq92
z`E0A^m!uD7o=LH$=dFT^pHco7tZ8GmKUe}U^>i1uw?x0L8PV>2KV^w;3e{}J!e
z_67>Oi$&||ouT=E?1QR$%Kvsp9f}V9-ZK{lV(5PzKL7&QC7uc642wqX5vBU@G%Emn
z5HC^EHHGqz)rkXPAr1+m&Jx|RVRaG7UA}X$u?1EoxLjkxj&n_MuOw?P>3;+W7&s+~k6nG$cD5tO0V!m>6
z)^Vdb`{7;rZmvEh9k@;z@{hDbIQ_lB+KkiN!%K>|diI3w{6YC~_u@F5Ftu(SmmlAQ
zWe(CaPU~`bWQ5!d02$mTZVPuJ&^ad!Ienj`t_?s2yM;q8ho9~?M05Cm#z`v!E$V;49FHO2@0Jjkms^^0_00G%i>pU?m?M|3bo>rSKirwf
z6N>-KhjE)rZ~Px`^1u4fZUQ~ev7q$T^!)$7=jQ+9nr{o!7la6DP@M3&x^DU)
z{KtF3^i#C$fpS*g7xHe?g#2bBA^$uG-52tQsX0N_==VTghhR4N3+TPX&#e*CchsCn
zzmHKtw3Fa3@(XEs0$#;SNRufU=|}$84%5&ZiKnQbB-cgDTF`)mMvmUTMqiHg(6GB4
zzx=pxy#0A=!=-N&!*Uk#(;xq@KfUbW_(FSW;dA91IGXf?8VLK^TEhNO(N8bNzUpRS
z|H-&n#yvM>iLhVv{zNwNTfYe9XX7>+;$78f0ScV>zjf0~%sRv;Vz#4tXzdWvo_)gp
zc#^KOM824}5(!li$2i{Ru9S8OoIZFz-3b>R
zVO~G9rIymi4x5OEf?Z{j1~+d>1bT;4dM@H!S_gUL{{!u#xngapKGClwRm-F^%xrx;fjA(fgxXJ2*Y!
z`0?a<(qmHTJ7b0R@+N=)5P@$Bz89aATJ;pBZ^^G74S>S#^{+2?#GD|vhYUlXo6
zn(TRlCJScMdA7%wKmVfdiA)L1ruSXy>KjlXzQNCn(1buIu`TPR<
zuKoKWf6=<{t;d5`)NT)p2B^~WLA_>*-AM0$(Ds>le<@c{s=v*oA0G*4EE)E)hj7J{
zAfbj6JX*#26-F11U;xw{zYA%Uf@)f+G
zdddQ_Be!O?X0}#mY>K>bI%H|1h=h2&0`=R6mns{cNI&WtC&;dqzu4#{~?S=emxH6Mnnn
zYxRZbnrVX{QaLtp`(^3*kMC1HQ9DOJS^c64v`Y8DJHnkw+ib=1K~2=GluJMxV|wc8
z`4K)NC(yb)(WbCNhGxc|RvW*fI&zv_HpKeg*j)$QI?Kc9DN
zrxU$eG|Y|axmzluhTgBLyI6Q%NK|GIy{96p@-Wr^a@FfC^dB>WHcz4Fc75rcLGM|2
zYVeDCq)TNUc}jS3{kPLo34bdcia$f=koH>-O`ZUHa(YiL^|Qcp%b(Ety~WErlHy@|
zj#9x$#5wefPiGNM-L`Xi7SYE3SzJF5&FJ9vU4xt+B~m?pt@iV#ew0_)J(KD?R_DDT
z{f8XSR%`mc{@8F_;EZuT;Sp!&*thq_LaHY(=y(OunsCVsqO!&
zKicmow}<}2{h20kAb?%c!ISe#k$(Cwgz9x(MLCAPZ&{4v4I01W7ZoN`x@$+bAC;@3
z|L-xSGn4EXs+Zd2Tem3v`0`jF5q#hIa}Vf0d=6_`Lg|+RXLk^dnyKhT^ttiWgVfI0
z{*$Nv!5`D|oyvRZwd@YPSN!Zk&OYt=VN5&K^T>knFvLNxd@cNs@}(~SX|Ssj)8$bKpm
z1p2Fg0%yNYagPq5{Ph`GS1Emsm`MTAg7+$bCRk^T_2=g8>u#lKG%jX|zE*?qLXp|>6@3Dg0TsbSNa1A2L>-*%u;qsx2Zgvwt^NI_Q3+w_R=>B*v
z;1Z@G(DYomQ6oKh<1J1P*dNN`%2o2fF<9isOcmPeBYS)~{a`z+iqmtJuR=LKXz*Jh
zZeMtcOFwRWjHAQ$F6Q*+NDnXWyAPDF-hUEiS+0f4E2ovmKQ`-+d^T?ptceQIWX>?C`zn1OzeO
z<_K|+bhI#CzYFRQ>5DE3^J|6)^Ap!Y%_DtAH3(?GD4G%U!%88poB2kFd#+J+kRFOr
zLiN49Qn+r2%|mws|2Rk~#no$$
zA`WG>u}p{spDnCTj~uANXTg4I0|RH{T;VLSk9Wh+PNTkOAIpV5ey8j>mur|WE>|et
zgx~YPH4&1vtJ_(jWU_TcA>vDkzKc4u9_L;z>~u_pov^Ri^1?j@KKsvSpo&|26B}aa
zv*t`^8)NlPPm6DYANld^@*m(gt}sCJ(a|Y<=J8p&`J+@Fo0(s3_q3;({Zx1vR(z?G
z-IE>_aqBMp4SwBwUQ~>Ih^`=~(PB+(?8p5}KX$jW#mO3))`lIdEFtaY(ydr8B#$|z
zn+AK%UE{B7Z}8aJ#n*pdmT6}Wzn(Lf;&xWjzOTj`_ibk%-eS7H74ZyIUQc=WyN%Vp
zv(p$lwVnB@cehsHoZX?j*T1iP(811XfBp2h3-L-PT78
zWrK#dvG?T_C60wntgpk^o#DO|?W)-QFN<(5-iocVp%sX?`6?o|>@V&&k-h3WuC$IN
znyAmQ-~E+s)U6q(hq#~atL}=;c5G!I=bdWs$8#xX{-yV%H89=1$I9LPnwU|xgQ-Gd
z3sbS3n{#McE1Tq~WpoO0$bIikR}f`wEIQ5i`ba+R_glB1G^M$fJz&b$2X1Xe{20fB
z>G>@zWmCV+vmuBdc29f1L^rK3Bim!fZ`$`o;GlE_y|n_|V@iY}A9TxxtrPm}7jc{x8=S
z)_FYj!v3-r_C@5+oB0*Z?9s%@y&k6cKBp!*t2NE+${yQ)ZOfbC5A^X+npP`YruDmI
zr%DUU$XfqTVA#wm&3BF7ZPLuXZJ!YG331(9X7*>Q>9n$H?Hyx=?`mOBBHd+nxiqt(
zJN6g8!~HkX>%85i5%1l%x9;%iNv-V62CHKMM_X8F;YeG5+=uh6G5y_t8i>Qr*N!|j
zt(o038ML`KpoMwE-}NLGD6@D2AvJt`e^{2Oy^Em@>%hy0qX
z&J)XUPhsvNrB6q}+nO$l?Q3R{Io4Af#s(+n8&q*(q-+-Ivxs@*Dz~r!GD>r9&u?J~E2fN2M?8Bc
zn|0P#5YP6te`55++b!(In$NKl6!HHK%(>M&vxT{ScK!5Ua0~le`Jo^d`Hm;oUvnsF
zVICiA<*SVmAK(1S!H#7utXXR2F5TD`COB5y6^}Ub$9DQA43lbQBSLTg{$HiJJ%xfk~s4fgQ&W-Y8O
z*5l)#O?a>M;RiTs&h`!ZymATpzlzS(cVBQ%q*LvYvClEi2D|V4V}^S)KNj5R94gbs
z9?DjiM7(TaJ8F&o{6#;Felg*GH~RNi!ThPUpAc`?@lnib#Fa^tPvYOkIGHAASL$rh
z%D$f#(WuF8WxwRLTvGUKOdh>=Hl~ZnG*;<}v-C-@P@y%JI^g}JUugX({cfW`
zt`MW$sOPdJ3Zg?7wXukaQL7K!Zev*m*OD6%XJ@u4@6~Hl+>;qMrXv9LR0;}c8fuF;
zI>&b!>O5>?^G{gjb&le(iThTrJ%qSAx(DBu%t1X3PK2mj4QylgM*43!jB~?RGA6_y
zK>o4|YXdsh;(Yhd@JUgKTewL1*)7eLi0`MP_H)HWoIn0MF=fsG9uth5@_zIz9vd3#
z*%6F!A2>}ru4P{}2e34aPmILmgGBgl*=Eq|;Gj|vaPDWh8HVcEI>pZsn{@AHbklTJm%KXP{
z9y8s%bk8fq9dtO6cXA`-e?HQ5;RoUz`K7f%MSFNG>(xK?&4{=7&U@#J_rYyUwO-+l
z3*rb}^Ya|{195*uc?CaT0yTD^(OeT6%zeZdnFE6L%X*h?@SD3hHACC
ziq$?E6yQHrPChDu^Y}*!jGGp1#CJcxAYX`juNx_9Xo}~w7IaI
zz54N8y+R&4+VoU;3gQQR-nL}VWt^MOw!0G9Y>hZlmhJi>tD&#kmx>0Xoky)0bU5K8
z%Dek}v1AU9MLw>|Px^&(?~?PjeeP&ui+@Xr>_&X)r3XT-0u1r|oXLS35WjN8lXqrc
zW7^me_Euvm;$#LWwRQJjYGb>`DZNZYe@#-owZj_y(A&V`SH5IB>pQpN)FYgmUwC|Z
zSJ0z&mb}+NGy!o3)>UX3X?S+v{|8hKLp^!c6)QS)+L_*c{hW$3h^M3&uYDWm=r=^H
z@GG+HV7pWPWIw%z_iu4E-Ua=i)EPZu_1t#mU2V1c;>C8xuQ_Zg{jHtt9P?Wu6!ETj
zoom~!Bi`2S{OXYkmw4>^&YhQg{_&XcsvW+$hLE3C81Wk4H@x?fpiq&=j66l0O3Zo8
z(#cWrHR4Zc$UIzObqoDE{ECRQM+dv&lb}2IdI$4Tf4<_b1^mHUCs@&WNP8d4KZ;7g
zy@Ds4CJYX2X=f9Xk8FOZ
zjPJ_z#(lya?CKan#dpMSGgaM|{&o-S0I4^gPewe0IV*oOGGD}Rm@ck&hTo3+$-{Jw
z5zneK>h|`5xgG4NG4Fag#?uY^*zCyyjQ@L8V(SqP{J&bOm3MFAp1|QPvWx$=vj=n3
z)_unLuaLKoclKGrUUPn2*{&4$tF<|*&#lF|JCk
z&a<`)#QyoCf87Ux4?~s9HLe4nCjPPUnivGT%$Em)-#^2CYUx2fe+qEt(h=n%`(X!4
z`|-5;9`;}RUaio$g}9?@Wu_|(!@AsD|73bP;u0puEA)@)VyRE+v`rQP*WRu8)cmuZ
znUzPJOf#o`T32*<4jVo2RkB%0-LM^go2}evWgJy4n(c
zijZ5=Ci?AlA?(E^m3tgsAwELE#VKl^v2HY9`oZTf;wpu_wG=}<`O6lg=eRG%ep_>R
zY|K`)V{=aXS+rlfOmAQIUBuUUdidbfH;B)0|6Y^s6Rbn)rq5RcPG9}XIqQ=l>}WT4
z?!USa`{p}_4-Cb*+0lhj{9#sD4{Y0Sr4`I)>tnx2c;Q@U>egIe{pUEJSiCZ7^mxQ0
z(QbTNV~%x+_M?DUU#v5ZKFK|v$!DP}2YR1*iSwM6C#JZI3fRlky??TBu5+g79!1e5
zh)WV2_V66egxO?T2
za{{03>M4@TyMgnt>2m3B;3s|I;FpaXaegv(!HK=mcaVSt+t^7`IS#K
z4bQ0w*c$2i_q!bsS4OY5O9pXb+6N48wfV+pCtx>&
zUZ$*f5wOlNbH`fc@>xpMqF4Vu^4X9xzt@`TVcn+uH1{#`ZL8PZn&W}^e>YF1E6?Dw
zp5sO~!#DBS&ye0tQbj1&X8OuJtoKe${FD;d%x5REZ6_@aMEzrm#4jBYu;Ge%TaOkA
zSh~)NO)fVD%u_VH_#oDQyJs%+d+aA*%sbS%J3_#UTEE?PzbjxD7M%)?>JYF?r`~I(
zH4B*0|FQSx;aETK->{`9`z|6PM5U57%TTgM_E2`AvXnxW(n7RQDydLeitLeWS<90B
zx{!(%6-p>urEs4!m*0IK_kH~Cf1cxcp1;11$+b2JEs74
zjK8nEJ?T4%lAql${GBw3D8W%Kp|2*<+Z+eYTVqh)aCfQakx7&o6gF3S3d-$zc47sz
z=kw9n={0;GScaOvOPkpinognB=KcP9@HuJ!-&1e%;ddMKE^Oyn
zE<1(lx#1I4E>kFP`1WA#$th&@b~OhxeBb1zR@br(!c!>DM7%6r8Gi5KMN1u>)fA#c
zKJI6C+N+O23V4!!{Wj{SP*NhADjR^r9X;IEUYWXH|&3}~-q
z)pYkbe1CO#(+Vx>
zgDLc7UHONWt|`ROu7=_fKmEqabTyV^SU$eeUpVA$hQ!FKmJPt
zwMC(73hA%b+_B4R3T?chl+Nilg`V&K(E2ZO3TaL8%Zx(3tEzI|w3SXFw(SC|zQXrf
z+BjQ{mw^A)s&Lkbt(ii%IZ}j9$xk6Y+p}T}3RCD{bgc*@l=~UGEg=Z(kbjq+AM>mu}=sZJ9?cfK|Kon9QIji(`X>)vrL4|G-AvR!
zez0Yg?+OFIu_>-RSLi&AR^{iq8MsZOB+nCH6`&t8>f93_C8v>peOKh+UDN2sp{a9*
zDbt8Ll6&{&S@<2cPkHmc)M@0n{GD^in`x9-zl=FBavB{QY{}xVh2NQ-o9L=5n?~Zd
z>&|Rrm_fr$shi}Qr{VWKdCeA1K)#TR?7oO;RMAuXPxj$7DzFxaugZb^nl-t&ogb%n
z@HJFVp`ve5PgDL(p$-s?anVu4LwmF`Y(*c0I8dK29S#
zdB5Nmg=xeQsUkWC^ZH}5wYJLHDP+#;x1E_~1_^KD8KM6)jgBw!n(k4VMm8Ct&eh`j?QYG11Jb&pS@U6GYdzA!E;ZT1)a0RJ(S(hDxaI$&O`rChx^
zjnq4)Ief#QpKVpW9WWn$d>r1z4fC?}OW&Qjj45Pu>Lj-X(=`0<<$<1eVE0gVL%2r_
ze1H4wp8XDorcl+w)YOTC@I6~Exrg1%U>>hg+wkk?6v_(K9S@;SBD%lwl%V)Y_}tqW
z3+o;5y{{Y>a*Xn)P*088lR}uE#qo4V2-e@`3&jSX)8Knqx%i6LpM$tJC-3h|dDG{H9;FPa;(=&J}z9LHuM?mmQ0Td7eI^%%=jsXSd_Uu7yXF
z$XoBJr9=BqL}&BDLf{j;w_cjR&kptlg+s~3vJfw~yQcMj{_zvBE%#7rQ~3$c&zps>
zoQBWE(~%2Ex`+;Z&qzn7P5%GC{>MZ|d*gtfi=H*m<*2ud>tT-oCxsF6*dEmA>6qa^
z@()so{=f9`#|YssiB|l7A|vGE9By$$M_v_x78P0P6c_RJF8nQrRGiPiz(6Wku~Z);
zeGU9R2mbr;oWRomc$R9xdEa>Md9cxmlN?4;6V&DF;^RTb3!TLG;Ita(=mO!P^&X=4
z|MveFmh7d2r{4ejAFZYT-k;mV+m;OZ|IeTI@QX(9_y5$N*q8t1Pk;FL|2O{p@BOuv
zynT+X|L^k^{o8TofBydCf&X~mKOXpx2mY@fP}zq6s7Q4Q<(CjGq45%`FQMlDff`G)
z+7c=+q1OMAuCyf2DByokk@S*mwuJmk=(mKmiufN?G`b}7E}`8LhA-jbX8aE-dc9Py
z{!;q7C5&FefTjH3OLEJSJiH_uZNdMbqL58A+_aScc`2W1N$w&JqoUnAX?k@_`4_j+
zWG~_XDvFh($rdBHAoO>sKWa<$oL{OZdl^kn)0u`nVl+HN98N`Bmg?ar9iyUF(jh9k
z=T6h}SWm;X+i57dR8Ep~kcv|O(%L^y2AGN-V>PfxoSTMhOXXcTXtF~v4aWy)<#NTa
z;Z*eJ3a)^P@CpsN*3oeB#73gm
z^MNK0*UkvF_$VqTIX%93+MZd#W6L};R8%{;DM`Q_}udO8c7T;5#Au4=cjg_RcoY_UN
z<)NVkQi
zmv#ClNgr6JN!qnMX+iR{b~O@gT$xPr4`e+g`O$~h!H82)P0mYLl2mkb^8z>o#$~+*
zIDm?lJJrFArXo&jn>hu*1iv97U)=J9)b~XQ4=|)(dJapTii~+P*e)TZogcRXHrEL8;`9n0MGuTuVrPeOzKQz{2t`yhdH(bh
zRY1SkTP#>+=zMt;^d*83j+
z&;_U87LC(a2E6~(lN+ZmuRQw)<1e_s$N8SyYo)P#CeAx!BVgMJk!f6yplXbf5}^2F
z#_BDAW~tp77^`$;nRfwl##P?l3AoyAh@M(%W%8
zzquL>aJ{@LM}OdY8b_O$aQng|8Ou}vPgQ-1#Cqe`dg*cg&W4Me*sdt6+2h**_fRBR
zasGk2FzyY2FhWTsQW!?@p?{SNCEpg#$#j%*eH`SjrNc}>6$q2zpA&f#JpY%$QE
zA9HLNC%x&KaQfv=9MGt!;5SbLmbo2x%CY>I@=6t>M&LjNP9Ho{(Sfn){tP!xw@w?|
zf#p>v6DF|CVtRKI_ESZd)p;zd*S(0q;~=KNErk2mF|zTIJm9*sToJfGn;-VNWBql1
zr>wBvRO$8o*xvSD=~S$@TH^vE9(ODMI0y-9g{gD|c-K@VFk>?4YeYcclv
ztb?tZif*xUCt$l$%}hwY*q&z`!0ArX?*uVYmD^!~!~B0fdLN^Z+TXWW|3h*pJ(lSY
ziTYs_qrjCOU~Hd&H@06z|E(g%qnAqGV}0Shx2|CHVxA)Fxb#Eo$a;`9b^nOsav>MJ
z_u%o>*uZ`a_jkR(4;zds?|zf`7V-5B@$Ue8I*BvsIflgE*kDT{r`Fn#xN)Iqkfghs
zN0a$?v(=ZxopY1DB;8nW))cqL@z{yPv5eoGB<`7{9)TU3iZ&)?U-4o&0!Ipw{L!TM_%T3};=b(DAvb~-BBM=c}e
z+b;zZ+;_#Miq?%S=xwbnr0Q+{HF4;Hk
zYU>jWbQ>f7r85N;PT7CDOX!_UE!8}MWjisd@`{hLB
z*K%$I&o0pH<=#l+5!yDe{eiv>ZWiz^2EePZYhnyt;%(4*BIjc_Li(r$4L8H+V|f}k
z3#g|}L#ZX6wfh^*E(tSQdI)Y7^ghBY#~8nYX3yo9G?@W*Bb>exW(~%TdNfSGNShby
z+-UkEax}eTu!~{64X(8Qo?o(q2WLZi90ovvCFEYhD};yxk6F^sUecdk%7@?te^ey0
z)Zac>ZCGAhO7~qtZsHg!I=PaTuPI1FdUYE9!tRA~kC*(r+ni?S;F8|FI4$4IfQCA2
zXgGiY$`Lg{GMxZ1+rsJDOZ6UHlING~Nm!DVm*gKy?Fuc)f0yztmgI{|_KPo#ThdZL
z&n?w&j@=IJbuaOM%0FBXc*IiqvrBq0OXc>jqUpB_(=ZHo7xc@Q^oN()d&Wsi*9)NG
z9Xza{r?iB@OMaim(-zXDSJ3=s4$%Y4vzR6WZqlV;HE{s&6e~QSus`jlr85}OPv2xQ-o1
zMMYHXP*^{oKN7vkZ9`B270G5|$3g#>__|F37m?i^I|#;0%Y)%m)Yr*Mbh0;ovp
zPdlm4Y$ZME-=Mk$(aX66bP)1mz3@pDc;9+O(tfUV8VnfhUm9=7eyi+jB6z0p7z#cRJwVPuk*m%#hp*quBo#=y&X!-9ET>`d&~oco6Gj(z`ZV5F#M
z=K3rQ2o?Fai@|`yyq!M;R{^j;DE7Ex`E!c?SrUPIgBGte561B)Ro~R
zN=3V*U`+=mRHA;1#DB{YgD77F8H|Z~okk
z%l&HFE`s^ty)L1|J%H`Ef8<~uaw7LbBHq_2&*)ll{{D%&K6Su}`j~iLzyi_X2yH;a
zu^%p&f1!XHKa6&+=a>xudFIjUTH9JEr*+As9`kX9ruuPrwQviEsI@wlzv
z=QU-uxL-9z4_5Umfq#s*Bo9G;o1mE0*A2iEujIE0f!vgC?U&Av;AHm?+|U%u?3VErMI64ea?5bT_PYO#ZePH9?|yq3~@iWgS6Lw0lYeA@y!s@
zU-YLR!}CsZm!2m#;LmNmi$#F@1UPb%0Hf%dx9H<^oA9+3fT_*NTm1pW`kO!B0(AF#
z%Z2Azg4(V67^wH@n;F$sbxf0b@ja
zjZcCd`?)10u{>zuso@Ifc6+gC0@96E?aBKM$iH(7U)g4u7mkKr${_1LKKncs`r*=B
zCw>g{hwDw+-$QzegV=;D$Wh|MJ$N1VJF?x!>&R;~6t2+V9LgZtj?<0Ls@viIY>ja&
z!~R?M&IkfH%(II20epUFY?37Bj@Xs4hfF}Wwp?!pAI?Lkb;}j7Kbi~WC2>FZJG?Q%
zae=-~d09WSFa3+V9=BiGtlEqHQXyy2jP*7ylXSxV+@izPmJNDqTFYGW0FP}_E5!Ac
z@JI0AbGhmGooi}J&x0-mQP>8
z_K(OK!UhHW<=BEL#zpCmBp$Zco|VP@DzATW8{2F6ag@Zvo3agpSTCu}mh20L$JjgY
zxKh_|Jc#WcQ696x`YsRSS7SZP^qb2u@^KWH;Bg#F%p>~Cm@0y>e#rDjUfj;wAA!V9
z@3jUb4(G6nlKrUtmpa)$w3n?TakazCoTLjIbpsQDdl}|AVz0z+IOc)8dKGOSq!-2n
z9@f9)Z)1!_0pdI_8q@JwRxdde{^j{V1Yo|KmXQ5MB+QrWOQlRS
zy~V?K8FBusL=`zlSq*r@0f~yvMR>x-4d?iZ3R2Em3${U!l`3F+0F-IQM2w0QgE29O
z_1Xd38X)D2Awgxezr+s4Qs^$E$JoR61=yuV>xcFkOr)UPAWSpBTwB;i0RNn)-S55l
zU_s8#w`k~>OT&jkG|a{3LVo-b
zZ^fr+EN7Af3ETsmqvg-?(9nOW+``g*?zS>o`+D;9*iGQ1q(d0j{
z>T&w?lD-MdVk|en>cuG0M6)Ad>3kuzJv93+q|;=+W?DX4^3SEE
z`4hNao0Rj|O6z~^51RhoCI1e>trD)+=M$}bv_Wp&&#HzJCK7R#J$`#B;6jO
z63$ncreUN9P2YVd&HkGY?MVK!qqKGTXAf=s#xd=Law|?kI0Fn_`xOepy;mk}oo04p
z$5IjdM;h;r7{CW-*x!o!pkgZeI>15lZymsnhW+&d?Huu8fS1V3{7ldQ_~`^~{RZB*
zCh2hwD+$Ji#}a&{{tP+}=ZwfQBH!S`69eYM?g%0exW6N^R(Bd%uPj#YN&D(=Q%OFz
zBF!$-m2o7!@<%!>5V&tW$g@h{R1zOE!`|nDd9kmxwtzV$ai9F;eeY`wY_0vGeJRIF6>C4kP)FcgINn
zLfb}CexYm;Rv!52vVIiCx^roU7~dT_4}k#I^X@Fdqh-Dyior75_+CDY)1@P1e_)XK
z&Vc1lI`_|D+~H`@OX9ZXx+{2J{4=`;j=+$wWod%*pRc+CD-q7+EWu6~FRWBJhxeP_
zxO1{%)htPfelUOfi
z;*1b(m)V!;JKoRcb2ON6`WA*XEnIH7OU7=T&v$i$C2p_&?$_6tN60N}f|D$q-z5c<
z@VS+SE1&b@bS;A!gNuN<
z=a^j}oABU{w0H|lv38Uz*D}mSAca}2#rl4MZ$H8#m#g9C{xgpuF-)ZRl-w-=TU7@Ox*|p--tWzDmV^nzsYCyDcx;DG29b%j>s~iokt5
z$7KfvC};S*wPn8r+!w4YSZD7DewX;V{s7onlPtYyC)jDMyT2;|+WoEOC4lE;&e}-<
zdcZqstJa%=U9yS%zd}HE$P{?u4Ekzs0uO3~zDVWC*Nc#@(fCb72kh*Q>OI;4^2(I&
zUHCj5L19l{PCo}<8mmUq5
zfqKGYR$Q?IyuPYc2lM;iksj-uz+Qb*wu-$Vj~tk1!uF{c*HCa=p_<(Y!RPsw*!{=t
zA)V`Bc;p9=Ycp1DOaa{8r+Jnb=xO?6>yvS8KtdwoRbq)Yym7pe|ffp{*&{O
z=J?z)#dZD;UcbrO+nRBl5%CGt$8o3Sf%|)WJ~B&@|B26ek*5?%{2Ox#uEqU`{e1>D
zT6jLmc=8xG;Bp4KWE{`ds?n9=_~)hZ>KfM5=6W%J^;$VT>azfTmdHB933$^&ssP_#
zs8z=T5d^zb^x^6m&eKVPXE%YJJ}Hr>4}e@dcG3p>xn=6eDtw-ssV^#)!Z_f_jLW4R
zSObI{{I1hY_MMwYGs%9ryj_6oYaxmB!C23KeV`vk)_w5U2=Mb$eR59PT477>r`RV#
zZs2sYGi%8>ncVjWq5$*eMAlx6D~;%RF(&*`|BC(eIOh8%9N(oS!}s9!ebm?P!1>)F
z<$p0QWW3T{2Pk!H`Z@N;QyX;&tiQ1C@EYvz=t{{qSZ{;R8nU0O_+*j&T6Q{t*f$t%
zO7`jUG}xvwS|H*0X%yB#hA|CW4IXn(-dOUrx7uKZNbxILyBf>lYWi
z2f`2gfQ>EDOUiCIg3Gaa(s%}kJMBEMJaiwidms%HRw{bgN;`KvRG%gG9=izFdyxL`
zssPEq@y3DR=p|l)8ShLYRKkqOM
z3#5NIQAF%4ioZbQlCNM6q|;U4gAm;F`~Y_YZhKQe#-(x(?cSPO>M2Pd-0+2z_f4Ul
zxA5s1>rVyI@NEvwexLI+`+xVrw1@nR1X}(xMjCGOqTy^U4Snfk;UB;QqJb6yMq-)^
z_yP|XU~MGL&TnsN{r{{_liRP;xkqv?+w2k%pyrPcfFAnm-p!-95Bei&g+^u}&-6FtHBodgZ%X!DM3FRgrN
zT`-wPyAq8_e(GIVc2Mu%`?Pb3l=EvMXF{~X=}dEHNxkn4c!_*{EQ4U5tSLAG?)&~!
zfMdXa?ytcxxZheaNb28Sl1uVMygUhR(mia6^YwMvNqN7bQ$+8*WCvM?10@w?eHOd1
znPa_mi+{-c4chmbls}Z`N%WcT;0G$O{{8N8V+=TR2qqHnm&0NtpP^us;Hb5|I8Kj`
z<0tuR&x;bdruQ3(A1(2AFr#5Ut=Jld(c)W4Azs%?C%3v`dECVc90KPnCIJ!`BW@P;
zVfo9qSOXlNpF9hJ1qkE#YJ(s~y_tDGj5TqJa3cWsCK=MZaD3BuJoyUCGNQ&7IG-(7
ziP*oVAcMTOu|m1@Ax?j3C`;n2MyCjg(*=TFf4LG
zXBFU>lGY<^XU>LPS$_z=PWswbufc)KKQ}J!fiI8@~(iHXSST+M|#@B5Z4Y0dT5wF)|eLUkWAMV}$(t
zAH^l{u-~W6$p{!h{*`wV=>mX9)c-QzbAgq>T9qu&w@Oc8YXaWz_>yoB1>~EWeiPD=
zU!Hq!4dx+Psz=SSz5E-w?972DO}Y1fZ2>){t9v~1A$^A05W52K-5(oie827>B*OR_
zczd62<_B+Rf4Q(w^PQ~0$=iv_1=}v2G6bbW-1$keuYPAmX!?n
zVe#0LbJ{Q-*9;3AB|yHpyR`Ty1u-)vHXQ|ieZ}C{bS3cR!=)!!5~01ylV?~;d4QM9
zoWK8&748+9Nr58`i-*~}_un3w4N
z1Ot@!J0YnjxfjMaoXuyu0~ICeY_dD{vkUFrvUSRmLO~)fH@=(i27S6K3*9@w{`NIl
zN{2d7OASY6iuFq*l%RRp&0KfOqs)&3J
z=aR^YQiP4}xHvtO||N
z3}BzXbMM~P?{KbX9w^@J+>MeYe52k*bR)(Z|4kdryU`Z9$-`l+-SE3shfBUW^q`#U
zR+m@u_M#2uhj{zW_oBh{Qw1_td*FO2bg=wMH#)PT{-|Sr58B3FSHYmxhhi%GnMbo<
zBhkQq?rpt2DAPw*GX636CG#$e+Ci{`T639B1I`buwv_^Kj@`vD6v6=Wr@vL`Q5BpE
zetK4?D40S&IgL_+K&Dy;uhD|}+We{T*&NJk83v}Oo(SGQjH6!4W$#9*A3mvnNrZJ_
zJ5aUwmVy%Y&1@5Eq@ehY7uK65DJaBpa4;nsbLguCE(Mm9`3^L3#qona|q
zSZ|$@s}}l!*LPpJbhr;NS-&fq0qpbL`p;DXuu&#F_V7+vw=K;7)S-Wzf|~>Jx%~E>
zj4Ss5i==Nv7s5K(V4c5M`Hh0+7}+Vp>mi&vil+ejX{o>d|O}zc6XqlCSy2P^FI|9NQL^39#4R6>7urE+a?R1IL@Qaa^2uU(g-st;G8gd$|s|
z51CO=qH#a3ZFSSc=gSGPD6I;x)7L)x3hr0zK8?Hh{Hfsf@Fg4I=|I^XSns>;
zUyWNZ&sUt5>;4A%ZR@|gk3;&gy0o!x;1~G~qXC0}nMb~8V*B5DwkczD9Xv0GUkY1TpJMwrM3f{T!0Y!lwQ&98htI9Xb_9IryM+B|CZ1~h
z6YN}AxA0^)$Z@wg8nFL5W)FpJ0DL(k?S=a*q`xHq*PlG(PV7FTqfgE|PoLr!w%}f?
z?H}P=t!KgyV)?6Epb5t7o^FIEDa!)g0GZL;jOaT@%D7@XcZ=O1eCfl7DSe!ue=k)U
zeeK7$1kF72tF9;&t(#xPSA?
zPZh8}w}WUU&Tn;pVT1A47kEt*a9@HfssG6J2ZT39xX%&!pPVF-m7L3ny{6LZN&4HZ
zG@iai1!x?UXW8|Mq(ASsR>iWmO$y;(@elBYJMg4tN21peRzh&i%Oi9+{X*tLa*jF_
z5JUVBX&g=bW_r7f=sjHbOc3kw?c2+WQTdw%;fZ=1g-O1$qz5VAl>)RD@`p0Oy?_Eo
zXna3k4rndN3RcX-j+Qx^-R0cSZAiaAt3yz}ev0JZlU5_>^>>zF8v`sG&{J()NpQzS
zHG*BA4iJ0GRG5f-fQxpHVpPl^<$_&T5r566c)>vf=3QwgN#Du=G#F&PG1_yr5q?F&
zqeVRb5`XRFS0Vl15jI8AzdblWc)63<8&ZCJEe{#D`DOS42=K3wZ$xjs?|xD)J(LL?
z0q3XwANw$}{5(SJ{N?kP%v1I_H_}ela${nb;7D2>o)?3obtIq1&`A!LbFv(QCqBR{
zex(1x`O6zVjAGtfnk-BFEmku}+Pf<0N5F6ldA)pu
z!?-74?JH|?9*{iQJcsLHBWnj&XL!AH1(9qc-=9uZd^b%5Wy_V_3x;K5A>?2*Z^PyM_#@+utiwZ2RXB!WKMW#&9P(CfGVeEJRW
zO*eCjh_c-Xl{q2y9nZ(xnaPL)cKhqJ~J#_RzZ5Hf@m!&T+q^^bM$z!4S<1Jxa
z6yw5w%Xgz@PU(6XRoE}r%ui=PJMYC$s=Nb!UtMB;f=w8Bc2L$tM+60#&>OAJ1Ri-r
zOTY{FN8!Zv_$uI22h3O`3Wq2t-JI>O(>MifsC)E@O90;IEwQ=7;t%)fH|P{h1}R98
zN2mEa@H5rLu&)|FdQrOI=Emp1!%o+%tQCj*|3$y_$e(ZE{+|*k*__dh)c$syX0Yi%
zE~3r}+J)|%Ht*UG
z{IuV|x?yZJU8P>4
ziKN8KHjf)nX1T;&jR{sFYnje+u^Le*y9=B$+d-9z2$gk%pS|*p9*|HFAoe8UT
zPAEouiX8mU9jQSFqcr!73)Ulx8zKE17ES2m<lw%bmdT28MikD8HkPPn8P_Hg$?M6aR4bL@N@Kdm>u9r5iE5ij3R|
zL-w0KTCedMDTmo}X1`aQZK_qQ47ASgyQj
zK^vJ1l-#5`k>feW+4gr76kHZ`RQ(1#mki!u{+PKLmG*w*3h{41Cj*3oPFuc2a@&*j
zD^$9W!}|28geeMgQ7q9O{?~x6Oz%6t*RLKGO0Jvv@v{M~?egn&p6Nga%|7;%A{68l
zO+Cp5&k=4vuS~4~J}GfO<~`}d2
z+OHH?Pl9VWcimHmbs14nc?;INj_GK@sYWr6C7U`)$d~N1*qIa_BA(%+pqi
zH+?Jg%l}+$$7|s6pUdgKO+tLw$I{iq0qB|&77l;l2AJ)TgoPr5S!t
z@4NMydIPXWAXX(6kH_AB|9;2;dL3bQz;Y&EU9mgp=|5o%-;C(Rl?MsVPzz^BSv>t#ylf9P`-j6hPH-E=G{%lLl1AOl^eCLW;HrR7MJ;uHS
zIc@YtvO@Cgdw!3rKvG0$cXoEO0R@N$nctFa!Z&ZbH1r-%Dp
zTh;(d9t&Z^daQR_>G8gr^ey!m-hUaQ!_Qzln-3p4iR%kJUeJd96`QuT2HT+)t0s!;
zO~_*bf(Yl?oPC5Rztr1H_$vEhBf_K1x!OrSMffbKpWSzu@Ej9%JM!EjDxg~nmv=h4
zmGB^rToD#*M^H0IEl!ubTE~M?=gtfeMX+mD%K)SB@rKjb&R7-p16Vc-%p?9%Ww#?d
zcx6%m5KVZ$%-!}b)>~t_ZZGy%&WCfSaJmNH4ql995p{W?
zMY0pgIjc+G${6Q=oc}=TO>(?P{L=r9Rxj7Zd%K8zqjA01?n1?4(tjO0xb6d)ZqFqm
zyIip*n0Y`s2j;oF6rRm+(d9FLm6woK>1Q4{E{lK!-hfIKP$oHhGU_L2)D$>)Ac3%fWn6
zs$lFm&X>?%AbfJ&;zA6bZ`V%7kms46@e`@~APa4ZzKrXeEWb|bzp{B3KbA!`H+bWD
z-CWGK0hjCepgw}_Od9I6#plgdvGp-{9pu`KwP4v~%rGee@CLhx5niVPesj<8{cq5H
zT_Yhtp^x$>VgZ@$er(={;~IbRO1$3Ju3FB8*DuQzr_=bJSoBKMyKF$7k7i~U0UbV%
zn6Lmod%A;%2hd`R2*XuCsmtSE{UQGKxryuMLwp;HLM!om)2drv(P2L+-|2V8=jxcF
znx$M2Kc+pi5>lYuy{)RHQIMYVNF)@mcV3MdSIlSkW~48uK-?MezkawJcwztP%BfDs
z|6KWM(E!@9(&R4IhWsrf$Ls|luIF4o^4pUM-UIo~LLUa@RHZbl}q`sPC<~ta&W(Ww{i(>mPj}{*I!<@az!%F4Ib=gmz8uzSv|6_mrD2
z^GJ@`zx^Z)(DuaOo}2{=x_Q-rE7KO<5Bi%rAtD9$8HpAvay%(0
zM>H=m`d~L2_kKOP3F^5jzky5k5b)&ag0~Ohe(bS(+Q8FU3c9KwZy&ez6$+vTom2_x
zMi<+Bi$8Eskdm#WXrDwkdg8k_)7hAU&hPywaHJ^%>7R*|wS#`BM6fP@&q+bw`D->d
z|LaE4M*|aBni`OGdbtD7rqt@+%a5*PsOUr|*>j8@9qvKp%Jm-KMLW<>{cMebv|A|r
z86Ta%t8&Et!*Zrbs|SgV^|;LS!+j}yS(-&?B@(Ppa@Oy>K<8D}5~2G0jCuRcZ|5-a7EKUAZP1A@(c-1X>_
z`p)cRp9a+L{({M<pZtc=Yu#
zT0Wh>_2K1WlwUk~_t^5MXov5QhyNlg(LH9#4L6QeBkjTWJU{N`p%Cx$8>p{~(bws<
zYZ(4KMuJYu`xAebBBkT;&+JUgQLOba60Pk)RgRs>wm-Vi(8D;r2lw02sy9uS2e-XM
zKLc%s4EHpki8fodWvN~8doVt|4|a7Rw|{psN8MUc{JRv*S?OlP$9%b4Tl57oE$LHv
z&eMdVR4Vxt>>Clw{iKhd!4EB*4f4j{Umy`_&YTxU^=Nm9&@TPk&r!ZNLnfnQJ^U`+
zr-s1?^(bBB=|)MOI#lxHVwL9F8u;Ay9aOzokC<&vwSMWULvK@9Qd`99P=!j8px@yd
z`2KP8uhyR{PQkpv7_vNu{P2g!m>dc9P%&WW5*2Q$!oQ3B|ye?U#Rlbagy^3X?q}?2Lx%9Bbe^{2XF9ojo{F{MKdS0tfcnR@q
z`lRUeIK*48jV6chKs-&4c<9qKM?uv_kF9qAj~%IrthR#la&>qbix%+Z{?}I$@VUHs
zL!P__#LJ|J^C=Es@8@*4bQg%TLGe!vq#<4!{#)kd4ZN1i+p}l$ao{V>7k%cT{Z?s-
zTOZ+E9?QIe@1iT6Fn)i@RYj%>--~Q;s<93Lbn`t`b`a7nET5k61HJtU
zJ}>e9B+MWDZ4T>~FALibGWY$cs2k9K%TB{8Q|OoN@cg|6NOx}ue7Oqnx8eKcJ)p15
z@scGB+8t=tSKkNr+2#4j<9ia<08e{)sDIT97tIooAFK}14+T3L&!&5=fqf-GjlNM6
z(#wmtj|PCAXa40T{Qgu;K*k)l)AV-LCd?=8!@G`P{1y6L5A(@eLR3Y(PiHs^^7{a~
zGI+S){V$)xPzCGVS3A>-_eB|&ZG`WJ+@fEJd3A~2pWS%>{Z_T!TO08A%Z+9Dp6K`&
zxjUE#L>IP2>;}CR4s&ljxu6zbR2WUu->^=qg+d~858v$$6
z4-J_^`t`W60?bR?Yfj9G0)D!D2Z$WZ+p)z!Tz?!_R3PRl>oq&bz1;CP-0QI%xKk_$
zqe=6cKUmMFPgo4gg_~;Nf)wt}R71f0(iuL
zHQ~X!X`cP~Ig3Tv@N29$HpAVzAw2%`Sv7H7|8IqGGmQJ6)e}BxYh1{M+xNFRN%*OzoGam-
zcg3X0{cC+u$WvU8iZ^u|*5g0oNbcz?Ti?jzdQPl=OWNhV_NEW(sUF=*c>9TSn((3#
z%yV0rRIDeO^Nfr`H*WyKKk
z5nw~MiVW5h5x;dD`>&g=Tz-AB76%@W2d!Lsc%BZg
za5mEc%pdeLKM81iLzffRI<0Ut(dIp+etDR^aN0eCy;QUbn5{Tw#Qik~wD_9zNA0WyZhy-tAn
z`-jEO72Bz`vWcG=aNypF57vOpRA0VAz{|dr=R&Zq-8N@inu9E%{_6W{D0g2yvZn@Q
zrfMeLwIEAe5%c7R`$@J@6^SX3=RAX*I3R!7`SrW-z3BHLJyrbv@rpOzNAn?G8hCRL
zb%NeGN;P#C#22;;5mVX_?}V0zo;HK~kV)ygkzEX6ueKD$8}iGBsxDOlK0Z~_a~$+~
z_1z9AL%b}o=aU}cg!k2i{wgH_e;gX#%i0I=L_)Q&R~+KBs{8sReTbK;MJIRLnBe=@
z3#Hzm*I1`2zY5~_*s~9Oub@4SyUWF#p=
z0^WEyU-qK_>_^;rcSh!b$EuEJRhhwkL?EaBdKn*BH^N;e>%UWwu%DW0&ld`6%{`dD
zUi=k$G`M}*5Zbjt`*K18VcZJ)v;E;Ya>LNyTWMjiF2|M`W^eC9RcpR?@3{oLU(b3p
ze?ROSiB~`O!+Qb`w3zC{le^LOHE$WV>GYyK289>8t-4Wlpj%MM1qvG3?rRnvY78{5anj
z=OYP!ZO$OB9CT~K$;r_!8~yXUyK8i)04ZMAI+(m68!4UmDR*;k4(y-lS_Y?{q4~WN
zopTnCQUCVqm-MAGQB3_ue;xm4s84Xki7dYtD77~^q*kUFd6zzL6v)U&+H{LUs?Q#x
zOVi(_MW;&9`tu)^3w`Sl^Gx`|H|ja)aGR~3C_9X>eGr7p?mQG`-PXX#tapj9(%0+aI%N~gx%|S7ge5TcZ3z2xxLTPSwK2j3j
zrWUY1A60n&(AM|KLw}Cla#8x1gZ%o-_|SnoR25|2sFDU>A`_pCMpM~1Cn&x1a|-psbL1k!
z`~J5}2XYz|v^YNBg-(TT?*E$s`?c4bX
zC}!C8!abTWOVX?X#8bwnM;`M)Jl^~H)n?EeZFOt6f^${~?~4rYfXb2Ke2-O-J|f13%L^
z+f#lJcv;(g|4sguH18{j3AdJk2>j=YZK!~wA1wRB|pSq%7o!LF
ziQ78A(*rgZ>At)N?W-B5bCp89=ReM7rYgg|+Ojiz_TX0~t6S5f;HTTWte-VP`m^e7
zQ@bJFn(BN%?G1LiK61F>19tNCmC4}GGqWDjjEk-_WPpx?^q5ON0On_r&!U><5dA=nGsBEzJA7hgq;}
z^pI;ceqX|7RgeVUS3ijJq~m=!f6K-&yk8$G)_9526WP86;r9;$M&pL@d*@^H502sX
z@_LPvuwPieDrDe#Uo>A;#`nUnt#ur5`;|My*W-P-`_Q!)xSrC3e%^S0f5<$MfbEq1
zq#eHrP@DeMU7XIXEoX=I5Bxlrg4@4UuT1z7Lj^zKFAj2>3Ge&u
z+>iU0YT!j=xObe3C44MHP@dd({xV3&#CB!ry(I64aK;@V{B@^n%ynE(7Vi!=jCuhF
zC$Qe(po6;jemc}Z>J#?2^5cRUoFDe;9(j*6ay#8c%(sVFOGEH|S>;%V6xKVq$xH<6
zaT`<`U<{S(*T&;=vA2FNC*ZMn#&Nj*^h-|$aJsfoiVj;cll!OSLrw!G~3iv$2sgArdV4i0Q
z9{W)P{386@?n70OAG^9j9Dn}iSj{;5U0DC+%pt4)LL4cO)_lGic+RQT3EqcLo>G>U
z)ChXcTAG&nP)~D~s-x=$STDAndASe|6E<%@I0*4-<^bo;yI_aIc8Pm-a35whv9oy6
z3F53NolyM@#M$9)RbGgTwlZ&bc0!ym`*6RX73$qtcXnAKyicbp-j{v|+V$GqXmlR>
z-LOWcjvsiJiYJ5XCL7>8&YC;E!RMKjp72-K?CnBl{$0++
zIjV;0`dTjF1FtCR(-zX*=*F*z-W6+Mo^B1)oU(`KJC!bPFEGP>+*s70EAS?tGy9@U
z6MNA7r!5|@nBX4l-r=&OUtMT8@YMIAtZw+-+Ts(I3hn5SWWDNJySVDt{Sv1f?@-X0
z6V%mSQScnpWqfV!g>JO-iOSb|J+&zF!A*gcm*T6_E!@8<>vbWqW7aWK0X=AKK-LZ4
zA8qJu>Goez)(;Rj`g5>3y9O1`$)rf-cB3iQ!p63=ZglRXtjVsm@ch-cMaQ)DAsQ~t
z-(9I!gH*i_S-(y0K_y=mf?L7-6*?k#l*-@I>o-`40~fN1xlrInF#JnU=fz
za$6xXQ$IqV-=2%CC>K*4Tk??c;wRn!=BMbb<#PdsmO>;P(A<1;djYEI3v1!Y
zL0l;kpLVI|B4h76QVyYc=-->`UX%L@knRwD=Ufha>S^7(hVZ-D$d2RQ<*EtbtuNoL
zQ{I@1PH*~jDi;-?q`T|`^TLJbQZmJ;NTLA62$wuz-~9wxSm&i_o_UU}<2gni1AkGu
z8vU^}S*Po+Zun2WfjGZq#K1Ph2KP|9ES0j7b
zz{@AK8qg}I$ah{XT}bFm(4I4aun(!G+lB$}ODd3AM+N?peYaY}2F^dM9p0mX$(`uc
z3U|sJyg#9}W4>7dc%+85)qI2iykBQ+BqzWEajijr!{xiJ$g7Th;qIy~1UK~))Fbd*
zeQ+ja1)O_=-^S%10Da?`BSUwrn~?4aWSz9K9fi*NS2$)+(D0LOl*>2aeV8LDUA+(=
zqPAtnsqJb&Dn|!OcK&KXYb`%=-xr|3?@0J{TgAcm3HUhB2|@h^EMMAZz)pqgR^y+*
zUz~ml+v0mAu{E5&kr0=*v39QxhB&^rZ1=&N(61Y8`FSxAFWI}xK1M^F6G>M&gP&VF
z2M5oe-U^>H9A2Nc9Nrh1KB8?f#1HQeh*xXh2c8wK-Zz&F_3jh5zR3jf%AjjK~UEm8@ihh!DvQ*~wm|5{V?CLQ_P9P*D*|38_f-p2=QWO`C=i
ziX`;c+K-IYY=-$Oa<)~j5&W&1`9J+BkiHcAQ4=;uKcVx{?PMtLn}DSH
zHkhx^G|9=MeLC2pVR9VMa$NC%G=y&r?|5bodUN`xcyJ$iaLw5#xX(WEGkSV2$dj+)
zgx3Nu9Q`iRdIjp$Be%kl1@sM%Wx<9R>~Gjb=e+~W=9KNj^}NAGf2I3SuY{t&QAv0{
zLrqq57SP!&a&rObRi^%Sn?~Lnv4h(g>`F@Yv1RrU{!3q%0m~7*tbh9-kgdCD_Fy@E
zBQ$l9zxYkn@uS^sP2I$R_Ex0uVLRIA;2JFl4?ynk@W~n=y-Rc@>bnJdilMzb=(K2n
z_V$v-``x&1&^yMPtpRd~Qu0%5kAaM3Up4}6Rq&QYeXp{c(+5EB{xtX5?|?sU)!i|L
z@T<1l3{X$Cy?zq;+40aB;W&s7vq-VS?;V(aUduIx@UqSZ_+TF1!>t#!LV#&rGN+N}
zXg#-yMxIfm{ec^?#yYVV%OCEzYZ%v6jHi;UP|xJ7@&}CnyV^Y&TgZc=*?tG@f^8=TBomzVtD8HC&wt{zd>#@m;{}0{vF~&|a(3
z9(^(c`D>tAA+Bq_(QJ;w_iCaMIgIJF#?ya(0sT#B`{s4Xvjo>1zJ+|(&B&Zt3$P=2
zr%D8%m~%uo59sO3s%zPRd_(-jNj|_IbT0W=4o5Y(Iu&q={^Qwa(BFD&%DIux>sH7K
zKZECvaosFF0OKWew#WYr@E0AfuvNuuz^7dA%}-;xM+GZ5LH|JHJ{~vFm@d
z=V{!;3gbt!nxR7<{H=OV&t|TDuudm*ZORh-N~_%|7TyD$r1pFE=YiC2625urYd!EE
z=M%pfj~oOZm3C>%G<-L(F}O&vKp5t~4MimnBzj1K6!*4BI8Ub-y7gS>={a&a&4W;$Afkq$Utr;)sRdK%{aV;4eNVI7HGD8gX}d`IzTnoA-J
z_HFEwXY+b^D&YROAlCu$XGCW!lTG#40wSt+dea&DI_kfDt#8_1;naEq~Pz>=S2FQKqK#L7fJF}Vv8S!
z?_M?*?u&0NBs739p(CvsB9=w#QD^nkJW#z>@C9mltY@
z_4+}h_rA}FgU#~nz=&LO@^0YcUWYtFf7-mxR1)@MZ}zW!y1klg_$2Jo#FPWyXZY#(
z$mSEvoSBB~wgS>HyZIN_p;97x{gWrZOcl9o;#pgGESE4=u43C1SwNnv-d0xGUqqhB
zC@PD-E+tY=92AaE0-u!7-LlT8j+_){zZvzbiTpg@uD{6LN%AK3|GI*`OsZok^YMmr
zVd}m8Di3Q(WcwRN+BJ=2dh*of;$S%U_b!pE2G$d$#Mg;9hQL0)&ftoDXRFCFqhAji
zjbD=RNuyUTf$bz{YS-f=@H=?U)0>*zgn7dLZ6E8|Dw4GE;JvM(FG&7Xv)msS+Q>S4
z&&5B`FUAh~c|MPU`Onv~J8pXw5zUVG_kn%K-r9fkZ)w_yMc);^7jPbshEK@s_jT}7
zwgkVXgZ#5~kN95$d&#)q)>I4qy
z8yG*QHE0qG^I`vkohk0K8o1K~=Wzpt}~baDb|YV5#X
zzSOsnLwmi+Zr9*>;0pyiVvpf^HnYYfjmLnZ?H71sK~IbRCoO^bof3Cd
z7y((xZ^NTRNY^5vl&cK%xZQ2u^+LER2NP$l9{6Djf{UFn|9qVcY!(K+nUwUo$Dl_l
z4F{9~X>NyaiGlSpVx!=m0`s2E8t?o$u>U6t?f2up<+*$Mnfn1J*Zvi`0p%=RuCxr-
z-8mhmVs}G6VjV&GXm92PTLXnqW*T0>y%prMwD-5;ykD3ux|bjDVsZ8PzktL2RgO8JcRq9C5AH|I
zCCj#;UH|;_Ho<4y2`%5dAyf)b2Hlc;-I#pmVkpd9~!R#wC}U0?Bxi(
zZXDXnvG=vdu>6afJ9nYXUMsGF_R`TN13pxT{hl8~*%;0(w5W=D8|78$(0(S2${X(i
zbjpg{iF#KRnj2AXHhjZhh((5Kfv_zPWJ6U
zneB3#y9J=qV;61IqvJ1N#qgE=qQ)pICaUg7Jjt_I!W
zTh#GS85aPPudsM$!TatAJ|K>ERp$6!ZEnB|7b14!d;4wrr`ii}|4Oc>?w}X-qS+Yv
zgk8PdPY%G0m{k*&fc+0&aUTGD6=gBC4)iXk8MkA*Z&XN#!uNRGCB;n#aNm&u-qV1K
zW^HVY&~F`fS0BQBo?LDDhwowK$phZl9)rv;y<-6{Mbi||fnMurZg*VQm0A$JlLO%w
zYQC?cf&MzMr`Yls*vssHQS6E^t`m8cc5DPa^;a1;=Rhy*E`3%y_TOF2jyFK>S93}B
zF&IBS#R?{i(0?_i!`|IP=q+dY_Qz8k
z{EAl@U0oKi&N5n9qXqWz_>}x*)y__mxfnwdr=eV;M}z6EA{c+Vm+iHlj~eJ*yolGMXhU=FfS#rrXL19KYn^wEv7C~
z*3ZW9m#c?_Y*pEE&=mX#$=?pGkT1`dYLjfR=XcIeTuW_(eC=*m)cE!gy6?la(K=A>
z)wwp3uzs^CWyOMnyz}%fBpcL93l}?cNYJHz;T!c`gds(rLtLksJRRZky!E$`$hGb6PIM_E`_~_H
z+B}d={Pyx^KCJB`g^#@MtK4cP-@KJ93pN##pR)*BHKCPAJ?52jt}i7@>}Lwc&Xtkd45wQTsuYmXO53dwwcRA3jg94sWIKsz*Jqjb
ztRQ`X6QP@KmXST{B)-SX6cIbVlzS=1bII2a-O~Qo3WzCVb(E4_5jpy5ibv{ADKW3#
zA}BpwNu1acG`eH+h!=*%iNh8IsSqJtqlVRsvUV${Cb@f-_;}CE3jrzg_=zI}vfSzvBV@?6vUp!nRmgXBV>y=g9?r{MfZI
z+Y8Q39|)I}1bgZJb1@|i#+Te{A;XJcFQYqUJCmQme(Wc|!mJa(j~z{Q=)kWokncMa
z3gh|^pGL>xURl<%8Yd
z<~g)&6V&$rr;4sV0lUvGv?BuSOk4g6j^1UuC09LVNj6fQz#T=Cub&HhDNt|E=}+
z!g*g~^OQB(OXda37UU^`pCA9g`@ER0*~5j$uphMZLlOFQg(buP&<-5DP__l<+k%9@
zzICAI`f3%$2TEnvrJ|nx%%U%@tM5?|yn^#QSJ>dS63}xRWTL)LGk)oL0__@KTJr-;
zPi$*zHrf^014Hg*pf^W1V1oNzaod`%pdTAnI9`H!L`7L2?c%E~KiJV;_H56k?B}?`
zVYt8tc+WY<3{1CSq>!?op-%*_DS|BWE0#LH5%+K+4(+wAkjZ*XKb6R+VEkp-DLTa4
zWvtY`MIOI2b$_hV8%t|kpI8$qq=WG%eJ8gd@-p#I`z&&S8&gq^TFjqDyl-@Z+Ha|Q
z*F}97Dii-~C5C@*?WWE*bH9wyQ3iQDUi~AkKTPhAHOBWqwhqy%0B#TCI2Z+(r*YTe
z6THVYHEnC~JaUv|A9arUeo=A+?n7w*kkC&Bd{u2Ai1oelxYY~G`D^5+iaOw<)1UTZ
zy5FmQD$Pi_9v>LvkEwiYY#_Bh?D?J>XcwEC|>q?c9z0v~|PtFtp!O
z{|<)B0Y;x*S%`dX^uzByNkCbFTjBWLYXdwE{)2vL)pX}Cem~rkxI^VWj5C(;dOIPo
zlaWSqN54tHdd8Z_ieFHl$oxFzaX_)H{*nRE|6SZ)Uo!^(JjT*uS1uRm4c)z|1Ldre
z8{V=N(({yNIfDB!ad#fp+ym6(*>U3SIv7`Xm{gsi9AmdbxI+PVocjIB7xec$l_-mc
z{F_cN-*pFmz;<#GZit0-C&s!rX!o7Uk4R^M{9L#pTN`8@Nru
zxi#1WFWb5Ih86hb@AJ*%++tz9$w#4^)3JwKQ}w=e7yOf>|6ZAPI|7ej8l68i*-5@Q
z^a>t=bUQplXP(OU5Y1DuuNGIrc(2k87J+%>iTm9L%HS6moNp9Seg;1=X^Jrl+5_{E
z3+?f@FyEv%rJdded}Tp&&6%qPu>Uc%K`aU8qwJtgtL5K1NKK9s%>kGX68{w4{{ijg
zVJmt5H9Ys>#Fm6%Q^eO5lQokYs%zPC2Cm&FNruQWOJ1u}?LL*nUMrXRR8}>d
zFJF232Jjc=zToZqjVj5|z-^6(u-~OML_2gMshWt{kC+VZttF}Nd}{KZ0-qWB>UEtT
z+Uv2I)UBLa63F{?v(wxQxEFn+fyz5Lw_?>i@M2Rn8P)ajN!VWt{?lI5>h^lFQ0X9%
zT2W7!6;;BXJJ%83ibIlJkS_hi$}We52J%{PxQ&jkm3)+raa!ZiKz^2P*Oz!vN3>7*
zxU`nl5ig(gyl0KgBxsG_%Tu=P#7#nMb%AgV+081vFf;j*@bScGy$Akc@=M4e*a7&&
zpR1msQadB5D
ztar+s^<#LG3+u8c^L(mceXp^6Kj#;Co-H%ENe|`)y(cOcxnSO@bZnQy^E9T}ze~(Gn^zLcuSA=w4%t5q%Q>E*ZM
z{YP_=Xm2^9Z!l}aKFJ~Z4edRU&WPc2qs=g{RmtoeIRoWOtX(-V3;QZJ-aWFPf$u8}
zvJ4e(!FuLqb(3FZ+u&T|H-p;aU@ul4{lF&z_*m9SYz6R_qkERQ(Lwt6=q6YNz&;#L
zyKwXdq-Qbs5iV?k`lY;Te2nGD*1ez(@tf5L=ehyMoC=vwgI;o0^5RC&`)o?*{}a;P
z`uV!?C*WU3()ub7ApEAiT-Fu92GJ+%c>ht*h9(zJ*e7Wz5Gn}(S&#Md<4+Lp8F0J2
z4`mw-3pS8>!VXk3Lwa2iY913H+k1VE3IIHQ-;x%;Yy6#kMaB%|@6WEVB7f;xv+`*L
zgm-nUR78H^aaR})Xu!Ge6EQ9VfVJR+pxylPeEwzwV0Oa~YQ2Pfdu@-rK=gtAMYO}`
ze+LQpKzu8m?M__pD4c#Dgm(OxZIm9`-`pPYWVFkh4Bo@l(O`G$TQ(q{5I0V-LmuSx
zPj(pNudDTHquti{>agA$Fn9g)5?r5&@@g)|brru$0v2fJ?Yz>WG5_@FmDKs*t0|Tg
z-*|fT6ZQT6mETRbk;jC&9gRdiPeWE=l%?JVQ2goHcsa$3hW(Y{LIyZTFtIR%@gn7f
zivRa{BXu67eEyXhhQ~_HQTZ4}XI?{nm25&i?-%w-5#uFI@Zx9aCw70RbHe(XY{n?B
zk2pr1Z)Qo?RYLi=^F>PbY!jxwlcVizJBs1^|4KZ;a{9WoecS|?{d&+{70@JL)k|zY
zt&o}*$TO8biB)0!Cv<&ZVLABvmY+v7FLhas?{7H=OBL#;YNhPK@O-wjG&l}2Q-i5}
zo$CP;HYn3J@>C;F>)UdUI*)yUiH;e|!O_Na6>-Sgx(oSM_f!%?mJIX6uhK7$=?)tB&LVuHSz62J%Jvpd^1>zf<9Q4MY|C
z?{PT~3_sra{y3iBXE0|^$8zy*?@Ckyj2f#RLOyw|oUPa#aINOQeXKC%lE$w^u>9qTSiUa=;IVnKpE00pHT`6k!iBgLO0`<~c){pLybov*TfW
zU-gK(S^?$I;>doA@AJef4q8c=w+wRL-BmY&`P8h;xD4{=k$F940`Yv$EXK@0pX=vT
z`dg5*`55(b1>rp%^7qB{&$0qrQyia;ew)8zeG8rnb_qi{Vx;x1szAHUUC816H{U^`
z)+K$g20w3ME`0J<3HVXVBX(Ooc}cw9N%b)PEg^qG)W0SMLw%FpIZ6a}l2;+Wbat=n
zBIkC#XP2>VCvEn!Ib7pKWTrYj?E4j1w^Y^S+XB4q*!^8k2f1N=EvYo*Y%6?+@h_lF
zh!^Hc(%xGEzW{N2aK@ixnT!ac%w3*YTH)ky}^N?#85Rg)ISg|~+JH6)AmrQWz+9eKbq
z$6OjxPe#Z8JW0ILL{3c3Xel&R5p~OfysZ*7Wc0XYkR$LJ#(%vL>uDRvDJ@Ofdt;5n
zabch(2l$_|woX&_R0F9o&7^r~58wS1Uw+-~-%N%~I2DwaHIiqBdkTssx=4gK!^uBT
zes0C;2OM!tBq5x4M0alkabFak?{9A--*xoQqkq5S8{Pa9oh~BO;aE}u_Y|n7ytw`0
zS3N0vn=^myP#Y;8&3m(#z6t*0(K#ZU*hI8koSz+2Ya^OhMNhm==pgmCwid0s&`Ivq
zwu{V{H4@GKQ*o!4wUA|pm3O3@x0C)a{+mSJcaT0>H%oo^J}}gBzhcpsCbHv9bqgbD
zh5zjFsEyxkCk+DXfsa!=$+@WdzVrsL*U6Hqj6*F%>+8gqcIQ^2obbhLZdK(W=r{9hH%!lT67$s|
zA&(+^$&TjYhu3fR5GF~H3!6W75hC$C2*>p|-DdrO{hcJEg(>dI^A55q@NAcDNIS{6
zFSZlq!L(x=4>u6{E#2W}z^7jX=lpn7UPtae>#AAZ
z(@fmIWQP~*YbWhOY!~Q%`!L8)I)`gvC*26s%
z+}1aDYITyoc6K4lV12D|n~m~yXW%n>N@IIH+TnZ3wv5TkaE@;E`F+V8u-`6N%)9$A
z*uC0I+x7t8;%8|M?y03MR0Dsk&_qKQq^V-W&gTfOquk=<*|9S1(
zL9%7~cVD9eKRx7ciRMYLDel6-a)<{z_Qo}JF#ux~GA;4T2=
z*6RL{*|GulyE6^NjbQ#3IJNAtk1woCIMGCIhJ0kwUq?m2dO*My2Kpl4p(odAH%G(#
znfqIByAsHkdK1hTL}0z+de_Z9n725&ZwB1n3hRYq7B#gXmwXJcjb7{|JIBM$WWYSh
zzVh`P1MpU>ZvK1Tkbj7|Gz;E;7s|7fJ{`*IE#Yjd3+e7tRDCP~`FtS_wznYO&TiFq
zJDAtQucb6*!uSi+&m5Nmy_nrP%W%G=^M2>!2JsIf0{9PuJ@A*Gr451S&8u0Z4nzF*
z`#wYYU>^*2Ckl^)UQ<$_LLsRsEM2}t#H2i(5wzBlq8
z<$2F4%rA7G{zr`GH9qT!;ZrQf)`|dHtNgx!=}d;{gkU;8RplYbL*6GiQ1=21wHg*|
z1o{3;EftJEG^>z~e2I_7E*SNAwoJT1IbBrmH1fUeDR{=E51`M01b=4@So3I%(mTl8P4Q&!(-tWhE-qO)jqPyIE8Yub&O0sfh#wgx
z?Xe#W#^Z%*z*8H{srAW-pKzT#$n|H{9w7Gqj7i3JT_e1KTE}!zyH4dZC;l<
zvIDSqp@KTM9(!aX)jv``=1~1X;KLd>%-^wfcs=GPd$)(GM>SI|)z0S)w@~eNn&&cg
z|HOs!8CKYTzu#0eyawa)%m$7Vt3aNW)Jwbr#v-ie9X0t3^dpP+<{IKYEvxyV;^&|r
zQ>Qv01oIAmlOi2~@blu4>#*O8)sAZ*KWCWx#q9_9b>dSO?guz)a-G5P{#(|V2mPFf
z^AF2WkJDd83VHn3122!VK>69&cN-`P!T1W#Id=wl_>EJBcdr7k9vV=LKLX$NC~1_h
zW!lzBw5F#IU4r>|e$S-`(=xE0V8Iv32=z>wbFG&J|Guj*IKKh*DV<|hS)PD&t3Q2I
zZ3mRj9Vx=|6ZbwIKKB&bV}>cVoyQLLZ}fO9Vxd0%@BdggLj4@BCtb(+P+OCi!3O3}
z>-0IJeo<)GXKju~5Y81-&l~`Hv<*^&*P#3ZRR<0RL3(>tUB85aKOmKLS!kfJldS%h
z823H0i~I>n`djd`ofKEC%Z%AxPnvwI&co4U>&FQ9k5580V&*3
zuzuW1uM`XAx$~;ypq)0@31N*1S+JuVRYsW)0iE}-%#?M2KL+Nbo)i3WvwVITu%ES=
zR}Vdg_VoT>5;hFyX>w)Nf81^AB;~3x66)Y-2izCxxkk#lo$L%&6G=|(Aa{K(iyyNCo_KbxYHADYn=q`n5tab`SIFvB
zRB|WTW}JD*8tzw^Tg`OOQ6Kol^0u{m;6CHkoiyiq&%yYm<6!5{g!lb3qV)>wUwWwd
zh~#E;5{JBh&U6A@M17w5OA+ip|2$@(t@ONy&~VwtynotDiqB?fRs8HF6)P1@RwcoH
z*|w0yCr~fh>2D4U{czvKv@gw;Cp{!eX6w7jsUFf@5#`PxRu-_;5b>dsF-J
zo=k&1dvvwj2JT4?tU9}R4Zf!weUj*33+4aOx}$*;-b-j-U#*A>jLRgm=>rqJWPRz>
zN=v6c;?%TODLcN8Xk>CfSkBo;B#4XirRZK#l_z~<>_-pTN4QQ2nDmk<#aJ&1pI&17
zyM2I(tC#E+;}Fp+=plwG=X-MDp5xPVM{Dw;yUEJmm3hfU-DG^`ABXrk_-IAKRbv>
zy6>=}csCJzl;v>saTl>L8|=ur4CD8-iPl}XU*)X3l8rGFoNrifyez2+*6%9!=H`LF
zyngvx`uwcL<
z1LqPuPT90yQ0ySI=T9fDJ<(3iw695)+6ebA&z0(I8t)+C8~nUHRztnFR{dn??I3Fg
zU-lKscarm$hRXz|I?3MD&*ONHd7A6^kvEm#hpP&=35|gtX{2^l3f5%|1F}4HZD3xy
zFC_JOC#)k`OfBlK0p7tAQbh;z!G^UOFIXU6a=*{G1I(9cQtoN9;1^%gIU4vD%9Fcu
z=ut1+b7wmKqiPvD@T+f!Bwx9~{7)k@Hv|1}Y}tZPKpotdH=|N02m4C1EAH;!V-4#m
zH~R`N0I$uQWfSj%^vq4uyjx%%rZ1{dIS2VYjNc=l2>d$baIfY%oY!Aycs+&rlZ7wa
zs2S#W@7vQ0iLj0<;bW5h0pgh)cZl8q9xoJf+5R`otBu~(RcgT3*Sl<6^>IC%Z*^39
za2DRn$hU>weV`XK9i9`WH$Q6*`Wsc1wRVD@3ZGiT8OY~3&1dBh$YlM|_O^+i5dVpB>@D(;U5!lM
z%utWA6jcomkgXZUBX|J$bD#de{o4ZqKSwnHZAl$f&>!C{vo2h`d(QhyFa;
zcbD^HiO4tp4rq0ue<66(BLMAsw1b=y^7h|?TRM<0ml=hY$pg*{%H=%(y-41^7ii}l
zRCY?Do}2O&>U;lTGkM0?i_xPD?!T6e+BjG69zTZso
z$G5%o6kpT-VOWnmbbgc52*%q_$x`=u)eR=(WB%hW+&AF+b4pJ=g5?;PD5B2qRVgM=
zJkK$Fn&M@_rDq>w{EnGI>VH6&tSqK5y;w#TPfX{k)_WF2nL0yusvPyB^N7zkf%^o=
z(e_IWvG<=n#aEO3z5P(mwe3BCsD4^p3h^xorg*L3#XvQbd7kfSM*Meuj^dHmWVcG7
zyzS(EYCT)Tu$JQCC)NUehVb1NVL2VJ^}7v)i9c&8-u
zJeZThy})`=-Yo|J__bq6-!@)(Dg6C53b&37QswOFO@|10UngG5BHlC%rM^40qHhdG
z`S-I;ctHmyJuH}
z!vuEhHwPt~mRnP}_8lFCTpZ^qv_IQU`6q2we3Z<-8;S+bt$1g!gz#4|ly_7v{h#HH
z)vW}Si|13aupj(1QJz6r`*hMultXP6`_aGf*i>>9!=t`@+lZKOK9V{QQedd_24&Ci
z@4<+?D{`rQ67h&o>b=V+?vuvwV{%&S5bvg_4IzQy|8R(ud)Q0)z=}pmlIBtb)UznkvV@vn@F#1syhLa2EXDlBaHKqkD7n=Biql6QqUBqte047D=tDos`&3;L&NFL7Qur}GrYf=n%h6gjdKcxFjb1j08*it}qkm$1
zhKs6)mvLz?hDYAce24n-GS+%nj#-N&YTqTU=Cx8kl=G~w@Hu;!w?xeEpYnqEZ#nhW
z++ZK{I)iqEfW2tl6>{M`g#R)BUFePR?|&F%fPQAY_FV&bp4DJAuQ}MmA8%xQIU!zr
zY%lzH81@TxO7L0$h6i6##&wUZCkIxm0>%ow`he{=a*=%u^*Dd?%j3Gs{-PE&oX-!o
z3ou|gJ$XgGV|!-qSV4UkG?%ln9Q{|0Wzx)8uRMAl_7#x-%BQE9n^_?J9~U00fnDNK
z%xm}q=S^Z6{`FJ>U$(opuIV%I=~3EJ$!@T(HXEciw!;2oYpHCxCX_qU;
zxY5ZK;5QXb-;V*kkHSBGazpy+i@z7pjz#R9T_XkQ--*%^(u4Fr_313&ekeoEM_IHR
z&Mezyw}Zam+rV>K;79ErS*?_f_J63dEfK;~z4q?Hb2K~n#e~sLvT6@Zra?Y^90KJP
zpr6FU*=SV@-*ZS#=&}aEdoS}7tAh3L_2*qKE5C+&9_W?BB(8T7QTG783$QOCa!gEJ
z2hZcocaai5A7QFm&5vHbi&b-fvnKKt-Xg*fUn%j4~Qs$^~j&<
zY*S<4{|yO91&czvvkqK}umS!Tbanr79ACp3e+4I@{o;-}4X{Ev2NjobdckuoG$$ST
zp`H^qCAaYZmpe2MZ3C1K%IRK#?O6LZ&>r5CN>qP4AN0rN>b-62
zA>J?J=utT+=O>lZ96te5Sh>gY0G)4Z?cNIMUC6T>xC{BvZ+n?u3FR!-(-lM>vGbs^
zhd$IR*E-2$0Oa4tlBRI|rPAUX6XQOpmz|YmC&cS*w)R;G=LcL4JQZqVg#GOkbQgCC
z!#pA#yn+ezcbm%`Q-k-e+jNTW6|6gR+dqiI`y&HePPg^KdtiMzYtatneB&1|@ekTp
z;Fj;{BM@&Bd|P}o^q8+DY~QwZAIt3q6H
zFJODzb!Hs@m$Ei};)e9zSmsP!g!jPpWKSfnt2RHK=aq-&g}1NBG3J5szv?;DG2r)R
z3B|P=z;8M3GWhuyw95g;;5a4V%Y4QU{~iQB{_*$hWp)GLf5URoOweB8Uh_W^pkDv}
zm@htn{_fo{O|uuCGj{B+x&h_bygj~o9^Q*%v2;Eol)ql~PMZWgSNd_BRt(;cZpV*5
z{80Wy*2{ObA^haIklmx8m;d<@6Ye8t2fbSJ9_n#@oxAEd#BX0`99ahOztjJAo`mNN
z6#F*U02bX)_Ujjb^%$Gm|0;p!&ji%7xI+EPj|3Z|AFxbl(?T5RJ>F{jMg#h>=4NJ|
zU6AfYrgMijLHurUmS>8PU;o`@E0qDYB4rQaI=UY{->@R2GbKOJ6$|0h$7^!%ecn)v
zwF>}jay#bW1mU;hJUf3u`eXc2+@cUJy58>aNx;OcUwmSK!)Ey}@tkXefyN!YhwkuG
z|3{A@{K#AF3?4w&L)yG60bNc7hqpjKkZfeu3kH3@oiyql5WYTOStZWjYHQ>&XkYKW-S!3T
zuCBaDDbCYo8?EGVz4ky}J}g9oU9HcL$M0dP?Uv~x55LXfvWWKS#UrcDxGoO*c9aXgu{s{cb?zkyRI1k0Vw$7G4|YDea&dzotv>g~dEHd)6sVth>SpOYAG_HjfR
z?Yys~eF@6r8n=I-f3aBnkh&kOVf8dK#RJqGsQct3ZqRMO`e>S&7@!^WC;rs`EDz}9
zb>Z_G_hZ*$yj26oHN?+B=?2KhwP!a6BOdJufCXmoD?+tJFuwF*aTnqr(Sdx7hvUZP
zD9;tgy+^&hatFOot~TnTc$Xqust$%{G#vki$S|V_gc#~`-&GQEyvK{;=PO6II-u--
zM{plvQ^-z=hn;YN&u~GGDEUJ1_Wc?BX(-?16sLH5QEm>E&Q)1eAiB`+wU&4ye}iQw
z%5UYaQsprJ@|@!LcE6gpWB6*_Ix78$-D%W%dGhy@>oMHGKY59d3d3i|Aanm(LE*-a
zT)?nlJ==eX;?Ju5&}<;*!D(nj`k1BtgZFxVR5@u{A5-BX5(N}qG|Zs%quG(5gPlCm
zOY!fOcA8ZB-E~b={GXC#6e<*dr}XXvTSfgf7ng8B4uV1c)edwNaQx2>3iYA&Q67fY
zN8I^tNq!5hiSmiV&^&;$!1fUvf%Q^)7EAbga7n+UaEZ4TK&ddkA8La*4XuMnkM01V
zlKzt4F#^2{Od#HAG}zff?g{20+UJFHgej1q^o#rSQ-H
zc_I>G98jdjfJnH9Bz6e+|NRA*Plum3JlI
z1QqXUl0@Y*@dn%*h(B*sLyZr9CwD4*cdHJC4Fj|k=A7-M@a~BYs^5Bt)ltv&{xYS)
zS1QBfkj_`}SSq~w6)z>rm*m~W@SGr8YX9!h{N`OaZ$v7%Qup-u@XZC_JRbS$&?VH%
zSi#JVIH?`F6LD6oFbDB)qcn9t(L%VFq|iv>pxKbN4k+7Ro<6;HPrY^{OeA&o8O^z
z`%vFZ;xQl2k1~9^hj2bE9J8Vo23)&imnr%Mp6>53pb<^YW#DGvK%buZnew
zLcb_1ZSpqS2fr`_-c^7!Hw?Qm-Ak>OnW)G9`7xDm{@F?%od2%|2E9R^A^rRyRX^v=
zw;!T@6x&=O>R(s*RgHE_FyhYw&fgnK*eHLj%`AyJcgK9v;0x++*y^2dNpE74w62e%BNU)Ly;sD|e!XD?{!;5mDNi^2lN
z5Khl=yBzJv+#6O+d{6XA?6!vhd-J+g&@S%oJQs%L)lNy;it-SvQ8@b94L2FZu>JD{
zXyW4mSr0h1)&l=gvuXOV2z>VZ?#@Gjz&mqP{>|WbJ@wPZW?P{iqpV_|9if~+)c#tx8=UhX}o2e^45
zKpc5jmS6v!{UHCy>FK0_d>BlH*K@)9sJ&a!fa`@auI`F`P_GT0zCLNdS9}iSzmSIf
zS=lDGjR0@jl$F3Q2IX#IztDi~yl3a2=sd*JSl9FQ0WMhYbKVd6Y&uA@TO8u))wXAU
z!u+`e%+dfYyu4|B0AEi&Pewi>u=8(KE9k#}^2Hz<(BnhlRZa+>9>}CQ%2lb{BZI2?qu{e66Dhl#%Tep~P1agDy;kgG;znskz+FXE<
ze>y3Dc+l&QQYYvQ-+Nmb2B`b*(3xC7UtYF$G01<1#p5+ffI8uS=Z---c`gUt^#Pez
zNikx1UoMF<{nsCU$|2^_=1Q|bIPdi&e-1zmVNoA*)Sr-dMIM+Jy@!1{yqCq7Yie;_
z`9@w|SuDISOY?z8xUTTT4lJ*^wO4X6)R$)~eAo>r$scnY$MIZoLn@Z9
zBAw?-4B(UPf!;Xp>9(i7dI8TXTvl6!_0;wCTY>XO5a;p-Um?8WJstfrkUK=gEO31J
z-gzUo9?-DnrQ2G-J)xS^{V=-0Pqf`YE?|0PgXM6zqR*@W`G5XYbn+4O8v%nnv9}=e
z=6uTd4)v)Q>+LxT^~*WqueBHI!?Qw$Y=HO|;(zUSLB8|#2jp?wHVA*8I|%8`e_rj`
z1@h#&W!Coq|JvR2l!fr&n-Bk<1w65Do(0=Y_2&~?d{60{;<7ye>l4iyzF$jja(oX
zTf{}b1e|VN--Uen(5pBGe6K8OdU26}ZZhW1Nr2wVMI{vg8@ijUu)nm(x~JoO(px(|
zXbV_-*~(TFkS6}s9SOiKL0>X3-tYMr>OW)(r=IO)208zp0sLqU)~lux<8hvi^Ip1t
zPi}S0JDiWEpJw(L0Pa#Vj?xAkd-IF3Q^p#v=y4u1TCGut^ITk+Mn4VWo}>G4+zQX_
zWx)IdA1*tD^Wp5rPs>p6vFpJ2PQbK3s<#L(==F>**QT9=||I#$-tvhbvhx5DT
zq+A=ubE%7hA%^gV-Y19!$F&=={$U;8vM|11?8-L`k66wQ3#Z_x@1JhNcnbB`@4ULhax6
z^{hTd@f-8MRD=_v~*QRxuDcSgzN!l^MKX>+HvVq7-j^;067-CZbbOH
zkHUow03cT&Z3espr9f1sUqUCO@gT=Q>mu4gZ4ncJjUtXgZ$PYmy_D}81d!_y05$)Y
zC;xxG8mkZC=p`Vs|Bu)@C{x=ffZoviP!2)>T>d{^_#c6cQQ;QDx8A2gkl1bqi=!F2pdi`B4wtS5e_FXqWO;{ohZ`Ke?or
zwr{Cj`u~Ev4$trSdXVy0L}Cl!0eG*Tuc+}(=X1*%jY`F=gNAFe(
z1q7ZqWdaGqeh-(iE#exTXH>c$@|XVS7qQ}sL&avU;-c?xcAMdZtJp>Xu&
zl0Pmwd6Y`;o}vsi2=X%c>;ur_4ipFQ^i?(rX{?s&p>*UWrLUz<`xfWPwo4Mv5od$<
z8>1a~Wp`E!?M(fa51-I((tiE63)6Y`^apfp;6sxasPsz>Y5Gz2&q$`$cXoEsQ{ji`
zfj&UEeDY>${^Nf-L(Oxm{$x`AgHGxURj*y8;5UF?ScNz>KdQZMrRE*6^SV#5JdP`A9^_Ogo9<2f97^GPz=M_TyD0YzQ$XY9F*<#d0>Ws
za{#k>+K58e3AMhI>9MmJ+k5>$+wEzV^w?VZAw#yTS}`O_KgHeEx&NGA>-Nj1agcgZ4CZ-Lo9@&*)YPFkt&+gg)d$z9Bh4
zL)|C5+_YW~{Y@X9y(1XTmw#sumRCUMyDIXqP06?PupO++Iqzr!7A&LHM!R1hb+`xZ
z@Ru0g7g&$2zI9Gm9?cL9s=xnj2>ikT<-8)zuoroiikX`S@}sNQJG$5adp_p+pntdR
z%(*Zum%2Woz9ScVWHD_Ga@XRxJ%DJ7f-4c1(#5VwLwKm$1@0I1t8b90*9DJ%(
zf#vh8jlF^OJt&sEpaFP6O9_6s3-=nuOXgubqpBVCKa70FJA1MGb>YSp*k1nQ@hMo3
zZ8X*MSZ_x~7cSHnlv_uwFB+#y*&$EOZEN}ka)qAnug{|n1z;Yg!We>pe-D0BU$9fK{uj)b`
z?$!FK8RIRK*7{*NH`H9G?m3jWcY?z68#3}STxme;2jbB&C73XPw-kjLVE%y$F?tx@
zts=4sWv8i=CK&&Dg6$RZ(q~aqW_*B?yA1NMos@>=6?_4i-&s6Dlu&TKh3{Rl!!-!o
z;aHTkf*xRJ|L`dC=H07K!GUC`M{boZwp-iPUAGWjhPS98pFKGHgBpifoGe=dO?9^LjCIM(n`&LaC@U!(9$*q8T1Mlrc0K2Y7mta4O
zij%#J?agu6S{r%rl)&>VI6lt)d18#|>&SeA1ytat9&eRq+p93IniY}GEs__X|C6Uh@mQk)4j@pFZ|vwN%qP{T
ztmO{!
z^n#Z=XJdMcX4kdw`5OZRE2w-oob1DJ`RHlNZg5K*|HAMi4-MEb-BS|bTk&~cRpq}J
zzIHbaSazsyfVLUpqYDdc82^3wMas@Oi>>;Ndg7aGsCjF<|3xscaG#g7g&elmcE-QU
zQJ-E;G6wbHOA~!Do~Qqv2jZt!aaR#{U1$4<@xOI-sd>Lva3>G}7{_fjSh@
zUD_9Dj(A4t|Ky>g(qDe1fqHIKawVn5GzZJ3p#RHtiMMw)A#sCwPFRRS^&Pt?Jbj~J
z3DZCw;^}@Y@vkAPCEjMYYN@==7VoI^vN@Od5m!1iJLrF1x5QU8k&b|T`~UbhWq55M
zC%*&$WQ5zx5Pe~H2XW#gbThz+R_Km^zi%(4vm$6I9i|^Z3m}}E)11N`6HCuUz$}m9
z-fyAWfNbGQ`Zjv-x$@@qawtZjbuq;H@G)
z*tDeg&2vd!Nw|v34ZATNas;Om(JJ&*{
zx5y+XO|_8o3olvztZyQswS9?E@LiLDBe^JB%o-Cd9`o6Si^o56O{+;!Zvoc#?s_pRFF
z^jIHRTb|;^!`4f94cg9aKG#LArTZRsP3a~i@A%~XRU>40h|P&V<{hDTU$E9ocugM8
zC|-HU-9seSD`ecLhx52=4IMPzju5E_y6JMC-V(31e0g=guZhoad5G1r9%9AYW7r_v
zMXVjVguGlnl6kvO2gauZ#81tBeM)8@F}PUEnc&|?#-1A9yL{mtS)}7(>)1L%76VhS
zv1h#}2Y*Ut3O?^8MVfA@qs^}g&+-7XLva4N-rO;|UGg>Ie{RkZO501Cf95c+uIMGp
z=ELVK)4NHDO{ekK&~BpjP;SptIQK2K{9VQn{8y&IU+a$N&mO{($*dTE6yDnxhx$!$
zzv()ChvG-K;D0c}L-t8&FNw)=t7zYuYEo*h5@odr_h5W0TnNy3O(HDKv{YkShz@V>
zg~J0i#J_^ZiTOYcDe$T=G2*Wzbvt7Q_cXmH)sC#h4DS7iw@(i%noJ{KcOU$FfPigs
zH;im8DTnWv0tD|~XeYYiHmW^Gb1ST0TzR(XbV|j~O*{+*jFoT?82^#Ee}%+$`o7!x
zvXd1-8P7g7jx<#GnGbe~zsMupk4E;Xe<~yP;`{64;GXpoo?T(fMrug4z?x6z;6Hqq
z7VPWznX1X|10Em0Iad;aW7k9l#!KKnKvpf9{7Pb+nizh!s*>=G#^vQ{Rl)zbz6)5b
zt0bPs`wlWZD<$GA8@_C8s31)r1Hb>iQBDLbi}G)_l#xjPCb!(aQewGM`sy3`5;!zw
z-~L;_gltwxk{-7yA{$sI{%!OvCco(Si!6LBCdM}yS3F)`L{vU;nbV5o5!#r%_g{Sq
z$rZgz`!=N(lbP@&+981=vhA^Kbi&I5GNqO=k(g8p-*3g1KOD*>I{jpNh!gH}PfRo6
zG|eKek;Mzq;aTK^kCSTC=`tc1D=BQcGM}i>?|g9DI}_G}|Fqs}$RxD-o#qY!S%mpA
zIXCyD1pa$`{nWc}`Q)O|`qP!6S>&*8Q{j}>Q(|w%$1v94QHMAAm2R5_35
z6Nk^{9tvNdk;ht>XdD@ylG7uZeHWdc!hhXUg0+SU;D0UVLpMb8$gLAqW9K(LBit1w
z@%u*Mp7_H_0egcpNjTe-^2m`a^09s9vA{?siOT$sqU(;w>igqLRI;fgeKS*$kuCQe
zo2ZbHol!#BGowgUb}1_&BYS7N=VnLt-mA!}tP~l)=lAFH$8%nvd+)jDe#U#e5ZIy3
zgYcCwRqusdxW`^}#M~{1J<*Clm@`v8hPt1Iau{#KV@AgHQoNy!
zyc(DD;OzL_ri|`vxNnL}j`3zdhSLpxx6;$0+r>XKTekTyp35?hzcddXxksh(K`|S;
zWfbZQkAHxcFS#Tgw$osw_2SmDPY!(dUwY%EnH*@;$KxO#Lek6Ua!eWUE`;>lRs0(I
zMNo%Znt`e*8|wGCEB-akfsYYC|GWD+FiOwdfy^%#w(T5|6pPD;O7AV%7PT^9vS37V
zzG?;}6E1I;;?97**3#EzY0&SMptoW|3N#ih
z4G}SU4=;3E|9k&A8SV`Yym2Z_gP)I;mG?bLfsC0SvIfW!;nv!%P?N|vu*A~Qeo-$R
z{wvP9#doC$+BR8vhVT`^j={J8(tCUIb(E@!#!cC6LYaoygLe3|JI$LF(td3@F0#Vd!;b2BaKi
znvU;{zayN}7^`W;8yFZJD-@|Hh2jpX5
zi$J8JhI}|A-#@ylRvZBN1uhkaYzsj~rGE3(v1bH%kz^zm|BU#zch^5cycD*~Zcg0{
zD1)1#{Hl5ntKbW#Y0JgapCFk|@1*k2T6pcA&+GoXCGgnB=7!JYN0`*m^OXEY1uXII
zVE#?g6U=B&61F*73su=fX6`%W!udo_mN@%-_{?%OSVpJ>mPfpV6H=dh(4KRRm
zMR`H833`lV#_88I!@_@RA6uwuA>T3ED^}^XkiODdf=Yno3;WAHjmx`viTJ|4|#p)xlv$%27I>43fUQb9R1I9z3m>u&HQQ3h#%g
zZTz9cVVwHlz(V;acqpSoCb*{=eyy>hwfNHlwNIUOEg|`S+^}0U*;(j<2e3qA&#G>C
zLHg*@?xkk9RqCK}U$X_4#Gj;L&TKbsW-IwecB-JGm|^gKH!_D(6;0y;~n7vnF4AAvp*cxH-+h(lG4*-9;HJGzxF`
zawv+D=BM9oTvK#9&<_t!bAQP`Jp>2PE^Xc12sG{N`EOrx460e!eBHc1NSa4ErnHp}
z!iezR7rVa(;rS<*I&JC);kXA~-^SBHxE{Ty&O$c?Z!?xN?K~KQ539qI`Wl8HZGzs~
zPSp_H99wI@5;X+lr|j|yyoRCvbV1sQonhG1*(N%hGy*e97%UHukHAKHd;2>hBhWqI
zwfQ|#oOwl_@XXvON#BxQuWoG&+V8}6MC6XcKNRs5_}VyJ&%D53ATAAS}XnT-BD=tc1vPya}?_N4c}C(9fbz%lRYo)j6%(V)Z6o$<8WE%;7sh33Fv9%
zV&QUX5{6A5M7^66a65}XyDwt`-uz+ZA;UQZc~Wyrk}RiS(Ya6G15>AW72;eDTIsqp;_^W;E6LQ8*TK
zGfj|g1pbQsVAo_b1X&XM$}YSegt}Cl;*_})kOtbvb{S4UiiTsHbZX;pTtb@w??&MX
z>h5rn;t@FUP3F`VDb5$<(#rHVr{L!gS_*tYlW=$Z!OOMvaaa^1HXpz^4pT}**SW;{
z;PjCVr}OH4kbk~VnqiJq*D@Ax6m^hr)EZN&w2>s;l{dbvr{((L>txwipRj&t(e2%G
z`a>U-(#Xv@4LNhpZgK!#HzLYzf9r<_JKjZe(Tu}P
z;>~*i#$biE;#*qF5$H>ObM_GFeTxrNkzeEAy3|OHmj}PVu_Nt#)}(U*3p`VuN%BKU5-%w9+UtbE0TKq^x4yuR;|=FH
zNax7WVQf(2K_5KGf6F(4G(UACUw!fhRp<27nqY2F?
zCM~TNSe0XS^5;lxLe>URb4q7(>`vSNi$@eLU?IF2A
z69Kp9XtG1K5WMX~Wl@D!!vURBHXgiHFu#FO=xaZM$JlSWa;;axt7Cu9_;yynqkUiH
znS;xr@uic!r)J9F3tgrfDTylhyS?Dg7EL)kRe1xLYk{@(ANcnZmKTao>bT}K$q2Okh4D%e8<#y64;73D?H{yHc
z(8!+Uc!zQ+6cYHJM&C~IZ6YsOuJJ5^IX$pB)4gXo(`
z;aCK>Z?nYY*%iW?E+x@R(M6DYvvj5RY!Q^t5P2#7q!89xRMOY56p_wRfl2Z0Lbx?`
z%PWs64^lq*^^VGsl+PdS$nz=9f_>9An2=8v96u3UrL~>~2gK&K)5T)pOXJOy{_eLV
z-K)Ky27QU})_(4}FwrC^I`^SMC^s20CzFexJa>n9Xx>N9(9A`^$i;fj77uuYV)_S}
zSqRBDd4%><|0}q6y|qv4R1plINszTwDk141eal){&4&u+CeNNUWWj*4Oi6v0Y*-U?
z)|Q&C2)fxRTXpB>Lu!uUuCyoFFuQHC`Z>&oR#fC~)?9MnBjGWj101CNaLC!oy)*;1
zc#AZxoyml;o==O^-)6xvhY&00@=VCR{aY>GG8r0(q#xE{O(F3z#r0zjr9m6vbL!Ot
zsjxi9ZCC718gv+S_O2#gLK;Dhf|&R?NV$C8M^f}1yyMCU+bfcx`j6v2a;B-!sxV3_
z)jb=&i`5pU8OGBl}-UC)E%=M_=rN-o?#_r+iENG|+i86wgCDjP2A(43^z&w+FS
zVlQR{a-c%q)hDbvx$wh}1^*-8^Wj&cM8o8jOqep|QLRLJ4w3u!l>i^fKdj}BzZ}We
zHPrhzPu!h6$aGfVKxkeXoDm$7)iKI|!<3wX?Il@IJCfsv3F-dD?Vxg}rcB7%&#!Ax
zk^+xev1MCjrNNL;+oKInKR}_^>b5RyS&)6^Ym@4oY-l*X(PAo^0qflCJ`CP2CiS5m
z&zC$dfDGPC-y1?dz=x~vOLP{pq2Cv=+v6nvuv=jA}glTo)m|1N;fY4@-!>0+qX^=e$BxD=lD
zzVuj`vjQgW@vQ1xA+0;W|L6j#&rwBX`BCC8NoUmW>MH??M#yS+W%PbiBm88kz{r21
z1|GLMkb+Ttf^nj!nCFv}GD&z|s@;7x%ANUBH>~KE?q>U9QGtttQZQHL$P9!hb!f^@Y3D(BD0NV_;5=*A%=wCd$Hf<{G78<
zW`fSx|IOh{nNDN(83FS;nMTD!{+)SgGA@)ehqkQAu&JYkbB%_HHA9~a94y0|WN35h
z6|a{k)jSJ6MaG&wCDYY2Cgec2UK4gKwaZkcN=E5L!sRL9`kIVGM}$YWyJbQuELpqU
zh#IyJ_36d7>oTjwuKy&C)W}e}e=MNTt(5t&X(+I16D#xU6V=h%qq8!8pShI+txsTE
zH{!k*+V0f67I9YJPu!D9k?fUwEliF-=lxHx8n9yjxVg^`|B=8;mppQBc^<}u?$w;p
zD_XB9zS#Y|_a#TAqJ0m}W4;}bb
zBQZxN$>+(62bVq4861(sZ7Z(4t1Wqy6nKhD(9M4=wO
z8>g@`tD-824P`uxZF0me?=5!!M6v76Uo$+im!mep))*U89oW!$V1dVg)pIpvP8i2k
z(`&gx@%YpOW*U(ZYb+t3eRIgu3O~VD|5#zm4)fpGKSC+vi=S@hzE!~c77I3(UZ%^s
zi&wW;#2S2b#9VkX7%~$c;&B)0#k>B6VdsxC(sI`)W2qE#trH<1aPZ7VEzB+zTTqLl
zIv9|JhcB8+f1|0!?1BHq4)rnII`Ubq)t7keT!=)Y<-u4yW%GEt+>dwIRP7ckxDW}+I4!(_igyXKK1x-@16g6c?3}mFyy}nyDoEx|(khb#!(;1ry~`nb=`Sxio+u9#9W3~1TJEEpZADE?
z+2_D5WxMC&Ui>KiWnyrEnFI*a_-jw>Xri$xJHtbhk5Sg!!%Tvig2=z}`qHCE7eQI~
zsMGsYArugW-QAZu3#?CGdH6>A0q8VuEE={J28T-=!Y)Q$M6A5|%{mA8fE|j{=Jx`cI-?Wf4~Jt$9RS^^?X6$B8NdS9>ljw@MKsArz0D5=0XX~?
zyB4p5#_7bu=i^flDrGyFbuA2(Qw3`zr?LkQ$Z3;N4
zNK1r(G$7A%RPu6|JNTq+$}p85j<&9Cir+{~0~3Ui^8#}UVq47%-iRzm61lZ)!zaQ}
zv4Hw_1KK>mm0{~p7FCL<=-(0{dg;jBPj%S!L>Gutc|CFcVkVL(9SZcLtO8Cp%(JnT
zW5_;O8q7TY3QV61(VoM80Qa8B!(PJ_Up-nEieIxdo`7w>(_!-iL~4N|lUEN6@0(V@Js)92i*W`|u9+0*_Zmt|q1|
zqi?5__iT8YP%5wAX(zWiz~*de1!We	)TN$FODeN|H0&N%Rjg;^zNxP=g-oIGlQm
zy`+P>7Pd~RY(>DX@cphC_8AEGpGctPst1O7+3{(%F9E9v_2zFob0D)|v3iHD9atTA
z3AgO32DOqOrI%Bp!P3*|HQsBE;Lz{mGpP%2P+DIMpDy1RQZwwn8+~LD9sH7ZD@VEo
ziKExN^*cG}%G!rn?^7?q;GNWZLBa3f%
z%LPF4nf>ITH9d^({rgvt`Y=qi{n~u%^lwnRIU&m1)CRhf3$O9vUyzK|=O0g34baxX
z|LzRTzX!i9sQp|;lYx1|{G8->JwzT;c8DwH8B!d0sr5NF3Z=fdDo44IjV@nS^$E-f
zK(_YkCRdlEfZ=;7&T{8K&}!ivH@6fC(11(D-$yxse&xjJk>EGTrc{aJb3zKra?US{
zPW3=z^0Ip|wkaUWFycx~WeyM-xSRJ-s2O}svbm}mkq;Q+gq96oFQj7?GQX1JfZ9j%Ql6g9MXj6J
zilsT($S63FD`qAX)Mo5yw#vK$O4{EtLb>$-Rrz9-v{MJ5Ubn5O?5soW)F;L@Ue_SQ
z<(SLY5(>bzA<;gIyV)T5S5V}!&sK=!y3|6kqKj_UGhV;TE(F4v7!Ot$9R?@8oBVAL
zb0EE_&=dL#_rYW{4}F=eBuI!q@X2oY3ew+GcRARf1J-$-=R_lnS3WTe%QO9?jE-~J1JEp
z+2y!8Ij9P}5{^tUi3_3psH{Q}lN^vFp4EAz&Iw$i;tA6x5-e&~
zSn*qd+vi-}ut#TDcZE*h#JzEM`}${%*wNRt(}oSjn4uECc8bL#JbGw7FJvG9pU!*p
zEGEkrGxNr)@Py@m9Csk32$RV@?-i;EaU8N`ODs8qVL@d^{HqA1`69dN_|)
zAC)!uJ+*<)9Q}7e{d5LC3)es7pXneqm7%7RO+(^SmXUvqOR{UZ(!f{6;{rcRi`~3@F>hm0YVAwz_^p?&tIl*<7=^`Q$;H=AxIWXkv`Ky$zU*r~
zXh@ZUH9CG!9expnhfJ@#k5r_5rOF_(gbBA}etd)FkR=v-y~Cba*Z`Z-
z2=N%}#BjO1iJP6GT$q`N$Acy#0jw&9rU+?7;tJgqAzi8N*aFwYq0QsNGX72uT3M6V
z@Ne_eBQu{hu%3D89v6QyT%684pvh=YrtNbr|Jc2QxER~&poZ`nta5Ap(TgX_c(8Hw
zV)W;68B;!^gTKmGYdlXK^f8^>m*G*Fvb)A{0yq8^x^Rl)ApX_Pra6IgO-77v`NO-}
zNg446g%2v~S8B@U&Mp39W5Kc)U8Wb)lrW2;m>QoPO(j3_JFTIrxn+1uhis
z%oHg8w?_T#K$1|dE#C5LUGkLLlFVX%e~?R{FkU3PE@&GjgGGwJ6u(|*fDf8Z-8r8fYYw&7t1>s<3@wtO@CL-v0SYql0|*Y*gB82
z21x)F`;jiXbF@$uCm+v!RV?m=-3zvCOCPt#Hg~|n;@C}WaFVX-PNxpO{z+2fFQ+*+
zr)KfX-rpOa@8fWeN)p$ydXMLg
z&wTzR@fKryn9nEWmV$dFi@fD>F2~A>PI1n(&EX555-#)piN_xH>{VF(dWUB)7~BhG
z&A@Uo;?e_sgvVS+DE0{G7)q&eR^1B@UqI(CNo%47<3eyYSyb57+Q~kuGN2pI@=x!{ncQ
zPmJUJJA)#rDZBVxyScNaZ1g~WoUNposU3@bqrJzu-;9xey<#C4`W+V`Y<->8LCARnFs38wv(gx>q
z%*D{-gvYO+rfGq+3qm>lUrYeI!s)O}!F(u^_1}-PdIDhSXsDZbgBa4Jk_Qok(%^%}
z^<>3C2V^pQm4?<`1oWov-S!}tKs&rwXe-%efN%3mP|Xv74wjNp*%q1rMipqK>MDek
z(tOl{4oZNAq|bLvC#4X->a)xB=cGZ?=$)UwP!v4s@yZhXD~eYpDwlNT*1cz7VCIOzEuN1+=81;a8`Afa
z0^^{WIIoKg72Haaujm2txv=ev7lJ`ndj-S7&vYaswOlZE+!CB!eirX_*$jCwO{ib#
zu>>^hUf&-wC!>ORUGHAu6tH$5{b;!S8r4#C9~Vyu1~F>dw3Ru*$UfE%A1^2eEc_pf
zze#3*UaGv`>tsGa@T2Y1fxepOoE=nM7`bw|%0W22qUDxdM%pWAI;tN?c7vFDk|Eims>^BblTF
zog`&{;s)+x#(tj#Dhj{nm^Gt;_z_vw`Y>U_T&rK0H*yhNSYZE?Ci@$;-}+-!c4Z8`
z49(B<`b!V>TU+ybyyu<
zD$n^x0n)PE=KV%>k|(ZS|EW@g^u9ASaaJuMacPsmsptk&_{F5kLuUX`DZEe8PU`~h6CMKtQH?0___Nd#
zG0EVjpJiZi**7FK^Ohs?RzLFc7j>u-L%`?ZTTa83aj;R6<9_kw0+6!addDcY4t}tD
zZf@_^fR6a8X*Z)ObdRF@>Yd1!NW!(0bJCEUFjtD`tQuQGPi1#(3f2Dt{*LO3&BHrj
z!u-{Dp=f$ygEFfht79iD$A%2nE4hj4bSmX)=ilht-HR>{|D%EV|5|&QID3Fw&czor
z%f;w*i-h&b;7@1>t?$(Zy$5PZ^H-y*KY&2x&D!SH8lbnhtQYsT6JY7DEv@;zfy3wW
z?i*K1(OBT0^C!;4q7d6%^^FVDsMwb=K+SRh>40#(-7h#u|1S+*XsiT2R=nz^WBzD8
zG;-8re;0V&FBP}F*S{gt@M+>g=7-#5Hrc`5|lPu!9H((DEZ?ZoSsS#3~t+YyFVn+=>I
zG~jz7cR0#0h+nO@Nk&~^Kc%#9*&++Icd=t7jzDVaJJHdoiB7%VB&&9B0uHZkSxr9s
z2rAhojW&ayqb^!2_AzcZ#4*ESN`1u@%tEqAm3zgge@f-)e^=T-T&xcNx>_M%HN5-r
zzws1ggI^=N=jVs8dgJ}6Ks`_}_C?*_e-%5{lRinu&WnuPm>YTTF@m3gTtc0Xeq&`1
zBnNm6`2gol^*sJrIpD|270u{R4zy8E2lpBiO8fTtM4JD1%-W`JukVU3QgE?z|4n%f
z9V*~tvFD-!Lwz64MfaZsxBKjh)@k?g`uEKl@``RiRW&um@|79Ts;u#(wN^lF1Nz6m
zg|MP$>dy~I!#NsM)7!#4r_
z#WWC6!7ESTmmR9{)bK9+w$iO&m*5`k=7}8hd-h~#;(a=U$^#0-Vkr9Mr1e+KhF18V
zX~bs?R|=+jKKm7CT$?@b87dCseow^@37!DPlZuDUpVFeTsw(p3^gmduF1Cm}{KgF5
z)$hs+ivf;};+@~P3Ru`*PwHR%4>8<7$Wir^25pZY(77GJ1*8kU`8`=G#t%MI4^Nt}
z!^yWO3$BHwV56&(Ligy+ao>q5wlnU@cw>K*p^f7RcI#m6jOLGzSpE59W$D{FxOV})
z@-re43#>lrq;W6+3*|VdkWsmSOR{G@%Da?|d!9DEw{+|y)^hAULFJr>=RWvXSrQY7
z>B|q8dEDN_y3{kcZ!S&aRSuR-7kJ`vr*Ms~o6m7fWW;Z2Gou#2>-E|GP%1C*b^64_
z@@*P>ee-s5MgJ;(%S`a^LDvOL1U;x3S$i1+Am1|eiFMl7~nE9
zV}u99hrAiDjK}`5uHGFIjKl(e*G27|b-{OiZd6_N*2hdF=5C5Svc%p<+J7fM*o@ED
z9y-u^=q0XB{&_w6W;j;#``@_tp8%YrJE7tGo&o+iGU@LXszF>I+H-`zY{geJjAb3?
z^D+CnOo5f&C_KiRrS!mwxAjSwK_^QKVQcROLMkjV8*Y^MiCik6(yyB`jZv0#!y>HJL+ch!$WqdRgyF0rXD!?m`
z+n|vJ`FE%AI{l;F=o1$v#-)MeJ>?6BN
zMLaLA_Tj<9VyBz9+NUVL{r{pd&fZEMBXwJBf4MB!Xww$!I>7JYv*UrYmC3aIKKTaY
z%MYA5P|=Rv;WRvcV_FzR3k#^{OvQRNMGA($m6Q6{)gRAiR^q`4cEfCDNw@(KvABIL87q32
z^&$O527W*9zP)(UN35Vt#x|U;75nCEG4aFnIZnkfy_JkxWAUkW&}lmmx43aeSSoxOWKes*6+C1ETJ1AAGoCSnCXeU}37)gy
z)g9MMZhuA4-_d6mLtS-1z1_^v1ZDyH`!5YP+6w?U7oH~BD}oe_s(Uy90NuAFcfH=L
zqGE9ay_z)(WPi^7$A7HSC@>+prcwC{7@~~)TPr4qf}85iaz|MK7M?81`qCOW@=ub#Py<_5|bEeu2Dj1fAc@Z0=#B0Bh6(x#)`2@H>x&{WTPqkX<#Gn%GPfLJ`+v|L3H
zQhykFZKmi6(pF*bz?w9XO27>TL0NNPFS-A$mM}t+ECTF>9;$%tk8H|yt7Pynhpa77Q!nLp_Wd;&NPbbS(Trz6kVR-WoTe;_o>&-hCHHM*p_
zP7@}X3|>lB`UbmxK$leb`qKmQP|`mUhSV?SV6EMs?3;543S#zuS3jN!u5(Gmr>Qoe
z7LMlLgBw(?+|%#jrqaHPr>`e&$3syyHKnbm*=tGVsvHonT}#f3JR%=kfo~)1O~PS
z8yzd}0b|!d=}JxyIB@q-+)C3TNYgmq_RXdlDU^oC=NxJU5x+Z^5|g^o0n>+DJ5$Xl
z;l@(o`wm7T)M)aTXT&hl?AMyVf9pGtt!s@>Ik$}xO&GRC`fJgP3)LgK)W=9T%Fb~|
znk3YFT)8>XbQeXCaYg0XGQ+)lDp{p`ZY92oc3A#?f11xmY!5zHZ}7eX
zDJWn4lF^Wa0>l0i6VxXO7Ju(v@ni}j)anoIiQ7^zh-!=Mlf*fqqiUz4FW8Lu!9gFc
z|5_MoKK`oa^OPOlwt^z}6i~!^&E+!3j)@omKu{xTV>GZv;?u=
zqQIeStqRTEt_3_65P`+fouSry0>qZsS|zpdIto1;o-KGt8U8#aHeH^`NT_ATpAM+e
zA$;W6n_ul*Bi3H)(+zfs!Scht?w|V;ngr)7poVp=Rm|ak)K>bt)zP1v)KQANS0_$u82IXrB*1yFsknYl!x7S0(aR{Wm;=M2OLXOT%)@jD*qaeX#p{8+|z+
zdG1)PF5G(?5~ZDE0RK#UWk1NG1&{bQ>{8`QLNf2oOK9AUxz+JduHOby@y*C8#Up^!+K`<`B~!5l@iGtulEuCxl_n2l~jj{
zB4gz&YC>#ARn>@>2?{I*q@Cve3!40HWuIU=O~fX}PFS60CFE~PX1X(9BJnJPe7_!M
zhJM#ye-wRk9*#}_{qnrwJbXCyic3R+8TKa8SU9&pB6{dm;K`2)#L4N0)a()R#5>ax
z_2(I6gzfd3%9sBHA!FGQo~!@q5s_l)>-_JfN&4as>pjnC5vB2=Zniplg!0-YyZVce
zVDpM6HQ)}AY`639vuTIn#isg@xBD;ANZZ#5)6gZK=owYSjgI-mg}7+{04$nWW=tyWU@IK+EDrgI{%fAlYZ(niIZj%S>T;aH
zS&953X+_gWbFzVR{N88K7XLCSoU$GK2%bJApYH>hX;>3ayemS5Pfz9KjWQ8ZZKs+k
zJPr^kE^Q4940T}GKK-Rgt9{5U=R>A_
zGL7%3_SswDn==bp!ISrrE|lP;qIjkC3<8VW9@M?NX{P37NuI}@tdF&
z9`zK@ioLdF-%=0d_og1h`U$+Oh+f^_Mf^eD)NK5N`!cRJpI{n5YF
zo%rY%mUeR0Bx-F7?`U+I#q?Ey{i8F{!k8E+j#!kMfoBlM{=x08OXR?XjqAa$T?){s
zaDD%Bk{%-ZsZUA0(gBoA7=^y+Nwj~s`=kOH2l5!7KQQSi4DMXLcQvtK3STnT)1j*D
z!GE5gwV|*s#Xk5ATX|b1;79v&H-BDC!CTa83Je_9u=)#0Vh&-Sv56;_G_j~^{Op}S
z(9frUglExgxNi6gvu*jl8%jrpg2etRbY+&{N_rn8`LH@n%3L;*(5S}yR(vUORZr}N
zpq$?1S8@yA;(*Tj?-*J6|V5s$dPH{eXou<6hPQGl4$fOSDP
z<{-3ke{khD?#7TJwW7O<`DJ;}M__fhfJS&5?J+$}W@cH1J<$VulM!h3>5)G!YA*Zp
z&#x$q_g(HAYSDh|q0zB@iVb~iPk@PC{jMkeHRW-{%|0)T#*J-`s+AMdYJCNs6t&GLE5(Esm$z2w%W>vuY15O`#n@=H
z`e|zKCM?#9yg(^#0-JxtO%W9{feF#@he)%G;X-%9#p}M$VAS3gMjI7B@HDO&<(u#%
zlG>tgIiVhdm+gAin@kj9jGsHz?B3+z+ZkgV+FM`oP~GGGH_SM|N{*PCzuycd5qZ|1
zGQSIFPj${{F0aEjgp?8@d^WI%8mb4!ychAMx-VZv3uwT>(XslYzUd8ghiW^3bCR=ZLSJaXgZMJ?
z2yt>GSQ6593SY!(Uz|>rW0eFHm5Ncv2JfKUksRvEwab8Or~JOtkPy1I(jMy>CfBG<5c!X|1*(8{twT!oksXtoT9{`5Tx
znB6dI)DyZ59=2asZ8sA{0{_Tn#QUzHe#P%;QO4Q0>-Cb)x#T-3$d
zW{#--Nf3B)(gBR*_(S(KFJKi^wB!#mP&p0!mF{{KNnBs73$|23ugLlp&IQ>4d(mHJ
zRzw_fvQrt*R7?RjlNybkSr37yf>W=;H6Ow1JQqr
z(A=|tOL;w?fUE#_*+#_(l4wqJmNK72+9|aG!wzZ4?*6^Dqjz5+_Tj+Kk%!9w(@^dJ
zZ|x9b4Z58B`q3v6KA~x)#QG&@^imG~&$~Le_83p`3oo+
zC+@|Vw1anU7YYLJ_aMoa4nprMMuChY^
z2TJdr3W>EdBV&io@8L`F9wVy51hhzY_8EM$ln`tWI~n
zf}n<~w9enY&T~WE8m@$(xh}-xHexOR@DK1eKYnQYD05GSAh8SbwY9Qqb5YRC&UYzt%%Fcv6jdB9f_(?iFke$8Mq+LIC=IE
z4>ZX=-)vy6Pi*i8eAkl_C1%bzQJ5A6!gE26FPyjyh=Fqzb_UOn5Wz>H$y_8<;DsBJ
zJ^#+Q!=>s-W0kcSf+6c%(KM;Z@@TO|8C~d`&JS$jnCLe7dU%
zmKxNh@WLJT*wNVV3*RF?9oAwvIO9TeSGVgQH>x9E+Pla%JK)3%>W;kUYL0~T&B%gQ
z6F2zgr0kKiqONeVZ`AzW@-t$JjRW&prJNF-3OsXOwaq-*aBliR%*ElL^!zGsZ86CPlbtQx_IXvUm3>}D1U?Qk2
zPYS00l8$hmruJ_nxlHOoJ!juF-~&x56lMPaG7nyHV3I&oA%pNUUfo1hZ<$8Gyp
zhUk7;St-ab0^jPZyeG2lph9i!qo*wQpun+lPlKD%gqHefZ9#`FQTR4;#CaYn^TU0t6IX5jb4H9)!*;!f)8DwW!Kr~#fE7Ro2MR;bjyHG!h(bsx-uX=
zO_+vA|9f~~+Sl#ZU>p>z(+u0o$bmjsqK!O>&q+-d5hDG8#4BIhR`zVlf?_?h4=7Hg
z!N0Y*^!F2T#2f!#i*$!MVaKR0zeWKsWFnKv{nGdgWflJEe(+}-^|OUKNB3o^>_7QfNeTyxWgV;jM7G7)5&0OJqyIFY+c4?@2{`9@CKW-4S
z*2{T;5^T`PR}^qZ@{r!MHS+sdJe-fZE=R@p0_O53F}!c(B@Umasf+AZAxx@GcBk4U
zU_gR!A~ijUKXLlon{e+q=yrO9;n=NvFhZj51S&KpyqT@YYbdTj%S{cgl=CY{)~>-t
zSL82JqEQy-i|s`w+Gd=}j#Ge@LtpLN`8pujeWL+8NI_gH`gur&s~Z)63SYTY(1Q#k
z|F(D9k0GuBUn33aJ`mYROsgdDLYrL$%TSYV$a(hZ{8JA<5WN^c*6g?dm>#wA#Hg&I
zu6dEXvNsUwCMAoWI?N3}OOV?wxt=CY(NYH;zfTT(@BXD2`$+@EryRKN)XT$Oo|~fY
zjg{cN9M$SeABBin-m5I!!<0lTrWsV7!4DTpUr)4JmI6BJUlMuw9jGK);9qFmJD~Xf
zXZOsx6d)UC^?{rtA07RCXXen#7?4!#vx^t(22XFdUwEm3&>aPe`_{y3aKR!r=+j&Y
z&1BShSfVreG@NDw@jEzSz%4n~-+Q1ut?1K_dDLf0O}ha>H%)kf+F&^QUU_gfAlW%mv=JJOVL}gGmQiNkj3MEsw0Z%hCGJGzQnW
z51`=0+5aYYjlh4=idRcr1CYl*HUTpb2zvfa(W8d??!~+H=#5+q%Svz#FdY>l
z>z@rp#&3oOgtX((TBztoCtn{pMorJm+&hWr<~_A5NVruQ#P79ko(DR=*&%TSt2AEk(E6P6$yRq$}D8d-pbzN
z99#BmsfZ$@Ldb}GKYzfzkH`Jt-upi9*X#9s`E$vfMY0!V-}916AW`qxtnju-P^4|}
zg@)b=%zbBevL!J_uA=vueXTEnz}pMfq#=KlU;JnL53Q#l%I^CGfAEK%Y57ZB*
zOu!?i@~4fpQlQ4?>r;lAomS+#JI}82tF>6m&L-mHs`879>i6vrKeu)+5B-}G5JLI&
zUwDT+*};ZE{vBb5iB`lY)_-t(p;byX&{V{FUD+l{yn8=M0A)lFYZVEIq4%avtT_dg
zz$9x;>i6A`t%sTK({76&wnhfMIlHoJ1DsTgdPG&Fz^32Sf4|s;5ZR$sa!~InP`0<@
zRw4gKne@K-kK=YOAoZCK?UM!zaQ?o_G9`x!dfH?EmW7TF+?XXdefC!jI4q7OysfBL
zruA=#omD7rT{WIjy7whX*|U`G!>QlH%AY((R&vu1lut^&X_K++Qbw0@q<2>)T93UB
ziwwQ>S=lSmW?fTcyLEG-K1Cp8OWCK?zcz;Ht@4!PbHS8Ty{)31IU!5Sy~?{*ACg`q
ze{bby6!g)hJ_ADCSQXw7QmfAncp4Tt7h9KTr;iCdUsl$22tRwJ`%i1uOFlj=APYYC
zB&lBw*i(L0ee^p0ITG}`Dnr6ohYBQ5@dfq`aiCd3fua7g80f#2t#rbZ72H-|e?@(p
z85LJ}gvasA{=U*EI?lKd&d3fmfB)UZOLu#F#C
zn(Z!`^3ej?3SoWEmlqN9Ez!=P?nvN$^57ozq%nBvPX*2%;RnSBvwdwxuc8(+3AI8~
zFR)&)y9rG8b(zJxN76VyUku9NNH0{tMCspJ1?pgl>|jTtp)8
zFMeuma7Lt}U%+NV0`S)r^0tVa1Ul%Ji{fS+Ad?z+#^&}I5pQW;Uf>J^?cbUgxLy^4
zl&?G!oBoxcNpSO?*g*x5mm4MBOs_;%w4Z(|RX2guFrNd;4}IuQ-Lc2vmwqGX2;Ry&
zTe&DY*I{0+yAph{?scQHsY9O6j$KmoZbOkQ66f0TcYxtKS+)M3R-ngQyLGYN3Y1*W
z*3=SeN7~J1rDD75psCBbt(u|D=bpwco4SCmA+_lUVfmE3SvbvoUNLa$Yj@x<#9L-J%
z|N4Ovub*_JYb}?AJ?qrl#pa3_F_SfrIg>%Lo7{|_^;z+IhC9DFS%u)((_~ADKM>D#
zy}umiuZm;LM88(h^rHLoL^tPO9fPCo*Ioa6OpWh&T6mV7R>faFPFN+D-h}HckAB#E
zmBveF>vhPgRG~?eyPW&D9;RJ>%&{YE0abjA{qAPmhe|3nZ!&(LhC)Ong8pAlK>6d6
zz5!cjvC^(wW|q1cWVV)7=oGn(=k9yn*z@_0=3JkLCmq+u7i#c1x`G=pO+b`+HtQ}9
zRek7|1|9K7y{EQ)8mF+$ru9$s(ikpVu|9jNW{rI^)^n+sui~1t?6}}-2{7@TyVYp0
z0ZwyyGQ3P;4};ohmj!72@Cyt7=@{Rq@D5?5-@1~Ihq>;N>2@q*tuRVG+KXk@2B}vVJMbNoi^nOP+OY=N^DW7YHcWFkJ~(!=3;*CY
zAjH)=aPEy$w^{g#;eCTN?~DsCA&Y~TN=8E&^c7j~xEoUp$@xtR-lpY3)}tr4|dg*+M!tIUT1#*
zg6Hcxr~x5QSz%48mmygXT}=|d-xBDB-cMgw<(T)tAKF)GQhs*A2mxXFD57?FJmJ4j
zNn;3xjIP?yHw;1FyAS%-JNhBBRK!kzau2-eYG6M5qysuXzA^v%XB)hpqCgVTp9>9U
zmq}P}CBtUg)%g8O4$!NH_s?Lv88+QHHJoBC2OZG&Ca0ZJ_^F8E$4^3zH#GthKT-{X
z+amVo*~FYM(iS@MNd7wh{qRwpBS}5fuVgtcPFDqWR`)t>-P0jOwZ5HS>q(g1e9T}a
z^&Xz4`0u3GGJ?`IU;0LvnxHT>vCCmxDV&Zq@nuVogmv8m)&eOic!5~NMebxfBn?`l
zrEqVBzCI4F2Cpk%o6~w9r
z-p|aL_7SJae7(jCK6v=jV(Z0w3oOjkUEO7<3m>G_9}}gJggRa$_s?1>;scs4kK$uT
zp-z&!S)|~7Tv|Z!c(dLT>p9*Lzw+AVlGftewnH&t|VS^Vi8|`gl>rrK_y|ov>
zA$|{+GtWPht2=_fa*0^I=rn>4*=**f29&Umq6^`D3WXQ5E9R2~9kD6e*7vs7nJ4V^Saeek2#{g6FHG>Az7iLDh5~$4`r}$%yO#ZxYc8h+^nad7Nq()
z6G#Q+jdJFXK4-&cjCQAm`HZ2-aGi|+n>wZuOgy&eC5D5I_vPp3#i6D2zPp0k0eYP@
zxv_s%9trC!QvP}N21S@1Jz?;+7#Pz`$=UWk2d)1}v{WcmqQ{vJTk<@~VIj%AdpLCv
z)J__Drv4m9XY51=B3F9AG-k&FRk^&@|{BGfm
z?`uRZz3wm4Jp$g|H*R4k;lh90P25RC*U>vV^X_Q5pJ30J*Y%>>Cv?Xk-o{i3qmloP
z&2#A;!_{y3F8pa@#2Qh|#WyEbfTr92RW6TFK<*u4Gw->KW;>5L?SxmO#Fk^JZRXs7
zi!%4B>7V=H2x;mSN9DUHPq(j2Z^s(YJO&i?vI&ScJM6wddT&N-)SZn{KV`
z)&(39tbA_^f)MLfZ)TFg36O8Ltl~?V3$7*dd=it+K&^@rWpfs>U{(=aGeSy8PCTbs
zow5>H20UiUHNhZ_TwpGsArJ{cpSa0{F!X5a2)}CDb1)FW0LhaT
z9G1SX1EGQ)iF~H7(3QpG9cz>SfnJfoxgt(ykbEZAc53wsIGH-OOFJotiX(FdIG6Z<
z^ETz%!9aHKl8Ry6eIpn(c1Wliy}JO2mDn5C@2LQiPj(YTmf$3d%Wu_Oa|1TlE+)nf9MlpbAT&-iV?Bn{NSicE4Q
z5lFGrKJz+pvo(EV+*u*=pR$jD^7^G&9>m*xloYn>gPyfA*
z4?4ouTfg^$QsrzuKsmBt!q8;|4otEmK?DcT3apgXVP!+jqZA!dgYg=)a~hl@
zQ*tDw0-#M1ZVy%S0wyV0d*Rc<=;!{|aUG{Kz+$=YNI%y(^raxAGKgCRjS)pgvxjSf
zs{igvu#uieBHjAuj4%MQZ|!S|y)2MdsHa_5dLUronfbd)YzW++*4(dTC-^|9nT0zp
zoCRY#yhJ8vEkJ4i0$He{FG&6NN9|<4Enur-_~7qhhxP(W4Grtv!RHB!SAz<{i2fh5
z4A0Xm$hOo^`c0<`VE2lmGaqw7Bl5c*b(eiWT4T?iQ%@o~>K-9%#8m`FWz;MSFWG_*
z#!kikl~>TPg8LRp+f4$JX*<%>`59t;wk8yp-3sEg8E=_OJ_Sw|98LGb?jplk`nT|s
z9N>PMV4vPr0L}*sIBJ)SgTft)FDrMGK}^!v)4$Ks&|B^3m=wbPIQ8xR;18uvWOb%A
zN%GhrnwVOxop8?vlr4UrJ_W0TOZk?^r}gik|I*lN<-1!@mOuZAF4{G~Fyq#6%AppF
z2L83w*$Dv_N+Cx{9#^8sca|oDa^oOR{x$2D%D+H$URt`Ks1vx^5(V#hRG_(6qg2x`
z(}Dc`WCJdmNl>zUBFTv2J^Ft4RFuZ$_h3DZ*5&=&95UyUSMd4meF4uR-KUD+K?a%^LLvSI1M
zaTsy!YCiWgExxcvTQQk8xgY?3*2fd*Oh^6xc{gf92oS!b3+I}H}9hg5&
z&!0Gl1H)Tdnst@%;oFP#SN^HMgioUnKX6^b9U{STzkeG*65YHuzXU>S-dkVAw0bTq
zv@yF7%)<@oGyZYp6aG^!KQKqqBNvE@Utbz{_YnJbr;b&89Rc_2Ej)b_)UZ^b
zBS@wFGA@SH4g3G1Nuf~>#Ac--RJQvvnC6$Zn()iBkk$3&gCAPbaQCh_NH6!nLh5Gw
zQs2U1rAvF~D7P}!%z378^^GYscyzdG^3NWVhYW=Ly5|7{i0}6ebtOPe@Oy{B${F4p
zF5j|R^2C=uj|87ovBM$!LCO+uVqr&?Zh1m`J1!KY?V@Ay#*bd6mk1nxh<}ifG2}1A
zLM!5oPLYRIxH8iD{;7X$_^Q0a?hum=4x~@Jm4Dw43WXN#^>jVM#1w_57t-7DZgEG)
zbq2y5=(DWy&h$T6@Ma<7)$6D5*Y^NQhVKpdRVbIbj9)wU+mB&kU1-7?;{y-EJ03x;
zhpVLw-I4f!XlSA9CBlu{W*TqvY|!XfkPiV-1+{X
z4gYb;9*mJ;HXW_;dcCvv*Opr?=CDRCDzj%-;LQ-C0~x`)nkrX
zssA{s+OQ&v+rJEwcAS#MUcs~8iEp)UH-11p_{RI}cG;Om{4z`G6mNVBw%%SphzxAU
zK`_Tt{7om0EWa82L!}#Ey_XUx_hS&=Bx92zIo%Hr4#qVhNe`@;i}f=0>3}Y;X!D8o
z+Td&Qh}lpAhnDB!PW@)#5d7hP%;OV*k9T|^S-Y>Y6Uxg}>*ZK?z_name~r~)D6AT(
zS3xukqYeza&R^?;u7`Y!#>w6ANA7KvQI;;q;B(JHC~O2KK9w$oGXv0^W8t>@;%g{D
z^GL(Tpa)hl7%ym1cEP`OQx9dRN8p{~kI$Cp55Tv6&l9z7zJ^<+SZ?)kH=Od&e`dPh
z0i`E>ug{9M!}`=1-fAV=d`eY=-O$)`nNa8@KO>-~WBd7!N8r7Cc
z*|b3s)|Z@&?lq9#j46Y{AqUEyk+I@5jD?Hp0V5>@{u41ks!Pl5aE$ZHi<35ukcYg(
zlXUeJ47pfrv^ACkLA~GpH-j!HO0Djp7)9V1y3=bWy=#H^xKDUwW(_Qk?7_W79Kx#PI@NCW*U
zeR#G^AveFp0tRHCP7a;VfyuXt4h){eLgIGKnFv2y96T<3u3hLRZhUZLEG|(W3iI2S
zHO#++9)H7)&aP#{rJLkjtAqF8PpGk?m3s?+45i#qymuYfu=?3BidDid`Fw%<5(SVf
zNXCLXAp~w1hF4MdSYvOak3%J#XM1>2U#Q4wIZ%QrPKT&ymU)JE*PcU>@{|%e@7^1pOxpm1+7|Lz05O_8W(5<29
zwW{nPx~AlBZE)!w;K&U#lCD?*P|8Bf;nWVYA{$d5v0;UCEB!Jj?RpT+-pvOqx@$VR44y9a1^GC|NGt2QCWhCB07g
zhPYI88fTVcK&lX|X}nwtawEO^C;!VsKF4k>%sh$$KR2^oW$UI;S>egd>1U(pg?XoP
zqGJV+6cB3RrN4vpzVE&IE@&Dxr6Cs4&yo;CH$2{4$uE_7DeBR(oa
zDdikFaE?NOv6U$grM;9tBymVUEZTHt>b5?h#3kmWG>Z|Mf5%tW^uqyNrYH!b@5lu$
zIip0^=MoV#_`s*}K@4TPP6c3e8+^Vy`H$wV3u^Ul)+RfW4$i+Gn_JWLM5E#1y&2`!
zC{LgL83(;RXi;gpp7`kkx;VeUFulqH937LJ%=m?oa7zWhG3yzSrYr2O9eEaYlC=Lm
z!LN&!sGN2l#BhVti;5Fr_PnT~J%jEg6(Mw`}aQa?vVzrw^A4_e=;LGzA$>hZzLe#9D_3Pud65{MT9=a
zH6GZmUHOsNV*}nq{IgdP&_lT=gr4L6n_yu2+{RtUSme(~*7DV$4I~fQ@6(Ci0bSli
zU~b0)eWHFF>NOGoruavPjCV58pxVsSds($$;WuKVn)3mKOvGQ+{>$ir;01-cbPe#{
zCa!b(bPVc!kV#c>ZVcR@n^e$G4F_q$D|S7@dPvi=;^3;&eH2E$Na-X~1XAc8eTeoR
z1z%TYuDv2p1Y0?0$4+=eq9cmq>2XDmK?2kH?^Tg6f$aT9$^M`V?Qw7fDTQz_gycljhhe&}PVRe(-7>{1!*cf(@^c_xTam68Uc6J)E(vlr;@j*>>jRs7Nr?
zqbcd1DxZN-$oWn_vN?3^Hyx*V=`V26^l8zM3^{)CBI_0ZQF_e$`PDG3$I6@w4}qAp|CE=J6;%X!&&TeT^;{>I@~&~XaK)F&fdRVa)8o8T%I{?
z@WP1mvq5Q2;y5fkRiX$e|Vz;X>zl_5Gvep==D#9
z`eqLBF=qZgy5Iya*2wTz6>`H7&&F8|WnI|9N80Ik(hT3#|6yS%V+QT*qGBf3A|U?4
z9+_Lfi#?rLuGL;54C4!#>~FYq@SE{t`FS+#uw%v``1xQgG*_myTAfjY@!QX$6u1o`
zQ~R}x#1W3T`8vfFI`aT{>kn;*V#s~m(&<{hJE@M}+;`w*J5KNq(-g}&1zO;$P07p$
zkGvuMD?YY5tp;qK=aAWF2O!HLYrM$)KurFde(PCsC_HgmE!Nf{4f}XSrSGM7;non=
zYnLPfAp3wlyN3-}gIjTTDd=bomSOkxem>j@?WOi3E+n+U{*q5G
zB{dKXyghCGw!I$CJ>YZ>)vtt8gFR3Br`w=(f1(u|;WIh1M8Zf;;9C9-f9tkI;AajO
zS}~<8R>P}Y$Y4emK@}tWX77cK(^T${hr`
z2Gls;{cD9?`0^9ZO$3)lcY@4B+u+`Fvc=Zx2!0zKK98#j{A;x8KSS7)Dtt4@DV=YH
z7yj1m{OrfDDB|gg+Hb;rGIl6yKqXvF34BA9UjX-2bsx#Khrnp#?7}zSZ`2D2MQIyo9`uuJ~=SGp9A7$;c)-y{#))V(!>9@?>D0RCz2bkBA3z1@8@9lWQ!C{kRWc)X37#gL5H6*&$b_RcfsCxF7M{;
zFQ`&`)0wDL4fda;cp29zhw17Uh6;L5}b-kx@_Bo{VoJU@LyoNx~2T|gA~Y4LmQ$DgIuN${`J
zxu2+RoyDP{WJQi|Wsxr-P?fIIKAQ7zqBRQj?Bk7FbcH?)(lKrtGXO
zj!b|Kfp1cbiAzYFR4n2%C;`xoL6(u!2hDFOA4@fNN3@a>r((xE068OT-N^gPsBBhn
zRh=G#(0LA(jtg0+GeMS}t>r$Frz}XJc^Cptu-teo67GqV+?rn8G93p_t+dO-gdD8H
ze*+41dRgeH&(Ci5vS=Xpk=elPUm&>N2J**NU!x0dpNwk1v;tq>h2lRRwt(nwuesWG
z9HK9MsM*vU4eXLlLa@v)G*tF#d)2iBapY1z#<%Lg$v;;p-sBJt>iIrZ9n`+anQgM0
zQ(6|Ve=bt=-;+ajGuxt#TPlFyM;}-wu|7y#=48KPN}GjF6&tP&>y-kb*)wTwPdX3<
zi~A1e{1j>(kXJ1~mjDbE(tI-49D!U_){L>oEhJ4kc<+lLA?N-+F!4po3aH(ckK#!u
zoxC+Ezy
zGrf!UoYZ*Te@uYfP|MD~^f&0hw)t+B=R0t1fU{-w`B&7&D6N!WvjIXQU%#G|XTo~7
zj97yE`0&P_;qPevF%%WLdU)o)1;9hKn78Kr8^y9*$50wZ
zPAYiO?Sp2ZI1!epXJGL;tAXd={hd9wGYKly+lUuV>>`d?@G(l87Ft75y_6@yn4FK7
zMeaWiEUbOg#&1#<9}U%?JW9}`8tAU|L+{J@uv{wmQOr#k;&5k1ea#d8RX2w%r`phE
zqgj>&3JN^NCn9gK$BC1d`|lM|-Gr{Wc?N#H!8oeJ_a5uh4G>(brew_va0yWz)pWiA
zjNCeCU@mdOk+J{!HBG&+jKO|B{TC^0FqfgsFDC*Wrr%wyv6F;h4|?V6tvoT_KbSTD
z7Y4=e(k-r42x420&4v{9i_kphT$P}QA$}*%d?Y>g8dUx>Y4}eu7>2Y(%bA`wf&ZdQ
zu159~jxfuX6x&CGVBEit{Td@l*tcJZcWJ%@U#V|rrlh)yiwrDf>Q_AR`Pd}IjhZOf
zR%)0%)l`d{b{6N4FtlQw+~Vy^#YVVew%~iYvp*!3)^_oxjmOok0~K9XP55!fiS+k(
z>To6ZT6j9eLl|=X@S*dMV7MCNEXL&j5}U7yD?|}=V5QC9Rz{0u_}^VOYKk&n$erbU
zlgIfEX3;FJ9%x|TZdsP)lVwYgUOJL|d2Y-uC*
z+WUU>Yjq>u?Y(j}&Y%vn@?0Wg{?LXO6mq^##x~-3ifI+sGYwdSsN}yCy++&`PnKUt
zQ;+EdzedYT)#E*T{U>HrwRmalph8El2H*cRuNg&OkEgm#q^;NMaNY@`9{bKhZ18%M
zmH%}WCi=)w;~`dsmp>?o(?rx@TH(9Dtad%lw?s1)e(5Q?}&953D3j-
zmVLuOuNF3+x?;@Lk_YXwrK&62pF{4(^Bw!9jc`N%(-TdjX4oq9#p{nk1LUpzwC;DU
z49?&0GU8y#CG^`$?Ra8q;p?W#frjPv1?x-e4&tHO_0#Ql8?IX>+ca8+1g
zUc3G;@j?!S|NflJH}f0t4tXeGd!`Q=UF(7)0}~)WSM8bjyxXu){qW!!mph*Io?WW3
zy$IJ5-e}%H=5S{4j)<;MF3c>@n>@i12N@@CD(^Fy;-RaZw_^WVU?U6jAKaCfA=}h(
zQlVo-F#34Nch|QspuIO=%`nMbxN%02z3HX{zHnpt`SCATahPEF-u2cZ*vt~tca1v>
zrcMOBblUWX;?ik3u_c7^fa#Rq=c){`d;Vix5;p^A;=VG@7Xq-eCAp#OfBcx;cjt0~
zv>;S`z}8LAL5Ax!J}jMMwT9KRY)1wD2*Nq_VY+8@a=5~(cq`#1FZ@xRV-cfr5;~kX
zNpiXU3QT$Lz0tL<55H*buRoy_#46NGZ_6HW$aBE<%?dSp(Ry^(@8`CbxoEyr3CtZSP$KD9l(IGQ
zy_t^#&dsihY}kR9w)w|A!;oRGOZP0(cj_ykn}A|p>6Dvcp(UnJ)%y(*nJe{
zo(?mTnA}GZAhE5_?h~-}{3^Un`5o-rHJ4o?=*6TloO0CLEbz-T^?%*96X?&HZUk5E
zFSPFOm&qcs1`?7bV<|td;{M`bX{~J{_{NHJgK=jWXxKkKMfd9`dPb}<-39spxtr>D
zYSno(U$|E|suBpe;;Ajy_Dg}=$#>$nr}EH$*FEZK^eh0<_qymv^ac5PZv-4EA3~EG
zJn<6mDnZ-l4=)~vMxjw9>AUJQFOZW&@#fCyB@jDi{ht{JMu$X?b2>{ak;lfpvs6X|
z{^zl=*Lk9MQEF1^wc9D1V1kET>)+-$;H#dxyW!T2l-fR59BrxucAO47O6$eoiiW6;
zMA#(wGi~uXLT3U*a&C`*TdG7^KdFT8J@P^q{2lDGOA5i%bI$HS_RC-qc7rh4
z6LL3R8;}zpaff<260rxd6z)E(1mw?&U>&^?&`D05*COoG^69HL?JwJcd==Wbv%}$t
zFD5}s)UFb26uvzZv}+1HB&
z`sq4nfnzAQvf6V;RQa&9&r!YzJb73=bI4W))~Iv^h$iCD>tTt>%Jb>SsXv>VBZ;uD
zRMD-sSsw>nn{;RGEbBn*s4IQM-&OWfW
z@;RCdBJ{5}kKgbE6@RZ$s0ZUTsSB58w-UcHe)c!_)f%
z3JEuOVWLIoqH;$Lx|++vj?VNW1qqSRmGdir+Ma^S<$wY6(SA7=e3J+2_;Z$cj0)rY
z4@y_~_|#yOBF|b9i58yD?>P}Etd2*$-rs!S6%18h+>~?GK7q|Fzcr_iD8czzl|5Oi
z3-~&18C&olRw!_PZ7nP&fY3kI`)ceRH>CR~`fI5{5SHk5C9`)ZV*dvjPFJ5?g0C#T
zym_YO33KbtelCtC`0#c+vMvTQ!V8}-|77}j7E2P`)#L;nVI;`*hi7
zQ)1?9M;>5eW>VtqrwO=)i6w6!J^~AkAJ_Yp7mf{DGi|MirC){cl(^(^sq$Cwb*Rqns4+n=(+c&O%`b%Sx}SBaIHp2c
z)$GSxA5x(9EwfjMCI_~hBRYL$Iu8~LpVWwn&w+B6G(u|Hb?|__h{s5;9i+)134Kx@
z1&8Mp6;3}(gdeBFEOMI?;E7fHnlyz-Z1P*+owY&;o}cgg8~fXtun)3Y^BR5v^Dj+_
zzIhT38v}>gMB0krrFyHhGhSH)lAat*Lj6NRZ^Inr?N8R&jmT!~@2V
z?GB#{tjmIHw_Z6@eGVq{iyF_w6SxK+wASB!B+yr03=pe*4ak9!&BL^Nndy*WN~_4h
zAQYB+&diQ>oxsI6{)@p^ui=?Vzb`Dw32>p>_(oBAEOg-fQ91bD5B?Uh*USD-3Q7uJ
z_{q7bi7TmR)ZcrBLsj2bJ0ohLP>Nms+f^SAD7By!cCkquj;qsG|8|EsA~mz(_=P!@2Z__js`Y3Q+0h
zIS-E(pG-Br&4-t@HLg@@5|S)3?;9%^O)%-FNf}8?dCc-HTKvrCvry-}(`5-VYMi>~
z+a{yIjfuG0$oxmOupH!DXze6$ECQv=gX#I8p}>ry{QE`#}_HC~S%GT`x(rk_3?oWTA~SH3N(&_b>|*WcCa
zd;^ZUogHTx*OB~hD~4zDpTQubHn-9fKFsA!Y4YR2A-H<@=j4s=8=zaMB2r`XH;R-h
z8$PAj4|L8CF0`-kVxmW%>N9k5QJtgP)lVLqV9W7$D3$m!nmlGeddIv0%~Ffej|{M2
z3$3VNX~yq}0V@DLpB3<)#8)Tk;|Ih*K5lf=`8A>@Np&dKJ%LFn3!N2@tf3Q^XDIp(
zR>1iIGBwGPcc{0WDJ-~n6on=0W@T-C1HW1XxXGR00LjbM$Hc{((dcOsjsC)CAj#od
zT*+cFAgQF9u~mBq1Rf+!%%{8q#iVZzbB@#^S#E{9QhQ#g)=BrZYI+Hnvt#Dl&r3qS
zA6{G;avTA10a@D%G^ywTSqW+6&m>fr8ZHdzvOvI*pU=$)deN8AjyEsm5a3HJU1Wd#
z5I8Zt8;IabMcE9W?wk9C1DfXD;QRDrs7|8q()MyIFmHA;xS$gW`sH}`_w_Q7gKUXk
zk7NX3RQ#3@w@L6Lx#e}4^J$`LPc74%A(ueUoe%ySuWutL_%TBBY%Vyx+t~KKUk6MD
zM|^c^*IUWLs>-Z?l?!Faa}>=t+0b*RN0|
zCRK$@B}-+OCIC1QZ)WlLh%E~0ZvNzCW()R?_mpNmaRHf>DrQU_^(fp-Nz0+R00>a7
z?OGqEpwu|0At%KIa3Sn-?;%+s2$mPp{;V*DHf)TS+E#J^iGKM9SdxaKds^oev{S$;
zSvGd8%?HOlUX6X!okMKO6G{%=|Di&LR3f9mOJHi?N@K+hPq1k$t1MWUjy~&F6O35^x+9b@HFh(WyZ>Bjyf7
zs>R^AsshvSZW3bcGhcoXF@m-kuV@^4?}8g{RNqI+ivhQ;UhgfXadhQRNGWCUGGL9b
z(}pw0a8x7xVD|}TxIrfna9eN;0n*l_b%lF?zq6^roUn&Hx6*i`*oYR#>MdwyeB**m
z(E=k?4m{}O!zr;NFIotDxF0dJv7>13S^)cH0{1>ki{s2yr5)66GkH80?^yENK1;Zj?_%WhXbpvbzB_kN
z{0;SfeXEssat)|Q&+cJUX?%R|S5-8JEUXsqvOlRnijQ#1xG<&4VB!{ynlkeXa7Kwb
zP>zVeRk011xj5>ASGd`})<>Vh+IJRR)-vU>ZlC+@RdOBpuz>Md?tu<|7!o?@Ipv3o
zevIAT!X_}w6pt)Yc;anSA{*}MM|g{-1jQ)XVk~sy)}`lnafY$SJ;x$rXq9*I+up02
zxQ}XCk^j8{PI9!1{a|PVqY=-oLmn$dTM*7Qz#Bj8bzK`;?uA&L{=U!Ix4D?U{EgNViEL~&k+V4#(ttHr;}x2$tFe~p
zXTyewm)K$PS;&w~F=lMefx`#+c(R~RPI4$5Uj5Lw^G7!nUfJ5P3yXG#gWtt%*6c-~
z=-s>XxnxSX(wO#T+<&g{bkWtHwei-_s%#@(fJXxorE*FeEPV%w0p2|W0;lm-tp24`
zk{HbQW`d#bYy?&eV-Tp1ypQ{q^vC75Rq#;*6Yt1NDzLVb8P?KeQw!w-WiJ9xkFw1O>|&t
z{D<5N2{+*5_O@`>>kx-W3%^`pC-|*ze)s#m&JL^EZ`>I<*$tkGnA|OUdJJEw5&qc3
zD1_q%k2P7`<;K@-P2|hHso_#S(dA|GA7GL7T{UAC!3T0ZkSNwm4L)MqyZ(em8uIUs
zzKC6B!>?d}=Dg!q#A(*#FsVuh`8W85ygf`IxnE4yW2gc(z+$3yjWoWWmApywff!Ps
z>8`2Z<%d0_ynj2Dtl`wp$91u*=iq)xjiUQ5uAKjI7r~y7TCF^Fv9tRTVeym-#|Ep(dgXNJbHfWdf|PZx4=j$
z;H{QqI|!}3wq4RM2*q?Rzmd+K2Mkj*a^;`@qAb(%-BJo`po;qEspKo;;QfeI05M^&
zaW{H!u%BZcWyF6?70KO2KWB|n*$Mfg`q$@8y;@_D!lOqoI(&Ojc4%Y3_PYk4lxq5U
zFU%iMxI~2oJxWAGX19bgbV2~D<7JUE?4u~4F8bJ}Wd+EgEBkmF#Dd<+j;4~X6!h`f
zQhU>XK47+HP1l_r0oKunFU$TmBXIM(wb`BnxN9yEbJjW*+_|kf(^Vafu5FIVc>Nm!
zMHkv{{hgXbKB)ZDx>+2!YRwo=Tu}(<$OVft-SSa7b*%8Da3}KrkMsOLjrVBhwy9^`
z!7H#5x4sOmQ^3X5xD_|D5@dZ$>fCObI!Ig`ELnAQLHx9i@55_6z@LLNebSUJNUip|
zsCQ*b!f~%RNKS}{=
z@rBaagd2b(VYTVmJr^X_5ldPu_6T@;o!VfaNC%|0zeQzA=)ij)@tA9|l*o7Pz2LJU
zRbW)gYLNE93l&m}J!D$&G@ZX(UF`FOiNY3BkG%>3V*l(zOH}
z_0u?Fm4|3b4*b_hG3YemXeQmAL!&D4-{bmf!9f%oqCEW^Tv=jtsTzEV-ikf_N2JyW
zmisP5-~QQ;v{&Cgjgs6zvQ-n+lwpP-iY;t?-8>G&2r9LB7Zf2gxl*EVpHUEhys_kw
z+kNmtNo(lZtyq-0OmT#{Egr}m+R*A6q!NT^OA#ug+u%uN5YsyA3aH4HKNGv~7kIdS
zrs@B?ijGa(*w8TT1jqGrcwIJFU{t2Ov%iZ1e*5eYiQ0~$t>&lZfs7<@j#x4%D~bV+
zy3CD)`isCFR}_>wp$zjf4JZn(UE_RASE^x}4a5^+9a`YVM07UaI;923X0RpBd%<9s++E-h7)M;)s8
z(w*x2d<`@BWhmSGxL|`{PS*JYiZEs{y3Sov7vHz*kk9;R2&Isi3d#0asNX+2w|XoJ
zA9;34;BVhKSekI>bVoitz7rH1p1ZGzw`~qj`nleQ_H#LL>%SKfA>$4SVTBOYigAb|g>A9d`EQ-miY
zhEq2G5Oi&&D>3YAzHmGHt*OovPn>_)lvqd5Jv@xRzGj@vg8~9{cJDdM;M{3}yZ;gT
z9;+F;Gk{$PW{FAAIJ0X9nWU3sZ*P}CCaND2U(daQ!gY2J3dyTr;NjEAuMoq3MC6JtA@X>y*#7z%>im<*u)%I+`>|i>5m4F-G?t2
z{_O2BH^4HM=TSD_8lblyBZsu2Bji69c&_BU4{V-VIk41@#?Ptuj}QE-hDV%t0uu^b
zV36g4qp@%WT<7bL)KaR3t{%fDE_T$z!`ZEg8~?QudIWW%g+5?7bpJx+6Phw;Y~7a=
zUQ`1men?7R<8Fcq9W$+?pcS4XQ?7klO~|QD+^gc`s)U;Jk7$$g>mXibi`0*3g!r53
zX|+8Jh5BmkGHc>mxcP2ly#ook2B
z9@Br7FUL@jq$T5-O#@t6i`!d&(+XoA`~F;r#L(AT!{Vo48|3dYoe0$J!7B|Vce#VR
zu$Sl`WWmvn+etqCWK(Ow3hg}$1q5B^T=?a=TDBf69xKh0&D@1e={AqwEo#H!x}+)h
z44QHK^3hzXhYi^Mpq-{Gs25LF*ff}3?Z!3tj$an-YR7-W$4PDqAzZBUZ6_wE8Nbx-
z>o9rPi&L`wNB?W?#&$%-e@$jO@LLtt_eOgYlD7%Lxv
zMwb6I;~T$%!$?#b@v-vlnE~y3+{2T6%OSrKTV7-9JMya-D}~s9{OpOaiAMUz6~k8i
zo}HV3JZZomC{<@^#>z15)fZg*R=M~m!)cC>RjpXqIR5ICSu0kjKKL4^T#w&MTr7J-
zU5r^ix;Ir15b~A}{hZ7e@^Si!h)tCrnK-B`pYs@bC{~sjI$vXX0dmGWZ7#T*!r*ZA
zbEjpCar1vC8%PARv7S$4Z2$LQOji6U<`=Iyd_lz8r8r^@7ccF|^A8nc1>UaiQuz#A
zJ(}JUUUwT;DO{-x_$y*dsaFuu6NPY4ge!gFm%zTLj#`vQ|4MI-h9E}5Y0GTc9Xgze^WnqUa
zNB7G4N+2eVjyuMuBnBB@{ZG+($5Z|Naoko^vI&Kf5t&Kqo?~Q|gvtonDin%BHrbm{
zWRL8Y9$9*N3IQmU_=k^0@=U-ysC<8dG7kNf!CbI$wqdOnSRPXwQae-EV}I^TE|
zaukxwXIW|BxKoB>!Er+PL1OjkH{p_y`_k<%xal_Po=NX>%Iqg$_a4`!o`hh6335f>eg{q6r4PNA8;V%i!;sy1QQ~&%s+yd!9ee=fSs&
zm6q?zbK@7>h9QQOC!w#^hQJT94~U@N|JwcICW=^Y?c1R?0qoH&JXGsS~Y_Pdn@aL1r!*1OJcc=f&hNY6&#>VC)-}d<%9A_Q*9{
z1HXf!qvoVe{yaEoHnDsqdnzG$ny&a<
z(*mLg_{3w0L5PaA@MEubC9(?n$u@4XfpCS%8Q;-%M0T3(_4T?|P}tlgsLfu5<`?9S
z_GYCcG+v%>_hS_43+n_##SVfsY*(K!cbwD*O8L;eo`s}xA~4I_R8aQK(k3LhZ#1j{1nMK2ho0AW{2nv~5%U>IC>_$iGL;Ju(Ke>Ox1C7Pz1k{vY$
zqIbQmro){Pmwhaw%NNr9fM{eRM;seUmOg5@KeJD0rN5ZULdSuYLjP+r@zw)}PG#L5
zi1I}r$G_OiNF6}qlC$q^(*|B&~zw+^`Oa;)`m)B|vTQS8?d*9qd^
zx#!byy=F;%>bV0x~(?U0VOcRw5^VF|8=^~+HWwlI9zlnx7b}GNc&;gIZO_TluLB1EtZ$<$QOds|$EA!Et|zt^}Q%OdPdo>jq`lS*`TH>4J1U
zskb5fE{NlfW!HSQKhRD}y49oo0=4X(9{Ur~gIre1ewqxLBdKru?ppiK;Na2S+9?V@
z)U{^#zJ4JMJlUeX0>nyy-$c?|?L`96Y4|SLe#acWp1z|~x8nrbE%jJ_cEqDygMy{9
zXl!dRKrj8m>IxXg(e#E;W%f52=(~_7c+&PJ
z7E94Kem6&r38zP7OO2S|u%2AD!EG*_H$Y*`DZvM8W~Uz&F<*ygwO*v~{Gx=^On#9X
znv9q+KSAP;EDxmjUU>K9x-`~2DyLCVqlsTLmQ5Whu!ZXL?OX>WoAgQ@H9LCWZ;+oexkF!gs~B^8&K@GULh=aoZE
zkSky|KBGnnnuy<&6rH(&J$o8zb3N|Cdabl~&)x^%<-25MVWvsYd4w%#ZAS)Xy5fC!
zMG+rdHx(m$DGN{MWNW;ob%ftaK2LlE@i;P$Oole&CVqVOi^L0UV`#VL=aQUiiP^a$
zqL?%8!Ojn_J}xIlz|X3@x{dQh*kdL;a)7kIN*Z~&o7;~eO9h7;=e=h5(dKR3m2U|C
z-ll%c<=zMnbbo#;o`oUXp9RLyxJKyZMc(7LfS}Z?OwADP7Fc*b_7*c&3{>KGXB73S
zf^pwhw06{+A$w9f^ZR267T>q09CB!Z{XsuUf6fNrSC*;Vr&B%g=9&DmqNF;=dSqVz
zxp*^dNOAN)O#jchk}DKShLgC==MwAnB5@q&kHVvk=`iM9YNkd)Gt99MxY}`o2&K>Z
zUlZ9wFmy_SUuC@+)=lj)-yUg&2N|R#lt?`*{fNQu-RIlk>N@|rR!s{m6OQXWvDE_g
zkGIr*7i@(x7pukjV@RL3{mS@;THvd06XRaH7U(@O>YPjB8xViJeVvaimAyt*Ab8$J-U32DcB
z5|8CLjN5UxZfMsf286$^o#?9W!Pw|qwZF+&Gyb9cg!0o+GycxxbKWhd8Pi=hVsa+s
zt!3whGR3^>FdwU@;KTR^?7v!2azUXH+pI;m|Ilc}U7yt2J=rU9?c2OIp0m~1QheB2
zU9uK8{A;2KK2(R*RBvo^KCQ!rEH2hpQaW*Ni;3pRp>~{jZuu_tPAiV*Y+pHN(To#!
z%M`r75OKWV1YPK*c1)?oHIx$8hS|qWgHEnAbn;-@241>`51u)(*15`WUZ7x!so4RSW(WMR*`l%J%1
z@U?Zn|0L@$al*(oyR#CD6`tU0O*lJ!_+UFl-m?>a2-+(I3jR^Y4V=7GK%
z1^D_?6tmS%#HXoNc_eq5uz2j>2aGRkaPgXz)bhuC{26vcI_0L|Gb%Icb&f>**(?9U
z=57;iV|0HyHc^GO4ylNW6lLO-QkpT*1JRhf(RVH3a1##TzS`X(Uxzy^TE!pD=i_HQ
zZx1js2I5TG2k`>~TJVUx^uPYvO6=XNBCc;&h&^gNpS?X1f$Prw$17553x5>!I&jV0
zgVN|DR97y*3p_oBubW@vKJLiXW#$Hi?>4u-a0_JIJ;sCmCqSIix?NdR#Hl*JgJqjK-$~I&NK3`
z@Xm+Xg;WQadgu9>8<%#FH<|Qx?R-_dO;IF#^@u$D_<8yFR^3^+P4DrjE=&Zk#LU&*
zMmjj0;<#?$b6%LURG94fLJ-E>Q?-zPas~_0CC^09nBy3o_$=?siukF(HM^@W@-XF9
zX4P0N;0E*;8%5hON3@?ih?p!f8b+?M!CUJUdO2>AxgioPY|#3ddG}
z;3%s*cJceFAjo}HS0vyVtP0l@P@(YxdasYN=7NM
zeF7J};#pX8+d?pxuEtxGigbS;3QkRKAbO$^4cmYOW;|ohT&2o{Q>kw;)wz?N+a%3z
zvHwJA-O5=F!avcljiXo2AtAh6T5ZPDPX==WN(63wUIkZ&mZf~%*N}(VdEMjdAHeGc
zs!4q-GWg`*HQ%d{3YK)V!hZ5;
zQ!n11otZ>DD&Kw7&X0m*`+v0y9)%?S(QX=R>ONYeGb|>{SVE-Nc!CMb2++PCqE^&b
ziKb%cuN~&V=<~B}lyCYIF*Ll|$vH3vu2NjMEi#t_7LCJbH43UxNnL1ZOkO2m+u8lC
zOSK4mwa{V&?HIU`UwN>ztR6iTF+F(u&nsYJl}Y~mOEyUD%_Zw)nE}hRhpA}A?SREx
z5;x&RHo6;-3N$_#p9sf(@g?MbbaQ^wU9lCxa?2G#Ke#=)%%4kK
zJi$wY9Dg2gfA&=t_%7A5?Z+FU^ohoRc%w~XQ@-k5lh@?nXYa21A{8CFnxS#(8Mh#i
z-pL4QWv#Su#ByMqMVR}NV{6(0$gGG9?{JQhX_LJu-7jtvpl0v5czH0B95
zVY2!++?R;Lz1edCkGBb|f9;~>WC9J_ZfW$vYa;`kq
z9fjJKfO}O;7KnpsmJCN%h~e^IR;2k9QGJ?d-u9>-n2Abs$w+iToio;Fk9W8t3$kdh
zmlcA@&?$-X%ts;cc%p0ZS^Rkfcjg=6_8pQAwJm%7nj4zt^XGWoB{cImrZVx6D`~*F`=r-d7(BYbn#gxs0v-Ms^;o6uJkU9H?E~8z
zW%Ob-*oU*t2OZy!jID79N8;I)Ju7a-Ku)mtRZd|GGS#Qrd!x4qm_MuaYhGyrF+H}s
z)YXAV=*+#eF^UvGMizZ>La7ibxcu3T5Nrow{4s<9W4r+fRebmwOstoN+~Y
zdd#I9+>zku)DPjATUAJi&g%I;+bXc?c(?h?@)?k167WcANFEqy9~ZuG+W{4yNZL%<
zCC!J6j*+l%?Z~saQYDbq7rB_5pLZYg1?ATc)yEhFp|XjCFv3rf@IV5VQ=T;t6$l(H%Q@hwFqQhVxOK2bLWBquIu_m*xUFNwC%ShxUOAUS*K
zejA|p@5}3=nHIG7TCDpvX|v9*xb!mm#tcYO_iNnoWW|~ZTZWqj*YOQ=nO%AbBY2bd
zi>r#56+R0X-Zh0hf{rPk6MV&?@EKcA+Y%=gz9r{uU9x)uo^njwlUC!y#$>@|HSYm_
z<>~s$Jo6rubPfA&Yo8Zt$!v{@5`^*J+a7v3Hqvna-CVBWJw=QfPE~OaYGcFBc$500
zRCu;*Kw{z)Cp6=_uX0jP5Dy#@NR$;)fQ=e_?(l{pq-(j=UE0fz@2R~VK!*9KR}
zO4Qi`>!D_co<|#}2C0{3C+awI59TzVjkeHt$Acw*=EDAzL-z@>7f*GY;77}mqZ~%k
zxY|$Q*~38t$U3k1%PG_h$7Qt;pO5(B^^t@4x=%cu&p)j7s^c02vNHaB?00bTT3=D~
zo*iuI9+;fhH-J$mDG%*Fj>OtDk7-2XVF&(^~Q4yk*)?6q*`?(NUs=VGDYdFrlo-$J-qH78~n
zRR>#bnkT62>mikV#Rhps9lRC%kD6&W1dpunN6L9eV?QBntBs3s@UYdL94gu(*l~&C
z&aca*Fo)7uK&#yc23GC3bc|VIyWH>RzTAl*yUhs_)Yh97I1{`ub@liOV-k=4_D+@4P!vcC|AQ?x(z05K4>1^=3&0wllf-zxPYu
zMG3|DFZ%e>#UTR@gu6A#E+jy&kLPTjXj$MxDVDB<=aaEV@C&OwjZB=|D*Y~uw9fHK
z+aHUaFX8+#BEA8S>FxT~fbe+~fOjPxmcD1X-CLR*=*+JE~Y?nrW{7fyDb-NZIAW)0{+f`2-
zQL4q#6B;jXpD4z%rx*YOy_iHP`A|Ev{*9IHX9Z!(#%k4Tb6}aNTjCepFtFgKLMW)Mjh2v+F{6
zacnKFkbTl21FCRR&#H2Ea2`$_G^Myrc!d*=#Kvpdmf(jjO{-sc3h^Zgvg8M2saVsO
zVUuyg9b33gUA&*G0P|+Ab^WuB#avMvbSVabn4j)mYw*fV*y@pSO1b4B3~SywCFg7o
z2bS=|WeRirFSslP7F)p{vT^^8O-HzAJ!x{T^*kgS-AX(B{U-h`6xDE-w9ewbVwv6g
zbj<&uaP*PBAS^yrZ2Ln!)pB)O>SC>D6slfF#4*UqY78fdTe#~J+Z81w&`LuS2+
z?gxhill*Xh1v9i8_D$W@2X-b!tf}TFZS|;-({#^9B4EGCtY|hcr!b|#o
z?TC;3!ogxtJM&L0b_o%Lv(f3qv*yZx%L~j^sstvZSQ544+wG-ygTbQt$|Cd`}g9h3}9(RKW
zz9TEn@tr=YEtEpu)n>hY4kxy29{a;|0w>>9?ww^g3{xvS@1cobXy@9vKHr^P^lijw
zVD14gzU?AG{U&1;^lS#~2dn-E5;ZtjAJZ=(jdiBBV~yEhfZX3;;|wiyHk$<%S!-bS
ztf6>-(=;*`t7B}cZUA?##N^TG_Jc?Ry-Vs7!>Hv#*~eeRA4ugo&*a^;7O>0l^6sL2
z1!-=0P49+wClcx7*4~xbM)QBBT*r!sL1Ai2vA{+%*sd&QmE&nc=VlPreq08uZ}d03
zKmP~)>PcgeD1Qey6n)6mpCHhk;cQ*N+=j08yf|9BUkk1b|5t7Qy#*Ya>gT2<`Fmto
z$iKkvImnNgpFYCoj=qqknP^#MgEPVvhr^hMz(Ok3H@6GfNY46UQp}TB6m0kLySQOI
z=$BOt71V!@d=|ntUogBycRr;>&Xt9uB!-S8QM&|SB4>9mAv+8atoLn8W3))~{v@lb
zFY16MM^lorMh=J_R@3+t7lEFnHx?KF=LE#W1a&rJp906VJKP5Ent`6#vGwev3{Y+|
z9!vZXiG1#SIYjB=0hE0<9!OL>g0&TcMVDuJpii?}<3O`MVhQ@e-qmv-ReZWUf5Y1b
zR7j;EY$J)(BW6q)6rz!fh@(h0*&~!%t#SUki77bh%vi%kp@D>M%}ue)DWUgY9kV<$
zAAr{9gQ9PA9)QK&m=&vSO+<4{+VW*RH~8zwdQp8E0@cKtBvDb&8Ph?aIkSudaiqO4ZtlEd3c&kuR4Cyy(pP
zBb|JN->Y8dC|Og9@d1`Xf6rwTV(qLVRn{wsu{D$ttVZ96C6RX274>(B4E|hVvb2c=
zrD+}NUu_x0b#BG{BRLI(?=sSJUpGGz+=K0IO8Va;@)SzMI+ceK$Bi?mI~h_5!#={6
zRNq^OCulRSw3D?HWsXHO1zq6;;RVs}J9kf_JG>8n^IVbu-^M-WQ&$v_OH%Gh0&+%q
z4K(?M|E&=NdYJq)qgM$ft@8Efzy2X!iw_UJ+ja_I$+NLLW&pilqWG(CG*1+~>R|Jr
z={rH^sBubd$|ms!SCIVe25NBHvm!>UTNzJD+Ih+TV
z=Th0&1TKN7_g|ap!otwSlJWFAn0t=2IQBmz>Zmqw}}vpQnLNMxV|;o(l9L
zo%KrD#}RNd!K-Lwe++1C_gv6Ow?OYM?f$M}HwS@F)`rNEtwE4^hKNRYDJrjW;kgny
z1U#&rX0E0%fyZKRU4BY9GhhPK&*03Fg!4VF0NdRI@h3y!Urt|
zu*Gw*oMri_2gM_;EJE)k5mGNCHT+$I!U%Zc^m1&Q#0j0#;Pz4^&DEH5zTjYHI@mgO
zp~R9;4%@zDDl%{vz;<%u;SVR)fB{ts_sPaRv_YS)Q9eflC(eHHQ@zT8Kgu%c$dn7<
zR}JnLr3EX2Zq}o+;l?&_`Lwp-jj~w;?rg_V3{b%t)>=fKvIZ_LvIOu*45H@X%zk#J
zRvc~@mW3=LgKw2iT*gzIo?87g=W%F2{!=!7I~X(&`t3v%4=jDgwv2O6
zWA_iSE{D2gV7eLQ?BhyB{ApY;@hhi3p4hq?P*!#jlhyAvt%R^cxqVg#_DTU9v3y(b
zIJp9hi9O-rpK%!;>+#>=US`1J2DaC$rvCw^yJWB5q^zN#UgjU}kuvxT!|VL5>k_c}
z@5{CP??Lc}vDNdztD1PEV<+zBl@}O|RvkS);s8~%Pgf2bynrelcMJt$t>F>-Er0mJ
z8K(8Ey&lPUh>L7@rgDz?;s~1%-{AWXvDyXqzty#RxOa@oaVPj3E|Twh`ums-TpUtj
zBXpmGwy#F-D9M@Q>_%GuhtkS$)iHDG*aHPDyU(BFc0&(S*kscuY8$}#7r&OTWmv$<
z!FyT9CfH&3H~FOlV><9o)|M&tmNH)K=nC(7c^ls>`Td#9SOF_$<+T?Yy&(DD$r{h(
zXF#fnf9L-m%!O4CpIA7GX2C;$20FaIr@>m0_L5J!5Ak3cOO?XDI_y_8>&uSxgrQ35K|LJ)!3PPeT}p0iizBxRv=!EIy4U?oy7dY;L}
z{2U}({@5UAb`#p`DJvdVxCQe#4DTl7>p@@c1R?6_G`O+$lg<2G5`3_CJv>Xohs5E~
za0sD&iVu{}uHNMi!u}$%Co1Lx;pXCCq^G+t9AXEi7u;4rV8|8jh|a19Uj
z=|%s&ZHH@qf=*FkN31=hrO$_ruyEn~qa}hS@c485eva1`N_M71^Qvn@3U1hF_vR*)
zxgQmNq{Rq+Zoa?mRUd^hvc^EYm^a6P`q=h?tH_&MdS
zUUK{mT)U3w)xw;h09%;i
zsZ{KFwrl1J%^b;{CpHGGyT|5D{|4Xh=cA3AsP
zFG!*Mb#7ux271xft;tQP;B3Wtsl2a(IN!;jU;pzZ=;$&V*-R@2Yh4F3C8nsbZuxt@
zF@iLFp6M=G,&*En+hn&pFa(+^WiBSi7PJWhB?RS<7^Z8>%SFvm3Oykp)Q2KX!M
z+VD3Z1^>n03N+3ZfEUj?$W+IR;~e&@shs-P@bRmkKi+R;#&Sly$4__(!!v_1kLY6#
zlXyW1Dlz;__(2A^RA?m&l`>-97}T7BsSbgY|K=!gF~d8nn_27V--ednG4K~icOCGx
zb)v>u>FQnFtE7I;u>I1%D~rfhN`6C7atAD^eOaK%o<|176w8;3g)uxB;ws)ZiUf_G
z#5R2(hfg$T`>RCPkcqT!lmB%mPF&7QAm3oY|4qvL<#1XCwlFn5aP=$r@Kwl=S9<|Dn2Cs|9rQu(i@f}HvyhBf%{(HyTEwBDf{yC-_T`I0h8jBUr^N(
zl;+>l1zH@)Iz@(30ZT%c7eiJnq8IC-EV};{d2E(G&q(?JbZ@MP)88rw0yUdka}Qb&
zvwpO#2_nGr~i86AS^Ydt~j#Y!{fjaqTxzxC`l?N=Rk*Dh9%496h1J
zd7!87?v0_&UT`YeH@9N(HR3o!9xJ9FgZ4++OZb0hfr;xDmY+6yLASkYY1)$v^ojIy
zYJhRbIQPG6=jh@==Eq;hoyt8COXh_qS>By!Y0hFWJ2Vi998~LkUJ(yA|2lJ8eF#A}
zA9kKeSky%cd|FNWLUkZQxRY7^U>;b-x2`stM5DBEK+4`C64Xl?GZhcY92W`)-`jwvD19KiH
zU!|uwBUit8&Bb+FaIomJa7B(3s;#_P5NQ~MCJL@{T8p}(c4;#j^J~U{HgxC6=`sQ;
zXudBd&82}J_L6fcbZUaIySfY?%pZbW&?YA&b{WxWoz=|$!H3E(B%~~vk@(Tad;)I1
z_5~~~c7cq;Mrg9|ROreeK=WimCyO&L1Ce8q9`{cvpu`Z`*xEl{z+9`DtyamHl&AQw
z-km6kvcm4HT;c|RzPom*2}`0)n^qnpwX4XGL(4(VQvqk={VG
z<+=S??${8qx~%q+{lO!ICsv^yY$f-IXA(=f#*R1oaftQtQ6X2Nn51|3RGK%jmHC07UOoeXFY5#ETo@}sR{q0s;#Wv4xnlbL
zi;5E=Q%f-Y&w2>)v0#-1rN|_)@Gpm2p#K-b!LnHw-St)C*|QnkPy7!7yG~!@u{sg(
z?={;G7WN9_u@cGTNW)q}^`TC$Q#V_Q34}lOI>SSR3Ms>1753!d;N9bCt|G4pOENUo
z{VVxIj(4LoN5rcM2R#}O^4ks(JKqHSrGER9_>ei>#Q16uQF>|Sg!*(U!O?r2Rj8ab3W!+&`QEX#;^EN*i(qS9^NcwXc^)+}!|#9;N8Fg~n7}id>!IG@57+ahc8OWXgUe`U$-wrE{aae0b8qB+=5B|ElL6{ONXR4UpCa!r^@N#{o0z9j8F~u`HD7Blc
zwV<09eJTogEZ6Xn*i3h`s9|r3AUfOON#*^U=+rUhA6Rk>TvE?8?Z?|hYt(Km)&f&=CLbaL?U2rhop^}G1Cp;&>SyP(aUk#UNON&-4Aq@*<|n$o
z2jYGx+mG~Opqrr#P-~>br`-SgtdzxqnZp@>KA#BzB67pbS)Yp$OGy9x>eM@smue@!
zlRgIKThb`6K1@JbzsF5w&Ei0N>tpHQM=?n2`P<3YA!W$oY}`&*cMCYvD^*3F6%HP~
zU_RG&D+Y<{a=C`=CV=c3M583@2gvg^Y8b9{AzEG?@+>|_RCz>pAoW!Q@X7-ILt8J=
zWCvS?mPawT=rYhW5IqdmSnRy0!V##chboe$e^G~wbBp0!8dx9Er7CS8kLzA*W;AB?
zg6fPX&2-e8pwmrBGxWtSGH~ahpLC;!N4g}gGJask^u?VvKePn#KAqvUU#~h)7B}A$
z_LNZ&ejn7
zi`Zkr;`nPFoHieP?<-xhEN_H8asbOlhIQJW@8FBkrTwujbk8|ejhkhd}
z=8B2i3c|2GFV}Xwju&3AHO-V)r~*&7#7C*a8DZbO$8Wsj$gr$tn(J9RPE4#d_+zzdm4XE;+;N*VDT!T1hm@wAus#<~xnRVhtf`B~y0gRV
zKJjDSO2Ic-1~=eus`IKFMTQV+NW6@g(T8qA&M$90zKnxx*P2C%Bo4p}O6q6X^0>-t
zmO-JE6<5|>ihefp3vf>2_>n0-c)mi3Hba&c(wF{BWWLCU^{LQ-ZE0c5d&?s56Soi+
zb}F062^vEEP)R*KnF(rL=23eQ
z2mkPR(>Bg>2s_C&+SWQRfa!;%VZ@)`fNMf{JL;o8ymP$v*S@7L^t@Q7y9;EXiPqCZ
z3RgLtIpuZymo3S=p%Y3G+VKngcQb5b`0N4bS+n}@nD}>)RsN%_xPl&A>~eqdN;`lf
z-q3ILFbd&?W2O_wR!P0*$Nk|Mwv>2}J=4>`lmVtus_b;$VTa?wD{FgvJWy6aa`|XC
zD|DLr<6rS`3)s=cO@%+-L-*eN#Zj`fIQe?vMCtu)@VaZzJpIu&ATu=-NVR+i^dL{X
z)8rCG&&!m^Te~|aXP|w(ocfjx+O`|2zZp9t8yPGcrL_
zy6lUs+9xo-5Vza!ge@_5=!U?YhGGj1p;1Sn&0hKQJK_-J
zd-f>>q+uXuXuCl2EHJoU3lz
z%#J^D_$dt({L}Ew`}qcnS;>WwIiJyGcrTEjWgT2~$#0YQr-1wnDZ=8p?Z7*`X?5=5
zIFd}cm#Ayk1!8xt%txMXfWS?c3q>}=Xz<6w&(-9ksKk2KJ1DFLl;qPDl`7=`aL9A%
zZsI!fGvMdvZR$d6qJa$CPQ8HBI+zcJ6d>kj8?`Mb%8;77l?wmF0yuErD$0^^7X0_?
z^tp)>Z&1t+rkck!nP@57ZSo|E6LZ3u_VUSV3yAMM-zBY6Gr%2x%X*xVL(0JfmAY*=
zBYvmnk)QIbfWQlS6d^DGitg(-S6Wa3I
z!`VRE>qXU7>o@4iNhXJbVbWmlgzX;+rsp7-cW&Yoy&rm~4uyZIRG@s7I>O?79B>mb
zk~zEQ1@xxKVu(Nd5RI8~68zevW`0VTF=oqYstK$-fNBW(N^QA(w9xk!HyYPc(#{r$Tgk~wol
zn7oGMo7a;6{c2hXeLu!{_aLtlvX49@l3nfy>UtwJPqE$wyv3cZ1q4k*nS7w{=!>IZ
zn&IZUfW9KIt5K2n2D<=Ct{y1jK#&ug|AY
zFJPH95`8Yq6umDC4o`h5i}v0eHxE};0xN9KF7dc2AWqlanJ`Tm*^H8|Qx#ioA-sd&Ftb
z+oFt3+SmhNVWfojKd20TsTCyA^&UpIAD0Dct5cz&VSf)3O)4;(bSbyP_&4$UrbBzY
zy)qhnvon{O%7xD4oq1ny=O~CzVen^VBS(8Sr&^Zmr-*#6oNk&dDkvyYxfP9YqM-ps
z6#eQ5xaT;z9&?HuaVRbGh3L-``{pFRwo&n-UF{<)wu)l>|sxb?GxX-Fe82H?$R{!-b
zVeL`sx5@poCbORmId(>SO$7JilY$o+2&*|u)xEkh#4Rz;`NVn=;(5!1Sqm$}O@A`V
z5-jYNFQs~4n0Q}IP9zt0<#%2cCX8BFX7?WvBQidJ9~k{?@Y2mOQKz=Y!%dra(Oy*4
z&r2zVsEzw2Kk5H#4HNmL5P5&tIftM1*-Af#qQuFh@haqiNEk3;ivY0`yj%Fh=d6C7TLap==(WIt#6Zv&pKuY{Uk7PeC+n|3?NU8
z;S!-z6*V9TzR$jO;Jgj7MXNx*vF;n58@?=|M9-Rl@a)e
z@Oa;WD%U5SxGEQXFTQe}IHvKIPLDKyierrZSsO-%QWc+UDzk6_XNJ@T$M;mAPwR_z
z{I@JAM=CzrRV2{h!@dn!0)1=OieHyD*S!V%o?>=Fhw(pd7`5>UaR~ql&JZo>PyK_
z8^m2aQV+}z0Vc82yrs1CsF6)KD>cXh9Vgn|TzOYOc-@pYD-r&i;BdO}d9=+zl(0Y7
z6}@#5gzcFcaxYnrJVkS=PQOsCXA9~Wlz%03tX8S`xKs1AAb=z!#UEnhI=
zNGfvG_L4$Zuix;tvc3#jv)emD@^p}Orng@{wIPsm8fx`L{TNrt?A?t4f;&{RtD}kNrAg|?
z7@Y#J-SJ#tP^1I&cDe^?e_ln6pKj^R$S~j|lT%%77R=CZ6dro#u>;0iXGCLOu;U@=
zGtZh%3;?=A96v+Dmr%L)kvFz~JJ5bun=jABZE*X|{r@EJCt$hj3tx9-g4M+FXNCX9
zfQs^g#xKM{WT)LP;kw<2nyX$!T@0KAGx1iMJFBIL((%cEKCYdJ>}Obn@t;1x$qk3{Y8_KzxEwzUVBOg2Jcz;0Ni}|-Q
zdOm|+=fn5zv;PNV-m1$}-5dd$A5_MNyt+ZiN=U2>Wiepx>5b=N!zZ>0qU@1)TD+;&2*9A6ibE3=>CDlPCBV`=uN0#8FK~%(tRJv)DqU(usth
z)yL%5$D;5fysjB{TfojW3M%S@tswQ$*i|#s2fQ{^mVZhN0S$UAEgLcjqM63uH2o|C
zh@uF*y_bbzLkV5!e<T&|i7U?Jp9vGdd9V;}+_D#L7l@;wQMGrtWVs
z9!v71S+JdWHjhSZR_4c8nvp}tRRKxgF>vf9thap*rN9e;l1@=0WXJ&K%+Ql|d{BZW{c->3jE4v-m=wF8#-2S8zBVT7#j-bfWn^yhSMFj$l)=8wOBC$sj*mTK9;QjZ37PjR38liEi1KO9cG3gs$;J6
zfchJ>eLZXV8GjTQJMCflOU4%j6#9rTK_WgJ$#i>jt;(zz|I(Dw8ojb^hQU`RV6l9-bS
z+FX13HQpp3kmX`~NW}}SOroOWq&$2W+xNzK_3P+S+S*cRc{TWzms5V_wNKrc$f%J-2ave*2#a^=EFFc_`4lf>gU3@y(W}!D#8*T>*10^zh=$`UAce
zVtGky)bW=agtIhP*$bw&h`Ka_eO^-pwDKRGH+2{y=!|v6UH~*e6!@PEW+F$y1EMSU
zvnC1G0>8MA9#cjgy{z%>>vX8A&A6KIj~@J+GzD&@EoW!+QJA-VwjtH#?-p`<`&&Q@z&XHd?^5nKn&kTuf+oHdgxS
zH9&m&PwMMKw@!lR&wwWJ@FxhD_9wwp@OXl3!aU)q--ANe^)}*mo{enL#SB8a_17AbxrX2v;Bobc
z`x1fu5cnPY5BrhM4z!KCkKWWkJIDNrgNtHV?-
zW8O=(?{V#NLh?U`5j>?*G!NTaF~?11UBjRCSb_c1seiA_FhdbD`)QibcqV(tkg06~
z+Y()0q_{L-ndP?lwwPL+?~(=0A{)?BFl2E6!VQdDHr=1iOaZa
zx2>3`;d$L>pI6OBp`*m*XOlv;P?js8q|GN8W>HJDa;nF`Q=;}SDd{FiT#*z>AGKj<
zmY;Ee!ifkUJZF`A@V*MZ=GdfmJDm$f$6TxKX$(S!@e23T$z9Oz#);FnZ#6@znsCwa
z=_*LQ$U`19@CIh8&OeaL86feR$M5~#8G=9T_ic|i55SUpca%(C3_@gYX7o
z{N&&;0oMh8`99fs2gUo+L&yC`;FSc;vb^X0aHU#8AeHKWTwQrQR?qiOq>!XkcG)6A
zC3~MnNt8rFvL=zpo=UPsR6-=z3L;fDZ
zZ(UE;Di!{p-|54}93ss?V?#@&i)Ay=UJ*H@t<(x?GsA+F-Zg_eTb+Aoyc$5cxyeZN
z(^|kBXkui-Tn*fJ&+4_Xy$1|J|1`aH;a{WBGtKfHZUp&L3`%SHbwDUrHpM=r1{8Gs
z^tU%^0S{|b-7^!Kfkpetm@}Amlf$9(k_
z&r+BYOW^N{<@_6z`QgutZUUP*9s>R@FL9K+sUD1F!Os{C)qurA?PnJO{QgSz&Ln^M
zJBBx^PGdvxe|n}oE^$1os|Ojk`l$@HYeBR}RPFYv3h*!giRB1KIk=uC>1{1k4){y9
zO!ILOU{cp}x=*egXfg}*rn8oUj=+N1SoocI>VpsTpBKLc<65Ub?Kn<=DAQ{`iRtCQ
zwYdPqpLdu21%DrHB0PEIGW=fsw{z70&KxZPS}&V)
zHa#x@Z|!$SiSibJ6Ew%Ye{z?D%sXycS2f=OgZ)O*)Dy*Ed#V}Dr`!S%c8to{$mlJY
z$!?Esg1;kf6z*_!ga7lGG?~A2(x?O+aI_d3jwt{flWcyC9&do{$!)rpiEqKCDp}^H
zmv2Dod^p49;Czrdv*X7ju3WIV^DU$Q&n$2$fo1#Vl()c5$JdJ0;tlXpOZ*!=kq11=
z9{s%Llnp*ngDZ*!8Q_q@(*nhuw_xn?kz`xmHz0s2T*<~g7kIlxOI>0~1A_E2mw##{
zfzG|XQ9pOR1?)eix%`D+gWn4-E_a@1gY5IW9^6(*26~%LXgu$U1?S2o;X(
z^Fj&h%JGr{g(Q$DBeK|i;T7nYKY#1MZ7-ngxDD*)4I*BbQKwb*L=$3~E>VUhu|VYK
zdA6e}5rFt_>X=loEnwbgLhZ-#oVa^!LAUI895G#JI-Fw~2Z|45NY3j-fam%KFLORS
z0K4e!Phls*h;z?9zS-_cChX?^wH|u%gy7}ZI~3a$1NVdH|02$36X}1sq|Y|JC4AU+
z$Tuhz5y1yp_`jEh5P#3zf|sPm=8grvl7{dnst
z;^+6;vhcmNM8qglXrFcwA#L0`;mlD=G+)`}_B*kH_(Qn4>j~5lV+Pm4{_d(H?zPDf
z5(XuN>$IH42g`Rv_PGZOth1E_Ta5H4D$iQt^sD3R%2(xp1@t6wv>>IvU>k_Z30TSqvJmkOQ^uP1D>Jcl#k
zcanI3rNeY?0}=de-p*O9k?@L`Z#?DEKqUIUT#g%RBqp~MUAU3jM6g&qSRWi}CS;q=
z?i(6wA(rR(CQ@5kh-QVy93vlEh-3H9xOm?9Ket|J@|
zI>h>R)DgW*vKJNl>j>}9af-a!@V{4`9Ed_V-wdj=kp)zl(x$(KXga)!r%AenNVqieBde^5;9MMx
zK6|f`IJk?mF$#Plx(_$B6^nl&vQ1a#sog#jBc(GzgRUP4Wfh6#@mDQ`XhrAgud?k#
z;h^nFi+^pzg&Y3n;O{3w;Q-@baP=cGK6v!|O6mvVZ}2t7O%G~`Mv*C5hlACG$_w3l
z<8B=UGE^ADjFrpi)6tJz(8e*<3_0
zsX5!dN6J`jh_89K4GwGjPg
z205#J4McxO8Wqp08seBvb*qwE6~X-SWU}Di*F*x%$17(Yvxx@XvG@F(FNn*ja-K%d
z+`vr7#jl(t5x{giyHhFrf0Y@kBQKq}@`%HRuQ@B;#1i+nEwe0KumPK@4R*xXhJp8&
zolM^-9@J0GA$I%yq2^NwCN`)9zK&SA1A4EV8Rb7$MLZ0Tzf0p=PAp}(Mcc>b6Hhk$
zq5l>dN%-h4Kk2`I1ZZVvhs<*FEgOz5K}&&W@wxX2!y{69M=-@2rC8O6N~B+!vevFQt2lG
zHF03$VS;`9V;+#Y-&fh0QVh77D^epm-+_~ok97;^T!63YabL~c7~rF>a$t@M{twH~
z7hKmhOF`qz#gi+31c+TOC<@UO1R*W27w+;#0=pop~u7l*;c;;Begpse(tRRti~FR7ALE~
z(x-1O@4ah4IlbEI=b>6a`=YDkQ78eNY*RgK?n9nKTz@%WSp#N7_6ud@!2g$?l?}i8
zy&iA|%Us=9PyuEca+5zqRRMa*pRpp5wIEpG;t#ov4S>5Y%R@q}5va6^C*Sx6|9^03
zRQ>tiYOwIm_SAxKH7FoL#oY%hLDkWv4{j_KfWPu5_d&iIVD2<}spUZpnEi9%zVe|O
zFcR|ou7YVbcsc#I`dfAt=vschqBu|ss?;q1@XFMIxp9H?EsyKqe3j3Sn$&{nI_=N%
zKWaetF~zv+-|7K@r>#9Jo1T+5`kw);W;56Uz
znMSbZa`B$riw!_%g7wLuN*%BmDtRyct_FlVFx^af(E_yk_x8D6Yypl>|HbebG=YR@
z&GXOW>w&G_fmL6jI`GNl%#{2E0#xcpn2F4nf$s7B$4`GO1RRx{q7&_21AQx78VUAX
zpeo|=VG({0P>&_UvhySS|HBO}QogRmKy3N%e_~W`K*yFvzYW&8AbC?m7;SqM=ng%v
z{^(jIux_I+3Z!`l+NcJyW0ML&>etj=yu=&O_yUaBRn`DHe!VM^#?@fp)~4<9FDih9
z*A`XRKc(OV2UEkl-NnGi+^58zwGN#3^EU`utO2h^v@S}!RRKi@u}a%|0_bl}<~|L-
zFC!-+@=0xDF8E0s*RlC~78pEU^;vTz9yB#we0xU87i5
zn+hstfg6g#9lHe+K&)ST>BHd1pmw_OndBXPVx{izFYSkK!2ao?zq7ycz)JAa-h(X3
zVC0>+b@IC)_&diC?;b}x;B>d)i0#iJ5ErYd%#`{T6tdn7GU7}FjxRn&JeUpyhflpd
zt7>El8it}(Tx`n#^N5=CczGdcJ))Wyur&kxKIXe^G&dN2hdJa%BCrK2a|{V|beD;R
z&0lC3q~Q09O^hvgTw@7^Aik{}3~5BCk?lLF^%P<&!zaldbVmrG$PXK-e|i(PGN?a
zv;1B6Q!1gDJZaRmau%G8(~&t{>`5rD)bf083?p{0NZyyfA5Z+Wh2IxhOCq$dT>obD
z#1bfRZ>TM~bDLn4x=U3s9Zra}@CQjsClb_8_P$frOeLB>)bn#aNhSsr4g`7eW)nAF
z1v?Ih
z7GX@cC0X%wF5#KTe4?p5mr#gZR}D(aC*+kLxiO@?A-46+)N0Wd5Y5xJ9T7PN#Oh|g
zDLR2dqEE>wt~;)f=uOXN6AgYt$W<(A|J+wV+|ivhE6^$=E}XUXJgHYie5!1+-&9mg
zRNdE&qvLu_EFa&uQ|PDuBTS}4(%u>_A#Ud
zv6{Ujh+t=RA)aF5So`%mHpyj##tMJi+(H@AyVc8c@Mt-)9q8o$ORpdl3DtE&_&rO3
z?GK)hSd|c*WiArEDrH1ty8H9zYy{E2(pJIXQ9<~c4AF&Xln}d44V9>CloD1`w^sP&
z-x0Er&rZ-el@n504$7yxVgDL6UoJEh5c^x)1g{(`B|eX9`c=7=6Wy!D0Wvk^L?fH~
zfrn3F{wZ35%|4=pSWgj8QdKP{EQ@7h&5l$OYkR7F?cP@tOy>otO(6fuOO7t`x&{9=
zEuYI!*t&wayLWf*qH`s2O4s^+%xDeq@(x>nTTv~sxViO9U_}Md!JGC$IHH0mwY;z=
z$^(8k^mAUij9D!q-6)eOPg6%+91l2`^}d`~ITtKhb`O4+snSvUd`1;9mBAByEx3lT
zzfRD-qOB)BR-Qllo4bNwG%rz8t*#^vneMAn8muBXXhSY4Pu3Dek})Dq{q;or4bk|G
zO0~qPsQ16Rd+La-ql#u*d1?uEdE=Oz(mFzOE7$CRV?80Vw02BSsE&x63I1zVRY$Cw
z>D*9Vt0%0CJTJMhH4v6p;+xz%8i>NE=)}g2b;OQ-uLmC28;HmV@0f|eW@3)hW%Re(
zdqQ*O-IXDu_k>L)b@a18b;NEDHc`dSMq(*&OZcMl2jcEZv%lFwJ7GWI;o{)iMKF5a
z$?u+OAa3(BF7xiKCl2ND1FrZ_#QiX4;WMsX#MqXvyYd_QiPttg1vWh+AeVwr!1Djf
zgDCPF3KmkZl!D0=tfOGY|EvF>ok~i1CIxf;AJ$XkXB1?j;13Gkr69d6>0kW~5?Ad>
zbfVOE{Xczfiv9|vT#zC^rN~VbET@!9IFk0{Ddit1*h9fCYqI{&WfF@heoRx^zo68s
zIFa=)5=p#ri^OCr5?v_z;}1!h-<-r}*GR0kBTT7#_jZ`Zc!8d5-`YXqS_-LuYJk+I?&!sOHHD-+MN8`a)FmidB){43f2
zeK)dxIf`sIx&8^)|2|fM(d;bQ&+lH+K96`CE)RJ83*%3_yKo7QfH*2r-agoh%PpP0
zW6XOGaJ|QwqgZA>(}_{0nCy4=>nLaML)$90msMbo{fw^QfVcPvm|0)O
z~)azEF$F1J(D=
zwECg?zl=uLtRYShz3@ePzZ0D`QN2#zh$^aQsd&_hvgt*Eo
z(T4i(Ww-4<2Qg;j9J4M&Rc;TKaEQ*E6Y9bsn$(?J?)wV;kK5$_3!*A#=?Ub&p2Q1!
zbbUUF_C_HL(W~o4iR`AE@67#
z09S~a5v`6c5VyTsK6n$N{SL8iRNs3}TnO2NKandy{@MKJP<;ks2HVeCEr_cZ4vnCG
zx{NK|kp0tp*8Yt^KSN4Bjj}#Nyrj&PDb
zHU+hpF!<(!4LfUkSUCpIJ(zH4Jedb8h!P9hBAl;%S?sG$e9
zaCm)VMf8x~?cIlm5mx`+yba5oZyk}In2zNHnlHz}xioY=uQVPlM{?^Z+cDISnEg_Z
z1H`Ts)d6pahfJ9RP&?}Yl|5)26C9J;sGUjDq7<^LaPZ3*s-HOZ46pyoM>fHu8qUX>
zO*z6GhfX|?&%bbDz00SZr%*ZLKO=m9i=O9yfn>FKHDP3r^|yvFl8ZjB`XdyN-m{2s
z;jPkVgp9Fm#>h|Ft?U&DHMD7DP`O(j4K2dk8*Y~&TQ|JYyV-!{tN%S1`uhCC|ZQRYO?*
z%}<_JyHB@cKhNf#!}~V;BSc&uWkBkeT9N0pOlMdc^7F9VZ9ETl`>e3O@remMe-a#Z
zxE;GYc@9^nk>~40M)ExLP9|}h!dos@k>k0%PWIDtk*t5FNa7<3FDeZt{iIDI&vk}C
zvc9pM#34Z9eC;sq-;bHJ%jrexx8{(3Xc?2|ww>7=*3B#dx!H4vp
zmcqk4OiBB46ker4;gM7nzGdY=)*pCG;>)Wf7EyRxoHHr&Fp=mjPvRdR65A;HbXQ1u
z`Wo54DuoXQ?;*?IXOgH$@x%NgIZl0Q(*M%`DC0~a^*Ch7dDWxvNVtjOadS}oxV{8Nd-fAuKkS`;k#pM4EdUr(RJ`;_ZhPvN~@6u;OM
zNk8?ElX_b%Nz8E}QOS(NE{gsC#TkPX8E&H3uS&{w`Cq&+Kq+^jpg+a00t)_b+%c4T
zc(`CY8I{u>47C>S10w)3LsIa2(Xpp2vGK3TufmBbH}c`%}ke}tkR
zd55h3NLkN+D7;?fI$3^$a(&9xNO|HTdA+lZN%_BfBwnW6N5ToD?3YK*Z&Lv&7fO>@
z@Pec7P9>U2Px`)^mTrXrM*K$AU*Y~nUH;(y7Rn>B9}M5cupB${T^^lxkAzdh
zk^a7Kw3tW8($3@auv6^dFI4`sVWc_9
zo)7$r=jYO7DeiC2?oH&phDaSpdTr@85eTQ9%%Tw*-gw=EP(ksq7ebn3jREA>E#^Dh
z5w92(HNKBzHplu+s6JzBV+fLWbd6h~{tid~-fIMrJ;VDPs=vyd$cgG-SC*tAzd3hb
zRj`Cu5LnUf3-N`ZKm{tV@?4Tbyz#_cS|7wuhqAfc5g&aBvKh=EzR6$HM|vX#y?;@?
z$n?u-V~E;i-L9xzuwmxFONj3MVm*j2U0>PPg?OBx9x!u**d@5R-Us3Yk9PeLh-pQ)
zdXOKM=Y|Nx|3vFe$K4^`{@gdx4f$=A^g%{GC?6VA>hyw0xJ-(4LR6$G>wO688MgJL
z@S=L>@s6e+kl%Otze4%rTdoS%Ik0(>aB-t
z7a=mll|DQR%LTW2J-USSM4QU?KtG-Kwf(0E>mSWkd9%av*3Q?z&^Xnm!@r~J*u3wU
z#W6Te>%jEmf8jXZ2G9Jt(K`a3tPh4Q!}4hTD-Yc-0&8PgO)As0{zu!-jH4ok(Rp4y$F`~hJ{Q{!Ff3Ld$T`UN4ulh{-OKnX8ey2
zXq{v;J6u4rbY0OhT6f#EBFd1Te!q(n(fCA0Rq(#zcduU?jW5jM2uzW|dH3!&_JsI{
zj^PK22a`^#?M3nB?pOwA)L!sco+Z*N|4?`ytph*)FVE3^qG8WQg|6$bwg1FWzdULO
z#n61@%!UNdK|fbLRm2A%cGbi=-G$|Ymp^Vo@vU9jkE3WEB->w{MSf-^@8L!5_kWMa
z@ql6uGu{_>f4-E2%9$o+{zLuWV>_jcvgL`
z2a2EkeRk9%eQrtG?a1GY)uqyCKcP=<*@)U@sm>)J{qD+K>^FOHE1u`na!%}*|F}0s
zc6Ksenu-j==fcsK4{>?(unXRYwKzx~PMykEjs3_Idx!9-n>N-9XA;G@tCT#)b-y&>
zIPt>tJdQ7poSDMs%vG~-Y_FI52R^U<9^bx#>^%OK<%Dp);~t*ZcAn}+BrDjy!~Q;s
zD8lxc9P}K6nnnEvHW*D?=|uRoan~$_x=_+AEW0Ias4zSIZvN%=VCc_t16BYXI!6S
zyWgslF!nh{;`Tl9q@R(zWIW(Z;bXUD$npvXlK%{XX3`%Sk_TBFC+Ah^63L@_Df+Hu
z>sXJ65{EspBl#RvKdH~qKH*E)d97ed*wpQYsA?W`rs<0x@MEQKF$qww3tLUNqq
zt)#zUNu)l%E{S1!Bp!Al*Wn8(a-QsxN&Vzyvfhls&%+kU@eo}kf6VeD?Ki(8^)?NY
z?V~AifztnYImPc@3U6nk%=05kIX7jz1{D3%l(@-mmGonTQoc&jgUJs(kB|gml(iuH
zt)#>YwG@3tO1wB%L+W{Hkp41ilJe{^5_2f?xQmJGkII6KZ{enb?Ux&nxST;^^()d&
zQXhG}g>A`pD-`~}0!TR~gsgv7O6DEZ*hrcA09o(4fs|uqNI7Sm%nNvUkoIea$^K}m
z$nh1`lK0scJzBheWC${kQE-do;am3AVt=XcGUNLFOqm#gD;bBZXOY*Ri|t(7{R{7(o3`)9`{41v-k6skT;jv~jYmuy_DiG*bRm1+R^%~{
zS8Hm(`*X;Jy?8&pGbw^``6Y`l(zDa`&PLcIjcquG~;AAtLe)<`dc?f4MVJ2>N^iBPOXn;PK|
zZ^HuO9m6*=?jkhTabW|2>_^Ts?u>z06y0eU2E&WBwG$tuaDA=GGKh?E!gaV#7stU0Bbp
zWR)real_A7WvKr>Z+|OVLf$NBvcN+J_20(pN~1W!MEIIM;*T6@trjt`{AN(ywkXK6
zE%u$w{Q`Nn&KZ}-Vz9g@s(bA@>L+mNTf=8)H);P-320B~beV|zZ^&21lfK46UhJxp
z!qf!OIifb9yB*G7Q@~Yjno*z>PH$JrFbX!=vfuQA?VL`XNHT}^9IpA%NOwYiE=!y>
zfaBq1*|JLqj-xuUJ00;hwUgt_>9GEwqs*fzy)FGeXJdS*}b8Ch16sMw~)XWI@e1)*eoEQPiQC6a--@-Vs
zXkIp`0rB0cPe%vgbC>@D{CI!CxOBQ-a$~m)!G0gFXUF%!`H@*aD2Tj0(*tv
z{8arOcHRQp?{<|c^PU_515ZQqZ^Lor=?0s$!ui-}d7jf4`eCQ#C3Zs{et=9@x2;QU
z6!daaXH;c1Nq#M|fTxRc=Z2f{-+X|R4n
z_t!^Xpq#$j^I`$S?RSp`)WP<~r*=7_c@X*?F?kaD%j5ZLS?MnQYtPs!N
z@_PM?9+dBj=bb|LQE$r15L)lwKV?(tK|ilK3@3y@`6;_(Ac`Mc{T@k*}&)a7>uU>+B`jt0aWgvElI;U@fdc<0y#vzCbCbQ~D
z-{0wGHi`o_(^5quH0tcc@k1wv9wXYH_AKoiLGjW00*4&9eWgs&y&P>TdZ%_+l9{^
zYr!NXr1yiP9zU;9UNphjTRtC#${n5KsL=QW^;lBMFs_&UwR7c}<
z_~SQ@#&`eM@@^DQ^Zsza`7fbv+pS3dfuiLPq*woG7w-S}mJB=(MGIdszu1<_gwOxB
zX!1FKamxWctS>3+gU{>!!x!**J%8^U)>jS-$LC|t
zxeeI<-q0Z&-~KbaACLULDyxFep4J&*j=cwcn`>9^lZ
z{D{zA&l&TGjvO(}ON_Jx`WP%HG2EIv2?2d5nL_cnFx2
zymGW|6xSbAIft*~&xxm4KC*)s+vi%y#IhGR>1QS-UQwg)`!63!f5s@h`qF!{-y@Xg
zXT70heVz)5jp8I4toLKPv6SZkCKNmG%E^9YUy$RL@FV?P6(-AT9+9Z~hUCY$sXk&m
z5`Qx=wu_MSFy=+hgU%aryxkwk^>W9}5|4|{gVYN>K=N$?<$BzH)m8x4D}5*P?pfKS
zzs;1mVpN>GP7F6DvA)Rx0gT-xWIn#IfsAJ-CdY93r$lo8SuT?OnclpO%RN~%a6i^w
zWP9YG=H^|%Jfehj{aVYyk-6R)du0iI;E-i`@8!sWA!zwrI%&o_$g
zsK|5U`&8`xO}s9hOf<3HFBWorZ=5Q@vY9CxUY|Kqvn|M;a#{+u<9%Nc_tSaiDPI5O
zw)S|x3tDW&_peiz6h`sTUl_Zt*^_+wl3@U{b4{!B2ilMRXn(}{oH{K-JT3?6yLjJ_
zwjMu?^!5ZRze4xFPqrW4$2N&g_n~r6Q@!g56_0OsK=t&+|1{8k5#E!5RiZvM@)7u;?tXQBln>FxUi_;7vj5ZW*4|n`-|VY#gN_u_H(aMdrSWh5$N1ad0({G
z7ozC+Lp%@FkJdbp9ifxEVd?&L@j}dc`hgGer7fSb&bq;RC)Hdl
zZOGd$aG2gfc{=g=dTMlD@3={S9F1G$$3wP6h!x8>_QXakgoZ6mgBgeak)rT+zDZG`yhPuQ<^^CsGtu-~o#
zUmuiTw%oU~#~YT@S^*aq$hYoT^W7MPd^&nx)dpS2tIW3xu_C?&_<8@UfjrDSi`^ju
z@_P$)&(2?v*D9GG;PjUo1q>Fe^zo3dh`w~09fN38+Fiqe_}sj<*=@#Az}UfJadH|y
zS8y6D{`emze=$DwurdyEDsvbYKY1b_nm#O|`$ry##`^bT4I@nL_
zfx#FT7%zB>$n&)2Lp~uUuk$E>6!>uN5E6yF7I2^aJO%k)ai{oxwOGiHYsA0Jc8mkw
zeWnA2qmzIsEb!ieB)A?6Rrc&Hf$P!T;PlG9QILE;THp-#G{`kz*}*qG0aUd7S)^cp
zFQUfT%lzSUC!L%i#%jpp>fM3B=~3`c{m9Vs!=s?O%}!$d8T9kmRLi@x5fH^cdF1;8
zSns>M-R0XTF!7#T=Y2g2yoZP6m$OEJe7VGy+`KWseP+`&Y1whWshCT%Xfg)M&-Q95
z9v=hyD<2l#IyDK-6`fFzs+s`&@nwfI_m6|w31yWN`ePu6Su4DWeGE))voWr2odkc1
z!wY4lCcxwBDT#9pW8iRVvDoYIF)&xYHqzEP3jXrR`V<}?g}=j*e&!-S3c{DpM``el
z0x#-+T}L;-c*~)f=@*=@2y3}7@pYr1z4>u|YV;_4{==T10AL*B`06*q5`5p~?fpEo
z&JW!v;q8X~&u+L@6w)*bY)rDA#VNx$M@s%q5{!?Mtydqtg7ZQD&(70iaulevxNK?j
z8U;c(gO^eK^YBhd(E)uJe;WDA(WvS9SV{u-R0+CLkO;dLul*9=HH
z2%ke`RZ8qz8V1TIg}7f%!TWQP;E00j@_MV{z7Du~gLyN<;NIa6b)m0@;dde()aufP!K^o9
zxy{fpxK@?TJ>518Jk<*FyE2AB(2b?}GPPlFQM5PJ_3$vLb$_yduih|tp^!s=2m0yj
z;P1E%-fwDRSC>x0`#?fl{yn;X3o|~cu)%Q|JaUp}gZG1~v)5o8iZ=tQERR6__B~I2
z?txeudb;H;)TgcoHCy3&E(pAFq78g!E`E6?x(PC7clwncOoB
zW;oaMyUxNqjl_McA=vN4AFZR#a6YWoo_W7O-N9!fY{}rNt@oKCs>`!6Kd@PD%j4EZ!(SGZ+b!HbjKbGGZ;&UQ(
z%WZs~EtSb-pmV}9IUhf7^WW%Pj`S8BhJ4VuSJC_Z6gpQ-8Y)@hA@&{&3q|9zqSo1o
z_H|wJe|>0vwg?P*BK*|pb>1ENIoe%+@EMfptj7*}K@=~`anONC2p{othWLyr`lJsW
zr_!OUpl?v${le1I9#}5)sNd@-#H-VOohKl^Q`s(o&dclF+Ak1F)f+xQe%+8{#?Kp8
z?}g#K-7(XIPBg!q2g-+0J>3D_-v}Kq9>MuPy$`GSIg`b%-8laHRyg(>jqA%+8O&2I
z?7X`b**}yTb`G@*0<+ENJg(Z@a0~gn&tsMf>8n}u;kYE`jfD)F*OX-k1~rK1iMQ(L
zd~=n&fjf8*Z;xgwmdX1k9JtiA-fW1STRouovy@uC_X$M^N-3ZF6>9|kVp^G
zV+#5Y^CYi`7<|6Roh9FY2zI!I`IL^s#{Ecd*uggiA^+D-{CvuSp%?SD#=$3;m+>7x
zOpE+vEI46=^eVL*F+b|Mn*RaGuD>#0BK&ZZxP{j5xjsAi>JN-tEw6~6@?!Zvn~?m4
z)}|K8)7Q%|zn7b)!S=rm9>V-{bT}W&ddp;dq;RAd^GBK|Jh;5sh2-;oX}_^7+435r
zjNdpOkAlfojQ85f=V}~+tGJ#uRSes2+dcan^&dKKM8=7o6(uoU_HODc(!SHr__ziM10?;re0
z=E<$gNq%0fK)w$n%dd;|E~``W_6=m-TOyB)2THlgxX3BM17G(&?@2$^_L6bp?&swD
z^1tt|;&JS_*p1hPNlF#=_wW+g{#+y(cRbx*jLSFM?8Exv9~ZD3de8?WLya=_M^iHe
zBkQ0ho(J~D99%DB#eq>S(F|Xo{E16=eR+I6g7eZz$<=s%OoHZdy-adF_ItK?BQ6(?
zsartv+Hx-(`$^o{hVvL+%|G$F+GBk2E2`hQF(32j`X8aVy}&quukSNmTiotagDzfw
zDqOemdN0&W#r>y7ZpZhjZ}TEv$F05jCdl5JkZTEA$E(vvaJj~ONEa$kqj`$&Z+%;7
zRkTj|H!aGb`;BVTEZ!H~_C(^i;asE<&L50sipe88?J0-DaXzRoW*pf$drk@G$=`>$
z3*!5w#%v7Tzt8O^LJ)suPOx!7dWEdi1_-x=1w^6#y6DQjpuG34<_Kq$Zx^#UKZD9Y
zeMnS6XcrjojrJv>(fexmA!_{XnMC@u_r_GwKC4;y!58)OFGL-_QV8#R|JTiEpYdWl
zvBm@gA9=OLiL(gWxhU5X)ePG|ySPXEE#wWY`35!_b1*+O
z`IqSrJl|d%iu3J<^|@9n1^1vVw>J0sG2}zmM?Q?}K^`^z>f!TXI35`-_Bjrir|;rS
zh@6Gz{n+SC{#VG0e(a09S$YoUP0ws$MS0tdB!`cFF#o9Ad313b%!{|KNxjmCdEE+SN#8arHwuhD`<#%>g3n1#w(-uu_U^nNl|Bj2
z0~5|yv*=v63=8JD9su9pzjlTxvwIY{iSV?((whSDPsO<1GbaGYCTC$O$ZI6TCbse%
zf%E7TeIdws0=Sk=um5YD0?Az44y*=^gQJo|1y}aN_in~?twSNdJ{jK?a&UbH2r8>y
zzgayAUJgaRR}-8Bw(b*}$!ZgzTjy5XsZDcWhI_03qn$Ir<(|DGjl>j?)!%7P7|ej;
zM|tN~Cuf25-Vxg$wsRoT!l6%OaT*9NR~R)eO#$Q5$5DTNPl70kM4#hH!ywhBFkB&O
z7<|reuUL>D1~+{c3LlCMgTR%tYjKdLh4*e^(t|t{*w|)p*$x8+eJ@|n4Z|RrsiB&x
zXb3#5_+#|9eF#KL)0sur4uO-Q%YSS0hrljjE%ye4A^1BQYR@wrgTPZ?Mzk__5KLLy
z>=UIL0-n+rBKy_`fj7;6aFs^U%;U>;!P@8a?*d@dr*CqpblUVrL-
z182)1;0ic9S=}%Q-U*8|t{oT#J#XK{?k|MAC&RvcZqF$2q}92$wQ~S;ID7OyxHA9*
zcFe)QK^_D?iL4T@n}&gRWLa~>YxrJHZMWL8{Qy{RF+K0!-4Aq*U8}N#?evetm8G!_
zgGj}6VOto7D6D4>^7g^?vN>uw9LA^jSewTb;r((y#$x$CydNBAa%=Cvbw|f>R3Z2u
z{C!i2UPln*<8IRWKnn8J*uAH=IKuY(-8Ox)gY7yCPON1e90iWe$r~c6MnPA&!|q#f
z{bzhPUU>}f|CB3F7SZ!5HY)p(!=>;!h;2x24b;CdJivVi#sexX-xJ?f!snj$H?-Q}
zefE3zi*%*EkPqwL(d61b3LXxxEYZ-7f~;XvbLnCYNYLCPI?7o!@pMSNJIa2^IPbgiIqt_!|Un1Xr++b*1X5BH@PjC?B>;kfSK
zcU67@^;D098{CF|s5v_&8Nhj5{dhf$0q!SFTlDE_VI0#%eQuy0mTQ0YU^@W)R{wFB
z8NDyaDm5{P_H#3R%Sktg_Qnky=sX)0)s;i%74bRQ4#^*9PFKuA`>eaNyb#ZCzGXR!
z@;0L%RrOGwM~rs(3)+`oWlM>_hIrXAi=7>!fTnR1I^QIp)9c)Yxa%+csw2c?@71fb
z(4T61OW86gmpRa$5rjC|o;CUmA~&zU0@APE_22PbP;W$--@+OC%jIg?a{!i)ak=p!
zyDxYH$0HyHD!%r1f@rn)O`p?{-MV>|0CX;94i8}-{Nh!2Et0eL
zav!|{(d!5Y_RCsf`ZThW-lZ0c&cBvZ5>}BA!>=0xBqtoZXpZKqmd(=~`5_p70lzQ!
zPhuGJ$EP%Am?uBxVa9yjWs-bvBf~id^BIY*lEcXUZJigGCncpR;`bd??%U!#rNZ`E
zT%OLj75)?#UYDXRvb&ib^{0*ocB}t$r*+G((%r0LiOgV-##NW>r9D4
z=)ijz#|OtAPJc#v$?M;7Jm7RmZwS?U%!%OVO$-9fm@jcNlYCQyf#j)6{p5QpYWf>7
zPdqIA2KyIsw*vD+*9R{#23eiJ^=D_v-ybDz;2?R~Np@Txck3ud=4e~Y7wr~%@VI5<
zSTP@b#+!xh>iLk*J)C8vv3|q&%`D_6_i^@7gwN~J@H~EWx5D!f(#L|^eZQKI`Qwh#
zBdln?BKE0aS$}>X=97=uT(C?hJz9zE^qu7z#N&6`iOWkT~G
zye_^_{kVnrVEBO@8R&k0p?nm_Z6>mjm>-wxw_rYhp^6jdt6k6O51{=(SdIff|D5<0
z9*Wij80X+c<#axqgpgmO6W1RqLM--=+kou%mb~AFu=ZL;47#sv@32e#2hoGikQTjH
zeL?)=1md%u7Y!Yef9CpEQqgnP@F6D_35Xf#aSgf<|K>P`A^Vp;2P{5-c;J+yjub?>
z)Q8*d!~Nl*byz?N+%JSZw~3*8&oHYols|dp?x%|SWm8TK?t^&jceUAm*stENK-No8
zPSpR>ucq6C7Y}w#@8fNv)XOn%PKPR|%
z-m)JBi$N61W*M%FjY`zJ_U
zNs2kAJqiX}qBd+`h4XfcCik-57`WST`H)%6JXoog{c-c_EXc88Hs1Go7;N5WVP5nN
z@|IIPwJcP#VDq8ApN)NUz}#bv!Ci6&%pIHku5@Y`Y`5dE)Jqx#yPYobh-S_KE&&=B
z8|gXlq}X*&?3-C&`7ff5p?L9VT+^|RodV~t
zeWjYonuFiL8e1#4)o&UFJoQp7f4i!9VPXbN!b9sm|kHmJlT4S=n5pLPEW8UT9(uKFG|
z8vu{0^2-`}`+?^V*Qgw$0pNJ{UooTT0Q{cWx|iYh0l@L5vFh%Je(?NnrMuF>e(-do
zD0T7q0kE&#&Y*m`A240c+#EOG56WAdALe-WgVR~k9P>SW;Krr%2AYs>ww?Ow@$pMP
z_~v|1O84Uccs6nA^#b=Ws2eYTu`UPsfyMzN9o+$NjmQhnSM3KUloL0aO!Whs#*<7M
zhC|@^6!kTm^kFat%r{aW830ap5h-p#ec(6!vo$IAK5)|dt7`4k0JtLmRgmr85a@rb
zd?>)HAJ{hgY>uYw13QmCaQnQY7xe90yK*b3A6VFJ391Ey;9$n2U%^Tr2;9@-eK@HX
zc;3_Z2)_9P+_oG`JGZGHw0!+?#(j_UZ%9|KV1j&uJ0;)-3Yj^Y_~u4
z*klN7{s6Y5%Tq*ll|ERP_N6-U@O{pnaV^pUcvoo^=e0)c_ZXaTIa0L^RTAK
z!=}~Ht^?iijWTdQbNT6|kO7~s2yW9k@&mr7MMZf0-3jy1;+bDmm*9T-<&nj}BGez~
zoNWIB<7R5Q#A!q5M|QAQtMf;gZ+{zOItBYT^XOXffO&yFK{eYZ=+|cdI@(JR-E|Xe
z<)HqjJqGIN9Mc$byn*=VyG5<2ROlyv{%ai1kL2*WAbz}++1}?2L>0NB(>@RfwN=@T
z;kXJDlMHmB{vWkpqUd}HHMM+!o_C)3;DxKn*;u&6TXM^
zMAsW($Wm0G7(^ivvm(@f_Un~y^q!8Sex48F1>Uqe_&u7l$qEC1pg*JDR|Aj3@<%j*
znJ90qF`({@&iCfNw@WC$U(a}ipB-ZL{-Y{rerZchCGJ6Fhgd>d$N}<;`Q6!WhJ5d`oy4cgu95lUyZX`v69*H}P|Y*4M1K-OI+0_Q$W@{ZMklq&-b#>HF
z)zNw!U+BNyhvT2w)*n5HcZojxfS*sOsZ{5q`6?A^ra^WjKHQ^2?O4BGuhD?$u~~Hi
z$zthpDo8KlIll^ui`4E7Z9wtm*|v}7st`Rc-{C@fPnr8v5WZ#KH;4S-X+B+M4lypf
z@FKd87cUCKpOQmf3MO2UJu%~*kI{Ptoi}OD9f2sgu{Kf^;%%$v8_|7Qo%J9K^=JOb
zMH}T6-wQl`gz}JGAtLWeAr|g<|N1<{7rljA-Vmch?+K$gmNR<({Cl5gGKpXV>7^F)6qo#V~ejqxiXiSH$Wu#K1JOfqh8N
zJJhKf>e&cN%dbIu0(|Yw>99Ppr6c<+EU(YBJQ4}_kJUw~5Bktf@rh%SBanCM_(vbt
zhxO?z^jheBK@~y2=KrAHJ@XwVK`_r@l=Zp)_!Q)M@?omR(C)>a?wX6J{|L?~Z>XnB
zqf1jW2J-^Brk>_-pM4rsbGZfX#~Mn8B0(_k^4gbq$_wtdcUU&6?ST5!2TRmAZbE)8
zLL=M*?R7omN@ebVd5v^$ePo*jSF3clE9I;!gHHG`BSFYD($R`iJ%!}oO=h!dp
z2Hk`I;64%{O%o5#tsdXq|3}kx$8+_)aiYj5qO5F{6_q5BW0XXhWhASNB%4w~h3rv^
zY-N-bl~EKK36Y(ZtdG4{BBS4P?&tga>-{?SJonsl&pqdP-sd^bJ!gZOX$Xbs%h!}Z
z{Z-$}C)*dne4;OYFckFqYQnSc1+Tyl@Ob)Z>2&U|m{k^V2I*@SFwM&?5Hm
z2$D-rt&*Vw|Mo!>|G}%U{u*IXDx3lNO3o7BXh%>f_5PC&g6jqh0dNe$jxd5(ehgr>VQ8p#rIa1F2fkCXWiX%Pk$awSr=^&cN#*^_4a)!
zzBGb3dW4*BmCvFDf5vqosae#yL*VOXi5c{hj_-l~lOc5JeuHT5_AwMYp7SmCA+yl0SXw@%%%Ua6*!a!X
zvq<)qSxV=jS@ihw<%=iiXAxhvsOE0Z8T8iertadlenh`Eqw
zU+6xFJa0b9$c_X5-nNsnSs+Kb$P)7NMK7x4RJi@*Ngu4M&)vSUHh>sEUw`EQ{^ko8
zeld5a!Fq^`b@&76UL@af=d~DLA3FC#S-Dqn0DbN*@xL`Qh|F^srarR6I{6_^hY0Yu
z8mo!C_6I(ZmM(N;5XPZRv_d02%=b*#(#DN}7wDH+R`?zqK~u*WWdy;m_pU0?rxW~<
zWU>t>;2;&Uvk}pbrX?{CDMJU
z2k-i;x?rfkBS$1xHpm@55?tj4*kzn!?4sbNqqljmZ|20XXuxq%4gIUYgXJ8*q3
zzRQaD0c_iCRD$`_rJITjn7{D8Nx#AYxULWpzy|m{g{vFOZQHjy65Eq~lq2sjAJvrV
ze5?oc6b!n(@DIvu87Py=>t|6W~a;VE6%LVfX?Z)pzV)rF@S^CdvB5r}(
zq&^7e_g`cqo)?j}V-UYD?S}pj%$sV1ztG@(6}x3WjFgd4*Kt1U1pgK6rwiBIdmOiC
zK~L~6{NA6fT(;x&vpKf8?emf@k9-trp4uXG0^f3^Av1pwgCvm<0#{aOz
z#QobkGV1
z+#;~k?;%{5T@-nSc@gjMuh{#59cTCHHvn=?Us%?Mad-7*b`~BdV>eko;q`^jf;Wmw
z0UfMA>Kni~j$N={qy{_u!;judTg@lFADQi8ug)P
zeV~Ul?Ko8&F5@uU$4F0?0+r?i)eO`(eZzO#IeaJN*)d|m0?qQsD2i_NSNX)eM
z?GQ?3{@{0K1@udCU*X^LBj|$V#HLvh;G>7KnCU#m(8-Ixvej?Qq2!SZTj#*PmOa3b
z5DsydM||GBbZaR5@H$)H{b>sYKms0zW}}XqmGLE&P8z*6XldH;T4<
zwH`K40{@eaYo{~SG*ZY-`7uWY{U;|s{N_&J&G}|*b1x@Rn_=tf-^>{#J)ER6Qa6dd
zD*UUA${#`-JlSG38=xP5AM}#0~vP}PIESNzdYriBLN2gIKOV<&Vy}+Y<
z6jrRgXOQe@O5Q!48MIumIpB1}46K{%`hJ|d7a8w2unOPRi_G`GOibV1i*z{lGkVqb
zpm>`LLjn7HQIOJW&Fiwg=w_t$N*l;=DZgd0Ki!M!=0+7-clM%|TP-SGz<;`5(g@om
z^&&yx_?Ln&d(pDmeFyQ8Ui6ZOVUGo4AEN)0u5fgt7b*2;^R)~2q0NDaA@yV*8f3^B
zdg|VXn4ZMdaK-jPobboi-jqJ%%-gE?{Zt>4IKrBI_;Md=@B1-<)kLm7G}^M0mkRhxmY?B*$e%$p*GRuU!3TOUNgd82>_sby
zikB*lf$uy?DQ~#hkE*xQZ;3iGh`3EWKJx_)q50LX7mf(@p#vXP3eS7@qD|iE_bKl5
zp<5p$UmKAI(AGWYC9hl_L|LCpL|H9Wj0A_ujw)Pvbsx>`43RfZ2W&rz`tt
z!V3IV;<IoP*|=RTL!2hrvDTDd*1hfq#JRr=yO%)5-sPd)zw<8DCo*-Smea}1a5
zT)IF)soy!T%;gNCe~IdScVOOe%IzI*&~}I$?pI)Xun2J_399s&%LC||cG*aq@*onr
zcT{|nD$HXtCML&dz;Dj#`R^3?!6d~FkBfkRFYXGz!g1hFZ6d>KVW2lEG(Vkb0sZ);
zA={`6^j_h6waLX|uy0*ESMWUcr=8r7Q!rnYK4tNN8T8Zih_bH)=rLQmO4`G~3#nS>
z>z)CRoIT^OF%0};YMN`yH7K`_b~1qm3iRctSpF<|
z;GI5uf_d>iGmfKidzqkJuMSj_D%yePh3)6!Y=$`9_9p8Lc)nh)mcbt=|AYhAL^{ZG
zsmb~yKwVb8vv@ryY%Y`df0@Q(I3yC%q>z`d9^-ndQ_(SloQ*fh0Nb6f+q`=*4~g5s
z{Swc|XZ$y<;`pG8``8L@fqbouC;Kp5kGkW?hU2LO+i7+l15`QKSa2H3QNF8`69Vt0
z>Z~S<{RSpNf4BbtJpGwVbw8l;!r`kpPJXY#8*gSPPs;YX_cCDkP1hU>xPSIhq6xNV
zv75M}a6RoiKd@nYsj6eS9rGA_#hOlRhZS$v_2cVopUlpM0lw31-i7UHvXljE;Rf|e
zZ7aj}eR{;w7UQZ~9w)Z5rrI}iu>Wl&a<$t9@Tp`M7q;71+nKbm+z6FRJSAZ6V;?iz
z4jD)Isjz?Yc5=KRmh&Fo^9$Q?UJYf@Ir;!ql7;(
z%!y_%ZcnYZdl0Ut*>tQJmU}#&qK(rJ<30Uskfm?g=ebK2pCopd?oya2IJK$Qe
z9hQ5l=NF3Gn_kr5FaG|v-d@KrPcbgOA%*$-y(>SgG0#*wyaHQN18*_W%E5fUKIznz
zIKYwGEh4y{zP>^)v0o_JWVH-`UoP|a4Y-_cWA?Xupg+*d>(ya@Q00Yl)gFKmNQn#c
zxfGZxDgw^Xk5S=yN>Z`61eR0W`ILmeL-eH=uW>zQ)B;vj0jcQ^4dZ#u?{(|aFi7)O
zEQZQJ`f|3`(MIU6u8+K~oq~QDz+FJS6L=1z`!>RVyc*q2Ll1Ik=7!A`Aoo*gG~o}7
zw=@=OdkgTp6yTeQ)9a0hAH<
zBO{Reo36s84gG#!N4d*4$oI)Ri_{_CmRBKQ9?+8B^O^|Al`=nQ$NOJJ{p|nH26DxZ
zstupOc+}q?E0hU5sKoK}i51W@T&v7Y1F-&?%f_|$;ieJfYOX*hnlpsSZ3O6{_}Mp#?$=4NqNh~v-O-t0&V@71aL
z?1uLHDO4}VctpH-8dZD0&rp0Zg+}9AiW>`u;GB|6Lg7XbXH?U7$c=Rxsa=v!X?{43
ziX9#{UNV_NDmvy*F2MR{byctTQ{7RtAUYt$1OA<(hQIPdL7u-nJxa553VEJh`Pw55
z-@TY6re+cTM|q#l@$<+uI=dxLerLoK;_@H5-kvvwC?Y+>I*foni+xFN+&hL2zP-^?
z#Wsz2F1p$7wwpo{HU<30LB6AxO2rNAyNrNpXu;n;G_>dROf%U1JCT#~TVC{|eeDuW
z2f<#BA7qMT8ty~9KfIVnrC~kH=eNgXOCR#REcxl$<9^f}>B7m8H-NTHteifq*N@iT
z9q1Jf2Y-pr+Lo0w{Yb))>(*#mKhixPtC(stfcn~)e&^Npqt@hSnUrJwNQGq~UNXNQ
zODcC&oCeWM@U{}@!`!7+e1Z6@8tNBfbTddgkhiGC!rSI{M0
zdjMVeRMNf-_LBOyLfNRoAR7MY{pG|7;M=pzS(?xvax&8M|AC)ke)Mzwq5J_ff8X3$
zoe7)WxxgVbVXJY3
z3G7weip|0VtfMoIRW58mJf4PB^<^`D653$#yvcnS{s%EyiFXAakwmkV;uZKAmu(x3
z+Cfh8z2&XSZ6w&|;a`X$#KASEf9q!^oz~6~R7mkg)e6S{oh@68b)er$?GtH|gK^H&+$0tb
zILg`Ry2=gfV4BCdtw28*7427V2fpypUseqN$5E*GsN^z?Gi}{_I@7zsAF8Uu?+<*1
zx9u3e4A^yeskIIGMaQ?*Z;@OhC|v1UZTDZW+gx+h4?&LZ*!RILFy7^}
zO}8EZxyjM-EE=d+mv>goD2&gQkDZQqUpGT@xoT?A6Azd+DQQBv^c=HH2*1~IQW%a`
za|vBdfIvhOf9L;98Sjb>huJF{&>wz9`HTf%MjunEqLiTcqS*;RHQo{C-tY4BA20
zRGtm%Q_~DhHGYs&+|HOw1sHYyTmmP^zcXiQ#r7h~-{Lp+({Bm3EW!5VGJRGL*4G=y
zDA(}%SA$Va=0U(~KhPO5At_Pc1`NU_CsWO%KTTMW?5(DE#{HOhaE1_VVGB2w^X0{^ciPdpW<>EQIalYWcat*k1CU
zy{(F`>x;pTU&VZ@EKq?B_kX+UukRKBeNL>~?}O{U{d|=n(4TKe*jBc{c+@mGXk!d^<-$n;
z;my!L4^EF~Zvwf-^rNmHLGD3FPE9VzO`7fPtOeeE_l7X%R%{m!K96aJ{O0P7kd;2~XSE@L}=e_+do1;|lbK3a-~bYgQ!C-x(*
zJ>;J&gmU=kPYF@L^*SCa@ADw{B|0Wq9Wa%Nx3eF{Kg-6sU5~+D2D60qs)D_g35(V6
zxC!md9P?Y;YzRHhzu;&khHOiA=
z-es6uTG&_x`X!pqPXOjcG0rcBW9?yGafzqpwc8N#ULxi5!McA%?w8Lp;0H0J3{7|o
z<(;m2NY&{N-$9Amavb7(##+u^TY+(Xd_w+%9q@xC>Vpk!5Whgj`-dks8T>2G*PwsYh3M;_97C^_FMFiaO`#7b
zSWh%<1^fKevPb&f6!K}~8L1bB?>(2$z??sdR8#8yEW}SC>x@|XwbV(ZU1l$&z&wSd
zcjuR?F@PO7B+9FBd=&l8wf@*_F@=&ST&=oaO(JQd>R0l;lW44SVRsS50DAPbG-91?
z0C^S19F`pEM|~aC&hk_J$oWk3`b&tn;d@PUg%;vp`tH8u$U8HDW?6<)^tc9)VB`s&
zpX>eTc+qE-G>}`px&8QJ2*j7Qx#eg%Igg=YS7EplG|H6KBL#2ScPfqC^eQ3txcP7vRo#Pgg(m4pT+Oz2FU
zhmi8Dp(2hGNtB~DErI@6G0!||bbkPKYgAI5xDC4ojung*zk>g16X=B`W&~;y=!F;`qWa4dcMuF#g;OZ;j)9q;|SnQ-y$BsX@I_JFH)cmCj^m
z!urqZb4BSAkYo7aaMl5M1{d?SrFyW>TfV+`VZppTjP@YjcS_~o16gb@bI
z{w56TwYxR%GSV1I$Lz1fxTAeWUtwTRbyP91s9$pq+R@k}}%y-w;(75-IfWp+p5tbA^dPoJ8q-rW
z_`RzWuRp-^KQH<&D{L2XbogShoa6ddVqanv##6m-AkBDe<{Rb*3h|b(H6hqTzf3=D
zPvV;fxNvzvmlZ~D0LuMsXu5r?6bbUqSaC!JO!)-8^w2=2xHh=OPlHM|aCu$jD$
zlErYb3Cp$bZM};z`TWvlT<)`EU*f-9?WCi`{oVIsiFK4W8y*!{?sdzaYZz-6PZDv*
z=R3a={({60i9}pZh9y%2&VTw?nvfH@bzL35zuvq=BDTvDeJ#WI{Y>A)5^+~PRNQZ{
zeKHJAW5xS;1qeuJVm|_}b$J24Pucn)J#MEgfBqYifN5{v5c@7AJKo%e-z$bfJ_`FM
zLWk_1;BuqnHVfl++t)!QjQ?BEGhJ=O_0fC!j26rHJiGf5^Q)h-SA{UYd;82*EdbC!
zwEZJ~UYUed6_!irAt_-w(TSy1%uBPr);OmD@^Yv)IzT*sDIEHJ_f{v_+c3^pJ9qHzodBL;NU~r8
zx!v2g*!%*th$^30fbw1(<-7Y7@HInoV=s)ygww^Z?m&KS%eDpV*EyZ~O8pOD&W;D(
zcptdWAFSw#fghxDe-256>k|7!s70~7({<*0z%i!YnXQ1F`4Wf(pj3WPb+8aFbeaqhQr0bA%3HeRH{$4X$oc3k7ni%Ormw(m|xV-CeVcSjcg7r=pSdQ
zJMy=K-ttuYahDSIlh|>+%dTt+DZ6&MKedPd+%m47jM_JX?nPA4T6sY~+FEs&+7bS%
zGY%T=1YVH4U;3pH#T5Em8}PmfPnae`KO~kj2{dwcpYHz0z
zK8e)**Ff$?>yeM@ooaq>iizB^O``$Oc!PU*-oJ2
z?eQgR+Eb`$C6S$V^Axh$vAA`UC$x(R>z?d46X@riSMs}I-;F(?i>NV{gw)^e4ciDL
zp|DuHAJV2Ilz#NU@4bQ~IEQMIwT+d8CMr%V$9^UuT>~2b<}wmuNDxX>d`?2m+&d1O
zzD`2(7MJHIElDW!Z)~kXI0-Gk-=D7ggoGYm2(aLNKteUV+RZx*Noe(51j${Pgi@S7
zM@YJpke6_r2%Q}XmGybE|5YHNi?F_6N<%_ziMx*LG=o3j0g3xvC<%@DBtE=$gM<0DjUTQSY`ffCQFR=xxdeP_?8-CI|2w^E#2N-h(8hu;Z8Xv*!a4muX*eANZD8
zEK{S)-2v3)%hSWdM?x~MTAD~UB-o!XI4bML0Mb-HsNTswfRgqN76pL6#FD-H&LMgd
zdgbIRUqDSladg5u*O&%SLSaxU_00hkeCB=R^P40jMjO-Z4eKFt`%~je?hL?wX_oBs
z;=uDHj&8;8SE4^A^e1@`eUgh-w{IChihdjyUIQ<>p5$}zam)Z>*ANp5J}`(PD!=z1
zRRf+q@JprHZveFk*U_B;{-PQ-G*}FNH(%=~4r?>uS2nOPgA-OEZeZs&^MV1SMf>^!
z%Xbo5JQkbU^m_=M-+nI09O7hR3fFGCf**u0U@EWgDhcgA%^djt-hKpTO6K>W&^eL=4p)_SU$Lp%+l=$JYOde|oPsKY+cQ~T;YZl3@>#+|Nt
z#|?1l9~aMMc>h6369IXchiuH~v)h0lhI^}c5{_GbOXD9K4A0LBjJK48`OdG+I_FlP
zzNgkAoi0JStIe+!j(}cYmfoe_2YM=GoBts812M3CqOSt|T4c+@AqM4CtQ)d#2R$B7
z||3_!tZR^DLN~&1N&>F6%nF$@j|4V1p(8j%l>l)pr+prz@qoZ_|
z4^XV%uk0Yy(}~ey7W;XaC=M?Yeg~J^MR>mY;O0`!c|b=?s;YZX&at=%%UVc_@Z|U6
zavFNe^^OBJQfWNL^YP0QQpI?_{V%kG_b?!-YJ(g5(Q6{#@nL(}Z|RYM?eI*Ro9Z#Z
zt^}6$BY-Vdd2QJL;q%$V6WdvZ;jhI1>?-;D_;9^^mN!pezxzGzZPJ)8J%1apv`f<4$f<~42`TGM>i1^`J{bS#7{?D%`i2X+@r+;yTK-w>Go*LWvdyO_wcmPk5hp0^+C*B1M?yGnIj+Z_pO(-T2TV*9DeMh1LK3@
z+>WPs-H?f%I}`KtS9Ce~m`6sh)<)oZ3nt2qVcyJdm{ow|LQUOL5^%fO-7q_hdGe#=
zzd=}j>}sJN9v70#^#XG+&P4Crd5GU9MgPhOZr@AWckRIG_nh8uaeuvKQzk-%T~EFDdax<1}NFF
zrSBK$i3Zi_nZKatF5UNZ+XMe^&EDjgJqiD%BxdeUD%%76H#XlSQFaoEl^@HLZ=Xi1
zqC1|-f?m$GBK@5MzL$T|iRytB_~k0xJnqMiq7dfa$DCPakS2fWo?@tv6Z=2WXEmUg
zN|oy0f*+i1>EBZY6PSnm;&^2Qew#X*;O7@XubL>EZw&3-ldxW0%Kb4B-fM4A%CIf8+h2jIGA}6i@!FX%OTH15PAf=b#y)~VxGLmbtD#~G>Uk=)lCbbFFZ#~o&0R9`smh`^e!V_qH*;|wU
zAMhtv3(78Qh}-IFzp^O};tCIt4xT?Yfy{j_-lMk!-q~6AjO+0@I=wApG6d!WcEW14
zk&{D+-L>hP?&v6zUV5hHOg)JdGT)jn-y275DPG~JR1?VXtNc`g?>PGHXu8~FK8`r8
zt}f?LjH742XgNr^W9a+xHn#k>akO*4bZPhRapZCNu!VcXI69J5n;TIvhC&m+)QO%M
zL&IJFvUIg4Q1OS|)@s}nh^OWHH7c!f^zGluj<{!IXyfR+dCh<^loWWyy5jBx%Kn+O
z&!K)C&2Q(B6R?2tSuU-#et|e?H}-(Dx?{*_V)fx;dEkLfU1R)kN?sh0Pul_9USS
z2Ob9(h$A_qz2}h->mWK@^57Bk0_>;ow=+$ni3IWg(`}tEf!DY(y?FQ>;!@IG{AL1S
zpN7RJcHcE&eaM9G=|28Jh_7H&DPS5z6-kK?TUdt>^ZdQCy7@sw;W9V<^DSV*i!RD*
zgNXHatg0QvVYr5Qa-?eyp_B)Y7m|fxKQgDe`K-Y~v?d
zbAn0e2d{DVQ$G@-z4GGCiYU~NW6=BTJp2#m;h>*YJAg)ieoOidJhv}S!6_H`Z0jzY
z?qV+zV)fglHqHk7K(I~ayodkqYK~cqu~H4f`H6NFF})-d?MrpQJ_7i#M1}XW-vcO;
zeYlzhcP!r`t5`(*5=esxQJA;@77`h?3RIzq)g>`8GCb!?eARfkj&UpMc%rl*}hppT}U#74?y&%+^qRQ_mp9O%)SAAZZ;
zpw9y4i#29po_zM>wDs0H}g597GW4|)W90NHi5i{inkRBG;YqskX_)!{5syM*j
zJ<$4av<#lpo=E!)&-Yh)dS%rB-^>`+ghG86Zg4K(_-mgDp|CP|PaRSBTwTa-`x6s~
z?by`yu@9P{A3vq8mobB!Ve!xSJ@A}uO9#$`L4NF^w^Y(l@2=AOizZOdtB)L&MIoPo
z|7_kq$Y1K?JO2~#rEj>(3Bb<*&vPQ7{MI9uTK#~*8Py-Lzr+4*#V%|IDT5PYb^%^~
zTQ;H&`5z8&OAJH$MbU|`Nq`!!TOP;&%IW_u$96nPT}g8XJiqT#PzfiXT(rNL23&6$
zkQu=CnELCGFRtg#nKEh|$HQXBf5{xuG=2%0Mu6KX{{>;cUg+{iNo<#Wc%*cA0LKn}
z55?t_ujpChdTny!{fF~uee+x~-)S(*7N7)>SxoF5UrPWa|aWf7+j!7Km0(@wzm&7Vmo^4s656Zvz~`-rrpwIso7vqEhM-fY80-rtN-
zmAvoq%)c2zPX4GSfw$5_2uwOd-v8Lyk-X0tO(i*PcbdFU+v`O}vV1(bymOn$`;qOX
zCGVF_^ZYPTkFCCY2+UG&%)<42D;7%Zd%?x=fViH-F?Sqa_fj){f!iZKotxn8zORfj
z@pbOzyF`1iNoHH&G-+YGGyYCC@+|~!zfS5P-tYS>F@o2wb5{}afnNn`u)J>C38J14
zLdfSI{spwPUh;q-ppeVxk|B*K%cztW~Cu09N!Mw|Q
z_?mhD;1NYww1(8ko{&NEP#6IL~e~uF8`5|$;0X*N@
zC|p)e1{63vO6=nyPn*^?5ZFU#|_?_Xp#4dlgxXvvP+HVY!MyL*g93*wv#%ygP;BUqgDh
z9-+eb75}#ulrFU<3Hnm#We1uAew%C4x)=@l9zIH?2e8~g=fyI>s0Uj%;dyEWS9U-=
z@VyiT_X+%6*ftfrc>z`nNC;>H-VXWWh2?*BXi8;5`KN{Xh;zVaDwglyaYUc{V=tax
zX0)C4Mu6i#Zl1$DHzA$23AaO5std6%##N^s4OouvX1};0VBwY{1~?9WqH2?$89YC*
z*=zS@sKHkw?Bk1I*B_A5l$IiD7bSgsrA1|@i%OKCzzrJZ4+GBEXl+O_CQIdyQ
z@iaWYQ^zLG9Nx?Iu4&>=kWY;3q8$f$>)9v4>|lp7o#{3>p}ukxobJsH^FHjG8^tf%0e;c!vGethJz$^S
z(jwQrOTe4aX14~2M;$2+U1JG_b_?AfMA1Hm3R}z<6Bs5@zsB_^-QdsFnzm-UH3|JA
zGmtk1{2>{V4KL0-gg6Wi#>lf*M^Ns*4|nc^eJc@DO~?nkw6DJC^Wi_x50_~=Mxnm0
zTt7XW&0!vNg(nL4>*Xn{90#yRsnTAluHc86@h(lgDL#TsQc|NhVg1Wv@lx-WNUQ;UVQp%Sy}T0+BZ7cZE<=L7yM%pG^G
za}3sb|BP4Og!P|a19#7u!8)7viw%x@Bd9EqF50ko6!}Cb(O&2tMZ@~ZvVQ+y{rk1w
z+-^UZ$F0q-J--6q{{YFcQyuo-xX?ov{u<(F?*CG{p+AP0ZECjqERCYQmcF*FupjXa
z{hRg0PheifY#_S>{_h%c9XHtu`CnVv*y3Qnw7~K*$`c}^h-Bt^h!iw}xE9;@u%8$~
zaeA)TSlY(W(}+-(GuOtDzO2>fw3nkuG|BCv5-rTPzMN*B{Rsb+`SbN?%|bnyPpBA-
z!*$2)kFUQOMaM@^$PLSnqMdD?T`998h=J0Q<4@WMI&s&zZ~wvwQsWm=c*j19*jv~}
z*M>(B%RknWbx$v!Em#11#-~XYZw&k=w#XUv4592*j;oJThM>Imtne?l&L4q6q5^ho9Aqo16eYz}bx=2k@)r@>G02
z;0o(Jd2oeO3Q^n#y&-O?kY?#Th&s+xao>b_OJ0ngZxD@^vnnzk8$@1%mZ6KVU)*6=6fe^V`w8yU
z(>2&UgxdRWTV(PMA+b>AJ3K8zsOsNI363Q2_r+{3wE#bBV!mU4tqS-WI^?5;Jt+zZ8u6`96jyRU55PCyH0lgz_Yh^p66{B_4HyPcqd4mC^bTdVT0sH4oe`!N|xX+)w
zNegm?M(-Gx;QFi%Ndoh2xhxq*ZopO6iQ}K3p7s-h%#R>lsdIZB`-^MjKb4#W!)%hL$^
z1)Ic#C@`Pfoti<5`S1(2Ss6CKgTX0=*q_HRdU+k&^OuXGQ<&det!`?%Zsyy6VErZ-bC>GuZ63`ydg=F
zE(fNSJ)Eu+3+4F4O;YU9KNFM9jN-e=9U&xhp*k
z;uvib=ZSqt#D#kaxsuDB#QMb5I)5Tves8c6%N-FKdFz~zRE%3jCgXq`IdDV9H`)!T*V8Nx2NuJ{8=9lvUZ
z>rpikPJCazbMwT$fC-htJXo)I7JV7U?XobuxgEe=>r@YG+BVX
zZJxZFK_7oRkR*!zNQ+vU3s_%Qg@)M3fqxs+s1VZK$d;V9d`h2L};$+0ZweG3c>sAyZ29zaDZKNwiBd?fqB%`
zCn1mXib1a#H+GG1k02#lljX0a4gZ&?SE9A2Y=12Fx1k2)fGu3->r43N#nAo;@kn52WRUd-#
zjZQsw-wO4q^>G~P1-_eJQ@2(QJmzZNNsHSe;HQrnGAe>~3kJLS8m&a&nd^%;PA`L<
zk`*sbg!OBtLxrci@?hO5SNqaj7_^tF{*ES?N79T)>gfZ&J=H_`%opl?zdKo%73N)3
z?C~j>pNwV|-o^jrQPWwNeS|bi@YNhOkTX1}{o*#vzi66o>ErVzo_o2gO2hi~>`7s(
zcOaL+lw)58d`D~Z8hs1!luNR!W-GAnUGawX2wtCkWH9J_CTav}-1X^D+XUyN#BBa`
z=fo&-QET7&k6{$$zKZbT>V$m&KQhkjDFi>Gd-PQw;Ae3utV5$haQ+D@PO~~Xg3i_z
z8XHT&IfqZC>vzHYZV!DNb=gPI7Y(9H((l3kypCeK4*paZ^WI4#i0d`0N^Cd=JkTpU
za|ajjEDEzn=U`JFEGE6e^fXWl})J~<pz-lu%zYS^xSP`<|ABll%szW77_mDsv1
z@UfuHWl=WZ&#={BdYuOI=eZ!ta}dWp*TWZb^fT1=EN`_he3vF^f0YQ}jRw7Q0Se>b
zM_%TiUtNIlm*rCQ1nSAU$2m+3+Uv@r2RF^)F-*VtMVI+v(3zOo$|noNBlk))4;gB*ks^83H4;IsIpQ46iZb1-V0b+y=+wr
z_vfe173sozHCV*ge1mrT;?pc?whsC!ursp@*6*IAmXWT*_)J#$7j+Qo8<1T(-VW_=
zoTDz9wgk`lrt2)*pj=7CS6_ZWeY6e^ov{N|j+o!OaA|&;g8R6~
z7~gL|KBI>t1>UbofVI4;74Dnn5MuBJRLQYv6RZJVs=Gnm0eWQ4NsulT+Zje206)XEp^Xf9P*YkZWn>_kM(o*<9$RPJ_|L@g?7o^T<4G1Thl*WEA4@L_9|{z
zD?@%u(l`5cxV}dF?w}^zx2Zv{u?R3XyXq4SAmi68WqdBHsP(7p3y|(P68#zTXiswi
z6&&Z9z4>C_>u1i}kZ3^hK@g!OhF!|2!92dynI5AX}1nKRL>oehiLQK1^;QHN(Ps+gjQ10#ez^NSba{njt9N4~IKH@|8-CbWC
zvBPrl;uc?UK260u;m0hNnYYFExGQAR2+LcxUr@yJi#vSVgK?Tm_dN06t;@WPE>7p2
z<0p8n@t%DIPiEahX^QQ)=~}2S&e!jhCiwP9dMk0x%ZE(|bO^gSQZ$6+My)22p1!u(
zlXx$IOEg6KlA1Rm|Ka5VQO`BS7li-%`^00SydP23L^&k4p9Jqbv4jY}_8Gh5L_5*V
zd?k1;Ya#jOEWVtFeGVYEe<5@#8zUe2gcfesyM7cL_k}>518Gk${$%gMj9vL;d}d3YN4@hRpIc)blSA|m
z4pDugKk)XK5aXiSXCfK*+lS_lZ(__UC7*xensbRLKQzgfsJC$yEAgJ{m0cNFE;5&V
z9;mA<`MeM-y0&>4Nf1bR)MD^QmnZtuU^IFD
z74vAEC`UK+Gl89cp2Yn92Db4{@x4*ny_{~qkdzVzZZxSlj^p)$DqY~vS|
zcwQv7e#QvTqr`gH1$_ZMzKIcW{GZaMEU=tY)v79f-sH{VFIc{B)^aDd6I0bbsyP3}
z)>r|YZ)nzRgVXB|ISIdlu=f<Ogt`l1I=
z`_-whG4P%dx4Jg{277vI?gFuXb}mN4=qBVlZ9lYt?X>B*R$e8b$6C-h_QTay`{rQ#
zlFEMSCaxEQ?HsW#cG>%d)CoxM%Bow%?`NT#6NBq{^(Ey|Jn!PV@b5luHwN1?tGL`=
zgI1kwfI_n`I#r>bt)fp^Yk^m;`+jyh2lWa2x?aQuJXS`FJMsSl#cyKR%rvmBRlQG9BD^#UA+5c8U&#EXdb4Y|DKE@7Hmk0
zFFDgO!u`i(2i{`&SV?Lo9M5pXqW`Nc@H@f6S4DJiJw2}Y#WK9lzbU8WqoXjNYM^vB
z1-|b>H-Gmx)H8PWPTWJl=t;qCvQU5brZS5aC`a&Hp2KUnzmoo4P6j-uw(odg6{NR^
zEB?Xfg^FC=#@hfq>G7{yse^$0O#+ezkpHJb>?8hv=b3uIR{VYk+*6O^`DDH4gX>0s
zGVh&baeHV!;*)v^7-8`%i~;W3(pp1#AF#>w+ruQloxDf>DnUIPDrLkU4njMm<}eIE
zyD?v|3da89>d+P6d%&OHnYlZag8U7&UDOkRPq!M=1pz`Gy#+0n#~9A*IrQAx#=@Z$Ld`u9@73
zf_6@bQ*7*jG%e2$2Tv$pU!$F05%{v|mp#w%d|Kub=_B3;g{E!aerL$%5)FCv7~Wg#
zS>zr~z|KgHKRE8COSpR9Ilu!NB!-=Ues|7uslxsA{NV@idGQRpD#hp_Z8Te3iNEK^
zj1(5Ue!0*@`CbaHM=&c{;c;=(?b{|ie*HvpR}KT-cNr!2kKLsy6_gF>kIdJGasbr@
zn^8UV^E>YhJe6TxVz~9k!L#t(pyZXe&G0=^`{y-pLHlZ*OK8CVsr9lYXW;+D%D>5)
zordQ|Oh>=N&u0xT@JxX6z80CKV*iWH>{=C`=kSJX*Cu$Ss&e*6$bXz@F^cub-vu|~
z+%C8EJH^G2etjxnCKAx7Hb6-oaJZ;(A4VPn@u6Hm$J%mX
zeRe6NJj1;AOWGwx-2PVP`-ZU|6BG(f#CVF*o%bCKAHcReR!yvbKdz>F;W&b3Q1!*d%Y*;@pE`z^Je4xT)RhIIKS$I7ZbjIJDJfCr)iG7kHz^;!+Qz8
zo6}j718sRZO@P95vjSeiARd$i!$&{Z4
z4&$`6FK;|Xmo-#{k$US8!PkGTd?w^99x@UBL(%xMdYpgvVjhvNQrSt^Z%=AV!Vf(7
z&u|*&Kbz6X#&~K!`Mf(iiC7~4#HBGZAI?!A_(t89u=i?FTL|7R
zR})F_Cm$Vag8%PNaw2${oahCDZz%rMAoz-f@>4=y@7oC?-z5JVk-z7^bG^+dO^JGL
z=hY+XAv1fF$gkZLL!jf3Fp=M$mr10LzV;@{Yez{$`5CJkM1E~vK9PTRM>c`S6?_R^
zGr@A7kjuU5LCBrc3McaEOY;acw+kTh#gasbe7QC<&wLwFMdZsfkmGi2Ufv+`Tcye8
z)bKKq&t<2#A*ZWC$UK)yA%eJ%?@%3q4vz*1Ol~0i*B|kdI$$Z!4
zKA9i;ERfd?*8b!5vQFguR%qgMEd{o
zRu1HJ=T$PAS&|XRG$A+qA71zmL3t3@CI72W_kY)cOcK{&vPa4hslI5jD
z$?X^T-+ajJzjllABG
z@xZ`nVlKWonao=}xzfj})liWucbq%Ew|NTa#
zZt^@K!@ri~YqjVUNt079kaC+f?<6VkJ|^d<{O?@(K3!Q&##bKY?z_}dSCacJmd`74
z{t&d{%4__Z+b))!I+?$xrj`5;KWTICZC)wVll2wd;O?UuJ-Pp*ehiTL4=&{T$JNde
zj3DG~6d5)o=4a_k1LXceu6r@L9%z_nG?V+2pL1<($^Ft<-x+Ble%{GMavrks{Xp)E7RQ;A{2q4C
z))MJOJIdtrNX$_K(W$`ld+#^Vho=)Uf6&*W9=*LOk>r|YPvK`FA
zi1jYGI8c&^=h+!FycDU~TLbbwczjnSxsK>r`&kp~SWW57B_jXUdTV>)y{VW#A;@B>06!#_48Be`86LN>YzbQq~n%*+A
zlSrTH=-`xrdeF4H5D}kie|$d?4*&TNiT^uZ6rxV7+eS)jC~8O#W_i2MpguGbp;0S=
z`cvG0UOaIa@3HahX#!W|hx`Et;=6pj_eahcX?o4Ccf@)>s@uIO8|k;el)Mn6=PJ8}
z2sv$(>Gs4v>ZHzx&4m4l@83>4f?k)o@QP@!XPdMS(XP4;CyI&te6{>zrbNFz8t(Z{
z$f*+{(@y9$UwI#U6a3`P>OLd(TivfZ6Nq}Byz;$E=oMo{Ayve_mA9t)8gc(eS*0mU
z`0FU%6>usWsli0wQ$im$q~7WH4=MkxrQ+w1e)Vx(Xo}QroBA(lr0U)5{?$m$4ArF-
zkU9mn?jz*CSv6)6iL|_Q?HqA$D6h1HPJFjomwmiMe}u}~C^#ZraQS;2p`ZCSN>&l=
z`6=AiMf9)!pVtQo`NsNmg$Q|*L}EEa`;?VMw-V)eZ+Jc@$_dXZq!R7*HuVc4!Uvb{
zp%Cp(ObbgS&Q&L+jn5k(?Jla?e}l1;{QidrN*suBvFQCdZ9<+8AD*TWa@uW9mLupM#&k51
zE}wEfnW%qliz4YS#CFl8e4-xl*N^uT@>^KsuOYspVsx$uQBK_N^#($}G|?40Z-O*l
zBl|3&FWx&yg~XSNSyfusovZh*?6RCm!tM1Azl5o`E@GN8xc}NVMv{1
z?oAT>#n~#yiSPf{dj&ZT{FC00_pYqa3==|*xTXIp68&D}(LO
zxrX37C8n9a5ou_sHF-~bB&>_{^Dv`#=?~FgYZ5GHt&r~3aU%8YUaN`>VjR_!@#PcW
zp>wIk5cwadKO^6z)XRJ>C*)~QoYN)RXD7Jbjre~bX_>Q#_%frH_XxfMC7(lx_<+2~
zy+nIxO?vXg_?Tbkah?b_uL=(&`YGEl*PO`z-orA9NOu?%B>kFnF4t{~!?5zcA{sG1
zvhKbrB>25!DxW6m^(%8F&!@Yp#1e^d{j%)$E`skNtJEffZ)H;}d9SWqdT1@d-{eQf
zcVb+uzo+j_l>g7!o}7n{sfv;7)8DR>RqklCK
zc9MHU>Z8UTiR8NG6|GA0oz~MO^KDP(p4T^7HjsLksaF
zO6i+Vrq?qS$#rEw@A-SO-aVR`1Pzzwz6bH0X(rd3q1_2&KUxh}WD)7lUEh&+jdAs?kF(tP
zKenl4{+}zj=cs3*xakgUT)k}0kgG?P`Ebw2U8J~r-Q+H=UgcWHrIm(U{Y*-VyY9d5
znb#wVtk8|}to*Hg?xHz}m+ET6_m7T)XkAK;6^N*Ku%OCi|O;2QV<4i_Q#F2-@gW`*4PwEwT~*B<1?x5sel%Kz)ZjW0h5cHbI@
z#Oe=wl4uO5-M01I+{_qAcnMi+35>)3BxlQP_i<4FmL|I?jsdjAE#7Nxjze_va^XO?
zF<9K^zhD7U3s>nariEj$Nj6h;S<@K28}9PhC4=v%>njzM@g4OBwocy{e0S~8bj4m%
znGT}MWqJ~PFg)p=lU+RmTF%QQoUKOTkAy>l>6THj9+<2C@_iKIf9iQUqFt6r8rR}AW~LBH4%T3h1Kj;pn_L1f*JQ7}@K5n1L-!S!Tw=b5nqFk5*ep%C%UpCqjM
zeHY(dm$ZM9j#`E9y=PAJ*-C$a&ri!9F|8=@Ml|2^uOrUGt5$>+7UKEe{pq{yF{2Q%
z`0JUnz#b^s^Aoy$BF^PTvzxLzhrqKZN%x}y-Wxvlds0|{>s|f9)RJP7UGqXtUZ
zL%@^z@_bhF5Dcm_Qyw_dVAWv13V+4`#O58>ZO9*hH45ufz9a6`-OxRRI%Rwp86
zag7F^DU12tme8Qj$}*zg7ya;USoqX<{s2hN_v(yUJph+%(oS0)=>xyP%aY^4{qQl?
zTFBu(6})9%*3EuF|1P(fRYzv2pjmF)Ep+4~NOm;*@vQp-LVI_#(g!Kf-{}|FivIYn
zg;teUA)c5izghT^g3oaOhQvLI)n7s6&{VsDXP8;)y#1Fdh_8A*Ng+#nz7sr){Tq
zv}jQwVzaPF$xSMlPc2R{zfA>+`N-7U{*si3v+
zdGihQ=NFW2pGS?Sg36(R3so*uc;O*+#aNCCVFI#VP1v4ke$~gy{1i|gF`fTxVLwQZ
zSoCRX_JiP>m=t;We)#nFib(!$949S*y|RrdFnl|1zw(bhcplyzy&Lm)i#(lMDu=k}
zK1cYA_EF)GVp+q|WD30Hk&(Qng?R2iSIZsl>4WBQTmArk^}$cR!h~0vRM5AKj}XeCfKi6O+vA{qsNN``pf%J7
z1*=?*`u_Go)5O<-m5PWvKVjY;jB;rFrn)e0^aItL716IkfjcRVS<^-oxK+*Sy1k|!
zwouwG=A*wFR*C9J$IE^Yy=TIV-cEtZJ0a_M_fg=d=HI8O%lbh<_;KA{i+;4XePJIz
z*AJzi&sDVOQb4vlqAuEy0`7Zwg&n`6{~xKplkfNU!;F5Gz|?F%Z2aihpn>`P=f-Nv
z91wT5W_2y|S05bN8Ko|TeuEl!E?{?X`XRpeL0X&v1;QzzCfWWJP*CSQFFuI=bDEwX
z<#VS1ZKLD@!J8C-lFweT&na-M;#6Q>H{#iDagF7}_H6vcGRknqa?Ib$d=c;7wzGB{
zb5t!$DR2)RxVZkLfJCgmNS6#1_UZni
zcwl{O@5Lh*u)g92w+hY1Dd4M|q$8|^ekm<;VPM!)C6&FHm4pE_K_;`EP2jrh%b;O9D3btF{KF{?=JY03gqa6cOIIb6x
z&X51apqZfB#W4yvvgU&GC8%I|e8a`jvxuL&=>qH8FpgVm-wb#B&s|iqwer?d(XZ@7
zS2@Jt^>z&i%Fai>PEylVjyNBfYqVLosv}+=W9h1k=2YOTO`jAGp+bFfZ`#&3_`iI}
zca#=E{JXhZ+e^0%fKy@O`VZJ1yFuT{ic?g$*wNBh{e%j=F(ui4IA095Uf16xI{+)^
zeY)?7a+;a`Yx2Pa{nMW`-Rtud`DjN6?Zx@7$gbqjk}Wtt#YSW%qMX5@fs5zRPiJJN
zhg)?m;uPxN?r=d|zE^WWQyVl9&qr|ExGc)a_t|BoKH?2}h^e|pqkZ$~-j5GV6lmZt
zG?BRYFb&Qbsb1cxIRMVn@^2q`4M4T+(T=&A0nj+eigQDJpG?~|36)ATFc)VQIVmA-
z;l?znL07~xu8O|3^w|L7Fqo~C#&zOiS>J#V5893M1h3HDN&`uaG#xQr8nBE5_S_4k
zfg(j{3I81$PVD
ztMg8Exyc&QV6;xSYBk~jxu_bq@KgW&yMn>JTv*Nx(ebR
zdE}M+EBurOCN+A09^Sw`jfpUUz8???HJ7+4)k4vILgD_{O{0S
zy)MtVwgYQspd;lxOZ+C7_(eS=ucVIrsd)Q~?a09l-;?u~B>4!9kicj`Ys7Jf@
zd2dHu`>;Rd`?W%Mp&wKky=E0dw2L=+J+uE5NmodPeLQV#kN
z^W4}!G&}&G3U>O4{2qY%i4*otXvZ%;nS5Ilaimlj#wRih20*yU&A&Z!074mw4h5A1
zVE?99+z#8Jop3YX_{RVgNUkni{EY^`m@fu&v<6{y(z<_z(C%F;OdxEp_z?V+(M!F1
z810A7dM$b1i}}P{90gYnLcsWE73;`B^eaMTY(YDEQ*Q@8I^tG79$l|zjN^-wv(nX}
zjRs<$c6z8O4nn2m!PwfkLAbd7RM<@OAS^Na+E)QIP}pzqaGMS46BkPzvk`~uMDVjv
zWjc;q7qyamQiD*iixzVealU5+PTrZq|JA`_O4W4-j`N0V8Sg@AAnP!=mYz$4*NfEr
z?-&h%;56-w5BjH79-3Dh{_)XuEMGZq-hd_Vx
z(igfDLm<}pPoG^Bo{uEf2o&JGR>n+TwPNcal+h3KfB8EI5f_>)_M^WI{=N5ltaP)o5bR4~8bkqm&+BV%wd%S8C
zmh_xyPR}0&E35fqmK*3`JbPS|I8Uhb()AI;_pS4k*ALz5M*nG_Z>~P!ISQh0b<{o4
zjxKwVR|NA2-edVaHa~tJ&l3WBU;f2+vD&HX+AR^dDleY$9n0%-Sxv
z^C0fr^XQeoZieIePD#_sJBTY9`88*`k@pDfN~PD&@DIV3$iu97+aZv0n#amUdob(x
zXrcZPJQuJq80N=&W~-dNoSKb@!x5kKK<6MG{9h-{ZSti<{)_7d@rX~$6Ki<*TLvBU
z%6r6`Sfj8nOFTSEmJWjwVpncx(IJ_hI&m8H@JQdh-lGY0XtAMdEwmj4Rbg|FsBnA_
zJeJ7!J$)2XT(wKH(XKJ_$}ex3^>lc#UTHk$G`_FR{_IRI7=g^h3Evzs><4~Eu(Qo5
z_?|KmJsUj=dW(Tzi!|ip@Zk_cmnUyF}OPIl-AZU2GK5Rzg9(!1M_+E#zg^i*ll`x%Va0wnKWj7
zF4%(m3KxwoVZ;-uu4+w|Uy67vPa2*Hh|;0&(UW6W!s&49mAXOPLlfKF4lNslC}-VL??UTw(3aED-MeiZ
zaj9I}_hgOZ`O6_o=@kq}dT^6=T5=ppHyM^AY^K%6{h=TFPi5jSt@$EVP>
z->@C5&!?9+Bc99AQy=fs#t^T6Wk
zcBYMiM#Q|rn`YxMlpiow^9t*??H7%rAim7!RAxJe4riq}=NzKQ{yOb1)gV!)lvt8_9labH*-;xv-Mfa5;lVF3;d^y9n0
z_mm5w~%}
znP;vNOh|kvefRBe+<&^ueDslKfc>_Kx5MEK*b*0)Wr?_rpEqw+D_?~8g$VK=f%u7S
zMe;dosu^&hGUr_nKNIApriNb|GU4TQ9=a>~Uyu^B{-+6XAHN>DG=E^40nWd}Ui{T$
z!d0b{sSEEhL2>iX0yXsS5NOYLtsC(htD;`ol&)ri*)7#+`!h@^4pNqQ%4S0D^{4=G
zRTeDjvUD7fX2N|J6T7ip7{AmYZCxl64t5^7X8Mc?BC>gtZzix_AAy`m#DgAMYpI$b
zj`c14)qH&?`h(wb;fKX_Cb+5wv^kDpKG&vbIKqVa?e3iwQHVp-dZWRB&V&-XT-$lh
zEV%eNgqe+gOMee-urjY^l;9&0!{p1C?`ocJe&>+uEVm`=VdfH{AyP^(>%Thr2j5vYH;e{1$msy|{;P*oo>(?@w^|uyA|6Hf2iX$>?Shf9usLKvEboD8^
zT>ZiV`$H)dQ)M>lqh4;ayV#)kLPpRA^D8|s`@9GFhn?ITcG86n+S)n2{;%0!;}svF
z^OFtA3`tr4X*TSc|1hg~F$eZ+ZI+Nc#RiFQaj9uf+0fE;NK%8%25Yyk8kHSv=$P03
za7Pat7Oia5%n4`1tsQf_Eg!LgC1*Pq@{$cs#_Ct==W#$+Ur_nvat`PRoUW_G@Ar>~
z+Lzp5Hf#yp7`Xiz8v<+1^=$b$&>2)7F)Yi0E;-H18UkzxOQrgJv0+2;wOP5-18mSg
z`K@R_gAFScOEjL&uwh?MUGNI*NA+)&m7>-hSW-*3l0VFW*2xME`3#
zO>@n+zpx=D*DEmPAP3}w+@<%bVEmlSm^1oiE6k9%bpA32=tTthOUtSZ>%`Awmo|NREKRw&KFL*3e@V}3q?;PR8
zfS?D1`P*>bd)Su(qsLkB$k}{eFwP?lZSTjJZy3-q({FqY^JSac`-kslKp4erk)06>
zw8cF&R_8OpUM?s&4RK^I9ecg~mB%dtvCbrmK5`OHqk-HDRgR;
H>@fTf+||f>

