Prom maxwell local greedy algorithm #249
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
andersonw1
requested review from
ckendrick,
dylan-copeland,
ebchin,
dreamer2368,
chldkdtn and
siuwuncheung
andersonw1
changed the title
ckendrick
requested changes
Dec 8, 2023
ckendrick
approved these changes
Dec 18, 2023
|
@dylan-copeland and @siuwuncheung, would you finish or continue reviewing this PR? |
dylan-copeland
approved these changes
Jan 8, 2024
* fix issues in Matrix::orthogonalize * add Matrix::orthogonalize_last method * run astyle * add another test for Matrix::orthogonalize_last * rerun astyle properly * review n.1 changes * review n.2
* Add assemble timer and timing prints * Fix issue with -online_interp when running in parallel * Update timing print * Fix minor typo in examples
* Bug fix. * Cleaning up a few things.
* initial commit * Updated ROM operator class definition and input arguments * EQP hyperreduction setup * Added hyperreduction until line 1000 * Setting BCs * Finished implementation in main * Started ROM operator initialization Need to review before finishing * Gone through the code and implemented what was straightforward * Mult_hyperreduced * small fixes to make consistent with class definition * removed eqp_lifting code, since it's not necessary * Started eqp coefficient computation implementation And made fast calculation TODO * Updated SetupEQP to make tests run * Some more fixes for the compiler * Changed rrdim -> rxdim * Started SetupEqp_snapshots for hyperelastic nonlinear operator * Added sketch for compute reduced eqp * Code compiles * Added element function to G row computation * Testing element vector error * Debugging. Vshape not implemented for this class... * fixed bug * Explicitly adding model to eqp loop * segfault fix * segfault fix 2 * debug segfault * actual debug fix * perhaps segfault fix * small fix * New test. * testing * changed back * testing back again * Hopefully * Finished writing the test. * Fixed compilation * Debugging elvect size * passing by reference * Fix for element vector * check all errors * Synced with hyperelasticintegrator. * Divide error by two * changed points to check error * Clean up * testing SetupEQPSnapshots * testing errors for NNLS * small fix * removed testing * Remove normalize constraints * Fixed segfault * Reduced EQP is building * Running but wrong * Runs! * Fixed some issues * rank-myid fix * Syntax improvements * fixed bug for v0 * small fix * Implemented time windowing * Added command line option for time windowing * Added EQP command line examples * Cleaned up code * Sketch for optimized EQP * Continued fast eqp implementation * Improved regular EQP * V-basis speedup * Storing DS now * Solved segmentation fault in non eqp hyperreduction * Adding QDEIM option to some examples. * updated gitignore * Matrix multiplication optimization * Tested precompute optimization * Numbering works now. * Testing .hpp fix again * Fixed .hpp style issue * Delete the RomOperator to remove a memory leak * Updated RomOperator structure to have less leaks in timestepping * Cause of segfault error found * Segfault fixed * Removed EQP bug * Add x0 to setupEQP * add matrix row/col rescaling methods * normalize NNLS constraints * Added x0 * Fixed bug in NNLS constraint forming loop. * Max iters increased * improved eqp speed * resetting temp variable now * Moved temporary variable to the correct position * Minor formatting changes * Removed duplicate lines in examples * build succeeds * Updated formatting * Astyle 3.1. * updated .gitignore * ++ increment * Removed ReducedSystemOperator * Removed unnecessary comments * Added consts * Added more ++ increments * Removed unnecessary HSINV * Comments * LibROM native read / write * Added pointer deletion * librom runs * New benchmarks * elementmatrix first implementation * Construct em outside loop * Added element matrices to ROM class * Reduced EQP fast uses em * Em on everything * Debugged EQP * Formatting * Removed unnecessary comment * Removed more unnecessary comments * Implemented subsampling * measure elapsed time for EQP * Measure elapsed time for all methods * Update .gitignore * Optimized lifting of FOM dimension in Nonlinear elasticity EQP (#259) * implementation works and gives speed up * Removed unnecessary variables * Final benchmark --------- Co-authored-by: Dylan Copeland <[email protected]> Co-authored-by: Chris Vales <[email protected]> Co-authored-by: Dylan Copeland <[email protected]>
andersonw1
added a commit
that referenced
this pull request
Apr 2, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds a global prom example for the maxwell equation to libROM (based on MFEM ex3p).
Example run:
Build database
mpirun -np 8 maxwell_local_rom_greedy -build_database -greedy-param-min 1.0 -greedy-param-max 1.2 -greedy-param-size 5 -greedysubsize 2 -greedyconvsize 3 -greedyrelerrortol 0.01
Run full-order model
mpirun -np 8 maxwell_global_rom --fom -f 1.15
Use reduced-order model
mpirun -np 8 maxwell_local_rom_greedy -use_database -online -f 1.15
For this example:
Elapsed time for solving FOM: 8.870853e-01 second
Elapsed time for solving ROM: 9.554200e-05 second
Relative error of ROM solution = 1.88768e-05
Speedup: x9284.76