libCEED interface: Partial and matrix-free assembly support for H(div) and H(curl) vector FE spaces #3626
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
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
0c5952d
to
62e9796
Compare
3 tasks
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
62e9796
to
e1f9226
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
e1f9226
to
126845e
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-coeff-dims
branch
from
0ac2395
to
679e2de
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
2 times, most recently
from
a6c8419
to
bd3388d
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
bd3388d
to
89ef57d
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-coeff-dims
branch
2 times, most recently
from
3de4ebd
to
3b42a09
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
89ef57d
to
1695883
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-coeff-dims
branch
from
3b42a09
to
ccd59ba
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
1695883
to
2360ec3
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-coeff-dims
branch
from
ccd59ba
to
ecd35ba
Compare
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
2360ec3
to
7b6c93e
Compare
… VectorFEMassIntegrator, DivDivIntegrator, and CurlCurlIntegrator WIP: Element restriction with orientations for H(div) and (order 1) H(curl) elements, vector FE libCEED basis construction.
…riction construction
Tests now pass. Note a tolerance issue for libCEED assembly showed up for the fichera-amr.mesh test where errors of ~1e-10 in the multiplication appeared. This was temporarily patched by using a coarser tolerance in unit tests for nonconforming meshes.
…bly after they are no longer needed
…or-, and matrix-valued coefficient variants
sebastiangrimberg
force-pushed
the
sjg/libceed-hcurl-hdiv-dev
branch
from
7b6c93e
to
d020d9a
Compare
|
Closing this as it's more suitable to be integrated directly into downstream application code. |
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.
NOTE: Based on #3625
NOTE: Waiting on CEED/libCEED#1156 (merged), CEED/libCEED#1196 (merged), and CEED/libCEED#1168 in libCEED to fully support diagonal assembly and high-order H(curl) spaces.
This PR adds PA and MF assembly support for:
VectorFEMassIntegrator,DivDivIntegrator, andCurlCurlIntegratorfor squareBilinearFormobjects, andMixedVectorGradientIntegrator, andPAMixedVectorCurlIntegratorforMixedBilinearFormobjects. Arbitrary order H(div) and H(curl) finite element spaces on tensor-product element meshes or simplex or mixed meshes are supported. In all cases, the basis operation does not make use of tensor-product structure for these vector finite elements, which is a limitation of libCEED and leads to different performance (though surprisingly not that much different in my testing on CPU at least) versus the native MFEM PA backends for tensor-element meshes. Unit tests covering this new functionality have been added totests/unit/ceed/test_ceed.cpp.GPU libCEED backends do not yet support the special element restriction and (non-tensor) basis operations for vector FE spaces, but these will hopefully come soon. For now, all
cpu/self/*/serialandcpu/self/*/blockedbackends have been successfully tested.