diff --git a/previews/PR1624/tutorials/out/solution_000070.h5 b/previews/PR1624/tutorials/out/solution_000070.h5
index 9ed041a1a9d8031ae21ab4c2a8a1fdf6a152902a..30b44e01f6248d5722367efe2da672ab36139268 100644
GIT binary patch
literal 11976
zcmeHtYao@|xA%iw5|V^S>YwP3U;qHws
z6vVJ4FgSaZbUoZWy&S#0UF?oHZt*?h=(S~5RaFn>oSmQN*ZiIyM~*u<9{S4&W|IZZ
z|FGkpiQDFwt=C@%KIdnozyADTH*eO?@ra|Fqr1;v+s+(s&|UY_0S7j-UN=W?XB(g6
zo{m4)D+GhH>)qq-?c?QZ@8j#`=wJ+5u*yHj`)7M6mm|OIx&LHOcvf=O_&MF5=H*MYHy^`Q`X!@8I))wGI_7lOi$W(I0u5
z`zuexyxu;Jo`1P9`oAXoxd`rGk<5fc^iAT!#`a4v@{?iMEON~B)xTO1p~^|D8IYZw
z{ig$-8GjsXGYP;_h}`}D5td#y+GpFniBR1^cmfxAH?!U{g^v~e_HNQl=
zm|fYiHGf|x`y&^B{`T!;$1R!%h99kSPS114a1)ugAu2~P{19_o;t9yJyVB?QxnZ~g
zeKGq=K+*jU#^(SF)lbS*f&QM&)rZ|dud_1#QQQOziR?6+-@u09PwEbCS;K|l_k14I
zi_ORI7e-N>onjb%BOo$uR0_lW4EzQ1+how~-Ah>LV=A@Daw7}5
z=h%GW*~>yQMH_b2g)&iXxxy*SV@zbC=$LNn$3*vUPcHxL$VBEFpD}kIVxp>!FSSNa
zOtgP*KZ^sfqdxI+ImqX0%ML{!VInYhs%=deF5(;!o?lQd
zh6g@~T%{EPaq>RuxA(&Ev7&FRfheepmgY^?Ft1icwgU&$0eLvX!y4qg9t^eP=@{;D
z`bE0tRSdV`3wkpS^`evGu;%Jo45uV)Xg6Jr;bn@MTHAJD_?>!>3mnIxjw&^_o(se9
z4!gDo{M#^mshH5S^ajnxNV<`(U7QP=djbfs&o;Ngx
z?3hT4L=7&eVxsDl{V($6Sg0@ZO6`&TEM%j*fb>)m&H;h#CeKdyx@di>6J(jiLJE{&tz8rrI_kr-DHPV-{g@PU
zS`fp7iXC>dd$Z7G+xILM)c0Z;e!Lde!<#j9sk(=W66ZU;4J>A%c`lXdt#B^R#7JGg
zum{8IT-PweA&(=Eb&_>;G5qE_hh>cgETne3Hl?!^>iF!5v3gG?x_kxe4~KdjrG7y^
zaGoyGri;DfFkJaD|CcR!7@oJmecMIY7p;Ise(9qa9)AhL6Y8ejJ1Rq($s)$b@yrP8mh}Le#}HsWhwj>
z;IHz$-0(Aq^N!|Ja}MN9`LWw(NjR6=13F&^LOvGRMv}&_vryCdg8L~oOoXeS!QMk1
z#Mhly>Ir9}=Q^Jo4KpA<0nTgrt9(!WSS(5=QR4KI4LTn|5
z3(BZ|;(>mEnF>wEK>v{HsF?P(hdv{YCd+miQ%E5P23gGpIii9GzXYr_~VJp@6Ris4-dFU
z7n-xs{bLVT<{CqPR%SIH8e*coPo?Lwdoz)8jDw`iQzpvRY|W`%hT-<g|`YWje`fZabV93jtYB{j8e(0{D9MI7a@
zV4(tq;*TQW=lXS*=_a@jtSdi9o^girxaj+}OVE#>hX$24LjIaY+u2+E|K<7e=PUbf
z{XY9{x=y^r;e-GD`^O)y*>|WQ+dTN8;9?6uM!aoQDKkHZ;HC2KdRL`L{fg>uc~`0a
zZ@sJ7XY%!b`j?Eq&Zim<59^m7R`eYi#HjIRKRD8Sw_LmM*
zj)wsXrp2|f0U~DRoI{{jt9B>a2kg>%cTSQ4Gv3&3^98wc&7DSqADt5JjbIL;%j$(+
z6~N9e$I$LAj5pt8^c4Ua`H^<71gyDY!o3jijmo0=3&1Y$fh1Qq;LUBT9H*v-kc&Gt
z@-lFNTOD^YL9bIhX}S;4P%-sDHi4YN^TJ>}*gz!Y8eoiomSjB4+ZJ9UG6KBJ?Q_s1
z_<6gS>ts0SuUnD%RS7WP!_|=BVjns#-xA3MbUP&2KMnGudsO4*1K(IS
zRZ5@()voFh$dh7CPBa5jbw59F2l?9d(OFA?3+{9*J_snLa7H)>*0mWssx1t8lUlB3
zX9@LrJ+RkfKXCpK@&RSY!|}O~MpsM?p`_hKqM>|1Zqgc4n
z5qQ7-PQDGmn|8FAt_8k2qdq7Scx`_{Qa$*SinL#q1akjZ5ekAZuS)$LdnMo*J00n{
zfOn!hruaZ#cVxThL*U6t87@)4kA0vWCFYwKCw1Bt_-0*kt~l^#|1Cg!1@PyS&53)#
zUu;ylFE4Ovy5VC4cr&`JnW&RQX6N(wfF6lI_Njop?d9I)*TCPZ=iMaE@Z7smGTFdaTSR27|gRCB4Xo6wC9*pm3FaDSUTqD8z
z))-*+?i0(GgM4+qQBx#vVP5`>aX@3`?lM!5J6?+APzUs|xgkKT@2t~AD+lnlyt>nm
zfy?YNYf=P$TdSpw6Oh0AOy?><_0iJS=P*9`xl@b=xS9Kr+&+*Svciq-0`I#o=0@}d
zHs0GSi1WB!nn~n!_oLWdpMYQR
zt}vhgMi${Ek{~ZE+BqQ($d)^j7eVM1-B#NS$gPvq&;xo5_iaAJend&9moNbXUkpZl
zg#GZc7BwQy)tma^?}tHue_yrX9N6b&yYzYA0G@ryuER@k9ewjO7#FJaG?oGkYmAY(
z4R-gn7s|!Jn4h+l3^sZyW0zs|ER<1Us>%a4t@d`Q4WVbgp`e9|GiA
zaiGs0kkTS-_zLX2y3@Pof__$*@MdEFz7}67mWH?uT*~JV{n?LGV&ELa!_Cs75dAH5
zdvCHB$dh|5;;sRT#tPXJ@i+W%kG=%{oR`|JS_)kAlg983(6>I|^G*ijHVVT=#UPh$
z&wo<_d?)LG$5QAIjM$g=nxHSN5VIU2;&?BaxZ*U7_o(owSOcn!)$`|p{LEex>y;4K
z@B)z)@xaA8!n2eB6-KlbPXluK&zm6fZ6CLNd@tB-yFyw)^yjgrq(dgah332Ga|3?;
zB9-MV>Q*vf&RcchJIq8hseqKu;NSs3mkj4_;#@4O
zcKlfvXZJcwOv8NMr8hRc0z|$e&-R18Q0IlpNFjjxOdJT;CKVj_)p96iR
zxFx^?xQT>mVh-qAKRuk_0oe0YkpCs*tFpgtVfOiSqz$hy03&`u9kuZjS?O
zn3Njq0=sJis+}&7_c_JBq)hOyVEo=+4dmgE6SGSpKG}Mnh^wH_ZADH~fbsa(SASCm
zY}Be3(SbNOPImX-g?L`Zk9}N7tl!1hSR1%QSI>?L;AwS>mBInz!S2fKj>m&r)GrH^VSv4vhDNiR-F@pN4|Bj~;`ZA03N62y*RDKctC$
zm(LE;3I`n4++IZn^mIL0@E!Cvhc?ZzfxHa!bi$9>VU)Vf`N|Z;$!PczO9w8I8O%lW
zo5&qLd?x{G!lb&(V17&S1sn^2E3y^1#e<);E^BXM|2AL_Ho+jrTJ*&q0N1J$^05OR
zbTNld1n@K0+I__PCv_@ApSTyw7aLgq050!%bnFOlC5?MI#QC%|=zFCAIQEgpgXm|n
zbI2pay~=OilW_oeMeCfMn}Mq&hQ}WPv=4h{8U#50Me+P~h~H+HF1s4&^G8mo9Nt0Y=c;
zLyf_&>)d_)XQ6JCwI4=ag?ujymG*fH>sO!zet!XV5&16L<38v$3M32@aq7?a8MsKy
zub*4t2X-l#%?~rkcP;kzjttoCHdq(!0Q&kC)V>?Qr(0I~RfF9!`!ueFAYXzyggU_f
zZb-$nKA>@Z(HCN0Wlv15e+%OtFD_7c2>r^2`13G6Y;|w9C}3Z#Oq3zWMSnBha{_p2
zSwbMuKY~Meb=W{&rJDUs7x??YZHYvl^ULsO7QmIB^cNC!(5r0}PTY&8Wjd)u{+IUb
z$#DjJRe#zl1?JVhloGKR#^coaOHTZ>zqf2B;2itdh#0`Khzbkh9Huyl_`d-?vsX4<
z#C>GXsMvl7hx!
zfi3<|jEVa{)XeI+FmMwq`)~Px;un-6i8{I{WP=g!!CRM3d{73=8$0Jh)Dd^mZ`+A`
zH9z%)0#Rq4&$7gLL4I4AnlTAH`=!#VD!}f(&1`ZYUv=4j*E!%?flp(Idp9;ktcA#j
zh-&d@EuhKVhttIU8Zf>sSz&X5H*y{NR^vkoAek1nfne*s`EsRfk)?OvvovzEL4y6Mg
zxRz771+ZY3MlErFw7*JP%?;;LrDR|c7oh7l*%;zIq{Al`W&nEi=Z2M@fS>)cqNj-W
zQE{RDm<)__TU~!k{BAQjIIZjrx#~
zGY<4Eik%dRdrmsIvqu5+pB!vVYXdzA$NZ#mn6Gi*z|F;w2Q%A>VFu`}>S;=E2fNX=
zWiKp&+kc(NVcnylD0OAZ{--p=F`@Rcx0!|(dN!JHRnbtl*zJ)t^m?s9l!(TPE%;=hB_@7y4A*Ox!*;Z8!VS2ss^lSybq
z@Kdp9BMEK0nwqw1h=dlYt-sbR73$Y_;e^2_RAGAi{L?OxqPMiq-MuxCt?
z(L{6W8nfR>$h`g9`hG4F%B{?cKlN<@X)S4^XY~!B=iDWW`A7pO_JOMGO~5l5swAv`
z0F^9gvrGLnfZ&If2>Lt#zh@g1WTwfez54vHbtxH1zhHGtIFeD%m{moM4H?y28D}-s
zk&)79e!(*p3i8UADa=))APJf0=EZNxXo?!zDE2Vwrhx
z`DzA=HQc{?~B~q=Yw8EBKyHpkCy3^buS?>JXK9o2*k
zYG3D}qoXTbNAA|qklH=+Skgv1y7t!0Up9k|bOr437TAAdp#q<`nsl`9Y{#5|dvsLA
zUs6H8NJk%++@tjs&`^<@_vD6*LDc!^-iLPYL3GXDbhuU$)~hX)#ZgE@`O1weQ$bH`
zwL#SxSsHR&qwvPAlZrm&c!-3|p&*+2R&LfY5_%#js`
zeXlSND2F^hU!G(5Rf>k5vqH&H(St~Jfu?+o8Wo)zm8sbnNJh3nUkb%kNJ!N4OP?2Y
z0JWM=Idh^sO4Vr8zT`4>iPbhJe{jr7J1O&T4<)avLfNi5e|1O&>&E^Ej=k
z161TuP!PvqNI|+^=#;&2B>4X|23<>`-t>6v)R%rCA=6#AO33{bv`6*fo~5S-QDnPu
zh1mug3d!Cx`uQ#mVfVv*B;O69Nf!6C{W}U83Ea$So@K<-SeGD+*eq?~{g__da{rsV?%*_xx;1B0HjW4FNh=MpDdYn*XBO~pk
zfd_Zt9B;T~secakh0%NE;9c4PVlFo?yz`WV%8V}F&JLv_Mxoq*82GE3IBTi{ajx&`
z+hDbZhN6AN1@WY^jJQ
z`Xo>X?wR#g`khgO14z~-O4rkxg1qsHqh;1~RKH+o+5S)ly7su^#;5`vNnDhFd|ify
z+}CgKz5@O5!qRdV`?G_Hqg8W%^As7~+AAMb5BEv7vF~$Bs6(rFTk>AQy}G(y{znh=
VW2Z^^&$8)sWE8a5DL0>n{sUa;Ja_;A

literal 11976
zcmeHsc_7u>*Y_70GKFL+sbosYT&e7D88RfH5K6|#m>~@m8KTG#r5hC*5fY*(bEe9a
zic%TQ!8yl}k|F0j>s#D=pXYbq_xa*={FHc`=
zF&~C4gTeV$((>}~_Hpv{b+vbQ((!k9^3j=BRn~?%=jZ46HNUr)`*BC7Lw^~;e6qm#
zA9n0>aoZlX^ZDz*7yOL$*PlP^xaRGg+?_m}JpKOKcCNiacjHe7EZE$7J)C@9Z2gXV
zJN;a*01VErcdw_fpO3$TpTCcjqY-GqD*qhspY5Gp-GAA$|H+>4tl*;dbGkpx%ZdpS
zdG>SlaKiXkpx?Jgl?zj4K$!E>*BD4yjQ4y&^Y-)kW%*_A==XoMJ|S2vL88Z@Kk~Hj
zSDuJ@ef^xg|8iq=e@*st5$wMrnG1(FH;E4u(=WmJPljRh$TD}X{?&>IRaRonfXvLy
zKOJz)`D0<4O8}Nky6g`lChXsq-?^*qzsv86pTYfWeu?8^
zeq~3M{=Sp!k6iru+xOG$cPU;N?pfoKmhFk*MpEyB6^~%J3;p}jXCS}cnYJLn1H*Sx
zMVVy*m+W^miUrJ7JuOoK`a8B)9`*#iw(_{Lm`Mf_-EFq$ITMDL*BsPQV#Dw}zsI#A
zi!i*)Fp{-H1jBEfjJW<)0>cCJj`8NmVEE%E6l1&5aA7MQO*e6V3hbn#OxE`ugH$?NbJ+3}MyDg;si1p%mN8Iw>R^M_Rt9p<
zwC&?CXCT@9EhaS~bktTNf5z%49T}~4O0x@~qq=)jlEY4PWWTkMzUL4fHMWmc8#>d`
zUh`fC3t)R~!fz!Ym(Y+NiE^hSU2@asK~n}=td#ii-UbHP-462Q1x<1Fg~c?W?yF11X-Z!q+-8P=sr2t{Gw=+F1#mbWsMPdq;^EDq(ok$(#UdSpTKI
zBdK3_F#J%Z9H)mk9oq$v{&-y|NecHt=uAD3zr)IN{l6^orM-2*XPTC;V(m<#0F
z-r)c$!SFr8EH@MLFnn_%tIdN9(67mVxi1&wHZnF#>M(pJBt&sX6NbN|k2jP($M9IO
z`U?F93}5CFe%`Se!`;GL->Md2c;R=6V8IFu-~PolG^8HG_p+cO=NQZ``tQ2drQ({&hLwKijuzA=VhsP($Uava0i%GGwnhGO`n
zed}W$eGC^95g7Ug`5CLIWMVLGyT
zbL(88?mRO
z<)pz=c@=cjl)V2{jtm36jJR6uzMp{(Y4MZl*J8LggU#SA7ltn#9%;xAVIaGyrJo-+
z(^1h0%7JMQIuaaZzndCQN4DE~G<=~h%Ih_QtZp&T%7M{MCIbv~+>c{h2&}vFDe3%K
zJ`4{paNNV}%RpD{J}?+i-^-+U@M>5O*R7E&mEBP9i=5vD7Qi}O%hQ^nFV3BpxN&JO
zhTnBlqK82qd!K41X=wr9=(wsbkAc+hRVQ~8(NX?|6BD)GbaeSD)*A-($TT#D{Ggvw
zD6<8=F&M7;lxIvQ8^f<|@zlQz`yzg_j%URY3_pDv+lhZ+prnjPuKgv@7ai7DyMyVd
z@adV4ruP}BNl{x=4EDcy-7&`(H5ksVE>m9yb(Nj!@oWtCdF3Z(Tg(r`&Br;im3kQH
zK!`<6=5;!96D{|+caM%NxQ&Z@g`vN{pV@n*0K*;I9UAkX&SWcZ^sfT0R$u!*gQ%Oy
z87bTp!)tV%4<^JgP-x2XcK>5^^f^GRQF@Gy4!d79YKQU91=5vlP{%8}U3yEQ-g(yL
zMejI^;nU}+Hg{4H3J#kVxLy>!f=ka
z7pf*;9iD1jbsniuzsxA01ni%Fl!{&k^_&*eJ+lw$D3WLVpcFTT_sMNv{Md|vZkz4B
z>wJrjHr`q^!Vl*~?iZge$1l;5j?s@LlTe4NzTvjqk_=>)v7yBF8uXWMy_!@6^p&E`
zOreEe`kT->=9^1vCFO5!iya|MS@Y+R?zTaS=yBG5wC09XRtDeI?KpnW$TvX@|
zqoZui;W~qKI+EvQy`u)}Hrejp*cprABzxYsOqCd(*o=?!R%7_t!NP$oI}9%?ihN_3
z0QEVn96nM^Mz7YY5!3*$e;G%aCZ*amuT8xbRnSW
zq{Vr!CL*uK$s6Gwam!MF>Tw&xj|(@jm%{mE!uzV}q#4xxWX8;kQU;3b_l$aA0sVIL
ziENe;oUaOurb8oibfA94LS|n&T6x}4T&kXq?x;6sRL#
z7GHr=ED0F?MmVEb3i_dSe^_vKSj=6{w?;*C20PW6WGyMMX+oB#fGzLX+a=jQqAan-oF
zpACO=H2=8!CjPmuvr-`eZ@gWSF9ejUM&ObdPrr;nf%)dyeX!R}zD
zf&DueZ@NwE$ph30AnlO_th&03eKBCS;*v%DU>EdQoUIG+hW;9->DdwF=s6hi8*uKs
z?e{Z4uT?B@wui7^n|k05;AU1qu22|1swWhD4e%oGCh<6!w=t|rXdHOC$8gXT_~{g7
zI~@l4TUKX$Q2@;Iax)+}^?c1DL(uCE;e5FTaP#VTZb6XCCCBmZ0PcGxK#T;azI~#l
z8uH@FwQU}$V&a1qG2OVMMf
zfH^`YNt%$4FX^pL(qQLalXtWM;_EQJ`6dzgNAj==%wD;07xrWOj-luCbCQ8M*;N|dN%TbT<6K?ica9sOuJZd
zz}8Un=l*~*r#yn#0e?6q*|h+sghkg8`|$d=yFyujUWfR4XF=|^S2<=8aE;>WA_AQU
z?JLSao*ljGL=#|-*6?FbkZ;}`b!!>$X=A5?gMbS1=L9ohT`JfS4ME78kfe&e71U>1
z;3u#Bz?p-|2i8L#ycd>zT|GU5?z0EKBeJ1SWR
zVch(n)zTc`z3~gVCV#*@HW|T>!4K|mGg}O}lhgB0hrs^k^G{{FL0`8(u+9niXNTR~
zTY$eXerdV^xMF&3Pz3PO-n_(G@Fy7IuqF}Yey_vj`Cwk9+WQU)z_0CgqGkc!j%uIg
z27Qz9ox)FmrzEDkMgmX#NI62xCln)b)(yC}mKa+M_;dJnQe!pn%BiLVbMSXQvc#Vg
zI62MWDFRH0DsCd`9>!!Is4QZ+7_V$pH`C^gMSMIOC-9s#3tG4W=1HUaiWaP5OW*`j+Yq0EXt{
zh2kL3$=^LG21w5w&khINnSW1ZJK*Tn#OK|hNA}eBBlaU^MOq;pF#6R{Id=yVueHO+)DRL;<04hz?@??WN&U}}REW|a&FSI%i
zxNv*eEd@Y{agDWS0lx)sO%nOGjoJC#9PEs)l2#Mv^K3)npk*;
z`9X~xk%9Twf4DLu3-W~a^)HFKl|Q-QohopBGa>asz|oFVr}_as(_OlVzF1W0^s_F`
znY%2Vh53#b-Q4yXaEt$V<9@Jb>9|yW9P~a_K8j!lJg57;QVGWUo-y`?Ef_`X3SOS{
z0=|2xX+kFG+tweB_X2!d&&SgO`6}EEk!=xh&=LNFTGC0vqO)6!8?fi(n4*Q
zH4wLAdCUbFmQhqN=;h`L`x$E9wU+P~8e0)ZoJ)$`#}l{!ZAJp}iTXdBHn=keuy#sf
zs1xkc`jtCeA@9ru{-g}>zs~5xF%^&pK25k&1o5q`c_A
zwL+Q@r^ZxQZ!W~s6gTmCG3aZ$8X0K-x9jXSE(M-aBdQPv7#+6E<}P5&nhB>%;O|>V
z((DfKVAV@xQx9oaU<-M{7Iwgo*->;w-{tBw#QFXCSTq&5NX98P;=H+F
z?8ki?usKwss~F}#D8|pi4_t*Q&m#`}-0ZaRCH7APbF@7La{3EhvB$twYXto4fgib?
z$t?sp%(h`4@%~AlPS+*wg_ojwRzHACIUSjB2d=92Fq7!dy?Q;b6VVh<#mkVrug{7(e>z(f|i=mGb9t7h!zd`r#g7z@BKSNCS`yt~A|y0(eew
zd?0atoCxOBWCD4K@||y5z&{1*ClGngFUA`!fp2)$`+%r}VGYAD;$GZateHyWU%Gp5
zrVC(w?OE$&m{<8qa=0jrN2u}?o&b44@xxWS0ofg*!_Nbjg_l|qeVF7dbnGqY>A$w^
zB<`dAw9=jTL0&QFoRbYaWmnd=Zt(XiR%$XEILx%B$=6H3Y59^n7
z&{hB%54{f%0C~i4UlgS*!PBr`^#0n%|yRg-Jg9<;4UhUvL;}C$Y#|u
zps%^Ic~L50C$ses@vgDtaEKx5O|ZHnjRx|OhW%nofFB--mOKq{X^PcdJ^}c)Q=+;^TV2tch4xqK?=TSMDV4
z)!ft*@MfJ_~Bl$8zFDvYCXjU7`s2K>LchUmR}Ct0C??0zP}~N5B2JA-U+zl
z&eL?_{a_Nn`&9_Wmrhiud;p9OTe8y(+B;vl$Q0ev}*7>%mecc}5+egT{#5-f_
zTJ6<(5LbLQ!b)Oap1FLTw1e^Q-qk6@yVFl{`cN96
z=e5jo9l#GQ;UXa~1Qs2=E5#*QjH13$LX!e@y0
zQQ-rJ2`L!ovcB<-_}yl8aCW^fjQ7aoe~g8`K5{>^GZ5lo4IQ6|g>mF+mu?1jTkmwG
zw}Cu5LokB4pC1O~=C;H5wU6bessXE2kGc{4HPujcBXHaTOjn=r4Bu-?R1@PACL>lP8ur2KRpv40yKRr}~LKD18%
zb`0q6DR5p(+;f7bI=baSzw%(+^;XbZ>6DZB9p-!9f8e$#ONz`egrW-=a9P^9Ymf&KLq#5Ae$`BM`GE%&b5#a2N<-6Ho2ONuCHCbd;pE}w#W
z6KD&20XLnGa4jyNAaT1zVf|oFQW?im;s((wHpPUF#6je6d;PPw1%t>`vD5NE?;sio
zzFFPMGlYBwTVg_#hLDqbGUbc;5R!F0Jig?_5K=j2%o-3oguHd-E_iv8(9G-G`SKl;F4D9TOhN6C+srEdd9r7M%L
z-hNcf-)f)Q*N?XK#S5X~eiRV!TwZFHjM^$Mj@lHF(dt)>_DLr)`aEG>T4hT{?bb%O
z8fwT$@@r0Bqv8NMkt6jWOJx8pmU>}P@DAd$d@B4Xo{YX_PQA1Z8A6`IS&nX!Lui5{
z_pE*EATrcy{!Z|$IPW-NNea5}{Dbo@I~6g#UzEGffQo`9?JRGyQqk}XmVLO8f{t5d
z9F$x~L&+Svw2}!bI(F)Ul4>#N3Ckoc1|B+8wD-58G^C}G-}@$phLq0SQlTHGp{*-g
z?g&TI5alC_>Od?FIYenEXp7U(Q&S7LYimHBy^|c)6>z6u`fCrk&i1@&ROFy+2UWb`TXue$6vY8$ttItTu!F
zgD4;`FNVcn0BMg=2h3wg=vn)aTM^XT1`d1Gzxa+&h3=l%MQfer5=rY%?k~
z+d@I+clLfA&ZVGDBmko(Wt1uleWtT_J!VR+&K&Vy}X6|%FTZe@jsYd<(NT2YMoj?
z^uwpiS`T(!%@w4f(3{g2kDjKXPp08!;frbL
zU1gEkQB^9MRIeQAlBb}rv?!tT5d%oPeO(rni-hvzt1e3|gZ`Pah!I^&L1RJoa?@vM
zC?nx4M;n8R#8nTdbt+TQ(F&(SSK(ZWGpjArjvqql%04I6Cdeo=-j0Qt+K+f{D^&Ty
z{?4YR+<0G1K}){K@GVK8p|PriM;<~QtGDlxeh&7bwa>Xj0aememkB~0Hu#L)8EhLw
z2M*xIP-O)C>i0B3W9NvZ(2%N$%|kbYP<1lO96?BSkBnsP6`xsv9nZJFK+-{=5Nn1
z1UrX!dn7DJ`_TlMWy>pJ66&R|;x&Rk(U-s6rCvfo9gn7x9`C23t&8t&Ftnnf2_@LT#IvvK&Hu2g^9Cs;P)aeZ|gzV>EPaVYw^Yw9GY9%iCrES>vTgifyQy3?0kEb60Wgyd1T2%&12_UWHIoelV*xRDUp_g2l<#x-(NL6HKk3pFLp58xX2Uenwd&)b-kxwxUd
z^@0*A`I|YMthE`P?#W^4w=J_q{W#c*to!X0#zAlT!l#{y9J0-7Mxrls2-fy%V4amFjJ|7L1Exv`o(Gn$q`}Avh@rW$giHA-pnBK
z@6l@qzc2`UTPb>9fWzn1-84~!!;-IZk7qhE__Na8BgmOS*Tsf0FLwr`_WQys_Ayx6
zWf>phz@R-Xf6%-L&j}8;W{TlC?bFx&<#D|^Pc8N_uG@0*kNe=glvQu1G@W2Te*{~4
zMlsm6z)(~DCWFqX9cR?eGuWqW8!<1EfpDLkU5p=t@qI6|>hXNK$^W|
z273Py6|Xi1Fk$|xZ;fbAp#S8-*9@ehWgMP$F*qW2?x4*m1B-zyb88_E3+^@)J`&~N
zEBhjEa3Y6YOTDk1{DFDAVcgnN&cLpH
z!!-y^>PcC9Wi_Fy>OHd2HiQnAr~hc!M(7#ge`P9Z)`SMMY|S=YN~nqP`RY#m9(~e6
zD~1w!ec;`(vwDOoXus@MHYT)nMgG&}n+d(G_10(<`QY>Yq=Zi>p?0$>>rDP3G_~zz
zb>lTc=N0Tawcs|Pa`h*ICg%`px~50HD}m7F?PaBD*9leWzyGcIJfYd8zOpefggzJl
zeRNfm_tVXh+ou!hj(M(d(FYYCn9{P0*A?rro;
zhN_?9d#?=W8P^dyly<6V9e!6iSl=s+dvZt3;T-ODdo0u|@!n;DV|{J-Jx!_n$4b1X
z^6w{AwGu*qz01=0dWX=ToyQg@UnW%Ne|Fxg{xm{QEK+j|y+r7S9yt&zM*oFbj)KT%
zt^Y0w-1(c(D~TiiKQ0n_Q9aA_SO%ft^H+On#1r~W*FDAm6rm@-#jPpZNvN&Wmdp?n
zLOt8GD~i+!6=+WGEmFmNoSmBHXh7(VO7{f8;}}oLGP9@vtjErY8@;e@YOC7YSpAP#dk!)ea&CXCK_=wAvNQSGKWZuXrD(73_^$h
zaXuA`JYFw-VbK8wi#jZu9w46se-~}FRX)pLKby1_o-%kg@wU%pyr=WXThlj+!#ATC
z2^);-@ym_Fv2QqhX%JV|M;?rxbCI`2z6m+Ime^iqh|?=Y#>mU;q}q~JDQ*U(%?%<%5J1WiEQ7>+dY*!0s;H&F%
zOmHuQ?KL-zo_H~c@fRq_e|^Njy)4QiRExt6>*5thOF0BJREZfaB-HF`&dh$qja6#G
zZi;x&sy*!Zw@_GD&u-zQAhVr{Nr^P1N%MFTPfCaLi#{X4d%tvAm((}
z2L|R6-R1J
z3WwIK+0S$_?$f3jR=Q~aVArapvoJsF-4t|d;}|$z>Bz~S#38@(rxeRU-J07bG>Cd|
z-~@du6-ekToyIQ)1!#A3;}`a6sH_SUluBx0pfH=J=&W
z&K#`GTgC0?658AKXXH{}LW`?*+V4Q#UXNLpd=>2*)S9HZeq*}FOxYm&FaZd&fL#C^nx)DFdUKt8EIMmM$X$@ROsQe_MVY>_X
z&b-491W{Ky3oH&vVt$;J`5j5rImzhZb}ubLcir?`>3oL6>aCXAldwO0sh=^sZ$I)z
zT=d!AXa+l+f27>C7IBzn)6~01kWlln+pe{!t5K0YXQUny8tlsbHsiZ{7p!?aoP_!m
zR^C2>bH#@tlbVv>91OgRBt?NG2I^v={A{l<)Y%>ZIF^rge~UgShR;Eqrr(#
z$WS-jrWSk++t46BjDUHCqr(1X$Z~>?`}!7VO(7ixo(c=|8=~=gD?1=
z!eYvrJj$diG%k8Bz|}<>=?_E{oni#Ir|Kt;Wt5h=bBs45D$R5fpbeumrbTW&gk24T;Xxu*C1Lq)_fAN<#f8!h#BH6ST^VgfSC{7XkM%+Ue
zj|*7;Ub!qaBgEOe^>^}SV4axOd-0wfgwC9qDm;5Sq24N8%WlPT2uLe9JA!`lmQD#%
z&qDiQBGXs6GSFY*w9XXc5mswo7K(N9O!#7Y*N@O!gZIv_$|aQNUtAH3evEGfWyWJa
zf3y~k+oO&6`x&vu>WBlObDk8*?8biY{V{v-3DjXpRhRfU*gtPO&Ij-PO#s8Yne+T_cA`Y!RSf?X`pZ~m!ytIy5ZWV(1s987m
z>LK1+x5E5=ksZ!6Ysw1mBHp;o47tykr&rbv^IYyS_}PEeXaKP9jLHaJ_{m{Ke^JZP
zIMl-n7nfA{V;{GFwNSebasIR*LRk#`McwFF-kOLyrgvei6mj6@vvJ(>C5Ja!2L^Ux
zU!7Pw^msPnR_wi}RX5`KMfm2%plk-+Hd^;O^f=s{HDqgq^HoUU?rvkO@42HUa~bNF
z-bZ)0HjE?nvx(9h3AK(;mEgLvfZ-C+i;1SV{Ef&9u99G+M_WfIu;tEgdI3-UvA+SC?B#Bt^7-?E{I@2Fo2
zL4V_XHMPD(r*R*lGVV1Wqw)Nch?;?Gs0#|;Pi-Gk#rZHLNZ>o-_NKIlf6h4elf9DO
zCHyvn6Wbb8YqxW_B=|}z6LB!L`Sb%1=7{mi%Mj
zmnP0L*5g~9@IU5gt&X27{%`zx9~lse`o)vQo!;X7ZdFx1B^mS9_I*Q#H|pI!FRC)z
zpTj+Y^SRa7Pt@IBg)heZ9LHLJY;nOktkTB!{wd_m)1gLzXWoSV4{@(~c$3~T7=P&s
z00000004NL9hZ4L*4Y=wQPM=EWJ$K{+mO9RWg;OWVpQ4)31KJ-sYWBR43ZvgvWF~5
zc*>TgM^QwPEx!~Yd5p3~{XS>@dcV%O=bn4N_ndpq{pNkO&FuNkHh3KNk+Gd~+@QMP
zhnzguxB-u}Q-C^Tfo8YYwjARIYJbn#p6P;Z0ka-QLAlFW!3V+IW46zgKt85}{W36o
z)=+x%C)*%tQb@!IvTDt(UwPrjXsAy<2Y!3U-n3Ujmb`GyI|ICtQK9Au7R~!Tt$_W`
z@DMvm@Z{banO67-X_Yaq0zGy=oXY^mwlw<7WBj)rg@R^~eHay5$5st6(tscoEwA1ox#MkRFf{uoBG;b
z9t*w~|5{!R&L3N`c@G#i(0`yG3_t5to#u8x<}_AyH-~>8Iq7}dF+b*vQQ1{czoRUT
z*7@Z&PfQ==f^C=n-3Ho==rl1f@1^nmgNfiH-qF&xpufipFF#P-qBN-y{xyGn!49W$
zYV?_t1)~>bGZHcX*H2CZhrrhd+WV@(f!~kLQd}p_iuGKCUC`G$hJP-_O|tNK9@x(1
zC$tLVL8&88W+3O5S`_T1@0q?|w;vpt;W#0LefIAO%yfs{DW25t9^jG3Z(nH8I^i+(
zQn1_lba_$`_SLa(Ou7_mycKL~&vVXyDJ~Pu(4h$X)9!~+oHm=f_z&LN+dj*PP
zi+_m%hhyi%>LX&(W
z^wunn+0$Uto69;8uz#_y!kUfo$ifxN)`6*a8A{|wWB=`R6UdJ_kBZYd#ZQ>6Oad*&
zzpSMD8DydztA7cqE5r+PgH4g?tos<}@1K;p3m#Xt5GFq+#c#iF!1(oxQ4Ip%z5NYF
zzMxoPmqZ`zu6U+tOoCh@{y}2UAD!kL5dpV`Jr`O6j>k7wQGC0-Ioz&;H+QJt%K|G3
zT05*UuBRW7Ax7)&b}2prE@SmJ(|#)2e-Rv@b-nENZGXo7NOGMz$&LC{JSi$-3A^#+
z4Lds#Hz&(8p2rrA8>rMfcI<$CcCAjC-5m1g!3_f$?0?rvx3hHKpK^mcnlL^$cx7Q8
zdWTCq{DF{%{hyfCgUlMMo+9WCHx^!Ym}eUtajMqifvj8n{N-}QkDJ48Z7JBy88tNs
zf4~2I_$E5ui&37xTrh_SS+h*!2CD$DCLiZY+I8T
z<)yuQ3dwn=cE9t6e{TuTq7u-WgP)rV{^fE+V^uLeDY(DX0qoElu9Soyxpc|v6o1>7
z5q}<3_okO8YSBIVY%{@6bk5FBru?mPnUbY@Af>o?1=S02&Poqe_}5oWDc?(dN7CGV
zCuk*qaWj3d``$@gY1j#u7%}U>G_mMz8|ZVMiCmYZ`cbw*p#XYOM*B6o7Z1e@5~=>p
zE?R3zc`NZ?aBc|xZ61pseTs3vl5>3e(0?yIkUz+zeeI++LQ9f(sD@jov4=bmcfhS%r8E&U~z4t_)FF2PVY(n{A5%Jp&
zE2`&zIF8YImMi>Jc}4zJZVXVIw{`y!M|pF4q%GxTk^tJXpC)QA&*mk~_q5H4&DQycVv$TbS&enu0r7t^`3B5z0c^>uAb+!Cjzfk>&Q|(<2
z{dxu8O>4mN8D<9Q?@KvtrhZ`QTz{M~27R0S%PTj*HC47pK7gSja%}Q%9ocP8eSde9
zmJP+HrgNN&`v2@$yU1tiZ>NS%(>tU6W^W&z??b7RkyM9O6%VbYb67Cc?RgKD<%Bb+
zpPanAL7wu$uPaoK;tabT0_nn>nrQ=NUQbUB{xo!V17)6-!0!FfK4_rUtaEvguw$@1x5XFdNx{eRh7pEYFz?wJXELit_$x!<1ZQ~o!5Kg#Q@(r|6MN5`vz
zV^W}>v|qSQ{c)7{a0nOJ;4#(`h50(yrgXDV_lnp{O$AVA-7j+F=FGAUB77JQ?x?4{
zEy_9+7dK6dO#1FC{R-UlJ=VF$Z%@IG)S)j;XTVZLEpNJ~!(E#4^nZ@=|M~jIF4O_*
zbQ>8b@MnQX#aHN+&S`$l216E4ZO@{3ektKG)kEhqyfm_7(*c|=uJk7#YDE`vY3Z3{q
z#_A6lbj}G@%@0&yr(9ysCyRC3U8CZ@Q~$4el_^2*!M;bDREMMA_z1Rw{F6SB55bY?
zPoi{R*0Q=Z`f=_8Wzi9mIQOzS`Qm%@{$@KVg+LZ7*zSJ<5qzC%_8zxpIz^=hl`9KlABmI&pi|))qKsAxrm4lj+LU3#PyI?V|Udt8QL~I)A-GLVYcbslTbEw9x;8uHJh_
zHum+FQJX>MP<%{aObzn7PSX}keCOXKo)Yu0yKrQMx*zqIJwm#}kPAO~+@!vs!uv{&
z`b-69zSCyv2TQNsqP}dwe>yS(`%QE=Y@~PV;iB3KMbI*-W{)fM4bHh5m*_h$n0wGZ
zr)x^XxCg-h0}k;8aFgCL7=P&s00000004NL6_UQpvLs~pG=+MY
zu~$kIkwP^JC3`=V?32hg&m-ij;JO_Iqw_lTOV5k1+tLFmdGZgRVutDDeS5sZEt!tZ2HvDxS{1`M0
zzR)~)Pab;yzhCV+4}U+!^HTR0W;u-RmZETR=tHt*Ra4>DEq6uA5AcPv0n-FDACKF4
z7V8*!Fw^E?_r^EPIuPu6E7}?Zzc0_Sr@r8Oejj2U;r!q*HYbWV8q;@ua|jNFKG40o
zeg@xBSI|Ak!Mu$b7rigTZ{+TkyoO-YGg)mv_}`f@7VZgVsDC|Vgo8DDrFVXVeMM+~
zQ3w3YeV18z6a2&eWNs$>y=`FKkHR{3@rB(?$eqoDxo2U&+*<6OF8ucMdI?p+5C5dm
zhWF4XnV4-+fvmB$L$ndRGR*6<7S!6mHRTihTxjQ+EP`F`(GT@kK>g?bN!9qClrtBj
zX}t%I>d8&ew|_qt))j_-t8z`Lt(b2XS7&Mjy`o5Ab}*R3mlLZA{e+^{vCmWw5j^j`
zz>8_hGF8xfdvD6_0~Z^X^s?bM#42Je80%!Yc7^amZ(bQDLUBj3X0K8GjPXqEZa{pJ
zMvl2FAg^lUXJ&%kGsBICK;solmYIRi($f@#5ZAPAn}5U@>_rV#t}&3W68_PIN1y~-
zysiUw_g4n@QhYJ`9vl52TPzV-Sp$2;Ub8w4st@_#&zHcfZ*p%^U3}#pj|#+i>al$F
z2Fwd93@hY(NBu61^H!(2llq{Li*etQ@{!Y!9g764%RsBt?i1U<`8@$vuQ5LABVKI)
zIlyf1s(&O{+a(WsH@oTzZiNk_meuS!0v~{#qLU(fNw;TH8kj^WfOIa8Zg|;D4*$Eoacwu|%EDZxc_Iq!Pv_{wcHNu;^yzObe@e?hqpnlsWz3o4F*liY^8nE
zzqE==`7hBte}nR^nILs<7xuG%9By?(&OFfxS5Q9u>-x
zZm?@qaP8}cKTENI!3?Z-Rd#p!BKYfiCs^bPT8NRt|jtv>HJFcWU2i^b)dbqjDOba|G3&j3x1=D3?tLPz*1gbUW}VWeHN65
zTv~EJV-%F*iH%qUJy#$-#RBr1q~sknm~a2b5|{OmZ9}c!()hWO=lxqKAO0HIe?ec)
zTD|=U_@md}mHhr!E7jGDdHRLR9q4=H$JJjbP#$YeDX_sZ_kD4+Uy-d=bv|IBPk-(+
z+Sk%(&WUcY&h+c(QTQ<$+Z1pB{QIruB*il!yLfd3ctv4R@kcP^w9iqhpKt-G@p{Pm
zMKSDtFtftNS`2y}G3ybkLygH(s*}*Kx^w6^J;;7eF1`ieu9nb-yWsGhi^5b#E#fog
zO^ExN&7Qrq{vS(Sx}w17UEY3le}5eQ#g)4rOj*dfDFoiEm>lD$x*FGbNcY!Qnf)^}
zU=Cj+*A4pgc-E$BFzxZD7&@op+Gd%>@XtKJrx6PI+V;qB8<1W3a_kI!Z=BwQ1t@9q
z#b^U4V=5j(`+X)u+BKBo*Ait5fkqbI+b_{R?kYawOYt2O`ic61e=WyS1AltPb>8O*
zRLA^*c9f?t)eEu|x8M!^a>`fbN5`BU)NhK{T5h6#QqU-Vi_XpQ|586AKEq9+-v5AW
z_>1d5L;u=4K)@Atc{A#@ZICsOlnPM27ETV?N`mI4!b=CJ-(1Laqw{-kIMRXEa~;c7
zqB_x;c;L2%`c7N+SR=**9)E1_4*>b3_QhNUA68p^ScdT}b!rx(;3nBW=blkrS5!ua
zgL~Abmr_1X^ku((412NWmgiwq2fj~U-2k&wn>+;QeqVI*g*JHUqMX%cFvIy}5beKl
zL7z|$?3S+-^^~Cd@_L(-6Xa_boK?u5q<^O^<+3xCj`?{X3o1P@!@
zH<6?H>RsAYzpvT(MGEwN%=su;FmBoG&pe=3bFZBmczY`O7~K=RpBg`>!oDa^`B4(p
zh2(a>LHd3kw*|U)V@yRlmecn#vc_7V|F)(ssDRFsfobj($ceT)r9$U^K8EuFvSDab
z{0NwERl1Gd6T>$(lYgl%v7Y41Q~t#`GY=_`Y_0WjROhU|43;U_zgPY+?UQfT`YG~X
z-C#RT{lqy-i=)5>d5466bJZv-Yp>=wOE(uCuyrDd(
zpZefh?~{G>zBzLwbNf;-B}h6)1g!A%ka-3k4{n#C{t*|S#(&BKr-DE@@3|f%cb>0lC=J!yve)~ZHmPF)|Ibf)WD&@oyqsWjWy-#9O+!fD;H3oH!jmv
z+6dMriusjee7(E=k15=z+*2{@=srt~&vK&sUHRo{cWbb>HCvVX$dfe=S5xRdx%sCC
z-5c`u$NN2@-+wKXUi%cBNMxO-bGUWx`|00s4wNp8t8~I|?E&#HOY~jkFt5sX%-g(6
zY!Q9Odewc?v5=oDw9aROUQu?PbZ$Pd??0e-fyrvcuPJmMHGf`8v}qe;QJw8q$+Mb;
z|GOWCy*RL6)AzQ9?&HbW>GUd4=A?~(JDtaN$K(rOaeuJO|E!?zmGc~#0|if9F<3|E
zckEF(?MD}n@(9(z%P$_r*T6j0hH0u(PRZ&o?&u5b$@=_3&>@;HhR&gTn1x3){Hi<+
zS@QzC?O*+X-Vsf?*S)A7!}69tK1lU1ykRlLX`HlM@joy$w%pDf`n!3_pXi=_-ZQ-L
z9n3nW^MBM9{(JphCFy>27|Occ2>S=SZXGbB{R^2}OL?i+Qux#fR=&5>@uzx`_V76m
zdhTKAQa{N|+1Ok`=UP}XXbSWHb=m!{5Dc5J&nU+}tgVQ(vxEPaeg1RwPB|*{!%zcU
zu31>N6Z0(eMPleZpL;Hi89{w;xUPB>d^~Eau77~|wD&2UD~9Z|ZiOnnlOu$yncv{|
zPM@uy0?0HL6-dN*bLF~vHe~4?-}ZFjoj5mkYQ_`m3{?4P(Y?uKZobk(`L*{;)PrA@
z=+P--Q0~dVzIW880}>)#z}4&Klj6W3NfUkZeNXEnXwgbNdG$Zvb=3UrjwjJYLQN&LsEr-8E;zj=gKAjW_k48_R%6M;_YqV+F
z0xD`fxjh2?iXsIURZwZ>!z}%`Oz4-l?*|9}5BE6g54mF)KmanYlply{9
zqFlkjm%7bg&dYne;zfGtaDeW-AX3wFT=7}UGiPSbzFpq
z?5k|@&EI$ugDFR{SGg@=@Io(ABi@67`xulQjbdQY<=AX~oxyhb9MeY+87LZ^3EuvO
zf$d6_vmC3Sjl->S2U>=L&Pt9lBHnB)4xqsDCdXB|1=RD6lIusnWr#J=f
zqOh!>?1WDUg_HO1_8Xn0Fr4JSB`}>rlFJ~z6jD$Pu$w)hib9avr|YWC6e^gN*)v!Q
zafzQFCy%9ZqqF%zoHdQG|GYI2Ph(B*m@AQ=MNl+Rmqfb=HD@fx4t^Ektzm4FCV4)^
zC#n4TB!5ECG4-5+FCusjUszW|iI8RPSoebwp}Kfrw4@+{?5fR?*Ly{9dmQ>=g93%E
zWeo=gO(_&h%WP|oq=26}nU`!RR6SijW6WF%m#hw4$e2XIZO-$Te^n?H1^=9TSDiva
zwStYlHiZX~A2K|3DP&%{rPx586Lw_IwV6!rr+<91-bjirZ!mr{Fney)_qv9Qm4x3{nrk^epqsZ8)
z?U@sYFU4^-X*FWFDqI(auZdC8_i?rQOEDV8p0ANG<&9`#z|_e&mwZx_9-(-AO&mQ!7_U%*05KXZdP0b%Y>oR^#tkT))I
zU*Jgrb{C9BTgMCdr7S14J5s>sU70D%qXn#Yx!RW;E5Q3^)wJjc0p@KppGuMcSF}qm
zxUA>~a)(H4(9)G`EVzhuOllq*_JmleBQJJcj#=~~%bBnFDJlvaJ
zHL2?hSgPQ*xRLa+`L#chg-3Zbj<-!aaG#vl>cL(2J3MZ#(=g8W=CQ%^mzsSqhl9Bp
z8XG(~yifCymU}No*;I)&=ln<>KRxd>+bhNfXQhI!S}`ikm99$^aL8NmRDb94Z4TR}
zM%Os!i{U=_JFqN?Mf741jlpmRKAqnqCy+iI+!t9^T1&w&S-S0j5ryr|-zR>gMc6k(
zI9|tzpt2$B`Xwt0*Hdyjc0Hp|UcS-4^bC#DxzAdA>crUM_rhB)pGAIvcX81L26tzj
z8Qqscqh_VX{-{O@8GbvnReu*zaAA#FYrcul-4vyNf)!zGe6>T-LJCcatB*!iQ!wc2
zm$yrxA-i#gS!xUe?_7ERB+xF@H*il-fZGmTwFK
zyhl=qb@J+ZM)=U8rFObUlfvHW6IKP8q<`mhGAsT^n&>{N4477v0J>z*$bN-Dq|AIQNgc+aIZmJIcdB+h0Q8DA)2?y{x0gB{}s_#vfgZe
z)G*=v`vb4V_}tP%c&#hL|CZSk8K+-eDl9UMo){-asQ
z4MPeYrA>)GWdHPu3bG}$3E#%uAC&2*Pv*@t$#WeonzLjC$HvkR5Mpe
zBaz3}%lfG?e>mvmsib%96eB!Q=R{35gYX>925Eg7wT8|euZUjyV~S)wJ4Hwy@~IeM
zOJS(p?$mw4vv@DIHMf<)xHE#a(`hlNmT2EFO&%SR^V`!51S~KekXc6by}q)2k5!`>
z4{R1}dGLdQihq1{M~p3v6p71kqW+-}{APHPczVLu^blM>q#{jN9wR*Fd$<&
z4a@WPZvHbV>}po2vgSqTwfCBRBb!25?$y;H1`J#qbbs^ob;NiczU5${G>`2brKvYp
z3YhM7@2V5o-{#`yp>3r+!t*SLevRVr@A}WBkzxi{1IiUHsnO64t3T!>OXkh(aW^iF
zqELALn(HHJ8tq+@1It=S9;f~Fn#zlz955JTpT{G;TU)XsM!=<P`8XN0kUJLmNcoe9^hMwabSkLTfE>Aok0_==|UpVqx5^S*kl)i2vP
z8X1mi*&(E^6}6N9eN6J=9AOo4m+--S%(?hcaeoZ{>8>^|B>H}rzL=wlPR>+yX!k}5
zI2K=+`IT_Z^!1uqwPfGWC$;k7WIbQm>-8#EjNvUxugXXqFZ-j;MPUrgu8eL}B=PP`
zM9j+~zM)-t{M`VeSMBd6)$W8lE%#JM1d;eHDd|3UOep+R9K~gka~xR_V!P=q4bz5y
zzJJ-0Jk|gGUfMczW>eb40It9f7fv7mztPN9wYH0t4SwufDA%b-A-D
z+4VBf-@&uPvrBIc5H2{#*=O
zWIe>A_hZ|w6H|Fi3r)>1`y+<>DQ=&DOlk8zWioH#WrJf@
zlKeFtvh9#jB%ImXZ=jsWV0Cw+Q5xyT1CLr%er}{NcJ6^wZwY57*j62Kie=#RNSLgt
zDMs=q_S^Oh4j&>HCgr~2VR@%5e}8F^fWvcjbZH*}oYdaqT~$1W?RGmgM{*c=QeE5F
z&%$qTPSz@->q*Nd$T<@qH7WNBdxz-Y$Nc-!Uc?7w$PFnp5N;Zto?van197`&$uZO
zDhxI}RYt?i;TMou&DcMIP
z&V}RzCC@+s;gh$fhVB)hc=^UNvrrxxl@X&H2gK;U@=!9zkj1TsLU;FR8b9u*KN(4M
zD>>)An{5VYd^nc4m)}?~Rx~|Xv6-TfzsIyg@M((rIDSPY0Rvs4nm%Mi+
zx~nz!f7iT}_=@?5lT?V_QY$*&c@tePHfonsCjFyY?-7!FipHwetnJ$N6!N=^=c|)A
z&E<=9*B+wKd(y$M{3VTp_p=Thy30bQ%Jo_5H4f8COkA`S1dI$(a(~=)N0me=;UFrMVCZOi3g)gT^gb`R^Xza@sZ
zh5d&V;(v-gK4+gNTu+cGiRca|zT))*{o;Bv3TG{z=>0?bCA9LZ@k>&#P8~aoVxqUQ
wX`Z8Y7Kx#1*D_*x0uMRgXq}^kYlUeCqcax}9~bECk(0;a|C31zh?Bia9o8?WJ^%m!

delta 12223
zcmV;wFF?@Bl>yq70kEb6lMDnclhXq@0XdUF11*y|1UCUPvvCC37Jq;Y00000004NL
z6<2v&j#(7GN(;?Ij6}3))1uO5?c7o-41TLIq{I+^lf*y=I4qvE@bQ}@4rS&wV=)&woYf0(
z)Vzkj%f%&Ky~E*Ff_(Ve2ORe9+SFUwz@bwAq`{|N4mwFu{(tg>Q1zyHLX!d+e2rbP
z_QrVz4R0ehi`-yP-rBsoqkzF#M+@)f+YH7kiZhkV7(5CdK6jy-!PQd{&9e0j)Z|x7
zpKE3ibb36;HG0q3M`*R!^NL0CdhBz_kyYPI%$DM)TuD>EH
z|6*X+yD1^eo_|4Sday+a?h_qu%bbe)Rvx<&D35EtjAq<@T-WCo9`eI;QD!aEnhrDA
z@FjGUcQgZs1uJ#5{$(&2z5Tf683y}QY){EVF_=CmXBQj5K+>r(s~-2?G&%6y+mFH4
zH*)7%gBXM^)flf1W-zt|+B1M}f~rTwXF*7$ar>q?du$)RQjO;|z3%vMj8GILx~K
zy!hS}4k5D7ZVgZ7uzj&_&XF$+?p-x*>wmz&u4}{1@F)&B#kQJlyEx2gXfiNf%VD-c
zqv~~Y4u4loW!z(JIqX_iTzcXNhn||4@z@LwX{)9>w3cyb32rn0fpuNcAy7%0IlS8V
zr(XGe4)aVCH#i_qvJO@#*H0yMwVlsS-oZgz%V~RG9Y@fYx=c%6AA6uR#BdQg;0&pH$OI?A@pjw
zzkh6OETJ#NzKMq7^ZYv>%uTZi{h1r}NTra_UY|J&lPd|8u~3!2TSMqNqjk}5Y6+Ek
z>NQb;&#=T~r1~+TX_tnU7}pUxmVUHp9lqChsUMKUXMuc8`;-6g>$lXZ!gKio2M0Ux
z{YB*mUsmBc)!T2?HOp{cYnJwhd_t%69)HwGxk%`$yjy0U(+PD^({u~}2l4KggQ=x>
zZ*i7`AnIA``#%Etr!j8QSm2ingeGcbnI6m_^nk)@Z|wv^>kQpf1CJ8A_hbB;iXDX7
zY~GR?W6P
zw7N5>Q-IJ#!|`d+2^Lc
zx>qD8pzbz?cIZE4u&pUGJVBmAc$?*6Cye(`Qf^w}WI{c10(GrpIE+n7%Q&vWA=)y=
z@7{9;*`sHjj>a*NUN3oGZ9jwU-G7^!?x3Di)VA8HoM5nr32%kR3?5F-^Sg-W7TojI
z@sH*(Y7{GOjd^u6ZWxVw#X-73Ol2ABU~Z_hye;ZY+{v}f_9BBexKv_{y39+iEo(zv
zS~V5RWulIsMT>;{7Bg5rx$lkeOb*kkWhd4JbJ!gaP^dhK(4vsI@Y{%=aepTr6nep7
zWc*oH?j{a@oYH$;)QvvZ8@^LAmO-zg#)UZanc&ZtO37gGdTw%G4eELA
zYa6?ueH`-aH2QPibFfc!w>4cxsA%1DVH2#A=N8Y3;m;g=3<_)ww{uV}7?)&Si1W3v
zZJLn`{0;pM3hrgFx8`4?M}IyH;sXVW3htqQR76{b>vFhcRl58@IfsPj52qR}B-G+^
z?wrra8>6(u-4ywtRsDBe>oh_o!b|0weKC)Rsy~)|KwsWB`PY{bSg$>jTPfCa;_Ogv
z4dP<7BKBBs2ZMFupY#LVIF#nAExLt1pz`wH6`-k1kpQRA0{eN0}&NAd(GYq`{
zh;i*jv)ijvIdojkeqxAu`A{k9wVefp0w1&aQ#)J4{o6>@lS%XD-f_KC?N
zKSK|NGRSf&6PlfHVXenhJM5?N5ur(!2sQBP
z@Vbb4j9wVIa|CrXZ*Y-(=Q`xQ`m(254>{C-J2W&I@7EGsm9yeFgWsz&_uC_H%rr)v
z=9e;Xwn5w&qD^e)d_voMXGSgdC-nNm9Xq$9Zm-1Zr+-|=`}J#0(%n(d8an3U
z9%~3q@lq0mWDrwB`hIvbcsXLax~Y#rt+7vr;#CgC(qV1E
zW`xcV78FE11k`{VOj6<;7^r8>p9Qu@Q)Q8D&SorFE1K`}Cm?8Ijc^HG}*Vn{5F|QY=3fvqp
zeqDm%o%i@XVVOfGD<3gX7^ORZS-?R!sCcC~;$-N%v#S+xigC?bsqIMUL$hT@v8w2Y
zb6@REeSgZKeNT{P7tTpO3Zj0^SQlkW!TIqx&rDa5x{~dMxXrN7et>bF?XnygL_XU@
z*J|&0#$Zj}#4jy~*Z25>lV__5ZHP=8cDzRDubZ#Q-8@d{JIH*y-2(H?h)y>ZCbVTr
z$I$B&nBN@>^B-ayqLhRR&*A)0Vvt!9bBICD?tg}DvpgBNN&RTFKF#1km_*ZF#P4ge
zTD%hWjhMU6p69XtJ#tx^M#!_~-}7(H#yTU{ZzGA|5_Y})9FPg
z#xPFJ;%O0DSq!8iL}x8`Ww3OS<2qB!M^v**KOF1ipZMPNdH|vS4&OLqRzT=T;f3XK
z7=K6SYDi`R_Cs5d_&s{azbR2`Hver;sNSOzsomJ`Z}enq97Z2DRd-H^N4+>^xGT3L
zG0>>C{&v9;ahh1|v=RBh9C@hZJVKSbq=bt16IxkkAu^Cf==v?ghtwY;Zd*$G#~u*c
z=dau+iv8_X!yj*dbHKbsM&9-RMyR8obAPTO@;ZIfKNG74(8mY0vjKJQ=A7H~6nQu1
z=^2
z1J~U@v;+I9VEM>>Y2>YFyZ7ckzzKM?{Cm2A9)NFRxQW5-tsXM$G%@rSH`!ZKFoic-l~K=Zdv_P
zHXQk#{7o_BG|pFZ>dOoo{=zxSy{0Dy_dh#TGn9kAF!R&VZ6oTa`_vGDPk+eU0!dGU
zAJ|XaC49?x9)kmaK3A{Z#vxVkg>EMDfHWVwQcmc!S!*M&SK+;Vnhq9m
zgnpLzb?BQ8p;GuJXOrG1?TraA68FGLEN5xTHo!9
ze&^ytRiy(tJP|lkP>ub>#v$!Rqz2-jZPnB2jC0sSYyX=^Q8$-I8U&vB68ay+QE+3E
z-#r+A>d8N>Ye~>q>em
zZj#EHwnvK=rHvM?PZXtycA|vzBub_7`^^38{hB#*=6lYWIWy;*@y&tN{gcCZ81_D4
z~JU~ABn&mK@A;!em8Fms>76BUrZ
z)YWMr7(cZ^cK8>E5kD?0Y64lg`sS)d@MAXArb
z0_)3RZyJ8eQ3?!ju9j+m|1g3r-t`NbUv0Qj*O(
zF|Yqyds#HtCHbxF5jeMRiLpHxJh9BKkacM$Er>+@$0R
zX!xomg}yWY7T?7_$l2@9e_apSi0U^bU|zN9y&Z92`l8|D*Wh8Fr+$H;?xy0ajqtCv
z>M182^7lsnX?ZYeenvtZ=4*U#A8-SoZ*T8=1P=VUXNuxFa#Er@6n4kI)g~B!fd5M6
z?au~VxC4ckVLT}LQ`IEo?BY$i&LGd^z1nTy$4RaO!dT~^eK5-#cENnfKYhSm4_`mk
z1_#40)=9(e&wBo=C$O$pE+euP;B&#!Avfr?xV1HV!OPw0TvPP@HM7wkpp^-?b}8&c
zb&{+%AYWEnX>8vPf2D7lxG#Z!+iJ&8{(@auT!;KWu$y7Ad=GE_CmIp)cgWv8?2F
z2hL~CD0a6#{E0+~RL8(>RqWXx36TG3t6^S%Y>;}quMPIIwK7>BAgd+T%+h(fSUvf~
z0CUbXDKybmMr{(@+Un(h5Tr2yOnBA`kf&`u?(YH@vU{3oJ=yKQ3k`swevU4G#<(BXJSPtGqCPzs
z7ZAoCRdN)S-F!1BEzZ_&wt{RV6
zvRiPt{wu3vTV=rM(=6V_(938J%#Ey|`6l;_Wu@7{Zv^2_e!kq}U4
z-FD7?=v6+ZjVXXSIogtR&u{v_siFAuJLcv54W8S-;yJ~COy%=_)gcRg_RAM|MX}!+k*1KkL0uNWqU}
zs?;@#zx^|RrVrJ@KJ|4?TXcG{5%i;ft~rCHki`U+n7;>CdTofLxZEAP
ztVj3SdNkRZ`iji^x+41S)T1+NZ-UEob5x`$FF&fLSc3bWE%?|DJ=6I>UjVH$R>Yk0
zL!J3QGh4_tTX&Dpd6p>sQhN?LSMB-$#d%}b)=QK(p`Y5K9zZX+Rdmpw>Z<>XiXp1+
zGqG%c>a(ob>&D7-o;%Om1wr57#CKo?WIl(E7jlrdt9(dVOZDfs7A`txyI}KtijRl_
zt8+E=ouWeJZs@yh*$gkRlKs6j11w3|{NMLQ#cj2yZX8Zsc!WjuDA+oi`lxJ;fbQ>L
z@+I{ie&`LA0@kdc{Y;jok-k9M-I)5p#=ZA{(?+PTC_cM%SC-wc1HSAnk�YvSnaOeRZi4^pBZ;
z{S8f!Et5a`RfBEA<%zV9TQ08E1>0VBeJIvBThIsT@73vmGiCF?8>=>K!ePQ9ar*vzX#NAj8w5PVa_0
z{+;yh*!-t+;RSH+!6wy9VDI=z-+xhme>CgvOxS^cO$Jv{eph|&ccS`~_1!6u@;b3N
zT#xS2&dQLBNzi|Bn!QE+v47FdQ`}&c&qzxY<{QON7~DbKtL6M=A&5E~63Ug8Ifc0R
zC%Af}o-S@t)u*_4=xk!qzT^9qcxgY5S^IBIP#<;s(sUe5RMtI6_q3x^N0HutG2Xvk
z%x$K3U#gv)J2;c;Q~nKl#nU?9GQeXz6C3YK^6K^~^-EE^xAX^Pz+EF8!>RS`L~z30iyx_gw8;W{SQr9DXsT6I>;nYhi)@
zFs!VuP5$g6<&&t78HooRT?hVuNRhu=3BK3q_twY!^tt+Le}Mm!u(zZ4ZOv!Pa5c<(
zEi0?kf&IlTf0st*{I6|ufg0?Tikue9<2x@rqb~iV_eJG%)=G@e^*+#{IvnxRU#Jxn
z9QThb1c!fo5U2Yh$?nwd$GMA?TwqG!+^eS*@889_KjpZqoPw;LyYc9M0gPuZwJf7|
zO^=3HtRd!I;TYyneb8=}oV2F;DaMhbcXoK-sh$=(|M=oTEzs3d-_-;>J!DXR8}v++
z9w?yqs~%6;O6q@KrTx0_ox5QJrzqZ5RqJm$!e4r@MQ8;4{obo@-c0X?M2%=6Fh@H!
zn~ix51&uoNzE#t>sz&{PnYEe|pNw)E7-k%iiw<8~gl}=zBM|jLFlzT371H-2odM2po%KV!d(R8ye}|>YiUyt_(U{t+w}sp1n6q
z`#kK4E={;N%VW)?+`Y97c4s`^N0#24rq
zj86K&V-;#i>Wm7Fu;Tu%(rUqcIW3lKZ^-=bzbDziZ^g$kuUU*YRl9eo5bn1|`fPMv{Yr*EqDF|fN!vN;NIpP#H8|4QHWnZ2L)lgH9ZkZv%8
z{`>cBLBY@mdQ{X+;ydc{I{JASw<7(l_XWfacUmlD0@glN)b~UDweiEDUf_QPtrAu!
zSZ#Rg_Idc11m_mCAaGFe
z{w1_etPFAH@=aCWLmzLx<~I$}Rghl*KhA-^x{IJm
zjelYVzUTajvyn7kv4eI}BlLeY55*0{5#N%lE4v!w&0=dUOrcko$j=I*@xqs)b)lb7
z(>pi_`Bj+U`+Z<+s=7is^as3GWp#t%CdKb65%+{u*l-Z$$#8E!Edsq|S%?I!`xIyD
zD&UePMmUj=&DuNPv%->0V*
zWNy;D6AG*8+>FmH8KM0z);)8b_FE@j_U;z2m+e$QdEF)5J7f#imN%ay{|A9QbqUZj
zBSW+btkAenwF!Uwr5yKRi89W;@r|1!A9*BZHQ=!qyuGHwOa}HtDMnH?;Jag{Tn^TK
z?<`w%JMR6aQgxYK@GE1ucXuG3t<=%}bgDm!PH6&&*ZX#EUJ3lmGXD{-rggnc+a(Of
zNj%cI0sRNVwmQlWr?!`hYZ2dXVNN#P-(_zywf+OWyZ(RbM>OBF5Ax=Eh)i
zx+NqeMBg7VICnYZvf_K`L!h=`beI72KSa}#H$v`7Oj=)s@eWt!xT!$43%2ee`^n;(
zp5MU7{yJHILtny?U$-BedS~xWae39U?e8$oB7ea)`W~H8?Pm9eLSiqky_uBt7s^9Jet;$)dimRX&Kpv`I=cDhfp#x~>+gVl
zx6g`G9#u$BZfL~1Gi;o<)BL&f+}b0+vs=9VsD6L={KtJn1-v+$StthHDCG}}&_0an
zlu-Tpq_AUh63h~A7}-UBahz2ZVB*7%QFKo~>#xZuLj3hRg>`}{{<`o`8?d_U`S1z4
z$FYXrHi9zdUrklPB^J`B>AVBe<=lf2S4>Z`QVi7H=)LY7=(nY4zc1`z2gTM>Kk%*Q
zZPI^%-twAIO+3vfa?FnQX`o_Ok=8xux)GQ5tL%frrS)J?(XvgesE*||NZ){;>!E*B
z2C+WfU+V*b%W5idmj~bhu@`1?dlfD#`|xHQoiQ%KikSs-@7F~uNV5Un2cR?
zf87Scx6yn%hO<{vo-n`N-?bFfY{?pKfIWZw{<@x{pp>j@)Suwv3ah^Pu!r=m=GeziNDV?pD$
zFW*b&8AJ1q5T=_=7EXFgys)R)cc$EP3|wr#&C=;U)&j`1$G|WoOBmlUd2^%
zfc;9GdLH$8<3|Q7SAZo6Qhr=IUk{_7-9jU&rdLHOu4SsCS(xg68
zzI5B4$*>Om<=sQ$2T_>U`FBr`;8Q_r!yO(rt`YZ345ae9UygwlC@5>)>}dC>jU)9fl(pjkgbc{*Aw|J4J1p_X5l+YdTN3P;gB
zbPL(&8Hu=xPfjm=2Ilxz+^2U$WA?Rul#dZP3m)zUeZ^IOp>=LfbW-~ljF0BpZGiq!
zPSQuJvrTUXW~b@C9b|t#u|@nYe|H(GkItVnZ#KZcaLbLICh%82{bL#JOPwC$<4fv0
z@9miWAXm=Q=M4C}Gsl4X$@SzFO{L%sakb;)^zL+XdYcaheX~#J;v5u8qwVYvzrEZ4
z2fb4~#eSOTfb(_pA8p1sYa@v$de2`zmC6pIIxY
zw*_b|r1!Oeo?m~2A!LJg87B2pHIEnjzCb_L)BjMF`VzNu{3+Js?$YZ*@{8^}8`$ux
z-m=-o8}XigyY}@qIH<$Uqkg8;cuSS`J(A_9+J?Nf@+g^B#&~I=5F2{8WlBsnW??QGe=mlixiUf3yq$
z000000C=1gmw8-Gdl$y91{DdFL@HAhB2ua7Nv0@Nh)af0NrQT?+d%0wNs2NgX(pi(
z(x4C}8f7db6rDYsy>}FmG`zq2*Z%Bt&R%Q()_T@=J;$fJH)rTJhdU;=Hm4gooVBkp
zcdzI0v}4-J)m0qkTAUH7lyf*tf8At1-{J7$=$L_Tg&fW?ZHl=C9BigOjT%O-Gghfz
zPa;=lVOfkG`Cij`bcC41RWa#2Y&?O%#D8{-g7PCrsB-j%`D5hxCbVz8*o{-xPv
z2D=q=7d?EyK-=h4$o4l3)~!~}IN!-Y<%Xh7nF0$-h3!9s=CCOF`0{k?e-;+W#n+ok
z&$3wMnD2R8mx8U%B!}SL6s&HQ9rF`Wh`n3#$LKT#nZ(0egD+CZ+&O^H1r+pxY-WzD
zq!72~<7Ks%6snljIn!7Qp$VTJC5@ty*75RwtQC!;|9ER5o`!wzi1f%$Vknzv$e>M(
zx>Jis4SW%!ePPTqZE~N$jLjsjHIysdv12h3JNtfE2oW^O(Ean!1=5R
z6oO|xeeqkBLVd{h*(DkjGOCnT80b)_j(nfxsYf9<<@(qLa-Xmxe|z?di4@dSPShF6
zQaJ0VeB>s{n|Xel`=>)R48u%ZUaQd%CDjzSG8B4?-e)Vmqj1x!x45H&LhTZxaDQ1E
z=Sr_B?$D=kvHZ$f&vi5^m~HxT+i67H<0lo`)3{)MrFHBS8o@6uCS^XM;3W62uG&!s
ze`=jn8pkkjz2WORf3ls%MtVZE!xtLgbH~XIlF!e!$+R5W&Y-zq%e15-2J`nH416QQ
zV$u)$OLI(FI8BNlbh2W>S2}kbQeqKV>1taM%^<%vW{Y7Jiw?a-Zmt$AG_7kYAPZ+7QMJ8D0;M5S(UEQme;ifIU9_!;KI2+4wHhxZs2r8}Sb7nM
zVc+$RSl#0gB5!QrXTf9HX1`-@**tv8=KNj7@W^|{Ka(L6d!f2Qs54d(If%h2-)JIL=n_WNjt^C*ydt@ZN-kIC9eno)VBeI$?kC2?2}F8x6OL7w}qTr0m{E0iC&M7#(}#$dM)mkxZLIu
zw?WJJhBuFSoZ&I7
zHB-JUe|?L?j(Ih@sW&-(J42^EbX{Kq%83g4i9zIIWm7+go?Ty7O=HAkOWH*_?
z?|XjUhiB0+os=c3?nAir#xTfx7=>dFUS0Ks4{xTZ$MtAa@TodxaVwkfX;ufb>VGuo
ze-pKjL;M*?J#SOcO=Drg3~lZRkic+C>*M$HSv0*lqh3BhUgp;ZyjjR6CJ>88MC>ybwW}Ykxk67v@7D+Jf
zP+#NVTM68EUq0FVKNdqX`m;O7Gk8?(f7zD*`@eIa`C4gAVYIzcU_0SngQQ}@7)1&W
zs=6ZW>lEhf)adx9kcQ4iQ;(=22Gd*aTMN4-2)lalYELPL$?NA_TI#|fr6^S^{4#Y`XbNEQZsy
zjkAMDza`zid!H54xNo}He2+hiR2|DBI(8g36{%D`Y~az-GjDU)z?c=+ll
z&oUmxVcwmdG{ey>ey$n_%34XoEYWt);pzYBUA5AR7vsCF*UYOq6teR&f7gl(8CW;y
zCYpVrldELo$Ao1s4_%!HL$|EFy@!$_BQn!ttN+Tr@XzGmUYfiFe+3O3x6S7f->oB4
zeq2CK?s)m*^94AW-C5uMe=m=zrjw@opC&r>!?SN53>xG|DftvrXy4MJxPs_Xs%1xi
zKbg?4vY7LxmIcw91=gS{T^W~;7g*xLPx^2t;5RPG~#*qRk*k%
z6JPNv@#BWqWZyp;W%0v0mPVGndX9+n^+EN-eUC_8>?159N(dhoe~vgCFBQwcsJqIz
zfats9q9sQYoupT`>-0tup2inse<55me!Xr+HHjPkq*^hYT$3t#yps-E(RT;_S
zX&c~J7{*{u`tZiFBwxOC#GEU{H#Dh?DG4Ha9skv&%7t*J@{ZbHM@as<&9vmy#+oFwHrY=w>tCH
zhXYan?H&2tvcm#)UI<%aHOS-Vht})IG{DFe{9Hm!oQBTYk!~fVqm`b
zbD{J`!pH83gUg9u3TRDfkr3a|=NWS)jPS2}o4zqe`qh15U$VkE9#37o7B)KwxLF&h
zy8N_&M&qsNeX9gyRLGx|74g_}OUq!<77lM+HiY-|v6vA!EvAxiR^H-3+KOE?8h)I~
zNbsYevDRcyf8Td8*2&&FafF;_!G?KM7wMDT1C6TmF+7s|jM7cV3J_iG8i5m0KeG><
zd`mbx&brdqA%^&phr&c{Z3%KdvYp$rIMheZKbQB0f9&1ctv8k(5wK^rt{&|tpi9>M
zXjdhVv9^01UPf||s;;VT{KLX$VAhp2MAs9Sj~nSoe3WipI(wVw;OpGGa$dv-<%}Fu
zY9QQP88>e6i90OzYkGAD?d0%PsHv
zjAT*&TJ55~HSsm4DTRk5Pjb>3WzS#%5fitig}MtEnR>O}G?YhdMTC@nzXZ?IAIRh`
zB>aCMba%(m_;L5*lVL=+{dz5n4-&uXE8Vs8f7?L{QJ*icJIiTo?fUdjQ3MOqT5I*m
zwPJRPuMNI_kzauyH^hQmatH)bgoak#KEB0Waktm
ze*q&z%J#cYlKv+@H*f3b@xt=wOv`Zc|3i+mwmw6ALYlVh6yg^q)JZwiWzhH{NwB&>
z@?4MjZFh(8*e^lfeju8HFy6-Q0;y}jj1|wkH6=)|DLauin#YrgTz)Ol_0Q?9=f{w`
zJ>QxdqIQIb$KD}>jn^gcGPix7O#DxoRqLmmM8XB-;)w2W;w#$k8x+-sf-!QER-@7MJ1tR${3t-aRTYp=c5zV=m7J9M`#Wfx$_
zh#wmp#)5JEdYu2mnETZ!{Fcn~`ElF1hXr#OIfplxem^Y0_%LFOZ%)oN=Umi
z&&*U$kC4p!pT{|~f5ZRh8Q7txV@#yLaxM>Sb67W5KXdd~tr5?^l;?&24Vc$bsQ=cI==Cc23?$?fhIlJ$x}9
zE(}`@gY&Jl)zjU}$H~{%)&7W+-q9mYK6>-2%DOP;{QPXc=lAkFa?H`m`EMhbPZl`;
z(~fm6?t}hzK7SuL%dbd(|M}By$-JG@5hr&i55K>+oojE90uptYwqFe=X2D-@2HQHqakR)D*qbqU+oXM9{Fw0`WJh`vx1Atuj&3W??Q~1
z$g`iTyA#GOivHN1jZ3hN3X53%tt{C&Fjy-Ugo;&Poo9czbwCVue$##zoNf_`#f66=v!>2aDfO6PrNmFPT(AdKQG+$K==ZNpNQGy6|2fX
zp@W*YD~%Y)+bvq2C5V9%)Z&CHqZvr`{wJG+JO=7~=(aMaf`Nj*Tt6w(!$5Yufg38B
zG5oQ{q(VFwhF9<08tmyvM^`=$E8KLYqmx(rO~TL6kz`uBv{xt{oiktWIem(b+TI1P
z-U;|r-<$1HEFC@4j+m~=r=zQH>T4sa>1cZJCH3`PIufPW3YA=^BdL{c3E9j_-B_o45myRY)-duLLOh;>sKaoA+7)bUM
z=WSze2FiEPnqf1Abu@|G?RI7$spDxp>a7eUvuluCBZT3XyMi<53K*W2nQY%6h~YC4
ztL~Rz7_RVq#r;=)7~Ub1TxR0{`4_u=_ni@jXE`nP8-)CyJ4d%+O~7!KT|YgOA7OZi
zDz)o!6^84Dt|hN|h~bqXx@;?NVmR)%e#ebh7;eaMS~lS&hA-87;OJeA;YCfcTD>nY
zyf|=JBOS&+Txj=WtAKslVoh&={hp!<&NNkGc!rew?t0Lp-yw}DLOpc*pQ!x-_HJPT
zrqKpVS7;dnoK6BxD3}^oCbFwu9=6CH{J_vD%WVnAw
zggD0v!)yf7FkJpN*X(tu8xtO1PTS=ez9UiA;S6G+HPP>73X|x_t#-t}+n0`9_sepp
z2h!1Dw=Pmp4%PI4^kjMVT$KJyzFJ5oSa)}PV5Xc_)x=v
z#fOdx7>_myg8id}kQs?YIuf!n{2pIQN4H=2u4h19w%G0Cl=g&mJuIv4m%(tqwlMaV
z5C#ft*WCK5hK`gp?y~9F(vjSP-Deg+{cOqRxz{{KN7uJ?aV-;rcsGsvwZpny+s<59
z^MyXrjuLMa$MDcLG0oQ|3>0CumqRIDyq^%LDu8JJh8A
zV=snNwIr(eRTya5^wX88&UEyyxQMypAk_8QO>Jk6KwYHqA8>{`xmBsi^b_h!qO4)Z
zIGi=FPG91P#37;?A|;+-xVbSxmz5y`W6OXf*B!hM#sO21>EmUSP_sX=|zR>gH!
zHe&ciKY=1~@W;|xs+bFPRxuOy!uTSFFR)RjYI9o0IKg+!{#fclv
zm)@I6VK@V+JavqGmO@A3wcOolhv>+jR#RqwoQ{gG_+__39i9^w2%N3Q@WK}YOP@l&
zh?7@kDW+ohhL^@JeA_Yn+Nb04t5V?n8XL|YZ3O#os&_je@7LTS8tTEHN%V^xcBo6~
z?e$rlcQBk&O|?l0{7kOg=RkqH?G~pG>RrchuLJ3Etm+u<9h!SI$Ay7X%i51Wh58-U
zjvUj5cw|Z~XIzfb(c{};&zxZ0rU&(d)~jH63V)=7?O6fV(8$%#}P35bE@~_a*fWan%qE~d39obwV8OlI6
zWk25ZHpTG$4}x_oAkOFgKa&Ov;k-EZm}hJgh7TJ?G8)=Oz#t|q$;!y6|~
zWOKv05Wn&Od(bir7Zd2fQga!I`vp6t9P+8z&i3#*^!GT&M5uKD1Nn}M`g}OD^W#Q)r=i|pM&)hknF0ZEW>y$uWdF9@klsgaSH}jfA
zSuW@k{gl#zD#%xD5L*Mh584h~-!p|_IFpKX&aRsn?mVKqFSQQd!{wch@z76yeMkRK
zKF{BqGKiZv{P16Y|NOIP{vOJ;UEV*%UL%$0l$6hEdNdVFtW5>s;SEuMqw&JV@lWv8b5rbJ{oXOakR9FTi5@jesCcbxdF_W|r0!m5|90Qp9?@GIYd
zAKdm}u`=-6EyYbnfR^M_RAT*ar8S>D11wNB@eKjIwWL*|3H194Lw+3xqxI
z0&!N5Yj=;xZ~`ZP5n$p4Oy9Kkn-Tb7U^H827_Y%94&}i9o>6$y&jwse=$*1SV3z8B
zt8UQuxTGLM>`Q^6pwVqW$AoC{Xqe|~`&qYW$fLZ9Q?)$k7gfgF=o5C-!_xOb@0+a8
zE@EGgUJe{F2mI>4N@p1C&dD~QRKSq6q_t~6{wUu^4!#=+9vU#N>)OR}9#HBS
zW|snf_|q)(h`O+RAK!Z){5?JEUw0Skx;=Aut`5rxs;~J(KL>s~yL{_hK%X=-OX&rD
z6|S=)Yhm6c@dryT5xjI#w*u@vH}R{u0cR_-GG_rl-aKZb%OIYNX~Dy(U@uykX5s~P
zQ@yd#`ZVz5d%|0b!R|ruc;|lL)n_t}D}ue;#S({YfFY{8g^4=kvN)Y(3VMzXl+zx>GJqAh?%j6*KOJ{h!FjcCGnGXB1f0o>3)
zdWR4A3u!zRG7ECMK(#R)K>heE%_JD_I;NF>7dYv}-44Qk!c(=yuN3gQ|yyA((ClZ^BeI}%|RYO{;TM-=OarLi!s{$!)ZGT_@j
z8XhLpQLj@-~db9f9#5$&?iaz|VaU
z5#s=koq04_3cUHibNKh!5!7f~^MUBki`#297lJ%A#&*0Hu=6rc!B;?~b9Ul%Krxm|
z9VXE0S(kU`0dQwqQ&w%@is5c$L|-=7-*?#yoSDoOe+Doo^i3Jj2dW|BAp$e
z0`4&JxJnOro2RsQ53El&$^Fe6;5^RbA;fvsWIQ0h1?t!@Xn5cPAghL`w-un-)s_Rq
z`D0wC?^q9L!TZ5%0r-1QcXg=(993K3DEB*druYvhDMIP%m;bz5{X9=V9&12{!AOM0fn7k@>M}Qwi==*?
zc@6U1ZzHc-084MH3KMm%l{3a!0^QC^B(WdwnPYB*
z1AZ=kpDhY@(`8AAI^cY(`06dM2J%xH9tOmFey3e`!*%eNN2YzW1$o-*h$-ScU2^<@
z92u}EGmFg?kb3R*lSas2zGR8b2hbZfU2!B2{H^P^sayg6UuHcBEhq93aL-EGx<
z=6=wNt<5cw0sCyf>CV@X$32@!0yvEOSn-g5fLz@;p-&0qw-d9I**a11#ixI7r+#Ua41`6YqI$3%b23@NEU33+{p4mhv)AqE8h(dpn8yk+&j`
z?MXmA4-R?a{5hI1yKXITK7G+W-+{00j?o)|dHpq3Jct1PBWQgJ@gC4p%6UV?`DCYy
zC~*PCEzjH=Mv3eT;Cut_t$&5qM}I_^7c(|2_odLbkDRH18Q`B@4O3f
ztv|E4*9YuYCq6F@1V7AdS9Bv`+|g>;#5zEw(Qu*vLONsd9pIclW4Aa1CbT_|xJbn35HCX98_EMXipxQ6b%}C)
zIf1Q~l|vv8Eq>0H0yw%~(nbl8%)Gx%9*`7oIhYPuV$<2`1X%pEK2;g)d}0b4iN26%
z(0GXhMy*sUmH_k$*7RQuctclvg)89B@_~4w9=z6gQ7(g>X>DeQGVoa41Uq716bqg;
z6Zgd91+|7mod}s8zp)a~yl+rG4)oO=O<%G?o)u4yvh;xbmh<(70lkm-?i)}C167^X7$v)N&kXv$xH1WPVy3bmR@E1n8axe4hDbdJ~MxB>0RG@5RHLCTSfYS5|qLM4b29&Z62!
zfScZlktEKUlN-hAbrkk86;_9&95KY{>%IkwTFu*$A##KZgGO4A-66a>LmUO~v;QL)q
zQ;B@a@D&LY{bW*GPZI>ZTePBvc&{%y$;z=C*4@Q&)dK;Vu}%gP_u5@21;&W`iowS#
z*NOM{KI@iZVx6-{<;gwZ`=$2}zXbgLp~9*K?AodrDN-=*@1NuJZ1s=q_EF8QcGk^sR(;bBTkeU=5aemC+yY$Ki>@MtlSY!!2(EnKR
zG>k76St+?2&M%*^ofDsc@6y(6B=TXy-crQ}^ViUxP-g&{4&`ko&UtIiq#YBW-*uL@
zd8_L;irY3oqZbg1@Rm9Fog{-?TeDkO|0jJencNNY8%i-%
z5qUGvxM5cUXzz1lfH)t`{c387I!sTmzf9bJS6)lmPkd+XRjzjl2EMFNCFePCX2Y+&
z#=u!ZO!JBR>I%O!Z)Na1G1Y!S81S=mT{rO^>me*XMSS19jn-S#4D)MAWjy=_c8)d&
zw-Wo~@YHfD67tY{;Iui>@Ar%K6a3*kKN4i{eIX!i0?%ZDyt^x(+F$^vu9p{9$wxyG
zS6rVMi_p-H0}FF90JQ^>!#enBNM=Jy_tF(Kq@+9DE5=JhY(L5z16I-y`S8y$Nf@8t
zwGvbT`^;oo&Xq6ZGUWXs*!?Hs-pc4k5W+C2K||xOR16&^X8W0;M@1&G=Qhb&P|=gOS|sKAew5^M^td6tAIbIDUHD=|Mok0j?vKTg
zkwSs8`ja{`B7Nr=x1y7g_*$|2#YO|DuZ=D9QT_npQfOQu^JM@@Zk`l7+0%y}(@fes
zseS0_DVvORUJ^>kbpFgOLP8-_{MiF>5|Y>(irN5=1dpobC
zd-_n2zhe>EnU{h*THdba@`rq{OW(6Fh>E89QinII)6lrxJJIq3G*lTuTgnW)B>8+)
z6N8Gv7KGHSjisTRmHQ8W38tYsin->584dBNU;EVLMMI4z%a8S3pds(?ucL10(a>~7
z08^zn4Q01$wJ!vJ%~!aetqY-{Ssi2D38;gJ(x%GZeKd6A5&M3ZaG3W^X|Gi%4f(GT
z%pW4^WqAetrvnuwE&o0{rb9(4YIJLvawRPNhcOkA%;7)BWhvGk^`YtF%II=k5*j*IrH^>~(F-1jvU`#JNKE`?%V|d{V(vI-!2ODX7Uez`71bU@
z+Ev^4{Dk_}@N7D!xtWA$A-&V;EF>h5_99?Hsvq@_FBP7>NJd5SvkUicA3)xbp9+E@
zKkC801!tnEs53+Z>unoE>NS$q=GZ4SVBTSq=t$FJ)qu<%8N^4$>=0=R{D6<
zAPRFfeGuwKL0%d0HjBcb&(D8)a!-_om|L)m87>s0_%`(2#x-Pgj4Ow2WT+2$J6GpD
z6YfWgc(K7#8wZhsf83iyYbwHo254rBX(;xp>uh=C+lB%9*FP
z%|7Z!Gi;61{k?tY?c2?YFGR@bd|_XJ`7H`+2qBaGpicPpp2S2!{du^H#O$)7A$@;W
z%c18~w7Icg@Nn=T+HG<9#1GbfR1rDC8plIILi#8qM0EfKnM`zYU#FtN?^6M3aWtf5
z%_Mpr?6uY>r)a}Dw0`Sadca<&|A=+HQC$OQi74ygC)p(U&SgtsKp*CK7k7y@4*4fO?f
zN|Tcze>KJyFNOzE)oYEIvH~(<#gt0E&!K8}RG
z}X?3n?8bI+rk0)AF29eW6mT>Wn&}Y^m&JNJW-QkN1zE4n5
zU32%TXPn
zPQUh^mzSfU)Aw#%khi5Fui|*~;BQp4H0?puwgRZ1%NzJvZ;{b+c;L9zYqZg5@5-dB6UC+f8x}8SRc$-=O-Ogzosy@Em}?ewv--de5GW*2@kG
zRj;R_1zg8JEr$G2_WH4{N}?hSsS5|qV1L!k1zFk5si=5agEq@v3bOV-WSLn@MjW^9
z9C|KGLaVv+SxaA#(4LmIWy_KVkyVpJM~4XwsRZ!#E5rWoQHgmV4d+(mj_P3y>Zhr*
z`ut)uDzdk_XkgqqfChBjK6B;5d2ua@-VXUoJAbb;;`IO;@^!j3fYA`QYYmGBoR^j2
K>1|UyH1r?Fs0AJX

literal 11976
zcmeHsX&{x|*Y{;85)mmwnW9V~70IwKMFYwZij;XQvy@~iN(mt)WXfDh6iUiaA}K?f
zlp>CEI45H$ob#-!xbORY-}`-@56}JlKfLe%eDJ$|Yp=ET+H0@9u6@XU8v0^;G9_Rls=6($dza{g0e%xm6foTpS=P;l7_X86qfDvN?b8?{b8u9!~d0zP6!1;PjTKL<^
zykAaC>6d^7TZnOD>==`WgR8qQ#shKv4h{b!l>a3{P7I&3;{ZjlbM!uH>*wm};fwL{
zVAx6+oIgt2Jl(x~9DRLV?T$F=9zEjdqdTvvtOIk-&(HpQelO1>0S=DNe;dJkvcUPD
zc5HKTTl?Gk{C(i8zastp=TE!E^LCC$9Nir~{Qll{?s$Xl=3fq2u(|cRJNmj<`vrJ8
z{#q|D49>6D%){5u=cv8kQ6EQ#-Jk`l{A;{_wRdto^4p&6FZP6I1(!{~ru)mh3$SHG
zp8Z_i9kHci=#TB$xER~WfH3E$cbDk!d{bCq4k1TV2_3u_hs4gJZ49LvP
z{L2CRoIe((xddQIMDG6h@y>|}_l;`xV%
zoqs(26A$y}&S>dxZ}|V_l&FUJD2SnX{khl6+$sChhza|b<#+B?_h02#>{oFAXMTxx
zF~2hZ_5WNa`!g4ReNx_Zl-^c`;iENSQq`px-klQn=G$Ei&&vs19Sz9QFvBAP@+!ug
zdVSzdk**0b_b^;T&p);n>;mg|AK|XR@Tthv<@@b0{Dg%2w&TtiuAn?q>V6o*z2bt(
zntd?*(ALINW`P*a9CkmJITXV$C`E(|MPPW!&4Kem=P|soNUdDt0*0TCRr88dVW7|f
zjoa0R4CLb$EzcUrKvz}cS5-$dkip%LR+sY_sPn$t@|;Hubn;7fkW3E)S@#~_P|bqj
z57j3X5_mAYX4kf0PX{_m{xGC)!k8;7?p0%bL@gc74m_v6ETJRiL7P?gvgt^5xm#sI0v+l5zdDu<
z{*qp@+|bOVqdk%A+3l5dB%js%^7bP-dc>NM!IDcyKa5{pvcE(}D~&#qJ>nUN{S^0Y
zBX0(}X0JKJZo)wAEutme&I}}eB8_izI|K3S4Up?rVR+)_;0(F~hF{7|wrdi`aLzL;
z?%u;NeB)F8yD$7Oyj3Q-!pa`QeZ+5-eqTng0UAEjUlhUVe_@{JP~1-nAHhza>ty_Zf!Y
zK0c(L4&x&i-ubaVf_m6$NpFJv?xPCNv{Ykwhi%7~f}d?-Ex$hP!TU%v=x!^)l`g)SiLi9p{qmw_?=$=e-PcGv{WSBL@QsJy5r2
z^`WBx#sf8BurIzGGP62~j+R^O{+>`qM{Uo1*E66lKieMQmiEMO-}@D{{W2Jy)e*+g
z7Q#UO?=-f(sG}nb^%8b%8#>y?wC4;H)Q|CXzS7oFI?CAonP-VO1FcaT^Lq#D4(<
zt?G;6PqfZSG)rK3L5H};OJfF#Fx|(g6ir7VNi|zikI@l#yu1bL7#&eR8N8^8VIZj|
zheiiJF;M0Bi9QOPKk_Maft+LvFW0e_y{nDkg{#FYhb$Q=HLbgf0``1g)?Tl^PDf78
z=$r`DTj=zVY>yI#pRA7SRJF!%{CIVV1MJVZ{qvh^%V8g89UX|N+UYPtN1lS$IdsJ^{JiP6&sSqHd{(jGykZH4ABaqlVSu0R4aU8Euzz;n
zTJ(SH!|)-^)ir{f7-;$QL$_edOOtNMG>LnvyE_Wo(XY1
zhVvppUXi7k3Vrdpk&D0%43GJELViUG105Y5x<1kjePvvwq!aQUu+RpXG2s
zT}tg}%;LU-;mcH2T9m-g)ba!NgOIoF67&JxYzz-NlpfEv8O~2=?$I0<20Bsk?!;rL
z-)CBpqgoITXW9N4m!ov_@OIb}M_9LowSM6GO^|=VNPC;J7~W!hQ~wRb>9w7GrsOt;
zZ(!D4CAc5GL^>iFs2SXPK}UI!t&gNa
z8K_kD!wqi}3^yqc)_DYRKI{K^b)X3LA>bk3s2c3o?noA+YYbHKBRtRn^2cT`6e|Mz
z;$!BWd@cg=*hHZ$@uDNw;H_PP@(e^0>n(~kg>!uT(=814C#r4|O~ZQMwDPLTZioGv
zIC*_3^o5Jd4{-!8fp~>FvD91!;(o?4SPA*u`i}knQ#jw_oD-pz#~?2w55bu&Qn4F@p}y;QJ)W+y#PC@4ZR!W%Ju`LAW|n@Mj{L)(
zC{DD}k$aeSk)%3?@6KJUvJC20R?pne4f)
z)U_s>5I=kK9BOpD<|!POQ%MtrT6$#b7Wa8vnL-K&yBe@FF?+^dxT
zr|wnEbNTwe{Y%E*=W_!N7wccYe_cM6*8cbMN!(HA@2Ot!{O7mKZ@>S{-@nh-`p5-y
z?*D##elehtPRkIA{}aZxxoT
zoijt|IWL=Rg##7QIeQ)!8$;3wuanB&`S(ge%%6kfe)XpX90dd^(6O2;F)84M|J~8h_r7P
z@NLT9l8bEnc09QL%DA^BODQH)OB;?UbX5$SU?B2)Bsvm*zA(6~U!rxZWj>sfVTSg5H3v
zj~=nF0hf*sn*k2`uh1R>yLj0qlnNNJhO|Z!wUu^4ypXh9_lcz=K7g43Q#!!
zvrPd%OVZ4BiMlX4SDSSh(ds_Z&x4O3OiQ4`$c27J8S!i*LCIPsZ|EP;4#P75DS1^eaIX~teq
zH}^L-Tb>4&m|CqS&ht!zq9dFET`LN)<;{#dFFQEw-BvRwQD!!#}qiNW|vbvE(O
zz^y9;0*N~47rr|(1>9F5=lm4JHDY=(o;Xi3(p9!3ke>>*(j)S5h3!389E@vLZB$-7D&7XxOC0B7shcU}SHZ5ob0LX0O5@*4oZ
z_+C_;6FBpk2a{#M+YUX2f1e#jFKz1H6X)~7j=C*HAPd3VZzyV;nqX#tlEcdH=IWn1H2mwmt&l6ex&02YS6svyq6wh)OB3XD^(
zTk@;|?mF?XMi+Rer?htutWWQ%`>R*Lxt+&Ch(0d!c`JKorI`YLrR4Ic%utuZ%66}3pstky
zr)tM0hENpS?o?LLJMhCvmpI=!Hx+#M0L~7lv+n@E%%w9m;(&MBPHi-Vd0m?ON(7*;
z$&!K#1_6IcI0mXfKTo|Xp(Fx!Q7l(%PlEneviD>$;K|6*=LI0|U%dLkYOoj7;&`G3
z@$jT?`;-ImM1Q5a5$~CB-j=2?(ErZQHca@5&=9>f3cOooLP!UA4Mm;50Qd$w9Yr~?
z-w_k*u@U6XSI#`Y1^7d)=LB(H(0*pFQSi4>TwXf__}fp7GerM=b5r`G8`vEWTUp@-
za`Du!GcQ4&J3joP4Y2&SiU?8X>vKlA@4lt(OU4
z{tpXjVm~#0Y`k9!eyv}{Oc}v?RH@oWcwv4f3yxV9u(LnTJ)r=YdS>V=(N`aQzLF~l
zoMYhvXI4O=G}k6p(9hwGmLu|X*I(OQ7V;XE@hY(c^0mOPCUX&>tjd+DL>O<&6rc0~
zT;12j%no{do+9T&fsb`I#}ntEbvDPt9MJksWegeO3SL?(%?|UdSf~H32Jk4pROu_o
zD_ct$mVimT1v(LcdnVZ|e8Jzdme9HzFkfd`Yvx=Fqs^%x+$|?`71V7`)O1(4y
z-@f+JI0$;@>vQkPfPISJbk|GBqnR2>2#0YG3qJA>kZ&=%+@}Qcyrk>N@!&_l-N!=_
z#>tim`8t4)csQ03=jCLc+inuzwX3hkiM~!si;VjUcHcG~8E%LCEYDbx6a?{QYIt}D
zfL%$??)xL)zmSa<>JR!Rw1-8qfVh0@y*xlRw?%wfFkhm!nNvO3XI;w35d*u3W%A$N
zfgS0ycg;3pp4dz+KfrwM`vb&%4#0lqRMNgDvJmok?p&ecViRVLzqo5|eW4&bLXB3NA)xUkFh&!d1|
zOg&aVA+EKGkKFlS{5XHh0VUvf^35?P
zcYW_Ffw<()Eb8?EyCq3ai;sgJcJ@S_NEmmrSTeB=P;n&Oune%|m|jdapl{DRy90of
zRgG!(VBfH~(_|3rjK*?GJb{-hi%jhWK6I0g83JF8WO|A9S}Te)8v{CxQZ0ycC&)%y0_fL#sdJxL(n`ywgV3b>VY#-cmG7ypdg>I|6P@iZa^{BqbQh!XdPN5?pe
zD?#pbaWI<~(7b(r^*4|Q7e8fB0USIiWu*kz&2q3q9&j{a|3Ermxm8!YBVfbh##Cjn
z^N%fRCeDR)llpTUFlM=G@oGT#U=9D3fay9~{H}odm6Qac9(*Ld1}}k~MSW(cGVny5
z%eKV6XcRtaCGLqQ3hQ?hbs}kUB7ZsHp}qn6c+i)1FnP`fc~%V?VeJ9=E$8ee3gBRW
zA2+eilFqJGRKVWn>zxJx@7Y|iA?oq@N4q#%Kwhbx(!~2J@PMTz;Xi2VeexL0cfjGO
ziVEnLhpV|0b-8$tg_j&~9v2@;qW^yyKBjXPuqgA(v>J>HT=p3y-iz*PleA8dD{Q)d
zmFV|doW-<`05`r9D@F8~#OoKf6ZeKyt8(*+_fos$viLK=g;S505c;dRgf$60xBH*d
zH~^*eJ$6sRc)tImBW%FCEvZsWfVFhvC?c*-yA9FAz3X9PASVX+bXoHX;v7z@q{R_^
zGfGqX@=M?cT~AYqd@2as5+Tl$QGFv#7_jgbe_cO}GY7G8?tyi8vR?5(fP2^`gNb{s
zdXUg4abMB@keE%pzb!4>iivg3qD@t$z+I#d4m}4Pd;iFy4eValFjCgSxb>ISp;dqs
z*3LKqx4XSLpa=Aw%nCh-d50D+3z-3Y*fiZq$emT!vJ?Gdy7b};E3k{(
zb^q3W;3572#nUh@D7su~5A-juuw4@$f$M2$G!yx-uQWCoM$%g>2Y6PSiz4T$Z3N*vsOYsY`(G
zE3`1&18CvjUfTyKx>dvF0bsXm9`XmAzVQ4_JNSFz#3{uKyvX)6B@>YEL^!h{_$kMV8UbGrVp2fdSGoMsyp_Q(ZR*_x5x`H*4c)|dY@mqr6!Cr29<95u73SZ%
zHsk&{*x6ZGZzJ}{?(zPqNXP@_&}lQ`yq6X0U-pN7?iFb8eF5P22|SY(^6sX5YJ&lw
zx^7-rwEztTC%RS{iPF%{Lkn^;0JV-Khjj|lkoty{ZZ3WrlGB;)6<@$*SIf-F|DC1}=+5h|?Qu4A3bIThAQm<64T#Mg)yEHb2^FCFZe4+HKM_C4izqUh7~x;lu0lpTo_bX8AL&Pz
ze;%yMQzj!1Db0`f%rZHhkLmM_<``F?|L)}4@0X-LJDCzsl
zbNRY7wDj>Y=4uHVig~B`ZUOjfOkDb8T?o{}RwJDWsDqfYmg?REH1y~J$3d5H8scdw
z>$M1_A#X|Hf^SeS3-}({wSW
z$hr8v4{eU;byeC!LOk4MCmODjQ0|`au9MrS=$+nP0nLg*q=1*V-gF#5jz0yei`~iS
ziq!X7(*hD|%26#!`P_#_i>sq6bx3F|phh1p>qnJ*_7$a({b;Ae^S09tupgb)21{QI
zB97dLVq#hYXnW0$y+5J86+Bx4G`5gXe@O52W>yjsOM7-~Vr@VAGR7q`8$(7(^0NyL
z?w}ys$d84=ke_YAzJ+I^spx5lI@a4UfRyW`EX|gZ5#8w3CiBH4#3%jjmaqp2B^Xu~
z--{!|cTHCM*tr1|=WJ3Q>Nbe3Wh7WF45OhvQ6H;H#b{`(6^qGm8AQUbL*HzaB%=VH
z9QNUFeaPFnHt&f@Kcewr1E)3)AU^;2S4oys#I=e-GhIYOx2{NCd~%wGnzpR|U?D?8
z-}By7xNRFmRV>rnXCL&VN%rRH{@yn({%=pAE}Hv=4+jq*ee+8vf3WqV+Q?zHcs>#m)<+>BDioAxJkhl@
zn~FNWPaR8(hjmyoi$#I`_VvjrTF{47wymKb+ebyS5$k%-eWoC0F}A~1*Gb6PhCPJ=
z=P=E?__KH`1?{9tOK~vMP`l6ymabEf*IMPfOQ4QLTH}Ob!B0|)KptNW?3-l5xZy)M
zZ_5+ATd5Qzw1F$U;7vc0>EFnF%94b1MAMFAU;B_OC->>o5GS3dUD*xhnH@dmW(a;i
z6=f!zFsGrf$GfD-$yCH%XJr0tXaH5bRFADFB%?`*D@IQnNhtZ#9{YvkedyZ#CXUmC
zedsBth`%`x39Z*4uXzS>UK_oft6&0s*1J}0Jcfo+-dip{T~9^oZEuh6&xL&LHQ?}&
zC!-gNj5PUp61rEgN^}?_q4aNZcHf8l&=+0JF3A=OO7M9&(VjAZPRFo@OKhZ~?Uo_V
z_Hd57!WR{OpP-_i*7Ort@&Iz1Nu%ACA|os@IJvr-gmUzX1Sd>LsL?;1C4V;wRlBxq
z;(ZS1j`xxr)q#vEzxGDS%MBv$()b=<{RHRgI7rz$vm;IGc>r@}o%d&8d&Fx2;%YB$=;7>p%ebPG+`hstd
z?VScHstg^tAT&IHzTVc~H!ws&g)4&P%Awz{yW4StTt`Mb<2G+lc}hZ+{xf`s;9Nhw
zp5
z+}mk?W<444-Mr)URF;H9mKLy;JtLt5Z5>ONTpd6zE%u$A#xx{;OrT#G_E&vVY`HY_
zt+<`FLm1RgbyscFB2(yZ7BL1!T@=)#?e>W$mxOL#J4b&9`HPP#?TUCwK{Q{-n-q+O
UmblijdO*Ldlt}NG;-jJe0MvN>7XSbN

diff --git a/previews/PR1624/tutorials/out/solution_000080.vtu b/previews/PR1624/tutorials/out/solution_000080.vtu
index 53ebbd93839391114e47ced14ea9579ac623fa9e..695c447ec9707c2d9eed7e79a389bf86f38b077c 100644
GIT binary patch
delta 12218
zcmV;rFGbMVl>yb20kEY50ya034FoKc(*rpHGLsDiV*)ZVvvLI37Jq^a00000004NL
z6<2vY)k_;6B8g~qBb2nDi%Mw0OhSZ+3Q3fZM6$mrOA%Kb7g=&G5kf?^>`O#s_bH@X
zQppm9O6$z~+`r~?erL`zzj>bTc9fJoX>S=KG8d!`n~D(`9=i~*r4$iWAJK;^VSU$h
zG-w{4FI23lQH1qZhJXAk46A2t%cpm+j;aI*-7i7JJLZM11MGM4kf~IFdVbA@=Y-1<
z(Rq4YTyHrd*7dxitx||cUY742mqEnx`xDtz1w?kYG-%2x!}sY8j|=x8(&o^^u
z24s~Cv=K?W!+%lo03xf-?|Rl~j0m4~vg&dRM6^SX{r0g%M1QuUHf<|6!)hT9g<2IE
zPJgw+*4lt!^BrAd`;IWIKE)Gu%8KC>&8-JtIWjCY6W1^7#c+*$h=z0k!^MdM3K0vk>b3%xA}YE`TN^?zrNa0-14TpF5)c3-&DtP9Dm#>wo{{9jQ^sz>@?Rk-YAL#
z`Nh1$gD74aC-%D1m*QeSm(xF7DNe6HVk3W^;wEftt?fl|YHLqkhabhO91ffIUZpru
z&{=RKnBs29Rr46u%Ixv&g_kHE4=!lEbph@fr~JEYh~l3+9+r31QGBU?L4^$T#X}^G
zUoV^D;D6DMR<|OGkLGT0i*KV?O#j2yfl>Iqb-eovO7R!Fl~Hp-4F6hSYFE61VHE>~
z>j4WG9vOd7ySa~I@!)UK{(1~stk^EObS=Z%J?}l~Tmt)?_q@8hl;K!D&#ynWFq|iE
zXsvXN;n}xu?_&{mt;c
z8zwJJ;Q7CI>%zVjF?{Wdu0(h`;3wFRJ}6}P>dw(19oT=j_&%49AR-$R0;e715z$aH
z6%5rwq`L9+>UYNxnf~&C^T-a7ufkj(?ph=ACeGX$??$AIPkEpnzTf6mKbU|K;oLY9
zdVgO9ktZ_8qFn&zJ%QobVrGaOTjAE)=?=UR*;~*YiAYuTkXTwKA_{}KnfHqj2~)Ky
zR(pVmtNddDS>Rp3eybaH{~+=h}0!^K1;cUNbbDuvf?a6*4xhJ
ze}?_8T$ysR4RF3m$x$6c(%J5mSCCY#D2_p45;`s=EpQ!cWmnb}MOm-Kys6nK6-v<_sco=PcLHZw0-0ni8Ef#<1c_;p9`E3>PU%-Qt1yuM1mvU|l)I=9_mo#R0Cz
zR&pXefH!_K{64|3I-M05UKs;;Kh(vOdIOQ<#5&#rs9PscXilJR(v4yJ=xK(9Zcd#l
zxytbLE|HvhafU+|brn|DP<%(?@qcw8Z;BU-)mYhDQM@O|aVE|kaDDo3MawkcrM9-p
zqd|(jG|@ZRFhp*A_E&#&7!e_7ULD2Nh#0REigg0~m~WCw;MD*3f4D8#tv~cHphMWGDmHG%c
z(Ft2^&(aXNlQuh13p`S~c*tl9csqplW?35`!oMLfVEZG6lhj0(n{zSj+Nx((dIfNE
zTi9q2<{B@Rmvaa7R-h`@JnAg?)MhuoMQagRx+);aJ^+!;Xn$7!eJvv53nX=2pnmza
zc%GxFh}0@Hl<7m?tQylCM{+<%dxR(R-%@-tMUB4|@XfE~`pFOc?)|8T(9#)-x5!mD
z)>km>Gimp+M<0=v7W>$S3`8Uf;uJKY?%>9%s5{V~OPul4XP{dbb|!OK$|3UjW$CvH
zXNC)ZbdTtlQGcx8nfY)(_?N$7(|6rO&~=X{UW+{pXO%b=_JgiIut*Rtg?^7D_I^#Q
zL!?r1Thte*%hOk;m<9S1Yt(n|$!XAy){+&1pw}BB10Qw+F1a_vO#TUdiRKEw(c(gJ
zm}ZGDeNOS#ds&fHw;A>JKV&Bz|WYxucKXCi@bU#^r0
zzku^YIdw!E?pF{dZEyy7Q5frSyASjy(po}S1J2VjN#Fcu3nC3&g2fqyh)4@J&&`D(
z^2ySdoPRom$U3*}#%3!J$r8{M$j@fDB5zTY*bauv#s_w0^-?ThrT=SR1kC@A=F>Bf
z3&hJO<9B;fe4S&bVwaA{0);8g)rW|5bucyw=Q%mI+vq|cFPtd*cPa^TRrg4IuOsBG
z166%_FrN!8d;?qC8NQ%jFFT{l@SCd2)B`UlzJJEajZBQDIJe@wW)jrrQCMsJ3HlfK
z@vGo)5ah#@Ph-O1CyO6l=kWpG$m(s(>4P~%1+^8w2HoeM^UUmpIr5b#E$UZ7BkrPgU+iNUO
zAAkI7*m7vd2?f|}V@a$9tA0pjyBA3!2N6ObxBU$j#o%<`Jd%%Z`
zErV)5!<;g#PU(r{BjWq>wWz-*`0!FIuYZfO;FprKCT75!qr;EL4nc;+nTJX+;I4Wf
zH@7k5zR~lWRJ}AAKB9Z^jA#t;A4AlhAYRz<@Q3KcAqmIcn-NQNw1>&R~Gm~pnq|w
z>7NWAJFZqVI1D-RL(h3bBScE&-h~w;BXT%o?9NZ{@sU2$qwz4Ol&jej{lI^Tx|Hu~
zrr_&q|2Wn@4mp^!$!QtzXo16Tts{R^T(2tjT-ouz+!nXhFdf#(Eek2+YAs~ho$`_4
zpsif@N6ZmSxBmF)4;uhF*I%ltA$!{qHy%a9mrn_Ddn;#6#cZdSlfY&P^Z8I>`+<@ERJ|hCKDe
z^_y5e^rxgpgVzJji5}s3djjUKtq@$)X^)8RUu}Bdfv;1${@~2b0*+G-xqt1G2A($P
zG-bnmh81INPlFymoe4-%0v-C2yDfbe;J~G0GW^GSMBZ--f7Sy$tdVu}uxLWWWbVi1
z2$;V{!1<&Tpl9*_i0&gmHE|Wg!MuYm*Fz~DTz~g=
zG~{9OX0`hSn~4v$iN#5pS5wT-D3dF_3@$57%>M
zHk0i#7=QB$00000004NL9hY}J*ZUX8iEEUZWZsC3BsunFl06b4q?A%b*T`rJ
zEh91_O>&Lw-<70+G(@^l84*!mWq+;T>)gMdk2Buq^B(7%_xoejWKp&-$7I+UR(B81
zFd6HD3|eg@18KYzEn$dAF^#kaI2X`acj%GY9+f%Sem
zmn0ULjA-s4^9gX&??6L5+Xy4vp5N64e0641ND;>fqwj`3=h&hVM#;|O=flAuUqQVw
z_7R5PQ9Dyfuy60fe>~wI!_(NU1X{aK=4*qNg;FOqz_l@g8YKH
zg@0PkfDIV$`-vz1CG2G*H~p)?2VXDy?ggD4`um&VKdV=t8V1U4;4FU+{&Jqb@(MI*
z^D;CCFRJ7=EP(F!@~fUA?(M)e>`oYOrg`+;H*lTk{YqJU=i1j41wC-(@TAH`=vl0x
zjGGvj;uFgZ0YzrT-pzxTRxjBa1M)9iRDbiPd5*1B_9K1in&tY~&+K-Uz;cZ5V|i*D
zAYSI1gv1Zv-_0jw;=$oTjj(gDZuFPo4+e7VRGoptcL_pgF*2bffulk7$Fzs~GQhs-b6z_lL!sjrnv-hW`c
zY~$d>Gsqee>x=k+urUlMaxxBdoyhFJ)=D)i%pqY*vf5YLe}{z501Hz%A&`|^10
zF@XmdcmC?Fs>kpb`0{>tH@HoA9oqru{ZshQXM**&O;px^$t!nj?|@x&#D8*;6_}x`
zS)zmd-Wq#fn)1(Gn$=2q`?GVp`~`Sl=~i0-#yuFXl-LLV@!E%uEbvlzv@Hwz+#Q$r
zmcvh%*)miD+1icw6gya2H@MaXepOsHo;wi#dUb+>9K~1S{-h4QlP>RPI)~wQU(Esd
zPsM5W&QhJ`N{L!Q&e{3Rmw)P@Fn0Fo2XG)GU4xzK>W)FdG~|LDUeQ4KpYG>A#Dn;%
z+TL}lu)9b+dL<9aL?tO@!B1gX#cyxH6&%k_7}37X_(gkz1uoxn)Ie984-xxdZy(I)
zxrBIWmV$g?;K+ML*T0}|t*a2DeOPpL+oKQQ$3@Nid0>y)-L``A%75Z)wBRXlQl`P=
zGV=I#0;^U6a$m1r{1ndZh}+xfTj(d;z9~9282|g@`gFr7CZng0E3pCo;&+n$4WO@?
zAG+&}?@1+8*enCH=2wN2{`=wGf)!M^ZND6hke|*06+xR>OvbR_lColaCr9VhIu*oc
zt<2Y12mZSy;ZzFZ@_#KJ3`oRx&dR@FgkpThp44qU(C0*sy_-b*d5cazZty72v+N+)
z)0pXAT%cHOSe87#BkMNnMgGptZlk-P7k5hD{yX+LKS7~<4)VatyUVu0USJWp*$Hus
zvv*YGVBEy0p~6boPrJv;{)B#4N2%Zz$nW{2L_EOigWlR_sDF-AX2oXU&nf5=Acb|b
z%ACKZ1^aG;#BMe08_)fsL0iaKt2X_pfxoHLm{2X%w_#^jC>YCkIh6W}XP#%Q3V2(&
zN1y0Z_$Gq#cRw%RwWsjgx%GZVI^t_`Bvu{(OWEryC1Jm*AmdDR%Wh~OZVEZo@3}tR
z1GB5SecIvoYJXqlg%;{xsReA*XQPxV#%Mj#Z{>|BZ%Q@1is`=a9?v!7rTr-i(Y}Lm
z#&3kfC?EbPOp(i_e$JRHu?CH=jZQ^@%0iZXly9=`el|wnWPD}-_4VjD|4iy%A@i~s
zr1W-w$kxZ{{h6elxdKt2N?RZ~E9veSiC3y{IR&Zx{OWxM&`Xb>@fc
z!Hs+!h2-Ba_N!n9bWnKKHb{9Tprw%m%5=T6XM>+!lm@RISi8F7w^d;BM<2&I#8-;3
zeM|R);KU+fP3VhKnv6`TuE#aSD4)z!9xkDKu4$8~3)OYqavOE3&q9|l<6H34?M!=7
z37!nMs(%xs{?B{-GWiAPy4BM6#Z;9A4uM>;KE@lM_cb)P)u(xm+_QTKZue3zYo|Qj
z!nu_C%XFii>k6=6R7h!x`tHGklXSj0I&&*1pT$Jp9i#cTN9V8RruE0>IBL^*4Xl-?
zejvWDnd>p+e6=PEX)rO=Y5E%0`BVDk>;LYFynmMjGF)Th*AJC^rCA9-Q-=oj!)qR2gx
zH-B)4hY7=8>=4sR_nBzp&gopR=J|<>bWiCSls3|NImL;_=z`~K-ZxQQpK6OK8~m?+
z|rO4^|5Q2(K)tjG3dQe5E$%Ac^7AD65T-e==O9=YsCGl
zJNuLm`rsjp=89jFOh&IyvoY1F0>9)QimR&s^zt!U-%^>|^sbmnOJt|>(vx{U5r2v}
z+PP9{8z7fO%(BsXoR54~Hbp;p*z0!86AYTQOECh2A2paNfv$|~xyA4kInc@@PW|Hi
z%nquzoU3w;bdL6Ng_ZO^scw>3MDw1Qj_cY7`xDoLTrAqJFNZZ95y$(F2SFv^1#O9q+a6B0p%-SYDi)H*HPbw
zc{W-{fdOybuJceo_`%>k2mixzL8mD%Z;XojU!i;LS*DE)
z^-IA5x62f_`*jq}v&Hj3*c8oIx9Nx_)qOBW#UaG|bvNos68e4K)t0%BIDb!N-X}ue
z!AP0ctTyOHZ@xK6^*7egQBC=#ou1Wm0BjQpeH=*oFOEv0U>CdAb~Sw8QFX>F7IAIE
zBFnu`L3@PZ_C6usp%J9!Ptk%;#7efpo4RIu{bCjoBb;p
zew9bpm(#oWsL1lC!C-`LTZ1#*@6q?}Q9d^*UfF*hdavTv1P8?X>bC3oN9q^pFJI^&
z&Z|t1UJX$FUGR@4dM}75+@kjlXL!&F%FE_U<4Yt#Upu85W9UDNAAfYAb9V4=*r`B$
zx7a%?o9^Eu8$MA!p40n#8TD_08+;~1$ah^1hG#tXfi*J0tp>JaogI;eUPz!w(-HGa
zJ+Z^yE4~J2e6G9&gpU~MX-+$_1w3#NmNUA3(O7D@9V_|HDX#F$DWf
ze&fYy#P5hV*|8gT*^9fQFPDrN^7Cqm9OW1@TrFaJK@9fIsci!zh%=jwVg`c$^OzRc
z1;)IJi&6k5ThqCggBtV08+5@<9?8FjfCc)}M$MRi_t2Yo+q5~RA?x1iF6yVI+8X`_
zy3Ju)^7a_VUny~LAMB#4lpi*TpK^H;eUH}lQJb+DRDH!PJBxW|1_;G@fXNa5R$npS
z`28Xy8sJ3_SB=!4V5#iw^Ycu@*tApm^3Zwe?>Mjs-TE4Sju22wu>Tj*R}3zZXF_-R
z=N+qmC&9qylb+V#HtGMXaY5hc{bki+=qa$obvJR(D(A`OH>7C8T#30K#n*|l|IC8@0}K9LWlYG%uRdD`+Ewj
z<`4K6yH|R+!ES#WT3LzlLYFjc3PeB8RkNLcRnXlj-*W2*Y-J;(gM1+Gbg+{$sBt{L
z!O?>tE$gUccIfQl{cExsk%)51e=F~cjbL;b%-z)fqlbN5?Al>1(3oe$fQ5dR!->`D&^1mB6+Di4FBsSrdqRKDYOgvM#I+sEUZ;Y%
zv}~I)f7-_fIrH{H{{-K*_mt;_;(KdCs9sSF@^gHJGR!
zGfTSRTRQI^1$6`TSmT)2Lz((ZhG_rZ6ChOwN+wvQIDlLw-3b$jGhlQS(YaxNS3JJT
zk>($3&3zDD9~b7V1iE)RPR1kdidDm$Bp7_^@1E`8fOEA>Hrfpsf3{KF(1Pu0SFkeX
zW)GcHN7td~D^w4>+tbA;&pr(TE5Mtz5;{(xPxYz>;tQqd!0XU64~}CUL4J)a-ev5B
zEpi~c|2x``vJbk<(muRvwV&#L#`1fCLZ+|>Mq^fQEEHsJv}QP1?C7ckG({k
zrf`MgQsmV@yvuSQzMr^RZP{b^qXR|HeL~#kzRA8qTL0!=4a#rY#In&bjBn}H6HMRr
zb-tatAE-aOHkZ!jPOb3@%8N%usPA=vRJBjGjI8?flsD?+@ee6kv~0w0bFbG%3Yy_0OelRxM5AF4}XpVf=nT8Ma4v)aBOP^+|H
zcnkd8bD!T)U93CMEV>VWRMFwslS1`On!T0kfZdr1%NHQeVNIz!X!koGu6-KJ-t1U+
z4s7XW#IFHcc|xq0fuo~WhSpO(%2`?7jDGh+`5TjI-&Gj4d!qeKO@%x4pNZ5G{RV&I
zdesIQu$SFo$dvNrWOJ4B*AU-g?GDDBXZd)8r3*L2?t}a@Pa-{kz(IFTE7~9LQuqwz
zK>aY|FK@t@igQ2dUUJ{)J$;tyv*qwHS*nNGE|n7C2m4AFTQFF2WG~HkRqqQPaWEpg
zN;(ugmN64f<7{1{&$kzJ=`Guk4PF0NT~iD&&s(Ip2z27@%`m2MzOC(~xJLs&Zb*Z_
zE-&0Ygm#^yBb@Sou+O!R6jEI|@~v2+22?n-PJz~W$>LAJ2&(gLj0W0A?a#+gQ{GME
zt3T5BjZ<-YWJcpu9V?;l=op;Rs|uPNV5nb)|A1%KbQRqXY`t3hDL;x!3kKmoW)sT1
z2gau!vZebks$4_Bm-cn2x+K*v=0sDL7oFFDD)(WGlOdXa9hCw55c`S|T2E2Pz=^+M
zcZ@Ey2;)5cbnrRF4TjD6u?i_)=I!|$U;;nKZo2mpmOM=R3M$R0{zl_vh4A-r(tH}6
zS-0rC#mGGxrh2D;#f9p~)y(U{R9D=TPK_U@JaBHWp*r>X%8FLnA6|b)&KiLW3^83g
zhf?u6DH>pZxxs`p-E*IQ)jIwVT*b`Qq5AGS8p|YoBfT;KsyA%!^=v7h2X&{vzJ|Za
zMoOgsd^|cQ6ie@my-CGtbg#KRj#30y@H#}^rusO)l`ov`3x$0fxZsx&Q|*xfrxueh
z=D>fcv1>Wybv?g?!4^7?75(Y7e+2Ju5??~=J<~3KMg0Pz%)iegjy*+fkeBLERGHB_
z@Ko@v2s$T5yKS>N(e6;AzfucSw@8+sz<1bvl)Ewr_7d&%gkKPM%lm~o?U$|};|rEl
zhny}tQ=PJvXkSqVE;dIvk$p!&f44s<_bfH%9dzz&6>y>R{l&F8jq0Yf&F!*Z&@TCd
zQAqE9m2T(AT`^$TJE5yDXgxiFxpZGdysBQ^4LdTiZ66PsS{Hs`w0J?Z~{wmicGIyti)6?E_49S*Xcho49Eb$AZQ
z*K##A0rmT}vr$Jc`f(&YIb02fx_mIW2x`ma9P$IReC9(bFO8X=e%!R);3%~eaM#W~
zZ|VM6P??@h1pj#-`+(M&6*Q>12l*YOib>~6@Z5T;htk}3T)tp`
z+f4U3?Mq&ZBl6=^hZ;I01fka}yR;;P@@m85q=~re8`?a#K+mVv$eSB>?$&9qX|z9U
zh!C3sU1oT?*ueIX=a%0Pmt+tT_6Bhm%wFn^!QUyqB#X}5QMv847XN+6@1vGzS85g8
z6ohzh*LpF1P~f(Vpd0EK$xAfhC
z_Q^E!XY@T4exEqCK=Vl2BlMHb)hS+oemVy^ay|53*;o+$l=2v{Tzk(V)$1=OXa1`f
zsWk%!U~d+Sd#MNg%naGTsIDY#EHZ0@-`me|qZz#$T6V3er~2`taEmLwJLk23%>M#~
zm2*vez@*wu0`%U!w0y%5y$_dt8OovkB6`7igwDX
zt_|!)bAL@1^ej>agJo&oCuu%^rSp3!eC-*!cMr)9l&Vs_E7&JQ`@_yVX;v9LJDeF{
z2J$!BoGAh8cAfb21UlW9&A2+i&#F=urRYCADp@H>-@($Y|1jPiC++-oWf1?;I+n{C
zx%&Inm)h2Pn~p3lsg&w1uN&-e12mFx2(s%G;@iY(52G@r*%
zz}DjfYCNo|62B-_9^-oiR$Iur_vwGOUPg(>s`Wk3g9SWR7&cFuMB?$@Z(&DCoT~I)
z_6J!%Wcb_Ko#gu=!DdxM5*Md^V*eRua3ScPp@spAkb~Ra_-te0UD4B^5Y9rWYx0X}
z87#6U@AU~QXJHy_^3uD7MQTd2vw&e?{I|^F&k|zPhMmoPs4m8d#sU5KjbeWsPVi-Q
zBE?v2wZ%%=o<_}bsgZRBGi_?
zbP^%J^RB+sJ`rBkjBg2z5usUL=4Jas5sp~RdGl~0g9-gcjI=j{xQU`eixnvRtcpv$
zWKALX_OxRCgA{yH11^-Gq_BTf-gdYkmaID}G#4gQu+h}Mm3W!L^Q~LbEpsTmb}U?T
zyM#iNoBEZkw-g@#bed`^OGEx@6d!EF81uZ6v#FYz7GoR8zDAgWm5Rk}*z7G4&31NRGcj3jcm=a=UCnq0;N}O>ZR%(pxF5$&wU~@Mh7E{uSclj70l9^1Z9`k8_7)D7E;jQI)
zncbHtINey5I~YdxH#dJ&Ru82xwB$h8LQoRr*#Abaq^V
zxPgQA`8DMir}LN`kou`fo5$+R^0C!~XPHJdr{|Y)cof&O%*TL-UzyxFr$!DzjVTfi
z1suj+rsm7narkjo%V+&m9;F^{rgS*+xGhSVSs%fJ{+53?@DCMDGe
zd1S3F3w&+H!{b8utpzNH-H|N{)^Qv*%+IncThBovxjCfLgoFFkd}H-+4!-ZSr7@Go
zeChd4|HXe~%}-xq_j9OvrdvNLT8!4^XX?F|wy%hpx)|4yOighiS&!RMj=B>}5eb8tI$bFaU?7#ba8h2wH8)>W(!^mK_Z
zsI~3s(dRa9wXoUKfR#
z<%xfaGb(5ZKMXC*`;CLfZ#qwIX^Zh9(X)pUF(|y;-|CVpLUeArUE3)dRCs6K-Yg2g
zk9HYU3P?T9e*HU==*e}vmZ;9z6aw?ltk`pdf`M%Pn%Yzve{|_?UtA%=F}cW$y`)c3
z&ZIuX^~Ca?{kpU
zC2UXY;OJj8UOZ9XXt|9+$EddFIzusv!fUIKlyT7CyZ6Ck!lMqA+1`^nlc#%U*e@pLoH~0sPjKe$2UkqVp!5Gm768QONggHM?Of!e)QB*_12CL8Z9bbJ&JQUhe6tC4{fK$~OvJi7phr^*T_|#p0^}ALpIxMeuty@9WBm
zG(s|;*EUa}@Zs`_J@=*(AM3Tf(V#}-T>H-Eo!>+Vy8eCEqbXt>@Ui)&sg{F+Me&P&
z;(5#|NWVIb@PD3Sh-si32d#hVrf_#J7OghUDuob1O)=I|iul4}vw79?2_8f2H|p?&
zKap1o#{*u_s5R({`tXr~*|Cq=zcONIKHD5puz^QKtiqGzV?4~3r=_nUbA+UX`00`;
zF-kVes(toiP}*r?l1KVl;r35a*8<|VQ9mj$C`8ubSI;UL^8Y)lG?jlJi{N`i#qA`)
ziE+78Q#7f!;R!XhsR=w>pRTwoM|4fOIY}jv?C(@%R&V0Vp`_j^B9hD%E42?R`i8Rj
zFtbKCj^w-B>X;enEW)}|x1W@F(^&J}-tlHO;lt=%kw3}b7(7v8F7a_cf%L^`#24Qx
zyjii&j`*TWcw;chx7mMjexluB9{0Lcqa?>T7trr4f8(qZKO%&Nw@vTl`yBE<
z9`4!g&f~62kVpGD9%BVjtN-V~W1541)BCGLR|Ru?&!men60?8fyNf4_-JidBD9>ad
zJRohB-A!ZXhG)lTle#InZ-4!T=#a{=d3fD`5Y9cvRu@w&QZj8CmS}SDkXbu)h4_(0
zcAiYf8R93+DrQEZWbSKiP}A?>@a)T-Rrfo^&|l`#_wGE49a#>wMN1j%T)*zEGV$?~
zVbQhaXtWl?IxTsQ!JmH|D*vZdzQdbGghJF$Cvq;s
z<0XY@qa1cP_jsCr7DK`)L$Z$4O{-J)_`H`Q%<4O1?@DwlPUP$GgW&GihZnzCrc)^J
ziq<~;jz(|Vm)Ny_WL{|Q@im+!Mrd}9<%N?R9$YM)_|1#QCU>KX+(;e^+SL4|x;b1s
z{Y}Z^su+LDE4>#9CbHP0_E<6T4-r(dpO1MH-fN5o$U!88-uJ$mnqUfJ7g~*E3TccD
zb>;axk$F{f-YGH3bIo&TU_%NA-F25;mk08wzGbFYOK?|K?kR1Z!(sf=-@UCmV&vZ*
z*gv|9ft|{wYl%ePKi!fLYnW5$dp~09Lh>AloELvPV>gBV_|o`&*J#9A?Q0T~^B6}I
z$k&p(sUN=h%!bSt^)7QIJfaBRH1jGy+Ve|Mp)@I5I
zY-4F$74)S)BfLm%YYOIw9^AR5WMmge;qVRdY4ldnTPvr5W{MkmmpB(;bo?)gycp6pOGWWlGzzeol^6Df*y-}`%5Op{LdqJEFV)xlG3Epa@me$99{JcGyo0a2p>
IPm{b#9b|{LApigX

