Move remaining scatra to teko #972
Draft
Conversation
This file contains hidden or 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
|
And here is an overview of the tests that are affected: Finished besides changing the results ✅
Preconditioner too bad
Iter 140, [ 1] : 1.369391e-02
Iter 145, [ 1] : 9.107705e-04
Iter 150, [ 1] : 2.866800e-05
Iter 155, [ 1] : 2.058302e-06
Iter 160, [ 1] : 1.618146e-06
Iter 165, [ 1] : 1.317104e-07
Iter 170, [ 1] : 6.158300e-08
Iter 175, [ 1] : 2.994942e-08
Iter 180, [ 1] : 1.332235e-08
Iter 185, [ 1] : 1.132661e-08
Iter 187, [ 1] : 7.252641e-09
Belos::PseudoBlockGmresSolMgr::solve(): Warning! Solver has experienced a loss of accuracy!
=========================================================================
PROC 0 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad
Iter 885, [ 1] : 1.202636e-01
Iter 890, [ 1] : 1.197272e-01
Iter 895, [ 1] : 1.191979e-01
Iter 900, [ 1] : 1.186756e-01
Iter 905, [ 1] : 3.360798e+00
Iter 910, [ 1] : 3.210767e+00
Iter 915, [ 1] : 3.150261e+00
Iter 920, [ 1] : 3.090965e+00
Iter 925, [ 1] : 3.001797e+00
Iter 930, [ 1] : 2.973915e+00
Iter 935, [ 1] : 2.947395e+00
Iter 940, [ 1] : 2.913978e+00
Iter 945, [ 1] : 2.861224e+00
Iter 950, [ 1] : 2.810215e+00
Iter 955, [ 1] : 2.741332e+00
Iter 960, [ 1] : 2.702915e+00
Iter 965, [ 1] : 2.684234e+00
Iter 970, [ 1] : 2.548493e+00
Iter 975, [ 1] : 2.151351e+00
Iter 980, [ 1] : 2.017465e+00
Iter 985, [ 1] : 1.765070e+00
Iter 990, [ 1] : 8.722739e-01
Iter 995, [ 1] : 6.895343e-01
Iter 1000, [ 1] : 2.255960e-01
=========================================================================
PROC 1 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad
Iter 865, [ 1] : 9.009990e-03
Iter 870, [ 1] : 9.007772e-03
Iter 875, [ 1] : 9.006525e-03
Iter 880, [ 1] : 9.004372e-03
Iter 885, [ 1] : 9.003852e-03
Iter 890, [ 1] : 8.999490e-03
Iter 895, [ 1] : 8.994626e-03
Iter 900, [ 1] : 8.993271e-03
Iter 905, [ 1] : 8.993271e-03
Iter 910, [ 1] : 8.993271e-03
Iter 915, [ 1] : 8.993271e-03
Iter 920, [ 1] : 8.993271e-03
Iter 925, [ 1] : 8.993271e-03
Iter 930, [ 1] : 8.993271e-03
Iter 935, [ 1] : 8.993271e-03
Iter 940, [ 1] : 8.993271e-03
Iter 945, [ 1] : 8.993271e-03
Iter 950, [ 1] : 8.993271e-03
Iter 955, [ 1] : 8.993271e-03
Iter 960, [ 1] : 8.993271e-03
Iter 965, [ 1] : 8.993271e-03
Iter 970, [ 1] : 8.993271e-03
Iter 975, [ 1] : 8.993271e-03
Iter 980, [ 1] : 8.993271e-03
Iter 985, [ 1] : 8.993271e-03
Iter 990, [ 1] : 8.993271e-03
Iter 995, [ 1] : 8.993271e-03
Iter 1000, [ 1] : 8.993271e-03
=========================================================================
PROC 1 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad
Iter 105, [ 1] : 1.866550e-04
Iter 110, [ 1] : 2.320085e-06
Iter 115, [ 1] : 1.532587e-07
Iter 119, [ 1] : 6.792557e-09
Belos::PseudoBlockGmresSolMgr::solve(): Warning! Solver has experienced a loss of accuracy!
