Added global prom for maxwell equation #246
Conversation
andersonw1
changed the title
andersonw1
requested review from
ckendrick,
dylan-copeland,
dreamer2368,
chldkdtn,
siuwuncheung and
ebchin
|
There is a little red cross to the left of the commit number 34936f7, which indicates there are some failing checks. If you press it, you will see "CI / code-style (pull_request) Failing after 41s". That indicates some code styles need to be fixed, e.g. non-uniform indents, too-long lines. That's an automatic way to fix it through astyle. If you download from the source and build it, then in the parent directory of libROM just do |
ran stylization and removed line of code initializing Hypre
| // FOM solution: maxwell_global_rom -fom -f 1.15 | ||
| // | ||
| // Online phase: maxwell_global_rom -online -f 1.15 | ||
| // |
There was a problem hiding this comment.
Does it also work in parallel, like poisson_global_rom?
There was a problem hiding this comment.
I tested in parallel and confirmed that it works, so you can copy the comment from the Poisson example:
// This example runs in parallel with MPI, by using the same number of MPI ranks
// in all phases (offline, merge, fom, online).
Changed saving mesh/sol files to calculate relative errors Saved mesh/solution files are now in folders titled "maxwell" instead of "example3"
examples/prom/maxwell_global_rom.cpp
Outdated
|
|
||
| // libROM stores the matrix row-wise, so wrapping as a DenseMatrix in MFEM means it is transposed. | ||
| DenseMatrix *reducedBasisT = new DenseMatrix(spatialbasis->getData(), | ||
| numColumnRB, numRowRB); |
There was a problem hiding this comment.
This seems unnecessary. The same object is defined in the global ROM for Poisson equation, both in C++ and Python side. Can @andersonw1 @dreamer2368 take care of it? Thanks!
There was a problem hiding this comment.
I deleted this line and everything runs with the same result, so it seems this part was unnecessary.
There was a problem hiding this comment.
Thanks. I'll take care of the Poisson equation.
|
Looks good to me! Thanks @andersonw1 for the effort. Please run |
This PR adds a global prom example for the maxwell equation to libROM (based on MFEM ex3p).
Example runs
mpirun -np 8 maxwell_global_rom --offline -f 1.0 -id 0mpirun -np 8 maxwell_global_rom --offline -f 1.1 -id 1mpirun -np 8 maxwell_global_rom --offline -f 1.2 -id 2Merge snapshots to make the basis
mpirun -np 8 maxwell_global_rom --merge -ns 3Run full-order model
mpirun -np 8 maxwell_global_rom --fom -f 1.15Use reduced-order model
mpirun -np 8 maxwell_global_rom --online -f 1.15For this example:
Elapsed time for solving FOM: 4.906529e-01 second
Elapsed time for solving ROM: 1.010420e-04 second
Relative error of ROM solution = 0.00044228
Speedup: x4855.93