delta 12280
zcmZ{qWmnaKx2;8_OS(b2LqIyDySt@Jy7MO`ElNsBcXw}6K)Sn|4eU+F-o*Kjd&arr
zo)>q$S@RRD@ys>nLMp;aDneZ`;#&dkWUP1W$#aN82!j7nR&RNE8$I4F69JK!Z~o8t
zWfSWAi|VftZN)McFcS;V*Gdt~&VFzzC7_vQCUq&HXVo?v)sV44jU#uJNND+>31^ubM
z+u-xbJohef>FzEQN_`-y0*3B(TVYQU9Cg}Bh*02rjV4BR-QSCr>wBrgsyPqW(zGwB6YIgn{qY(#9OsTEd5y#REU^R+($PGNXj-*-;ajZ!A6}E$VvF{phRw`M&evD>4ieWfkXs
z9-KvP4n&2rekQxeA;>m!
z#Q~Ptq$&EK38U~O`C9T9*kNzpuBxWZ0|GMs>NBYt1i-BFUdC(W#f&{eQp1q_7@fuG
z$;Teu*JY5g!~nu1ds;sOC~SWMKvXrD2HE?IZ6S1G+tJF&HklgHzw~@qxwG#F8h|uJ=shiJUGaaUvWpfCzxZQ^eUDJ7;80eNzy8#+}mn%bbgFC0M
z7*rw$+?yazw!Z99-Tw?Nnu#hl4K|o!oZLgQ;@;Xskm7`^0d(=Q@
zt1DFM9ZGXv+tO>w?(>ByvXi;nLvzJA7w~fcNcp>naVUQDa7{G7Hb-QDPfZnu`*#WD
zbhS*Aq;ZZr&}J;n7wJS)=ds`EvKOyQ0znT92}^8vVuyw)_(q*C;gNcb`FlTP`yA5U
zUSXr@rTwUcugqR23#3)V^x53kwD~TRwSy>eXwBV+tCWe*=A*%x^mg0`@#RVYKIkKn
zZ-l}@9M8qzbqpxZ=);8Km7*CS*Si|UHEyc!kASnYl!@y&k|K692wq&Owi5(mZ@+x80?ameNFPgTdkSi@Bnzng
zOM}v;2f1FHPFwJWr*!QLI|wHILC3PhgHxBX=$BF*6xd!ADyJuXuT1eU5;IE;4~2U1
z{>VNbW6xVtA57agl*6k7;(MVIQ~CsAkY{IFH^@P4%CKLGA-rnTDf%l06+
z$F7j7BqM$^=IDKk3EMPW9XqwPA);Etb@Bo=>gWX%;^^o?E+tkL4C^<*etdEPf1AGK
zK^6}kE^*}8|Kg#d8$bu%coe}-lR}H2DzADESgP8uOUIp&0`JLwoz(GTo^C0vKjq8C
z4*>Gm{wN#PKaCe$flj2`U8&5yu&?uM2VbZoef2LzZ9OqM0p92Y+zZ#+y)R{2dwGA72s~J)
zuEfb7{(QTnaWhqx-HXK*|G&9q_;=+EyipoGaU0@hn+X9Tm9nc!5WetS^4mk9#E{@)+yrAt7(BjK;Rdy;=Xg)_@?;%>SUX<-%
zBpHoEjf%YLGgUd|nxk-hSMcY>;uVs@T2^Fa`+4{1#a{$l9KAa-g*!E2$z{V!e^ol&
zYj)5xNCh-qds0l+DZRf9?EZy>%ld!?h2E|o}As=^r
zM|P|#aluOAdw(m)Bb020D^4snh%uB4T`U1nTI3p^JeQSf6@oe)Nijb|Zq2jN{4;sw
zT1^^6VC--nta@zT$Jo4ug-7zVW5{M*|6N)Oq^!mScKjjRWq@!8SCv|8jZl(eTS%7;
z{@XFPim@=K0qm--K}8#wyiBb?5vbtXs*zqW(cnRu<@mJua=Joj9R`*QX8+Aq@Cq+7
zneQyEa;@hDbH%F0wW#4)1-%WYTv?{mwnfQKAK;7_3AMwv=jKXFDh19`cumpfDxm9;l3lU$6V8_Pc7coS0mibOmtl2?y!|Dadf*oPyWv0Pj9i?4O%BJ
z@!<#_w@9A}K2S%F9-B?7%rfT|qyXI6cxcG3r7mf@*$jY&%&rOMJFcrefnR?5!%4dU0c)Yb2vDI-hSi{*;Pc-KO&8}qR+_Xh9L
z!&{2@Rv5XesBInto3AJZoYAIhRSE0Ji({$1+eREb_?f-gheij&H^nFQ3V<7eUe*at
zk1D7qo8-0(Asr@<2b4Dx)A%mHt~lLWSV_w(Y|ygdNC_*7E|ji6@gt7tvk=u%+SG@&
z3y%}p!GZgo*t}*n3L+OFZWGbeX*I%sd;fmZEwE;k^T5i$B}vD6H#EUxd9hrArCEA#
zOoqmvuM+0U@Ic%Ok+D%Y@S7)p_+V(Q#x`GkFuFwt+9EwDbGAdMU!>C{_G5f_^Oi%L)3-_^FKIBXw)HZ+ZWw*|^R?Wp1
zCyJEX2q)t=ZKBR*&V(BMJbTgWq591wu-bYaKOiLgM~V7O=H)uJ2r9gW%8~kNHMNIO
zMO{B)!+&+$xd+7(ndJ4$CB1eno0;zaM&wXRt44OW;&(&v~?Gm|p(ln=7WPqXdp;bc#V~
zbzjj$d2@ak_DS^KsK;Zi&~N+%P-T;;Nu0lEvB)`8F=91Pf_i#bbivI1+fiSQhXlkCgb)vYBI08lKX}hPf9}JXBDoZ{wp92Lr#67@;A3?A
z0{B_gZJy%Wy%i&*F{Z1w4&CY^@;10)o|FZh6$fJ;=Ds2l)Huu{{I^MY^9}32
z#sBO(^#5@$l^ilj=m8LUba()&8|3BsycDgFuts0_0Q1p%7*ml3qm;1uPg
zpK(g^;e0#4lN4M5X6YE6;feRd({N*``PLNK-Rjh4;Cb^>WA#wO{ylwe^TwHl-Iidk
z4T0mTHO@+(!rBq!owbiy8KY*?Yz_94#X*VCyTJ4!6RRtN+=G;!6!W)>uq^trr^80e
zK`9QoUfkC=X4%tk*k&DCI>`+J1Lk&UeF
zjky=HE_dJmv*LOfj@I|(`P^yRbCn>OCDn{CmnRwDg!}K0ekpX%;?K>-hiCs|bGbMEj@btSO5~&+!YWEm1vsGuA3G;I7
zYeHPuhp`ss__7Z#U3-X(V>{`$<{T?}+#o*QzvS**PZQAH&o9m2m^z4FG_m4EYCoFQ
zbhsAb@o3vPUm7iw$Bvr^ed{t0K`bS|LyqZgb&)2r-52ovylHKkd5zUTIeg2TlSL%Y
zB!(I!3anb!B6ZKzbkzv&E6bF)%waz8PT<6Gd=F`JyResF_RLudB=yqN&3VMS+YgzV
z%Ku4iD6w4m4I0!isLd2MId@?*=+CXNmcR5So2BLa@XeD=pnOnjFU3Sk(n6lq%U1+y
zf=_W~ZPf0JH3GKep)F*L&ck_7VwQ?3ge`t=5+H%@W$FWABIlrIa-6@fZyBjG!TF0h_ZP7VA{FDl
zE~~Jo7OSm4%8#vfHF+^tK4s2lAsl{Dzcx(1x&Iv~W(~QZ<^*`rM}Z_ed+W!%z4y_XsBCeBu9#`oHmjxZ6>5ma
z^_!fFSwD1ry_?5{;K+9Gl2Mrab|1c}lgtw@5bzRa#UJ|=%$YeD-lM0cl*e-O7tnrg
z4dSMz`{PvmrbLY+mOl6>x3joCFYmGR+HwDA^MlK=Ekf?CYJi^N$HSf9*(Nnn5BL~Y
zs%hcj#=Ghzv^DeNpgj6<{A$B7^x+4Pn1gX=9+#;2%^mTsdaL6#F(Tq18?Tgw`o3qE
ztmLn&XfM@q-VLk%tfy)dF=SsUE}%!uW=-_WQ?hIES#e~GXkBX=J+Q@Q$
zl5t2CGt+_No0t!=@t*mym=RRbHFTZ6oDr%B2YQDe_-;zInNo&!UZqC`6tGUq^d)?4
zSk53NiaFnxTS;5vU=$xpIahBAF<_{w?w&G0IG*XwIx=|
z&pW4uDLs97N&2kYM>gJY)+k7$f|kP@Km7;Ua@MmGBXEtE>#(Y7s({qwO|6kEv^S;Z
zw<|_YuKCHMT)e&_+xr4x^5JdlnE(=3K7H*?(N9;LCHK3%ZfWiB%JjXK1yT0Cnvcn8
zh;w6T6WeNvH!wn%L#x=rfkMr^@A0>GqXOrqFL#wJwR<&4nx=_Au)5@KPv=vH7A!7~
z+Q4%Yd@pB{pMeW2lN`T~H=TdKW}UsyZ7uK40{g6umZdkj&WQOJ+)&z%abOFFmLye2
z?!TIy*!2mv=~gzhCxW}?wF*@N^o2Kt(SDhy{EDD6$REq8M;&**2fQdmF@dFMDRRr#ONnoVmsif?6B
z#kC7OlLV6#LP{G&M;%^l;=GHn!y1=CPK%N~1LOS`gx05_Cc{j5<@ZTRL~dWlyo4&a
z1{QumuFK7sN4e$lfm&e=x3J09X=@MhgT{`hHE!6;{mf&BWgel(++Pxye>ZZb4$dKb
zAx6AZ*w(;Ghl&P+BB?MaDaiTK|)A@iHyxx8BMji+`p_5eTN
zi1FAw?rF5db7pDNMiClru%U0>psvSltbMZK6X2+Piuq#}u$l=F?LKmM5NowX%3aKR
zYPAl~dgkGTzTUI9?Z?BlDz=`qjD;vGa9p~bbxJXpt35kTxfD}tF#9U8xTEoftQg2O2<{8x^w>RL##tpZ>CbZEAs;>Hz~BGj-VDLCo=a#
z7X1a`u+Vj2ueMEZTdk6lbQJnNAwDF4Ym$uG-C85qm$;%w9iFDm4Z;XnlNfaokq(}m
z5%x4v&waVTHC%p;CRinldY*4Nu4_Zxj9$6TPJo;KMd#cuG#6po5K?$vB(*yE>zwjR
z=TPDWIm2>K@KWdZI>%+a3n4Od#0Gse94!c;VcQ3OiX3djB%BudN3*pgSUE?volV9k
zgUZ60I8NInfJ?)3khjzEAMm)|hP3;wo-Y0O&=n7`KzS`mTQB|1Q3{T%lY30Xz3qz3t3?7-Rj8>A}!U&ci;hCQA>GX7-3+H~+(
zR})Z|IXUBOVZn0C)J2x+aI+Cszl<%hoZC+Qp75Ko#PdHd$?Tjl$oEYI^^aMLrFnlk
zjvGvjHCwH?Tm@C#@9l?83;gs6dhY4N)n%|&v~<#E@oeqFgo|(G97$fG+v1x98_Mli
zQ|!G%pu9g@220hDWeLx|Z2>AEPWp8iCk+@hd_dvA6qEJ$57rR6pxtfJF>BN>eJsKd
z#q0CW!P30RsqK18D|aGZQ-;=3BJ8UW5%~Tx!lTfPv2c^v)G_<(uz#jPSIQ%HNreDv
zX1;#cCG6csk9pT(REq0R_D#*34z8=6De2?xhy8wyB5_r}6Mkv%08wIAV>wr%pc#Ot
z_-u6a)10o%%K%`!l_ko$PHOI}3^V9c8{~3%ppo;R0*bapKl|4ngQsPKjT=nTC>I{@
z2v}f~--!}2RJhL)#~8doB{->BwKK3G$8H=8YPq<%qMC5*nrurOjf|X2UZfDN7o)&j
z*e>FgCo6WSyguKx39g-|mkZ~CCy;aZ_xULAp2a2BSFVeTN~ChN4^J()8f|yp{Q8Xl
zZH6n(b6KdFff-vCj66i<_~Yoc98_y2#H|^^8ZxuAA-|djR?^vPo8WQpLL>Bz=-5Q&;xuMi-wT_4O6K5I_q9;F>h8P7sI65e
zsJT}fNju6z-Ko8u(g(QsQL
z-X!)W44w%53c!zIvG!Ta4*Bz@%|8ax3u>O!FL!+}8aEF&?{+zoCn#F`IfH{ekFj34
zmB0RhBq4f68D1$b<|c*a@Pq2@etMPe3DRr>L5zbf--fps|CWaray3Q}g#|P|lOgBj
ztFKa9Y&%=Y&dH5;!h;)v4>6-Y8Pzodu3`vXNanfOC1J+#G3h&~qHup-Lc
zsEyn~;tw@XjZ|}}CjYrW?iAOPSV;<32wd_gt3d-n+Ffawj^SwHe^yB)gsSEaSOwH3
zJq+x4>0;;e<_L~wuQ56j9c7QchpJpcwiL{TZfeL-I<9*JIVllwx=Ge-Fjkpgx{Xj_
zu-r)=2=G}woXsf@VRV_I;}jE^;1*fHg=Icp;(*^n=gA){-gk8I*C!z7Np;wl^WXSB
zco%c`;+D9(u#YG=nJq;6bxxu=V0(drqq}z_{1Zz&of7+B8%<}2$$d_vE-F~#eSg26
ztiHeTG#1@3!ZP)2E~!keOI1<5=X`_2C}pZ21{n4IQyrGP?oafW$oDjG{G!Wl?{etc
z_6D;WqPrR8<19X5%QW7(!WL+Pwr3qgitUHdGct*sfJO6ZPiQI2ZbicrCFL!nJ~T
z4zK_YpC-S_eUEcA%QLsGNtqBEd>#HIy710tF3l}YyskASp9)F*XM3AD@%jfI`+ix8
zv6qxg_G6>sh&U9h%{fMFUmr;TKD%)XiFT6sUD<^0zS8bP7+Ti~A!J`ic&Nc)B!+dgyO2EqCdsTkaUb*OLB>Z|Y1azz@<~I6G2cG-CeidIG05d5&rAQ*XfD4)1|&o{$>oo9vpp}cs+B9WwBNux%vvT^5x3M=3u<<8
zpFYyeGQG%V%Pkq`vwTg{Eag9Uojrcaw}^+c
zpFFwP(%BDG^zLQwnTLI12HMQ!cc`1-x$(0i$hM^I^SN)b)-wK$_3G1Fv|RVqN*}>P
z281FN<5GUv285T#a*uCI%KDZ_mwD
zJ(ADWM&puAB+QMdp$#e@+uXp4@i>A_+SA{!tZ0J~lnK{~+FwosfeWdFm|QQe$C}F=
zL2O6kdupA0&tIoz3@CR(y6tP)rVFI&lGmoQk47>?qRGb9B-m8l{DiW5zqxqL>JZnL
z4D)|sZ0xzN4P|U2ZZY_d_MTr0*H)JXJ%hSZcd2ew%=F&n5U07EOLyLdN=klOuAc}m
z*3s_=ZT+e&hIXne0y&HRec%p1KIW4NYi%m;ZW19mIOMm0(Af!AcmI4)aw3VU}S2RWdp+{`{Z5Kpq?1u|RZifsUw_f3DFB=_7n
z{430@3e`P*v_UAk(6eE|rDD&$nm+6uV?Kq`a;#G!FLEE?`F*V-`jO(uUoI(<_-wT8
zw1VJYHl8yk!o&rht+Ul
zM?7}!o$QfwTsy++uN5qeg3s#SI9N0fsWv2(quUR)O>_Pi!{RQ+3h)YAR=Wo5mv5sG
z_EGA~-R6d;?e1k;B(SgGLmA(Z~@lGznETP1+!KE@-jcql@DS^RT
zvP3s?l^GO%jj&Temon91coZVNPPnyXl)^=sOegA8aa2>fQKlyJ?-N(Wn^Lku(|?nC
z0-I#0tP)craMYY}EK7(N2t|~fx7j;3S$X6kKxahB&7HF!cp;Z{MAvPy``?tEe-?}VWv7%%yV8edlX>b+$;xWrf&VD+_PE)
z#0?Y`wQwih>pP@x#UYJ5OeyVP)xx
zvGEhL9}5-d-^J~|3ilxStnx>S6yZ|~UyCeAvsj?e02hxsaMXhX(MpL0x!w?;xNJc1
zrbG7~6d6q4qj}?A3TRFL=bAA#WSUQ
zsY(Ie79Hp6Zo&@t(?r9Qkk{sq3YUt;Fb;`Fs}ZvEO7qbwv9O2FQ^=%N8kgT2+Az9B
zRzw2Jeo*tCHWqf2mcGO-VE~G;WXS-m-2jDYwtbhY<<7S#;HSf5HmDT5GlNQ%ldNMG
zlY8S5`ShQ7R>XG@FKGJ6}OGmS;Rkpt)(REnhF_
z@Nien_{;Wi|E{g7{&TkNT$&A*4%95Y=DNstx2ooVL&L`gltKO5tX1#p*((4e3%(I2
zcwSAHf20j`eEyJ=MSSLs&0%1NHhi+^QrGyTp@27}X#)%7D)Dzryk5|H4){+1EPK52
zWM!xZou*$RljJ;DNc`?=vwa?GC>bzCC?K;m>-)?{>o^
zlkp+F`Jq9)?LP&URHQ}xJjz1pSF~5e@zZUd|Bc7MoB#V$F8P0&Fb;xgh9oXDbdf&`
zr05h`(Me?mv}iTsRf>NqX(=`3u~@S-q%zLv_{7-%(kh@O{HImk6n_wSub-ktH=bcW
zbsG$ApY+T*T3<8lV(02|Y>0jsNs8|@q6IB$ubt&qF_Mx>ryflUbyAc
zh%dX$aT-8TMB6`7TE4bX@570b5A=*j0#;&9OMd?p7C0BX_x>$w
zw5^&WN^X
zbY_U37Lj0@E;K35S$?SW+;hI7MRv(HDo)}J|1w1`H!DIiuZL5M^TS4?_5-}>96LO?
zMcWt>Z~d{rmxgU`g4Tz=(Y1v(GFO*LYpN!%?jHPrSJ+MPq5y1oR)~g$rG-7;KMA5g
zI}v1|YcaPS9}_bDDZJMZ`O=H9UD^J5wU3#4fjSpXXiZK@NIPuv2^b3OmByc%6X?`lU<11r1|J(#G
z)imDmB1M%xO)7vBLN&{!@TKI+%Wh5`%>_p^<&i(U{~}AS)pL>HPhd{xuGp`B8DAye
zc!EfS8QMcg51xsZn%6kP>4s?t{dscnMV0gI#1fdm2m@af=UCIt?S^&XQ_%(e-cd=(
z3c>y&*%225{Ku&L(
z&&gzan*bE^PcRU8KB7b`>){>^5Mk}*!4w7ae%8nQxpX86++>41k_`B>Kknw{cNnUUY~2E2rF`GPB;~!
zSA>^DN2gsKH^l`&aL$~{h7_BZwr
z>B&2#qL+!x*wif;4DRqe(jOrse8ei17dpVus8*iPE+fpNsO3E}|C6}j{M;gK*3H7&m5Hx2bG}wp9mFh+
z#!LQyn?nqu+@j*x1>^7jBYyV26dEnjGpmHWeT>zJi$7u3($t~yp+`aRimsqC0fbjR>rN?1#
z7q>z#!oB>VV#Yds^N^04HNayq8^3qBp<+&
zr|#8S>$v4Bw`+tf@exej@I)-T@7Kbf4g#n0$qLfS=(DsDTAV9S_
z{al>vb#p(ffeU|KNMvUg)K-QY1#aI7=Bo8F41%WNwLr&tn9+TIQda6YIm_MfM
z1G_`*n|ezWVN=;&QLXY0O%cIomlH+b`mpzu2+^H~C7$sWZ|Jx3g}{5(=x;ENk4JbwJH
znlHr11uqImYg#DXRd}7Nk81@+vO?Dvd)A586R4%(Cpx#$pDlNQlz9Bj6f?GFm}yX;
zaU_JDP-6%^8a?Mqe$R@k=kBJ8&zD|wq}j+s`|oF$8*{Ctk>I<1vOm*XDt(Rt-MiQY
zu6=~8^_M&{@;8Cus;-*bfM=#ER!j{F_|aLgJMAy2^00wZ=0M|TP1dySA2aUl7YjeP
zH{3E+bcs;x#!lCskdP%BJT_IArRb!$%7y%~W*kT;ry=wKogarnmrC>4mPD!~a*lue
zOEJ8)Bk8KS@@w!%x@7Vh5K%`7GwGTu(BX7
z*%R0bAIh`2ZE9C$Hp!;r8mEDe$-rGVln<|t?iN9P*D|>fr`m=Ug$w?&qxSiT;ljGi
zD3q~x4=~s!NZxP0NW9(CEar{FJ^z3ck0`s-S>>3uVbkDP3bOH>pW$i#^QAtza=c}7
zK+I%TkPfH6p#seYP`gktk`kI~JWA<_SfIR#Sn0Wsq{8Zd{J5buIG24~Arw(N>L~l@
z6vMd>x>z-x>Dvi=rDie(lbrdLJf5`v;tS)UTH;IiAnpzsxy$k(81Fcb1AnrQXmxg2
zy^8Eh8+%TM{5Y`aUVx+zks+8@EBy&dAkI(D4RIIU>B*i25U$?{AQ<705AgfB`&k_o
zn7u)sOWpli6vH)MKOIyidqHyCrese4B=qFlckzdb&oM!yz^uo@0ytJ(tYuG>Oo9)Thc=o+ZZ*jpNmzFYL{WwIF(
z@S(VJzZ3!9(<7R5(TRbTBiW?tLGB9mHhtJA=m|A|1(sw{?iyLOOw=>&l}M?KA{H7q
zgd}*&7W77O@uYVsL)2*ASjzF$W~cjfuw-H1>;*#ETG5g+bU4hKtYjRWXjZOrE>K)o
zXqUKUlZ1h3WOB2ILa%Y%a!Tc?{COwq*;JfdX1^Udpot;ocr?-IsWdrSQ|%?8mJ_{9
zKLZT-vgbrut{;Zzt7C^!uZqp<7&bxE>PNsu^#KQk1Iho^kr!P9Hr2fB^R{@Z?i%0q
z_);2~NVzkQ)K%sd(tSy+T4?Qwp`Mr=O%3){8LVgYj*IN0ZE0O}BJoD^+X``>sZJDK
z$218v`?J5Dd!nN)m>vmlDCU}vPNc?(sn-X-eh|K8A3vj-Irq?~nYsojlST#B9EL1L
zDs3B7I1?Y2vh>se*6pW>se9pQXf$px9Wtu=SEk_=2szO-!EsyQ5HZ!KlL94YJd}Uk
z^+;Fv2FE88jpT1~tn{{{H@W*k+g~GlQik<

z8>U#o+laM!$_{zI9@!Og6{jj3X<19s{U3uMKfAhD<&)GkJ%3=w6EnP7^)*Wy_igNC
z%8#dAS+{Nr3t6=y)aq8Um)V@RoU#CraqRvp>dJMqQAf^WxyW6e?I^Vr>8a&ref1H6)-jF)
z(+L$4J0@nMXN_?z0ay2}6id>Me?J2}T8
z7WhvSVso;dJ-h_eHOd50p+r?bkQ7!%*31iU;+X=-4u}$rCy2Q=vtUfanWwN+mG3QObXlE
zpG#|?`Ja#_=qoH&0@(>TgUK8g=azSbtg0dWW1|Wple`JN*7c{
zldsO7^pwH(tUHID)X0YCmGd19s^1P8jiVQ3#mMPwjOt?AO>u+=b_$OCqi=HBBKSGu
dd*ghKhU**BeR0*<5dHe5wC6=MxlM`kzW{Wy$O8ZX

diff --git a/previews/PR1624/tutorials/out/solution_000090.h5 b/previews/PR1624/tutorials/out/solution_000090.h5
index 49f84a2abf1f62596e76efe76f0aab8b99aabfb8..268bac75d1d68bfc86eb55f5cb186d9a4a4390a7 100644
GIT binary patch
literal 11976
zcmeHtc|4Wf*Y{@O)7ItY6ywB*
z4?8=?jPd;Xy6}fS|EX90%~_xq)(_5qVVK8L^LUl<_ZJ3C1S8f&=J`DH_E$3fX1LDF
znVIV75S#`33plU#Z}|T_0(v@HCPWDK&BuX#9_#1pXP*33YsB|o!VAp*1}@ZVI_qyE
z3w9S_D!&*^7%R32W5*c09Nj$qFdp#hw`=$xuKX_^F2eA6Jx&k=J*UF~cK&YOUVa!q
z4~8v+#f7W1-P_a0*U8V%&ECUFC&0tWS7$+Voi^;bus{3n{e8SWf*qY){?>zqV1e;J
z_1Nb9K5)d&_wOBN{^jZKKY!|RE$BIUIC(mG`TxD{{PhOejlT>qVe{wpbnN~r6{MKjti#}ml*>(M|?f$Ya3noCs
z+27663FBRf{z#7+7p6uBHy5_wU)jSZC_9T5^cUjG^jqK2|3B64;F-F>@+1NM87Ge4
zaU%Bh^LO(3+r(J@-t5;Q*nWF5?+(#7i4OzAZ^q;=3d0tVX}+)i-H8ZM7UIl+jEs!G
z46x7JV`7*O0G3X~?vD@OJev^ycK)N&B250*8q+UJM19*q{D=}IGW**n4D)sM{=MuyI8vXx=Ra&TMYT_Xl}sp?ASbU8Niq0g8Al+7`|wh^h^uTSFb9S
z+Jxb26dmO4K;Ey~4cBhP@G!3ZY%c+WHQwli-oo&-qdf;jN-#Xo;)F)ST?{{_7~tT3
zAH#Rw=|YK*F}#C`y2`l%!&Q7&oRqD^aDf%4WsRyaoJs1?sunXo%tTGR$k1hJrRIIoZz9kOlw7=@g8PqJwgVc^T*kyNK|9
zF*@>98L4vLMn`W#jL)BOrlZp}&4C$*Xh^BQ)=bTVhGIufDGT}1P~rtl%G{BLCU>W#
zya0^N+E6m%OhdGMt$JCITeVVxi|j;0r6og1&xeMbRm0wPex#xPsx(ReW*XAWuTgNj
z5B|wk(d#R~e*5kZt~_+)F|B;x!-I|*3TC@L$J3FUx#;y~S2}7fn?3BSNk^MWx7IoF
zVYrVg>vPE;bd)intg`n79TiGk+TQ(*j`a4vEEm(j@Nj0GD6wD+-*mS8$-&bY-Vnx&
znFV1uUx9?TnH7dxGxs_@FT!xv`mY5?av*MeTENHkJq%FE`W90<)RXWPcfXd=1Q5s
zGYl`M&faXhgjo>}?#R9KMl<;e~OLO1V%c)63#S&NR`HPmRs0oKYIu5Gz-)
z_ACt@2zwcv2>G&7JTYYKLPNC`8U^pbFZ!VJTs+iq+sw3MU<`(zsma-8DTm=Aw$Vp<
zUFhgwbe_)7Y#I`#f8V(Y>cleQxW-I04IQukxTZdthI|W*o8ur}?*@lAxxsmchn`&D
zaUAYJj~GMMVhraUy?5J6kdEYBrgM`0!5_VIS8p7pAqmF{mJgq3=wWWe=(eSF#B%*}
zPcX!>aaEJJJmlki!0DHkIi`Z!P>1nu-Gb0p_?WMn
zc|qPA!+Xox;xW8@=Xg42E*%MErHHnA(NKB1hj=j`9oeS(i_Pj|xIK-hUZfPm>&NGc
zx}c7($fSt$f*-rTJ(>-J^L_Qo3V7oS^?Asl{PICMTJIv`B?x&{D6;4mH=rY)uc3V%
z&aj@i;&c?$iHpIdn-#Es$dbhHSU|CwJz>h=7yGQ;+QauiE}7H5#1F&WqvS2kSfSqx
za=GtUg8SO5xViKa4YB2JnxnzJXP=s8ya09NwR2+WP3UimvbhCBz8;$}kA#DtWm@4}
z*afIZcdiF7)iFFjw9A(qN=Hw27SR^9(NNh2OXFJ*_d4z?E^bgy;dA13HZE|k7pmMJ
zqcMC5*O*Wn+{7ILKIq{Z``~!Tp
zI6=R;e`!u782Z6F*;O_@i5SkP(SG9X1`K!IljyXU4a1jin+nl}yi9r>8xLcoBYBh3
z%6>Ta)1tNf-5E4=qi|DY#v3}~z>QDwLcCu7Eb)tk^Xf?PQsuHRyz2hS>xPhDbvEm^
zBIpkj#x-r>n=$;vE&teu;P0%l%r4vGG<2!4)HV?ObL}~)BqodDvSRN(8r*?*?(F6a
z2k1w=W3~Mcp^i*hw&(5vn!{WqBH=2Af7uXJ{{Y^ZP`~W;ALulcdRje@&y$8;`8r1k
zt?UZYinH{@TcCEv^u
z-kHj>a3y#TQhbCqO+!Cle)Eb=67=J-!tS{y=%+V!eCNo6zVp{N`hW8K!rXL`n8e|O
z|N8sqAJ&CAl!t)}eki!v!H*HY1M3Foe-6Q<@~@h!(#3y!^^eR|>;9+aD#rPE{onp2
z+`RRr;6NvFP_AVx-h5uz;~11=KkToe;=>4r	q`TKRXg!!Kh|FAdz
zHuvuc{5t~wj==vr5m2@AXqDAIea{>G_MRFQG5G6!gZgoi~>Jx(LP;QrZJSyD
zVh_eKl-u#{+eeV^W>I{q2e|%pwbK;HPv*UD6as8v&ulHAjiQe4N*h>V{iM~*#yf!8
z!ndy{fc>6?(LIX+8L8b-=Rn@`4%SHnWT^_(S^{!Owu|s%fIZ`ZbQi$lk(e5LkS{hl
zQ-lFVd-pe&0E&wWibn$S@#R~Nf&T7Ko1Pv5d6VHCh38=ZGvBDf>_AV~I?a{;7)46g
zvHRX2m(FFcmjry)*W;gkL7zD@_0Ce@FL|=!O9trO+#}1-0kmbWmUR)(HhB^@S|G3Y
z!)Ed8pK#8cC#N0*owM2K=N#a^*~%(LuygVG`>b`KM>Qt3s{nqpEsN~|jMiC={J_qI
zMbpmPK|V$ce`NtkdCwtX4rml9I@Tbf`RHS4T5<~X*h{MJe}cTqyEOFg1A4A9Si=xdGA&l>6v*qb
zypK%;?BrXN{T)zRt>MKzuqPW^<*NyHGV5K}2!ec7{P4Pcz?X~s=voJIg-a{D`GJ2s
zyyk!^U|Uyp_fc5CX|%KU4ea-RhR5O=(2}YdA1DxqK1F7l4cH@7qWeFAetGg4tJ@I&
zaRJr!KS2MvpOD&jh@(uRmka~Q86_M^+yk`iw)-t)SQocUJhcZf7JS^CJfkrGg>$e>c0g0wjEhF`hb5c;MnL+kk1O^7ZCznO-nl|27KPUc-9wypCUu*RsetI>ZMPg!Otw`
zy|2W8-sS&crv_k&r2Q-z}E2mZ=Wn2qSQ8?e4LC}VaHpniAUYGQvI
z>m%m30IBDW2&sU4)YsJaGJvudCQ{?zoR+%59@VhEXr0LFV}L_vcbO3VKxco1fiuvX
z)@Ozd0Y*j*g;WAEVnOP}`kk6qm(75oDLd{g0Y3K5{$nE`e`84&k>7`Emj|~&%B#swe!7%C37!#GzcsVYw*yp?&st!9$R%@QKqH0siLfWl}NV
zhn)`XBMtEVs_|5H;Ac8&7!&=a^*%qRDj@4EtDq>rRq?OIM)zU4{E(cUxC;8PJ+?mP_t~{Mq)8>t#Xi@|B4u;{Ju#G!!2JR1>^0Mbt5NLCK04
zY$maPV-a1J3$}o_P-_WqH`?Ags6Rd}Jp2qe5|NBNNacPPw)$2mQWsp7=DN
z6SrQHDuVR^k0%MtfUU)kJ!OG^$nJA-6!=q^bkbfJ_?-3ay2QJn;ywfa+&6~ylO8=J
z;(Y0m@TLsFY5A+ii2jlvCG%Jc&ZAa8IiU`;`HP-V;{D(hl`tmy`$KCoZyC_4MYio9
z!QSGVA^IL5Ke;|B+Zg1FEoyy;cemjzJ(d;v(7oBg(0dR!(~-LDtAN`d#&Sn&leAy<>C?xI);8xP>1h^z{b0
z&Jz9l++A5UVZiV!8w#d?uhI~r+6Vq8&sbkE1HV~wdO0=#|J_>JWh&SiKGgb<=vOrc
ziw%kWwOd_FuY>=`Irk=QA@*@ObcPY)(s0ou#~bXNFDpKg4|akZve|z1CojfZt3c11
z#UX}xhmU-FV8IIeCoVIqy9j)@jW)us0Mm-*@qAu;Z*zx$h|;
z3vc82Rge?q+H@`yuxYUCvNgzux=gOk0K3k0JG61ITiM0ue-G}9NviYc7O>-p8MYDc
zqFHwkGm)=S!$rk;fN>VLBQ3#h-=%Si6QIE;Ut22Z8(qPoRsfxxY9;?2XjPe2szkl*
zHf>+|0TrEqL9InEzbZK=Ztfm0luNPs~idJe6r&xAo}sZ(n)qb*gvqJW3x5TYB>|$4IqE?
zdi}RYKu1aYZP)~u|GB+D8u)AWuj;b{dOYEASQEq}&P=>52>2r|F{g6Cjzq+gl0@)p
z`C-$`w_u;8*`xf#JdxVzPTdE5Ik5)CXvp8{J!AJ?0e`kmLQx#_Qphv*>0sxKVVl`)
zK;0h-l6kQHJ%9N=qVBdwyxdv^^w;AV+laYNp>*3bb&xBn@3JBKGn0r~Uo`M1qyjcr
z1Am&F9-|ES;D(9-26|^)4^IaHwr%k#C+4k|rmkY5zSbYnn5YLlrwti*Ui-NWQc>FH1^#cJM8-ah*n`Czx^qN}d9CQICp6zg*gME_7sWwD^ZkgSqvm5Y2R*-l)pzkATSq{*T
zQzV6S0)2ojU-t;eKc~e?5q(d;v+(W&AX~HHSK@u3ez|_+IG|}s=;o!Mm%VMUfQW~=
z+CyXFKHf{bHq!-s&VZd%;vGG^#!WsQ>=}0P#|i?UpXqoX2l#dAZ9Kx;|u6Q(nm^W_LCQ1`^;&(Vc&lTuA
z<;oo59bIH%(oNK}l*MQgJD_jv>xwDZpV@UI&ox;0uYA;D0Q`#T#xi0axxVCnV-{fC
z#@?+<0l9xhej)U1&-k8|1H9Ae?@$JM{FXOW3BM+28b9{|y*!Y;j(A5)B3HE&?|vh7
zQi7N>p9yE(BjyRFFneJMK%0t-Ppd$Fs9!{zn1e3cn0gcQ&GnTo#y#NIPL=pmxxhd7
z{=
zBxytu^FaCjtA-YU@zst}rGPiO74nFC_N~>!ikO${&0cLI{K>s@`L+Sz&q-kt@jFGt
zezN5($gSN*n>`8n=-*`^{1a%&!XiGRe@hf)-X-QA#p-=Ij)2{~)e6Kre2yX!LipSC
zGN)M(^ho_Wj6~gT5KA{9`q0bVV$yY34{-4-oCDmvHK!mI_?Gq8k3Rsvo!%QdhXR_-
zx?ENRzEr?Ht@l7LlVnL~gg(dpiA&rU@a*|*GL~?U-9D>xae`c{*@GA_z%|2&=@7_A
zK7S@b%qJTAay86=FLXXX{V8CZ`Xf>>`13prJK6*|)q7Cq6recsOs_e}fBCd`xD@2{
z$It&H@)mOY!ZxDMRPzosIl($#Rhv)*@Gsott+WDEWuYo)0iR55++z&;I=_E%Y!T2g
z&F||hVcqS10x1~i)JuyF5PeifZeLY9;PLw>ZV~lx+oZep7TD)n@zr_v0V1HE3KfK6#2SN)iFmzbAYb_?(jbCR4FLt!?YUwVb$I~(x-N!89#V%}jtRlJ2b
zSNhMEtO3Ar-GjQZ=cwo^_BG<>aVjdQ2(4TaPDR0PN4S!J9z4$+nGsDz)OurU%UCL!
zTA@yp%!3>VjL-{DL}Jw{i4H0BSXM~71rQ@V!G
zVD`_qm(+$)fe)$MHewi^*g2gV_h=YtrhJjknHWX^%tHaJ_y}U{ej6H-G=g#h)xDW(
zMv#;{<9^)~5-K$-yOdc&LRLHZqF;!T(RlvwhDi-FQpqY_^;wIIauw1tkLi%ntM9y+ZP}NCW8U(cB~7`v#Eb%C}>EBLnD$|FXMlJ`JF?
zd8x^By#pvFC0BKNcmUNL;?P?|9YAI^veTKL2avv|#8&zK0Tfu4R$@9zLZmda=@<7&
zh|9v~)3SIHI?PyY+;xkDc0Z(`atAW{u-$g~02)M*v1*#VX9iJ>WT7w9mO-?eo~G$H
zNJj7ZWy4h`hS1p}dH2d=Lx|^gjH00U5USXw(ZPKZ_=i~IWS$Kp_2QR*xpyxT(%#89#ysTNEm~zEV)%SLy2a7bs|LpUM7X=mN
zwRrc-QPHONdfMeJqvI#w@7P<)M8X8HXgG-btsiR6*d;tM%2)w
zByNB0fr$ac%hlMF0{L*QQnjoGe=qrPt{8QoqMT`GkNzqOI=@En%on0g>;7>vXwR!FZ}dorm*KHo
zxEB>2oB8=FS&E9T?DdS^!%so2ObiN%{XF|;!dNW;AL%71CdY{%jonO=r86bf&rPj6eR6qK3q^Th&;laGwdb@kdi?$Yl#OL
z)y#P~*V2d46^)RqEqqiI8{+IP9YIA8x7oNkLLZ9C3iuG_LPh!#D@x^`ji4Aq=~Erz
zgNWVw>X;ndM=lYK5OFFQNv-xLWt|&AK{hv{NzjJ^$SEp`(~M$ueH5=!a3+fxB`XgQgDUIO-6
zTOO35ZAL{Z`~gCb!9NBzkt?~-S5?-sgult4AU}h!O~RLkQMps9<&^axO6v2DHfkUt
zRXc2FohS)O?RT2FHa>uqZQ7HLCyb!+U1}D$5NRrt|Vkrv^~wpMNfc_gSl*C%VPeRL&mVd6XB%z<~V(cRuhY(IF
zWQ|CoAl4Y3gr#uasDA#YJLwPytxnU&?a=49rmYUGhJNPJXD3rWj4U6#E?Q$eh;HQX
zQn(5Bde2=ETGB>BtF5#{ZaWVnRdU^K5m_nKzj?8)V?`QV;jV{ub61A7V%tm@yMjvxi4<&_Vh4!H;9udapo`Mnh=zC27uS*GE>
Wd!Y`ur?$&A?14Jt_Oa-Pxcvv*Lk1WC

literal 11976
zcmeHrX&{yD*Y9nNG#E2vE<=R|(#UnI$QYGmEc1|gDrG9kkd(|Kib6_gK$;ANkf{<)
z2;0WqnTG~@zv~vy^PF>@^S&S6=ly>;=f6Mr?X|8oU2CmtUDp<=xkqE^Vxh$t@nK`b
zm@&?|uM2x42TpM&*y<{9&6|OXP%tvHRAi6@B;I{feZbb%=+8Ng54$9
zhB*ck#)>V$*f0h+TPIgWV8iDAO9
zxNwwqySsXL*?D_AS-aS&`?}b9sV|6b+5!5ZzEVp78w6iZ_#|*
zM~_>1{e9rfbCLf3^QRuif}Wj=ovWRj&)@6LA8(M|I%j|ho8PaiowwsrpOYSTbNl6n
z#fAMIa`X1_^0o2t^|G_o11Z?$U*r9&zP*#nZ~aAo(I+e`IBuSE_ZPn`7%!1$A17Bk
zj7uE-Q641@Oo5M}#N~v1dR=M#ebM)+Y0hWv?x*8-!WmBm!tb17Bv6(_D_6S
zIClnIzpdf_n^B@07NQ`Q7Ubuzm-$onrxFA9FU#-zRrg=zS9~tG|1-bDxLDZP@s0o7
zB>OWLfBkmtfk>XiYYaa;;Jvru6^6%0W{ODxwvX~<9je1{?itbxH9$AbibQHXhAYU~
z$XbECPo)#SrxC-?aF{H51L&vnUOnJ0h9{l+Xem&H;lW1dR9@Z3@KbWWHqIp&u5+&)
z#XiCCHYVy?`)3%g=p`C1U5??rqUWXcDlwccd2H=#K<24BZfJSDi*-si4u<7MQi7bb
z1cu`)eH?ZuV7TNsQ+UD-442Vdx@k%S!_#vrb=r)e+#J%!{2VZxee`zLEe{O;e#Y0E
z=ROTRjGCBy`N;1gW(~855!aGbhJ;&r7Yz#
z4PA0<6?#@iLmORP*}B*44if{ku8RG;i3R>)09=7~N+Gg=sKPvfi=cmVcBXY<>kj|!v`1UexO
zqwkMr0%3m>ZmGWSy`XQ7*_7R|q@x`UQf_=uSLuACUNLPtS~3yP-D;2Fdt*hP_D^OM`PiI5EX|75d0ocWmVym~Tqb8QDa=iVc{Df*{UfwIB}cDu!n|bCkYO
zhVv27?$sYaN5#7Nv?Wb65RBZatwxBJW$iR5A{>!2*0^opN={T)2F^Q($MrRQ_|HO8hU!AQ>O~{
z-LI~uD0GF61X#1T`>us^GI+bc<_dT;+HNpiTQtc8ZGF1ql@wzMlRn@(zLh;;Xo*GO5t_CE|HHP+!GG=1uu9AExxHnu4}N
zAKvwed<5|>HjvtPVK95xQK7u|n
zV%eQ>5NPH@0>Lr47(TMaueKEK8F%lr&0pwX=e+U>ZdV%W^0E)+&4N4}r>M?Ap9)r=
zuBm9kaL{@Cg$R*B(`59`c=-7);}s@GYDwsbV{yf{c6?t^rX
z72Bp@p0B%e(;^P$@%NmL*?O3#^?QG?XTse1>mB_+`F-Ktbd7k4!v~-H`{y6lg?A_?
z0|)$2aI%6QBi=_hk>`I7!As>|^{z@5`yJIk@~+zSKlQF+oX^+)?LQfRpHD>`9@cZ8
ze_cK|$o%*6NxV@P-l-n&-Qc&mfB5g;=WAmG%elNAquU|IWa_Gw|;W
z{J)a{c{6X0ENldsO^UGlO%J0m_l;UxCx_9qlw0NLK=Z~tyJhlY7?oalwwQ5t7%gE$
zp1$;9)FPST)dPAi4EGI<8Ai}$x`O{O@RODpYdA8EpsP%dVwr%}+s&4_Fpi+~)=%HN
zK)#(t?ztx5=JS6d%QrOmXdgHCGa_(ihfH0y~2ai46Q(0b*h=?18tNkVW9@{>OU=(z56wcM$?~0
zJOMg!yWX!^K>eBW3P!LK`SNq>CeRzuCpB*XWI0+K`4KQoeI4=!JLyZN?018FxElV}
z2ypl_yTlMi1epCXR^u3~7aBV^o&-8nnu~J+
zXvTgYehsL{XNLOa5umrEsG{U6)K$UZRj)D7Gvy~$bO3phB3DI#yamhW$XLKn+)L7a
z07@&pYIz9uq#`T4RKZSCt>bz=kiQx|xXBp!!XaIbH6V9=WqAh=@QZ?~jw%9nv{!bV
zg7t@by4CN&A9(ueg*w6@lj?4kpBr@#mzrKzsY-r
z(hta^T&$ZE1IXFN9FIK+^oCs}ulr$L&@?vUAmG8Wm__4YpSz_w!wcx!7cjvr@Uv1@
zGLGoarnFu~RiJH_%7z{RJIP-UlnQ(wMoOK94>p1PSAMgvOCg?1BfAThf&FK)8oj>2
zf3VI*rWy3aHeGfW09tces_iMnr%@vmBnf@q=C?gb4g88dU7>Xw?CoTHc;O=Wd+vgj
z#{!?$L;4vF^rYG^C4RusJE8nZV25h0cOV7UkNc!L^Z}p!%O19^;Qy^;wbmQ3Z_}E3
zyBqiqeNPYTf_y?CkH8AR^|YjLA>fN-Mzgj64u<&GhytIId;Mz<#Ch5N@LM6Eb$q_)
zssP@Sw4Uh)xs=O%?TkR%M24vm{mh@dQurvK!aIAFky6La-BZXtN~$-BJTUE4c>;SfQ_$XxWW#fIbT1Z97f$
z*9kqVR$*8#qomdn@rI|>ydm;XzrsdD5aK`=3$h=9T#3g8l_B6Or3tSJgE%bIXk8=*Jf#u%epyMPn^HNs#k@_0p<8^PZE9nJx$(>8RT5h6455$
zi`%$3(*ei50-`AJ-Ko)-eIy8{zm`1oXShIHQvYKRd6l%7^s<
zm!~nzfL(=8T&013)~cuQEX0!?7jDfD{6)3R8pOS@!Fd|~-8X_vNRJ;8dCqvuzbyrD
zNH+I0F<&CiNU~kzSe=Qe~
zr*4i*(+Bw?qiT;VLhd3xk`?CAof&e#L&%%aP)%Ac;NH@VH7%fDhP^w=4e|n#VH5m-
z1H2;5-vC!@n;jtR4jP~DBLgam@$19@_P9>hO-qe0!~+ns1WClZsU8CuoH3@KTYWCk~uCC^ZN3AX(fI@
zznfdKCxO52mA_&)#2-Ixe$x=*W=`*9-vazk8)-MFV25(7@ewhvsH>QAH>zciLaIjm@&h7IM&Wm26{qPR3V~XiC5%;2ThaWRhuTq^Q
zg_(enMg<|JV3%}#lwt>n4|6vqg1+HR?5rrzS&3${KY&)2TB}I(+r9(M;w^x0GiT!o
zd%cZ68)&dD!#hhq0W^!sM;;dNcVZcRdJy8-$o;CF$eZI0x9=H%wL;7~#9UmaeSF6Y
zK#ufLjws-(yF1E|z|M#jdp0qTsVm3XG{L`bFZ*_Lpq0|c++TsbcV6xH$3UN6?Y(6i
z;H{qK?A5@RHd)(k1$1xBlfZh&N0y;jlOOP>9Ks^f!H#0^@}gLXi{JCWjl0Bt(uR46
z_e5-)Gu0UQibAjCLZN=D2S*;h1^!fxgq#@Y`S(v-Cxab-ohHKqz@tCcNM^!1^9>PW
zqVH6L-|Vaadg4sVF5+Fc`@yak${?3l+ipS3XG}n;I~4fit9-YZ1An4FIZOdC<@N?%
z4D|dRJ*WHtTX%Sr5$~;*`u0Mizm$)wjMW0BD8_0N_xzpvr_K|7>4Bv=5p!dz(deKi
z*eSLuea8X(P+ig|3ebC#way0vYTij4jfXtwtj2yggFI(l%lUZVzYrLGvK9Du-APtA
zK(D%S)BU~hS@3hNKhc7i&$TlL)%OGDr22^^0|q=^EzJ)4QF0{zHlU3cWoaA-
z`C3}!Dq`;Gw&mO(17vQ{nV7pDzU3W11GGv}!1k4(m$8eSP2|H&>5)Eh9;;(-O}7J|
z$5)q1+@l%mon({2o=H1TBp>iun9g*wLtNKCM6>^Zb^qyZ`rJFXc{pDVXrJiVeZ>9z
z-R501F*j3>uNjgDdiu4&FYKPb~Vu_Cp@Dw9f2-TC{HKu(ZvP^9YjBi
z8V$#>0lHPcdo~IFe>rUByanr)<&RsnfnQKrS4_M|?kz8=O9hPD+PQNj;IdyK-w3^<
zQC=5i0CU@XY>GjT*Yu7e5f_D~^2-?L6(`tghzc8wO3#d@eazcIN(maZ;Hen^glve2R1}UWB@KVY!`Z$%d#CzbnNv@6&
zU_zzsst15M9cwa)bM~###f*3_zcYNhi-;%b-i-onz;EOHB;t39i1qmEiy$Yzi#8Ju
z^=aRy&HoE%;hcPKVtz~KU%5}be}pQH(`^AexGL8W_waR!gg+5)^_%ntKF}lgsxuOO
zi-eL5h&l8%qmYyb>pl+NIkSK}cBW@10^g!G?@TGgZTeZqJ^;{Q#^Ht%@D+R?s(l7}
zjU-D<9n3kwuN-1tfXSD4NtwbqcIr{)SPXKlhNWR{g!~|4ItKCqFJDLy?-Lc{3>8D*
zFTb3X{2cJD@?+9Th^IObJ5>)j)M=?60Vu^h-FXP)DPIo{J^(qL(aXPxx}7Yzx{H`I
zFS!Ql?O>g^qG?4i@U!o5m74+XW1+5513vS&x`X=Q*Y@+%(@TIp-|)G{6xN+eVn`=}
zPPxA1C^1JhWsEDD0Z*5lyG!&#wn0brU9i7Ybi#Zu;7EXsLIT*ani!2Y1ifw#-)%`C
zSM!8+pLj3T@8{(t-btI?7;@5J|La8gK3PEgFDrD1iT4gqMBxr%UpIffPVECEYglST
zUZSFh*hKK1GgOrIETDXO5ETVE9p{Jxnsk{tBqbEER^Qw-l8R`e%9PrRRHWx?XRa9s
zxRqOX1E5`Vx}M@s3W_EVurFhyB90#(wPr$8Bzl4&ZmB#KCAx>Ya3U%)iCHyNq)kO<
zx=;1E9Rff6V|%_GrK0Z{n|JTEr=lLcn?LKa2GCQ_#Mp%P0W_5M>%(=WL6qx3>NpZS
zh!S+C5-&U+L@Ej2R;P~*B2VT4Usikw@pOC$2#Xs+7f&d=Ggl2EDQ89#jRX?PGc3M-
zrHX`(>T-v+2==2NS%X`~Rr=AU)WWqrYW=8WP12Rq>iwv>`S9xCfPNIwux;$a$A0wd
zq?}oWBpGdY^UqQv^`Y0NGLHY~?nD0KA4a-|`cS2h@cs2)`_QJ$#Q53HJ`|pip*S_z
zhu$A!*IZBSLwZ%xQ&)QW&|X!EowB`sD7ZMO=)f=u4J8>)wUm$$w~@zJ;b;7kqtdb)tlUpx_G366vqaTUB0Yy
z`RM^PbvI0oPiz3??NVu78a{v`j=e&FIy};UN@w
zN14+!dI)vDUh}Mdf`SGnR#$$$Nbo=$D@D!DXF$sc3ekfa`=g
z6|J^*C@5?_U(`3~lTn$g50?uMAhBQD_ja%9M~8VE#a7Fc
z(4j2`hMZ8RUDdn^g9ClYZ(|4R9&0L!-pg>|iU1XLG~Qp!zlnmnMD%l_jt;{74e)c9
zA)^ecOZ_TWsN>gDjj`lDL@62;+{aHsWe4Bhe6LADvJ6jTg4`gU>0fW-S5Z;%Vb{=u
zJQOs>#IPo|cK|8SX}@A!`%!qnRIWC?53TO+c#i}~=)htnmmgCkR4J9>To`n3vZIm5OK?>>1gv7oNpeT{N#SOz<|R;KEeY*6@V+PU8^T
zE&6Hj81$iyp5RiR+`Q?;5Pp2gQ(0d(R9+BjAFXoL-k&f(5xkpBRaKI7g&PTdyL?1lp3;v(O1(pkbI54d;U^DvI}W0H-tW8p!ewBCWH&l`TeOB)N4(ZZe#^$-~SibKsm@V!AB5a*5Eh_dUuX9W+m*`t(T|%
zUNQx_skI$=(maGjcP6b1s2oCMmu@SmvO%O*`YwOHJ{gr{?OSsP>~&tcxng+}3GFu9
z<6mG8{`+h83rIuWBz#!6Ks{=w>k=7Y&giM%e^O`%_uK0Yo)(=HWVs?%w;k+_FIJVh
zuhoxy*(-FXV86>4Hx!Yx$mk~54s~3bf{v(E2P`^GMYoDi`fI@4B87{*MYw&kwwXHwODr)PW$c!_bEciOn*v4pLF`QV*kE
H$lHGaP+}F}

diff --git a/previews/PR1624/tutorials/out/solution_000090.vtu b/previews/PR1624/tutorials/out/solution_000090.vtu
index eca7fdbae6268acc852fc864ea7c29b3f5789602..5b971ac051575b4de2c99721d4345fbb17682acc 100644
GIT binary patch
delta 12223
zcmV;wFF?@jl>z6K0kEb6lMDnclhXq@0Wgyc1Y`m78zUc60*$qV
z4tHItn{Xwxa$lzZM_)pnk4v`e1Q6+gdpi-=m@fx>pJ9
zTUGO0FBX3|5*;GQ4g66u#-o<|!^LX0JG7kPRj?B>*;6QKM
z@nxrG5PG^c{zlB|Xy_pEBil9hz|$lon^yM|B=
zce?@iErhCtJ$CEVBy@U8{IHQ3_Wxd8BkMNRITkf5d4&L(!!5}8_MVVd=okF
zPIJgQ`I-Y=O+J-}K@4^mjb)}~GN>Q3vq>#wpjEf1^KmhQR_&6D>z*)>_TbgJ_LxCw
zOv3Wyd5>P7^(Ox&G`cYO)?=pC>&U=jgOb1gaXc@|b0q%=1FK3e
zHBqellHxSXjxjh8am)XV1K#(Nvr}{cgN4^uHC=RPV6*O{gPkh_!cjM_l
z3{pIeLl${4`1(8ao_7`l|IXowl52SWQ$%zfo=d+wc(Xp3fk@dxw~VU{>;yB-!yn?k
z78b@ewlWwpif*W$#QE`aE#9IW6pys}@0Ne%pt5jR@dsh7=Hk&OKQhR^_%2r!_qs@=
z^Ua_0UlgY|$^2lTJ+wJZaEJk)(>2)?Jm0lag(pCVgJcQL%pe$Xpp`t|0&!<06=37=
z6LDvttlTn<&;#nb<4OlPNO&_YC6UU3XeUqH(02|bl?q35tK?
z_*}`e;$UAp2mM-AV`1|NZPp}0Z#NPu+xqJAm=U3WU28oiV^63;U&)|{8=PUmLfupa!*UXen#j)KD`e&D+t};|Mk%Tp4-R!ufGuTe`o2OIR{Zs
ztcFDWT1yDEcUY7rf_39ycdHMc&#~x#5nqbD<3IXvb}^yp*(%o8PrmE0b^dxT!?F#WRSU*x&2pV|ovLbERHwU1nZb6Xgb
zne&c=2B&6sT^rnw#;Ng2+<$+tMc(Aq6b6@#Kh#F~Ge{pgm~XAE8s<4!>Cv!ND4m*#1L`1G78UH$sum
zm1Xk04#-2@XzfA;TSDb9V!cEodWI-xTi#fN9${loP?INZngH?>t=|5V38Q0btt
z-%bwx_xD)PjCux>)0Tg|sX?CR+uznJ#P{+pyL@xpmw~GC$?WY74BQ^|`h@FpP`2iU
zFc0EEJYtckW*(sr*CoEJI8SKs5B0kz{w7qhQsQ`6EeF*f_zFyvIB4E;$~q&D!PDHO
zZhUSGKF!^cMx0TXo~Xo3q%sg+q1`7XfqeFT?sDuA2iG6$?9#fNhW8Ofe|CyvmYs@kz;I*8{lQ)2lz
zpViaL_fMcsly-mnuD1?na5y!6!V354Qt?6Wdklk!)F^4S$e$>tI
zyCFldsMEJ}OnGxk7_?dD#p<|lU~v3rtc*0Fg$X@IUjBcCPO9o?{D*vPa&k_3g1&tz
zW4XHUHA3%L$JQySqVG4??CXi+;2`ii=e99O*l3dEj(%{irt{ii9|jsz@0)5-7muYL
z$%(7xz$W7P9F=|On?({mchQG0=(cHeqb{tu?rd%H9CJa4VABBd_LafW)Xt@ZrnCzR
zkC<{W^f`ZS`#CclVt}A1ys?R%IqeXyUqnY^8JP+f_d#
z6mg%o?Svs}aU%zZ_XJHV!~2;iw)DiHUnohIDMTV}
z^1NN8R(@kZ;*LeIe4O`TvyE$U-dmg7sdpNoN40;GGKvs4-t)qJgRs&b0ZW4Se;Ya?
z$&*Fs63bj=C(I`kLEBYdqyAj9xFrxEL+BSz55Hf3@but*M)&_XuuzYdemRA{p>9~y
zfnkD2&AjW+)NFw}|nA>Fq`u^;=lc^PHc-%dW>f
ziHgGdzrJE$K1ri}@j-;Hmhtpz%fY>yrf*WgeZQ&b
znBgh`LZ_{d6!!4J+~e(VW4D74}nDYI_Uyt;l1~xqsCNB|oep$`M!f3$?pbo)Bug&~!;J&MkYW
zzoG^A=ICm2Aqf3^c)HXX1v$jKmdl$KoRdv__o*VxJGRYN`0wL)j<}4lrE!0l7v(iS
zPA%r3Eg|DGt>Iw!sZiin4MMLwozM=(xhefrcaul`D%)h_>}$$xxcVOIKy2xJ^KRtN>t$DS%6m9)
z6IY$zjQ;0&(4uz<=C+LFN3j){7aI*+3gt7=n}M%j-(RZ;bt}2(IDz|r_sRCmYTWyg&BeEt
zU_EdAxy%Ot$6R@Hpe8?q&;{p}>?BQbj_rE7W0*qU$E#CgVXbdG+Z4p2Y#;%Po1TXd}%B{OE7<>YmU!MM&A4%B({ohlkhSke-)N_
zJeFA$$0=pc62i!qgttTtO6HBC7)ymrGznwIlu$86izN(_CW$ObM6zT_8)eB@leHQu
zqU_{-EoBSyJAZvY_ndRjbDwk1az@uzXHnF2DqW(B{pn5SGc2Z=L3eQUEQ={I=Bu%Gf1brGi<*+~`p#k=vV3ss#sZ6ZBr8K$Xo?dOYp*vYpX0kRj;RKH_xIf8bpSIk#IndmKFM82PUu;#RmA63GKUdkV6)hSL9!<+-RM
z9=tlXT>V%VE*fsE#Qrh+6v?mH=N&OCwjS$OHWf8Npj^z?)x4lgsCl;AB8$1y?i%X{
zJEgQV<=HtF^VO1c1_$&p_MFka@E2*i#-{=Ny(%)N4RTq|zdC-Ve>f%ydELSOOp_h8
z_ply(IXm+s^xe5V?&q-%5I?on1^O)Au=;mcPkU^r{ug;5%GT2J!Mgv@#l~^i8(Exj
zW`Uox4z>qF9{gUSgFS2GIzXw*u@frbq^XZe7W@YpSBfiQ
zKUa=>)CBDH{gnM&f3dz&)X3zZxSLr9J;$1-D9p{hE)h8_2S>6zMP5Zi&*KJ=0=AKCn{RG;bXq)|;
zzT+~{{UhDas+BAU$j6Q`xHG}R8`0Ce$dl#5QT>!>UNXH&w;;D&?j1$-7bK;txCZ<+
z)%sB@Jlfl`@q4dmf#d}AqoJAAc$`5`mVW>?q4EK?;Gr=&KXLo
zg7H<1zCPS1`>8@iGpMK)-!+PO_iW@^{1xA^DE+O0fAYz&`f6b(`pSyMy3T(1`4lMl
zhbh>le#}z_``nk0JUj@0`Hd=hXR+TX<4vSL*8mqKY7C=AO?Rt96eITu-m>VI_M?zAxDG<3vn(_rt^!9
zU=O3$f7T1*@<-Q@3&kMk;eJ{%#u!uqyD7QdT2BHA*XS=>mJzenlv-cM86e_+&5wkKO+?d
zdL8ip_bQbS_E^7d)?Z2aJYg64EsXLsTR_PGe~j<2OiX~^6@EiauCQlcmux-&JB6Q2
zv)b{Ug}Itt(V)LoYUU5n4;Hz~T!dcwen74-)z6M+bL7{YU%FEtoQq?q>rmVco8qa@
zR1}B1Jc7JOFvIFGzLR*p^KYu_E!yfYsV~QDb}XR2%c&nWN#~w96unyna*w$iO=930
zf4fVZ$06ssd}cd0cxjx?P6d5dD;k+Ve>+z2=nn8iRa2TR^<}pWfBs7SELFLh6YJf#
za$2c=JTBGMn}cyzhi&-5ch666v;$Lf%A^IU&isnEY@`0J)hB+C`fXjf$Pj3@b|GH@
z`pRU@0bwvLL)o0ZU-qodT@QS%Zhw#Be^Ii}DVXYM(;8Xc7g#5#cUDvX{&IG@{T%4n
z=+jPlzw)q<=40s7EtC`s!OyPIU;Dv(Qy%e>&<`hO8HAz$xQze0ohjrfET
z<=uR!$8a0nV~W;!>fhO}?)8_cpBELqyg>21t1YZX`PDMD&`5p-88>OZxMc;$e@oE4
zuX56$__{ZWjZ^;>6M1)+=I?zw>t`~MPnSm=m8H30cyq=GOx`cLjP9XJ&pe9eq1t1~
zMnkIWE&azjX|80bo%sMhd?vG*=F7#wUbBthKIdi6b0Gf^z|GAM>aYY=sQ(%5)ZRw@
zQCe>I>A51hz;r0w^EC+PBa-y#|{^57T)h4xh{eyS3lU
z()>>0)jm%5>{Jvulm&257-9{*7PQY+PTdqd|xBlnQbJ5nSsLG#c+Aj^~LG%E7_9U;)^k?t?i&=2%{
zt;q)WI;HE9U9f;%pEKyq=`gg6`p{@aIr&w6T*^WJW5p~Kpt(CdAHlJL;{0Kqkq>xt
ztSW)H@1MM7+~6<4fuGH(e~#4_PTGJG;oQCkpb1N-yb$&G&xg!>%EO9QF*lm?4JyL9
z6c4-VR(r}<1(T+zA@G|2Rog*uvCGvy0pv^eu|5HoR(%t=4c;@$HOT>^ziT#8{XS;J
zJP|{mC|R$g|BUjh*hNSKdPd3$e)2o*rdC%r9C2*v&=G}LC
zPd>SIn%)OWy2%dF)JF%}{2Rf7hUwB2$m2#=3n)|ECV2X3PB6P_C5u3dk+0&9;lE%h
z(O;YDXpK@4<)^bhf1`oweUXuHe1zs)_2nG^6aHNJj6Kt4$RZ(Q>>=I7(iuuE*0l|Bi|Khhm)0cApr+^ODP
zhp0MB();4X`@kLG$=vS~`E*`)My3<&(k_j2#=tLsRlj#Kf8yREY!Uk}7;|(?jNS<+
zpEgV`&^c$Ul(n#r`)p(by_4^j?y=0J{5JjRk|(JD$F~%!BiYl*saJ_%F#&{fOJ3o9s~GCU9F%h$hAN(A!6s
z*1=9nUDBzI-Xo%;jT&IKQ(4_AP$DTI=@`AUru?@0!oD?CTlo*z|K4!jr2@Qh*+rl7
z)WhoBD|!!}ZGPHS4Eq}P3!d9#@Au_6_0>VW3)cs*e^%w`s-k}RLf_^z)oWVW)BwF3
z=l&W_dJXo62Rx_#*R+59lQsNsac$7FK)kKoOjMs^PDkI??wbMmq!g^^J!t;mo}U{0
z9oJvKwE*l&;#KlT9C-Qb{XW1>)-=V{6Z_X@Ctk3Dzv`?P`>DT$b~AJ-?@wG|L=1x+
zL#y={48Z*(HA$J^|71Z1Sd-8v7=QW-00000004NL9hZ4LmRS_Xl_il-k}NNwQnX0f
zDsN)4yhfpD87hQGNwQ~8gk(vQX+g47yr`EVrLsjyO4^jrOAMkBS>|`aDGn=#1u61I7$#~|rYxU#+(|THB@)CGz6MvsA?3Fr3pMTt@2bfjI
z(}x%E3^Lc;n2Mbbe%ohzEdzQDt@5w_;NxGKhi5_Wx6q9l06F@~ttM^Aqj$8eZ-Bw^
zvoC*~G06P#^hAd6%t7XS18qAC$YPH~EA3!c;&OxoG0v<{k*fxpZ-|KX1ttG2(`*6#
z8&(NYd^MAVyAI$lUWbR9v41}9!ekpe1ih=^g
z^LFxMzBv6Q%iQ4CQ2o7H9o#$mIe;I05m0}p7xNsjbPiC4{MIgbu@3Uk%vLuJM}Ag$
z*uQLq?3TEPm4i5p1-U=Pcn6sr#(=C1;&9Cw)_o|AgslkL!lH?vI_L(fVc
zXO}{+w(QTcV8jX26@TLu!ustlSEZal{Bv~`*>TY4?B22XBq+6Q!#_db(bDLBC5W?9
zIWkxSY?pYkhw3GGe|b_dYec~wxbassSU{CoGU4I=$^O-bHuC4h!z*I7-
zJHH$Ato>fMxdr128q>HHh}(ACMc*E>$(;uqOyJL6p{$xi@glT7n?T<(bx&#t@{|7E
z(f@$)GJn0l3EDE&yMM(z*B9iKp2vPt5?k^JG3Kzt;&_yIJDM
zL*y}6vtN4#e<-ABx`dr1kL5vd|bGqn@_;MAh9}S=%(vpu5
zhO8`O9Y^=+?&S2dJm}9iZrRobUa~wLmISWjnDPDv$_kG%3Sjp-(fPVcO8GjE;P+2Q?-+Jh`wq^}gf%ZmM
zSxw-TE^R4Y@M2AA_B7Vjv{K>MF8HU9Bqn$vo=@j&U0O%@s8p{4C{d81^BU~ka(MpV
z;1B+gB{!(9N(v;o82{4K=KO8Q{n=T=bf1sN4FuD;=Y>_;YhmxuFb;V`b?zJ)bs6y&
z2aB>3v40Pj6VCDx$h(?x=9UL^|5!ZCb`cT!)(^c@Z*z)$NBzP@rQQte*
zrK6SxJ_dcN%-W;C{qK7$qrtJFj50~Y8DWjs(*3Uyo!c=1T5Wb-egu>mb3SqferK1z
z-(U{%?b4c)4E^_(@bCGE|8?mj5$dZ(Mlo$Ily@bklj4v~{23}~@V^mX{+Rl+c=6*B
z2Y<-D(oVH>?weB{Kc{mQ%QNK{0MABvjr+lm?WyKk0D2#F45fQeX?4cEmHMAYu|fw}
zzuafmGH`O&@_&VSs6U?>dy2T);h&HH!nn%Ix#A4W&u%@wEEV=LTffP+WBA?Zu=$D3w}vmTt@H4@V`-yA|a3Z
zY}dbxIBMN5AJ9HtEmupf2S=QFOngC?L9XgC*tfo2FTM}lY|CBi1U)l#-PCQmmw&#e
zY;vIYuwj;;LA)xrahqW11r$3@ZiIezOW_lWZ^?R*GY-4|M(#mb*vGTu`aHm(FYP9Z
zpkzp@ZW{b1iazOd)4dDky0R!=9flXAK{xMo_I-?(ZHTS%ggx$apDg^IHW%^OxrYBO8
zFG5b8wyvHF`bhbdE<~Q(;~Vs>K#uLXMRK6R9M|JEl>Z-d42S8yb$Xo@f$V*&(N>)L
zm|n#Q#b2an-r@kp>rCbgL4Q!x;2Pbxxn6(VqIbr7#N%NQ^{r^nB_}|KdViTURDb3b
zLn4%y8vj3&WkAl-klfi|m%q-^R*Vxpq!zISTz08+4?lP;U-$GIu%_p^mp8>PkuhU{
zd7_;9OQ32u&&)5d8x$3%OHu#YlKqhK5f^++M3DMmP#uBaz{3
zI`@DAx9Mz9Vcz@?RNwa#COaiT277@L<^MpfQtklGRb2kaCMC+(?@E_X;JgQZ_2r=a
z+ABl!J}W!7){H?ONcmxL7ECcLsdfb4@HRHNfS)(WDaTOV=pD>-27gOu2`f^)Y}JaE
zvj7jRye>@ZP}?JPmd@L||JvbukXKkv)lGu$tTt|^{*}2*rvDYFr8cGX9@JU*UC9Tm
zO>{E!2E~GmgXds;OC)3UG4PsNbT{>b(%N|;=ji*)*6}^{MdQ8soR9GT`q;JB5Y%o{
zoTj?9Nm*P^d5mo4M1S#6eV+XGlinr&CNu5->wJV)jNVT}h1SUfh@N`yr%o_Z_!lh0D^u1=0
z?P#A(Z2a`}7HDI7fQQaA&BAmpy?ctX0=J%_Z_Sd;uc!JxlYdpMqY1mHdM9r+^$Q{W
zgR80U&fQg50R~kI&7*qOf7_Q&{Uf|Bq=)v^CD64l3-(t6_r^KsZ@D{P4;{t(GgZd4
zcm~R!Pkv3`jn#^>$@IPGw4ZZd0RDTXpP!-kE?+>HMd!6KZ6!b5w-36l>&>aYKivKQ
zo|;)(8Og%$TYs)_`vb%eIX$l82&xWwn08?OztZ?@MB%?8{(?50XFost$^i5ML64Pa
z9}H_tJ~h~Njm*ZVpXL?6+v5qwzCWQ)??HvSA6E~2=Av?F7oCTy?WZhyhh}&+`~c^l
zid{naiL~x{LhHBXZp@(fOjMlDD;+FwvD%~nd)|T7EPr}W9~KNJQT`K8zw;`m^EYs8
zaK?F#cs}nL2hY6|;M2mpJ@n622l{RdHumW{fpLR3{MX!pe_c7pf$}kDps0x#@dd@6
z94M#nIfb^%$&mkybd7#T^=tNKl=|-iztHy2;3rFV)Nk-|Uzre{dzY`%PP)%*cY94i
zu({3hCV#!tsYkXo)BnRM|3_V0L3N3%f^<&@L^PP;)Hln7tm*yniBEK*@1eloJ0upt
zZ=L+U&S0!VW54l5E#&{NQCqtz&|zdKc?aw_bFLnH34a!|GK|hKaL{+y5bLdOo>jL1
zes>ohu6~Ysa1TgI973EKPwro(JUbuVEBFa{2pCXU&!GR0HGCrbY9OahG+ycg{{!i|
z>oSwjCm4UK3;+NC0001ZoE4XOTu$j5#!I&Lov2A0MOj8sv|L4-79?#7QIsK-)}d_e
zL#k;RQQ4(v+Qd&td9BR|MTV4Rde3>!7Hy+R@_Xj5^Eu~p&biO?+}HKJ?$hra?3|S&
z;Og@IMq|kW27+C`7w;Evam(Jc&VvGI^R<-=4hw%6JSVsROp1VaH#L+S(gbXYt^2N%
zDd3=Klwb8P0;VUt9$>QtoRs!_%H#>~v{Lg}lrNxi-x=$SP7dW=Lv=B7JT9#5{d#pi
zj{%?jp!=>o_Eyt*7QsA5=jpY*j^*KeSo7ogbRK_N?{lV#c$9xz)_3F*kNKOM$K#uL
z%dw#ihgMp@&f5ZUAK!_UO@b^>}
zM~g%I4d<{Zuz9t0r9F!)sW;QBHn2E(-Ffa0$t<2uH_WK2W^vxuF(QAM#gI;J(h(Yu>@Ss9KBe(`a=d?a+h-a}COMouKS2V~c!jIKvINyt_o8%X
zND$=T8{uX|uAlfA7icX(y=uUZZ961znrP@LG)S_4{^YU{Pmfh+sims-h^s{BIW)%fzrOj&`FH-Os>fIbvN@2i@r=Jy42!E4*TK|7I
zg`2`Qd79k!u4=obRu6?EgX77)HZ&N+FzMT3F(z&ft4#c+xaar{RRV?_Jy_!U}=_92DEw|m7s~pT&t0I@5pn!)9{d?{Yq^9v{S*l4G&+`bRMy_qJ)s|My&X
z)|F>dDX4AUrf}GR!b|UI$Ci-w=S3R%SXom@+3)3RV?sf8QC+I16opN{q^mA_~?HL;M{H1@pLWeP@4b8t+`3
zC9C(*@THy{vn-@>=Kh1XZm~2*wMFi>3N(h=j(l{_AxERs$Qg5J7)`4gS*=LpX7j`}
zUt=2HKOUMR9Y;e?rq9vq6OA2?CUzqSCGd)N*VI;IVCCPKYLLL-O=I1bBNrGL9~8ZL
zdw{_|?sFSv_DS%~SZANyP6mIqNneW-S|!+erTTID1qlM@9&s@5l)$d_dB^A~2EEmZ
zCzo&xk`=Tkcb{ZY>y~2{xtzn(C&53AR&cmt=N@r`_-K`Sm6~!bhfR%l9KLID=oz>5
zQWiLrUKwJuZ8*5rxnx*O;t)bV{xtj}hk0r)QbwmZyt?kHx7(PJWz5>eb>D@kQC&1h7zT&jMc<3hS#Lxa!z)h=7d0)Z>tnVDU
z9kWtEvf+$IymAPKX49!cpdWvWUDNwI&VWIRrBWzs
zDna8v+&6jhUeVTc1M7nniZt!U!^wNUhWlH&@nR^2l(gg%-_+P}*I#TB5Zo!YY$3oI&A*
zN?g4m@o83Nc13^O6bfI~JB2uHr?4w5eV2KNfcU?eDk7W}BhXA*Xe0i#iISClA3&je
zw)h3BLgR*G-e%E00o2WqN$C-Pt?gw3L0vY?Dt>mq*4@0{jA__*^>=oj@|0o`&L
zhDinj{+V}PS6h!qitSXJ4Gk;~Ru5-|&1Uf8oA1BXS7~f{RlRP(Ny0V%9ahPNEBOl*
zE8i1t)cAkwt8Pi8(9l&nd1E?_e$%24mj@+ya`tx1B%-6{+tKqohk{g&V=XOuBs(Oe-V{Il_q`^nxk!(tUgM^On>ZRfjxtrcZ#;BYICTVc=74X
z)g({H0?k`f1^lM>K6Gm(57|FbeM`q#bZ>r;{gX6<<0BB=B|NJvcsDjzm%`oFJ+UQ3
zhuMEsdlH=MDU^ANT;x|u;5aIIci4)>hM6+{jp{sVn+~>E6Mt8xi1Obb5->rub?%%T
z9#M9Nr$tR9_sZ-era-Ya
zT9d02xsU@j*rS>MhKJM4r5f%(2^f{1s2zV8CO~G|+3fTMJX}V#KWUq>aL7$kwkEll
zIIcD)K8b=s%L|jg{}v;9)ug?35ftXT%I?(~rr|y|RZ2yPMVHatMHbU}1SOT9y0nq_
zb;ZUVp=6G@s#Y(`ALFsnpmnjAHwVu__u`YL3{FjWQf0f0##;?F(=KJQ?_mdQy`_IC
z9RHM6kWBPV721Wk4@e;6x$r;$*>B$huiA-kdBo9$i!{lcGG8{f-JBe&eF!
zHSet@b46RRb8n6VnM=N|$W0=1z%$*U`}HeX_~q85T_$yF0=`P5^d63Np&r^T8_*)kbVdQ0N*){>W7oL3miPSqwuQ)yGCm?L(
zkjzibq#x-=@=d!q*mP=qIo`~|XG7=nzm73DoFntRc$);rc9jNQ3Z}8h(pZrxppdAc
z^XP#lg~zFbu?uuKD8`3AjI<&9WB*M)L|wr6>E;m&qNhl^f!7UWjJ?zJ1m$9?ur2mFy>d&%9%E{Ws$Cl2yt%xr+pBHmi&b_vKOf
zXrwycki)^xyxA!cgJ6fz6Apj>tw+r-_aK4X5^XYn%KTY+QeLs??EZaXFyl
z#KJYCPbDXc|Av4MVXJ)WNREBp7gxS*0r9=ctfvV?*K4Mvm(-K~%=?pb3L!a^(*CjZ
zf2tH*UmZ}tM|cuvbgeWYo^Y|lOzC7b*r*HmM;BO!TY)mPiGi_}re
ztlyJ(oXkz9q~sYt272y!@4GC>e&bl$iAKVi7xm#vNQ@37wd<&dj>Kn&R?@ioJkX%f6eqHbt3o72yG^H6kASb{}n^Qv-((W&?_2Y8PDbZ
zyv3kBVuQx~OB{bb7lle2sR{VCxh9JveR(ORzibWRflO19T-PX%#^GB@uHhUkEmTdY
zAO>-*QXgI7Nu5kzlR^s$4_}nG6_Nh*7Z?bA6%;hSudO`pEWxG16$|COS(G1N^+(Qf
z9<%*yCR`;xP5ZzM-y=L2$@t-=8kv7qtuRe;{g=lZJ(o+zPHQ*}PQPm=%q6*U%UHCX
z^e6ey-SwqJ2ga&;kE{uQB09NvZzUR$t#9OJe}eK+2i=NMzP%*3
N{trH1^UIUZN*$0iy-ffB

delta 12235
zcmV;+FEr5Sl>zLP0kEb6lR*P6lhgw@0Wp&d1Y`m=F|%?6*cN|&3;+NC0001ZoE2Aj
zIMrJeJsHxZdPOKo85-0p4N^V!sYu3D$e33&7(x+IiC2{LqRb&f10hmG;%3ZLQj#%*
zCUu3(>UZ9Lf9>yn-}#+=_TFo)y>5J#Ov;g$gqrMilzd-7Xr)zw?7>Pxr(Zv5xUq`R
z!u4Mg^f1~9)J=aVVcZNn9UqMSXRPaOs|j6gEX{X$L8$ugis--ao=s5T+=bD1x|v8_
z4WT0+!fGd639a6n?)%Z3P&a#-HeEkLU5>l=%UmY(nWnt+${<3oJoss&5=y9=sxIh7
z656}G`lo(0p+h3_pnH?hpMyf4L5YM8OG`w_CleYUFw=i6`aYp|qMlBAS<2voacGw2
z00Zqw_KQwVA@o#D?5+G+goY|Ss!bLr^nt>rz=P6++O9Nvle&b^w!+G73ziXjTJe5f
z!=Hq1cRxDdzKKu>e(KhtMd*~I*kL1cLV52sG&6p3pdewEo}|Da^~1jPJWU1;q0{E<
z)MpU+DoKA@W;cT$^o&)}L<+NjAv2A|sm)NC9Why#N%A_!!Am)H<*ObXXqsM$t8?L?uiIR^=NJd(>y&*B>^U%$5IB^3
zh=aowUg{DUjS^xs&FwhYA9Ba{tRn|ouR1$L_;Db4bNSm#?i{$T`RI7mm4hQM4KDWM
zJ%)dM23ErOoq@h+OPDPOcD@IF(($`)k)>v%c>eu_XP}l32ljjim9(q%=0NEq886(-XFaE`Z5P{rE}fVB02C9
zO}DuI81v1|kEv_spvNfUP1P?BUJ0|N(-MCSibk4!cgi!U&EHYfCC;G1Vt&Mlj~wtW
zwPvf~T>qHY@$UcgtxA*M%JpLZhBn?49pYf(*bVt4Jm0oVRlrY|L6|f%H@wUsT|05M
zCF0If*3Z`Q2ZP;)Dk@Er5Jwt2V@d`Y6u%pnl}=``q(dNP=sSb&YhHW1WH3-sA8UWO
z8;kGoiBlTK-}wSd_w(BjC)$-`!Ltc1(;{;k))V?i^Q&uPMuf`WXttAcAao_ac+kTQ
zdDIr36LgHwn=SP#+qV-s*(b+%wI&FRve+OR;%yHJS2a#C-2*nn0!KWe9IoxJVTz#Yu}rLzth$iZ!NBUMQh0>9-&s*r)2622|XgD-*vkj`S1Jn$pD_y5xn9%2l>CV
zWX6pBs3-125L3iTUaq@^p`lFF~Py|8i(f5
zM;vVV_WDuISq>Z~-rKA?%z=)osiY<5`yH0ut^WY^qTTGvrYnT%>-nf3w;%(Ou7$!B>CHlj9XQ{~XJlUy(^1WEgkVg!>}?$m$`%Sn+?xYkl@pm;-*9m8Nza+<
zddRy!Ux*7J9^^yBB(!n}EnO4;viv-u5xpAsk8dVaqe9v~xQ4;cE}=YAWd^@@owP~I
z;h-sdp_`B!_IKvCd&C)as6aJpBAJ6#OLh3N(hOdBKXRoodcWX)QN*QPwVn7{qIp6oKFs4tiFW7-omPOj~47_
zuCqJr!uveN&3ceWyxhKE1>C3E%^xRkjYs^L7vE5clRu2&~^Jmz6#?&u)p|4
zAL?f0{lKAU)aevmGr_E44yvtkqIF#uXxaaWmRn5di@1MoBQIYcWbf&NimcaX$x&z8yf`mKz>P?pR3Z
z)izP_5i@@Vy`N*YUP#6Jr1_r;k8n_~-0?jD=kWMIC;t@g(c)9)mFAL!Mz8V9RQ?BJ
zWv{dn&ND+HOtKyEV;R-jD1rRFcFrM28hL9X4&hRD*w0-7lS=V@=1NW7QMfNuWJ*_r
zB5t0Xc9mWBjf1ghyAUoHd40fq{c7y@+Qv3|`W}Cw2XzwCc!--*v#xswVBFd*Vny)%
zZ-Pc-1TqMnXO*qug#KhSV5?d^>dytsJ0gB^gtmKn`1B*5K0JDu*7-LBYmJD-FMo6J
zRbx+e6Yi9~BbJ
z%SV6R2rtq-yw07_S{aiP@4pBY4mK_hm|zg_!e#m0a|{%mB_#}IG3YkT`LGxFy?95`
z)|IFemHDq%eZ{)`5=Qy40qD=k|4Up)Z%S_grEi_U(UWz|)xwQfeQ=mTV3bg%!rF^BBbaS@z5b
zb+T;fXV*&HCuxP5QV|J+#>THrxQumqjq=C)v5(4=H|HrlC3HNgo%aZR-HS7O#2XE9
z&s0VYSBem-v^G@S;|znq(~e;mac=sVsWp>vP8uPr7tcPyU}9-}(+)iMRPwMyIqrWy
zk~`?46plDuER3!Wi^7
zidr9kOQD~MOZ!Z#8H_xe({a*W?4NzH
z(-aFrA5F7R?8Z9Yh_n>#nTI;8H}rB(C$CM+L
zJ~KE_-=nq>{Y8=0xX8bVgZO`68wYxkKNYI3?#@`(w3EYC4^RgpOJ-YiB7a^jip(nO
zW?(m8ZFVE>zbpGKdlsN?%TIg~U5(LjwmoM^A_c=T8{HCzI1E?=l}fE;j=4o?sQ|(
zodp;#7=JFc#s4vp@Ag&arlBsJS3W9ZM(9|ZzTO!6knE4!H+&Xk;HsNY)r3p#?S3B#d7|d$Lt1gwP~AWs8g$
zYbuqLt}P*^hKgtqQrA|wgx}}<^?sf8`+T40obx$!+5bedDnF(p9Ig
zgRg_{1PT7&uugWf6m+IItR2QZA%C%44omv0pSJuAhxIgOQu4%44(p&zS$fhehvkr;
zDfA7u&G-k7T68)a4f7ic%ZJIdlY
zVYX{I_*C;@j28R@`oynCAYLnPMY;_bv_YP24QiZt5U>_plq-v9ngHKeTNV9JTk&#G4Rh&Zrf)Gp3Jh_-T;axd>rJ5
z{fVxk@i63nv=7wN_rmOp7j6Z6gY(w9B2P{Eo6-}IoxiAP(0}z5S;u66tv%-%b)xv%V=-KlRKrWjz(X2uIphJfAH^g~Ia>Z9eZnS$^cM(*K`z9s;s)SkH
zai8O`<~!UH{NbmVovq59=CB&(Wpfw6p5(w2>xX>NmQvnLpesA7pnn~DIi9nfJK*1#
zCVb)+;%}O5uKx@2;0x~52KsJcuh)6VkrJonU187F4{v-6dE%H_?Q_%tV|2ZNH{|~g
zTxuSJKg-(Jg#&ix@9zkK{=z%Shqv*cXCJN~YXtuuQ8}XmJN6@@Ccoth;l&&Be!%$IFWB!V&lMk?T?o#F?*(
zKkWqW(f<3W51nIijyD&4@JE!J8OXLvyOssJ(0H%b2;`2GhpXE_d51)9H}VxYMkT8u
zfBqSFlUc|C#?>loz>>*+)(GMRQ}r(V3rZE&e<(!UmWEB|mw&;|4Ull$MfH7fKT{WW
z5hedlama-!`U7uZKebz8^bF_}w%o}M4EDHN-v}y2thvvD-{vI7KNQz`bJZqZu*5ej
z>?7!TvYmSw_U*3YJsZH*Ws5kD&`*0Z7w3Vul45@dpiX#2Vg{(rqU8G0(xG?aI}=0w
z7b>H#EJgK~Wq(`I27Tm$t2G|rVR>V5+Gl&gp5)cYd(%;?`6XDZ7n0Qv2JL^;_XPIt
zY;VOF!1&o@=KuoGxF2*TMB)>l7YS&cv3i8ZyUIE{-@3&a6~fE
z;5+>1&Y4KB0khc5{(hY2%!8GYt)S-mmEzjd}hzsQVe0M+c-EUSe3P8L?Xc>Pr$euBJuoL#vPNjE0f-(-V
z;WQV1ZPJZ!LcWn#vDLKh7k#_$?*X;lLf`1X&Un3qA&z`Y7xc=Q!cTi`?8O(b2OJU^
zEXBG!Sbr|hG(irt&;A6~SEQx
z`?D?b@H{Iq=tTbLW$N!8Ab)Q)T153cw=e2@IDh=F-eIU1gPEN+sVT_I<3H5m2LGOT
z>DImQll#>&zXRWyo37gy3&!ot%3DGEQSK&piO%uwph7>a&vNslY0B#&DBEQOE=*!-
z=^^isNlP-#nOBt&t`*R~6UyCr58nxm@48HVt)Q#*g68t&bxtKTcNZFkPtdxj4#b*^
zLVxc&ozx-@3h%qZa};_;wC|?HVB*;5Xco-}G3700u;(TSd2FV>W4C16L(k`~cI0=g
zt1C;jmItzadO;iYkH?jUMk_G>+OVA<*!9>)b03&q@I+P!Jm6pX%SQP7tnZiD4@xvd
zhz^0f&Llyl(y&ja=9@03^Y3~m69K)U
zbK^p7P+4MJh3by)^0DD|$YUAn&(r+Qar0`724&03Ut9oVZ|jO^P(QtzoNcCgB7elZ
zLHCPCen_$;&BJBR+O)s^&EjJ;e>aG}y-oM;<9)V|yY|XXy8}vXtj=?mu2>&C)!Y)sbcffzsg}ztB4U2V0gO
z1P3G!o+<)+bzgJo{!SOrJxb^NM|tp2HqGNqtsI)~dZQP^3}OFvE3by?hJV{WnMUWl
z=)Fr%BPiNfl1k_G%)hjR?rlTgV}H{8T3w~%{u=UaU&Xt$-Y?}!l5`&$GxCp9pI(i6
zcT1S+p+f(+SlXZ7Z*_M-8|NH-@(W_@>vsXYc^rp$LA{Zu)s&ZcuWAAPk1eZ|LHF+9
zOyq)v&|AJ=wZ$7u{K`%t?tlKXh<7p7b4c)iR`65Pm_20&Mnx?4GY0o^^r}nIe{t{g
zim48(+Qi-Io^MeXDWv`Qy|&GP>Q&9GC1wbW3%q7O2>$4Hb4a24^`^J&N$@`VJ0p|o
zZ+oFx0T}sHr-k~rj1zZX9CPB~YCWSzki{xpg|%t@84Cp|?~J=fLw^-`uXo;Hy?4&nSP=0CBh9#!iY)kk2ztMnvl1?I!ol=hbz^sb4I-C+z$xO#qKz%Dkf
zH%0$zsZ^4p0)DQkd;1gdS@+X@=zXwJ|E6Os@``_I4{QeSHvOo|fIeYFjG+qul<_44
zbWgCl>!r)7Za;sMxPOQIMf0hFx{%GKRLZG74+S!tsNd(9DMvpek8eb8trMMd_U$FR
z=pAr?>GKw>(Cb&O1=qD&Uz~y6fva7;lJ5V>Hp61HcnNvGe*CTFGKvWaGlXI&qYzpbVwfyzUaZu;*
z_YCSI=|W#Fy&J4GY(utyYv)%Uqkdr2L~2u=@Za;>J4k)sI>tlu>EEXlK~%TZH?l9$
zTrsTSFQa>3(|^Uz=p5{vq=6IW^uCmzQG1AWvRCPe(R=EaNlpp%Rpn_>zK!@Vg)pPY
zOwe6nsB|s3sW()I_V>`6iYu$&w?a$WxgBvC%SW2E>AmCpq+uB-mzI*|35ren8~VY&
zDN9%NF#MgH;$5GDsnM=RRHtWmo_k5}!RxILx+~$&9)EqdBoqGa{$GyLTpc#J5dVq(
zcS$#!=H+uEJ0I%TtS6J7=-tRYJDm0py(c1q9@G3|?-~1OOYbT^H63g0x23z;n#Z`O
z6Eb!Cr$7Ojl{@J@XjS%?zXtNT8Lc)f0lU)#R06RNO9UJJ-@{MVGQ;gS;-&7SUa+He
z?|GFOISe)pi5bm;dq3Bu<$?bP5wQrwl0F!J_zD0300000c$^)V
zc|4U{7sr(hkw`+qF-vq685>Tb2p1iN>Z-_;R5Fw)GZDA0p^}73#*(Q>q=*obTOp!^
zC_;2e$@^RHU*FFf_S$=|z4uzrV_24^^z!!@^Lpu?1MR=Ym}1Y5+NjQrF)ikW=B|Qa
zn>j{oVXx6K>^VAr4|=9du3#Hys%HI;T@Lmiw7q@<`VCqYAN|0F8BI4X=siV_GW{V(
zrQK=OhWzWEwpAABe_?Uzy9MLS?pp5~ybH&f%k;Gm?1Ieygum(l?D9jUya52d%B)Ccf!A5jhu2e
z=n<~Cv8P;*+uMUMdT;-L{NHI*s~6H
z?LoCuG*8N-it>KQf3yi5e~dilYL03zfSf;n9v~P4E_pZG=1B9|%jHIbjDITMj?+Du
zak}Dz_;QuX@ARSnsl^EAg}i2!)!+1*`x8HDd$+TT}6IUWn)2zgU;n
z5i~9OS&H^~_lC}EaA2F;@@wD}N3eVr{8=S;g@-WywTJb^e8_{D8I$y$?Ip&7XxyWM
zYHc;xzo;7pzrp$a=NuW8iume5{Dtv<*oTL=Gb0>%H!`}p?GeV?o3JrAK)*mNT%sI)
zHM#j)u3^03Kxnx;?mUzs!=qP#DWx>BVBKNr3v6DHLelNY=1BF-Q37sk5b
zmmrejJOo+cnSRwm&@bi8ZQ57CzY$9mF^(d5_0llkDd!s<7u3=U*IA3(*fiU7snM~Qz`>*F;
z+V>0Gv(;JYINgsQ&d1Z>m!j0Q(G29Z(OQxS{m84ZFL{XHsqlo4>gp!Ln9f(wFO_vV
zvkdZ1Kgrd}@V|ATqK4`-S8+{$$zjODVovpR?)sN%TIgJt=NfZxg6ZL2v%c_q?xEs(
z7YsS!5JK;u%<|l^cCgjGSgH@KQaa702+r9k{l^RT)}H%Oi@0iG?>%Q|d}_`zNz8w*
z-BU4%-jR`-7v-xosLE^#enfbMW
z?v^-krLGwx0{rN+*B}b^G55+h)l|QCg$Pi6PaaI{?EqhO)Sh<%6&4otE`oho=M3Lx
z@MDtn1L_;>(@{?%A-A8l&`U)erNNhvXdm55Dv6D>A7{3mKA_L|kn%~`Ejl!o9R&4k
zht!;)-<-7JcRtlGp9t%JZ0L_!Gb_#^Uft1I>mcZPWcto*qV@e-@RZ^kn7qiIg*{-?
zkfS*4>{)-0xP##zdUwizQo%{O$?)SW?$H?p*8~l@nm}LKXK-1J`r@gS!iN~I)D&Co
z0ei$rb;A_s)o!mGruSsfXqiCuCDP5bf%<1miJ=Fb$Ip->mvm@#3&6*sH+?B@5_6XluRy*%Z&kMxbP+vWCW1V9Txi;236|Ravr+<-SmNqw4gZF3
zOAIEd-}QT)=Yt$@r`dKHaZBYC#aG;7_UbU0q%)Vt1N}*U{p@L?wJ1sC>0-9|z?<
zPxtH_+Q)E<*D3gMm8>$A1dI4f8?RE|8Dv}d0rnk5#VMkY9k*pZrhLQ(o#f*NBX8;}
z6vE!0)hW~sW{Gp<3qj9qab_{yznvk%jaFcCM%y>aXI|)kn$~r+|ItZl`kqPFjfkT8
zRIe|clc2g-e_(V1{!M(7nRM=MIZyXk)4kDKck21N>2$Ws0t>t%DsaIOr-
z@y)W--+xxQ^nk*Td>bo3F}1V_>dy+!?azKdX21Mx*Lg6*prp|t+ibpcy9O31~4
zLR%bfI)ep&T)Z-LUszhv61zapwYPX_9gKZE=jpryN3Xk;Q+(F%hB>gsa+3wsuPnt?
zqyK{2ResBMfg44>$eso(;++gm(ftfG3R;5kFC!&YPl6dL(SuYE%IXD!|Dk?q>(E7Y
zamW6=(s%Ihs&TD1p#68s%+tMPU6O93JYHxkjba0T7YBa&PJQW2tEo2KbB8H*0qUQd
zrP}B05NFejSZ@kw;hcS^7@RD6%_yY4X8$>01mmvh*)ys@jvJ2YcHqU?hn(v4Jt@oc
zo$5}{Wz!~K>X!;m{`9@Z%5<>LqkJ;IeNAuHWnY*5YQWq0n(|2Q|jCdk_FFGA6eaK1YHQ#cM`rS27UK2Vm
zt>m>F^xisj+cnHUf$sbNf2TH9mWJZ+`<}yp+xrOdBhJpQcA)Rf33uZ@tbZnXku^Wn
zhYOdr={&!16sC2>i{I|^p!d+_tw;T!e7*0`7OL|u
ztB=@#M#{E58KB_;ucmL*Cn93yDL-eehM&^`(AnB!u(Sqk{Z
z>ZeccP9{+PW6!?zssKav9h#hRo--aT!?WO(x15W#(6>jwrXQy7#&Dz41Kt=nndP_c
z9{igtN)J;$mW~y*vLimXz*D;l>_=0oGc^&id8BLfTY49!Z>FjKuJ8@%eNW#Ttiq_D
z;O&tIJaq2!K2Ce+Jr^Id*W?DzH*cf=hiJbi13N%9p>%F~r{jF;%rNlt
z10E~tKfV{@o#=Zg@aJx!mGIMK9PAImIyN0Lx}t^rAD^~8Pz~BmO(gDyy(l~V00000004NL6_G#ZE@8^6z
z=Y8(ydG71_UiXnY`a7m%2)MH}&S>PYfY1KUKl0)PT;6)@y9l^uv_>|7ei9A}YVjQV6
zJc_QPIvv_hewatFdV*n?Duagy#zZFXXVCFs-^s?i
z45n)OhV)Pj{tX*l|9KLNvw1<^3}>+@vVIk?+>S+M!o9?@^(=DlIL`U$FpEdi43o-4
zEJQXAA=&*bKIw$YB(LSLKGQK~+8i3=o$t97r_%V_uJ~H%a~dBeM1NU5`%1%ly#2Y$
zW2AUITH@@hEJa;eW7v%8QbhQ6g}4|=@$778
zeJ+FL*RE`}T}**mPPNXCrcm79cyQ?@3hwvCXTn7kG!9y1PAQ|XOliw%hpQC4`?|Jl
zETAx9El;=PP&oD`yMItWmqMekO`az2wU%wS(0orJzIXI+mo<&AhQYEm5((tC1dHN_
zB)CyI%`icp!mp~V!`R6bzTGuBrSwCBg2+pf@_q>r9_a8tL4pD8x1%>`2|o5IEUf90
zKv#Ms){i`wC#D`U|0aQ_s_}_EZ@XDGPZ
z_wf&;6c$Fl(|>ogqVdMbQMxLI#wMyc-6Ds^^*@7rPVEktg%jv+<$Hc>9NCk(M?jg-4H!Zyds6e
zoK*WoVkzb|wYCqfWbj^e;G8+fAW}hl!n<=U@?A1aLzi->Y4-nVsD#5+Ti1}g#7Ac8
zWopXh9ClSdu>YaS;mxRxhqA!o)Qvti&66iXr$g`T0jL`)S+#P4V
zy|Z~dEPwg6A*Y0gj^!vbpddiQspjg}6#|5|n~%nN3n*&Tt2t*Yz}e+b#i{@BFpSlS
zn)$1MPRq?%13Lul7WdWcUoN1~uxS5oHvzf5m*bN|1+>&UTezPhyl5#mT}F7~abnga
zgF^yp6hl=-{sLxL*ovz+3h2u1<{AzN7(8vs?|&q`D1L3O-k&bOA<5;;^}PbjW^IoP
z{LbV4P=0Wq1CN}5b)6G@Ib1k+;^kZ?7QMtsON|+*=ap|=LwwelnbI6(LE}hUL{3&Z
zh4vZYI?PlG6Xl!!$|1T_s2g*jiqzv<$KUIiEP&aYos_kf$H>{4^2&i6@=YcQ8+=$e
zO@HfdzX%427D_>^i4?v6M^rc375XWIl(3GA#uUM#5?7b~p&cHLHyWmlV6tBYbi?|asHjBDE
zyh@{iF(L$lpZ@|b~kAmuiucsA>&+o-;s1MeqFm>K}Uaf;dQ?_`j
z8{y-wV?hJznF7AaB^kyVkiP$Xd4{$gj})6p*6XWSWQh7xf@d;#^WFPjQ6&xUSE9A^
z&Qb95-D!E4a3y=bV(CZ1jeA}(qJR1W6mE1BOxTo2gEcwx>3T2ee^E{Sc%mb5P59h)
zqFn^H0v1|*sk%jS$$y}y$?w`2
z4vzWRNw>-VpNe@A8%{%`cCE`NM+ywLr_7w_Z+Gd&Qez(q)!{bgt4`4{*B`lh+Kxfo
zgEQL`v^kWHyZT&fm4JgI8x}Ps2so+tF({yvM_*Hdcflx&<}HuXcF8hG9Yn)J!n2~&
zA4cZPpwQ5?FCw4lFuiPFw149*3R&(APV&p8SU4p8aNLqbfYw;wYIWkH+N1SW#NRdX
z4cQ-$36O0Fm@_Mb#~xe5!iHKFt>UH`)P^JfVTt{i`;B80+H=gEgO`)T-&Op;MiV$o*w
zaDmx0vcK4C7m7Ct=vcOCXV8Chgxal!>=7OT22Be+)^S+f>za4Ygu%Hn&1E)AXoxk`
zOgfY)NP~~qtdk{v{eO~j`Y_S=haB5L*B&Wk+~*(hBlo@bv`6{4w`AXR&VngqPBESw
z(YA>Emdoc$Tje(_
zmC=`?$JUO2P41^$bMW$o1PX1F^|dTiD6H%~ouyCCsr9qc)qmeQcnl4`Y&yS<%u83A
zhj)>Bsr0hKAs+$$gU811s^bxon3`?U!GRoPAh(XiruE|1f6^IbW{hpk+a^Wwo`Q|V
z{xrr}%vNMhQ}EQ#dGcrqh316bh6_PEvREaf4CN25hulgX4Q!*fg}&J#L-o0
z6fQ5)b*ld+g~|&B@l0=tuS*fFSjKAOkykMy4pxeE|oSehb7^zes=r-HJzfVnTjdxCsO
zo~^x9zDAzpW1Fhi`ZNZ4OXiFlA^A|PT9KzRhJwH1^H1S7D2%$Xj>E5Mc(tuvQ|rrM
z&)N$G4u3q0LcjL2^H=kDl7BYuGX&fUUg=#ya;zyP^4hj}0_;?DpGOm2yG%^XzeW0E
z@He+2kmOK&TW7%^suZkV9Z_#2JP9|tT@W2bLAl*UT_ul3L9gZ_&X9qjzjDaTRu&T*
z3hb)fc=$KIpwgTOw|v>5-KqcOqg~5Hl27yFqkr{}zvZCyB&BX}1&i<{0ScZ=Y1CD#
zN7yW+uzb_C4I>oEA9?-x+(rsztMd=Y#M9`|-)!AY>ZJR)`25DQ&~?=M?;{)m8*!)xl#6TJm$kLc%lk-4DcNFhh~=tOh52K#BOtZuz2=R)}M?Zf7Rt$hb_X($xR$TL0+mbhkp>@eeCg7gz-t9ktP;?@o&@_q25#Ef<)
z95yH#&`e&&<9u|Y`b`fG?>zgryKi8Sk-d7QB$>u?@o=&`sdH%G^q@LYN4dT*?VtS=
zJVfc48(-0gNotk*`#u9cWWC1RVh*Ecf`4R<)C3%_yP3j~zSPU;FIi1^AX^(N*D=JS
zvj2Xp^9~M+%~Xx4jSLPp$#gnJkvhrV#yMsb{(5ok*%{KGzS9OmcL@cfA8Sf;9i@1X
zvuwWHIu^CLEC0$^%0t`7dQ2tp>Gn@de;Kt3@aB`sN*xzn<1+vN

diff --git a/previews/PR1624/tutorials/out/solution_000100.h5 b/previews/PR1624/tutorials/out/solution_000100.h5
index b9ee84c65e52bbc1aca3eb550c5fe597ec764567..fce83c79f2ceb2cd8de20f1e7fe0670bcce96800 100644
GIT binary patch
literal 11976
zcmeHsX&{x|*Y{oi
z5^->bC@RD`&$^2HzTfw~-{<-8+|U2R`~J@dzw_H`*lVx7_S*Z@DI{XLvKT@3;~T@M+|t7_=Oob&T@{+{30$1}vm)$MO1m`@fs|I?0R
zE^f!5eTV)&aMoXu{{Hi)9nZX-tEa1%t9QWP+s++t&|UY-0Sh*_UN2XFcgKJbU)Nvj
z6$0V>dab?v0}cf`2Lv8+buj}iSmj^i{j2?c56|EB9DlJVJgd5|{x#iS=4Hbc5_t~r
z@N&fjrO+SSqs4=1F(AzO>2p%ZyaF}TXx@H4zbwD)T>}17>&(b-@#&Hz^k<&9e&>mp
z*FV73_is1G<@aR27Qyj5lDTk*zDazTn0^n8|6&+6k1TV2_3u_hsIn1j24rSt{^fvk
z&L0cYTmrCkB6oj$gyx0`<*BnDT^C@Azho@GJQ4N15AsKpD6zA@W5TdQ?mmBbF#6-$
zKk;S$+*t_z_J;p&PKj!mkAe`*>(5;;bEoW2BPQ%$mfyLn?!U^f)UV+F&-@bYVt!>o
z>i=9P`!g4R{q{ckgzb}j3?H$ZsL#5C;kmqQnfkdH-lVtRryO|Q&NLa>JPa3BeSxKb
zp4sq#u*^LSpXQi(rU1AhzpVaZ0fuMQGtFkUVtC3%BLiMQ<$U{*X+WDFh2E4F4EMY5
zDj*I#N9VbaDDVw+DPx_D819gJH~&NfhS#`n-&OGp!;f2rXC2;;;XN`|WRfR_$3`yD
z7dZlPgec}z#bUV8o8$2u=P-QPO*t0&B@AadzQi}{CWdpWKTealkKx815#wV;7|vIy
zEta{9fy5f)R~LpdkUU?G(5gEO#8!K@EV7J&q_afqvnv_sv%L@(UljxS@bLQtyl0?~
zqn#_q*)Tk_{){$x35FlVJ(roq(-CKka+~XAI+Bh5=6K;T9Vy!=d=k7vM?OB2dU!G&
z{kVV4XjvE?H8#olX-3kKN?mfl_Zd3UYLDMBd6$laLwD30hSHH&`R4kSL39+VSggb1
zPe;zdVJ#{ibfg#HuyY`ej+PI_eRq3GN8c~Jvy|4pe~FIZrBxJ(Vh9RuwIj)1@qfF%b;6UI9^K2Ge
z$8gc_%nKhvJxKI(C;ME$@KTM6sztCb&7oV@brAbsn%<~)8N(NAh+K(2j^W!>E=zUB
zLESj$1s{QZ@tb-o`=AcPEw61q-wOM^n4_;Fu>!-ZB-1vEWMFu3$G1Jvi5M<^+jJ2I
zb-?WvWO5SfBfa$2qV=$!p)NWa+r2Q{VDfuAWsHHwUpA_7sWVVXoQi)&JstJ>)@RLL
zr=#&uC)3ta4A*1k?h2{L@aPz?#PiJ<&NW#6X(|iDOXQ_Qqs%azUqbj>O%VePrecQt
zvvd?x)Y`0n1@bHFzeG8Tjto{LzTVb^;nKm)!AAji@Ooeut1w)3t9|v|6BvGSMVDY8
z)CaTsCC$^?48)TuolnW7BazJfC1ZYc^vcNYrMfE}?W*HLju40L?&X8^;AineiQ7@I
zZZ?yat2(|I-X|)wRi~SQ_P&$2_oS1KPPwGml)2GS9@p86d=YeXPPbVu>>M4@GsY+r
zkVmfP%Yts#V0hV;;=?wO#}yvaMxFi`9(7*+P|rmM(hN4P2u-D<$DKZ6#~}_YHTBc!
z{SepFHF9h!3{-vVOjbVZOTWEhDjm-A#OsYZMQ|=3Bu3rr3Wc`0ibU?)&p;vHd&>Jf
z=_rPGlXoBFcTGyQy#ECTatvR1;Rl?L>04vByCN}MCa7C*c^>dO&6Jmr=gI!1)(S2d
z9&kMEMwTc8?K&u{arXcnO<4JmJ{iHj?HQ8|Sc&0+E^9jPyJEOhW56OssGAs0SAM-G
zu>U1{-nu}Yq}9DL30jBYKZewAv^CJtRZWG1GYNDw`KjDE^BDtO8Qarz`yhtPo829n
zfLW^SGGdilhmvKKdla3rq-&<8jLmy2iao525>g2csW2!N{
zH9j`+Jk+({0X1ETvlxD@u$U!Y4#U|avNfcRGEgL^A-BpT9c>fa9cc-9*LiK#!UAy~
zO*P%0uYlq8L4oe!0T}+#!QFcV;&b6>udu1baBuCv+;_oH_wK3ffjk&quD4e%+K+*R
z7DqJl(H+-5(~k($BtTh=iQv@>V6p-msdaV4we_a4G|t5v_Z1nTwj
zBf}TMP!HT{r#uhI!ufLu_dw7$8qTmr%ES7)#DZH|4$~1WVAiY1myQxWeGjL>e)oKs
z_P(u-;ridC9Iicp{$HIMCIb6u%xdK#sK`tNO`4#%_$sVoK
zt7vqzWxF%w>wY>C&z-EL2r!VJ)P?$8h8XU8U(_-c)*H!V_u@X(y@p&`xpOgKS3r>j
z?2A?LW2XLO3?K8Wv~QNe@FEUL-1{>fB|ROPYKJ)OLy9#VPcRTeQX$jI2g4(FPmX*o
z!|>%Xb{llT-yDv6#jepB;Kg=?^Iv-Zt>gQKm7Oa^Q9ikHaE}T%e9i`em4BW(fr%p
zzccXf4E#F-|LoasY=(VyDvG(kTrBJP0?;6UrGi{*f=QB^Z}pg(hi
zq1yz!!fu~PC2;LeUt(GTH(NiXSi<oIGf
z%T7RCcQ_V%UxD!!7ydA9ugI>5bNL{7PEW?c;(i?D@0tc*VRUI0p}y}
zRNDgQ6H4JS3z&BhMSRo4BKlJ=3Gl)B9f11`Ch(pC@
zp|B>PZ__|4u`Xu)rs8XWMwg`zodEpVFrGo!r41fp4+TE9kf+2K(00plY>Z(OH9@Mr-?QV_)7>Z20+3Alc<)`|$2H_$|TtqRyHXNmremVhi
zVX^e3W)R2nTMa3z0i&z-GG7J#l=BV37EmYBddp`$S;o*0TaIxV(D&AFxiARp-8+~d
ztN`+ensVutz^zBl8v29ZRrFWm>tOzYV1CJOp#QU@dq^C{d1h3fjDcS0?)d3g;JhC9
za-v|K@bp~Hv%s517tUS={k?bdDz-x$y^Tg;d@#;FV>RUlC`wu(oDL`%|MsE+*fB;L
z54;2YKIy_cihu={wyX|-!Y8+#vH|}&{gS>HU*JVKGMX$d6I<
z=2Msp9rV}l8W9%)`+=aVhl#xI_)0!Y2mP%
z+E=PVfEF$m+t&cfuksig2DwSC_e&!{9}`yD4}ea39;$L67cVY2yc{rXzxl&j!q3nO
zaXa9ba!=+>0_T(d;F|y_@7Xo664sL!!lgC^`@~~q|3w1$Ib@Q+0t
z^|HX5gLoSVzPD?Ih%j&!xyBw6V9_Pk2Xw&mZI?P4VBbTFzkMP4-MDG?T5&*|1WuN>
zV5cG`VfPjA*jJHFY@k1z@wnp)|vWVr}-3&i^~T@DuTW=TjSREz$Im`R-Oay_AL1uQRl60FBT?&-%MTm
z1|lDuf*SH=z;EEABDHG3%fqX05qU3VQ9DM|&DR9U7$v}#yi09Fyg2R8I1#=?54@?$$r16PW(O*IESmlU$*ckr{P-bNt^^tH}*
zWXS<iz
zX6<%;dK9hAxF7BZ;{(&nyEjAp?Fx^URf2tixm;Qp;PsxF)qa38?>0%i0C`M|&KCPqfO%be6d%HP|Bmc;rC>jFsqZ=sa6_G(
zCs8NM4n1Mh2f4`jvWrtNZ_{@9snsx_|9idt-N2n34;6SppJ#0-7;^)EmM%IJe$c-{
zee#SrzmzHSu5s{VrpsBy1$;O$r7#0{dH4O2G2l->tWkG^b@e8>udf0>n+&+V4uK!8
z(={g^0rGlWR@w-9i{*AKBKH4VEY$@g@_5rC<|Ob_0$i_-gTJkcq!$6OKZ^1l!L{J0
z<;zSy739xVloyJEpIpaEO%^~&!+q{V-@ixt&LRwUqPoX)+h85(H$39Mfc{EyXVDDA
z^Rqojv=!tTJ9Zs)0R3yWnH7m3XF2*_L=VOHohM6HKVdy;xhOt;;(+t2gu(&
zo=WuJp^E_$#C;N{VNkaW?BoQjg~P$`#rRV7hcLgltjsGSue~v(6YF7qHIpt@4zP>M
zs*)tuF=TTvpa}SnGwcS0Ki9;Cad*J3V{%)sF7Qp)vp6cjF1M+Ej~L9ue8zBT6<~0{
z7*7j~2b?gwiWBp(+Uu(Ts{5Vo5Cr)R#V?b@J8oHGkqr^wrmgYSL_bYlv`xDn{2TU6
zGFO2AZ&}G=nSlHzqKn;#{Jf1MFQdwT*rz91K+&$^X*vB`yg{r
zjuZGZ(LFOsfSschOX7XGZA3(rkXJrzmmLK@Uel|t4E`DmvJGBAd^#cQ)6F2iF~Yla
z077xhV17+giX>mUnp%z)th?fKnjnF>dto
zdjVfAa`fOO@FR1sMUc4vPhRoxGy-Jh8xvUu`@2b2l&=M_ER}jz1o}l=$ioNbV3%WR
zbt|UF4tQ?$tToXOKh{2E4FZ38#~M}DVf}ZfKXY*dXYd=Xs0O*3Sl62z
z@T){x-M$#sCC}ZSxO0k%hEMMhEPF#mpSH%>dUa5d=)2-O+#jh(UB~pzCNdQT*XOQy
zJwZkFBaGz{vsAQ~ZAQ6-nTEbpRD2%*jJn(OIO!Y>)s&Co-8+geHm(~#?qM>WNVK<^}_t09hpri#u`
z>-*UZBZqvk>wNXY$XvtWMEmS8s&lOFU%@njx~!awudt3F
zJHvYpU$%^(s6_P??vfGY-TG3=l#7fK*Sp`3R3)L^Yp+!2Sdx$dhq*?20tx*vx`S~<
zlF-B``xgg865{P$b8SY1ggoV`k#F)
za(@h>Vhf4$<>Q0sRA%{>2ed)7p8NWNOKV7|D?xos^Dzmr2y+|nr;w1+HSCC95ed03
z$gk=-MnVb~*KR4|9!5gF7A5Cz4WZ2s{I@9C4WXQw>;toTL+FZq+?kHnVKi%8Et)1w
zMn`6B3mM^Ll-GSXn|U!A)opVu8_pg<`QkL@o1qk>`mUw=%?~o#`18#v>V7hs6dQWr
z{0Q_-Ti;i8QqZ7ej=-Zzr~_10*h!Z+r6`$M--^YIiS^zX@-iv7s+d%
zyi7%xw${j43R97*l?a(bhl&*Ww5OgisOWxVoyjRb8j>FVR=o8n4f#b>+^#jIA$qdd
z$35Sv=(R;P=P{_mk{KJa15QJ@O;Wut)Cs$2{`7VQ8p^x0(0=z73i8oo@Q;O$AiP>l
zO1NYQu?%d^P^loH8!i`LTy!I$;NjUwwpq5wwKC-KzJAgnG0Fvnwfs
zDDf$`=1YDO%5ov~av6}&*6&}ZwlLAqqD5D^9QV?YXp`B7v=AD)pVF@Oq>F-@w-$vb
zQL|!Xz6B3f}JL?xmA4GR$?_~#0laTQlgA~@>A;hU5p6~(oCD#AQHqUV?!e+l0
z_a)KL8n4FJ{8S3Ma3NN;S#204d%bA7^<@y@$E$ChDkq`PCE`(<*>H}w+$rZM9zh#N
z+3W8uqM|PgHkh9opdxJnza7@`G;~C|Ctvm`1wHW#d+!@Og#0A*-rpw=qLauyTw!CsS^5y@()6AiI!y{s~wMMlYWwI!D;
zNl3H7|JB(IB-AH4a@>G_1Q{CW(iL+l=$nGJ9d8Eot2Td32BRV8gEcwV18InB0c*3f
zG!2ms_PTB4BqO^U8iO{lzXQ+2rdLIg;NK$z_Aup)AiB7-`h*e{Wu3j-x9uktjUDGY
zz(drt!IE35`ZN@?Bji{wv(hoSvkZcxji02_xP6;
z^g*6NCRZ+~fjZ0)I`YmC{E95w6wVFnlTIbadfL)Z7G=SLD5!t)?f|vAJydk#!L?O+
zreu`%xgpHlb{GkW#^)6(4I$~Pdu-`Y*X{!McnWTfphfWyN0!@xUTM2`{7VWl;(GP$
z^JX|7LKW-m4?-XI&ULP)P|)`Di{5PABj~_KQBzKf@xK5JMn7A>wsTd6%%{(8t
z=K!3eJe`(UJ1R2vS@%{1^4DLwZ1o5S8Ktg_7noofK?$3OrHTcI(3$!Y8uQ>FN+%0W
z;yy!2B^TM1;AD*n)PgCw9R>i(KzJqn?i4a
eJlr33Od5&;u)nP#ez*tJOJB{SeXFLa=sy4i$OE1L

literal 11976
zcmeHsX&{x|*Y{=0lvzcDWNJW$MrvP~lZqlUB|_#gL?uHqL_#VmQBs)-nIjQ06rowf
z!EsI$72=#{UB!Lh@B7~G^L%*j=l|h-|L23>`R%>d+H0-7_S*Z@Sv@TsE{^3K81Z3a
z!|b`7aFf7&VW#8GnCaz=SYjPH29dciz4P)9(SNc{$^K
z+SNR!IZzBtK
z7h~&x4KQJgu*Db~#^7%6?B;{5QtbfrbEGsx|__f?$)@8v2i9Gu{
zyE$Nd66lZZQRT!`>EPzV^4UpbF5bEsw4lF`U#8#s_P+nA_La~O(V4OY^k<&ff9Hu<
z*T>hv^KUoC{`X?Pf?)ma$-Fy6-y}W^48I2^elZMNK&JV=`gai#qAWzrfQ*cczYMU=
z+hbywPXLxmvmC3g0=PZ;LyQEgt1z_?YEnL*`u!zs<#xp_PN-Et*F>DuLG=NRgDv#c&>l7g!R=
z=?@PGNETo?Birm#SwP*qiiXSi7@pn0FqhGa;fdSyw7CEk^Q=Z@0Qdj6?>^ds;ogM~
zyrRH!)t~bV1K(1gG~U&O;U+ov^1>T2yuxXpY1LB<4>tXayZfW(rwlRnyUTD}HwF8andit5_EISgtv8eX#CMm5JUpf}@kAP8
zx_@17MKBG$X_od<38kT3^@;uN=V(a1J!b#ZJsJ`U+TWlPL_@)qI~vyb)6f;UVs$1T
z8VU#qZjpBed0*=T1JNKi6#d=tF%6AfdS@i{mWCp|M42UOXlP3}Uvo2p`(yhBHikx9Z$Aq;0-taGeF5yMS=bMHF<
zk5DLl{d9tkw3|{HB$xG2697yABOLB$u(Pi
z1H&c1GYXbKJuL6%Nc6ab;f=~wHA`S$9tZ8(+zI=CzdW@`GabX(l!dNFoW^i-`E-e{
zXjtD`GvGMvi{TP
z>VVnJU;hl$M`(HWk}a^G$@c2X``j?xYU+FY=r|osbTlclE74JNw7gGR1LWPaA#?5q
z4PnRa3|h-Ed^;mYcVGjC`$W3MUwj62GEn(xIupb5Wh8{dc0)X30^jP2=!iE3)8Uz;
zq0>dJ&$O=6P?D6-vUOoJWUxB^)!t?dUmx%+-~`}aE@$j=4Tf*nWmS7G9K)kmck}r{
zeatytQHfQfBbkhqd80Wrv?3#K*|--C4eD8TC^^v3j(Ton3w|K;Rf7#+N3ulhP8h^Z
z+u3qW-4nyV3G?q#@1Y~pcVY#Px@gGEKFO@Yk%r>g&tK*~MMK#d&!mGd(9oB(@zF`h
z<8=E9|2uUU{_tw?F*C@cob!xcmk*q~i!$DCFVoSHfSpxA$u!j8
zbb)KTdmrRiDXCV*=Mo)Rg$Q2y0q291J$|P<6yovk;aimp`_-V5)B$;(>|bsoYY+Q<
zI^|}jFw~igl=8i!G(-E3>5p&Augm)STPqJ65|nemj4vd3+j@3>&N*6w>l)6fSB
zKgd3chJJKNAx@(i>gVddPbSMtp^qIdt)s#|)u%>ICqsRl?-Y2lESH9gc4rFhV#4s3
zWfiQ}kr=*8Ibw_#>Uaf3fp0fpruWB!N3c&{rs9HTbqeC%;J?Cg?h_4d)PA0A5=lpPx91w$v@m?*TH{p(CD3o5C>1P&dd+&M
z^FjdXK}hketG5(}Cs~I$Bj_9L=a@rfAiftO0WB@ZXo%`N=T_tieaY4HSPJa-rw=pk
zca&gXze!kMe}Lg?waLLku%Fu(>8pA`y!UUO^K&VN8_R@reTDve_N{8{
zdMXVW?K?dB^#~0u&zX8M%1cMy5|zg#+{@^CSRD|h=A
ziNU^@7nd^hCt~=xSGCnM2@J1d702B_)6m(+Bh&5Rr$b<|vTZmW&5Fxr7<)ke&8J4b
zR=|EoT5i<_Gsi7q;KKZ-WDmdtT}@NU
zAa~qY!8aN7RvnDfQU%;ODl`yJA46sgy|FiFW5_x7$qEfXr#jNhc9`hf@FU~Bf}O2C$|n%P{CpT0@g
zXa-(xc}S=l_|{KfB3l8GNy(@YjJrg(rI!I-SNT$65BB(dVoWyyw{Nh!8wK$gH`1gg
z!LK~lrS4Z@yxE>7SPk@hq6MbOAn(g17W9&y1NDodUF|TB?0gPEX!
z3ZO^xKr0a!O|!Z9I-q{~O7Cz$o~DU3LN9I5dr=VZAwkYEPe7e1yG|w;KkvmKwg*t0
z^11#q;OVB#vIGy~wI%t3|4%*SgFXRQd#1Yj6s+s7ueM1Z^kve<+YA5;7s-7Jhj}@X
z)aARu5Ap2Aqz!;)Y7R191NqF0jRJ;HCz~}_&ABp-qgo5r2}zI-(rURh2=N{qj1!QB
z`QWvYLW}<)#Iv3my
zgLO`)=CGXy{(MYuE*<0#-pj4p2Y!5L(hKH>arRl`X-7a2(rSTJK(?4Smt{eZh1_)X
z9mw~rynj~?FyF|6*&1-^nZ0MtK;L`kCyyHN!z?3tHV@9n+s)ovO92lrb3NGsaylNZ
zhl%}|+%|3+3*6X@@1P>+Rj(H-FoSitD7iVWKwhh9tri*ht3=l_X|N|LN>3d5F@}ae
z1>4g=Ue$C&lpplp`(Hao
zZe9dDd)D(6QU7!dkI|2$Ri$8g(
zzz=9(Z@6zGpxk=rp<$Rec;eon2k5BJEcF4cnSmA6{lgJxQ(s8tc3B{G!3qGAkWJ1wYdv;ZxL&Q
z6!3O`u1118bgvc?0Ink4^p*rza)tQ;4e(;ymCi=k_rT(BUxeg!=FRcJd4$p1_$t+@=C)jYJ04u0f&yj@cZ_0y4+cxwyD*_BMW6Y;fZ>^$=Y
zFe^5RQ3B+9rEWj&2VN0v8us>^bOWEmPM8mP{iSy~
z*!9rwiC+(KroUHzb{580$oPiJfxHAu)2{cx#ig!QUjTmOY2r7c&TAcC2qu8tObx3>
zA|HDGjd_w_H{fBBVl80C@P=$6@1;zNCyBcG5+@$H7O*k*N*ht{H-ZN`KES%#0Vynk
zFn+JhK7j%7>h<95nV@I%R0`dO`MI_T`6L)`J#Mw=D(GE){Ne+#UhmdTidzBSQ-8J#
zz;V5eWgo7n2L0SU(ka1!H{Q-}@B*a2+b;G3=1)h8Upxx-
zw0ZS1R{u7!{2c6Qzi?C{`p{;7+MGD(3C;NIkpVlR{l(jf
z{*Y@hFui7a4B7hlaz6sT|C>zbIj~oKpvCGe;Jt1exe^%f-=Foa9P~%8^xdEWZmpMg
zCF*3A_ahc9m|r@v;_@`C`*@$s^afbZ<-O*S9^i*;z4P6m&oegWk2`|Bo%ZUZJRqM*
zdGwSxzhl#Tx+lQSUJbSycHnQ~lkTShFYPHT8wXzhVWW~G#Pun`X-f^**{;q0bqMUR
z#ny#CB+xm1?KY4Tl-|FD*#D_0iaiGFQ*IeXo&g@s%l`5-*wd0Dz3_$oStZjM@C5AC
ze3{LozIzeHulVe-0!UQOwbI6J4`{ktXOah=MsBu!G4dS?Z(>dk~$Sab&ie|x|
zxpsfyR+vxSZ+gNS$Q---7^nlOI-aZEMQ57TKT)CoX65lK7ZJXwgE|2h{x
zMk>>Ychd!Ni@pnBzm(addLzu==cbx+fW0Pzwc~2Q_e89_Mf79gvbM94FfOaQn%NzA
z|MM)L9GFjuvy3I`?3&2O{N;dL`z6jffPWWtwpb(rcc?8jCi?U9b9XmNfPA;{4MU=j
z+*L804Tn6KF4I-N1mj(sezvwj-jcN(pYy|g?ONS7u?6x)mp2zn2Rju!wGa9LS$igu
ziT*ou*;kCXPcoFX>sNrDG_Q$32-wY!DOW0i^}VDdUlMtJ9Z3q`0_$(n?`CENz0;XB
z;zS%HW-h)(z_Iv6+JrsFc){qqpw~OKw^swW(v3{kYS7DSZmwG64%-@mwGDW=Omc+KX{
z71*E5Occogl+hPn>In84-oz-(f!vq-6MV$G@tNqJC~=Svk9cNn1#)Kt@4X`KrTRF3
zOCldb#}X9mfOihIYlDA`11aS|ao12Iu`uj>!odI!=hr1nrS_AWP-C?ZefJS64vKoy0Z&Pw0
z?$>}a4tfLNs!hv+iF;kupi72$C(P_zxth4Abo+`95O$P2SZ)#dPg3K(cm?p@&NBxi
zVSL!H0=D`%!;Fy9Kz5GpmUz@ZsH|zJy*Ges~A5U!!fcbHuwR
zN3pY)2{2ggr#;cfbi|b2)&pYtR$@hfF8T-6KY~1)k+U%IUWgP9DqjoygRbdUbwC$V
z&zuin-P*3D#C^2Np!7^7pyv<%-PwTe>hiZ|0;b}9Cz%0XOS5qi^}p*4PYvc?iRrEWXgFV=oc>nOCIb2
zy-Wk+3nn10`tMu;ypS^d;sDs?PMyr)1m3PQRNMlHwqj9}u)cIT$)5QAalMXX=M%7F
z!(JFG2K>7rm+VSFwx1PsIN+rXb0$PT?0-_i><{*CpKMZ4g7}MOKC^QGXX4dcT?_M@
zM7m$!2D`eX4ed)IE)kCQ_yf}vG#j8ueGSA8D(octq*KwJLwlyaUZf)P-Dk2H;;E>rDpzVz7!{G+EsP`rsK{pj!@6Q`D$+YM+T9pE
ziWU{0qe$nEqICPAH7RqWXgHE*Zx&8LI#Of#w>&6lZLk&f=4A?UDLT#+_MC#Es8%GvYnl&|DJj3X
zX}!A`YbXhk#}<9D)*+##y&JF33X#xB>99-=5fYLq>)^P&gM{+xZ+`c4At4(FrpINh
zBxLv5!!YN^AZj-hyI46fh%z!Nbsta%kuJxLqgOVP(Azkrag|aM;t=52d1RD?)UIR4
zHH%0nba7tI+mj@;>GCGsB939i)@xXHF?$HG3
zPj*rfHBscF&36i#Fsx-e33b>wYeu%lsc5NLLW3vNNrh0}%syGbD}q+$S4YuNO*+qb
z$OuxaRg@4Y8$uETJJRHj=7s!wxQ4U3{8Hmh+>
z=63L#&CMZzYZOUUm8A
zd94e6aBmXz)ib3cms`UkLhC4~#w>nNi<^wzM4bH2`Fsc&w@y?$`jb#kFposX)F7Ir
z?u!iWA4Gl@!}3%gsPEv;_h;j%$fS5+R0HA`va_SyuOCITeVIqyHjW@!I!CMKClcyX
z9n7j89Yn^DIaE4$Na%(=sh3@wgtmYGI<3n?%Xuk1>fpUEVnGxbl(
z(yK{Gv(e|}`K@q1#79nR^Nb*CZ4H`S&M1o){$AhSpU?+S
zvmfPzdY;x^maU*gMUndhPxeB7++NNe=9htbXb+Tn2z3}C$zx%AVgzYgMiQsW0cU>p2;z?^8ChjXMw#X9?lB#sNQeF9
z)6YAoXqc~Rvy}_v%OmG-?dT}lmwMTqrDp`$d=xfdGag0^4C1U};9o8+uCxK-HdiEV
zQ^SVP3FA{TqQ{0&{r!UFTPb8zav-56O`M9xr@1Smz+SCvTU__#C}R8>&w35~OMI9A
z)CA&9&su8M19{X_d>wR^lZ4*Hbv(438ASWJu$D7kL&!Q$?iBkOGI}k#lfyuSiZnAW
z`q><%qV!z#mMBXKGV|E{MhNoPQNCis2rC&St%>2CWEw$dcMMAu^9`Z&hB7MS;2=sN
z^G)F%Lr6I+JeY?}Mq)udA!^&Hs32pT*TiwSPbi^Hu~JkdmM6GZRg8)hE}rO-^oF|T
zSB(x+BcuM^lF#@Bhft0bH9a5B6@@0#V^v|OvC=}
z+H5nI0sSJ(;6`#8)Ul;3>Ck7eXE>!zJIMldLU!E!3eIEtbPX0XM?wLNq%=-*GSUv}
z$QOkDm9L=ldBA>r*cQGe4C;_!{i^V%aPE`$Ev`EWd9@oqtJ1(iMz)7n>rFuZ80C8N
eW#Im(XHb^oh5hXg^unE?Uh3-}9$G&`LH_|vxcx2w

diff --git a/previews/PR1624/tutorials/out/solution_000100.vtu b/previews/PR1624/tutorials/out/solution_000100.vtu
index 491a18ac86fdeeae3fb3362d5657663c45f6b1be..f42b25b20ec156292cbc4484a369993016987330 100644
GIT binary patch
delta 12281
zcmVz^i0kEb6lR*P6lhgw^0XdTm1Y`j@vvLI37Jq&W00000004NL6<2vY
zm01`*LPaPdJEgKzq-c>iClXmxL!%j$rHF_W6C#PSlx7B{L`u50xYA-@qO2*EElZ`8
z2`OuqbLQ=@^ZV|--+9k@-e-AlSJ=6IW%m$q^!xcQUj`z<>xwSArz3)&6pRnU^KHYH
z0GljC)>%u8jeo=Hz+MIgV7!@DoJ#e^@UG<2#_xJMJm?833AXL_C
zA0oVuN+U}TBjRgd*B)~W5rgOcJD2(*GBV!p6B2-k&4~RglQW3aEES322}7jhMK$Mo
zBqB)yec@LBATnJ0J}o^F5iQ-1QaN`K`QG+p`P=&pOG)c+&-OE%>C>=ekQWiNyl)%V
zEI?%C=YJ7XqWEat;GL7f@V;2sSocYaQ#>R7jyOuOTcVJu{2_{;J&Rbm)tO?^;Kz!c
zE)@TF^V^xzu%64lC!8Ka@zr|m98nL7Yja6_<#CDyju?H6^`f{YZHHhUynk_0_O~bp
z_}=Gu!4`P0{a~-70sPK#x0X*O)RnsP=YMxkIG4lZm6?ioiWN&&luJcZY-mVJ
zsmPt3S&T^H)N+kBK}1%>>%PvaV0fL1jl!rDA~L$JL!zG;9=B7gDQsXk;Yw`oh8~7_
zH@s=$&O^irH6Qdn7xysVp-IT?ybj>BJV75^f#??*wK5C@SR
zv;Hog1&CNhZL>T9x)x2|s`(mp?Xl%%iznzhHmA|=2|Sl8#_LIe&i;Bewtqh?29fs0
z1Fe=X5DDPcdZ)qn3ZC1rM-d`+VcIXF!H<$vhhthmuUg@@MGvhJDc#>uuwe-z{Nao0
zmOW&6R&S#mzX`)F)5!<}zCB(~Z_nPMn8VvPs`d<#+ARwzguyRl=K76gS0TbR3N-J9
zeRWn}_LV)0$kI(GCF)>bv47hKtgU4bd4BLKo&^2WRX1@v%^5Zgu|DQK2>dy#@m3V}
zO)@{(dd!>RiUxK6x8PI!FT0aEVO1?(?{op~-?g)S>}xzCA%D5q$GRZ0c*kO&F%3jC
z9P(wI-Z6|?JBqqM=Sg4h%241z-u$pzZW$EE<(=AB5=e1xoJvp=@P9k#&9iHF;rvGo
zQ@TrG{{X8_-~I=Pc3^lbBl%0ln#Ms4Z?e%V{4Q&kNjubwxPKM5f|r7RA~H=~fp_5%
z+xko)AB2`ID%1e}DDwI9cO)R9&c8Zq51ikl-8~L}0l)llFh@t1;ZE}pR)@gP-qgt|
zECxJU-W)w9aDrm>SmYdBPw{y<4^7TNhBKzPrFFpXMisp#hiXLr7T-Q-2X!*{hUa}x
z20Zwp7pw;#zkjl|?__!_!$1$;?8P*qq0d~uh+;nI1M
z44-ng6~6M4VO8&f&o@Cwb%GT!&&olck_XbqVgKPPqY8nrzs2>5_zW+h%(1{5=V
ztU98m=O@J?albnXfvLGWMQhr@9f5%UZJA{Vy_Hn-i_`w29&$yTDhf|}U`TPYLHY|xMj%;IiXyQVk5#%j@v&p_1&@aC}vdIE|dQ3SxsZ>HPW;by|
z!B;+f`G3n&Kpb@Nk^f=W4u*sK>*sqV0}ex@`8t5hl-!A!-4u
zd`kOL3E<6BC94~8z|+`SHH-UjkNn?iKX`*rd^%CfwE>?HufHy|66%Vt`e9`TdV8j0
zyU3)FVfBp0xj(=suttLzKlnsdZX(w|h2m&>diyV&3}>9(2SBfDM&GK~=R;o$TeN=h
z$bW$ASdaqsZZwh*4VaIV?ijuM)tAHbgDT`5q3-DBtP*8Jx+i?h|7l`)T6`q!$8Ls~
zTpM$r{7iAw6WQ;XaIds`8`IR)VGas64ax9;x#o3pzfcMyTa}H*&7tpNk3aOpfxah;
z@7y{8`g1lm7+ISH^^a1u63Bt?9~CSOSAW61&HDN_)G}P6uycne%p)o{MTgX9DPH)h
zX2l5TK4)RfN?id&kVe$)BY;EnT{Evw;1i<9L!BmJ&Qs8e_M3n@pG$@G*MnY;)dWmM
z-bN(ytcXB{J0iOsmM~kmkLacdFRjN6KeFfT>4rW%9`D-e_YCHdU?pCW?F{>wQ-ATf
zkblMY)}~5&h)m1A6}kaBX%7e
zRf55YSY}I-Bli)R>EbF#LZ8|!xqr7(0XKDB{oiI{LI0Or4r)MN)CT+I7XAl)o@YE0
z8cH#ba!|PT5sJMt8&3}B|BcTpg
zJvkxtAL!4VISI-F3=bOJYkmxQsQqQv0c~B#`P1n#8JA$*Wve)1kjL5HSHwM`&N7kM
zfM4@Isr<9zEBF_kUbHy{bgd?Qvo%?s;e~eRpN4>c6)O8$7#*fK;6{>04m@A|Bq`aw
zlwoU|l@;e=gW9$9Ga2mTf0i5Q7P?%aIe
zA0+CE$o-C1-<|OP!$aNRyg2BUGngTz3-^&bnc~(7_Yspj^3d20@PC|sA!6FWu*2%G
zKR*aD9M$#cTkkoF8=_2S5?m>c75H;ls1p1?zhG7?fMLGW3-$GD5$UTqAEpQWt?s;Q
zsQ(-h*OZwRmf#Pyedg&efY-8G$vItVz+Z!$3%_zDFm=)UdEh@Gds8nDe208pqN~yY%;fnkO{UFfI+D-RspTqp~Kc%u%?32zX7=H*0
z00000004NL9hP}Km1z{mktN+!Qi)`(5JjjYTwE23vP6+2lC3Z)RF=>bNyRN%WJ$KB
zT$9M8C`{49MG+}lEJfE^$~M0a{7
z>gO>tdsuPS!#t+Zw|jjpKX}acD}SMGcB4F|diLs^ieEftTzUII!5EL(SQ2A3Fu`LM
zm+sJz;_GJ?ow!hLz~9d_6#1;9guLHAJnk=mJwQc4S!P;4(_uzvo);LbvtfZN?6-Xf
z#+>2*E9rQ9DdHOj-iYi2uSWd7vI({i9?l
z$BzTQ$Isq
z`;Trj)h&M2gDwX0HuL2wh=Nzs0J{wcE0F@$cJ7t2a>OD+0g4RBr
zPlqOX%&s+oVg<b!LVUrRkcZ@dru=8obmU!L=I-SI`Ea62z9sDWN^cXsK=zMQtlog{?`b~XP4%?T
zYxq7B`D(=Dv?dX!9w6+X4oY(Dk(
zdRL%V>9Bjf9dXZ}h=1EkApe-qUG=+=$EwtZl99hQnRA`a@%MA9ly%S_dfO1vjX3_8
zBl_>a+!OKDe9+%HF7?Y6j0$$NdV>6CSr56>z$A~Q$sW+3%SyXwfcyt~liR9|5t7z#fjadFwLh#N_e5?2Bh_VCS`&j22uYl0djeX@H4FUlA?2EN&j5e1kTA*+p7Ui4=K24f;d@5%>6^C
zkGapGab3t~w7w6?Ax~|P8n*%RqP>SyB*59_hWZxJZ|t_qHbDF(O)<_+&{EH#cntPP
zru|CVm)IMFRDUL@ws-tMC-mm2@q-VLzv!c+VlH@obn=KZ^nsNuZF%H-`%@~&3Cz-S
zc;*eZ1_WvBLmgiFtvLD#a$TFAvk&|~gsg0&y4Lm^pYsIs0zWI#J?%O)_L2M&jZOs}s9Pncu^e37{j%58t-_36FS&4kLj~Euwu)hiHRF4O(rC({x
zr+#T@>Z0#-LBp~_5qW($wg>bftL|1dcR-w&z07Gv>~sFS&qK4IzvdURg6^lhnBHGg
zbRT-2pMP9L-)TG{JsEPbQaF1$;@+Q8y%h^S^2k`fANFXOjoxwC*RmQ*d+OsYxAe9+
zgZqlJe+$ul*LJ;}2!H3f^XFT^&%Rl99uxM5kJk%FVZUKo6IGR<17llQF#OKB{<+i{
z=UaMz1^XKOt~>6OU?cAC)nM}^*cF$z#?^rOPJe+dhRAOd78P;}e!tm)>QtZgWhdq+
zftFg`-9fM~4yg%np*b~F=q87Cbd!XlxR~GUYp&I@h?}`*v7{~h*Dg(;f8+L=Sk^((cXE!^RAI*aY2bWH2BEMCkzVt5W{l^v@`vqB<8EkwI
z`#;}#2|misfrQ$&dJ@DmNwu-*oMF%=+D0t
zI2=j$Hk`T09vt!G(^&}n=6lL@IUrlVs6-roa+Qo^nh%*59t>Whb+_b*o&n8rynn^0
zk5_Dcb94fJZ4>E2CUic2>r|TeQ5DEB+uuQs
zlzi&{5%NdhqH#B{o}DaQ4*v4YW?O=;s%=J(Xusbs<M;HY@f{HRW-tuqCSHP#D_
zBR?ZLWQ69FTYA7!THowaMWzZ^mw(3kz)$r_Nsz5Xo;`ij^v}`zW~$ZW0;UL+WITZ0
zZPs93Hh5%Ewvf*8>$UKX!{DIH;k0__)yux04MP5M%c{lkkpFo6HjU~f7?OB^=6Ot?
zx77;hwTn-t4T45PYj?^*ug^$YD-6aR&bZl5{VeKZHcEZ#&OfyPdI5uj3x6rkvA1{T
zQ9T%^^PH$|=_U17e}bJG#a`3=df}`vHv@9?CDV%8kU3xPZlU^%bVS$8125LRilP6D
zj8p6|UGR@yodL2NT3d{y(tn6>vZ@(4gFoNq9JrU|wQ?!hod)g1LH^z9A}c80i8F^t
zZ&tqfGR57{?mHa`o?i1&fPdCeJ5b1>xw>+K!`VykgGhHR#j}T0Uv$9F>D|6~dPgKZ
zQ|nymozqLyr@Bk`+$^R2l+4bXDu$d^aAMO;$g3qh|L-TwH=Cy3)4SU0d%6)6jk@n&
z1TJW~IBNpDl3w9?2lq=W!KHB;?y+)Cz0;bDKlQXy-hWmhZ3dkq^oL9e
zn6Aw#rv8dbDOK>Jdn_riRfG0<@RiRxd{2cWbM;;@%XO}Z8SF2Y2`r%hmMU?rrq`fT
zRi^tiFwiS4`ZeNML935R()nklbJl`Shr^6Ip`YX}dHo4_j1=S>XzmN9npvoUH+Bv$
zIS$G%)@}3zkIdf2$A87W=AN6f9tHoZ8fGgZ&(;M&-!8oKvygiq;j
zr@l)SQGWv7G)~z-|C<666LvSL|BfBk3KRfF?_n1=+O6XchRIi}{i6SYqCc(fBtCpB7fA?_;GEn5B>MrsFVnS
z(OLuSC9pS?h)&%{T;bUJvKLrqleeSsJ;blhbUHxuQQN&Yo#wry=;MWSu5Gt_KhZvA
z+$$|QsjnVNv~fVY0;NuR|HG5aGJeB<$0N@UX_`N;9!dC6-=vA9Nn_pC>JF7)*mX5r
zZfHU89xXU2Mt^x{?h2>*(8?YvqWN4<{XX~^{JvH_m7(`@|7E#ylq?8hABKm3yWy_Ej%6m9>XHy^SLX4AX8`@Kgxy_2#!x##J_e}+w`?$iC)`1d9aD}ONOD7Tjf`|9Aa2o~Lw)0=kW
zz|URf`%bD$@$o4Zy`S&(hc!4;oG7?g)k5``WTzI{*0}#sA76Uz}FD+E4I5wjJeZlg=j?e*y~t
z000000C=1gmv=nY>l??BLkArxAx+21CNtroaY8l?8D&I+kj$eeiRxsBj3kMu>_{>*
zDv_czB%~64$|F(nl;7w2>;1aMeck(Wf4?d_{^=iPPcRsgr;7EzPcUBmW`*pYn_x^I
zuxs{(KXc&Zo2#InfzwNw-xCa9e>3YzX$}_S()W(xYH($xa8e%Vx$Ly%zhEDift%7J@e5AnQGw5OcRr~<*t_x4v+oIjNpV7~Te<1!Je^(Yi>NBSk
zwW7i3H7?cn!GN|1b~{+q^K79t=zlsr`~unqdYMRwflKVp^l*dOYd75x0o9}`Uvh#H
zlB{tn@cJ!(W;W(OKYV6*3C1&3qx>`t{?S~mm^EB1hR@B*etzJUvhnX7sIRBIJG}?-
zKZ6!LSO@N`^2_E06MYu!e^3J}WVXFh0L_}0+U)_O1FApEgIVUbT<6e_++FYBAZB>ZA7!ELE%}}Ze@~DlBYf`+{GwUL
zt(KrYueq8i$R79-t&Td)H%m2qQJ3TA9|QRqZ^pZ$x7}c0E!nP3`y3@5uz3{gboY~h
zJ?-~irMd)@1uVvxiPC)=)K4s~-{1p(FzfqYwIJ_aW=J2{qT{I3iTeLIi@6*I^^9GD
z**J%)-d@WzXg@x_e_dOT{wDoD1uaMYtpbrx_)u?Kd$!Fc)c?VIx^e(@q>mT;u0y_`
zwL^^p?7>pb4HaPFtdtwsLPzRCEI_;7LE8q@>3FX!n2!A3#8aZSsJ}K-cxo${v_HH;
z9C^!~qf`UX&mR3uy(HLfBRey~k)Km|I4}hKs5qP3M(1BDf8%u}@!h)vX2@IpFfWky
zUq7_Kbr05ulkqI_4k*+ebzBPl?okg!*Pl`{dAi?p31;9tGD^T7(VQ@nUlfbK(Sb(4}FjYC6@
z`v{$%`6t_k5HFcGcDx6>zd`X+HaMu&dSMFn1>W7aH^saTI2?16f*pME{fs8!s#!N2
z6G7(U$grRAa}21BrX#P;F30L2%~#!u9f|d=XS!W4f5(1DU9`IyjJ$;>b){S2U%GMb
zp*rZ%IMh~*cHw5mE%CDx46UK$87JgTe$H~fhB`T&d0S{-O}w9r`D2`)GLEgbM86E@
zcK3Yb&t{wR2f;ojX6eq4cE7(qyv7B4zn;W?O|&cfAXME1+t8=rXcOWY9>?{cqQ9z8
zk!;e1f7CPn-igQ?O%k(L2i1N|Ile(Zk#U@lMv(t6%W=0NY=7-{xplBl)%P2+!R^xd
z?fPiHDqTmo15{CtejZ5rz&K&I9r=gk%R@y#hil`!CGfxCSnQmE`t^cx&sM_LWDa&Q
zV2h{txPL?*bA_UzGblIB**J~yts4`0#EX7Le|H#Z=D>eKf;F)Z?cUEk4_gTCsjBeE
zLA+5SZa0m`gz2L582&Fc`{$05KHbrCWQNLDwGiNpKtC{DB(l_{b#=CnZlO7oy_!@_H5yN_S==_?8
ztmmQocW+Cj%rIzP=+C}F`ri6u;u4+He=aAVVziI+O07vnKQ2Zk@eklnwlM3Kg#X;G
z&S}y;l`)@{mtfCrUhrQYY)#IvbYbLG@4nG=ANCqPNBdszVeur(AO8H2Q!g9nTn+kY
zjKhDY_DgXv>Noq2r(J>FD{QdP32|qZx(2%*
zY~I&&ucykZ&(b{}HFce~pnGyWCv*&T3);_xxP$*(*(~4+Rs?3KEk!)mHr=#E4YU`q&`HOWAIpANv^@0=W
zuFw*B3(~XxC&Rm5Bd^pYf0ZO`OXKqGdf=LK?#!p~mkYFAlmbul?0rG+f1*Xzg(*|g
z)w}%XI$_^1(vG6>$SWm}(>YT~is=w1{W~K6>lfld>|8-6n68tjxe;u+-f3_djC;nt
zSq6EB3I{gO`*cdG>{k{1;RR{}2T6amMZ@U57k_%EfX*9Vp}wFIZ0;`stN1`SljWJ&
zG_L&9GN!c7@i(eNKv%7jfA;01(=kfL4&WlkMVzDqLEp>R@`y`FcuVXBPu6Q+sslZH
z3|$|AE9{mV(7tF%obR`W-R1K{jJ}g@HwT$?(7gM4uaU0DXS#f&?~%Gj*+sOTo!5j@
z$bZRZ{~7~Yx7b%*qz`?2-lo$#VK92gv`?ESfziEw(%XgQwMM8r-I5v}0)C0v$W8a6
zPuoj$E#j``($7g3&1EkBJ%N7K+~7%=1NU16g=xT_Wq+Pc>#uXzgoDy&gZxFMe`>SFWeh36M?u=?3u|-F!_XR
zCHbe`rB;!yO@;a}rmvO0TP^N8+0pv8ipwcH2YGy0f7!2SKg@aFZUN8O$9TL#
zUf+&ne$t-{_cooFML$a>ID+1y-oRv;G2OFU`Hgi?>E4aVT~!A2gSv%O5wBx*AE0xs
zbEnbt8u*znA)mfeYn|7JkS<6MzG(>sLsEuhN!J{ThUE>h{?B5ksz^_Qwk%eo?@`4c
zj4aZ%=CjU^f9W0T;otL_bg{p|+4B$hbzS-NXn*>y465Fw?-BV`GP}@@!(5?V3H6da
zUk|(le)9nF2~oP_nC
zwf%5ig1+~y7Ri#%jGfdgwE?{z6ul_`d%q2>rvC$Rr<@dSdUw|Uy*?G^VaDf3`U#M^
zDZ`)MAS)F{#m!}OK8E;=i$H!RyHpMRDGX~q5yATWAEBe;IFrsN7=Ne?00000004NL
z6_$HARcRZ>(L@f}Nlq!pK@LS!Ds-EqgX!QDQ7DlR>4exmAw*J$i4+k@q3MW9Bc>Au
zH9MuIh`rWcd#_c|M@Y^0et)g&z1DTT@AW*-egB@@SfTTz*EkEg+?tf9-7GAYC!V$z
zu=uY1G+p{17S)ONqkrc=uy7p^q!rRE{vD6nH_oz1IMCAWIm}|gZ*;%gR~9P&eR;@x
zl0`x3o;n9<4nz0vj}FRkIDfh?HBFC0QDjk|@oEmrr^dSnd^l)W?p>7-%;E3J*!MP(
z9J;53YCk!|pkyCN0w*ki=oCpmpz=R1W6mwNuJ7IDgD2o!FCog2U)>s#gx2(3rf2Fku5l^iuqK!pRR16waQ5T^n+(*?@K?J@aTz^-ECxY5>iT`#>5pvfy=ae24
z!MU8(`(;Fgl*+1Lt7r-vYLmCbv{A^Hwz%p%PT@bMHx_43QZR4r*10UAApKY^;uQIB
zxX(u8@>L2minD9}$)zCd(YY`!ox<)T)Uw4n6dG@@^R%s}kib^T*v_IM;jvSC{}tl-
z;8W)r=6@85D(vG$D=9?Jt}JY|r7+Aqok-N7kjbmBo-a%8-MsDMk}pEYx#-T=L0QuoITOI2Md%<3Pz_pd&M>**hEv}sUyv$L?QYz_tAp^PZk6%_VW
zWQOA?h5nwsfq6m-;Tesed^XVV>AlCDzd$3k-;7guLPKkR^{!)8Gz#rc!D$ic&&?u}
z!G8i8mS3xmb6PaI8x7*_OVMbOeIYf^kVZ^Ta{go-jk==DIiWHln0effy;>;(rRf`+
zyq3Y{e?2r`v@mFB9C9aewULW3h$21UH74rM@eZK)y9Fe2GPi!
z-z|wB>|1|`ze}*FalPG%O%m)p*byR4d>tO_)UkXk3lqz#ct)Ou%cu%BqRC=ijGcdO
zDDlaW?qIzx7DoTa?mt!T&fzz;YY4^pMU=-
zi&w7A(-H}PZkzU8TpZ4!DG{1aR=Rw%yi8e0Y9MBip6MV&gHZ|5_8@
z{IJQ;Ix8=RO2tC$)?pE*Bp#gG`;o@qMLP3SXV9n!D(Wr@pSAQ(C=JSulX*P6awsbLy=M`Jn8d>U8`D_?
z9CX)LEtEh`&1_{?q8LRFv!2mz40ijyvcF^@f>p|s>hgs&9G8Sw2gFh6`F|0Y<)uWS
z>CflENyHB}pS)k!ByeaojOE=;=kTGRWEJmc!q3_B^IW&EQ1$$~?8zz#&WfJCuvHMF
z+D>Wr$44T#o6Z_{KTG33f>jyLS14pBCM2m7U4P)&volyBg5EuIe6K=5`?T51i#s{^
zOHok<*&G}Vo#*IUbEp_T*njiLlSNqN5miea2^PoJ*d#YG;CG7L?E^&^+VJtG5i1%@
zuv~sR@sI14WL`7z@PW2Oj}9SKd>-co2OR2oZBqkmAnYe~THJ0X;L
zO&f!Mr66k9bTrS7#{JbT$M%xF)%8=rv6n}Pu4?wJ8QjBR3To4o%pdUw{_!b3i|zCy5Yk@v_6u((?RBGR&;1t`D60cZpjPM7hy@;
z&vA)lFVqA$6|LOG;eV?BL=KPWI4{yrdch?Yz6ybV1ZzrQd9d%(COZa?D^4fN9;NYe
zPQb0L9ux#(qmH+vFI9mr1HZda2zo6Y_2B@G^0vI?{$>nHl}<|2S}gotII2gJzI>Y|
z_a;AugOjPCF0q{KrP$&}7!gAwd+6kbQ4yjRZ_kUpG&cV*GLw8La2Tw1d@6H1(s;+E%
z%{wbbk`Y%tJxzq_JKIiJby6s)we5)bB*f;*FQ*jBT|q=HY8WQY^YF+Ad^{FnFkc%YLAb#uwFS
zOQjtYELFTG^N7Aj_SCOgnL+l<$}?8$Ovs+gnmA(QEXLZ~Sq8ftS!n%rK*~Uh!bG3idHx{3h<--&S-@JL_+X_hzd8Su7vVgoJ{B5gs
zm_LWl2^R03ykS8+GEqnPbGE=~9oy<()ogZQmHrct_!@JCv}a{}Q{y)ysW7aonn
zV`fS%gnvJ}l``(Fq)%qoH^~$eKXq)GPQ6tpKF}*sIYqePHF^EZndA`C7(FFm8VSrn9FQsof>F_=XHTVIo?kh)A$YTvYY5TDrRMjRvhebfDC
z{!bSr$Xq8#stpsP-g_cgSD(zc%etzSqhT;^yv}eLjaLsAEp*={!cxb)t*gmCYTZ;8
z+iM`fk=UF=8#a@iDBEH5cmW5)3X|8L2uF<7&I)FRaOetuk~jN4$%nIRQ?sf`F6;Bw
zNPqGc<6L!7QaI@wHJmp3gUspr{0C{JO%&Aj3WrqQ(a>pB30dyU;B)n$ep)l(W=TEw
zkod&HGxS0W>7Uz?hN|Zz2O7Em6LKe%L-t3@?;WJxm63(TdlDr$r(GR$g3KqNepGUa
z=;3|++t$q_KWM*hS}uQ>0${ptTP6*d4S$9V5zf0epOoiki}84=>`cokgxj*`d^~4y
z_=n$oDTlnXy1A=WmFT*+DpE5yhK2Zw%pC#I!OoPid1Wdz?($2vvPKkU`uf)&COPEL
zHR-SIo)ok)>Vw+0(`fj1tKa>-2>KEA@(a<(H1s=hP?1J_g>S5v5SirN675F1TG_qePjIg9U5
z9~VHw-)F#f$5#;+#7pL=G>H*WF)cyIjpX#8fRzi$+|6X$jKxPeoNH%dMqZQsF>qIE
z9pPg^_^{F<;_D3~{8c6?G|V^4dwC5rG$N{b(qc
zSm<}oBl%O#{$%7wA&!qM@#w!txHb}g4Ua_EfKz

delta 12283
zcmVz^i0kEb60Wp&e1T2%&133XSlMMu90XMU91lSgTd<*~p00000c$^hi
zc|28H6y8dS5He;+B{a|tDT%#PH!`LkUW3X|M3fXSWTr^;%J8HTm98nSO6D;#CMqFA
zrIbR-RA`^K`fLBrJ!h}+`@XgAo3PM*<@XUe>iz2~Zw4a%>x(bBrXzx@3nxe6Ic;0J
zuT>TzYL*fclW-n?;IYuzn~li#k>@gEa4*iiD`063BJXyZE6mA7gk9*IY7D30{o>kd
z@cXZH)!dfHh+N`%5MN$|NWlsfvjyddB$dcJy{tq;oFg4GT8)U+_vJ~8YY}N#CAV0#
z4iQn)XAj-$5$W3N$G@%tku^QaoLM*xPV6?+Z9+uJF4L%grx6ji!Tz~cCWs91hsd7X
zhseCgWszk^;dg_*oiQg6(Rt>xQ_%~N;mJYIAYbq^X8Xb@5RoQD;TWDUL`t65aFQYs
zN#Gv{x444HNd3FC^y`Rd>wb{VxrfL^$Bz|n9xzy4Mu!_b$l&{)O_IZN5i!a8roDC{
z#PO-xGkOt!A{lf3ZR`_6WYM8VF-8)I*c5MC(JBRYa#KBL$YD>kS#~T_HBDWzXQE>)bYRCk87Crd1dnVertyHs1?&
z47TSN>vY)9;1gNMp~?#0n}lv3v03DdbVS
zzuU_BXEwzR&fMfBk0_1^(oG+%rr7w`JgMn1iVs~F$lT>k@vFNgpG^Sns0N5k;lGXWHrh=on`o}oCyJ>uVp;}oB|E@&)&gyOoV5i7SiQY;-%u(aEW
zVu6%zf#=}-H~YR&dJx4|8?|yo-6(#ZOX90eQM~#X`$OzWiu==c2;{;2q@S{XU9_XP
z-Sbr8X1Mq9aKF6)d}p~^)3XZvirhW$-5uh8vi+GbTNzI=E>o(Ij-vSBe=7!)!YM9I
zzqUI)k>Uq+uN+vB@O|L!!ll6!vzN#S_rm?o_dl{5E>W!E`z}@&>ageG(6*r06xSbR
z$JEYcFp(U$7*%C(hhK2S$YusxxEXHNU&COhCanjYS&9?cpC$B^8EjJ0I?S4)_}AEf
zleIO2kiTw+@488f*N2?28U5xF|OLcK!(_!+PJGOLoo>r|{1#-$ON-RwLf`jNpO_O7ZcYJ$Ee#O7}3WAK#L
zs}?RFBG#zwu$L+#DoJyw7x0BT^2|klE{HtI`>m4VkI2Te>jIC5Bl5-8e@(|KOL1<5!wHvutSi8$lBjN
zPVR+>th>0?{50^IKY0u5CGgsB^R0Gw;B{h7vv)N-w=a#?lLnp@yqGu`7K2ECL-V26
z=FbuFiHOSpfqt8jxPB7ET@4a@Q&
zPd#n9c}5KWI+Khr)VJ$o?d{*G6z9zC7+>`ik($j5D}_KW7S8vc$bJI$*nXz{V0X)!
z%U-gxhAG}0to}w6?53KY
zd40lz;@T!PpEsaWBCWkQyWv!=&~Uf_`SnjpEv2^%
z?1x%(dksE9`*_s~m-?6fE+eb$If->DID0=?NDXExS(kHPNquZlIS1Y89w|IJ$r
zaYyBE-{?zm-YKT)raSP(wMnuN{0whWLZ>_B1p{0p2D)
z?ZD=UEFMXfQ3M>Zq77MphW-pTv(jA^I8E{8o9vFXSc)&j^h?+}0I#&&c@}z6oW^Dn
zb8{)Kw*0nl2JD3g@D+-KZs&Qd%F6})zw+pKK
zpl`XFCzJQIQ>+rt>sJDCdPy3L+1OLuk+sXH6Z}P|#wpFE6zfKRWEwjI?#@MQ9WaLZ
zz_(1KNFDH_$m=t&D-n_P^VVeTf%wmMy2jzp3}*c}oTH=5U;(rD7DqtOWEy1^7DGM6
z+K->$KTWYgEOHEJq&QN}jm0_4;Jj&WSp(p^RYkAGt_G3w;@gM!f}bp#@VxKIh%Due
zTBrd!esjyfne^9x3_hbbZLcrK;3nJR7efgY&qzx-rhqP4`Kby?fG*ndA1&jPVz8g1
zwNOGUgO__0e!2xb>J+Gqd0K&pmeis2Nw7bhFs|SS_Dzzel3yFa9F=%s?pwm(V>JX7D>{ZnUern2Y_qoJiZcda=Z41ROcTMUwg54{;YdK#*
z4~NG;@%jkByeYj{64?PdGj+j_4fB?-&1hd8@E7SLt1Q5$)3l?5N);k2vRk+zpevo9
z|1swmhrQu{!@T^S9Sn9EY+P_MnPQ8P@rOE47r}kMUMO)vCztVF6o)xi9C&uj7vdiN
zI22eHfe5!(r92Y$37Z1*gaV_!|}5hluPdYS;XOBR#Ht
zVG8(vOKK)T>aZWJxyye0r!T<^{43?{LD$DOWtA!;@@C4@^hyhZUy6^V{n!nCxHjSX
z^Ap&wmi?Xyd2#M*r>UtS@*&(fD8mi(?q%|zU<&L%%Irx~==+?4_kD4|?{6h{Q%?i`
zTnr7y*4;$J;y7JZ3UgrcgM#_C)red+85r1qP|x5>g`GRxVINVtB|4(^n_{6CbxLEv
z`+`L=D|Pu15mvu=`xw;W@;#H2A3-PhPK7x9ggtMOW|a37_~|VZ)YkyMoT&4ij=T-^
z3>N0kaE1NHPLi>PeB?GpoYX8}@MGJ#eZA192a`QJy`Mte0+!De-p=4iQz||m=3kwE
zt)=mDJ;0aT8^IgElNNt|8~uFPGbOy24?^9nPkXkwf_^pZ^T_Ohd~DAnD`%+NjX9ZI
zUg*zK%Nr&L@UZXFX)7%kif02BZ@9q$-j^R7XDw&&N7fS+Gnk_jcT^jiv|#V-=ByS7
zKxA9C6gl<)`rgA;kb*vyn{!j|K;0C7b@l(6jRl-UI~`VsdGS2JJGbZy=mnqQY)A;j
zbftf|)-j6D2=}+0yAHUs^uunV3^wI+A#NcA9p7G3WYt}@$<^&LI#U1)ATsDkii?ovR{UQ{%teya*YIkKLu*Z
zi6K`g7N0XGQJEk1ME3o*0+@#zKJPlDrHjaKzw{*;(TF@|R&&H)9w&Gth`WOyA>mlx
zKl{FZ)q%>dpkHWKWK#<8dacl{*U9n>R@i&KHVE{qTG`8teUxIK8#mQ+;Q7CgQc}H&
z4F20{W#xI;FX*u6vnNo8tW)8CU+SU$O|zE_2BGgfS`jl1CD4C-dN2q0b9LOhveO3e
zH*0LW>pg?nj>zjV_&MJk{#qIKvZ@&@9{@URaP}?F)f|e?J~m?=1pJlf3A4pv?%aAe
z=r8IFe!5od=Pg^jx)&NX2N
z-U~9=t>^K#{!rL+FB;D#I>Y?nKQJm-Me)GH!rz*{4CcFYp|Not=zQh*Fg@sRe)qj?
z`p*zKnKG+n4*F0&V4D6Ma4o5soYRvAI5)_-@Mo?>-Vu4n2l{i<=Fa7z?-WmpDBmaw
z1bqvreRCi7%dJgDz23kbpV5jZ%WJ9_-25<6-ye9x-uR&Y8SFp*1IwaTqLa`k7=Hx|
z00000004NL9hP}Km1z{mNvOF=QdzQu3K4agEUAl2Nl}(kjEq8Ddm9odR3w$XMQF1$
zg_LEO5GGV|iKLA*mZEE`WXb$ae|>
zO3HR>O7RV{ijH5XSjRueVhexNRzlv-?<)uMVfR*1P?ixGWZC@|lH&m$)izz^4EsH=
zp$SL$PbMGjC_{Y9P^+*jcqwfB(rR!yXJ^Vh(6o!Msd<{mTD(}cTo25$U4M~b02ams
zhT0=Ov+wK$HTYEwt+?F)dCqlFtwOLmRpw_$8N0wb)Rjn_W`GKr5hfBb9%~RHL=eMkMb-l$a6n)BZ`sts3Y9HVA~J{247jKEbo3(qda`U&zo(mq1g
z{#&<=>K3!Qw1+_m}BKgSMo1%HmdXmR7QmXKU49&a2E2am%DhlK@N^r$+dtzU+Goce~=HyDAt+c{Jl0!_EJ6V
za+<%)LB1NX80Jrk^B1yH2c`0Cm-r(tSaqP{KlFil?_2G6kX!7_pBE#}QLMvS0Sve2
z*S`e4YNu_>cEmk-B!6Zjj{M_-1?mNmztyOXq#*y(l)S6IzT=5ysvh*7ubP8;5jS=E
zu)!Pf*74XnKIrcrmHJ@=o(iidV4~Zy6gTK&Zlx!#Lw=9`m?etTFA?J2
zIbf-;O4(jeb*1sl5IWC!Vscv%Hxwr&rUWYT^{Y_7Xf$-Va(}=#%@^hCpnt%N-ggn}
zU-9{K=}*Y4@|o|4QU5Le@;UMFGi>sZqVLL<{;v0u`Y1)y`X2{BW4ycHO;uY%+%%NxZ+Geh!d$LL-Zx`sHZ&x${`$^Ug
z4cZr@bzK?@lz-hb>D>*zQCjS9De~X@AfcEIri}eO>f6)pzuSe@BqU3##k$L8Ei-VEVz2igZu82PQs{-#Nq6LgL_?
zpv<^G;IDV;zgl1L&k!SVqMro2i-Gm8C)bK^Lx1bCm4BwZFc02qY;6_9d6>MIU9|^x
zIkTa%HmuK8YrgLXj+^^`(}Vs*oABXV<2)8e^rfc;^3^?Lm`A`~d$3zQ7BrG>&|HZ9
zj5WXPp>qn+w5U>~bI!BzHh`?QTX~Zm;ut$*PAk$pS@3aWF7z3`K`ZHgE)~)LYa04n
zeNQ46Lw}x}h|5TUd`Ib6!YssnJfnI$8mx4?wss%v2W2*JVz4j8dW#*@$0oP+w>W}(
ziZaIqVfWT@z8DXG&-n`%TEfq&O>O}T_Pdp9g~G94ALdj|HE7D%78(e@Oy`r!9Pxcx
z#VZr8ke~f7@dU(`Uk==q3_Ekh>zI1b)Zt+JdVl113k?st4ZpdG2i2)Qtn%aYlt2?^
zZ*KtgSx~*d6Y9KiCs;Lm1V9WjJFvOK7E74mbE
zVGF*!@hibw&`v#Qes9Q)!EgS?Mw5Z%4gZnRmi{nzJc@~(4U-G6#N6SDJ#&(hwAw{
z(C`!1dGF2*p}BfhW`)f#)_E4yzwajE7oHSvJV*CIAnq*9_1(VfHrvCG8Lgyd3t4`R
zh6nZG=%>!FOQ63{dn|_L?*89Co#{ZF{eRb|IJb(L?h+dU2Ek>3U9
z;n6eD8!~M&wIJ6P_%)pZ+vNkzl))pDjHoE&|9GnEE1h#{uZ4v*C~(a5;0W|7&t@Dt
zfq2a7z|Pgn#<@
z_pPl*rr`H(Dnrm1@&Fj4
zkNrMC?(iy{+zU1)qzF~e{cz7rumE?dzBPPE_x5vAF3tO?JQWX`9}RKO8%>CsYXv8f
zUm+ssE6pjt4F6@czFAV$4HdGdvwz?7Q(aQyj;67hpyf1fPQM0r`Z_2^Dg|;i|9KId?YE)p?_C*7f?MU
zPUkpK-R_q(UH%S!*dWqE^Q+lWVg5D95lJRhqLA}H6>OpU3wK7;F974}8=~m{;_9h{
zP+ickUweq`Mpouu(-2oGl%i?|3i9V#pQCQI
zv5Tm`VpGc$e8Fgm8Cx~!oP8QR^>Cg_$(z*s!3^j5!e+2nFQ2i9{#(k$m@iwX|7vcy
z2!Mee=@Bh-t^vOXOF&*RHzQ96tQie8?1p~U;L?^ZEy
zLF0eK-{yf9`AXgN{)Z=Qay*XfNrv*72d
z@?{s*rReB1o8C`347Kmz|7z;#9h1-}7=Hx|00000
z004NL6_q<>8^yi6>nHgd2ScV}CM%fYoJLa`a(k(Cj9{sw!v
zB!jMizu)VwZT~&V$e2-y?!dU!F1H+5oGive%im9y!EUFv+^`sSZY}kzf{+*9zN)1S
z*=x*ksP^Y1!^Ow)9}_UxTkyF6#uvJ0n|%VEExxQe2ESaPDO)RyQ-Az^<<;;UkD1nVA}faIl`cl
zRLL7oaFZl!!W_(d=pCGj@1GmKFuVfmVO1!;yaoM8x_aa~E*8Th_l}nrm{K$`+k*IN
ziu)6~VCV2%(69-#Du4A#SbHN)aN@(@q_)(NW&#BV>iY+eM^uI5-N2)%Nex`ijkNk4XJ(ga_x?rWlTw^n!D
zNWuPGd1~Uw4W`HI9~FUKT+)s|6>&FzNCqAm1+7twHB}exr}U
z81|`ZWQi@Ew}V2}Rigzg#;}n>y(QvDZDx-iVTZ#91Ab
z{i_Q8ZWaeCldK&J`M~Jj#8CLBJa9eZ4}Xqro=HAxZyrgdeN%saa~kn~57gTlpefWs=d5_8^9;C@5n{a?{_b+c0m2|NbAq=3`fiTp%tXYm6p(!-
zPUjmu(9WPbiSuy&4}Qm1Z`NaiV&6DxzJFnT8^(qI<;6TBHu`F5&<9GeCJ$p=|4e7#
zB9K}7=x7@3of6UeX+4&~4w|jdk5(M{aSA+`bGY&!@J;(5lL?0GR-93yJ}7sNAwlz`
z70k$B+?4vci+8}KHqspn=zA7E*t835@_VW&4L@D0Z6EJpJ(Wdc{tqEf+h^S7z<;{g
zt~{;!be@~puc<$2#1wZ^-@bHl!}oAduvg~z9ObKzaw;Df^{ceF0Sw`{NR5Eq>`|$M
z6u3Y%)$bwVC#uc&P`;Z4{7^MO{4x~-&T`bNlFO*VSLpR(oKN1Mex!TsOEmS@VeK-?
zUkCp!JXC+}J4&R7!MzW>*>|BYcz-@Oc^g!1xA(}$_)90_E8;QFVf})b2I!;BOgbc?
z58K!Fjr!i2aS!p^kjD%bT*`o~!Woz#MEkHmr?#GCK0DhUup@tp!1q$}7)_TX!M95z<{`5BolriYncw@4e#8JD^u@Rp0zKD5G+ZD}M{j^6Oa|
z10LzKlnDcGZaQ&QfX?@labP6PACO{4-{~6sU#TP3ckOt}v!mc_G7rx?+V{h<>2&|4
zuzBB7U4JPqzesgFWbF9O4BQ%&7BG&uh0T}!ohcvh8Z2=HE6*e<{{ef9Re}}eZFtb$
zS*n{3uYwit5dTS!VdOE$k$(*CqVrVm)rZs9B7TpllTb6g59Fi2p9T*%?>%fm_h4BV
z4-d#17x?rC;tPJ(E2MgxNz6Rm3R!KP(WaxYOX%EhtAYI1{kX<9#4~%nsgVPE^_@5E
zsqYG|kTaux*8O65-&^?IcgR{J3E9}Nc(*pqd)YboCG?M$G$l!aVShY_Uemp3R(f;V
z7yEU^Yt~((|TkTZcI>~Da1y$tO5(&<$nHzJ&2ty5DcbjW~ga__4nI!?|_l7
zxDBM?=lr05E8R~)Qbj*Yp%2YgUUD3i)DQ`z`(E^ASvKVx*8?2^eR>yt@>#=2?^TE-&FGafs^UiN@_J
z>Nj0_j{kzIti^QcT%;wg_gO%0@^~Ri@1&Y%zD6yS@9%pusjpv8cKAx~k;gSMOKCq_
zGKJzupK5tzoh~RD)!k10q31wv0^Jka$4(e`VZ6?w^Tqj~s(%}2`)8b2tBpp624pv{
z{+mVgjuSUmtfTYe*_D3}EIt$Zjp~5odQycoy-#%YpZ0-{i;64tDgWx?L;b<&NG)!v
zi++uhBJ0U+y0Me`qJebMhDqu-IXqW?fHvm7fvV6a+g@kW{_D6JanQch>I$FO3x4h2
zYQGcA*68`O1b_BfWi82O>dVeqI@EVeE2ms2Kj*jP{XR?g*rBkQFxb~cU6@J+!$V|B
zNdNgmd@1#{>3~6P>Q8*8d~S4(vFm$(6v8iUzesx}`i+n|Ly`ygmCK2&I_igCN;+fd
z9r41MLyh|RUt^VV8{sch@%}lzTMg^&t!aOsu9B7Sq{$k40yMyY>Hu7jU
z{3dK}@KgV}@^pL1Jmy(G$>H0Jcmq>KhE!({vudhd!mnaf_MRem$G1aJ3HFNMj$@S9
zYGpOXncxWD)hv3aRxr2tQ(urBcvlwy2E+}@P+zmo8GWg(rvBcC>%PxsvcrCjPma%-gbQC^$KH!C3C4VSn5
zZ@?`A(m!}{Z;xBG78t_sWq7?B{TIZQr}G~I8(W5ls7|(~{4%vfeD|+?yi35O>fL^~
zz!l5w=d)oiyCFT}1b?F!dQbA{{c@@)%o%K09Dkil_r_TPCw`*AKg)sK#|8nBt3zN_%7=Nk^00000004NL
z6_W!Ei3bIFh*Q8HwV6cQhiL=??iijWFvE~#`xN;qkd
zl#_97-*!r|&&~?d!evwbpw*&vW0u=U$rfuaCQrvQRy7J*D9d3(F;mrz~1o
zjB7TeN&dy6B#}3Cv44k!OK4^^>7KDGtS_x>678aJU$r>pOb|he_F^Z+bmAD4FkeNbu*-JRaM*Ae_T{
z=^)LQ`#4a#9#uB+98_qjCZB^Gq+9EDpG)OnGN(eiD4oNUf`75lV_6)&ogLPXJ(`MYh+hZA^^xLUqR%lUaCq
zyt3J2$imE`bxoc8RT9PD9LaPB;I~*Fc@P{UT}D9SV{y#^_&uEa+A;KHxU5=hDgo1)!pNg6gT_GN-Zt_ALX${d=nYQ
zh}Bs>C`44rRsVU>6l^Mzx5PA3I4x;f?l?*z#rV#mjByIH>)y<|CZsS@KP@bq{I40g
zK>b=dg~@rxuK#s{!a&=sOB2&5L?=;;7v)f>y}#Pcs*=KPwnWNmDvd5>R{GN~0bcbr
zI8L5JA%Fie?|{&bLc)v^ew`JCA*Nw0aTbMC<;v2zvJ{jyc@YQ1
z$4e>XbsxLcKZ$~R|0QeHjTDx2k39aROQA8X=T4(K1$HODuuzLa@VE4cjmt>-sXx8ytH*j8Zl4}=7cbFB>NerzU((RsQyOsODvb(WHtZMD_;okeuQAI6lyDEc;
zK@09aSZ(g#M_}%>K0Y;9+mLU~xP=m$J7#r^sK`eTb
z-uUaZun2ykw5i63gI;uDLHrI5b9VBMEjYlTXJhPkK_Uk;LtAUrU=G{~%`WdD7JuJ2
z)=W$!{JC%3R=g;L!`Y9!{5-P>2fWmr!+bdCTUdE@5g+nbRK{@ht}Qzz57;FtR*H;{t&T<2HSyEyC~N|3pm!{PJV
zD0xT1ov1)Tz7fU3szdQdo<57$hkx8}c+C~VQqFoGUrmIx9L2r~qe57WwI7}GjmAg*
zXH$2g_xFM2nx{f2&`R`tS4j#VH9tPuNH|p{v9swg;ger(?ya_k95NI6dp4x8@ZRg9
ztI8K+XsU@_OQHxl&oW=p&J6Z%`^dXuDui>&%hICxG@L9$N`2xel>Zx->3^L78Y~e-WR@f5aWih;hmMd2oG%(
zclFl`;cq;3)a5*l)2$Baj%5_G5)%%p5nWd(^VrF(0D)g#*>cz@F5L-9@yNfK0q
z-Z2hA`i@%K795@p>}{)eV}FrclB8-rOANQz>kE>r8Sq~VU3k7iNN(u=ZEzlqVgCuI
z(};iEwhDPm*+9SKk-s<_y=g5a74pS>4SN4T+aF!X7S#_L+aO<^scQnOls?%w<
z4`g7zrKMA*iN?~F6LlL-QQ&pEYljR7AgY&r&`joOm%DFq(F+PjnzbkWbcI;g_~rk`I8^J7<$ov>9rMHWCC#o7AISUu;jbaaoV{INoopC9x_l~G_7IH^T0Zx-
zx)R=q44OMgUmp5?@cp@lLTtNaL{Air{Kk_@yiFLCDjt`lr?W_SXR8)X`Z8H=!spW|
z96XF$ZzLA6h>gvw$Djxr$G#n3KO}^oS97z*54^JGy{3>|VbvV=RRD*Q@2BJmN41+0%{n*J@US{le}Ts!)FQHPqO%ym
zrZ<<3d9(PCe}BJqZ#0K<#!*^R+&Rp;yZ@%21dEo-jh8Y889daz$LsB)@l`e2TyY15
zMJk)ePZE8PhgPn%OQ+ClcW&NlLm>h($C3;jMF_s1sTW|&Lha8e2|Wo8_6fJ|GsM4N
z``hkGkX*7P&~QbcmKdQWHq~ps3vp>%;POwz_cO0GDSri$IbK~Xf3Z-R!gM^Hz1NzC
z&aw3DWYX`Z-W1~(E@IT=ly^J*VDWddT(AP++ZN?fFFwg3jXDK`X5^mEvs*2Ky*UUH
zOuxMR%;MrsZtMo(&)tHQM>OHj|CSA2b|rqh5>qAFLijWIUfCXmKNpnZZ+=&%kvL+a
zSWEb$Q-31mQb+n^blXWPkND}$mPu5H8V75ge3fj%jfM7c%t6A9-jT=7E^|3NZfcCkuun1vNL{-WR^8jS
zPlQ69{cTq(#L%#}SX6L>MT51W?f7O6M?Y6y-+w{sl1r)hT3U=i+a9`o1MWf0b4akY-4p*A{uwf@2
zlb#qyVsrMb-^3zQw%Oo?8Hex1hV5SoMe$r(>r@Mct`W|oQ(*|k~y
z;D2TjQcH6WhLFB71BZwHC37m7`}A-@H3hZZf^RBcXqZ*01TJx8AYIm{d$>jfetsqQ
zjQC`VThOIi(m%_jhgaW{95~bEufPXE9M1Hc|7<4pt_kPog(iw|PO~&7i_9maa!7oI
z=%KB$qiz$)4{GhzOXT)bSPA1uuM8UNO@I0VzY)&wtT`?he@ujz*0NK~rCF?(z2MnkzZv9<7>|I#r_Udso9XPQE>K^kCV;#sp*XUblB~-hbqt
zBJ-LsS90F1&=eOnm4(5Fa_jd*-_llt@~WPME4?>WY*$dIw2xR1fdE?!gF@G=qA(c0
zv@*_zhQCLz)sA5yY!8UFRH{XYxjZppmNN?tKOejK|IJ;dakl6XhnOZNX0V-wbnhdH
z)r617Lk1KV5?`+zjCU|hp<%g6&VNnLg~F&@vTYmT*H3okP#HPf&r*EUnnpuu!I=nw
zJp=8mc-|zUgCYL0KeMir?}szN23rXqE!OREKe~;Bjrv{1qW|vU-cEeCMvRpGn{SkF
zBDp0)J%3gzja`*XA8uSpa+St|rOKoprw`}eW{|uZ5TJE#PX`S>w=Z*Co_{eAKhl8I
zbup5f@_+A~M)t_-sQ7I}&j)rmhE|Z@7dn26!xb^FjpfG`VJiK3ktU*iH?#C3Y2SbwflYQ3syvkHKd(nDtKdR3tjsu_n#r9no8?
zjqjddB?^4wicUpK8mknji$=x_Dh!rJ6%hVBuYIs%-WZGj1B;B>K?)!sJTF0TaAj^}
VWn@HQbYVUUJTF#MOGasIWj=}{vgQB)