=========================================================================
PROC 0 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad
Iter 925, [ 1] : 1.836439e-01
Iter 930, [ 1] : 1.019956e-01
Iter 935, [ 1] : 9.533681e-02
Iter 940, [ 1] : 9.278695e-02
Iter 945, [ 1] : 9.019854e-02
Iter 950, [ 1] : 8.938341e-02
Iter 955, [ 1] : 8.677166e-02
Iter 960, [ 1] : 8.111195e-02
Iter 965, [ 1] : 7.566726e-02
Iter 970, [ 1] : 6.452227e-02
Iter 975, [ 1] : 3.363782e-02
Iter 980, [ 1] : 3.265104e-02
Iter 985, [ 1] : 3.239676e-02
Iter 990, [ 1] : 3.214833e-02
Iter 995, [ 1] : 3.190552e-02
Iter 1000, [ 1] : 3.166814e-02
=========================================================================
PROC 0 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad
Iter 10, [ 1] : 6.436398e-03
Iter 15, [ 1] : 2.486579e-06
Iter 17, [ 1] : 7.672161e-09
Belos::PseudoBlockGmresSolMgr::solve(): Warning! Solver has experienced a loss of accuracy!
=========================================================================
PROC 2 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad (This is only for the scatra field during init pot calc. If we allow to ignore this error in GMRES it also starts with the first TS, where all contributions are solved for but does not properly solve then.
Iter 75, [ 1] : 5.272179e-04
Iter 80, [ 1] : 1.027570e-04
Iter 85, [ 1] : 1.949314e-05
Iter 90, [ 1] : 4.298326e-06
Iter 95, [ 1] : 1.500318e-06
Iter 98, [ 1] : 9.234936e-07
Belos::PseudoBlockGmresSolMgr::solve(): Warning! Solver has experienced a loss of accuracy!
=========================================================================
PROC 0 ERROR in /home/schmidt/workspace/FourC/src/core/linear_solver/src/method/4C_linear_solver_method_iterative.cpp, line 174:
Core::LinearSolver::BelosSolver: Iterative solver did not converge.-> Preconditioner too bad (This is only for the scatra field during init pot calc. If we allow to ignore this error in GMRES it also starts with the first TS, where all contributions are solved for but does not properly solve then. Not yet working
Not yet tackled
|
38 tasks
c-p-schmidt
commented
Aug 13, 2025
| beloslist.set("SOLVER_XML_FILE", xmlfile->string()); | ||
|
|
||
| // required for adaptive linear solver tolerance | ||
| beloslist.set("Convergence Tolerance", inparams.get<double>("AZTOL")); |
There was a problem hiding this comment.
It should be solved differently.
what they actually contain and distinguish from the node_block maps that need to be specified
and related implementation.
is working with Teko. - remove old now unused xml files - add new block preconditioner xml files for usage with Teko The BGS-AMG for elch works with the following exception: - the dof-split is not working - AMG-BGS is not yet working - test case results not yet adapted
- AMG-BGS not yet migrated
however, several not yet working properly - BGS-AMG with dofsplit not yet tackled
There is still a segfault, I don't know why yet Info that is set looks good, so I don't know what's the problem there
and the parameter list shit as well here and in elch
c-p-schmidt
force-pushed
the
move-remaining-scatra-to-teko
branch
from
d5a3b99 to
32d4683
Compare
|
After the session with @maxfirmbach last week and some further work, everything besides the dofsplit works. Thanks again! I'll separate the individual changes to separate PR's and will close this one once everything is merged in smaller chunks. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Description and Context
AMGnxn should be replaced by
MueLuandTeko, see #754. This draft PR implements the first steps to achieve #754; however, there are still some further steps to take.The BGS-AMG without dofsplit (splitting of the matrix only happens along fields [scatra-solid-thermo] and/or geometry [anode-electrolyte-cathode, i.e., a separation of the entire domain into non-overlapping subdomains] works fine besides some probably more powerful preconditioners that need to be chosen.
In contrast, the BGS-AMG with dofsplit (i.e., an additional split of the scatra dofs [c, phi]) works for the more complex scenario of solid-scatra-interaction (besides a too less powerful preconditioner), but it does not work for the pure scatra problem, when Teko runs into a segfault within Trilinos. The information provided to Teko for this process looks consistent, so I do not understand what is happening there.
@maxfirmbach: I hope you can give some further hints on what could be the origin of these problems.
And finally, some further information on the individual commits:
compute_solver_parametersHint: 1-3 should and will be merged separately
So to sum up, each subframework or subfunctionality is moved in separate commits in commits 5-8 and 10.
Related Issues and Pull Requests
part of #754
Interested parties
@bwirthl @lkoeglmeier