Veselin Dobrev comments

Results 452 comments of


                                            Veselin Dobrev

Add an interface to coefficients to be evaluated at quadrature points.

> There are a lot of these build warnings: > `warning: overloaded virtual function "mfem::Coefficient::Eval" is only partially overridden...` Is this with `nvcc`? Travis now should fail if there are...

Debugging lambdas in the `MFEM_FORALL` macro

cc: @camierjs, @YohannDudouit, @tzanio

Debug of Mumps solver: Segmentation fault: address not mapped to object at address

Hi @liuzg-23, Sorry for the late reply. I suspect a 32-bit integer overflow may be the issue. Is MUMPS build with 32-bit or 64-bit integers? I'm not sure if there...

Evaluate basis functions on device

For completeness, I'll just add that both > 1. Map the point to the reference element > 2. Evaluate the basis function at the mapped reference coordinates can be easily...

Modify GridFunction::GetTrueVector behavior to SetTrueVector if `t_vec` does not already exist

Please resolve conflicts and failed checks.

Fix memory leak found in HypreAMS when SetOperator is called multiple times [hypreams-mem-leak-fix]

@tuckerbabcock, I assume this is ready for review, so I'll add the `ready-for-review` label to start the review process.

Run pcuda mfem with mpi and multiple GPUs

On LLNL's Lassen this happens automatically. On Perlmuter, it seems, you need to do it explicitly, see here: https://docs.nersc.gov/performance/readiness/#running-applications.

Run pcuda mfem with mpi and multiple GPUs

> What makes the small difference here? Is that related to where the matrix is assembled (host vs device)? Yes, the difference is from the assembled matrix and, yes, `LEGACY`...

Run pcuda mfem with mpi and multiple GPUs

The way MFEM is compiled when CUDA and MPI are enabled is that we tell `nvcc` (defined by `CUDA_CXX`) to call the MPI compiler (defined by `MPICXX`) for host compilation...

libceed solvers improvements

This PR is now [under review](https://github.com/mfem/mfem/blob/master/CONTRIBUTING.md#pull-requests) (see the table in the PR description). To help with the review process, please **do not force push** to the branch.