diff --git a/previews/PR1624/tutorials/out/solution_000110.h5 b/previews/PR1624/tutorials/out/solution_000110.h5
index 5008d307e667678c02a4b1fa2d2124188e920f35..0af43f74303a2f4bbd591ec6eeb702e3fde99f57 100644
GIT binary patch
literal 11976
zcmeHtX&{yB*Y}M?l87QjC@BqQ4bpWdLP|o?WXjGwPnkoRN(d>HC{8I6LP%sNnJOoe
zsD!;Yd+&(M+51^{an5<)=e+Op;W^L$!~6dChyAK
zb7S)`{-2L?U#!_r-R57MIeJcRG5f$ai;=Th(bt+}^S2SqB@3MYX~#7i
zx4E~a``-sX?`NdHzW%gZIA=$4Bsr0sJ^$Wzw!J}j>rV$9*zA6tNFKK4p1!W6pZi?`
z!nyq(a`y0a_d4n6%
zVFD89kK<8Zh$*um%(>;u&05z6Th5?4`?>sb{Ia+9{J&b)(aCD-n&Qx(d0Ox*PsF+&
zo+Q`5Js9U-i~ZaL*RM!s!y)D-@nK{8#TfZXVb~mU%+A%n+YzD4N$eSrot^!s1MXRW
z9Bi`*z^)Lv`{T1@mQ5&6#(yU9V(Wg&IDUE}>e~|XN0caWvcF=&Fn3#*KRoFC@$^qT
z%=Mjtz%Ot3|K^mahPfyR(VYJ5>t(jf{xo94{$=@{ebxO}`IYz?+<(q5F)rqI=B@D0
zO|n08@z*EWFQFG(tuUOH6L+i80>eiayrm|dz;G#ViFJ2@mS~J}P&tX=Y#pkhV}Smh
z%6V=i47WPh9&*VV!!wq9oEZlDtrMZrWyKhtwZk>`!6OWB5bOMwT7cnt9A&H4=3scI
z`L-&qEDXP3M|IZ9#Bg-I{k+gk3~vkO+v}Br;XA}s9iCmm@abM3y$5+1KD9=OR~B$h
z<3)Y0LJU7f`Bb;B62tkXtcROxFg(=jL8S(}l
z&zQB35(P2*hzYZ_M;gN;+a(Tf*^1$kPn^F|H8I?RG^owC2g4^E>{s8=#&AUe)YV~(
z;Z2?qj<`LB$8XhMKJ0_xq{52Rjl3)*muhrMWFre{ig0?U9b}<*d5e{g>$1>=9ge<*
zVl1?3xk5$kcP2_y>EfOoXCkN6c~-P#EEKA36RNPCg^E36%*0MJ(Feuick|9M(e6`U
zn(C66sM_ws;XVFLq;}7>_oFovSvNd9&#+^nsruaoY3@w4X7Q(URklpjnb;|H$d8G%
z3x0UG7c$Yese>mk+=O-5GCumoGEo5!?fnGM+u}?v|6nkYO_E@6hb0SnNNah>#j()w
zs0@Q%ptqM~9*$6FAqLONzQjou($>9Ov%Q9e;`F69qa{&naFqlNuMrXCJOUyJ^z~x
z6UF3Yl0HEm*Qe-ZaK}Kuz0XJ~hCYxMnLgA|gyA+LrlzlyAy2RTgiW9xH&kZ1^*b<;
zNJihf%s?he=_ogzxWPp2?NS*&v@K-wh3*
z57%!Bxtb02Q25;>A`Z@5EYzdrHRP>rELrr9JJhBA!{G1G2g|JtyyX|L(0W=M>D_z`
zf1AF(tPboQH=Q{`wZ`x_nG!MyaPE@K`)++O4-SXK@MimB_>*Pn6vh@7dd2OTBMg2z
z%6YVUp${-=QP~$VP*)i>Dx1Lm($+-J7C@3@>mjLN41d<-;!TGAXeqiUr^1}kv86xg
z&&P0<`{rT~a|}QIiYF|4hJ|Xb(tfl^P6K>++}i;?rnFww5$}a%pJt|Ez6xTUDjGT%
ze*NP4Wh*mS$hlqh$F)=@YI76PYl>zfsWYVaPad<7+I}7GHt6dhAAXH5kiVA|;X3j#
z=O-ohFWLg!}orUJ{+DuMB{ckb+c+IYZiB=xid%+s=
zcS^>sr1UJj_bL{Z`DJ1FfrXln#xG;Ia^`~Zm(UN9Rekg;FfX2R)=IvE_{?nchD-XG
zNQQT3zzOJwz}I|=mO(6Ztnlds7tBYkw^gyH;C%gFzoYe8!MvHdC~t5T&ShRY)fMWd
zWyml0t2c&!5{NpN0rkPdvvcAZ^tYRWv6u{Ep{eG?4h+u!%8}pRNtyho>&stX+5hPC
z+`TD{xQW9D|M~sr*ZjGAC_mdm_@Q8D2|q?W%oVA#KZoF^@~^sAU0L-js(<8OrT9N}
zuVSCg*Z=7+8GoNoB^)l+KR^Gve9FuG_wq^HQRnWduJCN|%iTZx{rCA&h~%7I=kMjp
zakD=g{^4l;?e59Z)-
zAj;=>7{P-Lp%=gJX1fP=qH)*1d;|Wt>)ikj;2&TYD18OAd4QDK4nQ(iinoBC$w;4x
z9{4YEj^w`!^!lk5Rw>XaKFOkA!S2DHaV!M%w_H3~oCH{}x^kxy@b@WRkXZ-F|GSJ$
zJmA_!QRy16lh)a1&kabL${el(bf51L*bZ26a_#p#&{ORUuWSbNj@7=c2skW6bx{Z0
z)4w)70{qX&#AV-tIG)KEVK%_8!4V$?P=Px?Aqy~H*Zoxx$bZnDkjp_oq`cL-1L$*(
zn|6ADJZLPppdRuv=p^uV3#=QQMv8p{7-3@=;S6Y~$7J&dJ@t+cr?SCb`bbns3DD8?
zpK_Fd<`E*Lgae(XSRy|S=hG*MbrJhZ;?%8|0o)%WsT2!n=_PPz6=3g}OqtLOoKxr0
zcnPp)p4Uw=fp`+-HKP#7YmIu6hajG}SQfVy&?6O(5}BY^^-w~`B^rVKs|9QH
zQvfff+G?|bTu(UWhXwf2(%=l)47B?xZZZS(SB@X|kcK>X$j249K^++iJD*3O$K(y`
zd;@4=ohdE>^4wIdpra7CT{<<@9r$mIUAD!5e!)ApV};;ng?}tx4ET-wtnld(=zV6J
z8lD9DDOJyD2IRT**!>!ahduZ)FCF-kpN@3(KwV3Dt=T^x>c4yIilS!Vx9oWznGSeI
zfBO?-K#41d+&{p2-V*bqiT%n;tjPHa^oP=c>#pFhcIl4wCxHGQcks>x;0WtWeG>4m
z4M$&F0QjoMg|`#dmv-H?Oa$zUx=)mGg5HZmYpYKK9V4<&WD?|@v3tB`Kwr4|n}!$Q
zma7TQ{19KoF>*9fU+eY`<{AQ?cp;w62Xcv?=PykNe|!fN{Q*a{qTl#~TvU@TlM4Pq
zSH2MOhx`)kxiTbugbuNdgO2K}x8K%G-ruDyVX#mtvW0R44N
zdl>>&j+DjagZ_5$@NJn8U(aS6b)t_?Nw(g93;ZNzc?~<@`196SVorQoIAVPjg}(gW0ZnvtRb{H1HXZO=kpdB5D0_yK%z
zUNfmGz|&6Yslp(yl3KaM1^kS)N1@N1G>ZS6LkhKsrK#FI^ZXb
z%2b5|{jsBsgQy?zdXH7Az-N2y#61eg87dK|19&ndEIkzT$ffqIV4$5UVyB4xe)r$j
zQU~ijs=sig4(#n%%O`4p-_|M3?7$}nk!g5`KqH*D`{CtxxIw?|Ms#r(=mO52Rh!EWWF
z9FK?Kr={nMj}+K_t(V3!wJ;j{O{f
z9S#x7mq4FBC?YikxRi0SSRU+x%LFKI0i`s~r58gUJfh;{>p{+x-mu01&@Jls(QOc)
zX(Nkq9sKkzdCB|exlmbY1#_*EqbX1oFK3LRw6{qxpO_w-`<9N
zI17z)4}u?^7rjEeKwjt;W*ZL39?0A~ACT92{>h7g-*R1Ak3-%Z62(L+VBeRQZk#6k
zoe)1kJqrG}4hQv?zCnek)Su?f`B1foF8;RzdKcD-(j#<%dhjk
z0s3B!b*{v`v1AWC=LQ%~ZqiH#e{07b2d6;KMxQZh1$G;_f6sdW`gsCt%Qgco5U^u!
z66ik*Rz6Pj<)sKI8=?=Mg`KQ-2l~FK4P6-UYjixHJ;eQ#>Mh*~blAy{vG+hPytsQw
zH{kM4Bt)EppxVT#O$6SP>^23xRnkjR)M5RJrA^7~u%Fxrem&y7ZV(`sA_;nGmE;uS
zJ|GdGq({tUKKc45w*V!j(-!T8_@CZ1R`&rtCF=dxePG9)-838q@+m2vkrzO3TNa$=
z2&lLHOWhsdC#vynH2^%UT%=0Oo2FINqxm4$ugxzc-ZM7yE`B%x>y36BlX?IRj`67Y
z0KZ#*TN^RwJ1=Q=Ed^9+>>2$8I4xgx+643p8zgfqfz}DQe|8nn
z_oU7Hmjdm`T~tSgIHnU6G7bQJXoRai2IM^+r56)`UNri?XEDenbKi(n18tZ?PuvXh
z!(Kd}i8@@T`FxU?S9TJrw-*5p%4U0Hg5Klwbl1CpW-^t9QK0{|=7dNx_^HiWKJf+c
zT;|$k#C%KC(XS-(tX0x%!DWu~K_`xmu*e*cPEdw7`gP+jj>lrrz7bknVkpLSS
zzcdl|6H%!(F+?2NOQyL}fi^Zv=(+-SXTI+&=LXE*NC~nA)Xr>;d=Kls;51od0kq?~
z2o7V=_mpieC+cH=f
z6^kMI=krQR;8);3ektTlyeFTe-B%yqA_u5s
z7G1pw_!AZnoQS!_3dm3+^0N2Q!ggX_DZl2C9|0_{$>0A0Fy7Twl{i<&?F%N~0qx8y
z=kx^fk
zlmWexgLFf{PhS5b;rT%0I#OK5a2^XX0zT9M&D`#=>;l+Hwx4Px`kK+6{(`6v@e>8#
z5y+R_y4Q9BQ0H6nma8C-#viji1bLM%wT!q&^IJPEBlr7wi
zAjmDgS@aS2HrZRbMqu<
z1h9*?Gs+hQTBtR9rzX%0!H}5{z}mMk3P!X`BIQG8lz$cBc}HVC=)
zd(qHxS<^)RFd9p1UiEc!lxj5v~Y
z^T!6z1_57mVsZd^ynhy@u$6|A7E~7;xJg4%zDjN5Jv3CDc(j3PLO~jaxc$sE3i@~`
z;Q9Iq3i8rnCNF1D&?GvpZXQlSODo@>UN1{QHmjGv8rVQV!B@ArxcgBMX#11j~P;pDd?iIa#h&xEz?{g9vUFsFnzj~OAWVp|^^(&ImS8Li=pA%%%kdE_S
z6QLk$e+ze;wNxY#a`IbzG!+TdR&)l)Qqg9&{Ak7t3KG66H=GvJkMy7Oj5^HgM`}Y-
z&+5FXNM}adOs|HD4z-ND(;4qa_{W`k^=oP9%0{(}bFMVhteev((Kmo-U#*anOI}DwTVIWi?Byu(X@GYn}%lOZ``i-Fu?S*ATC^YoDh*urGs2?%wj7t~3p`b3G
z`okgBWRy_UxSMWIM%PGX)G~KE`c3fcdpjB(rJiN}mKM%{-#a{XNl6B3I$`o^yA14~
zIiQyoLPbZ%m+=;5lhJ{?qncq*F9%(?D2`AUIm2Vx*P!n(c8#xx_tKGpw2z+yiGe<+
zZlXxbGEh!?4PKQ#fP9w3Cy}5Y1z+B06ixIYjSGpLwTsAzt5BTN5bDF#sLi`O0
z#kzgbmxhiQ^6q*E{Td&AC);8R12qpE9(c48&M$DiOzUGZDp-9q&uXv_tqxELm}Ua)
z|C<=4je=z3mpP`vIe!>W%h$33KR)`rj$2`0K7!qTV{~K^yp(!vcRyOcOHF2%BpF?t
zEZbythKyuA3SM0BqN2TAK^LrV_M`SsMaALJhcEf*hpOk(k^k$OqcK6y$9L*K4_eSs
zwCM(I1zjp~7Jm>~+S3RB|75K-??LD{0kcEq70}TwyBtGvWO>AAH}xAoDGH5)}BvVwwM@CaF^K_B(B3cplsq#)V`S*3|%1L#UTRZ|V>
zcoTcTP3?0u)E58yJ)Z@1v^uHv@!N4a3c1>z9v#O({h@-{13WO_F2xooI?|Bhx>Mbh
zOZ^D@rrz>s4;8KQ2_@-59ZrRr=3YB7fG*wSp}n0RK&dT8m&~?7U0XCglsnEq%q0(b
z+wB-=f4ic3;7U5;u70VPuRDM`H=7M@>!l*9sh*qMUMfm%a?W)2+*H)#!u`8h=gPC`{*LMyezmuDcs-RD<
zhK{7XA@`x)n#`|0)Dr#O`J^y1R%)KFZv9F#CEWfPxAi46xI9$Y|4Pg}mj6jI^ht(q1G{
zk!!j`s!jzBF?R9n6?b8v@XMUpp#Kft3ruTNP!UP4y}*u_jMm+&a!C3>My6u=kex&F1rYt)O0Ws3AowU|w9Tj;$y`K1X$|Z;
zfm~!%Um`uybI)*r+U_(j@}#@xyg~lK8?NLM5fB$P|?#Q=>vu
zroA_N?}##H?|0qB^E~Ig&pGdh_j&&x&iU^T`**MLTGzVPTKBby)78}GVLSERqc{#XXohV=q^oX7}sv=<8(d7v$ymYrg^@
zoZGLNhp(THzrCNokE6pL(1Kn5HQvA4A9i;8ZO{1^d&2WZC#7G@{bgNtOo+&{pR>Co
z#xIHfI3C3Xm?9IxoLj!~Q0wY2+Zi-xKbK#&-}VlE|5xi5G}4YW&B^G`Jk9@|Ct_V+
zKS!^>Js6MQi~ZUJ=kG{n!y)D-@nK>4%^3YfVb~n9&Cb=o+YzD4PV5f09bN0caWvcF@(Fdrw+KRoFE@$^qT
z%=Mii|8H;j|K^mahPfyR(VYJ5>t(jf{xo92{$=@{ebxO}`IY_rwU;Z<&aYt|-Ou8!BE&x9?+kvv}vX%zGGa#8$Cg-
zvflQDGatiaohcq#c^IyCsqK_d7KXnITe!gcA5XUMZo^^oB
zn@$^WKEUwZjUC8`BOoVYsWFQruuOhSNMpraBri
zJS1`T{S-kAw=!px_ex{)MFxV6}O-V
z;!Lz*$>!>$?+lcq)WtP9&OjlV^X#a?Oq8y3Bx3V+CMxkuJS2XUfx6{OKg>JHK!%5V
zn(I>;sMY!7KJ8!zGPvp0_sM~QoEz_+qB}DXPou`YD?SY5u;}y2Cr%7BkkTn-c8q~c
z@BQ@kdB8y9Q%3d)Sqv1-lKUwriGl8OQ`;wiM#<*qf6^JqI#n>N!TW0d--ra^T4-442!Mw}lt#;c9F+WfA1@Q>S1|0`R59
zdWBRsVtCF_$-;$B82<8=K%B%m3~x(~6?h5fJ;dLf?{osg_p@GCtcASWxHT?{2K#QQ
zkM8+#7_NS9?NTqO7sI}yp8N(3PkvW*TnOrF!H(w}=
z#~r_dyG-Qy)=l0{iHXKmX2^eIVIuAUYf05h7~VKv|9#GFsK@a!`4*@*_s%I-9jLwAWzK;Z^z&~w8yLF)t-eq{W36PS%BfX
zCANVI;3sZNQ}htj^-bCf-Qz|W-fq1}LL`QXEYdESsMIi!`X2kht{?^q4{AMi?Fa)U
z6y`a8hCE87>*sPMVz_*JZh9&7!A8+(v&Iq(KQy|3|4T&-zxMK&h&j}wVokpHfGYzv
z$5?V@0zEQTkId8>3oKU>(eU7vo4;RTXEe%MxG_yv*1Tw^f|
z*Wa;Y_!{iv_UR3_{4NaixUKB_w>}0k&NiSf!cx^O$5fZ{Y~vH?JUXHDhUF*L|3%*x+v1cj$u+c7_3)<}=ZH>U+lz
z98gc$Yb)x({;=$c0~7}gf0HLEa}Lg3oKfK25A(nzJdx*C5Qbk7&L-2hFi|y^U!e&2
z>8aw@>VrPukQQ54Cj={6U{iwG!$d6?sXyBw&VzA0F6OX~N%`d{r!^Q}EU{KZMGwP0Pk-uP
z6o=vWT3J1npFmx*tXom~4Cc?4I}&jb7#?NogqmPZ{m_tXF)n4Ik*4noe15R+cZ<{y
z?O>u46|c@IonWF((T^pH`Jw-CyO^#}44-Ea+okjz;uWEdUMjGLc?dIAba7`{;b5ZSesN#QD**#Ns^z9d?~w7F@|hWgB_ix;R#T4396_<;M^EP55xp__{I^HJp38F_X+du{WYPso<}}d@$of
zG6U`Maggde&qV#<&ZTAGFM_@GOb*nkb0D`*8{mN{t(W!iUf3i2d1e~st0>l~q^^tM
z>8DQ#FUw`3z&7Qd8JP^!?k%L>9M3??Cmh==A2N~r9$l{Y(ARqd`P6$Le`(dxx|?9m
zf0Nw9y9LfUr}IR$KjicK{>=OvaBi$|8zUUDnP?%;k;w_D|CPp{GMqaYXzkuz2@a4y
zdl~Pt@+b`Ns^+aYmXF~E3pDSKpT}^8y!qoVpdU^=>8D+QdGUz-*@_PkpQ}^RNLfDv
ziSg_Vv4VaGc(qX8Hk65M9z2@hg!!oQ?n%;NINzXGAE^ECK0G^fdXwQrc>m1Hrg%Z!
zG!7rT^ECj&NBHAT=0bfe;NCg$82USEv#GcYVxsAmlnxBeKk>k|4^rm;>H6~5SN1>p
zJa=!pLfpjRga7*e^NVBd9?HkE0DdSq+rp0#Uu${F?9U;%sr;+%RTq~3j_Mz|SIPfR
z-K$t<^YwrFOUB>lQvrvI^{>yrE}xrZ{(Jc(?x=J3R4;fo{O#@^{{H)XZH{H1UFYxR
zipjG-8~)*F{_XDH8TfYw{+)sUcQQbJ&nvi{Z5TZ&Z`5W^3?U6wqg?#k5Nb;3ka{^j
zg!pr7k`7M|p%T@?!cB}Jw3Ea^UBEJo)Hv*N*jb0s4bSUw(t!8rmgdqwhtSBS{g30A
zL#UYTZVWdzjH=#hu-pVYvE)lV-+({mbt8lg_(rV!KhwA|DiW;xJT1JlI5g53k{aV#A4dU5z3Zv(8cU;VuZ^b|UyYgzz`q^hpB}q&uL1Hxcjte%1=bC|;+XUrF!qRXj0d2xK7%C~^tN|=JbVl6B@e`vl>vRK
z;d7w^(40b!>Cr%6moM8i4d+7=#JY(6C9vx?$N=gku24tE;km$|lV?1oD?Ay=lV`&s!{?OABb0n)@jX(0g8|J>~$kyPJPQ1Yqjz
zQpqN;6V`|_Nd+{#yUxrR^!$pH9=!*&P=7b$0{DS_;z1GMsx6i^W3X<-oA)UuVBa)<
zl|edSNT!nx3&_<(5`Wr&A0u`4@QpzG9Ofd?L0@#-(pMVtV81E(o;TEyr-;WX1bRPt
zB0FCL9&pH$kOX;ardFs0#JxM4lIa8dE>q8KiJ<@dgZIG);71}jX<;Jxz3^r8=lh^n
z$1*iC3G_>fzWWTwD;uzaT8M`)>>&>g_+LIB=<0>K7V}@VhXd-rXY10E7T~|tZja3d
zylt?((iCw01v8(Iu%5f*JZWOT8j?#3zXJWKeDIPN_^V#5vepXdZ^=g2Cjf_;Jq@YA
zPaBEPm=D<7>&eqe#OI|ZEDH8j-BuOspjTnG`sq=iQ$%-*PJ)~zN!x!0^cS>zQ}+in
zyLirn58|sjNQx)wi*47?9b-VNI*D5gK`zq!{DnE#KVP_4J{XWiJN|Vr$X9C8WHP~D
z*s?m&V93wM6IxSE;3rq4@p1<6U8SrSHG_S|!XSqwfD&U)Htv8`rOkJGzz>$VPsbSdM70AR50
zQGa8=SECil#h|Yu5xp%B;`_Yuh$_*?mMdBd-U0s(qpFq_aQb;`5-}&b7mPYw1i5kK
zkrh#ZO2y1>Ucl*ezdbsDA>Ae4h9RyNDV{uHJynV!?>fM?%Hb76d{1VO&vo!4Eqe9|
zAJA;d<$6RtN*ERyEe0GN{Qi0aptj+n-e%BSe(SuU9nkk~?7c1pSX`jGaTTD%b#v85
zi0hH@#%30v)3l$~YyqUn>`@@{M&HX*$_{9`V)tD)&|7yt+(ilSBw`;R<_5QJZIuwn
zJ1lHXDb
z=je)iF#*3>QA(219sWEr5*q~&*uu_`v9va9yxx7bwzpD$3_7|
zc9!+}1NKeK91jAF6h0>01$Mh-HI6HRe$L>rtNMWIk8;!1fzQ7>z$psy%F~l2`4jkS
zc@9ZE0nBvI&J+RpfYdSpPw?}@N8<`{{)P7w_7M4Pey0BO5YTUreWuI+o_qH0VmE%2LL2VJ*<-te&p8HAnA(a5oDKu0Et4S9gxgGi-CzQ8}L
zk}^!-9g{MrERf5Zv)c*gWzG$9ZnpN9Ibi2f
z`z@82*NyB0Q$&54mzV{sgWd}ra}|7Y2suV~@HfGEd+b|soeT6^F2|R4fu1_cIUzaV
zTdfK+nyo_75m_)}c?XzIuG5Q;CUmN10%C<@z5t-)^j
z{X*Zn;OAX$PoNao(Q0*eN{kJmxWytis^Bl}{H>%G;Pd8(OA`6xF17tpY*IplRbc%YdiL^
z0lsmKQ9KL!V?&}+!+;`m`_fHd7g@nieh0Wp{bY72
z#An>Zq+bF*p9OCB?F7`3t{o-np5mzSkhq^{_IB=X1^YMpnfKU1{`gy&5HaUMd!N6{
zfqb|MjdKlwAFaARAvKVfct<)#15Sl7c5wh=4jlHU0ekOwwpv2o+)~6vt6|@1i#JRY
z{til5Q7pj!`jODSGFZ>7KfZY(_{*p~TK^5!5f6PWAxP{;i~XZ6=u$4q?@33rnFn2
z8&J9v2@&TYxntt6EP*#ybngef?a~72s<1vL+?>V=`>BcH(?Dukh}I`0KJvRRcj!iBj;YAM6BfHIGDsoGQgVS_ibMaM%?$
zK;!K__1A%acE`f4hJY4|CCbFSd9j>gQ4DgEXT=YQ_l)Pf(;uy1z42~S$6i3agWO7i
zz$Y1Odr!>y&u2Be76U4_jhy)in9j4I`!S%bp@=(iuLxFD{`eGVs-_<$2=J7htDF@0
zIk&atei*E?NT9i`3TVsi`!;U_`6}a3CmW!X!rh;T07g7-DK>>Xy2Mahp1}T`)XT?t
zfPdUpK+_fUEjf69^?ku2HSv7lXSjUgKL*&kRQtzg!0AmDN6kU6sBuN1EzsH_1yRI3
z?y|J?z+#|Hxk~Cu5C><<=G?tNTaR)!B!c|AZ~5tSK=b}+?_C7)Wp`eSKLy&Nke0F$
z)Fcgic*0?wzTxb`%r$UJ<$V@Q-ci!y9t3J;2|vt~8tI|6%;NmMZZ3
zlEf2<{-G};AO8ycvKK-D#Cx(T{f;d0KFw8NQz7!?#`-L14CLE*M-LP4nSi4k!i)gd
z-u}Q()Wzltd_Khe=xyw&1frfEMLau0-0PPaxpdD1)Z;%XivgZ1(;Ok{FGqEJiyWZh
zq4=k=!2fA;+ntzOlOee~h`gxWU(iO(E16f^n??a^YK!-L1U&2IrA(Zw|MvNlAAok{
zk#nzve4Ki7^%D`7fC@*nBH;0=buong0c#%;kq4!=8D;b~9Oq;J@HR
zfB6u|Ex+0H6Zf{YSMQ9=0|xPC3oQg3W_`2O5BxPfTPQ)Ci?qz7XgQ!;N?~UOVdv3R
zP27v@yX#L9eRQaS{6r7@mK#t6BY|I4Za@S7$PN8_siIHY3;Nk^
zzV|#Sh_BAHwc#?zrJsIo#KF(-Jx=qV5mfZ?{XSi$B^4db|H>s|PDO7erlp3DP|=0;
zLS6&@RJ3y4{uI7QDpJ_~+%PbKicYC&ymm{XA{*7In3&U4l)q|ERh~EwF*Yl14x*)L}B^vLA09jf{XXTAlk@&b3@7ZLFB}i
zrdK=$G=C7XnjAz{?T_O&Z>6Hx`A_fd&7#8pDHYz2_fpY=6pKcRIT>v+#$9GI$f(mS
zda
zQ8IE4w(&W#nu3JG?Z2Igr=Ug8syjp0QIMQ>aXh^a;=Lg^awTy9?Rd`p!*$*OQXZCi
zTpvI|nln0w^lK?d`_1SF-SGjW^=YSm!)iG14LfpAdQs6My~2LU{y|jbAozalz#z&A
zHBi@!q9WNtuWpQv3?f}Ir%q|LK@`k)z^}l25M`H{l1oof(KND@?D3|d%M2c|_;WPG
z%jH?)%SS_Z54R`o52K>B7Oz>(Lmg(U-BZ@ZMnek158cj;(vWgaB#KI;Beu=WlST9C
z$V#d(_@E~h`B4pCr^gSVA|;Uv?Yb0HoU<$I=_4|#Z{N$Vzleq|x|=NYh@!#0bjhvz
z8g#__YUa_=I64|0T*N1Io2WYl9obKV$h_gxd+8Ahs#K4PR<9u=jXb%Dc1*T{&n9j8nTfNJm%_1M?W%U
z$f@}*1J4|%Un$nIqoUA210J`nbW{*1*nMma`aEngB|~EXEmzwiqqc&C
z5+^HU?M{%8yzjlb1b?V6&d>yhtN~Q;xui52`ml|UX7-eWhMZs3S|o*@itI!v+9e{>LyR-f`*5KBd(SB8f4
z!f2??^M_o`SvtCL=h>re{ZwSfLgtUHhWg+Zvb_R*)YmHVLb-{IM%S-Xm^e6y?w+A&
z?tnU$W(~>GIY~vWXWrfnoKHimQ(GUt8>gYri*4EQ$#g`G5WF?WO-C(flkUm8QIXo3
z!`YMx86tprh!chHP8qT1V(;LX$oH2K2F6UUa5?AJqN$NyA$(e?NWx
z?0>e3f+`%mUrt++&;ymnyZuy1NIE+G=d~XcbhNE#pppUS^R!=!?EoGAF4ijc=@cD_
zI|_a2gmanh%?VoTMMs9q^Af1g|GV1xr`4+|$X>4Po-+^lx%tF3^&<&+i5Eu=eH}mw
z_m13J3g=N_&B0>_^;$>?FIfunBKc{IgB|qKnlCvW3XlhrntUzI@j=wP>vAW9myAjq
zS#^(dlF-X)F*`jc3hLRfSTs2`h(s;SJ{Uv&>XsQX-od;JVwkse7=>ma_5$*0c1yItau6vbvu4>t9sURYND6}h

diff --git a/previews/PR1624/tutorials/out/solution_000110.vtu b/previews/PR1624/tutorials/out/solution_000110.vtu
index 5c45f21acfcbabda055fb32a4b78ec353f34fe86..ee861fe3bc5852ff9844f1fa077a79427e0ff671 100644
GIT binary patch
delta 12270
zcmVoBl>v#C0kEb60XdTm1T2%&12_UQFp~`gV*)ldvvLI37Jq~c00000004NL
z6;^pX6kHpKDIz^h$JyCH{BJH44?JJKJZ!H
z;l{DM+YwoC#MAmW+<)^GJ^0zi0}(6R;|ghTE=sBL_47nTSSNkhZYLr;KHi>V3TLL$
ze*Sm3e@$V2nO`F!W5mf>qZX09g;8A!b%@LxwYbfJGya>Zb6*W2cXRIj^mv5GyA3l2
zE>|NWxu7#s49+{6#UC6WBI4{n>6`I=L~aF}hVjc0;UwkXD1W?-NM5Y9>F0Dr#3Svx
zq%z_2wixBDIf(3wxzc*I5Rp<<*BIkFh!{%y@!#A<Lmrz-+ho{dMB4ZN&2?%(#ETareUXF6t3HvXa~%;8{Ml3a0P1=kbkla)
z9z@3D7sPz^M}MSA|3W!AfXJEJSr20Z5eZ+`_j6?sBE{yt+ImM3nRsDU-xiLDQf<<#
zNpXnm(%qw@c?A)PV{f1Bc*=0~W1nCvF+`RwEST|S1|oi{OQIKr{!zn$n|E35SPQRx$Sa6cznBvfxVuu;-7rQ#Du#@3CD@|fD
zH#0n-lXYl#1H)C_;U%*645L)j!qa{X-`3sH7nZ?r)COLsXcxt!J2=04#2FsSirbd2
z#PG0h(0^s6=?n{rNIl#2mtuR(KldN>Q|uhESYTgI@p3Y`*6WkI$#V1yVt6ttmv1^7_q|q*lze#Yooqrw_r`NO!Puoecr1#ty-VTb%!}p`V
zLMY}xQGUhop}1^H=%Oed#oS9{8D57dcGu+UZ}X%0Oo5Vl4*dW17tMkoPl`34NZp`&
zDgL_P?5e=y6z3|gS@SlD;>|C%{0zyU*iyfIwErB%K`&a|Js|I9!O|(`&cpw@t|yp<
zQGd)E-C@yjl;WQKLgFr=6w8?jLbu0JoH}|S${*@W`dB$8*9~=BU{OhVh7}iC4rHx{
zzB>N4O>$;JPU?9a<~HpvJ5vn-ymN+7}og^KDm7k!~11C`sJk;$*C@+C3ig>>Za7e{kBGG6tU`-}g*~rqYI%Pi5iP;b%B?Agw55*cNCE!0318Ezwu88P8l)pF5Pw-c
z{mw`&{H`(k%i-`mMBLZL_g;p*uUZz}TG|A>bKyhop9hFoR4aZmguQ7#Wm<0y@5F5{
zIjLcW$d_olLfI*>_s7$-WYZazP^D%E4H<3`d|Uqv-s@j#T`dQEu;_ioiXPZ+lhhxB
zez4#5$k{llt#(TeM#|)(_)tnTtuWhz0Ug~@KSbm*3~YEyV>G?um$j&GGF-jgUOO%deIB}YQ=w3_@sj2sXg6m-oc#znTS_adrk3J^R*`N1b>QO<~qF0
zg8sSm9_D*Oo@x=M%tNu9KdPgC1
z6nXmU6!ia7`v?0m;Mc6Ldw(mMd=Lrr^|Nb*xd*Gp>g5zM%#+TFKB~rW;^4Nvt+y%u
zyX*tU_6Xp&yuK=UH^oxD?3kq^kQE*to*N+A833k67p{Js~m#3o`1WSpLiY)-wRI7
zBTk6ac=gJRK%6|2fZZdYGog0{nP;K@f8uQtv>^YpYHbaoa8K>)hj=+*L>{-8Jnyq(
z*m|B##NY#pPmczt-SVZlGLq>phy5l$<5pW%LVr{g;tc`cqq~2W?Ky+U#kG00-=XhK
zn&-=SFlWs-+!>DmH-Gz56P~=W1$_xzyX8_h!!G`tw6ZlAmUnD?e*YrH^Gms<_keef
zKXR)yK20%FAMfY@evI4a>UDcQBDQu$S^w$7=FEuAvVimKr0eO^A^u5sud*kXU_YX`
zoo7}94yqcGqe>XQ`O*HWG0f9zMrX^qW5AzDl4isI@mR33Yk$@wik}_{4&D;P@a$>!
zw-V))ED@`bK|Dp4>F+dimYP&8bqRN
zBHUJGF}&At`MKdv=)*2Y?MN=gg4XXNX28=cCkB(_pnu;rhssZkL*DY4)%w~HH{LMu
zz;if#0+(sOh)3l9hWyXFz*h`fO;LFT_*r1J({Ka$m!?_TyAi<7=wo4EU|>#Fp`}FAcG+%EEvP9g&rHldTbn
zu$xuyB?EjLu3D^{27F($OG!+c;SDyyt4m`T=7c!u@x>A0-5R}LUms7LkOwNOd$Rs6LfZ4)pj%OC0;!+R*6R8l6fG>76q
zAAgV6lfg&4uY358m(6f@jme@vYKY95nrmOoMP#a|w8J>OvqXM^zX;=hNN)YJO4yT}i$_1%fG<$Fq`7e&;BQfWldqZ(!wadt
zaA-2adaCd4D#5;(u1o8X-3xrG>QwhR8-J0gUm7kapkTm3qB+}-?H7C;_9;pKRTWPUtthp?KS~^qr&RCF8Hn=>f*M2
z@XkNU-+p-QM}#cZj}r!+;H}B`$$)#KrE>vqeO~tY1n~B+OZ0gHbp+a)rHpNY_w8cM|MW2Y=C+;7Id$+25_Y#FQXucD
zvcU=PKj*%M;WF?w?+$H>a)SAAJb$}PcK{9s^h|tF{Y2ye*J3&IMHGSGQ-h+5
z6JdXb_GT74flgg$ew;Vc0eGR>c%RH$hC}T~b(Cf?EEXVbQv`bFt(@@wBk0m6b!$s4
z@I%U9HwsKG8NL;t&ie$oW~EIsUqFYvZQO+HZi0TFd}fgW``T5u&Oa9Vwtuy(qt6m>
zA(+^-dkE^$kJ#z4t%>1%Q8KRK`QPA7zy(%>n&kicNvLXWN
zwx4{zLk|)9R_3k)@Y&&~?L!0L_qSGF$T|W1vv}%~p>lXfO{iNC1-u_^>b_bIaID|r
zuF?v=qnOta?g{?wPKQHC1|{Gv>e=73MbNKuxk}mfGZ_vK`75hB2)yZZwSC?eME(yS
zJT%>t@G=;G0}B8E00000c$^)Uc|6xy7sn%wEt0KN_BCX0N|v8S3E3(XS)!ssi>)zY
z%2rIKkTpzEW7lR$vP=;n#e{w(4|+(EACm1CJ)iUZ^?sdm&pG#d@AoWsq+38_+wuv<
z;DyC%o3AWJTXjWX^CXLL?2A>`%r_Q8LMHUmPxCB)#`B3SFIF$G7z`uF?lCZHUOMN)
zcNSx-yjgnG4;I5wI6Q{AY=Y6EVZN=DV}g-?Dlb-%bAmCIk#n>Yc6HB~zdgtL;g`ih
ziCF(^ESdfQa<1LL?Fsk`JEYA#2Rq!XOTL3Hxt)(2z~W1fEq?Ig2)Mg2a5ct0g%TCPUCrd73`EXZ+!b8*FB%9L710`!}tZW<23{)kV~mI`q4
zXg0@7aQKJjTQ00i*9i6P0M|29T=s)vf(`F)fh*TG*j~dqd>Z}y+aN2A^|QVs?kl6Y
zx^>WBc4yyrg?zd?`&JFslNw|7uY=;|yzym!;2HZjf(2l2gl3UHII4R^+ZALdE;TEI
zj8*S`AwLJ^FFBfsBc$`fsR`8cROlZE?O$%^-U~X%)aPZwu4Zg3_YBDCBY!>zoJ+m`
zQv?`qq-C{;^K@GapIHuBe_-c714}H1QdPlX8QAuuQNC5%^E6+kZwK^pWl4{sgYJj1KPvbKWkkVIz({=km9-&-urlfrX1K#?`D3
zubd!%jF^t4|7m8RuQUNVW-Uv6P4`tNVe$>@%?cH&k?@cFEE3uReq%g}ZGpbBYx?~k
zpq#m@S1a`WPs$%Uz^|Km*tm$zVm#@8`sXgi>4+77aRT4_dQzlFALMC@ROx|!dVN9@
zn&2;+{X*B%{aA?i(s>(}UHX0sc^#;`v&#nY
zmTZ1bN(S#8*6TeA*52pL8i!rgA-?`;_<1R2>|6=vUGRM>2EF?RyDPO|L`J!PvK{5i
zQ@80Ee1G)W#V!lT6$4&i5!daa
z)%aQ5=Z@#G8uOr~oRyj<^jm{IdQzS*H=p&|kNBOEu9mMME9)HL*@U_h&E;pCB95*6
zpC>OuuRp1D!yU5r(UBvOU{+**x720mlXUe&yAd}<+j!Oq9K6Q6Gzt33Mb!r*?vCTK
z)(^mVxy;gj*c<5bv*|zete7jm2RXO(KMxxrR~Xi6QogDCul$wHHFi$c+mrU2&Ha=K
zyH4vw-*d2&Nk|fq29GekReLDD==Br_tXsFRTNbeIhFO;YMci|M*3}1pXHZ9K-i2vv
zp^r0TC#S*AE+wpCHP(eE#cU~WHD1MUqgs51~2^y&@QoI36~?%f5l#hmBMI)E|1{@rcCxU>Q?7o$5oX`g#5xu+LaC=_}$r
zOgg-g&N=vgyJ0`%*C+2<3P4WPC=}tvK33i%?(}~mE=pwZfrI62t8~~MU~n1Bfa$kw
zqo_Y673E%D0xhRP%jo{5^&cr)fcs9Q{G1FPtkr*KNcH2ZZO4Lt{4!+Un^16M55sOZ
z^iFJlfo#hA1E!X1=(|;#MmVU>b((p0;5;|%_vDDdPFtg*R0sE7y|}cQ@|)-L!T{xk
z^hpEr)sXMG^+lF|;)@s0dLq8J$FmSKtWRyN;ok)P5Pv{yA>y0)y$tva+VrR&mO}jH
zz0DKfVRymrwPXZ;xXE_tsu1?y{^Cx(KjIWccx|Wra+Nn};fB8N^oMN}Pt<*ld@=lM
z3(QKNf-&}nm#F?d?=8qw;{WzmZhf>9eqlblHKP$X&~IC?HT=o~1f6!n&U)Z|LNw&i
ziix{lU>9f;DzpxIC+&HL9C&KJMyC>f9Hz4wv-rPd2E%uMBw@ezrdS9s;%f(~++@Hm
z!K2`QAM8#=NJm6L=5XxKr8+NOI>l56B~7y`{NY#o@2y>0kYz{jT#BbUFjSg42K}{#
zgVWRa{xJ6$?gGRqD07LWdf8?v>HY}3`uV2KZG7KZKC_i`)K4pr&78&s|W@
z&OwOgtFPP%Hud|-|Euf5AENEl
z2PT4l@3hlAJZoXn4Y?%WHjnb5yvRsH0aTl<)vbq~w=Q2zklrb6BLaGqPaY%N=zAtN
z&Q-{O#seksqBMW6wloTWhxo=szrZiU$2l(s%)c_!NO>H^Sn-MaadJ3!5!q!mM71e`
zAr39O+$ew6vdcDt3kz$CZ^AA+uPr@}{GDZgIqAHby5X)T=sm$39OewJ61UBx`^f3q
zvzg|jLuRl2D7g2AGXFBNi>v36q&m5&Al!yH3nlaHF#11Lnybd>J!O4`LGO+ue?9h3
z0XMGs{WQ&oH&PsNztQIk%K0yh!s{O#RveHQf7ZFBSFIpz5_P+T+xB
zjFdHKZnuE)?EwgahK$BQvYgy{Lz?*`aUu)VKEAOwgK}S)u~0n2Ni3Ym!qkD
zbkCW7I%@n_e=ojmH^q5o6<-wxntYTtorhlX?;uHfKgKNVNTIw<7qWKW?(oUA`t|@B8s9Ze(8mG9yIKevboK$Gw?
z7=Qi>00000004NL9hP@I7HSm7jU*xEzDA{pM9NIadR1gMu8gh~*&-=YGP^XCE?sg%
z3Q0zYq*Np%8f0~Eq7s>T6(Q^XPJewrXFTUT<2>hi%-An66lX@5p1lt0nPVeN5zZm0
zJJTb~g^UC#>)8=z!oVqwKG;tSWmz-lMt_(!A3cxzvay(!VcNO|>?~$9JD;2(c#=0l
zmIJ&SXnCXzb~8QpOqux+CV#Jma}MI9=$&ji&cR|9DRS*yhInk7xFVkZ9AOp@SR3^~
zR%*yO;|)Jyu@idskTnt|;wCVUM!wmu#n9_Hs&*eo{u$1M%BzUG-Ee8w736uUpMOxK
z_+x}A(U_jqfcTO7Ywg})9;H{9pQ2$em~=d+0$!Axeyjjm80vF>#CTJM6|-`nS&OXd
zG~!;5xOB)JvRHd>b_)1irM$Wl{IXu6O&Iy76yDESL*DtYLH{uF%>NNPssd)X8!X>}
ze7dE!+@BHuL(i%16fo)d#0DO4cYjNrlo7^doiIMz1r~fyF4Ka(dw<)HWN_a5P;Uj;
z@hr5V1M3dD6|Z80c)p7KQA-d%;sKXM3FJo(910$wODGp7Cuq|9c9$u}{du%p#S3wH
zCnLfnTxqg`I(5#(S)OKd5z<-MiWpfH%2EKw0QTTEeMQ~bFpF>>B2f3rnq6H9tE=_rx3*=Ca4pA<|
zHSXrSB?x`~;h}Gu(9Z^hl>Cc4-b%7ZM8K@hxto1k0GwV#DRN~;NEw#JIG&pvvW1Cr
zh_R&w^2BOwYVd+SI)B|Wa{w&Zs(M9<>g8d%XduQ<9pzba2dwXm)69iFLX6|?QH;xY
zk(GB1vfirVr(YmH2~)MBIxDa-{?QKE&onYW4f5anIL)%a8%_%vZ(|-FHcp}*Sl52f
zlFFUP=No?2QU`X05BKg0AWu6FkF*rztm@lkFX6wugUO&e{(ntt(oG)J>dq})NBza?
z)>e7s;S7=!sHXEAO%yl}nkF>t4uQVvTYrH){ASI5M$^6+4TekE!Y-!PTCo{&bXryk
z3-%>S^$*h_%io#}GRM4@>$F>6f*gNvQQSS)J5LWTNkhDui_;~Sp}+5-Q+ot#-q*d~
z9{SvU{mS9+Yk!;5Y|(}6uH?<-1pE3M4r}W`mX+Y|=fJpIXT$Ep&YD%W=YV`K-DpcF
z?UR1XssM1x)(!f6Nm=^X3C6RD3LJ=ZWQ3x={aul@Rat-f;RG7!sTRH30ghXQMLDQoqQ{I_Lr>*f`tq
zfh$#6{EqN*njgGhjrgg>@AFQAhbsIt-+=RyN;ywq7d&Rg(1ZUnYdKaY9
z&yxinsDB4O`O<<-kT>_T{g?%dW3#NR;qO{6D(+14vk8AD2;S^|V?p!p+_;fl1l*u#
zo{RYIQw%Bet)wK_y5uYfBV8>tp?u*_h}u5J&@0H
zJvZVV5h!5hfVDo`8r?wUqOA&D;GSm{?fUS)HtZLF3G~TJI=cenT2I-jouK~T|DnE@=aVHL
z?|)b@jB%A!HRH68yj+5V)EEEx^HK9vs-qF!3`@$(n^bHEKjDO;qvU7oyi<HAu%t8*=@JkS^LkrxOoU
zZ-Qm6P7ELDZ%ZZnHNtOMkW&EF(GqT}QGa^(_}aI%je?y#-z8=yB?PDRTTPOD}fZ+hf2YL_cP$}BKC77pmd;xlMR@EHapoL
ze7S2*cM1H(Eidh{r~WA^&2Iv-C7JZo`&IUw5x5t!(B}T#i@|rYZYStmI}-)RzEgj2
z^Y6$3Yc?zrqIDWIw)D}yDoKogrGI{bLr84$LC2~0yr)tXsil*jXTFF`nR|~NPOW%_a;$e3-vW!K7Ywaxj6sJ
zU0W=LAe-c#k$4Pxv7QB2fakmV+3A1qD*kYD5Bvt2#9K@u3;kW5P3u2q7@jwac&4e5
z%Jd(LJ^hw5gzE5+ZC@7szv@-RUSphvlYw3vn0xPXk}C8uwY(1_XnkiQGpgtwT+~*l
zb4X}CT~7bI$A8^6rM~x)bAQmq5KOGr3Y~&q;rBpu>bIGiQN!$@pKJf=al}39v_>qR
z){)*SB?-T3?zxQ=Cw?-Z?m1ZdL#~Ji^=B+=x~PWYhsTvsy)Ed>lHO1C|4D!R*L~f+
z`_8ZXemmiy82#r=^_?r}oy**+X$8h-wRurLEDVwsq4#t9?WNMjh=04x<+;dykdebF
zK1T0{aRm?E*JkHbx1SidTccNs{ttyo>rF30|Kn!xtM&9BvT1xo?`{xpld?NF81wN9
z?TgL5I7bJnZ~qMpy4SX1%QNVGJGV_^#|3auI@_7%xmEJ=x-RPb8UvT&G49`gJ9KG(
zBFa>@Z3nBOL~nRNZ+~#qcj*}7ziWB^ksC65kE!bvWU;tLmwHf6wuZF|`W(Gwy!7AA
zm`*=N_o4pE;6o$&r>;qxX@Gs}+OGlh4n5(IT}E~M+H}&7`k$=z`}@>4B`P^a(-5zi
zzx&!AP|ZJBl>z;0vy>?Mw`@OT&WND;eP$U#ePB`3+ah7Ge}C$LEA{o%I?iC~FO0cA
z|iug%pp!PBw@BzpKrXrt|Ud63(2^76x~{2wO&Z
z!d@o-`mZor^K@yx6H&_<^gg)`J&9jL|Dm1_28Gnm;!ldvdXw@md+T7{FZs$^weTKQ
z)_rSP3`S|k4jd1GlQR#*I`BT(sW>dI#ymuxb~WfB|Npa`>RXfWG8lig3;+NC0001Z
zoE4XQI2CIf#?xV^5_0N5(UDS!P)T=k)`=Wau_Ov9q8*kT${`Y^Bqfoel-i^Zk#x^eksC>MOccKajAbli!ag%vl4?@tSHYvP>=g>ER%UC6in=hN#fDrcO}3-mdDjk*CPtU
z|9|}`{YWqmgRKABZ}8_)KJ{ltj2Dj-v$E0?t~}n$99e(WhKGNX`H6Xl4S0Cp%jMFP
zdE~ro&v{?S!Kye(p}v8`%vV9T4h?cRRM_s=H;c#Q`_i%5Iy|!8uYGH0Lf)6N+r8R~
zhmT&Y^aV#AS~Ffe_~OZ9XwtRe{2(4pS1#(=k?#_}7-xPbmB$>DGK;n@3Hb2~jyv12
zuz9yI+BJoRTu^^a(9sGO6JMW1`fCfVQyno{PJe0*`a=xOMLYCT`e{fPc>9RQiSe>g_gWsrxXo|Vv8fhg{MYSqZzL4P
z@3SbKx{-f|r?c)CVJHm;y<%2nAB}U9+oLY6p)q1A^`>?#jrLw{_)Hyzmfi9J?HLq=
zORrnL2&b@Idi~HCAz5#0w^We(vW27e=G_!Nx=c)L)aDygBrXbY#xQYvV!c6Q4u_5udaV7VMLJqex-k)Lxj?Dy9Ox85t*y)L
z)>jcy*WXl&BJcMpJNIC*Hig5_i$bcMDd+|NRM+>XV5ewTBiK&idI)zpb`FJ%gX?V?
z>?wa-m2nBuP$cJw-0E^thQjps^{Q<%C^T66v|a;+pWd<|$CD|<$l6%18KO{o+aa(!
zjKOU1Zrbn#$v6pfbPvv&Kp
z(g-)pSF0g?$h%2TFIdERkTu=kpx9~b>FSX
z+*Amce$!Ya!ADbzy@ehO?nP~0ZyU#A!?Ekh)tM4(_+0cKp`ip%(}a^3TS;)g=|!I6
z9SJt~%dg1+(SdvJP}Ex%(&r7W8*Mmj3~nu4aEimSdG8*ZWO7hi8(N>cjYHYq))aqN
z!Vi-jucgit{;YLAwyMFK!;%@h{ey_E&Fh}~^b;;LTQkB4!VAflhdv3`JnsK2`)6|{
zhXYT)`{)vVg$J1QhLbsW=Vb-PFBoWM1aCZ0Ax6P9yPPL~(=)R)M(~q#^QweYMVu~CFpN84&2vAbQZYEH~k}x9ixA2&#Uhg
z=H2~xaMLCVrhA?>^^y9;NuS76GUst|a`&B`r+JJHUU`}3L%8bYr(*Y(gTQrVlf8(=
zh;m_vQiTM=_bh64f*5?5J1}O`3Nf6YF`H~OX-F(;LMd+w(MppuW6Atadsb3fX3Jq%
zDmlYTlZWP^x@NBw9%mYrTathF@UYcaT4Ywi;oF6SWrq1I`uww94%kajdcb~N1}(;4
zvVUfcxkU8+L`m>HlLB*DH|fz73P!_oZ)=D}5R}}18r{U=ltAgRdJl*0p$BXH<9Xy(ZGG*o?tOP4r{;qG@F0MvjGiv
zL*h$jX>Z5OB0dz=Ql@KnnS-`!-J^q^RRtBLf-n!m+3yLf+Hod-jByKVf26@=gWdac4U
z=2EC@n&ps0>fKbVI;X*t!G$wo3lqY%Wn&|9tm9eClaW6x?ax8;zVhZ@GkB!NI?lC>
z3I<5WoA
zXxwFgI{grjMThMc&;7w+@!AhJ&d=x2{6vm(S79Nl^myaSFj(R~Ye7||82%4bD|@`i
z9O<5VR}>?LVrIC5T^$3Dr(@3d$+3{ILggp=91MSS)s)Z6@QD8-EY6eUu{H5(78PXm
ztu|K_P9=O53@`a`)3di>76;f=2ou=-Gyf+P9u3>qSiOlsX`hF^7*O*^et%LgL)v%BYQmR8z4fUti4PtA^7WHF@r8bcfVj!#q~CRM
z_R8iIURCb%^bv_*rCk43#h5~w?W*KpZ5rR28Z^dT5<^QgB)o1Z!OQjYTd#d!G1Y%M
zbc)$Tl0#MeZ^#h8`|H^$yDE}X|5SgqAez)^;Q6s+OLYd0U6&$dqiMV=*xoI>lR`mC
z>bHf2hrgwRd;AP3>_4dNA|QFoSek9S`9cg2TkjipQzg)HwG5D6#z8T|ByOo0$=fR;
zXI@I=A%7}VcOl_p)dbD0>4aa?&7^-?Gwm7JySt?Ioue^Yp_$abgF@DtTJu7Ze?D2Q
z
zGogdzAC<0IiM0$igb7N1lKu}q>M|;s&f}Kb0>4$NJe~&^EV3s$T~kqiPrj9KsQFq_
zH91e;LVe4Jq;5eNur
zNhWh(d&j+~lH`I)t%2NwggBdCi_Uiz4`noBKa${^}D?z3b~G5+Dfs#s4nv
zEMhGmQQ{Sf^++{>yx%_Za~{(i2I|YJNh7ED`9um4`PfCABV@_7*pbde@5vePoBjyB
zPhhbum-#N0)V}$ESm|_|5Z^?E-BtZfLY*);bpK+(SvHI=MDLpJ<7a&)xmP}@%z^g^
zqg~BJu5HE*t*BR9b#4JUXmLnwH&PjP_Q@!#o)&I+ZAmH{gTaWd`2yNo7lgl83}@Ez
z&Y)CBFZ?9`EWN;Wfa9VDnbA3mrw-qJMfD*EnX>s~i{yCuefPU_d@3L|K
zx;NrC^N)y}%b-cX&Q^0z5Vk>g#j=8zAd}n;(np~OLSDX~pH)RF&K^{Ml+ofV{L21Y
zuFv+y)H3vkj11Ce@f?=5J-`%4H!B-q1OLv>$N4J)Po?0xkc@$&&2F;q{aY8qWe5RX
z)YmFH?NJO<8H*yKZx&i~3=#_;gwP+@et&u`wUtvl@+IU7ka(+GPLtzk`@-k>(|z`W
z^^&T*XA^=oO!3Dj6b?f2gWEwwX3p4UGFlXElsTEW5y?N-Y5A~L(tV{!8k5J(s+H*$
zq^0+lbT_lHU)zw2-jPrTvYgQ=j+U){=UyvFH?I&jh#TW@`>4QsNf)V&BnXpm-FYC1
z>=m2rvfp6?$PjRn*A_{RBANJhRvdyFqrQRstX^hkO|rz&c@C$`soS;
z{ROO$woz%8*kuZ)fHRi_pmxi-=?`D7b0Gh${+A8cj*mg|r?ZjzTeH<^QqUDIfQF
z4aY?|(3*9EWVP|k5VPNk%L|S@k@raZx4bM*ktn?5bP4jx=Yh#-r`?@=d*+^UF`%s~`D5DNEAv&Px-tRL&3dLz@(l37u$8#C0q
zj7EAFPQFn({2dY=u%TVVy;d1E#3K?(Q1CZ`o#)IIJ3+dScd27Z9_yY((8QAAG+f_E
z{kkdp>~f=vsa_a=3uRbkULR`OiEM-JhLE0m
zDcL=L;NC)6txb$Jua?XSWe8+ep$F>TPcLZqlO^=_k7sr7chL^hYo364{4WwRdPOY5
zyfkE@hT4iB_d{jL2f=ha3G?2E_?HCCpu|Nq7o`}xkl#Ny3#FO^i`Ih>Yjf3Jzop*U
z>PhU9&pE(#ZuVcCR`?99{%N!en}=>XJ)^sMW+}#}1W0pF-khkUzEgvprvPX6xVMuY
zTNV*)d$y2E;p!^#ggZHXqp_DxEgU~Uxb3kKptkd#u@^GJk6+eWKjp>IqlM|qk;kg@
zJ#xB7m~vsYD%ymZRS|x#v>EwlyTDnIjPojDZvmmBBg=IRH^|q)a*`w-fAyb)9(pO;g?N5wUYl=V<-ZU0
zbAsEe?E
zAmk`wp1Pdk_Jb1}vn+K=+y>t7WA&>^2kJo>SvnSXG*y&&ynlLBW^$^*tv!0
z#R2=>g8m`K*1(;e8L+CaD6;}@ycqGGtjUQ>HtP`W{o`51Wd~Mm1NXG_ub#UVJ2cD%)>AM_?S1;BK$;^FX&O_O$0g(k6cvrJL4vxx
zfBGr&xE6a_+QS+0kOqnevOm2HP?Mp|;@nhXkGWnPmpwL?e5u|a7!blI`bORUin2~}
zoq;d@W$g861DL*KHa_<@M5orT4ObHDBDZIS-)KU+R(p5SH$39Kh@7C&(|U&w{Uhi4
zf!O*~i&J1Nsl8ma2sra4=?R|FC?|`Y#V7Cqlgbu~HCBZ&o?BO3MD20*>SwN#-fVY4
zZ2c8@fNdq(wU|u_(YBJy56FYO*p_W?3trjP)CDOsP(CsKd85MZWG;#BTsEd~mupM(
zxFr9U=cV^#;o*toC9!mIn_Bx%TA^7K3Z^aNm3K+y23>%~BJimh12)T~ZMypcD&E?@
z9nN)lnlnV(=HZ|FfJncvTXq(I*QG22-zMyJ#g-aJZAD-HVI1>0NzOR9K0i
z&pZ8>$F8Danild2xYND2qIEL6De-!UHj-01C_WO`VK>LOU)ar|Nna7QHCx*wuGcGC_EyZr?B^+9sN
zewypTt>FPhU9|s#l=fBQ`if5SM!U4L*J;*1p}Fxu=@-R3({`3yll?O>S;Wt`udTR&
z$B5+tIa-u!avJa=;--oHaiFN_$4DD~-rB21-pkm-@|JjsOaz#+UcBmOt7yTkT{nZw
zT)uU1jET5xQ#(I13XN5$SL0$=PNA_7yr0TH=v!EDKW^9NLVih39agTBMhqKY@^;k>
z?L~BN#};+^dXL;^?ZPw4&I86B5w;6}3lixs@BHlTR_;6Lst7FqDgQrK
z*$l7Nn7;Je4SuHr8C$@P-}W22AV6oFYE-#2KMwfhTnzpm#f)#Y9_**+;P!~rm2Cz#
zQ=>@yljq36ayNTuWQe0W)Gj%J7t3~`032i_iYWdV=Z^=s<)PaO)oad@Q+Evjy-#9R
z1re`^TmigBqDAqE;P1ruHlD*38w7j*9}ypAO>16*M&XslOD>tyc@NFbFh&N8zdBna
zTW5`#O876^=7ql)GN}&AE4JU-vQ6Qi=!9&-u{&Gr&%EVhxB10SMEsA4Kz;^KrnoSb
zktW_|0{`(tjEb**X_Lx};)d2hS9e6Yr>gLll5TgjAMw^dE`Yj;?DBcPtF-Rj5~z%^
z#qvH^eRwhhQ+tJ>E>HQSbW0?RYv~(nwBkqhV>S~bOkW6V`r{Grp&B?Vy_VsTej6j&
z!q7adf}rkI$BXKg3o&9UH0l-0^M)=cKZAJuv
zfnoOI^~LHdP_{S1&YcX_VW>m_p}LHN&2)aL*rKj|bCJjx^uXxxKpyoX7DxDA-Xr+J
z2Hgh3?F~|E-GMR-#XB0X1I4k027Zd3L%%Q>c6PmNtGPv6qI_=Cw)djHmgqtgUV8aI
zfb!xe&i{!2)8Ux^mt0c*BNrRe!S6A3Md)0O@Q}XLHG0l@OiJ8C_
z0xZ)@`LbVsWymy|Ni|v6B&ZoE0*&@Q00evi$%2SNmM5m6rO3|7Um!Uz)0CDuJ|vPuSv1sZ^}Bfe)+T&C5mG5(WU{uqfA$lwC!^&{`oG*doUexky4*->K^;P-za!UuBHnGnO+=IcmBnOohA&!31AJm_nzDX
z`?OdOr>}~1)*^Nx2oytqW(?^F+Lg
zw$*{?fVY%k(u8qt#Bg02S}y2}GSB}qAWbh%LNK4WVjEl8DAP#?kAfX&N<7f5_uSAZ
zaISj3{W%wJ6XO5W5
z2w>Lida>-e0nRO4?w|6k**a?zL<=1AA!#r2FsoQmAsxfvt+w*__xO3v`$IHp!>uW2
zLYV(BtJ~x36VTjZb$1UUbJAj2C5wIH-jtD|=(ZgwDE)HCCNi*z#285v^y8OAZqCkx
zd!{`^dJNUX^Ny9CdAp{<8ssAXdYg7qGctq<=Y9@oRwKRTwckBvHHs3?Q2zS!F33Si
zbZW6ALfIGv>>a_Sh2mxR99G%y9bTp*klmEd=6Wx<1FRU^dxaCj#&sMbRbF3Et*o8B
z6#qAw$o;xiO1opjWz0?O+bYKkL_jwLr0L#aoL)O`VxxED%hdiLfi}oogS^ab!BroM
zglhHglB3aQ+kzo&S?K#8}VxhLvCnByjMqxNCy%=(L`g
zER{mnQpx26_Q49}Y1bf$T|>?w!WX)(G5EB(Z&c1c@W@eiTqLk1m%17%k#;TkPKyN<
z2!vHm40q7C;uIh>XvO6qQ7c_>QKnJ)cq}A;y7L4-WG@(11mBV9={hQMa~&w8hp;W^
zK09k$@gF>BM8KQyccZ|a8TDNRDefQk!`1eq+eA#7^7LS2u=QAphM)U%p_8thYCshGSO<^FogLra
z%k|CDiGsW0+-&Wc6d7mZzY`(>ze%AsAH(UJ-XQOpURzvPejLy9+Iz&Z>gk*aWaM=@
z)z&Ku{F!+iZq++uD5&!9Y$LP6q@+Yot1{A$D?4+s_|fRep*U{GtO}Gfj5*W-p!xMg
zf39sAl{~+s-MbfEBHLA8rMYp+_ZIxqih&q`C{#Seru5p7nUJ
z*DT;q{`@M@FX63M7>n~sZ%yLhOppWk#Ya_RJvsi5c~%clLDLwH0-ANxI&nO
zv(uO2EU5fbqg*RD*T|~D83PQUcs@qeQv(UUckjxAJ+#NQq6@YC4tPL2ozM$K2M29v
znrG0Xp(;r%un)Freo3@V)G)qj(!wpBl35ictNM*Ez7MCZ|v^w5|9OqzIGCa?MXOC
z<@+z&F&$l_e|t)?^uT(?I^ZQq3;qFO+H@fiC25+@fyyXTgHN$x4bfyJEb3w~Uw#Ji
zwR=;QgirHDJKz3>ry*`TB@D+k{~g-_$6X%dCT;R8
z!m8yJD1_R8(ZhYhwE$2Z#aPO2W@t0StD7`YFrs_(-*9@eyL5-{B3@O_E_am|x5eI_
z-{bw<*hb;aL%qSJGRz))#jH`V=nxW^@AJ-^YVMNb@E=w=@7b@iPTvt9m(mS9+F1(s
z!bcU@Q_@6cTi)O}2K3g_W6jtZS3Pt^hvPUA!>>Z7)L66D@;y)rCSkw)-Z649Brug1
z*oD~gj2c+2AWS|q?mkj3`A5rD6<>;Tuz2k;>*rzosNO(~XkQBQ(&IK}9&-0%-uokZ
zVh|Pn9Ew_wnY&=|SBnGAW_O(wO4llYhwg}4wT9|$D`8GqJYw;=wP71z-dOAYSv=r};s5rzC3feL!F*V|eDEr-E!4tmS68N)3eo)uixG#r{gSBk^jl)A?e=VcLd+2s)2bB?R(n
zH{U*0VWy*=NQ=LB^2anEguT=_uMs9Pk3Xt|3Kxkr1OwM9uLL4NB9id!7$UD*S~`6X
zFE)wmhsAl)p0uHmUPE4{kSzz5hqzj4b_sM(GFXZ%euxW8M=Vz?*XwQ3}*MWRjECE
z=Pl%s0FE5UhU<9M-%EX<6qItqe#!Y#h4VOyu*bVpT9uO!p&+cR5|6|TLFK@3#&Y`3
zkGJ!+)G_az;(FE8=I;^RDe0|_1gi?iwZsSA;FOSIG51N^j*JnQ#$ccO_^?Wq>^E7y
z=9B1#X};Si0PX7xlH1Wro1ov}%B(+L0jk{sL@nI*0H*Cl@>r)(H57du?!YG-+bY5R
zle{To={!(l?f2YfN+cZ{4`vElo5jcsyp@)1YapOLF?j^mWr!bpV`gIP(Cg{+yXh0M
zxgzLkvl=T!^iAjkTR@(dRwS!}`3%q!&D)p_y&W2?NHVT^R`xN@dqj6~JTCT*EL#G;
z`8X1~*+oFzTA;dqFGnDFl=Fw1wl4paONb2)F`Jb=iboXg(Dr3jLMPRaA2{}(M-Pqf
zzP5mo0_EZ2{$`T7-Z6n6)>q5h9bmG8DpHDJf4JZ%IP2%?*|Z}<_>HJHu_`_%icvR=HqDa9{KcRMYD9kPh`RFh-Q&_3Bb|E0pL(inJmyOe{s
z8^r$feZ)MFd)5)Und!D_Sm5nVFrz;Ey0b&BNl4h7`UnvnK7dbZ9*Ht|eTmeks4;lv
znlc>B)uVxT{gq|AML*T!%^%$~M<8EPv|+CN|046#|BcMlDLw``DKTdg33J
zB1s#9nXNuH4TZuWH3hyzoA_1#MC|YDGtn}=)87j5->c7gC?3PsE6RV)_FuS
zOCQ=PkiEZMLsQO!%P!j%N=a@ySt6$s2d(n*l3u7BZ`pb&QLK@RFdbTnapLp^pc~eZ$
zvQG9uH^uTVDrPHQ8
zMJ6bC99bS8e_yZ5(mt~PNr@W)JNXGs@>dzw4i{~URgr%4!^!E$NhYgv%~v~#5rMC+
zD2cmAjElG%*g9TQ^o2;?#CdSMH2GCmf5yEhWQYv3L*F%yb!pMgqcNJnt)7{(c2+Yc
zVpO>g5#dh&OiX4E9)cq(X1u#>u(-mLohLmSC|zQ4nK2&J;7ylqt&%zM=lnlj5PHhr
zbC(99qFpz4FH78Kk$!*jVi91IAFLWGMd+bNXv
zcUyR^&$Hz^`R$i5+oWtsT+E5?(UWjsY@^5R4jf$kN9{J`*7S0EvQ9X3H*=d
zsE=~M#%5F*9>8jEsVuS=V$4D69kO9ua^A8d%D$%R@5H$GSImIWZL628G5He=h-*)d
z_k7i~{HuecsLk+#!v+Cf07FR7_BE7dj||HFSd7l!u)U|4OboB{
zYNgIzm~NLcydD$gT&)cuI?Z_nINCqGos@l~VMAlTUil7C8R#|}7NPL@8ZO(tK(^6e
zG<#2Rak1(V?5;4L@}td&-+5^Hg#2OFi#mc86UWF?nq*1xY!2Mr;+Aei=eybe^L2ic
zJ!uEJIY_*W7VmiK5&8(Xj!K|l*==y^eml@!5h$eWVKDHP`f;8V={@5OL~jHoygy1Y
znfy~2?s=)7qv{<78_OO`Izo(Gf$^83+=$3$9yDKMVf{n?cnxh2SwG`L?DlX;06oS|
zm#%m0<6NI76-loLCvtJiK~5$u2Zom*^Vn}(hok$B-iy5|!d=s;6A5oY^DOOtiuzs+
zzh^|6TRm1?_gKrdrqT+5z^h_@L89MBrF@0ruFGT2u^U*8{xx6e1k0MHwBO!8<~)y~
zu7nd3zq)m-jyH4*h4SkQ%WDKpC^mKaQt9Mnnf|4qYR_Djy<1RpFz}?hts_Xw_dWM)
z;i!(f^6L23zMmJEX67kL>U$v##&yaQFNK{)zP`1qRJ7&D)%^+uYF<}ODm*gH=P5KG
zXW7Y&h5~BQHLmKN8(v;~v0lG?X?&5r$@Ch)7<~)~?|!T6BWbQmH{2Yz0j?AH)`=3I
z>Tt|!j2);ZO!M_wFO>xfH}*w~eQwyF5@~WwAggi%0@S(YH5IM=)6QAFt{^+uAMMtM
z_M}zk=X~Z~u-9V(o$DB#>C}MVT#%!Y6IOFBFrqnG|4IXkTA|3Rxc)5sRTY>XoIgai
z#y5d(-No5_)EGqHu8xn>3GL4AD^atpu^#F$Jq)Qbn$#mWL^3kj6=FP7lIEa8o0?OY
z7h0q_zs{{F1Ssi$zL)2`N7@Z)DRTddYazAB37*JWgnR(H`(ni%p|ow$BJ)c)ojIsS
zOoT%WfR@EH6z&yzFh2dHh@oAAT$(S%Ymr&6o1w!S)2rrdt;xyIGid|~wjWt)g_i8L}n%>*&?)<$BoK}VO8z%)uaF6E07
z<)DZ*{2ff;PJD@1QIJaopalM@e6Qqrxme6amT_5XU
zc}^nFYtc_W9n)-?|W5fR-1a?y~TuU4O-uxtc
z6Cg-7a@CRo<#)Pcz`b~_XJAL-{@-Z4+cGnuddELFPKE8m0*at2TDA*`bjJ72u($D?X9_9@ck
zmhA2>iDJWT&R7P5=fKk6ylGW&_AJQ-?52i;gPAb%>l!*2DSOVBTq(Ny<9b@2q_c0MK{?JvVrm$Chcqo5b`z@?u!jUB9!?-}
z|G&Ik|Now2Qhh`#N
zI{0*7MX+uP(=ZeNV4~+_>_avIEeVI(RaD6<-5;!eT=wrxxSg!C_e|bB8L#XY)hpL|
z4}7tF1XykX*b@UH1*B@T4?NtT_hl56Y7r)4hT3{A?$}wLb-gX>72zwdZ
zPrm4wvc!>Y%-as4_^^)vu*G9-S{h^~w=3&Pwp@9jt$Z*eUf$Rh^`esZ64^U;vBucJ
zZZsi-a?v;uF5&u3V%H_DB*Bur$lkVrpR016F3y~%4b-Z8-AVHO7Pi+BzD;K_%UU!g
zb?aGnzlVJm`o0=A?&ooLBBq(OX?kZ}2X9ck+uA%WW5|UT2Yuy%kXBaFhNct>&)bM;
z0r|Vac5-m)vWlwEjeL(Ro`l-0rv3F^b%A^Yxz5Cg;z)rR(qU=lpw7we}-E)*N{*7QepycCOERAHX+UY?8rRTNAe
z@{fXV2trw6D8Jn05wRC~{#JUc`mwEGLrZyMk`-oaIi;A_Z|ZBd?TNYi?;w;Ws@Nna
zxzo9y81tGQpnkc4Pn2&C(F=!tF;z|X!qruvFKM8wcrJNM6WusR!YfsZYeY$LMzRx-
zu52RcV|LsB+za|;(>+ZAF{RJI5)N;$rElQ0p_(Kom1^4cP9OcxallzxqeXlhlBJXa
zPSUKDra|0q%7V^s$mYObiZ@9h-}1r18TAQJJHGt!|yYU
z(fu~zW)Uu*Nc_aH`!-iJHQ2H_rq`Anb9D0)i^o{WRL5++7%5S)?-;glx(vN#@v!nb
zH|DrkCEQQiMl>!%a~B!(Ws9Ro0YevG{P_vtY-KZ~18VSV0u9ifDZf9bV1PO8CNs+8
z^Zf%}O=do=Org(D8tW(?JKxnAsmqGQ9KM&9uE2}DRCr5}$u;=7+9u9Y<4s8Qz}J@e
z7_&miJPrln<{mm<2MhRZf#&3;Ngh_n8u}=8aV@)6R&#DK9fXg{vP^s$MVzQmv?z3skl+GT
zJ$iL8uwr6hpvMGE?IWJz?6>yRVNub)9?dq2G{%17ufK{OnAkExtOdGy{J(t7TCE|A
zJs_Z16(J+L_4#jwANkg0=lKItfg)`4hsUqJheam8RE(i^JAX~kI!c^V=yqMmFSo@-
zYIS{%E&Szs-{Wvr(Q4h-I-*P6^c=N-?xrCq6x>acNF0r>z2X7G!-+82>0D+zqoqJd
zEO!h@V$Pl;-THchdpzT;IZ$LevCDqB3!%_>l{{>$`=x}hJlJ!oiTUic__3w6Z@)ck
zF6&nq+A)6WacptrMg*8VYViK-ju<(79eu~}OVphGo7RlCgAV04>f9|B0WmQk?9up1
z*G0aJHJp|}K6Oo-z%X{#YqYAv%KxQGZ;7QwC=%s>oRAp?vB0Y_@f2bJ8u2
z_13EPsjAe{?nfNg;_HXsx&@%HuR`0CTq(ynW@4C{4t^y|zEfn=KLk#AiOl^hhoSzK
zxEevZg;9HdG>4#ywq5}wNzgZf%-P;D0xo4(UWU`&lhQ*KikdQq$
z*Q4XBd!0%|qj0)}xOo0$oxQGHb`|=E-PGB3`A$qb2jTbfh^4IlbQ_6whzUEBJV!q=KJ?VALr$IcJD2^h*jxJJb4#@zzqOo+dsE_0$DnPL#
zu)jR{z{_$m=Hb=(AaYxEf2p=F^1fHjjYw=
z#)Mp1YseQdZuemg4lpNMiI$%*KgIkB;Amh=6;NXj19&G_Ef9mCj;F9*Fqvpuqv?fe
z5&CFwf!H?;fIVy(%WWtSv?Cq8G7HdEo_o2e^KtZn3O67
zo5ALt>{r{@rDG#qdYa
zLG0GAo|USlBxQbFVThkMId?qk|evp3sUa;
zy()KkRljI5uhX
z{d&Wc@rVk+y<=9Q9)*tWm|)ycirH~MKIoXQxRIY#(JI1(or-YnBSC@RVBP)N1A8C^
zn&OGd{&ew{n|jz@_IkZVy#h}Ojqyw=g}gh3cOb*C4F<|{&hN|g1U>}o(z=KnS7@uX
zS_eC`XqF&H&7>bn=oI5M?u5lgU6PrI<9syTM=|G{z$mg4>KIsI;Pw>YVqXb_iQ{!c
zFW-=tc(X8}*>f=aSLU?MFvVOdN*d>jtw(h`DdU!*klK8!)JYd?2$u>wR}YcmLAsMW
z?NEGUS=#gV2MA@P<=T8|>zG1bfZ1Eoc={HDCLHgUsjkcFv-Q6Qdg1P`xik6je4@{X
zFX-d`r98QJ;AeVza6fB}uUwYIrdJ
zH?h?=OY88v=v~s9L|=>duQ{Zf$K=3{k0|CtnXi);gAcJzO;6NqnWIl{i0&EH4?Qd}{pP2O(i2zIbHxCLN2Wieg+y>apHE`XJJV2QADSs)=rB`)=@r6~Z%!gb&G`(E#Re?HH1KldNcbDckY&l>kyYp=cbKH`iGH}i2X;l_v$
z7Z=8k3H<&#_lG(AX;7Zw%+Yi5y|Z6fW-)daud&X2VZlT(LMA%P7nt=g#Wut6nANkj
z*|dq^%=w?gS+l?3e|ZIrHyKzFCD=Jz2d-IcnC+ig@^`Nh-+wF5G5-rV*RScEGfw9G
z@?t8#8EhCQ#*1-bEWWOuJ^`2j#5EHd{zoYP5Fsyy&)RW=BG|e42RQ|L9`Fsogaj~b
zDG29|(#8Wmeh1wG0z93)-8KbzyB*v#r>bTMYtF6DHM73o0q;;(H}^k{V6Ir;{BL&i
zX5)4Yb~^ayfwTXPbmq_B?0Dwv+`Qd<+
zT?oRte7E@q1Re}>2@E>u=4uXFkmcXU`}g+yJiTY^=lz{M;aS;32moD$tI7X=}j)1SRw=C<(uC_3jnU
z7+xPJyhFea!@I{;pMGtF;fWt!_2rph_>h!`&)W?czV_gINp2j&@7Aa`6}`f6^PYaI
zc?}pIXL`x2rWV7Wr$h*(Jq6m8PjRpw!#{kLZWjRzd%{~{QH|k6ZItEHM{Hd2
z28M4HPCj!e7}{Vx`%0xK4BwR@o0pV;;hPGp3#~3XCWAHyxr?5#;?Vj|g>7N5?JFcEJ2g67SQ;i1(pQ+DuRc#Qb1
zro3JzD##a#8Y*KV)z4|R&gD$x-0zjQyPJt5-DUIzpiUjdmXW7B8K^Wxs=a!gf!3`I
z{pl*qM7126@&=;}l>BIEjN)|$D(U2?ia5+biw;z29P?wKq^y1MYGDkN@O_8;R2l
z)gA0n*+4spZC(DPoPnYz-VKwLVL!QUD=cMaA`3Hn`4t@uq!FsME)e4RymD79r#;A5
z{fu1F%tUKq@XQeChkDcb@@y4MbWir%bN+ZH8fr~C%*J3Mqdhw6r>dEVOc7yTImASU
ziqAbOPce}bPh6BX8;1Lbo%hbs!SL+$g~@9TF??cuQtK~C3{NkSYYPB-nP)uDgrSd0
z*2L_JSdQVP1Mes9I%2rQbGb>U5)5~K!QQX;0>cF+g<2TbG5nH+)Q(Gg!OuLZtuE;I
zO_p$^LSOn~(V{Wnch#NUzuqDY&uX}-{`e?{zY(;Ixm=0i;ahEPzI_ROMoZbQI{Ji(;!3mxp2`m!FP`ZAHkgi4ixHWT^nl(|}^jp5Z5-b^nC3{SZu
zsUUF$!*`}`;W*le;X7Q1kI&1+aB=E=tr8^+XF1WuYv2w2rgm{MqMd>CPbeQ0%V(g+
z(StlmP)9tw+do6Z=b+;vAxV6`sEPKdD*7cTRQWA#C?HXsOK|g-`Bo*A=4E@j{
zzJeUWK)%uGeJY*|BoY-+u`!N`a`k*YC86%Mi`ET$24VO{jkdA7O&EUh)E;f$FbsFn
zx?89Td1!R!9cc_=pv)AH5p(e4UVJ*`)-?voop#w17Q#de8jr6U8DpaG`~Daw*qQXj
z3cEmFftFlN*K9Fdp{Q<0DT9G1>*jaxxH6Cg$JSpeEzswmgE=l}G0{?d>l@7~CdxAD
z9oLq^@O>MdS*l@QtQG9hx@QHF&jbc*I05h2Jh%
ztcE$G;fPnpf?q2kRT;>SWFYV){2_*ih`#mLIga6t>M?_58!@~f;AEMqG=?wO>#onp
zVW3lmr!P|8Fc3n8+V{3Hk!MVIX|)7~3tO5ud!57ZAEi?h7vX#yb-Y-{a|h}!u~)M`
z66Rt>goN5&41cx$RU&Hw1Fg0_64QHtfueHHN|pFCkQr-~*f_>SqP~OtFL^PXTQ_%L
z*Le(2=4Q%2X~b|JYrk91YN7v&3tD~_VK|RJZ&VB9*ZiR*zBHSOZUh($Ne?j4^oohL
zLokmNaCM(|Jq#3-{x&3uAHx;HRZwX%h9|uYP*r&X^Z9iazY6qghjV3^7tGfl@A-@*
zn7{RFVizve#qbs(GrohWOq8R$IWH3CS+H=}GGbmX{H!m0OB}%Ya
zsK0mfcm3NX7{1VfXSLB|4A1RSs4IbbN{*efxB}-+m2W^KoWwwaPt5f8!(23Gk-6Ky
zi-}SjlKov_KMcO@mG}ttsh)OK{i%=PMtKg3)Ibb>wot!u8{9WOVq)p{GckNokqbv>
z9?Ype%lT@MXG2fCk7g|cZJO|G+kX%GFFrxC25MiI6+hTnYUsD17x
zhPxaIGbKa5UlvP9GC=-zegd`OJOg>ZUHjow1p|GF>y6+`U?M%cH9NAPzW;DZ|2GnI
z@1`rnOB_D<@4tWh!#Vd36=30k9}1pM@M9#vQEh1U=McP9{$1~?bjg{h{@J@#>bKfI
zc~`0ZcfG4vXY2L<`cKB6*HaybhxPBzzptJu@-r*`XX{D4QRm*Ne(>FD#@)a8@1NI8
zHI`#`oj;eaJu~~W;a?ohU+(^@z+V;ks{;RXDsZZI|K2v%5kyXj6}`$BMna9dH{Tl{
zMlL^1TI7BWqrQ&=tiPYtLi?N^A=VW5xX_UjA)o}XC#W*cC!4B3aLVf`pgOF_mq2jiQtFXzrx@O_*cWtC0u}?$fj+g1~|;B
zYe&Q_F?O|=8{}0_bsdQPzv^ORK3tUIW~ZjBCW#9f$v<<$gu!uIfX6}
zVjY!guG1@^PM1%2pMDSdB)Qic-vjyAyTQ#7K;N>uS5gG>cWE7wp!Fv;WC{A4cR8+20=h%pqA3;ZiawfDtOpEkWGj3I@^jlm&q+dmr*SrZ<%4+k
zi`Ja=22>2*ZA=F}Tg}L%U10C&SYWdeu<}W6+*&}+2fie0KzXg0%dMdI^&@}ALh#4U
zRYWEFoA*xX>eGNMm0Kp7K(A;^>gZmeeb*P@M*$O6?jOAf*x_KmYl!e`rpIFiC}L~i
z-UIazPfzG=hd9hkUegPJ=9Sf3vkm0&x-WO|0-t>QVcG|k-!GJflEz{2?_gl$2d=zcIkh_Kmkh*MNPl-5dV_
zkV{s*P?QAzja}yjF9KRM3_B8eR^3l`A>vcB%{kEkdM(YI{bfK)vGKOa6KJ$Dbr}4t
zaG&Q$%)2YHkIa}rS1GQDS`YLEJF!fDK#hzNQ(|3OVOViL$W`}TYc&K+K5)I234Wa{
zBz}AZx;@7D*ASqHXNz+Zpu^!@17co%cqUl61km+S-^*iwt-iU{CqO=(Fji~;btj!a
zSil0P?d$zw0mxY*_64&6M%#o1Y652IZ`Nu8{bf>p;slliH#OhuUQ?^&E~5c^vzeV&WZmwXV@K+Mm!)^?LUfO|)tzs~`=C-3B=^FSNQ2iFtl
z^$2pV?}s=d%(J%y0`4H`Oq~Pyv5)t%H-dg~8jBHe-UBB_BO?LHHqtIzz)p;LRh$8O
zq4UDTmO>moFZBddL7sL;brlC-_;ntCqTc@cu|KYYUN+k?H3z^xR!75Rus=FdGD>^nw;s9(aMg?fU5FR-ikm5(kul)=k}eBpmPo
zmxmdVXR1@PFCWOYzWX$X0)2IFPP`Gw-$(TX<^nFuc1u?T42V`rs|R`cr?fH)K-LR~
z`!@sn_zji4159gI#d<(bHuaa259F2K?{tGWFIuieB07Zp?jvyow3Nr%>%@I!&P-sa
z1Ex-8Z9fRud&c=XF<0v8$)VW>Kq>~_FW9{1#4Nk
z$~U+Nn1T~*Kqr2D!AZ=wa^2PrQ^K$NgqSA)@x
z`ra{ZT3Y~1ZKBy%0sr<(iA|yLZ0s8u4y$6)y{Y3h}2=FE@KY
zo{ys>wWUG6S9;~`IKUi@$9=@SNH0E^-3I8;WK~}Uaco{UtXKqb&?xz$-+?b`Lt0@7
zemcq5+%TZ`d_k3Gz~7l~$xdg%58c<^;tcRH!;^-Ofj-`nTVw=&SZ_K-5%+SD-Pasz
zkZVULgn9s$B|hZ?&DqP~fWrOrgX
zJPd+$2!H-!UcuLa@0~nvlBlO;lWF2}kh^)U7EJ;B!&?MXMF9&u7YjH6KWBI5!xqrn
zDTYoFcJ}kud_O?YL;JKI0jjKZuP5F)wr8%k^?|-3UG6X^&}X|ik4?jRx#X~=Ccvj0
zXEi16nVUz`mk@KXCtr1sEg--7lCL&^V~Zht!1UcRuRB4XzPxu_1?a%1Wf@lhZR^8R
zs=yz!p3y_hH%rpZm%%cZCFXL@jr}t_RG2TDw32_z&$^$#jV4Fo$v_
zaX(J)lpC}G{@r`Gx3j}~=SISXiTAE#Y>U8U;FrXnJV(s!$+{Qg#5;zTJ|5Z+m|(jj
zOA4?s*pZ94{}tAyixc;gsr~qJ;{Gof^?OhBN4j=+0r7t2VD2d%1-=qj^=IOpYNYGG
zzYnnSWc8>JtQ)&)I&dD){jVbX+Ck55lqKLUpu_5K0>pW1$uYMV0{OQFOk)hz_1LaK
zBm4~AVmEgJoEqM=n*_K~>HSUO9?j@j<0=jqpDSNw2llpS1-ho7K6+KXmKs1?w0238
z0gb-978rnD>Z+CNh8ECG^HcqF)P|S9Z#R{Pl~X=fpZ5mky~D
z?}m-YrfeOcM#ESv@eZ+n8{kEpgCj5SA5%aF2`KOr_mJ%7%NK~aRnDD~I1chIGm`i{
zz?}Xa8{`0Yl}R_`1DEgQf=2rm6Y+U&wajn@
z{ud*+E+e2nN#^(y@19d?hvm3|Z<9B8bPV{4TbK3418o=Xd6rnWNY&Tp0q8mCUZFOE
z9Ap2*OYD<+z@U*l;E$*)CPx4R-foGx1Nt?$-(Mp3ow8neKk>d*xEq_g2Kb*UEqaK1
zQN;9KCGn2@CRH291b)UG9s_?=@j!PqwVTvy&fMYqt|W2K54kkN+I{9E6DAS0<$JGzY9
z$w)ZWqqe}FjMfP9IF$_yp{w5;oR5-*(7+XbX<;)m(o^h_zLHHw@%DKB8!8Fq^&RMo
zR)SwAhnti>9~eXfck&hUFM2GCnUS`a7DMt4pt
z3+s^3o6AlzEaU-{xh~`W{jLG@`ZK${C_iFKpf5OV33t>cFNd0`gy@c!fw
z;>0dUR9FllYMt8t;lo47v?^#r>18sqUlqMUa#s?V})Zz6O%M1_d2I+43Ma
znT+^f+m>gikkNV4m6j`t6cqZoOw=rhilP(z`jtpDbV>U3QUraWx~;fde1wLi1T%s?
zf~ko6>eV!LT?&#*uFWokevFPts8a#IdK<@SzON{#hYPuQu+z|U*Nb{LH0fxfTB3H(
z2|6D
z)YnTwT2G8w^MgpJujiO{7o4;G?zG!&Q4};0TT&Bb0eRh!mzfWK3^<15yq#z$G|lv<
z$WAg+(J%D~{0x1x&Qq_Eg@i8lIIXtpA3#M>X)N}KhEV7;_M5$@$f&RAtg&GW6$v#^
zZ)u&OBb{=I(EU?Xl#vz`Y?e%@n&4!S#
zq)8OxLmKq~WL#JJ=4jLY!BHN0z_Bw46Dys>u37;H5V*vrW
zeA5F+NQeK+lCHILx_D4-D|s?)J7UWN)3~lI-}5Um(QhKfH~RPA@6m>2lDMa
z;KJ~uqf47_>HJEhqabeuxzp-&bfLNaQxMGGQg+Xed*GaS)PE1rDIP@3#jaha!5{To
zLr<61Ad>d;4bqmSpgZmg^9{9VsAG`)CJg2RUwxLU;5V4#AFk6ad@Go_>J
zl?xA5ms8MiazN(+F7RiS7GpOxfTDW7jjU%ILgISLeWPA*zMr13)q?s2B=)z)Hc=3V
zM}Bf-I~Dbfipt6O(otpifb}J~FA4{y4|zWrLTArT7x+T`+QT?+9chMozT!p(tYkFu
zzRZH8L_t1=m)h4{gn2pqa`BSK6l5~u*0S;k4QW?8Y+3>NYD#U8pgvR2H-^be{#L>y+FC&;Q7C8T#
z-GaHeZG!Cl{yK2>Uy=U)`kNikyq%MWlc$qcz+cPNIFv9rHp!Z1Hq?>{~0
z{rT%(@ninnnehMihW`&viEfyWf)LH?&s{I`+xVvy3-)iT@9&NKr}Y*472E$@Ut(O$
zXBMRJubX6lmf|0N+>Cf{=Q_H-nlDVG^G=;Z3+TIXrr-L1G7*D1
zu6k^iiJse)rmx2^{L?y3kuabuHuN02Ji|m^-8weM`C|Bn#fp1kJu$p$@zyu}?ikKj
ztJJ^V9>ZG#1P<_7VE7*s8xz~jFg*F)i@^eY3?CJ9^?ap+;TnE#MY(Ydzf-T!R`LSF
zcMc4hENI5?7=tYL`UVWIPL1Hpc!J@vj=ZuXO&I?6i+GP9V94V|Wkz)veyf|Z_U9c8
zXI2VL$=<+l6M>Yo!9f_#&}5g9i^g!v>~#gn2^hZZTHQ61ix{qLWf^!d3&VvU8RWZU
zLpvPRD7;b&mwmzYai4%B_JhZMQ^#Bis
z=L?s&6?}yH6)lS%tzaVQ4;g#y?=X?Wkb8kuKNBr?k*X+m43D+3Mvysfa@nSpdF?8cKk80hKY{N6jj4~$ts%?x6o3?CG`suYG)nffB#J9wTpJqAsr6(yQ5-=%Asc)cQ9Jq!6O2836HokU7-A
zX~{&@8-7Ht>R=+iSUl$_^uwURLMgUNCc3)r>$AmYn26k!e2k63M95lQ>2w_v(I|q<
z++Ze>l6~e@eVU0ZdE%qh*f88B^t?yDI)rn3A5j_lqM{V*G
zDq^7em=T_2sG~++zi*5phELDvY3xhFa0~a%mn|WmO^)TV$;lWlb7+d82>mECAQsfq
z!9YrHgx8agGLS_~=AgVA17XqrmD}T)=-gH>H&Lj2&2r5#w?GU>%H0!p+b}%)w6&U7
zDAZf^?ll$2L#n^vcxxa7g{QiH-UEL2l_sW^UuB>NKOL+?k3v7RoZRqvf{9{kd@)Y2
zGaQT)aDcqrjk(&c?uEQc8V}rJFc4L9;Ts-D2HL@4J}ckJKy@F2I4-Ed`M}Lzs?;)3
zn*PTrH8Bhi-fquQ2m4|mZHY9WV))SFNn8Q;S7k7>Kfes>U~|+oC63Q|+uf^~s%L~q|Wx;SKuky8$i3~*X
z_ugwz23Kzw(u@`bi9bp%gn(u^q#2`k$&$5Qn&JU9=ea
za9*z^UZlKaAXRit?VdRkdByhMsuRI*0pmR#?r9kQ`PR($i*P(5FwmC0$74TwGth}E=fui-E@ef)8<-%}&eJRqr<1xHT3@_@FZY+a(3QwFi%7t^M%R4L>Mq(i8$A(*vz+6Nu5_g9V
zF;Px)imxN=hwfKfkzT0J)1QtCKeXUpC^#%j4Z!fnOSM|}!F_W?NGP)=2g6xP95~(=
zFcCgzyigJHd^&K(Q>B4{Hhg#MK5~zVKAlbwErGccA~W-$ALiI(k#59Z9K*|B*r=u5
zgnQ(8r~w)B?OP!t$^iM#g$dN=^9&UEYRkLRl?*fx|1p9$fr*SQHXpbQ_5FuS`oEEw
ze>de4FLC(bzrO$W#X0{DYu$^#eS*%
zlXsQkf7iQ;b*^6jtAEM(>v}5T@UZ^%`S;aRUh4OX|EKjN-l+5MR3G?l^4s0N`1`Nx
zr4Yw4x6WV7x162(v*BMH&41kervm?}z<(<6|4s$Yd^}>?&H5RAN{w4t$QVNctya74
zO^qSDANrk=vt#IE?{Jqm@L!*Nnk9gJMk!T!ozg6yk>4<{krZHW{WrTWz|YO-(BAlS
z3_aBErQDbqLuXAi{MrB?H?Z`}{TM^_6ZDB&;AivmWworJe~0}^nmb^Bn0LG(;7j4F
zMS_60k9kbXg5NJ6kGkA|^^5aM_Ll*lkg~&b*78~`pat}{mRlmH!Owf;*!md2!E0XM0|DuI9QO|ca!IH!r-Pn}+`)o4K=af`
zC*uLBm3Nf60e^N}(7dIX{Vbryl&gjT
zp{J{NxEbiB1=7(=fiCMuuU-Q^`F`?;GSrhkdsO!r(0*5j)Q166zi)iG5AcQr*^Thm
zy76Rq7}SkdTk5AI(7`4j+}fe;8xw9lwF7#ba&K!A_~95{ubm72H{g3j9Dtv|rlzk*
z$XPWkh`7Zj3O{m#{P`1&!^HmIbuiN=_Ro%e7+(YO9fG0a!LZ-UM2$n-LGP<)1setA
z#S3@wco2Skwa&i;Tt8uUpcLd0d+wjn1$p`5*0nc)_Km!iZw9C;*d?$Q_~(jSIhFv;
zFWo0dtRqv$mAD@2l%Ck1_!jc1c4^YP2XfYXK^+l57ns~DD*?Ha=~|WzKY30jS6F%!H3U5Qn{fJG~g_)$6uy-Uss28Z8GF0iSyNLB>0^tHV<$j$n@_`Qz%K2Hb6YkcUlE`p0s7X@Yz>S7by)9(YyrN$3pp(VusCxltQOEu
zelzG^YU0u;Jx0z!!M;rS%@fV<2Ycxc0j)Zu97s+_dNMXO9=9)pcuHa0H1Ap`O3{;Uuf~t
zcNpZNwa;Zmfq&!BdH#!lM$Kb3M4k;bnGQsJs(bUJnnCYX2j@@)(2{J6x}^YRb!F1W
zz|VS@1vbRI%U$=-kO_3X?E2_!KxbJ9)*ZbTS~>)BIqR!kI{@RouXi!Q
z?*Su`$zGs)WA$c70r}iI?UMoRj$P3v=GDkk{_0hL4i5)gP5?G|U8#!#`R9a*Qf;Vv
z<9WYg7C==mkLOE3jzu^Ju>qbk3k^^K%+lJW+6MZo#Ri25ED!3i7Xki*rA}lnK!MvT
zoW%T_9+JqBh57!P#qI{NzmLSva}oN&_hXxh`T3@+N8cLI`t!55`5<>$H2v^A(2i0;
zO~iSPLH12U5J%LWyxjqS79{nVG>}L1-pkt#`ehj`y2N?+|2`fW3HaGe++jD^Eo2r7
zGe9qFL8#DbqTgD!@~4CRYL&tU4#3dsJibJ|1GVBN3qh}t?S$fCz^322RU+v8z?
zwOp=-M4q{J9bUX3-}cS3BLwJd+x#=SAn%MG2)F_$p68S)3+NOhm(c|BJAY(U7y+_h
zI5xBk(9>tM;x%Aaj{-ITda~)WcAk(|-jLl5;=HVJED=-(`Ng{rg%QxJUAJ5(?kh89
z0z(NfZRYZRKfr;r_ScEI@{*on&3R=!D?Bc6UHId`DWXTpTO2?0K4
zt+*rqbqw8P@_%Oo`dsgGPGY`QYIMCZ06*z2--TQtj=t$G?yEqr$tzGI>KNa6|ITyJ
zFL=Ri@*3p##gh9j0!AMcG~N~Ly(PE3xWXBL|WNYGE!M_NmX{<5T*9?6BcXO*`bmV&*|l^?T*0at9n
z@=~GRsvVo!iGG_
zFLk&=o^{cpYT_Wb7MHml516I=Xpoo}8Kr)C-GGj5CQY>vhoR<}Yzf3MOetFV4fq0P
zr1d+%PdE9h69)9YPpJAV_)GYjVs{Sw(7h~;&H{gW`>7p|fcEOVQlblf*lyZI6Zdk7
z#g}|jkZZ&wgt!7$CO!7)0r@ey+zH~I{ifi;bsy}kBKP|eecR3^dS)BY?Ike|H$gt5
ztCT>@*&XM82uy*$2cD~ABY+<4T)K(i)A_a1^1$UW#f@cmO3OcV7y*k+LQ4CE)>H?B+t`=h)0(^mo(yRG1}1Ae7d
z&Vx?Svl2q52|K$5o4>dPtMY`lMPN1XvI8Xe9^@7Qv
ztM!5Z^(3nSanDqTXRadV;HM%5>%D+{dsclh1N^=M!UxQ>l4ySq`Xg&UPRRr9^`s&@
z7tp9FEVUN=F`5_y#C+RBy4fNCaq4ziK57G6xu!qO2I!Fs4Qb;*4_8@E`vT?!Dj2DP
z-R{mk(Zss5rW?FH0bK$pOhwpNwWD8n13-RQ>s~Gi>Zsm%_scfGj3*6Cq=DaJ!Ahn>
zJV6|rs)_rNc2IJ}4EW{uZtrJ@^)f$)2@vmH$+%9wO9VgeR2nh2zcfCdngp7ZITbPl
z7`^wvWii0&AR8{?{#Ve<6ejK`Gs~&9#QlF|+~+OPAIWNA#l-t{hGAVg4*V@#bsva#
zs=kKrkwL(QQ+4CZVBPRTKLZv3J^Uhaum|)4###LD0{U+J%14~H?)*KL%RoNfj44mR
zx-R>bX@sAVa`ruTfQ+%7RwO`0xwkiodo=6KW=CPbf-6$B7GUplj<0V9>a(Zzqp>p3
zx?O$ZWWXKY9E-Iu~N__&%47qWU=-1*7nfL2J-uk@c8L^H>
zR6`FwZ%@c9NJH+~xzdLab{GQ{JGeGu56<4^$fs
z5clFzgL~D)JCZ8a5YGhusq=Etv(OLUHmUB20$fGiH?s@ykI4@-8rW}Yy;(%$6MHPq
zqzSNwTO*Blzbw&RSP}qu-Tj9p(ZB9>iz*iZ`kwY8rGx(n|JJcvAn)O=*hcJk{0FDW
zcYyUJ7c~ewEcG-$u@8399vO|Gzx6?UHgP}7>ZI@z?`H#lEsoPb+n+vVs|$3yWKVzw
z&;mSbA3Os(KEUu0vHu-k%wPQke@#B^%B#Rn?UeBwF_16j5|dVeIm;ViEq?`QzfE1H
z_W*zBEPj*+`f{H5WgVciBX#b)0J|iS7ItDzobL^=YXo^4P1K6GS0?k~BlQ9I$}ibY
zyypgPs9hxLEB~2#ftp4}L8U?(&x6RQYlG)|Azv~wRXXSX=p-3+c8__+CzDZ)!m?|<
z7s!awW~B4xG8q{=_b}v(TNAxW{E3QbnU3V
zG&O*VlJZ|$2N_Zk{iyQcB1tNG(A#7`G($m!Q}wZxuPMkn^Oe$Q69rujsN@e%rlUHQ
zPdf{a(^0dB@@?|~I#O|6pzU##jwCZZyP|8c-Z}nTBkOm04KxXy}#b#}JfH
zLkE}GH>kAI(C0GMhbp{uB&c2{ET9PM^jerw#KNKtk`JU5SwPFUhaFR3JyJLo+a
zBR7InkG08t@E$>}RYkHzk|QW;p+#$D6baoW=YL}nAfba6T?Tx!!{{47EszsvtE$9J
z0_r5xb;(YGg*=RMHM47K`i4>O2X=8cpp)9UY*v08Mh?!d;~RKMD18+_M?pRb9oJNr
z2^b}z^f<9ENGz_F7+XSB>IT8(th$pT_&=)-WO8bRB(~tsxc93fj
z6>%39W+-V;(9GF}yb9>YsEC9{dGNb+`xMRV1qD6fLJqF%G<3!B;?^4~bacN?q`^80
z`l(+yLNJwv>=&RzXUP<_>kWlH82UPCiS4%@b}|}Y@BPsV;%svxJ3I}eqO7n|uJjxl
za&Xgqe=U-Zl!SVdWT8LsCq0*T7gNze-R`8`n`CroIQrrXw5781%Hu-j-cG>p=tGgvHxN0HZ4_M0D1lTm-kIlUd7R3z9;
zEmu8DM|yWeLXON(k;VguzQ*mND0JPc_bp3Eh$3+%bPG2Lz29D~ItB5x^fV1VGlV!r
zmu1KMk&$iG$Dvi9sHoDg!+sz5F-hkf|D-}iCkwLU)@~a?8%{{AyE8b9T+eA#M872=
z8^y6UQvC=j^4X`Q3Vowi$hY*xCJM@GB(Y?T{TM}0Tq{!6Kp$o)2^H%+ry#EGZ6~M4Y3TjRgGXw&
zz_e$Ye|%l+426cK`C2s#Q|QRo=0wpLorZp(jPa2ZRD@Zn%WJDqkdk|w=XL0J
zL1$yuao17QzeR4LGxXDIwdd2vJt#<+td*;(0dpu^yyO@rg^6!hRp)LvDnPgK%SS6mwf
zvA7ncMD|e8i}96`5?*w4uYcGy3+{`Y;h(`C_eas4b3co{pnl&&Im?fCkWkGFZlujh
zM#FC_j7V}6NVm{LWkayZa;)Rq$kaOH-$d_FA8e|@c;k-

diff --git a/previews/PR1624/tutorials/out/solution_000120.vtu b/previews/PR1624/tutorials/out/solution_000120.vtu
index 458210b30d560b8a660617fd982a9bd7df0cd1ce..8e5883e18e3a5d15667fff733ca75374bd35a8b7 100644
GIT binary patch
delta 12291
zcmZ{qWl-D=?B{_NC|=y9xKpG!Y;h}YE$$R|mmgBxN-6G8+`Z`H?q1y8ZINa7p1J?r
z&D@L2o8)z7l6;bUr;`y!lM!o+-XMJvOu~B0o;3GH5J`~#zt9|sho`~y?J^PIgNg9J
z%{z-w$4hi+ofsS)t93xvey7ji75yfNQj?7wO_}*@
zT(C2fM8Emt*I+03_T{we!aB?E!Y9EE9d_&{B-nRleowr6poz)FO(paJmc0E(xToXm
za?LUIh70Nj#Om!LH5}`b5)j6CHsw$>;O0ZDs9`I2DZotdqa4ALi|i17PgI}IZO|5L
zp8J%v%{YQS_b3wS-^DLR;f95~c__z7Rzou@8fX=>VksRcl$sNxB=ZB!mdH9sVlPIK
zHc3fR`a7;BC_QL+#L=>Bk3^cvM*9Fuj;7#~@1eKcJ6cDk`vJmu8sNt5(xWPaUwUno
z%+GlKLvLMjgP?khylUb<%6u#7Uk|^1E1G+s@IIkd<|ZuC$*%
zy?R`&O%9oW=k819a9Gbx$kyZ9WF2wFj>!BmZ~ex`
zDcj$0^Q^h);nakt-r5Zwjle50r7Qo%LIc+aO)U))&|w2do|NSvDVtp
zax;mF5Pw?F=)DR8(*0wZy-sgw>wUz~W6`B7F7i?;zOmD;;sAlz;y!4J-U$~W$h?g22|8Y4ZzmXcp3o)@7S|7_X1&|@bFld9o3i@3fu)bw
z=WMWLh1>rqGYNV&he=!<_GeUq8|t8J>^jT1MajYO&}jPi>z4r!L}&@0SSphy=?=4)
zUyl+?*Rxe4f(!E`K1i^sFlz^wa-)?)fLUjfE&s4a4PgJ6Y&YN)v%|sT1f?|Au>Y{4
zC-CNyk_lz6xlzNO|H<0VUSNPqVenMulsP`*t36Sc^0jTkG=y}904#QQrJRq`uaq_A
z;}T>RZ*)UwryITAD>EP8>sx1Qi3db&5)-qb~!R&f55>J9&yc;h{7Mul>-b#c36n=m7(gb1gDF_i=HX!
z6yHilI3-{;E4(*iJj~5}iZ`*+XHM=vvhKen?2fM`#&5)yv?pH*8P*4`FAyPAoFn84W=
z>};&(W@9Z8{YRcYed(;z?ul~qk8zQ#_W@v(?|O&#k|R9v8|GeVh;_9#Aizt
zUZqqzGunr(z%*Ri7kjCjK>N>oO}lS(6;Tr#k@h)($ckmY76m+iGw^gp{aZ0h75oE3SUd8_3sB&G+ZjH2eNuyxEz9P_-GGwKLNPIRr6Tud^PnNW
z`-8N&!UNfWWLADUQkT|M;nf$aGCskn^XsI*YE;?+ay~BP;t#Hi0-O!#?)ux=A3I5R
zTtMT`=BU?z&D*$z5TgiIH$Q$uIg2(sz|6gqnl>rM#RpC%z37GMs2k?LE0fX0ltglX
zOJ!u!ihBWhxSDx1k7WO}+~SV}-EMffdE0|>g?b!5wX>y)2-`OPO(jKPxtbH18F^Ic
z`4>bHKV00dU2)-v0&sscx~Fg#b~7{n2foIv8BHi7O0xdC+cXfb7&6@Qm^j-dj2RZn&Z*5-iImS4sM+F_*sGYl
zdvdeHY?<&QnW(|?nEGUMV^jQ<@gm9Q{SkO^+Zl}c80G%EyuX}c_~q3v25lHTu%SYO
z2X$FU6?604l(Pt?-#DQP8&*UHT7u+)>_$IKb{`Rk3nfl5q?}s{@&glnhH!NXJ`=Tz
zH=>mREFW?a{imDHAZ)IPAY~ohb?ku8>#$8inftA!Hv>~)c^6rJr)LZ32*fRWLKQtz
z;-m^*!xeNl?Z?!cJ*${87aZ>cnR=aH-{BFV!0e(tnnA+S*R{`xOS#>Eh=>dCW;aJ9
z$-%&cQdnStmJh=o&f_DMqp7r5KV*}tZ5+?JE;rk*yW85QP*G6p?G<6vEYGdQS2y<-
z@9(GsQBJL7OH#}*c1Y2uEYr+WMi3l`7>g`F@c*Z6
zy1(c&w09wjJZN+4I8^}fdXpfj(ZT6P8VKQr%^B;@ER6DX<9rg!x!5nKF>^DpIBd`iDs+<9C%LPGis_|x=D~d4NX0MZ;jl!ew>4(c
zXrvh*xm5nxwuB|qJOA1-!%3|xGl`Dyf7(m+4suGrWXTLvyygIc+a|gKF|C(x%ADul
zZ?8rM^nU=X1T8+GJ}O>eL_w(|^rIIGl3XGwIxaoV(DqDeNH0uI`mPH7@n2K5w<=B)
zj3Ge@4pjVxcih;}?*|hOB3n%AqU9Y>yuQQJRlCvPk~fH*11Ra!xf!2w#}K%PFY0Fi
zEMcs*17w7R7B`i%Eic!d%%#OSD*6^Gqg_JzA4&gjw-n(fIcZneFO35TxMDlAtSD_?
zUWL~DD4+%w!R?OdSJGpOAZ7HE^Rm*N7s-@ulXO39Yycq1ad0{q#kfU3RDYNq+9IEc
zz9|gc=muo$vN5(moVUa=lcD0AY$hIgc+RC3@zpy)UbdNb(nO*qJsfY~kTS|o)^Qr!
zV9d#hMRPmc#IdjC-_T)6yv2Exm>#ZQZzR8fl1)LM{yq4Gs!M;&dRfsqpHjLhJv8q4HCAS>^2)ZIDu1YzLe$uIkN5TvEQ
z)j_m3(qQ#|A>#C%%R-9YfaH2H=32WU@R`fWw$dFZwC2nDm!wSDhZ|148bjHaiBEWPQ
z94Z}AjB<6x=9Gh;JRt++-^A2(`YH`M6dc)AEo`*EC
zhhw$MrV{mt`aW;0S4tm2V64ER@!R!QLR0rP7YK;fGMZ%E+K6
ze)-qKE&Z#C6#AIGR`JN6zkL#;ad4wVpXf&jC(-=*O}@W{WWAeX(Fng}>Y6jRaS4h^
zjaDDyjm|!+;B}M&p*?N@RB)n$rkdT&7Or5Jok@$o7vp0e9AVj0xeD=r788UZtp5rA
zk3U2IzZ*-6kLZFv@t_AS)-V<>&w;e!$j`F)4=f%tc^a%z((;L;?zs8FEzIhpYRo*z
zKeA}Uv<{^CIwJSN&E=WA$*PmN#<-I#jpR?VWk{<4Y~|76Lhyb2;1f_1Qczh2ALNr~_l?
z$0o@v7<|zN%7UnsX?O-K8NUDSU=K1XQ|L3qK0}@=VgJ!%qzPbLW+-sR
zfL>R+5zdCqskQ0z+T(gI)^IjFnT^lTs`tij^xky}QofRp=JjMK^+o4tndTv(o(@>V
z_KQkqHBXfUYu73_dyM$^KH74x`YSu%d>bafS@yNmd&J-AxHWD<4zj+K6Hh?#;%OV5
z{|I}Yz4KJiD@nO5$7Y`U<^WWNLu&;OeLN+XG=voo_yab>IRsRKFq;!-s=OxbUe;Xt
z+4jU*eBZ?Sin{qcs{B?q5w7#6^6|rpj-$45HJI{5a>5^VmzyoRAnyrCf5Ph&zzivTlUF7th>;kaQ^)~tQsAh67!U?OOV@dPAJ-AS=W-{)!(av
z?RU0^%5%G@%oq#_tLd5#2`s97SA%>u+IFjXn-6OIzE^@doN87o>ob!~W$l=Mt1U**
zgYoK|M7cf&nCLpZXf#nAYFpa8jy!1@R|Qk)nH1cQN1~OwySHxmeeAZbHK!&ya&QHq
zwY4d@+&DQ;;)P~g3H1fq&3(&4%kZ(bNTM9ya1;ymV07H8a*_-DVZjlQ&|7Wt!WtGV
zRoDBh(S_@JOKl&_AP@4hV^WuQUAy@zXeGdVF24e+mAp1bC~3J+P7mbM>HN-bhpUl4>>p
z{Vo9TB96{NITYE9_YHHAlgMABny%`)X=6Ni`p3p+O$Tn~#(l{68sRo~REP51-SCG&
zAF~@f*~Cd^JlC4Fpg!m33~C1^v#;w92_4K^n!n)TI{Ne77E4`xZlHYtUg+Dgk!p|b
zd%lhH8sAmIQX8omE(%sOn1LX>xR#ACbNT~viv9Sx3d^T%kNgvF@P&(0|9!r^=U5uo
z%u80)`G7Ba%dmHD{n6eeUHl$D2iwTS>x!~r&y4M0gpZAda>yG{S2+?>bA5`iQHq^f
z>!Nf<*u|2vP4(VZ&EOgrMjc!xRACE6cDvy9=x$r1cD#ev%PhZf>WCUSH>yjv7pdl*DCCA>w`%b4Aj@%&JFQHijzNe|VM2#n3Fs=UMX%8@6r8W9
z21ig9Ft&x@Gq?*qP=2d5J=i~C_{aGLLc#CH6J^z#6uf;
z34oT*))`C6pGcq0qCb(ji#vU(Tuw!W8%J9`8=DeqM=YZShhKn!$T?rr7GFonnm=Wu
z0|=gX!+lcXCP~KqBvtJftxutnM@sD?&qU^(xw2U}^I(4OyXrrIDXHFfcYsyDP?t*8
zei6*Btg1T%)T7Az^CEv`f;ZWLPk^Vu;bXHD@Dxhg;=i2W&r4Xg5z*cxgnVTTS;iMaUm2{+
z{MX2NspU(}>@d3n!-PI7E$2KZNHzRzW%XV?H>wu)R)a?|ymDPb7)RQsKfU8n$ow(+
zLs&og*`jZsSjl1Q@RJX79hp5bp<#{mea4M?+(|GSp`(0&T>Gg>5+K0r8{o{aev48M
z=rJd_O}H7vvQ(Sc*@zw(Hn6IWN5BUCzU3bK?OS2+aWG|+^}BM5uZ@Bub06-#k4?p$
zD=*Y9Y^Om|PdzBLMZ2ygqTZ~=Xc3=3%X_@14%d9
zYu{R*5RT5FzG`77yP&>yx9`*o#jWlM3kE&}pCSEI>9e0lOJZb?H0qUzhP|f!n4O#w
zMP7%*R`)hFH?ct0G*wC-&&}>dA4-`S7Z(t~dVZmj^#Ae4LyjA^QGd1|!h7EAo_kT%|g{Be#QY_cL6igG-KT
zx9h(&oKy_$;#{)*5q~&&eSeR(o`d8!I_2zZ_3oU`c%Lj3Yw*_nS~s*LmT!WubVG~F
zxk};{Fa8%m!7S7riQfBIHM_%9U~=EGhlpFtS!Yj6@l4HTSmNz`QJ{1RJ%-9DE}!Q=
zliI6=7=PK5!TtAI@66
z*x9v>ur-L|s}`h5Q7_hF978a$DEma?pK~F7Couryjn0reJKeuqGKbn)soQg|qYhpY
zmjr3hqLv@86Bz$0ydrZ+(Aa7CbJQvDO20+E4)Mg&y(`6e<<_wdLTbwJIuhUVoP@b%
zU!b+iaNp&SQdh)RhitTU!jnY5+<#yDF6D7A>nQo_8H?AJn~8Azz(H%5VKDQH2o0zC
z&H#XM0&V2#)B#Ed_kj)jzjq+~rnKf2SyPD-slJMG$U9zNIS#{+>)`
zF@mM-mr1fUC$WZv+;izc50F$ubDz}ADxeR_T(s#zINk2^M_D800!^JMs)+=C{x(qERdAWjkoUDB!%B}2Fw`+79j&Qo3cCHh(6v%#)?xzt#BDpwguA2X*VAMnyK
z_|zwi#Ygm|UH0{8YIy~P_{i4&Bwa%azC`)WZY9XeMEAV)*jJ4g>i7fk;>FNNC(0e?
zl^RjuYXAR7$ant_Arso-4{_M!%lGG37IAWn60yqR#W&7GS&V5M6)6EnI==%Hus>uG
zpzB&1M=L4PD=rZo+p5biX{^yJ&Aw#*$H?hERyjV+=}f|(Z6!|*nKk_)LHAb>_2PKH
z0(eiWx|QWV@nx`a;W&|liXPXVWad#L+s_2MFM3VavD}s->+Y&o@!Z-{2gvrw-!xM&
zPV=nC*g*QI;qyAlVp6~w+F0xp`;{EEzHE+D?6>kHHMpnkZz10*);orShO%K14nE#d
zi+LhW-84$elP+wpwew(pa(*`z}qo)4V~f77gY0v8t<$NL^Q
zUIY+#C*qXYNlfHfF0*m3vTuZ%1s*x7HJ!Mi0`h*^gdt`#bhNp7PVugL+j647+~xkJ
z1a(s>Lf}yoA3)o}AqySXKUJcyGfveifeiHKn9i>@r
z(yGX=K!9)LgD3(+H%7z)Puc%&m;G)-I;%h-0vC3i=;^J|zZXzDDRHI0Q8m_G6%28q
z;ujSzZ*=PX6tI{V(Q-4Ucq9W|WD^a}&5bhsNzT021>_atVOH`k>%0-sP3Lg@@!Bbz
z`P}#ZW(c--J#povwAg`%**vcwaP^k32rE0TdlC)Nnxjo1|5cbxLRn$^-fKJ6U%{~tt
z7WZ<(1UiG)mHcTvnfA86`2|>ibYbk_!Sbaeb%Yz$%0V?UIVKG?r61jK&ko*1Y#4;G
zf8Y}HL0U6N9uP$6ge8>y_}IVcE?sPfcay}EBgtg60&b8Y{v7cyU4*uPKS
zI$8>liI*P6bVV%|IAl?W
zGHdb-G0QczG2c%q6vNhkcV0GdpT%a%{-eBVaj?ZPN_d=9HYB6{*?KGOXK6~TXS!B{
z0{jziwtGN=TE#AKu}Fb;owYrvchLW6;}CuJB$b~ndFg>WOaDPToTW@`=0dp@QuN#_
z>2sQq>U`#{ec{+vuaoM4UWAvS59396zd8RLub4jBfA78A-%-6r48XmKa8?ls>#f)jl8N!9FuX}@P`_%OxgMuJ
znc#MZa(ZQ_vkLdGP=#zPGABOVpy#ceKE@l4He)fzT<-
zg!ugrDvQ~bPfuG;7hZ%olEBWg!m&;*=%gYDDe>RE5Hb6OCASfm
zYT66F>hiwZaDHqC2={mYs0Wfmw#*$T%FwUOUQfpy
zNvM}ClTH!+e*ZV_-whHw*$&5}&iQLuwu;^vCG*dUe5*`vk0LvI#i2r6!;_(TM7Fk
zLxWePI0`l&b)@^4MSWmUKjvw@7|uM5&1Crlz=Ka>47U+u6#Fs_#=@izhQUI(T13a!+`&ODij
z8=FP{ZLIV6j)g%ODE1lQU!hip*qf>ed_6bi9Xup^?drtGtkr7}jK1*bv8TL0*JU$!
zvFx^#T2iD6b_BjnOpAF`TT#}aep^(spJRbtu%(Z6FS~Y{l#5J=^fg!i@vS$5@p||U
z)hT(-wF2}^B#YIuS&ZqT5VOkL{*F1qyNgk=HeDM{ql%~KGHkH(e5=A`?O@BC@z2G%
zB-+1}y9|?{gDJ?+(JRTBCb86JJ{_(FRqygeIaw{fp^T
zojK0rtQ!E1KTyoh6MjiRZFY4DcefHm(vT7g1wgBub6whl$dl7z)fYYm3Xt9Kf-e47
zJH0;r%6*$$9=Cm}*_YMkCPTvdc(pnGXW-F>@VAB$#!&7)>oi=Z>J)jQZsj_|lehK~
z^%RQ!Z2kK_HqNZ0xo$~_<)}Rg)mFNP`GHzYB!8*a%xRBC+$p`?6MfjENSdOg&AY)5
zC=(l`Y!_!>ReG(xh_h$Bzb7JvRG36WIVKVxmFf}khF?>qGCIxpPKi5dHosL$P)(JV
z0yp=t95UYvxWH*?c>~h8eoYpAx+BcC9zWIv=hZT4aQ;!
zVK@RGdCdw;5#GQS$X66go#+kMZm917LsCgM)%OBl4jj5geli(2BX<_&o|pKtmr{?U
z5d5C?k%J4%!3RCKt*x4aBmt}EmnwcGKUC#}n;4i}9Fu9M4bck1MZ@-BJ-?`5%(N=c~z4jmO({gVCbfi;rgZMQ}ZS
zI@RbCt?={cusg6a`J}mWkUo{nM{IR!t42Oe_{zBpnCpF@#;TG~L*o=8ezJC+dg%*~
zNWWp;!|XT$lyW9}xQ&YMV#1)z_MSW9`J}*8l%Vxd8aT@)s;!pRe)XC@5vp3rg)}V%~EX@F=><59?s&
z*VF6$`KeYwuV|xb%<`L~KusN&;rDMcMGb|D=Z#kfc?^aFm`Co5C>WCC2aE|FW<$cj
z-9vSM4eScxb9$i(`aeDio41N4BbpH7GB)jH@b
zMCfb3*M2O`OzkhYxIli|Vq+hL%`O_MELvy_QCBFj@4V4=-#R&Bue;zoaoZ56CFd-)~u(AC>V=w3K*xnU1Dun;^
z6pdtI+hA4gc*xBP2mZJRGA3+|jRQ$={lgW4U&|;i+vidn%PO*YOj^}
zkUn%gWXD%v>tp||&||bRb;$eE!gHkr6{L9p^R55BsfIbO#>%+)x$}YBe~F4$v?
z(OIxDp`rRe=|NoPLH2UD1kzXUj4D-}r8>sPpM;h=}}aj+2c#q&l4`B>jG1vjUE>dY>^rt>zm!
z`0F8p`bOR$3&PNPvyDk6+(u?S%a`?B1}b5kd-Pm*(X2&|LF00%Y>Xz0TLWa&LRD_07R<5&3){GR{|yw=MjadP&^lC257EIq={AN;e|Rs-qK7c(Cp
z6TO(zumh=|7DLuVG^~~${}I4^?vIwK>T|sw)S)WI=olalaqefauyf^V^Ser$-(`$#
zfNfyjjFt-Tjs}WY&QPQ81Mk+7N0k9gAk?w{aX+i5uEIw#5otSO|3iFDc*)1iyRetE*BR*A9C^784JL;d?b6
zV=R3)N0=zfPJ~X4HgdWbZy
zfZP)GyithP0OzIi6a&@RNpe#s5;8O7B{ymwANO25=0WsB2zrwHD6Y{@Ly=GFz|fa|
z8|BmbMgl|_*@%oOmyhV8r{BNy35oST%buGuL>!}Y?oG5L{pq62o*ynx0@Tm1@0oB2
zi;$azn~*xjL8IX!QQmwNdkTh)?~)L!qQ~?{GHOa&%QCVumw30u((P(I>JZJCB5L^;
z>l=Q`$jHUmTTq0yXbZsC>GKokfJ+julEl9FkuW7IEW+API9|Jgh?DqUzTgl3O@FUt
zV@0=Hn7PXq(yvr?`qCU%l)8h;?<6yL6Hd
zYnZ=cy)AA-aV9aivTsviNQhY%<{$V`W0#!H`Xohb&?m`-6XIdOAh@*8v}V0jauS!
zJxyIsUuyksYf2iUMO2yqF3>9&wr5D^m29V#5G3S8k@^Eqyn$E2V`^n0ADWb^xrB-`
z^shVh{KZC^W0<6@D3Wa!&vg4ZgT$yEW5aF`f_X4L`k=5*mmD#06
ztXpdtUo^uOM2p+~{7-3xR}#;3dD&xRPu8a5HaH1j9^X63);Yt1v&uMmb9<9sN^(e6
zKSnP9%Aajbk?NguNvX$hMzAHf4s02g)o;eMTC8kzsR8cNYbuMr|(tD*F~%K(BN$Br2*wRB`{=?
z1_Bn_aMU47$X|$XYb^MBru>CWOtM|A0uH`Rv-eo;Ok=
zeKnLsxfetPzV`OTPM8o&YNYdVS_CiYdKz`l%qbTYQoH^Y6)017O_2bb
zvt_GGn)>W#0lkX7eFNA66y^$Do7uViw{C)Q3XE&Xb=T~@V)3!(v=vFEU8d(^*w)aA
z#UYE^v?o^6z6}z@*{~dQ544SIv$$NZVG09hF2TC!yQG-G2g<=g+BQqkt_xams{ox?
z-xYJJz+SV63<~U+^-b?$d}D$p*MV-rUnM1jeSL2@0q&o#y93P4u><1eH&-E-6vao5
zY8}Nid8Ot}sVeOVFWPbS!hbl;!@ohYM48N5G#1l)?tdgu-K(nm%O5!EwC>ioh(6ov
z6E|A9M2atZ^Q7!ejpi7AYRHycaQ^3>adh;Pcztqiu%T9AtUhjKVB2PU+pRd&r&9cx
zz3Qs%Gw=v3G;5wp-&waWE4)*wv`-o>$5W2%vDD&C;%j@h!*p+;|%Xst{rZ`mR%Pg6N%OAGD@2`whOC65N&;D8Cp}UI>
z1q|CuGz^z`A(}#FdoJcXq0A-Rdd6r@)Qdm)t$<{WvX0IoK?Le~D2)
zaTZUr^sU9}mT>1A2?;Kx^uUJjV+`bBdi)!{`X-aRHPMOR?A~Elfse+l$#L6@i*V=_
zq(A4^ibfyxwTtQx=G=uQ@0x4||KlsWobRpJTZ@F47vkVjywMyR1&Y;vGNSHVKWHrP
zTm!3zNkZHA8rBo9Z<)rW+h9o?373J2q$j$_YyAGUe`@8VGhfm({p6_ypA0K8=-amJ
z4oC!CNMb7F;AAf6f^l0Rg!V&@J8U;@HEgy23EAdH;H45!X>+3!#Sq>Tm02F%n;iUP
zQjreSuaEo|@lcQ8h@bzCtY_J`bgJ22a)9yI_UyO^wEm&a`NUfY#hBFW{blx7aw3@J
zJYMc9U-_rd99v(_zaNCA6Xl@t@DpXP)DsO7>xLJt9&a*0&9_fkVKccF)F6Al_1-we
zyn>dgJ2(+emvU5|q^zDPlulQ8*tb_@gf<*#S#$C?R_=scDwX4cz0oNnhx(VSJ5Z`=
z$#wXp=iuw(+52AtO}W(JN<`{{dmeLNm!bw^1;UNv7QXJWi|{P9rAmE&5v06lMs#&m
z4v*@W!@wV<7wMBJ{VVJKeK8S^mh2CmXmiW{mB`l6do5i`zIVV4~D-9VwTRkuJk-(Et6Q&Iqe$=wW^=yoS?U!Q8_P-F91sQsA2N9!MgRjc+{6pcXE$pHQsK)MgBKH
zCm+z}L?EU;vfus5ai&wnzo$UaRX`@1bhTA#LhzDvmD7rK6=JYh!OI{=Wc9MVj9L

delta 12272
zcmZ{qzS0-wZ+It0)k1{DC|jpk%f?ih5r*Rk@)x;JW!TNfKaSg
z|L6R&iBs*3N9^&2^E8c=W}C|7;30Ihve(H}a>WEW$%Qdq`N=aKf}z|s-50<6Gu;23HHfNw$9o6}w7WE%
z+BwV?h@6K~IqYAgQ}w7G%^cJ{AB~5jim!Fa7c79@BG4v8){GtAu=R9;i
zDLk%x*nQAXnqj|9SDU?Hw%S39un?nqo@Mev^2^bB$SGPL3gHQU6QC{cFBlB
z$2I#w(IR=-qwnw>QhS{2zr63}D#z$(Pg9qXsTbr;R-dJQ2l)QA^uwJ7!U_{H#E%y;
zJ3fWaV0Yt=cQfO!;)yOzYi7+H|CDfeIU=a=f;QGf;uI*?Was6TM8o;b7NFue>YfX|
zxj!_+DP(m>H-7d0+q~>Piwgu)77CUP!kaOwF`rA+j;iSvyne{yL=$K2H&zG*`QMkd
zLEl7z6ks?}K$Dw3bxl@NNobgR<-UC+5zaf4)xl9Is!T4HD4$jLGM$TwD}%`bX*UJm
zYK+Knq`SPuHP1`LZWaFyO60m{pJKF@F#3rv;~(%&!eBIJr?c>f&y^i&LodO>#H62f
z_S{r%k3DIJw!wPHZNGo^iIY!joD!gJ-G=|6F|+2D1UeF){C|Zk+NGvknqf7wbYglr
ze67{2=yssm>cH}4tEsd#7lSzt)ajy9?)}6rt|0TnndZh;vdupuo=fvcmfD)?0K*Vu
z>_3y70=sjhrW3Qs6rAi?#6h{5}vo^2}
zQTDm?f!g@KwY+lYlvb$sS{jc(T9~8v)bW!G77_}iLPh$qv!r40K?qqZ=l!rlxw%g_pUE
z1>r!m2&EbGyU!2Un7*|4_932!N;yKP=0!S@tnWOd6veoB-BQAa-p;aSN#K+=`5eiP
z;svaYsZ(}rpSf0|pZicyXI3cK$=4VvzhB)jp*N+E%+}|+D-`kp
zFbZ&;BPX#y=o@*o9?fK@dQj&x@-AL!gYn@%2W5JZTS4t(8l`m(r(c}aZO89yOaJ96
zr13#uenGUK#Iv*q^P_4k?FRise1xwtw$LPcbRZ>4Z~Bz~dD;WHw43
zhT^K_0#jblpX(JK-+98Zbqh_`lp0onC1ctfz46O$7Oz(GUNe})V3;6&+lHY&P<|}d
z_5F}L5asuRG`{BliURf&S_RCc7c7Ig3v`q%bikmm3c?pjiWDRyLmRg
z-)#sU*81xt)kr5m&dq{cA^BP`d<|FZ}~7b7M5&z
zVxLLq3Jk|Y@74!8ZO)xessD{w$nW#52y;^LdXj8y
z+@wmmwMhM-S=@PIG$FFczUrH=zYkgAMae^>hCUe2cJ>i71uWZ
zsFa<3-p9W}XWRcr+rXCwd-Y@vBmouIqQlN`g>J$_Zu8#k*{ayXcQrO6w^Pf55x+5@$
z(~>5W)_HfeHE|{?TMUdGK^2UWJl92!#p&OM6Nk#Mfqu^u6;{x@y(%QEhX5eEI5;J62_S+NAw=&gVSxXAqxwRMFkr(RBK3L#Rg1*uJ|
z(?qqLg90m@pdWz9Ew3*@wX{jvxiQ}I_XqzCVMEQ|BeYFd@z{4vWb6VSk565KVWVr|
zP9Ge}TC0!0#L2%yH0pkg`y;%!mrL&$QOkw;;b>D=p1R7Q*-RYcd_2qABlKD}&-EzM
z8+Cc9%(09Ag!d#67?#?UcHlL~in3EU$)wFRWr3LEIQoO^sRbBYqm);9}Jcw2`XRP{>%Exc#-
zhsBHF^XyJhXY$NW+j)j^ty(N|y8ZK0Xh^SmTN2ORY$JIyeIAXc-(N}%9X`r7iUS(F
zJr%oHW#v1cSUpb*JHY$Rq2*u_N7|(X#8s6B6b4)?x_)x8*b3GlIVTZ@l!?C2L-yn=p4#rT
zWuXWr^3M!($g255KQsU;$VSg?Fjr^$N9EBmYpQ1X_VCz@vd3^k9o22yzB!)>xssHd
zF$lnXv_EZBtd!)r)WK7|ULUr&adMcph}{yQoDI^O&j_2-Kh)lD>$e-R6l`2e7<7`$
z#S55xJ||P4uGxYg7D+4)_*#U7EW2VgR}dp~QhVQ@{PA<@EUMJHdH;_TeLuYtEp{K-1D{#CcKyZ_5GpI
zS0lI731l2a7jSB+${rq|KF9t*0AsP%FD1M6V~PqljyIoqEok=`w%l4ZmjHKZ
znl|3TCl=oGmmuSpt6`EV&uL<1Iza_qZn2g1`)%YfLLPB!nq8sK6(1zT%H-`<>CsZS
zmP@-kMY5S=ZjA_f+B->3MHKmrBuOAfkqy?N=VVI}W*0V0SRSn+Zv5xyxz`?BQVgkM
z`Q=ZAn@X`+iGkznuJc+k*?oNmNq}%GL_S%6!PD;xpakZpzfoDa8E(-awVll4$_s(C
z5|dQN4~fL!rD)=P!g;GW8lmG#chjuC>5!*KMK#Eu`IEIqyQHG$(lPVI;kHUOc;OX`Y46DKH>vF;gP6^4EZV)1=SV|KqN~6*)slPAJ3SWdIYF}pVx+4rC>^%uP;rFc`d@k_m?!n&#DHm
z|F@yM`iA|#;s5?i%>P!GG=H&FBfQ}VO#(V)4BGl)E5%gP#Bc&{wpI2Nh7XM^(u*h6
zA2hY4Wp7wlygBe$!`bcem0CxY69Mhkp8M0s49~*}VcT3`zH8$n?*`WcDvL{yc$*AV
zFl%SxV?e|nX;xiZjYr@%jlkaMZ2bZ6U#B6<2);Xke5=HU5O;VvQ=0*Gc&mb$F`
zQLHv#CSYkHzkbQRudJ}e@?)TefvLflEB#QP;{>B(YU+;UTbIzgF3RZqixcIogWkc2
zkow|fE(TBbJN%m)wN4sSjFqqBw&n2jCA^iiQj96qw;8*yj4VsuY{^X1gYTSQ`P~I;
z4-&fxLLf>XCwf}>ox4yv7XYEh3LNHHqaG>jXC8L(;W}g=Uyh%tN~nQLSsRx~5MDL*
z{HL#-)sW}VZY!6F1$wuev?)RTpJ_YIdzc|FdNuz$M{09*_T5C)l}BHSmEn@Kos2l$X1oPK!hqgaXy6O7%c`(W
zsrMW3d&Z8B`OK1O)0`rJ1)_z2eR?X{Ge5C=Skk39XItDm!lND-r)`PqHnOzDdwey-
zhMf}hm=Cq3D`R?Nz{&J4U)ay9_{1Of@;CQ47v_72A&%MG+O2~h>tr`fm|V6w5mhdC
z3HSK(k{rV4C>?sJSpcuZCxLeFlHczFq6y?ckG+3Wl~$y!A~k-~Bc%e5HO|pj;oDzH
zwMNcuhI)AMZRuBBg?H_`V%>=-^yg@2ev@>mC`qV-0hZyV9Hxx0r3gKl|7iWRAJ
z`;wf789TQx`4`}X?i9MH|Rgh@VO{?cZpGfEW*HaVNOmnh!Bc7-W>S#2#y`l>!
z?yb+~%rI@t0pBOP4}xcr6s}MsRak0GJFy>_jUg-UTAK41>t+VY^&*>`uLKlVHRD|L
zPl#u(0S)Vhr#la&<3#a_m+NxM1(hcE0HtlRx#h##b!gK4xBcrcr#XDf3f#0y`IOK2
zshfv^C?3Rr^(3&Y%n0X;{3!DR>g(RJZ#NoeU|bkh1Kf&`)fvDt3;hPGUoX2kR9raj
z1dx&9`iYnhCL;}4JN=7ULk!}O_CGl3dRRe9@E7P7&iQ=B-M-vGc|$It=Z|zv1tCT(
z^u_}N+I;z>fxgK`SYO~bf}1gc;W(UQ&I6ryG7uEAF-hycb-MblhqoL@OhYYc`;M5>
zn^%fJ{}9S)oIm@!#5copZ^G3=u5%KL9C@IP-+CJIc00=t+Zh+{g-4ntzX7;^#|O=|@QV`1Ma7vbS)T$VvPEJT4B5}ZuZ{YEFqhhe$M>>GdL=@&iqJVNO;Y|e
zTE77P>5AC3q{SkbS6J9w@-R|ifz#bmEJZH?AE#uERFy6C*7JhxNIhsj$h!!d#ySB|
z1;0i$taZ|}%#qIvy+`+(A3o{gyMI*tql&AJ*!~xeKVh-
z-o*BA4-s(qWdV$DdtwLf5ALN^c!B8IZy8bj$hd^ZYA@GNb1#8jXYKYd&OwZ@5QWcQ
z4IOCr%U_kh+nvdyv=)ikVV}Tw#Q(U$ca10N3P?Wf2>*f<&
zkB6J*%m$!4J7`*LJCd=96R#JQg3fVygx+l!T0Ga{-4UIMYx856i7Z=hb?wSSAB|RLyr4D`HyLB-ioqO7;Rj8wD=dM9E_vXI1GV
zE0?Q|;NbnqMBbZeWI=|HCdS{cJXeZ&waqAoLEdSWU57}AdKEHa?Qpn8w+OjmgHSds
zY$~#)KYtJr;$twg6=woM2ZFjUI(qQCPGK!Uruxtvq2U(FJ3WbP_wg`9ub^_=?vqiT
zkNxW*sNY!}1={&u#`%{DnsqI5nQiGk95~PymlukY4BR!wWK_;PtzR6Jg-rzvt{ny1
zH#RZ&dFg*>|$-}4otiRXz4BQuz12@9?%3X>$3w1;3!>_fOhHS==NAspA
z3KBo0O{N_e1T(LdjO29hEpqTWC;Q3_+@{qG1g(vG=*R$W#Ze;=n+l_i$$E|LQ;?24
z?ue^<9#Sb@TJHE{@o~_j{+RFH5M)wQxu$_Ha>PNoXwi-*l$lE6iX~P
zogHikV8rp0!jCm*$-i{;)&;qz<)Uqg9-`=^t2*m>`0};H^)j}2NX*CBclui`C#}S9
zvp3A{A`zn<-WGP83&N+h2(9<0kqlwG-3F#H*Em4Vwf+~u!Q%RzA%=M_EHb%i`e1VQ
zr^WF5)~Z5B;Tg5{8OBUZOmrx?w40?Qmf86a;Y3jxTuUoP>9&v)}S@HlWsXbyzuT>y#%+7&v?b2>HK}D$H89hX5_BJcu*U)co8=$IS!nl8a)|9;_q3e6@gTXnn-uBD)fCc*3+pdbr;wNl`m~0HpdH;r6+yD)#
zXDvM)x%(B5{cMg~1D@M#(Z8Rb-q^b*$CofJ@^{PtBPGu)ANPTzx{r*LN63*B1EpD?
z!lG_IgXRNqT`SkLBcIWZE^bo(f0PXSFD1WA^%YxkBpN=qVyst?Hc26(qn89ysw&-&
zNmImm+ld&k=?fEM6H{uejIl>Cms({F+jV8b`*q}_>MI|G;b8b_EJO_QCy-9*-#{KSc-dAM$rx7My*C>6
zVz06eEL>~1`sS#B&3pHu`pihSebQeFOK*I6xKDmb331sSJ^h};Jf-iNtcP2?OqJWp
zSVTcH>%M+Ld6cIV85N$4wD<5%8}CTL{g}N
z78;lx3p!-w^**hOZ!nuJ75cYEMyekj26<}${i&HdW}DHyt?!ie>{7M$;aKY#FFBfu
z7=aFw2`79jy89+m1xWi6w&{Q$ayCb{rdat8SX|^Cxv-
zRG{H*2aI7quf68*o*5ChTx(Y;oqJw+A&K68&)UNhmV3P)0Nj`A`;PtL{)3!+Wjtqy
zin`#kRQX+8HG}1A__d1Z7X$8a1oYGHJY$vXmvt%{%;tFm;1c!V1GY)*=a-*<97qqx0epe1UX%$DJoi!f5|`B8CR3^pKld{4rF+X`4TFl^{GX|UK7~D
zLZ#N)nWS$(VK!4F$Wa~K9!v8?^~(cS)v4M&JOTD5+G;>(m=lJ%oOGUsz)hcI_@u7P
zvwh0s`Ky;Val8zqXM0JP{`@329l(+~i=RtV8Ufo`-d*ljV|hf~wvM;8Hhk|vsuI7{
z?^H5v{TEb4lD^nwx#xSVsioXHH^X?k;_^z19aJr3BfSmMz;jl3EN@Vq>cx?PVc+LnX>uw!smjF+d;ZfQtt^>4))jrGS&PFLytF?)=^x&d5HwzK7eB|S-XyO3_#oMcA3wJ*ypsE(Qk_7UXV_xrdY
zM!>Gc-kE>lv@x+xkmjG#qB|X$JHdsuhxn@lF`DXCg-2=$))kcCsPuiT{DDdc!(sI(Kyj0vQ}P5&g3Qd>OP++2Eh(>
z!`wnco<3Ji4(q!1aU-}rj*d~t#x4sM08|Z}LQV9~oFsFg-6^>HzUp~=X+oM3F(p3*
z7`7!N96}AeXm&w}DgLE%lZJ9G*AWncF6bxE8!kZodKQ9g`rg<)U|;)+jqKc6$v~j^<++IxAL}aJgatd!A`r
zo8x=L&Xy^=1Q^AjYGTle`tIwnGyRphQPBrirl%GsSm##3#YKABvk=nvLzP8FO&W!<
zF#}6b)6MHJu>Om{(yB1w>2Agv5pI|`c8^?jMDC-4l%^yC73A}LS=#nq@Vx@$9V0ar
z8B#}}o|G!`nXjAyDwvKF8uB7<(1
zl1XU0T`sIQF%w*S9yv2(@@`O*BBh2Qq{fmVX8h94Z^8no53%3y3jzmPgrWGhCaxO8
z^3gq~|4wbUYqD_Yp7ITVq-8Q83iTpY&sp=^RRWFLnte5hdX+1S9bu20w-7H>`TpiJ
zVr2uFe%6g1aEXXpf}ox)9=f`
znGUjo(7&vY`2Tvyynnw5`H-G)Wt}*6oxN&y%U{~fEhEezWl%ENVQVg^Wp{^irq;O`
zKhx&WzWRFx&~kN2GyQ0FjVH=87IRKJe2X)JOH`FQ(=D2)3Y--Rs}HycB)?U)X2qUe
zpEK~>+2cRZBb@y0pA?{@nF||leeZ9Ku|81CJW~|7d6vQO2|`(+yYT9zLw>lqIb^t6
z?^;gXWS?_!4(L!A1ARg@77l9I=2_StbkdJ6|sIXTgI4>oId14d4yE~
zRhZk7!%&)c%Fysds@#uWJuwE7ZRIXlAO}J9oalStLAY=FB&_Zj(=DLxo;7D}s}6El
z9`|wrYR@>K_e|*{30U$nwG(Sie8we*A5VU7S96rEJYqlTLhy%xyqbi;S7HZvjDdu_YK%er~)(D
z&=bj*ki!U`zyS&DinYO$nXqjaJ;#Q%nfRN_snJit?3*5%NiZR9MhaB%BuK6o?P9}k
zF6EL(7XgB$Vz^<6UmVJ0xlyy``c3yF#VF4c>WNr4b2p6`^XBrW7p@XLxfkK8q%A|T
zF7)O%!T-@Co^h^?T_Rf%4z3d|VSsCWPhQERI7yeV4jwc=)6rl6F7?jx)CPy!piwu6
z(0G~iG91R?KJw^~|Geu-j9b1bi7N)z$?u+d-kibM3^G}gddf^?4P>UgbF)|?2>`A}
zdDECf**q$KTlb~s_-0;TZTaPO91MNfeMZxT9CZG_oD97B@2s5jUlRsf*Iy{jfXw+n
zgK?O^7=Z@wFaBR%N*%H)JOvR$n$n6f1P0Os^~#M2(Il9~pXKI$&qbf@!q4-5NdKC6
zI+fVE?%V=F#o?V0ssS#9fy6I4>1qSqgGuQoO+AY+u9!DHze}gktqR&^tO5afp%|--
zd14v)1oE^7|rdeXd@jMCTz@Oyhew
zj^lvWFXfZtv0%Jf4i1rnElMPp%De0)Rf3s7`ak?4pFR~SJ0F95Uw`!U6$|b#=q<^w
zrOv|)XnfujNfeA~ZR%Zjmz`Ris2LZw6QBTzRf(g3?qIeSy|L=Izj}4Z26E*SJi8OqZ$LpP5O>
zueJxeiws6)`DsFpJJ*Z@XL@`>qd1O59(O)|So$ly#o?V7(PQfQ&
zKHd%giBQsA_e9L&i9|4VeEV1MmeR+P)X0{UfHDJ`c;PXn+j|)?YL^&Fth2+qi!eDq
zowv9hJQI+4xL8%U8q5Gd{7~>K+S&N^+qAwy57hJP)za^kJ3_)YAI4V$4^!*&>utzi
z^arFye@Jw*hKBjTmRvPis5Co;PksBS_A}2(sJ8y}y?-(OzT$
zi4bP?G!@~fP>**{i2Ly^nQ^q5I9}b)K2yV@SlzF9WXtf?+(HZg>v^m-ABq&yj%wf~
zr?2tr$H{zP!@E@9L3M272GbHKd&WGN0{_KKy2Z
zVWq8AQ90@>@Ba1CQIn>)T^Pc+!nDdW4qb93^2)2CSZz#~UQHor^YYlNdlJx3mZgnq
zLT|$Q?dBeKa~g6_B(Qp==&{7YPV>MCUkwfQniv5}CW^Q3&T?N|U|qZOT_0>E>jgN)
zpC1D;IVUd&0q<)2P7Nh*y~ocouuz*n2UzW0h=o8n@6q~9gU>=9RyXa1-3|+d3&-lg
zUX+CXcTsW5q(Sr4g5}HTdA&{ziMBf~)G3XQ&HRKAso5Hg97luJVnL+?yj-#k$(F@k
zp~#z&qIVLLR9@W&PA+)DvF)pT*YfOW&{|&x19x^Jz)^tIlrDhot~~c-`Vj9?k?K_TW9b
z3wcH_K#|bIu0AuP!k)3;Z1TZhHTB#3g;>}~sn~_xS-`4JuSJDoM71x)+aIBEsSoz3
zn~c7!>myx5X+76x&!aL5!N00)s$T)uYML6Na>fhbAaznUX-oC42ZS#E14F%o4-`d~
z1g;8FI~1{!3Cdp{v;agq3M*sfd~7SVz39S#F~JqCa{{{$I6@4juDja;MNS#d%!M(7
zt-nSv-0}|39;1&I%v93$tfIQDV~(9`WC2zR#^53l3NxGX)L|>^Xib7`(8&FR2bta2rP^S}FS}+65X*ukS-4%TaH692Q=Nb^
zI+5G5RWK~1Gpaf=?UB_UedZy({Mv|3Sdb6Q|2KYc%}dpK#S7JNsJmVZjpc03lg;Ba
zkBG{YIsCKiE@4>xqj!*BdTb)8whP8ld=zm_K(yVxYo|L&-#fX~3!W|{-EI#AW|ITqUovJcr2F&DK*i%xmKX|C^b7;aHk=_EhdqV-blpSN#lf?blaW7XnoNKY50@kqh#_3jYD>GxlvVVG&crv_{RB!cdu`(GE
zhk^J?(+_FbO#LN2CXqmwA~-LcH+y)v;MzN)F|+po2e0#KWJ?CEfRPM<@M&OXyryA16R*|CDhfu
zDO5w_0(AuawgM4@lylE?qCF!K>@MK^uc%59(=N27l#}N(RUL!ZdJjN_5^-A;Z%lA`
zYiiWSd$lFprnB{EL-=f*^%a~qAakigQSLN4x
z3Qm*x(|L0ju7%Yt)CA17u#So@lBF)-y>oHzJR?%Oe9o-mgLo4CY-3!HP*h}>7ROaN
zp{?vFtG9nr!BKo|L%E0PM<%Ac(#WLx>G~^J(eU?H0p>#9(kN9noLSsO@!x)RJ1DZQ
zF6!b}?#4?3RfCYEDg#6USI?q+FH#!kTHMtCAkjFv-sg7vO(>uha+O}z>xm^F>&F`c
zn)*tknHWySTA&<*UF6K@mwttnMSGjrd>O2G(@nnx=vD}PmptW>B
zNIUJFm}Nljek4F2?=i;vCYja1i0N2m_Caild9W*w;aEi+hYewkt}NiFxuxPaOHXwF85;>75<0LU
zVa0NGk(|F)MEBdCMefqNK1tioZjmjVc!d|eL$Z4P=eFS7-N~t@XCc#q$Gz@vwc^c0
zt~_U1_W99wU)1n;xSP-9u&yGKU0m1Vo}qU$mQM@kG54!L-?q)ST_AVyd3
zOGv#;ty1}xt{czpFj>f%f3!p4nG854`&m1;;8glHLXSZ%V^5zq-a3;ZPkR)U84a#t
z05yV3|6`WG6|35s27D9b)Mawl!3|=D*O_sdjR1PO&i+(4ljqBP+bC_sersCh+-uV$
zzuz~Voz$)ayUU&qD?5(-o0Sn_+Sd%%>9yd7^!;BNx;}!bjvDC>djy6e5kq!LzOgJj
zCTx#bm
z0@!3RzNE?&)hfH=LX-|iB^_LZ`c!rx!#o(u%xCaXfyGtDf!!31mO9>;=9I`o1Gu};
zB9DOh&T}m2~`>Q
z(+q6MG%YumbCTob)hg<_oYbN%}Y5!guXAW+lojMR7Ci#DPa%<4ckiNbd
z6s7!M^F|B_UHNqjqo%g*Uc|Lkz)+Hpwe|MCy^s>mmiCc?_B
a7$vdtPkI5A@<+`R?t3PNiKJ?!cmD&9cf>jX

diff --git a/previews/PR1624/tutorials/out/solution_000130.h5 b/previews/PR1624/tutorials/out/solution_000130.h5
index 3a5244f511d1f565c0f7d69819da633f976ce2ff..52856d72a9c4ca3bd3984004ba217683959ff703 100644
GIT binary patch
literal 11976
zcmeHtc|6tI*Z;??kXfp0OsU9BhH8H@M1y83iG-p|8ImDGgLw#1%2YBFDKbPQW2Q7{
zP}1NG&N(3sh;yFxDek?`^Si(2`Qv+^?;pS4^~ZUi_u6Z%z1G@$uf0#NBF&7~3(Q?O
z7bE_7c`;5*=;znj57x|ItI99QET0`eF!P0N1|w!LpZ(VtHcS#D#w2ItLNoqlIerP8
zX7qOLFfkz{v;Jps#_S*X|GWZbChN8lCD=1l2i_TMo!LJ#^ygk9zW-F775)o2yI(JI
z|8g?xmmgF6Dd51kF@B5}V{>zGcJ;!9Ag*7b;lGFSpCaVP@EN3xV!i{I3E7X2xf}~
z&i}CEnTgx>xSi);2hRC3(qA8c*zwKUIl4HyI=XrPwe8ID23_5s4mhxxd|e&AoNT@Q
zJRE=KD*}VF`C7Pnd3*ZUd;54gI+%kNWclZK|7?HA+2xl#&!6lG&niyZKd1ZCyj++V
zQD<*wS4T`(7X7|FT6~xm3&Nb8zDm1Qe(P>FeAa%pz8t^o9lZZr>qgbucDj55`lC*B
ze$|PX*UQ_{<1ZKH)UV0llMUnf70FCEMBgO-*w}sv#(pvkn?;V9zWP@g5vp87&VcOf
z>^~jw&iLbCn<)Txk*MA8eyzY_H&Hmrzc{6+d=(^ElRBHub42*)5-mJ
z4`#oA`#ZkOp1WWzB5)Z#WkWoJ(@)!&r3|O49y5
zc0_}PPViKbO|)1@*(q^#I>>X@u69@ma(q8$wjeY&!#C~@#k?3E?NYAA!-e6~^n@h}
z-&n}-O77FFcPun_abQ00V-~7f9KPm2JqukkNtNGL27X6Ia{KdGsPMt0PvWw!^d}5QwwS^oZh!&zQjEY4=8i`(02#JPvK9E(ptbjH0-HX3{RO?HhJM8hCk+)
zTrkvu;ZZ~HhfTXNJm$fU*7=6Wxb<+(NQhX@W
zb-Ra!o^OxoJqY<*1isDj*~mhDOV>%VDfzH(Q4UNbKi<|K=D|b~
z3qlq?0Qu9v!IwJTOr#L`-9`5j6Imn%4EhL<8X$udY#p{qpp}b%SXp`gk^ET&tCd
z-f`s`AM0hJxtA78%+X_^Sk(_Z95Y#{$$H+h;9D&8Vo~9$_9HBmpOkQEVReQp;de{hG#6Q@Vg25eBS*^G8z0NU)p*0RSbq}2D$R{$YA&mt})vS5iHa`
zv}5?OAqx!#Cb5ilS*SKok2I3VLK)9qWk;z)zH1A5z7$}%Shqw=XCuTltfsPYFNSj;
z7;}3R#X^-)9(Ol9XQD{!voDB06Fbl&i{VbKr>bljOqBFyk?m_&Cfa>`rKk$*qt5pA-j`dLXvL`f7B%P>
zV?FN*6c=Fl{2-ZYm*M`PrnlL<%;VOJ{T=ukK_!_B%?0Tp>#{2$tcPfTg
zrR|YX55e#yikT*RA>S*j@wRPnj%~0}Er}a&j-^#NgATJ$*hSOPW%|%(lMeZ(#5HNH#xlDP@$^6PmW24M?c+0%D4avhl3N~6nW$pOry$gaiI!B_7?+-fbye3g
zBbl?%3YwLwANZ3_&)94V>$l*&d`cng&-ARarnTT#$$SwH1?tG}cd&b*0pu?g?RPMX
zh3t!jt6vl{kx_1l%46swg2xw&2XA4aj!;8mWbixcB>shC9vOPi{A2s
zzujE71&kp7+T!|+14$TuHY6Y~5zd>{)Zr_0%9#jzTi<;n2>OhM)7|caEVS9v*C%*6
z+yk6-ExT+meD&L6?g8izHTCCfR>om?GugI0s2#(X=G`{k3-vALD)zjT&O*(m+^+|Z
zGm%H=@lD;(*MwsRw{k|aP(;}2yaQ7#bnE0X(fP1%;xD!}xeH_XcS$6-F&@@+W5R1Q
z$p4^?t5^OS48KiVOfi7-BcPdJ_Rx-rlokc!?i41fW~B4J-@`&8&eww6VgJa8HL$93
z0gDYgg%4u*x^h~#X%~jyJTpkih5KM<%l=0d0#I+8Q~flkS0J-B{5$m1OVgVb@4-6!
z`HueI{670`$|7Fk@Q45W{Nsar_8lt3#s@zXobBMph?lJfW#;D)yj1>O@79YlzoPnE
z-c=g^Tkk6NnR@--{w3qD>#2#u!}{mnzptKZivL?3Gxy}`+i*a(``
z8|aAxIo6ruE(@q+TQgMw^7r;t58VJKoR##SGl$VmJkz~zdKfM1y)k|TP)go9X3_XC
z$}=I$ocTVCK5mwu+c`0eM&s7IDuMpp(b8LnU@y1C%HtEDlU}0z1Hg>to|Y!SX^tnG
zA_41PSns(rHH@%b=h|h#PT6eUlTq+5%dY>?oi&W8&YC_dfJ-*uoJOF3#Y%XpcW}eQ5Vq1EwGdDThPikP0Xu!YE`Ts$m8FR
zZZrg*sP?dL5sb4ncuoyL{s)xB2gX5uJmr0$AIRg@^fa^q-(4=nvIl<6%sAp2)GO73
zA1CIm3FlfB4&z>5Jua34e{u8ti>M#N$h|sHzZLW(0;P-FK<=$<*!2W6z;0E(9*fxUR1Xp+Le!zKFuWlZNaero-1`YhYB-e&mfLC7(BSMW&u5DW6?p3_Trf}f2>V+#XdpWN+?dm;z&
zt6x|8S3o`*xoa~liScvUwT8eQPds8}Kpj{yJDTQ$pVQTiS8@TDRy>Zc1OFs(#Gs1pQqhG6(Jgs@i=xp$g+7O~y(s5MNaCb)h3L@3c;`<0#0(
z$vH}ipm#MsI#3hvgfySRO^CZ%wBe=;_$wZ{X+Qz)JCsU|0R7tJ?P1~|moA>B>;(Nu
zRtJSRe_1^7yneuyujU=g0zIM7!t7AsrtLc23_$%>^Ja6f)7Ut6BLi?@s^2_fosQ}7
z`H6tOc3X{`1o+wd=*Q(ppufeuG`<()k)#?L6S!twwk0p{`1pBxivV+N?KRc{R&edK
z-~&IQa~wM;pm%-U5@riw@75740OQRIHJ@n%(oCu6j{v^f>eFilXkz-&iW9Ijx98hMNqMF4{&np>u)vSmpd`6fr!)7dp!6)V1R^U*C6P<*}!X%0(!3olNP-K
zlo;yi*$${co=i}K@tqZ#aWcTe7ONzB0s1Z|(z6G7181hP9Qesv9BD)BFDG+7Ekzjb
ztB*TPoL|c$F=E6%$ayU{UI61?D}|<-0M8A)93##fw-tXlu^;jdi=?Ch2E?x_CC-29
z6m{+-@W`>D_X~h?$oeHq0cuN4jz0(a-f)iV;=s*++&}&mc*
zS8{_qSs;~zi1+pvyYz*?vzvdo?E^el7Jt<8~KmHxJZTfqafhCE?LPd
z0Z-}Ms3n5lwvw>hL|)didq?;|&rRGmn+LFEnd#g|fS2fY=5qk+9xf7%2mO)hho6c1
z=7lY(LE
zX`PrP1MXBSv7Q6^-&4fYm4G{)v#?fx{_(^zTxc7hQ>oDOm&svt1<$$82jgE-TK%1&
zFPeNPv?b2v{vcV7dw{&d*;lFnb*Ej8x&bQ#&AN<0UuyH5=0{MUH|LBGYXLru@mWdS
zv(e(&jXW@~V}-9jaW5Nd`g%8kUm@0}7XzTzYzknt
zOPdBqwt*i0yv8a8{3~*GGPS|qixma0i2H0LD>R+TF@kQdskSE0Q)3X9I2+jcYi^$Q
z2E3{5SLXyMlp%Bc8H_hPP_)qn9ARH~!v*GheY2)!4~*~YC><9EKS`P61vW6>tzx^{
z1u#!r+##wY@E>$-Ut+!4Um8(P0=lgbq^E+MnccNh0rVEu$5|5lhnY1#LEIk|7c|3&
z{38mJcM$KB&pM`;MSe~YgU*wy*z8F&MV9Gj|G4szj-ori}Z
zzp>)QBi6uU)Ly0eK;0z#YhNhA#Hnup^QA0b(?;akZa&ZO<-SZou4mGdG
z75p43ulB5kd7sahF?|SjrB)K^o-p1c){b2Vd1&MDssxB5^MdTrk1(EjBkQz2poK$l
zb2aD*&gUH+0={@vL%=BTr1Pa3DdCfdgL0`5Izj
z{;~}^$;5hJ#g^pm1N7lCZLk5mK(bWrInYxTpQ}ckv*E_QzSA&nf8*&Wac_I1x6X+M
zT<+`O76@_yI#)p$@a`e&4C35~v`fc40gTk?-ct|$Qmx*b1p}JwPAJj>Jlwl*&LObV
zxbsB}1HG34{St)V9XStH5%>5~Irco_eCFp%UQYt{Z*0=uIEE=NZopG2-J`5NGVY`>U1k+-IMVjDlm<9w^16aDHuk5b!QX(9v+i8*cWO`I3O$gE9bKnK^uMommxWe?
zyf^GTM>%lwivwXh0O`TUMCQYMDVxum67RGe)fg
zUE-Q{nKi&i^eWDM24qX|mL}e*y>}__f0f|fuQ=q{0bBuZmn7b2d-<<85_KI|(X8wU
zcx}Y5L$80Q`xxnb}+5E0C>`Fz`Y3C(6WrEqc3XDe*4UPztmq`toV>u6*L1
zk-S4dnCKH0$?AuR{x=^5;}5v>AB4vrgJOs6*EB<*78l
z+_xh}z95(5et(wuoiWC{%dH6Hogvu=Jb@Qxh0z58=Zzn-Bi>`uU!@ybfiK?V-|PfD
zlNM{;3FvXj>`fHNv#j1%6YmF$UBU%4x)~ieZ}R8`}{&(Rhl*+&$0AtMBnhY
zvcJ9!cx%iUYZB_Y%`eK|8u)Yx3aNoS_@7^Jp9UP5xGE|I=a}zCZaA^7r$;I-YzI^t
zcx_(|cu&Zymf(d3%9i0^U-IKk)(POF0&R#mPon+QlB1B<3ER)Nh~H_)y#joRI33eE
zE3?4=Nc*E*ZeZ7cU%PAwaB8xFYdK(0s~RI7aQ8MQKLzyJ-fA{X1MXK|oy7<7B-hk+
z5dE&WX5*qA;O9(pY+@*k7m4*r?FQU>!XeiR<`qqKn|cd4RkT5tc!z&Bm+jyH`{<2H
zDwV*0Q=TJb3gaR@YFT-}Z~F@=*MZ-%Qu~A~!0=-qRJMY@(+5gfJfLU5F6SZ+dDM76
zN^%GOG|c;nFQ97bMpM#FGTQ!aFIQMD8R;&3wAwkFjMf^=dCF5pMpmDmYp%LYMxHEg
zyNDVx(wlOB-&ajWFmP>#HpMuhl*sL
zEHN(frlPf5mPEyFq@oX-72Ft0sAxjRLb!XHf*J}=Bu2fVpoI0QW>?!O$f)3ScDMio
zF?PQ(_V`Fgow?d8m&w}y~vug8u*J%j{8fbX^8D94_kmM4XwAnU8!@IhV*vz
ztg>#Tq2-J1qI@i9s7CwrajuU;$fEZ8ddm+(NLO{i7C~bg8oZWccqoB}CN|%UREwh^
z%Lr9+G?jv6y5)}~>rhec>&V?bA1EkN)P7pvD;deHQem%AAS3UvsQZ0GB$VtDv)NRU
zjNFxUcQTK`IB(9wAUdj_Sn!M{H~YX68vEhDa22`OalGu8yP$GmVlk7w88SJK@|Q)HQ!Ja
zxRcGQM)2!+|M>S78(4q4s&^b$$*6C=$IuleGRipX6kXCzLSj}{%`!L0=uFvt#Y}B7
zYTX$s?N>`eL2OezJ+DZ}x>@?ZEbN!)FGXlm6$O#xmOoY^QIIq`5qIB!f;gkQucm|j
zD3_mAFzlm6V?}p0btuRnPk7*PFcpPpD3(p_r=fXv4rbA84CJ~@bcbLZ1N~r?7Yu)+
zqm#dtzit2wri^pQK{91V>P%BXT;G_=X!@nriT1?_cR7abHqMPAE>
zTkd=sLM0DsURkEmQD59>gJ~oK4OAQ|ofc=HPQS~|-B~pB_ZiwweEb45Df)QDF=m<>1bT#^dwIh1Fe65HZ5oq9qByg)%?7C
z2pv%qyZ!1k*wM_D4)&8#0!6A$3C`ikNS{q(Q4}iN6<ocDzN
zd$*2|5&!PCcuv^I))w9+<6b1BU}@N-uSZ5Diz%vg#bh+L_V?k3ffE;MA%
z-ml6Y5Ak$uj|h*aA@5RWrnegz$xft%XTmu+Xc?+;3HGsll>Sm#=!0VHg{Fr96G{5r
z;tL&Qj66`s{_f>}bfm-K}veCj-$Y<6qtfzn2HDT@qwM{ZEFUSt}yTVH&ewBr-ZlUX6pAr6d|0f^)iNpR3x}ky)`IEUO|R0
zpmYe~C3%;kJD@*0i`?ly$3UWsLJbaVW1thYN1FI2=;*t~bUFv*-TB<9BdU*tf-MIE
z-FZo9;lbSOr*b6pSl5Ps*C-kF+Nnn^s~SS-9PxunZFH1clxh+O{qn<1A;0_O64%^mDDG9Hc=9efx)HW!Q9Yce
z<&K}ud6A@t}F0DKjf=o*u}gNO+th7`32XwQ&1)69Sw3Q72TpB{9FeG
ziCZsxN*<#kv&K>Fw^np?Z{m&X9?1X5aa1B+IfTf0uFEE&e$w)9yQE+}l*^dv-mA&z
z{uA@NUAh!>M{}iX3iPvyQ}fDKv{8^BT^~z1N<)il^3SoLerCmeIp3hqgzqYdO@Z}@
atkl!sdpC#%n(i**cnJA6tYa9#z55>~7!#-f

literal 11976
zcmeHtc|4Wd+xKlIghJDiA*DiQi8|L^5=|PZBvKS*%8*Eg21E!^gp|yrltR%VW5(u^
zBxO(bK9WYU_jBFFIp=-f^FGfX&v|}-yzkE+`@6r_wAQ-TwXU^xpH5mVUm-bR`UHad
z6BQ){3F#kS$3J*ue~k-%O2+xPe&^U1zA=m#!(9HKU-*cr1f`ogCYK(IuO#qO;5=rx
zd83&bB^i%Dj$>~BhX2nqU}3g&J(Ys3V|frA!^W}t8KXaHjr#t>cwG43!0~!LC;T(W
zcw7lW`-eb)5GEuDQG(Ci(Z$W1kj8oaoErJZRQ{J4N)Y6j-vl&7Kc@q}4n8g(?%u>?
zX@d9#mE%Xr)Wglw%gNi@WtXdynXjvpm)W?f?s6P+e0K}b0TDv;E(Om7bo<2IL-0lZy7YItlPpzj{A@2SKz0=qtAcp-KcfDnXM8Bf96T-
zXP&5Wy?vZK|8`@J{u~TH`3RApGZ~u>)iKE9uVp&yJO#!+Cbum0VNC{rP7&4~Q`
z{67MSj>Qw;8%qFjj>_F1f0M?7k4RK~J7E5(5~X(b=bQ+_%h}_P
z2o`^Q`zOAPpS$%kKckWVTTrSR#%DpP#_h+hm$6g!r_;YMzhhV3f0bXQACvn(^Gmgh
z@s%Ca`R6*>pSk$!x0o$bi9MP;$mSn%HAXxU(ekBkE)OEYSDp?=E;x10{b(H?q#Aq~
z+NaBdK#?-KnLZC@Ima(fLwWwMiyfz}8cjnzVsCt1A%=IsO-e{W;slvaYA_jFF5izPv>d!d!*V@r@PqPD9$;5fA+a?@-*
z2-2XUHg3mOg6v#ouYA{(APpmvp(?@53)9{OJ04A
z2eS7S+ipGJLDz;;Pj}(^w+Ft?@?FJ)&vRDvFh6iX{p~|L;sqC6Nkh#OpSa*0=)UBt
zG!O2;7oDsyF6j3-Z0b431^3@>9+bVog+l_y=c>dcE+KAT(;6Q5
z`W%qfao|C$w|T{0TOQa8pEleu%!7|64Dzr&K^_@&9KKROkc#d{o+RNuoi@Cn;(CQ3
z7c*<8dj%8ZqEa?7$B7`H?3B1U=+1+?=En3nc{uJ3$#W7j2=dB}LLKn{f*cSIwEV4#
zAny&;-sj-{2ihrz1*H)0LB-#ZiJw1&9cid6&|Fj8jmt%L`b?6Z}-a2}gK=|3-D
zf%~{%!!nIDf{asfidTB4yKw&Qk=b8RRA*qDYY5LPPdv
zBY5zVwQ=C#G9G*iOyDgy;=#k5MYQ+vn8zobnUOjKS@vt*gO7OxsoXB#&{|87X#?5|
zR&66lk)1>C9g#dJjr6>`@);MxZBt8UUf{yI?yUZlCtS!1zME@?=l0W8HLI#g1leRT
zC#@Lc%vpW?!kS8gv~DGKK2XAQ(s;DYp2LM3uV&bHxp86VLCwhvu#Q;GD}2s3U>yyr
zSZZUx7jLPfgEjBJARSuVU{6wkTsE4plcBo}&SI#g_O=fchA
zur|+hE|k>?<`-k#MCYxzr0m86Q+Dk0t9X7IlO*JW@SKWOyjrUoN{~U`QD+UXPWrrr
zKRDp{>qZ5>C}m-t8K@j_LjTXJHy>T7#e-+bcFs}-JW!bvxscz1Amd({JC@)%-r8rG
zKClP(?L+ZyWjP)^j(_T;?!<+ynH$O9?{h(8_B_bJepYSj?6({9qP6tW!iW2Lpj7|6
zRq+?>clu{eUfzNGvhuL}m#er=A+uuvScfTf6Rt=UdhYfI^ug}uoRAC)vT#T7y
z+0KPW2}`A0xLn9-yk7{%@!TJfTP34TkO@;pzAZaLkdws4T(>q6B&iUWS%rC@BHG*4
zo=lJpsaq9wLI`rE+6A+1xZZ-rWYc;)$F{_vzWg;Rj|&8Y_VOV5ocZ8fV}ksWuqPlj
ziy*g%Omm^(ym+?SJd0|AZ13*ynTqqP!T~U
zMHjr5z<67PZb_Qr`l}18R`nzhWN^sgoOnEM8%OqD6f5BZ-|MROYeBfby3Tjock#f^
z%g;A>9zkXcRyJ(5!}I^TP`C&CLuJ*O3e6aTyi2z)32G)tm7H7V+c3XbLWN!_X*^(<
z3wQM$)C^~Xr@)WF_xN}W)9x?>^Z7L|QiY3Tj
zR>gH$;QAfy+`My_5af;iSqu|AKXQ6;7WW;vpgAL$^k8t|3@1&rcPkI3xm*tN!1|dd
zTf-~M#=0q7)+)1$AT3Jz+s)eu^49S_MmF9DHVr#EN+o%)%I;`)Kjtfj+Zg@@`)R@G
z8nxeWAO3nr|4)7&e>Y`NFLC@Me|-M=Aw2#LmF5%29||rG_+!M|UY9ZUa|mB5|EhQE
zImMr|`bXYXy8lz}D*my2{onp2PcesCN;NKbecLx67$-wfb`ED+J?_v4nAnz&s?*Xbz
zj;tcy!`DSU4`NU*(3<6;gs5R(F;a^1*SpK^yCZ&bQ8#|Z9ROSMg2&6z0a)<#+SiMS
zGgNF(&G7s^
zK|>uPPvFt&lZXw^ZMWVT833uxr<;|~Z@$IEM}rt&mEZWC2X6p|T=aYwAkJP%3YwyQ
znzxbGA>^EI^=Y$_2mE&LPCufynqox)&i~{o(bw_Fr3!}VfruqWB@=28&xB8u<3qeM
zEV1Y;qI%a9zT$djWJ>`IVZZin6??C#GtVQx55WHNK;uTVlMS3%*o5)}8q3<=
zq20j#tpzE_S4)ctWFj}7c-Foa?Y;}ONAExsv+2CPAN2{`RNa1z^PGM=#0u?V`EpIE
zx)G7JJ8~1THg%68RWDy7!W$|3rddPNi_pKPSN-^Av{&_ve;bYR%MXst*@SVnnGQ`q
zjCIo78uLgQCD^9kXs(=;H6_8Mo(?5n}Tt2%WE%YBhD>-
z7+Z<)`#mIcrO;0P+V@Gm$log&H&|jk4c3x$CUQRYOIxXR#Wz2E$wW-89z1*<^-tY-
z>$OquQ>%XS5+cb6u$heVOkqo-7UT+xs#esZy=92v&bx@l4sQ=>p?*@`a`gtBUr6E=
z>3ulvXTwCNL6je%XQ{`dU14lgpdMo6OmWrgIPaFpHP>A+-u3s_O&G}iS;@=@w699s
z5GIFm_^RR`xND8X%+okWe2qG4Q+&R1$v38c{QW^Bql~TL86?so%*XIh1`#mnKhMK39&)48;#87#swm!7$
zS}AIhgm%yS5@vKF%Ca6j*np@@ACA*T{mxRo7)9hqXDx{LM%+8CV9{=rUl+Wfp^S0z
zW}UR7>TBQHMfz%}e^C{4j5@y?_MMWY>L9;M`D-5P#coND)FH<7ycnX+8+V&TI8_e?
zdnYBOA_l}RET+!?xe?}sVdN)=SiRGb3o7|1Dj*su41axwa@%l$D{{!Uf3H6H3HiNE
zzc_Z|JWRiE9u^{B$@{vKs+0Gt8#L6A7k+z`OwI3PfbT#tVq<$$<}TEC3VR-SfH--!
z_^n8kr*F`i^c>@SJ=}a#80D#w$pX}T%Rf4#O-G(q|J{8%qW6&vCe(gB+-to@9Py}C
z#IpeOdpV!;s1UJhT9r{U`c3$x=}YOAI^D`-5p$L<9irAX)BF0q1hkv4z;91OzMrg>
zIDq=v~y8>MaUCD3laoPDMUV%=Qx2_1-OY=^aC
zh$Z)DOpZnS-qHIXsQhMz&8z%|DC1~qCWG~O$<(HJIr8TLAA_j%cIe87zd*b(f7-8}
z=-29FJf#Y8j%lvxzv2dgjNaYuI>he$}zun}zqoujy*Dbx}VQ7nL;`^>ZZ~
zG8GX+f*)H__d2^9Efym>s6T6@DM^jD=IP5Es!u6IZ_oXR`gq@8
z!>N8KKd04l3L<-B)EsJ^DXRRrVQ7D>#W$Ri`v{li+v7S;Y%ABXM(ngqwYY(D<X4;_d|%|$v($a1+W2jlj%cl2WGjaDyd+s2b>x1ht!!1XfAm;~ORq=tE0!Mp
zI6MG3WY!gN)PGKD3~<4|xbkhjJ#{W^f|LY)LlhgxyjX^4IO=BFj#wFJ(PoPFN^8XG
zJ1{?;r5N9;J|t4ZCn(Q=u!A~^2$QojJ|UN+P7^QptQ;=I+*d(iIwvKxjM(4SX4
z_su4>Q(NEqaVMg|$E=VYh>htl7Cy##h3`v|O2v1=aaI0^bc}nh>x|kiyzgb-&=NP}
zeh9oPFp|J{syg>q)MFlI)%Cq!k9Op<3Y#R1uO`&WHNbeC^Yc2X`|L|bXc|)h>vu`H
zEp?t6gM{Sx(9cV6&8QDznSp<$GvdT_#e+{!UwKc>&Ipmlzx0|bj`#3-Ma5Rs+qV>d
zmBTo(7ry4%;dn)b4!5V_IF&Jbm{XAx90NaUzlC3zGL9hn&X-~*qnz(@+a^`CQ?81!
zrs{{A@%0;Zf80E)7e=i=BtLN@^*&)5nxCJHJg>f~*$VmXu50`Ph+Yy6eun6G&wr7L
z58|W)Bc=0DF8;1{?*OjvW8tj#w#ZLtcc%Jc-lPIo&Pu)`!
z;vAlYBM!tZ;*f~9Nb#Cz96xrYVIsBPr-<3v+Yyfonb+8%Uoc(a_Gz@!m7AbVowIMX
z+x$jR?|SX=Aa!paN^2C0LR9y2bPq(i6k8}S4EY0=Z8~*sCN$4H^$78-Vf)r9jCana
z*CH5k-IllleZ=igr;F`DKgByAWeK!vJ=`r%#nsHZw~)HWWt91IsPmbUJM~He^8JMS
zk{Ud}`;sH-LXdB2eWxyh`p30Sq6O$@E$O=dJ+3E0=whxe#^>yiRiW0c=@H)~f$~Vd
z@@G`PIxnJMvl{uexq|C=PvQp56b^Hi3G_P8{eQS%zr;GVOK@B5k#q?nA`WHeDT5{cd=*k<*Lu-kG`>O~82J
zTLb4WLb>ezrHiQk*XwXzdNIno!_EklAh$Tz6SfhN7JOjR6dW&c%^7p*opwoMnp+~`
znpv!F>OE<#AAZ~w=X<2NflKvuwXN@-0P>e+pMGyZJG!U55p_T4EuQR0^)XAqBhT8<
zF1k8vvmSD%-D2&}kstd60-op>xujWf3Gy$CN>6`46io7&Nxf5_-DTkaRlh|@f!JDus4-KPS<&)LDg%<
z>lt&XccGzrpgq->qt~|OQtyoTjgm4{pV*eDvzO{$M)x;xsr!paFi=xM`L|^SNw-n{
zjA;XFkl%JOOnrlSNSZe7!}>@UV!qpkj$N4$cr<=*iwl6
zU-vjr@3EPmX4W(!*Ipe^?~FXHKiakxaZifHt4NezwCOFU-ZfrTUc_6JzcM9Ss5~b7
z$*EBH`Ps^{R0Bkz18JA3zTt1P`^tLc_f8G*hB2RO{UZZxk$*3OkP2Lf$g{H^qlm0;
zmnJLVITpW`9Zv1*vG=8CHz3aI>Dpb6SR?Iyo8p&EG_1qXKj-_aj6=wol1)IJCy{Pu
z(SBT)tNn)?)bF%I-iQ6DdG1PWy_tdWKR0)5c1OSNYJ;mRME>tJLh}$0H)?ZY5v|s9
zB^YSW>(;9oMYPpeoFR_$Nvx=Bq555M#i|(_G0xHY==f07=gYoS*n+tFkYlzDjyoaQ
zedIObP{B$i>K)EltJETZ{t>GZ7TiRxsv)Lej(YJ2+8H^>O9P}eDlzWW;@xo>h*1aL
zE?9^0LU$JPM9^*tzp|?wuA|0tGFn^0Ib9rUMnEOyDH
z!!i@G$0Aqhu<7?_dJAvSA%G|B5K%#gUq?K8UzXD$Oh0T|K?xnQ_Xk?sE~Z0}*`bh(
zCMNJNO1SW$h6$@eoJ;t!n4tP-_VNNBCRkd|j*MQ#gswHJ?wr|7;2K)Vw2v~NBkxdr
zMvHb(&vEtjC1c^T5(`fIC1wAPMKk+B
z{ILk%VYhy8u)THD@NPdG-27mnZEZj3%yNkIwd#jDgJTDU-m$>?_LUXZZ&@%;Ynr9h
za-7HItYv%R`ayW@^^@8$4A>f>MUP@KK(k$CU!oxsTDneddGMA2M2fo$VIZVqB-{gI^l#=-}p1_D0|m9p0|+WL;FJL;4Bl
zsG?>XOti77SG-1tD_5)4E*Q|^;ik};{A
zd#FxhfGiw}sWxGNKvessH1vNad20M=-BCK
z5yi&=!?}|;O2u$MfViGF@RU#-=#vj^7?+ry!ZQyqLU10E9VI;c|!y12fw__wWC3i%=Y!(
z&uP%`#I>kDpAH7{1-JW{Opu%`a-wP4DMuP`>OOh7Y13g&kBuZ3>%o1{d}{7f76ctg
zI^X8Wf$~J%EAo@sa6WFwwsr66z`vy_RuJph!pf)Ut2YhAt(VmqFQUWkSq!bpLOQT!
zAOFzX#(?h=R%;z`?T6j`-CF#y9589y5D^~R4<5xXTpxEj%=(rTegV(HZtGCp6s%*%
zNaHz5*axTZ=bP_AOr#mN%i(+;?l74$bb<-hRhiQ_JM_bnX7}24f*g1|9Q&dg;}-Q?
zPLblW!Q@Ez@j0?==$IN}b}qUf_WD|J3)*OK+SsNs0kE&?=#<_)zyPHTZ?~7uEci`b
zv1qFo8-iBo&GHT9z;&%q)zY;bIQF8VOecm7zb(FIVTEVucCP0@^zD6h65rS`
zq&u1>fa`ws%(*4cGiyKj;JNG06o^v5JYK3D7E-;%g5joQhwxGc$fVSd4206)eM9p-DSoW8
zn@_Ek2I#OoUcGBxAq%puF5Qyu&W0no5L!iWKzMtmd0aFHrsY}6$l&^8_Wcqe=*xzc
z?9U!PQB3%E3
z<#n3^?#I%rTpgdqbZC3D_HLUI1Mcc+x+P&h6FfGtWPTHl%QhyG_V>fAirmvY%%64P
o%dF4XXU=cVi%!D*NV&O4SNu&M^w!;-A#k4tX*EkZrg-oE2XZ?Q!~g&Q

diff --git a/previews/PR1624/tutorials/out/solution_000130.vtu b/previews/PR1624/tutorials/out/solution_000130.vtu
index 4710cae87c3d5415f7ae814e07c3dc137b566d6d..a5a6db8c760c27a8dcb6793278a2a61c1815aac8 100644
GIT binary patch
delta 12226
zcmV;zFFnxSl>z3J0kEb60XUNl1T2%&12_UXFp~`gWC1s`as=2Ge}fDF000000C=1o
zS9v^_*%!VEQHH+JmnH{^9fb1etZ0oe-xoPv$w({5kf^XUP*ah
zC$wH*M7paC|6kPI8F!9QeLsVI|+@Mf2bMbx(IdtZm8ae>vP4CvTS@Gj-BmM$GzpMS*9}GgxEnve4_thu
zm4|UAqxa{#T_rSK?qSzz2o@KCPLqhe-^6twGz6|)LezTN$9}h{uR4?
z3H=c-{mA+cLU(Bte4ZJH`TmpiaY!LFsjoGPmrUrHj~)7(|0HzHY%RMZVT4*)e^Rdt
zCe*?9;F8}03C;Ejhnc?-I*mVj(lDCPO@m$c8?j#N#zellXcB5;lyTjEJw8tpl9Sv-
z==e-ofBu!tgdW!n63yB`Xsvg+;4wWy)kM$ON+}b1s!BqvWiFw9s+#92X5qa}GHyN*
zCDazKQ3*bwr6$4&-?9JVMQ_&DJw-f{)4mhJgz8pheSauUsQjI}(n@^)SbJe1onYXk
zIWym&g@K7o$+p%M25|0G-$54!-^FOZ%4!C`f7B_(uUg4KXJBo7?Rp082DX2!`i+6o
z$%{vG3>b8nfA+oN&%jf!PycfSgKcTjf`9ljFt0E_J4hKM=Kr_5T#P~TlPSf(ehz}0
zziP-Dl2TiWpQ!=?6XaIzf4_f6
zej=H{2PLb$HDbu0NsHKQGsJtFxn3*c@N-FKh!FBLNTU4t-7AE4tdv#zaEnmS94BKd
z#Ch#&q3C(lgtqg*J~e|-h3a=i#huVQ`Q-tkn9qMyo}V_I(8drGv2Y_otKCn997S9!
z9{78nJL*93j7R@DHWI2eOX~Y+f8^53TSf9*akmn9fCQM7i27v`}JRd}%J64vpHGe{y|
zc>8$M{Qw
zt6RfRSI818*IlS*!7r|;e-)u_*}Aom2aY7qu_QF`tU;byI-!TvT3q`uevZw)2V;2tkG(4V
z1k^SA!_hYGsKa$8f~)pWtXr6H_tgPBZ+dr^x-kR(p%llUs~q&)e~NyeBhTPa?L76B
z?F>$oew%oK{f*~)+z7||tLh&7#U1OMDN;Gvgzv+wM~esq~zM{Gh|@e-yq_bwoG
z3$HA=H;%#MmGfqgzTsf$#TKXdSPtHsdMwI%%YkS}`=n0}gXfmP{2v-vmpb_)nRvdC
zpw)iLEr^$}@yHJtfBzmd#(l>56aK_Y?VU5B62l!qJH9cn$>Lo*pvRzcRnTG{^1V5*
zr8(b=gN<_0c85_H46OpMnjd7);%e8o9_QU1O(*G17~g)uasCsWH`k(~iX{<`&bnU5
zyO9SOWkyZv*iY3wYk5|I_)g6~NXD@qp#|yTp&Z2eS?a1GfA42h@&eAaa?~
zE!_M}F+*LAoiDPjI*rh{X5HFXy-~B(2Zlv9ofmlMfs+}-2UWoI0rebm29tJA<
z>pu>aVO`qY_(G^hw$Gf;Cm!M8KEGXi#!C*&zZw=zmN58Sru=d>&Uc$6edh}}Z&yT5
zoAn&)^5by3e@I*<@~BO_{^A~-+Y7}W@!m3c{qI$kQxXiq60@~qBRQD2Loewc)W1ln
zZ7z-W?41^_037}kxV)2qk_A%aBUjuf2HygC`SKMzt$^s5Bij&?Y$Q>
z{$em7a_lk>^BCsM)5%{$X!k&K-tG`WO><7m|MU$ao5iA@BVNtpyKElfy)Lhb^6o(&
zASdTvb>Q8-gR?VwS4uS$`@s4L4OwCQR+q{lJtICBS(Iy6C
zCvV%Ae=1Yk7z>+B#X0T9xA?aT@if_((T?#4=zemK3Ln5=;QapzdIiWT~oss|IzxAELT
zYb5MUe)_FZ>CiUB!zRgmD|?(j8TUR;%tgIul+;(Bj&-^oEiQTh@wSVVUe*cZV2$q3
zpN9}H;VI<5)#wxK8cI#R;C+_A^>R$Ze|h98>MQ4s{pqq_dm#dS+2rz-b84{Pg%OF4FPdu0*-o=5P;1#vI
zjts8m@C>u25qf^Wa@jhEpMEuJwf9Lvg;K@$?Zv(;B^e2po+i{Rgh8s&A$r$&3C@+bc-4e=8s%|^JzZ^w}+0ddmN5^=@RLi=;NTpZ+nIq>b<9iS(!TO
z^kpKeV~hHC(qaD(VbsUn*F?_OVP9(WMD6WSzq+5d4?e&-BhPe-597SpRsBTY+mJ!;
z?4y;Zi_u@~=`D+M=OE6z;~$MUe-4J7>H?=B-g3o<6ztH4bsgTOVT*pSVrONjKJsVY
z)2xVJ5w{#>o6p+>iu(JzFyinJ&AAYZHWiFlPi*5%;be{P0(dMZD@
zUG>w4cideRi*-|`!)tA^KhMtFobqgBkW`=RF)@k0OzX|5H3v8_U0A-r0)5kOL3))N
z(FZB^ioE)In86(Hvl07IAHp|X^|+1S8xA3&GX1|3y6A*q;X~AYi+Inoy;wJk()Q`R@_lK2gi#3D)0o`{J
zPm|3j7=Pvp00000004NL9hY}J)q5Dnk&H5;VT6Vv5;sKV5f#dclw=lVYe+_AB}(eH
ztjt0oj#WB!O7n({6PK={+Y~OBs7@W^exCc+`}K_P^ZSnH`#j&Dc~@z$KmRBrpRZ)2
zB{;eG&o>+RMj1D+{+lWb`2o+g?ipwsJTtwXdCru7F<-1lBf!r+)6g
zcOINd6|zpB9$`3{<%PbU8ev3p+db`njWAY>oUyj$jxba#9Mt%Jjxb)0E#Jxp{pP5a
z_wq&Z=x|D7CREbaVpj1B)Ax45loh*NgO=aM|)cC+^0xeO|(#VlZfvPZM(Tj~25
zVt?^1AmgqI-!Y6^cO-d}2gaR0<~Zv;WMS_hi%C#G`KN9j^q;dLbvi)H$Q=H4VCWwY
zH66hr|GAG<5NAvAnBo9rtHDKb)1YY)^O6+S-7hBgn>y&|Z}m7CepXgp#lL}bH_Tra
z4QhI6)YXCeduC*MgHb1q%5Gu&Sk~TZ4u6>TXG!92=&g(7q{kpXF`X)T2XeVW?;B_x
zNgI7>|NRsVqRkWQ!~1bcs0#P(3!H)3^Lz^1$`L02%CRsVqGp|*bYVLET2l2dMwO;jb<
z()UaKwKv#Lr2$JE>gO~k#}
zkQ8^H_OF<8b2rAn!2mBgU&t(L_Vm?>5yr*bE!mX6!w&u|I)|RnJsST&_Do)-_lf*&jA_3E
z&0_*?w?pq6+q~c-;(wMajDI)-d1FTA?}e0~zFQeA=wJ1D?v#OltWfC`3-P-@-;H&K
z-BmtDB?g7Ag-eR
zYrPJzc8`5&DeNgP5>%WJf0M>CtJuTh<)+Q@qehAh5cLayYuHk
zP|~eA@BjRZW9x2*z12aw{v_hRN$yZN2RkQTLC6;IUxXRSTT*`9Y(IMsa)+0QpEI4K
zt4qi?$l_0A)+~YDvmrrX663?`+~)oTR@gm~Dgg)N+)eY~SJ*P&_7&=v*ZSn7Fz8dv
zx-J{jJv`KL_yNWrN`Gg}=EBd@*DoUu_SzVI&K_{u#^jS1Xg+#%-cQ&CedBZ1z<=Hx
zvjMuF+TC_s@^dqL+r1L=RuffQ=?c0DRAgn-y;2M^=Epv0Y%6OjgI~V7Ph2J9ST>wE
z6$*YaiO(aZTNqjnV15STv&Xl=?`Bc2Vn{6YJbRppY;IqVFKdoXnI&q
zLAICIyq<$~6d7dsT*bJ@LIE7AL#@gdvsb`=FtK=U1UOM0HC~1IcXytC!2{o4S*>~z
zeru|>2RI^rm&Pad1Nv^t@o+x)dzZdE5d*rH3^ODUpVOhWZYS)go(AeseO%_kc3T8_
ztDw$=8+d2>fSSCKgH4y%SF4DuP$G#cXteenNOCT8Q>DUbNv0*5~h9d%uYGPwwPcF#Ila
z!q={%`WLUbhw6WV*4h~&RF`xjFZ^Ez!$c)Mf>O1=*V;mF6Cztg_b=^I{V09s>6Q&W
z?TDYxtbfTl5Asw8j#k6pqrBXc31%NS6O&EzX#FTd_0-GVinE{kr!aQ`^|?dSOBI`8
zU;cUWuqxz_mSwioN3OqZ)lcVt&VNqM*q!=HIa*6yw2rsl#=-lpi;uf~lVfxr!xGpTD+OI*s;;9VV{3
z41anxgYj58*S^Z8Xko}NI$tbXL49uMK2wck^*Q?DbpEvm?$f=wX*hQMZ*bS)@8zU-
zI`OfP&gZU^g3lYwTdE?(_&Mx_<;P#pzU|P{-c5OXdsj{30=gFx&w2FTYL1dkT?B3(
zR)}nYU7{nI{hH=wC~!!V>PDZgIn`BNlYeHD>(H+~YdxRNPh2!1PlD!CFCjwnsd@k1
zUZ3)IYAoFXa;J;UF@=qzqb@lZ(>3@8(
zs?9~2v>$EHdg$B>rPBr}pHhN996kj93k`ir!qgXR6~w+kufX`@Z|aMdM^lt3f0g?s
zHRye2CYXOK5Ui3h38r(|=d`?%=CR4y_5{uAW1{Eicjz;9-59^XM
zOD~!-DPHL;^;17+KU)OP@IdvrMStJ-!mmAE$>}y|#N{$6Put_nFBQNZlHXpIf%%mS
zTjvf?{WxV(pbPtlSrv8k-ZZ__`gA7h-1)=jSbxAj+FnnD-W^{9^SkI>STV#2PY0hY
zbe_0C>n~V5PV-WqZ~j<`*0D%hx({)K^8Cz->Hg#^>eBs_Er`B+7vxn{o`1{-=bGni
zr1c)F**ik{-0UUylHzo}Y0G{JnXyAjxQ+T%v&|~%3*J?}Tza2)?UCI~?<%Fp>fQ7%
zSfKyPfbu8Jv&H>1>?eNvp7}TGQmNah0KMmQazok6DKCq++fly@7OYuv6Y&m(n-0wc
zGlHGds1CQsOqg?F->|3Oj(^^Td?Gw;dS|(>;|WNUzQ*cV8sd&Ux7a~_`H7gtJS%!v
zU2JlscYO|{pP%yJhnlW*6}ZyTJA~?>kW%PodY8_M%-Tuo%gN8%NBd;U%-#`Bech+o
z;{upwH@5sf)o&g7q@{EYU9Y`j;TNSD!C6h~J5qj-?#1b<@Dj>zA%EX6>0B^kly4u+
zb4Kd^qqIK{DxIAYD6YcxJ@lV3)DzJA57oWeeCZUN>nFjOhH7w8%jw2)>QmYoMpQSO
z&9t;AFB9ah=TaY=sy4o51HHj{#m)inuhO@wKfvL-$bEVso2xSGA@;+}Gqo%j@0~k(
z*CK>KKjxgNulVoz4+MXi3--S9tOLGa(n6K;QE=$^!IOWy
zh$r0US*SsT2rJ!@g@3
z&9eZyO&KYREkR+o%X3&5_u2Qr^)F$&>0C-{#dz;KV~1ui4{x#c%&XAbB{16k7ULa~
z@9|H?IL;5`mOh97b;2oMQPhVR%CKwj9o9ngGU}nzsv}!#iTYdJXNyc=XC!|!MZ7_G
z_j}?N(C-Lc=63~gQBH?=Hgq)8|NHS8R16b5`4#$aj>O27W4x}>ZF9aLZ}#unm^5%u
zzUjz9V1t_`5xnbtZpupgr3GG
z{zxO}uL$2CyBGG4smV+pfJ+bJO7kG|q
zv|1egpfATKBfwhjXEmoW|C;wUH%M=1ZqBVDu0h7h#qSCwV7I!g?)e9KfqxW29)rsS
z>u<`yU(#4H=MikJdlHjV&^@DTbF&51kM67Yqj@#+=K3K%!+iXR#*=@$l;ukM&udSu
z)+p-b^e#N~0iX0b?SN9oYS`s^;%tS8r|&HUWn&II~v6
zpIRt+V;kx>JF6v0V%;?2+-_b6qgA&Ya0L@%k3{$*?(>(_mRHDcX!IY6gY7yYXXuLm
zI(*nnToigzRi3MNA?~B}yygV#R{`GsR*0vDK9iAy{oH2T*nWS+r3w;m_M-l@<1PM3
z=wCLj=O}~i_~ce;F5(qGn!9Z1|7#bVO+o!w{GQK)Sf`q!pKLcEF8RQ!atdTG=55ad
zr(!s_XG8x_S?7uY?1SWtoxiML_xA?0Rv@lq{X;MrdQJJqDonw^ppS2rK&`%%diqYM
ztU~vvsN-b#Gzx#d4n%)Q*m^NnaN4n^H2XI|SGzN21kot`=D0MScnoQPj^
zw~S{%Z>{OU66#Nbl?GbI;Q2i&muTMyNV&QWBmYa!7lCrHZP!jNW$K?I_oafs;qk&+
zC$Q08Zu>LDL#n?$qWLrg39O<%mK_$BD+K-cugXswz=9WD)rO!VL(EZ}`UB_N#(3oI
zmf8?Q^$dT75;qx8xQw@s_P6?8WnLG=AKcef5CeJjR@CKS9vK2J%tc{OIyu`N1W!#j
z%r1mp%hO$c>Y#SW{3;r!qCj*T7pO0okV*SXOq8R{47zbkFAppSW9ENOm811)IOpO4
zzFe+tk_!DOu0IA6#K5-Y2S7E+#Jjwprk&|~Q`CR$f7{hC0^apV|9KI02^mdNFQ~u&
z>R-MN%;jR+k^Y;8!&(`jn1ubcc3O{unp6ic&%RAI7W>mUV8@-);1-@eX*uY>V|D5!
zH?eQ7MJimI0K0ErHLbHp3h)$f9zqe?iG
z&i#KA>+j=DsM}#1Ua5|F!}QW+W6(4HreZS><5}NX8{!F@Eq&458g`WT6}OehE4o*l
z*99h(>{}#4|9h3@!X)TeMr9t|0{ukQ*4{+u@4qpJMe9C&=%raL^+$)QhEU{>UA^cN
z2ky9c|63;NPWQ`kX@g6)H-x?b#nNL&?Gb+uHI-(`f@QxsqHN*+ui5W#J9y?}pg}$K
zqJ31SXJ|fR&ofDX-}0nFI=|AJ8y?;P7hHL2l}W7Hmqz>Jr&seTRnWP=(0moRQTF;2
z-5*CPb5D&zH)z9pzxklb+RP=?Kg;wgLe%_Qs6(>XyZxy{bj4k|NA+|=!0Q8FT=V5%_Yz;44X>K!g?~8PT?!5
zU+w&5)riOIm~=kumv;)&IJOcw`P64*ZM*ID=pM7vadM&anb{~&2mTVx
zVx5G4d$4eW68LbQMQ$hbuIay#(no)Ma9u8w&Zklle>?}`PtwZVXJMZY{O<7q+&O6e
zvIg`TYYBKm`GR(Diox&#>AoKDQ?XYF^>_B%rPxeo30PIp>F5KxY97j`?_IH7
znPVxqX0M>45ZKxNr`H|zM?S_Vc3|FN|AicQ4Cc!@DE^@Pe9Ghoo#(27a6Nxl>dW&p
z_?Y0%lWvkl^#4=NMaW@Y`BYu34HueXZx`rx56w_-E9m
zfI5qh7CwbPa&Thf3FIBGc+N+C?(wovG5YRpMhsRU@*)*4jLueA`X3u!@5&_qp1b}P+iCrS^j}jyaQ0c)PVa%kjolL*piSAeyVTcs-k&Hg
z1eZq3oRy(>NN;Ne-D|^TvFjqiI{mTLbl?Bh3q28t^~qhAmp_60Ilh10Ec&0$v%W%-
z^q#o5C7DTmY^z{53*WQ$eu{&~xx93Le4!
zlRMQX^b~fS>M;|0%&y&mOh)Clr34LbHwzn>ON7oqh?WX0&yJ<&F6O7G6O
z%?qQ>(>R^^Vc(#?CTG90Fv#truSe?@ne)5wEA8J|mqhvxeJX7d^j=mCK45*B-U)VJ
zG^t-$nkj9hdy2bzNkBFAr@3byp8?;d{M&5_Zj!d(SxdAUF6Dm%MI0lP=$?#Y_W96x
z^K~-boC?abD!pj^O#-I%>D|1;v$5Y1T;{(|f%=NrxqsX_G5%~TxBPn4ch~*%g5Hx~
zk8f3FMx000000C=1gmU&!F`TxhYiqeAYG}^Q%Efd8zue2)4P-$c-CX7;AZIz<5
z2vJcXAzNva7-|wlsF8+G5IudRev0)Euk?Wf2lgb{2q;n^m&cr@6))j#(Jdq0gdsR
z^Aq&TX}Ha+QAsMJ!K7E!)!e0V*)L*cc@YhB%cWL*YdG{c_Iux+&%s0Pg*i3KLjC^L
z(YRI?X&EoK9V})mH107fW^fHfwqQgSu~o4Ue?iPvAJh>
ze`VRI7#npBinXi6=wz23j&>C=EqI|-x~hP`?w(fc9pX{rPiddox)vG+p&A4!tlRUx|B4lO$=3ytjeJI|L#|;_DGQam6yi=0>
z?BWGXuf14#;+TMN!^<7>;sgW<%~>kRe*zBA{~B`gi~yS-g*HQ}0`i?MS&fkI1Acuv
zxb>s}Q`K3!osSEMUsgTr6(b;Sd(tNJKmmK5`l2=L1Ze$NFFUc4$CAK7g#kGMpGMOo
zJv9Z~ZV4$bwGdF&N6(3}7LXBLt0QkIV7Y5fznzJIhjh7_!$JYu72CVJ7YPV{f2e%1
zce#Li<)QkOn*_A%lW$mhNI>sm6@Mn5!|LAdU8iRA@NEp2TIj?>{*k`Yyp25cd)E9`
z;=-e&W|7_*Jsuh}@11Y_$suY-$jxsp9NgTOj_rTW!KnC1iQQWcuS?E9teV85zam92
z59N`b^DMY6iNy!m3a2PJ4u0*je;HGka9AE=8RKWlLFe(sDQl%TJl#z1|5n0c^4%yW
z8xIy`y>7jEvMl}{FR2{ZD8|++W;rwJ#b`d4^hQsD#f(PJtE&@P$VAQ#y=^DP$4uAD
zlgRH1?Q_(%`o!3z>Gb*VYcW(7AKJ(q5+i)K{(4<`F*HR>2gXh^Xiu-3fBs#aLAUw=
zk6Z6)IKIei3a_N0^Qg<{QWuSixK0bHzZf)oEL$N}N5jYE&;#)r2Ei5@zEAuaJQ3fu
zjR%8>+OGcGHX8ra)z;P8Nkb`NRnO1>h1WKzXP0GCh=1w!=MHZQUv3P@IXh8sa`|~;
zt0x8DIrcj*6i}#2I%vFSehi3hF!<5uxLFv5m+NZ^
zU+Pi_IW8pK7!)Bdp*nJ~O$674K6()yA}szp{;UTjg1o(B&WkP)nl#rm{-i~q@2n3E
z?-3!pE24O^NQBnsdY7K?B5W8G1;wx;4B1F!c97?k(wi@+4U3S@e^0)7Rh#@zH~Fkx
z{uGQy?`*QYL!tLr?>puVh1m9PzuWl~PMM_*c-T{z>giddcAG}z9!saHAqpQqdtEax
zrBH46NN?{$3PJ1amkdkMn0vCpp9-VlnR3hM^B4`UI?w5iM;L@p$oHMt!9a1&{L#8T
z25rIh!;+^NNGJVweXpYkPb=h+yhV1Ng^d6Y6SeYeqQ~4U$^Dzd1x)CD6?b;J04v(~^}YZe0o!@Qz>^%l
z?K_{DPwM5)V{LckilMjL+4oa53+bmXI}=}U=ngHi-gcSCeu^scUC!T2?$&j>5(4B!*EyCp4-VB&buGOsyG$_Z%q#^Abp)XPREwcW^rzr
z^7mdV4)!aic~{!-__sbKrbI@-Saaywx-bFWt?1DrJkb5T!eo=JfLR8e`?9lnl-BsT
z
n2x3+CK3kVQn+5TpwduflyK2JvtsEq4kF3)S=25)C
z_46rX0d5Tzifa=ESY%i@Xay4vI5{kt*vrF7xE66ko=3G?iuZ?h7FG2h#+IjvQGK|q
z^x|d)vn2T|hKCvGUh{D&7Kt$yW6)z|$Dz4o@cluee{YxF!J~~U1hDxz*6Ak%_(pD<
z?7v2U^2N%Q{uCYujXd-%gE^>*i?(G?W?^J`Y;5^82F^VnlCSvEc#x%c`PV27WlnF!
z{0K2zmrUxhlOcSFe6m{Ci^sFy2c90W7qC`t)#H9*r*?p
z)68N=e`4;NE(bBLc~{LkUr(bU{bXm^O$tvZRXgz^G^W2cei|hu#vNMynyB0i
zT{55I-8&s;e&P`m7+J06%i;HdlA!n$F-(dd&)8B+!`M!j`xZ_?PJAF@H6udwu)^-M
zUD1m
zf6ZcgkB@n+h{4^{f}V{!jj{miN1`bd-Wq&&p`dL)%uQn6seBOZ%|)RrA&
zE)7|==7E#RoJN$_TOQ`1Z(u)J+eeJgMTJLV30EhLbbS)k36F%1y42BRN3+nYzeuKjgvsrGLq_e_<{bJ#&X52_?SODbkC
zyyz~kLh3X+CAzeed}1?GBD#ar(+QthrF)x4!*Z9rJ;b*Wlc7J9B*0JEBX|YzSsV3J
zGMe)Q>~oO3@|fu6<&0>%j1wG0f3JJTPjV!F5Z0*ws-MEEx=-JgiJy!$?N56!!eG@i
zMH8omEPgZvsY<(Z(AX~HzQBV=kz#?V9pRal_OI=Vq`wry60-=BJBG&YD&*=3NbTuL
zP~XGDYD~E9K>AX37P5UtMJUzPUp9Lmg^Z5*^?A-T&ZYVKAKJxWPun!fe<%53#225B
zP90=1bx>F2PV#`;g@#bcpF9H1Z0w681aQl;|L9B*aOcg0`Jksi)C&-W-
zc=_tnawEd=(b<7#j*@#=cXpGpKjHL(!)Lt~M$!1PsPq&QL_zChe}(x2L4=@yN?Ff+bc!`C2Kj%pwj;ADjN@<4LB&yDXY6iV_J#7lgkpr_pRFlz_7
zgXWUOFLYU48{@B!Yvs`I*6y}`CE=pN!ANi7AAING;PCf6f1Ib5G1K$bR-Vs-DU=reNTcadV#&jo4XTjiO-&Dd8QI(gzlM
zwKhluwR4cVDMa}bu74SF^wB3A{$4iFu&0Gbiu;cOchXPq(t)yf?qbNim;1Udmd5M`
zjen)=C^$q7f8V-5c%$}9a@2&>jeNEN~OZOQt7-4E8~k4s>YSE)(Q@n-O5oMoFL(T~sM(h-n+))p1FF&HqL6A=A2I-KFp+Oo+fnn(G8e)do&#~X*
z-QV{5ver+y&TF0f=|~&tNDbx4sqfyfrOqP@z82(7#d>3n#MSKaW`zg{eD~`Aj$bZ?
zp0DWQ1u-z5QIBiFtyon}M;$1ZI2K#nSxjSNomRtWot~VB%SNrCsjEw)NM=V~L{t3x
zFemN!5jG8lF5f%_o%bB%9^~Fo3V7$p-_JUrD$T(Ac>ux)e`YVAwif*my;Q%)s$
z(YG;3WCx~h-IpQgR+RHgY>d2)4a;vfY1zj8!CzK*F}Z{QIJWB%7kAw6)0Q`Ra%{&=
zN*-a%8)9E@Vpn7gYFn1xx{WLy9iH_5fDkQyLwC#DWYx^B4z9--FQt0C&0T74f*Muh
znU78PVRG803^XZHDLE0_xc1(iJq%)RaKRK@-bWNMPD>q2?YopT#^(Y9qtpe)mHK*P
z!IiF6_dI7n~jo~2=
z{s=iEqWQy(QzcGJbG;LlI6A{E8=@Gb#~m7Y2q`c$(L|epa2v&v<1$?ZxeMK)(pcn@
z;DQ2V$7NOHhX;hb&f02lx)CqK79O_vGzARo%=Z99YPawgeU7+fsqMeMRTJ^qiZ_2;
za%f}xei2N+9iisFS6UA=;-RU?0f8FKST=`Y($3q4B0R6*7!WM@o}fMfRtwKub=ghGwvfT3nBD>RKClGjgYq}edvUukIfX!GGDn|&e0d3y7{aky4C
zGD!)ewQ
z)sL!ali-^|>g01|0%T!E&?Bgkovctfs`O%j9InAU~9ZS1^oj7?A)Qd&Q6xBk%
zE^H4?_3-;c8Jm|4Ir0o=tfR0QUcV{#BLsID`g07^_J|352x6qX=w`ENPPDub1aq4R
z5{r_tznr0YtG0|y#+Um^rs|~R$Xx-%AtqN)^@HZPl;&17CRN?`2d$>$?Px2*gL*muWnD>?cf2G3{LvSoE^Kzd1`p
zRIwi2_=o6Tfp3}Lb-49)pZbd@Q4h^(>*YBo4_x-qFVL3WiIePdRJG>xgVg}aKdSUx
ze$xE`X0`8W_UCmQ|Lw;-v=}%*!o*-Y8KjfQf>;E8V=@X
zCS1vK)JW|{k+iB!y;@%il6LZ4>4GT&NMpoY)Dwz3;SdyZ^eFO=9#in3ONAB3t-*St
zPGp?Y8*hY1`}eVMe>?!=RRy3iI{9USIz(X#z9reW#XihO)7Wsqv$NWZO&MdCaH!<<
zO09$LLtvc=4(tG2V#5>>lNTYz;!56)C?zkU(adGBtbyw)z!0V45+P*
z)hQW&)7YhU_>ol>TRW#N9to!RSk+Mz*`EB`+fj@Iw9lGrA*$A`T
zT;ui5S$J-KM|q!laVarm4pHcLWen@b&fFyv`2JCoRK#J<4kxHmlgo@~yZ&ue2R=sk
zK+E3%DEnq}QX7`ya35f#>*5@{UxBM}t#`V(5-)#xuK`;8+lP}JElQw@Y`Q;rMUX$U
z$y2V9Fla<#>@TW1_kwmhxP`B9Bk*zCur#M4EkQn4bK*QCrU3K@`e<8OaNlHr-sjlL
zG?%j;7R^EKe$s^}`%T8b%LR+uPcJ~PqKSI*Nxw;ElPwk^`U7}LVVT(M?u;7GK1-^H
zK^Kg&E^sqO!ET2VMl7dPzK5Pjz?~QVQ05$A@Gj=fVZ-n54N*PYN)JVyu3R)98e

