Veselin Dobrev

Unit test "LOR AMS" failure with Umpire

1

I was testing a build in Spack with CUDA, Umpire, RAJA, and Hypre (without CUDA) in the branch that adds mfem-4.7, https://github.com/spack/spack/pull/44010 and ran into a test failure in the...

The FMS unit test fails as of v4.7. We need a tolerance when comparing the node coordinates. Also, in order for the test to work, the mesh `data/star-q3.fms` has to...

Matrices with zero diagonal entries in `hypre_ParCSRTMatMatKTDevice`

1

I ran into the following issue while running tests in MFEM with HYPRE built with CUDA and `--enable-debug`: the `hypre_assert` in `hypre_ParCSRTMatMatKTDevice`, here: https://github.com/hypre-space/hypre/blob/0dcae3ec7c069785ea25d25aa0bc0c7aa8b0be8d/src/parcsr_mv/par_csr_triplemat_device.c#L604 fails when the diagonal block of...

Potentially incorrect `hypre_assert`

3

When running MFEM tests with HYPRE built with `--enable-debug` and no GPU support the following check fails in a few MFEM examples: https://github.com/hypre-space/hypre/blob/0dcae3ec7c069785ea25d25aa0bc0c7aa8b0be8d/src/parcsr_ls/ame.c#L464 In these failures, the memory location on...

Potential issue when using PARMETIS column permutation in v8.2.1

2

I noticed this during testing of mfem-4.7 with SuperLU_DIST v8.2.1 -- one of the mfem examples that uses SuperLU_DIST, `examples/ex11p.cpp` hangs when run with: ```console mpirun -np 4 ./ex11p --superlu...

Add support for SUNDIALS v7

6

Add support for SUNDIALS v7. | PR | Author | Editor | Reviewers | Assignment | Approval | Merge| | --- | --- | --- | --- | --- |...

Add a patch for MFEM v4.7 that adds support for SUDIALS v7

1

The patch is based on MFEM's `master` branch as of 11/13/2024: https://github.com/mfem/mfem/commit/81ee483d2a249eca2eec05d3143cc559cfeea1d1, using only (essentially) the changes in the files `linalg/sundials.?pp`.

MFEM: add new version v4.8

8

TODO: - [x] After MFEM v4.8 is released, update the version number (`4.8.0-dev` -> `4.8.0`) and its source - [ ] Migrate to https://github.com/spack/spack-packages, e.g. using this tool: https://github.com/spack/migrate-package-prs -...

Add new method: `ParFiniteElementSpace::GetTrueUpdateOperator`

8

Also, in class `TripleProductOperator`, add methods to access the three individual `Operator`s that define the triple product. This PR should help with implementing the workflow discussed in #4565. | PR...

Add method for gradient computation on GPU

3

Here is a sample implementation that we can turn into a method in class `GridFunction`: ```c++ void GridFunctionGradient(const GridFunction &f, const IntegrationRule &ir, Vector &grad_f, QVectorLayout ql = QVectorLayout::byNODES, MemoryType...