Will Pazner

Thanks, that's useful context. Let's see what others on the team think. I'm definitely open to it, I just think we need to strike the right balance between supporting specific...

Hi @sebastiangrimberg, Apologies for the delay. I remain a bit ambivalent about some of these changes. I know you put effort into making these changes and preparing this PR, but...

Can you provide some more information, e.g. which initial mesh you are using, and how many MPI ranks? If I understand your configuration correctly, the serial mesh is refined until...

Hi @Heinrich-BR, Sorry it took so long to get back to you. It is important to note that you are using the `--full-assembly` flag in your example (probably because you...

By the way, why does `FiniteElementSpace::GetTraceElement` take an index `i`? https://github.com/mfem/mfem/blob/87290fbf5674a0c9ab9164fa7b5d30364a164a54/fem/fespace.hpp#L1186 It is never used https://github.com/mfem/mfem/blob/13144287705c1338561106b62e75a780d15c8751/fem/fespace.cpp#L3276-L3280

> By the way, why does `FiniteElementSpace::GetTraceElement` take an index `i`? It seems the reason is to support variable-order spaces, but this was not implemented yet. Should be fixed in...

To create a coefficient from your density values, you will need some way of evaluating the density at arbitrary locations in the computational domain. For example, this could be done...

So if your mesh is arbitrary, I would create a custom coefficient that identifies the cell of your coefficient grid, and then manually performs the trilinear interpolation, something like this:...

Yes, that's right. The translation vectors represent the translations that line up the vertices that are identified in the periodic mapping. You can do something like this: ```c++ Mesh orig_mesh...

PCG should converge, but you need to make sure that your right-hand side is orthogonal to the nullspace (constants). You can do this by subtracting the mean of the right-hand...