zbv`V5gY>M@e*6^lKyEBOkynq~gL~O~Wv3+GAc?Qs{S{3fpk{6V&vSz>!2h;Wx6&hZ
zz?nnRq{%3BlArx?nhvkq}c>-)c7digHkN8*h3@_ZoZ{$OH|
z`L_U&`O1lt{HOv|UUyuaRpUU?6v^Q#e=^GInEMq<7mNHGuXCka*n?9*rmv^4A2ff<
z%a+~j11ASR+9aucN55Svx{a_})RIw5(euC_+@5u`S{8K!s(K;p$)=L%peKVbp2isc
zb3LK(tmq0lBHA#X9qf&69=+CizT+{v`y%tw?%o1&XBs$($S(6gG+UY*)#ZOPDv{2Aw40A9pr)&SwTGwjzPfg
z=PK@xhob3x&mPi~0!Z+y(Jk644Dj2X4J?>D3VatAc!is3P@aHZ*4J`!ki@}%_nWZ>
z(lv1_7b%oO!ffzG*2lBJh>tHyKcX6=blqB7ilsoqk3ZQCs2KxJxivFOGDV=?TqAP+
z?IpBE?c$pL@+gqL>Zf_xng(dA0ESE%M^t)sVW*gf9-U)%;#sCu0!OJ6XVIn*dU(}6
z|Hx%Vt
z$eV-H50?d%UOI{RtlI65XdDCSY_^Mf62GuFex|Pu)(7#}+j|kepEqOWPfuhn^j=4~
zR{Lbs>~u)oefVYj33_nJ%h-NWb{h{aCrPPWo5rORIo6Ci9%4(!g*5nXsN&aEYFD%9
zt?*rQTO0OWG5r3IRBhHP=9t>OkaKU7Vlm}w=~TaY+_BKlD&!o@Pq0`G*SU=pU3}nJ
zc)zHP34UzTIyZJa6F1`gwxnYffIIwVu;aaBhyQz+cTs587PC{i#CxItF-9G>A+_IG
zh3S~b1(aImVgGWTozb1q#1o%a8*j%1;st?(Pp&PVSX>!NhwHI9Jburhxl^hV%hR~7
zLw>ayD+p2pFziT>k&gHTr2F|<~ba-?DXYlA_@6(sYN&m^)uTMy0
zY_-8voK6C`tCP#s5S1smVufd5Vu(7nwdTdBTu+Th?t6k)5`4ISSAArq+Eq+QdFG5V
zxiKcE`N_*#(*zHQ6c~=0#;_!{zN-sNw3vX0e{t7s1zcrTcK#1ZFgCwAHaRBff^qL3
zFw`hp;4xO3;&s>qOw47~b#e4MHWbF$SN8oap6s7B$wiicJ)*Xmyp`mJ8@p7gkge+A
z5n^r|I_{nr{oah_BB?aaD!d)9bAt)5>S$DZ_2xJ>_w39u`<K3VF4~%iwtk%@vMZJ^m}nk*x1HP(7EGkE%hM;i^=VYi?eeWWRuR8O7Z`6{*NR(Car8^?H
zxc?Q~h}CS_$NTC5aFLuJqL_Nt*InZIh
z)WaC%?YVKWm$yU`;zco^h}hQ_QaA98l|rnf)eE;T=`fZ22yneK^RIX9#IXr4+s2V?
zaa>OMJs%65Fy8Zts$Z`58paqhSmUUuf<1PYQ$JTEiN7Im@fIzLVKIC$32wxE;>YTk
z^tU1ltoH@{+^cvRpV<9-$0Se&-wpOP`BA2U!SORs|LniU`HxO01yjAl1!=WJ&iQ%a
zO=Xuhhpim3h3`%|r6tdC5H)qVChRpX<$iOhp*99z5slXHihPIrz1Cp6B={aH$h}d0
zq$3!AQ9eAc?f4!0=X*T=?@9yab*U>h|JeszCT#uVooX-a${7@?H|&c$)L)q4_!@z6
zV;AI@e0y>46ce8*xmYYYK85tASvbD>;ddFIZY;L+M?rw3{C4
z&c=s3!X4|JbFd?~hV#p|igDTrxhO;bR_yiVz~#1Y1Nf
zFCjtRjmy5WGkn`OgPBg;KG~Jdge=}T57OUFz_Tp*+V2$A;8V}WLO3*kV)Vrq9E_%y
zaC^n&S(|!Jln};dsngnzo74OVP;uzOzS;}h>-u)%4lZBrRT+=qMF-AiBbnqts>E4M
z?DjMU?max<*y+QCviVoeleJ+C{DVfqPWu?kX}cR4rxURQ6@#b7Qq#EPS|iVY5{>xZ
zPiJDjMy+F1trC{4ZJg*n)k;hD-bM7;!Kcld>pZf&XaC&s5d&blsa=&2LWd3<;v@Tw
zcoD@#!RVZbKFoXnsLb9wUi9Mp1H}S)Uf{qRvKg4ck8C`Ta)mSTfz`dKaX-5oK)TIM
zhWr=;FwkLogq349+`kavC6CMp(>5nL6M(Az{L^0@_Q0&t>emI>jQssnhBDhIQbIOjXa=xnCeiE8j1yEDurBU;Xwps2Z?}
z2#Qh#Hlv}WFHG;<5u!NynUM2y8kC&V{?oSS3a(B}o6#O*A*C4KXg9|K
z7@yo2%sLf|g!`18NR#5hW4h%52f+Yz)LxssLcJGobG*Itw{8}(hx!DcGZ;rcUnxJ^
zITFyF$C;}!QNO^hd%w5lopt2!RBpE{aUMupHESMYr-0?xlK*4=m5W@TN&Ea2TSTXp
zMFdW#93sEw3+5Rw&%vK2J$_ZM8DO%T;gf$_Js^YQ@29)Dv*4V7G;5dBKP2h#<^{=C
zFX(NT-WI-j6sk9>#+VA#fn&%#;iXk%5H7m{a@1j7v)XaUP4$8cTweh4m3a?I
z=0?z_TjXll#1;@Iza7JTV;X3DuDthZoStCW;^ydUM@bOQowNKtxC3^n$nJcXTnCMV
zgg<-M7YK%WSjh!TO8EAR8KIl?0$f>eN_X93g1m~7TONs`gti+D=J@0|3a&lQeDv*2
zxF&Tu()+0#{2_V&^GBMa(7se&!=H%+THmIr+JRdjKA*SplPwOYDU+usdXAx#dt|5o1pY!3nRiw1i*+Hs;s5$VcBu($=4CtTx)g+;
z=%SHm_!^RJoKDOwssYJ-pNS${oKS8sJ|=2*
z6O=Z@X`-EOz#>L8a8<}dwJr4;ZgWxCkY%I2VsQrk;ZLk-3S9?3249^iSf?W}CD>Z6
zk+-6UzPd&dHc4pxcUk(;_CAn@WFzhk#{+uL1|tcQY7`k53fsi$kea0eMiDfM)Que2
z8m_b<>5TUbAmyoyKL2UEduQ#~=COS|BVXdgk>xDNUL*J70XsTiFSH~XaW
z{s*`*FIU8!6a)H0-F#Yy-lNISdK-6U8$pKVOfK9^K%aTUE`2_okF3>19I^0Ez(VbK
zr1W+K*w<%E{xnvHR^IQqyqK;5wL;z|XR```>vqs0VI>vul+_J!J%~kLZ;Lj~O!cEB
zMeF2eUnh}f`(%F|c>?&?z4M8dr4Fz%*&k5Wc>~R&^F7;L|AFK=PQLD=>qsSFLpX$Q
z6=|)-#b5v41Due4vO!Eg_H-79hpp*fAOu+yV`GUQXZxi;3I~1`A{wd3G;+R2hhmY4%q!DT3#!FP(I0xt9XK&8fjm>bnE&|;g}bTm5$94C3ZSaICcj8%(atq@o+OfR!TU;D#Wq91%xg^^9LcEbeidA!e4rBT^I5pzkiIaAwHC}YC
z#`pf{#-FgM!RWV(1$2$0@Gs#!zLJ!y_>Y%m4X?h=VJb!@od?q8*uYMV_+z?u{8MJD
zIoPhoX!;GcZx@|L6{lp-B?~Haa%h3B(ftp`Iz~Yq-(HT{n~jeIQ!L;V$#&d!0t#R*
zobTseATPR5?BEJ~VQAtsT6_o>>J@UKYr~**<2qE_bKlkzS&f8efAjUw2ZsbJOSPG
z!gep5+J__fQgj;@w`RmvYo3dV;{$KLe~-iSL|F~VSw3MFdJ}V6S+B90r?+{>ra;J#-{fKI3316)3OV?VIB#x|v~L(*CV30mP@>2s{|}TtEBJ|z;72d{
zC}#Hjyfvq93tpB!Y8lQ~f@@b?YH~jkge9|{+je0W#_FRce@ge=!yPW^VNy6xCkf*IOFj?vvfd@*szfR22)B{ooY0^|iaW8AxUgy_Sl}#2kv3pZI{i4wKjD
zF?fNmvN9C<Azv)E{D0apXrXUP{5kmL1fh>+&$#Q_rH;-@L_7-TM5zFG&|4
ze)pU@JkA#zl`mjbi{rsTYhsW!KQqQGkRcjY%!%vJYJ3)?P{53%zGT)oc;eRAc8W(f
z&S4LeF1C{%=ffS9~)JzxR^zT%gB&Y?P#6vw+16f0Xo~L*|w}7UInJp6$se{JXZ*$(0^^y!5DF
z%>&?%Q-|=9&?W%mJyaknRv0t=EC;xevHR-JcjD}dGtKiQtjKR4kD~b?}kTE2d-)R&R|?>0PA}5
zu}{Zz4S!KrRq4sa0KO_2aLsnq;3b|H?$XC~V}Gvq=x5Z82qB*$*kx){6ml9(EEpz=M>N%5Pq=KL_mVv+j;XaG=So
zJc9@hX7uc5gHdm-9#|S`J86>741T#S2~jj`;U_ef{{~c0pzMQ~N`VK8pz7D>NGl3@
zB>jYLdgtm8-bO#ibnX!&`nX;ZSWGGmR`he;N6D(AE*e?U!U_zf+29m8;FCX-oqKc6>
z0`C>xg2uwzK
z6Zrif*T`c~m+JP$tK%seZPRd&Uv>i48|WXIg$HV*v;6dMwg2AA9ponIf-Ml-^z!n;YA%YRpE0PYP!^3kpGiQ@?uPk0O*jU3B<7s}P+e
zKl}Jn!*^sHq@fi%5(wI#m_Gac&>D5xbQj->&j-rq%A3yZkAQcM|I~u`W5El@c-H5_
zK1h{lGg6*^2EKcS=oHYzAfK*-r>AtMP}coB*?-uL5c_Q!nKCJV+Q=+0))tUg9>1C}wuf5c
zX8tnW4n(ybk-x>R7JzPnf`2YB5$qJlo>dF~5Aj=byWA9HC#2BlbobzWAi48R`=yjM
zV@B3%4C+V}mW1YzU-U&R6qZbHwTu%GWD
zGvS@o#3L*78!&zPR~|0~1|7~M>5JOh5%?)ZP9#XPLJ?*O{WdFcg6ww{GOZ71VSfIe
z8No^!mQ@MOo-dSuR7uOhF*-a1p4Y3=w?FxoV=Q
zxgo)&pSs~*i5N_%C1F?Qd^%kb`(X9SjC@+2nlp3o&}
z`mcJ@Gnjcc${&$w5UO|=H(#k-BhEdg7B`vkK}oko;l*N7!e;&l3*Vq~yfgwrvVU)ZcEsc&iPSUon+18!14J
zvcvmdA6|t%KdO_o`gjSezm95(n`;nsIW(tRqOA$J<4bn54{QmK>NnFbu9Bdc4f_GSjza7=#l{YwGqDQ~Ji@!A9zTHmB&tj7t|
zjm6@+=`;i{_x84cWmdwM6g_e)PBH?eOu#jZrEO6A@5z_iQ=Bk2?nxP4q7b~~6Qwn)
z%?%q3ckUF;{)Z;|zf5~_G7@&C9bD`t1PIxj&*)AB$Pna>JC65uF~TBSRTJW_KP(^N
zO`v$h0}nq|7pEDWCsY(y8L}4NCfI};Sf3fuAOw7%VO8GefdP!Ym;Csz!~NE^rEZpe
zko@9fT#xS@x;bfW)kV6545w=G0$a%7*f^P-zF3kaEROkAJ&1{U&_hw_Wxq2!05
z{N=2Ui0oOOjc`gG;$KLwj?}3|rw5DsmAJyu5v2a~#`GuP_DiXKwBr@{i0Vx`?<|3&
zo2Lughxz#@X}Wbns_e
z>7Vz3^66lOzr_CRU3+t@G_^aRqf}Pm1iKEfecmg!gnBUVE+qEauI-N=jB|2GoL*&_FnK&H?BbWkRzC(5hvYPXuq`04yIb@pLWhpVrPY%uhj|8
z;6kp?3U)SzvGhch$3l_C_?1UIv!CDcqO?tC`^R=u7}F!G%Ajj=7(I_hnYqv`K1tr@
z_EU5aTYSl@6>y&kz%lo>l>$z5{Efu2SM)NLI+A$roc%Zk<6Dw^>UQxjz0Vl~?;3#G
z4Kb_ayF939zA@Rkn+>JzPmHbCu!GnQZTZ4P1|;)N=7ix!F@CpU|6OZm8CG|Jn~Lp^
zKdv-k(sF9d8ovlH?bSvFW8vc*o-}-I_^Ki!&17E%?qeR~&wcef=6tBeG#L6G2gX3R
z_2WmJmeOxXh-4a*8eF|*2dc0H_b(X>q}llBm#KeOjpMMbKpWNI1AnYTV7Fr+_W<{w
z4BXRO%)?q;`8~XsN-#I~9jSJiV%(x?c#QFSELPh#^w%YZ4K3rXujij_;$|iS9W(z;
zW8d=R93#&3<1frUbnXk5;g+u-3R&LL#@>(0_?aa<#jnVz7NYyE7?9OB|NS8X&&XSR
z>RauM7rxK%dS0cC2~2m~n>Bolo&2Ryz!2ex8|}5O3)g+Y=DDofUoe+odD(~NkmE5H
zAgQyzZSR4zo2iDR3Ho8Pw%l}oX20WMy!Fkc0-1PH=#8T*onNpvYDsE173Mz>XrK7{mjdF34LmeA>x_xlev^V9ZS`OG5z
zN5lC<2qQIkVtTunyPp~aIX%zhy-bNFm;w%d2AlzgHE9;dtpyNhiYWBVT)>qwb(H2P
ziO}SX5P{wHGbnuf_snNMF3{^p-tsI6LoeHxOon++qCxKG<5ssuu+Q^8Xb~GZE%=y`H~SyJSbL^?$D!A0L2gF8IPvk2mOPi
z9*0KeC?fi+*r!Zi;I_kYJkKQ_oFUyD3O{2E*2+UJ`_Wh;zcU3@qZb{)VXo`QZ2=#o
z$M;Haup$;w8=o)B=U>IHy84WWUorwQyeEUGL@kjfTOiqRj4FEhJRNKi=NKs7zLP&;
z{1H%+{>aKHe1)DBlKQO1*a5l3rKGRB(ct%7@4;)wVh}5wPM$-a3dZ`Ugl?bvfh6|o
z&Tz100C$$r`i<~-L`IiyC~&$OnNVJ%=Wfpd2T7U9jk5*lY^~RfXj3_OOcN8qaitUi
z6T?ye#0A9rz&~NFEf_I{X`9^UPXJU7J;DJpNvJ*X=B{5%1fo6h(}lrc3aA*>d{ZM{
z=cu!{%J)lYsBHd>)2)|Z!J_IV0eA@qGx?1gIUT)V?~AIVnauzQwflURmi-qhR?*?_
zJ2MT)6lLDnE^nfx24m{`Z@htxm!!F#^9bTT0K14M^v4n1U;qkeU?i<7{t>wahY(*?+<
z*4(uoa+83YvOGB=EKX?oN}y77mxLygGu;Mq%7n6|W4xqiAHY6(*=vju3IxslD&`sQ
zQxM?J+-=t2Edfihk8_Uwolbol0#q&&h5V?Wo^
z92QVQl8YAuF8bRL-Y@K3l*4Yp)j)2k@Lg^~%Qu4_F*OFlXJ@aa=o1`5*9idbDZJ>H*IdTdw(9$iTWF(Ibap
zK?G-)$Qw6I0tvqiOeHg~Tfy+%4gG%lAcEzhl63+_8X>c%GoF7tg#gtjUP;9MAUv@a
zK7J%8hmeM3_*?dh2(g^p1QSO(+pMZ`
z!q*Nu-)24gq`c>%*oR}5*|LeH3D_L2;Ksv@l}x)goDpPJ`SDK
zaD1Ob)JXIP;VUeWRbc&2SRq)Q!6*F)7Q$QY>i&+f@t>Cc`BPrdEqlhXeute+ewst5Kdx#+RhvY3>D+&d>D*hW
z8~aDBG9Vf9wdZ6#(@%xvZZXssAM(JGt+deFmbVBgq2HzQBh(37F6uWiV`_q;vy^0e
z21fK@Y1GSUQ$qLPC6Bdv4QLZEcv^c^p1`8ke_V}9op68pKWWDs7@@YX@vh_%Pxv2~
zwEX!oKd9N^dvKJ;2X@Qbocwjc1nvb7{+&2-gFt@1BX)r+9=0EmT;4vH2yI1LNQ61#
zVQNU+nIlx8M9%oN6uA>V@a8;mblNf%e&uwL=gdrld)iNIrTWvLUSeh8&;Ge!vyQH;dcC}NaFcM+_0&}wVjgGM;*!Lk
z%1H=)7e>9(e*xNdmsK3#E9lMLT*bXe9~dIma5ChWK4ev|uc5^?2xc2$Ijk5z1Q%Er
z@|VxUt17IRC(3@
zv{EUo@Peir__Thyv+}7P2#d7g*}9~JLmswugHO$f=9~2WKF?mzFKX<=aBcz#yL<~1
zx>bM<2RAL#E^)(a?(<)?92p3A-efpSRQ7?my9);st6M-I<>rGhzyDARH>IMu#!YCM
zanx%=Ux3g@cEUF01PfvPeWF6-2nQ@tIax=+wS$hW&{VNT{X~TqKvE)I0^p>T$x$Uu
z054N0Se1C=(Q=V?OoyolC>e$^kF*-W%11Gw?1M^VyZVjhMuf$i@g@g1+6?~1Y7f?gu^-d
zSN!+K(Gw4mx?zPrFv0h|tNTeU>J;%_<+3e9ZVNV@j-SX0vg??Elqm(_asR1t_NNQL
z$l#*iGGz+5Zfqs7jLo2L(oQ^mB0*r{sanV(i!V5ytLPhX!Vhh+RPaeB
zsfPgGf`}M{8hwx#^S<}YQAtE^GCB5ykq)^roNr@%#0CUo^Odf|u_Gta;~e&_5Gn3{
zK5N>28yuC77Rwwui{4hmeFZ7UQT2I;Yz7@3aQN-lZ|gP_RLjqvY(k}ryl^$$Kh4r0
z?m%&KJ>)F9)M1Lh+!R31(jwNGM1#P#X=+iIx;8L7vNz|0Yaz1_A_DWWsvw%MTr@Sw
zi;AZ?ER}`gz=HD6SvipaFw0&p{3p}_VR?~He4p!phq@uTVzCcEXATcVGY=Qi>-A+8
zF<8QPd8w%jsAjQqkM5evwhrUm+uuH^{|zV3@kc(@dcX&2XJS*%x=i8M0~5*N@Fe~`
z>E3R+!X(C#lfZITwF94e!$?wofgkB;FOl|Qr@>=T|FCqu5^Kw<3dqly#BJg>Us6Aw
z!BSLGI{qq2BagRH6(T(X=szk-d1e<%;O3xY_PKl=U-=cpJ}0<`+cFs?1UIXL1Sk2H
zlt~4&XB4*;+q;id5R4x_5n$dW!u;*-f5tmk
zm^b{xKVg%086+7^q~Ni1^USx1T&&TOS2M-45a&$ofBep}2xHiEt1XfA!gt=m?cJ_!
zd~Ujx4y&lfUVY7CQQT_9sh7+mdoC_s}+X3yW&e|Wm0TDwHy3GhvGqSCQc1ZlQ@T^fjZ2qfFn6dS!*fd}XF
z!RmjT_){ina&LxZe6>{Uzbfm?fa+=Iy)`jCKo_}qw(r4J@G{K{8fXh4tmSMDOMn=#
z;<_(E9NR_?N)xrb)^~wX<*D;pQwkt^uCuFW3qxy@DU!ON!?ChMfOU|vX&*vTCaXkIBW^7x05b$Pd-KZtRWITcU{4&
zn>MrpM?#Uq^)))@j%c*<+=`r}!x1?0HN}c3nW0jRM^0@YEy4J~d9&%?L8xurg6eo|
z6%Y|@Mw2()fcoo6!{EEdz+}Wu{D_<df>;NTpHhw9w-P*Qbqfq$a}r-JM;3v^$yYTy7X!k{%E){sJIs}3#2?^eRl+^
zewvKFlGO;}Z2H2x2}Q^wGdG;)eF6ZCxBKY=W`L}v>TQjq695g4|N0#;3QqlxqVo>t
z^83TMEvsZ?Wu;_PNGRhRdt{Rm60*tOnc0y|QrQ{VGb88NG$@2J`Vz`Y6p@kN=lADx
zU7zbZ&tISCectzdzix5LWqr#TG%Z^Q0=_PSO4snpV7U$S$;#q%q81zGcd}afHZu<>
z{yFeUny;dpLEcrJjeiNc$DgEU35@s*a!b^9IFCaaj>KEjtsqez$&;>64}hKEl;vz2
zHP*afKiCv;5-NPE;!k=a1nch{dTd?Y1nD`AQu=?1uveb-*R4P@Xl0igHtfR)qjFz7
z)#TB{wXtWGD;gMKF;nv*g9bak1v-eBtOQ|@>k)dPFclp5mH5P;5m9{p)bf6(%Q84>
z=j~;DoCtdTTv%~K|A3U0ZpY;jeS9%JT8QzzCA5=aC3;~@$gtD!_&llvaH+fQ!Wwq;ukU>Y$F5j)pGpd}Re2Sj09x=l3n(tvHm#r^6Qv^^Z;nBV=a+YC4)P5
z+qPqSEuqBUp5bT8ckx_`Z?q2C4LDEi>iDiO1HRNh{<<{#CE25b>kb)U(ED;
zeS?2+0^&Gvj#d>Di6B`_drr|G1}OF-H>NUm^toGC(IY1@%>bSV9$k-Uw1CH
z5ID9Q4ij{5q0!g4RPj@G&n9@gTZ;`7jhogz}85g8w67Yob1uoBCigCafUtbH#Xg{^Vnr|Yo1$p
z@6nmL06{`--Eworw5JXhyXf~l`}zbr?=-q4Ju$^!D`bM4-xjfg7WyrjF9v@Q$|?wQ0Yl~S0MhDV@Rs`
z`6fQSC?KS?cLTC8&X+uHy@E}sB9FFAt7178e8W23)BBFy*S!v0Ii5@va{3VetNru~)gRLnSTOpFj&ZVZX4Gy0HQpZ?)EtB8lL?y`Zw(J@FsRqL
z)XERL;&{Xk#ECIiks$l#iEoINqb63fga~$~wJa^kUBx+<*Ek+oQ9`v1!@{R*?C|ER
zcr#6R2CNlS^AhdXgYxeTv*nG~u)-GC%;hP7Z%(?4_{s{yG?qP!-ot02h5+>#9w
z{kcw&NIiw7!c(;ONy@=|N??GcLN*W*xv*H2n2S0_)B9{Lq=TEkSIl{0zo6ytiz?=o
zpHb*pvw}BMO@Js>u#%}Q6p<*Ao;g!pjkZ+tC!F_5p=j=d4Uz9ILrIew<4}2?^Kvk3L=izlAl4|K@B0Nm|CsrmOvkFH(2XL*oTvBc*(ppzs=)
zG||#^taIS%>}x`s&lUihZ?dryng^S|Y`vH+&Z3w9^b91Xdcl4Q$I4^Tr+{M2R+l&_
z5a6-`t~aC($bL#u(zERb_^j%md86eDNRG8j_*z|qSf$z-{PshEoB18Sp=lRj|I|4}
zjKm#v9*HXyXm9|~dr;+~QwdTNcxzv>QHUByhQq~!?*m3DzMX@#>u5IqHysI^ruGZ0`bp8WLj$Go4FkgavOreUcc~04~j6B9^FC`~hpmPIT|XZXeaj)P9sYM+fG
z{c;g;94Frx)1teEZ>$S6+(BH7&EeFrDv;x(BF~PNLQ!L`ev!gLAYsnuegq+pa(X~P
z?&yIRQm(U8)0EZ&A3~yfS^7_d%(s_nl2JF~#+s3D^UG4A{2j?eP
zUR@emk{hMTd24omr%k1CwIM)`1XKz2kCn-tL{B4P3$-QBg9WWW|LITe$-S+XJ>w@k
z)5c4e^LILUQEquyGIC0g75RNt@mZWT1nW^h#AU+hK#p&Xh?DYw+#N4t4TaBpZQ(DJ
z&vK@50)CnS>g^VDP`b7{O|!-Vgm)X`k9BsgSXh(`BG&WZSSSb`gNm{!PHuwT=oT9n`u2_>A{BF=fy5%lk3!D}
z*k6QB*a6-#YRbrmcIau(*lNIOPw;rgV<3Po5m|pZ$s4B-2llwAjjoaSg1vv4eHKab
z;55mz^XtX-XqfX6n_RX%SS(pQ=j1&M;?#z3-`|Y~KHoi=k6Ohc9P(qE=|U=C%<^H=
zei)A~>v>Qf@a6$)Udz_%2dTi|!Z@#8a2m1{{xd@Q?m6(wB7H;_(vMi2A4kWoyaSvr
z)7C#Ep8-Sr%-H`99YBU<;=lm6JIY-CNxd`o3bjea?~>_kgY4hd6r{SPVBY^%ljUpz
zxYRwIDUkC5-M!#>DoU&$BYm?TF|#s9zaQFFOPgaKMULp)nt@&sj$FDk$f^mYRts@
zi{ks2_dvLl)($Ry0m{*u5B7sLk<)vDV>aE#pqHE)`{TVc&|M;E$IEOROtK%0YlwYA
zc9H$(q&rfO<;%hFqDBt5y)wfqOT~+S&2Gm3Eg*%O>H&``?J4k?nigSpGkU1$^qG++
z@&wjlOs=Zg)xjqeUuGUsZJ?h#N&XF$2T1N&NpO55FDz2`YyTy0h+91}1kR@`z-ND=
zE=Y^<;Zt;FFk?X+Iv;0q5ztk_N^GB|1>fIA-VX
zkyxSd6NNbBN{)-|xTz1v-*8u-O1qCoZl|^yY4O0E6#m5wsEBX6KPphNGlEAnJN*tF
zJ+R9VS%V%$Ag;fDv$~8<9mm@D^{Y*&L5eT$B2U!={QL3kM3xRWObEzV*WFDf^m@DO
zyM9u|sgKW$v5k2{MWo+#AE5aCiN&V}(`oS5K
zoUP9$rLu-C^R}jK4tlVpes*hHv<=^U?DV44%MB-e`G$w99C1ly(a7|@C}^ZAv@-du
z8T0uJR%;}-14ty=ERa=@Z5w5GqDEr2@VH>Nm
zO9HO#SS0^wp2KD*K22Bn&xRbt7mhY#wF2@4lHPWFW`0
z6}G8(uRO}zhS#4>j{b>k$LF~RN-8|NaQi2pXicIX?7j4-a_>kxj@_oH{N0W4wA+>V_FrcPtYGJD}r;
zhi;iLg3%fs6%;22;kST(%7Ui1koN;u=?Ggd9MH=bu6f)ESM6mT)YsbKhmWs^CKLuB
z5%u)%;@!6}Fnw(A;k`a+@K1$xkgprQx1J1DlkTV`kB@TW=u!Z9&=L
zp+0!&^~v@1V1&BIgMLXJb8VyhrI`A8bKFX#(0^(1jf51z3Ew&z}
z$k5a4uol2q-`|_(Z6`r(@m&iE{|@+*&+!2fQ#-tNYKE$Dy%~n+tO_a65xAZ+f6oMH
z=0j4XU?l&t9rj93eE4shz;UAaVK7VF2Je~F3daZ3!n`>x(e&OTs63X)8!OYi%(^?&_qUu8YMpnu94B!d4gubP
z&|)oEl0I|R=fnj(z&qZJEa-5rOtVPNQ$Fb9o`Fde&f~yy+thP%mto`2*@U(s}*Uz2LggC0Ou1@-K{5v%G@z?GZpgKa!2J1$_uM*eTAw`0&IN8c>UHcla
z%2-l7j$wU>&I52L(quClgHOmLK{jFNAi
z2%o@X$(F5O(5&J_Bge%BaP>%{3Q5sNG+v*wyA^#FURMiA^p^Ms-f@$$dYJx2PHXgx
z*LS}Wmgcjo(_N$Bj&3o%n%_Dg4d~SIOCf?2Ly>du$9&RK4haanL-I_81Vp
zrc<(NGz8Q}+Z#K=>cPCG{k{xYHrk~=;pg%;3Y;!(YWlS92j28O7``DqjP&A1gQiuh
zLER3u&ufx2U^*I`Lo4zKt?ei=_u0pTkx|>=rHUzZp5)brvUDE`J$=;M{9zUFa7{d~
z;g^I|_fCMGn~zca=jrVHDLX!%FH8oP!7`(vGw<^oxTm*lXT<&}R
zPa1&Ff$`Itj{K+d5Gh^L5wPoPe{H4s+jlJ6rWc`L+8F-_sWSxA_GUq-=B;?
zVaJKY+P_oQKO^W9zeg9xWeGk9
zCRm$maRSRDl?*EaEGVAb^gFfkaUg81W2NqO228cZ3>>?w3l`0FY*y+8fwb!!+p7X`
zWY57wf1{dk&*AEcl{C~sktBL*fgi=u0DezGC_DfzZj#jX$7E2A9N$!arZSjyJCw1z
zZiN`-8gI$JG6HAmSj^0A>;T87IT6_edt`hi2VYb20>4j;0t?S1L`oW>k}dWCP}mtR
z
zh=n!p2ZAmM`}O6$C+K<(ZxE|u1-KtjMkf^E1-hSh>8v{|qMibY6htS9TAzu%u%it^
zas|e_qUnQRKUsgZQHO92ooyJ|z6Vg#nNBc{jzwKz!Tg0wrNGTjs@OYX41~Kehn?yz
z2mcnN1U*Y1f;$IeBu9U}K(^BwlGo1k0UQ7N@$YJfAS^29n2bz0_%!|WFYysylrMBi
zB}&W@v@GpymZ-i$6wfEb9=_WF>lJ6lu03r6OYaw$T_;k3(Yq@}k=ZTCkDkHw;_DeO
zXp$K5fMy>AYW}WLlp6-Tj&h$ysNbT+h^w{iPY1w?l8K~n8zHQ5eATpxiwGx02dFw7
zOamTuNj}}Wujs$!Mjw68HE>n3xp(p!5q5g~%<5V!FF|i!t(Zw8f|ur)2678X@sjJt
z_OoIdxDz8zleca)~m%e1>Yf=0Ka_zN$;*12K)zMsC^)73y<}y!pWhe;JP(7bmeQ*j|
zjaHqTJ`Es$wph2}DRr!3^>kTuUm2^hE2nFlB|{zJEPv533H*i5>pF9^q?+YcW@k1Uds
z0ron)IqY)Mnm7phKWaZdt`-F=+%2qY85;1fi~L!$o9*}`>5A-f|0|p#%vO?Zqz<(_
zPaS(g8Vq@58-D+J+=`uD9_A1`wc|UTZ-dQTnlYD5O-=6c`%vraQL(*~k8yHqosnZ{
z2mVj)AgO4l75|#EUAj-0Q^AKV1zRa!`1$(Z-mK9k9Qr)&-oo{EtY!7Fd*rYM@88xu
zLq^?+8M*wonFGu4!IgVy?f*LPdhT;tC*x)uyS`W1nbd-n?B;qFN!oB-&Xqmx>+MkH
zDw$cn5{3iW?_FPvzk)Qxt+j6oYoVp_o*5l~G5qh;+$O;`3F(Kq*ce1G-1s9|K=gv}
ztYVYaM9~0cPb+HEE0jZ(xG6;{0R*R;#!h7AA^5cZN{R%5XROGWCoQMl0$s=Kl_-@6
zoW=yXAG$;s(#ndAX+3R&pQ7K((i3=^6FN$AdgBPVg+MzfMKS^m$7gW50KHz?XVEx9ppoahC1B=FLfKob!?Y);W(B_^EQ5_H}tZ
zY<$s|$XlBWEh|E9QyOW)>EnV(=8X^T@;^JV+5HM;Q4rrM%X9W`*-n1&Nv%17KT>dUPsL972%I*h6nS0&8U*{J+~X4rXv^)^M$T`G}iFv-y(*^M2Na)7Q*)UZq>M2wUY|`
zDy@6$=o5K(%iapIegT-VFC^VHj}F5i7SZ}nf6OAG^Ea5*3cGz$GctN^1mm}e65qqq
zgn7?p=wq51{&jh4;lFxU{I9#2ZBo$(XZLeAQuAp-ca8WHd=Kep9T8KjhwjRIEvYn-#*t3Zgh+*n4%?SeoM)j-LK}_WcS9Q}WX~
z3^c$Ry(#CYZRGLFLPhmKivT1(`ETc(`wsGKQFK{vV8c0NYc)L!v*7P2FT;^14^ftG
zj)8Dk8QK#ZXWqy!0#G@}e=9ErY5aE2?2B0d-Zu~ouJY?~hNVW!~
zx?8y2p{+ty&zKUMPmsdshosWfWS@cKt9ErY+aYxR@E=2!TL&25`>(tDb{@E-Eh~6l
z<3FUrT^a1O^c!rEyOk_%HK5K6>U-SP?P#Y^=}$)XTL4Il%p^8W;;b!l9i_2Ju%YB}
z?rOmTP?U|?Ah;h;=nsFA|IT#*V}-zyb>&R7A(hg(RTzVIPgc35%eew2OOAl8VndV<
z@Tm2X+o+0B&1lNL9#!{S2pDd}qj^8tS&M;C(DTduz3u~dG{huid86$XYS8cCd^8^o
z_#(UWqN58@NlrsVKOwhcOLy(X>{CnNc04hFXw3%|(oQg@rhEV*b(!Xs5vjm?k@wxv
z%NfW><+jt4!#LpE8K_G!tq;y!ZR8ZG`~XUy2Q$q}c7RV0*l_KWe3Z9)SbQrm3{+e-
zSpDsg2wEM_NCf=72#U4LBYdC?vZJcg7N1rGJcpx(b?tFOmb9n`@=AlCH^xZ*5L6W$>VnQ4dX{E$#!_wN|D_Wm2GDLwt9kKm{$0esOck
zQ2-l3t1tAB0eGWotHYyw6Zzbn2)*qq3tZj{+|L;`M`R=h8i`gmK$5Xs?%_i(q~i@T
zcN7wl$q(yX!Gy~|K$TBS`QSDhI^vr~%IySNejVFOB#%Z^Byhmp#SXa$@-%#Ry9>6`
ziaL7~Z2_CVJq^*@0OWb1Y0c|;67ad5;Aycx3VfH%tgpj3;9-BhZ^kwP@mBh*JEz0|
zuR_*OT=wB$8WHbYr3y!`2AThKiV^`yW0|NKLkg;n{O5Y_cRpZSiFos&h_EQASF|Lj
zXaU-A{ECERIPko5RUS~@2f}_y>=BPc(KXZflP-cSXwdk)7-;2{d8TAe03LIDGI
zXVnWcX)x)K?uVbAikP{eh^;Y?4d?odw1+ydLgzal-iS6a;MZG7(zQ?tn{v-yx3YJI
z^5@@~mXh<}*xvY27O{&krSJ6f1a(E+pcSWfgy%M-{pB{nl&S}t%c-VPLMUKyBNZC$
zJ%^XGpZ|8Slz@!8)TPPc5bFFD%JZTO#pahoIm`dBL5H3Qub_6(vZ{aN-B--x}Dxy@dV
zLksKeoJ2e!nQi1fZyR@LW!o~QLRE-sk3r@?V>Q@uJz7@AycSb_juB%BslXGvf*oD2
z+Oeg1g7dMt7TkQHT03H>1uMryjdrxY!aB4Q*ZJrX4sIE9CmL+S|2YecPkd{}9-b3_
zue7#cA^HA~+Qx)ADYf$Tf-uJP_Zu&)5WU1dh*}%2-fF~2>O4LX%uP6WI#H_TaU(X=
zFZY>bF2^qosn5|<*5VfFr@e1p*J4*j;%?!_2HbK_Pc*^!B^J}=AmRYk`0!g!Ur}Tg
z4!G9#O`)+C7c3Tb55(8wx0YRN=WjP)QnNpf6BP)$eB6}nU2B7z4|;{f7%>E&Jg-fE
z!0Cj^s-Ws=o?
zzJk#q3cQKjZBXP3$I2b*S8z?ExmVch5-faUR?K^;Hejp
zJ5M_rp`ZWoU|dE$Oy}MoSP!m)1^%gXHA9UAU+@0P=HCWLn}5Y3F0~Fe4fDTebgLn7
ztUFl0sZ_&_{e%^^KZJ946n=(Rzl5)jV^#6iVtD3zpiu|GpBf=HG^MqA7xSLY);A^0
zYi{3&e_p%L3bSR@Eb>#U2t6Rv-Rm}~uv9+o2#>xs3?Q;LOd4;6{j6)tIVrE;6_OI>
z%_|L1Z|k8y-}yqA&6M4(?U@2)zglSD?XH8LhXZ`-9UGw(-G^!Sqz1^-(|2=+sSFYw
zQt0Q4=R#A77rrV9)sWlXWZF==21+*3op@|l2e15^Fm-4wgTksYZwu@5Ap5QK2Y$hB
z_*uF)hwz36-ZT23i}f@i3|%;~ORf%M%QoJT9oNL%grH=rxGjw8h)G~=amR&CRb=xT
zI@r%3Y;ke<29%eFF?&XTfz#9_S%DL&u+MmG^53K#)K2>)BBJYnSMPt?;(nnH2~`2X
zk{_=@xkt~=_!H($Wlmq_xXEBYA`tu~g}q}xMH-e~p65|HzSB>NYcbVo{9k$OXGHzXeRC7oU2`t0F=d9H7ccu&
z{$Rk7d25&47jM9v@UHB`a0=*L72X)GD2j)w3bxNqalu^g{;VNaI!JWpU&my(AvB@u
z*Q2aegF`Eu40msw$2Uy54c#6S6MO~JJVM5&pzv78lQZYeKrSXhd%Ic)7ccqDYLfjx
zKNIiDPg^}p}s{|RZ=pN#Tg
z?IY+^^GtyvtoW*kt4UzS9{P{}ClEAP0&}CTVv0&1P~)%X9DPUMBl>K;+p<-3aJd?n-ossF_%X&i?-$7ss7x=A5bRGMZo_Y%q!or*s$Q3S>#p8GUi
z_ySDimA=rLe*o^j)9xCu5jdltbfpovy_k0DTfy@=xV*|9QsP7V4ABXXMuxHepvp7157Ul1FTq@~A;z#5}^NPzs
zRn2Xx*!RguF75u&YpiLA`vwJDUhW$}WB=VMC*~cZ>>t_-oCyNk_n%q>Yv+Imu|9zV
zdQZ{222XP;@vk6El5WCqDHZw1P3;`(Ekjh&$ye_V6ahZ%!DkI)27t8u`Mnp;G9Y+m
zIZoWv02x-Mxi|sfML~SK$%~~K2M8~SD$Lek&Qz~#~q|X+d
zs(&6`bT$m7UNhGXb$0{5&WV2v6;T1{t_m{sde@L-x=V&co*5Whdcd+b<&A74E7q)@
zCxL-VYTK|ZMKCUF_GA0F4$@1z!274e2;{i@y?W%K2a4FVDk%Mw4D_b;ry5kQfH1nE
zwWEDkk$RketbTzJcx|$f`IpNZk=$t=pEpSYpX{Gf3SCJ8um0^_SSii`Ww%=1XvY?#
zfMJhm@5V+TeMw}Vn&uC%xXL}kEAJW%2120$D;L?M>1}OA;5z2zqP74FHjiOnjKO<56pa(I@uF)LF3O~
z*)ew`QNu^G%lobFU~tBLC+fx%@JdHZM=WXt$a9fvCT~3kHphQTJQ%YfoX7NuHX;IX
zEtIk+;X2?KPp#Xy{2IJ^mPvV|aR&rs#9i6t-$CErwR@lSIsnw8<4LDf2hmBYw>{jn
z(m41CiwH6GCNTP>_^+n!H!@VazTh1A7hIe^?sDV;m5^i1SM-%nrj>fjUd#
z)fdIfXy(5*6TOt*fQ$L?a}O&S4j{AbdWhKZzkusU$`bwqYJb~Vp7zb6bV}ksoX3BF
z)h<7}^vlHfM1E`WZ0j-n(wgC`v-mhz8Bm@{C!a$MJdYwqBeE3xfL8S!Ctqc2@X+Ly2yE3fUCr~~|ul{9YF(iU3pSbQkR
z_lGgJifOEJ3vjC-^~1dJDc=+}H(1al@-J7LN-pj>n=bzo`A?TReH0E)=Wlym}
z*jQ}kzw0oqpxkQH0$?gx(0h$64NJb7l)C&r6=xqgp;W99hrQn`+w{`XGbEj%3hE-O!y=vgCgnNIF`PD(^-#Xiy
zbmb5>r#W{pE5nOvu*Q9E!6g0`Bu|$DIs+y+Ej#VK(hoMQwmdqjnw0}>cxb#NPo+Ug
zu8HdEY-jl2W%A8~*lU=07?F{m1ekB|;)W*SJA7@YZM)3_UnE|LS!xbnf{4@3x&8WDP%*UZRl4IUR#v+<2t$LzXJT
z*aA6t*Y3z`4?SfVI5sM(dyx@)b(s%#$gtwme+s7md7Xf<2YGX|NhdLx(^$Bxce
zc_+K#unpK=a$i-><-sjyPOoz=9EB2P_e~E;6!4K>Ujl<7#h~kd^6Q(6tni=R(J^M`
zlep6_DnvD60Zg9~jCXLB!N)J^_ustA3JKR2qV5=G*nGd1G0=)0zuf!%^5L)Vi0V$6
zzcQ4;NJxo=zTh~-{}s9lh%>{*78jFq{YSB2N`*Ds@ERg#5cyV3b`oma6jn|6kV2El
zw#4<@8>r7VnUA6D1DL~H*&WpcJ?I4cZ@nZYLNB_dV}2YZeCO~s6VUuXif+d%%p__7
z=j6T6UhP@Ho7knfO|u1%LfLF}C^1y0@c(w7b_|J+(A*nb>_TDUBpq?_pTJeklzK*m
zITZM7P1}(5H&8xeKJ8D}Ue4hiUBIBIl1Nq3?vzppz|uYA_{nrO?xX_b40
zy^^^|^axuhY$fOy2}9a5guR82V#y~a&SCI|bBMtwrxKWr{P%&OC=Z#_JsSJvmIVC$
z*U5P|!-3>3KOiH4pd^w;NM`gZ;?8nc$zL@DF}ET*vUCYK3$+Wh??~doc)tUE$>v2+
zdhT{ZfPflu^{l3zR5Jv|B~tffq&yL2WYgojJFx^l!A7!I`z0VcNX}(Q$%ZbmFtQt;
zI*`+p`H=B;i4SPXo6ibbn*(wMiuj|?<-tV!g?VVA3C_NJo|HRoj^6F>7$+!&fv`&H
zn8r0*BsmA=sQbLuh7sxV?)>!LS9JRxIheWj9;CR&_kO?j8ND5ehmMB}
zfXlc)DarFMs)>!baovR;>u_5>%h$%Br&zCPyJ-UXGFLgfPA`F$#E`9F4`S?`f7EM%
zl>wV*8$YvE>Ox0rKF7GSwSn&%4DP&Y9q7=pyi%z16L|2Qs`t?RE2`MFNqoOMgL;iV
zoln2B0n9eusoWdcN1qI*c$;eI;H_Vor^q=jKqqyB7K`)Wk=Gl?0onW=V4QNC#xIHt
zyU#8hlr}NJ5zCIfd@(6lkma{@kL)i>72#z_En|iSTc!HnELd@oQB~k|=ZpAIw&ID@
z{vDWQc;m-6USe$C(X+ah#shPADm~UrU8{?k}kPvMbM}gG|ED)4F!qiBQYuDz66$=Dm&GRckSGh0ZlWG04#F66g
zePjf^T*ej1;_bB%$l#8(WRG9VdvXajT3xii?Jt948fm_52g^blcK$-eV0Dt6gE2leU8CD&jQv7%emalZK;5FQ
zGG-D@>`one`y#6i9wj>Ul%Bx>s_xCT9y*4=z~7_M2hZeTq{Ilzi8(pA@D=QDJ14?<
zkEfetV;Qh#QrxgAG6!-Tyry{<`FI4;KvW^O1wQ7b75!|OxMyXf-4_pcVa%h
zfTb+k4Qi5Z_zrVx_h&I<9IJ1q#Iu?YKdWvnD@Rqq5Q73S9y@|3j52^Maz7Z$oo91P
z>9fH
z`LUqm0E48hbr>>i;I^{X2;+@1{8>NSLyEAUk;>0{(N3O;rx_o9VFbz8C0Da6FFGER
zQvIU;;arc|%X+E?Z7Okb_|ub6wK9
zE>s%($hX-Tl#AhaCc|&~9c`fRQT5piZ;ap*e~pkMy~5D!vRA{GH~d(BYS?jEmLD6;
zDLv`E8IMQZ0yZ54B5_okj5WApi*1r^JHKei6Y{dINva%Ipj9OWOGw`{oH12Q^UF6C
zEB@oLZHc&xWk?)f`JTNFIg+iK!}OJ5q#lbHg;7497H@1dN=?U^gla#EL>uhWq{yW}
zX$INf{#$>XECtoN4?Za9T*4q_PRU-M0gh;t6OB6@gA6uRk}Y~v_*V7Z%tOcd3^KGGUF6v#PS@1K^ljr;EibM;>HnDEc;ry
znIU%^WiZJm#Q3ZMR{iKhUYdOL>?2=l+WjuX9c_NigR>X>?clXbdA5ue_0D8(!f$|D
zOF?U{{TcXpl=j4VnI4o%R%zR2(g|XISBqRYKMk;_9a(DhPc)mSd4XxC3@BSC{r9)7
z19Yw}5|>c*qnOEuEcdT{Mg4hcHy8>j@nSRg9lu+jfOYa(fC_=5yOcO%`&;lUI^{*I
zb2^(Aiyn)dJ(^F2kAAE>u}ktD9HG-?;JDrfo`?~(@#lR->b?&Cr%T_0Rb`us@uotc
z+Z%l9kTVOJhRB7VW{L$t-HJprfq{SmP&&tL_XEbH;;tj01hkem6{p8U-
zUil4i`JS?)@vBAk@u7QRt#x4LWUxmPR|zs%6-q0rjz>usJr(5*zM-EB!xA1=kI+et
zIr}T8YQe>NsebU4_+!p2t?TONr`IV@QYnIL2P;s7`2OfZpzY9Csq%Pm|RQ96|=i!ObG
z7kt^T0@hG=IQ!lj@wq%f$>@7wDRr1g}sen)#7{hNHwGaL~C%*;haH|?E}?tswl
zWrIk>r55zSxvmRzRpgsLpPE6;Kfh|+&1G}!={w^9Wy!G)#
zt3?)m4h$lRO!*SeLjf!_ctTsyeh{Vq
zDpHctodwSf-Hz`oucKs9V@v-+TA1@P*Ni;l3^W)EdHD0AI8?iniOhW!ae+yzskGNs
zh~jao&6Wx#zw=*?t*Rfcp&U6TeMAB}o28`Mji1D+>cs~SXT-63dvzVwaDql&^SsLZ
zfjH2UN$Z{^0}P5*+Bx6Tgpo#?9ayX%H+F)eR!sTEZ3f4`n|F&8#}1E!ZgRl@*ylT_ET
z3L+Y*yv)}wc=f;VGt{1;cxOjhtvdQS+zoGKr&{F#t3`s)^O^q0vE
z=4f9W%VZ@oCzb>mi%({nFE_#}5}6|e?Gjp>=FY1>xC`m%Co{}N?cp=ImvldABQV!|
z1Ye~+VUHf9n8Yg30^iICD}0pE$K(F3C!$Y!!+o~nKhHY*U~d?DA~GT#&LrPXV~%Ts
zMYsMP5dSEHG+!DIrHyJ}?i&fq^pKZu`&G2P9a|gJ-{cNC+lS%jR`KT-juLv!tEit`
zkEn;}c`fs*dlQr(c4t0%ir|;1wR>VO-46FX{+Nu+H^MC17aoEH9pIxuqA!;~GYq_Q
z=2y`>3vQkBx?Imz7zL!#Gu1h2zRw6v!{-?;^xz@
zW{HP-v7&!qsP^3+{KO?RX1AadYk3fPhFodK`mNcIdNmM!#P;S9F6hC3sZaVk`*-7|
z%qQo6wsznm&$gV;au`z*fN3)?+Aw2{RbuvHBVNWSEfGpJn4Wdd=D1S@t}!sqc&br?
zTh8+lHSy=;G#=}W@Cz?7$VO~lD)snrXmQl(&{{mcNnuY`Scc^~9-sG#EWpp+yDacm(|yO-2-7~wQtZ;OfL}>Q3YEesVQoXQK;^w$oU~Npn?msbx0Rke
z@^;q(=5B?I{_?^P(U+JIu@!wFNn4~o+
zn2;FeZCWmc1uyI$thF%V%xX&N^TS+lCZX=vyP|a@UP65=4^$v{K@%bw*Pod>Zllb?{ym(WO+J#~9D4w{Ynb=Ca4Fsl#_ez|HPz|7|hj#6h$ylB(wUc@pdda@Efe_Cf-cnP+8vwg^2>`rIE2
zzazsLsfdfMA3@o^*wW|e4sg2JGPC~CSF}9WYqFMCjru)xdJ`pUz!Fk-BatpdZHLcp
z>SZOOEwcIL&g3N&GY>1`ms`QtUZZnErz^qZVDg*{nrdWA+PUtu8v}ey)A)Y>R|OWs
zQokRdw_u@m=)7xsE+RFv;9%>DKq|q)M788>*K-nUJ=Bil{SRDJs49(fRHO;;K6>OEDBFx0-xW1JFbx3r>nn^ehlBz06JLW5
z3t~~3tn|+tq3p=n{M7ZrWMObE;Q8RopK_?;$Eb)zp)KI{b>2Ko@IqRaQ~l>v>Cq%}
z+O5dn!XPX@+lnYb0G
zwy17$MkEr|uh^$Ny^;xznLK;)k)s{0hH1x*Mpl7|TmSLot6u~0Hwu;$lfBU+y{}5)
z^%3C7RBGV`-&`aRat}&g=|ML~sLuv+X&}9(q914Y2t9$33Y?=UuIR4yK#}gX4B)e<
znC-n%g2)2hct!_wkb)NR657%LbhaiDGN;r~q_itC5OznU1L;jCPL_jjEfhyLp6G&|
zdwevE21Y1uJ99VE=QgNFEjGKeYJ(a?H@^%7xt0Ke|;97~MKeCx!iRuP~yUf|^t4k<-;0(LVuSPWdRAaN1t>ixqw
z*&4Qx$R637`aBPj2w8<>L`G@YGNMFQ!;Xd#vg%v*KF?!kC(6iPWu%NEiGH8oU!Q-@
z=eo{yu5+F9KKJ{+?-$K0P_ImvH6&Y%>_ziSyW10iE{{-Xi|P)TkbM$3^i=^ytL>f9
zqXby+TK&t5z54KoRoH3YZx^xapPkHzI!CPIV!&)}MGwdM`W=ZtcAQsxE!3Rl5Zw4F
zrZ*o<2f63U8Vj<`F@L&N@*Xb=%qP~bBY1}ys)QBHyy4=#w29-X9IJB2P@c8xVf8ejlFPb5+
ziS$~pJx`_l1?ug!dHG5hV7;ulz|LtcoL2KW6SN)20?iU1cVGHLWtVs7=a$Jn7Q@y&
z*Xl&FKD2Yz$h!fi-=){S4$i}jqdEb1gUxa8pYUM|{mU?J&?H9EFckBMPPTl@%!iR7
zJr_FZ<;dUfmR7%inql8Bi@Se0FTy8@4q%4X8J8qIea*ZOjVl7(HR*X3aW3^~4cj#X
zcrMWEo>q`Gp6q;Q&`;|FgZ^CoR#@Z+ugLGqEZ(by7i%d;>Mrv_8K#q8eyB;q!RNO%
zXcnAtWAtFCrlJ?DwHdn@*9ag8_Hw!A6GG*KXaxSyp4+`lp}4
z%X=n&@mC?(aI$9qH*$_??lB-eS*wTgF8ro%6?~v3b^VRn`FPyAKQ*UuAQw_4T=&03
z_IZ@4>&-qQ_XzZ}FjWLv=;2(R*P3R%VX%2Rgicl<9*b%+%!$xcK#zhOoOv}E_VIT3
zCFj&a`pTnv`i%|HT)=gd+>){&J1?yYG^>BG3
zo@e-dJv?M%_w{RS9n2bER~hd>PP!aij~dJZ~oWag1gIEi;Kx~+)sPbH;HRaxU-u3=y%~}eA@ng(p|b{Y_a54
zo<7!u?Z(&Tq!b8j|5#^SWej1jd+#?tn$=^MK1E@BYqFo{ScOog9N9OWz3}bPKsk1u
zVUs;iRASng11SS9tMPRwc1GssH8^p~&|P)328+%o4matM{XnXF`gE?A;U8|&Ik#a2
zR@-|?VlAr@hlJknrj)6|+>Wu)95F3;`DVk-S~{A|FSqps~I=rQJs7E^5*
zs>33*g0Fg;nlSrokM3Y{ZZGQMv#Z}M>#9!eLQeq-S*m?w|Y~H=41TTCKU9~XI$N6^D3-@*^aXrsaE%TmgtlFAm(Mk#TSB7%+Zhx!Z
z%0ofn$n}Tvr!h;if^XD5Q{2b(BjQ2zQP}8zjMuZ8531YHv;BR46pw%ZDp59m9{jsnJ)CgQkSr^chAPgRN_mCCcw6II$_ukC6fmc7b;kb~-mdKCV{FudFBC4?FbQk`
z@zlI9emXuZp>?|ZQ6?vhdn+~yPMjp?5$?V*c_R6aoiZGZrW`nwu6BmcBm)i^Op!#J_jWDuFRrvVmpR<+Uk;%EcET(7War;{H
zu{9w^3@R!ul+RPafC01f@?5Lv+s#i7$qO?`sc+z?i`P72mQ?*c=e8GW*X2`{>kgny
zaTe{2;6W6>&-#{U+Z?z;p9)ek;>NQIBbEIG3mxWr8|jOx{){Ae5T^!^RxxpP*$
z$YK_$zS*}H`?Ukq*Lkoz%
zf_t9R{fAUu@uIT*r6BCZs_jI@0!Z_s+CBHI2<*?wDLGhQir8XvU+tQbb4B{z^;OhC
zs5h1JqR3J&S$}a1us&6fgnh*5SMNEY$j@q-f5MZ1qQb`hd)*1Bep_ZX?M4fFt{t%%
zy#5B6-qd`Q^C$^qr&?|QdUp%0V9^ghrh|~rS$9*T$#V4S!CGZJWh&?sl#mSk;taS_
z-(6Pv=Zl6q54JR>x&z%NO*w<=O0exQ@_m&x6hxJ*I5f1~MzwLOLlWnGKxipd(%K3M
zumw!lyVb@ax#@})qw_JSt#s~>b=Vc)TpQHJF{g`C^+bhVKCwgsm|ZLC+#g~#zgcAT
z@&Um44@4L3mNBy)-2RP3Lh{RnRAlBrqc(xn9kQ#I9FZMiMB2Jvy
zw{&V|i3DyM?kyZ(1&8YjRoAaPBPj21T172&5$wxd@pfE8q`dnXmtgZ6L2;S=)t5Z0&A8sRwH7biR5w)ti-VQT30)~eU@n*au_XNkiO5}KF2qG@ecK$KH
zi%v*~z2K1LN7CW)d9l|;f#{-Foawv*8fSBj<$tda`s*4`oY`^(xBoq_aj2sKE^GqZ
z1T8OmTW)n3Z%dFlATt(JUb<*AcvJO8+C|is880E5ONW?_ymvqJlLssm`?*?miKAhj
z^PMVM`e3QEvFw|#Hc02@ZMR%~feboxCv?Fpa7;?(*V)c;bX?#CM}d7KxD%5XV^jSL
z+;q(>i~nea;NQ?7E+sv{wsOPMb=Va!2Tb=HU3!iFMEYCj^L3GZ+T6eGbhN?0wtqZ7
za@-M2uxrR1x&>zIR988bBaxfQL)XJGO~?_a(Txl`q6=NohqA=H0Bg9{KMT`4h=&{)
zTnl|n?pvFWJQiFE_WiIv+`)MkWXihEl%pGn>5&R1iH88@H{!H%!s$q7vvv9n?Q>*%
zu>W@nTm_NWHN32=exlIX7OMcuEpW8@uH13aMG&2RG3;T51}=Z%o8l4t7>Fl3{YP`D
z8B|7IxG>Z_ia5+34W(A?ff@V(Tb%7wu;SoVy7*URXr7IbwVxuu^dW}gecX+xLOqY$
zAh`iu|1@wfrHujpu+PhiSIh+Iv@J)BTWXO`<-To&fmU$-Wyx>dfIg(KC_wXmiw=XT
z#}|FL6Vd(vmxAs#4CtcO&+(b{ppj>4=L$-{0xf+9bG}=Y@bTK=l0Rc&_(`ZxVzeqF
zyshb0<3otRALYvJJmme|F{g(Il2qK`n%acu)!S0gdbiTPZ@mcPLrsrch4Uy$_=!Sq4=*-ngTSrdj0ZGcNLsv
z)DUKOh93JAt(0~9)5CXXt0Yc`G2-ABs@7e${kYKOxe@0lP3V=WeeenwEo?MnlJ>1<
z!Bj0JbTKwOP%!vo%kV!jY~#ik-s*J*?+=zznCtb%KkA>YdKgDTZ#ViK8vF$RHhg82CG5z|A(
zYLGBUIA5P;gkcZ7KIP{RyBnWd>m)fsEgjhlhsb%HD2rUr;xQ9QlB^xGC(hvyNVewK
zq9uMNe$Z1c&>vE~xN+)bQxoj!e6eD2sUAMlvPHMFDxkN+scd()8mLp%`Q(OC72Iqs
z4pEjUhQJ{$0xXfPtl%Px$Hdbbtnzc5^B01yB
zeg@UhTI6JLTwXP_Ij|?LoWB&V>wZjE*~*0~E!Gr&mY+idgO5UcZdb#QL4Lj`92HQ2
zd4y$?r3ik^o7hyF&xXHml-Bh~J%t|fXH6o`R>EwReD$o7ayWL+{8w&jF}!@~gd|IN
zJ~WHHHe7iz3&z^^vL|zA!HqYB$+dSc;Um{kOWEjLc;fWU%%^%;aP-tW9xIbH$TWo+
zQyDBVUu5Orr~N_j`TQ&MoW(e}--JC!|64rF>VHhf)Da66=d&n{DAHhdd{$A`t$XmO
z=GZmGwN%Wnlunb``x?(pAAW#5N^n*g%fj%K*|zhv7DcrTOwFvG^ss`FdJsEUfnXE?mVH1)scp
z=Fr7*8SVx%2v`FTd_6hAv+%#0*uPltU#><8zVbso)f|*zlVcIj8yL#5?x;+tr%VMt
zC82Mr%R<(56~3*E(Y?mMDe7Oxs}!kl|(m?U%xFi|=ar=6ICP2!`6H6mm2_bO8*jWo_`|e8%pH6quJ%Euy9X$+}(*XT-sz6^5JkkzAYfKlezR1GrrGSmKiO_
zB4K(5ukV)OwQI*su4}%+-vl*!(uxys{_oN~RVLn8TKm|a+29;(rp5C7l~FntyyM`m
zJ$eV9+t}g!;o=N)*c&B=FJFOD9be==v%BGhV0pI_p%)?d1LD#LWlz|4RhrTJi#C*g
ztQ!b8&9JPlw7p1-2b>Y@Ra>$4guL`--vuTlpzpSea^=VKn5wVLNUGHUf3;ews1Ill_PED2pdGI~7V&2bx^y6>m)XO0JgpIhv)tMfi>;-5EuI+{Ch^RNx_jl`odcUb71*G-^g1zG9T4^
z;&_}aH5B1gp8!A^vc`HAn3DbPSAK9>hh9f35!oOfe@LKu>`+OE}gv>C}a(}5gbzug2Rm4?QYbepF$#{mUL^VtK{bok;E3zTDfg8Tv7#2
zeH(kr98`@=s(Cc*MsmQ6vh=z#`wCF~+F5Gd(+VVYXhgb{>rplv*I+SC5r{D{qxM-(
z0cVD||1w3cfxk@SYS&J;0u90R*v)i|R&PE^Ull3_<-gAt9Ipulwu6ojwIBBa{#uc=
zrHFjg`7G>Kzj_>csQcQh=TjoE{sD;g^%x}2I=r%Ysu7*P5Fwv(KM4scrOO6%BzPp*8EuMS#+qd`ggmo9i9f2M4I*I^%oJ2Z41
zx)RZV-ua<1_N(Cc0`F&cJx*|(Gg;jFuM;}BH_nmV@{evl-FVn(>wyd`)vo!w*nwk%
z*WA8&8X?WTks}mWNoej<0>jB@6_9FT;Tn|W2wv_mm-xTzft2y6u<12dmQac
z>b5{D4j1o|Dw70*9}V;p7XP&FcAK%Fc%?P#$$WC3T;=y)jP*k3fPA=4jEKFwKj9cfEfIhGJ*#nU2dUXguKH!~
z5b>!pc}6X6oAmTe@pyaCHzMOwWg-t-D{=JYUiX{k4WyxW9?Aom^~Bv4<4C^ZQQ}~f
zUGLCaJ|N|65+`Qr1iU}bM&~$OLrH#bd0L7Cz|yBBwZwZVNVBUx{7!fRO7q=0$jdAZ
z@~Y^S>$^3PH~8yq*KZ6G10DZ;lyN{#DEpP$QVfdVDye<)M+!+keQRqUqX?)c#Phg*
zX`x-8ljaZa*aORIYQE5jlc0C-)P_C-FZfj-OL3!_yRh`F*4~D?*p8LW?7<~GtjbM@gB?a62Q@Nr+i&yFRXg}aAA)xGv4%ep&uDN
z2tRr-FR(fu!JGxt2f{3LaKNnp-DiOlK!*S1^}U}zA(vwhf(}b}f}K5EsjKfP;lCIq
zR)(>8B(QyGb2OwEW#x!#Pe*?QQO>SsEttkoq_f|wl}$g;KJ-`fs68!waEDd!UvwQ{
zP+yd=sBA}ahOx~a7e0aKX^KE6cp6=4uX^L^PKBkD_C`n@t^&h7jfJs%xnOHHlm0+V
z3+heF8vXiw2@INh$s4Qu0Z|;xBLY89;J;ct@3&SoVMnQh@B0ybymPvqV|w-?9(Z{&
zxwXX~dUvY&%1DbrYaZ?0o+DD2tFmI{wTLn_8@}Jew0IVK?X4Vro?wVmGZmjlPJ9D5
z|LW%}h)(*5qc>*@HqD3>p4UEPK}l7U)Ub0q=p^&x~j94
zEV#OmQqtu&Eq1>_ZITEz;jQXFP(9%Qq%v|;=GxDO`wvc;#RT)fmA+=V*m?<^!p8J%
zs#^;u3PO*aE1q!6)^OxtXcTN7KVw!@AC65<)_qO>k$^vZvXu7^_QfLa5Z9Yy#+Y*5
z{M=B>ML1b!MV}t)48NT2QyUiZ!B=d8o@h`9;&N}n`2Y6H;Sfwj4Bj-yJ8S$o>VEpL
zp1DtV+x8NyBTxMN6sE97SgzA*Kp8GC6~%kfo`mKd9RrN_jB(}x#p|Q8x;S;U?mTyh
z4!q*zi?(C{KCOPx2uyImniKMB6BqTNc9it=@je5r!SSIf|N1%n;5+n_vG&7Hx7YbU
zJ70s#m5C>ouE)Xgfow85HxXWGG)oxKj)z8mG=iHq1F@!`s$229Kur0YUUPBQ6-y(V
zL_H4@{MR!vbIafyYa?u8#
zuuBu*`mGIHn2orC-=2l+6&n_(m@Ysrq1InIKIdSq()Hp9U
zl(H&0#-<*UK*&N8Kyhst=5_j*O%Q(-(T
z{zX~A#Rg}+|2FVr`z#K+bbeTxT?q>|DJUr1(}eU%Bh+@ndeAOD&(Y?TDIAk`dg+{X
z3SO!mc6Z_yhK_n(^l_33IL}on>(P0D4<1!9vfir%FLcd28Q3bo&dZyssSRJiwc;2mbW2j7^T#alRKIS71>p?)Jye@F$vqrBO>~*X?l9++!4P`!yg>Jjr7Pw9LNHtJV2OY&nOk$GfRd|66tX+IUr3S=!cn}hbeuT^v1WfG
z$2@sIs_e-U+joK<3e@Gk6|$1S=^U2wJMQ~2m#;jv^;K4Qv&Q-%Rr*;>bBmh$QN9v9
z?)D#_@pozXPW42@NlJ0N&G(naU*`|v`F6VTfwVk~v00u`X_beHO)n^r;2PrQn)||3
zv5TCBLzmwL?t$J^clJI~5QO*Rm%{As(c^Oz>_Jvte^L4C^bdB*zko!Bz6I-I5zzSP
zF-Wyn7!$4~!=j2mkn!TF;iQD$V0&@@O3d{+ls}|0aL`so|)^bU?Bdx)YxPfSkU~Vl|0*kY#nLb0*@4dwy(7O^Yod(UFT|O
z-AE6ZIP|5`x+e$qFj7S%ze+@|N<;-UDH$A{u_HyGG&w$KlDQM?=YoB@rEIm^}y96c*-}rksJ)26cxZjTh?WwC$oW
zPX&*mm$U_KG8!v?t|P)|21e-z&b%e-)x7par=)p35%b57?1h&fqE~lf(q0AnB2lrT
zU1f4#vHVP7frI@?WWpv`(|J@I0o~s|(>yk)rsc}y>WBlFE#j;{{Xh&+2Uv49MTjH2
zvfh2f-ecfsLt;eU9e0o}@J&+J?F>4z#U;huE{t@v=qOlJeQ7_!nAm>Il!eOBL~+Z##iZp#tb0*E(ed(j;$
z1HZ$UqNpY0kZrrN~wpXYX7^}i+}Hs#`q*#RNoNL4jm7RCmuIJ#OE
zJAVBO6HB^fGa4+Y
z!b0?3>vi5cLrF^N{_sFMM3{7B_lo!kwIdNj5v=z&#+P&@t!a~$b-j*0o?$UOl7TpB
z52EPBf7YemJnOKlX+k<_JN@32xJ4*6jlVuB%1$~!>1*(8uOP|d^`#ft^-@I5Yx|m)
zfsmA-8d|zfFp5NdQeIt7=Pv1aTbI-B$NvZ~+P*FgQ%u^JzahMoxJc+;;Am;{+a`s~
z{;PSI{*vUiC{uRF`YkaNVuv0zRFgWdi|Wa0eIjJTLl)kS%n=dYdU?GJUIeB16ican
z80o?1Z|NPM)T@VYA-n0xJPEN?D=t`_qrTx`$-yg!xcH+M~up}1w(&9yy9(|&J#
z*@gdSs?0;2j9Ag99V_*@-_qbTXRf1hE+bH#4$qSinIg&58-#j=OcM@HrdJ#US4geR
z8?$4#S&>XPozMH0Ltx{K_z4m6?LF%;pT;8BNE-Ne&+ZWVKzR6QUb@{v0fwx~)W*!X
zkW{$Re6{QZ;j$&OJ$Ye_G&nM6b7J2dQD%EqwcmS=BzuqL19!XtXcCG|%J8uQVI!8N
z-D6i#UtW^2Wc)3Vkoe=x*_w3peu}F_c=7^Dj-E&js}(>e*h2i5By>?k|1`Hjh!tq_
z;Jh@vbRM6`RkMyeh3lzg0bV@xe2M9N_P;puQhguhMcJB0L024Qfg{AMG
zz(Tv!3%?*wAd}9`ZSgn){g*QIZZRYoba)TR4rCRe#h0&5zqU0a9>?i_0>OqrX;3lm
zT8bwKw49-foxX?8AGA8;2y%d><#CE7y()0*>t|E_fZOQSt$>*|+H0VIR$_da_ZZSm
zYHhxzmWiGYcRdkj_5*1Lqsxf_W?1V4>TTDU@r?Zi6DvV3Nbh}pQ+Zkt$DMN#y75F4
zk85f3>a7#Ny7WaZhu$5KGH|ruzBAx1X=>r+%C@NU}?)al7V7a&??y|oX5gpkF+H2;4mHIc{qo)@^
zHO1?Nc&v`~lw(?I&PZUP!vAXS2AM$5g-fmJ2HM!Y+~?zyElWHW9nZz8y@_76Y^=*n
zi(>zgs2l4YLhxt-gYuZxQE0z7HJq=)hwD2ApB0L3gO~(<(=Dz&aDPr)SyL9jd_^zwLZR&I@
z{KMZZQ9SMxcG<%&wd%4(&iCG7YIt%4QtZ3u88^p^Go6L_TQ`;PkHNyF@&bT+c7+Z0
zv8!V{tvWl0qe|E)Sa@chK>#M7Qf#}sDg+7av#(o&3pUMhd>v=p1kP=0U*9Bkf_lN%
z!|mbh_}HzJN~{S?Sb0+7&rskb7-{1=BPK@=Jwsl3Td*_0ZQbJ;A4aA?&gN>5vjP>K
zmCXx{+h)ZNDY}oHxX+APs=nQeOy7&M3)A$jwjF`i9t*{XTZQ3WF3HJ~stu6lcx}#Z
zmaONV9{A_YA%qX9uzxr%tc2rF-uy1SdlYLl)$%V5P~+gB{;hqt4??}QGi*HOf^hfn
zU&S5R4$c~8TwHtFgA9EA^tST~fCi*JD1G-kC>#4ruWs@c{NTQ#XeP`6lQ^<}7>rQC
zSntc4=D$CI%Ect*$*yhk^_ELqdq;t97pmsa+}(@4K)3Aeaw^Ov_|^zi4WqwlXH=C{
zSHKaqiq9b}n?V2Dv_SV?e!RHPc!KrJ5zND@wRrHc0$fpMr`n7IkYY6{yyUV9ydC4!
zC7vb-S8wI9McfgC6R$XLDyc}qi5}}{lSO9eo#AOu3{YV1yXSk^5~hHZl)JK^_9QqE
z9HJs3OZjJ{Uv01xqucPb;4kZ$9WMMB~Xm^0qwI>b4R=-Kb5
zeMnb>@j13z$LZ;?_s`z$pT`t&z}j;sJqijO!TKqZ_0l02`ufK1yxS%ka*rr;_{obM
zdnG?f3h%*cUurbz@6o~@{q|nR8Njs3if$U(O8%6X|BP%M1!^8>#rP}|z2z4Lbk4ewXjYdbej(0z3Xm{g&
zJ}`f^}N6;x1!7MxH<40$$1|!>I0@rv8%!#_~31k>)Q(~%upNj
zyx0<3K-5xUVN=7`!5PIjGks2TfUUh$Me6_+q+52_(aEd}an2SNJ9i4A_XzPDk~FZhIeViQrl!Q3%&;
z+X_%?_09D7@hq@^_?Gu${zx=6Icb}HU;(Hfo8?x{4gwMbSq|GzYLR)vgfx$BF#!Gy
z{&`mSfXH@b$hYNJK%KYEfM6^_BP%(h`47oHm3VeUc_;=%>Le{ep9ti8;8>PrTM=6I
zDT-f~@B+ntdo78DJ3wI4FzxNd05l$RzT%$pL6BenLQPUS4+QxrEW;aqK!3kQzc2_w
zl~l3~I&p3wEOb);p8k1YeR@uAgE;|NcSIFTKtCqa>xPIb&BsA
zd)<(g#}B;k!DYbCd*H>Ptz>kiV^P10KrdZ77UVlu+?AriVG
zLVIKRI)Zie0k&p#Al7ttgXX#(`c;dx(^Md$rp}(tP&@+~eU7_Co7w}19hM^pB2~~9
zcHH1Uz=BQ+zG3y0R{?u!+(?oW=RiWJ4b$c8rYN6TR>`+~lf<(u3@r|d0BOV0V=*_R
z(4tZjK7aNGsvTcB^zjHA5bn1t4hv8Q@HP9fNd|e89(dW4RK^OpqVo*?rg#_ck1X?N1U1dK1t%Y@aA|4AxkS^Dg9z=(hU(6N&-1mO&
zR}EwVyL&|pbeet=pTlI!G@lQVQp$KeL!+lilkZD3N0miDs#N7$9xY}NVc5LUl=zc4
zYx0^h>C*&>a9V8Jp|VW!kNUwdqaYqc#vfL
zl;5L0v4`}>gMqT?<{LtM)n>2L_c)S`?Z(z}Vj3wlhs`u*aFpa}?HC>Yu7x1&GcK4g
zT&=50aldqZ>oH06xX0vI_H-iDwoTlMk%uIE>eW+8y3OMjG2a6G4S&>iyR=1YmMRca
zFNN2+#%_^bg=|652OV`j8Uybv9#7W^wP&iScTe7DEy{_TmHmyBIF{^(ox
z)zX7Bgj(aX)FKN_f-Xioub$eM)a9aK!2a2j=+&wUyD#8NIw|l=E8*Ww(#^KdH70M`
z3EzubYgt!y2|3fdqUn5=Bx;+yU&WW4h?GKl{}JAMqzo!=*CMSLl8~iUl&F<3adziy
z<5D#}iNcj@^N862l8G9v^
z))#}4NPUbgt~GLvgc6NOI!y{9wVdl1%`0vt2-0b(#O)EH>D=6>c`{Gayunz2Vyl8=
z>t-kTy}yE(Wy(JlV%|vFIHK~YENY4%G7ju}`|~qN-E47VKfO0mUD~g-sOLj6EcO=?
z)DI=z^>n=y5U3(mzDhO+erd#L2T5dK#i&rbGI013IG
zraf9LD69VT{(2iZ5S9_>Xk@Gm`c3@eCQ>IzY(soem))m`6FbM|Ui+?)v}=qusBf?#
ztDfcSly8MWh?Z7PXwXMup>4(fd0hudoqA8ImHZ&_@l@TCM-C-;DKrqGJ|%)K=C)7%
zOD!fSC44ez8iq)B_!)Ulw#*R6Yy&qc|ICma%2$I=2Frsp+&Z-N@ouOh?y{stIYdFH
zpK$*A>j^$Re;Mhj^#VOQdn`%2p%_)oF9k6G7l}?J@%g51Q_eid
zNNx^g?pXt3Vedzo`JSV2+EQ}YKr&bw`5^Zu_a(B*G)Xa#$^nyerCr?d{a{b?g`lDL
z5ulCYUqacl2qfJ2*z@L{Sa7RywBdnqK5`%0o^Y3~L5*G0|5aY}M{~#Vx96QfU{;cJ
zJftK9?cBXQdu09vIBzJCU_6@-if;MW>O0#YMFE5TT^0^#h9Zgm$)`ggtWVqOXzmkq
zQtkBZsIMo1NaEOI!SPiv5-m-Bw5%iXYZv7OTeeY!PP7!z+7HF|Um<@2P{P9!(bcxk
zTY$!aTKQ{h10ebR$@3pZ=FkHnpSK5|^aHl~sw9f`RS=g+`+|||uT3*@Su;<32<+`;
z_CBR~2WDpuJT&R-Mm*xD%s4)EBJh4OYn~+!%p7}Byf5oN^wiPPkBBKjoo~l9hWHx6
z)$f~kFDax0n}7AS@;mY9t91M$-QarA5&5Qc@>Mai=qCMnxmk(|7;Od$BAbD2{rV2!
zum@TNhLkb>SwPtj6{g;8u;EAfgsmD2C$_btavC>g!ymp7ruViI`3!D9VbqEfNcsS
z%X_+y&`yA(%whH_C>Qj-f6FJTjZqp&;n67dM>H9FzU0(YKawufbdwbPfE4wRu-3~;u%}=A>4}w>sEIDEU-8L*2qf3+z_V|l(3j=#r1
z?3{8e@7NU3cPl+M|9c#;bud3mF&YG!&t{M3Gra=Bemp*G6`iQi^P2D}zCKi=HIiRT
z`vE%k?CuuC=DO`u+W
zX_I!c1*G?am=EXP0#<*OK@QdjsCiz>kt(nR1ZA3hZXc-xFT`%xrE}d!hsEeJSuT_y
z-3GhNWZFV>;Jv`5fSx5(M5rV>HLjx#SEtdH*3aP5n=I)km0v(CxpTq1vH=uI3?{nm
z3?rG!!oeFW_t1q0)hD`d_JZAj_cihUkHG$nr>wc3Yta?1NSE=q6@cY?T()pt83;Jf
z7V&lXHKL&V_S`EvpX}4Qx-aLw7dU#v%Hq=P+h|XV?_^KeYjom#QTKKE`@rck$KQm*
z*}#KZttRSCByi=3X{K>0Lem`Y^`>;Qz{wZMh2Obzz)?Aarw-{U$f)S-OqWRpTDzeu
zUGCEWGVQNX9zN8DLe{S&dw0J?qX{TfJR=iS{ywq3w>lX;t&vlfI*5^_|Fg+|3vZDR
zc}TbM4g+yI-h1BM)dSSmnuC~~gAwyvSN9cTf8d~?w&{_U3}kC{CFZFd0c&#dldJD9
zp*>$WiZ(4Sg6c=3tgfvGG41dpQG|IJ>#j`BT8)MW4ghl;enII$YyYyC9fY=5b+F6aiNv6MqCo
znILnX-#21Qbx?SMZBSyr6xjHr%AoP)Co!8t#knt905zxEsqgL6M`_tLT5a(s!P={=
z)N!UTVvZ+2Ys*Uj8E+ge&-f((9=7}{`>F4W@V{q@R~(N6W7V*ATQcAA%~R?nvt21f
z9U@?7#Vi1pW{*oGAL0RAbYW?%?`1&FslG6m8XBbA8*6j<5m~oGxc5bNuoX`oMkR`nRH2J;y?9;dE$e&)*g#5LTJ=>Pn&7&B5BNH
z0Dr$YPYR3LFQ=IMh2Sj+I-GRyBPn2KlDj5R41D^qtGZ4@3FagOvY3zlA}%n{Ek-WS
zl6ZN=)f(8kNJ%nZ_r<&sBj=ibvwm1+0QXmfqcU865f4W>T}+y$Nj$Gl%ZXloPrAJx
z=n&>g4=h=`_^mu^20v<>C5kF&6-w4o6kUEFwf*%{UkR)X}lw)<@5YIM+
z6Zh(*kVs?%7ZQylQCJ3XMxH1peOGBWVifNna$MN`#VekY2+^@$Z$EU-Cui4k4v1tD<^WhqUGx-S=}^lUz8(
zRUV^mPSmH{-a06oL-^&NG2UTvCY|0A`Me%vM~WOS3m6c#BS^&7lu(fosp_Fs)Vatm
zEKtbw_Mb^J&M=da_PL3%n^C&)Bl-q>a344ILM6hP%6hZ0TJP|K&l?9%pK8X14t&du
zQVp0ZUp_Ia9btwix+elS>#$Rno8Lv&E<80vnm8%dh4)(CWNQy^!Mdm0-+1WN;g<6z
zJ8>?Rcm(Zs%PS0F`rjmRBk2#=$vuSjO-uuB`Cyq6IZ=gkZHz8t^Z$qYYUDoL<(k1V
zZV4S<#0GKB7uS)u*Q>C1yt|I@as_V3H~+}GJi`Z8beALshH$ueNK
z#x!+zH$L5{z=Z!r<2bRGSXyFF_-kAO&;7S!-^uq1AL7vE(N^%o_2H+6ua3ON?@hNt
zw=P8D$`51X`;^C^Fhxu$)mR^V+48+0vbG%B&N)**S$GQZ7q{TFh!{A{pK{rd=@Vqo
z4x*O(_#XE2vm_4%*Fpzd^CG98C2;NVr8KLqY{-<$ePC6l3oa%n>0WPZfqxRab5S{l
zNo!w5R|`tuyAzL!jo#(LR$48!A@@#rC$GHWL`n}lGIeXOJ9Rf%XB{h@UhjsP+Xucf
z%ydDjxc)?SyHD_Gl$#6RrH|0}aOrK$j$YUTy@oTBy5K_V#f%mwGEds>xm#e?Fl6C4
zD+mItq0%1
zJS~N^zhfOxpE}baiL(Q?CC+Vbfqq=p{BQqD(P8Z2+rLPXdj}7U)$5Y%I^o5UT3Sx|
z4%m9_?WdqWUC{n!op0BL4#=wYM1e4U2VK5#q9BDPm_5o{t>aD3m9`Kyo~m8&aNFym
z-<}=N_nF+y?qh9G>8eLtk54n?efi?PKu-g-mkHYnq^P>*}@J;+?R8l9h*6u@hETeEj~bs2v(iv#SM9w?f~_1=cu;
z7Ffo4_sT+WE99!WphV}@1~(0J2PiZ;;DDx#e8-{}6ZrV(Wg6U#xg4tlQt0n_MwNo1xpZFj+x}oWW=m?~zMaQc!f?#TjrRL$k2&Q}Id%Lp}F!}l8UX|i1cTgsIPqzHF$Py`m
ze{J*(;pS`TF{O6A4i-QQ*v#{{tpcj&{or1+EQg+fZ@1)viedcjwFaIC1<S
zLyoWD+8E~c-pYV!zR})Ks&2xr!xfQ~KSJ=KZvDp-T5sSZQkyA_YytenKJkG<;5pQB
z%;a%oIS*|FjC0(xqp(=0;^7rg0R4VQ4(3qiz>JsLqCbb?;mAtbY&zpj99TK_mbW4v
z>;FBmKki8m%r|@?;!=|dAw%_AWXA)z!gBXV@z-#CkYk6*-!m1n{rm-E*K%O}#Q-uY
z@db?gYWHB6J03dg;S!B2cd^ntucZBF)3DwD==$<_uAc9IN_ZCuA<7a`(ymfTQ%I#n
z5h5j}NRpxuA+m>7O9-iu?Aby$`@V!yk)4+{TS|oAnYpjeKi|jW_t*1r=A1ioX6~Ic
z&vWj5aqimP)2vQZ;^U@UDoX<4?+{
z$w#yD<=uXH6!pfyMaLnGVvP;_Rwy1zUE0yZwoW30nqM&zIq~Ns#kl1mk`q=yiM+i0
zt?x=YwOs37`i)~*6eFM2+3p-gxxptDQj!If^eK}=EF49Y)8jdY?okWoh1hsVuvSO0(U|DmGI{%|vQY
z`lIkWdc%q_2k#O}DfKmr*tcS8=&qr*d_xhH`)yx~{k>wUc7Szq{%|pML414SZTKPF
zvl0_It&~zK$<#=eQCvbTzO%L~L#K>-9T_VnnovgB&7L^%{9^^h_pEa9_OME-(r&fw
z3`Zr!xuZt(#NkRxrZ7%0J)nw`)AxF{i=&p>|8o76TRpYZ%XO*?e1>bOHxI%!lsc-Y
zFPjv!gR*KUt3$&IogSa4Wt$INVackaELT1=to8IBoY_FoKVqfuOv=ma?9y-LIhre5(aotI6txPVZQ1nE{~DjV(=>%@vVE{}M!1;T9J{cid$54|lU}-QMFK_Hh~0k@E0;+v*Qjc0^m<2i
zEEQdRpr({sIhl3n^_yZUUucn-yk9*w{J
zfpcZl%9Ir2%M!&@fZz2)<|BEO6L0+L>+cQOs4rF2!)x~Y%r;e0du2}^Zn#!T
zJr|tG5mhRr=D%FHQ)4ub()=Y{arklyRjye5z-Q_$m40o)#dP^I>g~dRX`9&ZqEFuY
zpZ@3a4DrOMcYeF`jv8A1Mz2v6`}O*2rwL<|fsnfhb~A*}St{
zS=99>T9b)y-%(!P4il00UQ@X$x;;|TR@954VD~q|50C)M*UjqJD9Zb0?K45X_tam<
z59V{EPE7i$4`FQvM{vvRHaKXoLAGir%kMu
zWFy~WR#cnii|k$k*z#+txv)U~+yX`a+aBwDY#
z`@BaMnty3W-}h75Xguo4<+Vv4kwtK9zt5h0WE(k8f0t`MDtcA8$8PF9de`sVIu?_I
zj5I&1?c0=xW;EGcx;YBa#f)b}A|Fk0>z6MvmT6k(E1cO&m1WVEcSvbEKWDpXAsQ)qKI+v2&z)*eV#6g0*l
z?fa00uAZuv{Dd-4?czE!r@uw$a*F1>PfrVxNp;}Jg{(aEGvdmbf+HW$iQL~2A1bp^
zx$$7#<>I%<{M#Nyr1}=2Rh&X@?2)Lc)%cJErzbiXy;i=mLjWz+m1(?lHVOXz<*Bvi
zj`2uW@tkQ4d*Ggdy4BLc;6wmMUIh+_a&n{!lPC)!BNO6
zes_xES68Iu>cwky@-$VX+~}RJk%7dluTQV%e1~4F@s#ROi$Re-l`6bDywJ8M-)=2k
zU`i#P&oy%Xl#LDxbk?}7$v~cOsQU@0m0A2CX=KGZ2!UnE!
z*9gd`=U>|{@P-mnI6d=BGKrEjkgka8K2QDHbvf1ksXNv2b$KVFCxkNCl^QRi{hFGX
zkoIfb0Imzq41Vm9iPh1}e4#ZNCoT9(Z^~6!vMKZlz#S^Ig?Ga?77bGzoc|(M%(4}hpGgsIi~f6UPB_4GsmbkuS}+T-I^-u
z+TKyu^>=VLxxJ&}-=*7okG!G&`%3Y!kH=F|e|8o8QBS5uC;jC(B;Qe0(@;J|EK8>TcNM)-zGN9i(Y(HW$Q>f9Lfc6u4a9&woAFy5UmO33gcW?d*7#K^@sJ;n&oeMG0*GX&zrK|~8vnYdg+`88FSrj|xB?Z2TcU0hD
z%@})E8nyMmvVqlWGbw{}98xUr-c!E}zHI&EmQAS*{EBnBoJuVZS01uoolaeraj-q9foy!(Rwhvm(=l#O7T!=X1HsLxr8cN~@ZKsld|`*ZJF
z7IpZ%+)iWo|F#i|sS%p;6qP7?aFY$xhn9-hk6ZD5pxQ)z&-!o7rcSF~Tq@s{O?_MR
zDR25eSg*2?pXo>H
zeUIhEjvR{G6=-i5twvGvIE0+goe$Jh`|}__kTupQ)Wyc;Qyd*9{I{$wp!g4oOr|ZP
zsEMT|30r?aJ*#N6ZG`_LWqQcWD<`Ia>RiNg`#^Ic^>&@lmM;Uj)Rt$vPNsBIl!c>#
z@~4(O%1cnMC9oa-E~(`{y*8sFYV||0$6Y%75{%a96sbAZ3IIEDl
zuj2E3)uj^Zuh&Ft+^$kezxc~1&&S1-QLOqUmw$y+*XXs~JD-(M6TUC$?J=e_ZoNTcoC%F=wlv<`PvZ^dcoEF;>zVQhb9^Zc
z+U}_pja8RvRAtKX=V+NDg~m^+H1a&8ak&zWHeYFU^8HSp6W!aiEV-A)jfpf0b<_H#
zy$!^!0MYGt@6ed*N8{FnZ$v-+rI8?#rQ5wiY2;Z(rGzK?RioaME~YI-7kA8-TqsJX}6??=>J}5C3w8;Gr=dL9RxRRtR|RhJwmW9WKkyG
z7q@O!5?Ns10Ko~%VuE9H@aNh=Pu;#n5Pj(;Sd;UKplfd@!JK^BpCz)6MCM_x!&g6g
z9TFaNk#@zpJc4gV!eLAFpfh(L6ZzM_twfI4u1jAZo#(*${DkMml9?9;TL___7^3fP)Gs1ttDhjHTT%m%D?w&sJa(}V7P6?}!;
zokGui#i+uzNf#qWV)asNr>x_?3!|UR&T8DQP{r)s
zeC>__O5Si7N&{?*p3Qa!pUAUhsYdYQV`igJ4uYmoLa}3{$
zQKMaz(+seE;bN=G1_c)e+wv@AAKT|(=?Zopj@i_Mu%jVIwx3S*yZR>uF#yxx+
z@c2>&j>dRhQWB5#VE;E{PQnK$F#kTEP<);1_uUvJ`!Ktq{|}Z|3Tu&hY4tP(VL42!
zpUi_jF9wckSg+^Tc46e+xGMuAhxs0|9yR&f$^PBA;O0x*E`Q>9I`+d`NrddH5nb_p
zxV>%s8(D{^@XHb`XS^0VBzn3NcUf**6BbKKMvq}F&G8Dm;muSCqh3NC??XRz-{q6-i&M!Dax0_hk
zllGqVw7t^|eO?T`r_Tiw#xT(f4AJM8`uGDD~
z{Ob%}L+*3Ca~_$Oq7^@}Kc*>4W(De$AuXL$*_
zefl(w<2E#Y^P%yYKJCw$cv=n&q4BdCjrL4F!YfVNv2)P+$3VnO8lNADv?IB@QMZ^JDs7#?(jMqx*e%fySFm`=d+nCG{68Lim1(z^Id6OB
zynLDV&^hU!o4G!%GqnHWOny7a?7xr6XTLJ-Vwl&(F)@#c&zQJ`i8V}wU_-`fWa0%T
zjx*Q6l_|p}CU*8rG-T@6F#YXiwnK0ucGsDDCnhgn#Y7FJpZA&9sW9as=6x;baQujm+!IZ_@gTWEy#x>!8il&u6YrJd^J$=+kx{7in~3>i01FKg@hy?yJ*!C1!ul
zOgwp(Zol?{#?O=VK2a#90
zw>izr*O$@y98)@eE%BoFd7L+GCv}>(>+DD+am(cvtsiiu{fMioBK@@{EFyaAy+vc1!og~yytt8c16djXncOio^BxRW+qaab9@1Z-^3|KTohzhW4D+K!SM*%49BT23@RvR|y9g|oJk74cC}Vlt4)cpePYNJJ
z_Mqpn(r<8kO^sRt#)X<9`M7=SrH?BxE>GNIfbEy~Og_W?>+4ZVu$-vj*MA8R@$dL{
z1kl&)m>0G;^8LR4GT^#x@xhqaD?gX3#n-*de!E%+knwnt2Oej@d5_)=13bSmrd$)y-}X*=;&+(GRjJPxL0<7#`&$rTwPLtmB;ca*LKbb%M}Lic
z{uZ!KueJ^QvA$aT?|H!03Nojo0ezpVl;L*e*3h91fG=*ENn>7Gf2hj{_vbOasP`D)
zhuZG<{{gPBc^8bw_Xs!`i2b!`6#JI7i(t?R>L%{|EF$aM1<$jKJE2CRTh
zK9w^^0qvFdZ~OuJ(}_LP`QTS4PrG#KH{ipkbSzea{;|51z{-{$6!L!as`(&?)UV5%
zZGgPzzb6@{UwhE^bG06Opgq$?#S7bS(KhVf2Y5*Hm?&QNZTnuU;yhsaz?Dx3kVCz)
z4d1Vhse^`&fNk-=1M#{Zc9*NX4QMP@p2`CLELxVMnGdpP&(P9`fSW_~KH@xNFzR;@
zUjOs6w&N_|?){
z<<^2-y`S^|9@nBnmA3`(ZN*!XADHUqNZ|W-!MP+r_y>{A_&$UOX}-q(^Oiaf;CY^n
z5s|=h%&h-;TR@f&t_-|Rvq7ub@qKTYT((3HP+jpj!t3q@Os>OA@dmD-{QOj$#~zS
zb7ih{0o;0dd)hs~_d3$ou^o%dpICgItz`v?4;2SAGna$RKhERy0HfWN3r_*xMy?da
z`JiH=b|PN4kc%H#alac+rLkjNYu-)bYoZi4i63bW<0OumZjkzd=cRQ}^a=L6@7yL5
zfAX$YlDtdYbUi_#C^M2Dg}hrz{E$<66oBnro-ZPC^)w|+@*QE$4|BLZp6Vi_>QhQh3(&je7lY9ehbc1!PuWOOX50D
z%NRKi(t3B2^I?Jc1VPuQG;Z%%L(YYpcaM>Idf%t(KUcJ05<9tfbp5Qf@Gm)svIkca
ztXDrz^lxlr$#vg#XNi1-^AR~8e#!nN>(u2aLi9OhQRF(QaJr6Ge8s~6`#Dj3hpZF7
zXd$V;=*u}1St>-6oEviqbEMzZ!!(c3ilF&Q*kYO&XywxwV311e+J)$PoRlQZ54J6*
z?SCNpdY;p3i9aJt>AKt5d|F?wL0{)1Pv=Jker2S;AG+4W{>Q#tBA2|R*JGnP8);v+
zUYYoN<*Yne=fO4u(%x6Pm)M^Y=Oyy8SlZ8%`7{rrG-=+{%;aOmOn$~WP4lDW8)*CJ
zPqh8Q7qtD&92&XYX+Lzn(7b8GCwhLrlWBY9)wErk9S<3wKa_6INT&JAaf%*CgN1J2
z8b;rLw#A&pUb~;pvjtAjJnF(Jntzq7r2Ew}rtuOpA24C&8&XVOeK~^G$7#{%#^lGN
zOw?fV=uT$-B=MHE`@_tKvY2)&nDy`0JM=uwGH8G0-qZFwn0ezK16r;UpznJgb6q5#
z(d}{3G;%ZR)vzha{khNd_ka156SG~*gVt9v`S%&7KPpWA&VGy5i!kQ_o164^o2hrd
zPRo-_Uf#y^bBxLB4>5Va6_fAl`q1`|lj-~D%)I^vvyLvPLF=C{ruBvOG#|HnMYlKX
zrBOGD#x7PGCsxzz=5dOamoxMB!eCl%I!v$Uw{x@|n+)Bq&8+{cq|th52K_uL2r&28
zLt2hBp#6}wp`WL33+er1?~zQ_V>*?-UXq_i_|F-nU2!O%prq6^!2|c`{zen&`ArGX
z{tQ_56a6|f1@c^$_}3FT>>%w&>MOcVe@Bj;==C2yBl>yL^z-`VF#UcYg5x)7UmkFe
z-~g)|LBrkj^F8p2_G6_$1Zi*j5={Cl7kW#0JL=0N?Q-sWNPB^RD3P!IWg+qVNk9xa
zAH=O65&iyCD~Nt-G0n$q+LTEA@PJ5imbq6N6N!+WDm?U=o
z&**gWdXYePmDM7IDccz<3B+12KQ|DFIbMrSbGO!^%KQ;_ZG@d|l$}CT+|&WaV3U1p!mf3u|IOJZ^2gh5PCM4>e$Wr~eA>
zz65M3dcFt&@2aRtti!yiSZwfR5}?u<&wU2~CjuO7bAX2(^wZeg26E+0QYGd;K3}U+
za9w}@=yBE-(68Tl^6npyKL*PUwEqCU(;p#@^P3Ra`ANKhiilcp2*yG2oNuc@p6I?`
zya)Jedwpkn1jwR87B0s?pPKdLf*i=3dwRKofsb$B`uGbcw0DjP{I-RDZ>&=|brocn
z)#o=f*F*kiqj;kn#ur=1yZrBe48-`$<+l*@^TY4*&w+h>wY>KU;KyzOjoY?=0lv7W
z@P{1Ohx9>#5A5#f>pgaZ@t!M}jF^GGkvD$k3S2keXynMkHsJFTNvn$?AE<4P4Am|w-!4X?Gcd(fue
z^2c9SLLL)86|eET2RQ_1cXWI`CiRVKz%U$j*EGzJ#~Jc%e3%g~8vO4WcL8Ag|#^!+cRN
zzqpml#C*VxefM<84B+t2?Fs)uzb<9mr^CQML-S|7ywIPgOjHQYAMDBoCiIITKN@&-
zZM+6}_nCmPr(hTHJlOdk%=7WBNcLgqN8PD;DL2gjP2j}>4VcI2FK0`m{`Q~)m2UYF
zolviD6to-$zpf>JI&v56Wn+JpdV)W)EW51>KrRuRw|)s=wy3Z%E1>p#`O!E)jiIES
zj{&U?%+%rMK}07{hyt0#EwJMd;4hVE`Cs7Yz!71QIq*kS@X!+ukW(AXzT|*j=GEuy
zU2vVPL)g+IAaCDi9%cvYvGqRNQ9Rz4Es1MeoJfFItZY{mOFMfR8TdGPb1q1?*vATN2zP*w#D
z-(0v0=SPBncoyS4kH0r$R2}>jDjuxJ2D!Iwore@$r!dGarUP=qcxmQ7!11!zm%hV(
zDelm$SN$jeVUXXLIvC)&{JL)Gp$+`Dvxr83Ja~bJv!06C6O!5WxRtxeTYM7foXFo`-
zw#R->P1W||^?0uKz8Kq|OORw>e7DW`F2(`vrnPuo(jQuFz;eTGF8HJa>LFjaU*qc+
z<*N+gcJVAHat^*!yi3k&tvCmAPFMKTJmS{*NYY-TOy`M$V+F)s%V9gQ^RJR4_(gd$
zv5Sj}BlyhgV=x+yw~+NK$xkKcaOW=r
zvVI}@Kgo3w16}8^U4hw0l79vAmlHmaa8#1i)4GMzh~42aI-gRort^S(OK6^@kU-Zz
zIi?m6JBd#;&+7CVB(iP+U1zO!r+Hqf=mm0LN;;y69*vg}?Q9xpeG{`T{wD4N>96oPy&l*GI?{v96B^SbFwud@I%PvBFj*V0^m=k$qV2bBr0eeHyrsm>y}Fb<
z5076a61nEaFhRYGbRMAm(woSxJ>SXmGd1l(**#AqL=aR
zA$kw%lf-WimA}NVQk&K<_<3&)mLctyDQTp=V^EO9kHL!j~#}Dz6dN&oQLgHtX<~ZpuE3?54+x?uUOzKhm+nNY(
zzji>f2DeKEnmxp*!rt=>qlWaJNgM~?YGzeqIeI={C`M1k-6a0{42kw&xj)_HH|DFn
ze<^7!*QZ#Lyr5g~(J+=5N%4~J9HQzTj$r?MIUd}>`0>v-OMJd;IwYWk&o6~{=Xo%H
zW_;_2!MrtjnZbHI{zLXC@}0x!XRM0&IzxwjgZP|T%Xb)FdBORskxA-`3&eCVqYZ#{6(yFKDWc9goA4m+y=9EIL7c7{ldGWnx}%AxfLi22eP~+#c5r
z=k7Di!Q9~5u}
z^O5qiLK`uED4emHuK}pG*O5R86Mn$Vg{N>
zEtAoJ{*|{~4SNH2Ma`1vGvJjLt!R!vz~4T;+B5<@ow2@RjR^GTZ~lE&0rXLZ^R9>W
z!Z^q~c;_p)FUoIcK1Be3nK*r8?=YO-eXJ)_f&YEhD``$og>&2eFxLqa1`=#I(OWvu
zhZYGXd`$#i8@aae%mL^pX+`L&AGcxNVT(r3dH2IVG5_W0yb1oqT!rglWflgqYFZY*
z6Z{^y=gn0b1bNo<8OIUeWjhX9ozohE`t*}*nJ}1t2;=TK&d0!?ukEvS27f<$C$MdR
zadnT}Ur?sSKs$VQ#C27H{|@)~TWlEcd-C&UhEcufN`?lTmUb`7k#qPZFVKt9-I9JK
zy7Zw(tTI^*_xe%A@|Ph?GW*a%xr8r!5A>ql`+o?%Gw4NEZ*$)B3>3IcVcrRC1{yF=hHrfsNaJ8c?gZFv{wUw61yyPzd?)xf2`*H^X`>=^lye=teu&
zvLAH;es8RipRNqgp;Djx?|jHxDj$CLi-+g9`H5eN1>6@o&Y^$jz|Zb~HIu7=M{g|L
zy7LA+mrFRJp5KQ1_$K?F2oK0FnoSPIB)~q1e^a3A3j68e-Vj3%74s(F#x<96U;|cKhNoT+X9A75G
zs$Q=JzXitzMlrt+{Jptg6vkg1l-u?l_M74OQHL-<2Lo=tLeQ&BY!1Wu)Pw)r<@U70
zd(WHA8jd}{^AFs+9R~gE?-N~5fql~3{-roCc2?2antTcB6R9t*{m$<}?z%^~?)5=j
zT*5uSAM6Z9nychtKYmDDvB0OX8wL61nKi+@x}V=Z>;-nuRYnechv($bc(PzRrsR_qBj|E
zRv6li7oHi#{ob8AD~Zn)Zr(zKhJq3i`R9R)XN|JfU^ZFvh@LzTf}$c{752v@X*r|
zjDe;D`4}sd{)FK3x;bcs@PeDKB>&;|zl{#g7*EE#k~~9x(_v+NKSQOsO0Yfu=?kS;
z5C7e19Uez>*=X%DP9N_c8~S0?c{&3c;j
zA33yztV^-rW`Yy>*GavVyL~NLM{y?a-JG05>|56>lJR^$(ENVxBhA9L7gx*r3KcRG(pxioo
zeKc0l{J*@3&gWKFjFR!Vj_VR0ecXrU{o)_!JVI*?ZMUW{i`d0-&L{I(m>Wpsv$wQJ
z|80xr6Is%%hOE;Ch8d~fXIInym@?mE{nN`Oc4+|$WIc}m_lJDPvRufNv`4*NOi<7;
zis(ajEhW!Eeg6;g9DmDqBG22AzMEv64RXrF@8{27k@k))cga3NO>$&&3$a{jz
zL#K#*yYvFtk8v$~|HJ!9TFi`GC!StH{7`?ikGxNQz4I)oqqx0)6pNqF@Ol}#p2ByV
z%*#su1(`Q{!Bt!Qd_D`f@d)>)IR7*0FS_7q8*bOt{z38xi`gR|u)h3_*GC-h;_Tdr
zpRrwH!Pw4yAfC)4O6k)%yr0L_BQ$WGxIKrD_^A-IL>=2bzZYVR`%#g;Z;1Wkm`&$3
z0PI$OScl~X|FLX*{l=HUM{zvxmAiWi$2}E}UJ3j?Nor5FI=((KMMMK*pI!4496wjD
zcWtx-l#Mj%#PPC5KClG$bM>7n1>(r+XYa
z0iO%-?^XrBd{$H)xC!T)Y|y1-ZfIA$@$v*4oL4rt-)Z5xeejW(fG&P`D0*b?)9Rq1I9U=yzR#k;Crew32c?WATOD2R+0fft*mX{<2<9XJTw%~
zuW7{f^lqq6iy!N{S@Rz1%~hXL@%m5SVEerX@F}18oFSa|As;PcgW3A`{Y
z;+n10Kd3wIIX8X>cuy6#SKz^C{mAEnk(#UHApF1M*B+tZ->s8Jl_Q`(sZALQc0LSr
zYW}h%Lq&tAQ;+M#?3)2pf*$tlQ-||SvNosZDDY12l*yjjLF6vrA}eul2zBH~{-}xS
zLnbV2Ew4D?dF3r|FMZdK=GSmN&s{x&dUY;-`mR2R+D8>xUIz@J8$MTm3bOVhzf~*k
z=kbl8T3i^Kr}mwa13hZFgC3?TBuqs#VLw_pBQg-!d6gI?;``
zgbfPZf6$EZ9Eg+eRF)72MsA0~4K%L$f@9!DzLh=J2
zPIm#{wQst2%zP&Uo!jp=GPke;O&08NRa)7BlrPWL92e|DfAyv37F`6MmY=?Cz7PY&
zes?>x3-&?I!J&B(@I34BZ$IG;`)Vrq#CBnjrPgvkPX;{Kd(Q4W*k$fJ%U1#O8eMUX
zx()N(t(IGZ1Tkf5Rj2{6_RP-v0^q
zt!~Nl`3)H;)9p-
zs*pwy2cakBZIF1<;A^D{RDmOZKDt6F#crEc-&u*$1h7;+=6vjEbvRM7UEaNrs>>=
zxZgInwcAD^f1VWno!<_5z$x}id!XMZa%(13z;4ltiWQZB45!zgfncx9exeZbcQ2V!
zTX8*Oda_Lp^U2UzE>nD7801R0<8y!~CPBvuaLe88F*weBIkZLwpHI~1*W~v}E`7(P
ztHH04N2{9gdkq^#+I*aMB>7nX!}-mL)5{#NeW$CI4bC$zbr|N&1N`Hq#fI~m>n~h7
z@VHwBUcSft{0NJ%Z8G?&nX2Bt7G#YP-#glX9EUIHc>+3pTs3A0$UXjoyk}VKQ#+2&
z!zDLfeZk{2D_kTzV5^+iV=M=Lbl-!|VLyFQA1tSI~#D47bE;x>5
ztr7z4+tG}A-u^%){2aKUW6q9kE@z}X$oVPUJBu9r+@7w
z{w=Xnrf~a@yLA0MAejE%GL6zQgby
zGkc2dt~{|L`XVPWV(+tHEwK~w^Cx+VV*PrOzuX8gB)o23{UM@Xk=#lAnd|yX_Vp$DnlQO;dhNcl9fJ4!#NpWw51cyyMG3}wly%$s}ohLJ1OAde~+=7?uB!2=7;iN
zB?DcwOx)lK^~%w!nWABE{zhArH=DuuyKB@QQECk2@???8fePJiixLdGEn{!<5)|ZZglwO+^T)MfDiJB-5O7a?@ziUuYVSSdT`#ldicA~$p3|o
z{XD3*v74`c>3MY+8GoL&+;^RUay!4yB2EU9d2hLL{!Z98;hWwWC=a89KTp<&TpB|e
z&Ji&y!Eg9mqa)a(WN02a#XpJ|HTlJ
zcQLqk=f?;-^KyArpV0`47~L~=$9EVlyZjs_pZtl^)RH*1YYw4nDoWfpYetY;y5shF
zmqt*<(pR;4&BG}2EB`K|{GaHnp9+IT>nHMj|Kpch!3gqO*zInvJ%S2Pw)kKDIgA9~
zxNq~{1^%WPD$ct9L~oAe#+t`=!tb5cx+i@*k*dla&jF)Obo)i@fWJWpI%5(u7x%Xv
zX^Q9+9kB02jqe2{WVUspE1yNOV>fpqv*LL7Y|#$H8+BU#_RV%=zTnkMpQ;X2I5=hc
z%)J9yY}eRb6aoEI)E1Rhx1+glW!s$l+t6ma(J;;(9cZSU#kY{R16}_#SrWt5fu6Fy
znKNA9jwU%aYt)@=gO9ED7Wl|?K%82&E4HT{e%EOvI`Fw2g-o3EJtZ(kqywFkHY?E7)=E4c
z)?MhajKp^vs7KrNNnVe)YeOu{%UdcET9A4T-zwA1ZRoLEQsfw0C)ysl(?se?7g{Uo
z>%hX^hMHntE;)In1%-^p`K&c>MgCW}c$=zrprYe-((V17X!-X=Ja-frX#UXer}lCT
zB%bkj@@g5x(MPiiw!kNE-Sa3n1m3B2%K6~{)U!_(jI|Cx-X?$fa?5Mjr&hi)4Xoh*
z{$Qbr^$=eywANkJq`HyQp&8Xb6%hBbz0C}kGEmkT=ff|cU%fv`@l{0-Z$A~asJV2Z
zCd)78rS;uNXM@7&O|YL13+`2&2*`FNHT>|31$CPPmlZ)q>~xhcD+8yTTy|8el6
zxF%i33G6TEdfs#eKSJ$GO~-!%KZ+X9b%przu<$bPGl(yqA{}hJ5a$kGH;qpPoLic3
z@gl^PRXfi}e-VhH%tTfN5fW1<Sq$@Q54*Gf2-p=^Yu>Ygc$xl@qbVHxC^mT?
z^9-&N+nrplQ4a5!;!Tb|fPVQLf5hws`=lqYdp?8S_0}KRP#B-v@7p;g&}+PCeT(_J
zQ%FR#8{mhR(rNg7i=DOB7y&!`H5-MaK)z>KvQ7;!M%MK>=D$%#jYTnEc3MzadI|iT
zyTsDf3UagupO7yg-wF}m1AzCo8_nSP{r#i+^Bly_{_VfIEWkd&a#C;s^ndy^|Lt9%
zPySWbzY);hUvbg|=2PNgzxNmTyR9$#;2^YbX5FNO>(tpY51xww@-#?`;e6tX%py~q
zcXV`hli%}euiDRt&*A2WFLq$m(RF)-(L;4#0zQA=-#7h^Wg8a_!Ux13RvgCi3qQAr
zyMSspC5`d9`}>J%I6j{@4v)5Def}1E@;+64B3pbr$d=15T*d1kbTUE`^NVnuyd~HV
z6aHpjEDJ4L@K69S@XVhfyiQFO8{%&Qj;ttm!1b+i&jqhBFA&^zi14vPY1^lc?P%m2Hx5!N3Yh$MWfFawb~cV2Wo;c>%TD01D}kgpt=udg;+N8ZzO
zbo0O`=WtG6eNO6fovIUrXT8wioWg!7mELE=en2^%3)>xtbs#*gVlITN(-jHFSj@AO
zc<;tyyBmC`xUt;|Ikypv*JbO;_pAP;4#W@FUS7ic?k<%eJTRlfE)LtV?*Bpfchs0A
z@zX*1EU_DCKSJz$5(i&nJ0D+J(tde>uCGr;)89oaC(_^l_#cgm#CBf{))QV>b7MZK
z+vf~^C462h=?&rYLNe`yx2}pdJ%f4LZu7guZt%#Y1Gis4`L7cr${*Z|(Pd{1sY5pY
z%p%{*HBO6=-yOEzu&c)Ub^E=^bJ^Co?H9gJckg_g#r#tv*y#!$e+}2jer$J7yI?Vv
z6K;IJhQ|%v;{4s6GMudh1~_I;P=P@5*ma<^VK+z0&EZM<|Jp!G`+7aVVn9$LJ37I<7n_FV~_
z=byX!cQHR)*E>)riv3wIe#(3a*d_m!{qzOSkNEzAo@F2hs?MIc1a_|#L;stF_#xyo
zzeWJ!WYSTN7q`Y>UxdfX;eA`#Hm$G#+Woec+_8mtv(k$Co&xofnz+22BU2FHcW--+
z`6%o5!n+e-e^NZ)U
znYjk~6St6_9DNAu?7v~O0{ncvcrtD`oG+5f4v*eL9jcFde{-HP1HBqc+#wDAa;}ok
zwgBF#-{O8x6VA7u&BrsG;T&oGx7#&lE$~KRSDiI|-DvR^m6itJjpHL;tX`S04xuw8
zEejcFSG>lyS~%A_qPJN8XWWHG+^2rDfqdv+&EJeSP)8A+JEfokJoxJQ1L6O=(dHUi
zhPO*Ma&9jw9D#b!>fJkAdOktDO8Wj`VK}cJ8GDGPBy^+cG0&GlDg!9)_-Wfuquod=
z?t^U+yx-MZ6D&WIJA`h2{p?wEe-!CR+8wFiF@z5O`ndkAcQ=y%b6Jts0=|ba(S2T;
zJB)NFtt^iE5p+fAB7c+OPZXar@x&si8<}eF+?6ZPhYTG1uGbzKMQ>Q-`gxX&po83V
z=Ps#?pnZsS^5Z`S`u(4VvCG6yw1M+YDAhQEgv*T|T5KIb6PI?$s@jgCn;R6%U55tH
z)k23{)G>nckK5(uuOC5SlTm^f7muL9U8$d{fwx{*W34>5unn0TaeU0Y+=ANVzZ$=}
z-inkX%$k;{bRb*xSHJ%BbRz5hoBJN_Zbf@~?nc40RKTG46?6O)4H?Z_nA
zD*rI>$H6t>aVb)*C@zJKed=Ee5GdqX>loOR64;A}cr#2KaHcfdmTF}9|Rrd$CwxUBhxu-*?
zThL#jRh8RWTG9UB*~fxe+u-|AvwQm+JJH7SYIzRDZghVBo_-H_{tsy{Z)gGj(xJ5Y
z;~YG{tQ-%{^{?(kU+pG?&INX%m`~-WWFsLzz0LjE73z}aFWwGUKwVAZ(BH(ZP3>sa
z?^dzv8#>{;ps1{}ukhW3S9zp~1OwGa2e*rgGLT8ovx3pkHsp3&+O7VQPUdLqu~c2{#^@`
zWWaa&yfyX`yim84`NR^H0@p8VH(po`^LSWacdZ-tQ@`l-H}bGQbh}zxg&FW&z!~eW
zKOwKPm3*^d9NML=B{r`DUM5hwKQk8>^wK0s?ZyJg>#DXjJb-;rSM4v^47^F>vu4vE?A!4$ztxj#
z;JcQE-IJQ@8Hj@o%?!Z)T(!J8Iu7#C0q+}wvS4SL8U7#!c+Sz4)onZwr#8iv=;8Hm
zvhCFozttu4obdPApVBTH;JE%#Xyz}DdyLHPn};C|
z1;?nAMuYuE1xP|{PT=pYw5}f6p9^+AiEh^xAa_a%Z7mTNWR=
z9BAf!Fx_O1=JU8~#4UCL?8hGss8N({|(b4u@2GcR)$?pjYPr4|{yN
zk9p2jo)E$hOUi`U@cH_fw{Qj{%NdJljPX5!A=u8|`+O1}C(h{(d5>S+e_G9?x;?C-ytH^2S-rAN#c#t=LX|$u$bw4H*~fDEv{DDJ8
z{ej`r7IfccPi_F^`Tj9!%TRv1+gT-ipE7VoMDZ=4&*uYIa$$Z^RpjZ!aS)ZiqZs4+r@n5h*$23wRb}ra;G*}>
zs~^Gm)?vTnjCwxz_Sk{ENP9t>Ujm?SU33}V?@rjT7=iKcF12w5@|>rg53WT4_FPWm
z!RI`^spi{$81L1WrqZq;|Bx|Go(1`yuNx{^VVp>8X;||T>ftCTYZiq3$pb>kFJKp(IW+SR@R!h}
zR_A`;Hz(SxcU*+|TK_`YptU>j;W>sX-cXPC#i#zRP*2C&u9N;CuTYctZ5-xb?OOTA
zM_?bLbLSPl9w^@uDl-}n^(;>>4B~|OT=}K;g~z~iR=Jp}jd;O){BSf-4D`&Gw%H~Q
z&p%*sziR^Q#Ag25{2`c+&qpcp$iw_Dpw|4b5aI-JJ3aTw0$)nXQB0c$c4d+1x?Hb0
zFt4gqTUqVzAq%gBUu=W>B#5|J8YpxSp}iAd?puQXR|U^x8FrFuy&4aicKswm*(y$p
zT|0`MDOd)(&Gf;4%ZGk$;^Xu5zH1zzC71Meows8dHiaC
zG;46Vg)QqN@4kG0A*Iwyu1N1$v($Z%)a!-?Z?_yITA^19c5n`o%=4zRzuy}qGWqQn
zFYpc$X$7lQJdX!S1Gk4WufQPrX0g0u`J6#=I%#OIha2$Ve-76g2FSBHZvx`3oAAACFvfs+CD}1HWkH-v<
zi4OrQ466o6>>MQxA)f)V+#&Jg;w1y*#Fki>6&nXgNXeG^J-!2kP3e%;>dXOR_@P9R
z>&JI8UM#DdBh^TfgsQpQ`@fUIN0VLIz++ZFoO&i6?7Gk)&}W2{?0gR6VSb?75Sj^(C~C
z+%xLA?(Ne^J_#B&`9l2cJAYina-JsAVw?N%;?5=_*EhUw{h20WC7}G@4QIe+mdRh6
znuza1zWYLIO{6e3TrbYMiClBOqVqnjiQH8!HNTwOL>xA+@Y=-RL;@V%iktm`>s+6G
zx}DuhI9&ZMXY7aX+G+x?I00|?5UO@%LZgu!{8f5pw{jCXe(%h2uCf;5xi)adO^DAn
z2LG-<{0REb2f^#YIgPMS^a3}Rb|YCmmaeF4)J#6CcstX4ZyS-}I{D^6d^?$Q^6+lK
z#3o{pH5Hh;NkB~Oaj7409i5wFs#%jBAy$VaxS(#WSxM6pH1
zr8%;RT;#eWzCN;rc%OOMt+1|*sB8$eRe<=n*l=_32eA$^B)Uep4#wlM+SlS@(7(lY
zyz8`vai&)A&C2iv#IwJZ*RihdAQd8#pBvjDuCZkD*Efhmy&u}gz5(W|Z$w$}(cN}3
zRO-bS`3&lDF|*vg5XPfbb;kBmuoDNY%4=bqReU=;buyrhRPMejd>P`jmT2Sv`K^bGV
z*KKLZt%Ui-J-_jqIOyx8$&u>@axt1>xp+U#tIpf&K#%^0pykcr*F_)wxjX~-Vequx
z-()D)V!Qas8Rk!YRldUypx!d^dUgE%F_yz%6Z+Z5-)k*K`;z}*&1T#u#J1nc4edte
zo4MEPKrX9kgcs5->8ti?og(2xI34|Cww8l!Q6Kx+T9~R6m5|g
zytt<3hS%4ae%Oe1JZri>0o&ncUriadL(KZtFw7UW*=U6IpL1VH_t)~AfAo9>qz|(0
zxQOvLdy%Lo*j}|Sv$-)Y!urBI39n1OahAq?^do(~qFzKs+7i*wmE#lMH#z4S8}gik
zb2rpsKL3j?HHeElcA6q~FWpMl8~e)KSE3!iBKTz(^Hteqe?$~byLA*XKtv`S@8|gF
zei_!wFA_-4effFy6Xj3&=Uqb%
zE0h!AxJJ*DvUtW%*EJIyv>ot$EEu`+-4W2=_b4a6r{k6eN22!$oW_2ieDS+yH}2BZjcQ^!SHgK>ja%
zs`@_OZ_8T6C5Y|iHw97d@+E1Rjez>O->*)BT!{&X18w&@wH%b|-S)c36>!yvdE|OP
z_oC_r^8t^f3)=k$=HV9x{@I
zaT4}dO7#$wm$7%h!hIe*LPxf>LcGTKic_#Q$Ypg`-@)$>lw-H9?1pieUlsEM<2UvX
za>9ziFLIQr7rX*;&B|GVnjp9R$+m0wJvaH>&o2(+om;syzaHe|rK5k{1GzaW-%U|2
zlK12?^gDh3o1IXM^MZ@$&$}R3*z;#r8qA;UE7us9LwW6dQ^i`~E9_3l6o5%T@+o-uIrY9|Ut
zKL-|Q0^jfIFx7?c7s6u~d=5O>L-toZv=Ibe(lzFOJdCH8EK!$}R^^8M;b-PKh;0Tx
z@<)M3$n0M7T(LioH+g_W>73!$H|ZrQZ+EYHPzCFj3YN#`rb9oQmGSmwNiR9F(P4+q
zyMEGHwDI%V*S+M7oAKiPuQJEy-<$_~tm<@oUZG&wQ?u>w+!BN$JcQi?J4R&twMQ{+ed8Klt}RH|%@M
z-Z-y(;&n4wwk>VxZ;=+lo|NXZzO{wKoAY{4^R$s2Ykul)g8h26uT{jOFix=fhsx1E
z;J23yx%9#Oow`un{OyZY!nVG^<{CTr;vUJ1Pk?2D?9?(#43m=EJ--QTt+5SJ=l
z%W3Nh`|B(p$IJD=c;JwES(Q-?_Ib|dovYwE{upI^w1IKyR`*^N;}~K$IV-opxT=;C
zTI>gM={lL$9>O?(>N^nn5%PDP5P2(!^Ma@3!b9UQ&uJ7?yexxtpz!@hOKV_!f0Lbk
z0{FmrS)EN7M;6u2#tUR&~!)
z$9ZRyNA!%X1ny;QV%(8|;ABRm1u8;Ca^Na{R$_-#k}|-NkjR`h&_D<$%+#
z)~3z_95a_todbHVTCF{2AIK%C9KV3`diAGbt?z)!`WYcO@20S{&R+$X^>bD<+F#46
z=c~|OF0059!+G1Ra?Tjq8}h2_7As(b>!ne&LvPqmZ^ZS5ipjT`=73Q?B8F(MI7GU*
z(2fMI*e{3cb(hsUYLJJFE16i~I>O!JyIawpA6B>iir?+!3*3*=gmmmj!w<+`l6Puk
zpxl;u>5^#I6F9GkqukUsweQGlmP90k~w;~+0r
zrk+JT_Z~W*jrs?4h228?S0kED&o2mxj-+uDmKXOgpxko#Za5MO)@!`4z)`KRU(H@h
z1fS=x*mMk*U%AH-h!{CFm4b473oZ_$eulQQSWzxLk~v>LIc)VslyeoZq~}Bqg}-vZ
z`>MUNX~z7+fwAR?&1XJPyKfVJ4vv0>_}8jGu86fCm!HLUUtqH3H9l9PeP9{d%YZNe
zC$yL0aR(#ty=@+?J%RDFmxjkZQI7*R`ovL=-SXO#b%2pYyXT4ka%5SvVLS0GsY%55
zZ(CTr$OX_?<$abU;G3V<%Xb1!>sQRg`&kREhNFF9{p{`g87S8^_Tc
z=^p+B<2O84FZTx6%O8pw`Pm@nJ78U{3UUQX`DYSA?y9M(!x~tJpTFzL%!)p+ljD+lS;^XL4hC!;`4A@-&DGmtBMx;Hct`~=QN=LS4LPBX7|F`hRwWd1(?J;(*0
zbqjZf`>ydyYQg*E;?S0-S*
zoEzb1f%+;9g%{3){XtVlK66|L|0Fl!c)()tLsZ>HIJLOSeF{`1DRjyrtl|0Qf4px+vt5eIdi}A?2qjo#8_2c+~a_mboz{NfKmF0tWckTKYS)%V|+
zw81|4*$40cEr9(%M{RG2^B49I72V-)Cp+2+&;1vQeOb_NJ*^J7pXenvGi?O@R`-$1
z;yV3XGJA;H!MXRZ!v2}uFhTv)p=KiF6)Py7(@f@sC8j=reKs*4md;%W_Hw7HntQ=?
zGl?xwJ$O#Kh3Gt9oFv@VOya+}FR>DBA(Gpdm`Z@Ko3<9RHcooj0`>#lZfrN|1-q^GrOa1xLkp3c-%xvZQwv!U
zJnTQdw3S3wH#v&EY$4I5PnT9NZzZRDkJjWZZy~P~*DQawyM=rXIB;RTb}NZ*Rcu}h
z_VRc@s$NE4E1AB_4S#fm_aDZ};t1_@taJWyf3TNJ1ice|>syF&T}w+ftWQ1pEb4U{
z;#-9Y*IC%Xp3Mx~%Nqyrh{e-Ce!l(QLLyH$*&0l@5>=D%kwpT~ZxYg8<9dr|^9yqg
zsBdfh47;Nott7dxpArE`+SNE2Z0A1{(P(Z6Udcy@6%KT`#-8y
z>(LGS+pm3}{iX@(`?yBx&uXZTJ5#XZHOR$y7B_4IIb%=rk76)x8o2V$9f9>2U$=Jq
z)gZrYP@&Nqhfn0p);wod9SCsQO
z?9h*IzH!#M6Hq$i!iL3w`$w+y27{ck&9Qt7kPGm9z7~1G^4Tk$y8w;G&%-DEFuxo+
zOuye(ND59d0lBl;L6vtPf3^3PgD(NYLrRSn0v_xN6B>Z>TPbH{-2t<<7>*P`zS_u|
z$s2$pT*CY)mvzQ=vI^!gpF#J>XfH36bvUBE?78=p6YYSYuA(%~+kGFN&PV^srDj@c
zFQkp%4JV+S#>7t_lrxO#ut7VNr+>8XE6C*>RpQ2W`fqjPWt2PgbYnNl{rRlC8s&t4
z7jNzWxfQyZCk-II_FkPD+Wm({>u#W2YtaE~v_I^pmA2yfP4c_6bI{ID2;Di3_Je)V
z`AC#=FMOGc_Oi;lgYIA9{xyT{qbSdiK8bd7EKW)b?PPlG)SuIUGp#?+{T=&rkF;R9
z)~w%j-!R9dRtuKDIAb^$d6w++Sh_zXXTB@l-<$txG7;q)%~>K5OUFXmQ2v?x49Xu&
za{Q^hhFRQeET4G$obHc)eL0^4)5?N(ULv|L>!AB!)*q6j`&f8f)>8k!QX^Lb%Xc)(
znxMUGai94Y+bNQB$O)hKZ(r94#$7y=4fXIl2knD-lK5OjIlMbi?v={9&(45iH*0R`
z0PfA|5WsVc9bZ0rgnHhxpG*0Ou(U@Nw%0+E{2Y`o%IUB>3b-Tq5kKDlplmMu(h1^d
z#tR>yTt`z8{eJJkXOWHmkX{u1TLI;^?XabMNTJt&et%n;JKc}{L^*UY80+<4f13yG
z`)i4+k7_WkW)Dw1z;)P&7h%Tuy_YfP<%?gyuG$(b&&B=CDqA(KuLilPP|GqGNS8@^
zwid$mZl2oON?@lOM7Msx@u0k;qi!GIPa!#_DnQt;cIrIT8#>$kvK{c|Np7D44oC@NfBP!E!3KG(%bZL-HNfDV{>Qh09$B{z
zWMaFU6*{|q1i!l{bW;6g5yUrlzgj#7^FtiV=kVVXFm7`K{?<3O6SmQ^GwVUlX@7^w
z0L(x7d@Fw#!u27-RYS$X9mI83{L_My5JxOk*@1kyZbjKwR>#$*J*VbFI|yzL9YUT<7T!0ChWSCu(NxnMc*QY+lLOq~M=ce0sls`r
zeDk97eK1ey{hskW2G&bH4#b%^0bkg3d_?=(3wUqNa%2kPt6YkQG?3?<=<{|sBHK$8
zE#*Tsocag{+p5}q(>=r}!dRh52>3_NnOud95ZB?GjD8W&LuBivv`QcM5%oW9LN*QE
zB!H{sufrc$_Zb`45Rrv_abFt-W}NFJzxR1fmp<$#GuBqU(QE1d$l7*F}my0?EiYnrx;DmKRkVei)Bsl
zJBMB}_t;;%&)mJF$U0{2Tfhxw#)n!Hdq~OJ;ESQzedI{^gNM=_eZ(d!-n;cCMDywwuB_AH7YftI)5o3kKp|@vy$!=nA<>b*r`rB{b<-F2E
z?%3%_6?(OjgPgHRyWCsJ_5JfCcWwte`?)vMeID$qP${urH{L>)-tm0*)}@vB6?0rS
z^KKGecWxz$0>@XxDYp`?AdfAV_*zLmpTIp!@Xud*Z`kl%4sh@G@#$5qq%eGB
zQ$&9YX?rqKYL?YP_Jk~niaOj%M6Yr?KZO0v@gJR=ud%ce6Z2hscV4%Ui4{kmt-07j
z9$xkxm<8Y6jrW_3D(`D0v-chfVOi5kbid@!UihMgEWhD6cig^(Y+SjvDlM&rSf1FN
zA+@@d7z8M-_E`!0p?9t?odx{m#)s+DOs!V3Nn_u|gr6-$@9HIYMR>kp4KuDQz0D->
zPSF0NE-gfpb4}o!lvYyyhRtOG?593pF`!wqbY
z+lZtP_vGOH7ILS3Xe@O;?8hnT914Q^DrR
zw3OxoUm=G#2WDGVeidO$^LS
zZ+}$mPX#`sZ+LJH7tBxG!i!(8fOUvDt$N2(VO|jIJ>!Aaj|Go;w2bDW7KN3R$Qp+$xlF0P4L^@^;HYnCEvT_N943Jt~X3FQQ*4A#H!C
z6ZCm@fA3cx;9a5qZv|gKzCnxrCJDHIL5EWQ7MP!YTCtqOc+D{}@qycLeW6FzhF2gr
zel*7AHejS;F!C>MH{PgsfH^KwC$<2(=N^O*fb3=ziCP54A4XIT(6fAq&eU1GQjng
zD~&tieE=_ATJSp&%1!ir)_#I?W`j|b3ZP?R_8YVtx$^=faoxf7wfzd*r}Lj!GyRX5
z&a7l*EH9R~pFsZd_1qx*@(1D;t?dQK#{`TNGtU5O|K~i*5%8nL`q${Ui#&3E
z!+MO}=R~6&f9q#qk8%C6kv*&@0FN)8t&RDEyT=|Q4{-P0UV-(#dlp3b%P(nRiXjqCV2|7t!+vH0$o4Kpu7M-n}{4ev>_!v+()D
z1y}Q-94q(cbW8^aN2+5xo+=qt#PyQkmCCtTzTk)k<(-N*uhaMtcWLQ&ls^}+K?d>d
z-HVhj=}z)eewZqlLet?%@wf5%9mdWI_+FfrJ5as-U)-ns(e6%OHI@%#Us#8zdGxjz
zVw=*9|Bz>CY*3wx@r(3wJ#Try@vv00L(nh%*XI3K066;4XD{-*FyXVGQEp~+)#U?#
z-^aAa!vRlL_g{Yr<8}O3U(ten$qg%+Xcve|IapDM`<}7@fn}T
zBv8=+^-zeH+K3q3@kSB#*|@>*kS<{P;&i$XD7@C158JabQLGrp*|df-K#-BzZ7z+XPY@u-D&AZ~RbG)@5YJen7;;{@Y8
zHE2mk9Hfo-2M#`g>(A`b4dqXQeLEekKUYG%zMr0k?1KE2H+tp_g1ko1pF8p}uN`c(
zzf=o6O+nq4_KPH2?h+ig+L!NlRseS7>6}Oc`8mtm8*o2bo%cY7H{=^RtbW1|Xc?^3
zIRz-+6U(^~%2m65kJ^LY9A|clo&de){!B{#EYm?8hE3f0%pu;`s-Pkv-boVtjual%
z8X$*qGd|Wq+-1@DyY9ol(?arg=&%fd|C6>mQ6BhCS#G}j(DN=5?7^1m4CMn~>N9#_
zet&dq_vgE#@STHYE4vxQi(Gpb^3N}W@AW6LW$wd#>L_*Y>m=BnbNuNFQD8ULevyiQ
z!3uHt9@hD`u+OnFtwl}+o@=Xo&VVY+yD{cY$2V63pS9Q#Vg%`nio=ff`#MN-!Z|^^
zYY^v|yT-PBYaj7x`6?E&y_aykT~T<6rI-9zEk0fv(@j#_?OT(oyT}mK9V3?
z6x|5?%Aft0|7-gmVsEo(Q&w3osre*YEMo@y1-8o?HVAi;mx~Q9eOTT{{2%S9E@AH_
zQEo>)Rxj!ykL^R^dZysHBJCS*Z^rLu6I3h`JION-v7pNHy+nN8=g>9Vdx-7%ESqYG
z3(T-u_h}INhx*&LthEoJfA?6me0kALMvr?Q`S!Voq;$AwZ>j4hxreJ}IQMsxKaDZ1
zjqkh3rsO5&1#7y=>Eb`U4#i#MXsgy|sgfRYI%d1a1J=sO#
z?;E~c@v)0MY>D-i^64SmE~YnsE$<-$hLxED_qvG-Yw^C&S>0slfMkFh$Z1FCH|$C8
zA&Km*iG?eB$kjUu24|zX$!!@^5x$`=BJq*m<@fe3qTHnZ;_uELvf;*awi6mX~oh+SCT$p>y)6z_|i?_kL+-HUSU3D0gyURx4>V&QIDO(@KIf^O7YW
zwvjzqyhGZ+Pfi~4sJ^M(Miv*W9G=P6My^<`GBoIICC>w7lOOQ6llGd9Pwqe4$Z@N4
z>$fMik(XK9g(S4v2$!EgTjatv5`6o`gY|I!*nb{^L3h^zW>(yM@V$*(3}gTP*1C;6
zwwd?aQ@o9w3O{}0H0ax@SJIvayfv@Mb?Y$1yLw#ggSFqc5OuQfy1Rc1*|$hrRvh${
zU9_aM+z$BI#;bPw^&#H1*`|Z#SqoX2UjOd(Y2baT^G}VGwvraR^n|Gj_)c$az{+^3
zRN4>rz~5%3+~
z_v62e)LY283q^y4M`6Fu_rc<&$J$A`cSTs!0oYICS9*y5QVZcq-%=5J72NG0T73qtG%_<7WU2GtxE7Hhq&$cC+`hX
zfLAs=yit}n1m{pp{+f|o(@xAxhc}hO{B|m0fbAE=dG-JPZIyy~z;eaAwL8ASd2p9o
zp4tUNeCqs=4-fcht!fWUHo&~4>}xjN3iA(tF{|4QRoJJ~YkEw40_=ZUcm64;hiCK6
zFX|9K&fj$JuNm;kHR}2gJ_3)^blA|O3ghmzwBtNHufaXAa_cwHf9X`8gDT8ZYaAI`>(gy#``)24^(;N_y1BiY{rUtqaYdkmf{!Fb@+YPkM}pG1@x%)711U#*^i
zT;g*MwIV1V3hLSG3wk)6sOJfS=S;WsPFe(Vb40S1*+IRlUA}lo{{VYlZX|&F3Ujg>
zrwZZvrkwadJPgCgT(j0#dt!9m4hDc=TT>0VE6^nB#`?Q^4F+<
zJ4?3#&T1G_#Q502JN||-xW8%bvLuXa)Lgity&W)f^f%q7^X@!bG4j1z|A}@OgItxz
zgKfyajUGRXM;>x?$wC`-IICzogXq&q|MEkA1`ss@6fW}KZ7rOyA=x0C1
z>mxT9Z${o?F}ip;%KduNpMku==%w;C)~FrXBP5_kPd!zMDI9v4U7XCjTq@+`QwM2lvidM-g3rrAJ$L*VZAlH
zd2z^_MZP_!>u|m|3yxxW*{R1ph!?J}rS`r)X2cKcVb9j5-?MrAX1?>f(7{%Na_ZAF
zZX&L~@SO56`QwKvpA^2sM$=-upHp5MsoP9>-v;(Pn!nY8IZrJsjXBp#eEkKQKX*8Y
z%8#67&Kb_)X3iOdbx&F@Y{Q)AAY5ZhxID&NA5lWF0)9yk>%{z?
zRBpks^AuHFd?}B&bjhLd$rtBe>!AFD0W+$nWU@Zx(=yX>{M)HK>({^3Z*a_OzK8Xd2Xs>XjR%VP7L
zbqr5)0L;n=ufuZBtQ_ZKk5v5kMe?uHN`~5KXU(aYWTF<2SvU<#){+&zLYkj!K
z=)Ur?sm3PMN7nxoeg04DnRD0DH45o|Es4uJw&8WY*2ROUhpOGBV?lsQilR=Eh`p@E
zg@BgF4tc5qj-PsKoKyw>eYpPAQZ8wI7g=MQXtU82?_E*I)Pu>D_iY
zbR9Nk(WT|s|G%)pr;LD=p(&9#uD>j28AZM-6O#87`+eSHU%FpTP^W^Pr$~aeT=D&r
zr!}s#0l8Q&(f^=~`N{1>`E{ux)E`LdpO!%XVLI@gD7K4Cyo{k2%zqQx-@U?k?m{kA
z$B!`2^>(t#{DL?@t=Z+5k?@^%POOBfHR${Bbt%gwke|9&Gb;$<2PW3f8Glwzg!4vSC{pWm}7&VkF4q=y6+q3T?M#b
z0(OoRguOHL0(<%SduF}?oZs+kbYk1jE@I1aO+FoX_%tWGBR|A#Z2R2`)PKW#DU{gz
zAr$sGu&k@lJ`8&76?<_2?z@2$EDi^KtzQW$*g@Pl=CL9<0_%CnY(9g@!1uo%mdaCr
zc2n(q$Sw)>p53zY>vf3hMDF$4h5NT27#NE*!*^6Ew>^JsIowM&eQ~vSi0UDm9+?~8
z``$%fuuI18weKQs3vU$M{?kcXnjT4=4d^0kzC=t59_}IOTs-!x1$*Fs9BR82fDiL0
z+&nEB(M2@B3)j560`d316_?(?`mbsB-dtPVF5+#YCNVz0n*{!yUEWgCMRr^DB&mQu
z^-jjyUBw&v<38cJ)-hmLbc{|Megx}`2R+^|?KgbdmDq
zOA0!?x(NG)J;J9OJIVSgiNnpCJ4vXQ;9IVS4r1|ZRri@$o#gd`kcE9vPnA5cjn(c>
za+AEwuNa5(g%`;62<1ZjdiI9khtTePU#&}*h<6Z)PPX6PrvT5@dfu?+ix8<
z@YjBDPFHAH!T2eZ$a>J(P8{^w26iUEzO_mD4~n2?`?Sq&UXKpa&pPvbHt-$oX6xR+
zyP(})?`jhoZzlJ`RV4iJz!*eVF#=n5}_733pGcNxppex-@mR{!%
zdNaG7G#!d4`Wn$jnv$BOzQMlqDf@GmSYUra%bYE5+##--m91{K8O}jV&s$h7a-xj{
z-HX1is@g`lt252?hTF-Z;OlWb>98-2YthlkeQ<8^=-_Nq^>#9IVaArGEA1rpz_wjC
zEP#*Q?b>6u58?uuNp`yd4<)TUU3(q;*!5F;7Ujdbn8HxkkEdYQT#}ExjD`NQzdzd@
z=94h#^X~}sOLqCj@?7Xw7IjCrcOGmb@1}}YM6tAyr`>_J4;HkM?Kg#WpYg(eMK&GL
zq_6Out5V-U3ivl~m9-{vcELWsH2n?7AkKS=9QOzmhWgL)xZisS>!*`T-!BYoC&|eV
z^$u3IlZ`I~ld`UXKdpF9=qQZaw1~)na=5REh9B*wd7HXE5r*zRx}0~!~wr{4KophII)h;
z#k&^3eS>63%87(5ZH0J?LukYmsHfcLa0A9S$ERQF
zXu|yaYR;O5WZ>)i0aIT`LElw=(cnyKCa6ct|HiROZcu^6z!07off0WSclY3Ccyw{at=Sn*B=FiDz)1ocMQP2O+)3
zOYX}QlwZ<*C6Wbcb3@_JmjSDf@Hf>$eLFQ@EggXJKQlD4QXt>G?y$mPzLqe!%LK0adg^+$SXNZ-(;JA0Ln7{n({my+=Jis(JlB2RI-@M3H~bv=#qh2iVXZ
z%Y*hiXXi{YOlP%RwM8sX`3k@MhIZ}TXpMS0uP$~$Jw)a8ZeaQI2=)}@%NK(UXq>KB
zwEsQE+p>x`3m|W2*%&Q~_0?d#!J(C&SRNy=_zafo>qV3zuBv)gh1f7LHxu;;`=#iP
z=@pAtTtXE3^@H-wnQt~xo)&C!R}SmB_mj69^$?sHN#o9O$6O0g-sFnxbwqtLF?#-7
zQXtj7&SdIC0bB|LVd|SSZe(z|{^^x-1u^UyC_vtDh
zEI~QXjU$POx0GJe=j&Ht&QsPtyp^_h$>H09Snj0yE&)+&$jkTxW2}rHV0R~q>Y=uu@fVV-
z8GN6_)T?@lk$-I6LFGdK{r}~J`CmRTeQ43iTRI-zmG7LJAL>x&2Om8r|BZC1#}|IJ?}g)gJ0rQOA2H(UG5ViCZa9S=``gfq
za{63Wc6GEL6m0oL``xN7=N_PZ#f9WFlw%1rqVM~1m=pC&FRV_b>j5dYne@DoU#_k{
zQU0JU^B*a(>UH$Fu8x+`{#N;PDg6%js@9S?l$-mU`A_Pwpg1p
z+61vMaTPr$&wdTBChEfbXwhTg)=FbI$sIaQ#%m{Llt$x3b;r
z0yrN=*3J<{`8mzK!nocy?S5bk+wD;EvwTEf-%b7*;ev}e++X_!`!gG`U(emu-B*A;
z%oSw)s}H<@ohzx(1pKZo7VDz2q27II?Dw<)jlQ2S#Py4H0*zAYU?2U5SEcvE^_&M@
zp5=o4n3=~q(7t8PShu?q^v*fG&0+}f%&m!*29U2jn&@aK{y?*PYF465OK%o6JIPkh+$!S@}K*_L(ae1^d5L;l(|8Rp4e-7em>#+u2>zjVf&Uve@aN9lH`}fS
z?{nVehrX#Ge`njV&5OH$|JgMqi9$V>$3OC#$YPJ74I$5f<_a>~x&U470~z
z_otr4@Pj`ST+;r$6zS2}uP-jURHtCP{dD8Kw;%KvK6lbb8^)n_jgp=cv@7=y&ag|Mr?0uj?i-LF
zJLGwOFW?Tp{yq4Apm|pe?!)?7Z^4e=g_*tB*i
z9|#JQd;|Hdb&})Cz}Ii2NjV5YePIW7z+r)~FY}nYIy?09*#ByZoZ!9(1-iBgLAfMH
zxR4E`1+|kLE(3Di%$Jjd_f+DzuA?5(2^EQ<^6(sDdt4hd08hVen!E$oiM`{V%!hLF
z?Z{Ice}eB~+Ac!*w3OLg!PtP9jchlXf0j|q`W|txd
z`I$RzYefT!jq+Hdz4>p5M;!fzVudT?$cx|foGJ?k%y+bOG5~b2Hj*^~RNs3)<|UxJ
z*AMw&z>b)6^YB^v|FhcXVOdz8Fki$Pv~!Q{MONVTih)}4n64k%N52=f
zDP^VhUA<@)zCSodj8i*Ax?di+i)tBo_E%>-ihu5nKxcfv=hr
z+p8U=*O_oL|H<~9n@su2DY0tG(`qA^QN6V)EGVygBbQ2f+o=e3%F7m=&7dfBc!K&R
zwb_ShocmDUAeAdD+eqs<&B?^`PrrRk^~sEVNXv_Ky{TR^&oF#_;XZfD+r*A8rFxF@
z`_uCB`UezWEa9f*RyJ>FdTrG%TE9VF1WlWS|3}OJ#%IyAUpK?c!$fpw`hvFwE%$LM
zr|HUb%=y#z!hh1VhBqT0pv9oZMFz#i8Gqxc8^e=N{`)_@jx$p(tJK@|T&MQ0KG;Mwg2VAPeG_$+rR5s
z|M?{s)0q6u7Yx6D5X9)MzLCM)6->Ly7pBp6Fnh=3OSm!qO^XKe-C!2ZRJ<3dZ`@jEXuD{Ks-%l`pl&oYNWg}IC#
za`Tvavo16Jr7@|1-uKNiE~>A=fh{z>(bIzNCo&1kq~D!|Uva1Tn~ItL6ont0MbqN-
z8kAR`x^jw+3lZghnx7){mg4;pD8mtX{`YpHFz;79FRFSFcj}6J3sU|5#=?v#w{bllgzHW$Xref44=s{|(+IlSQv5H+-V8{@&&a`W&u-
z<}E0<%I1e2qMvmA0P54U&tMh$Co>L<+`xR>is(v2zNvWXe+q2;-i7)o%(_X}S!C-o
z+v_21^+P5R{ishyAO7O^cl~bdr}6vugAQ7(xxg>-TIH_b`g4=7+mt0>(cERJ7$^GR
zmAa`6@bhl7%sGJn72KCYJ+HT4wZ}M}YyZG1w0G=a<4a!C;`k+IsFOaRtu&s
zZp8W&&S%ZR`c|i=%AlQ|>%S;-9pF`|uT^NTV!nMbz;;$!-m?qs;GXW}A@pm0Kd^H{
zdp$smr|^6q
zUGlE>!}EsD8}%{=9y7yStg;F6^)J|kCxYBa*t@|zz*i5_RGvY-Wt(nwArEqWr@rhN
zTqijFWE8)1@M0b8MEn2WlkJL?fERv$p3e#Rt#~U-7F;*iX0rf3--^IbxoZFgKW`02
z`~Ti@r_(3Ey-V}Q@cKWoGaJ%?@72n`S$iC?D*J6rJ?LMh&L@HIw=0S@0na7TUghx4
z8tQ%5VOr}0yl_U^{Oexu{Kdl)V|tMGS+;v+2cX_-p%wUDQ?Tl-%vXTC*Tbx(0q;Nf
zIvaVzW{>7Jd=Glv#w*I9J~3fgtvr|0NW0qCs6N7oO{+#F!X
z0`Qw+api9yUo&-GAJ^p)QtGPMAisa~jUJwBvM$hYO%-74)}XlokgsCgcw+?G_u`|g
zoxC6)t+2j&5cE7NFV%wQBYbbZ*DC~a?`&ULiNbXQ^Eg-E1k_KgqyJ}J_t~1Bmlb`=
zi5J`BvzkJG5MYDy$RFfusncycvHhR3q$T3`tKDr^j=YC^&jOQ8fP->o3&a6=6fC@E
z14{k3Sq<-3@O*q1_EYWh83(TdmKspP2)dEi>BT$l7ag2mR!Do?Yd#o
z`LGwD+2>XK*p3nBe9q$e7dfh9^&*fyxiD_86rj6~$-R|;5{B)9#(;aio>YqhlzkXk
z83PzOrk5T8Xkc^Y<^#Z2`o`kV04s;3_u;%E92iK?joWgvc_0V$GP1*^ya)+KJ;k_)TA<+UgMdS`?%jV37~^E?g3qnpxkLu}HOsOv4%|sx`W<Y>pr}f@@3rt
zL&{@LNjXwJ9+G>6+QsI%b0}|5Jv~WzhW2aWyI9XW-^DaN_F{zEeF3rAln3OPGW*K?
zjK6xKoYq#(=ZLLYt7&_c^D}&9js5~!Kg7#vdFRP(RFBDj*Bj3XpnRdz+n>s{jO){K
z74fSyU&t_yqOAEsTK+qbiQ|7$3Z{Bl`!PJKBb0-d7d_6R^7(=c4{cv6LGw*!np3P?
z#PBxTEevmbG|K!pW=@_MeSW7mK~yfZ|0%7favsCWgxrT{x}YPHmIrTTu3r_x@Vf9x
zCSHC{gWskV`H)zp2{-G{GZ(RzgAlRO>L&!M2O+Df7Jf*
z>qS3lx#kn5{y<;m`V*I#@*i1D8XB3(75?)FU@*~i{lEPApP6**zqG?YJn>IH{{&Mm
z{14$})9c^=!+Zbm(m(nArj{H`rXi?JamGJ%
ze$J$ih%jhb!eFTegBi+YRQ~%vd2e?ne@czP9a9Xl?qb?U@}Hjr&N#hJ>z_X7beQzF
zfBuBizkCRuP`Rjo&l!G>Ngw5AaOb~r;eYbC{F@8g4F5?fJmw7?!`TOrVik>m`yTk5n9Cz{Onf^6U
z#MCRgoGJer#^3@u1{a-TVW
zllD_IyWURysG*~sfvAUj{zsaxbX=8=lT+^OG%c3He3v4~VN2I}R@pQE1uczX{4@0*
zj2~68kDJQBI2S;#msrpE%W8+2@$q|3DJ`#Czmb+(US-z(#8{c}-n@^&nQNGK?)x~4
z%2gbGOi{poA^`Oa@fW6g^hRgZVOo8+k}1x!Av+D}chTb7N^~Bx=4W|~<;s7R$`J=m
z8-oz7{oUPBzWkXkeSd{2sTBDuv*|kLl%*Y=kNdlSPN4kv54-8Snfxq;>OWgAlFBza
z{-X7}KG+eBasf+9sh%@cf~kB+O9z$n3u5*owhjc*dH+n2W((TA-QF3ekcW2YU$~0=
z!Rce`A>?8GS&6O4cl_SjT|_<4UcF7f!%>Y5n~UkdoR#O${@fhMOvZH^`-onCOn2n(
z+K+nV9Jk$r>HG4<@+cpDX0Hm~udsZT7nTR>_6;LuSDV`*z7A%qL;1gdOaG#syv2Q*
z+B=r=Y3e`iU06Z)Q7+TJqk{6SVgBwY|G+PBIuMZKk-NPd;L-#BNs5391fJ3vB
z?U0`huyz|EfADzt))DnxdhW#yMhbezfK3
z7k+m@({sf-=vNNb*{;Uxtu0PG#rIPX)O82>;lzBqSBSrSJ4Nt$^S3WwiF(Suwm63E
z5%+eVD?X3JxvC6&zsH8Y&OuE0RB#K=0n*c5!1Wuj_EES7`jvlVJ(oNL)cmc+odwvm
zpJf5+Ip~nQ*ci+6UdLg(q5dWdeSV&+D-qk0V=o;C?BM^|WeOP4^V1&neU!}WLi@8BFV)k)
z`R+pOKiXFE4-pSm7OP@=^93fnME%NytmybsXAzjj^lk4Wx~Ol-&^vdmCuOXm8J{m~
zo$FbAzMYjfrm&qqPmb>13E0l@pbh=9wI1uP`2ao|ADn@_oK^0@%p-u8>!Rs8`FN$e
z2>J(Yi!Mc22EG=UFKs`QeixXV=erZHx9yJ}!*baj
zeBUvS5n{i1E56?~zHW6`uZF+(JZuL$$sk3P=N&E0K!3ziF7XtW$E5je!g`(LSm?Z1
zvU!~DgXzgsS%mqa(S^>4twqs!c>S&J3v|A@cs-BCy`t9mabvkvN0ekOFxASQWm0RTU
zn)3GMN-H|=7sZ=W3|ePQ=lv6U_SC+}O?{{Nsya3f_Pcp*`j-@cX*LZM(%DZc4Q+%}IHr0clZ6>{LMIOVy7o;#e
zMjx%`r
zIpgQN@L==`xya~myqMt?>R*|7ZQ6`fdR+rMlQ!pJc-5E=!-t~(?eB4l>!)%uUm3m;
zqQ#V39ANlU?H`6W)s{27PvQf^%kHjY@&|u2xU!1jU7i!n^Q>}W_)b7AQy%u0!4z3W
z?h|3kZD%v-_w$+it>k0m1pn!`OpM9DHNoWj9AI$8KOPG|Tc!Gkgfr+a#2~*TgI@o5
zEsrUaHVbF&Cx3`Z>$x$g{Vz^`_#YqMb#|$(zi}V=n3me&^{$89dU&n#R8SA>&7M6O
zy#&sE*ypF2`m~39v$%ewMZK4t%;+@l7U&@}n+-GP_H~ncy*_ys_1%P}cBk~sgl-Z!
z%(q18NjF&*GTU$K|0+83xSGB$j2ER6(IicpXpT@R($XM>6r!StNJymlmrzMXgNh~+
zA(bXk8fegDs5DS%x>LiQ?-5B8?|T0_pL_2N4utMAfY>;d}m4bEZzO?F3{Vm2wuxdjEXRopfJ8hU~|Bb9$eU
z56Mh@8k>Rn8SSbL4VbT->8h#5`bh`>!BZvWF}>(lBpGlzlMF-ixjX*e!RK)LjqSe*
zao^kU!EQE?4u^dzKZn?iz{AMW`rb(7QBi%~UbRPmX#U-SlmE}p-DH)%tQz-^2_f(9
za2iPDg^LQIzjM!`;Fz$1QP{u^`W=JZuHVf2(ovXiYZbnyItuxhfesp5xu{>9N+@kM
zMxEP?^=aKY^p#P2J3f8t{TPfSy$!U;qJhq?T5j9`6$B5G?keO{LE!5X^C4#{i2mN4
zappS(hQhm8HE(FJbX{+}n+|d<$HqVO1yX^JLH0cU2f3Qh4eDA&kwf8@xV3~v0>c>9
zmIfCx%=hl?{4zvA{c7aTCJF_X&D;&^M*qD*zT6kv2KxT*mo)L@=OBFA>m?MaOaXVt
zHFS#z3Rv~(?BO7vV)p&(W;<3N#N?G;cYQesifXZI+<*d}&((F#4NzcYi%!1YA`0{d
ztUhp1hzgr_g+=p`Dc~R=*R`_~efEU?j^@{r!6okcl98^QPp)^N%u~5Xs>6Pbp9!
zVx;_*NFEusrzFHU%Wko@nh1r@%fzwVv-QC~$VN&D%Vm43~@UU-QYM
zKw-e{Qmae~D9-;-)$pYNXQ=qPE0#O@tlLNW|6J75)cLY;61a;!G1V0#!KI602gvAq
zp6Otn*eT
zLD=A*+p}-T&?rU3r3{mxH>&)PP2UJuZ#!8kkM(wlU-tU4j)eKT=j@!&2k~;&-ORtz
zm;w>MsjUwEa)y2<-f)>3fmM2jvPj^9%4l7KE5PL#b=QQb*~KU95%CPDNKnx6~IR3*VfPvtYh1tf5Oy}4wMCK=xTV0XPk
z{=Av4)~*@m2>=
z43J?e=Y+hG90fG?-;v|$ppU!mUjfT^$hDUtf34<`A@ypxQNm#g#KtJAC^X_aa2v9W
zL5|-HjW_5>!t)mv#6E~n;11LYKrlE3bn0|l}b#|tTI-PJZ5e
z`c%k&FP+FR4{un5NV7Y!$ns`v_)p9K6Ii^Mo$(Bk$NcU^ZnP`lPxi`86Z2r)rVb(|Z?bV6rKwM<)b%hgAzc
zHOM_woK&B?uZ!HlME7U>wy014o8d=z333;g!8N*>Qmi%eE&SZ`Q~5?d(-%_tSt@sWj7!
z@$w~HxwR;f29oRO>xV+IodkiI<@
zKXE3~Bi90XNzompC$PNU;W-<7S3JMq7PkQFE$K0m3JIe@bIR-lk&A_;M%5fBn)|u^y#^rG~_jw3z^lm?LuAr
z%Kar)cr1IOA`
zJ9NGyKdIAsd)zb)q(dJ2FU2@RXX(1L7@uSOJ8Q>G8Wi3%%j&A5K?R?Vs?-qb&5zm(
zN^0Zxu8Yhqt&z+0F0`i`_3VPl6AuIM`mU6Us%?ElgR-KGSR0(TY!0XR?l^KOO{B`?
z|52eoKi+K>a-YnKTwh7=qruyR_Z6mTR5-xBbxr086{uP{QwhkODr{`|TfwBl9je)^
z2F}0NmQhDEXR&_=x7_hJ8igp;O4HBCB@p`Fw=+y;42nN)eYN!v>hjw*WgjJBTsOXQ
zGtl2wStWOGP3$P_tx_+2Jv|B~lr8Zio5vu?y;@s08+leMa!j+o;`J%HE9`AD3f$3<
zQv**%q1gB3>TkoNurY{##&{16h`9LH1b-T+=h+1NHPT>}q?5))#q)G82L*T^#QPty-Nt=cW55*p
zP{%M`X_2>o~yn85eeB(dlM`Wfg+`Tdi
z_RC%whwzO-Xc_To)?y4EteCADM!jart#`h?HzpwArfG(jF&*57#M!i+AylO)77+60!_a;jhpIi35OmxLO+VHaYJw}*+9Mji!
zFIArb!ZG^M^b-uAO&>|DsA9lCby2GA0P-}v#a|sqZi(;{e}Ab3I&@4MJ7!Dr>yrFG2oU&
z8vJL5I>pcL!%ZHeZmT*=#P{n2xSAdmyl6{@Duo5f=P!8%}$U){b>0E
z<8$|2kQ*-PuIT-90+jVM#Wr83gM$Y*`qwKu7!5a1)uMhgS?7S!W?d%uWy`cbf6Rn{
z?ni-<`2S+R`mopeGc1TmH*IXk{|g019;QZoOfcK}@pDNQ6I$eTLh@I!z*90<{Dct;
z)I;4vozVstwFqQ!xhu?}v(**z9m>Ps3+A`it^h!)|CMe4gsYOgCXd2$rzv9IL
z6U*IGW$5d;ZH)CdSb+_CSyJK8uOpX5NYFj16UWW-rbz86)bk3rwe8@szNFc#q<)5M5-$vAYJ^%hD&vG>BJ%q#1HK>AZ=tr(bt#l4Oou8{B-lG
z*K7!duF=r7=(lchvR=oZ1LAh}m+Y|oDyMHpm6&V@k(lP6(dEEFfimx#9vsMfy;ol3
z7zgTYn^bJ=k+bM<*5u$$4m5mZ>b5=L!0^&CiPi-U=o#J+Fro{=cyFq
z*?JDB$XI6V!}8(|5y8>u6MrntUG{P#2e`KEhz;mRPY8OSemTp56XnwE1S#}?5dO3G
z3)W+EiN7YOhyx#6>HME}b7AeJ`tM~mTzHzs`{7Z*h32){Ru}Pnn|Vpsb^|Uf*?C0$
z{9P^#8QdH{|A7l}qFDts$gAAvS6I=n$c1Bvy;QE5bKwE~U#j7KE>IYkjGcSApd3J4
zFmLp)vy9QyhZ`+MAATRp~E+S;?B
zbQc#EkIhuyYv+PWd}V4C`eCIythhH|%>xgKPmMz8qcs#!p}1O>2Q{A0WYQ$?vzb(h
zOfdS^SKYBH$mhZB$I0v7^>abt-hQP4J{~kK%nR7=<3ZC&Z8_UK9+00W=aBf(zf@lH
zwb%$3Bt1kYR!?&w{mmbNdPg3xCjV##1o5Cl$*#}g77wnt7>DY1aN)C<*?M~_j$6W+
z-}zkpJTf`)$pU@9CbzX2+(3V?h3D~F*p9dmb!hVr9<1MXwN_)E3!8mF`L0XAgO!V)#_T6Rd7?e*4%U~T);F}bZqa#CrpCfVhY!dp4Z9UfV;XaP1$y`NN9}gBUw8Lv@0*-Fbv$98j*M{1V
z-rfZ4SGI3o8v2=eN!W{V%CY{r$jCx*^e6ck9Ui}pfLN!NmSe62Xjym0hTR~5zTUOt
zGPWPcWDOM@BS78O$0zwb0YPu8MeQ>PD9E4N_f-LpLwC=gcO;M-yvwrwB>|RK7aY4I
z2{^N2km)T%fMAnDQkWb8;>nVWgdPx3ytIV!x{!c;9{KREIPAAWUh14W0V45vLSL|+
z1K;b)Hy$Uzu$`yEa3)}zr2j;;2LXKne#N|91Z*GASPV`CC_dD2^Kd6X@`8Kx25dKA
z@Ip$?M~v@6M{>e_0&X-Wf8g#Q;8zQ4soM$y%<8MVPOssCn!LVIixdy^RAMVo{0;R3
zrC+ug5->^4{&+}-fR|_Zo9h92WxEERF7M>R&Xiy^nNR}s1Uuc`ya}iclXzjRLV*5{
zx`^L6FU~jrYmG?a!d9b>sVR(e@3)z~a~R*U&@{jNB?JicZ|v6A;lW(fde4R~F8u7$
z414w(uV1zKGxjY4-faBqbr$2>yOPho8t0#YoMTCJ5f{X6k3W8!!$odmnv8%d7y6sU
zlMQ{5FJBn0M8SQcJuX4~i4q%r|6nZ(W^iH6ZlxYkESG8|-8+K&a>ee$=DJmE(DAX9
zUVD-a`H7j=9kIToc8yKpxPGNOo=%E3vO%Rmc)b2R3vAZQ&KeFeA$%&R+&_^ECq{b&
z*8j%+aG|NO_7d*zGYy4uRZMsq=0ARChz`B4oSz=D=fS>I?T=}HIS~9)aC^Tq8*1&`
QA!3vP*@IfKo-X6?KZHC&+yDRo

diff --git a/previews/PR1624/tutorials/out/solution_000060.vtu b/previews/PR1624/tutorials/out/solution_000060.vtu
index 8b7fa15aacc06086097d49e59ff1e4c669fd8a5b..8c98f272b6f674ce70415c78c1b33b1bfc4bbe50 100644
GIT binary patch
delta 12218
zcmV;rFGbM5l>xw&0kEb6lMDnclhXq@0yj954FqHXGqZ98*cN|$3;+NC0001ZoE27i
zJl5G4zGEqaq|m|?O)60$#+IC0R1B%?6(!!Rg=A!lmr59lR|?sR$x@2m=naJ$m8C*4
zMJuATD!&xz{NHhLWh4;-wEw;-LeNl^`Kt~{p@V5-Yi9E9p621Hje-3=}#+~Oz3w5@y3@p
zx?J=29m4TP%Ic|+mpB;6*hAhnaqz9l=Fio4Igo2=R&(v*K+-bgmEuzle8Q~~bcQ(4
z6wX;}`GJF(P3GN3V;p>vUl!r@3xi8tt)ekG22ZA2>->MB%D^OQUWJbVgI&4FN=iHq
zmhakSH5SN$s_Z2BzxQyk`H43B;>SU<$&h-PHwU35`b>*Q)kSE#ABH=;!_~b?`Aw058-pL)J5OLI+6qK$ap&V?t
zm8@MK#6jcu{`6CLFO!F1CcYaPn6(T<56)+xuFT(SKZU`pO`C*^zH<;Oj8Rd-JUALG
zP1gFsL2+@v;t@#(2F-7sH%T#&kSW(o5ocf%|4@Ho?pF>rHP4Ep&e>+x|P_WvS9FE`b{ool6YY^IU?nhhd3PLBQcNnz%NocRpaqk9uLfd#Z
z-d}bgRHx21G626nkso;#yoAumOocBHNT{NFwN0Kgp%+BbQBk&pUP^!JsE+HWo0V3j
z@d#Bo=$U&soKQ89{+g&1LN%U^hg;+l8j^o)C|^-bXo)BwTX2@pxO4Fn-~S=BFT(7<
zs&qmJ_o}}!!|(69jeCu<2(8Fm^EF67sB0zP=1wM|2a}%c{<+^V)1l-1WkT-{owMJ3
zgHX}Og!YtX{9fF2#
z_8Q)mn8zF^xj4*6Xx{q^Mh6I87Z`u{e7!fuDY2T!+eqjQ^VpiSwS@ZFZTW}0LM;jmroEd@sKku0
zAn$h!T4c?wN-r=_JvTm64{`QhMbDYzGazm!{H=-^{4O!-djoaG-%CAWiUNP33!Ep5
zS!Xba9^2_@q|d;<+BP(@hl3)Y_NJ6{4o)1tBO`%)-IeJ#ID$Oh2FEK0;yLhW$;t6-
z;9yqe##&EB2Gz!|!}IqusGq+vZq<7RD=qoHD@zzaRntA~6%6!g#mi5RIS@$As2xXs
z@4P@hHe=i>3$ni33wgHUo?CxA^4D>la&}5R2Nkg$1;Nu8$QF)^^@T8S&OM^gGt6LK
z87NsTCv>U4gk>eZD=1?;L2o*Pt@FS2&!|D&`V>=jGlqkY-)x@kz_{LrQi?Y0I^i^p
z=csQB4ks>l%4TqWwcxjWHN@FvfoL}BKuF={cgAJdXDUkaRFQvAQWr&?;>v`cJ;J{i
zi*=XPkR~=SIQZPUzW&Xl5I~PQmyN&QB({VIPZd|4qvO8V6lPat>QC
z@1187vu@5oUaz^;7#BlmpsI%TnQPdG8zaA)wqPCZdNY650z$(-TNGO$U$+%(BMWXY
z7_PKxiO2ijlgrL_#=fLyuwhY6FbAzvT^RL@gP!4nRN*P)i`!GX6_}SJik5x1a|n&D
zug{OixCZ{Y5C5%2{azK=2%&_|vb#K9g}R_obA2>iz@YTsD8X~=@58x8gPyxM*j>5f
z@6iOr%^rX088ylbqFllRj-w2GjtKY7^CeVfV5`X{)YbI0t8OG=Uhb%k%Wum^9is;v
zcdRCKQ(|CSBHVF6zl6gD`CY2Ya_D=!#+9zuA8>EkmCEs_e|)7{y@Zl9WIL
z`;qm?@wxX9->sL@G|~_!x~dJm>X@GsR!jUI{6&B0isr4UjTpB?!+*~RjxQF?pB9(P
zAbD~~ljd>;2iIxa1b3s(91$J%#q)~pZ5;^5_ngn>?YM<}E0}ssN$Ma6zxTZY9@eFz
z|55faSid~NVi~4*oTlyePN*T=p22zE~LS^rB
z5Ipz(SRdBq@N?glZiug>PUEgjJSS^R7_on?ii65&-Rl|@Fus=Xm;>go^Tg)s6WFg7
z3dT~jupe3Gy$`nIBmP`|OD{9SKICW>e0&IbtFrP^z5|1a28~@=kFY+aabuc?@!rL~
zy6aH~Q+6g<=OFK|h>-6r>S{KcF)TidLqNCUcFaAC$QXc!?wOfDf
z-B+=n@7uBR+;Y?xwE>~}i!tQk<4ca|WYpT=H!B*b$YEG51xWe}QL$oC7}
zgg#f&xJ(CiZD>opJZ}r?7eBo(pq+n%ReDb(qQg;NiD%&s%-i;V`JHx%JGa0zcRTbs
zSsotJ+x^ieq|NCuMqlw-ZmN~#Pd{KyLx=t)^xLxB{)6Z%J{L!o+o6vUwYEm3y~e)8
zbF?luVPGwE6Y3)pvKDk(_^ozGo=Q
zagidPjI_6>bRWKZ{*o!1Ff@_2e{De9R+!ySDUQ#?0NHPfh;fw2F>m3guZo|ei-
zHX*Kg2H}Md5ziIt)Os|ruYBL)Z!9urAnD|!Qz}O2vCuq)G!Nv>QT~*_P`|wVHmoT@
zJe*P$oJqWbJbkFLeN8l>i57pEj>{LIkLzh(E`E@Kug&hK0a*VObL-ZVX;`1N*%{u*
zpHY=I(dl9iZn4_afh!rj3+d){h@<{JUYM=rgZKPV@znwSN|u@Ium!$n`S8(=?Kp2&
zy6tR?{{?j5&FC-aCrf!dGSMe=vWS9hIt(1uUqxO$$-$agiBU!Po_v49n-do2I5?|g
zF7S0^@Z;2D?QyKnOGVMk*8M;EQ82p@^&oDQN>V2JEpNSJMK`hjZ}J_a{xc+0K5ClP
zPV^%>E`Ih4P#@-%yR|MhX5g*)#KbF$1DPCMULu|&GynRW4VO7EdtcNo3PRi)ULI{l
z-}-!X#ce0V^^IPu7cCr!XUhi*5Ae{hY2S^ON=9E*4O6_QOeOUH2K{1_(J~o-;R*l%
z00000c$^)Uc|6zo6UV=D7m{O15usEhgu+JT$bEgk6>XLxawbaUs6@3(s6=vv)}5rt
zRZ01lqn2Eiqr$coxhv=Pd(H2!=VRV8@A-V*^PZXaoZsQHWnp$y-_`KMnPko}ea4)8
zvn|({e)hvdZVbpt)#a9>9Ao-_2M2EnMS)A-yt`up_PLI(`NSC0fBN?>-7Daw3dIR;
za3ohFV+F`{`d-;~u-N>anLq3U3#C>Z1KFY+f@iV*PoI$81CaSsys91KEAVge1j7}^
zQ^tRe>i@`RtV{;u#5OM54la&w&TRT`R6pp*V&@RZ1~HS7b&$7h8Vz%QUl`S|^Y@W`
z3I9`DJ7jI)=LF-up#vy5u$oy0O7!Rorh*^CEeH01m8XJPhM;2FwadcbUAvt4jqq#H
zz&g$cx-`c8sto_~iIGZ8^P~DcT&C64u)8dHd4(0&x>Eh8KAqno-G%tzj-iSeC~p*=
z@)7$dLV7ehz{tKX}b+l_b
z7!tiHgAG=3W(D2^P0!uCs7UwmVO-!U{F-Xzw?2YD|I_g~w%dO^
zPII6vczF2hoAaP%x?D>?DU*&c6
zwHvUjF&h@&3%%xl7WL4TpnFQq%P!dMeyq(XB>Qc7^-S3D>BK191tlZq`HD#X+n6-~
z-X5B)`2gOm8&f+3d#jq1V>a;r_Vc5|laO7!(@ZF?O96N&l%}vOyvf^0{&8kv-to{7yrPTaIRKWeMzmt*v=)Yr}3=gG~1z*8Mg#
zC+mTqyx;gpW4+O~Br*uH>VX74%I6FDkD|6>-LYIb&=+j;mP}p(UR~cFVGez@ldwb}
z<&_Utf;uRom$vFD^8QS&!w$O7pedJFsyDIzR0YZ(Ew%F#-LOmdnuvQ%`Q@BGyawxD
z8=4gWmBxSxq-s^1if0_?D$TY#ymCkGP
zA*T$?Xq#_YhW*!``>9gBM)PGE=Yq-^?|vBrd2S>WThjRtu(JcfM+am|T|mR&h^|B6
zUxzxY7>JXk;o_U2;H1Kp8LA%%p8EDKa8g6{Fy*s{o^X}tK)K}bW-f>Xh34Looaklt#
zVHd@hQP?X-b(RqB^Jyu?-^xRe^2+CgOWX^@*FK|7#Del7JJVPj-<|vISH5Z3o$W2k
zX~nvK=dvz#73BNAm2DJ<|B9Z9RTC|a$!dW7!BaJxq2F;VtfdEZPVMdR0=pchg@f=N
zqj+ue34AA~WcA+NSYM@L8>WKqu;h4e*HOJ3%~qm58K*g4<_!7LsYdrhJ=Ogyt
znS&*>;}^)K
zY{w5_kfT$XRlkGnW45elu(P+ZxRCC%!`_tYt6jJ>#1?Yj8>2a@!|d`!+~>hP>+S{D
zQ2wee`?nl+g8XR@DZh&MnNLuDSRP&eh~lbw*T{s2;%3RQBNXJ9&dH+uaNe4>D;E2I
zrDoPF-$ws8zW;CPgJ*|>tK2~I4u+pH^|8R?qPoz(b#IiSesCmzhuAMzABw!6MEUtC
zV*hwN*e+BiMez=gEi$1v=j5tSQa_3_Ejv|0{jK*bqZmwFq&GtE)b}M-Unw7oI(BV%
zi}jCwLjgu$p|2VjC)T%$DuvR0UYb3B!Tk_)=n55>raE+vR8gmQPV{KWD(c%Cs=iZS
zi(>W-Q2*T}&*$_L)brG>)Pmk_pPRWjxKk(d0@d;6ug^Q_o#20SIW&*@dA#IrWbb=H
zc?qq1Z;ku52=*x}Uldav8}>vc(fRt^^KYbrij)2N)Q1N8E$*m*vga=4Qapx#dB1QE
z)BCB>vVDf^))?5Yqy8heaFOcPxAWE->VvG8hLe00pATZQ^N`)oHx##l{a*1lq%Rdn
zIL;4!U5-sG)rr_2x*T-wD~AsnC_v8rA-h$Q?#FN+=U#fJYwaxP{Rk13E2ZJmM@eCCF`uyssHep%(+m1y@p3RJfrv@6@Qgo54#6>-TTjh4NEL{QhyJ({Csci
z|KIJNHhQ1!6ttdDzq;t7c)bO7Cp|m_=-%rUzs@$%`@$8*M|o!4o4I!#*m66(g#J(1
zcYOubsaKl)kqolWh{*f_YNTbQE{9(3OMyB4r)7=j5*AT_^vP)ItB+Tzg>UM(|^Q;+0UE4<)YD#&S`e
z^qS_rg8h2l3%!&Nxeq^A(Hx}45Y``|d8EIakLHqPC+_iSLN6%wsge5T3AUA40Q@zW
zY|y$6j%p`G9R^Ra>+~$a?rnL#cA)Z&MXIa8<-Xp_&tcyeRej4_GzYj|lcTv%*|ga`
z861DNR{A^jf5vcs&p)6Gdt%Up-j8rqD{R
zPMth12nzAAMw!J1#^2EW2UIi267GzfDxH}Ooy`}iBHUDw(7%B5CAyZ8fWeJrOw{A^#^E>m`_jAs3p7T8KdCv0Q3#xEC$IoS{
z2Tbn`5#q8cxvN$8K)%R{JrKmlWjVWTj7|lQZ@s{ObObM`ebfDYW`xBH^_J_N9$~%E
z@Ebe|Iq}cFK@+S;CqJyt2aj-W#>arK+rQA>RRno+nOiiQ{H_%2
zGy$W06D0=VU)ju*vl8^~9Fr5nez8Dpv3&R+yQchPJ@jW7L9$Vh%g;-BYD|r=wq?k)
z(RW(QhoyglpRC$W>om~j6w{sVT@iS8Hl4G7@OmX*HMq9v_2X{XM-9|Vtb~8rld-MA
zkbR98>oy`T+rRc&FM@2VI3{-!a!!$5jSA!k-8cQoZ=2_Wp(S8vl)3VD=yQdd*&Onx
zDdHIoea}YYlrhLL$<~iD;6MIVZG9^AzLySn*+Db;euxZ=Z-d=VEz3EXV2x+&f@PrAgI7A_U)S!bz+F)A?0H>z
z=q&_n*#e-TvrxSVsG_oCE9K$g`9uGIm_p9#UD~q`_U_{`W23e~bVt
zj60I){bD{GvRMHh7%@z62i6fy=8mu;%~s_u}*J%{2;!3@=%f_n5_;%W@KCtphyM;a`S
zws!Kwx(4%$q6WBt|RvI
zB<%lIFPxYKd7{JNDCO_UwKnp9EKugI4W|<8$L;n>mSg{`G|@fRK#io+mp(%;8FRwQ
z8g^@3KL4evN+4=>F#1OAp)I`~O~zg)PhKzZDMus`RnFy+<#
zJ`+RaDX;F@d`&QSUfC7Or(UBcg;H2<|84!k=fZrv#Jg<$2uf!>}usi(5#&p^J0GQr3nnk$tD6
z4|01w?}s^k=S#VZS=is78vj@j`X+Yw_&L~}^O>bsMRoC;VCWX;kGd+IUIu#s>s6O2
zegXywKUJZhEbH9B0n5H?i~kD-#d5#6VSUW?-d@Uw=?TU2Oz7EvrE_a3?^?7xl9|x|
zOlU8^0Y+x}R4ak{cM7eyz<+2w#RVYqnfO^Y
z^b76v0=Iwx-;HE{)WIpS`||WYL)p{ssBWaP<$USBzonZjrhCpXze~1bfA`&I0bi+~
zt=982q`tG2Z{d6}F`3`R1^R~OxCgr_jviJPl;0i8NB^Hwr>>mLrG0wOv@&TvRhch@|zWXIWgZSISbnkg=)SN?f&N;Jxw-A_OeUyL2k)4S&=EI%a%
z`Lg4R(#(3wp0X
zvxlpvULvoJv&uE;-U}jJRIgLt%4akvnm0_gg(
zZvn-5YIVMqC8%v4!dTqJEU=D@=23QWPb=8Z4xKAN@7Z9L<6{Pfw`dL1yoy-HCTQ5*k}*vvt>hN3Vok{WxpxKYuV97^%QrnE4`!ChfnVx=r97u
z2VbR9T#smJ^->-V&Rh)`1R45QRDOm2HqSzv=2&tVqnYaAwChvV%jgR$ZarEdNc|@#
zwDb(#^*51cJ)f}dSU1YLK=rE=o<{S(aaKYp#kJyXlCTdbuah!#hx*j6ndMXu!G(i=
zk9qKOh>=iw5Pnt-{~@W1K7FY%`g|zZd_y2x6@0ogvYg)GXaV$0#2=uliFRZ*5}-hGG5qi(vUFnlB|@UHjpWdEQr_OaFTnsnrsH^#7y9
z@uRw4XDjOD2A&Z4Ybu}m;@((wXPO6-N~vEF=X)7LHFUn(r=k1w4mP~|1N`tCXEmqp
z0R2ytXPX>>-4myfTf4yGl|}<2c$aDc&lbwZPZqtw9gt*J$j^AMdlE
zynkM7(e4HM&1LF^Q$Mll8(m0$d0;j;rVaZz-r!AXyk}KrmzgA((HgBz|FxxFCCN1B
z^pef558(fNP29&X%{XV>ncsI(-fJv&RB55P)1cZUNdM2iw{hvuTe#2PV*
z7f^c^VF7BUDG0Ds$e$@5XM|i-xe{SlVxY@B%L`^jlg-
z_q!swvxoA*9G(8F|rMp{@P|a;ROmtT^~)JS3Rr*7AzV
z?TcVPtXt1E!aevsMBnZQ0RIDS8plkN(I*&xs|)}D00000c$^iNd0Y?MAI8y!Hj1<>
zDp!^^R4R|8vZS(BsAzF*NmNovn-ncXl08IpyG7QBC|X2HMJXx1vwUYp<+hOMcYc4J
z*LUVS=bSm8&+~qsITE0z+gi?|yF&Edn`Yscu6g#{7Z!QP<)ljU#6wDxjPP#9gw;rTtEu!(S$A^+fUgURS;G3;~u{7pA^Qv7~MB|vr4#RDf2>G(szv{k=5Ou1q
zch)x%`aRx1YGg&&obJSBiA8v_$64e5*T3?oHoxoiYyK@l_R@uK
zKgdxq);VLDu1CRkC}LK#6NR0oUyu1;qcC0n$wn@a!YcoZwMWTyk~R8tZyR(bHULWEZC1Q8uBG;}EawJ%fg={73(<<}`wC-AjyG
zO5=9B{*uVmG=Ao0bc}PRaj#r|C~h&M5xsL==GEU6u1!1bn0}K&RHgi`Mb0cXZYf$m
zqmsc|iM!R3M;LVEEuHNX#^CyxF;bh4GFb1Daku|I1H-TXihKlef9rILQe)Ak8<=}}
zHVa?75KW88ELs#SjZb}IkguyODeuj|GPO^59V>>3$MX&Ahgp<`ch9tcsbZlhXSdS*
zEepNu`Xm==F+%@5OEUBjV-CB(!}^XGy^@+aJ0v)0Y#j5aVJ-)JobG$xhQs;Wku|xp
z9E_}&J%5oPMr+38d&xx{()vxcoGmz9tTOgIF^)rUq0J^WBMuTy{&O`Ca;TT=E$ySo
zJgb`&KCj?$|LeXx%d&ZYM4cP-9`&Awzx*#BOEHg|c_H^*>UmtsD9Inam&b`ri+YU?
z9;fW%KEFK3LuK*L9|3+mQqv~6G$-)bwY11`Q!S5b+bcJ3O%^beK3b;2OF$~ET>T_Q
zK-o(SW~T^f8Ld3Y^O%5Kx2&@8;{rPTTNHy22)O#*?^1NSfVO@D
zm}|Kuu{m5oIlb*|6L~)8vD=IMGyz({<09&U1SICX{Z(y2)(Oqm&>Z5?6tYT5qK${6
zUetf}eLN0@8%xUD3%H)FKKghy57lcstgGe-DD1Y1YI6}Ft!-}`yg-1FS&VF)gn;4D
zO{Ret$a4j)mu%I4czjb;REk>orPGPh0UmS)ms*N)K$iaP;
zp5q*%YrlwCrPLBJx_isR`pN&lHx;~Am*>!2C?}=yj~H#V$j5jXt2
zXxm2yd(8&q4n;7S`?Be^_D%+s`so#1H3KW3%zmjd7V=Vm-noWk^w!sys&I>e^`P2e
zvpY0Aj#;s|?hW&(hO8Gw0aq1MBN|$$8l6i&V7bcHKQJC<4P2i+`
z6gp}G#!e@C)?B+_Z0&3Yqg9e-K5}G{={bLd({3?;*8a8`?VZj+SKB7gJe)zxgzzfc
z%`~Kv3bv>xQs7-1`xIp;Y>rV#J|hxg$?5tV&kZQVDyt=osH8AhXVtLsF^v=6$4tJn
z43ZQjH@%%DhHLVw9f^J%JROSPxQFoY)$=qnXkj27vwg@lg9hd9I9}&8g}rVecPk|c
z*H*QEG%sL8m>tYqYqz7IT-e0;E7I6gyL;C8c?{muDygSWvG^W-a(V0@Fg2TZeJ|g=$hDzo|FweG<
zzDoLW#G2fOdQ%2h#O*zUe~GbJsUcE-bu%htdNp(MN)+4O8qOA6F>P!rXfFKJn-=6$^sTtnJ*4iz(a{;xrZ
zQN0-UD))DNImbiz)m#!z=6y&{nbKe_K=Qu!_XxrVr;`R@d!LA**yB@_ev;^aE@$-~
z`#K6E&C^B7-6A;uh%=R$P2u#Zc{Z|QrEcJ=rox;
z3LCYrZfo4cKxT=L_)?-6PSezXLZ1ZlaN2FH+C}>0*2t^cX9+L$6CXxqws6q)H;?^6
zxHB&LnO71S3wj+?XB9Ar&p+v^bel%v$X-Pw!k?zl??btSKT07^+%=LferKur=#X`f
zHd#KMM(R(GFx{O>)-RWjJ)*yh#pW)9|m!w>$N_)HA*
zjw*?dX)Ko5dn&n+e(h`4nEKnBaI4$lZPp12ug4$DR#&3%v)!0UB)X2u>C$c)N1?Xo
zqQ${u6pF7OXFAu@*z(2h_zL2``oBhMIg|HlmfoL3C-WHiqfj3n!=bHXf>(u|7_p}`
z|FsxoAS!**sb5W_#=3HUYs@|w?xZeO^r49BB!?XM`Cq0M$tR_o%fl|GQCNF2
zjSVI|rG>!B8=?g`Yws;^Bje(f!eEB@QTW3#nV+P1JP~Z}mz0X}-n}HsU?Yn?>5+e5
zks&(K_g8%_kUIVB98+OLc=xa;Qm2!wbE;e4gygK75?kyh5N`i}l9j8sB^(;4`E3%o
zOu*DDosX1>e>PlsT+CStn4?y|vv&=T{bea)Evvl5
zQK)lyI&BrvOHJ{e$u~ZW@GBN`bEZ@9TCTr&(qS<&{Zi_}hB?F`_Uv;d0sSKL(QU+c
zH%@CD_qV%%$Pl}K#3PYBM(IeIJksM3zo)J`{y2-%9nLAY#}a<3w|-mrmcl?#&u}5B
zM`xq8d<5~KK8s1!x}**n7lK~O$cj-qf9#^MH#pe0ooqZw@`g#wJ|Q_tz{t+rWkEi~
z*E?D@qM~?6$qb6$D00Zlp1n%bm&JAux2f(mM3?i+eHAl*C_FtHan@rCf|Evteo9A(H=m&92)JAF>hTBm4@<`&b!WH<>k0z>GDudG{oExM@_&=#hTY
zT{KIl%b&rNy^lvVd?NmA#lG?*dd=RmPI(gPv!rj{ha^HNlpWG8RcWGe@x>R*{6{SA
zpNoDle4WGX=-LDC#|c;(^kw}kQlG6`#tfPc^4R)+CF2bn&0%52VhcKzg{Q*fg!6<4
z8|JOE_ab@0Km#6ZeKWwOIdrEjO>!B-0c<@W(+2MVJKac0iaDVR7
zSV&4~dCY}^lf8LV0r{?&sK_f+lKPY^(Kp>g`rtsdyxR@ZZ%aRDy8k5kQZFg#bR)@u
zVJ=bgg9z{EF14I+itxbQ!)CrV;m-d68U^a73LqdnFF|l{Wo~3;WJFxt%0kEb60W^~h1T2%&12_UWF_R4hWC1j@as=2Ge}4=B000000C=1g
zR(U*^*%!W{GNgzWRLrCjB_dmLu4oadtSyvXQcD#($FH0ce3ilTlzn8Nmcgf5%XKST7>L%^D?7F`kXY;gR_-MO-$3(d%{L4*
z1hW^e{mfuyt!an;2!nUhOM*PdaVTzY6pl>h&@t6QbDR=~e`O)Rm+aK#;E@qGZJIj+
zM|>G9m^4B9+B+i{gp8Vf^~HU2%A+y=Fb0~Nbv9cBGmuO_*wl82fv45@
ziuHaB8pe(!U&Q~?eiC5ly^(`aL*LQOUEr&+C;tHJ*cf{eZ(JaSnqL#oFh|L*ge_&rPKU1uSLCBy0$Y3
z=Nmb2nO^G;tsIuSZ7!Wy&taug{n^UDIkd0JNbuRmp*ckM>l7_QpUzM&)0#o3d(-0^
zU5FQ(fAX-p@JWO&Gv6H-h|gPoy6e4HCG>vL&!&XsgpN*a(`{Hs=wtn}Uez{)HoM>X
zblsLvl}fANz4-j2^x)h5iwHfODmx6mgo-+*p`N8-md&Yzo{W95=hydkskUu@U&lP&6xnRK
zL#XgebZdM)q1*FnFBvuAJ|J|8bTgr%G4|SBjfB24Z2Rnpdrhx77RPZtwP^mu3wXX{
ze`|=pIjwSo2~jCzt21_pD4t0JN>`SxQ6dj$KNW{uEhMkMo-$75c;=5
zrdNFd=AUf3N57cREzvI@nAH+`V*0k|R?Mfq)+BCXIp+WHEm>BL{m9AOH0~y$A4F|b
zy-Eq)oo+ul0_$-w^V1dmW5`e6h_3Zse}wwvn}5&TNNBZbczNP#LPKpf%>In~D+@FB
z=Y|mKT==E)Ci1&TeS!Kr>|0-`+7gougeLl$UVCgwXk%u=9z6p>)w6VEKFVVqrv>x@s4whwBhk`p%FLt)p#wRoQ5Y;RxhJ1}n-PJ#cJhp|irG1eM>>AS3-KrVP
zFWp$-Cdc80!MnfyP_yD`G#69?b5Vcsk9kssG;?`tgQu#}d({`#E3e(~uQW5{o-
zE96T(#w|7@>pMIdI4!^L)QbGIe_o@I7GK36JG?DxzYK@(*<&I-{u~@KPRVw@NB$MU
zH1nl|F47TOTZ(!4r;J5wPvx*`?ti`0%NacV8hY(+DC*vSmakkeuE&Y^+zkg9tX3gm
z&iM>3DJ=Qc%ur{>pLDkI=J4`OlU(?74qjR|&+^q!he2Bqfc>%C>AF@fe~QDxil2HB
zsDHQrzNFiBg@GD8uyR7Z(y)JB^-D+M8z52w|9kO6?Yo2Z3`&*kb4HeNSTo=za#YB{
zsn*lhUWw4xp*;!Ak!gABC(
zH{7m@_3kU3>v%4KfzIC9f1-l7m``#4VBk&;<|ld$Cvy(*LizeN$Xj)lm2#rL{i`U=Uke`O^sFUz|U_vnP$gUjiA`
zF4Q-bsF;QJX&fq7Ce6B_jC19Bl29IXV1Le*j|PQ=vXcBvCFE0Yf2r_+r~;uEPleqN
z$G-Cg62!8b!AQHUZroAS>o13e**h5+?#&i8%3-i=rl`^SKRBqBKac5e!nz*{vKpp@
z&Yz#Uc=QyZr`^44tC2s2kq}#jJPz%5pFk1cxm2vt^)d%*%ly`FsK*YThgFVXy=RF8
z^rT~arHzHUIw>nxc8VZR(hT0kV|+)47ZaOsjvaTNCGK;J!TsFHwp+38
z)t6$<--Dydprx<%;an&2Ns4TLJScyY%pe?r5*ndO-wU$wNGMtf4Ks_-yD!+H^jmFl;H61-h@i`Z8O}2x|+G#
zvXaovF}_W~s6Rfr8W#ezI81-48=#TIAYiMkmI&5;ob6YmLgcBWfpry3f3pee9^8)2MN`!f8Ul+gK-N~efACF+O=S=Ohg8U
zl!^YeYD+nUuFx^IyO+x-^81*N|JoZ&9-dE*!uIA4bgwo3+nzG-KXUi+Jit*gJI1e#<
zNe+*{f2yd7qfhF65bJ6f!eITXI6ouI&(`Jq^PC_C78M^$Y6XMJFfkAsJ-jg#T2-<#)e?gH<$C_$cjIS;@V~h3cIJf2Ycbr#>
zNh9&XQIN$8d_n&>kVW#4Ws~2n$XVt2^&p$b?}TwtVKHVULf+>=sVUe`o%lR57?)n6RS6?RiQtqAIUw9bMUV}CIn04{JVLt
zf3@Q#&hto@6-7%?UyS+$D%~Tf>l1HF#X4~~csWL)@7XWDFAIB*{rEUx;7A+#LE)|*
zA8niuS>?-atwTJ#5>eZG;|TJR{3FvLAhe<^%p)K1{4D&&Qvu%FvR3TJv=l;fa>91`
zZbDv;Hz?FZT^rpRDeb-${Zd$R&)!y?e>d7M#Eu4HeBzelf^~EFH|&8m;?Bo6(b*b(
zPO7V`#10=qA0*E1G(cZ5FnNmkTEz7W3wr3yzl0LajNaquE57E16kDT@DQj#DNqon_
z#NE!K*pNezz)^U;l)>iSS1pb>m+Z5;-zOlx7g<&LJj1&7C20qips#M-eQEpzfAl>A
z=j|5A;XPZL9yI-jK4M>I$0d*3gpPN8d2cc5*}Cdx9YrHTr}`Q``h+xz*(?~_np
z3#Ef=5!X(-fjLhQ&o|a6cdFrB`MK4{KxoQgg1xy*I;rwBDcI
z*oym>X>|?;_&%QYeHa=?Xm)|SODg(=M?5HNyC#QCDsO{t{)s+$W=u#ff98{+clWzl
z5raHU(j%%wAiEn7&nq6yKjw~pO|IprcpUnwKS9b%Y6_wM2O>aTsgut#
z7=Pgk00000004NL6_$5A*J~8Ve?(Mt2@MT|l9A|2p?=67A^lV+Aq^u^RvAe&6-iV^
zMm8CdRY}OIl#t@exDAw9nb-ZCzuvFsob#OT^F7ZQPl%Sy*4e2croHBw^U?E$nT%=i
zMpKSq=KZIRb_}d3fhX!?o!c=qoXD
zcVupeIhoE_6%B?8ZsgYh7lt(^H2gQj^gOlD${TC7Yhyu`Sa01lQeph?>HWC*!lGImp|)mQ@HY>rmy5q3`+|_3j4?&Usj9g5t4vBKW`(vy`xn@M}_M
z;lmBu*I(N%3;!!4gQOZ}hL}zqx@Bds3+IhkX$-zur8vg~tIgwVh!uI7a)RVqT8_Zq_kYRC=95^LTId&iCV746&=~kZ!P<_2{XhLT$|`_o
zN-rmgfr~?oL|;SyKs~nS1X!Q1+qQ@7cS~{Z1+87;X>A+l+lUiy>%_h%@W1^-*`52a
ztJWJ7vW9-kRz=@cp#9zQ*MDuWvwW`1$OMIUq*b$EC#-T!q5%9mV1_#jT>3`aq8EHP
zFjn3TW>pR=IKp1P{H~V?{P%o+c5;mTIK}EvTwiWcWbwlP-G34+h9RGqjaA(>
zqM+vrPG`|M)pTT*C14#svij$K@OyfzCdDmBDYYaQ_C~w79xB6be}CPYPkj_$nZ{@}
zu-mEKc{Svvrny0h?NHps1tWZhlBk50>@7lAR-?*j~=e`d+I%pK%*
z;Rsix|5A$;sY2eLPc_$~`#e2v6H4_a+!G@~`Lk8w(HMIRr%@qzNiDsFHc(k$J$#r+{hM8FzHSy
zr2lN5Y2wHJ%-R$3RIkC@iQ1{4NPO4QVe)rBJjV!ni^DZZ?o=m-*W}xPh8_WJj^I_t
z)=~!I^i5-7vM)Fzach$5hj)JU`!=v+v;0ZQXUnjn-AADBGk;mHt_}IMH6GHUxHr2+
znhQQ)oqSU|e-pTGY&KjJTv?cD`5gNMR!5F}!oKbqhmi)zFWQ-Ov4O+BAJr+&rdMb8
zQhcW~I>o5YZumQY`wO%*KB7i><$T8G#w*0vAijBtA^cS)C1@+-yAubtb5Fo7q%$k!
zE#&U}Z8fD>KYw#6X{I;`W>*Q8(ft^CiEaj$c#N0-4ZXgXUsDHoAf~hBIM``E!FL+p
zVTLIijN&^Fa@Sh#gM5XYsh=FaqbJ7ouoC-~J(HxUPu@_PDYODlxh*?H_q%WEh4UBe
zubIr8#|gXZZI53WLN4LmY+gWd*!9Yk;-O!azcLDTyMNa;JMqCT>ciTWCdlOvn(pbR
zd#u^h?T7V+7*^>aux;4XB82)wXMIj4^jlkY>r#Dvxjxm@UVJ1;PDz&LibpEReY7nS~?e8_Ft
zEAtWZ@e>1XJHZqe1&(=;?-G#mrTe@-b%ygP$bV|{<(U9St%BqfX>JL4=8DkVlPUd8
zeeELapI+*}_Tt=@bJXvSsg`V^K5)>^KnOHbNw`dPEc>IjmF9$OdxUQqoj+{(0kZeF
zEW3#0F57PWVu$_hRj+cWjx{@iqv-!WI;7u^p?Hn;FsTpq^cd#J(Y;^1nM&~(Wb<_u~UTcnV(>-PzptJ=sxEL1Z<%RzI^gY&E*kF78>r1=rXCzenDf4%9i
zA?kZp!(00)5ArtbilhF+tut*y^%@vt{(q9P)cSK=b*be=ePG
zk&ARO)#;(w-KXNoe|$i~A5bYaF=hqylKmM5bWVfvWhQja3Go{O6;u~;S;ch!qJKg?
zk!vUqO+_rozieZ85cQF+%Eu0L{+LMKC3;}5@G-7kRKJt2&Qc%a^Hz(rpgzdMf5zqAF51CoX%2xhMq_j2Y;FK6jOfp3mdyWK^@6iUk>1)dF!=*+Lz``
z!yT0ibgm+XjV1MC5t|aa->UCRo`2B&n+gl5(|sw99Xdww)9#eqNps@9p~duk_^X~&
zFS>v}b84x&S1@?5PWSgw@aWsA|GuKn*B-5So&q*ywtagK`LwP0LVfbb@?Jsys_omN
zXx_D`ofw@5Iy9?Ko`>JkVIzTJ`LVg^dyXO~|Ku5xi&agp9r@=T*__?vHgM~vQu
z>vbC)qN#6mtzZ2coSGZx_CaeKs|OqZax@
zEFKFHdS|8?`4(e6u5DFF@3V-P0*_f(vjr##od;D5_C%flA6);Xmk)h0@8XU=#5Lgi
zc;$J-S+w=w78lUv=S=Uy%8gdFG8d)wA^82=$Amx}3>+5T^pY2}`t&4gYO@&-Fkg
zsPWmL*cnX8&{m3set+>#!5a#69`SeGocPXxV}EQzzyg!gE%ZJX-eYoe6nbCIzMwSf
zyIGkm2KEowip|k|Wi(cf%7W`xW(9JAqgQ72GAaLR6!y?u`?a>8YcrTQJHSQn3A-?L
z$4aWtgdjD-3fBu{L*wTz0QqwrcMSp$a#25JAt826u@1Ldf
zwJL}}-k~jiuu$OpQb0i;BdOZ?*5d8KudmsaPPleHv_hH$&4IvQ@FZYqga5{iT)DTLnNx_^B?Ywfky-p|@=c<%X?
zI$oT{=dARZIv6O-=alnTu{
zwoeUm{#85B8v;4@NoTJi)+6H|Rb+wzyuz3$uq7r>Ne%uklpc>Mni%Blzi!N`6y$T(
zO>asM7vgiq0vS(tLKe%>*_;hKUhA~VFfh8y(``1WvolIg3x3SEY@IHGbEFRhj2Gct
zS5^}*MDHpZL
zL5^d}>?Zn7L%Go8FYuFE)oPJM{?D+T=-ws1=l`H{=6~jv2~^N|b?=_G!~RTn4P!a{
zGfzjq@Q3Vaps4)-aoIi8VX*+RrNWqOA>`y+R+TD{E87bXli$}a^ZFKoZ6}SDcSD~k
zT<^vMBh*D)xX^cQGDsMMd^X3Y!-#x)<{FM6Of(7YR{d4-BiYvfgVuT*?xK&^tQv3YLR4b9p5eqy7E5^
zeZ{_kP0yG$;Gqz)fhol2P{4qc0QOCYbdT?X-C+&08L42kOZ2?Op!&nNTI65H>bYP!
zD0=>qwjA`vf}7n0K~X#58WE7CvS$b7q3@i&*MA!!XLl_6a0vF!<55QLkRQvuSoHwR
zy%E_H4i*};#MAq6KOeJP1|A#KyWj-wE_tSG4&EKD-|#o~d#sy&A`bj^?bwL`=x@~A
zk^hW%SLDBOdH{XUuCp5>pid8ecl~d$){=kpB&gfg5TOA!^Ixr315dKAh0;Czqc-oQ
zcYh3bUGk3J&uF`?P!#-BRIRSj1`QLkn7xS0UREZ{39^JsUrauj#M&#Q1N#{96->J4
z_xgK0Y49f3V!I30_1XOjY9O0q<=OTSBCWvt3#5-*x#5W>X`$o#zo%v
z3jN%uQ|1=1)3pCOVhue%C`*v??dtrfbypzo@9gBQh3sv|J+>A6BI9qO4!Q?*c~PF}
zHkNRF!H#=*|1$;W#Gmo*f?th^mH9uxZQ=`B_Cfz_ILRdi{sv2J4oHGaE|<$w9)Fj6
zT`QjsKCJCD)I*+jR_Dx72XD?Q$fkVi*MF8Jg>}{2lE2O1?{K;cgN^qqYdrf$0PIR+
z&#t9+a?qA$QXcKG>6SKteL`pRrexTsE&M9rP45@{)LkC(c)%84b)0Y7lBYUI{(Amh
zNOd;E_$9m=JmQsDF97>Dvwm}W;D0~$fKuTH*f-g0zNdUCsv
zO~`_2cW-w>eqA#>VhsC5Z)C4t!Tz4an5RO}*SWQiUxZz}yM)3$aOE1Ipl#3}b5M$0
z4Erhbl{YDVeD;N(Ea(OAw3_n3-0zy=KS1ATe!nBue>(hqkn&;lm%=4B^nY&oGpi`?
z8a15b+0c(&XfD13o=A1CPy$u%W?5{b`{(Hz-iO_il!(d!Fq+ZjGn?*VThAPU{PxY*
zI(ko$A%V|S_q)R8eB1*D$INY21kDyDWXuH%hTp5&fy0Vz=P4hO78FRX#eVzkm#@yF
z{Cz2Y-VOSB);hl1z>x3yvwu}VVewiydY`XpQy-~r7-_N|bl>I4hKh905%zb-{|fNaZiQg#EAJAo9iwwnjehLD4q7iwHlq0I4fm&0Kd%0R
zU3VTFJtcgM?k_RR=sNlFzptY0N&RWFg(mg4lGq#@s@ptgi}01;J<();Dauy^+e`HB
zH{_R`5rdqsx@VB$e}67xyDP=fbg1KZ7S`9t9+XW7M;MFQbU&kybsLl^A0;;DQhi5i
zIefkiejba%0Rwk5)~-L0;=!DORU@FAuk83;->v(khqSYRJ3wgN$`4-}^P)
zGic6mwb-H5*MF^upnYL=vFTKQy`NTpBKszT?=c?WGpz-WXuZ_)g$U(|p5y564(K24
z%#v>eGxblZ(0_R{x?x@2;I5-P>}J6(r}71h>U@pehe6u!bhxK#nEFU#nPEG4_j#QO
z)qg+JRFV4g&=!8o2I`;tmH8C+47XcO3e<Tb~oCwt!}Qd|QyG&(2`JAeP_(+keh%~n}T^K#fkljc}_D6^jGA;RG~>n8ew
zY|-OoLZI}up!{=q*Ht1fKYYRZmg-T?W!ML4g(cDae=2bypW<5H7B}6U{>!uy`tE{1
z_WfQ$^?%@#)%$c9e!Q6%^1b1QHSm|DHu`kt1Ma0D@bw+RG#1#nH=>x{A>cayhYb8a
zueihYqxrPn!?gj7Y8dUMeju7`X~qKet*XK)PSq!4J)*JxvH!g@_0!zuQ1NBpzlUzf
zS^sCR?n3phAi2l?73ev2x|{w>^0bnrDIc}_{D19N(Y!L)Q)W;868U#&u04NTagpXn
znXlk0`ajrSX0@0Pn)DrS$iO@}s&+)_JpLy$RWy>~AWzHqX-@gheYo(W9Q_CT-dCnR
zAy@6JO?BFn(4<1~NGX@9I!<}m`sZP)W2YuJ-%zUSSN?L7WOv*3y%+p#yyT(F2j7<_
zR(~)+Zw=l7s_QkIMYlVG!6N@mW`b=8qgCzbyC;+qhY;uDl)g$jU-3&{Exm);aBcTA
zs#o(F)kjG#PBRRK-G}XgMf<>;%k{el@h*P}x-?Ke_Ly||w?LLs9WLQwy{^D#DHr>0
zjdz+*-oMB*X?6wOX0mm{Kmqg4(fQyLqkrBpP1q+7_ZCXyJs+gLH>fdrgJQjBSfwvATBQF|CHC3NM4zy~Q5eHkw1N~`k
zttc(qy#V&#wQJn;agQM<(Y8I{{{;c`6vCVc
z{rhWu?*81f&)R!E>${$_KUB-G?Fox7RpPgPG>c8yy2lDWv$&k4B(3?LMc?Y0Z1xQc
zdBv800}ZcOJijO!(OJ!6Ro+;x@F|P4Nmr!J$@7P#?oqlVK3nqU-vqL*eQ30ndlie_
zE&<}=6BKsj4`{!>PT|rM*X~=-DI9VcQb~QH5W)IZntZ1)-!L*=U!KPG-#5l2YSYkv
z8v6L$WE!s5?`G%ErE$%>rTo4xjpZd_uh+$Y)3B=Wt({X!BgK59>3T|x+ln^7>%WQ-
zms{UA<%<{t-f!$g*4
z=hHaTZtN1Xl*aeMoDQ``G@d>Y%HCOj(n#F2BJbQ^3Kva|IA>p`u)SJ&^V~%&TtZ4+
zCsi}Bm%3Fmd_RMlqWNY4I~f$n%S*30z`)Nv=hpW-3{1POiUR}&_3I{5lopE@hGB(g
z%vh{-h}5+n%c509G%fcdgByk#!4sEy840!wRHT}EO90l++vcVOJ;Qa+Z@)fY3)~IqQb%#QuwF5s`8qkV*(04>Wn#RMq<
z+;)G9uoFBouC<-A*W#gnprxuApT?*E?fqy{Wcj-zN`qF-Xg{CBKGu9cea-G5a
zeywE7n?&aa+d_>9A7-TcJ6k$a_^*1H@y=`q7n7UW_Fb?gVkT&UREG^D4g=qYJb6?ORs_7Sxw`Mw8(Kw4+YDl=7pmcQ+VT%
zIIWTFyRA(8<3*Q$7>U;Bi=4(&cx1n>P3jWq|GV{;o$F|i`{Z4rI_ey|Ff;NRGk6gP&WZ$l`r{)iaQBZrk
zENsjU3au|g6(*8?R$6YSP;1738I?NuzB7vipIJlPwn*TA{?}H{FPpRnp7^tJ3hV-7@9)Oa{&L=!_$|q|c)dyT%VlQ1vG{
zZM`)QjsM1%Rq6|1jCagFw3$V|w&o+LKEkWt-;!b~DY&)imCaJ7@Oq-{rrTtH?Q^e=
zkN2jKJJ|bX58>ZJ`}*95%M7-PMp*`1OVId1#-QxNdA(h(Y6obC`iZ7J~X0vuH1UQU1!OALeXz?1puuYZ4aN`qv+H^yjx%JKMlL(BTLvw#W>iOQrTa_@9VaLY9w
zzSH}ZdbBy@MwSv0QWZx=!+_**?0eW}zzeW>2xE`LcbK64+qyjx|C!lhjUF*PekNGRno%hDRU&&WhRk!RPO3MSMVrQvMD9g2Uq)$XF%KbimtD#`3C@%~S3CZUMVYHg!q@9G8Vt4{f9XOYrN=_cp42t2
zc5kL^5d|;(bL*S@Nq@QoNKU0l;A5hHwf$iP5C1JTn%zVv_lBO+KSp?Im~t;B?;VFp
z!B+7<33o>At?*4H@k*bQ=9Fs;_7@-aP`^RyJG4)AD&bG{_OE{m34i1w-M9;~`t
zGpd5jzmDfp?=xA59DUS1|F73po$-JDXvFk5)#V=|yclutw6;2h&+XHg6w=qL&Ufp-
zQ={;(_oVfngA{5mW-*;Bi4XehkhPG^UxVL6^%jwHiDh=3r_*?J52!Rm#c`XfgVN;rG3H^!h-jNQ3C#^%Ilos$Z27)iK){Z&z^!JcqP
z>eXNKumu8!ob9}?LFVV(vk%HRkpOG0hE09Tc%)UPA6+j`xaH-cyfmFf`NvmD-8u~1
zl@?T=;wZd#|JP&*>6eu-esl&6s2fE(fL8?;PZ?3-QMus|)xf
zwvzim=B~F%liFG@0ZEa64k`O%c&Hgjo8LF$kiNCPW?vTZw;hYpZzwRx)^7VUr;b8@
zc<*2dsmI$U8|7#+hx)C@)EJUF6rBivCaWmHt62(j6)tgb{BXGGF!38x;&upWsRD*}
z7A^=65RlfLq@h=@>-R{mD6TRbOTvz^!6O@^^!)(tS5ngsyP%Y_CJ}SPyFdJ
z%HFqu_&ArE)nkXsFo2(xaYQ)_**dBDd^Cp&Qrp}X
zy0y?y+%@ZR%^4Pdlch_o?V>p7rfN@KPV)#64GAhC=Mhc4Xg+180Mli)MYp9$pXt=d
z8WFu2&7ES<9n4_Xwg!(n|%>P`t{1z6&hoR&eFd4B}#3la3@j!(dcFx8Bafp
zitn>1E!cZ^@FIshduw;URTHov{PW5eq&@*5^8FV5JT`uR&Uwx5<*+bko;97pB1q*y
z(s9BApP4HheTiR~bUn7Cb~%OSnaL4<$@f71r|{4VdnhE&>{ALE#z1|A%tl6Dg2X_b
zk%~ijEbx8m6zVGA)SJ1Ri;0gr?_sXsn!%&DWmVOOe>*7LaaMs$NtNZ^u_qku^JUHido}BZ2@LjF7*B(E6uk~JQ@5@okrRHM7(!xCW
zpO6sG7@i6Apa1?h%Ka})j-QJD@&C$Qxc}hg(kL!X{YQTy|LVDIck^-E5%53P{mt!Clt+*Li=T%(
zUmv$!9-9J!{oLq$Cqm)B`L5luE5JW+OF*E%+t!t!1+(P&kADBNzPsmkroP~R)W??9
zJQmXKpx(cJ1$ZXmItO_AxbaBLCja$$v_*KdN5KmH>z=8{{XGBCFF#X%Yry|$?YzR?
zJ-0SqApdK9$1&@K{q72I^ZOq^j4`)U9EBh=67F!&ZsLEuyo?}$VR`;hey&~ppJs&h
z1aQtk`S|!~147*P_<6Ytz|&ZWdjI~L$Q7gV=;=l`Q642)#7|qo{p||thg%d4H!~(4
z9)Azte+}?h{@1_$pZLeWbGM+BX>G}WYZQ0GzfoY(KmEoMMsW#4{_Fq$@2ziR71Qbe
z)%wn+YX8UeMRW0QWUSF|HHX!FGv<>`db#}1|3>j3^YF0A7p~laODA#V`CMLs%S&UgM#h6da4XRZz97t3*kUmuRd48HS4rk=7kQ|_n`69YMBo)7Ob?f=ocMa$<2O3-@j
zo3)fTxF${c)?oZ#EW&#vf$~CHxWFu8X{$(i&ehA5ub6H@dEVWrFaZ|1Gj0XtHxKL3
z^5CRhluu3HP0L4Ik||9p#DTNP?Olgp=dp<5;Rlr8(eMBU$|4nKSuij*k$>Gv^|Rvq
zXgm4q+?aNjP8GoN7)4*$p==_qM`-DNnpe_*>_Mzys4{SjxV
z{M$0zNGx)?alru6PlOq-dYlt27f#WE1!I#mt)=imSj5NekPPzsR{mOtwAAS6WTf6t
z4}QY(dzY`m0f3wKJEamFL~2!EAp!hPnKhO!KbVD{s0x7Zvm!&Mds|4T#4mvw*pUNxufZ&
zlgO*FHf`hsdgAOyXXK?E)*Q!rohkweyugnqhk%D;rl}{ZsarV{kkc
zOSY~?c}@PD{Wy+=_d4hN1bM-6$%tW~g!RoG%RgD{OhNrEH+*`Bkfy!06#)6Mq%30`
z$6tj^HNP{gzkgY`^T)odBr!}>lh34R&;b7jyrIc*dl(Qe>Qa;$MTQn4%@Ka$nmiv
z?LZ~m{>aq>&GS%E`vG+7oOCN(kBM7czG442BM$2Y1MPkjxHui?P~;iUPe2_-bB*wO
zEz%pjhwGMGwKfsk%{P_1gS0B?Y!BAA>poM8{44oaI@tg06*cBaU+h@-1nIswQ7{;6
zVv~~Kf_l~0?!+K{ZxXx(>7DiVd$66E)wQcouXVEe1*9CU6cwZf&r51(z4_-Cquy!B
z%^f)Y(~}3sAisLgLO6hI@=E9xSWNJ*8;bQsfQm0{YQ%a&J2S2OfI6(JjKz8j9@T-5
zW|OdMg42+$ayUo#|CxJQ;3-&S+>bHP6qv
zwL$86P=xmTE=6QF?pw!8Urcd6X^mMUsPFMWRSM@D9WB;|`)TQ!2h<+rm0hRQc785n
zU&=%&^_jbg+LcxLx52})h)2v*YX4YuAF2IvKK_o{uZU1rs^8Omemss(r2jq`bQW1?
zWAcmEyP-$*3o-%PLV5VX3~=>qvZm)bU5{5U&eHOCP9C&g+?Df`e;s+E3CF3oeg-TTl*e9br{mfAl<^<;{1{pneS*q8lraQk
zlijIcmVw^*4CWNt;du$&0>AWGA?0ftA}NiF1~JGR&%-hS&HB#xt8I3$tiXR8$)<9N
zya!Z&2+cpptM8#BWfRAYZ++PQB;l1*-}F0U?>n|gP<@kpX1ujwO#izGh(Z7N6(B$b
zaJbNpIIJe}0~eY3jKO9=F4@hLUcT^87cps
zOKrLH4dBiToDMBF;ObxC^8Q@>xsp5Yb6j~mmmcKmZ*D>Xi=2dAjiSf2@-0e5r9mw?9us_^O4~0hpXKA$8+_A@EAjT8)!!?BF&W_<+fwT{XXsW%s8qLcEb1?*$iDl10yy$
zNhic2FTpBddqPPJeIU=!#v(@loE*bXdd%?C6>vZ-k~)Qv2cKc&LtYG@oxzm*AHo5#
z2)`vGXVDJ;xX6s*g%p{3nSdg&|53v5zhanj-!05OT>vNs%Wn-ZyrLRYzVI%?M~}w^
z035XpKLm>`Uf4tFRbfVM<-+LC0&jrjC!26#*`&(Sn9}ucnR<&BG4^Xvm%;txV%2H+
zAyy8pw>g9<&o~C=4cajq$Bf&|U=q{b??LDQ<_q2u+jn$l^tMMcejvZ%0Mq~0om8*i
z72GZ83uxd%vWey1yOfvCbEf>e$Zs$Ku)mXW$3T0Umb83PL?Gq0^sZ9gO=&w=1Qz+V
z4h;pwyO&(p!o
zmW@qPB@8Dc|M^@i^}iA>lc;`wRxZ{5Y+&b%<%KV+Vi6aXz1&IHN9_4U+Rvez`)L1L
z6Q3MKy@f-%Uy+6tNVg!JanBY^2AizkYz#-AMW(CGJ%@aqPv2d{6|Rz>L;wa}g
zv4|D|`e`tB$99l=ed=EN2x5H0u-wbV?3~sBUBj$)Nsmog+id3H@)@8L-|ReO;ut*gH}A877cRQ
zIf_*{?*95he;uGls=g}Mf&A={%=i;PFJHKoy#eGl_m3nx0hR1KRrLYnyKhGnw*t*6
zPnqx<Oe0C-YCZPY|nnJf%=0^v$ta0@n)um)M;o(%j>i)#yKzIFByLYI&p{E
z(k`IqCSM!A^#|gIk%zpMFy7hkGM}b_ylLEK!z!TJimxQ``yB5wz2X71drVVzA&j^9
zdWg;+pvzkm6$?NfF|EVZ9_S{kyN0-)&S!P%a2(d`3PGegnikR+H~EUGPDR|DI(3E^
z>Rnh_DTDjUZ-whJVW4xx+}D@HnrC1(Z_qA{oe%PqXJ7L2fTlUism=s>u1$2!8la_X
zC*R8edUfC2Qwl&^zJw+o0KMf(<6huCvT#~0i|6R|>j`3r^LJW}LPiMuTDyfY_MaIz
z#Dn|q=G(e!r*<2^tEkI{9oR1?zF`-x?lkuD%`h>y%a({ZcY4^g|C>S#*GRrM*Ap%n$RT$i_y`h~+g(y3hHgEiG}
zED)g7WjcHYXxDMf9B|CAe{%Lyy$kWyly)2h0{scizXFgB?3Jz>^(SBKFrmjLt}+N1
z*hH)z!dobRzOQXI@_owgFOaW|jHB!7Vz(UtJ>>nSVKN-@FqL0uyCQiA$k{~V00JjA
zajHS!3j13TSmYJJ%>mzh6qX6-;YjEfh%{k;Uelyh-v|K~xJgZlw$
zwc-V>uOSXB@MJwq3#idq=oa*!V{<^uv3ZdHdJN$@@D_#O)`9-+VC19&ngO2k76{OJ
z1qf8Ze$QmeZT>QNuE&6Bf5ef|%dvV$^%GYzc~He-Mqjs{!BZ168Jr~$0%-_ncin6R
ztSquOhl!^==Q23UZ3ELj?^dS%9&8@;4nVYvR11DZqwFNm8lT!(u9;>zW14F9Erq4vVeJ{(0j0`)BQ7wx_cZah57EuVCana}M(
ztO$8gY!2EJhSP)mJZ@YW2U$4eSLibP;o&I5Z}MjJuU0U;v?No$B86FBiyEfOhk^_a=e^6Ew?mQ`
z5VOg(#U7MyoxsE)R&$u|HnJbg2lSui&+Nmx1)b0#i|7n8=gmu$*}uo&{uJA(@WRNK
zMP|J{P4ymSGV@9K48>4>vw^$cw=nx-|7yHoV3Vd|hR>RW2NdpA);*!^c#g$|fc)s7
z7RAS(vl)BN`MH9YAH0%7%VS6n3<%Ee&x;Ttv&j8Pw`h6LCB~lE8ti<30SV+2Za$*+Q{jFKwa??PD^mUkzbvJ$
zkIc%joxKlFL5Hv(u3{pEO&0wvrTwVSjG}p}tg|0rVIj}Dat_U#efK+0^X~7J1q!g<
z(G?S^pPTNwLL-r2NVgYfRBkDwe&0g=Uj`X>~Mi
zSuEv4$8~Bzh}v=2SZx!m_jJY!I^XgkHf#VEspJ0vmJ7zYbRqS#bGG06gS>`nN-dU~
z*WQN!3i3rqIeW36nft%{)#UuK^kK^)xowE7GB(TI9U$gHu+>CHPK(Z(X9
z<_9|wZ&}<@F+zMkc2dAa9LFzTmJya;I93i3GQ^o`6I2j?G#lSH{S4IF!wM`ho4jMs
zl|jE=Ex5r6@kYUkCE_vn`PyHlfbWgk_7(H<{;sw(
zpKTU<^~@CD$4{-@jCkerXSTdL=pUN#X#OhTrw;`4MFGC7xRBX{@rXcQ+=(YZV^mFG
zCxU&h%S`$Wdh^uRrC0;Ce|WtK=Tq@qM+9+i>Y}X8=YcL;PTr^i9zR;E;p#EWBF~v%wKwh>03HsjuNra$Dis^zv=ec=
zcAf0GLB6y%^j^w6$HPqu}Lid6N~Uw{1`L0n?78ta
z_Ee3h$#F!%o+?$(I+qXmz{ML3MS-r~Sg#+b!6tDkYKkEXU>}m2;6@lvL19ge8uTk3
zepn|9#=R|?H*zh=YbOb7I|9A?x>k7Q7+7b2vg#7(U0GQeuMB!NZ0Xl2P=0?X=e{`5
zvlj7=E*!WYZ0&FD1vjkXO(*DICia}qe>v`rV%tP_#
z(DD5+uaolHe{R9}GinsHaQ*a>#6+&@!M&DaMf`Z6PtI)Y2vcB_`fW|oPU37LTeiOJ
zIIL5de5MZK_{i8IU&Qxg%j$AtfR@btTWJimY?oYdIedppdaWfG&lS$C3mk!QydJaU
z*mICqlH`4UAa6d$zk4Upn&T6^458eH^LqJg;LR*G%)Njw4YHVS0W_nu&CL<&wK_gB
zl!I|zK2+%M1$5GQ&9qEdAFmb1Eli=F`;VogZV+cx>Q)C%6$1O6*+Z0|o?KClC?Bk6
zzDxnT0LI_?w@$nr>KRIjG~xXfZ_L+8i1QP&O4K@mIw}`hNkDsoF$+&(+@m64csUR#
zXLwcUM4*!APCtJNdao-~t8u@*I;S@u^%{i(7TW?pF>v$lmB2esoT{k}^y#5PpFe_p
zsi^axuRwd>gs&b8;~Z`n>emN)KJE}7e$O?tbuVK*jlB<7<2ehsTyjYizR%^%(qnD#
zJp(izn7jpg`qmn&vubc3FwPn;>nk;hIof&+*C<82GDTwB{c5$(aJuil~ZK(}ed
zhhn>|k3ZrtkFnwV?GqTMHQbJtMg2sZ2YEiv6+Xwz#`!hdw5y?g_xC(S^JEb(gYD2iO!iJZ
zi+U}0RcZf@A^ZrJ?BArBfk6oMaF%$IfgN8eU
zQ+kIHXhR(DkVVJoD$2w+cITOR>z9g@eQroCG
zR9+7@7gS`}ejRd=!c;la-hdJX58d5oKc9YVg~Dbud#45r*)jw#nY&EWP0XK+A(XDS)}
znY_$8TX5q)0cB>txDQ*@OH=tCO(jflBeZ!qnW(>5^h|Qo5S?)ja7=MJ9CGB^wDU**s=?b44#xds^v#w!_FHzn&gz0Z;AH%EH>(lzIKBj(+-%rY$
zIn9TGL*6J211WfyA@;4
z5zFC%0{qq9a#&y%`Os!dY4Q6Rw0uF}TZ*TT*ap{NdEy0GC8Q0KqBK8}n%hI$3)`Jd
z?ZzRE$)~aW_i+XKK19FzFa%DJ&j_7J@%`x~ZV!;pE}ORj`_(&M1R*TgqORnM5->J3jqea3mGUmjCRsXLB|8_{U4uL
z`fg%hqQgThfBV#v=9!=Jv@Jk;d^%HhI^x}TMnPs+ZdxgQ4{74GvT5kwD(BC%M!##x
ztE7ne=1z4C;2_U+p{fXRdDX+F0Bq;tF{3!_ulFY3+7_S+<4fQGLSAH`n7-@Tw)>GX
z;^|yRw^GFaw@o$}eFwUvdvzJ+%REm^|Be3A)I&lH-%q?s(3eI(`mAEwH1zM!dR5*d
zPWR1JmFR`??!~FHh>tyHI9VbtFP62N{uOYH_Rh=JSg#}d_=5YO*S2y;4C0pjuqpe-
z0nb+G?f(flxz28PGWHujDX<#n7bHFe28~TZ$f}LfKsir67w!Y9f8x$N%)8#q99|*|
zyyqmZH&=n5m!;Y!2mFcU9p};S)-89Kqzd%blZ9@vP|h!1vTh;p-nwUV4*~!5V*3*n
zz!jQ4AF>=kzPMVWTngj@Hw$(N0M5Gd>gv`Yz|$ty*Du!rew;gZraa=0)LFk@r$YI>
z2Dh(9Fzzn-E$&MJSH(@Se>FTG`d6OaVF>t!<2CnVFwhHpo-WBiJ=|-$#!dsjowe(-
z1mMuj%8Ma)p#O290(?nOo;*veZDcIqgso@t=K!9qYueneauMRn@jFwefc(_%EdNsI
zcYUB|_e6*n?r8;oUGx!fH^-!ioy
zg#9U;;&Z*1*c)ZG{A9Dqxul@>8w8432~!F`W>O<0$U^QQW!v&rGzU#|Mf
zvB`(^Q-o|L!o8BVz_ol>uhD=TVM@BH|z`8{&UUy}@MYUnOUA*_1VG
zvS;HPo+9{u`@hZATz&xTRKbyIWfK;eaK*%_E&=wB#@oAh7Qs6vC8dT*=i#1Y)#lvv
z<*={qmrXUA3GsH$nDeIq-z(^Rapr}6nP*aa?Iw)J$6u>03UI#Ug>2!?@cm_H%=PDk
ze!lfQ(|Fkq_ln-W`xSt`F!(aD2IRK|^()$8zNt}Cj$@v{`>1DLGuNB}Jp9zX+j~3U
z(kXg7_9epkH|&kIfbTiv+ovI^1$Oh_boKnPW9@4jgBkZou=am^--^<3c;D(Xh|3?@wyit)C9_d-!Oz^DEHbRpD^pHc-K^
z(oKk~(`yIYB4Izf-4AEk!Me;+;k<7J{V@)2&tjZ>^6JT@58-^>SXpm&;BOBZdX;0^
zY*
z?`qg*+2VOW-uC7AI_N*7`IH^rQ@;PT^#38
z9HllT^&Q&X2M1r3puL>;swD#TZ{3{dl?MH9G~S+pcHF46pc(J2G)BJd`wBeEOlL0U
zIZlLr5zmHsrsV8TR|NfEZv3|*L2i3PCAkRnVm(x%G2d~_S|xB00J%l<+5%h`KRc^j^b@bgXJq01=g%`o+cDnwv(`mc3g)>iVf6hDkULuI
z^yz?nk=@(EQsA48Pn=l|RP<>7A{@V2_hx^jy2AVDxauFOQNP!@rDYn*MN=Qn#Bq*3
z^nr&8a88KjY^42BmNfr2>@bnW4YgwHkKj1@)}8~z3+u{*4>%y6im9e?<(t{p&LcmC
z6vm*xy8LQaBaZjd-*d~c{iDzF9kJb0ns#|eN7#St>OtP+z1a%qS@~#`;*6Q?F_Y1+
z>#a!pgZ<6zET4h;iMtIcP72@fgW{dkY2PXSa7cmcLD0`yrAp&8_NXSsNr82F6d!eG
zzNUDqIh%Q}_H^wyf7E}{%EXDma}NN(1OF?^NAs*djRE#R`K7yQDJW;_+ulUpZ-7nl
z(Zh^Riqm@H-b26w{;*rF5x-+cpoJUiYqf>}&;wku`pjJ9n}sas`fcj@O7VS^-8|a9
zl(`BWSG*U%Bha6G4APrG-*@#;`L_<{zAZ-*rU&w&nUFq(_T1J&*bTf#VI-aZltjEi
zfcG)x!Bl?cJCl!&mYGNO-Dgesiu%DnBWV9m3z&82Z*rpR)>8`8hVsYx$uwSE{7wx9
z0{4lFmr%+h!7#p=
zS`0o96o+jFymUHjPoNqjOuZ8NOAyG{a`W$hF93`~J^2*+-K{N|aqde3xC#9Q{(|ib
zWJKYyD8eb?Mu?apCluKwZ@oD}Q?mG5Qr|osPFzeKCG@0tp71X2pT`QRJ
zhl_O4b=ZHtgqGi{UrOnh-aM+8mw24+ul#}vs+Vz+Mc1cEgL(flHjYjE$<30X^5-$R
z5CFn?3k{&-Os}{@arhGFuXJCT%L&o_`~+G_SNkLs$Uw*9E2b(oz1&
z=K5l!C8i(Su$>B@Kqa*MJzH+`BEN0hBQ+f7Db{Qj`U!0T^-4T{NqyhE&@R0zEGbI{
zswv^qfOdB3z^o%EKcnDdiT2n2hKCjAx9`Xp%)oat7JT0ant?X08PGwyTm5eOu43Sm
z#w-h$0seZr(ZD^x2_fSq??qhyHv9ZDNsv1ltg^ocJWu7+#7n?yn6U0#0llBvW`Q`)
zWTRGbEDwz8>9vFMu-P&ZD&Sv|#wC2r0sp>m-N{AhKlFy=@8Y=wULH|JHotz6tFA{7h%vYk(X0Ji@~ggdtBJW??ZG^m*nq
zxn6+sFPx?vHt1LGD%Q#c|G(?vjSG@5;hsHufx95+m4DyZWDWWAp-CIwd;$FP!_@4N
zSSO1VkMqBA%p3O2DVvSXP`|vb*j60+$-4EIH*Nsd`)Wb%F2L)v4&`}iLwOa+E<0Ls^B*-`kJl!-9uDgMy{!B
zh5Xz>(Zvy;AwE~=ERxfL`LAxizNiS|P%GB`P?*QxGu3B8ln01um;Ho(Snu`Alz$05
zfcv3RdvT)!aBrKdBVKvCk8G0sIotF@FIifA;OLz9fHOAh?dfs={a2Fv>I-^F(n^Vn
zg!(=rx8$Srpmi@%%@Vj5X$o=c)ts<118}ZXPqF4UaYzZ@gGIw`ePq~I{iy8kUXn1g
zW!@KvGa8qga4dH86LL9ahRs$E>FFE|Pk-7++>%T`xkzxx3(F@*v}g2_k{QxVZQt~h
z7dqwRn;bYK+{wfyPJlz21nfn2E$$-$N{Vjc?|R7jT_XH{p7fA55rqe~&OKyW%PgVi
zD|?7ZskoLh;Iu#4_oDm?d&rM$n{(_FJ>=kv#XCBjE2Niplc8arN`t^|GI-S1^+9_V@$l2RQ(MqQ62;}&b`}4$qK)n{;b-%{x_mBh)(PM^oaE{M9R5$|e
zy>s{4dZnLe7lQIV;T$N=?Y<@q`#iVo)-z!^$03?y^~E5r65XWuX%^trQl-t-Z-5F`
z%~%rw^>@b?wuM1Gn^4cV2xYhznz_Mr0-P7!3F2`AEEZ9kI{7QUr<2WSEb)hPT64_v
zZroSce`yAj&Vs&d(c|l5;Cvg;w$3tX0bCHD@C@I-X-Ai}G{HVie7@lQGHJM%m|v^+
zN(%1Jy#F4~t%viq-cI^m54`VNI-yts>_yG~;VH6UXX@VCeOL_r6-at3;eCaU;2Mtt
zp!Eh~<8d6}a;!W>m{-Hpjbe!NbyI_%&IYRQ5?6g4^eUIRBz}kZL8E+b2HyLb=6&^>
z4*H7TAC$s@UmH-WM}TUKF1T?I#@+Ky>Mh>8?Ji6-kpVqjj~msP2YG5)H;8zBRNW{(
z5!PWxd`Ywl=(UJkY?=@A3zt|IjCOQH_b$Qvr5^^f)m}rrIM$-+h(D|BE#Zf4Fh1#B
z5$LaEU1Pr>?{zti5AS*2JZULGobSDJ>kPbS(eP}(ia1^}J5Cny>a?Ps3|Wu|sWxrG
z-vN0ys?huGPS#RudLOIj)QI=7^SyVzTLR-bkZh+n7xd>B6rRflxuf{nspu!BOun}S
z^=E}1DEk0(a$!K}L7?`mee$`Wm;Zf9Hu}342iL&GI-98PyVQdItXiNn8RffH1Z+hA
zxjVE`0eORi%==ZFb1TzOeq{a;6`Ysyn4M3M-*PXc4(Xge0Rz-8uG{yx1NdZ>`*(0&
zf8P2Iqg-*OZw>mzP^&q;sDCuva0AbIYAa^d7{mZO1ayPt#9m$NH)Ue`|1@
zu}-t1kgqf8yo3I(yGz*t$H5yL2^n?32UYOg4EQ^Dj!?gUaOx$BQ^n(%cNDVkxe+J0?r>ny)X{&~`UQ
zX3)GJ-;3~O9GCQ#F8YojtNRZCJ-jC=qwg1X%~7H8!M;kf**MPQ8z)n|{(S2*crXHZ
zUv3R;H}Xs`09!b(7a2S`r46b(TPmH$EU~Xt9$H#N>B2^JTF&X6
zNqNPtYlB!%OBLV}Q2BKplzI&nQoZkiQz*WZ_Ya_WYtO)Ix*j$s%4m7{CM&A{s4kbb
z6K;74fHa%9oR6aYSMKeg>o8XP8|9@P@WKG@e>|3}qF%!7Qxs3?J>sX|W&7<-RIi|G
zDP0enARWrvO65}OZOYtVy;xmO?_ri+lcnn^6DLpexDTF>&~YXFm_gU~faGHuFRsqN
z0T)-0mruUZh}59x$#go-%3pMR-TS7~cOyyzLiH$Lc4f5$j?=c-j>eIvy6oX10e;`p
zwSa!l>3V^AsHdIE(~alz!Qsy~Se~unMelt)(yM7amb*~vH10z&>Dy}fo;vr|Ie*kU
zXm3v2*;9_cREB%Zt-gxb-o4{LXuD@w2l!B4_3Wkq?uX=M&9G6~Wc##|X5_bTtFkWx
z+AHnfiFPFPOfStNuPd?wzzsNX#bnEJkmucX7~j7fzGvO16y{3jXB_}Ppo8~9
z*WX1Kg8b3^%1bpsTMZ_8VIDd6h|deW_nrHw`CJb06J_K)&~Ap;PdL#Gw0_I>4`|P%
z)?EujJeztdcM0a*6gCFGLOWgY>%iUJU^g2LlS&ZBzBMVPdFsuX*R7{O`P7T23F65u
z!@(uVU@u+$(%tdhF-L5ba|D!^+b+7~1m&l)&%7N0UT_{yN{9&LWn^C-$NIdS!C}lZ
z=Sdeheg!{K(yMiGIr!%`&l8Ie1Md9ny1@Dhl)J7@7}Nm2_A)xSClTzlZ@~Si((kaY
z<)_4so`QVN^5$?{uLn0KI7dS}dRrEDg+eG6DfG7iIg-`!g*{UHziR)Do)I^e{Xh9a4HV21^-
zEz0-Qhj;epu5G<$1^KLtA`cwEAGv+q@b2hWc$ao5abXnrK|>uszaY@JS#{vF_TO%@
zP9z}h8Ti9D8NcnHK>kwF_Vnt!D^S01ivLDicy~B5hhy*FPaa=PNjC>PVYGix^M(c3
zL6NGE{i2XB`u@Tu-Wl-Z!~BRZI|s>y%>`^>PDPCQvLfv$Ar^;#5?BN#i;szvMo2UDEV|hsrmXMVs0{rqs#s<@wQGUth`;przQe_I(XjDz;3%Ln}0&mrbj>%BH_=_i^Bq3J4*`ib~H
zU&&IiyW+S>l@5E~3yaCzQUoi+tIBv^OZDi`XmVy3XKq
zlB$ddW8=-8WTDsmi7&jmNJq@cBlqTakz1D@G@V`FMKmloZOnV$NxWT~4}I3|Bo9gi
zc~*buB;h{tYFCpx39$p_sou-{6B+q4~O-~1LRkC}rwxfgGd+Y8P>(@cn
z#f2^R%V{SyJA3pG1j9I#YhTC!o?JRn#5%kd>J2TQ>TlFTf^}BKeA@u_WkB1jWNjzu
z^t|@Pc3}rmY{{wk_@jgD`s$V95!6jY%zvj|f;@71RdfTtMkg6KvomCpcRN`>kGYLXDd-5J+R^mC#Al3&*wgn))m
zTjz%P>t+16t^@*=9!ko_yhZ*5jd;XA`Hg>SF`x15XQ$vI;2*~us9-$*{*pu^;$6j8
zAIrahTyB}1CEl-&m%Xrw4gNu~JNsrE%&T(3kiI_1oksY+W8UNXc?n0%mxS%~yg3Z=
z!1StY0nmGwuyO(ZPRArxDcl(3LH=1!@IKO2vs-sEjI-MK@6;=xcSvO?Zy3lIJ}=s7
z4SLEQE%VV|Ilnu^!EwGc^pQqC?0k)Mp&!3H)n5kvF!`Rp7UgG_8~CE1+;}rZ3H_hy
zn@fLCPorDF3;pcp76W&rE61hQ=7Ri^SO@$N0N$O+&p%D`(BCQ?e3O
zj{~p9+4%{_Z+iIhUi7m<>i#*HS5nelv;_Taf5@OO;(w7B@@H{YueeD)IL;43FS=3B
z=G#Jw`-8IWY2Ix|LL9}{%QB>|A}-;mZ{3Z09~apBA&vTRQx4nJFS-MO6@O0>OYvV!
zUOF8|v{ov`p_$niDP{3RDWTp|yF@w;iM)D>C$*#SVKw;m?owLsc;`1W)ca<&gvwvf
z-a&C#eg8T(>P_3{MDv*L)5jh}J?$IYY5NAVeo(#nm)6nz;~T@5vRJggoUcvnBQWxo^}`bs&T@?_+E2ncXT@DaR&~FdF%x`Z~d#i5?~j;bhs#=f_D71
zH2D+&cUrxW;-H;6U>LK)7iimStFqf*KNH?wzqAnSb(Gr!tyh5iQY1Nk)lfcbV=
zt<~;tj)Ql;;dOC6(2l6IfWlw6A5`)fT5tr~Np}c(B32LiniG4YN5aAHjI19p2YHg#
zTa&Rrz&>B-ULgqnC*Xp#yCUSJ&OSJ~=&yVa@z|EO@$bF
zh?n1t)wYmddo)9O^VDAQ&0g8`w#6XnPzgDF@BrYC-?1}0K~I^rI`_et0rJQ^KHSK6%(C0C46ecDS%oL_e{x`*?YAc|U315V0Wza3Yy#tXOiOp9m)RI#+o2
zLtge?()l6|iEz_ewRW(bL@VFV^}gIjYE$H^909*b-H_7b?QSPm!^V9mINwQnYJVRc
zE^i}u=Q!m~uxTTkT;^5|wY8E!%lAKiY-uNBdLNx%+R#DHUm1I>OSX;7OtNj4aBCw*
ze#b)Y^|g|nv)tDnIM7D?n!Rq1i)tt1rWQ|%F=``0-!9rEdA5;;2_A1YYPJ#a`_J}V
zo6<&15B&wR8}dp_le)hFwJR
zbnCJPh#w4g?n=;weEROh&bf!@z`+c66=6UKMx
zI^}9Bzbt@w@4n)s>u`Rib=3rHgZ#e9i^B;+u>V
zo<_z>>m;-ro%(Z<734*=Hrq)eE_0O2uiB;#dB(Q+4|)NQoy-2N{t@CqZGDA~ye9Zv
zYqUzpO&mwk_es5=FFe&nUmnhJTF~Uoc(7}m@7i8p19l=Hy;kHM+C$euH)ZIjN#?w9
zDCp^bDmXC!bYrZYO&I8{4>Aiydwwr8NBSt-o1|+hERBJ7|K3?8KNHrc>DdjftDtv7
z|KM;WP`34!YuV6VvY6fu%=bRJlK*`?wC80zYusdLNB8qc#vjnjkn*Tr0yuW-k?y>8
zps%P9x^gD8tC%=(bQZL`EOBj&CAL@hI4B0{`8TZCg#JRRdQ2bM%MWW}1Q91*3GH8t
z{$=r^{tUG5qt+VLw}7&??VX7C^0U6L-;VxES)T6?`j@@i!W;a7FBrOEcn0`QYWv`q
zTi{=Af8LDw;n12;QOq~a&&XJ71>=doE4oYpv;6(3;@@phLS(yJH?#TC1rX{b9&n@fggbMrKD3U_UyM8gEhlsPVoF(qB6~DSqMm
zU_f!tM01z!PS8`AUq;^zguiS3jC!0y74-M@W^p5sxPx~f`P)qDLGR9X>0$gnWA8S#
z<2>8XmcdUE;T`K#LmBk9f}zU?5obO;5;BN5z~)ZNAl6Uh%YqARc<(fNHWSxTveoG!
zj`NC--CE=Y?u9BKy&*q^{+_2>?9GN`~e)%}SWxf924(e}L-^ISu{
zTOG|Ih?_2XtfIehiUGT#MESKBecM$nV3sV-Z=j|;<6o-!1^;u)P
z>9gFYV)=xQpKRQZhN-U|P@X)S!;kcl(8ygZmtJ8p9&yZW9{ypB7ZzN8-H6|#;6yJZ
zUSa-yci^X&fZvX8{(-ovMuf+2H&E+>dir;xs*k^>`P9afGH~I}CUdHi=ixh<;HIKm
zB|t-3f;_SQU1#Nqh=)dgoUX+Ew<^`M<09~Tc5LO}0W`!}$+zF{Q0zCoW
z^o98DY7EcXC;spbuIJImjW|w4_2b(9kcS&s`|IQdupe>*ZHHy*AwL#Ul+*=r^Y3qI
zdvakME04UNnhpJD&ECDV7S8G3E#gUa4KUtQ9W_~y*Av;dm0byX&igYiB|tlm6>UGy
zhd6if#Vf|i;E#$}>76+Nu>aF$U+#zUg!p}X4IoZ6w)3v+f^itf$_0!A`@Xlfah=*e
z*e@ZxYL=jPR<7;W9LNKR8RmD(z&pasKY3-zaPQ}}m3P44eGi!)5Usif=55+=^F#mv
z{HK^&zSa_c-=t|S&Ik9eckA>Ye?B+>_F~K3BzSlENPDcjk{aL~S>xxO3*h&|RR`Z+
z(BhCIp@(X2d>tTyX&Q4Cz;A4s@}g-k*!^{(LSr6s;GSB$(50md@=yWF(V{Cj?qUDB)L5B{1Qhe&x{xoj502T0Yc}9?(zj-qUQ~0r@)D
zlC&kBWBN&P;oDC~lKP2lw1cLLUq7jgo*>h?s-LXXF^P=z?I%J@eeSFA_LF;6y;Yya
z_Y<$2kbXUpe)9R~?XUKe`^j_hTM>6n`$<=g^kt)k{bcrdN9($)9OA`c*}WLoPwW~C
zx2gkG`r3Z|g)H=YHK`+fPCwDJ@Bd<-!XcJ|G7C!~pJx=pJ6}o+@Z6n^=NkWVh=%9`
zp2#WvWNE9#ggMr2B+E+No7bw1cK_j$70orHqxE;clw2;ZRD)**oUF9ZDeh-=to|_YZmqKHR5?~
zm#+R7)xzB{Afcx)-$>kiFMELq-07O#KAujvo-vmJ_v_39vA
z&Mik=Y`RG(=bplw+dX8re_E_@UmJPZdvCV+;Wk1%9PU1LZYKv0ZD>yH>Lf>+-(JYx
z-%T=?ELnU$rk(6C$tD{E+DY~)G4UIk?Ihss{#F9{LSYBFA(5aiGUV|1{^7WG65iAi
zf8kp@Ip#2bX^mey;eC8TH6XErJRILNPy@JZx>Oopl0L+Nt`$Lp5Ob_SJT-nk%9Rw;}pHt>9IZJ13T?DzHg3PIkx9)UXJm~LGnqSX?aVCc^Tst4;RV?~?`w+Z?37prysukKTHF2Eu8uSl+6fs{0
z>zXcWR$vGAdf(x!ZRzm4P_HB&k^p*7^5ib#-)FJc{4^~A^hY0rHF*GK9e;5S?~4{Y
z4;i5!E0j<0#NQD{zJDc*ze9X`K`CW5@S9q8+o0cbOGs8m9J1YT=`i|}X}M2VJcRyM
z3M2`Nfjp~ymbxLx`7Ydb!Fc?^tBOtNw~87keM5iH?9}^vCd@y6{{17E2a@U@cUueO
zx4$Ny$^~9vo^srK80V8KN-wX2-dz9dmC+#AHd5cS4D^0m?!MUu{Db=8?KsY;E9oxi
zuMLBqJ0gB*QFWIxO>Yf8q>^7bXu5x10rjQfTk3FF+@2bKnn`KpY4U*Nc;B7eK$dn%dLt`xrv?u??p
z>$_;kL+?S=`Jw@M!S6z&JxvEz}cjU(Lq%Ye^M+CdjW6GuwuGuHu{4A)j#ZI?X$+l=(Xz
z@w1kqs4>c)zn|%hdMcO7J&@04Z@-NE_jZmm#TQ9?OYwVl8Fj)R5P)+!f%OB|@w=Yg
zBivv2GS7~C4s@nwTLQkTUhZqzg5ODatOooEJT@`8BS7ARymIfeBvIhcHc#`x_X>9G
zz*G3$uEf4Lf!|FoN&XnFr{qG5WwG3W)jed@h}(%Cn78uWY$X%8U+md&
z(sBGd4l$ll+NUDLA)`YDGj4!C_&soBtDZ6Z{($phNF|U%KG*3ytj**Qi_7_&V@CRj
z@UM!BV{mWDo+kV8`g3?M|M%^+Ubt^O>#%&#dmo4F$o`!0M1w<=?qiGh(mG<_T`HkamazWCBI_TIYcaRG-`$?_^0}O&G8DoBw*vL!GU{gbpGR{P9f(_)CL5
zxxHCM?IbGaqQ2Ff4x+n-RNl)19PxX~qP0;S@Q#mFk@>Wfd~lD{;Wg?Y;k=f%v0)u#
z=Fu#vwFVtzC?{Wp5AMqzUtX|lzh@`0T={YAoZJp#B>8SaYf}fwO50;leY=DBse7Dv
zgFMSFh0Wb_Y&(et5infd-$C5W^1
z(F5QwrFmyMH92$>x%$&779L>tTviBp!+lllR6`E}@4@;%{)+wdtdsckkEj+7brHUc
z6SLFizUV`tD8U#a5q*sqh!TDI-bTfh}%fgK_Zw|dBWo~K!k;0Js?
z+tu#DzUWE#vQ5vWlVqO@`@&J}BHIS!7f-e8A*MqeyC(sTI4vMma0>Qgsm#8(aiFKF
zah-3ybQj4{Z`_up*F$E7t=AERdwZUcti|5oC-eDkPPGSpR$qqOX*e(9`mSeJg1=lF
zGizTt?E93BQ$D?c{sSImiZ(<1z#BFT{*W*D_wwiZ&ESVsGvh_Kfu0|~@fgep2f3aL
z(wqb5XmF`SFvJ5JuUX&qfO~$mo@pC?!oF3z_3g)AIG-YWWRKjM4SxsViSR~2z$-61
zHy$biy%XO8RWHJM-Vi#+tQ_jIUmX^BPqRj_XR0NU%FCp+T<|2LMdJRYm(3#(LGBuk>jk|m;)kjmFYBq||F
zvLq6rw2&epsVr@xqKzb3DujqOl{TV9k)%zQmwI1XMZf3V_xI0xK4<3KxpU{8JLfrP
z&YX*NW1&>SC|&T8ry_Oo(XM%XlIa@oHztp(p>@o8X>_|P?4#qORm!r!&kIXd3xeFC
zfTABR;3t&+h&KUsx14g>0dlX?@{*c}*SNfO4Yj8$R!R4rx^c^QD=OEy(vpMzr0iaz
zNP1wFgq$L+)03a6={KY0-{!vn9qN7Hdj;qhv$NaD9`+YXlhZhNcKRv~5+FWZWhaL#4rV;&;`j%9uB|5h9J~K2Qd!~r*}nnq
zG!9FLUH(qxeTf$DNMCNAQ1B=Lc7t2*NS`mT{-aC(
z=euidB7`SbZf-tE@jbh5+ZWPx(fh)7(7eq){=${UqjhQY*jmudukNm&hjQ@^X96-o
zFCBS1oct}>BVsrF#{A2Aefjw-yg%DZ>%N@9`>*gZ)aw%D8vpi&7D3Kk(_@+%Xi;om
z-($$l-ZbS3ANwrUo+kWU$c?+yxn)1za^U-lz_B52kZZZ4vW(uhOS>oh*b2FzD;8OY
zA(t^Xv-&08TitYchHle|=
zC&aX4d}E;ZuLov%-hy08mwzPU(0&`h_@v2VNfOZy~2Y
z`u>d+>@VGT|KXA#7W!BAlpPQHJlw5Ba~|xK=Z&qK3L#hH=fc043;cKGzWQeu#80G8
z{bK}tu(RuT6WM8rwhEzZu)a4L^j=8-vWZ1E>D|?s8m@&+1P5fB$OR&0b||2WJlKW;sqXi;p2b==-WUHH&5e
zmtP*RNu~>aHJhBl59Y$}Al4(U`%J*T?y67TR4-unPlf#WHx9V)zIB6F0dT3(iEw?T
zZq``5aAZ=QfNj%jijpfAu#2@mP3k|}*#Vsak-w!o*fSf0XK~HYpL^ykD8YV**6gaK
zB6c0@?6sU#s#Br2R%cXBSlq$lzBkyelj~s8U(Y@c`_<0kWDkw|YYV%+WcRHh`#YG)
zi0O^fFLki=PY+jWLoX^+#u}fLMLdt|K~ZJL4yIZ?xm~%ogT+j8tSyn^vB#rcCG0?a
zkoKc`rR@_s*cgY(4f}jMn2&|YFxT!5cE&1xYm60-jZmDM{OLjmw0zBdMf7)&{K%&-
z_jIrg_Sr+V+&kFF-Qvfcpf8iM-(K1TecANs#)B%vm+u^Z;npJP%d^2<{rPSkO#bDH
z=WC^T?A3?K^$%xs!5?0+(C+0|Hm6EPc`tBfe1+@jM#MRNc2M2vpvYtEk8OQewXloj
z-`0NTgMG!lhxX@wMSIZ>PWuuzcd$LtQx)Ruc}%5QQ$OJd;>G&zN?gNxa5GwTg&gek
z%sW~+<74oAQTA?D@jUj1sfGWp?qZRL3wC+|&lX$knH~l?nG3awFQ6}FZYz!`fWB0o
zImKBY`f|$|t8WMJKFuAJrQU@1X}#8K&4<|UnG^XeY6@_!qj=KKY{*4^pL3SZQ@(_*-fwrHy_wtMg2Ex^8_^JKl~kd?Ri
zt{ddu-L_I#2L9@{3JtQ4HY}61k%e4Yuvg4)@N;(`Teu!_i>rrAEd`%6ILgBY^wL2+
zn-kFcA4HPXNYD2qb$6bJ+(qSq&uM>q&&N4cGeCI(dN*6q&sjTO8`3)c!*xkq3*=^|
z?N)b2y~p`2566*QSC3qMIm<~;%pdai6s?PioBigH{__oB)pKl*lb)A7w+|7_(6=|*chLI3+*w4Co;&G8=NsC`I$P#&h4z6`6Lx+jIfb5^
z_QZ=k%Q~nG-hcRzu@t|m?$V8bX6T8`s20-u3%}*8rS`JQ){P~;yJ&(E^)FyUL=o9t
zVY9-IlRmlaImDafERUYx^yQle63Nug^A#((xNf7fR@{EYMwK*idX7B@eyfvSH5ott
z4awbVvf}nl)LmC|aWT7wxvZsqh7;E-xbnB%x43xVQ@6_=l6=cRW$wJ`XZiD7{)e~{
zVI1|Ex!bw;mcy0xT%5`1z44s=FLmWz5nbnf_a?XBu|Zyw+c%gw_6-+byjnbrt51ts
z8$){W#PCA{NMD**s!S$(`j_+CDTH$}tP)q%fht{*s9S>a{AI)R$&O#OEs(<@?^4Y}
zN#5?%fO5h;3oa>tr2VJd7VlVk-r6@CM5&#LCSzC7xP@GO@tMZWPkf(O7-*K)U#?#+
zGp2LnbbY_)uDy7F
zX3h5RrT*GSjd!??a!<=ak4?c3lG@(47VXYmc{p4F`xuR#lA{QBMkq{;y8*f7qB3{L
zPY`tOgj*VTxyC51Z@+=h6-w`(gM3D$gwH70$qoS%Li1pcjFq{%tL_f`6q6@qr3H4e
zq#tXSNkab@NM6i*3%k4SP5sR4u%Gr$+j7$i_R4{P)K`s>U2L<$*JmAgt~6E2*I&^e
znR9zDiKAY5NB>xT4fwHovg40qzj*w1HMu#!9l2AiPg~-B-!OXo1xc(!^>w=sp2d1P
za++z-G&{s4yGw|uKyJdO{r}N;HG&LqhBev@0^2D`L2
zr*!%J9awh;uY5V!MZnCQm)Gyi>}GG{BCn4=BVe~QSEtqG3s{4#tD_ER#p$bE!HEJE
zJ@<=`F5-)wnii_@w+q;>N?(hEc>;ECkJ7vQR|U*=QrD20P@JbzxMI51M!+6PT2(BL
z6|k1M&;OJc30UycoIge}0+xLF_$8ag0;Xs5YQz}nPYrcHZ@b9;r?
z9#t7$_rTz~;$|M(o>4E`bCSniPOqpNeUZl`
z?tW8{K8Nep^`u`|$77>rq@HxvZ}?%*yhK>%EOYO@9m1b+adp9&FiibJH$)p>g;_l#$$gaJiM1bL|p7hrqS`d
zgUx@Z`EkB6j~)1#AGjg6195M+=8u&DKH4!_<4aTrD_1J5-k{0D`N|uUMYf^8`_E^e
z*wMjsUfgrs4L`=qzd0om@I&OST>0H~43BvpR|iT@AD8LF`MU3V
z@e|L-@Yvu#ad``{AHs{jHR9WD;H54}(XJ^SOl$SDfPvA#J)0sL^vVzip}3o4)#RirH>8p%c*tmW)FVW
z#Z+Dl{;&k^TeIl!-cZEJWVTJ{{ae<-EE{z^vflGpT+6Z{GV-u@*Gs1T!n`dx^H?en
zxI5nUnztg(Emr9{&epE%WDocmLH5|!NDf?c$q_iH%;C}PWVHKs^Xkgh)1B;2;*H0f
zKX)h!2dqY)_X17iW`|_^u=RJjAbC<;H?2owqx=u!?H`WJ7&*eVDvHrB>9a!Hs
z5qVPDUHt_p*FU`}rUba@O@pJ2FV?Z)M)Op5gYF9*_wpC~Byp;)=2civ`sOO8{)QbE
zG2pLR9_lw6bUKg1{bLQ*U%Q3%ap0OJrC8i=R`%?r+d=2T8VYl^YfUau4u<1uyjQ#tFHVCP5s
z|LXO-W)gnM_r8vVgYf74(`g|+y#I}1!V^#%iAR$jf~uM29wgjjd>v2^w8O*fCgG;{
z&6oTrUZ~^x9X;}kB(%5RAh};hq9h2ngsgg=Px~vwhbL)~{F=2Ad+Gjje9O7`;n~YJ
zBl83DOvPyHkpAvbFv_FS{M+!l1=VMDIH+j9^;3x3Z*X)bQ%z_PPk
zz5?Y%*k(a_s0V%U1@qv_ZI1W1-Okb8w^sITBfJx@n#je!1Qt~1lKk=F
zawC#!m@(9ccoBW%zEe8to#NvfZ0x44w}0hcSoh+i_+z=8TRE?cvm%G-T!
zbLZcRR`)(9-j6?YB*_^Jh+jqhtXO|vp61&Y!_oKyfK%?wN1}b4FROZ2NAp4cPxRaj
z;FtLm$_A1j=hKWWlV~2qj4VWhu=9s(RHJw+H?>}i?bq154Q*_+g8T7mv)ur8pS6_#vwQohgzl0cXut2-(zoBRa2PpF?ayoGsc7zNxG58hEeZc>fIReTa`5P(Mby
z34V+}k%gmh?!RhP=ZRdbFV`KNWRAlBbH$;o%n7*j&4fX^;gQHApwn`EI{27N4S#*G
z&hAv7I>-Uzq8UAA*eLJ;lG2NcVK@8wZK!$;zrR=XRFh!zyY;qs{!rja?|Iv-e#4*i
z!!%LVWfJmXg&W0AgS|cSin55VdN;dxA}VmzUf`}-Hw~481?)ufXSe1SE{Ap$a_SJjc9ML{LQ*sV_@3{dub-we{
zQJMmlnH9BWQNMtdty~}OFb?}XU%Y;01E*;^=$-k7anRqha+%aatQUU7ZokC88J?#V5NN2CR;Xl`c=Ynq@I`#KyERxg{67K9ZQVG`?kk@?$uTQ^cn$3>`1dw&6Q2zZxc)v#m(R}a
zD3w|%FJO_A{2ojMZr(eh+#>oupS_+W(;tR@?N=D>ctD)bjvMav)gRHxt~y5_QT@ha
z`zMPmsei>|&)d}0&s^rQU0r&UgZA@S+WT5{am107#d>{G)#_x?mBSQz#X8xBES1z#
zEj%{lu~x>)8$7nh(ew4MicS`Fdc_cv^iI|zZTxz@cPASjvTcFc>`r#;W8IA_lRMc%
z(Gky;^kCPR9G5ng>|&N`>g(N+&tdPNkm?=Lo$MxWirrqsVN6cGKTBy-7n@<7F}};9
ziw*7CQ`$8E`#4Tj;=vbyzjQ33udeT8pQh)}*E_*ucCj}vbtLfE?4!>7gjqc1vEB0a
z_`kqsC#H(X14j-UMGQ#4Qh*eo;h~#}x9yQkTEoviOzs`VY+PL?xn
zir%;bJk~ESXxy2|V;{|5D(#l+WOW0)}{{b9ZQ&<>F4(^?(&SuR$Eo#PAXG
z-*+&H)-g`s{5sg)Z+4YQ8ax(O`uk_y=q@(+;?k$*rLd2(x<*a6y@M6189qFs4m{5;
z%^r8RlXaAzA6T~y@oZri#*N2&>o@YKMZa_>a}zzX<-^lX7Cbhu=F2LqYhw>}Il%8V
zwydnk4fbY(XzRqM$ZxS^^w@0=E4rA|ELo?GAF+RMWV34!;%}D9C%fxn{n_JxF6NS1
z7t`{V9=ROr!_(}U@56vwr~G#!rV95{h*lVHixugJ>GA#>7fH~
z?&$fov;5${maCfn-#*~NW2H`4LecK(wS$_*!@t;BE%s+A)<^vV)t-akkN1=+^NGhg
zc`STe?ML_(lS5aXU5)z;VWB=ss2^jmkuQt-ii6g>_F?^cG$6);^2_*lINsR=T-xt6
z+Uqjb!-E>LG+I#4-~Lif1?%zR)0ZxugTKyM-`3=%GF7w<-QQv|jreo$F1@iy1`cJ8pld`$Kwc&cx*@-=J5nzLf5$MSZGF
zO+zu-`DyAP^2G}hIRCOuwu=DIEEt9A2>!%5Fcep1P#eV%hp=d>yQWaq<%&(zNg>+H_@gXXTStW^uk
z3GYZNjoOq9Iy~~J4B?S7#c_j3&Z+p+Me;K(Gj`$PSuF2P6Ol&wi{DA}2yeW)I)dxh
z@bdXZgc~aQfBYjVE^*tA;)1?DsTxD=WnP%%O7}Z$%EuQ#5Qknr)Qiej_!;Q`0R8*!
z{VsaGw)cW;x_({H)-zPkJ2hez(PPt^w~~I|pYO}X>+
zy4}?N-Xv}A`&E{CvD>KL@xx&bUztYk<<6l^w-@5c_Z3Gt+!=Nb+jh8a;KN#uKFwMF
znXY?qzJjYaDu%le^?$8|`4s+nOyO|tdO3vIqFgE)@DK9O8v$EEtu1hz3*{QOunh=)
zwsby6-4WG7<-KBmIsWWqGy`6w8*WHYE3X+``IYH_%ec<=5!{U^k3I$12>!-tZH_O!
zg~vttM#(W;o`vT@;aop{{!rt}T>_f-l6-fTdN7TL`$t(0w{JSTkMjd=by>oV$LWF^
z?i|^lXLCvjH-9eb=kRr1dMNjNo!1>WTsyyJlTbcz1{W_LaKoMJPu*qVb1rgH;_hq3
z7oJxfsDGQY9}KF6{b4TyxOTH^j&SYTIz(`|e(3S5TpW1a69V+eQ?q_5x6gXBH-!7{
zR&B46lsL*4D4LvWME|Bvd%1v~Ysi7%K9Zk5Yf=K?`zbC~*yHr
zIfmblCHY9cId^`sD)9jK-7*`u)Xy~UW4gz2cE-)Ss$7462Y=z>**`sryF%9=z1*Eg
z~#eTc0RwVJ^Ix;tiAMv8>B8~GK)1xjl&ucuFEjSC>7yhl8-qVQ0
z3p)2(K?6h)ZJf`gVlle{FH`qc1Cl~
z`zm_htY&XG_yP5q-^yD_e%!2yBM48|dTBkS`{u>IYNq#S%=6!x^!^6yeDj9(VHdNK
zCq%s#H$+i?)jJ>Dpyx08PcoKU55`Qkq5bPRhmk{efXeTDkedSgo%eJ2(q+J7b5?ro
zw8ehnBI^mc;fTWsw^%9q4Cl61*MwHa;(2TepNCT(v)lnX2|KXPmFuU<*5RD8?*zTc
zuJA*+Z17Y1@}!J?|}TpcayH}qU-0(+VKhPJ4Nl$jlliGDy($2Lw<^E
z)#D9VSCyXFiIqcs6Jn2-K@YXM?`9Vvw<&7*08_|W)|_+P1b$a*l3*sNSnw5YA6W9W
zjZ8i8?;r(}A{}+a0lw6nbLknrfAD_t%{o5(bfaCG%*y$!P<>ce1L8aNBh+6vgAQst
zH%)fv-s@#EvFAU@qFf-!dHq28=
z_WVNFKl2v(?CQn2Y|lwyBTM;gv5nl7i^ZAIxVPmW(jFdkX%qZ@1-k?M8gg
z7@IwR;FoUvK00s~_HV_~on37DaLzX5;*IgZCng)8WLalIA03G^4vE8h628yIVix=+
zSMD!c>kWN+!71S=@NsO+Tb4Wsd5ilqkJ(>`9+?%l>YxkGC#uzM`lyNi{2bbtAjW6L
z)^ZyQ1c(p2IJkJndDw|!V+z`rL%%FYxjS7C`$Q@=
~+B(;{L1FXx*r&4)8TA=?
zs5h+0U7m>d?zE+H;zsP>O_Q=3Hk8k%nQF#Q73DLF7yXj0Ul1SDCTVVZ8S#FfJRfB3
z2X1n3zcRrSegZ$)zLpr+Z>=jD{ZGL^ShRL*OBB{y+sy~`VeiEC?T!_l0(?{OHfjm!
z|DP{X3Y?(F8rpSs=XbI5`~P!NSlz`g{Kvm>eJ1p6O^bsn<`v&r)cfu(tQ&WyOa3|2o`Hjx1=jv##OX>h9HfbAc;sH$<%W_J@DZ)Zx@~{9lJ$KmF;%
zd)sn#t%yGMljWaYs4BjM_~$gS09`M{A)N33`*ltiTlr?AL&jkEOC>6#hCb?K`q{xr
z;)gq#nq=*X3HQ2K;5xm`I_wYMc0Js1{y&`WbgvXSf31^^3lV6GZ-AZQw0!M;RoJyF
zwQl@s?qV0K7Avc|1222!Ju*kUyyYWX^T-KZEMVDz?}9+QkMk~%`n3RY7D@94@0UQn
z0k{3~Pj%sETA=N@7PvK6`E=9E%kXF2NcIRk2At{FWSx2le#m&^Z1LL|k99i-Ndm{d
zVrP!I$09$SeNmAZ?3Rw-E{R*;@mOZ?yd&+4I$3*N*O%U}o$O2*zd>$OC)=4(_F*;b
zfFCon-}TsaveUbMY`*}!>@+T=VPJO`6TOwM!Bo4L@5GIUb66L?^SM)s_q~gm--``O
z#J%tlNdBPY;A&_JE9r-ell$-+JR5|6!Q_N99rn
z0ACC499l9T^>Z7mN~3_Y9~3*i(HM-lw(ME=b^#~brk&ZQ($dKmS!v2U6(H_o;darx
zi1#?7Y*Ow6dG$40Z+qdsQoq)hc*CFUmAkL*+j0Dlt3R&L!us+m!%Jx(>P=+-EwKQu
zR_gA)^ALD9s`~&s?DI
zLgpU79E^JN#O*6Q(ay}UBYT1&w@J3}{#;zIBA@slm+#_{c{2HJ#I@X0l0b_uy;r36
zy=Qm-wZipx(_h>teW3T{xLrDENVVvtDX4$`yD6RgYlku$+FqgE3x^g4-2in^$%<1&
z`GnNC>EthM-Z*DdDXz16e(vrEjK|W>FICBqn>Q?Asu}9d|HW-0yshYM?sOS6r@20v
z;*({v`gI8(m;7{ACLGc){YZ@Bon9VDJ4kwWoA=(L1)zeJ1rntHi{{TL=HknKh-;C*
z_=WQtZ^AMCC$?`OTwgm%ZzJh@3CE+h)bGFh{JHPV=q7%!B|KdASvi9AV6l{DI{EST
zDvO^ZeZE5Wy8+c7nG}k|wKxagacd0OAxFoGT%z{%lV))FE#fu}l_eZ7|F{eHKkj$+
z4E%O^*ZIX~a2M`a(Wzn8u4dt2$#yc{~hyO!$J
zOa!lpZkCxSP4zDZ$Z$9~(fO1Kl~+i7Y#@4WwGW4HPkz1jgZQ|ZnE>j@4<3Jt^Xn!(
z73KW4Z~kuNaAUcHFNb^W(}d?EziA>3BHDNEmgn&J=O$&Yorjiz9BtGRMyzE{r?)Faxj|_y4&8#c1~%qW+0)dD6+1-(5eB<7ciGmh;%@a`Xle92uWuDncF4GjUG+Z^Wb{l~4ETt2?(hC}X9`74<*oZMn58IJx8KFGD>
z`~d0C(eB{qnVg+b@h#OvBFE1~)RZ_i^+-ft_yAyIEM>>d|CGO1TSkoasw5RrVi?H^9=lBV$cvxj#*
zWN%^m;E8mE_oqKAF{!D$LTlJXx_^g7M&qG;81O`9lbBp43sC*``lfq`M@0IPm7Pc
zQM*l%-fxLslk?;9BG)PYmL~tfqOhHJDIP>(`o9a*u5*O}H?AMQ<#X{iwQZx)8e
z&Wc^~)bDyzIUVY6<+Gnt_oDph1Bvyte}~g1-V;G<<769>u^tD>`|a6-b^iO_PiJVo
z-g3?+j`Y}{JsIB__}t36Rb#;a9{peW1L&2)dq(~A|CKw+|0Vh(J))QPGoG!Dlcabm
zzc|rU+IN)P_t1>|5x)k${6PH~Ff+S~p5yYq*2T|I{zQ3@J(ZvSeO8Uyu}v(|qjh&q
z*7M`!zj^k>FNfl7oBpKoHh{)Vo)(n`J8JnLgDx4^QTvB0sT+O3{2pQ1I#U_AbawZ=
zPGk6eRK*;g;<>7Q()F#Nzr(L-8ze#hK04@YQ46}+e@@a^;0fOF!sz|D&liETCgmX!
zFK*Eqrp+gpY&8w`UrU`57Bp&
zkMBgE?>-tgQ4p0{KcFbTPQ!En`^HA67*-#se|)%kzeHFoUv7!z%lij
zGp9)}!TPvOZLdA{DV`=MO*UPPxb%_xMSOuX?CVum9l`!}zOIVKhI-(Od710o^I;di
zx3pUYJFni~cu))CSqGZ_7bcBCzaeW3ij83&OTEJXAQGx>1zhQ+F}0dk}~9W4XqWSyb8SfnBAVhQzD7Z+Hts#
zQb*vl2Y9aX5VIbuRd|kF0~^hb01w^2p7J#xIPHw-`qRhYubb_6SUe2#L?O=j%3I*e
ztWQG*0-pt|`!CH@g@0k*XwlWv5D#2%sPiqx!*#mGx-9J9H2<6Y))s!sD4ox_G{U_V`}QJrBR+YcpYJrH58DH0O>{gqwiEq-_1oq61@zmg@6uO&Ja>Ti
zcAsw8$2yZWKE$9r^U>^#2Hbx|yZFZwpbQ_-`hn-0A2d%^d?)l^jHCZkSL|=!`khyx
zg!f~s+^p;u@R#cE=-+t(c6HkHloa%H%DOvU_a*^v6ujMk0&xdEWnDjZ19z?8IKBEk
z=HsFYg&H%&DPJfYvPBNKPW#QNu1bverGMY0KBHefBkRNm;J#ya-0qkF{(z*%z0H^}
zFP}L!D}xrtid9O1HZNmU));4Xt>3@fA@6#j-#8uftaW03xG1jEp6}4AhEzA;GWHF<
zRby?vaW2ezLR8B*$T{s??AD9x7j8~0p}e4LR=q5qjsN9}q|M{&XkRU9#Sy~$@j5nf
zq+hma{;rb;jlBLU?i$LEmYOVX$Mr9&y0Q+UJbUbx$YxN5lA3sRly9E&{{0E?bFHJI
zI?+B)cf1IlCv2>D7(_U_P#leka=y3{T$gY?OgrIW2lE?`cc0@aW6
zGZiDfD6SycN%;O#RhbRx#iL^u1dzUca`W9A^2>kGpTgrtGHr!^wMAQvc?zlQ>6w(vHXb$$rRQ
zw3v%exGHP7nCk8252aB3!+J~Zd}`&K1025f>_70F>W>e}$sM#k7!Kf^8RjB+sEPdd5r@(Tt3}m3`aW$9_Gr+^ei}h(kr)~yMBMm
z77ouC47>VCoyN)MTp!BGO(=cB`LDnBm2&O9
zaQVTtV?9-b!)GN=JvqPj?36X!IGN=Ia`>%hrqI8y9g@i5Wy6ZQ9Ns&ARQP?BPxFPi
z-b6K@yI*tfYz{X@e68ZzKYJsAE6<2WI0~M>><7}-q2Ih4O&ouD`bqBoGi6qA?Y|ua
zvkCQ|m$q^BF3&NXxDFp<=F0a3fd#*PkC1=99NnRDy@SUGMY0l3PWz2eP7Tp8)ZR|o
zjzE3rW;{M<3S9>JZm_Vv6{wVWaoT2qyyZ
z`DZp@!5eRNK
z`znWW$Y3IW?^mu}C}&Q7&0^vE*!2V3Kxnsq>{`4~(9a1pF))s&Z*t?fIOj0;{*RHx
z@j;YZ2ihAF-R2DU1gKra67IRmeUR=D{NN+HFeIV(bH;Pyo|bK;K>ggd_x&*e%GccY
z2_U=Z+@wMHunz16wV%izC(`G}er`Jr!GAmuKYa>#uN0T5d!bKj#2zdo
zyKhgd%5xf@4382$($jX1;r&|RQ$0>=>Vh_%tm~!cx>+K3lb$pc
zHXPyBbKQ#70n|^k9c7VZcOCf?&_d&;mHfnp+WoJ0Lhf16s!_KMeNlecyWz=H;5z@N
zw)`*X|FU^AU*-aTe~`;gF#}F&QjoV)nMaL{PC&#q~6vid%>3F9iLuyHcFw
z4!uz9o(4MM;O*;ZMr*&#o{pDp^;ycf7`smGrYYM|SA3+6mV|GhRV
zgvR1J#dBVT6vw!(&SnGUE6Dxx^ee@!5ZRE$euJ_yreM@RLj^J2Tm}5?P8)-LDxFIeRUdi``{5uBT;|0
z=%NqdB!^KYTj=|TWgnF990uOLSmPi&2ji^O&C4MFv2$=`x(sk~;niwGO~}V2dmSMB
zwNuwxh4ypB-+fxP1MQB#xbRIV=+58=3e?ZsHN!NYfco3NYAQhaI+GOxO(1{fWc3So
z(8h>5xW!h<~K^H}J>KDW9Q!e%Yg`w*}N>m&3(npg-$ce8>-6
z(my$Z;wO(T`@ZEAXxe{;qvJq>?m4<{1(niT(|-^2Tc45x^*3bZp(XVGUc0l(yb|kIXx!8z+;elxXYKWyb`CHy?T-UuH{FrtB#=3$Q@e{1M
z^)SZz58?G)|H5Qxy}0<_wW+ioWNV2G(4g!3CMHwdht?_2m`0ROiD=zSwC~s@knU_0uZH0Xae6h^Z_knj|p)N%d^y2hJb8
z=-3k|Z0L<6x~a7O`@T}=zGKqyZ3^diJ13hrj_QLGwgV8r&wcL@5FN&;Jzx;^XIb7u
z03Yn*@9;fI^+{$!phz+AHmQ#%diSyC4U%)1dAf!8Js%C$lUzdKl##^upAm5Pf2f_v
z;r~hI;hY}r62&oZ$fcu?3vJ)%E8LF{+<)br`3whxUKs
zt}Cr&`)Z2gi->B6N;|6a`lU(
zM{)S#SvbV6k6d8F^>g%p9{5Kiv`bcRt_>+RdxkWpyIXRgYp}nAN
zU(d-MmCE7jQ)DM|^?wS4II~@WfI0k%`1&=q&&CO_e~H~3;#l
z!;1ors@?;<1V1A|STC~%-3H%MfZKucHooWR-~=H~{P-9w%J0Yt>tDjUqjpvBOVor|
z;_*}@*Ykh-B9tB%{AD7b=Sd85%|LC$tJ3YkZ^FM%l+ZTOOsYmbUKWlY`+xaVGG|au
zRuEC{Hjsz0%kdxoFOQ^~+;vj_=Q$Q~^we}6COubQ_MWSXe*p2R#tx
z6W1Zkcc2~WpE|xKL*;QgLSA~7kpHz4(--C8G@PJQ=zys((f133$7WLYhv+C=dVtI2d9!65sm&>*l2k<-VJL=C1nzFCNwRG?a0C
zajx)vJ9u4)^Uq1ZoJIQ?w}tabv$l)lC0s^v*SQ{9&C%+6G(f;b8p8RZAtIbFE7l0-
z&w#(R+;tZg3+F}9Z-_&_8d*B2eamK1jz2k^3{ZTpMNL>=fYpHb_uLZzhy0sF
zVf*`(h5X-(YG?>?e-lTdA@~#KiG3h`v0@tMPe{z5fkj;5jUybDA7Ib5n{#jmN25ya
za_f`L7-4(1o=RN#Df7LYd|KUUjvt$|4L1h9EEV?mVLDA*tj~7;x4V+zJ_29(N)`VA
z_OX<{aP@yT3g3@Yi-qr_r<5+&Uies~&w$+e51tQ+ZpmD3MeVn(-TjaBP-3m%2JxR}
zJP{*(satLJj^t7%Pvh40|Dt8Nb!*@96MLxK_gB&ENO~p>M;YFk`Ua
zJxqW<3OJ&3r8kwk-OlC4O@1c*VX)gLZsFQ%@D;}&4*hd&l>yN|)x!TpH+VSrJiT$k
zy#HTJW^;az^q8TX9ta3wN+J#_#<#G
zCRhIjwLkfI={eGK(*KQHNAeerl`khfDz7D*L2?4+ADmu@FmuZ%en+IqMUt<7VbV|6
zcYpkVHwNv`4J@Pjk^zdGedfRL5N9v_OrC$5a7)3YEm*Lj7n_W>VWd5Af6ie3tLyG7LG&t%BjV5gggA$@NMtG
zzg8S7%lZ3CLgvzWw&p!K-1+5I5f8Gc{WY0(FX(^#kl`Ic`(q`F;TMU@K6$%`!wnhs
z*Xg;;gChTs-M4Ii+Y=f$Ge0?Hx?bh_Ci!lZ&v7eiB;G;6Rh0Taur|7Za0V|pyoYeb
zpH-=mMDJW`ezpMR+w^vX(DyoOc-hws!B@?f=_Y)c(LH@M*_34&
zf40Js{K%62GpAf8o^S4R4OA&+$$jeol%bl|WFOp;tmf6PTuuDNd4d6r{qKbqnhoisdQb}
zlD64&pB3jqt0}%?*cAPTgcrBW)|f{4yCEnxmChYTp3pm~0(yD)w?dj{Zlm~j$$q!$
zwA?oawAjPx7~#-#$w(twA7-8(b&=@XP=!I%-xTd{nlzs-s}!A}euyhtZK8S9GHTXb
z>hJUkL%2ARmuG%2BKdWD_AjLV?e?p(Cu(pYV*&Lu>weTrqK5Wj7l^)1SgcL|n`WXv
z=dZb<*s+E1f7?f-c0)Vvr_|eTf=)VeXg$4W8e_&g(LU|u2IX`bFF{1K3E}<_?G-$l
z2NkFEduSeo3v##9|FCS@m}#_LRatI0F7X$pEE5bdgzxTw;t-mY~%2zdQtId`hPx!T;%Y%
zzeEW4KUq9+P^9Z8C*7AOx>a!fLt&?
zn=$=y-Cy@APCsb;dj_C`xS0q7LfBum&E(qMB!*+KkSpJ9#Nq7v7NpYvZ{Ca73e?&(
zg5z(6KjU!w+QnzMa!;)Uj>aY|;q;)c)dudl*Q7YMQ~UeB2=TD2;!p1S9pZ)DI4;y|
zf~y!fYw36Hel?F~aP`-Q5g@=hl_O(0I(M|tPdfNnA9o-7hcKH^Ul~B2X5`s=Ih`v%
zwcrFtRRjaL^0Na8IANU%li}#2ub57d|2e6Y8?T`AIUN6PyfRm>DBr~Kr>_d}^tPF!
zxPE<_tIL%iSFq&d=9VM;9_@^LBlNGnoWG3Y7w2no^`*Z+aNpZ^g!L7^*vQ;`nzsW^C;JP3j(SqieUsj!8zpN
zAFaXggD<4+fNsNELiepl))=DyiZR_l`$L7-?SNJ#o)2r0=;%yg{paVx@@;0q_PmY?
zc`FW3W4$pKUVr+Ku)cksu>La858cL`eF1pfQ}d?toNW{0l%+Y+*x6)fW31&`pV8+z~Tjl-ZvxBFn
zDjr6@homI_GSPF$H(^rIqQ6bRJmO;#f^j}_)O|nAV8o5@sn=1t)Wl~lHym%6iQrt|
zf9K`5`r%yQ&o#9rh|kt8>v_7)g3lBZKW_0+!a2hiCx6W_2LHo!S#T8WJ9%BV&DY?^
zn|f{Cf;_D2FH1-5MclKyhw`3buAR)hB1zxj5|7DVeEUyo1>!GL4jht#-|c!@tlHMG
ze70g{ncM2F7PfYc^Veft9qi+PD@Hp#un%jo+j0`(#=j-Ko;V8U8=J0;oY*+BnZ0j|
z51Ia>m7UNrZOq7l9?vn2NiB!pJHJ~jx)Oe>0fV<}9n!+q2c7VC?QTaNRJDmMI43yy
zj(P0U^}sXjUENDg0XO;f*6+vp#M<=HQ>!I9SxU6OO9ApAj2OBiod3Iv-AJ>)F`*4{
zzq~$`hCet*S220>Rh>?zepoM69OvwI`Yg(+xs5pKZ2v&zuU+g!wYYW03Lbmmdi8Q%
zTnB4=v&HaKMhBbIv9~xjx`UNj`)2tf|K$^#QEwyOB0pt)Z0306t0=wyvrQiPCmw7J
zSahtnok=?#k@gve@B8WX{2Z9k!u%V4)hg+
zdsa0n)!Gl5*}NaS?$6)(iS0J|cxY2a8ymdN@lu>;JKOE;o2#1*x%s)Jb+I3rr7e>T
z*wMngmaTF0Mc(a?re+1N9NO8*9abqB&u|~_C}Vr$4wm?-Fk~6Oo#ij)Ra&>Uvyo|C
zN7H=T*)*ja7w47WyE8A=KM5D>U}*>Q{l7hHXN?IVzKW0A*(Yz6k{J>0Y@KRtPXF+B
zHgHOKrrq0iHg*52_tS2)Gu5(!`rI?^tn1M4)CC^+?%;z*mrE4dnVXD9OR{%6OKF)p
zt~{=t4fEPFFD<4W@t0lZ$DJ@P)!XLF$hR}iBS(GjXSK22xl7r)zin)%w7g5>uQqlh
zUhe2O91b|<5#uZC)yAyH?fJ}V+gO8&!LVnV@Ig+~rDbY($SXU&Nr5HFlgk5pC4UlCOV^Za`jl`%c#Z#X9Xc*Do-;
zk38%S&ii(*^K4~#&!m)}>$b8{>c57MM80?PHLDy~OlfB$kSurVp*A-D`_UWT_#RI7
z%PZrRC%3ZOTe_a7;{ROi5WVe6NE?&<_o&^Xs+D;q)n5*J`h_jrx8mEEy|1f(-rVhsO)K&|H~POl)XG8?jve0D+sbq#+Ux2R+t{SOxpx|nzul<+
z%W}gNt&IQ7&BXkCEAkRKF0CHY#(ubEj4YODW1SUNA_tJiK7Od(+_~FY*%C#E%ni?5
zS=f$+jo!*_`0mJaUa?jilkrs?eg@Cw#$5glb!lZURXzGv-)v=5UKo^47}>`5c$dlT
zp9Z-KY5!>n=;z42$%o3muo_>xm7*&6?&6|?V`oEK*@nxTZ}n%kvU^GB4*Y-4Lzr*C;~h|
zJK_{Pv<*UHb@*hTFtiGQWqS(Rk`mU<)Pv+dV7w(v?TOZSm;
z8vVVMjSTL)<R;|2S?T5JCR}8HtVmsw6l+6
z3Z%^O9nDK;3+jeSAWxsv8;wIx+L&4KAmw40r`3w_A-$O2Eh}sWw>Gph%T*74c_Ob&
zQEh%<{)l!K86Ey=qN(rnhGxug;j7*#q^_&*;qf>E}Xm
zu2|$;QUJ~kr_baExS-zd>6!zt@O{z;yLze;khgHwc9Vu#D>|6_#K#{uX5oDGvVrH<
zi0~NS%A|biTppWuezpBZy!W=t=LJ69-OipJYO7Ez#CmSzskZ{_fS(^8O^4UihBRW4Ftdj}KUdexC5M$qVGMvb@t1ej?9MW9DGzTX~RM
zuV1qga-*y-X^P>wBJX|?{e--OQMzuU{2{Mi74YGED37_wM4Y|819^*X^jI!OzD3hx
zXQ#XhgMO0|O!7ekIf@)q2I&&+Ea~i-GKU+se{m8
z^<$IoF4~OysjSpU!*^vLbzb?<*#~`)RP>?){d{A%wRz$7#|3Dod2eQOJ&(SdCDcd+|;wwGT6M~wG3&fSLm(kdg3<0AEO{rXj*
z9>`y*<6s&gfPQW-DD%_W-@&S8{*CyVgM6A*X4g_MUswEBo??l4n0uscPT`17_UPNi
z!238iANFi$btv)@=xmiWO9Gy_x4-WdMB&8>-4xTwUbFc
znz*I}=jih*Q{2lfyV&AW|E5%2!Slv=eG2-(V-rJ88#vDEWc4BYoinpK*)^q26NZ23
zWd98^y}ldyu2S4fO|ST3KKIq#8I!|fa++J6|Mc^i^+|pu
za=;xOf5YxsVE(jRQ+bB`Q_Bz5KD8|3v6$T>Y(@RM*!3~I(
z4{GaVMTMUq-B9R4JnLQE!h|l?7Soy7qXzk~$jvWzz>XQ;5vhlK2L4i+qLy1b+1ZSN
z&(HHaVW&JDX^4DQ+F6ECB86RSR>!)^=XjrVF9#K?3Xoqc;q*ge{RR7qkP`M&i#Dj(YpzJd>4jqNPZvt
zBUhgv*MfiEwo-FODB>UtLnp^PKz^3o(}sn>r?SsJ4_Dm@yFlFdSXU+Xhiz=q2O}=F
za#!|2!EWsLUYAz8brW$qen!J47IiWGy5tnOk?>0|xmJBS19rdTe)W;)&pP8fsRoO2
z|CVXR&wh2XjVihC4q11x>?}KXkESl>e%tZbLNDyQ$=Rm;cOQ0~eAf976OmumVRX}M
z;M!k{2Oho2aGvpB^2JZn`E0c8gsDH5@!288LaQ%0?v29{+&!hzlkiVf{-@AWj`&6Q#_0hy$OEOm$Flq`{96s$gIa;py?%;F
z-_eJ^=g%qa)CGu#*SVTnfOE64+Q)xxvgEUzgc*7kVSKi)K{=Yvb%uA2n;!QN{d{V^
z_`EXmto&Y|wgL8o-_m$#v#oq))zs3wH51RhF@5uB4%}Btx&^(FU(7qx*+jVt=%~#|>x!e|fwHW=}W8(9^{wM5-MFPuNHhflc
zc7W8jBtCn-*yQ)Qd_J3)mr?owanid6cr7o-`N@_^4<^pz
zvqh07hDsq{VBw8>qI!w~=BUqGpo9E({FaJIRtp8}U)%5{{{jW<(p&ST2ZsyTmh-Nn
zI-nb-%g_0&DqxE)y?-#zS-=)Dc3~~@#7rHu{^W=+d=@ru$Lu8nK2!6I%}j^fwU|GGqwXo|9i#GCLx5eCbRXhZ{O(B2gZPivjTRy-J
z_xHQ5bA8W0=Q`*7^>JPA_p5uYd#`n`>AqiU?fvX6#5i~l-k+!t|Fa-xXcC3E7X_+X
zPa=!FGkr47lPJhDY{kc@NmN=ocPjzjUp#8c#iTVdi2~o5tco}{i40}JR2bm>HZ`R^
zZJ*$KQ7-OwZB6j|FKFSgihyJRXv_WEAO9iuY~tKZY_&FggPcsftJUkUq2^NH(&p@4e-%;y)8M_))^YW%Pn4#Ri9&RM8oURR
zwe(b2rSlXr|1;u!*Jctu{kY+CAdHi2%-XpX{*#Esc1dMrCCsDjbVu0%c#mWCea(2r
zDP(K1_Wm1zDRl0~`NFC_@O^i~0?rq1Qz$cJkxFpJ6uR!cdgGw+6q0|s{8AYL<*ZQjmw!5iF8qr2Ka>sc
zr#^Okf}>;#Ne?TjhuTje^`&9!&Yqh>w;6r=F2i>z~CA&(t_{W%dEN}
z=JyNgQ(!;84`kCG(I;l`{j3vKpL>_WcfEe5$}DAGHjReE%|pw^r;r=(utyZfGkx=&c6*tsWWIe|!p6
zMa;}H!?>vxW$$8#`QTP^=bsUjf6=XY)>9U~C+7F4?C0ibIInHz^Z>hrW-K?)a89G&
z_r{mD!S}zkI3I6$v2z-E`-smd8BL?unKXGj^=ZU-M5!lIa~hrQE8OI^W*W74F=w8W
zn?~k`PeLV=ed|TkGFNLg#KypCeN=%EWuqg69SH|y3;qqUOT7n^HMqa)3{{MFIZ
zXsOG9V8i8U)WK-+=u_7;%BUMW@x6ZOGM(-kH
z?CzFLAe;OTowBhsI$7%Ga
z!TBX0^rLUMmD946DYXCD#Itai@2AdAgsQGB?V=0CGcX_bhrH=8
zg>^I)tx&=4IEBu)>@x0y_leznxl~whKYYjK&!O`H@c#M5JGL@xp~JabPu4E+Q1EAx
z${GhCj-1-xA9*hm){W!+nP96asWx8QR|1_m|q|NsAgtPBD~VB}<6?r!B`XLZ8ylXzKF+51Akz=WvN227DrmYuvh=VJm)`9{;gGD$ZZDXb~x3$wGNdjO7tj
ziSMh4;==z{E|h}v{@s0lij6^pzvpe!e=Gp881p^^9;oGO8V!5wZ|8xp<$)Bjtv
zU@utnKmAcP{rCN;AOb(uBY*$z`ZE?kXas-%yZwp#^1u4?Ed2ZbPJjOQ{k0{49G^q2
z|9$?}v(1qApTGa~z<+w+KRxiD9{6APfRY0Kqa)=7+`ND)3%G9qcP-$K1>CuS+ZIrL
z0k9erJpc^A-P0bLjHCo!0g+84_G
zFMGun@^37l=R*GP1-W@a9$k>NE$ov2X3Kc
zd(r`P^l%d;>ki@z=}2awJxUAZ1T2(ukD1c@bDYBJwG=v&0fFll%Gt1&N`D_fVbMtn
zk4R7`LI#A6co(kAYoO}A`NYlvJoNy{S4pWOI6W9m^5v|PNj^PM3`U%e92k3GY0}Y;b92xkbY!ta9Xf!H
z%7b%ZMnk`;4a~^{t}pyRn@@v
z!w>)+L`RPk0*x>({l`u6Q;T~Ee!3(~+V`uZK^N!Wj!=Sz(~*xH6G@NV#3_a4??wnZ
zoQ_ObjzYn7RP538ZVTYvbrDb?9i`OlKZVgka^Ne@4}WqAI*N`?o(a!a0u=ea7Y2lm
zoYoGyVBGW}eFW>>Q15_=M@Rmy<`r83W$XO8Fq$%%!ioZ9`3^G@#y#3k)Hci~^Z0^9YYwbpGXpzRB3cATFn_l-*m
zaJy*kMcgipIaRYAfUXJ^ev*J^#cWuy9a7Axbq4@5>2_AQKiQOmBm@Avw_aF;%SpOa
z%5DaDpx~#hIUo<`!1rl+I(lEVv2Yd0@3@!WP6AA3m_1+u_}J(n!u5%+W-!2ZzZCum
z9uVp?dDe^VT=iA`G{&)ZgH~LAS$^bCod3}3ss?V)$ze4kJpP92`DFg(ZF^ye?flHH
zAcXtRj`!Cl+z+FB{f6-M*INF}Con&TY@)z2(z5LKVK9E(F0vdM#_L(Waz}Mxa&L6?{6`+q~o}Eo(
zuE*&WCzHuMA5J!_!SZs&oyAyhnOh%OhZe^)i*Pz4OA>?@I(pPMz5?fe6mddW7CqTD
zhUEuJ?>6IpN2MGud*kyfE)6}|oVocGOg%D1M
z?+X47!*;Qoq>}ahXK$SiPB-0=&yUf8I5IjiZ_$2&WlQ-jeOUihs~01dC63_8Z0h+>Wv9#anPNbo9g{^$y1OvuD+?USZG(T>@}bC6h3g1l+dYeCAalq+4uhq_BtQ!p>
z^^_EUCjM*bliQ^HnNGGsJb&L^Rh_}@6LZfC#QH{uOCYeox(bP1k5S9HkX*k^+LNGJ
z7z_iX|KWgY0gr#hjEat&1u6M*#swk|FTsop#%mi6nBa$)sW_1`%tP#a_#vJk&oN4G
z(==6%TR0BrunsF9KmzztyJ8YAK5~>3{R30rZXjJ_47wMPt0f#Nfc|pRB5{LLj*1_3
zyCz8gjcI!kJsS&XHs}?4Krw)eaJT~WFaZEGB!r8OuHU2V5Uqo1A^p@(O7H$c+*f`_
z$wwrq_^q>Tp*#$rzcrRZ&maoh(x^DTl>^L!d<|PlzdIbv1KBf@swc+Pg2+dL`2khbTN}K&68dBl(SMD2(T)`tM&OC4bDJ^c!H-;Br1=dI6YljhYAFohbcyNlI@V
zRtwH=KS{OMXTc5#1|7WDiVuCo-t6+nLh>%e#sHy1E}L4SCmzIYBQoyC>HR6MMpw`~D^
z7yA8|HI*K`gz7h0@E%x?ZNZ-Pz?!h!^M|tQfe9W^Iywek4dYw9tN{-#)OTaSu7W^n
zJ=Pzl%1v4rhwLV*oPtDZz7>a2$P!He
z<*0lafC7LI&~!-p#9zuE7T@3`a%)Bl!JM|uU^pEuUcH>i_fBvVtn~0Dm^^=O6V*`f|Bqf?>sOB;PP4lc2>b9a!nWXD-RW41oA7QApBNQ8f%G?2kpwz@gLO
zI|nY%LEwK3n!zyGcU1yi0$d+1OYHIfWJctR;l(5#^XOe8@htcs6~`Km|0VSwy=y}B
ztQ(#{N73Q;!~Ue6ck0_o{ogOe!+=4Y${HX%Cig7+F)T}pF~Y#mQNN}w%osYF<~f8g
zJ}vhne8Se=L=?xpi?8Ns7|-Pg61)AD_mF;A>CsN&3Y(1+1LgzrEJ384+p@BxTxaQn
zByO%f$9fW%n`1n89rGJ8RqexAULNQU9u)Y_2=-ibWK-V@J5f5?b2L&8T$l0eEN4~z;RZm0B%608MoVNfRZj#$1wk@k`#BwxHvnC
zeIHF9@sC#)=IP7<8!ddjs}v^_;J?iewyy)U
z`rthh0ytIwWnKd?%gEZT0My`;^V
z0`59r9s39Ld*<}Vuzep6CJ*lie0I~@47Vew)avYK!1@}Q9BoJscE7R>&$~bTb_ra7
zw~ZGGJO|vap+K%4%Q*Uo@x9s~K|fNoMce=zU%7d~nm2<28^
zAO3X>@KX98j$FXqjU7>LGQgAEKQq)rzt6EXuQ~+dk>2d*=nna}_blSsE(dny2yDaS
zp|&?PXD8rRxvq9xZd8F^74G*VrYGEG0Hu^QH?aZUZ1Gry_jM0tMs0CBYE9O$;C4Jc
zrNoTquXdr*#Vdf^v+2MbA>Qc!rm+J?SriezZOFXxz#fo;*xDy9fgK|v>kaUDf8C1gsI
z<6G=6`kML3y13;+zorMWl|ZB+0{msJDUJI>Z=>c0+)n;GgL>E>yo&l;-3Rr<%PHG&
z{Skj125`TulIJPKdg)^f7Pvq4HhmmR2EC+`>gaU9C6Audae0~2Nql%;?rh2LjrYx|
z2MyL^I~_K2j$*sceKU5!>s0!{TjJMdxs8Y)><@M${)PENQ83oqAQB7#3hZ+2fyfMY
zx3>|$ENd$yjoW2wb|4bltD6u_&Lws?^8k~8@qDz8#0BxsdMy$lyGpg4!FFHF&N0FI
zi+8ZD#(LLdce7%=w3AB``>Wmyp$%Byv8mJt>wj4C5~v#R$U4D|STC%>m7vW8xJi&1
z%N0rdxYi8E8z6JI9U$p}ZLlo^IUtRN#Dj}%3?$uSc{9QJSWJN7++nI(7h~{CU0^~m
zzjZEihD#bgV`$PBwS!X>a_At)|evy2+N@`ze?LhU}5fVRt7p9Ro
zbTV<8q?;(faS`YxZ=&M&Vxq_}a&QG@kG?j=vnqoi
z8LrFW004aPi^6kVc;f*3zJo9vfMSi*b$?$_?G#U^uH$}9?RU(lsQmmJFx@~u@)cD+
zye5dqg}n`=e#ZUOc&lbZw}PHlAk`nfoDjZ2KC)Sxw6iz@jzd7^2z4jvCjz(#7L`!_
zWW@{3g8X|6{52SN8_2iRDXfKAhvn0e6pp1+cx@GhCD>fZPr&pG@cwh?X2AK~l)dM#
zQR&qyDZI3B-9hc$M9;R6s{iIZ)gNz|DEaglb$tP*Yf#?7CRKtV-jp8g6_xM%nCdS_
z+`XWOM+@UETnu3Ig8dWta*&1BQ>X;249jmYO$A)HpF;mp3Rw&&oLta1hFOgD3SsqP
ztf`~yxVW&t*s;)GQw#gWZEci&XFDnTP9{*Y+$-vOHg!ri+(V78tE3t!cYF)g{*ZAh
zKWL$U&k0iH+#jT_m$szx&%^4(_K2lY>4(!OnKwz8q(?98Z$3>?`ID!pd`Wf6{vA2g
zI=taR@#gb})cDyhqU^YQ8Qd9^-}?*S0D(9VMy*rEKHRauPx>g{ZPGnV@`viXp+H_7T4dxryw=xELy+qbp)~BLuvl=^#m$S-+GZpU7>3L7t`1aNrsL
z3W;3S3T_-^>!+8AoNr6{DfNO_!dI^xeoN{fIUY~)FMHz+J?xts?vV7}Cs~Bwrb#U#
zvf-QOr2LBQCt(Ib`xZ;WOo90GdIc;
znDN)fc(^f`#GNm3v4lrY`rZ!2>F>|4;=%ZOTAai&#
z-@|b+#jcqp{#$IF2>WPQy%v^WaBo2!B^x^cMygcyo=%cZ~9HocVCQRVM6Ta6DV|;QJDs
z?h(0E8DEzuH)M+QM+@Lh=74F+r(weZ`((e537BX2518-3aXZWB`dWb8_DT(9o5J}0hEzk@Lfm&2cP5hev+0+j>-#>q^1Jaj8Zn)SBJd@+R7^-1jzsHIY;goy6;azqXT8n{S4oj_dl!ks)2npS)1pDo`+ih<^IaV$q
zjpIwlj&m?q+&v_H@@=Z$}+abSp&tEA!W#GS?%?~iZzF_O)%?UA*
za6iDdf>T))&f#r2XiQ)a-~Au!s+s7BbK9GZu5i8cacQnnQ#gmeH=nc#x2IG~D_t1o
z9oxpEJ|=PKMU>XF}LcnI1Pv;AsBKMm#E`t4e*LPru6YSQjtr>n=JtN)LP(Ro7bw-JwQ&*=g7JR?Q1R58o3Ubbji8z1?+pmvpiQP
zY=?d-qR*+rK1a<|wXYBQE!pVyUQQ^_w~eu43i`QoMXW|X;Eu()9RI{%z5IxJdvhbi
z&t>;-7(>6GzAY3V1>^gCZETG^;F<32BOm!;|HI_+vrGv3b=h%&gK%AT*3k{_M!<(#
zB}Q3}LA!gs6l}rHPxmDH_kx{S{}kH8q23Qxts%$3&T-e(rvD&4^zRNI9k9#(=+tI6
zNZ&4;zVkSwuiCe+Oda&&pQYO}0?G?M+ocV5?sbwl-vxRM7d>RgL9a*tnV2!ywKPsT
zFAn63cf+0+K)t03AKB6&-EBa()EfH1Tf#Yi60Yl5^T+HVloMfAJ8=6X?8iS>aFxS2
zwI+2ki^6s5Yf8kdVSGpaSgNmp@ipzsFX)AGFTYt|UI(Zj^fXKy%Hh&$zi$rsVzyLf
z2Vm?;f$x@JXTh||Z!?fN9GpJjeW}}$Oo*W1$A7E(O;NdMHdrrI(Yr+Kdj~0pynqHIxT?#NASv_ovd*U=7iuG=*55LFd=@^8UVZZV`RpK}YU?uAkxIqj&9<6o6`A(~quHx&r
z@cojP0Xuj6oZ+?sd8c%qHR
z_cgS&zYCt0bY&d;V(*g*
zijqp@o=nYePPVBEMQGj(&e9E=$zaS6$}lFZfuxks-2hOj;UwZOMeHzal7~KY8QZ{=
z(IQlDkoZbuskHwmZ*?%zeOcAXJmTi?vj;!U2u+jJ(U1HsJXsfv@ywwPZoi_0e{EYB
zv4m@2ebGL-?D{+oDhZ^ZVF$ef@}daQ9Wxc`Z@Zp>(}ercu)!1kkf1A^tjSlWt>()0gx>6-R_nLHjcu
z45t%MY%sn^q&1HXhVp=0^usqKqi#Pguvqybh0+67ct?c6`qz|uzNI`}i
zn+1|;?(^xPIRZjEgHwGHEeZxL7u-zu&`e0A!Cqr*
z;FyYnZE7`P>~H@}(fqHN8ec1eU{Hem;O*28pBRLqPtcLorqG(daFaXn7o34ES36;3lpqLQdGq2b-wiOB6_TN;L>842dwbEnW
zL>Q#|@xCV9KR!xdbxE?fzAD%!s_0#a?JkHn0At$;7z@c!<$AU`=H@>%OkIEh1
z>^BCW2kK#=d!HcV@t#N9|V0sJU5j>KupF
zC3-^q-WQ22@l`5OiocSH-EV>b>uz)L_U528W)LIrN0c6VIV15RHBWU-WI@}VWK
zMUd;(qro5a7{*slSZPFm4fH;npP$MipI^W3({Y1__&&fCN>rOrF72rRm76Z?mwjTT
z?1`$3?WmnL2qG|ayKC^%GBgr7MEK#k;)&^e1tcc(_CefuxLcakuM=Cr(Fy5Ar0uNF
z%yd(3S@&h)jx$r(87VMuv;JP-OlC&PHj*Shg{+s0s28ft=Uf2Ra1q
zuO6sJ?la`jU=TM9q9iiGH#dSMzE>!hy(X}YV$;JMld$qpp1zX5)5{#Sp=M{U;(Q{w
zedcgerO$f|u|UPrhh-tmXZqtnQTJCti|L@HST_E5FI46{xK2p_gC?&+vHs`y-+hSj
zzsy3I=`XtAND!UONclHS_Ry%zKw5i
zsDowl<9L+xb-$q7O{xEleNDg9iZ~AKshhAmA3Ei!_-3#{
z^tfe`I#{y3=DJ6TVp
ztA!4C)>yglY)I!Hfkux%F_w#4=;OfMRBvOkXPJcc{Mh_1uf)Uz{-RBV6ov`1)0a#l
zjpd-~5L8IUSx%z>LP5~1v;}U8*-vBVQh2cSGhwBSuvq@Dc^*6ca$7bFTw6kXYWG;-y+DGdna`isqz_mB-wo@QA#nL4KD4PuD@QS52Hxn^Lt
z>O2YxEIS4SLKCk&Wqa&E>myxsfa}HbG=l3X$UHM{f!D+CdZeRL
zmdtn31{Pcnd7@!!*kNjUWDYB{EV}IJf+#|R6Q7=WuKi6b94C3lg1~#+a{b!idSid}
z%YCCCeA^WyIyu1+L5nYQN`W<00zX5JDDTtX&${lYK0j{NnxXq-yP=#l6lglx{g;HB
zU9*7BYpbmEq5JaL4>Cv2X@35C(H-uCu8IYcSo=J3U
zr+p!+<|-SkM8>Uop%F^P&472)P;2B)d^3g|lL|`A^wr`&OPO3O1&vD}R!|408rnLz
z$x?92m_9p2G=ATO6D5v2moAv
z)IO$GgHs31hYJk5_j2fAOlx0r%|x3`@c36OSDFRZq~hU70+Yoz#Z24lLMyob#3xqV
z20V|KR%qwbu~Bg9wX8OL0w3<@1}t8R4sP4mKFNLU>*<$z87(A*Q2VR#pK2i?M2KF)
z@Zf3%2ac=P;LpsS>>4H(=Ujv%fDofAvnx}yS-h&G6XEs)FMaMiw5X|_>+i(1ODovs
z07LWF1m|$7w)WB5wJ(VB%_V&XEzu)YxFCPWV^1VOFaM5x{|8b%S8N{{0OkV5!Sbb+
zP!z(YeuIdx0q7h0P09)k{>DJoia{b0!
zxhD7w{f0ojr?>6k#7Afhn95mr+^sM3FMkwYYJ)p;tf4AS)%vg&`qt;wvZ$o(b(?F3
zpvoINmqQ%YSLD4@*7Dp7j>C=A09IndYvpH!nbbqb4l%
zL8^-f5|O`NDA}47_b%O=h8_O`B~n}nBr#1FCYl13y1o;`fmcGFY=qr&siVthcJjl(EFDG+
z#0&6X=KHsKIZYuFV1&KxXDzCRzU8Z(k9UiiY$3|(ixn05+s{6$=gWtGu3DqZu7+)8TN5g9$E!g@aQ3Sj9fTQ$b_yHaOcmM^LGultxraZ59UOhDg*0VU8@*rlD>O%fMNT3jSV~S)hWNoceoRX2
zQ1719P}9W3X34<^>nXnCbnob;lrHf=NycoBkxlBkw~YE+&VYRdysBvg^B@sfPOR+ah)#m`^@8|8d4GH$B@nb
ziqBGK#CNVM-e@xf^Y`wJP+CR36&nY^KyTaTXMdu5Jw}%TfmgbDyezzt@WmhF0oelJ
z5Z6Fn3U79?>vvEtSYkVg!t6R1k>;#|CtxxBo`VKIND#6!Ds<^lbWC!lq_lC}|9YI5
z5Va)n*!{&N4)1swMOh{Q(2R=a*kD-FrY8=4H?i)0+oRCZ^^K{(e~kc9ELb0~#vvB&
z)M^vvC$eqq;BBlJ++QM`q;~TiI_;0mtH&g}d|KEcIr4DlFkX-vG83d}A<^X|3OrLS
zlxbez;$|P{`)^^z6@iSJ{{4J1X}7CgRT)~NvJmH`5G6lOmhx_ohYzOUblpvd*k)r!
z(Y-vC29=+J1)a@AK%*MtE7(t@De2|hEKZG+P
zPvowe0Kpm2CUwn-dEbAn+;w9<5HxkIk>m@U&MH_8S-tEZK*P>75oZRBUU+iH=R#AL
zB+ei3+gc?2-$lWSMu7gmF6RkRV+DtbEt6zBTxkcDWV%5ITp2aCXRW!LT&FqTbaE^F
z-(~2+?w>Zzq^u|CM%36~R~KPJuB=f}KaF}6uk9>@>FiC(QFy)7!+Er@ZX2+O
z5>2}5;4ZZ{mP+y6pO5d4)q?&HJw+LiG>@4jnNRo~Taj7OGlQ6YG9hE|_TLZe`?aKR
zd6}cmevzmBO?Wi*{T!0lFC3IX;$A0x5jti!qcrKO-?5Cq-Mm4+d=IucecJ5?3uHh7~J8%NFbXPi{LYTpQsrfk8tVNIk(_>#!cl(foF44mp
zwdVs*tafJQ(_Ohk0jdOb<_b6&Sm;3P`ay5`b7`nq8stNMqUiqq>=>f3WQpb}mw-JS*??z%`q3t
znk9Ej+%Bn*IGhZe$1ODZp>!?Oh*`4;di2M4VZ4l@T>JW1b~onubc^1r(C*7FIA6vt
z`H0|fGF)GJ14Hl-t(6z5+LA?p4)wkfo!$xuwyf~Ulhpq3kFJ0HYMh`F+~CZ8g&s-#
zWhrK~p@5mWpS|JVBe3%k4uP#yivNJReb4trenK)Ldl)wtfs>o0MEKnuw7QV5SPDPd
zrgEBo^g(o4oN65<6FsYV)lnMisfNkKMM~_3@8H6^nhFs;HIDz@AAb}MhWa|LAO|Cz
z5ym*4{zoxX0TljnEXw^#{Zkx^Qq3==s87?Oldoq!k$mdA1e^)`p_9w4o$KG+OaA#U
zrd~pB_f{}yICcX9amja1!I5XSNzm&mz44Ey1`>nsrFI=R8h!1DeXm!cP9^Y^%=~BX
zS1+k76qBV%9EK14ft05(u3^w;yO-(Uoub1
zM^o|T^M;|!bXCVoVU6GslU`yXCYHL!nsrtPq;^^W?N~L&aEMaW$*%E^(r@fE(?S}ZjlQvY|cZBuD{$w|0z8IljEE`F+
z&357*$0v&dgRaV`4LIdtU|RYIfIr_8`kL0pyHz~9Nc?JEJ>kXGJi3OWT}qr?ie$lq
zGuRo3bwB6m%K6eqQ2(2Qn2TK;9zWnYd;45?`sK3q)lAr}R!;6x#XKU}Y6B(-RSP+z
zwdG9y*GPkRXaD~4pl5#5rwGpU?+`S2=NMa}E|5#kEbytXA5@>cWUpes2^qf;U6|!&hRYQ
zeb#p4=U`bdl_x7-eTlF~or5JUuz&m+LgIE4F{DOHav)eqSBT8MzBm+u&0O*M2RDG5
z$Zaxhg49d0gV`31aL*~2M_awvmmEZ0l`yM4B6U-IGC(~N`7=U1`v{+^Il2;4hh5{J
zyi>0{mc&?;lc;tsBl2@<=(D81n63Ce8Nm`5c
zOKdc}2w0K8-m^CBI~Qy!D|ou}E(h|vUoSDOaNha|3FexMrn~{kRL2V}1}Z!+A)uZw
zbJ38)?#rXN--I%V&nw!-4EOUfI{_;3PmjH7D(_ad0qX-zv!0C3WSN6OjaEmbDXUMJ
zd2+#D`g@~);gAU(c6ROr!gvh#j*q%Zc{)CAwY@c%EyXU6HQe`-x`^$^L;;Y8M)~*M
z4uc#Brn@Gh{S*~{9?P(i
zr@|R}*>#$n%8>+KN?ma>Un)*c{n7oas^iu!Jk1f!GBwdk&#C$~XQ~&BIagesNr@(p
z4U`1<9okT|9^WNVn|U~4koc+2rydC_C#ziddfDo<-lEPwN}%-Qw_S4pk!bq4h?XA+!XDcv2{^cFTJ6z4dyjXyPq(U5yyA8ic4Pi%&M64URQW~
z3%qUr3VEA=66kIagTRSUx-0qrb;h;-&vPNn@E6rNq8gPcWi2ZoUU*;RM6a1K7j3Oz
zq0OkJQIb~dR3a@e9ZRD?pMD=nr}ZOJGozfQEN$;N-}BsnH6jHUbO2or7K7$L2h9M!
zfq-~iQ|^b#XWpucmSdMrX_-2%GatAP89v?*;*q_%v78r3SSJmc*x&h2+3B-yr?efj
zFO$9`oWyPo$nloj+tVFq&aKqwmZTUf%A^_SnB9bq+<}u5Y*2VvYguaoO@!R4|1)ds
z-B#Gm9KXaj%m({;@7l4%E?{y#Wa2A1%
z>7SdL&JKo$4UYU?0Q;c7W>_+*5p-AXTaC&TSxCUpXXUl6E=}2P8tCHqkD|vw^1J^y6@=!g_Hf;tzNzeSNXpUf
z-&qbuj)h;pCX?eX>HhsD*9je9V&NxFT2z|(Zc6=;zT71&JpI+OcSo?#$m@+cn{sT_
z*WqbvU7bYemw}>tmNOnmGv)921zl*Mku2S31o<-Z)en!dHtuquwpAGSAH6n2K_lX&ZZ;M
zD*hV1(M%gWZRl(89aaY{ipyzr>4jUfRd6)A(&=~J&R$GXqgz4`0}q+y3z=IFsZ=ax
z^9>|!x)>trv(ZE^f(o~^a_;zx!$&yOWWQv}@@6OaWKY2QH{&YG;+4hXP1vI;6^uz`
zvu17pvcAoqyZz&DIN$!lzXLmzSQJ;6RG-Po2)TPh%_X8{T8{u*)@DXyzmAYWY$}Xv
zf#&}o-Y)D9
zwB4R0N{^E7k;9v5=mT+z>${jMUrX-=$s7`K)ZTz(3QNIrWx*YV??o;k<~CirFVz{2
z^Eie-5}{(!w4K0kJ7X+sE;G>%7_&!7&7Z3qZAi+!h9p&Y+}bnK~8T4u-_t
zz77S($r6k81xBH{um>Y$gcMP>NUUS^j
z=94cvi-YntTVf3RsNKEU-Bhbj>SM|-@p{8{@QR2lSO
z`BB$x^uc@9IHQ^Lf-_?W7R{ekr`B~tQyK#du>XygIvv4tXvaDZ3QBChV(6=Igy2(-
z(S8*HxdCPv(Y{*;&DZkXqYkIzCT(>3O{4!1o_ul16&xa|4no~;+4LLFxN+=nc>SlP
z`)Yp)#}Ow7W@a+?_6ErIt+RueZJm*lt4lh0A|Kev&A^auVv@jrjJve($24p%AS78UG`Qwcmvp;&vD^+f9WC%(wkcs<{Db(_j*68C#8Of
zF&>P;NG9^sE?^3$0knXo0LqADp1Lv3W_Up3(vRf)q?gL^ju
z;#(F>rERAly&scIY+GHcLIno3L~AUCvPf6e7O&DJ8OJ1oBHg^N8xETb6{NK6oH0-F
z2U&+qe`UBeIZ-xO=03ddfB<{-38JayPC9>XItm1p+w+~_97m36YVa02%}oE{mu6IH
zKVG!E2h|sZX+SOMa%WWkG**Z1#JTzy&cytKte{<@;{uPkglSJg;|^m8#vfWx<=2gp2Z)mS%x_gf={
zX^BQS$4m#<<)y-O>tp;0vL4CZa{A1{|hlqyxcPufaS?^G4m=SplEt%%rc>|a>w<<#u6oydg<Vd1Q=h4->sx}_&8@c
z)MggV`U#jd)~86%N`U^TX<0WMX6iKe;s;}NST30Dj!5MT_egnVMyLDgPpMh=eyEMc
zU>DoH;?Q!;*&N$iUx9`R$kA`&*QOWN~9G>~{*`rzi%{T*e(>uW-sFOToxg(6lFwBnbXh+mce
z;trL&egC#Ueve!e7PxlN`|oMvR8vVGVq+z*{KD|JV66P;lP8gu{(0d2*LM6XxI|p`
zkjy(p8sW-6RNq1K8;pbt(^7lo?Hiv!9oJ
zDyq9wUyd^zlF4j8?{{$b)Y0dh2Cqp5Z?wx2?GYN@exO2K5Ul36X_Bzeii?(lkrD@c
j{fkN`XsUL4vnEZ3K+gMz`GWEBqbP@zn_KFfBH{l5(0|8yqJRA-?
zyAvWp1ThZ>CytW(VJ8n>qB;$o`>w*G$&M7Yv7y{^Yi_l-{tTjZ(F;Ae;dI>
zve5aTc2maVwmf3x_Vwh`mA;#D1Wan;g
z>EZ2S_iMc~ad2Y2#?I~@Zk{$Co^E!w`lyAg{A;{_wLjo+=(qipzt~frSJH{vLl2g;UC+xW*V`EgJDigpQX8rGb5XuoUotBFVAm#TaW+LI@~}%GR5#5
z{Fx{I-+7|ub@#Ay`P+>-^LsM<<|3y2j$}L>s&7&sF0S8#!C#CZCXi>mum0VNC{Yvxi{>;T+pVlp&X3P`M1|2;W?QX>LrO`@Nacl_77ME$g%!cpP&y=zg
z*g)Iv`R-yO8?<#l9bAk!`ifMF!SQF&d#4#BvEh>O$>}m09C&>ucUmHFz&R#uNwP8r
z4j2vd?vmlahNTI9dqp{5AGLN$zO$k1{nVtAJ~n)rzudToj((n2Ii7JK
zNL|0--lO{nQn1x$(>h&(d{`DJUZ_Tp@561T+9(m^(X@dXYH|c=)z-A4V>UsqOO;>i
z!$pvwvM$@6#euus1-X1R1ZiA(sZXntAWg3%zC2q(kQqWHg}UVgDU@=d(DflfhPU&V
z$0H6)$;~@bN|0r-2al=e5aj3e9j$U11o>0^SX5C8K`M1yq^ey<{*;L5%tAk6Lobiy
z6J)-0|8fq>Z>EpzZA~D^!Ti1(hKRv|$G6|eB*^o^wtAED2vV5-g<+vcka6_ea&06*
zzFBT}Lc)L`6$@OfPi-N{Z7EOH9#1F8Bkt{KmH8alPahGhKF@(;t9~?YD(1j0t4G$i
zUUR@pzRAX98VA0K*9Ju{=Rn(z_Q(9YIdD6*Hn11DXtRjxt0!z&Ie2x8Qve&zDw-?s
zyRpGNU2O|rFdH0VH(Z=^kqxBRD!s3EY|v^WW#SyzFl?20v)+phaaG4UFS@f~)l2n3
z8I)&8c<}fIvB6ZWHx+rK36=}j{D1yENSRxL(v&=?W3h^FbY0npOec5>rECOYb)4r
zfu3?@>n%1siI+6VQsjXCiG&EtE)MW{(8EHl2vW?DCw)o*L5lo{Ix*)3L6YZFh;O)W
zalZa7D%k`%cgGKdhvjS-TdsXOt_#;uw8-1pk^|GTo`nsH66CQl7lyANL5|2f2QI;V
zzuy@#xexs;n|(a`VlF}UmKqiECKKePu8c3-pV`0?rH$1naKPNf>Rn782XZq!K9p$?
zBuhHtF<&}CR?EDzqgNB;-JM=*7UDjCU%^tH6H1W!+PnN){~*Y|Jw8jcJJ|3@cK!)}
zNe&ziQjrjQ!U5|mRuKhe1W9YToU#||vT4$AZ~aq(lxu$9GWR&vgSDi39oEbJnzHhH
zQwj36w!rtPSJ=?zZ^OU3g$*Vl+0u+N99SHqpp&_WAlKeFzM=35LAo>~+4GI(PO#;SdFjtwWCTS#f)KF?r>n>a+X;k~uY
z)cY1#e|>RT)s6&NyMJEOYs^dhjJ%?rZ3MZ$V~bFw1P79GmvQBKvLQTcgV=H#HgsHU
zYsos!2K!EisYMwZ_Qm%Sso5Mj`BZJk!t(@a`ASl{p$yN7N4D%SJRfs-8`xV5*bwxr
z|9KzQ?{J&3jv(4w#hko#7VEId|9+-46Z?{Mm@})N16?=$_|x$mE2_%6xZ?gsb^2Q7
zzUM%Hf#6%G3s}cYq5Bet@VwBre~`m|rn!5JW{h~?k2*0EtRu_`HpZR}t#LP`X5#tjyc=&ekiY?hqJutHjR;cVa>&CA
zxc}z1S0DRe{`t-otT4ttSkEosbg>od!1&l!!_(MbLd~@_X>2fdS?9#Y^&f8KW$I?L
zAu3*AEq@9JRC7w_jO-xDT?0=yt0WR+C)=r##PwGC%F8mbKhCyvcAb6^&rRMM#lVf&
zx0>{m^|o^$>~Vst6P|kw?L+(S++@R!)7LdLE_2}LoXx51b=c2)t7l)g#&sGk(lEpG
zQ)C-waryzCr%`(;-)C6&jNXbrYzQ)%&TUkVbs6QmjmI4Gc;@w3_Q*Xp_yB|#YsqYcK4^=;Kf4+nBxgR_^
z*387Z8C+g?58o5k)2_~y#Xj-Q>}6=+D-Hy``s3*xJU_*gb$gcMxhxPlSoRF};ji!L
z|H-*2I$rJZbA+BloL&3oce~h?WuI?KDIfOTrf7QJ;cH!@+{wepW
z)&EoXD(>-o{onqQ@%Q;uBJpDV>+`S6=StcCUOuTi>cldAvvvn9b{o&Z9O~}=rh{s5w
z-jAGX{pS!r+SHg!AU@comh}~}U}~o0Jk$@dzPKR}QRnF`-T{<5Hn(`FVLZLZ0!8(a
zgIw;jrHB%e-I8L6_6<*LU!ndY70GjX$iq!OavdM*gM}aGJ;+46_Oh)|=
zj!SPp=)4Va=B)iEQxV^0C))8NwtO!;+mC*#Dh3uvpuX4vPlfrozZ*OAxprW^NJR2&
zZbvJ0psP&~vTO`P%{bnGVN3b4tmd|-ufqGRxl{OO?$KyKT
z^Jh@6;g*th8}jRf^PaOf?lY9#FNJpgcP4XdAb-CwW&SW?zIWu5e)JRWs+Bko^`+LU
zXl_Q-NEFCjfw=cWnA-wG<8uOmvru1EsBbHQXj8vCy$}6ax)sH;5Q}p8hfGl(Bp{tW
z9r5GPSm=7Rds*^2Z#VM62I&=ksF%Aspz{jizSz>$nussu?y=l(JoUh4qtm#~IY}yu
z?9p!P+(llgxQ=H2MZ?s*kDZoQ6Ue7VO@1SZoK}5xrxv2A*}8Ar(a&kWvummKACK1-
zdW3u}Bfi`Z{q;508@@-oKAFhr#;ET#5N~uCaf(?n(TMs3*H4|`qx^lyD5mDQyrb85
z0pek1!i8-(ekkGF-Z|*6amB0oH&HHnu=}7iVmy0OD^>4P))*|_j;Nbf75WwR-_$G$
zN
z)l>bKBXSKmt|g+qIadxP~qbElhPxv4m=PpfHC
z!TlG1RB_@R*7w(4$J5y;|GIFY_!`vf_iNgB2(j`iD@_IE7d$czn^0e_Kql-5@->N;
zEkZbcWe$Cqs*~;3w>DWIig;X2p!!nef$G7N$Zz?W-5tX9m^u>E=c3#@mEV(w*mJK@
z!yLy2&)&0Ni0E3}!3jY5*zw~tgs~orLuh-cIvcqc*6A`WqFq2oe5xX~*m{Rj|Qa8=AN50`oK*Ai1ufw@#
z=oZ@DK2)-e$}1=8V6Hl1X4>@i>8P(b`fi&Sa_y!oVrrsw>ZdKnw+Z^^7L@X(&aYj|yvZV{e`y~lClci!cGfJDM7vsY`ltf-LkAy&
zaOxcV*!q#ykM`S_n7^amGYe#BM!hIEpLBEOdBpdDJ4_ikZk#C+?uzyv^SAdEBd!c~
z{qY+8{s>E#D8%tsF_&n7`pvGl6qM1wiff|eJM%Z7JjrXzg@=g6yqEX-;yf#8
zFH^OUM|ic}qVg-UV9UpF#98%)k51#bdMN85^&Xup;Z!V-eS2NU^n6FGvusUU7a^2q
zFlv|1K>K}#OuO%hYvawssXDFb(r7+}_M*3RUJ4*LsB4Y!#e2pn3pvR)v`^dc;J!1)
zw>6}3NfyQ@7AHBk8Bx1ybdM9xe^62*w1|?QYpyq;*P%R!n3J81e$MJ!GfI$u
zxIA>o0R2zN4XL(9{$RalCv|@2j@D_7Vtms5zS0Mg&)(Mfy%zOCeOVeoD9`y)FS`l-
z^nMGuLe(q7P3yf@ji9Hyfda1^>6e{U8aiimbXQsRKFNmR_hgu
zIIsHf(~USzh$p_yUljyBlM@fr^e|9+CONrwy8#(>v6wZ
z9&wub`s&4~r?6K0Yb)*xD@M0n4&@2nys^|gpGV4r52F1c6Q1=`QNH?PlM_`x>UW3q
zryxH&yGT0^F~wlyd?(8FB-@zX$WV6S5KG&t|3sr}Lhp%jXf#@Cl-F^u1XIQFo
z3Sv>|PLHRkCs|P<6M^&kXwDo6K}_d!y}k_P@?CpFsXi)Jo41CJc{r*6k~AGbCr)osQP#n*PL+@<4uyyctfpIeAG{X
z%HuP$osrWJna^(0(M;;@Y-cI!uL0(h#W#pSSJxFRn
zd&R-2{l^e*>RBo0_m;?7^2ubx
z3Ld|c)OpOhbv!Tt$F;qA?&u@8$z(2~?&HCqvP-D^`pzHYsY1DPoe^CN`4qn9_f#AU
zy!7UqWBf~skJfHM`OH&3Hxdv4sqU0vuh4>=uL*P`zT+6UyRDYm(
z@{VdF-!|VSnH$k4OhL#C{XZ~MoVE)0`-|G+x70nR?>&1R^}c&~h-j@u?!`XrBZJrw
zcjsCPqCES$hZCYyUH%8^_so*G*qNu%zeUI_0qWkG%vyiwGTPgoIWL@qyu@>MYo6wJ}Z~P_d
zexRLjJgx%qb$-)~p0bN86jp>jct{G|PK@O|*)8;N*9Rcga0TRPks
z_PuZmQT5=K+m}$D(LB4!a}OPihiCBJHKfCdn+v|`Zl^PL+e;z;bp9#%Vxqfx6zuM7fjf>;%&UiVu)fE?;Gr-}6@S2)J2E5QRd!s4T1#Z`81*hdQATm69UD;zgT;!L(
z=PE^qBeyT;Hcz61$dh{`)1>JjChW6Ieit3iEB+*AcF=%nTCsI*9}VhsK3UHfqC=Rt
zh|}O04TOs4iSO;EL9Nh6$3K?RLG;I${kq{FmNq2J$KW`LfRrxa=34S`Q(^N*hB1>>GWSEtNk
z!84r@A=5|}tnwO|xmTM7b(dCnL``Oa+gJTm74*w9Nupy=m;pHw+Xmjrb%A(!h>RPj
z8$_0jGN=$-T8T=0^!{^e7$sf!K#(R
zS?Rz6xui`zoOUK8?pE+yhV^Eu>nso)+69|*TvM8g@f_}YTPchCvN$u7b9Fc7`SS7}
zyI}@6GK|XE8+w5^=YuKhGz&=KGPf29tozkzj_$NBc&%Nt(Yu`vIWJDXG1R3)pU>wd
zAA4zV?!qgD<3#XVO7BOaI54EUcfy>n2IQ?t;-LuO=O`!E>b_TRWi3g3DRT
zpE#eoAxuj3gm5w)%z(>a35^C@(`VM1FQCKDu_rN7JL&L=9=&beECv`=Xe94D+zsBU
z*FmC|2?h}>OLy?G;M-uYnx1wy)P~HOCB1_Ve97B0v$2jXE$41(F{6WCg^GR!;#P%q
zIU2X=pmp|!secLsEQD?brrUHwo1(iGcUUj9bk)hzrV}wc`ys6E$%&~fbqTe7hK(_*$oSNA}o}#Ze-Ov
zZt6{8K<ez{I>Ce$Q-i!8s_1MZe%9ckYfaCtCjRFa1dNzEgH_m9z`{ri={CpWrq
zeY67qj2rY6-eAx)`r&1)SN+`jI6AjFde4~m|@Zwu}P3GRbpMxzYq
z%z7w$xt|6RhpJP_Yjh}X{UY3QpbKhb6GjbfyJ6)F!_*_L-QYGFof#J111B7}jvc$f
z1hRpr$>0bJ7TW|btS;_>Furroo=;`K3bkubZRs=!)6x(5UPFhpzzF^ii3~8=?qm5u
zfB`Xg)0h#wUGTK4k<*0!MqYLAs>k<3zS5$d>PML%rFr$kXUyNgnvwE7*e|bhR~5XQ
zkM**?bdoyuoh`aPBD-+EbT9V5PY_|ix>A-%fPFVK?bE#^k98;`-IY{;`FkBV?@8OX
cZrCYZ5;xO`4!s#p$7^sOIymYoTd@!R2X4n$H2?qr

literal 11976
zcmeHrc_5W-*YAyFtcVJwR3a)3l$2VxLJAo|63J9NB4a|vga%ELF&RpV3K>h25@n2v
zP#O0&>>ZIJ+51~}@jlNv=Q-c^$M=5kALqRLkNvyXc&%%#b*=l_`=pk-<}`i@euDb&
z@)A=BpjZf#DOF&iVU+gAgMqnb-thXu`h~*KdaXgr1S1
zh6cr%^goFcX8(r&&oiKRHE^c=2?he+6>@++N*|}*F=*U?F1(6><-yE**Sasz3s&DM&0$l9B>g6`*pH&x3~20cCq`l
z-njfbb3ovkiv;V%Cg?_cc?I2`(I&+`|1%Cmy~+F#54WnFGUl*+S*
zgOeR0ECv54kCFhP#Ktfum%puH#GaAGNlw~N=9lZYy{*T8Y8}2)_gu361^6>he82NV
zt?TY#=km9OIrDol{N^Bden&D94mCHa4+qC@#?UW{ASRJ(Vy^z(jwn@bYR`zAoSeTL
z@J{&S;+RMP5kuwfAD@{MY)ZL(W|EyCvGkXW>z600zpXHTREtuT{T&lQxY-~6hX<{H
zJpD5sCeNL|@NaMA|K^nHhRG->(WL&w>t*7U{b}?s%yy%A0b{N>7U<}xX!RhTDUMWp70ZHzEHRmucoqy*yim+aU_q~*=ljc7
zSfH)lad0srR~@Msh4K%Pdj)nTvLMFzz@PY_p9M6@6~?_Z79^Iway;Weka~XK`;P7-
z$f+$ps_V1~vgUD+Sb-Wrc7)sT+b9yGTk7BpwWS1U(b}lcK8qkXr^v1K;ULH@D(kZB
znQX}H$@ZDemt0znSu_21M-^bHEsyO}|d*M)6$rsNXj0@@e4#Y%!qqupNGN)lw{3cC{$
zI|))H-^KdW7Tj-gsoJyY1ZnBsmijV}4Q{lNITaV#5WD(EgX(?Eqg9D@W<49c9vhBT=Hm>*RtG&==wByW7TeBAB
zvrPSGs3fs~=kSI5<83U+qDwt^Gr)q%Gh$jti&@}({EU5e4hxD^Ej-qiqhDI`)oqz9
z_!K9#`_4)>q@GBKuJuOvvti^;?{tXrtBf3r#!
zL2femvGeH@Twg)!c5F8b4;~%
zX$1LD=Di)Qf*`YYc_}TxIx`fQo5e#3lGHNtZ~22DdCYv4XtlH8y{zO3ecVYnricAz9M&Gg$8^`5hU12y)%czzqe@3DWaTqMiROf)w?Y8G|(hS-pC*
zpz0|$h%9%HuU^Xr)u5MI`7SKz$hW84;JH?I*X1?a#fIuxX0{Ex337I@dQCj$b7qYdMSbV%URBm^PI~(S#uwlXQ<<{mqfh_Rr
zq?=ki#yJ$%N2Fw7KbNW*EVxLJdtOUPzj;iMt30w~kKy^?=YGT5mXH1TVxX)a`sZL2PIa2tR&61?%PTc6ePPK^DtK<<2Q5$oyk#FV>V{|E)i}br}14n?*tG
zsTiEE*Mb%Kd~si=9b}{ISx^>xbJ0vZKfJkdW`hZAuqis|bIp(-W#U7gUc&kt*j{_)
zgZUS_kgs5jbFhX}(CKmu8w8DyZPP!^hP|QYni^fWUzc@GEZo0s3pYbMiv>w>f@}Gb
z*`StPEIwjDkXnPKn^dk4WChFVC5ij3@|BZi;CvLcbatJ78P8v?(#oKX1R2q&o1~-1
zhLg_{T%GXTt7;wEclQ?7^K`npdORDZNo-1Cts_Y3zKU7t*0@i@h3aN_eu`{kElxio
z$i6ZAMZPZxQj^|S{)Y`ghSN9=pI~2}_1(^8j(NEVX-N`Y^d$5UVa|u
z!pnq!IPHEml+a&PMy(=9k;^|mh~W7h+b1QZgLD7XBF5O*2Aumhbhg)b;{01xytDfN
z3pQ|nZM{;*f^%;9V>viy8oqq|u@n17*X!B(mxgSZX?9=P1pS#lt~9xc@!fFt7!ira
zI!D}n;}FS);BxzE30M!~BT*sNI4>993z1IBWP#g4r~VRrA8z=_`S~uc=XUVqcoTyS
zSwkxd3h_PREO2eEEY68{X4RoVui4=8`j67P)cKjJ-Ma$MWsk_g$1kuBe|<;)Pd-oH
zo06!TIDW`q-+z8hoxFz%aR}fK1qUnqG2(8yrhDS&5Z+Y&Rrl7I1;3;Er`)U7{7>Dh
zI4AP;fBPrn@AIih;>G&c=U_fP)*`+Thq=bl*S@8wG8
zCw?~klcV{!yMJfk-x>IK2L9j4KvT_8S~|NQp4?ut!Ha7EV)s|EUn9C)8DC?C=&Rk)
zXvR4J-rW1ciiiPNG3>8g$?AvINzA3+#`>ZDO!U{}AN`=6{Q68P`g!FT`QYbBKM>dY
zJSxBU!@=u|IkZu4!Qsud5opKvs80JB`YB(QBl-dHWHtL|AmW3WI-l8yd}F@jap;G)
zne}1M&whv`j(Pyvn{@`1sUYp9TzWPV<=N`9Djp+NtEFx6Ks)h7Zt+5-W9y;84Cy4-
z*(Hrg8=O3mnn(Gs`K%m(>-&8QmG?o+opV}tHS#|%U!BH*dh-f4WgkVlvA1RJLZpil
zubGobzd7HhY``%9CUGqdmk?J-i>{SLen&w>a}m;QUtI2OK^%yDy3h~##?NyU>eJLN8gTTY#C
znu=Jx$8z0blwTF)d#!J4UJ
zAGm->v#BzdK+H2zyYm(CF@J{RJk$@dzPup_QM)vgdl31aP0b!^7|-WpL9=y{CYR>C
zScW)Dsz+)LqWzoawy#lto{H3kT%^Nyf8q!nNBYyeM;U0>^tk)wT3mnTw_MtNluPLy
zbl#3ACc6J*3gX+WD|UQ{ZKIFR4xpbG<%9DjP=D0{PkBkK@4n7F4g>6$+2?pSwV|HQ
zPxlRJh#m(7RSa>RZw9SC)c)$EEfVC=eg_auBiIinPsE>=quz_3ikk?GqqtW1;u+Md
z&s4N-MLLsk-g6e^fx}q?i_p&d?o>{7q+1pwOMXYp@IJ>gfPTDPHLuJ={n_hPG&UjX
zToKGsK(zW8<~AR3*9E~KQPfuz>fc5ndeoVu^`pN)x1ty(VnGhyuqpD72ui0-M|?dz
z9=aawY97?*njp=1Bdy?vdKGH|I9&tI?o;brH|X=7k9d}m
zaA`Zry%WCe6-R&d3a=$^Az$EN&p~O#XqIXV)$iO&I~VIA?n!+W`W5x-s}=?)B0h~Z
zPonzA%|KNm32BGX>|v@NqO&%sd_{jpIM>WJKw7E!$c=-jr+D=p-wMP6BEY}_vF_pV
z^Ztkr3cjx~L|oO|V8%kbTUSoZaY0&i;oN9$w0GwzjM7G2veB>T70PE-)oyu+^cng>
zLIL%v$=psK%8%(+#ty`V8am2$DBtPPwnPnSA&=nNM@R=3-&PJr?B33;aRo6w_fd{G
z%DEFNOqfVdjqVPpM0_~T%3h27{`Cj5i;*^IZHpg4+~^{1LHP?(+c%(uw0)IV)+fYT
z<()q_Bmek0sh5;~*y-)I4>3-0U0o?bq-QR9OxO>k*Icn|7D9Qd
zIPE*tC-&BvsuqYM9@i46xpeYC#n4HlbA8P24dZ@H9f|34k$)tG&$A1$t*}Ae9OcB>
zLi+`XcK6%a0mvT+44fg1{a75*wU_F%pM_y>-l4o>?1-Wa(l@TW>!kYR$j1${2QZ$a
zrP(rzkx#D(b$fyMeO8tlHIKfJv?r+Iy0%J#b(c_o;|Z&K6o1@Oz4Z?A6F0xVCyD&n
z=c01f$S;etJ_|_KUwos|i}KlFyIxMydTg
zJ;f8iPsOnq-zMmPTyT*ub$)G{=S>ws{kVPX>~qNP*;Tb%3hg?`>0|OZ4-fh545!Y)
z$hJ>i18A?e#QZ(=o{^O4GVDYC-YK_MT|{gSGBBm1+%iKX+!gKJCH4C6Bd$8``lFuO
zUs#$%0m|#5qPqa~>s>SDmC?VOnN}Ayzq?ND%G-eac&{y&o+9RP$M5yUb(VKkr)VO5
z)~hv>%CF@7EuX>>Md}JlPNQ5Ul=+l;kFvy_?#toamTI4#=ZJllrD5wLg#0`7nq@Q4
zes2N8ZWK{5&ODsz)ADZhrbB2ip1Gr15NVUzmMCAmXI!#aD%FbiS2sL*=#24g3Taq!
z2jiO?D>b(XQLlS!j}xx%Dy1G;gtYL5rn=pTTW(8ys73w}LOd%8{lx29(;pz+5kGur
zC;I2g38}C~`u=*)PU`$fj@4?8VSH-`e5DT}Ew#O2v|`ovbSQ`SdO1
zD%GzveQik(#PB(>DHPfSok&`Mxcmff>Qc1lIUe!M0_l^&{^zOpen)Q#*H4T~d{9@D
zy2mICt
z3ENV0h|k8qg_?gwuN24bAo_=#i<*V}<=bs%mY~0_d#apnqW#lGYnuwhIUWyt&VDr)K2I@q=h@v72i>Kl0aqYILIdXZyWjT^^**
z%_`E$MNHf|af;H7vR4m_DEUH{?k`jy@*KXptr9Wp_^ACb;`gu=lgJFa+@yuWR~pDVDg-Bi)vlTGu5zf#S2lI4-_Bh|Va5YN9X5uy5{Dz@pyNsKo^_QqRkpR>mN1gShe
zGuw4e0CAw~wlVeIE2z%iN#�v&8W<;zKE?+(KOMLs9gXd|c=F^%%+X$Y0fKaQZP~
zk#_KyDB>lt`ax=)V}8brO^EM&?nzViZ0Y>k;fd($usSLmQEvb2jv+57V2L5Ifn9*2i48gy(1?~aXDR?`}!S2qY*j696wU`{y5|CHtOCIR6{67^?
z@(U6BuDdoIK>fVzN=GVxVRPlf6mb39^IjE|A?=~)CfAKv8TK)VAMuf8*fDB8&^@`w
zw2s1qhH!|l#$swZUCDPuk!#*;I^|5!a
zCnGLrrF%FbE~(A?Nd2B!78^73H2U8cA}UDTTjQAP55=SX-ZK}46OoR$+<0#{%H!XC
zQKRN=zo$23qx@OS`<4SJ57_x-_YUOii?t{_AXaS^5TWW^b5WR+$^$2-;hndLbgv8H
zA!z52DsCl%^7B2foc5#s=4EkB1t>q2xAHpGFJt}BhpGEll)t?H3*6WHbVF~d?gB-k
zr|PJEWt_a~hZuFlYt?-8!~a?5$wQRK=%%JAB0v4{$};MAkW!m(i7NV0@QsV6?gzUP
z0%OY&Tk;wY+(LeV_og%JkvaxgCxC)TUXDo_lDp`}+*ud-^mud29YxZ9N+J+NnzwHqwDN%jsq0D>|f{iSU1M
zmk!FovF0;Q(c$B#DF$VZboiWS?wM;yhiVs@*wpoO`0#Bc!B(0MPpw~^`!SmiX#q9+
z?V|f2wjrU-*`^OR`knu|VtOBNdio51sqBS3a%r*c*aTBqp~y>Kzf
zR3{^;2m0&o4Qo5}z^1p(M%T|XVOn!Y*ws)bG-@kKrkgP#e473Wqg_l0%vCPmXvBnd
zr6H*-eI{tUaxp)&jtSac#`4-M28g(gRb^K)V5h>nxZTefu<=bu|Gq&6gllg4Z}sTug%rjQLTY&*aI0QQ%AoZ^s;u5fExiXgN8jd_)4M_W`V3w5*lsv|^M+R!
zqZ>BAjtrfx(*u2Lx2{}lPKTgMVWp}9%%7&&TMeOZut*m@o|;34;PA+GkDt*XolmaN
zbrB8D-oB*WG=&Co&kIKcq-ii$*vClDhz6%u{v>9$cfp`(`L?nkSW
zG&J4?^X|_R+uPFxAB8qL{;`Y(n|^!=EWh@NZH&qiIQlz`b8M+X*^ws6SfWvxp9zbDvz0$e@8h
zRuX-DFAW~VO_w`iigC*rC%N6Cfka2>==*p&_*;1{BCUI%qEt5T=!rfM=sk3eN0bT8
zJ3@p^&oM#LYjEaXEhapRR`57Kl?g|`>ZYim-%&1!_90<9+>qEl_7!i-2_M33v_rPUisI4iZGhx=Em?6eNCS0zw+H-sw
z_S2T{C&I&eVcE28HCi=2FmfQQS!5p#j-DYimUhs<#dMJCPH;B}nB4UB+R+E5Eo}BG
z2PUjaROMo~F(A!E-fubf+aKD_g2zLFaVvW5
zzGI&0hEG@<`amT6qbc(=6IKg9c59Ykfc%TA=W!{_so
zPkmi*_p|aCApNM(qEJ_)8`{zq?iaY;2k}ooci%k41e$T7^@S;Y@bgf9>u!HK*o14c
zl?7>V_xr2WrPzNwUn>_S&7^_q9o-*q@6aIn@VY5d_1!S7^!1cOHVk;G8&fl=&4k-`
zlsniRJ#cE#<`cq6G}s3mJC}5I!LGEKwdV6_U^4zZYSAtlbkicY&l9Bsl&dH0JKO_)
zo6|v}h5@@GRuvoY;yL`3LT1Fu3vMWqdBAe^L^k%fKiY&lo8*^CC;%T;vC5qHY3
z%T~XQ^x2!H{>gN(7rGUcX43=jSGsF*hV{XR&Z?=q#F(&YZ^7kAtpEJR{IX+uOjvt9
z`*t+;q42Ag!l2E#hxI+s2TQJU?yOsvi`#sLZI&?$QtE@dTcY2{Clxs^!G6O>W
z+-R}>IFGdQ9o&vFAx(b1y%6^G44xt{W!x8Qi-PfLVH*70_^~$=_pQlYL@!Un`Frd@
zgk@$otjV`;l3msZS)ck-IB5*vcxYr-cAN=fi~CN=V|+FK)7CU-^uW^I2n%KG8=0-`
zw{&=LPCqgz+xn#oNUq&GJcDSEGdOB$QPd4v+>2M7mg|F}$QX^&R~S&AojLvy&z-MY
zK%GC%Gn3B)OFO+8P?z}a+>^py2(TP)PyI%R*rDJtDJ~i$H;n{6Jcern>&-*WBaXUyNI(#VrNI4|#VzRG_u
zNdv9*#Z$K8+}Wz_BVvU0(!V_LAwh%=>x!8o0ruE8`?RCwun(6=cPHj!{%T_9J#XFK
b14hyhVrLp+o^Ch=R$(3J?5!%>a1Qz3J0k8}N0Wgyd1Y-g;GqZ98=N5n87rsUbktj_o85yPUQAzizNOoqS
ztU^W-DGkX8C0T`x%7}!J^hsq?B$Xl=DP<+!q>RG*{?6~O^Ll^Y_kGX3=RD7Io)1f!
z-;(PFgFAPW)jL7-vaspxPG)fUpSUvd6b5IVlp1wTWw4V=mNgsV4%ZakQiws5&pvKQ
zV{m`H;s^5~h_d;iV`m}G)jJs%!*>nww*1EF3{H=;Ib7XO@rk|(KFik>`^3B~Z?B?Q
zVeRj_lm`^|yGT6Qm`!o-MEk0fi4^y}^4_l%MX}`8(i#0=ioKWWJF56oY}hL2Id_=i
z;vLr%g3T#jdDG3pL5t$blS-kQA&3lwoGpLwK8ncKJk^60j)?s1c=a#g03y-cOOA@|
zM}$Ale!H<5A_ezy8U_sz(WCXca=Q_ExBG{)q&6bH=uKjqIwC6)(#B0>5UDmd4G>?9
zh=UTJMS2}Xo7l+iQbeRacT6nFLnKF@ojC?^<@gHzlzc>NY;BAN;W_#FpsYzQA`O3q
zfzgw95V@)6{QF%NBJQ2@=Lx4H@=`Z;urwCF+uvGz;T+^6;fHFQ5K%G`Kl!vCk;r-u
z4V@Z9iihR!`vy4oFh7gJ`?||h>dh|^nR8jKk`1xUOZH<yc6kp;f-UQ*?VEYa~C3ww+RA6r(f5T1QQ_UdxNcTd~?*~vj3m7ClC
zCPX2!Cn@}{V=^N4pY|rG-$EqZ#WXZC6OrEF|Ke-l-uwN1_wdIfa_E;Ht1y2I&hv>k
z&fqj6v$tgwrH>lpMy?(-iX_}%-?rj)#BL>_T0&9SgRqVZDtxb`YUctl6v@bDp$%N%9i
zaU$|@r4FyP03!dj2ENK)i^zYVp;^XGE<~P4J_yp3Kt$WOn30zS++318d`KMLXDg&l
zEe|Cm0XeF&1J9|d*+^EG=q)&WtWbGGFYgLZ7FvZ_?lHY@ui5t8LG9`BVi2A*><)}
z57uvW?Y1g-%wV5uaZ@Gt8El(XSt4_d!T00z6#`E&SnRzk$J8MP->zG16lc$1o^^&Z
zx`!D&o-QBm;{p49NXvh|=*r*-Z~M_<7X~*vZ8k^-J-ByFM*M~^ga2613vqH|uqk)4
zh$O84&SYO72mt;s6BlWSWpL)KpB@*S>(`mIsUL8Uof2v8$THrap(g*PPyx^o01H2H>xHlIDJW4(j*tMky5vFoBK?3%z
zo%B6lu10a~dVy)4oIoMNjRbAE_uS-U=~
z$VQ|{!Q875^vWwwuxAYPV5z{1tOnq9-LnP*hZI;}%j0qPFd~Wh-#2#1A#zIBG6+FG
z1QNx4!we|?I9V`tsf%JB3$yiiHzQJ$(Xi(v)H^DDVX{>LBF)?Ru7tlp&HU`&m;0`x;57hhk~N44(vU*vC;dSsn1b
z%f3)n=xlxgk?oc($pr8sq#*f12eVkV
zk$-;?`k&{)y|ZXOBAW(zBkgr5u1jM@`4uzx6}QHRe0K(K_%=V3w~4{LySJ=J2%vbo
zOq4S(?630uYAOmxgnP0gwgBo;_F>|$C)9^SJIVE7Dk3JzOidc_LhFE3fmIpB$!akd
z9ey!5cCc;uCY(FYQZoJ*@Oh_+@Nq5J--CZO@cKs>=#SZ^<@uWs`S)eUUmbKgxAeR8
zV(6=2pp0D?@aduMbanqS@mkr|G>cTaa)QN{ngbZ3K(p4YhH-4GlNl4SDk|&
zgVp&=`;2}t_%7#(Cheyb&j^2Z5Hm&ON}abt5X?g?-G5RW>fUPScgqa$67TBh@Xmh@
z-d9;nXS7qCmC)_jMj7l@k;M^xhQTf@PEqR3V8?5b`O%<1O=gDd^`Mhe!{_>jgb`7_
zYWXdNWDqkE8yj0_sQBGc|`U-96y}^ILYdnw|x$Dsk63R
zVlkXsrQhU}JopqHtGRwlh-&w?JS~6jWAI3sQ?x7r-SO(27Q9cfn^*Lh8053PVp-J$
z^TmAO{x?>D*J+~>{Z8Q9orjGE?Nx|;U>jZ$g8K=cZyj3z=TO-sQF9IMDNwVv#24zX
zh^zbfKtI(Qb##<+;CV`J>)u(oSFX;(S~tM$zGCI{uN2$V7Fj97{8}7vJ^O!~2O_+j
z*F60I|D&uM5xp?C%gjT$r$OfyzKKZD1b?tAvmwH#l)*D*?(bDW-|uf<_9GH>dzqs2
z-mzflJ4(&?{*uA^UlrcJ-Au7w(7~JGPblsrKb6|xi;5ezk=J-{p-}`?>&jSKS87vwSB)A3m=5y6%-)YFhX=*f>a02{biqz+M
zYz7}v>gjh~L-Bc$w>ismD2{tPalahqFkN9a9SeDsFOPOUg8S62I*Qi;&iI`4*sGvV
zHPy3G=_M3D{_=x6Q;cHU>`3NJA%p#IAIokz4)e;td41b42A4;Fae03Zb1vu-+E5`v
zagCOp{cHyKMJ`v}R7pf8#bP8@LfuxBUReGb`orow@Q4L?7@M@Zst10r_YT(?nL&zM
z3uXQb)B!wgCi3s98C)_oB>ew8CJ)-KE5LlK(=dA!lEq-vxRrqut0>-7r?xJFP4W26
zf(wqI6H)CrQ{*h*FP48g6X?*w>sE@?I50bf;_#2so4{Pcf$qumX8?hcJ^mIPiN
z+3=3kLVs+9hIozu|Bt+Xi`T;(d7vw1tF#E_s-oJZJAmgud*+lXK@T~e1^Si&zOG+4
zlGX!X`}yI)9hjDOnz2llQREwkqtJmJQ7
zxOs}f$}xC)6#RdE^?re!)8M0>D@BfY_=0X*#plffA4xXmq_SEN$^U)e80#L)ac3vA
z+!pw9mnVRG0Osh(?$Cp)DPDGg!>tGW&#B8{#zNp*UOX@h{0;nh!HaY4_JEEJDXcKj
zrg)DzbKo>TB56su-oIOUbbv37=@Flrp>EoTLxDXoybxm56o8$9{
z;-`j}kFbA(A2D|7cR9`A)Y-qao}fErtYQ{MYc9e5lW~e4V5HBDWN10G8Bp|y>BVXI=?f2eLv@(x8jpOwKFebCP6R5+bF&%R?k>ZHTx`ahHHYVUf?e4u+f2bh-YYUs=dc-ny!~%~
z^>eZrp^XoJN`Qe%0iXA9uo+{zE4$Yt&V`t{zM|7CMsr-sk1G=_#+WfHZUF1%7K6;F
zF&5*?U&Z;R<1EJTv!za(e?vCY_+UH9ViaVom~n@nk?y2|80;4~%gf0kZdy#G`8m*L
zTgm#*;0f8f1&_hmIFGTV*x!ABr9rh9?1#M+lV_(`jHWESBWTzt}Ne=_2>
zFgwf~VV98dcK#=Dt!7N!aWJF*{m>myMoDCyCE{M(*D%8aS}Ad#)CT|jm*?gt=z~tD
zJIO$<)ppYCL%f6aKHO?x=|kzkJ2Nas%~KgS0q9fkH?^?e_$1J)8S7(zYqZ1EAjj8n
zyf%XVT3P?qN02wnmPe>{#K~UvDrpK#;^VuOfc5*jHcG3(fnEN|>Ljl`@j(lD^1b2T-U8Xu
z=cG;{>|2(+;mwC!ZlL~uB?@}y!%{~3!H*tpQdN}i3VFL7V8d+aSue1~K_UMP_(bQm
z1j*^;M-3Wi|7}IQ%i&icEP3oI)<4&tEB6E0q4h@+vHqZ5Ku!{M^I~B0KoaC;voDwX
zpzjX0mpn_~lO!`l@dT=U+!%D;^?jA0pm}kxZvFptu=w42$epKu&aQ8Rym;r}4YDg)
zo2BNAc+u_R-_C&M{ypOTSWh(>Ds}=@*V}$I27_jFMAm?UwB
z4mdxhHH&5C(_t}>Nm8mr@aR8M}K~>IY$4zT+ogW{>F-L^lIol
z3nz265&wx(+kGF%9T!thzk~kPRozi@$n{Mwj}t+)zs1+kxia(iXWgXpxFf0-hPWaD
zl?D&Mo}$b$>Wi1wj;qDtFE4n|_8?^WD&FFsV0N;)4&`-!(#BXt5tLHdF-&>#dNY38
z06bLUvE%^!mYr;jF@wBLB7yS^>U&A|!is#Xe=PmxvIgI&#&=LC1NulM*D6~upaQ7
zcm{oLyF~Rb8_<4igpc~8Zc0Os1N%m_M&yOSUcS#LwgU3Ly-l^0PtKj*X8hpo#-e+>
z5&!RFD&YZO{xLHbNvv1;nHg4sjv>dV_k$G*CeJ#5p?~Az^1~K<(sWO972;_M7W^3x
zIq_QA!)R}fIo9Z?^G-wz6
zp0OH#Jw?HB^#cc_p?^Cn%T`4mKMkxaq4Vt*;eWUt`FPGz+13r2`)AMO0PGjOvQ&Kq
z`|m?_arAvb&DSQH;HP5M7NQM#*{bJSW3ZQysjl1&dtYW~u@+=ge)kj>@;=98`_@X>
z=lZvg>LLEw->F?Rea0$hc?5CiJ0xhP!tbu(^!!xH
z<7l1`AJvcicL}0#!;6VVzN!xHu8?_$(*ajQHW6X@$bzbJ=8n
zi%aldelW?=8g%wqs892(ZoIgFgX+8KMG4J;%twjY)Q=8FH8&L?&Jy*bYoaK>rHYe-
zpop4*aS`a8b8V|W_LZvNdo~*Jy)-7cbm6zfD%o@ceSeIfT`=`!m6d54>{)lT4$|Bk
z9^_c{o9@%rbOo9--2Qit(j44&e(ipL3!3jTb2=$ct%_$?_apv?76!usoO<3^eHVU<
z^b#DXKaM#5;-|TDBZJK|Bnx(KX)e>UMUB>ngNa@$X-@MBm^rr7AmgiYT^SE`M5>!
z?8PR<>`&MZ{nC7|T+9EOJTlpXEJ?ZGT*Vg|Oj8YxeWQ=cx{?~rF
z+*q*2N>`lf@RnzX@iMx9Z_09Fh2h8ZvZmh$OzP7%pm&xBk$sZaD{5n7-eXL-+D_p*lgz)3=7MonSa4!jAgDg;|n+c$D&VVYX)y{<75-
zUMnEiCi8?+zWfIDb5B#BMY#yjJy{&NyZkNHx#Vr}d$4cax}vHKe8kC2r*j!n9Whu;
zb8a^J+aHjHi|T@DPDeP$CE9`O>^<%2KOScI-Z&4sg#`S11vXD#dvO7j=(O1;23k(H
zE~59w4P!B1>a%8ltMX-(*T88;KHWdUL9e{1Z?o8))^vXz{}KF=@)Yi8`~Uk)D#!FY
z?Jwiyd3YoAM>h1H6#!qp$XZKtpsT8<+Yb5+1p&ou@RBHlkw)>X-EY(P-gP*BMvD4D
z#5c5@-ox5rIX`Ki^S2Jr{B79n6P=F#ws8#5p!vSJesni~2jpfktu6H3+LB_{``}+v
z$FYLq&Zca*J|8T#*`9C>_T20QZHn7p(Dj)5B=ucV?kV_l{p_)z_e;tMZ=5;S2U4TA
z(EChdID|O^x(bf9rGl}Y9(fL+kdDq=T~Kv4Ln8-%@n4N@bJKgi!2IwdgQ=1uB9
z*(;)9u+u-|Cps7Y+EpAen>&6qBNzNyjZl_%m2HEt4_RL1+~KiXseAd}1|7=Hx|
z00000004NL9hZ4L*4Y=wr4TK)ETtanAWPOr;!z<>5s^wrp%~N1k}XZ5!q`GYQ4~f+
z6lD@xtWlD+XB0(}W_q4P+53IY@2~gkoO`~Xd+)jTo_oLEVKy<|TpVV?y*gh&(PC!1*`BRp9+$7XhR
zav9oyno-B5JLcI;2`}zt3E;1OP2P=&W7oRf@t$Hc`&6{Dk|x>A_Pr&o88d9=x!w<^
z;fQlFS|EQNaSii<>v*AW^GcsLh5D1%-Rz9<%YfC=k2YeS6R>xEq5=o|G{Q{Gv%}`on|vXWj9Dq%(9uiJ|SmnG0y$-9SU(^
zYJizbGrntTe6aTt^i~`VJmCxaGR{v$L3eq%wcIFprEATH0B#O5y!v9?V#NPB928lH
ze#U409ThOI9ek{-%c1}Nf`6;6Cv^D+8^3kJPO5I&y9suHVdB&&%~`iEhkoj+Er0WWK!4g_JyjI;
zsEadiGTJ+jx2Mm74K5y+3ICM8
z-u*!Iz3%OcU{BR)d))-v-W|5l0Bf%C
za#llcWrt{ZLW
z6h#sL#VYDrKX`Lye;(;e?q6c)2UcY4wrYoe;CZ0gJjj34e!`pbhwV6yAf^9f@&%hr2|B|7)m)7T0g*rd2_rC)}%5fGzqj
zSNuu!GB~}+40Mosa8Uv5QPi$p0(x`_iOk`<9ZMD!P=0j>BBQp0`s4hgiC~jEPdwE-
zwaON6`tFOdnA}t7N)!sq%3&Vdj4ikSh8vyEU$f`Us5bgyK86)>TFcxIPq0vNeyvo)Rf
z{eQ9k4gVrvX7>YXeL&^9h`6I@AGVq~BZ~IzXEMzF!QS%T9d)31&Eg=chqHWZ4^uv#
z_2|iL0?o8}HdDUTJy>t)JWHkI8@hwhA#)O;*e73>&X(|DAE|aK51N1@T3fl#gAV0~
z+#aIc%lzO18^k;O9_A-ab-m>vpE_78_s4;#;
z_ls?s4(-2~mo2?JY2Ebo#q>dQ)-R$W;84Lbb3Jgb;Z2hporjR+WqhD%t#sFJ=yr4@
z*wHxD3p+VH@Zan^X7C#9vV7-B_kU$kisuiizs}9qPWE75~y*qJi?`bfOd%!RhY**|L
zq5bajxYhU-=s0)Dh{o@>Nr%S^Y|u1%Dged_dZdi$jlK?hJE
z=!I7u{2xU;6~-{%4GrJ9qJL@L?2a0L(5b~|a0blO6a7K^emQHWFRe?o$8SdA_-;>K
z)(3g4vyd!Hiq=n*t6X~oeznR2!lWOrr+8=$7$xGBtptAxcl~jCU!9+dXnzaJTdWPF
zcIlJPojU2B-X8k+6PQ0Y`C11|RJqgEO!er-SB5OT!(Pj1xpQVCQEbOOeH{Y`YQ#*cXy8!=-1Fl?Z;IsfUg~oAX*RDhF!KdTZ
z%63#g7d|SgA>GIkQ-2P4S>o*drF1`VbuXs%*f_A-R|p(EysmFID6>%0VFkFbt)!pU
z*FR{{0S@w_5jt3@OZS;$Q(YGJiK)9wH4kj@g2r4H;-0(DJRQMze@U{=(L2C5NAfZ0
zMjG*Z(mcc~k~i-FBi8=eMdRPsx+p^sT%&y7LKWY8X(b!XKz~;^x8tr0y_e-PbPQ;n
zKNZi@{&*BsElumXC%(MuH`snFyE*b;#cMe`bE>->-za*ox9u5Kp>+^ncX%u9i_*rL
zl~q&^^~3H{y8-LpFvsM>m=_TB!_Y9+7I~&|La8Yb
z-)9d`-Tg*&L8LJ8IpX|8qy4i$N!`H7Otf<@`EscKrWP(|tI+xSG+-4CE)no6&jZaY
zU9#w%xodrN0o}7Lh3{QsVS97h-c!B2JacSCH0bjsq0Z+|y+VCX6x*Xuk*4SNMNMyyRrOYZ&q$$!%lk15Q^OxCEmgPg(C#
z1L&HhT~?69eVcYXSd)SHYx<45z2JW-St&6Fop-83@^*+noj$pZ*6-B(fw5;`T2CG*U^VRNhWeG-*;wqoRR>B2A(Um6D1}iZawCgq%ny
zMM;B*QXLJV^X+ecdq|O_qPzb3J($Ld4wwX!ujOtT9I?pVZ^D
z+K2c+3J&x9Ei1NBC`ovgm48U;?nR}N*T}dFrEhvMfC7IAYwi3f+*CHL`{YGo6_-2z
zp(}-BHo4
zIT|x9XU6VKq;XwK)+uBMjacQbUUO#}ZC?(YS+tBsnWH*eAWcJudw+O$e}@oT7fYlR
z%X84slBg6_uuwLxEALET5h#$@Z0W;7|Lum{GO$oL{2-U6$|9NW?-osC@t-$pFSZXb
zc)TSpv;8H5yVYUkA?Fyl-kN>F)`0<^6>8xp6yf>~!;Oktd0fr5ne^1=aY%AUhr|R2
zpL+$?B*an_iR;VD~CqEvKwkm9A+imcWHXeK|ScuwQso`
z9_+Fy`{mCe$R{t{?n;n7|$_&Ft>hrgltU_>GhKLh_tZ}NT0
zSop8YCwR=cDW)(joySL?u;EEk-^k^PUY+CtDO0XEjYoWXU4Pw;L!@poR2$jEBikr4
zRB}F#tZwbB{tY~04*Gt1@5SS^hV=Pg+j;QgG?yF9BU(9cWC8dD!INwJg8KSDD0)6B4?5K^8kh4*V)Hs
zSrLBzTj3Yw?%E@u9h0@I>3zE+FEtT`Hoyb>w6=NtLd7EpNQ@-CzK
zI0gTidv7W@QfR2-JsoBferKsjvqAxycYhnK=zkI5h0Ko#FOdL_15^Dnh6T7cXj)>=
z3XmZEwRda30Q%4S)~I-r`BOg!yqF~O!n==6Z~`>p*!Sy$0t`3|4H(N%Xg@Bkrf)_;
zb-&@-#mN*j_ROq^?4q#UTYkgDLLnNrm|Op=L#k%k)vNzfkSn<){JDkbO73sk$2L=_
zG=G1hYVd=CWohC>ON$V#D~3G%-_bZ2^~c;JpGDYP_|-m2iiPn|yWs5vi>Q6^M_1Qz
zsINQ``W{trMM;Ran~(J^)uHiyN`CsyS<#CO7IUnYO1OG6F#L2Ne9wCdEAJQGN%|s$
zd&kTL>^c$7jw}=Nxy_)%S6%t%cYhZ2(p8G%UpQPcxe?SI$HVgkFaG2pSwG0Yzj>5{
znRdDC>@gMxT5gT*EF!!~t!b{c5+USzr?hwygZDuj}=}<9v#U%jGk_>$=|{wXYfP)qf8F2N0Mc}*|6xUuYb$mUM_-!
z?tPtLBO32AjHXVmrVy&3XK~b2hRvY+d`kPcfN+uFA2ZeN8lOx~Cb7*$NRJ<$20Zn)sUZ
z>Xv4Xg6X_CiD?luW+)^SUVpJ;ko9!%$dTzBN{`-=fEkaC8s%cCnFJqEww9;JeVu6R
z7KEm=IBywh_U^F=2X9u(jc`H?jTSn`W>Oe+D~qcAEI^3mNc3-26gs2p6zyDSFuv<|
zkJ&RgKX*)7Pm_bP#jh(#9z6C-hy9X0!b7>p;N0K^4owcT?yqMUbbmGPxq44T^rRtt
z$GD~tQJk?xvOI;2iEiZ;i~u9IRhQe=P>{$8&pi-MBT&q{Kqi%(zrJt3Ybb~RsxOQl
zCVHOZl^Ue#$;0I1M*RdO4iSa--Y(W)@Uyve<;!_Anxab8-3hnDW}K0k10yAXbw?qw;RG@d!N*X9%b
z8C<$-b_vlRVXWNIkx3SvaRH4=#OH#qB`x^u%0tS^A-SCR@HdSbBgg$bTr=<9$|m}w
z`Tlucu|NR-7=Lp;S8)onrRO+ai=ZI=dYai`Lm^}f+$z2p&g3Z=S-S
zU@o^Q^%{%o+t&p}XLEReKffpnJU$x@Ny(CZmY68~p!e|Tbz2)S^%Mn}Z+at=M2BO;
zt=(zTU-aqtYf2gv4##}Ix4emhwqs__lcz!iw!h7F9e<+1>F&Dgwwgg<`1x5oqgjL{
z;K9Wt4l4tPRx4TYxaS_GF+}#sR0#4dCOq>s(@LGL<|RG3aYHkP$_ZTz0GM^g`0l?}yw!
z?RFu##D8|r!3G}&7M3pzrhK9?UAcj2N*1C|R^@;EQ3@j=>)&qqAwY*>|Ec%HpOhYk
zZjOqgpmr-o_sJf@_adqHE2?R@%(*+uz?ea@VcxysD=g-2U6a0n_`?wAt@jtn5y9dG
z>$tBBgmJt2TfIfd74LrU(@pfq*C}K79SV;|zkl9|lpy}OWS0H30RgH`*@-9bBmAk&
zb&xnL#I7~j?&k(*sMW50`sgr&KgzV@3&&aL+9+!G5Zooz6>QYK&0zumA>p$Vi{gFv
ztCkmu(18%ObdpyU+kVfmctW99exX?*!9nPMTBEv%e==iv@r~&ePNq5Zs8$IvT4uI=
zgMX_Cd6N1H)sie~*I)OZM{-=U^FPO9l6X|=e5i@+;o#pZK2dv!g_Uqp!>~pKrEzHs
z^BFY876+7-+@%ox{`6VZQ2`zvRkSo0qcHYt?R1$23LY_m$71wp*rh)kxjiTX6{nx^
zH{sEV`n7RdD|nn&zqIN(d3Q(MCdo*i!+$QB`{N;rEG%bAI=(9w;ZaR^+Q3I4LXw|*
zD5OxBsQFaH@_cvZczI)ydrKk6PC2%-OS=+#&U4Y9s$-qB>Aj$pUzo-Hh79bM!$
zEag$KN8Z+v=E66(as}yMk4l5VM=)7iYZ{p->Q2boMQaGg3C1QN}#-t$)}W
zk^DS8wtmKwTf|p)Sheo{!r)P{O1me)o8nF<_{$4{=ijya2!HlHzSWpQa?Zxy$yA>)
zl1J=f=eiO73AwPhyMgT2e6gcyCrbG*C<^NNaE0iAZdCpJKR5Hpd-aG5B)rd_D4M>W
z;7!kDBG}lU=%!TlECW*-bB)tJ_z`|=h`9QP(MJJ3Oy|H0k8}N0Wp&e1Y-g-HnVaB=N5n4A3jz>bwz1fxkjRCWYjrsk+RB+
zY#EVcq@M<5goZt{%7}!J^sOW^iZoQ@T6LAxl-Z5W`MvL7=k?+9Iq!3x_wzjO=c^(k
zWWjTp;`YC)o19_wHn(}#n@(}sWJ0w>2E})dOMi9Aq}atR---jPhI@uU6|5oC5B}be
zMe%=&b#Ki^U{xv!A3FsrYIZiRfbaSe4+t4&Q(Tl_<61w+aOA+0pv4P@PscsI*ICD~
z(w`HJ8Mhc7b(6fmses{cQ=O}hr!qYL%*T4$MTVs`t7i9vGVHu`kJH9LhIL=bAO7vi
z@SVSs6+_J!7Rm82cihHs({ZJ6tuRFT!%lyd`W!)|t9XljtrH^CUC(|c*&&j{yX1(t
zH6kk#9JU*qB2saquzA=3k)5nbPkt97&v$)sk3-(PJcWsXHtKJ
zNNZVe%=9%xiu7G3Ugsm?)jf|-BpZb81g0jv$)a(@@LArc!H^{Vw5B5y@)
zMi#-UJ)q)P0q>iuiJcH>KxANH$uWPcTZp)(S=MrM5DAkQoi4qP$eQ}zE!*l5iO|wm
zR}1w_`JZp_5pXe=p50PcMhsPrBA|k|*Qz?UIx3GPB-0)pn$1huX
zsH2gi{cqAmM0Te|UUy1I#JPW8lIj&i65RHN=j9605V5VCKf-eqk#_m_lfLlx&}*ABienII;#yj0ZjQ*w8(LH9
zhv57vSHFGoLFCaDJriB%x1@wt_?#p3+f2>)_8!19ZN{=~6C(0YW&CtjAu_x4>q~w?
zM6S~#^ffmkA6M!MSS^J2Uj=_ZEBO=Z8=B_s-~oK4Zk^GRL`2oUf-1;Co%2$z2PF{s
zT%edWJs**od4H#W?O<3&xKyY&h2hyLj+Unb!(+bP6}D;&x4LazYyKz0CA?)GZmJBQ
z5BspG$BE&7Ysn!lO@^NcEtT>K0REeK-<&+h@Y6=^YwP_O&Q%`GQvH96;h6ZM?ic(F
zk0i`G$F);DM=nVU6oDQ&vp3eoP<${@Zt3T6idCvP7V@zax8^^b`lp=Y%q~$*d_7^0V(|YcmU-jOHFJ>SdyR{Y5*#Sr
zyw-45&y`}a9EC_9xw
zSyir!2(mJXt~!ZGT4uRoHt1{0YkZ|I@Ugx0xHAP^EK$GlI|6ic?qqO}MmZw9b#qCQ
zuy5zIf5<%*hHLH?95l*dICz(}{(9hn!I|w2srrcc`R_hf2fVXSPY@cu3cMaJxLE%H
zks)c%56Yl3U4MV?h38s;-zDma65s=m6dH~W0zdl?)Ylb`Q+&9G)_rtjxY)Bxv$Kog
zJQ0({YU>c$akWu-ryKCJYr@Yw6?7qpf;vG@-o3?~A<$3v!FvVru&=eo|NMWzzG|uV
z($R5<+_!JpqUQj<@M7(REa102cK+KGXNKEl_H0xhVfcUjEuN?4>fkrn^78Kh(7l!W
zgW3W_t|^*%mw{h76))-=13y^0@JW6%=(_Devw>pP)gRl+~Q0PvYEojKpbFps(E`s-T}Y0PcjeH`$P%wCXgSqggGE_f;O2_n(iq~d!Qe3#O=qQ0yWk@2>^nxlYAbd>7MPVl9ppNEJjoZDQ3|H5kUajuy)
zMsN5Kd5Mk3?s+o2V*m1t)HsT5L%wYBX*K2Kg_SllqzH;I?w2B$6Q
zs=)8{4rOwp?;2wBpp*9Eb
z>Q#S`EI9}L&+y{iQ9cj&J1h|GpvQ1#mfFRD3W{fV)!&wQQY`s#UbsLj#r(T8RwM;6
zyhHY)ivaAe@hLVFMS}lM*T$Cu9(UhP{d*Yj`AyQ?Z)d_BQ>KryKo`sHq)RQU8TM6)
zJLmX=;(+0f(HuB;s)bbI574ubiHP4e*x!FnZRo|92;h(D=H(@u5t;mFHc%COxwPt&
z%wp)PSg@>Z59sNKjYwG{;IiB|{6YO{M82I4jvPP0@P2GI
z0L2D^dk2iZKyKhZ)~Zv(@OP2-j^cZPH;q1uXJ8&`%M2dZ0o>bd1Fo0?Uy|LO9AAIi
zB66wD{AX?_!ZxX5I;yzhF9$o?ji;%A`C+n&=6eF}(}-5x)Y1UyOZEOwFB>xR4}mt^jc4*oPR
zbo7{TEh1+R53LXb9>GymbijOGu4Z5!ZyhYm!mluK$-krL659aXvm6ku_q0X8MU%T(XeM(my!E1qMLeBb}
zRnVup`nikQl?+$^^MyB0oMG33XnL}Y;^SA37CiT(I5DtoeaBIX%VK~2aeEANF8Dm!
zP%8@gZ=0>dTrT8A9(TP=DVPW1agr+mH~y-#%U?i$GzaYNr~x0oOk2k4L*ARZ#&c43
z7^a|LW-8>PS(9`F3C3^iEhl#c~NNgQr$8EZwNGHj2aW
zR8HwxC-8}sPMjxt3UYr-JaZw??+~@fH&&2)>g}8VFt7n#Z7r^EfxO}36rf{f&aib$
ze5p+%;4@pTUjlVLtz0Db-J9ZliJ}+wLjUTGy0RqIArJ8X^ca4}FxmNQ!~o7K=A~UH
z3pm}k66gF6bU)p(yF3~6v^g~6zJfC%7A}>qjQAnnJxt>bw}O9M?fX)v7wQy_eBCMq
zy7b)enlwOv{DepN4}t#AziG$oVUE1j6F;Ez2jru5D(9~OpC@EjDgSo%7fd{ADbDz(JP`u=j
zpljO<#m;f~=U0Eo_YKwycl=!Z-}yujdHKVmsbI8XNc1P82K(2Lm
zM#~R?{}=HG@eY9>eBKpqzZ!h_ESEfDpCMdJE=dldrkNv_+{>zDitAaMz0G~xa
zh;>>)-Wm)Qb${3le)NAB?Lq&NJ{W)S3IG5A0001ZoE?^TJlFdd$0@tAORfgWjf!MT
z`l|0mh=v)rsEo3rNOnY78ObOklvI+^Eg>1dh7idpk_tBxeZMM{`FoxF*Yk1SXMEnD
z^FHUi-(AfoKVP3?G5RNuE68xM8CuQ`!^QL2jF&2Yi(SF4D`~ro!6BYp$tQpF*bISz
zwqFCBY(`jf`Iso^r*Q1^Ar3ZUT4&vdO^A~oE$%5a!(z0?rTxsEWHEjjvf>7@zQuGX
zIC7lD`1IGkLgNV*L~vIj=O)gVtqs(Nh8`xzdTbKw|
zYrhgDIleN0*@!r~yGnTE;8(UtEHsbyS9iYB3mgt=IGKR;q6U6xG1N`f;MTzu$S+O4
zWcAbk`df=7(YdBbex-Q)wH{6kFr%TrDhOoU>(gmKd`bHiZ#I8Hesnfz(_6>_`-X0i
zUA{(+vOVJYcZ!TAf=usTkpZkrRMi#&6RtE#v|ub0CW^*7ilbSau=HDVJbgm~0j^x|~d+rD+hafKB
zu`1?6@N>!K3hIk@=C&I|;4d%WYv~JFzJ}-C7Q65~VxObTH2@4guumP>9QbZ%M|eRX>{s<0#6E`X?ATIA`JA`U-GmRk)m(D_
zAmTfQDux{ci$YBt#IRoLWujjN?h8CMbAEGg6#aP)
z^6@t2xFh87Cnr{a1QYK|u_i!fmkA&B=ZSeEPclKr{19=fhg>VWR%P(z|D2an+_W#p
zCFs1@PR4}ff`*FM3rt}zWIy@T2wV`j($8?_$}Y_Hc$)l>h;gn$6>!jvbO3V>^*~n?x|D1@HwZkkoRBPcbiqg
zKF_=Jhc4n@_?_NE^JnEwA$L{CJH@=FJ7IrcvLHuygzB?iBKRcYF0qMMONZYr{h6id
zl*bPY25|NOs|Z645}Mc#;JpY
z72lg_o@!+G>@0)-n;OA5C&c|QyQbF_oaduhb^_~`#R)4Cz_8EK(yfS}=9*Ew2&{jX
zN;SO%|37_GY|TLjk7e6vo;6L}E8+mhT3(dW9LRZ;aE1EOB0x>A2yxbI3{Z&#=a2Sd+m>3`a+1=}kTG?&X*oC<#1&l9yD^gkrI&4cFP;#{*Z>X(>hA3eeMk}g>^Pq=@0H-4e{
zXVAO5QjhK#F3yL>;CT2#e(I0*YEwlyP*!dF7{&eQ`fkdC`qIrt>N#Z3v|20bFFBpy
zvU>d2$(ATfs_Ppg9RsTLUk9Xg>}fvgHME=qr9-4+{OI1es4GJGGMlpPp?ZJr(zreg
zboiblM}Efhd75Rx@JB_O;go-8BW``te=Dq1UXFEvy@Bj@$mMHH!suNPn!baV`d9x<
z+{k*m2fegRC{GI6etdL3nnk*rDd55YX{lGBSO?!`s*kX(``Kz#e~b+IZt#Uj#Y7EQ
zYX2-+8sw=O^rZWy?b?t#?dQ4-_j;PgFP9Iu(7PjF@v8w3Xf!uu_XmF{wfIW|#SMw5
z9(zLm^2x3A-kK}jVoLYLSL^xlRBw83+FiwA_nE1``82pl`@_RjFtX-Pb0?bHg#vGB
zZktKzdPjoItDFqzKICdw7+eHivehUNM!q%favjPVX)!d)P%
zAa610CmO$PKTq{*{7x?)4A{M(UV!R%w6SL&7{Lg)qCT(>*E9*J<4C>Ex
zET{KImZ7jG^;v)2p32pfSLYcBUkM)|;`z2w7CvFGUhts3B
z)BB7u92h(cdJBxdO$U>@T?%aI{?pbL*8x@MvQ_ipSMbf?HW%cWq8+dPg*+DZmsJ_Q
z`;9|(qz`}Vv4Pvxa#7v&F>l@g6O}6z*`VdVz25Xr89pQ$Nd51Btz}UU{M|O@oIVHn
zcjHtD-ERzuMn`v?uc7tAy36>F(0|ru(Yb4+?Tw)KVv?03o8I>)CfnWVy;XEh@bXPM
zcd1;V5b{s-5)y~MW(`O5Gsr%pv&PGz?`Sq?azzfjK0U#ywRr#hA3oFj%#uDBfBgyo
z000000C=1gmw7za*%!xAmZXG22q7d{5-Q6>C0miT8X-b4YRXd9nxaK{FtUVHRE#O4
zY?G)YOSY`>MA<6woAFfCW1ZjU{PljF^F8--w{y?EU;i+h%f4J}Ml9!ZTrgN()4T5=
zH=AMCsq|1C{C6~|qXhnhu~ntKZonS=ueB2q1
zI0qwc#k+{BUI^0QgT76>K3^h@OL4QQ3&xYNs>sO=?C5WH@qoWD;(M_Qe`p_W(3%hb
z&-%N+#X_h5L412U^nHsxitK+*Fy8HCw+hWnFgi19<)-E)7%hGw7oK69UJD&cSHL@g
zX08or*HHK5KnC>0jt52fgU4n6n2v(3XtA}z7?{(w=3^ian{l@C`V|4hU-}e$t_J;#
z&ILFrVP2;EtlP_>&q{W)fAxkg_i)|!PS~lH^#=@LhZ@99|Bm^Et9CkELwo;+wpMS@
z4<~enRZ9A5;}5eIu^FLSRdNIHFDiK;bcHy0bK2vJ@!dd9^e*|v4CI*|!&4n)k
z--k}MrObh~u3i}m=-1PDMO-6vRFb}(ehmL2RrMqt*cA!vqFOMJN4)Zhy@O|NyanmwT!ReB~1B_Q+;>&^Gp|9wAe(fgM0?T}q#SlMc
z6?NwW&3pD>Ht9&!IR)itu_MrglY@R5ZIs4TH$$QBgx
zijsW?He5dcLk8mr5B%r9Yp^5!9=}EFxz3~d{xtYqSN&ce=1~|VzcvKajC-z>y+G@x
zoVGycj}v;Ke;>BKY}{-;z6%JMxLiQHHCEyKuRyOkRqO6i=%0wRd%7Jo`uMiT3jRMf
zxqT1;wM|^bMG!CKE#kEq?LC$3J#t~EoHNd%`5rTQ^j!#S?Mg9Qg?^4t%W@l_-&7x|
zfzNdQ8g?l2Lg!{qr3OFrwU_w1ex`YJbxocIm3C}te^i97ovy`XAM*P~--Rm3Hp
zPRTO^ZKa=FR{~ken<|%rAzdp(=h4n}>Eaw(U*myuQME
z-Hh=ng_qEk$rV|vfO#yJ-Cp=N?6djVQ{G_ie{fuw6zC@SVr3EH^|amNwZTEow*6`7
zCn0a{vOm@5kf5p`;2T|*02l4~M#Z)JWJG@(4a;2_2X&nz~zrt|I=6Hxm<
zf2`<#C6gzfoUTkbykdG(fANcGZp`H1{mFqbtxBLSW9
z>gK&Jlvi&YTJ~Wbr(|qr7*scE)$=Vt!(Ve(GZ3fv>Onr$LrzNM{5U*kK|VlSf=c5HsM?=tA`|1=`mLie;%m2o@x!!}8W
z@-O~PW6xf&R9}CUKGkQ|Ix$ho&zxnu^eC>jzuukpEo6BSKX~YwT$eL+hdY>dG!Bj2
zPBt&yyL~6O_R{&c9Q3C9G9%IZe<#&nSMj#nztH~0DXBgKr%p^sH-U4P29sWaevw*N
zdcYF5JsGuN*uDImP2lkFAM!KNkMzCiT~?r>Xn4v7_-o{i+t*-ywI7Flqk7IUtZ2B0
zex4-Qf4oEIgg;A@^u^NHZFEmKxK$SfZkFB|6)o@RoY{9B-b?5E{X($2F8HsOZ0;Do*T;Q3a2|}|888S1yOckKP`;md
z(PZ2XI?ZPo(fHjBb$ESfyjn&t1!B1r&ovSyxODI2C(t3ud-`FlU{i(H|X4+yXTPrf7^xCrbzF_-=_Og
zAHm=DUe`<>{QlslSvNTISBcg`y5~pIM}EOLE&e>W-Jjl-$2PvAI$b_pWKQS(l=$O+
z#6g?;=PMqfU7e+pN(}6m7fnm8z~qi~t;w_xhupZ-!5@N*L>foBgTvt=@abfwsvY8V
z7Cq0a0w10m-N~l$e@k2}6GDGixw-{tKNtft{wu)oBN~0qpzI>q3XiH-kr=>TA;J9`$goc7i$+Vl
zX&$Rf5={5d`mev*MdLr%v^Z6m{HkRZ8_@oscEZ+K=qhA&e>`>ts})mqw$lCfC4YhP
zqa>gOtXR9h4Wvbyce?pdYOt
zRz~%9jZ>4FA?!THfx;s+fAOMZYr5}v!~E^QF#mFwGn9W-G4;~4ADpgAx(}zN42tOd
zhPU~x-%NEue~(#4?-}_In%Dag@BGpAQ7*V7dDmvD7wPpQ-gp?&(N>BrwG
z&qZ_Nnh_T&79Efdif<0$q*2|=@Mlx~W#%rQP=o*VmjSD2kYDgbaW=gNEM3#-oq1R{
zI*0Dry4)eR7}!VHtwU5VZ_J)t8BKNM-;gE^dM{U-e@|Yf_s3{m64mV_bIWlrP?y@sgZaw=jB-^Vk^pfurSHT~DJQm$klQTcK-|bVErJ
z_igg2(^|5K&(^PV_JO}mvRvXDbVfFeDB2BDP~mRBHe!#EirW8
zy!E@J8%F;l%u}5F6mX{a>nOc%%R-G}x4`~CK;GOLl0F!Jt_%PG00000c$^KFc|2F?
zAH^*#v`s3@FH}^r6rqudIR@DoB3srdOA>~V79>Q5YD%fJFi}G(5hcrzM2k{1*_rSC
z-tS#RtD^j#zn<6YewXL@+7FSP}l`DFc*UfzEofP&j1
zKhv5W6e<#bpB+ykYq(0eWD&XUK0=XzFNS%lm2N!?#R(zt@HF;OpRTv&caT)CV^oyw-uYtGPEYC30M
zSQ3pYbs5LtT{IGxboUuM(dc@=|CGvF8VwFB*#aql8k$`7t-W1BM68fVFOlP*t}gLF
zRKsGaVRO}+L>AEk33pR(7S1oN^D4pOSN(q3OjQ=S^l*=8CW|>QmOt7#!r}v*S17_?yYzok*v{ivuH}@61`mJ9U0o8B95&x6n5D{b
zuy|;Hct=yf!C$(y>S7OvpBA33d(gq*xo_q5QnZi#)4X(7TGt&c0%Cx!i&MSkJ;O?$xQq`&mp=
zX}9s7$3op{y<~1SgSF0u&t51p$bHm*oObfD2v00zf*r4mkd+x9q}?IHzXiiSB7X+9
zTG#(qb4P^W`EDw+Eg1Z%WU+i`I|IqH8|tsDXYeMag%vy(VblCnmvse+u+2)&HNH-8z^@Q7AjE8Z%EobGaecB>B!
z^Y+(M3WsUr_Lhyi5nWq-H<=vyPlzEGjb(X0LR36ZjVkP;pebvT^nE`Ct<81^j+;?P
z|F6)#Lx;k*Q^|G-CKOhk(7u@SCxxMQ^$Bxz3JVsf9r8?~;F@pX-BduKeaq|Yr$;I5
zob&tN@(vW9H1i(z^C?_8zD$aL6$ASy#i3uU!y!l0)&su@Xh`xK+EW+GCNj)
z(^4P$whs%S{czxyWgZkdX1w!%G$lY&WY3XFPJnJ4`EqSkfO!sMBZkrxhL1`u*D<1?
zvRB`HMH)F@#GIO#ZVEA8a@Lc|LJV#*w)js|h|X1&SFip@VRqRi;rDHS6z=EE(>UTz
z;e~Ons_s|9mGY#?_I4q-bz>fWuW7W!E?s!|oe1rPAM9eMvDh}&DR?DdnKIRG0Nb7e6QX{lXv9*pFR%^uH&HJHS^>>2Oe27+!>o}9&Q6V#gF5N
zuU6ko_n6J2BBWWu$%Vs6xm&jdnd?`p{eI0VGy?dJ>bFEEzVhcZckgCk|BtIMHS&ba0%1<>b%d$O5hN>LDz4
z5AVDDUP*{H_ag(VHEHxuudy%=5W%KqOZw&O3?4a#4&Qyl!eh5>*Sb~?OBZ)+%MakO
z=nuW)FOKtwu541AWyiy-#_OER84jOHWJHg$3IEJiOE`Nm&>q?!8S$2aMn&<>)b~QH
z@0z2?ZWiIp=e1&g-ZvQZ`>b5@{R{Eq>h%f}?>S^`xE|P(K)7&>7q30Y<5Zxo-_vmp
zff`jZ^S`k;+g>^zR!n}E(fG80^jL^j}356iSNFnC<)O=MxMf
z>vaadiIH=QRyoDa<)P_m;MSGK!=z!GW&SoE<)i&A_i{LYgdR?l{$$DGeM@ux&MFZW
ztf|lpGN93$Z7^eMBZVL}ZIhHuLQMCvvNNfmQF*TZ_kvIcLB4~t1QHw?(^*CWGMPZH9
z#`dQi1?@$D2@*4-Xe^XZEWBb%a-(7N@Znh;no@2`z=(&PT9sHv4v*?s8`Bfye#N!+
z2tu-0q?m>ny}mC(z~A+PWA5c*i{
z`DVx9=E84Fv{!NX+2qHSR5u=}QlUR&4)a)Atb1mE^gM@Z`?(cX41?OI5m)bsXqY^X
z+%>UEi0z!ATACaMizJt-8b*N88>)IXjTGoJk$L+gX;_MR6-Z|=$Zi?f>m0)2=1S$b
zkA&yuG`2fYs
z-vqFK9GU)awTut~C%I~|cO=Ivv($GkCi-;?k6GVEa&cqd^L-@0ypD$qu9@QS+VR5q
zRDZ(riwR0Mj77++x^U^?W+4`*sng!%+&gq;2MkUK;QGDklq%^HvPMi?CCPi4>cXN$
zBO;VEJ}es-=U^>AR$So9Bkf$Ce{eL9AG)u9@N
zxYcXtml6K-?UPOUJjLR5f`6+L$+@tiRK<7BJm#6%r&W<07GKq9;INnYI_FktF5!=V
z*4u{{N(2J<#~W)qi&K!3THsI=MPdH)nMTR_LZ}wF)V$XvdD1y-rtBtyL+b+_<8%gf
z3%RWsMJ$SUZVrsg<#4+qzc>~=9vX~IlOg+9ZIJ&;NARFr%>8GaApHNN{aKQ5_+X@k
zD^2FxbMiouk{X5h_%C<#+9()0!*984WC+jC%?*nqd`iUK
zi>VxD1dMG|G9!F*4OJT>`=rYU`jil#1ska++AR`d{`n10;w>oHo2sSO(gH|-nz(yP
zl0x>~umfkvInvz}!`hFKJoL|s7-}H?Ovq94P+{=jQ(b9MEDNjB%9F}up0R^}u0wma
zkX~XFaqzJ>gUzOobf*u|kY4hbX-gBLS7zB?drQvbr8k}E94w*Qi4P4Z!o^U~f;dPIcm5y?cak@^Q?6-gA?}o$(JhVEEOAsXQ7q`x|3=Ir#R8Pc|K5VJw_d({B_(ZbHh$cs7mE
z75}?CW4b#=+BXKcwHhwQDbE0$eUrd0g~KM{t1zur2LDTLL6X-0N`))X!er-gRF^F4^dNdu40D8^oDgOYzi12)|LnV8+L}&!j!WNEhW9rLX14nl
zx)A<^p8vh)G1>1=>BX+PF!Ek-EwH!$3gLiuY|G+b-FcjUc8?1nzCSrxJj;scO>4ts
ukf9%g@oDvQbvM!Y$uP6um-xdy>grO1K>>tQW*aO?@A*GQ{_jGQ$x0m}%+S>U

diff --git a/previews/PR1624/tutorials/out/solution_000150.h5 b/previews/PR1624/tutorials/out/solution_000150.h5
index c5dbf29fbb7116cc7359617424ac5c86d6b7b17c..c9fd4475977e8555eaca3eaefc1de01c25375b9f 100644
GIT binary patch
literal 11976
zcmeHsX&{yByZ2*OhNLp3K`1JeDO2vpkWxfM(SRaD<|&1QGS3y!B!!}qs9lr_nW89>
z6cuYNthLN!$Xe%mioN$a?|#qu@ZabE;eEFc>$iT_bYJ&%U-z&odYirh-+Zz881cu=
zjj>|_Kflg>FlYa2Fwa-<9x`~VT+%ugZGv>Q$>9kb8dX@U*mgvxE{B6aQw>*=86Tu|M26Q
z&D+M`&ik(uXaAY$ua7_cc;@^ZTpipT++N^g*U#I*-WaT4mVeIo&;Cc8U4Qv={mGw*EbpZ9bGSc^%YiK>>g?<6
z=723+iGE)mWgbkK32Dv^-*Bv{p^AqcpYxxqFWWDFd*A=oyO*K+P1QXa{ZXfRzv@Jc
z>*MR-`Iifm`fD)wWWl(8Wip!%ac&ZSEG)kSBR?62%^}xgo=LLwXCe<*YDBvk6FH1s%TbqSqUThT
z&v;^BJ;J*;thQ&ML!9s9!tEG{*>}@m8}!Fj^ZkVh-V9`WKTZD$^ucWdzLQ^|KMa}e
zT%th?B>8OBGVkLIv|dU1$usDW7-a2!+lzsA8eiJa3G4iN&CVG`CkFB#={d9QwH@bV}M7##vnDc)aT&10C#m^I}JyTc)lxN%AAnRpCWvprd63w3x?y-0RC
z3B$K{-FvRE5yO|Rk!hIEhT#fWlHF7Q0~LPTI+O`@5;2Q~UqI@Qsw|-=zR+MbI4;FRaj6`K#9*BTCHzUp4>8dD((TJW
z-)Ep>97}WdvtxLLvmA%{V+?n;4~lBZ#PGy5+96A5Of=-X)c#d31F@z!-_OlpAbDJW
znJAn)o{f3*eUOj1g5Q`Ije-2+Xd8HyF`PEq?=S@Gkwm?WefPoed--P6_YO>CtaY{Y
zbtLmw&+d`h+~#Bkr}$Nc_~r^}Wd<3-R%G7krE6^Oe>RE0+#_RH}Y
zh0vQ2PtuI}ja<(_=6VGpCYP8<>AOo_vH_eE_Nqo0oKwg9%Fb+gj^RJ}9u}lRKeQ$x
z-c4|>wR@Uao*iPKrhE>WL^w~`Xyg0MT^MLyUgq^w
z5%ftyG#(G8tyCuRAJt+g!alBYlDN4R)}^vEV7q<@
zhDY*NW}Gg^@PP#7(GYaN^;{G{I9aK5l-x9Q4PL%uJ^g2F_Y
zh{=>ZHv#MX(6)oGBZi5N49Ji=pq`vICYCQ=V)%pUXI%XkFg$4BnnHgL6P+%<$a)UW
zwZMZ}EBztf!u^9K<>weE$alB99`sGzw)tXX(8t@vJKxDgF;P&{VYRI%G5n=0XI>DT
z|C@CZmRdqzE;#f}IQR$LbMZ|Hg|HuOU;Gv}1@YwLbXOdLK90Jc6;kU5=S8m5t}0$tO)b{aah#vVH1Ww3>G)!gY_t?S)nt%l8JVV$7VEnKp)o}FldK!_Rn|p|K|6(
zcT*KSq3P6nkcW4#7+1-}P?2u;N!%f6Kc{@qg=G
z#X4KB|J%Q0{B=E*aClh%{QLLSbHkc{t)9dib?%+&3ExeBh5MU7|GHiqqc~>A`D?v$
z((KQMzd4(K!u?Z$e=6`#1^(Zu!0T&^T@q&okcmXu-9v1HsGCk6KFBeMTw8{$egl4T
z%`WUD>mXt|Bef_38$|SlhDSTW-t9xc){D~vD0aW{1GA|C6d4sy&H(++TK<>oehi?R
z;I4)9Kz~szl4}cN08O`785l4J(0ICui3!+I*$SnE0lN)^L{Ea;P)Sp|h=}jk(eMK7
zR~#q_*#qOe`Nl5o1Mvm7YPNer+?m6rByO-P*14%<40uFJHTpQjr7H
zC&-J^UJMNbk6ixZv=z%BdMe3OF$5&rMw7}we|oFEk_zyi8maXPfXMm+n;O`Eb%~0x
z1?)T0lh*=vb!y&bevrpl?{-al;H8T!+dl(dkn0T&1)LC-OO*uw-5sZ*%D^vU_0s#Q
zfDGo6N0ER!cbJS?Kx~7_v-N-v8@0OKAzyDJhbC|EleZN6zzK1>xlRnf1N}DRk=Jhk
zo$ux(?E@_KtjgO3*tU1FU<@#2Y)7&Y*l#^CajOa9Y~T9oKp)88P8Tp}z?pWn!Wp1%
zc1yD-=CME9@OmYndugZNQ;=6`p1VZEC*QP?9S8fYBp)j-;GOh>>iv)pK43i~3OwY*
zyu0pTUw6C9Q3tq$R95|KKyNP|u0tSScl7aZuYmh`#@Jqfb=_+b-qJcTfDV&Zr7=MM
z?#{aRED-<9pvx{Th})*sv1=UsQjZyFC4=1Si4Cd%e*0*O{W7qBHN52BG{_tByv>t<
ze?GG7V;d1aZKN3ov@Tk4NDb_>(t->=03V)E&D#W+YZ@%P0Mv>ItR)5Xy{oB`y8+L?*yKX&g9^*K8+?Ed{HZ6O0Cs{`R0?9c)X8!+x5)(chknk@vo@sm31S^+!wjy@L$xihPQ;}XCdAv)nXf@N_J6pDEy%Ra{($V13Q<51JA2lx9?IEC5|C&fI8(
zc!$4UF;N2EapmyDUEl*zRvU=@$Fg1V{6&bDDj^Zm0lYJzrI^c8>xQ#^eNfHfknB@e;grXb_vBH$;UUdx++IF&Jn51s(|jnBoNG(bzG5K}JJqs29W{2zS(S-43<{uC{4xEEgYO#{AJ4~?@0X(JN
zrb6tyy`^XC*nzJTIohQMT+KVNauslwOEk7}z?bgsPdKz}}9$d~9B|0r<^
z2iW;1tn$1L*xK`su@2<;&Ad3GUjp_Wa#{-^@z6MZm_qIs7B-&u(VQDL5X?VnV80Ch{-T@^vD5wOZ880;)>40{v##|OU%%>jHp
zNhMVRicMUr-3#dQxOy!ypD9lX(^%kPZ{xXTA)lgs9zORV-jWeYB+-BL&8ohq!Ol_W
zg%iT0pMb^i@3??q7~o
z@%4HT-z{T5Z5zn9uL_7I`cx}@{6RU)bGq$^G%;@d8%bSayfk@|4lCGqx=Ls1fnMtM
zc@=SxKaMwM7=XW|yo%g1(0|#*&UpaP+WgTy9zcT(yQeXr=e)@~ZU*sE*%dWf;J!2n
zD*tvK^fIIK+>}9Iyw2>7Dx5ETDtzChCt=@QyMjx>IETL4!G9Yz*4p`2+&+QiIZM27DTY+cTd0uLTI7m9_wk6#6Ii&R;Q(*j=QnHf{
z#C@jA^DGDQUwA@m-+ADi(^X1sfYTAJ>z_g#^*c_BgyFryTQSkT3*^d?TTQaTfAAbP
z`(fZIcO>;zf&WVb!MK&c?Ui0USPS~o*WLSwbE8)Lvm0?=afF)|c(D(ng+d3y=j?CEi5~
zg04qYgIuch*#yz&w{qT>%YeMtHS_)i@Gmzyuh#)@nqh950sc0qnq?AjAjYAdI1jY_
zFONKe@nU1ulNEtGzAIPZhdhNXKF4)|-_RrOE7E{0jSoxxK|h|L+DY`E)yNLC1F*np
zq`wdFfv!{=Cm?E)^C#AM&l}zAkHN41Y52{nfQf~gEb?K4
zhc0K%5Wp$$^alfcq7zmXM*&?G^)evf%ery+Hdpu}%Sn
zk7A&{U%CqA>;SED(-(MwK7T4H<2l4RGI~{*$V2Nz%!nSK>*gkp1%TCQx9W&{Vn)P+
zgILE>)hUB|;3wR~+U$Tc92yFW2;NokR1M;}FwTq<_rgV)o+re;$aD-b?BZK>pQNbTS(F#ycUBy}*T11>}iyG*mcnr2=sNJvm}TJPCjPH^h78
ztvT0mRlvd*c7f3#mnq}^Lac8>I?62vu8>3P;stcQSyoph03VLm
z^%3`E#pa|!;$4GWukaA}(w*WB2dsb#3q9_r1^#01V^J#9BaSbeHU(Vbn@ev2=oOUn
zbcy?%YxT1cL8wdQ=}zZ<(C>8}kbDbRHNtO5+%H0%9r47v?s}Z@hUg1bmX}|N`<`Oh
zxTO>1!OEG@F936YL?{ybpec9MSq%87cKdo4upghfI!N5x{icBh)*x5!UUsqufzw_x*2g~UG752U(f19rY2&_4i}t
z75tl5Bn1w_dW3g7o+1IN#Kr9(?nmhjiB5LFHAal@5$}W`jb|yu`L&a+){dBu$GA%+
z@h&;ArH;o7^bTn**4qKM@py!oK_1TK)9vE0f8JPI3h81lCI%3I_Mev|pf
z!&E_j!*l%y;vQ76*3otVeDlunEAh^6ed=v+46sISMdU`XzxP~HggAfiT^H!=2KkPA
z)b2vKhivwSf3gR8rRmeDaKJ^?52Dinx3*MEt$}!nW4e;_fnT_kbM7LHr?K)=^FcrX
z4bJ<-K0eAkm=+52{MI^rU>W4QNVQ%sALO5tx40*R+*q4KjQG8}hfzLA>>qKLx>v;f
z1KXX>Z2b1n=bL;M`Ucs`;MGrs_&YyJ>@N=KINL=GfiKB0I=v#YbjmusRuG
zx3*1r$dl142_ws_)nv5VB2^k+Nk-hKj1`z-WMuLo-~+yxjO0fLk5y&#!hNl_(!{wJ
z>6Q4t(GciGJdUB(S1Wta+I_asa*;hKSe@l$kx36ycR|rRS$hyYlp7N)q@Ze_W!bC|
z6x7vfkLSFmp<1l)9Pecs{6Cdo{&Ad!bYe{BFA$?4)m`B#^1?K9d`u%jL6C-~_4J;N
ze59f#qfK|!e5uIt=GKiVYpBRL+dkepmX5HU4R1~WQn`B~H;2&CVJ|byqBuH=7cR(b
zJWEGo5jWzY6*63LEsm
zDrm@xW5@OPz%P5{+D5fe5u?L7dg(wP+Od1_nq!%LNb>tT+Ye)X$m)^+9$`X7y0pfJ
zGfKTEV46p8VQw$V<981eo!5s-zVDAOywr=Bk(wR9IrgCB>^eq(DFqphZWTSxMnPe(
zN}s;lNkJj$P5t`$6hxV?-(})NMs2=BGGY&4T~5g6{t$)wt>x}s`<0BuJFBwXG|0#-
z(zmd?n}ptKrE;`?A)$SKISvnpNr>~JNS2~78I5ox#w_fDKJg|yCqSPVvTWM8d<7X@
z9lsx0zL1QndGq+cK;P)bB_k>;86E%V{V-zz8JXvQa}DYvA+P9&QU;lXLOj|Y=)yS0
zit21*G!hyRsK2TV{WGbdR$UW9K|gL#^b$!PlX{HW>6WOV51aD)U$4~pG^kMgbTL-Q;;S}D>r
zv@ItTiVV9hoG~{WkSMP6XOh!tvGE)(&D9A@qxq70b
z2MyoIR$!N=qW-LPDr@O9^lE9~QP&hY+S|1@^sW^h8R#dComosrnwMgIzrnmyP*vmt
zeKPVW^Cr7$QxIdrDXlYRz3BS8dg>Mu6)if;$$!O*j)tW8y(LZPh%SW0t2JqouAQjqa9KmFQ4
zGUBsX<>o^rp=CT(>2-kn&a!D(E~22jT6dqxr}QED!tE|6gXt*zZ6GRpPeu3ZO9s8<
zdr)+IE#IkKWE5QTqPk;)`YZ#o(9bMF#V0Tm
zGFq?w!;=(BLH)HZCw4~nqL$=_#)up$B0s5iy~0LEQ4&jo+(YST{3xv=j;ywvKi
zEg>Vhq~y$*K@!rO?B7fpA)!0D#u*Q)$;kIM&sD{nJ*Y*SkAp((L(~IHMjTUV$ZNFZ
z{F&W!#F0|=#5stLt`%2(JbR0RmX?v0q^zKz9S)^j7q!SpIO%50BB)na(PAU9?_^Zz
zcg#Fi3f3>_;${aUDtc~w-Ff634J{m>v{?xI#KBN5%EXY0J`d%{o{a26MJGF+s8{r$
z&eFPWiBdA6Zj--jKp~;gy%JBhvQv+lrpyGQ4*In*W7
zQG^X;ZPYQ?FW>jxUfoI75Su|
z*@MdM%UqdnM;Rz5!G*S
zV|*DIowbV$l!5btmS$5g{*sKWqy$6NU_F>6df%_aP>^>MI-3#IgC^`>Rg?_(A?^da
zi{>4sqgT2q0bfqi5TmdC)R~$dgxPs4IIT@a(!pv1L9iYku3K+BGa@63@e*Wv%ymioN$aXP@`|@b35j;hg{auzu@zP4{(Q_jL~{W~+gbAfF^3MtpdA
zF-}bQ=g+w>*6gQ2;g@8N&*|-Ef3VGB^eh&z|N6m(iDQILd{!1jq5+IKH$LyL@x8s=k2^X!|7`|y#e(2}`ti)>
zeJIf0_wN(u{F&*muRs0x=lq=9ojja8{r}!~c6)=Z`p*Cy*z9~gocvr4`5*Uo`Z-?_
zP|nTQ+SAY9H{h^;fUlFI8CbzA|C;Y#{g1f1|MKVgi$4)r!A0fgaDN$>3tLFk+27T}
z30trd{jogC{FpKe(wrN<=~!!X9X}^N=Ra3pj$i(c{{N|WA5*Va)iVYCS*Llw>O_p|
z=kMhGw+oZ@YcTj`!+3sWGMf%@ZW13hwqJs=pA5t1kYn~-{ks_vrd-6F0omEve+J;4
zjmN<@TLA1LQM*4rBC}$`xK||GNdQ~>Q^)Z$64BrGP(PwYiIx486NdS^c>NJ!>mR@V
zi63+O&SJr@X!!pYl<0=JEC|({{p|HJyUYG``WM!B_Nx1@>bvr1a{p(2iQ{5!W`Udk
zxs&YATKx6btw}BZE6z-0VR&+$;ZY{C>rAOi@L(c(R>Lm6aDJjE+zMovuhDd)%G3gAnM59)o%VWKlb4`U^>ndn?UmCr1LiEjDQtZpYU
zQEk<|QOy`8x{$oR_Cz=nO&_QYGd{&cujJj;LxP!TG|uDEF{$?vHDCB3yl?=O|}VSAs=^IWS12SxlLTl*HB=gdyCkn
z-b%7iLY3Hu4L2}6PGS0<;$;k1FEekBjKc73Gm(;gXD~drijj8!a8bDqXGu7Q2i{9t
zO$x;DZ_CU(+w3s>QTAH0h$V*mnyMu`H)8n3Fsn;_br^1JmT+Y2Qw;BE>8`4Hj^TB-
z<=vl}Fg(+1#jC;>7``cTYuPuD8#TU2oOy}iLN{)NUv0(kk_Y_fPR3*S!lE8pOFD)>
zXx&}YbQQyw6h0R#$;9xtBc+dr{V{x8j-kB57{kqtI<9X}#qdIb>;}O8|jDha;DS_?7!EbjaE*fiAV_gq!?U>5RKs4C#)W}%7O8+r?R
znW)lDB_wSg3k|)Mzj&xq@4q)Vb4S|v;w28&>y#V?k`F7Wg_dl=?0b12M>(|Pkn;^SkLO>
zkqCkHtXj3q_c#-&Dk)b!hW-dcdp&>iVIqIC%lo-uo!i&!m{D|LqS&D?)V<(;vg^5U
zFpP7G{}AgF*l+Nx6gqf;iJsc>_LnbWA+^1`(_ejKA;r&|!zf2F+#wgM@+-#h1Y3vb
zoedZs^7TvL5?F_<)r+@0fqv!+pO!U)am4GKkM})*zV3eEJF$y}vR+6hzn8%9r5#^8
zM9yNkUa0yqt!fNE6#BK$8rDy3m_z*?thd+*m%u)MCVDoK+V>@ui4wi^9A>^S(H$PS
z#a?$<$fawa*<!|;lWbap$a(?YwQ
zMf!PN<780NCaHUvbc%;j`^6SkQzN?1$
z^eyb8Rl9YJ)7P^Qr?sa>63hb^-jqkuT74Z;)Q}iiH$L8&kS>V|e{e
z*3SKKuFUIQ`hJGkC*6xx%VFN1KeUZxLOd)~-L!~N2FSl##(hL_867ly$3
zZ>F2L#0L6uokQ>9lRucKIH5JM1oq>;>WAUe5N}Jo-U|gKb|HwgTP
zhLe2@=7By(GK$BLIfS@A)EODEhR{%kxw$#mQ8`Lv76X!v#3W9Ee5;bCY$@=efbQmM
zuwQbZJajjVQ`5^S>j&{gjWxS`A@2C$3KB2a{jOV}WCmz2qZ)G@FjP#4vlrrIg&x&C
z23%HPXGT2u_lp~5nL)g9;pVTmARh(K+iN-jEuW?}uK;_G>eb7VKtI4pA~}O!PcCV-
zH^}d&SC5PWk6vDV#+GduJ(6arm;esi#gM8%pJ42$qyoIZL1x27K;69;In=?v-z_@U
z4zT0MK;b*Et5x^43V=MQ)NPuMz_Uedx;_HN$x}~;0ggz>r%8jqUiazfD)7r%z2vSc
z;5XKy-=hGvZn2n+fKwaIA8!CGZ_(=agnXS$omzdtPuxcGJvYSp%yVM&4d`{v#$LVx
zbibXSybtiMcU_?+V8@-R%G#zT8&B!Hhe
zG4Hk~*w_8m=d25SnM_X8OF$nVejW#qZ#-J|@C9%W?^wHwu&y@N5%1ckhL9U+RXP*o
zt+&>{WrO$;!)}&Z5Vt|A+j0{8l8>2crGPxV@(`*8e(h+g<1(;s99?whJIGrLeRn1U
z|9Hf*vjfm8eXI=!++MoEK^^Q<(?g8j1D~E!EmQ++usFGRKCI7VEe>M9bS<4j|w&12QHPGbIl#bXLIM+H412}
zx`!?d`ga`XY$%}bSxuAP1$eGn&5hUxO*YSS1p$i!X{RayxjN<*^nhQ5pY$-{XDg$3
zv;gd1bSyct1#o|3qZ`rJ5nj#4hXDsHGVnWqwRWbCW?(NQ(LGM&DSn9jlppYz3~zcf
z)U~HVM!Xa3K1Nh-*8)5&252r
z`#xVaR|4L9_3+ef;4{&-n~42~ZBx8(3F4(lNyT;p?@4@jk2nwBW{V5U0A5IRzD4ZE
zhdhoMD*z3r`3E-wR*AcpKLC5X;>=D_;HMv5E1V+g9((xU36SS~yyr~^v`{J+p#dH$
zlfK*uxbyvvQ$$~cu4?nt1V7ehG57btkB~m@a086>%daQKJ!Epf=OFOs3+i7H^-NlO
z-ISEASAw14VTz47V5CNe
z3bF6@SDbs!30z71XrDfCHQ%JCtAHQ5Oy{TpEH6s6{Q>Bm*>;K8FL&1EhmyfAMeD)N
zF^I3r+8;IqJdwTm%5tz1+3`5Z2XIaK{ojascW%Z{C;G)LdMSkq
z?0gefd0z*7Gw_DF9^_gDh4Dnc1n+ZjSqB`otxhM99;hi)21uJ9@1+F#>5m-DY=|qh
z$@{h-%Y+u=x08ybES6Rhs-Cq3xV6#
z7=9)CfCG;aSOQ${DIP+Dd8&1NP#pl&tLSo91i5<9szWEi&Ng?{m&pJ8$?C9tz?a`>
zq^E%5Q`Z{z06Le|uOsF&?k!~z2R!_B0L2*Q%s$@*wuYq~8b^eeg#%*{dtw)S^QGuk(4))#dvN`&oU-t5X
z%2JS5CYUjez+Xl|MSdCR`z<-S4*=Tk{QV9;pl+u9qgc>$7YIyRLA*>(#VzmPzBCM}
z>Ae8@4de4XltF)Gz11yMI9~)+1bbz_!M%I!DlP-#IDEE;|27;(6ZdzW`wU1MPjMFq
zKeN|U=Vbtos#)dj1gsU<=W!GCM|4Kx+JR@=dtYvcI5M)yzuCZjpG(@%nF`|zRFYkE
zA?`Ch{>S-{KmQ4>eHVa#pRQBt0Q?czzTpwX(Y@s&x)|OoD{7~@EJ3auWo(`Y{(a|p
zIS&KBd`nt?75FzBiNvo2?x9qDe;w#!u6qs==SI!aj~>K*H4$!6?87;X=8GK&R{(kP
zx_PfO08Iq%Nlt*iD?uxZ9XKWB>LEVh+u~}MJq6qxEOwNb?+BONH(RL7$oyWVhrlOX
zqcU1xe$keT>au`q;|)PafJ@zGS8D;j|KkaroSn?^Rhqj+O@%fuNsERP7=9&u(lx+74Lb
zGWKN<@QI#G2REQet9&4_&IexUT`vPa+M|eqEWo4^O*RFHJG@(MJr($xLoY8T0wycQ
zihcrnn{V7o#JPU9d*L*3&hnr3ERF}HUvnNN?z^*##!PAff0TxNCCw%ztbf2w=A
z$P@RRz?1j&OCTO!+UMd(h!b29DNLLXaVfzqL4Z3H>lTrr&Slm^_Hy8NPor4p3Gi(}
zCBMf)ec$wz$lC)Rxsfs72lOS=$(c_ej?;M7Vj>UC>ew-TKo{*+ulaz@={KJf_XLN8
z7ZhbmBZxU$VqL12E-!not0sNY8=B
zWSEE7V4|EX*abBC1f+w!&tKwO4DgM&LZzv|7p4g-5a;Nb#lb5#0-wJ-Uy_KoBv9xT
z@m^`$$#Yy4@Lsija16*-SMh!#);B2w-KYV+HlN-n0C?b5z_v+1wa5AlU(hphY_CcI
zmP|e$B<{&t?c@^TU8C=Qm7lnmZrnIoM5%)R!>c?XuP?wl9J+5Csf6#qM`ZZwnn2-r^zex0SClKqpr!4ao(HC-T&ps3P
zJ;S8Mum|L4l(S>10gHY_DiZtP^^I{?N#K(@T^roM{_9NEFmZ1WSp*mF1-U{0vQrHp
z4{xKG{>+C{Uo-&x%JkovzW~%8;9S@Yc9j!84#c_z@4Q?>>@$O4nnxbs$G1ZU2LMz3
zJwrc3+}0Jz!Naf~j{VN3Nr1xf@!N^}k-9O-#UA*UF|#|wJ0W1p<5c4Ovf^m8C+6cb
z>GqU(m)ILV=l217wJ|svuAI-teBd2j%wa>No+udE@++nChrv8QwT~WH2KmlYZPG6S`Cy8nX9~#8b+{ynzgPD%YleyaBj)z}
z1u_5KT`uQ0ft*8{lDU^eLc8`u=!+0@~5(;Yf
zTb9QjNkJdm9r65^bX0(qoENx4M-o;dMaOYEGK{t0n=eU6+LjS23XAC|VPZ?-MiDw1
z(AR%F)=5LnrfRp<{b|Ukz<6`&8XD5hb4=J9$3Pus&96=X*6|KRX@@e9p^uejX*>fR
zUtFBsa*lz+^~!8prNU5*O(aEWIX+6qL1&w}0JdGE(TN
z%kkJkMn|LkOZxjss9!6MtLqa9oes!%dN4{t(wD?@6c>}x1Xoh*f<6*5_9eR}LZ6th
zscl}qf{daj??%-uAfwj;g+ia8Zm}y+yt5A%i588WUN;B$Ln`
zua5hAFpi<32FC=QgeHZXvXr5JL^ajx8$u~a>E>SU{%9&nD7Ur#EtiHYU$!Nru`^KD
zWxKfHUj$0@gxQktykE(Gm?RZEal~8Ax~%hra%ibGSY}!I~}=-
zg8UVg>!-R0&~$FzMow88dXlqVWgUZ#f|djybx&m=)xLFMw{00n!ytL$>_P_OxE$x-
z3-i8+>Z0Zwkdb?pFWFs(f~Gf});eoNMOWW6(F{p6#7N~9y6VF~GcrQH(&h}rD~pZn
zJ3vP!=Myi;J)@zUr?@kAL!HV+GS=PrPC}y36dW@;A)feX#Ro$wlDCU5tg)h_qOl`~
zGNBHt@8WgZ_S4~AXfM!|Fo2qAw
z7*e4km%T0Z%ggAfQK~U~hcyG`<+wasH#~?6rFHIEBn+V1XVM0&ObSwHG<_2Y`!{OW
zi@X3>f2MG$Q%DLKJt^`{IDeChXqxxOXzLm1=-S`63zgE*i#2Tuj}BAOtroA6Z6Xw;
z|6Pc2?I0QPTd(r)qmhsTe_h6Nz=P*Fw%CYLP?6T{#|o)~NWWy8+o_YVk6s6(s<$+B
zzo~rKM*;dfp;7R(B^d>mSJ!uskkHBR2`xiCBvfsJ$!H3YQT5FChPQDPw8cRAp!qr)
zoKqh*DZ@TYt{QorznzBEM?)1}-=Lsq{*mxVF*3?MKkh3A^LX8s9})#PX}~rl2mQ>aMJJzskWtr=*sIc!n|1`ZaosZot0x
z^-*uSwup?@C8uQ148y+o_C=dCMnbo5m}TCtC!@gM__GuX1`vh>xhOQ)zXukLIj2EC
zj+b9JyNiKFQmQIlL!kcm>N?Neq#%(h(xTKA6tv5!g6EPJoSVr7v7%6~p3;S;l3&S)
zE%4aRI2kHBo_tB$$&`j3nO%1sJ5NUw6W^dB0Qz8TD*cxNSrsA^IMvN@Gq65>osox^XQPO{*xhaF7R)$^O2>)3EQJT)5`+
zJcWT04^h@dA7db;uX}RYLg~m?L(uEd5EUJ3Pc@Wfl9Bea?VZX{$J0eyT_28iM7q#W1gGlhe
zuF`pj8K_JzHR#hRsKa2_>9Y+3h{xV*{uv!IQaY(F90Kcc(%m@su_+l1I6Hj4#zjVF
z^KTWmC6bX1=QRti@B#F~{JQ{En2rvYs4cn@1m`}F@Azc&Aj)<%X-{7c{quu)k9UZK
z7Tc7&jl+5b7v62rf%|crX;6-!C>d=y{^N{y1O;h_@AXe!OG7RbrJ4=UhpsbqDy2JU
c$lf6^iBw2IZ(r9AjKlr6W2FF1C?(<4gdfE

diff --git a/previews/PR1624/tutorials/out/solution_000150.vtu b/previews/PR1624/tutorials/out/solution_000150.vtu
index 9cf3ee47e6801df818d1a15861fb4804beed6f31..aeae276bd45a5a9961717b73b83f9013cf6e1553 100644
GIT binary patch
delta 12224
zcmV;xFF(-Sl>y)o^#gTYpw6FE^A6lIPDb0-d}a?a$P7EtF==~cBS~sz@*#^>`RShYXh7q
zULMOovEGS);(2!Aj@gbBbEH@8z6q;(DxZJrX^OR;>8Ce1P^{whYIN~Qir;3=+LyzA
zy!Juf>KNKgPa^3m<8^vv>@816UnPR`-X~%b66kk0dts(e^
z;%mWD%QQI@56v&R6kbR1F+Za7qJ&}%etA{t2NX+x>bjhgOQ+Z>M3cu{8j)Y`MHUo@
zAW|QG!Ygx*!2v!azCT7799*Mt=vWVfx0k34+qNp`h?J*=DrVk8
zWQRifAa4#Lk2|fldFCSGl-J9(hSlJ$oB11mc+bAgv8f~pzUvpzGYdmxos`_Is%At)
z%eNZ1*CH}B5Vf|d3hvXw`6yk7$fNXzN}C2mF2)3=7`{Nn)Uo5yF%BZBb9-tOTM*HG
z``4>&t%%%nu@KCFy=l-6^Fny;?(okOb|WHM`g~UNpCNKoh~Hib*38x#JMnr%2GrAk
z*isO0=8EqQOE}-{xc}1vcz@B&VIKxJtR)$FWGf&~*5=5Ss7|*6u
z!rpn#c>#Rqz@F3kFYiE$ZHAr?5s|WTy8D}r$OO~(skjIc^E9P9OX2r()9VTgG7#Bj
z+NbUkhe!{XKiM`A@?9O976yImnXgrUeZUzJUhc2%xig4F)LEv-L?I$mFtKjT5s}vY
zdt>YZ5Gk&WQ&S2>*GPZUw;FNKIs<7$(l
zd59<*A82?v!QjUuz5@QRPff~q^IL^T=k>BoH3396vc%gta)|hvb@143Mx-Er#OHC!
z8bnm}_KnvILENDeEKWOv5A)4L1jjO1U;ozQ0b>S-{~SbjWf*Lf8)~1ifWeo&`8Rm4
zVQ`GGfRlt7gR>@{Etv&gjU}{Q*DzpkkIcg20DA^E;?%wv69)5V^BQ%mW^iwUlZC^-
z6x%T#+fB+SR;ZjerFDbiNqj?p=(``qd+HY{1bS1Pgf8h6gi&0;oq4EyhvMVx#61+nm4DhRHm+vJ`2=u-F&~zN|K$!Qt
z&?msT&-6DRsh5Zh8gDW<13I=)%(e1kGUU0zZr{;E$*v6
zLVhc}V+Gp^DOOXL;!zN0u#>s))lNSKEAPC1Zs<3ITYn5(Ct8U7OfXZv0QlSWdG`S|
z;CWIok7ohk{l?T5V-DO`Ajt9Z^(I7I!iP#CA0W~c-4Z(t@ojy7`>p?eLh;MCKZ!k{
zrw&(R{S|L8xS}FU*;f>i2T42DmYO1R^?<`Nn`lJl`LWHjfFI*KO5W!J4;y}uG|9v2
z_>5DaUWiC;%xV8sml4_T6UW;L{rHq@-=CF3v7ks_b*eb%W#FP~6^RUv%O8I&20A7A
z-m~uhG0-Pms&ATqjEMZoLpPIZfQP$JHim*O5c`zH<&a0y%j)NK@er49-I$#vB6=&X
zN>&0Mo=7&&`o>c%b#&+}1Nxtl@FcJkaIUdR=U{*YBAdkdRVN(~*>j;onhQFn+ug?!
z1ihFvzV9IcI>5w-!4!9>FPjeh
zLh&&BsKD}I2DdZ`E(imj9z5XgXXJs1)#IldBcQJR+|YAspkr~Jj$c&)FVt>mG#zxq
z*WY)&CIQ@drRwo>7@W{*tN7cC;`LqCeA8zs))uh!Tl|`0o$A2Z0dMe&7m4o<0?&8e
zbZaCR5MiZ%UJzSW3Uzk+c-;}|;$Hu3aUj(Dz4#4_ZqSi;7Nb{wK)>qia2(%Z27hJg
z_htc)&P^o$8~~ioIfZeR8-W)lPwvb>e22$aD`qVsn~z8ce+x%snoR|yKtJ`S*Y4wi
z`+M)nVaWrI0u|clbM+C?f5c)Llz{(fxQanJwd?2%#1AY{zqlIwyO47$;Cd^?
z5=T=sMy@beUHFth!ZwIM({)_rCL#;G_@0&n{<>UO`J_O9E7v}ld=Z35?3}xqAk=Zv
zT);2t5Q7&*^!~dp7x>)VwUO;faZrHgU?tS2Ib^ZIPB^#gMcmF$O$^=|-FxD&IU^U=MoZalD`TTANY;ra2M`4%J14A#?LentEj#gzpcZuo*;NZ8)g
zvjqG!>KWdd$OIkAQ7fqcT?}duchv(OSTS78RR^E@prjRsa*q8V&b$TighA9)imVqv8N{owFh)c67qSwgP;;D?KH;;#4uZY94&bWK_@
zSaIpWHhV6`#g1p$Dv2;BQU3+sq2Ec4aY7pbS5dc3WNlJGr$PqUozS1nkCabGxiI+U
z_H((Z;|vZ{lYE}93Hq}lY(f^!*<;{;JAW_eo~e+%*#Oi}R4uVu?IFxh&dy@rz<;=x
zSULKjcP|fl@|yua&IWXJECC+=H0*7c0G?MJh>SM}zo8c+1m~XwU%PL?-t`0WJDu2^
zpp6KB*NO=psN2E3y$72j5qWD^Q!B3v^>eCG4dybq_*z$hJk))vdd&)OsvOnr8c%fcf_rI|OfAIZP8m|eu
z9U^6xhzxr41;AtPm3)9-3Z>RBs&i0thCzlum^l8
z)T`^y^c9NrYUU<3Pf*Nyt|IJzAHiT7e~-^=00;4&T4igX-^0pp=3fDwx)^@+(K!|(
zCP~MuZo#|}nI^Q|2=ue^^T`&?^9)WO8`3_|L-D4OGZA}$mv3xT)=vQ*c9d?3Z%w24
zoBYnF@yi$-bt0|tJ@{P=%cM^m>a;s@QI{LciDm!PD~|)N=5FjQ?|`{~r>Km_;2!AV
zM=j-7u1gSE_)sie9Ol7ZWOMLl>N1*#nie-6_&(op?mE;A8ti3CK
zE8T*@)At;ObRZtfA=`9+0{URnh?i8*79B@DEx|9KUYnqDsQ&q^|Xhnx+^>K=MS^}n{U>-_8dHKi|;H%;l
z9cerGe-t@rU!@EE**3Chgazj;G368ST@F4+75WlDmnFML1NA^R#r1nyo4{wYn>Izu
zziDq$2y8B&kHU
zgcg$~@g60SG`}-{eLrWp=iYnnx#!;J>EbT2zc{Re-*s!I__!>WMiD&?elBaD+=SmU
z$R7LTRULS_EM>)T&6aK(EjY*vD_PUOP;1SW@rQhSWgczndy<{^YQVYc9uCV15H9j}xyk
z{?)8;#^v7}mWIV|&5anhJ}=uO8T@N7zbgPty?L%}9vFRIL-&6%=%&*+F^TbOXP@~Q
zg?W3Dq|~otUUQ=>H7baob+?}B4SRZ<%F7_c9q!1jlEt`w>0A{VaDmK+@0&orO8@2)
z;IyN#y(B2N^{P_@s8nqHE(dW>rgV4MAdYL+_KBZ}FK4;+<{41CJ!d)v_HP;zs+Cy3
z-v+ky3e3Yg#hQOChg_O=b94t}&x$XrZ4oak(IJ)(Jm3*`YcBE^G%ztP3i~{VMFw{8
zAFKVO%f-3_d$q=vB5tL6gt-RvlEqIf65%JhN;Z8f80_HUavAzR8u@G2LC%qz6S9rg
zEp8)0`T1G9O~@A%9^To;k9qv^RTFd|yPxHE3&%LAydQrLTQM$-+2hs*{T}zY6Vu??
z>+!~v2N&z`-XE|FHwU-Ogq+BK^0XBgGrRf2N^rNX_LKF9BO4a#Xo&c&w_0zdLLVSA
zp=<;uY~_j>axFHpNiN9+yN<;QtUn%^VD
zWThG6v(_q|ilO{3)xX^adHKh3YeUF3Eey#eh@YIgwv-1}nb9d5IWzeJED$
z2(9P4?x(ZhKX6(oXre`c&Hru|zpR|tQY
zVO=U0YW+Jv1?6E;H~1F@u?_6NvyJy`&A|xm@SW7RUYP{+>>?%}k`04ix+s}*pU%ZR
zD%gwi)u)tt7`h+N{Ul3=^hEcEy!nL{RX=={|k71z^x
zUAKzsk02jjRf}H>z>l*+LU23c4EK$>QT;a-jfOOV_wRh#bcOEYu22g3RUrubHJ`EX
zhS2AS9!k1^_~S0~4Jlt2h4X)u$gahc(n0kk?8?xj{I2e7uAn|%=etYk5Y_dcrnxVm
zH~%7ZiSC13yjlMey7v=rJ!yU&-qZxzm%?*q!6NLVSAJhaJp8iuKCzAgcMtozFMwbA
z)Y%fM!?LKx0XkRSD0u@qpDTrF@5{lO$AS*O!F?ky>#1(^oF~5rQeA%sXp7PNBA%W+
z)`<9KJnSy2*OYVOl6}x;Rv+EG4)TlxZb3I7->yn*um|fagTp8=mQ#VCHgV$J{a
zq(q;whky0-IT>%z)auRobm+eceO};%^);2VN{1kSW^2tifW1^fzghtGqhJxrp?>x=
zu8*k*Ce%O7IR(GL27Z4V6L3^^UtSBi?7+0I67_qLq_IV`t~>A33Zb8W@1>a$xX$-b
zwK8~gKyl#;*!yDI3%1g^zP0tz0Hqf@hEu=&a7m?K3bd>;`Jh7YQ`h8n+UNQuV;h*@
zu5vk_8k*nvxO*FzEl{0E`dQB_1A`&gMeemb0bcJZIPwLQXv%**H3~NWZOZqI-s6>i
zTC{H?Nu2;%_v$qt)f6zE_fvzzX1cEl8~dsMf6B?WrG5UG6?2L1{|K|nit5JBsIZUv
z!k=1KeCRyFcEq)b(7N91Z>6|5y1PO(K^`aJl&hdZM%?yQw2!4e!qhkB=nm9Ves51-
zJZKCaJbFjRlk9(GUFYZ>zt+0d%pc>9WTr&i1HTmfV!on%rEA>}1-Z+WXH$Ml<(lL|
zz~?iVhv|K8Uw@ksOz~A`^(E1Ll+8Lb6X$18k}75ZzRMo;y-5E6o8W}sIPdI|;@n2W
zTi*Iym+qNbazgnjs;|0&hDTI~zkE&UKC{f@CB@;l(`tV}q6l?p!Ok3Q0Gm{VN*7Xo
z>Mp&db5BcDj9y4_LY9tDAM@U26hrTh)4PbrbUu5OZ3Ewme741-Z(>h
zAL7i)zRQ1Yfd0dVUQa3N`>FM=dLY*-t^O$N|8l~!_Q1bzcU$9WP|mf%WB~f9s_qPH
z_%(*iuBCTJcf_^qF!TYUel^mRM*+6U66))Uswb0Rf2b{Nln%yaSo}lh`)lrj!8rKE
zE5v6p>E9=M|1iBT7QTb?=wD#$oE#$u9$hODNbi5X#L#8=a(ahkC)#?czbvYArgM0}
zvTvn#)<(7GBK?z2GQtX}ujWlXk)wX($^2?U|FzW*?$sy3&taFeJN;)F`rpl3!R2;+
z=T2bWh@q1s_Mq#vXBm6p*XAlVNd1#5<)^Mo??bj;q94@@L)VDjg-lCjcRI)6tEn<+
z$oqd9&)HW)K)G$3w=kjK@3vum3jDlBoUYNnCMqU(P=BsXXXyS#@2triS*pWG>6J!t
z^e$;~-71iutQ5VDJW!5@)aGX%_Y@o$()6W!!zU>J^$Gl1
z%1k6iG5<*8yRWqVm0=#8^sg~7X57*Sg_VD&XPt$9>`FyHAg5_KGqle
zVdK?)x?d8fZfb!(!CRiuKiskMOCP;^&&;08p>w!zbJ#PV-nY3tiFA)uqh54VUdweH
z{plTwZL)Krdm^B(z0_xCv-=4z
zYr^k8y2tm5lg=j?e*y~t000000C=1omU%qYTNlPfNTq~Ch$CYcInsbQ=13)Ec)3Z1
zB;HafQ-%f=WvJ7HL?}}OB|_bhF_j^NLfvG{p-d@ypY{Ir{j9audiHPc-`Z>M7p^9|
zTDV47r!w(9@BMvMjfB%2H8F&A1Sc_6OrHF!x
zjGgP;XF05!v93`iGaOcT*pu#Uu&=527;?`UVQpF#l$MS-w=mTQ^6)P{s+=R#*=7Z@+GuKL&cKM#)cr?CMDm3Fv6Z_cyNv#hK#+d*HWzo76uD
z9V_>i-@Wi>e`Y7^Cqbv`RlR!GFAgi@`OB%eDGqDW&;D`+=&vbhmk$3u6`8~IeYd@y
z8f-%RZ_kplo8a`uy_E-{E3Px$Ee}1pGG|>Ea3(=ZW&o^ypvKPxr|Jwvw;}GYm?tu4
zra7!NJ^mFYlN?sv6@mJ8%%dN9++6fGhn2rGNo)I&d_#-~*d{~?aAlKUXV7Zlo6{N%9Q@2OAT1s%F8gQl>K
zRo?IRo`P;txOe1rn#b6e{{tA4a&r0&@yn65aZ?tXhqxO@+A16|uaaKuHbF2@sLSXF)tN_dehS7t?FQ9~
zXdY93z6h{SWimDyal4(2ubDz`wSjFO1Nws2ZB^2kj~yEm{|j+(y6=s@f}FkL6WNIW
ze{sUUkMf>dhpY>qw_Hn(W}M>UQ2n->PCE)c&OAMY;H!~Hy(V4e&hSV5zM3X
zaW$_M^dB-g5>(eY(L0WuM_fqGsRPB}GKrj71^82~PO2*rZ~nWy6#jUZ!252X%m3mv#*@=)PKdxx&fj)64|=a6ccw(auTd=-FbR8iSVHay
z{^zKbr|Nsq=ebF*3;a>5tPa&;-OP9Cx^({e)gBk{V4ZEev4Ps4qisdR0r+40e=O*j
zOZzBz&qN#fY->i&JJ^cTqvNyCQ7?A*dKmWW&YxdBL9Op$1wRm1WbA5i1M6IqyuDV0
z&TqH{+ZKBLmQF5@!Gi|PL+aq-;M9mJ)ZJX%p;LHe=z9_jyMW#
zQ3baqUaO@(`oksW9S8YCYgnfx!6rAMSUN}QTJ;S^|fDlUy>E*
z-|@Mj9F*BEyn)se=wG+>4b{o`$BkX!@te{Gx!_v%8qZ;>lfAn%sqZ%T_8(2ae;erO
zOEH8$?bmZRT8C=z-0?1qe}8njK7O0}yz{C$Zt4ps7wMRQ!OzldLcyl@jW%CEUrqP!
zzoGwP>0VoRF#3T&NFuE#qiZFT?whatU5TJ`R-f}V_(jZFGFF&ZJ1EK{0rhe7VV|-f
zSS9r|T?ynmxT|p|DD$V#TpDp1dROi9kbkv0Fgp$T#q+WtnbkL*GnupL<~jtQ4u_EJvJP
z3hw=v6*)dX`eQWT{ByWxXs2nV%eaSg{{FAFzhL7
zPakwfmi#lM0si-le-N7soTyZvas@TLCWAkMLu+p@Q$c*s`S9clu(*>qTpPN7mMnj4
z26op*8O757=QE8z1RnD@K0OQ?D#q`kdt6N8A#V^EJ2RO|_i(QDw7eAE3+aPDU&8jX
z&ozmK?zH@(M{~f7TP*{q{{Q5cT~R>X_Sv}L75Gm&HU^jee}TWXhdGA{e`$ZC9gX|#
zxBN+c>dwiGS~}mw)eON^h<9pl8K?8LFSE;W706Rz8@L1cp__GeeCR&r=4W07MZ0#^
z#A2PEnYT9lhRwyc@O%ay+`Y%lhx(c1X=^{~6Jc{#Q@{W0b~&2*3PW>Y6pahLov=Mh
z>lv!Mb^>_|f0Y!<6+p(ZMQ{1xZ?xqPSO*I853Hqq5i7K}w1#aOt|cG}mOYv`bpZZC
z%dvPNT6cD^t{gZxrt5Vbth>+P)j(YSfu;bO$7ZNGas&8eqIQ7l>X6{oSRQcEhHM8(
zQ2n$3KYdrqnWS6lU{I-iI=zp2jB^X5!2JSVimlk^e_HMy2MO3$F6r;1^HQ+kjXCvs
zqYnz13&ArLv0L7Q1MRL>)`yx-wy;=@@7oRcafN{N0
z9XqEwt
zR3>X@1bgb>#4Xw%f33H!)Q^7eX*7%U{o@dR=Y0-iQHeU$~q^=Uh@^;>R%ZlUl+W{9p?;JQ|=ry5p~swmaCLu_NDX
ze!*@HlXcF*_<);mI`v<#54o~bXZ~*Te^v9S{*$7;OEA9Tj$;_Tmt}%X8p@!nQD0O<
z{buNK;EhiBk7w{@(mkn@T^iF$?joDtd?R?i-hE2KTQpEY+mCj*gPIe*+Ws
zQuQyuzoYJ%Ba7~t!3vRT_-C7H%TH09QA7jv*<=eYS$a>X8d|dHJt4KjH)c2DMMmAE
zi|D=E?jg%Z?~XkdF)WO4EeiIb^DCsomq+iohBY~>|Dt}T9(+~@^=q*9)CTHb0!}6!
z;>bIiAEvGaoq)6TCJSJ9tqe`rAW!ewS7IgfewH_NE&K%b?`<14rE}lm$<~1VKdATc
zXOqq+7=Nq`00000004NL6_$B8SLqwZE0wfJWsk~EC~2{~Cn>UoN?NQXMJOs-NJ9ui
zRAfqNifA!4OhgMxWoaS$np9ML&pF?-6ow>|e(&$E_qyKeTWYlW|*Y5m}o~1Uoq%qOs$*SUXJz8etmLtpmC=
zJQmNHUO$D#4eCgA?H~owz428$dMKnv>D`3eN6#{
zS%3RF#?_@$II7lj{XhhTk*mG?R2HzXIBOIsYL;N%ip(D^sS+rKmCHQ#mO$&qsfyn$
zB%mJ1SJkOXaBp*OpUqDO-34XlpSv04d>3{dYk3JmCrW)_GaprK!yF6FhhQqS%J@f4r
zaX6%ETI{XOVXdVioj;XBZ;t;{=l3kkH-1QL&@&g`6&N<5s-QuyXzqK*uHjkqB`lYDmQ9V{1cBX;H_?Clf3|e@6Qq6eb_V|+e#n{Rn#|*T9z25P@Hq6FSlcy2KuuM0|I7#hY)$Wh9$x_~M6weKw+gtm
zIP^mP9s%3)GEavb5%APb_MO6U0mnCX`Q1$?-`|RF*3J}A*Du>8BIjg&u{H}&6kz|%
zv{w{B_LT!QRKo@Igqq4k#R<^&Eq}XN`>O!cH*XqWrU;1C8ot|=Dqy8miGzKNfa&@V
zJS+VL6i>G6xVv0Hn^dIT()j|SR*SwhXbaGobm6(Om4Kl>z13!+?u55Io;ysv
z9ynXTv5R9Jy
zRIlN%u|>*1zJS9-H>ojE^&Awhc*O3o;lVXn%C}$V;MU0e`FSpn_(l!uvc)|9OIC<3
zR^(A0Vwm}`j>Bo4VOj4L9NNAP7&cyGar>;pxFxe$Jcw-lG;~V>m1eW=neh_LsBsL;
zDU!fTSLfC&TNcGW!uHZC34f-SM8sV8XW=)oDxfuih5m-EJty2)tbghsBY7i1w#l9Z
zFG~sD578CICm4JjW`=Kmqahdgv7gJNQQwrHbZQ5U`$jvL%uk_FAS&8Fs?5MWQF+tU
zbu^|eTM<$`Orz-Vf**-oFI1y3L4W7+%giej{>!q|
zYH}sIs&vmEr$)hD&hwPxbP7FXs`nXf3i6dMKYNc*_;(Pb0~Z-I<#nn?C-Tq3hHJFAKrX_bej_6=9Sqo+JqyDtJ93qr%`y4c;v{#
znH1!7Cd&s6ixH8Mbboi0d>6O-&276LF7r43>>Vr8mp7aG6*UQm(^c3!AADH=Bp)=Ty<9E*_G!M)!3-WKOyEEnQ@O
zUB0;*cM+~7Ik2pc4~5d>Mkk+1(MaX^YuqOqxt)6PdnF7+ZGXW@Wn))_!VaQW1e$ZU!C{P;3kjoovpJPoCW;6
zy6L_-IrmP1!GDG2kpg_Z74~oZm*{=toR@Z_ezH@wnv2g<2+5WSvph;;&NCgerI#3Z
zzvS9X{J|N8w)E||t}bn~T%D}M}Zb;db6O<~dfwZKpK-MiF{k{A=WFtyBcu_O#kD9SC)ICBw5kIohnw)J&=FvW(Y2cSp3WI6x
z=cq{+U&W!bP3iH8@i!6Ik1
zyjLu2CQHJC)g^fLH4o_XmO@Q_aDeM1IWB-CB93h!fZ;iUDJ{l+u|z4a0a(R13pX#ZX0{0Z#e%3EGg
zn4UQ{Uq6Atog~j3mptM_zg$&vjOKAJ?ewFt!va*x({9wR6EJ48R=ATakCwMa|5}nh
zT`8h}51z*#0jHPh5e1Go?l_`W5yP9k2^=Rl;s{Cx5DM97Ql^TN<
z98OfwYMK(_TNV0E4~QIZlSG}4R&|tP&sg;qr
zKRS}${H03_^T31tmSGg`Xm$OxB|H!ny_f#~+>9joxU~|W^EkJ!h9bUVefYg+{(mk3
z)nSWv$XF5oT;bg_!Gguny3~#uc?Oz)o!?+bxK?_4AX1TVP1JTJCYh|IooA}E$5UAS
zI{QUXD23gw#bf?bq|souMI~d1hV`%Z&kYHG@>)(^laPCLuldA3(c|!VHTU6;8V{SU
z-qQ{rc@NSt&%!K9%6qaZ#NmOIR*y&+;sAV~YAfx%7jgx3xxpA$b
zPM1Mgl$L+`AqgIa;>V{B7JmwJL~CjZw^{pFUTZCQB#sn3pC@5qt9w-ER)_@M&*vyv
zjM6AHUE1`eor2>cd5?e9DaSYE_3Jzi}g?(Pu{B?i@GfgW)2t?MGRADS6uF2u1ewA
z?1{rO$US8VGHW&we}8hd*WdLyl7_VGqaDsa7^sEk#6^<%_Vx7EI%_Uq{BGOjEkr+c
zfqg>^$vxi`!`j;`IVfLD9J68qi$aG-x!o%neDW>f=zl0=S3g$rA^qtNh`Dx-=s(Ch
zbXNXF3XgoPz17KF7uIFPjwkOpd@5_7*;@|Aqdg{VBv*x441bh55w4fWHn&jY1T;ue
z>NN{F$X`nTxUWruZKlbJoD_psGjDE}y+GkuP;v5~pT)RvGxVH~GI^Kf498Rj8Z|P(
zGZzlh7=5@UM5;@IH}~tEuRh^0VwGW3NH}?8W7N_la?b{Xy6j3)w~L#+4jl6*`C9H)
z)6U%z7`#?6y?>NTBTY^HREs+Whw(Ms2BIIi2WE5ZLWquyItTTFDU9lSCE61|j~vll
z^z)qrg4Ft*=esyq5529I1M$_?RkJQ63mAphLfIlxPw&3#+sM3a4{>P@_=os}^p~H<
zrqOVgOYHO|^>Vsw;Bt=Sk%()h&9-mF&=2D@3&Sa7-A$UU+{qIj*q6*VmSYj)Gx5+L
zBsXsErS*Nt9GFY5x}8J#v(&$L!**+u8;nb&Gi_LygqMZhCwZ;()BW%i!c9F#t8PQm
O_x}OwLG;j*ze*h|Er06(

delta 12305
zcmZ{qMOd5xv!w$B2p%lB1cFQBE+Ig0cY*{67Tlc=x8UyXF2UWM#vK|6ZjDIO5(MQ!oMiP%z(@HpYKhbp1
z$I0S^2cg5B7hzVwo$!5SIH0VA;SQE>ljM}VvqH=?C_xt?i0nJA*x$AK
zwRjN{XS_nc8SVmL;(=;W>%dJVEi$sICrFiRs_xIaQ_3ALWwyG3WP>LghEZWFs_QB2
z=K7ehwAH@#);)AvaHSkf&aYPt)uj%z0dF|Rb#>)7aA%0q&DQ5pqE2cn@ny|6qQItb
z{$W#@6qmkEJxlzO4Un;rCL4_IlDPIj8Eort98(R(lLIVrgjq7a*GGT>hQR{!u37&Yw#Gpo5IIYlToW(5^7a8HaY>%ci3+c&g1AT5=iZ;
z=aKZEckGjGL^|-|=*CX9Rine6f3;NpV5QQ6)(CO9kziww>I|21NdyQt1aWX;Gbm!H
zj*NVh&=5Iey4RN|7mT)Z8W$VjZgxBTQrR-MasyyIL$xb>)J7E~I$$~iL%m_rvLmz_
z+|(QfO&Z{$PjS)y6;|ctA5lC;JwO_v(%*H(izE@Kvg6=JpTOAs){xh&_sDV2kHXxj
zAZ+#%y^G&J(5UXG(l3kpeW8P7#gZ56L+2MJuGCa(E*{a>qLKs3eoMO8i}T6EOf#Mm
zB0za@Jq_iicF;Yh(+Jdl3@QWnYX}FGJEL(x`}PJcC&F#U4PrspGCw&wO-l{Rbi)+Qp2|c`9p+@7eC<_
zQs@7?QMt9|T6YTLo#A*FXLM9L|MPIo4)BdS4vY@_fc|onCK@X3oaLdrt0A%YNxqV-
zG4%QzLSnfWBW-T;H^TTMl}Lk1nxT>1>|vMkAYxp@iNw_XH^Z7GK6{>0DvsWuPmeNR
zMTv_r{=5aW5t3TFv-fUY8QCRMoI=+pobHk&c75b^LQv7-6dv-0xl>L%#Wg
zf3&=ww5K7a!*Fmy1*W@-w@HqNCDF65%gu0e!N>c)b~6=2l8Br;bstwhK4781%9y-J
zjr5ACO2adD2b_)4d)?Cy5jN)tVkF8`(RDKVg-E3_CcZw`Q!}Q&nU^USc3R4t7J)?0
zyz|O1jm&qS@wClw2_MH080C5Jb$DJPlMzDK*Sh%QOmI=jq279WNYMsKsq!Hs=l1=v
zi-elmPZTOF{!d`dBkEb+pFno@yO)@R2vY|H(1}YkjRbtP4ce_!qd!*L_iq>zoDw{L
zTco=}VKx_GJ60_D+CMY`p6$4>_d9V{vohSAjBD74#|r6iS8-s
zB5U{^bouMqBKo<3M<6qO`bGFixmyWRQqvNf?JV`_Gx%i1$d45fU#JcmbM)
zJxw?0`z?x;V`3VrdceA9G%ABTEYf0y2g#q6v|jNOQ>fQXcd4%?0ov`hj^J<8=&-vh
zD<(!btXrn^gANmhWFel1V7M1>G(b8H$43{R9{N1EO`6{^)_wMI=E^dK7@llx;hH@M
z>&AF~E4za4@avE4i0c!f!{+#I@He_Gx8P~
ryG=))ppKj6(9N1M=t?xVMy%aMAI
zc?%Yd&`O6gP%z^mDA2z|3bXBj++K2bx1WxrFSq(dT+Dzm*TjhI&M*-Uq^222ZJA3V
z;=6e=k3!XO&LGf1isbcIn{lE=t)s0y-Tas$g14U8XV5^cQiF_*cdK_x0gSsz4xfFp
zZg2ZVBn+?k0SIGeksO%X_K6gpA8dUs(|$MKF2^-hiR@fBXZ^*P^l3pgw7?^pjgP^w
znHSf6mp|^o>Ye@S-bbOrdFi_H?SE@?A0bDd7sK=r1f10ME~GwCaaxbXzkB(jW_9P#
z`zU5P#0aSwNfJf;98Vyt1?4
zDN!hBl0^T!aP6I|e6R^Cm3Pa*Y4ZmSjtnU^MKHS+UJ09ye26v^{`(%_O#(L~_U0
zR21{qFTjfpLIDOJ#rnza1S0H`yU+kL`{aCrlf*W}^W5Om=cmw}zJObY&0KO6jZtNq1SbO`7*vU1&o8M%itDV0j!v
zxP`hozOe%aVs}*2Gx?|U2HC_&qL`|04Ii(L9o8Oi5g=4C=X1R9-P+BHLc3qKu60#g
zz%&ivS&T>{c>ki;964>F-Ad}W()TWkz&$-`gs*tjP(=Bu0
zL6g^BVI&+_LsMv64f#cnT<9}`$KTtI$9o{)oalo#3D(yaZAO#@#(Sd7lSBF|gy#b%
zi&uU9>2@<*UIUY@Vg9$fyTP7*%a=_ahb@nn-toHPEnem~HWA++NEGenLz~=FTP5Z=
z>HplQ`<6Przu_L6CAQxy&zVZB-dg?M``b%F_B#&rYgC@>sK`^c@Ojdum<}LYW3VrG_h6$cQrOI1LPYqeeojV35>6(abkB
z;tI?*GJ4!&Rc|zV){62)%t%#(*7R?rYo(T6*6L>pwuQA>m<7&PmCFUmziGRQN&tAJkjK>@BDqWAyN=;9+tyLKMzlYW-+wD>;#ZN@(z$oWb`J=#{+k_#
z7d~B3_T2kR=JO0v+94F@ICaLD+`1_0E*`XWxDui)}
zZW0Nz6NJw!;|5=Gck0gSqX636H(En$*#y-V{ldOJxtfQ=xYj)>6Mso3U|Fid-4bB%f>ggs6{Cg0@pcq
zD%Xj59qQ2{TZMRtQGChtQ&D+(PJb57TUf&Ybc6cb@e@`)t2)xJ4`e^3pM{5wvqbTg
zph|R=L}F%q1r+1fB*)>9`=PGE#R!|PSsemDGzPZbBet6@ut%60JSpi1d(h9_jm=E<
zXf=LNU|@O>X}efxlO=g75a_Vrqqqtqy8Pe)3Ve(j10s?2Kb;VvOv7zF|F{>I!+o$e
znj5;%U^B!748}Cy1Asm5a?Uckd-aRGV>O|BOXE?{l%QP18`ibHK?~e}Ws`S)82>f=
zcfUaW-=QVhLuhdgcUWUlnh{(O&+OtNV=7a)$52lr@mZ%n@mEDynDq+PkKp>GLgo~z
zY9EDq4cYiFEWVci@}_CVp_~BFX~=2b#d$C>k#=nZ{^%N!Wz?W3U%I5^6kSWIi^#?6
z6jgirh1TaSUYAw~34cUCis=K8z#vxsRlDJgZOiK|zC
z(`3C4^Hgc%h$Roxq@z>#C0+(M=@wCX`0XX2ZD&olQaMP07mC-L`vROTxV_ctMBQ3C
zEQHmZZ26vJbV0XXC#5q?8WJP-6WzAkZ4g^Mm&DSb_g>Gv^X-@Vay_Nv84DDFw!f`+e6mY1F;KOt6;Y#DFh>8e>~iu$~cGC
zHIS6sm+YWN)b*KX91uf!$-pG&y*2d`-R-q=cC>3oe9oK{;d?}WT_Qe#U?A)n>`l^|
zyo{U0MR6~8A1HFpV1yNrjii?ctlR%P2aCrFR{+*^0o@%EYubv_Gs3Lq>{kRFe2+gq
zo@7{SS^ntvUFioMQ!l&S2qs;+T&z_(&)9R`7@M+u53l`E4Nzw1VTCLU@E3`J7RDB8
z$*z3DB|jr?xmDE$f7mlG-tPHE5eqDL{6TYohuX<22c81q<23M)QG)0DT3pB=8%(b)
zx7q2BATQc?NLPKz)s`{w<+*4j=;(tlJB1_a(jyw}->JDv8FDTyEt?fM(Vgu)SAmN+
zt`bf<{HJSUAfdkPr|xD!-{drmtPBr8=HU@g*bhKkwq0D$%}|Thq2F7A9@=rTO-C$E
zd_r$FM_mmh*J@NPv$Y|eEZ6G!hcL>AdH_d6I_WnWGw+EwR_wb`I
z@O!FTqw}w8o$4#IaDpH-+y~AV
zuW@Jx??22W@aB5kC7?w5TX!m+Vl$%*8{
zav~q8*fbYV40OSap%(486QAwQI?weROSS|7)AK#9Fh0H+UGuDW?VFIn$A4SJ5$mpW
z2`#XZCP^l&F(=XhLo)WF2f6<-n=(CClUtKR>wY;Q(^-U
zQ>;PRO!z&z`EEUUQT$D_+aB9;*j<@j?b^U}s~*NmF4*L6_<2drx7c?(jvJf0-w&FN
zY`3_3x|;NCE>4hWeuX0H|IEStRcwPn+w@6?DB!Xjd!MVh?e-al83@AFAMDQ~VCtOm
z_2XMYF9;C+EO;|sm_6_`+xx{-;NUM1Nme6tQxZ|KPG#h_S0CR>zZ?DT9&fdIbg>~A
z57a-HQieoa7p)=X599pHJD~PFQZG5Rm7D9=%WxraCp#%!8(Z-uy~_#D(`5o@B7QN+
zqz5v$AUS8P>$Z8q5bg9^4z>BLKtOIveJ(Rm_wI!4Hc4{eqZIEUC!xOVtArkK`$^?R
z$8}cFqFd#uPq$!1%DZZw#9%qf`s22sw9lgG`L~z1(i?MW!Bi*c__=d>%|1vYCXfOl
z{jbnPXJJJyX;w4e{vZpFJ9vn*65+|iRgMYdXLk(i+6a7$w=Nh4&wz3~BbPwA9zN~f
zr-bgMES{jr0xu=ZEcW+aT(Up|BYpMDk2P_N7C}-A2(jx3exu4JrgL<)DA@|TaDRjk)Gxfpu+U8Lx^Q4&ea-9X@eE^(f9FEAOj%J!vL9LE{iQOCx(0pp>bmM`D)R?iN3EO{tw&+DuetxKoL0<%Y_!3nE|JHVY`5omZLK2MO
z+`A!C?NaZIg;b`Tk^P`TTk5-UY2;A
zOzm_v1|51oWv3%|A}0x7-bAOk9Fsh33)WQb{*mNz`(vI8>Mcf3Le8V~>SV^+YnIQ~
zrwTfU3GK^x<`|wogkqyR+U=ez!P9A(hfYepxF(kXsFr~gPTerQK#>h9J|oOqR(jC5
zU2ykpmeRseHiLmuDQe-H|Ip69lykS4_uY&S4|_+w9-%9p0+OWcOlY#1j%Y@&W@|7$
z6Fn%30a|v>i(6En6`K?5QouFrjd?OO-EMNdy<<`yQ)Hd_CCknV63M$)S6p77h~;X_
zh}8w4{S-B^v9pfYTG~>%>8;64b=QjJx+&jWbjsg)(GDkeiT$^|7f|!_$~%zTnF4Z_
zUgC-8TkM1?m7e$3cAYgWJI&TOs)L)!);i;yzE3Ae1^aM{#?r0}n)Y_JT_g$et!(0*
z!*f0REQE1$UK%Z+l^wKcINfZJxQ|_vwn+?lA{X&yEA>5zYxr)UU;Hkfc`NN|_!9%f
z1nN($GOft(K2~l8Cfsx#K6o2`dqX6SUNZUewb_OW?&cf~t4Kyug(6(ZwX?vZyL=|@
zSb3uUsL*ZE(FDT%q;zE-lrpWyl5QK9>#;laTqN~o<@olAnl6V>M+JN~+U9l8Yd_Zf^I?fw&19}A_CnssPT{c6z@_#?oboK4M+G~`Z
zoQm1ZTXwagYopWC&-x5$(;`dl&O~+}t
zT+k$G1YO2?@L7Ik(D%{HB{SzOq|2GwT)s{jLcsiB{%&(ou-fr}vvmgV5MN2%>t7Zy
z;XnBU?K{%bpxmR+$kQOJ>=Dfib?w9@RJJ`OyOlL3J*2m&Got+5%u_g)W<$N9No_jU
zBPG^?nA=h?scQgmEG3@(Xo8-QB*x1h=48eINeVT*Xo)wW+zFx)(W(rcJc5Ac!(AIK
z6*S3j_(29NPe`vyX)K~B24!iEh8S5GSAkU-y?e*F;!_ttcndKx#fR6c&ZI=2EnAPC
z%BDx(YE&BU1ydaIpI2^*ELlxvAw1QnH*>MI+dVn?U(Nuje{#3p*0p4pLZNa@YCTY(
zMV-R`9Cc$w#N$v#WA%jJk5eXnR{ZCgK0cgF{RxKTBz6hTwX}d^^0@M&dwB%^pedd{
zw)Lrm=+%MJ`Nx-D(_L~t9~ea>nCDna>oYy^Qv(m_3lGeYU`DD73Qxp(^ksCF#Hq*m
zPgP^obPP~&Ynjq7V*gO#5lz`uRy$JsJ7m1}S<2;@$6{TJj4R5c^3v9UA7NvH4SJ1|
zn^0Htl{YmTTLgyV@lW?e=egDjgf4Mnpk)$Tk4T3Mksbt?g
z*gIZpojyC#(XB)9KQzq|FfA$|b83H`+k|SIFWNby%3g;iL5;>TdppaXc}m5W6?=IK0A3YQFel89IyLKEkFG8
zng?19(X(839l~U<+n9AS`t!D__iyo9{F`0)gj~b9gezMjiYfC{%SWn8EoN$FF?a_O
zePHPii-g{DU39t6p_Z9dW$VX7)^)2J{4?wOtxzHh(chevs_3sjQ7@$Ysb$sbj-0da
zNDMjsv;X?L_%q)iG7DzQ<#mtOv4=yjhZntUp2K(aH5K-$e3^IW-|Y!MQxSnmb_mfrs4jNOBI&$e
z<{v;j$hu^N8`L~_QS}p@qtesBmRIxu8HnA_e!hridN@#JnG@!UrcL|ShQ(occI8<|
zp_^79gMO26qCZcF)qUm@6C6Ks&mq7wjU{Mjw&;egu4Qm7jPpv{njK=P?6y?23jF!G
zW#dl=n$8|8!_ZND5#bX3@vbxZ8kuAHqGCb%y?k$z5zStVktjsRoxU3hE0GgMwk6ux
zyJJDNU$x#hQvd_yUggFTT%Q_zBq>FBwbaMoLj;1sJFFsyE+{Yk^J1^~aw{F5mMhGL
z>3qecx(3b@C{F9M)0D`!ojL?kO4xAc}OX
zs!GG0=XvKbn5!(+W07t_GLl-83-GlnedCJeF
zfeca4UNr{Q;Ky9WXDE5WNcYiBshCT16L8^
zh$`W!8KMkE=vI5c(WfQU7^*`>6g1%d;gj1|y1jT$a5JviqTLSoA3#yF6+-~F3b!hO
zD;iOh8`{7-euTTFt(=vq*ru=DsAAUlUj(G?krvVX72;#~BW|Wh9jryZ-da~%gyQXR
zG#fOs!KJJhn53Q
zdaizkzD?@|E*u}*n`~Hi{zhF7IEp!_Sp_mtp0M|Oyc5_MeZd~n~!=<4JL5*vjL+KnUp6ZPax@2wuaQbVDooP3xpjvmnMbj
z7stu-jb;E2ue=^}{m=7smS>yDmN_}xZ3S1?CPHIfYKfF<`W;pq)S;N%RDBpr!0tNl
zi3!{22#crofk&ki>J;^-xUK&CKVF_jkWK1e5zOSsc|$a@O;yR@64cy*a2gEwi)L-K
z_ZI$r@ZuSk`kHtv={kKpD06sth!y9c%)5Z@=~LiGKHCE~ueK8A=|k~B+PlT|IQ`s^
z7ULQ%xcQhl=@SndHlW@%Y-#~JucEGl3-e4*5Jj2co3h_f5_Vd8F=~j$h|?i(v&q-3
zj3Y#U*ji0?@*H+huGqG)sB1R}q|Q?Ogd0}jo;*o_
z#R|X@v%5)OX^G6qo(%E+v@VSJ3y14KPN!`N%dNzr_unp~i<4^S;t%Mc`LJ}gIht>8
z%$v~-0zYpKF;ITQ>i@lXe}p*rFzP`CP4{WKW1DRMIYXWvni{9Uu|aYosWUfrQG<^8
zI3lJO*(d9@I(%|`?_pQaIaHG8cO0ToeFGGdA8SSLtd(rmV(Bxiemmh9F{LY~^1YtR
zS`Vx(D4N@sQq77;_hpqo*q^TYi2oPj$;`8g^pe7B2cCI2ct%QO?A}Dnzur4q3T+r5&zco7*+qxxxKxlYqILewIuL`pa1zx|8r%Q?C
zHd|i-&{EbcJ|8K9rkM=?>${4A
zy7`>K!RR#bZDX1_`xW;J@bWRiob{n_OJW>)VsdF4zYv$2THy-uzq1+a6>$a-T^`X6
zDVCnRHjATfM9PqI76u;|R=mGw9230xnk_Lt-3$FH`~8jX*qYjj{S?beJOOjA`&zV4
z^+~jb+kw#&4YS2LGMy4!Jx*yrC=NBf~LtD{?U{Ir2
zTDsv|k&NU}V!bru;EE2_3T!h()A5m{Jow7NV8Srzw
z?_o(BKH1vA)ibf~i<|c_Q_Pp3?FNGP_WQx5F(^eY*Z#tgO_jebxC>6I~Q_x+Ko2
zVf<^!XC3&x+8^~w9W-n$={mm4dI6uY@R`aNXxzJt8^(`82qN9qug$JV$i6a&m{rEH
zCA0d^B1vLVj$dU1$O5I+CR4lYxQfBQ7tq+Zu|4~XWb2Z}*BUASlNKM+N~!D*4?F!v
zd6}cMOj4K!9d>*2jXPVLT`uXj@|Z-&4XB+Zju*Ac#ZG&C-eY%O`w!v4qsQQDcvT%P
zJU(xu;rgFt=|dkl;ZhcCE_JrBbN2-t<`Kx*OGt-rlP>I!Y%BkCjsDxE3se=_RiUB5
z6f_tYFEdK{V}uX`u-^VnxvOzc?D8g{(CjVPce;~4cuyXx*e`646SLL7Q#NG@ugZke
zhTnwZuDa6N(68ku*6>{W=L9h`jm(qBt>WF%+|ls2J5>6H9U?l-N2_97kkj&}(M4FVZ2i0hgd)2dhm7mUuyfc39RKZh>UrHVT=WskS6
zbB*|Spe+A(n^AZ(5Pefa_$Nx!i=KbcIer}~i!5#gXzp~lM#66CekSATFsl;ORQ9V=
zv;{=?)%Kf|v|!Xy)kc2=zXSw91*XFH+T3*d
z6fNWZO)8=*z*>xyp^csY85_BJx{RM0O{gCdQFn_s=5aNLIF%J&WXmo&?ytdr=2gKv
zYfPkf3+zi>3i~ShC8jz&8KDOGs_JT`TnT*2bR08^hI$DD=&ZF*4#Tz`_7yYy`B_X)~un1Cj`bc
zF|KSvyL3*L@S))S|Eya%y5>NI=V(=K9-Ezv^!B@`0R
zkd1tn!!0PmiH_7wHqI0X>1%)J^rB!;o?av?0topizki?2YY*J*jAwz`>AHp09nNxT
zCC}?PTVkGcg?kUz7(;tdxXMisf+ljfondZgf~~@bh`^>61CHL<_en!V
zfaXgJ@CDs7Jpe07rjE~OU(~JvCls++Y2YblFE3|Tn5#Khs(5X0K`Lw$_ie(q-7|3>
z>LOqH0lU`LDLYd5u|_WeR4wO7AIT!DOW5omB5AT&!7GN&WyY+#9Y@>i3NuvHaML~w
zPV0>&AY~@cezkDV3LPyTLTQYKuQt2`+C+2a6~=gl+WEfq$l4W=RR9+wI>JXu+^wQC
z{A4QIM1B3tD{Oy!xaPA@(kF$%JusiO>G@&g6%p5-DN-cO9jmE=cZ~HK+U)-|%l_f)Kr<^8r(srP;
z|1jRTsZwU@e5o~rTj4)Q3^#=>S)PO=?yuda
zmFQD1MP2H$ZkgykO~4F_(Q#;^x1
zJlcluasaEzg;TlQx-H8mvdl*htcd=>6(gC%Y`Mp2N{
z3FC=iB58Dw)bgy*X;JL~j1I=5PcTv48DZ;cpd2Zv(I#(U%3;ORbHD{FQZP}nm0Rlc
z&s#0oFTZkPHvU|#$sfNZtt;!p;`^qZ%k%;f(9g-PMqC9Zjpuj98ShktstZZINSi`0
zV47nK<@oSzoz8t^yBI_bI$XE(@8U
zLOHV1+r_>6dvo@ikiwqO%p@y-2!C3QN1!+_?s78_@=!E}Rp&sJS
zE{|aTJ2?(`F6LgGxXq^sF3)L9zs3kR3|d$SpB(j>k9pf7rxBBnUB_IK{Um1SWHK+E
zqf&gIoSxv*KmxgrpdB#tDs
zT4!zsQyEmiG8
zn6b!7Eji?`{Oe_!CVJ53Ja)R)mqR-g!I#8`!0EOt3%pV6wd_@qdd6K9-zGsYK(?Z$
zgb)qTdd%2MieEF6^c~%v{LS)J8IiWmqjF(rsn86rr0&Ao?=)7VqMzt)J{^!kAnuAx
zIUVgpe{Jz>iJ}tu{x~-3j|}~hGIq1`5XgSBW>>3z*RH*@-s_7NZ0c}%3)D-=e`CL
z>aeb6FZYwoOwiAP7BH6p~EiO*8T`!i=|6nuo3uBiY|
zoF2aQrrPn>me-n&Qvdi*${?BYnXVC|=f`=_TE-V`PA0Lm1fvVp?WgaL4R|;RXt>or
zxYYNahYS=Ns1z(iYjjLAEp#N6Yy~MpvBOs4zgDHYS`GiXKQe^&B3cd+pwRu%l$sKj
zc;X|{G~f7q@47thbRZ^*`Td{lG4Frw?_ckuW-G7xxn{Y!I7!y&ThOa^HRE2=u6nHF
zT|R94J6P5DvLRZEgWXAX4%z`L@H0#WUeRo?nA3)|_}-_K2pKMg*oH1sfwBu3Omu}P
zoa-V~
zyet2e|C>q4q)g$GE5U);*gFI&BIQ`{gU)74-1#S4-<=uHc1*V6L3vcN(6M}kRN;@h
zR04xfo;kbRGfS{-nV<~TJG9lPk@YvMFa{_NDkY^5NX*{e)b^L@F9{=OBN0>~kfyra
KcS~zik^cb>d#X?X

diff --git a/previews/PR1624/tutorials/out/solution_000160.h5 b/previews/PR1624/tutorials/out/solution_000160.h5
index 999089de3621c868579f5458ad21e4f3a78f1b69..39bdb861d50f9538072832d5d914d23f14ba7b02 100644
GIT binary patch
literal 11976
zcmeHrX&{yB*Y|CfWT=!O85)$Kfd(P#MkLZmI#H&`oH0rvQivoVhSV((|¯-}64thvz*15AXZmANFsrb*;6ob*<^T2-Dx7H-k@{kDxxh
zyaXp9^z-M`7i;oUzU&ugik@1xoczHyiJ_C2&i?BM8zD+iYoe2Up-KDF9KRU$lXAv8
zbag4tl>I53RQosle;xsSU9D|Y2zF1#fp-$iC+laD{8?+%@1Md`%>M>X)$2v>Uq+_v
z@)Ik5GB^lsf}h|e*j)EJxOfsm=+`gT$Uj{9Up(X|$Voju6hu8+cW)~%hl8%3#4I6#
zn1_o~t)z3%#m&Rk)6>D)*;d!v+15jMN>q6R?m4wT@2~yc4muy+Z+qZxJ(vm>8vj#|
zXVPzrLslMt?>OgAPk(*=sW*K}&(_)2#n#p9?{z2J8)a308sH!%&+B6AX>Z|m*vJb^ld-rGGm2f5w*@7gHxY
zwDO;uWPiruufKlGd+fhBjR_w5?}P7WGU2ns)~fK^OnBm+@<20-308#zM07e6T(ZS_
zX{k)8Hp|sFN@9Xn@|7}|L?-RGUAH_4l?2VtM88%dokhuxa~g2Lrk!4j!!$cj0Mq7
zHVG55EMOB*5?AG8L9CDPyultO(9BOpUVX)cSsrt8WQ)<>AnBlRlL_XdWBSuBFyXj_
zJR?7x2~6>(O_z2MNORCq>^yYkgP
z7C$G*A_qZ^QlyuBS9@hpPLRv}I_9Vs5@ZI)?)v-$f}AJ0M0Z0hLAH4uUa6N%kOn?y
zpFg=pkV(50CO(@GWY@zbrA7w`@e1%!48f}sXiALw0#gM*UDsp@(DYpaTp6C
zq6VCQ%w&O*o$QL;`AjGkEq7JB&V($_`jr{=Oh_6m=v~v#gh(!xvFRyH@Gfw9Kr3Ma
z=Ng@+Y2!?oXis|Bi0iAvq_5^)!#YxK6)kXMLU8VEl5k|gMMiIP^GPOnF7nqjj%C72
zmuHSzL8x#3&A=m?3GMr
zLp>hz(DS-)Z#F?%2@+dMmMCB=K~Qm=R=Ya0rtf_0LO$h~2_``k%<*ajYiDh~MIGHsHDYyhpRo
zV1Mu(&dIbt#snY_ybTg&!J}yfEdjy=dAa2gcgk7JTlHb*(-j0M$#G@Xcqu`$Oo%rk
zSbsCf$ci5r$FlG0k*Cm);^FUgaV|{Q)-iZB9PNazVIQzuMv!;rbTHWO66B}23Bnom
z)2}280@@WiKlfd#BSG$~Fp&AHR=n**630V7paxm=V^O
zQG(wTrGtpLr;3VvGH`
z$E!zgH48>lh&momg4BF^+}gJa>s+*dtP=BeY^ecnSQZNe;%>cuvX%vlTFuYh#kwgl
z^(sj|&V&r{J=~F)H*T-=W4)0WAEUgI^_bs1+oO9#F%C}@Yx|s}2{J~ub?>M=3%;#>
zOt&a!g5SDEK~b#3qbd_G@-f~UFHY!spTRj+cZtokh6Nw@Y^qSk{{42tNyR*aAXlt4
za@J)*yndv>#qTQ<_B?3|g9@zcRr+BM
z{g^O4aBW!(*5m$Z#VXrAqJLgK!yJ8tcUH=ZmQ=c8wdU>{ZwbDVcO
z!i4%qJJe)w?zX;T1)AMvLEv2355AlPxnM*$MIQT1e2-`uUok;?o$uyPC?iO@FZ+z!
zaSp}A^ywv~vS7||+3MIByjQMXpS>INX`sH)?NKumQk$5N`sUT9eYqGfOX){%&|YS2*seQhZ^+xt$zGcvH5%JeX%Z}u
zYBk(ijeV53bbRA!yr1gd7aj}4`zXvuUA@wk35@NPI@24Oz|-$(v{!;4mqp34oqmFS
z9``aq`wT&j$<&#gDPVzpe1VSsOD2Q_x{IaZd7Ho0wCCY{IK1Sh{9IoqXh`Sfj$$3H
zDNp`%5APF37`yf@ybqFM?c>{Uo-e7k&=1CWEaN+``5exbl=V`Q66cuk*E{-u^7quc
zDUo`KlBTU#5xCS76(_w9QjsXx|
zGR;j4`NMhPaFG}Q-Z#apG-f|MX);VVo9Kt<&HcXxqka;fwLLf605G3R7HwnoL$>NZ
zr$;EqvTwbwhI)a=V&!flU%7nui_?gS3&dsw;rgByYmpqpVU@I@ohT=jVIfoXk?}s8ApNi9nCk)aCk5l^}bsVlobbQ-n<%oJR=9k$;5iJ&+
zT{MK}X_n-UC_;Lx0sW*pVghs7{Ie+UZBt$|18F{Wk=Q`Q!5W#iVBCM?$#9GY(%j$K
zj$B1KJMNBcPccrJ2^oWRxE^zA9RIgA01vOr9uz?Qo_X8qF5<0I+=?8ypZ?hPY(u1p
zS0d*s5Px%5E)hY@RbPaEw&6MCIc^Lh7RcP%KNIC!4*NuK;JUo^x2-{lv7eF~RFPj7
zre|V`wBxs)Y9GXKT^_qc#IUu8$t=_#J}|3|%G1{Z;mm&IJD;}VwM4q#_(HQbqC}aB
zodNO#b=K+5Lwhophwg4g`pDPDuTK$aYZ%i$qn(h52j)UZ@1E&75#K$t9XM|*k8^t~41dQy1V79*68-?bv0iUacqy?HIl`6u4p8IF0r$Zk@mhx~(M
z4Q@k73k`R@`iS-ST$R0gKFTfX*N?c2_^C3IwiNk7gQck#Fusf@9liUIzuiu#{v={W
zy;_wN?k88mBb|x+7jIv-D+=w-af+9okNn1U1sXh5UJ|Vvn~>kkX1(VN%AE{uIC=mv
z=K_~Z81jn*Sof&7aI@*MZASWz5*aRqSQU`sOyynPuQ2N%V#;yF&OqE3x?kT{MET&_
z+H_IGN|i@me7N6(j1Q+aAm-d&n7Re!EesncKH$3V!ErGzM6q4|;gtP}S4vN~kv0>~
znQMj^uqWjtmH*dV(o3w6e%l(5xB}_M5*zM1#JMv*@&zMSj7ix$;JS#g;%yJa(F;6x
zPT=`W7p2W^LBG@v_@q5We$L(Adj^PCZkr}GAisOp=Yw;Re(h3VB#(B!zhdMsK8s0rh;|32UO4VA=ZD9{*o{RD!F)=MVh-x=~BvLfDal(!Rh^gv=
z*C{#98n2IyenUO47E|ZG+B;swj<~!!WI2`hJ?n$kJwQ3$P@%X1lsn44YLz_hdqv_=
z>k*X8YHncFqnygd)AHhoj~_^B*dU5XL<_9P^{pA9Y}EOj4F}%TAZ};-U6Gns;r=Sg
zrfAQ{{p5Qkqyu)!)gME9=+
zT&+TV+CYQhcI3ONHYf@rmJ$6@`N$8>V{zfHYMhLn+S@gzq>4)u;H2fUU*S~iQG
zw-m8O&h1qMqL-B$$9KeI-8n%_l+!J76r|#C!L(x2VMOUqP66A{Zr*^9X8_Wfi^aDs
zL)2aQO?NK(t!DfF;tbr+&fPX|6y;hwN4Thc*M3~zy%d?r9&=;vo|m
z0-QLXnumQ^NOQ5-UrfjK3lBKwiy_WR;5LpyJ1o5pwH%~%mbOWhpxheOZ;~@GKJE(&
z$8Y0)E7W*C@S&W)sg79)@>g(`Kg~i+czW<|75eq&Rpp+SsHZvHU`EYPy;H&Vd3bMb
z2uWnSg!g5HwFCckly8r8cA)Bj|Fx~v9bE7F@XUMw?f0o{d!di|9!^>#)Le>|&^brl
z8=ESuj+i5z+^^921hGYYIZ=!F-ersYBShgAbze>Nn@2>+{K05H6tdO4=|H=)V82p5
z?qku@RZ6`Z$_;KGr{d@&aeVGQL^sjHhT9Ow7K7C9OB+q@w=b6|1j?
zpx?DMmb|*?AIonzm+Gg|-9Hp>ARQF)J(c3`^X_~~&9D7GR&rAQ_|zEH*|nveYZcbTo!eBtLy7a2u7q;Qcm
zM?BWSZLf^>d8ZL~#n4{54QEy+;=973Ic=z)>1RA}3hAUK^Tkvj>ymqGC6Sg*SU&b1
z{qqf#RdqrBu6?r)nW7zOYsYmYVy%fS`yJ$$-ELh;z1xVl*$+M;^4@$96N>yL1bZIU
zuN7RkPLAPuI_QyMskkm@=Ixb@`t7avYuWK!8b;bF!Wi#o*V{*_cZ=Ubh8=a!Wws7H
zqVm!^*53XOG2JBDX#@JR@XK6(Y98(7$c&=q+WS;S#P3Mcw=GPe?&Z1Zi&|Qczr)m*
z_b#GUcSF!|#MraTiSvl-wnk46M>!KuuP7PZf5>xYODWPVDMCYEa6M<|dNJz09KE(_
z9i=B{5p?T!M8jm=^l6BXleA`h$8)Kt@$|`{-VnL8lJa+5Wq(*Z;@qcJjV5T9VJKmk
zhjfmVZv|y%W{Q{880v{V%@7PhI&xUdlDc1-jvMn0jA7{1kncnzy{G
z6lPO)eB21%7?fL;6e~@gFS`FVqX=oKu-g9)V&xh__B+tLu;judIiL1rmsiXQEq9~fF~6XWBU_Z1(0u)CA?V~X~KEF
zlQ7a*b=DTWh{lzAoxB(i!_<#DR9x<_C8MZ33p{HbqUICl0i7UjlpCWl^rH}$3r^pp
zh}c|E_u?)3CB#YWqUP6p=Eqee>W2+C4UHjPM{E$H?$fGM4I5R_j;O_r(S3-y$8-8D
zk)I=KR!+?=^W{Tx;xMo7t&&Pxkng7EGdhlTdiT~Zq~4*L{V&oeJKL{B&7|HJ8Ok@E
z7>Led2h*uI+;bI8c1Mg80b6Pwh@IN5Rf=|JZERh?3HMuTw4mfL)?Y|pbH#C_qm|Xm
zs5zBe|CXD|%XdNL!6V35EBfsN^)4+swdeFc^e27kff9bC8GJX7+(Un&>*dy%WB*hK
z+FV$VeEA#dw)`k}@!_VcRDYkVEc}hCcX;~gxeKmmMs_{(LO*s~^S;1=$S=3U@dfhv
z?f4S?kd9`mE^|Qnxq{N2+uq)&IZ0)vf^P=UU?lFX
zT(LI|+OtOQJD_036^_1_9U$Pn
z^ZWIN4meioJZSQ!0~8iWTHnC+wIRHV!+Seme0594-q8-&Wyy6skBtV+@(YL6IA~xY
z^00exRwtyaC2HitJE7{tN6p?1oxr*9*cE%uPLRv4_bzx#1HWt<>sc-hMw{wH27+lo
zw-v7n-9>{owze($+B8V^3t#Rp*9$L3(hi>b&;!@qipbB-JuslJd%QxW2M*;76{IQl
zzygiu)fsXtVoe{74rW3Lo8;7c!Rl=<+lG{0NN@kF+We&#G;NCd=EU^^pG>;qkZTW^
zEc0p$k?)2`c~{<%=vjZN
zF}oX@WCye#9_a-w=ZmKNF?0wlz8^3&LtaX`4P-)8U<#
zU;+d)Kw_bztg|}~UZ`&!64wI-Y3-b?A>Gg_
zDmY`BKK5s|a(0eC4Y=;S^z_6yi#=Gn+3P_k*dBVvkyF_Vg=$h?6+;=|n{6RD$FUdg
zniQYzZ|s5tN~4yQMV(+gM}ccW2@T%c6&5F8-o&&I=@m_QHpFPOJIt&gur7C^18$#3syek5Gpn(mltHb)jlw$=uorvSBZGSLt>`=#K2Tgt{I$o1eSB6Ecr2O#H16xxx~7h`qYJW-gf3k
zRyXLxIxaYd^Gy9kP>Gir1EjCX(ln1Spu}%%R@UcU5D2M#xT>%V(rtU9d^Xb{E?LCm
zC_#e(5`0dnV)4q^!pEA
zrQ6V+T%kA3tP2FEg(RfmdE?)g=_~ekKov(_YS=oQ^Jc##4g(EV5YBz(ZFmpedS%hj
z-35;?tcj3YM~7*RF4_mtzirE%9RuropmsC2P)HEgt%k3z-5eUsUSg)T2=mJ$6SL3a
zJNA>e_e~u}2Yk6Mw$K{ynaH(2URt4l7fkA+3@p%&*`bfod+DHkyGeR4=3~W6jXH(L
zG?+FhJUUlJsq|gl*JOhRXdYh;e{5FvXz6pFj(LWR9&t9Xx|5To2{ugkvUoZ0fax&>x
zkXZDS!AbBCf&?EyJ7DYJ;z@{LT)#pi{|M#3i%^gtC+zr95$){Uy{)|*Tn~5>(?tkk
zHVP+?lD?~pn}?mJr-O~Noq@Nrorl4ss>T{zb8>yYU+cTMI{Vq$?f=^dCXtg3=Z{_9ZX7_Wy
z(@;3MU$Xv9t_s62Z)
zxY!Y*a_~oav;+t(HikL5`~b^kO`bhBIcYzcU(R3lwqE~R>oVc&s$%y@_%lyae&vZ;
z*VD_+?QaW{_-irzOCxxHMKTc%H8-gbjrNN%{F5SxN#vZEtADp6N|l@1Ga?5E$4>`*
z6aF}96A2(LQn~x%Gi`!RDYs2av=b!e|CDk5^hEWyHRg|MQL3`PVj>6+d)Gfa82<6?
zpZGF)?ly`3@<#p-PN{B~jDiwP>Q7uR6Q}G?qyJ!jC$75xQ-0-s2KPVbml_w7J3F}e
zpPOWV=Hjp4zRZ4fbY3b8{0!fnynmYoA0139&t$Qn$UXVNs&p1O6$}uOH(9VhL!yVC
z!h&jx97B_9EO1P^Qs$D#0?}{2S7Z`cz*{Q2`NUopsJM-Ey4tWn;$a`p3@a7@D{t4j
zJuKLlS0ou|i|cs5lxf?~f^}(eMN?f_kns9@;XE%EB#qhac07pdHOHk!Enq{0)82$}
zc{T`5RhQJ}XG64)=sMFbfTn!4gsEo?W$PiDB@!}+Bq`_X^>K+-B-f_(onmlb78kjFZvYg(!j}!C%1SX+JvP{KR>=0bu%YR_*mJ$xY%mDi$J!Rc2EXtD=O5G9
zK-(w3=(ju;G>bn!u;M!Qg=g*Jv|1M29?b7q+Rp-SZp~4FWENb_cX>cBVZp!>{Wnv_
zSWw@7?O{F2O+)13bFQ*rheoS-z8eb?b7qi)BMSS2P+be%JhdK>%Xby&OMLxQxGD*o(%aW?b2WM^Ar|C+xU?Zy6FQ6O~mV=oIj
zczU^Dj9
zAE?0oe8FjRFOeX}e02A`(IUt;hq~!vU)kU}>zXBrbq>wHu33ZggMRj*sF@uLa&~%k
zuU*0hjufJX*OMUO$#EOsN`ll9?;owedPOT6@r9(LJoZk*<7I5v)oK}a7yIU+xmQWj
zaTWwh?&LX(b$jo5^GMHGHas!OEm?{6-M%@pTO9LHs8-$QB!~5qZ{77x1?R)kN4-|h
zS#WT9y|6g;VW8&t%RJ1t&c$&9?{h48Tyu$LUd4u%o$Jdr@cj1NaMH9)BS^JnCeEwb
zY={z^ZFJ@<&a-isZT?9t5O|udwha5cGORdNE|>+c-nq%m#(C9t`q`6p=(h=K9=g?R2nUyGhxk8xI=e|f($ogjH-!Nvaz3k)8&g+MtArfC_5JoIM)
zf55V`Xza(WQ;Ib=ePBVqm(Q@+T{fILsrq8NKK6^GfB&MB1nKwAtGDDb)=k!4H;lyw
zd3|l4Ak=%Am9SKHHXCfR&szJ2;Cb`nKQ85oeM;ING3H^z(;NJzYn$0{i6fwTRF@#_
zt;>90Tj9J>(2_rm`+ht)Rq#zXLH1QCr+5QF>cuWw)9A^DFZ{jfVt5W;40E1$JH&#z
z!YwQ2;M{%jmK|V`#fH#X^6!1Q2y*U-L9z;-vpGA(%lL~4()WCqU_u%8=jYwq+HnrW
zM)$3~mcoWv!(~fiqFE3Ve|^SpSWkByMYqCctc!QF)&^BJkV)e^g0L>Z&u48qJ)0mK
zYxK5BpC-s1Gu@sIW4{QbH=1AeCdiu=X_Z#~Y!E54p!4BeUmP69dlt{->aGN4-)Ry|1V!@pNcZpQo@29U-?YVd#cFn(~GRqg^l*`TehJB><
zJn7>-yieYSaIC(A_d#lmeOw#P^Gz?T3{T=b-rzgCISS`W{7PAAsVJNm|Dn(1yD5=+
ziQ|X-`S;I1Jd^KG5t;!0P;juuA0wVt8l4kAhw!EHpL(}ml>HUeKk}~9_`mh8;+V+S
z|LQ*(f1giH5+Bw-KmWOWE>ilh<&%1&PQFv!@cX7;?*8Gwf1j_#r@1HA`Fpum#Kg~r
ze>j?d-Tj+^e>3oJ2L9j4z@reZ{_wGW2pxfpdX51&OqkIUxCbC)z29ng&H<2|KgCS~
z`J=gK;36>q9F4{7R8~Jcdt-dlV!R)!n)`R0MEgX38+#tw0E|DAF51NIha1|voeEKp
z=Fob71=<}x5~GxbeEH|Uy$nT6Rge%mfpW7J8?j8pFPf=CTTxFT%}TxuX*EZ^9}Lu6
zZq!icjsETW_6c7@KEJ8b$t|e&RNd-M5aJusymKbvay5_XlwY#@jyVs~f{$)rsrrHI
zA5shRM-($YP$BreA9DBll!qc7G)f&jj{bHYb{wunbZmNK?TB`YmX|ri5qBv>%pJo0
zv`h1Z6;ZTN?RPA52`AhHwF>Y=G?KJhWfAke8MogLoY>?C!8tUXoC7t+ZWxW@-WrEw|N=rold;F^$gZIg2SwAE%Jj#>)eKr
z78ve){Q>*0P@Ch04C+nqHw?Rs*jRCvu8e&1!P1lqnBQ-YJ9>5_Uw@xS?J2|;wJR!R
zsklpcf{+&#AnBnovFGj`xm6UBHlc%#t6W5UAr3YtD$~ab@feg
z#3Ic?FMeDvJMDe&8pQN0#gvVx?`&K@{vPEXu45A1h;z0dJwy5Dd!zoC2kEtk(262|q2mX_YRim=@4k%X;Rm<`~9J|1q9f;(^-zW76@*m&rxo3nJm1Tad4*3n+Ke^6Ay4fY)Lx
zzLq3RQI8#WZ)ZR1x#fS4>_ob7{PM8^r2QHGbyOeSetBrf7*WMq`(Oyltxp-edy4ec
zb8<7N{8$$i2T^q!*1q=cE7CeNnb=swAA1)Uy}`WfyKz%$Int7qTQ9Fd6kmS+)P2;K
z_OCcpjeZh8yLYTcetl}p6g}jJH!p7DMcQc39$iyJ;pY12M8UCZSGo~7hsh#JF1ug*JU?RfXSuHm$mcm0XiN1;w?lLem7n{arwcSuKID8y
zF&_D;vNculNY4#CD;i1J8%0#n5TDp|=;WZjUvzYfKBB>`ABhybbX&rf{fH?#!q+K1
zt{ZzIXHfRm-<2<*zrytPV{Rx9|8zEq8~H71;*t@Fx8qK4tHL~7)w8+Win#KiQc(q>
z-P*#C9<&P#UETW@aeR245rO(Sin8kYh`jHO`yL_x$s73w0{bR@!`k8lh|&E?)9)dw
zUpKfWh3gG=zn(|!JHBVEk^@ocMbJX3?(0{cSpER@rkxgv9Y8%l4lOMeTsKClu=Nn?
zU2Cpm*P@=s}Q%-epjRBRoGF@By;rV
z>3-^+I?^Y$D%BoAzq~!4ZfWBAKXfimI0yZxi_vVU^E#(8NvaWXZt=YeFXYEY1;tmQ
z{fB`%IR(8!plnM?1u7`f>@}i#fAuSy=o;HXi;nLqj9WlV-SWY=e#oR%HFlt1Ua(cK#dY>~
zca~D`hDxKX<5V6Uq>j(Jhv+2kXS@mV%RGq$I-=oI
z?cp=P_}Ko#In;To`R#|=4WthTeNUnIw%&{<)ckV!v6zdB=U|mdjUDFu`a?4lKFY6B
z+^0E8&=lgX1p2$amn;1?VqL+|%r>t
z&Z$w{Pkry%kQ9_FSa^G7pnZGm{b~-}my*fqWKqm_>h<<-)Vsx6k-3k$=L%W}3aPrV
zN88)qB8HnKIjzBXR6fr-O3fov&fDSCT>F&54Er7Fj!lZm)V<8QF}I}!`J2pr`R*b*
zch#LZj(9O*A#om2(=<}x4C-z2^a`JY>koTQYbiy#E?H#gGs?5Ju9Tqe%h9Xrms57i
zRwwTKj%b@?aB~Xcr)zpb-*I26sl0u2(2hYWS5WaTujmhHN0fPDU2led8OBn^xk%^8
z`j%6EB$B;kN6}97Nt$pF(m}%#d#L-R`f&Rp%D+oc(U&pE#k)y}_
zt}F80=cv!$gZ>&Wy%b;~UukWCqZ9JmH-Aj-V`#S$)
z*QloM#p}YaI;lRiQtzl$MXWV%7~w!Yg~|a>Dj!?z15KwQ-!xrxg9g(5PBKoSNZ+op
zvFf4hD%LXiFdr5vAM~lb++Rk9Q+4Kl+B!tdC+_|FCwNfrE1hW=j;J9luwD(ZCBNon
z6UH@_i`Y)huUV`QS|r+g4!;>1MY^6?BSPJ$kAv&hX`>%Ss~g{TBW54Z?AwF*~}ht-cZYwkv$TjiH~0UA2nTJ5;y-Wh&)o!(OJUv
zCY6Wm1L8^Uh!@4cj+zG|!JG9;(eL=$)|KmVJp~hm5dn@cu&1cYlC-&?#6i1l=qhiB0a)?>(D)nC!$tqsU@DD=K*^!tVF)b
z4IMi{)I0xheLQu(BPt4ZQ2h>1K0b3ndE(j5r(PI`{#EY_oQQ+-wm7~-K5ZX=qCe6R
zEbRpjs4pE5#}S3|oL4-0$w8FQ?oyKfnKzS_?R!x^Q%JCM9pc+yqq60Qu5Z`X+u(Xl
zExEU-dDthoDvhdN&EBrt*U`@+sfJ!dq|;us>umiY?#qN1ONyAIHyP{l=|FOUwEap&VMouPv*U)#xB$LK)2*rZhKO^5o7
zkw?-lbP(?U9_4FChw~o_2!C7TgKGP3dpay!zvR)>)(*%Py29D_ssm)aw|>7~*8zu0
zod?YtJ3v7}+U5qzjf42+o$2X-A4^)&c75xBt$VnS=hEo#R7G)k1t%T$i#_a`m(GBk
zWki+I83r^3epuDBh5`JFN3Pg&F+d}u);qt64qh2__R}0Xa5vP54VFsp*
zN^9F_xS9^v{m(2ss?-BdM^asb-*-d4TM_xmxf?jv8yqj!?1nR$L;0y{-LO>m*^4x#
zZb({u%>U)5E*NOrQrzIu1yheQ%G{;9;J%0dg<=gR1Uxr=eU6(6o4swE50~|VwVLiB
zr&qnOHLrfFzBChN9+rERV#p+sK6V%ew;B$umJ(b*RDmOD}N8$NT(nV&XYH
zRit;A39N5{Q&;;j!6b40+wB2NP&U0cqh|>d=1V1KHK+E1jIr}E6W3n2cX(c#UV1Ow
z@=dJjqxZs5k%aY1Iz8al)I(d7+yk}kpR}7l_u%jMqQ052J@8%brrOYfZrHfMt1U>S
z3$j!W@J)Nt2~GAg+HUi@V4P}R15*k8wh1$y0#1sgiSSL487MoT9wop#`i
zZ)hi6D$Ntw?#lo@5tWpq33Rw1b}P=-iw-xA7gew;vF^#24=XHWfcRsx*=g7p5#xsA
zbBpOPn74hq2-eFpjc~7#V}N90lW@Kh_E$=ldJL-rLVE`4zYTRjK{j`{kun|oShK&?
zjShEg9vziLywBJ>?ypXVE4r8S6Fzmo(}A~km$d0{HdfAi$5jS=?_L>PpV0-J3InSj
z9_oQlP8ZDuqcJbV_m2$?^}<5&U7JRf=wLZ#!O+1U9S~u>M*n_i2h0-_aN{~f2gkNF
zt`D-Ez%kvT`kg^H5Fbqv%QCD<8=N!)+>d5+hfz>
zSS?J4)DoSwm&6&M0FCZSv%4Um_|vrET|Mw^>GW|4MlZ~)+_jqYWkN?*?AVcRD7xAi
zD?O75btz-^wJuCpoG~My68%N5msjWdoNrqFtiDZ?S^})?Od%vUGP?1SZInN1Ma`j
z$jCg3=Pv7&rzht5N3QY)uLlh9JNT9}v!VwoSIB-*JI#cH8CJqG9ed!OS#fB8eJA{;
z{%uc15d)lNs&Xro(BbXAg5m_M8|UhSYl{Tv;1HB|vHx@@oDx&qUmDd5w1#3a=S56-
zXAyI+PNN$vv(nS%Co>>oWx?G+?7KBI@3;Lp$9Uytr|V+f14DRfHtN&C-SxLnhFlk9
zgtQS~u)d30XIH;z>xBmb>AsgPbb)$hUNNc00IOt2|9P0-{hiDfBkX6PRp--osM2AY
z>_~WMdk5?nxt1Bb68nGUj6pW`VO&tKw%jeO|B9*J@-LRzwGrVc7E!Fd7@{wy>IG<
z9l9}`1Qb@)d70jaNk?mt%5~6dm;6NRwD0<
zZkT%6M#s>s6NJr+ZiE}tA+u%n4P8Du?2!JDK4Qjz5tkU3;E$b9Ror&&M|v0N#W*S)
z!Fgu<@lc7wOOfm?u9FEHaX)>=j|;p)5hU*y_)Kv1c`
zn{LqwbEgC)q~d;a-jx}u^>+Y`t0pC6Ii3TH9f`w02Tj7c&$5jUOYgk4s_W{6rx%un
zDJ|~>p?a6qt{C5{h0cxvwcYrR;}Hou!2pJ?ufe{VbeJ>WLT@hCS9nhJZmaJd!0`6I
zrO)gD&SVKi8@y+ZF8lGy8sj@@RugVyg>&8F;D^Xvy`YlyMs63@L;00%jp`#h%pVk$
zIgR}iZT8@@tpgpL0?Oa|aARLbJ}-Vj;yv}D#frL5;;XG>&to658m_O5#(AElyq#8t
V^O!D>EZnWmfQ46gHZpMz{x^;FG4lWb

diff --git a/previews/PR1624/tutorials/out/solution_000160.vtu b/previews/PR1624/tutorials/out/solution_000160.vtu
index 69c8d73079c61c5b89b2ee6f4288eb88b59bbab4..d8601e4cb9ede5c590dbabe73915d6030f1446bc 100644
GIT binary patch
delta 12314
zcmZ{qRacw=vu2Y(aCevB4#C|*Ah^4`yW5-K5+u00ySqEVCAhmgG|)6X-^^L(%$kd-
zn|l61?PssLPX4%){IRYW<`WNBA}TC%;sOjW3@<;&f3n#pPOe5**cChgK>6^0#ybmF
z%SR}6f;bybN;kebK7NQLP={4TmYYUXld7O3Jl{l|K$>c_`xHZSG=M^lz5!yNhA5GAujyIjo
zYF^jLWTxmV5`3WzO#LJFL2<=9N@tlibM9%GsU-#tTO9rg2Q!5#jDEgvD}tsDG4d=9^AQvNSq1@;!>rjx;ZxmK{bD{wkfLBTGpd(&pI-&EPh>WmFx8^h9n*k#T
zp_+ss!#BPF%@>^_&dLir-ZCIu_}X=NkZr4r^E#zX?ApTZB6hJ5zT
zpYc}S?TK6eI8n|gMZT6I7#wEdm0THmn@%@*4-0v`Z!0mYZCY~>qn#7fj1Z8d&PN-yNuPLv3I3f=_=u5
zbpoVYrDDADc57CD-Z-+)K)b8^XRkg|)vriZdG5Z70U`b3#_=D4l
zUJqf$dw%Vu8=m`mI{1oi8u)~3
z(NYe|h}ZW`m)w|BVmR>kT<-T^D0FpJ>()QxBh)X)Rgx~a;2Ny$>rT_UN;9Nc(W
zQi>Z)aM4tzQ?VZ-Ao{+#iS)>5&+Kh(Vhn!Nsp6B<+wWyr5=M1;JNf4!9hAa6fY1>|q$06Vf>a
zU~DaC<4zvzH`>-^SK|{#0CJUxl8?&<$q{X$T^!Xt$Rpk={oXVj9d#ap)@l^*{adK3
zRpUY*HplKyqpVg1GK|Tkzi@IQ3gdB@=#ss0{Ifzxj_z>30jfx6H{tVJgW>MSciqR$m2gW3E~|G-r4XA{ajpfpVcTpUUabD0=!
zW?M+`d)ogj3SJ5x=8h-{(O|FC^Rff}Zkfq8UlA_9$G=6l
zr%du4YQW+9%lAjJ*6Ee>rYGx~7xSmv$P27D?ca}aQfW4J((KkE+m5X06)`ipK^dh*%_x-+)
zM&TM2Y)!#bF^^X+4(2yi34kfs=pmgjsE<#~2HSi*AOufEOzXzLx1PhmP^jrvJc^y+lMTDfrz3feI}
z?10Q=WU63rOhx;>zcM~sA?W7N?A@o6hkf@^368~NfcABa&T4e`sKOHMK%QouF$LRp
z2pmY?7pDA}#0xgBk1ZWnnjk`+ouO-%+nd5D?XjHD>qdDp()MByPX5?>GTHaxZcGb+
zmaaJrRcmcIYm<3=Y1552&ton6l0(vaD+oUjPUmo?f4s9GaNN&g>a7^bhWx-m4%=>k
znEOk_OLkt58*7FeI_oDaHHjZCsU=VajtSPnj*5|LMkzZ!h&sT4V=9Rv%Ap8^IHtNS
z#B!|6v9U|zyTx|M0KF}r;A@yNCzk+A*St)=Ql;;(kj2@V`%kY7?bg)3A8yy#wSKUU
zez;|NZ}T%wf{|`NKE5O6^B#jm?cB~Mi+X3!h_z&2oU_dgdJY{P%JvO;{@jwuLe>*=
zEc{|$b0CYB!?PHjX-3Y-rEbtf?t)$W)^vfewNPc9dZd-cG;N=Zxg;Gx`{EDel!to|
zq?a~`P0Eo^7*>X`!)fI*M9NsQvq#%0TuHuRap@L~+z2h*?ESJteJz--N}i8cFG0eu
zN3Blll<=Vp2M%NB*QG5BL?12vyYnO74K90wLD(LQro;Dcs?_R>L$wZSc<^7fgsB%{
z`6H46JI(m8m*#~$Z~j-XSP!7lOmH@;I#P0=%k--0XzR_918ocKRhFdAM|4fG!ZS9M
z$x++f4e@_H}#^=Kav
zs@BNgl}lLS#FPl^TI;rE?_A#|U}^GVlx~C|zKj_0bNp44!_M`qU`l-T2p
zr?Hx!G*I9L`v?Sx8^R2OZpzGLv}U;(sQG$2|dfFH06@!sk0*miOu54+%7Ni%`3=Fo^4RDb9Lp6
z?w{pC-&MR)ERJ4c9a^%D(VrHn44EcILi{;f_;u~~
zAO!BQm))i+EH~RMS>WTVZgAUH%7=^Syv{N!*1OmGpZc^o&J@)pGE#K(7J7`!G9hIB*fQ%F#BM&D_EzU*&zjP#*Q4wiJVT^7b$
zN&hzc-MYuu>ITFX3*GI@mi!4Id=*yeV0wGCL~Gy~`zw=ebpwDltPDT?S3UXg6ZOBs
zfA4do|LH9$K0@cNu;W1S_=}Al&7Pr=v)7$
zS8hZ`U`ePyc_dNQu5G^okB~_r&U^j)Nug$Z{`5I!+TiAv$+AedgnWk%lk08$#NSZ$
z%0wOsJPUFRGgtYqo^SbMW}iEjJZsL(bIs-9Ew1)qyzrH#Ch8QvKHy<%2aBftJ~>WN
zBTr5Hy=Hjy3uEXIb6yGO*QQ`E3*+#t8&hwKFM9V_n%@c2ir(!dU~7O1eIN)uP0kzb
zYV%rFqBXOVpV&bfleJ*^^Oh%dG{cB}LJr>g5f!{WF}wdpGV3J5RSUszpm=7+CPa`1W_E5
zOA$wlV)~mF6fiTW@qL|=9BRuW4)Qm#zxC!h5Dq2QV-a^kY#!k_X|Fk1a{8>KtH6*b
zssw4eH5>!D2R1Hq#a6rFu~JQhGetF0^~0+ib=KX6+OXXLfW0*Zxekf$o^~9{r4po=JiMi(`
z8tp(J4qm5L44ws%jAO_sb!%_q)60tlDcfOmnBpruHamrzkr}3S&PArIKQkj_k7zF<
z`^WA{KYtkV;gQ8$AeF3+&upg5j6k&NJFfY2;0L$qcUqI6=iL33;6RBMEZ1JymoMCO
zn$;)(&)J%Ap+9QZ_4ezhE`A?T6GWGZ`M*DRdzLdsU(p#oXL~Ub=pA03
z!FD4vDc2GCtLLS%!{m^wOoQRVNb1gI;Y8hgNO@tXF3eGpPspM(mUO5BSBg~yeSu{-
z0~UtZI}3&7O|~
zj^tzFR}zF>J8`|e54MAB6wS^?)ORi=Lnq@RV^7-&rcuspdy^?AfzO}6LdX>JuCg-Y
ztK>tBGMA9Xy}ao^LsiSayhhC99j
zBE1+L(qOIKoD!FDm3AJ1(pD=++|4|2CIdH=^uC=JA|FL+{;SbXp^~C@T*(gyM(4Dwo9bUGIx>^KSFmd`X;0&9W5|KL
z%{FhU_qD+Xi>?m*j6TT*8pqDQ7+gVsT-}uHJd2Oh=w>87;~wtWr)ZsZxvwJ$P474>
zh_o|MHs;e|OvR5joM#)K1{>-NcSa^=AF*jdVBfaRK^CNjG16=3)cNgELErf8C9osc
z@N`#Z2;Dc6RQt3iRq@^&gf&B{vEPlHJdu*9HLXe2=P0=
zuen^%DrSaM@(!{}F}73xKUV=?{T+c6Gi9nhkV__!&EPh>L3SmvFVFNwQENP;k@!`h
zEcV&S@>_7L4w=)Eat{j}!D?Sko*IP%(~8FS=@D$lQmBsG&v(iC%SsHdw0&4Q7?iiQ
zbh=NX3cV07E{>Wj*0n~IyOjkXW1!SHp?U_w%HE@R$>v)c)?IVjv}No9Qi$2f3%*77
ztVqN~u-
z=56wP3t3Rt+>X#A+p+s8g~CmN+5;=R^6Tbd8`(4H;_03B0}$OFuc-&zNND`9pgiGY
zEX@}?=5v{%BW!qy8yPZm&I46kYMl5TPq!!i%u!+dY$V2p+{jQ7-d?({gJ}GCTaugv
z@4>6EW*soqZ|lfir#G#UBj*F{lD533tdkL3(IYqju|64owtYs_a_`?_yXo`))P0|?
zR_q=-PHCs9DuU#Aiz@@b^x`zMs{mWsCTY4KI&~D!)z3Jmq#G(ugqOSB6Y-_C=trcU
z=G$|cBQ{#de2r4mzp0GAZ`ED{O4VD_~GRF7Ke{Yqv6sH1^#67dsjH;Lu9u2oJHk6HfO*K9+r%O9C8+TBGT<_An5Gb}Vn$Xft6sWqaQ?AS
zPz>fh1t>jO?p=exbtkT$OPRLy-!?AQ&_v{lv^@`edWdm!x05!)=-e2_MsS>->HZF-
zWYG*d@{79e!$E)-#Yu_l&I9s0bq|=|ys6e8D-_#j>+GfxE}ayq^k-hK?J%Yf(<%N3
ziX8b+7yGv;HyX}!(3>QJ~{S%XjGIhAm
zpZEOpzrQu6Ft6`lJxulajHoCDOIXJ!OHaB!Ku|YhUC*uoCI;liV6V+uIO;iIl3qL)
z_3ogmw_5Z7E~m;$?E(GXeGe2AxZ&v!mAsjj>3P$cZMgkGkM8ZiN3%@-Sk7$=7N&#(
z`GgMb@b^CQ2Dy9@fj%==cLO|sjg$i6Hn;g7!>MrC+`6~TnfNa6=43IRrczL4)wO6g
zb)BdO1N*pl=YSR92RFGAf18@@aNL=+9KMzmPPf`W#Qfl!{Y0TVJ`CU8zLX*H3Q+79
zkwws?taJwGXXt$9)2OL`c^+HLd*&&C4zisN5*k47nh_JdD(#6rGwf9h6zrCab?Y9k
z$?FNYhh#q(wYl>qo?+^Zn1Aga^^6fN4DZKtYjnz2@4a69KCHIU!_UMt_FY<%c8{Y5
zn;rk#@lLuwvm|lT!^}LN!ZH3NNr~^vv}|hqrz8?SkF8JdtXCi7${dAlHijgDYe$Yc
z=W3F-g0F?IC*-!+`g~O|orQYBtlQtVP;n6cdhO{`hrSDlNJ1_m11MhLQKtCkdX*R%
zUj+|Wp=WK_cK<&>2LBHrlm2f&*2NyZoD2Iqg+&=sCX;_6zS#VbXH6vsP$q=nQQIL{
z(5dR$Bep81xL!BoXr;)?P_MEYAUb&4UpsB}_T;8>S{+UDyB%d|9o}jft&R}-?
z^NkV`vB?R;P_|~FDS-3|%*wzF&kw&$eh>%j_eBXBYNX$r{qA5ws+JbxHP~KLsWTE~
z6~fz(Yt;p`M&BHW689zmz7fJ6s*-#4QaL7$uz#!#t|`WapeYK&Kag{boMJZwdPo4tW0E
z@+)?tm3N#fr}Qst`Aj^BPkM?-JjL*J>Vb9zS6jUa?-db34fP&6ZOKMqPbV{TwtPxD
z4gM~2nOcc4GyO$S1^?WS088uQxA)~fu3u+=K(DO*lgYbNbmLxQKkkccrvL8pnR`Ez
z+Sqpz4!p)48UJg*8Rud>i5h9EVT8b6dK20sz&&!(W?qv0iw}-6HTea@$nEz-a@R!8k33j8aY=uM4n}GC!TH3c18vA+`>8`_xBLOVRiQ=^p;qcsU
zn{B3qo#kBvoe;tkZx;!71Q$_s>E05^f$juRKolRpg)|fJZ%33_S}PqHMTftK=A-Xc
zB5qdiV6MMhH8%OQcxOuDpE$ncg5Ga6{nIJJZc~aEa#j20532Fty>SYv=YMIn+-Al-
z#vGmI@YbpCre8lkiTjd~B;
zw$yILOD=&m$_GW`C@1ktk
zf>B&2rN+V1^|e(iL5eOXsT{*STS43Z&cFM5QN|j0*1CN-2>Yjqn`#1Vl*$dkp;WYp
zwtfHtMGtB%f^UnhxYn2`1d9T1Gd>kNIb~dpba3-|sIE}(I{c`w`j)j~K$oe|KjeJQ
zBT%b_+HQkJ0AM?xiy_nZ?8_U9uCU;#N`h8EjXGX*ZMA%U)g1IQBPy>%=T*mnI6SDL
zW*%|&<|a;+D-IH%(aO`UGlG@$hwv!yPk2C+ph?$6G@C@Jfmu0M$VPXqIbJK$LB=cJ
zxAVQ88f(MGY|JzIYVM&=44?1JhOHT1-*BqSP!{weX=Fib&Vf2bq-$jWVI_xqDA?!L
z;tf*7SDZkXgt}IHIHFh66)$+Bk+e>^+kZf2^VPfhygD<5N)$XosY(RmxHf1m+yaO$
z8XwEV(I2_h!t*!Dj4x+BufMOnWMn-S+*hZX=sn{45LajJDTnxf6Aq%oYLe9yQabK5
z%xGT05CW{O)(|cf)klOx;6#BszbIk&Q1!JUf`~$)g+-fBXR#sa5cq&J%YwJSn{`P`
z6Ye9ql1D`n*wPr7Dw;mN5p&Ximq2-&v}SmDU{8}wgsSj(rTQHIwQKhLwpf57w{qf>
z0obcGolv>rMmLIa@50XCxC%uCuu=YY`*tbv)&ScnZ5C1qV(ZSryZtMq2z7j*~EZD*WSGviY7w&5$44LHWVDxAAc
z+L5L(w#x?`+ZiXao}6+gn0$nyyOLI=^u2$X$!69ql~0l1{?5KKy$$7Cv4R}G$PBgf
z;*}RlKzz0RD-d0BZqB({#A1y-*8dJn^XKzEA$jL4w=fVrVDr8A00=IVDc(rPvjSR+
zy0_xFn3rw*=r{_{l%YQFwlPgMHD#F!`E>ZVZvEyG?0rR#f!`cABPW}tXL|435;zND
z)hpH|sC+lxpm2uSibW5G%LN8>!C3jp>xo?g5_ryaFM+n7UD-dPEBhr?1uxA^Qe9KG
zk9jMOFA!&p;s?Dd0aA~A5JR_8=Bt6a<&fD~F2Ci5>_Fb(*rP*q-pu<;*5;?Xpr2ns
zUwuV|O}~X%cPn(Mse2_rJA{j0dFFb!#2z7%uToh5hLcxv`xmvmOC=?W`XR2FSWSJ%
zRx4fD>`Ir81w{x4EccUrtq)gEMOU<z%zcc5d27UGGXP;}3
zUfX!cMM2_S$km(_-7@7X?dLPoZBFnM?5#^^8?)MTj;YgS{tWudUPmon=4v!?z^+R6|7ot
zm1)yfV9V9R8#|R-+IF~C+uJy{abQ~-pO?IX^9al0bvr%4iTDHtUN;!7ocv|vF}L6y
zGaGQkl=ML*@pBnW*xG*mXTfF-igyz9PjN<}pk*O^UD4jZ1vpylYSOo7Fv?PnJ!$W3LV+G%o367UVH_Ur+ru6@
zf8r}&FP)ynw^Vv#Q}1Ne1m79r-gT@FQGqg6A5>l0FL_=T;D-v>3xu_`#-MjetuU+8
zt~Stt9^yRYg{zbqU9?d(VZ;4;lqsS0uy49kr9A0f>!^B5!+1=NLfp~2Nn3m2Nqo=nuT8dmM+owa
zl>wAv9n=+yAScP
zBwE7IgyUIuQOL3PO?f{lamms!WET*EvH3_R5ttp`jmspFHn7pI@*4qundWm04<1!{S~s0-
z;ckYI-~t#4_1D&#d@zLTs*Uhy$e~Du1G`nKCaDR=bK-6#YP6bnJ
zQy1>fiV74^(B*2a`vwo-jg?b2113kK
z`?(4mahqFEG<#V&c{!~pvUPXaEc$j<;`pjhiBQOhd;<;GruvLrX@~27{I+deR-}rJ
ztmjs;@ICkDV7|{y?#6G3!04*KT>D$FR?XaH$%tI6`iNo}8=8BO_ZyX;4!TClWBLgh
z75`tbt^o7Mi?-(2q$yDUp!@8tCW%K@_$~Ewp*F0r_X}UO>Zf(jV6q7xEIz5fu|M7h
zC|mr^q+M{})9C`;k9E;PwGtT3v-!B5JXUGIZU{H4^{4J!@Y46HrFrKhw5c1N0e6_;
zV-?ZJe>3@;%3PaXInksS%(V!uT5AV8af&xBv-kf~lvP?SPq^Y}cSkH#5A3x;
znqHR7ag*l1;PD6cE9N3r@?ptZxAo_BNr$OKuO2Ug0)*ssg~^hKZe~d!T4Nb*2i?Ir
zn+}me0$L0wT9ifUX|JbO_zh?Hg
zjn?fWyY~R#77u~{O|5(yx*_+v#d&Dl
zBot0KUiyB^W;|D_%6fl55UttY*J=Z;Ls+hAjLn+feE`PP#O?lSFEZdkgq<+qiy8n(ti(;$8b1M3nj4OuX1e`sc!+w=s7jfzOIpOiT06-ZjF|
z*AJ+Lw!GEng7iYGWM5f8=%LrC9*s)^f3`1E^k%Ij_alwt3PiI(YK{th2Zi_A3;J$Fq#P`+%NnGe)_9k4WPe94ApJ$IWZG0}D!gp2cnL=zu$3
zd2HbmKW1+YL%V?HQs^isC3mD|TnH;0>Pu7?D!r}rsvCtBs-z<&+A!{KUZc4<1Wf10
zo9CAsFa3B9xX;q)EFiG(5?Z_bL^#>obSCK>3>#Zg^X6-TCWus8yO?^sy9BYVK7d8v
zoqBc91KraHT7SCdv-Sx394H!S2N3G<#1^y57U(t1`<}s
z%1i-{!Bn$)%NEp6aTCu0f$NzOjZRy6+Q{Eg>EHJADOgIx|!`;76R@alA^;Fx~Z
znb(NW&a8dwr7rk5ik9_0%O||WkNRH{TPOAFwlM!(E{iQ4(Z4MddHX~+$J1WX@ypNm;
zQBSVr&B|3(Gutyv^sf2w#Rei4{W}@CU0c!!h_i!^N4|7+V{KbS^1)ruBXRQcUBE75
zXfcncsdo}_HiHp$e<3k9V75l*7E#7|-Pslvj(qLhntG#Zdhizo$F*%=fMCJ3|F&q`h!O$E^)^>L;rI?H0*Ak}4;xGO7D0`-4(LjK9n#g>F^CK_U
zW5M;61C1>SxOjI>3x9k1qq$Y27;ID&@P>faUthJpL*ucp$@#L|=%W`ZXwTvg&yG#KVl>Lnp9njy>9V
zovaVz(*gpV@>!#YElU0IP$TE&@z`DFsEN?#{q%Yg{pX$9few_pnWH@Zi8nMgYvmrk
zWlar~VB=n44rFQoVM4!S%7!w0f^FlxFp`Z^&}=&Pk;ab(>NbMrfM&!NTFb7Pyc49a
zGCNPs<s7t~NR#|<}$P1J9WFQw{Re1dg?NTFh(94$(M_4Ex
zh?`{C!%r#Y1nZY6o=Dv5558!IeN~5la)L4
zrsUZn=@J8&i&L2sgSdGh`L^1v4;Qg*%eLAA8Uwezj;K@J-;Y+V?M`#Q-6;ia(H{z$
z9^};6X%Jf)A9)m>qpnIfEc159Q{FF##Jie2Ss}vRrF1r3n^CCzp4Kg(3R9dK@^g|2
ztPtU5JYM>O^aPd)!jcW)3$BSB4Cz#0LTf}yAG!vP3{yq2OHEE2zYZikB)&>XHOEaV
z^fL24aqY%y4Ba`;kf13MpXTaNGN(mr6duA6kPvn}XNi0ni91TmAV;>jnZMg7)I$37
z+4I3M0cHj)%!Ht2}%^llnjzq_~cJ??K_#gA;B@qVhTEp<68c)PEV
z5V~?(96r(oIkrfrJz*+1$QW71W)UgpaE4^}lUwuEa0t~@OYv&RI@R9Rcf$BCOz%0*
zkqi;MWj)8z^jEpM8Kb{A`=?%RS2vw?T+*uDwj}%@g00q&
zlnz3&)#C7uF`hny6}v2pxmnGlr>f7Nesi;2^ZW83aPso#O?gv^GLOG!&b@%(4?{iB
zK}QJZI{n&TdtIj;@8Y=%yv@5;UW2GTJQ+@so+@sgOEkkP655Ar?8YNshsBI-yXCi(
zfA@t2o?Ems*pqb;Ez9v;l<_(4>oy!qzy8&CqVY(f1JNOC6M?Z;r*u(K6h*t1lz~*K
z#m)FfO*J2R41pSTN)PojiJq4Punxd+S!(^jikBp2Hv97+wN}A3On`cs*A6b4YlDF~
zC29r|2d`#Hlcn^TkL=>s5lZ+^`X2d=DS7xMOqV2<#TWx!n
zzpYFeqp1?&WeQ~DPBc$)Pt8A^6UH|$B;U+9WqpAQ1fn^IfIvM2bj6P3QU{OC@A=m>
z-k#H@f^LxwRbA8nZ8Lb=GT73)F<#CIWg^dXi;?B+%j^-BT-%Q+1M398o
NG}Yz4+t^!({4WIztStZl

delta 12271
zcmVzUS0kEY50yi*|4FoKc(*rjHI504i4FqEXG&r+z1lSgTg$w`y00000
zc$^hic|4WNAD)CrNhez&97$AExQbgdLPXh$knChBk|mXdWG#{PB5sK~i3;~{E8@yh
z2PsP%ZI-em(RE4Y{e6Fb&1c?o=AG|6^E}V@d8cJ-Bi0sE%zL=?LgO8Z6SLp9`QM|s
zM|FAed>O?~eR+bpB@~x`Cu?ok!J)WB^rrcWn-pJY7!DkObxGH6N@ZQA*m&Ve<(8`y
zE3{Y-od1*JpKal!JHVKbHg?{B3q~r;oIo(@kDA#G
zdNBA{S&+HA8H1Z!v(|3WV(>`)1wlgv21`ZVvaJz=->>h~5ffmryXv5f_8gq|Z?lWs
z35o?r0wRn1DIVH3FZ1IC#dV^;G-hikt}w)+CkiPZx7_e_*BV6DwO9
zXurkai+0YtqvII-Eqjc+W*>uJDV*7ydjgTi13EhUoDdnZX)Oc`kp}{bncW$Pc$#0Cv$%*zKmYnZ$z(*X%{rv?q#?qoY&Q9
z&2-0AMDEE={`2l&=b1L8P!kM%OYC5syO_obm=lev^8%
z%BBjD;ygWni=qld+HTA%R>55KlI9w|M~DpRIaKsEA`+fjb-(l}BDv+KcrI;-jF?u9
z+qOg96^eUmV4fDu4*SSMM6Eu4>IwYs(ahN1;3t6l7xC}Z;iou?+p-C(SCI8xIG>l~ckEwKh1?uebr
z4+k7?^LrQp^{(}EEH4j(6oTvE52aalUFU1YhhcidO
zH#S|!#n+R;AG*xWD~D0M*Rp=qRzHe8Pk-^U0ln^V2hOtqimgVmW_qi(3
z@>TV1M5em3rJex~Uzti=jDvHjx~cVLxl=5mQ=+GK4t!)$^fRqyif?~sk(hN1{-7z~
zcrcH_)!ZSIE)hg9b*=r~1d(Iy^YwKRz>B-OR~qQ(h(t+om(sH|-fzPjhuaA8a
z1^jm8^1rt9q$8$C>Ns#8KRS
zxzqB*8;Ya7lDNJ58T|fvTH1<#&kUCPu+QFc2O=?s>*|6cL0_Zum16f0+2qDt6$U+h
zSJHJD0R7$d_dFF0=Q+cT|4Zo-#igDTZdLI9iMb-_@G}$}v`hFM>7w{$!BmCTK?Yl=
zg=hNxVz3=YUWEa_GrKADyAb%&^N4x_SEzgH_jmDjc*a#rK+R{3m1!uHgT
zc7ng!i);2TfqiCwlqMR$Grv+Cx~+4WH0bo?AF(%&fPPit{Aybue>9}^zfuC;@XsfguTw;1
zJjHkY+XUcwh6m?&xW|cyC4~vV?-RL=jQTl5{x&!}d4DB;BITwZTIZl{-}$5aPhX>0
zq@i?!4)poK1c%fO;0sdY8qa;bC^l2SvG@q+qASr}pak;Iv(2HYZ+}wU6t#S}KJdxg
zWumRNjlpZpq&uq>5HaFR9}BiaL_PD}A$h2yE#n23R|@Bv)2T=?XE1lT&HDZj#iz5k
z<>zLBZpC7MM@NB2p87vE9sq97*@hEr(8;0=A}9}ZVG#HH^qt=rY<7Ik?%{0)3+cLz
z?UF!*N_}0b5BVU#^IG%*@VR(tKlOMx7vHguddiS5{36#nZ(Gmcx3kK;dkqwC^_g3f
z5kWEQ*S+TFfZtost2c5^flfw*1$80kxU)t^bd(r>T)4K?d~*TpmnJwY$3i5@FwiF(
z{FLvCvG`HIPj``uVzB`tMdkemMyD9;sQB_?i8J^P6)_z6K=G1$e^sg`QrzFu;gAXV
zMJHW~c?$R?beXT1ctEjeO!p$`wG4jR+$3oMeW=k;ec55aO{V7a>JYdmfkgMVH;^l0
zP34w<>w`WoANj(k2X&ZQWYJK_OCPy!)gmax*Sb|$(s>k{Uh2pg2K-Vr+hu;hd93iO
z7Oi@UKW$yORu*vES?w;dy#sPrBq!oH_-RDtgAhLOr!RTm?kIzQ*9BQ|eu7TLZ8(9h
z$}sK@?A&>s!K(|sr8Gnse1ZMmyfg>;TdzTXxx)YRW{FLm1n4JF&aUEC2k7!fR&+U>
zKio1%tT`T$Y(?wXrH~iW1nYWSL67~-t|E%yQyj*7aalOrmq==cpBm&ed%H8m`LK`l
zN>&o&P*K^*K3&MiGQlzY4BXEqv-dR_fTP5o#=QuBzbGwba|duLPn~7+ASX5@_-Jc?
z!F$6Sug@uhuZ#tX+(eMGByIiFHlIX9DaR{n0Q3+nrlGSVp21lX+WPIlyB(6wnj0g*
zZ#@sNqkN!`4$jodLZ0sm78s6mgnsVwZ@sk`)TjSQB^L6^_K6?8@#To@NMAA;2Ri+{
z(RZ0V_?dBs@1y+Vh%BwKZ#X6i`n}A5J=Wz7dK&++vAq%aeBAuO2H^AR>~%Ap|IhPI
zVZQO8pLP$IWJePQe_v{zZZ8bENuf071njeMf|I)i>Q@QitTum!$WQG(Hh1qLVkcwq
z{G&S}=?6cx)Xp;4OS`bU19&~9S7aHJM)6Z)>oosE6#p#Zs+vPzOq$8-o&;Ziuw;it
z7H|Z`LTm-ys6jm(2
zrJ!dGZQ%oB@ZA&ft2G34)Bj)FKoxw){3|Fwn|hdHyDOr*xbS^rc6ewc*&6)MHu(=>
z$REXxywE(z!`pRVJ0$?mf=rSoOiW>9^$J?EfgjugV<#^7{Ttc7=-WmNzOr3jPUbSj
z{{wNcWTunRCm4V33IG5A0001ZoE?^VJk{wJ$D_2UY)#p-M5u_dq|n9Sr=})~Hrfkg
z$yE_^Q+8zuk)o(%os#TEvP3Gnl1LG`xT5G?Cn@yCB(YJxVPM@{p>X1-y@q<_W^
zA<(L5WMajvVaB?tglWzUk8$y-^~i@GJO;aswl?~e9nL`7e2z?!CUV!2od`gmxz`r0>(&@WBu9jg7T>jUdZpcU_c#-KOVvg9i|A
z=YW5HS}gYMAG={>i1S^Z;oVtA_dKK`Uk2GHzA$|R_V05Rnr?*b{KayJ{ZFM83Sc0G8`W<;I*#cD^*wnKg@F|Y9>sIq-e;%uBp^RK>N@An&}Yo6F=j(nD7vypAMzuQyIHR&KlJ1Tr4e5;h3(9S%$|I_nDT!`
zTR)}896Xb;qplFNO7^))jFVD)b_sFhPu}2vgM0KAzn7au`4(C-oeRB*;x}h0$j$lb
zK0RRCso&+y5x+m4&;KW6TXAMI<;(JM=_+;bv4Q;~%Gcmb--c(P=Wu4i56XA3cxwjY
zRV|(@o2Ggx(VV9Lb629ns{y=J>E(asMCYIOTUG^T2vcG@LgiUlUuwYpJs}uF9wBJgKg{Jcm8IQjC}_tbInTH$5LUSlO&|HOXB
zpLee9gxqs1UYqK;A+gk00C9gdj@|v@0C}oL^|T4rldi3}VGnu1sqGe(kQe3Mx*r2s
z{qTNA>Mwnh^WC<9vK37}-=H58SGJ(1^YG4LlnTzbCWH?V5fgqQEn{=PK^4R
zRe>)@3_ki}{YTu32wg5B8>Zs!6p($V9NJXjd2TUY_Ums)-79o37nzM%l^Kjr9Ohxl*i
zS(vVa|C$tSfmY~8XWxHIdPjMC=Sh1Lc)@oSBMQ7R?eSKO`u@~e)7`KuZWg_@9vrA`
zc6S2LJr3T*4~BNfHZDZmX8vPy7?hV^nsX)Lr_Qpyt&4SKrqr9+l*bRM3m$+8cHWsS
z;MK=kd#O&%+uTy}H->(Z
z1c$o~ynS=GaV_@U*Lwau{YR&QCV3WEX&b6%LHV@Y_#DkI_Y_5cUGSn))Mf|LcSOd&
zp?luq^<5PHX5)bkc3@~_aYX^>)2FPXiuM17E_o+IHkqkQYy-R6qV1xg?;bzxT1Ug&lyp?gdY6k9;~D7)52^cMBkf?4h~pZ`iXxtULM
zXU?vhG-sY3%BrM#e)Q5^MGASN93R$J_v?IN8W|!5kM&JE7wfGy=yYN%VS2Q;Uhqe2u&t6`$
zQi0}7{niE}nlld%Cp@$QPhFk0AroBwdNzyZ)F;ms?v#JwJ6`(_P<@84Vpij-`bCT0$GJ>!NZ`EVtxqqL8;ZZHmn6t
zT@49eN%hgKbG!!Y9`hyt)dB;)v>4JpAODWc@5lE{W_p&;d1ghY=xKoe+kK(yPrBE(
zUG}B0YckYvQKxy=#{a?j%e9XgM&sx9Svy+_FhMXDPI7d>Tg
zBUe>0g60P2ow*b4S#;@4Ru@<#lWhMA?CiMBr28m)t?EvFg_ERH{uXR4Wv~xW-?UIx
zqW*s;uBpZeq;rbsO76va9J^&r9_mO;u`-$ode+X1OTu}U7+#A=z`A*A*-rXDXRf(^
zrSI)AyS}3l{4>Jz3H4{C!a1>rzEnd&5SsbtJNkaudATQTCN
za;BbHfx%IA?v0oW%8@OCexO?3y-V~S(wTow(0vboGY6h%2-s5au>TqS)0kRe-{2=4
z+Q^zka=R?o1nYbKUBhYKDctLlraBw)R(9b*KXYW@qAToDZeuZ5|Uh+hhtx=6~XvT*Ix;N3fa61^bVCNGt=Bk@s|E3Er9!Fh}-iF!0$?G&Xa%C
zHw>B{{+jQr)kSN;bY}F~B?hSvJlVRWKmT>zYI2``{-if9O9y=q3Hwi5${9N_x}WGcDmh-v(wXuC$HC`cQuREqZUp=WN&g2F{4z
z^*RatYnJwA>U*!9@*1fQ=bmmWas#*Qf1bIC`b9))2KA$vwBcl`#}_)=`VM~*w_1J;
zg5A%Qn((XizuL!bY#^H+4jj6Qb^WvB{=e=a`AGUYaJz4o&T;6I19-jEFW=c1Fy@0R
zx?Ij#!QX{>_CEEk(z=y`RChCe1xu9S7m*R(PVZ3LfP!9q*zH~K5qSl9*POphr5X3K
zSbXo%G3vvm`~Ap2pnKej-hB+x;Zffy&;Abx78x{?(I*&x01E&B00000c$^)Vc|6wJ
z6UXh*&6XAv5ebncSxP;YbPGv^7J4jERLIgo%9ez(lckhANa
z!KV`pNLcTG8v_>K_9|!14KZr3C^Q}dz4d*o2f**!VnU-ap109vR15R9wZ|p+IffZ|
z!93Q-AfH=rr>#3X#HeQ(m`y@%PNxh$7h55Cn%*GmJxnW;DifV*nymh1$D
zre8Pr!2Vlg>{>(cSVd5e0^&Vui!}LXdWfMIz~=aW5^-8m=fhYtLyRYEUPo1-f8Td!
zO980Lv)^|g;=0&|OUPq=rgI;5sld+fVmg*MHgnqW`T=yl~jD;_ki1jx4R)5aYR|nkpI|@z6jWNC`mr}
zFg3&wIqd%Y>ckME;ez>=48&0mzNE~D_#EQ_E&142G0#G43&isf<`gLed-F{9^JBi`
zM`l?G*jS|Ks{_9UaT}Ogh_{S?_ES7$gJj!(m@4==k~&(N0Kd*3i^9x7%kv}Ry;F34
zk?-g{2PLNlo+3_0(cv-;_>axjSLOoGiwtjwgP!|%Dd!IOQ)s9@%7;9dRC6+N5vNmp
zOzaxjv9nfg#kt&Y8joBA&iWn?G=^WfxoUF;&Ch3-`UEo5UW~`VKGeX2BLu8|>T7R*
z481@&UuQP)1o
z8*+ZY(yOPz=c_hWXM&1jeo0h^X*&AH4#F;FN+h}ja_D)fyN7YEebEnEB%tSiHCr)w
z6Mh%V^miB{&cUn>XMf0*%a4{`g1wE;7cnO2TVa2C9vnPyQJ0Bv{!`k`rI2lte|SxR
z61P+)Td{6+PQ4^H$gejfS(YIV%kX&pZt#AnO%vs-{&h~OBgU)zZ(21_KiPcheizuh
zBDj+7@!5aOTU$W0{H#?bpe!4IN8vlHvzzX-F7kJJw#S6-ZLqZVUCe8$a57klbG8{UYNNhpF;VdhxB~N~Q#(fX(>a*<+enf>#&`|Y
z?WveY)(64N9}SY?prpK7IptaV#Kg9BnBS@%phM@reTVDxG${Z0T9P{A?0Iv~L>a7l
zWt2zp0_?tyS%TdIQKHmG9l|5(DW6UWaXNZ*UPl+abEo=gO8)qN0SqXdm!iDb?Q!{W
z5{zH7!s!-xOY=eaCvbh+hws#XQ{9VX6G2UGF3$qcr!zN51PpVy#4bd2b=pIoxc!Kl
zQZ87Towsuj$P@ZGU?rH5c}0Ty->PL+F4s{7gt;JL7y7wqu8`@J{nqM;GctEeQ)NKr893_R+A5Nf-24hJ0pajd+ehzPw%IunYJ!
zt#b?YyI@w>6Y8HVb8aP3`fm1L*IWQS>vt8q1>|WVWo~}R{2ozay>yT2cqgdtWWKvT
zrTRA6a-o~*dt~*C1geWm(Z7SJukcI;2d@S#^Px09btXQ6Ic@JU%*Up@2(MFo0P>AUiHwfYb2+rV%Bh8bk0UeaH5kIo!D(M#{9
z>|yR$+GmDfV;}XOP`%C1mV$gE+QyXsM+aY&X+r+@K&>n7TQca2CA|x7JGa(S-2Sqz
zPcxK%w-(i0;u_NsUFs8`Cz2MN0cAM&ZWVy+?B`cf{oXm;(fu8ELbXq3LcqO!&qPN+
z-+~9*9BBQ5H`*dVj=++~#-R9un@XV=-+9;lWi0lQ&s8LQ0sIwQu(yNqrRXK5iu_1u
z8*ps^?PU7~=b)FK^j}i}yTHu@{05MpzY$)4NBeNoR?K72x`d)b=)1bIXwTkXsJGO+
zM<&;S`(8-&W`o*kiCj|{ANVy{(+0mg?93F+!TBbQ5aIvlBRAC!0)3LFzvN>)d6ViI0r=G_eV&o7@O(7bWm-h%=j1FL(LU^KW8DQ`yKp?-l-+M=KY1BP|oTkCs0lP
zz6-5iKy5+N391V{#jE!~O}V*zdcQ1x@~e@hbLZ~b8C8Szev;6i3Wxl#%Ra>qGNV?8
zm-ZE}aisk|>{Y$xji@dc8#NS!LGQxW{_<7D>?64GhJy|;9djowa#;!wbDfG#c4OV4T&+E_e<2Ur);Qcr*NPTwHR&eYX
z;uyBu8qxdC&VywgPkkU>B4i7H)%jG8I_+2EmAGyr)^Sr&-JHJf-mFIZ$MpTXAg)FC
zZ%+2D(0Zc#Kqc+JTgj>;73}18)}TJpp8i;w>iX?TrMB(RyVglPt%W?eXzM**tn>N1
zwVsb@pK{;28X)s3jd#)h9>zD$c*8#Tp-Nr`)|=g&qF0A?T9rna+R;58V0-<^0PK!E
z#b<``=~Q9AM*0u2ws_bG{eO?j^)ZvtCm4UN3;+NC0001ZoE4XOSWRmi#x+PJLnNgr
z8Z@aSH0VyskOnzTGNclvge3EjLJ3idN;8KjN=ij^ii{~0A4fK+z1CiPtwpmm`QG!_
zy7qPL{XXk?-uwPN_nthtr<)0-a3Z~`Kp~F8+j`G5uUHCKTG`Ob6BHyl?;idz1*w0@
z+8<^1Q}A#wd3s?F1^MY7lUMGfkXvZr_V0EI&5vFvcyFbkKlyn7V}A-w&~M)qNMWVB
zVDzaa6z)2OUYb6W0%sWc`Xb~MvJZu=G`1wp
z%h-6GMzEG^hR#76Pu+F1W7pGYkovmM(vHR#YmX8wV;TWA-H$fUps_v7f6O)s8o!SC
zTpjvC`jW}5Zn;b0h)*U{cZkBYRjTU@mWx5jba!uQV(?^~UD)lD3|?k+MfrbtGgzRP
zP+q;1fq!$!e3SVMrl*wqkJ4cf|0`B?whDuKU9azdNigsgUr2U*Cqk)7vy*g`2nTbg
z&Tg3`g3--aB_9PWd;;Dt?cBoRt$Rs$)hrh3#medYXcj?I1Clo6sW!YjOnNd4BNxTP
zsv0Z|e9X!v4Ouwlg-Bqc!v>!n8{$uK@Sb1n6nKk+VbR0VGqoI!KfbcK
ztcgRdB1bE~;t-fB<)ELzA^5JHo4}mIzJ!>aQ-gUFY)emjXUapb_W{`MOL_?GYX&simxQ91rakHERardAt;?w;M&CwQE0U&W+)*
zJu$$O3nAarla3iY%On5&Rn^szJQ8D@=j9*ZA=lzrTXdQSC%Zpn%_SZW4tsUAp5xKx
zFi|xwj7OML+&sZ*9?gHLtE^M!^2qE`84gzF;dpP?^%O%MmG8es#%$qXQn=f@ioEaG
zVLDiI-|yN-)s)0W6lSzDPi$N{?31ig;pGv9jvi6cH>}H
z^UF2&TT@2=)yprB;D27#1nbE>Q21TArgb$80h|BAU
ziufkN5~atna^`;`?6A#`bG$(#F!X2TCgO|oo=dNf&8A_P@8q$%u3g5SfsOS8l@b5LvOjjowJ-bFP?zN$DCt#1uyFLn;M>3K}E>l?B-X-?3r=U{3
z(f9QP3dfZfwH@jaLi|C)EsOZMWNGos3F9dI7;pNsbr*kyLchmHRmk(WqsPwr(G*gX
z$DUREE<}RfzgCiCDI^3sPrG!0!jrgx#cSu$c-eoiI$l|Xk&NQGZdwdhb&JZ?&BeGK
zvaLhXmWAQVk*rix4o9m?k9Bo%=w5L>_lGQnc$4vNHmyQ5+>9zPBX#2ClB*-aD7ZU1
zNgAlr_$PmtbIg$vVQoyI!iYSB#2+h9>FbHHBSk*xj5UjIvWx#{&F5g^U6GK~$)V0P
z*>s5LUP|dirv>S&eCtToCZQ0w6*qR?F`#fUqE?Xql)|m_5E-9r8ZD7yJq_-PP~#n5
z82o|3K)Pl^&afD=eMX_G<2ZahF(%)X_;mm6pPzr-w(^+tq
z{L@9w&!7g@^;=MIX`lA4o}(bBh?q!mG(K+AnbV-oAn5l@OTkt#s+J5t7o@SUca%Ng
zMEp8Zcw@Yj3y-GHzxCY`c_cXRf4OxtsWU+5B~PT#n|bF~7MYLijq<7(V+#9BXULg!
zP%wYHcgeg>UWBI)JlkxyF>t=>@G1DM7`8@k{9g)$69uzZYRm9Y?P_!tpXAYfuP$iI
zLE5Ojiv;t!ou{92
zcwxKc$QHt*th70bBEp*pRk_APi5$M0-WZC|Vi9%G@Y}e43Zu;P_kJXM>rDMZTOEJ0
zr(BpfB`)`f?l)DO|Di5|tlq}!m{MNkJX?Et*u+9SSiu4li2(DC&?!wvxBoGiGExHjqUu7Qh1|qWx-NK5f%jd
z?K!&Sp5FYLzGEhbmaE|_^cBTO8hs~l*)JMC
z*R~(h4xzA3#J4yQzB!f8eEyc`MqRf>?;@!)-t2m|Jh|^z?HvXmNxeq5q9kV%!pTv7
z;rEq9=wB7@@r3v-_ksTe@ggy%vjGi@R
zTBZ##i1EJL`JU{ffxHcGB+2}=4<%oo_f-sG?sAv50}MXMpUSk>5MfHyl!xz0ekwmN
zn|KZEeWE3fHeq_3vaRlh6Zw0a_rdM
zBnQU&y($coqrl%S5|1ajt?_JZm2?#iZ_NV1TTKROp1H^(dN9=dy!vG{kHsfm#1#GH
zpjH(1+~Ng``=ftevRW&}nDXXqTzqPXtO4suVE3QaBNiG(Qr=j@A=
zX%t2@jIP!b-yHFZ81N%|$l%?~(`uIFzGuFk*0qG_xoK%}7U8yB?1xAlk~el-SvqG;
zFN?y@rKv{#Vubn4JzhfUEL^vDVyG1j-<{7^#gcryNcn%A#9YE(^BJBl0#gcsEyrS4
zk$v}7zb)V)(f6A&-)}DMEPRjr7=3L9kKi6#m9z^yZts%IE79a(oxnX|x(QdlHa13+
ze)a|hEgB>_v-xav^(dmd?IzZRmD4F;cvt#%azCNA#=jH~QrO%S*I+b7gk#Aunr9dW
zKb^FLn;(C$FiB6a`;T7+?sa`$q)wEaMsd$H9&Xg?P(?DglWRX`M-G$RYh-CBqbow#
z+dfOXQVKfqytJ?6trk+pDX}jxfTeNx9J@*NErScs^4``}usBr_
z@|XE69?Am8l~YL_rJ9f*zT|$a)qe}e?qbmppfZ1;L-I@0yV1HDB-gjk2o|b%P;eL(
zm-8PFW*eSOx=7|WqNb|0VISFJ2dd|ag*2|ow@-Ud`q5hA;qKaT~6Vm%v4
z4s1C5TTN~{;g9y7RXv11Mj758d6Ikn54pVlD+(YWJTF0TaAj^}Wn@HQbYVUUJTF#M
JOGasIWj??F+d%*T

diff --git a/previews/PR1624/tutorials/out/solution_000170.h5 b/previews/PR1624/tutorials/out/solution_000170.h5
index 9f62715281e9600b419cb2e5b1ef58a8b64e59e9..2f669fc4463f6d619145929db39c6f6872e088c8 100644
GIT binary patch
literal 11976
zcmeHsX&{yD*Y9o2ScVcMMH)y%g{anDp~=)B(jZfY422A#GK2<0nKCsA2@xVm6e*=a
zib@i(_crVZnX0|tbr;X`oO7P@z8~J_{eL*;zd!8Xz1FqXy4SU)>)M>$pra=&FhhW#
zK74!xHzD%tapH?T{&}qYn=?UAtnV9t;26iSalFR)`+Z`a5_T=`!-6e7rRJpmL%J=;TvEj=7uUEB$A
z5rUA$#fes0=jwdW&DP!B!OF>2_pp<#o9=|Dnhw5m;{ANT-+$26$;Zal?r%Mq2o@Uu
zQ*Y9^-xi*hZh!AM_b*R>fBmT^IH70jWb16};_>&o9_u*zvxqzmF<`Rdfi{%%R@}1
z;_TtzY)gpFfj`o-OpsW{MmHy3|KK^Vc6k*qIiWuhU#{Q!HXi?}He-fHNYBo*@MoO(
zf5(Y>ue*os!M{z+_1`bUZw_M8Z%@YEq2?y_;o$hq82CjI!~}AU&(*)%5hcn)?HQ4i
zlk=AWzHxh89OD5X&Qr1b<0CfCri4e&ezFxJ7XDh}`eli#Z%d3HRiad8fBQrbZuYK!
zSlIB#(?9Vr(Raq8zpat~n^CG7CcL1QCgjJDm+>z9Q|VtA-|?gFzl!giU(WrX@ukMa
z#Lhfb|G7!_XDt5uv@Sp0=7J6z8s@he*llJ*!WcLIqb+Q>>mwj?6ltCV8;UpRu|ZnQ
zz|eaw8v+f~yh2v9q3JrexfP;%xh-?|N;YURmws}JX8}j0rPGH57Pxwbtujbrfeg>$
zo~0QqP^z&O`arPnF7#r3vF1rX$P?F`>-#|iQ}{~YEXXM!xyO;d~6Ly!+koJ@yyTooWiXIc>=7q&_p0xz2nJ~X!
z<26Ap805EOTq8)kB7W`9s80^k2LjNZzgAsqXD&f*8r&KwcZMMS9_ZJfL_b}&?gCCA
z$nWy6wH6>fSfH-?-H9L}u%^C_pCFGe;0&piBgk}mxf=V_3N}0)ZoY2&kp=IAb9S>|u;BZp%25+xjC-Jw_|#QwxNkBozg~?E
z@>;6%4vDkj^n(k2im5DkabH3{)0zdsvC#z&vA!-Jiz~?1V8g~hS^C8@AlI^syW~*cWa;G0s3Qcaqo^6me93~1=G;rSr?cQ{kXXbgXBN!V)mxn9
z#e&MjcdrE5EEqK1ePCxN8*KP@dY525td}Wu*c3yMN0zVUsk=^)lUGLWI@^qOYBTx1
zQa0{Kfl(~v$O6$fi@Ni|u)g&-9BIZr^QqP64OSz_d&idUzMVvnHrx*3%qoIRtgHGe
zj`?TAEO)(yaV&f1!5HggLGG+j;wt*FQNXH`9?F78R%PbCC)gk~D8nz|K#)?ECC9q3
z-bP=2B)3%)q=uk*^!7r6eCJ==XNP%V#N{WZVc)M7KAe+!js^M6;z84cS-`)=GF7~i
z4b3b2*VdvRAIqlQnvd__nZuDf9rtx~S(~zwEkPbzoPIa-AsaYflLutb|K8x@)t%U<
z(L*%nszesV3~#T}5hBPhO9qRtl@a9g!COdywgx7Q~ttzm;&KbM9f&J*XhIXpg?@6@;9yjEOTuZtI~NwOeF`KF6@X1Jd;
zX1iaUJwayP8`?9wlnruI-`ed)eM6}hLptI<)zI2z0i1UUAsSz=vmmK^b(bp6l}VQ;
zg{8)_;O%D*cKLEPB#9Goo!{7SvOcKS4E-$B*=Jjdbu>EjND2qq-KC$izv&1Y7MxS0
z^^UM$rRrQyXI~bCKli)5;}GV5-%6L;y)3BL2#u}CW`n&3SFP_Hti!@7B0JFD=Tn!u
zta0wtSNk+5=@Dd9=UsawgGvh9-k
zy(?_+^t{*4!^C}6`6v0{zGo~HPc3~&kSiGp%ajrc@~+uOs~__*4*};ir^&Oy*`@r^
z0-UqudA211Sf9pH18Y;ivcSGXWOlwL8xGZ}ZnnWVu2L)CdM1P*FMg`2yih@qp-;CC
zB$pCo^Rn8JCpd3{-^e+BRKz;k#HF(s?OiJEHd=)Hl^C3_VUF`%^6dNX0X%HTj%hV8
zw!t}|$zS$i0YQGNKd&E&bz=BkKJjx2K~`Kn^^|`h`hE0?@m*0iNSGC7S{-A-qAXUb
zrVXAe8A*vs`&qC)<I%Ln^R
zRB&$0bdYPY8)gAdV`=9!?C+EqLluL!EZDpDeC8<5nfp51KPS4e;Myam)OPf5P_)ln
z`vgJCX!;8m-zP|MMQiTFwFIf?cYrh9mkr;-RclYn`;K|ngxY*O4@5XBJPzSJFLr78rh?~02geuL;&L`
zdE(x5iMom7hy3;Z=NIq9Jye845Pv8*SmKWncMCPf_|GA{sr;+%t>C^
z-K#jqFb18`z>tCOLT|Ad8`0vG&x}#3qQxD>^@o#hg@b}-xOErvV{5^la
zzAS3|XTv}2&A-k4I|BcXz`rB#|4sz*+9d_+zW2hK5ECP3&OWdlEeZ7J=>xYnHRV!V
zeW2%J&UJ^Q4|txup7nv~gM|0HzkFr)0@vL>_r-`DF)LyRSiQjAv#P-e<&xMQ$5T=6
z!CMi=1w;qE=URD)ca~jCZ$x>XoVl+nQ2vBU&M{@gJo2iIFY-?YC
znz6+(0PVfqEX{cpY1KOc4Ia48?cV3^j5uVQCuWWq^ogT|g?3)>D7EpU+uBuRoczDX>lev_hRo4|e)T``Go3R1y1{)0fDEZTeAI2MtMc1mi8@;@TKMpg8J0-{-@*wa?z
z&w4m7SP^OK;#pdJXwOLSvT6#_aqE)1q7WZ+wzYes9mV78H}N6Oj+CAH8Bx$jxzG>!
zJDkAA)RJAbBKzAjExL&5%Y7r{?Su+%!71hU_}`s%fM$N
z1NCDiDo4(uJ^Ran61<4&>IWU)BMv&1>EA@0H(lPN7VE&~COP>I+PQGz)_Q8cqrnz+
z?@?djcIXp5d~a4u%L^91Pg93HFN^D6zewy>Lv*_S;%N%5-!K|HsD|=#>f4Ikk-tN2
z@{&-*ID7dx67_y6cyE}Fa#G)t%=aJ)M-P9a;;_R|akDDQ<-DF-z(zUSna`vhkZva*
z$h#nI_ryyu6wxo+v@#oUj%0RrD!ykZ&23o-(l+XuFRf9&hJJ0kG}^1%R=ILE(kAXk
z83f{1?<4!GQGQ!Wtk^A#%hWi9$+D8*(f>J{)m
zqAf(c_xWmZBBJ=H^9pyATPL@25Kyk6T&Q#yajx4^dKqGOIT6c&>$Y0P1}#VjiYz`v
zNBsVVFHHgY*Y6uz+oPQY$4&MKBR=At{Ky|M|AtNqg;%)RhhL%pvr6SxE<^e;uWV@+
z>bc)88eNZetTclpe<1Fw-uN!XgX@sa@!qy98R-*&vg?i@79AAinuT^AB;`B#
zBJH$n=WKbzVE;8rZ7AQM^T_oe+MVu7)B*Cj3YD)tMt+OglyWK$XEi@OqmXa;_XesD
z-u2HYr}|){)|xqk6b^DJOh*0@&hA4kNSlgFSIj~Bz_F0TVx(JR1(R=~J@0-OFtL!>Xu1dURC@+ew@eirf5TXA0v#PFKw@$x9A@AZjDMt#A3
zDK0A!FE%VJU4-iiTANmLBktUKYRN3b&R0F{k(gIgK^~b4h@6a{>vK^4w1UP3D&C#%
z6|?CW4~2m?G70PZ!2K<@?P!1dQ)bmI#5CUyWoll&{w`7Ji1O#7(kHD#`Iekxx2XCu
zowh(_GTO_^T)U(m?K=g0>05$ye680ZsvaFm?vL%D{JDFrWDTOjvbjQ<$R8FSOCzV`{cH4Okv~!UzI`cw+3-NS(lRXXPZOTW(Igs|2>0?bn{Iaw9!g9pL
zb(`{dP){~pV?_zdNe;9Os8H{J_|<}Xzg=Fa_yEe~TA%Ww^7d`5)ir9KmV69znTLAC
z-i_bukPdC?%63M1hZlQ2)sYr%msGus_`Q&Sh7V$p7UwH5#D@M$-%SulcOHnN;xABg
zsKFWWF2}8~KEz`m?6xdNJD(QMad?e#Vv$GfsXmv~O`uc#wqk~pod>S(Hl0530p<6{
z7H1YC|7yQ?ni*nxy-$=N=3)KYnH!cM{d4+cULBNMRm}BtF6tdSkf%HY`7z4xYjqLh
zl*`@3a35m&WdUv6ebAeCWs(BY+~LdiF2MC7{z!XjPORs2mMld(MP4eC6Ohk}m~Z|A
zktan@JQwBfDYr!K!F&vRrCS&w9xUuh&c<_nYeXpz8~x$xoVx8C%2|b2Bvbjh7(JyS
z5#{!H|5WNm%eR`l1{-SA@>qTlo62{~
zEEv~crQcY39M_%rRl})sF)6d^*StHz&<-re_XMXsFQ(?+Q}0^tk0}3b*OVt6==Y}E
z4Dm&yy---yQF|Won(f)R59nvo@oVxP=ts$$T_QUX`O124QgwT6_UU)>Nat8~-ad@>
zF31-(voJ307kZ0*k-zg=an>fZ>t2(7!~yBz0nv;aL`#m)2Sv#DJk;UKM*G#70duH+
zh_9TqYZ9(olNNE0(OwsS;_L#n7s7A0Gl25bX0ad@59LD3QR;lR$W}6IL^`)AdZrs<
zWx<8jqR6+TFS|*db3@`9cZHDe&uMu2BicJH9l{fc_Po=Zn<%-ZF7xk=BCbAQl1m|vv4j#zUj9)j#MA`XGuHGMEY#8(1CtL
z&!EP9s-H@?KQX2H-fYR?hN+0b-F@LFalON1V38!!U+yZm3!z-Yu{Ozri2Xs!7Mmhp
zkAvnzAnKPp=*1$x&G3b}AkwYo4sXkmUf`m1vlB6B^}FyWF|$DSK|QY5=H^xLqTa0K
z_oK#8ZkqLQ$wkB;E;)xA5Sc$tHZMoHx<&=AKBN^c1`S0bN}dRvwhhraeU)Am##wIW
zQ17pK;5I+n5$y*LP7~?J{GaYJzwrtCqpk6?tq97E@D4FVkU!$OfrmP8i+;u}nS^%I
zWd-b`Fwd79DlAT-+!`4taY;nCZyqA^(B8wmn_p^CK2?6$=rf|L`eKP!h`x7h?)#zK
zgO8p()Z8mQPp&dYdra;8rZ&`zi8oD~i*&oOpZ7AP3vWL4@$2*ykI)ghjz^jHAlNK>2jP>Nb}F^eL&rpqgk1EsC$gS#Ut+2
z9Q9j9(jpKWYUop`^MH9jlasplNokiUQ}^WP2EKjN{keF1NdQ$(-al`sQF)!i?6pco
zdZX_OZ))DRK3mvffqX3nKtG~rlT?oo`kAwRckVQ#cSKk}2}im6ibE|Q(T~|0!mU)@
zcnNMlUyXcGnGf#Wh$DQh*Qt9((5-z-(oug>;(h;6#Lvg~PUAtoTQBbM6&;hEV>!KmAS*
z;))(|Y3d&JQRJXQD`Fq76g?Di3ODE0y@-olXmd;U25wo>R5~=g*+6Hnc266YG)0imlbfbOOTErI-Oj}-D?~k0umwYO#V&Qb9~eoneTydLu7s(LO9qkQM%OVYLYK8}4adD2lXYE)KD2-o?K(p0JWQx_*B
zSA)oC9b2r1`WM}lf~oVju+>eAnuj~iEg9+29G-4}NFVJ;)JK#}!M-h5
zJR4Pka;?++Lh~`c)1}Y+pyvIS8;`C~=j*%eWACZ^kI}*vZ9Cu5q0vn%;KBeM0=n&Z
zR*lhN_-aqsgDyI-x43x(x6mPSfpNI?TRIHW8t|B!T)J*TgXt-6ZgOs-!Szi?+^;R9fm-&XmNae}w9rmvi4J#w
zqv~V{&YliPHB88;ZSR0K*Sh#p#4gLzTz9)VAVhEN4ML_9o+OGs$#}&8Eqx9v#sLQG
zE8m!tzk~tR_127U3mN$T+qUJ?<}=`XwRYd4PC9rAjENi#p@TU+Q2Fi#I?Q+b;*`?Z
z4dP7m++7jf5Po2lxWtlfC=qc~nZBYM)Ro>{5XRQiOXzGeE1Jj#zVD=Mrae
zH`Le_zix8qhUm))oA=l-VXv0(s>D@HAXfHrFBNBk=aV;Fd-a&$3l{9T5lra5`o$|V
zgbB@hKKglHOjv#5qL=L{CTOyfCmlb>gdJ(T(V6dhK!N?T#7dG0+LL2e3jq_DeN+3F
z2r(gu>r}=2^*wOuSYv6JS~mm^D+NT>bb;AdPKWr#-QW{nnRX_&8{T-H&n>Fz#Q(dc
z-hQOt1y}rv1)sHc!P~Nz7VjgvV70Pey@p9ABg+y&!)FWLS?Enot
z3~e4dyrx0<=!j1B4mzZIw#w&Y{YAMlZP-{x$2zX5*Y2jn*~=?y*ZI@HV$Y{D*FSZ@
z=YZ_NvduJ@T6N1wU%`G`*gzUO=kqozv+SqDH&c%yKdMO8P7hAcHI>R5))`N
z*p?X-@=}ckMQhhAiu~LGsx9m8U5Q8jof>V|^$Zx0E8=dO+6DX9%Pzcc-wn$RMmMH>
z?S>Bq932hyG#KpaH*@u(K~RsYozAxocq};9DI-LKMYn<_rs|=em8JV%TXe#Jd8=@S
zK{q_LS-30Lrw2Y77tYP4^}wD&?Hh%-Z$ke|qSJ921h3-Dc*X1h^UuMN2ba?zX>7*)
zp%FSf4%Tlnis^#q+S(HR@jcMteynt_Jrn%*nJjyJh6#dKnq4kq90fsrchC$v^v|7_
zZzW0t#gNaUVZP1#
z8y=25&@jtSV}D3DfPHU~wqz%8Gq;6|4AY@t&5}kQk_P!VnblS1G)R^0sv3UN33E-|
zy0&6{&M}fwTs_hQJ9Y}aL3Gj
zQ_k;S(nyDUd^dm45$hGakLB&8&#DOsNe?3@
z@Z3*2DzFo{y0q>Mt>^)R?Z!NP6
zz6-QA)LC|d(XOL;s+M%HTVBA`s>wS2#iu1XJlar1NtGsF3FI@{
zyC7kBmF%q19*}jsz4e#|6JEyJG>z@;0!yn$+=|wCPKEG1OO(QML|TA6(b@rp-ntD;
z>}$Qp{t<849dKy5V*T(uI`kPFu}{N&SsTB)Qew`8q1NWsv4|d!I9=^FVATmlnmO&U
zn2$5p4;c3P(qPT=lCl)7gfx&$nW9>Eg;Yp`h)Sl+DP#zhhz1!NOwlByh-63zky08J
zMJ3AK8+%76QSJS%yLg`Gob#Ob{o{S!-yi4v{IS1#t!u4yt!qu!wfl6$)W}3+;(h48%UA``Ue(sK5F2;MjTzrkkMYW7@&++{Qe(%4_$1B*`#r1DJ7!MX2
z|5I+Ry_X73-o
z%jMU3C2(>4yjDB?{C)R0`S0;{aW+ROJmp{g{j2^Ccdy_26aJ!4Sypq?`L*3&_T?ic
zQ*rip_jVz~<>8O?XbTbA9CUMh`=KWgZ`DKZ!ZwHJYRiad8fBQrbzHUB$
zSTOzL+duJTyzeZ2M(HR1x799K<^b^TSr8rDi_w4#T?LMEqw2l#0IWf2d_6NZ14{_yu|zr8>aFt
zU@pvNgN8@I{aZKK;Bl6h-=>NUZz2@l20dnjoJzIP^w(^NwCb1m@SY8pd;In46WJh9
zB@%G{6dPowAKfy%2w-
z%{DmV$_D30GK*-BIAEqyQJk62fyG6W7i>SmftVs+@n<_Zux-;q|DhcmP<2ogFLdNU
z>Bk@1=6FuC!Na8UG!CexD<9MS!G?EVib_B3A;^2PUT2rN5~MV%)2hOXAiILSx*BZ=
za`mRvw0%AVxlz(&V|+M4PBzRA*>Hd$vjpW?TfGVLigBjasjUQgYntyJr&@xH*qj#p
z{1!nv1WkIUR7jAT`!{4gxJ{7sM^`IeKO)G>{@u!%57E9vF{9)@L9U-ysu<8nkoqZe
z2J~JKq|%U}E8{Xj?kW*9{DAr^hv4m{uq=_P3{qJ4`x#VC&(=$PWJUW*!PprW`8vY!ctZ)j>Ni9JGuS+l3)&a2AA)%
zs^h>)@1$=Eg&c6MKd_cu#DOQnZ?Cw#W5ZBn-WJXi%+vYW5i1c6+&;Kga`F-m+^~`=
zY|`RDfPv=h-I5&0xN|mCC4&vEHPT8sPHZqfee%X#tgo2;i8pffI1qnOfqw222NJ5~
zhL@t>`&y-w9&97X?;3ag4heBU*=GG{OOUQA`q8YXY_PKBUAQre4fPQcvG2XvAZcu}
z;A$WnMp9ot7viwtv-K9w&E0rT!OcNsSP!al=S6Y)@`1B1mRfW4|lr?XAXn;r00_VxrPIMS=0lQ5?Eb=mw5EtY8}#$8(RP<|tfS%Sd(*kluB}<#j<&rVP&lDN
z>-)xrMVhk$yhGR!`Z)B$rrkKFwlCg!vyTlE^rKH#=W^hP9#3P4JV6E&O%mIL_S%o0
z?{UJp(^?w{*r4S%aY%YH2P8J0
z*S&p_10?~s2l!Yx*XqK~1mn49%#+M0ze|vd87bPTsRUVS^Um?d9D-B`KcO$B#DT3l
ztM1LkIeWXnr7RrlbDPZI@{G?ohs(rf7V2|g*E7vE&KO4{t*Z4gQ3QGBeM9ZpYJ!Y@
zxPCCLoFF^28^1lkd9(keqQ^TGtc%q=MhnnhVR`S``FP$HLv!?OsrjDJ`aPTv`~Osj
zxur7)%Jc;*-^?Y*p{8WBIINSk-<487lo90hOGh6H&Lc?SeGe=v#5o{qQG8zi1H|`O~?Mc
zv)xJ^=f+fb#dgcS~V^pFd@xZvG0-xjZ>%1m{eH(Z&y{zHGR1k0rAa{p%O+
zw>3OOkTUvVBBeFhUn)+#smrmiLOr>&LO3utR?&kO=awe?)Jezt(9)=Gz2ZkU%o25d
zT!D2nx!zE!;5!?l^r5j3?*j?0YX99h&&ziAwM*Qrx7-K(&@TKyvdbcLe{O#30@~+bQ
zpL$nukHzc%_AeQKAJ2s(KCFNJ{p;eXG55b0PwI_2{!ZP6-z|Te`-eaOK3UWN!HXGhh<=D`-SX)(rw_O*`u!Fl@|{|AdXU`*yv!xdYf&zp<9{Fn
z;osKrGX~oYjKzBe}C)RHOV+^}PLRh}XzV&LPN;nQuF53Cd+1Jd+}V
z^s#B{Ji^i5n>DiBmyp)E72fQR>py+_{k##sxfDp)BJO+7)y_sckNH$P1ySyVo1nTE
z(j5D4ml!-(+Si>%5{Q`sxgq+9Qd2aQixB-Mi5|&Ex%-Yi7hO^BPG9CUQ?z$zrpgC9
z#NLaCqvl~eKHgHQIfm=2teD|;Kl@idYf`AkGw=8oB{LZl-{UZB2CZ*WW|(j|()
zPU46M?oOg7B5p2qbq_{6UscUAv{3Gah?wCA#HJS^M;mdy;(f@hO2p5GQ;qEr6ASzI
z`eWScOIJNUjC2C?hwK@|52oa@$%vEQOzc-cdvz_NvH57HsPSvzJLK1Bil0?R+!818
zumkyWcV|bcAnjZ_!$1J-trxnWnT~Y&inN}1#NzJGk3nc>)`69)1(5z3r!e^gB0Wg0
zC=~go4}C*q(2k7L5dK{41JClA#*%2KxMW)(9p#?0&6pLqE_r=slpW%9`tb*dW#mK`;{jT({a6S?a(eQlo!_BP~wMt
zdo59oXv8!(r9=|-hLwX%r=Xn7mov6o5ht7+{zAoJtA)xMO_aO%V%7}~%DGN|B
zX1}%B1fplq-W~NQzcl@{L@~xiAW>OV0ri##_3?Hg|L(>2W@nMEXfWZ=L|XsMD=RV7
zD-pa$n}=BW;ZkWTqT-16B0rQP6&DWy$~~$QE+0l5*}afniP%v^oaVyyEe4k6?MMfR
zE!a(`)?W%_DkJ}LjfImN+L?F2YO4rkmtXW=7-H@c|Hz5NB`x@l@@CwUC*yj
zeiilh+$xk>-5vfBfkj@U;pq+;FfrAPw_99;2CB!oW?bM$s
z^a??Gr}pNVN{C0pmZ^54{6nLAKD*Gav=8wNkUzah?ecx(zqgrGMdcw*|IH%`1$DkR
zQ+?1pFs+K}0~3Q~@S>Fp`2;pdm;_>
zIriy07bBi+o>x8}*UuQNUdoHOW&Kf&8Hlv!%#U%H*L6aCa%T~zGk&hjL-`}hdS|J4
zx3{X~(lH)VgPr6VtZ(m{buJ&#zU4z!T`}U7kS;Z9UN(J~uJu6qr1-1}OHlrE-u_~$
zzHFuDs*9q%+??eaO=$m6_@{miq%#@=cT@G~R8}*(3H59$E|)DswAY>`tdD$}$VkQ_
zl)q#WXKILiZutyyDcXI|LpiCz_Ca;r|LEP%_SF7c3!)_lQSQ3a(LgG1gUcN+Q}eX+-Qk_H
zQLiwl<@+Ipo#9#yN=Mspo2F`!CLLAxbnMlQ7
zq-=MyH(~)-@!@{Nh&Qh5bkGiSfxP<*l#`9y=SKDUY~vI<)o&)#yj=ZpeVg@^!8a)H
zdb%{H6!{khf--FouQUb63t=8iUrjgFK>G6(QGO$on^(&7a2D!?cowKlLw<@{Yojq@
zqFR-&1fE0EtTMcl7v&2sPEbag>$vu|xwxJu80SXKiM8C`Q_Im#X`s4j3i5|y=h*&0
zxp=Q{h%M*ipg$A3CvP}`a<)r?dg)vA{2tC>hEs%=IO4YS
z(>9T~{ygj2!UMSODX4jzx))P(>VD0;5QgEwD%>YB^KmIP_g)7z^1eg)Zo5ejy3lX4
zn+(bMBlxbU>uO9!OmRs_e1m?TIdECYAN?qQX(zTBaYAL^b*gSJ%{=y63F+$&-8c83
zy*#Cow``1yz`4HC5ahdDF1@-M?QUzx+Ut&V{-Ahv1LA6~=sP9I-?h6dgoF0$bHe4R
zen_gFU^fBR9Z37c`)H3Qm^$+Y+S?~+vpF37{NlVoh>C|wk;4deznkW&+O!~@+IDif
zFJjG&vrEO1?@rggPTg~(l6n=w$PePSIQ9JBaqev);B*a+*8m+#W$R@U+UM
z>N%P(aFnVy$zw@YE~tO2;B_SRUSmpBIo_c1mA=Kc9sNF>bW?2v?X|8Gb3Bjodpf2Y
zsUdw;=HLL8H*Mjrw{~-}pyGUdV}OJT
zVzcX>Y9m~~y!6JhAw*NZ=H``%gT|Ge)SNx%^Zrl(B1fXGm3rq59GH8GvS+*gbPCm9
zq90z@2Vs2cdR8Y=?+z!DFr@l$Q^*PLY?S+SV7tCJo+C2s{Vr$3%QqD6G~xQ){DL}u
z)DzUHi62EdX{X_`bBKICd3%}>zx_D!RtM!CwkY%TBR%C@#Md~)Nr$4PHXyoZEisA5
zIO|UT+DFZUP~SNxJrI3{q{Id=|0jEFuf50qc-`{BMGWOw{9hSj$nWtn<)iM~>pv4U
zCZL@}g^6zQnCC3_YWpK7XDR0;ITdmL7k{zYXs@{7`lm*eKdLmm_5-4~?gHuOh<>-6
zYeG@(_PYQ+YVO@nCYRWvy-$XPZJnr>kYt@X3+aB#&>(H3bFV)P3_`l*N9ZXzJeS~w
z^kY=NW-fQ=r{+`pjs2@$q8v|fvQQS%es`KKM|R!%e;Wqzo(7sCRNwv%q%h{amuKES##R;GfsDsJyDM
z`W!Qm)(u${M9uqmkLGpRBVUgJFn}o1Cc_j)KQlLO$(KTUYpm0Q<0yAerzbLj9j9HDS?+T?e*F@gZNfLq0ne-#c5Hx=+vgKfINUe61n#Aa%qkFQPlBzVSKQ
z>!pjd!BkM8`inGv+pr$#9dE<3U!dMDf9V@hm{)hL0)8RPlkA)lsd`*b_p9Fe73EvQ
z=(m`N3z?F#)H|w6Y?pfn;vl~aJsNQmFZcRwhzos`Y5@J$RoP?Mg>b|;cN`8$%dUf#=-VKPsda~2{(4QLKs=VD;m&0>p
zB&a%l+C5E*h4Sv#KH4ouY>s8Q@Z)+Up`SdIsSB$vFq^Q@&4@7=s*2E+e^_<%HGhT^
zg%ukRKXr^Q&_ex8U)4zJ{=M1ZYe3Dz4GHrSOwmrB3vQuk~5#?e;l{bMYuAYv;XiG?-v*b^eMK4HVK}Ugus-gLA9*`dyw!1J&Gn?U}qZ7@{4$Dn8r=
z?wX>~+{`Y>w@Aru{MZF_pJz$sh=UHtcq)3jAi-q$HA1c%UZsjZ$bQa%&1PJV3{M6)
zSFOq`)L=kklN001JO;du?9`E(!+>w~hW+!q>EJhURBT@q9qj1`)hbNspzQm}E4`%`
z`kA)*cCo#X>bXQxTB8^0#XQufEb0Yg)z@bwa(bbb|9*J72ovUq8#K`o%dQxm=PvDq
zYS+>iZSK7gaUo^RR%aI28Hg-NUBZGk!#>`Hk}S}9@RDbn2@BSNJ!e)d3q~$|3e1UO
zL9a=$SwSER3=f?PbUDfb6L#8!11C@}lmBGSYbGq^JS}sa$^t9VQ%i+`1)usS4`>Lp
z-~i9j>eiJ^NZ;R5epn0l9aasGYv=*H&)n`w3wj|XsWvm_bT2#)NX{>*>jsuz#?5fsJaRO?p<{kiY3|!67vUlr$L~o6DfX9*=6*!JQ0H4Bywe
z=^F!Re%U9kVqIi4u5D*KGC;a0|J>)v-B5OOrlofm10Gzv;W_y^9ol??e!d!{L7;{6
zUH2C>Xc+lsRKJN17Xvz!3bFpq_^_NgSVzaYF6%aKp+okC#f>Y%XyCH-eaw~jT_ABV
zcc^j=4Oq3smTI+hNH);u2@CFq)vIGBCco@~1Q|Jg2iIQkj7#DiL%V@0gQ+Pr8rbJV
zL_O8QI$XYNe%yyHSk=Db_QfO`?6}on=(CamwzEojJ16(R^_2?qYTSBZo%zVB^v}J(
zWx>_e+(d)ko&g)5KpMm`eO!&cbipm5(QY|m8rT&_N>4VS!-3lJ9WU&=LBy^@B-^|f
z&N|Pt%MWJ4eaoU*`7|at78zbE!gC9neMD&2jUG4r3aFjkl?q!e485!wr#i4z8}K^=7qOAFJK(mOLezIOryi-tl5Q*
z;xw4j-odz_PJ@B$XBQ%a7*KqwRf2V~7h)&r<>u{RfsNC)O7?CR6f}g%t;2fRXKcXj
zLu0@!U287&E;<;u(4U_$r$Kgh+E-T<8c67^Xc{KEVfr1`Y2}kl`04&JVDb?b$UC=Q
z^Y>sv^^8!x9Z|jD@77mhIJFxFm>Z(L4b!1unMMmAiFU5D>g#N2a8;qFZun(47+d@H
ztjGEkTPvfo^cxelZx(*a{jLYT8?QX^`g%9`p3vwqa%I4(4GnAQn767c`3hgZcR_X$
z_Y`Iu4a`sHPhZ*615%o5o{rcTIwjqN{xl|RxG+Cn{CGFeo+?fXcVNI(s}9AOHms+^
zCEupfX%LpyQS_~=3szg(gZmmf$gK9bFXh<_H?>q*hq2ED9xu{Aj(u=fbBE-mAqK2S
zPu`)?LWgRB>p$p-ca?+o7i^}(V%_I0KXHCd&r#mJm(~T=tB
z4wa2R*bR!k8yJh4d*EH-BTf$1p{}yD^uaF-{Qn=ZS#(_o%t}su^@xUT+i@k8^6TYTIaoP7jDKJ~ht)?MXh@xq9(rFZk{*y1ks$
z1H%(Lnx8pz!%Dk-1)2_YaMh`6Q5M8_7iH96RmS*Hb45R%S{83yj7Mb
zu^w8E%iMh?M~6vT-R0RebhyDDjbY+E)704!Y^=_NF5dw6!!|6S9Xo0H+JFTYBXni+
z>w4hovlH>2=dth4RK1H3ph0QSqxSXeE~rs$k%`6lolKHZwa2_H3fX+A8t1UqyP}pH
z>mEoSUZOB#gb6b|Zm!?2$Aa6Zo!dsY^?HJxZ0q`#^Rgd8nfcih_)r
zS=iT0?}x>{;&ef%j!M(;Y&wWo>~+h;bJh<$J
zp2mFSU-7i)3!%Z{$7Pvsdug!sN1H5Riuaer;O6SlE-(^`cz;8Y0rgpp;)}GHaPz2)
zmpb;-j9HZr3o#!JP)^!me$QpbN4~VB!x|$mU2mLc%oRnU%RkZJ^uwA%?A|W88hXw_
tXcZl1KRV<60ONO(-Iu^gU_rd-+b3yQhXzyc0kEY50ya654FoKc(*rpHHIoelWC1j@as=2Ge}oJG000000C=1g
zS9v^F+ZR15QOWC}q)2Hn6;V8Sw2zEYQiw=sK!!w0iU{?dB6(4!C<$fAm?33K8N!z&
zrAd@hN~kE`bAS8y*ZF+!J@=k{*IsMw_0=!T*qdU&f!wBO59ci$wABwPM{VW6TJ(7O
zlg%8I$?EBaY~&!Oe{+O1LLRk&(@`_rV&n}Nr5n+N1D
zgT@(#r^Evo%wFn#+%b?r$xipt#FGqEsvM#v&oIzT9i3fwe}O?Pu<5y18OY!HC6Rxd
z0pFo!icubenj_JZO?aKZ+hnlwJOkND!T%1QVqhMZDBT;vKw0?clKvnDiw8DuOh3v%
z{f3T!vM+--4nj-UV|`@px?a)!44yr&$~W7`;Iptm-DfWj22XG7dS$|aV8NqXR_Yu$
zq&8*+$a2sie;|}QSBwKU?dmP>XL6vtzP4g~ItL2sZJ!$jI9QLs!KWNki+1HkaNaGF+!u~s;LGIbKpGYU26G+g9Gp~Hm`vLixo;oe%
%qC)x#`KjNTfSaw*|FDe=RCG$XL`x6_Yr4ZsXSN>cqii
z&yE#8cMux)TP(YW5<2+@a97q6DkD|$fTvFAx#r2Jp}zD4n078ud
zwO7<0eozU(^(rFt$GUb2-6}$j=j8uk^^njmO$WEie`Z3ba*^`gErj}r-m%baC3M2X>t)U(
zLalDxmpX;l4RXxSIN`qai!E*!)e`#V>W>+d|-UUrz!
zD`X1Qazs2WkDLzIBXoS?TgjQlgu=rgf5+4bp%!cOvdWP^Y`$ElWGU)7;fO>?2M1ao
zH6(K85*n6ZtP{QgdHp%qGCi4t@p#XjtD87T_8jlltVbTUDQvob33adIqZ(n0d>trG
zDU#+uR@g{v;tPYKclLVgI#8E7X*171V!&DCO_jpy6`@$!!GQBOyVf4#dm
zd|;45Re3j2-*x$qHTbB*5nl3`C+e^`aj-Qkn1MpVhkv1z!NvI-H(eCw;8LIQgJtd<
ze&{bPaZcS?4CEsE7XDxifAZ|Eq_ynlV7b18%;{$wsD6H9?7WmvA^U;GWII9|
zVs57EpFtP^P
z`|0~^#8dr=sn|`_t8=cyqGTCDZ^r$Ih_fN|!D=pyb%!dvyf2270{mWDMdm0Dn^+|m;(uCHIuU79ze#WG<|MEy7^!F#l
znq_tPdtkbjH|i#*bB-C2;h@l?A;UD2!GalcB5%3l`-lC{x*uR5q2RZ8tPpXlYEAmF
zk^_Ek0LRehjNKIO7HAL}G|1aI5Bcn}^s!4!F`>>@IU`#5-jqt}E7KZAerrv}59K017tY}-B91X|IP2nA
zjyNe)WO`ZR{F(I>f27^Faqyn`tj(E8=zmr|#~!&6dQsmju{Mv;+1sR?)zCi-qZSO!
ztt7O3`^S{JKwK;N^6jXji}s%M1NQGg<)4k~LK&QXR@8GE=cY-I}v(Y$RK}oN5qI?S;s+ta
zN?7-mJ@-4N7H}ZkeN%KL=8GaZr-q?;2FE(HKi+j_aAZMcw^BZXO+!0gJVu=^7Cw|+
zqJVz(;*3f>>T-RlMCTIZx36R37Zvo|(mcO&#wXARe^(~j_nH|F?Y%g4*hmWG*v!E2|(i*9Uc_8Z9>;eWLdRZ$%m6w&VD>@rTH#4X@95)Cgj~
z)V>CJqW?HlS>Md;;UJ~Q`Q`$wqu9V{-yQUy?6Ed>{ZB#_kDn6{$lxF?TG6?46$dX5
z+pXH#f5gCgVA<;s)V2ESY|~=&yECsO`or+tc99Ul-6`02xgMwNAqHVl{a26Ka4=e^
zzuLMLpWhqz!wh}ynb%0q${_UrzfFc(aDIPEG&?CdhJG_6dP`Uu~pUA)s(bC7rhr0gz
zU`hLM9xn8Q>YOf@Aj
zSDC+Uk*URVWE94&mtMr&=C>)e@-qV;qfCWEh#R>oduR0++$yY;1Q4F_LB^<;HbIH;=J>rr39;PQ;P%}>zJtx~csTt+?z
z+P%)LO2wR64z3k5&EvL#PV-+;YdF~VExGJ{e+lM5&uItckD$U4T{k#Fq8p`Iqh$x8e5lmX=JYY&!AL@HYd0
ztA_MgoKt-B!~h$dBeOOo*m@K(Si_(2TaNj^c=cvIBVj^o6epgQxDpy({mNAV`*E2}
z1TYKuiT=*xb^k8LoHo6?ye|X&
zp{aCICkyLnJez)W9Os#ewfrGl%nN^(t?g31#6bLAi=|c~;`Z$4lWd%~ehcGte&ln+NUlmL=5;0DYh1?vF^bwV)pgSu49+lHMeuvZm4~c?vKef>av;MLbGGT`
zF5foH!&iHigbrbDov$~qCI1TcrM-2SABN}DUbkLUf_Y73TV$32<|=J1J_b?_{s$&0
zj8T&eJ{XhlCklTRmw7nVSs2GHDGE(el68`GgtA2BMp=eDU2$bCON&%$l#wM{QL;6X
z7E+^9ZiupkaAiypLz@;NQc~QjtFklS_pi_MF6VvE@0|C%=lqKLmMwTX$z$Xx=Bvie
z8D{KjKbP(yGR$Z>tXkV6Jj@XLGS(U+G|V_wQGffr;4puqePP?JbD%Q!!!M0+>bFk}gs*`#P`$dQ0`~Hx#
zg!qy<;AE)eXB*JeeEoh7?8(;NgL^<{_oc;Sh?B_O&;1Ro%&nPp1zT93T&GDt#T58H
z&0{oWY*K%DfbmiHlD99!xa)qMX%A?8huO~k(EocZ^5PM&Kk#UPJ80f>w2Sl)*&PR*
zLCfHYb?4x35H)Bk2I{uoe>H*l_HD0&+^{}&Q`*5A$kU;@L&j+e`K`@gM1bD%a;C-5
zJB-zsbKzfPR_>I8@kfnQeaAr&bzx0a#Q!zHx7~jT+~@YsRz+~cewld))@9L5%;MdM
zv*2e^>Fx<0gW+cS=sLJi+4%1~aDx!<+cfe&Fne$L40-Oy$e+qXo=g@$sSuPFZ)mYa
z9y#lYIAg@!H1YD;GSGF2ju{L4HlWH0Wy7yo*Z9E$#EBgVTvQFV#brxXf&n|a=gddk
zu=9V80vF-O+Ul)20J28}Hy#Id`sz<9A^y9lLVaz-EfWmfY6;n*L9lf<=6^R%B4q*e
zsU7tJr{J$-_cCQJC|TcO_8D>i6s@=T4%x)+m$yS0=XKNg&l)f%=U$@@IMbg$b02K5
zHRszQPUsNxYYqH=ZtnDwgM94uxQ`lmt|xyYWIm|qKmTnS?Q2rM_C~NoGd4K8
zL;E>?o8f2+ngx`^4Z!|6(Bz&9}mpE>XqiHA-`KojOQE3S2~p6xk6r4<naD4k1J
zgI(f?^XxmPmi!9yH^%V6BM*OXomIj3uQ{E6Btcf4>U^^X^AE}C>s(o3paV9$Ro
zd!P2>m6T;OtwUzg3fHxu))%2W0?-@qFPZz}dzi{Tr~iY0R?rR4TIgLSKDN=her>M*
zcQ);-LwUb6cwC;}Px~m@WvHwReyL{_ex*FT-|g2y{xhb!F`fEg(e;0O)R!lRR(+*&
zr1jy4MhN_J)q)o_fQf2#7m~nH=Mc?5V1GQ(8)Sp{VcIX>Edtj(Kh|7H{g|&VWCHqi
zZdADif5&x8ddfkSp=nRbUzSj#N`1RT)Ficl`ls1wz*~Gz`|Bx@GRUX3FS`vxRwx=B
z2&8k-YRYK`Q_8Cz%OQWy-V2Um|3PLPlx_MAY`qZn>>|ef;eR8I)`M|ThAR%eoI~{u
zH^_fyj!rB@Jms@<&+LUPYT>b`4Dz=3Tt#)r5%nYO2Jk<6;P};#piR>O#t67|z>yhC
z^_X-uk?QZRcIYD+@WiFqJTY)phPK5F;+TcsWJhDaeK_5B=sSP@daByA0CCxE!K=F=
z7mKMR)4mr@#`{wqS3h=a1bA5?LvB0eHBET4nd&d*r|=r2RIYnpSBC*R!m3C
zeT?`Py@y&$5U;eq@@5QZClO>s{j6#E%90ewONXk4j3940S{7GD{qdCk%BNuX&@MhcG7Th_(*4hu;9DbK>vYdc>hs%0k}lJ|AF3Q2NcYgShT-#cuD;A;H#yM#
zV8<|_d)oJ8mfSP&mQ*m`7y4d__1S}Tzvr!;YX*P+$kUF(#o+V_XPX&{8#w)q;txcO
zk5WC$yQk)+!k-tmq-q}G)J7&m>_fdB(oT+~`?x^X#XTDO3S~hby6@$5UMN+9n|cPeKg);s#U!J1HdKU9nvQ2+PtbTaUS{l-kxR2BTQ<{sh9hkQ+_gtHNL<8DW>
zA?knYVp^+SfC{Gj=g~Sk#p?<=(tB;at8^do|2HRpFb%S(%3Y=&>_ycqW-a2duJ8Vb
z-T{MtqVd#c!jC5OxYE00xv4ZV+AgYU-DJ7I91#&W7p#l3w-
z^bY=_Q(LYJ{S%I7`V{6d=QMxD7QB+$Uf&At_-EzkAlM5!niDU<|LjceZZWL;z$bra
z7VYcR0JHOjV8(PnB-K^q1`|!H$EBTSJ86A-BF=cM0cU$92ExGGFAs959y8iPo?C<4
zg6G<;07WuX_mTgVAN#!MzWo;PGw~f5Vvy>tihNO88@E$EJ8MrI;j&19^pyu
zrFY5Z%#7obbU#jxxA}md8mzai01tnreCVQgmhoT7g{4>r{gxl!Kj6EJ&6=yiA#1dV
z^wIhKxY|XF&WE;>Z%YE0&^oPShw(Lw5+7-3M;zu!{VZ>aoIfg7jykbuNGRLQY%E
z?>R_)MR?ZeA?7##@NjQE`hRe$bP=F;Qc-QZYZ?66gzVf$Z
z>a%)jUnHP^p_=pRCYY}Cw%G^@{{I6Cg#1O53_ci>12YqU^h%92$|XhCw5KrlS`t~3
ztzoiHiU~=iB3rVQHQAGENhpMjrB?`*P-F{Zsr1Ip?V=du*84m2*Z1>0=ULA3oadZW
z^+sP2&RJSYv*r?;dzL01QW=oWF-tQSUo!3a$)eq3xwZceSzF!nTzwOAPgR*AG|>J
z$vdEbyzv%f?)3KlgUB};sp0qnOzYvObVXb~%e~?;%Pg8AulEQY`5u>?ZmGe#Y??nm
zaYR13)Ak#@VPA4hn?)h}o~Li72PTCCNwc939+a)gMtlv^$$(tweY*|)rr~E(IYhBT
zoW;;j=kFuF?_BxE(j^w{sg#;R5M;Z3!o1Rd$nV-JUeyXafrar@HTXG}>%}p^Al|aV
zBJfsb;Li@k6)JdED~CL5oLdx)AlKScCp$5|RkHciVZ>2n_os;>Zt$ud&t7n1kxj|M
zdddUUH8vu?Sv02>fG+DZAR`45fNcnZPq@`o*&=3wzI
zA>T5vizl&|1LIBBD^afCjyYq8X6T21CPZrnF^{j>#_z_U-;@^D<%a#a;5nCEgFMmC
zZOlJ|B0~I?`N%_63#yaBIHNwigv6J8bs|~?lq%;ay$*kcVCyHI7|)3cQg{xzyV>^0
zBIf-dLSDZUdbMn&$Rm)YZZF#FK(xll=%A5q}-)ijQ0mF|7y(9Jhnfq$10A!AyhwU=N?*0nvJp8EE
z%3j|vZ$sU84yR$K%I+)eg`RnTO6SNu$Y%j4}!#DFm+nxRcV#u>8&iT9DSe*GEHlG0$E
z1U?)#aQ_HiE9dIj2Rk!eGig(hCtmCdFBn;RF)kb9Z4?QgBJh3Y+>r`@@SB}S7}@_3
z3P&p)cDgYeY(If*r_=MB!9rC9kN<$V&0JQtuoDS(P%;2bL`-{#|Dy1ViV4zhm5L2r
zNPqPdta*cVO{na;Xa~8;>5`r%P3
z{Xk^vdhnDp)!+l@a;{68(p?U9F|OT=uh-5O2e&;x8y$maHl)K&Vez`
zBLO?BWO}m?WNP+FzdFP%JHxA-2zzh(oi_~7*x2&e2KX(1OjI-!fE-f>71u$%SMp!1
z$^OSM#+G20lfXRo75;Tsbc^IMK6A*RGX-(KBm@>1llSMpAx)&OOYSHVZNvCF-VE^;
zu!GeUaS{50ET6AQVEfCjd#*sQ^sPPrG~}J4{PIrle{2w%z6adpbG(JTGreB?R9?uI
zHZQsSz$>?ZhyM}9dS#h{L1f>RrJSo@fk)-7+0RJ5R;o7IfC{>TxzXfa2S)R2lKXxs
zU9gbc=i+x=;*g!3y~o7Jy`^b2Z-cy?QdzeVwDCJoN8W2y=A%pgpaK5@x-uB>{vX~I
zj2lS!#QKuD`6VGP7-TjVbCEn!S`F%bpueLE`zgqOnL2Gj&U<=w50&(lK(>d$K~g7t
zTPK#F&#}pAZ~#qm<{N2XR(_7-V~{a0(?;&~ll{HjPe9=jshPE?N4|8yC{m}RH@@(Y
z^KbIhh$VgOBp*K|5_*jf4E_7$9sQx*LF&$kUQ2Taosv1HPlBfuYd?^8)x3Vx!VsLS
zTe4VxCiPvp_?|FWJT~|T;sw>E4dsFljDuJb&`-PM36lC1*KQb11#g8Vy}1H*&1Wu?
z_1t^?*9LO#*9P6yGjN`Ew~rNnM}55amx;Xq{rl|IE^*Q?{JU;4AY1!<44eZ~yb|;7
zf#y2e3R^I~z>1Gh2Um@(Ec3{D1=fwYW1TO5exX+PBi>!34jyCFA5TiiALPF@E5&5U
z!%js1ng9p<3Tf^kWd3t0O77f{>4RUGq<`$5?4-t%_wGXAGgokIylp?}R~E0hd7mPF
zg`ak;324h#aAFqokExjPJqh`MxnuHG*i-GUbG`ew{$wObo?lWCe_A+^u->R!g>bTN5SZne(+Jd-*!oyYmbJ6
zGbiNe6YcZlJZp<$p5;MK|6FcL`nd-EUMV?8t@Tf>N!_SS>YgJ1=crKa9P*BwtU8uV
z`d?Y3*6a6>d(ZB_>j!QyYl(9LDFt_bG)O&Ejtxk~z(2s=qKv!`7EJq3N}clCqc8663vK9_=h4F|D5VlrlnQ#cPwP{51tkegj^oL@nl
z`p${jWZ2Q{&Q;iB|EdJ0OJt#!cg#p8`>w!neM`>cE8R_z^v_?{JW+0d-8^@H@@-Vx{8qI!F@2mdLli3S@0{^O@``?SVQKS+%FrA
z^>c1tL){EEE?na_g+7}(y@%XOqp-3&gUB=7kygq^yuCeTAD@tU3-9kB{~_7uRa4~s
zFQHJPTFE==Ftl3^>q}ePZ88skeMqC};ZTy-P(kz?WC4bdq$7B-%sFKS+$l_@k#nx$
zoR}f|>1-S#cOLAwVzhq-cTDNsCI9H-NlHmZL#Q@QOZ@gp&~^kr3p!8xM-9N
zS4Bd&>1R$-Lc`hn>~r?65QT={``7!~pS|CAy`Qz7^SqN;ziDg_
zP;ZDDq7jsPB1oo>M&~?Uu{BL&c4_(De>-X1uJD+Df3$;!{hFY4zsSCWj?&-yT4>aD
z?@H=!rtzt*aK_TNG&uUQOGO(EWuL#q7X3m9GUVu+Vj*noe220)A;ORBikvnegi);0
zEA0^>Uh3J07Ryu6if=n^HkHDeE-8z2bqbYz7p`Q_qA>f|tY6kv6n?ipsA_VdFq)_?
zXTFg_f9%Vx(!8@23j23n-4;V(zif`wB5w+H4*r7|Y$>Q;29vd;UQ_&$Y#c8mAE%d!;A)RgMZ4GG~l`AM%YxnG4@^xb%o5Y5V|m0pWQ*fSV>
zq2>dPIr94B6AEawOyrqq@1fx|?U4SW02(G`e|(E#FB(_hjEv}b(6D?ZC1>G6Bd{;2
z)82}Pd+?~kcSRbESH&h?D}`GDFO1KnQ~3GXa*i9vK&<*gVzre)vGdM{k$Mb<@6FCm
zlV%XBWEwa8TZH8Wku%oIGw@z!!Mm-_fcf~MwwlKvbL&OZape20?#>%phD9iQ@lTKa
ze>D*reN@xj=ZO%OZX~|Nmmqxl%aH|*VkE9G+?<>&#@;Er%s(9v7%O_&9qQ}IcZTxY
z($=u3nldulL`k4|a--L&Y7)85L)
zN;g;>sN3nYyqU#f#pU~DTUqSmPwnZt$-?>UcY_JuELyuw56Ee8h}8cUbh?p+pXm{~
z+DaDRpIE8j8H>O~qceWvI3(O$bLkm4OjMFD$_nG)k#NFhQxb>z`q_5*860B#f7R4G
zGC6$Du#5^P&qf1I!FhW)_(-|d4ySVXvGa?}?NAOY!{l!0Y~--oExpA1FAhT`3*XZz
z}khe>t4)>(*N@;IQUjfzO~3hmmk?g|v1SPfPVY2zGMUjkfT)}|{IBVgsW(^)oRIPrsIEVCKte>w12D-8xq
zr6>Pq$cpg%rp6IvLk78RA5Nb8!k~B8<;>Dn8b@P#W#`q1Fd93z_Z!ihpX(0!CrS+X
zHMt#^N<^4Y^6G2O8g%W$IXUYoe04mlFChJxqBdeP_Cbil37M;ur&E|Z
zE48FQhr(U|nuQlhzb%N(^LT0}LWiE-{v{3!oPPH4rU{5HXX?mr+Dh_~wcf*>#Ip{m
z+ZMHjLh{OEiIGwi3~Q!Ow(l0AfEVW&tw^C!y}{#K0tKfk)$K;0fANp8Rl*u`5iH#v
zY`TKMDuUJ}S1(qwt5Xg8g$^h$cnP
z9lHqkoI+EQ|6V~sFexx^&H#nB_EYOBKhcOB$Lfdw6yb-r&)8L-7#F-dg62C*aP4HG
zREi^u7~RKaMiYqMe=K!}9=)R=b$4&Bra9r&%*{RZ
zcD)7%6Q%mYe^KPzf->P9doKj(-?TjPduc!
zjK;v)+*|8rQ@C_%)`uE0huQ{JtBlJj3u6OMhHaXM40io`pps&6hQf$XWS%J+XT*s;)4tU`3{l61<&Qj#fHn5w2g~XSem97-^eET-(#hIo%p}`Kw2W0sqc!iwp{0)$a{#7K*S){-&kD1~Hnx#Qd5``jQ$66J6Ow
zf9_pf`FscA?Di!Yby9?vdCltQKJ_thIeTi0#XI6#6s`@^WW~ASz1E`?;*Fo!{a8og
zo@K8bfnW>meD_>m=>^YTs-Jw@A>FNt?2{wAb6!OTX4{GOk$r#qAR
zEY;C!L*`;@?z6FC!k-wy>IXM^88kO4e?Kq1FG8K7qD)vhjq9TGxk^nGUMKxN`z
zv){SUb;NIQ$@`-ne+sc@tIp}{018{OT*C%bX)JkEHCQ`A1g&W2!A=7Pn18-DD)EkE%MBcccaky#`As!;qRm(Nt}nQ2%a@}i}w=lrT+BHPVna7tLxsHm`CPP
zSXlDT7ztEe+`8ZDieb_lXeGQO!u#N+PLDVmKb{K4A_yAdvgGJxw#)uR$zg`X$
zRjo@D<8sJSr8or!Mt|pw)E5)(JAOa7>I{W}u)}CA>g!mm&vM(%S-FW;988kfBttzjG(lB
z)_FDqmBz&P^`uVp=;Qf=L}v#}Z`mY~x@4Pr#62hL!iuO>rnwaMOiUf`v|fy3Q;TaC
z_erojZf3WqDu+7%t3E@d-#I~0cAPVZzM608l}lJmlv};&Mx7Ww{Hzm}QyE;Ec{!Yy
zM5A%SsQTVe3YT+?lNrL_f0t!ZT8=6d)TBIK35dSus`FE$elUox(4~sXh%dW&D6RN6
zi=v7dK3&H-JXF(fXB|0IOTF00&1IqHa{ee%#mIQ$+a22=g4`WHmA%$9s`H;s_9c2t
zuzBon{!xf7JGpNwi4Qf?u&AycpmADZ#QktR1J?vWv4rqTxzZ+bfAcREN8^lE9VC5f
zRdZyi$y^Rn`#e5}i6yA+n%rM#Ack0@QQpK|gm>fLy*xqYA-nb(uY%}ImKRi|*d@fI
zT0>0*!vD6M{^oen&jWB%C?IjvpGG-4R**Tp-|tb88i(LBzGHGEZ^8q{{n|rhUM;=i
zjRO+Ax{#?^ag0HUe_(^(xC$B-4;EPF_*1Z6ZCLv!SRD+j}iit2^LAO7q)`C%}h_{#Fur@?Qz0ExS1+nJ;QwL406gMy}5`!XFjE#r5|HH`n<~rtc?t_&=`u
J=lql1N*x#r+O7Zq

delta 12212
zcmV;lFH6w*l>z#d0kEY50yi*|4FoKc(*rpHHj@nmWC1m^as=2Ge}xPH000000C=1g
zS9d(u+Z#VBQAu4hT~fDU6vf4r=o}$4BZ*5wWn^ZjtfIP-CS^rek}@(glD#rAlM)i8
z#7z<^&iOsRzs~FX`JV6RJm-1d>#LQQb}UJgVKIX%E_Ox?ch`+=jWS`_n)h_-i+v0~
z7FSmf)n)jfZlMfSNyi%L~+TeqC>nJ#qq1^{SUiREc58wRvB0|zW%1u=0S15J@b0xOYw_k
zTIcu!C>Gr4eA*_E;<7`|-{Q|utXN@vMff7cYRTVL*4&^te;CuHS^vU&p8OQZc}y|K
z;i^S@vnj6iyE5Md&*>4JvF@uBZQK@zkgkkqi;{~1UE;`xEsS$=XHAqbQl)QZFzJ+o?+AE
zXXydr43~0ne`N{sF>Iq!Y4l+^!&`S(m(DF^SX#bq{22$sLMksDhi55%>-2MN!YIY*
z8sAw|8^z-_NtdlN;WL+A+D$%0ZV1Y451M2+c-o`L^aaCyxIa3(kzu1P(tbZ{88+Cj
z?d8zG@ZldF9Ism#Zk!Zf64lPIRGCp{F?_D7;~FzAe`7diT@R5;V7TAhsn^k#;XAIK
zTW0@886(mu<<@*;40f|CHemnW@GD$@wT5zTGKr$k1&tZN!;7}CPhq|Bkz5yO!8O(`#mzv@dE76QI$y6
zeTImVTK+kH*mvQhhur*Q=u67PQJ*|SR(PFI{0a5)FSMSNgnj*C32jP6ux?PRY=QkZ
ze-@f9bW|bYu6R5;pd67;1saCa@O#BDf#<)V57k%9-Lq~W646Fvcb!5cNJ_4~|0p6K
zEqSkM!yLcdk06ii5IOmh``XbkM7}MQNHv*4f2@ALF$Dd7NOS^qpx?%a&U6&}!u-%8
zqHF{AHuXCnt`2kl`L*!kdYDtg%qe+ggqbA4!ovx`m4vj*IgX2<-&^eo_?3
z6om8Rv{l3RFswHoVw#%B@U}SDLpz!nu6Ld5RjOmStxeLP_7=lmRXw*ySTL+MT9A}4
z%CHFcUcMjyQC$7rN_}T1#r>)&%dfUjyuu`Vu?RdbKWoG`0t;1s=}Ck3A0yp;#~O<3G5F;tOkZ4Q_HXd}Ua>ev>o9
zqxN@QwgPWf^jRe4Zbl^dwa~4m!-zD0HWT{{=gI6zF||mcINA6^v5Y^(mYbg#69;&%
zc`^Ep55?-OF}A(wutp63K0`rAe=P5$v>u21Xb6a%f5ou&_*-qejfeiFf2v$MDE8&5
zS-5tA;=IYXRprMhuJRlB!Eu7(w~BZu&t$xOgs==)|1;0gb`&SXF6&#l?KThE}*!Cgviyik7u|6FSc
zy#Huxr!U_l(8c==)^9`@f9Cq-%Kbfs;q1DEVRKQ?g}EK_BY@9KNgY33k`VduS*mJN
z4fKC>sj>&~=6d&PeI&;4J(tEbz55ieTDJPqBWH@EC;cOx-6>`gKI^CRD3+*ro-nfw
z{ADPBQP4SUC&{O|iir4*wI31!JRjNE<#4S45!VBmQ_Aq(XH&Tcf6akUQ*8qhQ-C{5
z>#%DrbufqC(1dSLck1{-#Yi~ER{0KENmi*vFC^rrg@Rh|TV%go14WC1?;SF_THQ-H%r2b&V;
zlVIt6H&gH*u9sNUe|bOffO;xqE=MH(faj?eCq!;(=*L%QBeHbAh@BkhhhEg$3BfW%
z$`5`@stJUCl`Tkg0FQ#LT**h6pKsZ3&vu@r`1-5-zVqOlTa&Iy#FbP0BX#WcPVk|o
z72uqbf?u)4P`BmhB1tQLL~KKgPK1Ll$hI$Pm+c0=H!IcmL)|xr;<$K%
zDUMCyq~0qSe?HYa%EvRz@X}*VJ|Aomsp8TMUk-ZN>R9zb7V7?aVR*_J^t`-2bW$4X
zzJ0W|b8#-ii+dmPZi9T0|A%ekL@eOFJLA(+JBkmjE$fxep?Lem-*38rr-Izx8HJLF
z=)Soq8wgkMBZqIgcb!%)PPCHe-U~!l;pdPnk0;pRcjQ
z>^Z~B>|>%|i2U
z(7%jNLaK6*w_d7yZZHCVWUp%#a*PJP%koAwf4ryoO?jSZF66)6qYrqCx)?rQGg3PT
zbC*!=@i~x)h+&|K|8d~EU1a%q9PnS)-Po%D^q@NL_}7p%koUAMjY&feTVt)KBn-Jq
z?p>=`HJr0Xa?WhyP0)8AgXFStie2{Jm-L3dar`TO_b|+9s>DKHy9p8d*n-M4z?XN1
ze-Wa^{)n78l5@v;J0f{cP5LdW7@j?=F0Lla@RPD*E_J08-&hv2?*-_&NmBZa+kofG
zmV;Rp$*`7S$I@jCzqN{thyXl1@DvE&{1ouks%K#Uc+QQBEVKY$b{h(^@Hq#5e^abO
z2XNnGyKr`OH{evY`LC!y8Fu)ZSp1<7fB5J6)mjPW(w52B^A`G-6?6CW5b)kFq;gv*
z%v-K
zZlyPK3gDNzJL4kP+@bjQ`!VN(0B=gPbJ`j3HWTdK*93jCRyn_D6mUE9XJXSZsl
zH^=X}IiM#OE_X$PuH^ZLZ}C1?~MU=BXA!XVDVC7)u|g&#hfA^$(#u}^(3HzGAsKVB6&g8x(wI7-4ilF*NU
zMwriTxYqsXYeTSC3e<@KCN2Ivb=#qEs?Q@RSL7)8NewW||ov(|P{Q9QSuU@ZO7ZkEeHjGzS-EZEKu`wch&%3SdXTfV
z_VswRK^{&RlIHS;+$y9lCNy#f=FsQb%Kd*UWn31EDuNFOv#|Uak
zOWih`$7ndLR?{KOV~7m=Zi*7(F<1|4?|c>HF+R?Je|pPCNk^X63>wZ_zEM$Cb
z8tipX4BCo;nl0rY#u4BC^~XaVSRczt?U3cj)2z8u&Set$tj*uCz+HeCzMczi5E}k5h5Qv}UoGDtPf66`xLoAPVDb|SK}qqtm$t~W$a*}+
z7;){#-@lOuJ!EyvnAkUt8apfxc28FsKYD~X5hKA;Rp7grtI`$Vp`Gos<|6LV3r+%m
z8L(qI@6zlC^F{=%PJ%k!wQ)-kzw<<)zBb~r1%sU}AzRf6Hu+=z`7sh_=RqIeS{oP#
z`;~U@&x(LbwXJ4-h*JWWI+)SH3Y|-T=PY3E=M>m`CiZGufu*7$gQC!1_OM~oeoowB
zIN5^cfh95h7#|2WDOQF2-}Q43p906%6}aMN;7(bR3-O(J{bm=o5sIeq3a6D-~?u=*kN
z1}c-^DF05E4)=55KPT57jsTw(J{Z}Bbt)=N^rQQ7q*Ap|9)5m-dhs!I-cH9lnS=7q
z0pfeW*0qjj*15kR>Yv7#Cn)s-lqK
zgFKg?F-;eP<)*(JX+IVeEsfmu^0$#eEIs8xXoH0KGB)l6e5WM@!lJ%q!TZ9J;l=2Kw#e-(J(bes8XS{VMGJ
z9m;xTz^KLiUfRb6ZHCILsE*V!3%|oIdw7%mT3Vl|s`^y$4cyeQpD_Y5`<)!4!OFyI
z33UD*Xoo$Q1CuUA=Zb+F)3q(85y$9QPF^JTo5^m!OW%=y+N{mWPh1T!HQdxg6n0WSLKf40r*E@A<%xLfY|NttAS(}RXwtpaQn1`O
z3Kkzcp-203&M_fakn)`mmdXeBYt+j+(YvfD>JbjsUNOzqr+xc;FOu%-f2zt$=(}Wh
zS@(0mkipP3Gl(N?l#;Lu{^Qkt!}NXzS1YP72d83x+GD)HoEv%e^e$UA75MtXuQRAT
zZvlAT$Ru+ZQ6#WHs-usjd3xJCq`qK14|H(+x^f}(m-ccEmBI8%
z;j3+Q{u~V?7J`5CCHU6J*EH4fp6a~s?8Gbd?uRLd1k*c|RmZzP_o`=3UV{Vdf7&rj
z=$-aIovHW+%$E+~`#|3U-f1hy@E~0ff_(PFX}eYahfUE4pXZT?E7&j>sjD4fVqVpb=tS^)c`&Ip&!}{Qp7vq2W
zBEIF@21PC8Gp#M?q<(CSrCvyNDj@cMOrs?zS*Bdcpm?ht`)E8xZ)Ws3^*#0lZZu!E
zo$am!=nEfSUUZe}Tno!^4AlL+)+K@Zg%1`s)E|8cxgSk^Lf@R_i|BkV-rH?NeQ=*n
zP1!2wtJvF9CovCRJb&62ypY^d+XU`=x~%UI^&zc|3724>Hl6J+hIQ}iab?nfzNQ75
zT_^-EP6bBLxniw1(WLXZkYl!+)`!D7w@nWm?Ud*b2XDVWz@_t;@;dabHMlKA#7+g2
zN>|%Q_P74s=SA;rZ{V+lFVy!KBzvkM-*GLg9dw?Zv?|J!xav^^%8gLn
zu=P~9NdFJc6>b9m?~`g`-AiF#Sm>obN$BqKBFa_7WQo#3mhL6lWeE+1qKG$S
zO+~gOLP+-1wM>IT$XH4x%Op#7MWk(H>sm_4)%!d1*Z1>0=ULA3oadYrrKdiE?9=p=
zI`st>=QLd;q%0tfZJKT@ykOG#i%HL6IyWDMynZa|fEbwKYnoEB!lb{LlWx=om5%ex
ze_V#$zQE?b5RCs?lhV=*a%SmKN)WelsMTW%vet4h>-sX2E+}iUC4pm_o}}ac*
zmaj$i5ht3B8osi`q-#d%7Un=t{W)V}e*wR(XSm}m5og<(>DYrHmua8tB=iQiD|*I2
zchRuBGSGkJzqm95`H{qt(Rr-rmulPg-$~w6HW!A$8|Mr~-XQ;jo;c%tcPp&tW18yw~=9sUk~ar)XtH8@(d_>e|i%#
zUs`jo8uE2Usye&?FLkn&IU(+n#UA14MJ8Q_+jEeCd_{$)>npJ?!@7wl4#>CRwA}_z
z*cV*UWKt22D@W(h0Wdx!NSp33UC%9!fBaSr!sQLHeRaJKCpKj4O)Px#%H|Dyuh@g1F(!wp@F_-gy=^6YI$eR8rlD
z_yU89%T_qgyhDWv((uco9OK~u-^P1+x?vqj`pUO+LEl$Bs_Lkd(R=~Fe^lgC9{z6Q
z4?9P_;YVS}Q^ju~`5(LY&eR~T-SjH|1jMDVwBApFPSt*$hhZO~^WzK2!)UYOjfLLM
zrIIrWoZYyZy&UmX-zfN*pq}0+`e@o955Gz4T_xnV9j6e{Ub5RQ6*YbDBo)hlxHluFVh+a6Rcf$@==@In%QUHh>pd5}`VR&)*ilEGF_+%X8w?LNf$$E1Yf8&{@gI=2uN1899
zxgC7!QdHdueMN$DBf0V|&Qe?`miFUTIwGoQ=B-`3b?wnG0_LPDj7)JLfJj6dWXUnKsUrvsn2
z`5Le#RqsYFcx0!4pAhn_O+V;R}WefBj$-cYuS9W1t
zWAe)9ZOMH)e&K*NC@JJF%Y!&urR3`iL5AW+zN0wL2*-pZWiUmkZ}tTA%@f)>wXiob
zV^6OFBcB@oe@ptDk76lc`tzCmVZ3UgkN_?
zN(p&~f7rXmx@n+O|1(M>^z}QM)eTUO8?{6|MsOb@w2JpU0E^BJvz4Q+O?P}3tj2iL
zL|^JxaK9v_#1wuG*9${S{Y+J?t#U-*Jh+PCtoJ=LPv-
z)+xVg#JzupTP^|iM;W)f#=%`ihs`&@Z(^jhe>M+X9@Q9fHxHa&D
ze?sbiia&_#`)08d744vjv=ys~+`}@3r`8})JCQSP9
z3D036Qr~oqy5AxHNG_}12wMB?t0wQY+|Icv|j%e!f$uY|JPbk)EYJX=Q=pwG6xQ)>@u
z+?jn!C-cm`1IH2$lkPDf+Da*^}za952beawZ2
zml_GZ+Q)I7d*Ip?%@$I328=4YE9qxR>|>|EQ`@ROl6TdpX2?Px9IRfjSOn)wf9F3C
zUKWlGULo_ZPJNRNmKp^yMM%GO%;6{XE23FDlmaG)C3al`J7+T%$$BbZ{k?&lds4rv
zQaaA_$Sw1NpQw*6e~H+0q(5XWw+X|J<==K=9I~m`=fD{-(IX+}FVI{|Q$`BoOU&yL
zO5n1A<>4IYmoHWix{^9sN2};Xf4l;N7A_;yA4hV?N%CJB5t|y9hMkzs6+Slj<Km9agjX`ssykpat
ze|YJn&nd_|%p8&~!(QF?8pnIsTUbkokUanMQqwR3CGs9RlK${HecF!nf5rXs5sML|
zpG+DLkae(0-Pa;@_*;s#8|jOkj(2Ou$v=lbD!CUdZ1z(W#koFL6*<8UIrdodEIH4I
z`O!@|kaH(WOh`YcG5#td=cuvvxfQ7!*-zRoW(MryqJ0J-a|
z%AbCq;Gz1sa{hXH%iSlHq8<0~&@L
z(@L^imG#No3!p$DLl2^O$2eZzKIvd({)UTG-8VMk*>Ge@?XDCSM1fYnQ^U
zp-%kkRuwti5Qjd4QFD^4Q-5{EaZ)ES=Zk~eVK)?dV9*Nk%14#WQCL4~`zqQNvY+3t
zaGH?)n;P3q?xj&!@$G)(8E#1}Vjog$v@gMQ8v*4^E>@+{yRJ7&plc&SM26lq~2X@u^x04G#
zC4X2=>lepMN&HBpQfYEeN=XByiLWMfBtsf*4x*^slA+8LsWb@u#j+q+@$RM?DSp
z{rSnR4KxPn>+Z$xX-xFnCbR63!2g&ceOo4hy@P*$1}{Nq;`S{QdnGVBp!!;OP=ZDS
z$FO`Q3g)rzkIo!Vp|D-fGF_8GRrlq*%&8PiQm6j1wWjc+>0xP|JB9D@nu-={DSsUN
zFGOB&j>6TRtv5F9p%AH%B{$cXLaCFe@3K7wtsBAtfeEqGs__ieqcEq!c;r(J3h9^p
zj(;CVL8sAo?{x(Vwi>F{7yBg$8~e1&pWLrvW^`4oRD#wCDJ3iCNs-%>Eba|3D2DSs4N=C7o2_wC@Io)-vsKH6fyK)q&KW|};ML{;;c
zf!|Vi=5Cq1T8TmM0!zW)nhbt^d{tR4V2}}V#e5`rU*zewrg1=u+*emS9e;01QR}Ck
z<~d7>9qFdBJ3p)cddokM;o@F@
z+s|KRDA6;_{GTq1{)F}KJgitqW4>+fH)OGGf}eIon+z^p+-A*HGNhU)ZAx3gA#2>=
zP#wiW0v
z$g^dyOYoVUJQm38<54*(1HccN>pe1#_C|WFFO%2hPP1
zyXh}&TL*XXh~8*1BYrWD?;U}!uaoTc3h
z9%V(A1}((zyYQs+J%1Z`9PjQhSnb2Z^{$U!pD7P6LRUGhg+sO2V7X!#hkf#WzXoU)
ziL2Xg?%T^E&t_iQcv8o*Sx#YfTUqGutiCjB1c$W91=`9MEcmQE_qS~nFrbG;;TJDa@*n!Z>Dn*EgcK#D$xco~klfRgv9#tw0L7
zf=Bz0>quc4DgRU{jm8@G)1yuu8h`kR%O(-MkNN#l-TVaw?rZH|A1Bbbw4+T>O89X$
zk`EpxoN(?;zkj)0oq6jS#krRsl~f`7(!GwZk&7E(BSJpAjvBLkcM
z_GNiXN&eE`R4fQ&k)N7l#Q1Qy^Vq8K^1l?4XMR*^nnnEjD?7fVCHPxq`R46}dkeyn
zlXfqru>XJS&lvYo7;ZVWs^k-mZ6i6uh@VnOef@@S2xKVoZ4H|3#^U73c)4U}4kG>9
znWm$8tbeo8?|<@+!lbHQ**X>!?6o#@){uMWhhtWoiM~#Ryitheabwzz8DB}AEL^Ow
z++Rc?;&Z`9F(pCU!#i<0LlXSn!H4}!rO-Tgu7AU&IKr>`j#=qXq!=t%dGNNG4Da68
zxm6|-K1{6qb6y??#Wmu_M56Qg$8x?HrSotx?beSXdNAo2At~HXVP|L5XbH*7@zd`v
zB=<>B`*{6By%QA12ha*9#Wb?|o3gJGJ>I?Oc52*p8HA_aMCcMd+vLviUaieznrhAA
zD1YMjRH5Xa<4PWWmyNesO(AuPNbXU1O5v_w(UKg(bJL=~iX2Fs&T%W<4T%4j*&aVK
zUJ_3C&5_FvXYjXsfb(g>x6?<@n=DP?poB+u!cZP-D=Kb$iRDpxOpwz@c=Mv%UKA;o
zg4J8L*GY}W3%@_g3biN{jG1bj&PkA2bAQYw`Xq(*oA+4z|D>UC=68H#EdvYAzBWIL
z#c+a(FrbgabvZqQH-sPS!(QyXXUn5pXcucg#NwIa^W51J85k>#Ii#|H%v{^?UMH7lYZ4X*b)Omtq@wEK|VEEy_~cC6V;@?u$HXY;^|hsKOZ=YLCt
zWA&3ypO{igxO!0C&_a=g-KqAH=06zt&RH&-M0D+uaLV0Ehr+k*Rr~Lf{C~xUqOeX1
zo;gz*ysk-6cWbXh$bA{A)(tLfNhjamA-a`=W7qQc<7+++p`&p*&ri==pTsH)LpHjR0cuYK+erH~r-&#eZ+Re!fAm50h*62#A4
zo*C;wq2=xJ2gQWH#uwNaF9j)LE2{E$5$?tQT%H-{%OgnNvnl=znK$9#Nn7`@P;>X_
zXwa8ode?eu$vr6=g6rD6VhA6e`wT}CUL-{44{^g13`EToM=%oHw+{+@PvR&(dHhs=
z5{=ReW8(z&Qn+eQ9)CO-#Y128(d6tQQm5AkW*qpzp?kx#$Ye6VUiV%oT@^3Gl~7mJ
z7-a@Cc4rONX96l@kaju3TAARkt+H+(&Xf3`lbOZTQB+L)l?
zw30*W%%safErt2^K1$`#-xmLSxxq)@%feWx>r)pBS+w6il$QUS=(%{ZU;9xW
zwHk&koHLIqxmRoX=^RG6Upj&m84kVm?>O*Sic$9hChf8%adKXa^(T5eYga3>_$a|=
z2gPqo$R6sTZCPI4OQTSE(DQH&1OGUme3tM_yToqGhJRliVq;8~9wdD#s7Q33F`euy
zd%Qk}%UFop$M&2zl7ZKLrZmG-3To85|BjP+*z0{$P)u~DAP6c|X_r8$(nLp@@So4>
zsgEW7tTEq1IhVwl@jS}extN9hzJMneGp!eW^7d)JS&wc%2i7&u7yDT>Up}Ez
zoXX&q&wrYLk;OFreQ0BqC8A)x%%t*9awZua(6=Z57YBBa6doe;@aN}J`&RP*BG5Ju
zbs~AoQts7m{K_KmUihAA3wg}EqM~`8?1yTSlOOtWI7AmUJr8b>q1<}uAFef0Tj{m85VP&Ehc+l-mz@IO@u$1K37)XC*0g6V%7GMJp3QMM&??R+)5pe
C&zKni

diff --git a/previews/PR1624/tutorials/out/solution_000180.h5 b/previews/PR1624/tutorials/out/solution_000180.h5
index d0c761172e1e2cfdfed32392c5fcc8a05a5686ab..e183831266be7ceb8cb78fe61729255be0852d12 100644
GIT binary patch
literal 11976
zcmeHrX&{x|*Y_nE$`}nKQ-hKSDWqCg1CffT&}7IgvxE{NREkiUhm;V~L}tyU452}T
zN}O|W%t?uJo^=)XeZTK>zwh(mxu5@s_x<~De&@ILT5GSp_S$RjST^T0cfPsZzKoc^jc>iaL{3E_VOC+anh`?r$`
zzx;&qF99dPP4E-E1c!^YgR>_gjB))Ajr=2&|0P0xf*iM-ii&7w<9^W6%fZ#flModq
zh&eboag^4(I=gw;czQZm9k9_pc)-R(dqP!p1I{@yKkx7P-CPg&THDzDZ3Gj^Lg#}(F$INP{*{k`q@@kU*ZUk*5l@%1{}c-otL`MTNs
zTCWHWPONu_i>H^z!TnwbJ#4JEpcbz3ukrrX-qzv3Z+o7<*i)XB?3e$V?l1Fl5!0zW
zdpS7U5YuGfkL^+8Bh=U!=EU^B?Rd!v`-%K={(tSMYZV7+pY0B?BQS*9w
z*|`1f#@zZn8GdpQJij9u4~OcT)E@`OZ^7U%Mi3LoIo?JHKkF%d__=!co#yEd@qUzfc^GB5^wX?rtA_xzA*FQYy{_*Xf
z_%dXger0|I_n-4iwTp?Bd8_<$
zo$SwC{Pi36F&?!v4uXssbDeRKlOV%(?VOCINIp}|ewK;cK|b-@Q%p#HAnc!n1XEYd>GZ1Q8l#`EFH>~I!H?YY~Tk%*k*jB|Ak3#`s4
zh_9W_hV5B?G4UfT$SyP0c-+PU>0@tigg<4$SY5QV*aH?sX17S1-eZArvx42wH5P>F
zKhx+x&w?J8=g})eS&%n(gZ2>#f}AQ}b#%D^LCVWKJvn874Mol|?_c+`A#SgNrK>PO
z>gTE{N-reHcfR+u8Ws>F%l&$Tvk*Z(*d~*ZKFEevM}x%Ku>^TFTx4wJae@rlI{Dtl
zLj*Z+IqX42I6-$k9nn%kAzG5Y0qHO{JmV@hs!6OLw(>RrBvB8D~Gad=tAmpxgnQ_Jy&)QKxOo6i2KVJ?7;NwJf;i
zR3038fDJwB$?KZ1jxR(#Qr;m$kW1@sjW!-+gUFqhXY!hCI2pmmA@Ya?#})WQ=3yO*
z?R-jGfc0_8tuuOLDI53{g$}IlWkVkC;>R5%K^D(XJ^IOtAa}(!`n*1l_I^ZrDX<~=
z$5MXr5*EaeRhoUaENJLl=6K@<3!XJ9dwfV|LwekB8s}!LACFb9yU!BjX^#Z;w^s?W
z=DqvncbG?0FSWgl3v6J&d|!U+BMaW;j|CNCUM79bNEB^hLEjHTB?$9aawO=cAMU&O
z@l!R1j|g&3N{U?**3A~r+LmJ67k%?RszZ3LcWV*7DtJyDjgKel?q|V{67yCIel`g7
zT(Wb}C&;MjU3Xiso`hZ2J6ob3{+Z?}^|=JOTKkL48*PFdCQo*R;r{8pEf`SdV?*GT
znS6m5&se&>|Ban&ND5_~GY-ajw$qw(qz3C%pM7QY0`Bt%w+Jg6g0z}{*zU9x)>8#B
z+Em7dA9tQ+$l|$7s_wrt!<_{RGbVp`!2QjCE4br+7D0A-?rDg`bC6aT$?LV3ATM2f
zxvmHIE932(*i78NrW?-|KMY_)+G0zinW=bw$MiMSu|Js7*2D{9|H{lTInq!}kcztL
z9)bA;8M&a~_&&h6+@4e!V;w)*w|JEa;@;2M9M3D*(DKH<^5#l5TwkKcWtPc;1QA!$
zE^8Jft5|#&xWa-=l}$yqScfy+O%h#Dzo=~GMhlFiM(ImyF8Y;wFg3s(*H^K+)ZQwG
z4K|h!looejU;DoD{!?cbEV1A`6MB>dyW%R|cmF_c5Fmc^2pe>-g_h1if2&eS`)k-2
zBHE@@UdFt&9m!qw-k2aQcOMG2!#dwp{?ftw9Sc@$^*ttx{XK*6F;%*UOfvm
zOQSz`IAGsx&ORv;#DX(ki7BhHSaAJn$|5z)k94|_3!gGU-U+37kDek(zNHccZ?XTp
z4&O#UgLQZ3p8LH}1{(|lTW{nxW4?Yah;_jI^pV$fki`0^j*oiodJ6Y>dgPSdS?rVIMh`>uObOC6
z%PmPa5BE1};kngXShst3ew7YngJh%Xd`D3>RG6>bQ0jwmu6bGPiuZ-y)SCTLD%3vM
z_y|12JbLvlij5-3bvrU*BE$%?$boCp)M7TAce?yM9`~OiQFBBnlGRA)n;Y;OT^=^%m{vFjn@~%?-pL$nK
z8qe4N=|35NpU-6^KCFNJ{p<3nyzsx5PwI_2@lJKa?}oqK{llMspD&eguJL*Peq1eT
z{Aa^I9L>Mo{W}By&cMGj@c&K*qI0ftEC1|;WwxU~8aVnO!{vEI?c_e-j#?OJJ*f}+
zT4&DQg7S{JO=9nfK8T)WZmo(q`X*;;6T27sytc%OBX4jrdl!p(YhH6Mw?ZuYbUo(~
zqFwMS+v$h{&qU|VM7`1*#fELHUXW^@QY3`-au)hGEm5u-=jXKwu|rzdy%EQ)M3Xff
zasCnJdab7@CpqTqYC+7;Td=?o@#oD(Q+~uKyF_=r1YKMSd)*{=;_EyBYJfeiNcX&+d0KP+wFe_Fg$+)RuWqpCDe7H80+be&e;Q
zhXpb2d-E12E#32dP8z5Emss@)<#_XZDL_<9N&fC(kv+ogs1p2N8Q)
zI}~&gzwU@xRDsA*b_st*_QCKACpkb2yE=2z612az^Y}_3dB*Zn^2xMs-oG3`Hc+v
z=tt>^9*6YV>QU=qI=Z$77|7quJ<3HO}IY
zJIW)cdL;OvJRm0Lu>j(kBArE)-j7qke#tm~DW!~C0OQ*hlu+!6eBm*Z&Gv{nTe%fo
z(f^R2x``}uqn4DxRc{!&H9D`QG_ca>oeO#G9yp_g>1JT*T*_>hDcL
zy%|3^XJsO{))or)MLgzGP}7U?gqANzG_CSwrHM9ou}Jn728Q1
z7v;r>4H3>h{dpJ_a=pSO3%yPL!|rcp}}1
z*maq|gxa53wE@S5ah*-k-kSs5}lz
z27gUM`6b?xCFzLDiPo$mC@&b95_=Qzg8d_mGN)a$7_k8n3B
zr0V~GPs=yTUi8i^J33->!TRCPXuq@B;&U`|t+iDvQ*a+{uUOV@hsZT+Y<3vhl_~zj
zpUZuaYq3;zJ7TcySY|Bh>s?Isqt?wCbJr>j(eZQg^mxSYti9}8SSOrSMH{Jl8QM{0
zOZDx+b9Q>%=vTnMU;h>IRcEa$HXw$RUSDx4Y*MWkii5bx~hSZ-L@;
zu0G&>#AR?8`I(WFBb7M5VaIa4g@``aPgf=5yjEMxB99_pSzhiCgt+?BAcOLA{S*5{
zALgqwqWzR4#!nV_v3fcCz}abl;|j)cMMEx=>QA&}2cHJizj?FXFb@5xI4;OH!nj8I
z%bM>azDl?oHH3D}5A;Pu@!q?c*DOorWk%&BO%mm|;iJp%3+aiUh6
zKC2P+;=)!mu0x(MUA>K}Yi)ajVd|Y{x7L55GWt6%T~jBIesn#VTVpZ)OC?I87AOzT
zC@t|oz2R*hnH-3EHY?WcMO-6jb96KM=?V!5EkeJo8SjptM?0RgS~X5MZl5XlRR=L|
zvtUXo>WQrm6q=8mE%UbW2I}uW61y)DacYtJ<`amTWvO&|9N%|@GhzU_>Zq{cA;gRB
z^f(ig+i=EOQs?BhlF*eih(;m3UC$6DjhBj0_jrhUz+r0r50f_TrTTB>GlS;WXqVfy
z)Wiib>q%yjHKKdii@cSH%6(3)!5Hu9qUel4#AW+U6)xiXyS6o-x`o(b*ILtudipSW#UlH223weP~75525
zMURD_s5*#xcX9DXME!^JZSEnq&TMI*?z2k$K9xndp2vL+x8Go#J{m%yJcviXmn@*(
zOX3o9g{iud;SbWwMSpvHH_xH^rnqA%rxChW5#Q?bn@gkIf38g7a^%x=E%z89uRKwayb|%7?-fxhPI{EB_5(!M
z{Vli3QQke2;!j|l_wN*c{DSiNbNg$kdwy$yu}KBSA@@{2Cj@z(f#*qTUcQ!3)ztlY
zOOAP+nm5?juvQ%HFKhZcQ+>`>BD;AC>NWXF(W!TPefxzm>b%9C=QYbj`H`WjX6l}P
z_wX@Y0?~edrgQ-A$DsrF`+IP{;fxEenuuOG+4XK>2X#cg+wPQ2(qa%i8T3TqYezkvvJEF)f$^>AjJ|A5d>gq*a>>
z`d#ysG|xc(P2x!U613assl@b0{=P2PV-(}5OyWpqVBaO&S`CviE}jn;M_f_9f6>Ps
z>8SrHCHasI)|KXmLbF$hG1UjFx8i&W_Q9X1`$V0WPuCeUI`R+7LX{P3l{pr1WG3sqo
zw-^pY%=OS$rrwh;l+}y#&`%iLCZmvVZXFc}#dxJ=^&O?|Awl^kmZ7-*S(bVmsru@D
zZfvZF_Ko@@@mEm)kl-?Yi6!2=+|P
zo0i@I5>fQhO^F?#{bJG`dDPeSHu_YT)B(H97pv!|bU^8$x}n{t=wMa!=0w{GI$XP*
z-S{q@4%1d$`NFQDL(x0kF@Zrk%)Py|vO<&r_s;f+DrzvmikDDuG+{tt9f#mH2L{}9
zyAyn0c((;O7+m(qK^<%*(x7Y^*|q
ztiV_5HHD@h|?glq*)zq_L$h{T!TA%C&f4d92YD~Jp{nm?T
z>k66RoO~?gojnt-id^w8yUv88!m~_Qd}D$(W8i0ER2Q&je>r!8lL?tEi667PnP7b5
zyoR5{f&FH=CFWyTAJ49bGV*c%=Q(qEtm_1e%IA$qwVjYWoLN+Y^BMewE
z99XIq-vQoZGnR$yQU$bG^^%57R~&uMl5|3E!bXnBn2#t3B;-HWZQ#lxd(JHgDCpEUfd?GL0194%k1c>X-1T6Xx-hSTr5&
z0z&hanw<{=7>5U|66)!2!myk$Nu@#W-Oh(^zO+MQ;G8Q4aWps+G_?PA1p^EV;s-LP
zbV0XFN+HdO3BJZteGIyoz@4xV5@D~xA5;Um`RD@opHSg($tHI3kc8=Yk-_3+VLk+{j
zU7fJwymPl(RR=uYsWdxmHysk(&Yh#T;Xd8Fy26Q%0dfguo)^TJzQEck*{5
zox`}UFL8OPL0#~n`eo@QjZXMBi7nE?X25!`{8eI)X|O@>im<{Y8mx3O`4~fVK>2WM
z;M4a^@bM7(wv6lsMf;qZH1jUF^Rc|SU#%0wR;-@(9_wex@@W@au#f1-7pUsE(BM4d
zSTlj=c*Uu2t3G3$Uac|~4N>lfPgM?GwL?tUK?ZZAKJSEmrAbNOunuk4UHX}g=z6fU
zascDeh`ec&{eTYoou_v`_rkvMa-Ri1*5T|aL({ad&-_&SJWz(?@Ont8gvEf#bMMcf
z&bv&ZUdo!9PH;2bWPN2xC-`5R9`5vn0rG{qeG(#c*rcyecNzPd_1Zgott4qM@LJUN
z&T=|*`bC~}w_?DW>4(Edu#U?mm!_`y%7pwMgW^p=Oel6#S*f_Z3(S3jUY+CZfR|N<
z#PJu4=%jhuYu?q-c{T8-4#h7FNNQTgP;N1<|rd2HB
z@azVuvWG&F!c5q?q(J4wHtg$HBJw`Gq`_WSlY6ge?eNfjVWf#P4U(leo4XF-`H`)7
zTOZc}F*2gcdB-~8nfNQA{>@Cd%ip@f8S5+kd}d3MH50!3Z8%$J)&UAvbo|ojH1J!q
z_~c%8JA52d^dzxfotkwn({Nur1xp^%%V@B=cW~ym;|$;!3fVcB*ae|SMilm+?gmYj
z{4&|;OlU35vz8O+0EetD0jn8wIQZz18NVkDPTRS0Sh~`nA--DMiiZXsiw9Z)7Sfkt+lE
zrip=R`X=6mxCu~vDRXp{s(4`Gd}cTGzU+eZq7%=m<}mIf>F}Mj{LWtlexSXvWoZJa9jO%x3(yppBdMxT@*~9CLhp{@>#tayjT@ZDaSh5sW7bo&Raa
zHx{?Km!;d^2R`vvq`x13+6j)^*&MWSvT^qKd)u+?jk+4Y9B>n3^L4Uuw>S6jIb`!|
zzTzkxpYIN5cMmtu{T`lfHr88E3upP)c>ii|>u~V5J>OsKDbGswYQKj2%eXwm6e`ai
z4o)`2WNG+gd6o+j%Q+b4`0&5&_{nkm@%(cCwzu~9zgy?)R*$$^!k>8(_?;(eTz3zf
zLw~z4w|@_YPcDM*cO+xsP<@m7adG`-4E~}BVjQ{0`s&}!h*IUD=8QOD!h~NA_{aQl
zbB!f{h@*1%$4`8WO)0mFr`iY+i+{qJm~)M_D{Tw
z?>ocEzrB(Fn^US9#-pG__MIJwMU
z$I*)oDur_Ef)27_{hP9r=^kv@k$U2co-Z4Ao)1mpgt9?q&x5Y?L^h~#opGwoVT1J<
zg&A5?IAEFOdoKP58?Kd_YE*ZyVeawPo1ryq;CmK5SF(}~XR_O7n^v&FtVO|Y=sFwX
z^q*+-pJ&66b6xbRU^ZmV+n{|^iXeI9%Kg-Y3DQow=9Ity2ktqYd;6-N1DEzHSh|Q2
zjeNnK78>Q1Q`KCkfJH>%@wU
zhY6B?C8V+}lpvj?+`u7^Alubdwxm@MWYL;EcEi;KDZt&NX7_*~T|Tc0Z7wB9li?kk
z3ZD>Ux_iTi^>qXpBX_tgzlIDe!kGbaqWe;_Yws8+Kw-O`dos1Ijf<#WdlL-
z<;_vn)g;K<&htB3zjB~(@tc9UAsl$gQ)J5L#eu4>Wiezq2Sg{T+mG^MeJ!}loh(R@
zT=O?5E0%CzU-M;mPYn(fTO_yz;(G2%VOeajVS~gYksTAW*>F~M0{429-*VoQ)qaly
z=l%25qp>a}(`noCD>%?!{Yv~>AO~Jcm3FpqbD-i!m*Bn-Hn`|?Y!NuXh9#R=S2i@T
zq0#YPP~<@l?9fPB*MxOk5cNoThcrQIJ-t2h!jl6+ciW!Gt>i#bgdmsrBR1qJ2#U|g
zI-I?;hPDvvQbx36kZu`c=$1*#kLnQa>us7uuL|xH595=Gy8GFnUTofOA;bZR
z&zJ2S^a=9xlB<_Rwf=GUky;z5r
zn%8~C^-5`d9g~6U_wMEs`6_=7WXoF`O-o@z)~LRQI`#)c+S+&#>|dGbCPy2K2y&@z
znp?m%f;_V@|KvU($a9Atmm6apSMHNvZGvd^F`KKdj04TD?eE`O#et$Fn|RDJ*bp!7
zV%lwueL}_Jn{Xx@ZmZ}O+F~6pbu~$JM*Y&#RU0iZj;BhW+H=wGqRL7Bt~kHkH6`{|
zIUG1(S*aw?WJCD3RS#>N*dT7feJ0qC4Q_E|Z+pJ8VUL0T48NmT|JQ>{W}!bwA?>eY
zUkK|ExPJxn#ypz4`mHfR8tpzDWQTQbakS(;w)z|wLw^RJKj~X0uwl}JBbQ=-
zvf;Z_es`Wa2dvB86Q>pvr2Q894@2xLp85S7DV!;?LE=atK^|CnY;F8|4lJ7ywyWY9
z8)QqOKQbM#U$$hQ5)WiUj7MVf>MS{Y0
zd<@-2KZA7_U*TF2%;3P5fcBfYEo`{)b771FuBV@zuET7skDBVoq
z{Ewt>MsPsd;Xd>10fLl0=HKFt{Y`!JlciArK^BIY&w2fb1N1rD_r>(Gp-sxG@&d-g
z?_I27b%_nFWpCX-;d$`YJ?g&PNj5Z<&?8duJQ-N?#hbwR(vODQoy9(>WK?xxlPN*E
zWgWVxn}_FC)S}omnj8q(yYtK30PIIER2LkO;6Syx)`k*qHauI~T;zi1h2Et4{c}{X
zPC9J7g{!bWJo*;LL=oh&9qH#HBnk4C1J8s>Md-)zN?kmzf1gzSQPE)RPZsyyt;%D=
z{*Xyc85kGS!%8sr3kOclyTK)W5BKE&uk2y$$B)8=p7-bwreDIFnL&dlR@k7DE5L0mRss2yh
zt0s))>;Lpm#^2|28HpF`U%!7{K9v{!_wq^IQOED8hw$C-x4VD%`S{As9n^JbA@2fG&*wg?o8qP>EJ{w+)7tHt?xtVU$a)pdP=a&w6!
zjRQFTC~LiD4f08@S-aX0bMh80G(`M%>xHQh;;xY$=l7$2k6%&&Kl+mvOLU7E?S*;O
zFPgeWdSOj}aiK2Kp;6D?ZAZP^=e|7CqsITd`^{9;pC%qtaS!qImiaZ05%Xoti#DU*
z7)|S85sbTZzI@^m)Ytkp+rt_2)D-@GI1T-#`)_lqMtaBW+s|L3yoQ_S(Im9@-tA#P
zqCI})n^+*E(3
z#Km^bx-#fT_ouYPCX|ak6G{Gow7t#r4t>N!&!$>kLo^z23)p~hHm>6#|;c3W!
zYZb>YhV=DH1$j55H`?z0Le&q0BSxxVeAaxm?{*=bI>&pS9Ad;Nzo&Z0uNhHU*@5{z
z75Ltl(wllRVnr^_|1J^8*o^eqPn-kEsQ+=fiq8?mO#x}bZ1gW^+H`FW=KJ_lowj=9
zR{=OzBY(xAK-a5C6Y?JxP<0a~zW#wW@}ozskIEw2OiT>kkj|O+tBoaw)Gc#-|mSP~?uZ#Br0&_K26a
z@+!KZ|3P1M6B(q9+L8y`(at>0v-||o!rw0sQ~A9q_~sbJw~tUwyoLJB70I)55gYZM
zRZKv=$=|uBXCQ5&EgI^B80ef|-;42uTzY2Hiu7K;gPVuYUQ5#RWeZ}hMaw*DKh>3%
zZ6{GaooB1;5b9U%{ifHASbF32B@g7cp4I8yk9zIllb%y`d-QA{gBoXI|D+S#ls~t}
zb6+41UlA&%)@N#i|M6j*=gVj>y?oS5A6dVcs&A#L?T-3Lf3lyxuL|kLxPrzir0@ET
zYStj_I{#_w0<@p?a^(f8ex6)e@GueOW)~0CQgx=raQwLm=lNFg@HKy=eX}=qB%@tV
z*u?e}#QdvEcp?y&kFNeIiSjND+n~#6|2Q&SUZQZa3j7Gg=F)QahdzGkv+
zJcy3MXMDvGUk2EWFi`(gsm7~V#1>lnZD~r*JYz@IaiFQA=tAV*s{TE^+rCovlkZNqqa(h|UqAd2?RU0Ve2hkVl~(zxWL$?EE0%TIAxcaiof(35
zm5M*{a~b!s#ZsB=h=I1F88N7@e<{V6n)ifr53EuV9X}>biAQW_@8#UaI_bV&xRI)t
z!5!tcRNo$owcEsteg%*9>%T-=^Q?8*Mnna3^E#@o`MEcG&P4heXMOipq@$h?E~k*r
zW_y<;AZlAq3vEY#XLrc{%)t5OCse)nL3)>rSpG_s&v`YtFc|ID9bnu#jkLFN#ARw8
zK|R++qmib4=}f2UcuU5%)CY(gZMJ8vM)~@zyQbR_FP3e*K8p6!9DN>8^NZm3naM)`
z_IAOm%rQ?}3wCZjh4j1HH`5AGUiEWtQ6%~~W~$kwjktb}i_2Zq%$>&-?~Eb6(Mi{tcMW3(&ac_37&D6k3n-Z_tz^iY2Z(e5CNwCamP4`LDLNf<}!
zqCUB4q2d%C?7xqA42~cj`Xl-1eUy_s)HW?b^t*AoJPF6O-eMN%hqU^=dk%q!t1b^R
zC_gtpaKid9U-XF1@Yxvux_l3IFLxifI}UJVVjNdBWHYG#^fAf7yOGkr^~^91{b?Rp
zc+CjonlMn>@({5(;X%|8+I^_h7ni_u?`mF)43!tL`x91@$ai56H&W+*|0lk(67;j|
z%%1HbXg9LZEW8oN9hf>A+<}n0NLw9I?g5XLCGzDx%SD0^TO)MSjL_ek#KiV9=*L*_l=1EF
zSkD?hQJ*nROSi(V5|p1Qc(;M7vjJO+E8h?&8=TS+q0Z&j{VN}$ydhYIP3=R^6C!cc
zxw_zUgk%=lpVyYLQ$;l2qa>S%diy$zb!X$e-+!E-@*4AMKlI@k3+YEE?*{~;-?H*E
z(NoZVw-j$fI~AwW%K!r$hgoO3p6at%3Cdm@(a+OKtlf*L@wNth6Gr=hmr{O#i1!lJ
zB&oi#)bhobwWy!MzjJydVy9nG?sD|AS;F3a7|~ZQ`;ja1cMdL_L7m5IK6g#2L;B>r
z`3X}Hb#h!9sCvGyxwlFh@p>iSvJf0MY-WqLA<`dj=Bn4DA0zd-64hug`EE(YK}3lI
zJX%y;7bxkoU!b0I$ch*1kbW{ny@RUjCH4lx)IHB!>)0Y?^cOg{{;3@LS?$i+8iVoQ
zELM`RK>q3Ul43X1`?<|6g9}mFX2rU_h$}>F{5GSXz7zhzh3L06{mseqXh)pYtan7Y
zb%yL09mJT;BFQDFH@rGPbOF)~>DK!~hE5@8yM9ol}^FoO0dZziwB38_KnDq_yR8|Cc
z>_R$2r<^woai!a$4^$n*y}2a65pi490-FlNkJH*3sq^fvexJ%>oKJmU-)V=-
zTB_d%8BI^0iuh_sPxA}%U!@YdR}tUn3z^SFe&{^u0yU&1bS?K7A^jk%ENKk#A;Ib0YCnx(Fbsn!1Tpyz>U`i9p57rh>E<*c{_g$Equ^-tRmTGFEy~5T06|RV8^Hyzpg?fo?
zggO62|r8
z?WG?s$hThnen%SWcO)krw!yk0-xZj>M2x8Qtlf&^CE5plpw6esVk+g-y|-er%WM{+
z{L@cClTq$eKI=7gkK3JkUFsv^kkiV^^(a?47-RMf`e
z+>)POYvfVyh`Pma0AjP7zA|;6Y*1D&%0oX9aEFXSdT0BHa4^O@YkHp_bq?{$J+=(S
z`HNd_+DO$`L!GhlCbTcr{}G>w`XF+gq~e_FoN=!!fe8|xx5{}7nXpSqv|sKX6ZSna
zYgm`dgmCwiyvb=y5RRgk=p`~??ehtD=$p+=rE$0`H53chnzRMqr!u9SaE0R{W1v#6rcSpp{T)tZTy770TTvf
zJmnJE=D>imLwAD?cXmR_g)n};!A@{=Y*Re)trPefC)*i)?Sw7&v!{HTNCTFDeo2r7
z4Ho1+C^A-|K}KSYgcAZl|gD?8ZnNJZ@g+(KK&d|U_Sa8wB?KBYTP!5t%qk$MRElXOm3)Dt;%;c(J
z0-be7YW{vE=ox4!>)0|uzA{lq<}DNM7`6-5Z|?%>q@~LXcXYuq!xtazdYMq6BJE;u
zfeC}7!x0Y-vVc26%|cOz1uctsBKzKUgT;pnGP9SkK-g-gw@3#Iq<7oLIqvEKwGE%^
zYCL)%O=sm77o2aLb1oW3x+2OM}+Q_hBr*}8Y%UHl8&EvW6y#D@yuhTH(1~!Hr;f^R~D>i4E#)t>INyPPqAU#EGTPB
ze4pjTg8etoYiO&p;9O`#aiwfG7;lzcVm`_QcI}N|#x<;u`A$4;>$JQJ>rPHQp7`ns5^wosOv39q;tY)T%}p>9N*jCSdOG3S^W54KPibH?+Gtj)!vLr2h0U|j{@vu{y35PDV0!E-pSK1K
zIN4*bQu2TfFD^JO7rTpfVtAzJ>gP_#3Wg*DWg09GnZNpL79I3bN;Q&wn2YMPP
z3uf~bTQvD~!xxR)%k8`wKtD2Ap74wg{)YDmlN1`zA9Pi{{?rMB0kbj<;%JZ%IJEyx
z83T6b#}8x(bVH4Faskbe1!l&RybZcpurOgwE*IwUf*4ojw>$=z)(aXWjMDIXwE77j
z`dQM$(Q59e!BBDkrmrnbh+1A3Rq4Wl#Y;NZ>~!maC+;&Fcq4n@yq9ZN4X(?+*&?=(
z&4hDxYR!dgXVCAMvF@*^j
z{v5%C*Td}rHE&tq?I!wl8QBAJ_Br*b=G{>8{$5M}@-7fxv1amHte@p-lP|SlAK50C
zud3rrgILD#76SM2itw+iKVqF`mm5o*Q0{@wa)<7QAr|P7L0lE96hx^xE
z{+W&F>{)Vu0OMH_dCMldk`4x4r+3zQ(15FXpM?!d;MoQcy2tV@Z}=?V|F(jeS1z}bEo9VDuqK@{sZtHbe}Irfhv2%ZmK
zJ)k|gY%!O64{()MiOv>d!Hy;QDq-80pq&|!_pX@+dM+jvuV|f6<+>=+WG?pCIovJX
zhiOnOQ`Y({jtP0v63ThUyP#&qOVR$#ET|D`U*UxHb@zNm+eK>@(0n(XeQL%8^-LY#
zR5}eJ*UF#T%jtyvQAKwW>(#wQ=L!wi#a*Phie5?sgWkbu+fFh-Waz}s!NhI|IQm0j
z|LGo(QMpzsGld1sC3)7e!c6eU>K3+|N{8@AkIaPJX>i8w5SOJ34W7i;YFqKq;DG!<
zoBtvjbk7cn^bBOcs6ptN8wM=U`FcKg2up+;T4zC>h
zCO&du0DJMYCa|T$U2k
zc85N$FBi|}6l*RfsOi7y6-#1*mEsystRn+bmY<+M#=MQ%TZT)lpuzPieDx!kKP{1~
etbSY%c7U|zQmjKw`K{(jSTA{}rn9xMPX7a1%P?mE

diff --git a/previews/PR1624/tutorials/out/solution_000180.vtu b/previews/PR1624/tutorials/out/solution_000180.vtu
index 3cf228769f1ccbfc4fe7bc7c88232aa93cc219a7..0a66edf2dc9153ef1fff4e35af6087daf6a913e2 100644
GIT binary patch
delta 12317
zcmZ{~MO2&(6Rn#-a0~7pT!TBoJ-E9=2=4j@cZc9^!5xBju;4C@LvVL*x;fuEXZ?3}
zYf?3=QPs2e+PmrR_tW1ul*7O0O2zoVnmP|J04E^$-)ITP!_(~cVVM93M1S{x%O@9K
z$JZ!Pfrv7lCT7*TRimhjkuK#%nQ>p9NzwyMfS;#WHlMsl$dn@Ov4w5u!6?Kim&}MG
zuG8LJ+TEN8I^e$iJUzj^r6izXbzG(phPdLFx35C4b^0YgsGhTEGzX9V5fyt?oJE1z
z_{Yo-8N}nrYhluG;35)7027%#x|uF)3blE~JJM1u?Q*CdHe}|xKDw@!yOlKdTu&tT
z{5T)bd+4%#@-VP^M}U;$tTp1j;6i0a_BPTR9NMEI4EyQdAH4Q`i{$z%=6CX-9y0&&
zzV7evOh>zgE>oe{E6!wE)dI$jxr5ttR^#3WIOu;1B8;`P02``RrZ>99ba+SgYpL)}
zGFGrxW8zic+el=A;f6pGpWA0l*PZR|e^ECkuk7CS)w^S&WP`(lJw$)66#7yj&@gtFPY50f{s*N{op&HhJ6ruk6nDtf7}FFlJRBUc
zxhh)3ATToxXnTkg4D^kSqu`6XqAzP7`N0^Gaut6Exqm?FicXIC%8A5%ch7eOaWmVy
z`JN9FMw;(ub9m=C{)~=;cJ^|h$n`O(qHvXNj0$pSc364*)Xh22
zj9Tyrthz5c@WFTO27!xRV1ZMkY~u{vcI7{nMyZehYGF_Rvum{fPrF-pX(>z<4=+-7zM$)R(ai}Drh&fboT#2skZx3lr_
z=_q<0pm}xW?uMV8#9wN3tw#Yr>#F^{6p;y*ld!pA2Y`?`xx5xppL80TOasvVW9fq^KyUZDcmR$WEqm#VRf3t{2ktBMeJxWJq
zz(1Pig8MM~-Yg;|zb{Ag&#@-PL-!)mGUWGH9a1q*?GtX211t-x)7wGqi}FEN6I6F*
zUki)uF^jxyWyA3oafIkjL%+?@Nzx5R%Dj#vZ}cA?*4F;H^=TirH6FhFMBf{3jV(k(
z@GmL#%dd^)=4^N*6zD+X={3_?fve1(GC-%<%^_2t2Gb6EDZd#*bDc`=08L)=
z=-U{%eHvQBENAP3WiV$%)JchsR30hm#h&)yn^d%C&X}l%Hm6vGy(}zxsESx~PCtHC
z1yDl;4-`5*gFo%LKP&(H-d_$2FKo5d|XM2)cN$)H8Hsx0A7b?
zO2d)D4b^tV8*HRkflQABP9R(+ZY#Bq=w8m4e^B0Q|tZ#awufF7Qae|
zs;-ric+ro0xr~kG#u(Z6Z0+90|1%Q8MrEp8h(MAMv(Q4^3%6fR%(o}ktB%tl%UVfH
z^c#a6)FiO|(4p*FNYXT4>d$i`X|b5g)lL-1I=TP0sVj%%qVKKu35D4VP)PGSg?aCs
z1vrv&q=wQZ7)X8nr8}0u!e-&J^;JRH_loo6?1#Gp&q*aY3W~e&-E1@-4Y#;bY2!iEHMyB>?mn0#|Mj($l
z)(>m(9bZVNE6$I|aXfwnCW7@G_H$E%e({$aUNG;9+JGI72{LgF_VVT$?miV&#F(OJ
zQ3oNtsmv)$iF{(tW*R)@(%b%=dsP_{N=j>?Q}+8!N!uQhup8)C;I&gT?;2=p_$c%1
zQpJJUe~p_s&M3_uCYL@6&fmuybyX0nl
zrL$HMEw%oedD$cD$-R&!qoODj0?vQ@{Y$(cbeOPr*wLuJ&}wFJxFt5Y>x}=s^`N`D
z=O`QR-tfx5%w?95YPflEF#pbraV?7unN&c1j|k0-Z$$8y_O$pZB+L+oTf8MiH_oen
zPe;{&z_Q{6lEB+I(B}|KFNem0s7(-2#36%{M|}j|`xo&z+GUX&B6QOLvyTxzk!WAq
zsy?7nCEpLJxr;{*tlMs?*d^Hn3S-ilUp6eNKNfu+oL)@kKfhTn*qQF0Jb4uKA=bT9
z-2VulIPpWbj5%zKXY8UuRMt+rb(j7Ztw&x2KtXx`X_8e8xF$bYnRsK>Dpp1&2d&w=
z*G5V7r`CMG54syQH0=8b(WIsD_=d1^L;j{Ps(QEfOxeTZl>T|y^|=725+lqH6o73J
zDJ0aU-6=u%$aW2DJ!&%y`wFG^;Jj_Wcp1@tpfb@d)v_}|H|J#a7%Z+OfnK~&TFZUR
z5H5Fad5aPOjQ5+_dQ7o>^8&T>P;NgM31{;(V0o%o=ky{y6n#6X_?7_)dm&v%o9zn}
zXHOtYF;X~6&mgzu`Yad5$au_0h2^=EjuW&SV3pmotwcD1-9@IsGh5K?P>M!))
z&x=A!AU9y1rfPq_*>PXKeANB=yd@HaO#o91|i8j0V5W;e>v<{(68>B`F2>jP$x
zcCS33(-kwnF%eU0h3uEYo#PSucA<6Gsuy<6t-5#Hzk*~kY43BPlh_V>nBu74_aeI}QR4m_Vs&Ppq`(JW|O#
z(u?eHuu4eR|4>KDcRf*%_*8ZNq{A3@dvXFCePpU}cMM{9^Ts?z#SijAfebqsDq7Bf
z`qzItYmd|N2m$HR%2Y3Z__zh)53^}VJeD4bH>Zbr?FAjMkv!cJBMbx@V9dYx;-2N$
zR858vApFNgjvFs73yHUL$@|Ov+SRN4`#!3P^lnhG4GD7J-g(;6=k~2l@=jk|=msF|
z>~n^^sd-tr7DioEkh7i$din9#d~PBeM-2ZKZfPi`eyKCsGpQft*;XyOH~mM*wON^r
z3)_2RE&TfhEk#_m&*-4?>xudwg
zBuFv#PV`!q(mP{xEwZD(Xm;T0(YGaEH$PFvSm6l!I>?6vDn>iskK#TdQoG|C*q$+?2p
zzb#jBcQjgGKNc8lG@t-P
z5c=OXhXRp-aa4TuMjw&>9RdlzN+)P}v}?|k(#1o!SM$3Jus&Z3F*_@Nft^j-&9YwY
z6b5*7Jm2A9n$$e5w!MRV(=F7_je?20K5!gzE`;J?4~?q-W(T0ex46IfEfe4xh&^wu
z6Fq5d6a8odM19X#LNiZJ;tzf?K=YDGgrepfQ(ApuuZru^XBkVX*;=kX8Rs(kdJCKgU3MqJGHq_Ji|J*
zi%{enZ)aI|t{c7XGLVUC{QHWRPs-fE@MR7TDQt~XxwlS)bIBaU@VKhtmK0*R)VfE}
z0UyUsFq34+aRvA|Q~*qOGqs`FS_C~+QU4%CmILNI+xPOr^E#yDFMF>v_hjEj8G*gn}tXN8RSYjAi%5c&zd*o?oY12xNY_#V7#HV1V
zd_qZ2{}s36VD-wV&t!z}iT3Co1OPpzha!i$w)yV)Ve2AaCrxqV9?0c(g;yAen@P@!
zeaTYeqF>cH@J9k{r)An3Rk*CU&YsC6U<`}YIekL)@`OR0KHoGY|Ar{YMhv#wYr3dH
zbW#g$apOiUb)v_qfN8>PYH?cMe?@C%?b#qX_&ZEr_$tajhQnPNetl8BfG{mZp{ilG
zB;hMQaaBfJsA*G{+qpX~I_@!hz1o%eS!sjlo#3cJbMD}sjOcDNZrr|?A_{yIFm8Bw
ztv}cZ;oo(YYP)8BzXVou8Z>s0bCz8nJDVmCd8zf8-Y~>?O
z`_C{L
z#ZsCB-|yu7jnUC~EzfOgo-)g~z27HO%M@?WBiQ9KXx^mAJhX@^su4U$wzJs8h*)pb
z)tP>Wj=x#wMsyamTRc42DG-QvvbYV>6NF|W9TjUfr5>PxlH8R5}IL;t7bln1?hkJ1=%W;aeqY4@TblulT_(j?mkEct=fL9#^^#FFQ#xEyuIN0
zdWaY+_Vzm$HOMj9xCO(t1HOOPOX@3QZI97!mlW3vzphKCCj^tF?hwyNJaT5Ej)!m>3@s1KL
zoNEIckUqHE}kh;=*fjuB*;AXZyrfvF}eZMgfOZ+Ud
zeELEqq-LX&cO2jC%P(aR;!fhG0ErN)KT8%v4XyxX<7T%7V1JZn@-zE1bV#)d*TEd%
z`#CJ~4wfCjwi=M+58;X*DFAsdyTdQ@kToVxL0u)h`yJGtA08#pj?&pYL-Plw2kg>~
zi1vo)bCJA4NIF3Y@^g>?LDokEH*o)ahejHQXL_+mU2c6RfMQ|fHqIf#?c$=L&`jV*
z9$!PZX|@Zv3=ZNORQwEmVm+wl-cTF=uzQ#59#-_|tY2D!2+CwoaE+m2v3Ii;x=WZY
zcpM707M2NnLz)_HU5+yAX3KP#<*_2&a$3~oNgSuHRCF26zh%C#0-c6Bqdll^yLk0b
z?d3tz&T{`u8^z0jih!B3zt>?8mtwn$zU&|eGh=|y>VDd3zp)T3p|CZO=A9_sv-;z?
zu1opb$`70fXUX|Oj$}-e4w4CBplJo?rPz_+*8iXG%60oZs(~r>p=$P4z0ds~={h*N
z)W61TSxLN;{|T%Zv)lREI%We+n_e@evMYD%w-4em2MB5=Y#OIyD1W-wgrLv1*u&R)dyA|S
z!^wb3AYQ=no4X06S9BisOLP)YlLB~>!g|yh434{T9ND%C`kqj$Gb=Gic*$JhHX0n*
zQS}+%V;;WCmpt-PQ7t${`tDF3Wz2Q>r~qHtm;Q9U{qdgUCxgL)lJO7Crb#WWT|Zjo
zpt%}j9()#ksQwdFSSJ#4;-q4k0Q$HHW>``>g2m6H*3F2;stirWaZYP_=x@*svZTMH
zZU4BAH=nNGjAYH)*Y(ErT6?aij}MisT?J?I1pPY^6e8>^g**oK<=9kpe>Q6pIRu8e
z8h*__bQ22f3)WFY2#9>+u=!Mp87@Og_5s5s=~Dh
zSr%y(j3I0hIVaUK*r;Fr&`Ee>fH&vA>+vM{d@FRj>i^y^QOTp2f8cG39K~tuu(C1!
z8;zNF`1qaa_EgC(&@(8;U}TwsDG6-mwn>a51=U;}T$!s4a?hsRBy@c#?Il^KKj)K!
znzy<}uQ~n43cJ1puM|CNB)E|RNJ7ewK_5UAc&Ch^VjkZ5Du{MmdCv3ip#1;*#Vgq#
zIC@*QkYW8GXOY5J`NajFAiVFI7`dY*LZ2;qyCW_X1Pe{@gCNT5LZ$(&?TcO9IlWWlPy)1JmY(`!S6*s>m;bP%e
zeLg(@l!TRE3o|lK^kHDro~Wes5FCAGl(1P5i7D*N^Q(PPh{R=m?te&?!*?Hm>ff3_G50wC|3oXeF79!^OHto
zG+N^H6!8tgEG79vJb&_)=*lh-iHcM1MKP@M9TZC4qn4|L9TA0Xo}>bGt8Dkd=SWAU
z{S*`4xetP74qu?ae~%*z%y`G2NcW)DbBVve==u(xH8Rq#2%LJg2}Xes_vEZ36h<6;
zOW-z>b~;#NS96kvY^ZV$oee|ZK=AtIJo~eSsmNV?zdjmis-E)aG9?}$N+n+K)P?n!
z>+|Gboy0aZ&IgIjoDp+m@Gd`H47LSscP5<H;lIrz=0A`B@xL@V{}>kJBr#i-+MtQX)H*K#y-Ld-s}MST
zt2Rp6Joek-*zGc7Z~?M@aS?6(gJoY-NY3pXI{E?n=1Ap#tlqdJ#){rw`b)4i@SiT6gG^+MipH`jg8S;E4h%T@RV)4!Rr
ze5bzv2L70G4a!|z@n5K`2S-XX{#D*Tv5Er(&P9OYYkmP=*Y0Fkv$5$^)Nk{Ex1+J_
z54jGei?BPh+~z_{&{w39PRQg?B#bHIy`hi!CD(P5%o{R->b$4cR*5p$MJ-4ga+o&v
zdz2*A$skdot>JxMD*5h4-T2j3b4~9#rXKis*;(dt^!v#5XP^5fTlkgzR`gewKTn0f
zo`?a<7modEmx@qk2Qt+)P4b`L=Zb}{9pg5mQbys#!sr}Yya{#I+hN|8*0tH
zPdfFh@6x0cN||$=ByBOZAeVx6CzkDUhg<`IF_=L;(5BKphO($rt%HN+qn4mKEazoF
ztd8dt^
z$o;9YH^R7cdfB{!l=UhKCTGck;H9TV$-A-^gvte4Q%>N0H5~~Xo`cvo2<~=8ya1&x
zpES|EP6v$B@n-62rOCRFKthjv>t%?3z6j6O({?@eR_%{}-qvQq=}%1GWCb>A-xJr0
zY2OtzYrK(Nqa@jJkg{*Lwlx7r-%J>!GZ*Y7Md!NJ>2&BAHYlcg8W=z=A`Up;KDU5Cei2|0dybf7;AG-)-#=Pv+-pMs#&ehoy&gJ<@I&JTnX`m(S_yeauJ-f!@jbFwnyqx_c^6Ck+osx=
zW!Ag6OIN{eqf*Vv*I-j1;OFszNa6^)<6b0rnnp)R0B%s@AD<>nr1Jv&;GILtQ&b85
z*W``
z@1a}})#mAG|7zH^5rLc1r`xM*o%zl2Bhyr{pUQio`ep94@DU+^<6$W`M9fgiLlOU_
zEU)EzgDga7xRk{6s3*ZPfGs2_BHSqX^HVur<0Q^40qop52yth6E0DV@Iw08d8SeN<
zm0J5*1NCKB4RX%}e>Yah@+`$~6S<_qen;pJKFBAotmwZyC;XU!u1cg6Iu#0&YZIKI
zgt~Znao=UZ_Zt9k+n4Uo^tzaVnwtn0l1KG_8`?RG$U$xDnF-8q#G`;@Ok_;XNAYVq0
z0#VVYSK2I&H4<~AC&$;nIc(;!ubnhfCoK_N|B*Pk;ME3$j@J(O
z;U*zU7I+!w%CMRhlR2@n!hiAY*8V$%d+y0Kz{PLYO_`|u{WQ#&VOL2IuF5EfjK;HJ
z0VVDoK>ly>z!=P~B3gHOWU6L-{DI~4ZynuLw
zA68;Q0s~lpmfqy|j;QORy?xF)M{7-}#%tRK7Z%HLrpdv7iA;^*rXRdq0$d9PBF_>V
zz?VIR;taGtx5ReCrt1U=t~JD@H6+v~^RKqWOhgJwS^g@$p%xuW?>E*ylaPPPFB$(i
z78dX|xlzu*Y?nV_2jOT-Z(e9Z>fxET-Jppb2NU39shc4TTvq0g=n8I=n1s))D6kPp
z?D>=TNbhIb_ckuHEjH_5)#`tY@Bl4_6a)jcRa_Q`X#(H%m8a5|o)hXn6AfeZq#asq
zd6fRNf>6IXWQ_EPCCxoY0Ni$r*S;W
zug_dIhYVw!7Q7qZ5|Eo5*m5@nV!7|=0Zp!XO^GSrfv!CxKS*6|9M<>jN|$O03e(U<
z3ftxKno;=bZG}!+M~b_sq)vQ9f+hd8)TeA*sSfM~)6KU4IFTNlz)#5M{9FpTvo<5t${lV~I@M19q1%pPOAM+0CG%C-|XyI`R|m}BqFBl`!V&lBx=Bz0pv%aa_v
zN@1Pso%o}%{iWx~8^5fh?2iTix-=ZLWi7qyXE?jNKGrNs*SBy*8I+dhPGIZ^1_zRa
zMoWl284B+dSoL{ry<8@6Eg!<%jjKZI`cnU0Cie9|i~TftVtX~wz>v8A&kybHi#Z~<
zX#WHEz+27ULYcAl8k$S0uRu!}vHe&nV%RJ_b*`H5d18-3GgW>B!8?(??9_}*WC6NQ
zzg;MNE<=B01%1up?loPp0AO-RnaSf%ha@9R4zo%pnxFFBu80_!*(WbCK-zHa4ajGQ
z?d%`$Lwtp0=JgfGox4rS5y}&;!<;4a6UvnZ@o`MW!Z~C
z6l6K=%iNJ(8!M+1F<*@TUYC7yn_fk);fE{1N7x$?ukKG%uJT@e-|Igt-J)V0V|>tMh?w?6
z>R#Sj_732eJu%J$iGp}>Mvcmy|F>aIV1?~p*_w)wuZe*OL|3PIpF32&^AdT=
zCBRbAhxH~+s&uN|(8!}LL9pA;ZWiJMAoVhb;57RlK}m-eJ@B>C91r%<)~>S_`)>4w8-E&O6cZi2vF(28Dmc+Sea-E_>K5s5^&*h{BgEl
zo|r@+Vcb+)g_F%wL)rUaYor@=k_x$P&~tZ|5C#dmxE?G-5FAE@uPprU<8F7BVV
zL7yp30_OBg?oKVWtxNVLxy8?(gkMXybx%l!ouJ~$i{{Sdqc}^OsM`M|Cik0lh#dko
z0KLL%9jRH3y?xlS^Vl=ux%IyEF-#zBU1tv-b5Y%%&04KgOoR|4W@EsFsXF^bBh#9VaiZ~5y+=e
z!*9v>yHB(wX|gBvDl*L>-Q4SAn|Z>qmuZ~6O(LO}AM4G=4}-;@B2Al^PYRdJ^v>Xk
zaMN2I3C{P|yz8uzp!R`^p}D$H)`BeZFFm%SYko7%$d~w`Zp&{-`z0sZ|G_ixe=+%g
zw(|da#s3$I!N~9x(ODz>o1Rw0lFVp7^XXU@+1ih$Qfc;6%4d8G`AYOzmUOikjba|g
zaAl0Yv8BJ66!&H4s4j^&-4jyKlV2Z9A?};gD-QL|;M)MO-y)QIjE;$wq^3f=rqZnO
z7U(^>THqoGEDLheerDBhfCrg&4(wPCo>Z+IvrdVu63!k)D0-GTG-Nkl?j4J!jvaF;6
zXA=J*{rb?sEdS^7PY2paNr69uxpnO0OU;Uej(ws@!V}?R)7K}NTI?bS_qEkCf!&KT
zY>$u-Z3?*kHg!F-F2UPzW>iwQoeiT5jU*>#`?sddkn34q#|tZ{mWErg8+Lfc%2b(w
zI}Lq3zppR
zgahhth;=1>XHM<{D*40FozqxPz?|OLQAiV|?!4&&D>1cjXr7w3%ouLBsk$DHz|Bru
z#3to@w2#ej+2(Y>Tq=5J@IHc!0UZcoK(Lz_O
zjj6+;r`ns4sq2y27~Ii^TOY8QPVbROVs1O;lKnpsW)DTb^7TFS-w!^+$ic&fAR*QcttrHr>2o1V11nxdHrt@%j-kxr2*f|G?bF8<&gb2us1KXD4F&0Ni(DvHU)9z
z#rb_dB07<@V^+0Lc&0w_PtjCXo$2}>Y09kKKIFr7dPNnDKV8GCt-=6scKOrySW_;OT$M0W11hMch{fzA#9Z<
z3$8929G>hACShI9r*ZQ|WJ^O58Lvsi3={^T)$vkCe8v&Q
z!w7dulnmL!gsQZ%?H99+FRthar&yERx8(Ll&2Rhzz)W(VK}MpjA-o>mQxQl^Aw+zq
zf^9LJEO&h&k`&DSE9mIXU?4>{YVqD%7Q(2Npu~PbhPRtH2T4*7bH9PycS_zdelj>u
z{wJm(@3qS5*e6GlW*)ZSV@+_%CYxB}nf7O5m0RcHQ)Kw
zrI=8|Ba|MrTr9K@UFR>)imyb4O%MK3kgx;*^nscIqh@+rO(!8aA4JdZh%(*;j(SlN
zq8WmM@-y8QA2WZy&K(91x>Ns
zA|sGK=Xo37*k2()+#c$zt_QY+1-!n5{Mvjy2!V>6?4miL`ynS;{e7@WermoM43&!l
zdeLZxc?#%+m`TC?%gMcIw3qzom!;G$8X-Gq~$b9$@E)ZZObJnV4kMGy@_I%$tg_k
z;ZV$vj3K}`CD}f0khYeyz9;92_oTV2U{r1=l*y?pCfzPf?&vEUi4?Wu&Cpa>&Dz0{
z%}I5Vn5E;-6De-SI&mHd|Gz)q!I_gt%szO!%owMikX>G&#}lTdb)uIW!w2brOH^mz
zv8R6V2TTAd+u_VyZOnMxuQw`WZ{$ax`wrNU#2E3qG_lu79?Ym(at*us6N-|>T37H$
z!CJR{CFm&i&XUT^S3S{o9_76tH?IlFM@&;U9V7qe{0}Fx2fN}Q#TuwrGXPwmowr`3
z&_+7IpQ@N4eu0;c#(&~NfAkwr>CtQ8p7K7G1C@z6$qckWu3!L{=n~)BTB5KHlipWR
zB}Dt}A-=QxM1y%@ox$;l13PQz6wFjuvg>N~evsq-1%|F1sT{-zw2IYGJ(8a^atQrR
zl588?s}-)iZ7c55EFL-8^X!hH203K;B?QWEb~*TYJL9Zmq_no|4xj*tB`sCjKqq@(RaJ-F;z>c(UA4I~c|pBAw3q9J$rwQtfI72hLs
zWHC
znxvWEowhy}BOx^@NE2I-|JTIWo|7fr*KyTp5EMzYx?|!i&Xr+b02P^xaPiY`k^Fad
zT>3V&f`vJSa|L+HB?Y|_zc+jv4_Itf`~wmGOgp*mI~W%|#!Hb>_z?0)z+g)7b(GIj
zhbO*gA3R_Y%Y@h6WneEau=!bVMkas4`R{-mTTtnyH_6oGH~u8E!ZYctVRuZ_OP<|N
z8gP3wtfpcygjZ5V&6Q>)KYa66y-kye+jK20=*Q(hG^z6F>TA`3ZD*
z_i_XIhWCLn%~2zhw7_c5q`&thxR-LCtmx#e(o6PaJWf7?p8+3MFshvk%@(eWPOsZK
z*?W-fIn(ng3z8R~rELXu@TL^1EgSK|LS9z2#7+K9Zs)AZ>$g_X8xP9k=C4v3#-LsK
zywbK4lcS2pAzqf7Azgxc5@(Mm(DoHOZ2KYH9RX1G3J~D!=V&0ipcNl&#=#Rd%>c(Ldf+HEruT@L}1&Yya@WjowyEvwGf>R#X_kt
z@mh#i4vJOp8Taq4yUZus()bljj{X7$5=e#}4vZ{H8)B(&Q1u2c;!L2AN<}5YA>ra;
a?PO(bC2j6uE`}y_G0kEb60XUNl1T2%&12_UXHPdfQxQIwe3<3^C5HnSw{JGa_c(C7Qz743&A|-A*ryJS9PDo1wNUms2lfkms>QlE
zaM<_j&G&a4@UrzJVUUCJKORrXOma}G>~y1P27~j0mha8Q8Gnqu2)?GsW3aGG{&R{e
z2P&dsCknhcP|$R72=(Tmn0)-P!;J&AO)iFkP8^i#Pw@`mcXiUWhRxQZY6bhj&Ty&B2ElyZn}ODG5X$tLAh$2nM`
zt7D?;!@&lZmy7+}IS>i4P)~5gdgZPE>Ajl46o97X5(X*xdj}UuFqrXPL8V)iLBWra
zt_!mnTrbEt^Z?&aj(gnjT)@EgjPI>*Tz~oF{Ps1141bNMLL
zqrWilN$ZxI*}&lFX$v!jTn2Xip`VSfGYE_3JLizY;BiKG@Ah&A3Z->1uB{Aut?Fwv
z{$@}gQFJJ=pMmp?eI2j67%WvUJ)Ty@Ak=$K~#n&jIY!fSxRV4N7MaZ
zzY|)!?|+4c$U;Kf0v<|jR3KDFWSNlhDnd`UR!Z><35_*77aL?nXnXphMu$Cwz90S;
zw!)rJ*MixK4wi%(%bmAs(;`$WKjK81KcTPIebf7bzkA3$H|w?`RA_H#ed=C9X-nO2
zMMnr#KAfAE5kY9lHr0ftctQg=3z!YwAoQVh_kXZ`0ig?87NqtUV;+XR^S9&?8ajOQ
zM&?z_vro^+r;yNvF}DR3lL(b>;|<&5|BK4b%pbT$D0wyiNiF_AFFSDD=mF|I@aNx$
zn+RP!CULl~h0qhFHX*~UgnqR3+4a9C*gxqq#u~WR>W@1x;~B1LIkUvI@Oz_^UmFRn
zpnrN+FX{=sK2~V<;~}AEd#At0KPI$#$?~2XxSxxtsG|rK-5>sA)>v*Z;
zb&s*`ZEmyq_h6r2TVefY>PkY7mGDY}HWM0abWHAbC~S7wguETS;)h+bC@?M!I$lH(=IQ2z#&dn=o-GC0HcDcop2gLe^i4j*%IpJfLd
zrIDvCd{>TTVST=ON!OKL=6^sl#B6)lza0G5R{6(yUj|RUCf4)t9t$68Xy=;~n(dX;
zsCAyuu%aHm2dLlWbt=$drPenmn@hlRX5d%tH6S~!6d4GF%GodNzCK6hx
zyJ(e78#G)oe@*XZ4t3+BVB2lv$AN}yuG%WhfBE8Pg+qB9+)w@zy>%t>b;Q-%xEuK?
zGs(w8J*QL^-&4i@NEwSddma0#()fQ{nsEP-wzX?dqrNrMrue!pF_`&Q(stEH)SqCC
z!o~m&Mn0bT+Owa7b${fMzw=!V9$7RyT39i-YoQrj%_*Y}$`aiHw-DHnh>24zQu7Zn4=Pd>&1C*^)aNe!83KaN``bnPZ?U2X%T=+L;
z+p=#zzV
zxDlb>>bq=j6*7o=o|V09hy&ewJ3t@nb6@P(g|(>5NBz%>f8e}b{Ew&Fy;258()Jfi
znh<*V&FjFu$Wc*isYY$goBzY#qSp5aojED+XC`uYowsadAL?~mQib}pKn4>6r!3m5
zIIz3@Tj?p(<$vbcTOIzvK9%>jP9IWa(3Tx3q=bIL@>XR0@L@u|m92FTVjUVq#;aT&
zA^-Fm0#{+)Pm01uCDsv|`lj_gpHrMUleJ~hjM|BVX=)3*DP)Nwv1
zRbM}neU^bik*%aG_T|Lqb}voTcQjXabGk?v9IEeG*w$7*Xr$6
zbZzjd=hxR+c#c8j@w1~^@;Gl>(pDs5y~nR#@N2+%epw`9*=_U_Gea}lFQJ}ngq(Tu
zXL0@~&*D?Z?~U&VR`QAzW8gMFdJ>oPs4?{r~xpH$g%#
z|HAskwQJERoL}K?Yj&vS5*n1ZUOXxQ=draJT*Pxti5yGtdC5TgAA{7KjSPO&$vz!z
zMSnNJuR4bQM?6=4l?&GIQ>%U5mJkkRU6l$e$GXW`)%KR-J+)2W^c}AuH0{;ygMl~)
z%YU=$N6X3x-DspU_Wcy0%>yHD)B4yq&F*J+1exqk_AIV8Mt;2{Swl3z3D9_7H=e)@FzOM|SlTyNi&%MycxJ&0AMqgpybWs1l^XR;)ym}V7lF%pY_2eR|+SOgq!n0+3(7@AnUNaYuh+
zrOBIvaqgTn>L0_p2^sAE$zdHMiGOX63+f;w@sx7XVnWZ)DgFGw59efUU%kc?4(2)X
zw#H%qSY@=T=X-Kc@^y{*>|zcEw|H7F#k%g#*C@J-^M5kocjsW_LA0=1*zE9x2R8?&?N1jY{g{iGVoemeL);M3{z<;Furimk-
zC)IrDPp4AkXUw$o11AQ-o;7|ZvV{7qj<4XIBvi_Fty&7+halg(&T#ZwYN6k}X2ud~
z^vfTV)Cn!HS8a1Yg*-Tr(Q*a*ShRKMp^hqpzg3Xw|2{n9Wz
z{o-Hr!&gIP3l3pFu*|pqG9<{c>Bz;YvgiYYD4S&6jr4vjO*5v>?7O*c8O7nAC%^3*c}6qAZh2qVX6P+*N95jtS?Bt04}zgX
z564|Wx2~&_eenN{As4?F{Pe(4qI+?gVIb>>iVx~#-z^9#A(>^XPtEO0M|OGH=F}|63$bB`PG
z44q1z@zH)Q+&czw9s)`WqpRSjuEVH_z&&Rg%-M%xo}U$VENzg_CbSQng($Lz1>XDo{l;Bzwq1Pr9>iVm7aV5=`EqI9jRNKwMpSaQw-x*xuU(Jr
zLY%H0kv`>M1$XcDO2n57IdLu*GRJCvD`(g(+}HA(0F#vqcIJS-%)@#Sh<9UYqkOW5lVZH~AV-l0JPv%RHMCcvVSMt%hE^F%l$wlkg
z{Zq;S9GQ_;s)JrlFin_UhM^zQA;RpsFq^D-q)
z3hYgp*^oi+v9@+74lFpx%kzIXI2+e)#*6XRfECNS?N9s7`E=-i7Y*338H|m-*k%I#n?1%$*6@ET>A7RO9&$?o
z+Yg%G|M}*bH@L5$lqa15uor*vWV8a~`)^GhptwU9Eajx2Ke454zzmd%5azlIvLEhz
zA`1V0`^;TbXJSLdmH9yzPjv}m=+=eB{a{(_bERQ0*;#*I3)sn*>qqa3MZPb`kMg9~
zmtz5O4x9-~%*K4*9w&b(On|R_e=KRh?rLwS#!bizn>IFO!hbm8LI<7KO>>if!l@4I
zi<#|#T*?~9*$=y>qChnQz
zucBbj8MDvW9$dK?=p+n2b96l_OL%|=-qKX
zTeZ2R&gU#*O}BrsUK~2I%+Nk{q*kPGGp*ouBkEA$91i)(l*rhWUS+pKpS4B
z!V|D}4qShbSfYJ+*mCB9x2r!_Hy~cHiL4>@gKk?1FKNh4W-mgiZkL>?7NvZCR_?em
z4ze?EWK|LM*~+0ERQFrldi(ys_~}kw?Z?#bp1Z_9MVxO1-GZM$b!{2B@6aD%PrXR*
zJDVp`=_JVXY_sM7{4CQ-*PH-TgWXSa!%j=ITVsDa$QmT>+X?+LFLM*Bk736Z*9JiU
z-e6Ey0scM*N|#JQzaBkTO)z)4asu7|hQ(d(k3h{>jX#gme&EE?#s_HdmMgVDC}f$w&Fq`|tZN?$FCrDV=Gh{E)vVy%YNK+rsLC
z^iO{Y;)+{M=h(J_l^0w){bHO6l#Zx~X@>m@+1vGLppR_!w{*}ls{JONLx_?3oB`FJ
z;Aa|6U}1Ukff48nSpwFa!+N4e%r&VWb6ziVpMXB|%*_F+_rqL@#g<^0$=K*1^qPVm
zM|xpDvF-NW
zwOvkqrO@+cb?pnI`&Q5kq&#YSa@Us5{j0fg);+`x(VmN02j;(R+Rp_&6H`(AMc9AI
z29{c=qJEqdna!g*7x1%Oi1PHX_vF+e*vaLu|Hy>AkR5bv7EJan
zS5ax3gRLJ~`u4fmH*N
zV1cW`gIDmYv8@XCr#jtQ#Ygw!_0fM!<`UJlj9KsnP#3wT164$j_l*F9+_;(GaA3YOverPw(N9-MHc%aOV0{
z3;k#NCc}B1F>XGwp(X+6+E{foBY^&2*2AhjVAJET4qli?N!xEXJM526KkI*_f5fI!
z`f60)BlNeh())<=%_*gE{VjU^v>v?&JeqV)7L2!tZoq&4x`wWYU{$Zp@IJ~XgHf9j
zF!9EzX0pp|ANg*H@gY9`a9yk;AWwEx1}MA5=F$N=U!hiIssqm@tURcmy>~nrqz2}U
zGHX!%*NPjx_Z#*rIJGt{9gKhf&FF(4{r_zrWhP)e?`=yY{p&na*QbXguC0=zZv
z6ZIj%6o-sStV<#0RiPrdLFybs27`wQTFLI0!qM$Y}lh&!%l-5mmXx$}IE
zDc!GFa4h9dy;->T7H~CBnB!X5Z3}jimPEe}=CyRE|LVeD!45UhZ;;%$u7cLvB))e5
z{*$BETB(0ch1x!@0aXh+ysGFw_dfu~66i-?4)esHnTp3lvTeptPe604V6SzS~N*2GZE54h@@2R3Q-i@LRxNwLb<=M^ZV=h
zIOjUo>$>i9o%gw}TVamNNn5UA#`~Q&{PlQ;8D*Qaea2=DGgRF>3qnDjIca|;!Sd`|
z@;QH?WMk7O2M(L@sDAMHh8Z?PZcvl;FYJ47eHlI&9}JV;a}qSl`8jwNRLkYrY6bo7
z>)z8#|F9WbWVe2G=NV>jT{mGHVZ9yZ#fgdB!whS~RyAkD@$=T(se*}8dqPCO=XD;*
zYoOED*)o%b^(s~+bV!0ZLhByIfNe7yk}7|pcSj{GQgw>W*k;qzbr$}4)74X^&~?nq
z@0$h}L<~Na0zX)qeUO2!`OZkq4(OUEc{sIz_dGg^4Z+R4vmL@Q|HZSphHBt^u3r(rPNtG(676sd(8)Sb`c^#R+-)KF))8?
zBl;q}WiySCS`
zmthOOzdp|v`M1|h_}C47o2i9^RjS
zW_bwf>xenes8JpSGM3M!eNT4Q{0DuZhWNB9tb22ZLxVHspLRc4qYJ$oNt|ytpnuF>
z<%ShlU7PaS0eK7l{;(he^h=u~`LD=~z$c%<_Y2SWjH%*T)PnG=qPw5eD5j
zWrK6VSTFg~`m45}T;Y$F{dB&EkFj*1llA_VOE&bo-s~P-0R5@3Xyz{XyS9HB=8HJ}eo;RO{2rMtNqSVTL-H(cTm0@&
zb8+bjjGI3WtQvxDaJHDbFl>MIm>+>M(1}_TW)}tBg1e6Q--1k+8fiE9%|i#89bn7;
zI-*E*TF*aar$gUaOX)^ktRoxo!bk;d`^YCie!IY+n0TybJFR$d8Rlm$*%4cXb?z=w
z+~|+_dYRFgv$5Z2UKLB*U>7NdXl@2|j(_Q)I*@UBR3(l1@)C)Sv!Q<{X_u040{R*G
zix%tB_aT2FI0|;pbJJxH5bx)ws!iv6pQ~YwE#-Y+s9pj7{qtWwQ3S28`}%!{e@$Ts
z^DlHaMAnKFQT~3%hFf8OcaMoGwSaFYKNMvnz9{Jh$KZcmO^2CQ=msa~wbFR;YW_1^i2p9#f6D;O`z}~C0sV_J2Af2&
z?%P{FwRA3X+}8i`hQ4@NGrJh`-oNhFUxz$#*@@l`1a+MPI3TfZe>AE!;y?$dYEa;Er7DB^#=wa>fS1!}ZCkyrrg
z%lEC-!MH+dP9XKcz8=N&E-;SS=1cy+s?3l=a3g2k6@A2Oaixgt07Ere9wjz~l)V?4e}E?WUy73VAN0snAaUe5&fbH7W^s{nR$wA=AZuuo~040^+s
zXgPbO8vM|>@t=Q-nD>+KP{S*5=emyMb>M8#FLK@gANPn$fc!};({14OC!TTR@Ut$M
zF7$-n-?Ys1GsH3TOVpPhJ%3#)4!#U5Fo8F<;K@KikLs}S>p&ilAhp>_uK*QR%C)i9qE#bHxlxEUs(
zz6bjsyDVMp4F8*}2TB4!X%VG4-IRY5_#QH{G)(X8`uK
zfgD9Tmnp;GWU3>v027%I{O(glwe4-F-++j_!6RT0vv}1r_)8WF^K%fF?B8m#67#A}
zdwri`oTu+h(K3v`YrLq}3kpWoO%xz*k)=@92{3=CCS}1c_(d-%1kIxFFF|KS3iYm?
zvqXaztS_wS+Yf5Ju}zi0`1XlPb$QJ5ILhhx4FBr-ZQbJVvs#nyWWf$tbW^#M_VvPh
zst&rV?y!E`2Ghs8cst=Y3Vp~h2bbK`m=^@*#w*L7r@kaGuI^VzNNp)gBKGX_c
z%Hdo)4hk)^@1*(^N!E1q0_V#qNCd&}bxcHh9_$C__eT_h`7317`M@8NCrS;#bN!-+
z|3RFw`MXH^u8qw$W{7|xuZ)?3ph{*?bS!@uubtJh59}DXHy*>hjw`clZ-7>uty9%t
z()tfP7DRO^6}sm}YCgTC@7Bd~%V9d_6)&4QNblk8a@k#o)7?^8Oux_1c(9eer!D6B
zX4^16K7RQi-N%)(vs!s!v)ec(Uy+Y*`vxqwBaYR%wt?>P!4->NKc#g%b2zk4Ldt)1
zqyboYNLbH=?x!94>gr&9@Py`X%HKp)Q4IB|#;=9c$9}1~Ki~tkAD76##P}OQKLvf*
z7RSrXu7ps(3J_v47J`v;O{^=huSNCJhjMYQjvc9rR9~D1
zdwT}BE>ZmS59mgh8R!Nh-Xz&I&zgVU9pU`;=3wwxrJ)?@&5qyx9&wlT0yC*z`Fh4o
z=sxF29Z+?GZt=B|iG;#lX5Pn0=iRsUX=N@9tr+1d*?38`RASd7O
z{0{0@LVWE119kb!>goL~j>JrJ8-@%s|hN0sXrdgn6JyaOmN-vxSomB0wD?Xm0W
z-MdW3gzmR(ZT=aObRQVKbFBdVe?7ON`*g*=SgS_pdTEHB?gYaZZ+m);^e^sZ?gpQ5
z+%MDb8Tj;6e(ry?#*p&2C^5|~lYW<=<5niUb68P^EbxxcKS||Sr{KzJ
zQ>shBJpoI@!Pb>Ap6(d$)e?M1=Q&u&QQr^$Ci7+yVQ`)L`I8Ijd)THGE(Ff@`Sa5h
z`RU(e+dhhXhbE`;(|f1)vqyR}{OW(q6s4#RT#Z&)BYzKtS&T{eg%y8Bj#HlcL*|a=
zg6+#kv%M%kb=TFX9(Edz9Bx9~%aSZxHtEhx7X8BcO1Frla^rm6V=RVapl71_en0is
zbJ>jpi{XE$svA-Ze|~>9AN6D9lE9tR@6C8TYv>;O&*pJC{f}6wpVH8Ub=1n-R>*<=
zPOJ#3(~KVV3IUA&9$bGM>j!R_8F%^wR@@F)8BG1^sCVWSde=0F_VUADn-cYa8+P0FV6qzQNOC@
z?c;U>74D@jeTQ`!Vn@u%VgC=F5a_^@0X`Uio(up000000c$^iNc|2EJ7si#8p^y?K
z4Kzq@85>ld6eWsgk}-t_qzDPWCPFEsfufsIC^Q*EC`F}obDORP`JMSFi9~q!``7xM
zVeh@pTI>0)b=3W{%DGe)Ppu~1tF2(+eMfy@LK%zRQa^>N0v6XE2Yx?)odq4avTAC7
zCX1vx?@`M#SOm*?t9e{tp|91f6G7Hz$c}UwLhQt`lF5;2EL`37=jNr8=gQ}z!(%8M
zun;UJpQj-AH!18(GPrZ5hQf%1v)BCIQP`>RaZ#H<;diQk&AUDdMn7wFpD5Ef
zY#$I`p-UrvfBIct(9rY{@Yfa^_Uf&FMXOHHP%3#e4dbLLpd
zBnF?H^;i4dq>&MIoU0C|(b&cBci%+g-+{wV0@u*^y|dZj1i99gbe!(-rLkP-RFDm#
z;oILPzdL}&l9`6%gtatWdj~~-_>HBp*CRjEhWNc!$)U7-KZ_pAJ^PBg8GH%N{=Ke*
zL86}QW6vB05(gDs?Q#ZJJmn6tQU;E56Z73PS=4Gf#&=t>NNkeL+vm(ezUyF3-U=2E
zn{I!y9mXQM=6QVPVFu;eE7N@QIP^7d2y`>#5OMML;QAgGvo@84{UyhLLGy9^k@^K3
z?(DGYj)>>*)59SvwS|LqS;otN**snfh@TORwWGeZu$@mRJ%y>#OO9G30P?T~eEXN-G<$J;?&DR>e!?y9MMoEG#!XE#UZT
z2dnf50oV0QI>%oTkg@t`>mX6U7bPdpfcYX)U4A^41d33QH~f@&T*NeYy_x%yL~MEP
zH2D^>XES>$)~ASg9`-rW^q7bYiP|EKWD!2kFCAQbKty5!Q)&@^AfmRa(8OfF2;HLv
z{<37BP*=Mg#dr}v>=t&)`-)gQ;Q1>;PeeUrl`Sa&AEH&7o<0;Hjc&c=@JWE<{O|s2
z%|tZEyml7z1@z>pBs5MG;q=e!_m%TRs6L5!N=*`Br@Xq#;K950sj*o`?
zdIpMPwq8*~W2SA5&O=E;n@F2^#mtF*2h_
zUB;esw0Ano!lpdIAv}}8_6!5h&?_`*hPrs^zozg;aUW7GDD=!}cyNI5;Cfv7hek?@
zFVntWT10ezoBds3<--RQs3MP8?O39(((i??;S8G74Ud`R&#W
z`t5B7&9$XbZui#Y0^!8BmNHpeB?>ySc@g|aDJqBN4XKo;;B$9PPjM{a+kbC$3pUV5
zv44_fwt+#o^T$&k0$BX?L-+F(4y^RiyP85$@2;AE1C48m&h9zB(z2xCx8(0PbM{au
zP&rrJ)=T_VeIGW2aBk*aqxLW(3i(TR$?+d4WH_#y_}51oZDx@&Bh^?0SPUKfcmsze
zRvvK*$9UYgXErt^3FNs?Y~@W7cNc|O^}A`<*=avFl~C}uin)EQM~aZ7n))_Qiorgk
zcf_uLrLg&H%Y;=kXjt6Wx$3)#!PpOdJ`2lP*xMi8n5DzRuP$zS%mx7=(!G)ZBM~aY
zLnYRWNI$VlF=(u#k@hBa#J(&F`UaDn%w#A$GMl%nf|sH;h@W(Q2Ziq30=dh(h#x*<
zzbU_9&~9v5b;gy$ixzsA^S?Zl{|SvvAbNg(pOl*6dql*A*GifDN!@LYL4?M921$!`
z%_e@J(C_(U`Cdf|vuz};0mT39#`T`rQ52RaZolDNKtp)tBalYk#oZ@c*OXCF39Kwm%aW?`^
z7>n=_)cSrC9TithjNm8+o0Q78Jk+C+T-?4nob-#~vP}HdA5y5j93pOiL?QH?ZD^u9
zgXfzI28+%d-ZY#Z*lm#H}}QPoLIYQ2HD1=XneL>nC3$>c>O$N>L0>^O)jyUf-3~P
zu2db+PZY8G&!>|kNM8HjyA97T;c?^I`B!C09M-G|T%@&&K_P#3%jqHtU!!_|Or8?G
zp44hLNg#bRw;@xjJe|hF13h;~XS1j{?bTac!DB?mx&X#R#KxjdkEUqCOI@uOCWOOR
zoRn8zxXnRr&Z3e35k7Q>>0S6k^i9289;`8k^ylEn&UBL3^O3p3MiWlV8rAl4)iF}X
z1(L9$TO4-FT<6jg1w?ACemsKsSTtFzXB_Fvnz
z`RkiB-G^{aA?eVn3KG9a^O=9z0|ew|jh^DVna5^6X4_It4yg~4=adm0S2m4D+=Fl
zN{q86Q}EKCzFtD+u&8^=);3y%Bi0r#?$#1NzU@7Chj6HO{oG%F15P4>Y%jHho)Gan
zz|;2PR1qDs+IG?=0u~Ox{($|=p=ABatASoDvfTSG$o`_S;?*(PkVXopC3zjpSPG7X
ze`^%9N-X5kMMn{0v3K5sL?zH$hPrz{V0q3-f9JJm(
zb1H2jeMXgiwt6XzRqNO2m6=dD_K(GqT5?X%vZD&+X%r3>IIFj!n&|j`HO@ugQSILO4Jibn4?Ci8S%)Lk9lqCbc{^WEm
z;`jKc)a4GMSMzNLPy8Z%c0->``=5By2W7Sxoav?!;F5Fj*;f|t8!xx|KjYDquy5P!
z=^`{Y+?x?V&Z(>p)bq${B6OXrbYgqJ_6KA$oO
z7vW;rUT`x|fK$`
   
     
-    
-    
-    
+    
+    
+    
     
-    
-    
+    
+    
     
     
     
-    
-    
-    
-    
-    
-    
-    
-    
+    
+    
+    
+    
+    
+    
+    
+    
     
   
 
diff --git a/previews/PR1624/tutorials/p4est_from_gmsh/index.html b/previews/PR1624/tutorials/p4est_from_gmsh/index.html
new file mode 100644
index 00000000000..3f6807813d9
--- /dev/null
+++ b/previews/PR1624/tutorials/p4est_from_gmsh/index.html
@@ -0,0 +1,366 @@
+
+17 P4est mesh from gmsh · Trixi.jl
diff --git a/previews/PR1624/tutorials/parabolic_terms/bed141da.svg b/previews/PR1624/tutorials/parabolic_terms/cda863ac.svg
similarity index 93%
rename from previews/PR1624/tutorials/parabolic_terms/bed141da.svg
rename to previews/PR1624/tutorials/parabolic_terms/cda863ac.svg
index fe087a3b2b4..9efbf291417 100644
--- a/previews/PR1624/tutorials/parabolic_terms/bed141da.svg
+++ b/previews/PR1624/tutorials/parabolic_terms/cda863ac.svg
@@ -1,45 +1,45 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 
 
-  
+  
     
   
 
-
+
 
 
-
+
 
diff --git a/previews/PR1624/tutorials/parabolic_terms/index.html b/previews/PR1624/tutorials/parabolic_terms/index.html
index f2c342ca627..526f2f24b8c 100644
--- a/previews/PR1624/tutorials/parabolic_terms/index.html
+++ b/previews/PR1624/tutorials/parabolic_terms/index.html
@@ -1,5 +1,5 @@
 
-10 Parabolic terms · Trixi.jl
GitHub
10: Parabolic terms
  
Experimental support for parabolic diffusion terms is available in Trixi.jl. This demo illustrates parabolic terms for the advection-diffusion equation.
using OrdinaryDiffEq
+12 Parabolic terms · Trixi.jl
GitHub
12: Parabolic terms
  
Experimental support for parabolic diffusion terms is available in Trixi.jl. This demo illustrates parabolic terms for the advection-diffusion equation.
using OrdinaryDiffEq
 using Trixi
Splitting a system into hyperbolic and parabolic parts.
For a mixed hyperbolic-parabolic system, we represent the hyperbolic and parabolic parts of the system  separately. We first define the hyperbolic (advection) part of the advection-diffusion equation.
advection_velocity = (1.5, 1.0)
 equations_hyperbolic = LinearScalarAdvectionEquation2D(advection_velocity);
Next, we define the parabolic diffusion term. The constructor requires knowledge of equations_hyperbolic to be passed in because the LaplaceDiffusion2D applies diffusion to every variable of the hyperbolic system.
diffusivity = 5.0e-2
 equations_parabolic = LaplaceDiffusion2D(diffusivity, equations_hyperbolic);
Boundary conditions
As with the equations, we define boundary conditions separately for the hyperbolic and parabolic part of the system. For this example, we impose inflow BCs for the hyperbolic system (no condition is imposed on the outflow), and we impose Dirichlet boundary conditions for the parabolic equations. Both BoundaryConditionDirichlet and BoundaryConditionNeumann are defined for LaplaceDiffusion2D.
The hyperbolic and parabolic boundary conditions are assumed to be consistent with each other.
boundary_condition_zero_dirichlet = BoundaryConditionDirichlet((x, t, equations) -> SVector(0.0))
@@ -106,7 +106,7 @@
 │ ═══════════════════════                                                                          │
 │ #threads: ……………………………………………………… 1                                                                │
 └──────────────────────────────────────────────────────────────────────────────────────────────────┘
We can now visualize the solution, which develops a boundary layer at the outflow boundaries.
using Plots
-plot(sol)
Example block output
Package versions
These results were obtained using the following versions.
using InteractiveUtils
+plot(sol)
Example block output
Package versions
These results were obtained using the following versions.
using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -126,6 +126,6 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.

+⌅ [91a5bcdd] Plots v1.39.0
+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`
+Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/7bbbc1a1.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/1b78a1fd.svg
similarity index 79%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/7bbbc1a1.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/1b78a1fd.svg
index dc703fcb89b..395bc76f8c7 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/7bbbc1a1.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/1b78a1fd.svg
@@ -1,54 +1,54 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/00c0ec1b.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/51a3929a.svg
similarity index 83%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/00c0ec1b.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/51a3929a.svg
index 5519e6e2122..94ad9baba74 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/00c0ec1b.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/51a3929a.svg
@@ -1,46 +1,46 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/4c20d287.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/5dbf4fa6.svg
similarity index 82%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/4c20d287.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/5dbf4fa6.svg
index 74c98df500c..ed2c1603d60 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/4c20d287.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/5dbf4fa6.svg
@@ -1,46 +1,46 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/0cdc6371.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/bdf59697.svg
similarity index 85%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/0cdc6371.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/bdf59697.svg
index 65eafbad348..562e1131663 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/0cdc6371.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/bdf59697.svg
@@ -1,48 +1,48 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/6c8c5e59.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/e8522b82.svg
similarity index 85%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/6c8c5e59.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/e8522b82.svg
index 0caf9d85a5f..235823e5f44 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/6c8c5e59.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/e8522b82.svg
@@ -1,48 +1,48 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/b9ccf219.svg b/previews/PR1624/tutorials/scalar_linear_advection_1d/fe5b2ffe.svg
similarity index 83%
rename from previews/PR1624/tutorials/scalar_linear_advection_1d/b9ccf219.svg
rename to previews/PR1624/tutorials/scalar_linear_advection_1d/fe5b2ffe.svg
index 375aa167e92..e89f8bcb3ef 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/b9ccf219.svg
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/fe5b2ffe.svg
@@ -1,46 +1,46 @@
 
 
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
+
 
-  
+  
     
   
 
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/previews/PR1624/tutorials/scalar_linear_advection_1d/index.html b/previews/PR1624/tutorials/scalar_linear_advection_1d/index.html
index 40f80440575..c2f40d04726 100644
--- a/previews/PR1624/tutorials/scalar_linear_advection_1d/index.html
+++ b/previews/PR1624/tutorials/scalar_linear_advection_1d/index.html
@@ -1,5 +1,5 @@
 
-1 Introduction to DG methods · Trixi.jl
GitHub
17: P4est mesh from gmsh
  
Trixi.jl supports numerical approximations from structured and unstructured quadrilateral meshes with the P4estMesh mesh type.
The purpose of this tutorial is to demonstrate how to use the P4estMesh functionality of Trixi.jl for existing meshes with straight-sided (bilinear) elements/cells. This begins by running and visualizing an available unstructured quadrilateral mesh example. Then, the tutorial will cover how to use existing meshes generated by gmsh or any other meshing software that can export to the Abaqus input .inp format.
Running the simulation of a near-field flow around an airfoil
Trixi.jl supports solving hyperbolic-parabolic problems on several mesh types. A somewhat complex example that employs the P4estMesh is the near-field simulation of a Mach 2 flow around the NACA6412 airfoil.
using Trixi
+trixi_include(joinpath(examples_dir(), "p4est_2d_dgsem", "elixir_euler_NACA6412airfoil_mach2.jl"), tspan=(0.0, 0.5))
[ Info: You just called `trixi_include`. Julia may now compile the code, please be patient.
Conveniently, we use the Plots package to have a first look at the results:
using Plots
+pd = PlotData2D(sol)
+plot(pd["rho"])
+plot!(getmesh(pd))
Creating a mesh using gmsh
The creation of an unstructured quadrilateral mesh using gmsh is driven by a geometry file. There are plenty of possibilities for the user, see the documentation and tutorials.
To begin, we provide a complete geometry file for the NACA6412 airfoil bounded by a rectangular box. After this we give a breakdown of the most important parts required for successful mesh generation that can later be used by the p4est library and Trixi.jl. We emphasize that this near-field mesh should only be used for instructive purposes and not for actual production runs.
The associated NACA6412.geo file is given below:
 // GMSH geometry script for a NACA 6412 airfoil with 11 degree angle of attack
+ // in a box (near-field mesh).
+ // see https://github.com/cfsengineering/GMSH-Airfoil-2D
+ // for software to generate gmsh `.geo` geometry files for NACA airfoils.
+
+ // outer bounding box
+ Point(1) = {-1.25, -0.5, 0, 1.0};
+ Point(2) = {1.25, -0.5, 0, 1.0};
+ Point(3) = {1.25, 0.5, 0, 1.0};
+ Point(4) = {-1.25, 0.5, 0, 1.0};
+
+ // lines of the bounding box
+ Line(1) = {1, 2};
+ Line(2) = {2, 3};
+ Line(3) = {3, 4};
+ Line(4) = {4, 1};
+ // outer box
+ Line Loop(8) = {1, 2, 3, 4};
+
+ // Settings
+ // This value gives the global element size factor (lower -> finer mesh)
+ Mesh.CharacteristicLengthFactor = 1.0 * 2^(-3);
+ // Insist on quads instead of default triangles
+ Mesh.RecombineAll = 1;
+ // Violet instead of green base color for better visibility
+ Mesh.ColorCarousel = 0;
+
+ // points of the airfoil contour
+ // Format: {x, y, z, DesiredCellSize}. See the documentation: https://gmsh.info/doc/texinfo/gmsh.html#Points
+ // These concrete points are generated using the tool from https://github.com/cfsengineering/GMSH-Airfoil-2D
+ Point(5) = {-0.4900332889206208, 0.09933466539753061, 0, 0.125};
+ Point(6) = {-0.4900274857651495, 0.1021542752054094, 0, 0.125};
+ Point(7) = {-0.4894921489729144, 0.1049830248247787, 0, 0.125};
+ Point(8) = {-0.4884253336670712, 0.1078191282319664, 0, 0.125};
+ Point(9) = {-0.4868257975566199, 0.1106599068424483, 0, 0.125};
+ Point(10) = {-0.4846930063965668, 0.1135018003016681, 0, 0.125};
+ Point(11) = {-0.4820271400142729, 0.1163403835785654, 0, 0.125};
+ Point(12) = {-0.4788290988083472, 0.1191703902233889, 0, 0.125};
+ Point(13) = {-0.4751005105908123, 0.1219857416089041, 0, 0.125};
+ Point(14) = {-0.4708437376101668, 0.1247795819332056, 0, 0.125};
+ Point(15) = {-0.4660618835629463, 0.1275443187232316, 0, 0.125};
+ Point(16) = {-0.4607588003749649, 0.1302716685409717, 0, 0.125};
+ Point(17) = {-0.4549390945110529, 0.132952707559475, 0, 0.125};
+ Point(18) = {-0.448608132554204, 0.1355779266432996, 0, 0.125};
+ Point(19) = {-0.4417720457819508, 0.138137290538182, 0, 0.125};
+ Point(20) = {-0.4344377334597768, 0.140620300747629, 0, 0.125};
+ Point(21) = {-0.4266128645686593, 0.1430160616500159, 0, 0.125};
+ Point(22) = {-0.4183058776865576, 0.1453133493887722, 0, 0.125};
+ Point(23) = {-0.4095259787518715, 0.147500683050503, 0, 0.125};
+ Point(24) = {-0.4002831364505879, 0.1495663976315875, 0, 0.125};
+ Point(25) = {-0.3905880749878933, 0.1514987182830453, 0, 0.125};
+ Point(26) = {-0.3804522640292948, 0.1532858353164163, 0, 0.125};
+ Point(27) = {-0.3698879056254708, 0.1549159794501833, 0, 0.125};
+ Point(28) = {-0.3589079179688306, 0.1563774967770029, 0, 0.125};
+ Point(29) = {-0.3475259158676376, 0.1576589229368209, 0, 0.125};
+ Point(30) = {-0.3357561878650377, 0.158749055989923, 0, 0.125};
+ Point(31) = {-0.3236136699747923, 0.1596370274972017, 0, 0.125};
+ Point(32) = {-0.3111139160522804, 0.1603123713324616, 0, 0.125};
+ Point(33) = {-0.298273064867608, 0.160765089773461, 0, 0.125};
+ Point(34) = {-0.2851078039966239, 0.1609857164445887, 0, 0.125};
+ Point(35) = {-0.2716353306943914, 0.160965375714529, 0, 0.125};
+ Point(36) = {-0.2578733099632437, 0.1606958381868515, 0, 0.125};
+ Point(37) = {-0.2438398300730194, 0.1601695719599709, 0, 0.125};
+ Point(38) = {-0.2295533558334121, 0.1593797893750759, 0, 0.125};
+ Point(39) = {-0.2150326799566391, 0.1583204890160489, 0, 0.125};
+ Point(40) = {-0.2002968728818922, 0.1569864927736143, 0, 0.125};
+ Point(41) = {-0.18536523146042, 0.1553734778363979, 0, 0.125};
+ Point(42) = {-0.1702572269208345, 0.1534780035235666, 0, 0.125};
+ Point(43) = {-0.1549924525477129, 0.1512975329264932, 0, 0.125};
+ Point(44) = {-0.1395905715122586, 0.1488304493795921, 0, 0.125};
+ Point(45) = {-0.1240712652914332, 0.1460760678321895, 0, 0.125};
+ Point(46) = {-0.1084541831014299, 0.1430346412430583, 0, 0.125};
+ Point(47) = {-0.09275889275279087, 0.1397073621660917, 0, 0.125};
+ Point(48) = {-0.07700483330818747, 0.1360963597385416, 0, 0.125};
+ Point(49) = {-0.06151286635366404, 0.1323050298149023, 0, 0.125};
+ Point(50) = {-0.04602933219022032, 0.1283521764905442, 0, 0.125};
+ Point(51) = {-0.03051345534800332, 0.1242331665904082, 0, 0.125};
+ Point(52) = {-0.01498163190522334, 0.1199540932779839, 0, 0.125};
+ Point(53) = {0.0005498526140696458, 0.1155214539466913, 0, 0.125};
+ Point(54) = {0.01606484191716884, 0.1109421303284033, 0, 0.125};
+ Point(55) = {0.03154732664394777, 0.106223368423828, 0, 0.125};
+ Point(56) = {0.0469814611314705, 0.1013727584299359, 0, 0.125};
+ Point(57) = {0.06235157928986135, 0.09639821481480275, 0, 0.125};
+ Point(58) = {0.07764220964363855, 0.09130795666388933, 0, 0.125};
+ Point(59) = {0.09283808959671735, 0.08611048839446452, 0, 0.125};
+ Point(60) = {0.1079241789809607, 0.08081458090718853, 0, 0.125};
+ Point(61) = {0.1228856729475325, 0.07542925321638272, 0, 0.125};
+ Point(62) = {0.1377080142575372, 0.06996375457378261, 0, 0.125};
+ Point(63) = {0.1523769050236616, 0.06442754707512513, 0, 0.125};
+ Point(64) = {0.1668783179480157, 0.05883028871526293, 0, 0.125};
+ Point(65) = {0.1811985070933818, 0.05318181683604975, 0, 0.125};
+ Point(66) = {0.1953240182159306, 0.04749213189240609, 0, 0.125};
+ Point(67) = {0.2092416986775084, 0.04177138144606024, 0, 0.125};
+ Point(68) = {0.2229387069452062, 0.03602984428372727, 0, 0.125};
+ Point(69) = {0.2364025216754475, 0.03027791454712048, 0, 0.125};
+ Point(70) = {0.2496209503696738, 0.02452608575629232, 0, 0.125};
+ Point(71) = {0.2625821375791982, 0.01878493460541621, 0, 0.125};
+ Point(72) = {0.2752745726282818, 0.01306510441121807, 0, 0.125};
+ Point(73) = {0.28768709681727, 0.007377288098728577, 0, 0.125};
+ Point(74) = {0.2998089100619555, 0.001732210616722449, 0, 0.125};
+ Point(75) = {0.3116295769214332, -0.003859389314124759, 0, 0.125};
+ Point(76) = {0.3231390319647309, -0.009386778203927332, 0, 0.125};
+ Point(77) = {0.3343275844265582, -0.01483924761490708, 0, 0.125};
+ Point(78) = {0.3451859221046181, -0.02020613485126957, 0, 0.125};
+ Point(79) = {0.3557051144551212, -0.02547684454806881, 0, 0.125};
+ Point(80) = {0.3658766148492779, -0.03064087116872238, 0, 0.125};
+ Point(81) = {0.3756922619615632, -0.0356878223992288, 0, 0.125};
+ Point(82) = {0.3851442802702071, -0.0406074434050937, 0, 0.125};
+ Point(83) = {0.394225279661484, -0.04538964189492445, 0, 0.125};
+ Point(84) = {0.4029282541416501, -0.05002451391298904, 0, 0.125};
+ Point(85) = {0.4112465796735204, -0.05450237026215737, 0, 0.125};
+ Point(86) = {0.4191740111683733, -0.05881376343890812, 0, 0.125};
+ Point(87) = {0.4267046786777481, -0.06294951494382847, 0, 0.125};
+ Point(88) = {0.4338330828434404, -0.06690074281456823, 0, 0.125};
+ Point(89) = {0.4405540896772232, -0.07065888921378868, 0, 0.125};
+ Point(90) = {0.4468629247542237, -0.07421574789251445, 0, 0.125};
+ Point(91) = {0.4527551669150955, -0.0775634913396257, 0, 0.125};
+ Point(92) = {0.4582267415819197, -0.08069469742118066, 0, 0.125};
+ Point(93) = {0.4632739138007936, -0.08360237530891265, 0, 0.125};
+ Point(94) = {0.4678932811302005, -0.08627999049569551, 0, 0.125};
+ Point(95) = {0.4720817664982195, -0.08872148869699745, 0, 0.125};
+ Point(96) = {0.4758366111533843, -0.09092131844134463, 0, 0.125};
+ Point(97) = {0.4791553678333992, -0.09287445215953141, 0, 0.125};
+ Point(98) = {0.4820358942729613, -0.09457640559161551, 0, 0.125};
+ Point(99) = {0.4844763471666588, -0.09602325534252773, 0, 0.125};
+ Point(100) = {0.4864751766953637, -0.09721165443119822, 0, 0.125};
+ Point(101) = {0.4880311217148797, -0.09813884569428721, 0, 0.125};
+ Point(102) = {0.4891432056939881, -0.09880267292366274, 0, 0.125};
+ Point(103) = {0.4898107334756874, -0.09920158963645126, 0, 0.125};
+ Point(104) = {0.4900332889206208, -0.09933466539753058, 0, 0.125};
+ Point(105) = {0.4897824225031319, -0.09926905587549506, 0, 0.125};
+ Point(106) = {0.4890301110661922, -0.09907236506934192, 0, 0.125};
+ Point(107) = {0.4877772173496635, -0.09874500608402761, 0, 0.125};
+ Point(108) = {0.48602517690576, -0.09828766683852558, 0, 0.125};
+ Point(109) = {0.4837759946062035, -0.09770130916007558, 0, 0.125};
+ Point(110) = {0.4810322398085871, -0.09698716747297723, 0, 0.125};
+ Point(111) = {0.4777970402368822, -0.09614674703990023, 0, 0.125};
+ Point(112) = {0.4740740746447117, -0.09518182170326678, 0, 0.125};
+ Point(113) = {0.4698675643422793, -0.09409443106501386, 0, 0.125};
+ Point(114) = {0.4651822636784212, -0.09288687703518478, 0, 0.125};
+ Point(115) = {0.460023449577924, -0.09156171967354482, 0, 0.125};
+ Point(116) = {0.4543969102408585, -0.09012177224394632, 0, 0.125};
+ Point(117) = {0.4483089331151018, -0.08857009539864649, 0, 0.125};
+ Point(118) = {0.4417662922553667, -0.08690999040934186, 0, 0.125};
+ Point(119) = {0.4347762351819332, -0.0851449913634191, 0, 0.125};
+ Point(120) = {0.4273464693498908, -0.08327885624791403, 0, 0.125};
+ Point(121) = {0.419485148335155, -0.08131555684993674, 0, 0.125};
+ Point(122) = {0.411200857836944, -0.07925926741086739, 0, 0.125};
+ Point(123) = {0.4025026015879757, -0.07711435198240155, 0, 0.125};
+ Point(124) = {0.3933997872536054, -0.07488535044544484, 0, 0.125};
+ Point(125) = {0.3839022123897198, -0.07257696316779733, 0, 0.125};
+ Point(126) = {0.3740200505167618, -0.07019403429336624, 0, 0.125};
+ Point(127) = {0.3637638373540689, -0.06774153367408606, 0, 0.125};
+ Point(128) = {0.3531444572451353, -0.06522453747557577, 0, 0.125};
+ Point(129) = {0.3421731297908021, -0.06264820750853495, 0, 0.125};
+ Point(130) = {0.3308613966940724, -0.06001776935966011, 0, 0.125};
+ Point(131) = {0.3192211088076166, -0.05733848941811218, 0, 0.125};
+ Point(132) = {0.3072644133633567, -0.05461565091590426, 0, 0.125};
+ Point(133) = {0.2950037413531683, -0.05185452912263369, 0, 0.125};
+ Point(134) = {0.2824517950208982, -0.04906036585632723, 0, 0.125};
+ Point(135) = {0.2696215354188702, -0.04623834349241404, 0, 0.125};
+ Point(136) = {0.2565261699769623, -0.04339355867155523, 0, 0.125};
+ Point(137) = {0.2431791400293651, -0.04053099592384862, 0, 0.125};
+ Point(138) = {0.2295941082432855, -0.03765550144139543, 0, 0.125};
+ Point(139) = {0.2157849458952252, -0.03477175724299444, 0, 0.125};
+ Point(140) = {0.2017657199439165, -0.03188425598348005, 0, 0.125};
+ Point(141) = {0.187550679854507, -0.02899727666564914, 0, 0.125};
+ Point(142) = {0.1731542441359161, -0.02611486151457043, 0, 0.125};
+ Point(143) = {0.1585909865622793, -0.02324079427214604, 0, 0.125};
+ Point(144) = {0.1438756220597465, -0.02037858016395433, 0, 0.125};
+ Point(145) = {0.129022992251319, -0.0175314277805827, 0, 0.125};
+ Point(146) = {0.1140480506645569, -0.01470223310184333, 0, 0.125};
+ Point(147) = {0.09896584761949168, -0.01189356587453844, 0, 0.125};
+ Point(148) = {0.08379151482656089, -0.009107658532933174, 0, 0.125};
+ Point(149) = {0.06854024973648176, -0.006346397826038436, 0, 0.125};
+ Point(150) = {0.05322729969528361, -0.003611319287478529, 0, 0.125};
+ Point(151) = {0.03786794596792287, -0.00090360465249055, 0, 0.125};
+ Point(152) = {0.0224774877026287, 0.00177591770710904, 0, 0.125};
+ Point(153) = {0.007071225915134205, 0.004426769294862437, 0, 0.125};
+ Point(154) = {-0.00833555242305456, 0.007048814950562587, 0, 0.125};
+ Point(155) = {-0.02372759010533726, 0.009642253300220296, 0, 0.125};
+ Point(156) = {-0.03908967513210498, 0.01220760427359278, 0, 0.125};
+ Point(157) = {-0.05440665578848514, 0.01474569380579989, 0, 0.125};
+ Point(158) = {-0.06966345527617318, 0.01725763587663899, 0, 0.125};
+ Point(159) = {-0.08484508582421563, 0.01974481207672138, 0, 0.125};
+ Point(160) = {-0.09987987792382108, 0.02219618763023203, 0, 0.125};
+ Point(161) = {-0.1145078729404739, 0.02450371976411331, 0, 0.125};
+ Point(162) = {-0.1290321771824579, 0.0267015185742735, 0, 0.125};
+ Point(163) = {-0.143440065923266, 0.02879471001709845, 0, 0.125};
+ Point(164) = {-0.1577189448447794, 0.03078883518202784, 0, 0.125};
+ Point(165) = {-0.1718563428491159, 0.03268980457290044, 0, 0.125};
+ Point(166) = {-0.1858399037768357, 0.03450385196323842, 0, 0.125};
+ Point(167) = {-0.1996573773370766, 0.03623748825421298, 0, 0.125};
+ Point(168) = {-0.2132966095779342, 0.03789745574015834, 0, 0.125};
+ Point(169) = {-0.2267455332406906, 0.0394906831577609, 0, 0.125};
+ Point(170) = {-0.2399921583489679, 0.04102424186233269, 0, 0.125};
+ Point(171) = {-0.2530245633834605, 0.04250530343879837, 0, 0.125};
+ Point(172) = {-0.2658308873846617, 0.04394109901707172, 0, 0.125};
+ Point(173) = {-0.2783993233102972, 0.04533888052223981, 0, 0.125};
+ Point(174) = {-0.2907181129514687, 0.04670588405019788, 0, 0.125};
+ Point(175) = {-0.3027755436824813, 0.0480492955198111, 0, 0.125};
+ Point(176) = {-0.3145599472847223, 0.04937621871394801, 0, 0.125};
+ Point(177) = {-0.3260597010456697, 0.05069364578437131, 0, 0.125};
+ Point(178) = {-0.337263231291058, 0.05200843025992359, 0, 0.125};
+ Point(179) = {-0.3481590194623916, 0.05332726256406103, 0, 0.125};
+ Point(180) = {-0.3587356108043638, 0.05465664801682354, 0, 0.125};
+ Point(181) = {-0.3689816256782782, 0.0560028872679817, 0, 0.125};
+ Point(182) = {-0.3788857734692287, 0.05737205908247899, 0, 0.125};
+ Point(183) = {-0.3884368690074614, 0.05877000537646382, 0, 0.125};
+ Point(184) = {-0.3976238513788748, 0.06020231838219783, 0, 0.125};
+ Point(185) = {-0.40643580495675, 0.06167432980291591, 0, 0.125};
+ Point(186) = {-0.4148619824472646, 0.06319110180426264, 0, 0.125};
+ Point(187) = {-0.4228918297057104, 0.06475741967717524, 0, 0.125};
+ Point(188) = {-0.43051501204915, 0.06637778599795482, 0, 0.125};
+ Point(189) = {-0.4377214417649294, 0.06805641610468524, 0, 0.125};
+ Point(190) = {-0.4445013064933708, 0.06979723470503821, 0, 0.125};
+ Point(191) = {-0.4508450981473512, 0.07160387342876083, 0, 0.125};
+ Point(192) = {-0.4567436420215075, 0.073479669138689, 0, 0.125};
+ Point(193) = {-0.4621881257395756, 0.07542766281688272, 0, 0.125};
+ Point(194) = {-0.4671701276898881, 0.07745059884734995, 0, 0.125};
+ Point(195) = {-0.471681644606229, 0.07955092452372269, 0, 0.125};
+ Point(196) = {-0.4757151179639407, 0.0817307896190848, 0, 0.125};
+ Point(197) = {-0.4792634588791559, 0.0839920458658267, 0, 0.125};
+ Point(198) = {-0.4823200712220043, 0.08633624620581726, 0, 0.125};
+ Point(199) = {-0.4848788726822436, 0.08876464368523246, 0, 0.125};
+ Point(200) = {-0.4869343135575803, 0.09127818988394577, 0, 0.125};
+ Point(201) = {-0.4884813930704814, 0.09387753278635144, 0, 0.125};
+ Point(202) = {-0.4895156730580155, 0.09656301401871749, 0, 0.125};
+
+ // splines of the airfoil
+ Spline(5) = {5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104};
+ Spline(6) = {104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,5};
+
+ // airfoil
+ Line Loop(9) = {5, 6};
+ // complete domain
+ Plane Surface(1) = {8, 9};
+
+ // labeling of the boundary parts
+ Physical Line(1) = {4};      // inflow
+ Physical Line(2) = {2};      // outflow
+ Physical Line(3) = {1, 3};   // airfoil
+ Physical Line(4) = {5, 6};   // upper/lower wall
+ Physical Surface(1) = {10};
From which we can construct a mesh like this: mesh_screenshot
The first four points define the bounding box = (near-field) domain:
  // outer bounding box
+Point(1) = {-1.25, -0.5, 0, 1.0};
+Point(2) = {1.25, -0.5, 0, 1.0};
+Point(3) = {1.25, 0.5, 0, 1.0};
+Point(4) = {-1.25, 0.5, 0, 1.0};
which is constructed from connecting the points in lines:
// outer box
+Line(1) = {1, 2};
+Line(2) = {2, 3};
+Line(3) = {3, 4};
+Line(4) = {4, 1};
+// outer box
+Line Loop(8) = {1, 2, 3, 4};
This is followed by a couple (in principle optional) settings where the most important one is
// Insist on quads instead of default triangles
+Mesh.RecombineAll = 1;
which forces gmsh to generate quadrilateral elements instead of the default triangles. This is strictly required to be able to use the mesh later with p4est, which supports only straight-sided quads, i.e., C2D4, CPS4, S4 in 2D and C3D in 3D. See for more details the (short) documentation on the interaction of p4est with .inp files. In principle, it should also be possible to use the recombine function of gmsh to convert the triangles to quads, but this is observed to be less robust than enforcing quads from the beginning.
Then the airfoil is defined by a set of points:
// points of the airfoil contour
+ Point(5) = {-0.4900332889206208, 0.09933466539753061, 0, 0.125};
+ Point(6) = {-0.4900274857651495, 0.1021542752054094, 0, 0.125};
+ ...
which are connected by splines for the upper and lower part of the airfoil:
// splines of the airfoil
+ Spline(5) = {5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,
+              ...
+              96,97,98,99,100,101,102,103,104};
+ Spline(6) = {104,105,106,107,108,109,110,111,112,113,114,115,
+              ...
+              200,201,202,5};
which are then connected to form a single line loop for easy physical group assignment:
// airfoil
+ Line Loop(9) = {5, 6};
At the end of the file the physical groups are defined:
// labeling of the boundary parts
+ Physical Line(1) = {4};      // Inflow. Label in Abaqus .inp file: PhysicalLine1
+ Physical Line(2) = {2};      // Outflow. Label in Abaqus .inp file: PhysicalLine2
+ Physical Line(3) = {1, 3};   // Upper and lower wall/farfield/... Label in Abaqus .inp file: PhysicalLine3
+ Physical Line(4) = {5, 6};   // Airfoil. Label in Abaqus .inp file: PhysicalLine4
which are crucial for the correct assignment of boundary conditions in Trixi.jl. In particular, it is the responsibility of a user to keep track on the physical boundary names between the mesh generation and assignment of boundary condition functions in an elixir.
After opening this file in gmsh, meshing the geometry and exporting to Abaqus .inp format, we can have a look at the input file:
*Heading
+ <something depending on gmsh>
+*NODE
+1, -1.25, -0.5, 0
+2, 1.25, -0.5, 0
+3, 1.25, 0.5, 0
+4, -1.25, 0.5, 0
+...
+******* E L E M E N T S *************
+*ELEMENT, type=T3D2, ELSET=Line1
+1, 1, 7
+...
+*ELEMENT, type=CPS4, ELSET=Surface1
+191, 272, 46, 263, 807
+...
+*NSET,NSET=PhysicalLine1
+1, 4, 52, 53, 54, 55, 56, 57, 58,
+*NSET,NSET=PhysicalLine2
+2, 3, 26, 27, 28, 29, 30, 31, 32,
+*NSET,NSET=PhysicalLine3
+1, 2, 3, 4, 7, 8, 9, 10, 11, 12,
+13, 14, 15, 16, 17, 18, 19, 20, 21, 22,
+23, 24, 25, 33, 34, 35, 36, 37, 38, 39,
+40, 41, 42, 43, 44, 45, 46, 47, 48, 49,
+50, 51,
+*NSET,NSET=PhysicalLine4
+5, 6, 59, 60, 61, 62, 63, 64, 65, 66,
+67, 68, 69, 70, 71, 72, 73, 74, 75, 76,
+77, 78, 79, 80, 81, 82, 83, 84, 85, 86,
+87, 88, 89, 90, 91, 92, 93, 94, 95, 96,
+97, 98, 99, 100, 101, 102, 103, 104, 105, 106,
+107, 108, 109, 110, 111, 112, 113, 114, 115, 116,
+117, 118, 119, 120, 121, 122, 123, 124, 125, 126,
+127, 128, 129, 130, 131, 132, 133, 134, 135, 136,
+137, 138, 139, 140, 141, 142, 143, 144, 145, 146,
+147, 148, 149, 150, 151, 152, 153, 154, 155, 156,
+157, 158, 159, 160, 161, 162, 163, 164, 165, 166,
+167, 168, 169, 170, 171, 172, 173, 174, 175, 176,
+177, 178, 179, 180, 181, 182, 183, 184, 185, 186,
+187, 188, 189, 190,
First, the coordinates of the nodes are listed, followed by the elements. Note that gmsh exports also line elements of type T3D2 which are ignored by p4est. The relevant elements in 2D which form the gridcells are of type CPS4 which are defined by their four corner nodes. This is followed by the nodesets encoded via *NSET which are used to assign boundary conditions in Trixi.jl. Trixi.jl parses the .inp file and assigns the edges (in 2D, surfaces in 3D) of elements to the corresponding boundary condition based on the supplied boundary_symbols that have to be supplied to the P4estMesh constructor:
# boundary symbols
+boundary_symbols = [:PhysicalLine1, :PhysicalLine2, :PhysicalLine3, :PhysicalLine4]
+mesh = P4estMesh{2}(mesh_file, polydeg = polydeg, boundary_symbols = boundary_symbols)
The same boundary symbols have then also be supplied to the semidiscretization alongside the corresponding physical boundary conditions:
# Supersonic inflow boundary condition.
+# Calculate the boundary flux entirely from the external solution state, i.e., set
+# external solution state values for everything entering the domain.
+@inline function boundary_condition_supersonic_inflow(u_inner,
+                                                      normal_direction::AbstractVector,
+                                                      x, t, surface_flux_function,
+                                                      equations::CompressibleEulerEquations2D)
+    u_boundary = initial_condition_mach2_flow(x, t, equations)
+    flux = Trixi.flux(u_boundary, normal_direction, equations)
+
+    return flux
+end
+
+# Supersonic outflow boundary condition.
+# Calculate the boundary flux entirely from the internal solution state. Analogous to supersonic inflow
+# except all the solution state values are set from the internal solution as everything leaves the domain
+@inline function boundary_condition_supersonic_outflow(u_inner,
+                                                       normal_direction::AbstractVector, x,
+                                                       t,
+                                                       surface_flux_function,
+                                                       equations::CompressibleEulerEquations2D)
+flux = Trixi.flux(u_inner, normal_direction, equations)
+
+boundary_conditions = Dict(:PhysicalLine1 => boundary_condition_supersonic_inflow, # Left boundary
+                           :PhysicalLine2 => boundary_condition_supersonic_outflow, # Right boundary
+                           :PhysicalLine3 => boundary_condition_supersonic_outflow, # Top and bottom boundary
+                           :PhysicalLine4 => boundary_condition_slip_wall) # Airfoil
+
+semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,
+                                    boundary_conditions = boundary_conditions)
Note that you have to supply the boundary_symbols keyword to the P4estMesh constructor to select the boundaries from the available nodesets in the .inp file. If the boundary_symbols keyword is not supplied, all boundaries will be assigned to the default set :all.
Package versions
These results were obtained using the following versions.
using InteractiveUtils
+versioninfo()
+
+using Pkg
+Pkg.status(["Trixi", "OrdinaryDiffEq", "Plots", "Download"],
+           mode=PKGMODE_MANIFEST)
Julia Version 1.9.4
+Commit 8e5136fa297 (2023-11-14 08:46 UTC)
+Build Info:
+  Official https://julialang.org/ release
+Platform Info:
+  OS: Linux (x86_64-linux-gnu)
+  CPU: 4 × AMD EPYC 7763 64-Core Processor
+  WORD_SIZE: 64
+  LIBM: libopenlibm
+  LLVM: libLLVM-14.0.6 (ORCJIT, znver3)
+  Threads: 1 on 4 virtual cores
+Environment:
+  JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
+Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
+⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
+⌅ [91a5bcdd] Plots v1.39.0
+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`
+Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.
GitHub
1: Introduction to DG methods
  
This tutorial is about how to set up a simple way to approximate the solution of a hyperbolic partial differential equation. First, we will implement a basic and naive algorithm. Then, we will use predefined features from Trixi.jl to show how you can use Trixi.jl on your own.
We will implement the scalar linear advection equation in 1D with the advection velocity $1$.
\[u_t + u_x = 0,\; \text{for} \;t\in \mathbb{R}^+, x\in\Omega=[-1,1]\]
We define the domain $\Omega$ by setting the boundaries.
coordinates_min = -1.0 # minimum coordinate
+3 Introduction to DG methods · Trixi.jl
GitHub
3: Introduction to DG methods
  
This tutorial is about how to set up a simple way to approximate the solution of a hyperbolic partial differential equation. First, we will implement a basic and naive algorithm. Then, we will use predefined features from Trixi.jl to show how you can use Trixi.jl on your own.
We will implement the scalar linear advection equation in 1D with the advection velocity $1$.
\[u_t + u_x = 0,\; \text{for} \;t\in \mathbb{R}^+, x\in\Omega=[-1,1]\]
We define the domain $\Omega$ by setting the boundaries.
coordinates_min = -1.0 # minimum coordinate
 coordinates_max = 1.0  # maximum coordinate
1.0
We assume periodic boundaries and the following initial condition.
initial_condition_sine_wave(x) = 1.0 + 0.5 * sin(pi * x)
initial_condition_sine_wave (generic function with 1 method)
The discontinuous Galerkin collocation spectral element method (DGSEM)
i. Discretization of the physical domain
To improve precision we want to approximate the solution on small parts of the physical domain. So, we split the domain $\Omega=[-1, 1]$ into elements $Q_l$ of length $dx$.
n_elements = 16 # number of elements
 
 dx = (coordinates_max - coordinates_min) / n_elements # length of one element
0.125
To make the calculation more efficient and storing less information, we transform each element $Q_l$ with center point $x_l$ to a reference element $E=[-1, 1]$
\[Q_l=\Big[x_l-\frac{dx}{2}, x_l+\frac{dx}{2}\Big] \underset{x(\xi)}{\overset{\xi(x)}{\rightleftarrows}} [-1, 1].\]
So, for every element the transformation from the reference domain to the physical domain is defined by
\[x(\xi) = x_l + \frac{dx}{2} \xi,\; \xi\in[-1, 1]\]
Therefore,
\[\begin{align*}
@@ -39,7 +39,7 @@
  0.945837  0.759744  0.610228     1.49706  1.43849  1.31317  1.14018
  0.859825  0.686826  0.561506     1.47995  1.38977  1.24026  1.05416
  0.808658  0.646447  0.53806      1.46194  1.35355  1.19134  1.0
To have a look at the initial sinus curve, we plot it.
using Plots
-plot(vec(x), vec(u0), label="initial condition", legend=:topleft)
Example block output
iii. Variational formulation
After defining the equation and initial condition, we want to implement an algorithm to approximate the solution.
From now on, we only write $u$ instead of $u^{Q_l}$ for simplicity, but consider that all the following calculation only concern one element. Multiplying the new equation with the smooth Lagrange polynomials $\{l_i\}_{i=0}^N$ (test functions) and integrating over the reference element $E=[-1,1]$, we get the variational formulation of our transformed partial differential equation for $i=0,...,N$:
\[\begin{align*}
+plot(vec(x), vec(u0), label="initial condition", legend=:topleft)
Example block output

iii. Variational formulation

After defining the equation and initial condition, we want to implement an algorithm to approximate the solution.

From now on, we only write $u$ instead of $u^{Q_l}$ for simplicity, but consider that all the following calculation only concern one element. Multiplying the new equation with the smooth Lagrange polynomials $\{l_i\}_{i=0}^N$ (test functions) and integrating over the reference element $E=[-1,1]$, we get the variational formulation of our transformed partial differential equation for $i=0,...,N$:

\[\begin{align*} \int_{-1}^1 \Big( \frac{dx}{2} \dot{u}(\xi, t) + u'(\xi, t) \Big) l_i(\xi)d\xi &= \underbrace{\frac{dx}{2} \int_{-1}^1 \dot{u}(\xi, t) l_i(\xi)d\xi}_{\text{Term I}} + \underbrace{\int_{-1}^1 u'(\xi, t) l_i(\xi)d\xi}_{\text{Term II}} = 0 \end{align*}\]

We deal with the two terms separately. We write $\int_{-1, N}^1 \;\cdot\; d\xi$ for the approximation of the integral using numerical quadrature with $N+1$ basis points. We use the Gauss-Lobatto nodes again. The numerical scalar product $\langle\cdot, \cdot\rangle_N$ is defined by $\langle f, g\rangle_N := \int_{-1, N}^1 f(\xi) g(\xi) d\xi$.

Term I:

In the following calculation we approximate the integral numerically with quadrature on the Gauss-Lobatto nodes $\{\xi_i\}_{i=0}^N$ and then use the Kronecker property of the Lagrange polynomials. This approach of using the same nodes for the interpolation and quadrature is called collocation.

\[\begin{align*} @@ -74,7 +74,7 @@ -0.5 1.54508 -4.04508 3.0

To show why this matrix is called the derivative matrix, we go back to our example $f(x)=x^3$. We calculate the derivation of $f$ at the Gauss-Lobatto nodes $\{\xi_k\}_{k=0}^N$ with $N=8$.

\[f'|_{x=\xi_k} = \Big( \sum_{j=0}^8 f(\xi_j) l_j(x) \Big)'|_{x=\xi_k} = \sum_{j=0}^8 f(\xi_j) l_j'(\xi_k) = \sum_{j=0}^8 f(\xi_j) D_{kj}\]

for $k=0,...,N$ and therefore, $\underline{f}' = D \underline{f}$.

basis_N8 = LobattoLegendreBasis(8)
 plot(vec(x), x -> 3 * x^2, label="f'", lw=2)
-scatter!(basis_N8.nodes, basis_N8.derivative_matrix * basis_N8.nodes.^3, label="Df", lw=3)
Example block output

Combining the volume term for every $i=0,...,N$ results in

\[\int_{-1}^1 u \underline{l'} d\xi = - D^T M \underline{u}(t)\]

Putting all parts together we get the following equation for the element $Q_l$

\[\frac{dx}{2} M \underline{\dot{u}}(t) = - B \underline{u}^*(t) + D^T M \underline{u}(t)\]

or equivalent

\[\underline{\dot{u}}^{Q_l}(t) = \frac{2}{dx} \Big[ - M^{-1} B \underline{u}^{{Q_l}^*}(t) + M^{-1} D^T M \underline{u}^{Q_l}(t)\Big].\]

This is called the weak form of the DGSEM.

Note: For every element $Q_l$ we get a system of $N+1$ ordinary differential equations to calculate $N+1$ coefficients. Since the numerical flux $u^*$ is depending on extern values at the interfaces, the equation systems of adjacent elements are weakly linked.

iv. Numerical flux

As mentioned above, we still have to handle the problem of different values at the same point at the interfaces. This happens with the ideas of the numerical flux $f^*(u)=u^*$. The role of $f^*$ might seem minor in this simple example, but is important for more complicated problems. There are two values at the same spatial coordinate. Let's say we are looking at the interface between the elements $Q_l$ and $Q_{l+1}$, while both elements got $N+1$ nodes as defined before. We call the first value of the right element $u_R=u_0^{Q_{l+1}}$ and the last one of the left element $u_L=u_N^{Q_l}$. So, for the value of the numerical flux on that interface the following holds

\[u^* = u^*(u_L, u_R).\]

These values are interpreted as start values of a so-called Riemann problem. There are many different (approximate) Riemann solvers available and useful for different problems. We will use the local Lax-Friedrichs flux.

surface_flux = flux_lax_friedrichs
FluxLaxFriedrichs(max_abs_speed_naive)

The only missing ingredient is the flux calculation at the boundaries $-1$ and $+1$.

\[u^{{Q_{first}}^*}\big|_{-1} = u^{{Q_{first}}^*}\big|_{-1}(u^{bound}(-1), u_R) +scatter!(basis_N8.nodes, basis_N8.derivative_matrix * basis_N8.nodes.^3, label="Df", lw=3)Example block output

Combining the volume term for every $i=0,...,N$ results in

\[\int_{-1}^1 u \underline{l'} d\xi = - D^T M \underline{u}(t)\]

Putting all parts together we get the following equation for the element $Q_l$

\[\frac{dx}{2} M \underline{\dot{u}}(t) = - B \underline{u}^*(t) + D^T M \underline{u}(t)\]

or equivalent

\[\underline{\dot{u}}^{Q_l}(t) = \frac{2}{dx} \Big[ - M^{-1} B \underline{u}^{{Q_l}^*}(t) + M^{-1} D^T M \underline{u}^{Q_l}(t)\Big].\]

This is called the weak form of the DGSEM.

Note: For every element $Q_l$ we get a system of $N+1$ ordinary differential equations to calculate $N+1$ coefficients. Since the numerical flux $u^*$ is depending on extern values at the interfaces, the equation systems of adjacent elements are weakly linked.

iv. Numerical flux

As mentioned above, we still have to handle the problem of different values at the same point at the interfaces. This happens with the ideas of the numerical flux $f^*(u)=u^*$. The role of $f^*$ might seem minor in this simple example, but is important for more complicated problems. There are two values at the same spatial coordinate. Let's say we are looking at the interface between the elements $Q_l$ and $Q_{l+1}$, while both elements got $N+1$ nodes as defined before. We call the first value of the right element $u_R=u_0^{Q_{l+1}}$ and the last one of the left element $u_L=u_N^{Q_l}$. So, for the value of the numerical flux on that interface the following holds

\[u^* = u^*(u_L, u_R).\]

These values are interpreted as start values of a so-called Riemann problem. There are many different (approximate) Riemann solvers available and useful for different problems. We will use the local Lax-Friedrichs flux.

surface_flux = flux_lax_friedrichs
FluxLaxFriedrichs(max_abs_speed_naive)

The only missing ingredient is the flux calculation at the boundaries $-1$ and $+1$.

\[u^{{Q_{first}}^*}\big|_{-1} = u^{{Q_{first}}^*}\big|_{-1}(u^{bound}(-1), u_R) \quad\text{and}\quad u^{{Q_{last}}^*}\big|^1 = u^{{Q_{last}}^*}\big|^1(u_L, u^{bound}(1))\]

The boundaries are periodic, which means that the last value of the last element $u^{Q_{last}}_N$ is used as $u_L$ at the first interface and accordingly for the other boundary.

Now, we implement a function, that calculates $\underline{\dot{u}}^{Q_l}$ for the given matrices, $\underline{u}$ and $\underline{u}^*$.

function rhs!(du, u, x, t)
     # Reset du and flux matrix
@@ -121,7 +121,7 @@
 
 sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)
 
-plot(vec(x), vec(sol.u[end]), label="solution at t=$(tspan[2])", legend=:topleft, lw=3)
Example block output

Alternative Implementation based on Trixi.jl

Now, we implement the same example. But this time, we directly use the functionality that Trixi.jl provides.

using Trixi, OrdinaryDiffEq, Plots

First, define the equation with a advection_velocity of 1.

advection_velocity = 1.0
+plot(vec(x), vec(sol.u[end]), label="solution at t=$(tspan[2])", legend=:topleft, lw=3)
Example block output

Alternative Implementation based on Trixi.jl

Now, we implement the same example. But this time, we directly use the functionality that Trixi.jl provides.

using Trixi, OrdinaryDiffEq, Plots

First, define the equation with a advection_velocity of 1.

advection_velocity = 1.0
 equations = LinearScalarAdvectionEquation1D(advection_velocity)
┌──────────────────────────────────────────────────────────────────────────────────────────────────┐
 │ LinearScalarAdvectionEquation1D                                                                  │
 │ ═══════════════════════════════                                                                  │
@@ -164,7 +164,7 @@
 └──────────────────────────────────────────────────────────────────────────────────────────────────┘

Again, combining all definitions and the function that calculates the right-hand side, we define the ODE and solve it until t=2 with OrdinaryDiffEq's solve function and the Runge-Kutta method RDPK3SpFSAL49().

tspan = (0.0, 2.0)
 ode_trixi  = semidiscretize(semi, tspan)
 
-sol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);

We add a plot of the new approximated solution to the one calculated before.

plot!(sol_trixi, label="solution at t=$(tspan[2]) with Trixi.jl", legend=:topleft, linestyle=:dash, lw=2)
Example block output

Summary of the code

To sum up, here is the complete code that we used.

Raw implementation

# basis: Legendre-Gauss-Lobatto
+sol_trixi = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);

We add a plot of the new approximated solution to the one calculated before.

plot!(sol_trixi, label="solution at t=$(tspan[2]) with Trixi.jl", legend=:topleft, linestyle=:dash, lw=2)
Example block output

Summary of the code

To sum up, here is the complete code that we used.

Raw implementation

# basis: Legendre-Gauss-Lobatto
 using Trixi, LinearAlgebra, OrdinaryDiffEq, Plots
 polydeg = 3 #= polynomial degree =#
 basis = LobattoLegendreBasis(polydeg)
@@ -244,7 +244,7 @@
 # solve
 sol = solve(ode, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...)
 
-plot(vec(x), vec(sol.u[end]), label="solution at t=$(tspan[2])", legend=:topleft, lw=3)
Example block output

Alternative Implementation based on Trixi.jl

using Trixi, OrdinaryDiffEq, Plots
+plot(vec(x), vec(sol.u[end]), label="solution at t=$(tspan[2])", legend=:topleft, lw=3)
Example block output

Alternative Implementation based on Trixi.jl

using Trixi, OrdinaryDiffEq, Plots
 
 # equation with a advection_velocity of `1`.
 advection_velocity = 1.0
@@ -270,7 +270,7 @@
 ode_trixi  = semidiscretize(semi, tspan)
 sol_trixi  = solve(ode_trixi, RDPK3SpFSAL49(); abstol=1.0e-6, reltol=1.0e-6, ode_default_options()...);
 
-plot!(sol_trixi, label="solution at t=$(tspan[2]) with Trixi.jl", legend=:topleft, linestyle=:dash, lw=2)
Example block output

Package versions

These results were obtained using the following versions.

using InteractiveUtils
+plot!(sol_trixi, label="solution at t=$(tspan[2]) with Trixi.jl", legend=:topleft, linestyle=:dash, lw=2)
Example block output

Package versions

These results were obtained using the following versions.

using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -290,6 +290,6 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

+⌅ [91a5bcdd] Plots v1.39.0 + [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl` +Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/shock_capturing/d2d35149.svg b/previews/PR1624/tutorials/shock_capturing/8985776c.svg similarity index 96% rename from previews/PR1624/tutorials/shock_capturing/d2d35149.svg rename to previews/PR1624/tutorials/shock_capturing/8985776c.svg index 263072bbf2e..58a680a8e2d 100644 --- a/previews/PR1624/tutorials/shock_capturing/d2d35149.svg +++ b/previews/PR1624/tutorials/shock_capturing/8985776c.svg @@ -1,45 +1,45 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/shock_capturing/index.html b/previews/PR1624/tutorials/shock_capturing/index.html index e9a63de5400..e4e99b9c0ff 100644 --- a/previews/PR1624/tutorials/shock_capturing/index.html +++ b/previews/PR1624/tutorials/shock_capturing/index.html @@ -1,5 +1,5 @@ -3 Shock capturing with flux differencing and stage limiter · Trixi.jl
GitHub

3: Shock capturing with flux differencing and stage limiter

This tutorial contains a short summary of the idea of shock capturing for DGSEM with flux differencing and its implementation in Trixi.jl. In the second part, an implementation of a positivity preserving limiter is added to the simulation.

Shock capturing with flux differencing

The following rough explanation is on a very basic level. More information about an entropy stable shock-capturing strategy for DGSEM discretizations of advection dominated problems, such as the compressible Euler equations or the compressible Navier-Stokes equations, can be found in Hennemann et al. (2021). In Rueda-Ramírez et al. (2021) you find the extension to the systems with non-conservative terms, such as the compressible magnetohydrodynamics (MHD) equations.

The strategy for a shock-capturing method presented by Hennemann et al. is based on a hybrid blending of a high-order DG method with a low-order variant. The low-order subcell finite volume (FV) method is created directly with the Legendre-Gauss-Lobatto (LGL) nodes already used for the high-order DGSEM. Then, the final method is a convex combination with regulating indicator $\alpha$ of these two methods.

Since the surface integral is equal for both the DG and the subcell FV method, only the volume integral divides between the two methods.

This strategy for the volume integral is implemented in Trixi.jl under the name of VolumeIntegralShockCapturingHG with the three parameters of the indicator and the volume fluxes for the DG and the subcell FV method.

Note, that the DG method is based on the flux differencing formulation. Hence, you have to use a two-point flux, such as flux_ranocha, flux_shima_etal, flux_chandrashekar or flux_kennedy_gruber, for the DG volume flux. We would recommend to use the entropy conserving flux flux_ranocha by Ranocha (2018) for the compressible Euler equations.

volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;
+5 Shock capturing with flux differencing and stage limiter · Trixi.jl
GitHub
5: Shock capturing with flux differencing and stage limiter
  
This tutorial contains a short summary of the idea of shock capturing for DGSEM with flux differencing and its implementation in Trixi.jl. In the second part, an implementation of a positivity preserving limiter is added to the simulation.
Shock capturing with flux differencing
The following rough explanation is on a very basic level. More information about an entropy stable shock-capturing strategy for DGSEM discretizations of advection dominated problems, such as the compressible Euler equations or the compressible Navier-Stokes equations, can be found in Hennemann et al. (2021). In Rueda-Ramírez et al. (2021) you find the extension to the systems with non-conservative terms, such as the compressible magnetohydrodynamics (MHD) equations.
The strategy for a shock-capturing method presented by Hennemann et al. is based on a hybrid blending of a high-order DG method with a low-order variant. The low-order subcell finite volume (FV) method is created directly with the Legendre-Gauss-Lobatto (LGL) nodes already used for the high-order DGSEM. Then, the final method is a convex combination with regulating indicator $\alpha$ of these two methods.
Since the surface integral is equal for both the DG and the subcell FV method, only the volume integral divides between the two methods.
This strategy for the volume integral is implemented in Trixi.jl under the name of VolumeIntegralShockCapturingHG with the three parameters of the indicator and the volume fluxes for the DG and the subcell FV method.
Note, that the DG method is based on the flux differencing formulation. Hence, you have to use a two-point flux, such as flux_ranocha, flux_shima_etal, flux_chandrashekar or flux_kennedy_gruber, for the DG volume flux. We would recommend to use the entropy conserving flux flux_ranocha by Ranocha (2018) for the compressible Euler equations.
volume_integral = VolumeIntegralShockCapturingHG(indicator_sc;
                                                  volume_flux_dg=volume_flux_dg,
                                                  volume_flux_fv=volume_flux_fv)
We now focus on a choice of the shock capturing indicator indicator_sc. A possible indicator is $\alpha_{HG}$ presented by Hennemann et al. (p.10), which depends on the current approximation with modal coefficients $\{m_j\}_{j=0}^N$ of a given variable.
The indicator is calculated for every DG element by itself. First, we calculate a smooth $\alpha$ by
\[\alpha = \frac{1}{1+\exp(-\frac{-s}{\mathbb{T}}(\mathbb{E}-\mathbb{T}))}\]
with the total energy $\mathbb{E}=\max\big(\frac{m_N^2}{\sum_{j=0}^N m_j^2}, \frac{m_{N-1}^2}{\sum_{j=0}^{N-1} m_j^2}\big)$, threshold $\mathbb{T}= 0.5 * 10^{-1.8*(N+1)^{1/4}}$ and parameter $s=ln\big(\frac{1-0.0001}{0.0001}\big)\approx 9.21024$.
For computational efficiency, $\alpha_{min}$ is introduced and used for
\[\tilde{\alpha} = \begin{cases}
 0, & \text{if } \alpha<\alpha_{min}\\
@@ -89,7 +89,7 @@
             save_everystep=false, callback=callbacks);
 
 using Plots
-plot(sol)
Example block output

Package versions

These results were obtained using the following versions.

using InteractiveUtils
+plot(sol)
Example block output

Package versions

These results were obtained using the following versions.

using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -109,6 +109,6 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

+⌅ [91a5bcdd] Plots v1.39.0 + [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl` +Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/structured_mesh_mapping/300da432.svg b/previews/PR1624/tutorials/structured_mesh_mapping/300da432.svg deleted file mode 100644 index 50461ad1e3d..00000000000 --- a/previews/PR1624/tutorials/structured_mesh_mapping/300da432.svg +++ /dev/null @@ -1,9477 +0,0 @@ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - diff --git a/previews/PR1624/tutorials/structured_mesh_mapping/fb53b05b.svg b/previews/PR1624/tutorials/structured_mesh_mapping/5839e9cb.svg similarity index 96% rename from previews/PR1624/tutorials/structured_mesh_mapping/fb53b05b.svg rename to previews/PR1624/tutorials/structured_mesh_mapping/5839e9cb.svg index d3897a6f48b..a028867a0ce 100644 --- a/previews/PR1624/tutorials/structured_mesh_mapping/fb53b05b.svg +++ b/previews/PR1624/tutorials/structured_mesh_mapping/5839e9cb.svg @@ -1,47 +1,47 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - - + + - + - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/structured_mesh_mapping/60cf2592.svg b/previews/PR1624/tutorials/structured_mesh_mapping/60cf2592.svg new file mode 100644 index 00000000000..f9a3925bd3e --- /dev/null +++ b/previews/PR1624/tutorials/structured_mesh_mapping/60cf2592.svg @@ -0,0 +1,9749 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/previews/PR1624/tutorials/structured_mesh_mapping/e3dbbc35.svg b/previews/PR1624/tutorials/structured_mesh_mapping/d7d4dc96.svg similarity index 86% rename from previews/PR1624/tutorials/structured_mesh_mapping/e3dbbc35.svg rename to previews/PR1624/tutorials/structured_mesh_mapping/d7d4dc96.svg index eaa982d072d..6c407982aaf 100644 --- a/previews/PR1624/tutorials/structured_mesh_mapping/e3dbbc35.svg +++ b/previews/PR1624/tutorials/structured_mesh_mapping/d7d4dc96.svg @@ -1,47 +1,47 @@ - + - + - + - + - + - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + - + - + diff --git a/previews/PR1624/tutorials/structured_mesh_mapping/index.html b/previews/PR1624/tutorials/structured_mesh_mapping/index.html index b846b47dc9d..0ee943f6a8e 100644 --- a/previews/PR1624/tutorials/structured_mesh_mapping/index.html +++ b/previews/PR1624/tutorials/structured_mesh_mapping/index.html @@ -1,5 +1,5 @@ -13 Structured mesh with curvilinear mapping · Trixi.jl
GitHub

13: Structured mesh with curvilinear mapping

Here, we want to introduce another mesh type of Trixi.jl. More precisely, this tutorial is about the curved mesh type StructuredMesh supporting curved meshes.

Creating a curved mesh

There are two basic options to define a curved StructuredMesh in Trixi.jl. You can implement curves for the domain boundaries, or alternatively, set up directly the complete transformation mapping. We now present one short example each.

Mesh defined by domain boundary curves

Both examples are based on a semdiscretization of the 2D compressible Euler equations.

using OrdinaryDiffEq
+15 Structured mesh with curvilinear mapping · Trixi.jl
GitHub
15: Structured mesh with curvilinear mapping
  
Here, we want to introduce another mesh type of Trixi.jl. More precisely, this tutorial is about the curved mesh type StructuredMesh supporting curved meshes.
Creating a curved mesh
There are two basic options to define a curved StructuredMesh in Trixi.jl. You can implement curves for the domain boundaries, or alternatively, set up directly the complete transformation mapping. We now present one short example each.
Mesh defined by domain boundary curves
Both examples are based on a semdiscretization of the 2D compressible Euler equations.
using OrdinaryDiffEq
 using Trixi
 
 equations = CompressibleEulerEquations2D(1.4)
┌──────────────────────────────────────────────────────────────────────────────────────────────────┐
@@ -20,7 +20,7 @@
 
   return prim2cons(SVector(rho, v1, v2, p), equations)
 end
-initial_condition = initial_condition_pressure_perturbation
initial_condition_pressure_perturbation (generic function with 1 method)
Initialize every boundary as a boundary_condition_slip_wall.
boundary_conditions = boundary_condition_slip_wall
boundary_condition_slip_wall (generic function with 13 methods)
The approximation setup is an entropy-stable split-form DG method with polydeg=4. We are using the two fluxes flux_ranocha and flux_lax_friedrichs.
solver = DGSEM(polydeg=4, surface_flux=flux_lax_friedrichs,
+initial_condition = initial_condition_pressure_perturbation
initial_condition_pressure_perturbation (generic function with 1 method)
Initialize every boundary as a boundary_condition_slip_wall.
boundary_conditions = boundary_condition_slip_wall
boundary_condition_slip_wall (generic function with 11 methods)
The approximation setup is an entropy-stable split-form DG method with polydeg=4. We are using the two fluxes flux_ranocha and flux_lax_friedrichs.
solver = DGSEM(polydeg=4, surface_flux=flux_lax_friedrichs,
                volume_integral=VolumeIntegralFluxDifferencing(flux_ranocha))
┌──────────────────────────────────────────────────────────────────────────────────────────────────┐
 │ DG{Float64}                                                                                      │
 │ ═══════════                                                                                      │
@@ -66,9 +66,9 @@
             save_everystep=false, callback=callbacks);
 
 using Plots
-plot(sol)
Example block output
pd = PlotData2D(sol)
+plot(sol)
Example block output
pd = PlotData2D(sol)
 plot(pd["p"])
-plot!(getmesh(pd))
Example block output
Mesh directly defined by the transformation mapping
As mentioned before, you can also define the domain for a StructuredMesh by directly setting up a transformation mapping. Here, we want to present a nice mapping, which is often used to test free-stream preservation. Exact free-stream preservation is a crucial property of any numerical method on curvilinear grids. The mapping is a reduced 2D version of the mapping described in Rueda-Ramírez et al. (2021), p.18.
using OrdinaryDiffEq
+plot!(getmesh(pd))
Example block output
Mesh directly defined by the transformation mapping
As mentioned before, you can also define the domain for a StructuredMesh by directly setting up a transformation mapping. Here, we want to present a nice mapping, which is often used to test free-stream preservation. Exact free-stream preservation is a crucial property of any numerical method on curvilinear grids. The mapping is a reduced 2D version of the mapping described in Rueda-Ramírez et al. (2021), p.18.
using OrdinaryDiffEq
 using Trixi
 
 equations = CompressibleEulerEquations2D(1.4)
┌──────────────────────────────────────────────────────────────────────────────────────────────────┐
@@ -122,11 +122,11 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                  0                run time:       7.42000000e-07 s
+ #timesteps:                  0                run time:       1.06200000e-06 s
  Δt:             1.00000000e+00                └── GC time:    0.00000000e+00 s (0.000%)
  sim. time:      0.00000000e+00 (0.000%)       time/DOF/rhs!:         NaN s
                                                PID:                   Inf s
- #DOFs per field:          4096                alloc'd memory:       2990.021 MiB
+ #DOFs per field:          4096                alloc'd memory:       3243.510 MiB
  #elements:                 256
 
  Variable:       rho              rho_v1           rho_v2           rho_e
@@ -139,34 +139,34 @@
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                250                run time:       2.13871747e-01 s
+ #timesteps:                250                run time:       2.09140174e-01 s
  Δt:             2.33008558e-03                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      5.82521395e-01 (58.252%)      time/DOF/rhs!:  3.74963116e-08 s
-                                               PID:            4.16031667e-08 s
- #DOFs per field:          4096                alloc'd memory:       2990.293 MiB
+ sim. time:      5.82521395e-01 (58.252%)      time/DOF/rhs!:  3.66958115e-08 s
+                                               PID:            4.06839751e-08 s
+ #DOFs per field:          4096                alloc'd memory:       3243.782 MiB
  #elements:                 256
 
  Variable:       rho              rho_v1           rho_v2           rho_e
- L2 error:       2.31290602e-15   1.94893683e-14   2.21188565e-14   1.44472757e-14
- Linf error:     2.25375274e-14   2.07972528e-13   2.75501844e-13   2.25597319e-13
- ∑∂S/∂U ⋅ Uₜ :  -1.82954114e-17
+ L2 error:       2.26960092e-15   1.94552593e-14   2.21395377e-14   1.44688368e-14
+ Linf error:     2.25375274e-14   2.07819872e-13   2.77139423e-13   2.25597319e-13
+ ∑∂S/∂U ⋅ Uₜ :   3.42148696e-17
 ────────────────────────────────────────────────────────────────────────────────────────────────────
 
 
 ────────────────────────────────────────────────────────────────────────────────────────────────────
  Simulation running 'CompressibleEulerEquations2D' with DGSEM(polydeg=3)
 ────────────────────────────────────────────────────────────────────────────────────────────────────
- #timesteps:                430                run time:       3.68036600e-01 s
+ #timesteps:                430                run time:       3.59835263e-01 s
  Δt:             3.93286656e-04                └── GC time:    0.00000000e+00 s (0.000%)
- sim. time:      1.00000000e+00 (100.000%)     time/DOF/rhs!:  3.74650359e-08 s
-                                               PID:            4.16202650e-08 s
- #DOFs per field:          4096                alloc'd memory:       2990.371 MiB
+ sim. time:      1.00000000e+00 (100.000%)     time/DOF/rhs!:  3.67055019e-08 s
+                                               PID:            4.06938265e-08 s
+ #DOFs per field:          4096                alloc'd memory:       3243.860 MiB
  #elements:                 256
 
  Variable:       rho              rho_v1           rho_v2           rho_e
- L2 error:       3.20962510e-15   2.14179782e-14   2.83584627e-14   1.56066183e-14
- Linf error:     2.53130850e-14   2.26790808e-13   3.36897177e-13   2.34479103e-13
- ∑∂S/∂U ⋅ Uₜ :  -7.51023760e-17
+ L2 error:       3.15789939e-15   2.14412441e-14   2.85317041e-14   1.56627094e-14
+ Linf error:     2.46469511e-14   2.28150832e-13   3.36675132e-13   2.36255460e-13
+ ∑∂S/∂U ⋅ Uₜ :   7.31182305e-18
 ────────────────────────────────────────────────────────────────────────────────────────────────────
Now, we want to verify the free-stream preservation property and plot the mesh. For the verification, we calculate the absolute difference of the first conservation variable density u[1] and 1.0. To plot this error and the mesh, we are using the visualization feature ScalarPlotData2D, explained in visualization.
error_density = let u = Trixi.wrap_array(sol.u[end], semi)
   abs.(u[1, :, :, :] .- 1.0) # density, x, y, elements
 end
@@ -174,7 +174,7 @@
 
 using Plots
 plot(pd, title="Error in density")
-plot!(getmesh(pd))
Example block output
We observe that the errors in the variable density are at the level of machine accuracy. Moreover, the plot shows the mesh structure resulting from our transformation mapping.
Of course, you can also use other mappings as for instance shifts by $(x, y)$
mapping(xi, eta) = SVector(xi + x, eta + y)
or rotations with a rotation matrix $T$
mapping(xi, eta) = T * SVector(xi, eta).
For more curved mesh mappings, please have a look at some elixirs for StructuredMesh. For another curved mesh type, there is a tutorial about Trixi.jl's unstructured mesh type [UnstructuredMesh2D] and its use of the High-Order Hex-Quad Mesh (HOHQMesh) generator, created and developed by David Kopriva.
Package versions
These results were obtained using the following versions.
using InteractiveUtils
+plot!(getmesh(pd))
Example block output
We observe that the errors in the variable density are at the level of machine accuracy. Moreover, the plot shows the mesh structure resulting from our transformation mapping.
Of course, you can also use other mappings as for instance shifts by $(x, y)$
mapping(xi, eta) = SVector(xi + x, eta + y)
or rotations with a rotation matrix $T$
mapping(xi, eta) = T * SVector(xi, eta).
For more curved mesh mappings, please have a look at some elixirs for StructuredMesh. For another curved mesh type, there is a tutorial about Trixi.jl's unstructured mesh type [UnstructuredMesh2D] and its use of the High-Order Hex-Quad Mesh (HOHQMesh) generator, created and developed by David Kopriva.
Package versions
These results were obtained using the following versions.
using InteractiveUtils
 versioninfo()
 
 using Pkg
@@ -194,6 +194,6 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [91a5bcdd] Plots v1.39.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.

+⌅ [91a5bcdd] Plots v1.39.0
+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`
+Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/time_stepping/index.html b/previews/PR1624/tutorials/time_stepping/index.html index 13cc19f0c66..e3bba0947e3 100644 --- a/previews/PR1624/tutorials/time_stepping/index.html +++ b/previews/PR1624/tutorials/time_stepping/index.html @@ -1,5 +1,5 @@ -15 Explicit time stepping · Trixi.jl
GitHub

15: Explicit time stepping

For the time integration, Trixi.jl uses the package OrdinaryDiffEq.jl from the SciML ecosystem. The interface to this package is the solve(...) function. It always requires an ODE problem and a time integration algorithm as input parameters.

solve(ode, alg; kwargs...)

In Trixi.jl, the ODE problem is created by semidiscretize(semi, tspan) for a semidiscretization semi and the time span tspan. In particular, semidiscretize returns an ODEProblem used by OrdinaryDiffEq.jl.

OrdinaryDiffEq.jl provides many integration algorithms, which are summarized in the documentation. Particularly interesting for Trixi.jl are their strong stability preserving (SSP) methods and low-storage methods. There are some differences regarding the choice of the used time step.

Error-based adaptive step sizes

First, we treat time integration algorithms with adaptive step sizes, such as SSPRK43. It is used in some elixirs, like elixir_euler_colliding_flow.jl or elixir_euler_astro_jet_amr.jl.

Other error-based adaptive integration algorithms are for instance RDPK3SpFSAL35, RDPK3Sp35, RDPK3SpFSAL49, RDPK3Sp49, RDPK3SpFSAL510, RDPK3Sp510.

They already contain an error-based adaptive step size control and heuristics to guess a starting step size. If this heuristic fails in your case, you can specify an appropriately small initial step size as keyword argument dt=... of solve.

If you run Trixi in parallel with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to enable MPI aware error-based adaptive step size control. These keyword arguments are also included in ode_default_options.

CFL-based step size control

The SciML ecosystem also provides time integration algorithms without adaptive time stepping on their own, such as CarpenterKennedy2N54. Moreover, you also can deactivate the automatic adaptivity of adaptive integration algorithms by passing adaptive=false in the solve function.

These algorithms require another way of setting the step size. You have to pass dt=... in the solve function. Without other settings, the simulation uses this fixed time step.

For hyperbolic PDEs, it is natural to use an adaptive CFL-based step size control. Here, the time step is proportional to a ratio of the local measure of mesh spacing $\Delta x_i$ for an element i and the maximum (local) wave speed $\lambda_{\max}$ related to the largest-magnitude eigenvalue of the flux Jacobian of the hyperbolic system.

\[\Delta t_n = \text{CFL} * \min_i \frac{\Delta x_i}{\lambda_{\max}(u_i^n)}\]

We compute $\Delta x_i$ by scaling the element size by a factor of $1/(N+1)$, cf. Gassner and Kopriva (2011), Section 5.

Trixi.jl provides such a CFL-based step size control. It is implemented as the callback StepsizeCallback.

stepsize_callback = StepsizeCallback(; cfl=1.0)

A suitable CFL number depends on many parameters such as the chosen grid, the integration algorithm and the polynomial degree of the spatial DG discretization. So, the optimal number for an example is mostly determined experimentally.

You can add this CFL-based step size control to your simulation like any other callback.

callbacks = CallbackSet(stepsize_callback)
+18 Explicit time stepping · Trixi.jl
GitHub
18: Explicit time stepping
  
For the time integration, Trixi.jl uses the package OrdinaryDiffEq.jl from the SciML ecosystem. The interface to this package is the solve(...) function. It always requires an ODE problem and a time integration algorithm as input parameters.
solve(ode, alg; kwargs...)
In Trixi.jl, the ODE problem is created by semidiscretize(semi, tspan) for a semidiscretization semi and the time span tspan. In particular, semidiscretize returns an ODEProblem used by OrdinaryDiffEq.jl.
OrdinaryDiffEq.jl provides many integration algorithms, which are summarized in the documentation. Particularly interesting for Trixi.jl are their strong stability preserving (SSP) methods and low-storage methods. There are some differences regarding the choice of the used time step.
Error-based adaptive step sizes
First, we treat time integration algorithms with adaptive step sizes, such as SSPRK43. It is used in some elixirs, like elixir_euler_colliding_flow.jl or elixir_euler_astro_jet_amr.jl.
Other error-based adaptive integration algorithms are for instance RDPK3SpFSAL35, RDPK3Sp35, RDPK3SpFSAL49, RDPK3Sp49, RDPK3SpFSAL510, RDPK3Sp510.
They already contain an error-based adaptive step size control and heuristics to guess a starting step size. If this heuristic fails in your case, you can specify an appropriately small initial step size as keyword argument dt=... of solve.
If you run Trixi in parallel with MPI you need to pass internalnorm=ode_norm and you should pass unstable_check=ode_unstable_check to enable MPI aware error-based adaptive step size control. These keyword arguments are also included in ode_default_options.
CFL-based step size control
The SciML ecosystem also provides time integration algorithms without adaptive time stepping on their own, such as CarpenterKennedy2N54. Moreover, you also can deactivate the automatic adaptivity of adaptive integration algorithms by passing adaptive=false in the solve function.
These algorithms require another way of setting the step size. You have to pass dt=... in the solve function. Without other settings, the simulation uses this fixed time step.
For hyperbolic PDEs, it is natural to use an adaptive CFL-based step size control. Here, the time step is proportional to a ratio of the local measure of mesh spacing $\Delta x_i$ for an element i and the maximum (local) wave speed $\lambda_{\max}$ related to the largest-magnitude eigenvalue of the flux Jacobian of the hyperbolic system.
\[\Delta t_n = \text{CFL} * \min_i \frac{\Delta x_i}{\lambda_{\max}(u_i^n)}\]
We compute $\Delta x_i$ by scaling the element size by a factor of $1/(N+1)$, cf. Gassner and Kopriva (2011), Section 5.
Trixi.jl provides such a CFL-based step size control. It is implemented as the callback StepsizeCallback.
stepsize_callback = StepsizeCallback(; cfl=1.0)
A suitable CFL number depends on many parameters such as the chosen grid, the integration algorithm and the polynomial degree of the spatial DG discretization. So, the optimal number for an example is mostly determined experimentally.
You can add this CFL-based step size control to your simulation like any other callback.
callbacks = CallbackSet(stepsize_callback)
 alg = CarpenterKennedy2N54(williamson_condition=false)
 solve(ode, alg;
       dt=1.0 # solve needs some value here but it will be overwritten by the stepsize_callback
@@ -23,5 +23,5 @@
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
 ⌅ [1dea7af3] OrdinaryDiffEq v6.66.0
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
-Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
This page was generated using Literate.jl.

+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`
+Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/tutorials/upwind_fdsbp/index.html b/previews/PR1624/tutorials/upwind_fdsbp/index.html index 1f1539b787c..4d99993c6e3 100644 --- a/previews/PR1624/tutorials/upwind_fdsbp/index.html +++ b/previews/PR1624/tutorials/upwind_fdsbp/index.html @@ -1,5 +1,5 @@ -7 Upwind FD SBP schemes · Trixi.jl
GitHub

7: Upwind FD SBP schemes

General tensor product SBP methods are supported via the DGMulti solver in a reasonably complete way, see the previous tutorial. Nevertheless, there is also experimental support for SBP methods with other solver and mesh types.

The first step is to set up an SBP operator. A classical (central) SBP operator can be created as follows.

using Trixi
+9 Upwind FD SBP schemes · Trixi.jl
GitHub
9: Upwind FD SBP schemes
  
General tensor product SBP methods are supported via the DGMulti solver in a reasonably complete way, see the previous tutorial. Nevertheless, there is also experimental support for SBP methods with other solver and mesh types.
The first step is to set up an SBP operator. A classical (central) SBP operator can be created as follows.
using Trixi
 D_SBP = derivative_operator(SummationByPartsOperators.MattssonNordström2004(),
                             derivative_order=1, accuracy_order=2,
                             xmin=0.0, xmax=1.0, N=11)
SBP first-derivative operator of order 2 on a grid in [0.0, 1.0] using 11 nodes 
@@ -61,5 +61,5 @@
 Environment:
   JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
 Status `~/work/Trixi.jl/Trixi.jl/docs/Manifest.toml`
-  [9f78cca6] SummationByPartsOperators v0.5.52
-  [a7f1ee26] Trixi v0.6.6-pre `~/work/Trixi.jl/Trixi.jl`
This page was generated using Literate.jl.

+  [9f78cca6] SummationByPartsOperators v0.5.53
+  [a7f1ee26] Trixi v0.7.1-pre `~/work/Trixi.jl/Trixi.jl`

This page was generated using Literate.jl.

diff --git a/previews/PR1624/visualization/index.html b/previews/PR1624/visualization/index.html index 1ab7b347a17..fdf79268dea 100644 --- a/previews/PR1624/visualization/index.html +++ b/previews/PR1624/visualization/index.html @@ -1,5 +1,5 @@ -Visualization · Trixi.jl
GitHub

Visualization

There are two possible approaches to visualize results from Trixi.jl: either directly from the REPL using Plots.jl or with ParaView/VisIt by postprocessing Trixi.jl's output files with Trixi2Vtk.

Plots.jl [experimental]

By far the easiest and most convenient plotting approach is to use the powerful Plots.jl package to directly visualize Trixi.jl's results from the REPL. In the following, you will find more information on a number of topics for how to use Plots.jl together with Trixi.jl:

  1. Getting started
  2. Customizing plot results via a PlotData2D object
  3. Plotting a 3D solution as a 2D plot
  4. Creating a 1D plot
  5. Plotting a 2D or 3D solutions as a 1D plot
  6. Visualizing results during a simulation
Note

Plotting via Plots.jl is still considered an experimental feature and might change in any future releases.

Getting started

After running a simulation with Trixi.jl in the REPL, load the Plots package with

julia> using Plots

To visualize the solution, execute

julia> plot(sol)

Here we assume that sol holds the return value of the solve(...) method (with type SciMLBase.ODESolution), which is the default variable name when you use one of the example elixirs. This will generate a grid layout with one subplot for each solution variable, convenient for getting an overview of the current solution:

plot-sol

You can save the resulting file as a PNG image file by calling savefig(...) with an output file name that ends in .png, e.g.,

julia> savefig("solution-overview.png")

In Trixi.jl, two plot types are available: 2D heatmap plots and 1D line plots. If you use plot(sol), Trixi.jl will automatically choose the plot type that fits the dimensions of the sol input: 2D/3D data will be visualized as a heatmap, 1D data as a line plot. For more fine-grained control over what to plot, you can create such an object yourself, which can either be a PlotData2D or a PlotData1D object. For further details on both of these see below:

Customizing plot results via a PlotData2D object

For more fine-grained control over what to plot, first create a PlotData2D object by executing

julia> pd = PlotData2D(sol)
+Visualization · Trixi.jl
GitHub
Visualization
There are two possible approaches to visualize results from Trixi.jl: either directly from the REPL using Plots.jl or with ParaView/VisIt by postprocessing Trixi.jl's output files with Trixi2Vtk.
Plots.jl [experimental]
By far the easiest and most convenient plotting approach is to use the powerful Plots.jl package to directly visualize Trixi.jl's results from the REPL. In the following, you will find more information on a number of topics for how to use Plots.jl together with Trixi.jl:
  1. Getting started
  2. Customizing plot results via a PlotData2D object
  3. Plotting a 3D solution as a 2D plot
  4. Creating a 1D plot
  5. Plotting a 2D or 3D solutions as a 1D plot
  6. Visualizing results during a simulation
Note
Plotting via Plots.jl is still considered an experimental feature and might change in any future releases.
Getting started
After running a simulation with Trixi.jl in the REPL, load the Plots package with
julia> using Plots
To visualize the solution, execute
julia> plot(sol)
Here we assume that sol holds the return value of the solve(...) method (with type SciMLBase.ODESolution), which is the default variable name when you use one of the example elixirs. This will generate a grid layout with one subplot for each solution variable, convenient for getting an overview of the current solution:
plot-sol
You can save the resulting file as a PNG image file by calling savefig(...) with an output file name that ends in .png, e.g.,
julia> savefig("solution-overview.png")
In Trixi.jl, two plot types are available: 2D heatmap plots and 1D line plots. If you use plot(sol), Trixi.jl will automatically choose the plot type that fits the dimensions of the sol input: 2D/3D data will be visualized as a heatmap, 1D data as a line plot. For more fine-grained control over what to plot, you can create such an object yourself, which can either be a PlotData2D or a PlotData1D object. For further details on both of these see below:
Customizing plot results via a PlotData2D object
For more fine-grained control over what to plot, first create a PlotData2D object by executing
julia> pd = PlotData2D(sol)
 julia> pd = PlotData2D(u, semi)
where u is an array containing the solution and semi is the semidiscretization. For example, if PlotData2D(sol.u[2], semi) is specified, this will create a PlotData2D instance from the 2nd saved time-step. If PlotData2D(sol(0.5), semi) is specified, it will construct a PlotData2D instance using OrdinaryDiffEq.jl's interpolation to evaluate the solution at time t=0.5.
This takes the results generated by Trixi.jl and stores them in a data format that can be understood by the Plots package, and pd holds all data relevant for plotting sol. You can pass variable names as strings to pd using a dictionary-like syntax, e.g.,
julia> plot(pd["rho"])
This will create a single 2D heatmap plot of the variable rho:
plot-rho
The default plot type and style can be overridden by passing any additional arguments that are understood by the Plots package. For example, to change the color scheme and add names to the axes, modify the previous command to
julia> plot(pd["rho"], seriescolor = :heat, xguide="x", yguide="y")
to yield
plot-rho-modified
For more details on the various format options for plot, please consult the Plots documentation.
In addition, you can plot the mesh lines on top of the solution variables by calling the getmesh(...) function on the PlotData2D object
julia> plot!(getmesh(pd)) # here we use `plot!` with an `!` to add to the previous plot
which modifies the previous plot to
plot-rho-modified-mesh
By default, PlotData2D will convert the conserved variables to primitive variables, but this can be changed by passing an appropriate conversion function in the solution_variables keyword argument, similar to the behavior of the SaveSolutionCallback:
julia> pd = PlotData2D(sol; solution_variables=cons2cons) # Plot conservative variables
There are several other keyword arguments that influence how the solution data is processed for visualization with the Plots package. A detailed explanation can be found in the docstring of the PlotData2D constructor.
Another way to change the appearance of a plot is to convert the solution to a uniformly refined mesh before plotting. This can be helpful, e.g., when trying different settings for a simulation with adaptive mesh refinement, where one would like to ignore the mesh changes when comparing solutions. This is achieved with adapt_to_mesh_level, which uses the mesh adaptation routines to adapt the solution to a uniform grid. For example, the AMR solution from above could be preprocessed with
julia> pd = PlotData2D(adapt_to_mesh_level(sol, 4)...)
When plotted together with the mesh, this will yield the following visualization:
plot-rho-uniform-mesh
Plotting a user-defined scalar field
To plot a scalar quantity, one can call plot(ScalarPlotData2D(u, semi)), where u is an array of nodal values of the scalar field to plot. The layout of u should match the layout of the x and y nodal coordinates of the respective solver. For example, after running trixi_include(joinpath("examples", "unstructured_2d_dgsem", "elixir_euler_wall_bc.jl")), the following can be used to plot the function f(x, y) = x * y:
x = view(semi.cache.elements.node_coordinates, 1, :, :, :)
 y = view(semi.cache.elements.node_coordinates, 2, :, :, :)
 plot(ScalarPlotData2D(x .* y, semi))
This produces the following plot:
scalar-plotting-example
This routine can be used to visualize scalar quantities which depend on the solution, such as the norm of a velocity vector or two-dimensional vorticity. For example, we can visualize vorticity for a compressible version of the Brown-Minion vortex problem:
julia> using Trixi, Plots
@@ -88,4 +88,4 @@
   -->
   
Trixi2Vtk
Trixi2Vtk converts Trixi.jl's .h5 output files to VTK files, which can be read by ParaView, VisIt, and other visualization tools. It automatically interpolates solution data from the original quadrature node locations to equidistant visualization nodes at a higher resolution, to make up for the loss of accuracy from going from a high-order polynomial representation to a piecewise constant representation in VTK.
In the Julia REPL, first load the package Trixi2Vtk
julia> using Trixi2Vtk
To process an HDF5 file generated by Trixi.jl, execute
julia> trixi2vtk(joinpath("out", "solution_000000.h5"), output_directory="out")
This will create two unstructured VTK files in the out subdirectory that can be opened with ParaView or VisIt: solution_000000.vtu contains the discontinuous Galerkin solution data while solution_000000_celldata.vtu holds any cell-based values such as the current AMR indicator or the cell refinement level.
"solution_000000_scalar_mesh"
This allows you to generate VTK files for solution, restart and mesh files. By default, Trixi2Vtk generates .vtu (unstructured VTK) files for both cell/element data (e.g., cell ids, element ids) and node data (e.g., solution variables). This format visualizes each cell with the same number of nodes, independent of its size.  Alternatively, you can provide format=:vti as a keyword argument to trixi2vtk, which causes Trixi2Vtk to generate .vti (image data VTK) files for the solution files, while still using .vtu files for cell-/element-based data. In .vti files, a uniform resolution is used throughout the entire domain, resulting in different number of visualization nodes for each element.  This can be advantageous to create publication-quality images, but increases the file size.
If you want to convert multiple solution/restart files at once, you can just supply multiple input files as the positional arguments to trixi2vtk, e.g.,
julia> trixi2vtk("out/solution_000000.h5", "out/solution_000040.h5")
You may also use file globbing to select a range of files based on filename patterns, e.g.,
julia> trixi2vtk("out/solution_*.h5")
to convert all solution files in the out/ directory or
julia> trixi2vtk("out/restart_00[0-9]000.h5")
to convert every one-thousandth restart file (out/restart_000000.h5, out/restart_001000.h5 etc.).
When multiple solution/restart files are provided, Trixi2Vtk will also generate a .pvd file, which allows ParaView to read all .vtu/.vti files at once and which uses the time attribute in solution/restart files to inform ParaView about the solution time. A comprehensive list of all possible arguments for trixi2vtk can be found in the Trixi2Vtk.jl API.
Further information regarding the development of Trixi2Vtk can be found in the development section.
Makie.jl [experimental]
In addition to Plots.jl support, Trixi.jl includes visualization utilities through Makie.jl. Trixi.jl provides Makie-based visualization options both for heatmap-type plots (similar to the Plots.jl recipes) as well as for interactive surface plots. Support is currently limited to the UnstructuredMesh2D type.
Note
Plotting via Makie.jl is still considered an experimental feature and might change in any future releases.
A Makie plot can be created as follows: after running a simulation with Trixi.jl in the REPL, load a Makie backend (for example, GLMakie or CairoMakie).
julia> using GLMakie
To visualize the solution and mesh with a heatmap-type plot, simply run
julia> plot(sol)
Note
Both Makie.jl and Plots.jl export plot, so if you load both libraries, you will have to specify which plot function to call via Plots.plot or Makie.plot.
As with Plots.jl recipes, one can view individual solution components by creating a PlotData2D object and indexing into it with the desired variable name
julia> pd = PlotData2D(sol)
 julia> plot(pd["rho"])
Unlike the Plots.jl recipe, mesh plotting is controlled using the keyword argument plot_mesh = false, e.g.,
julia> plot(sol; plot_mesh=false)
The plot command also returns figure and axis handles, which can be used to edit plot titles or labels:
julia> fig, axes = plot(sol)
-julia> axes[1,1].title = "New title for subplot (1,1)"
Trixi.jl also supports interactive surface plots using iplot. After executing
julia> trixi_include(joinpath("examples", "unstructured_2d_dgsem", "elixir_euler_wall_bc.jl"))
we can run
julia> iplot(sol)
This will open up an interactive visualization window:
makie-example
The plot can be rotated (click and hold), zoomed in and out (scroll up and down), and panned (hold right click and drag). Two toggle buttons control whether mesh lines are visible on top of and below the solution.
Both plot and iplot use colormap = :inferno by default. A different colormap can be selected by providing an appropriate keyword argument. For example, plot(sol, colormap=:blues) and iplot(sol, colormap=:blues) produce the following figures:
makie-plot-example makie-iplot-example

+julia> axes[1,1].title = "New title for subplot (1,1)"

Trixi.jl also supports interactive surface plots using iplot. After executing

julia> trixi_include(joinpath("examples", "unstructured_2d_dgsem", "elixir_euler_wall_bc.jl"))

we can run

julia> iplot(sol)

This will open up an interactive visualization window:

makie-example

The plot can be rotated (click and hold), zoomed in and out (scroll up and down), and panned (hold right click and drag). Two toggle buttons control whether mesh lines are visible on top of and below the solution.

Both plot and iplot use colormap = :inferno by default. A different colormap can be selected by providing an appropriate keyword argument. For example, plot(sol, colormap=:blues) and iplot(sol, colormap=:blues) produce the following figures:

makie-plot-example makie-iplot-example