Sebastian Grimberg

> Hi @sshiraiwa, Partitioning of `GridFunction`s will be the next step in this branch. This is what is required to support curved meshes, correct? Looks awesome and looking forward to...

@v-dobrev, I've been testing this most recent update with curved meshes and it's working great. Loading the serial mesh on the root process, dumping each `MeshPart` to a `stringstream`, sending...

Hi @v-dobrev, A use case I ran into recently was the following: - Read a serial conforming mesh from file, - Partition it, - Do parallel, non-conforming refinement. I think...

> This PR seems well executed. It's clear what the changes involve and they present a minimal API difference. I'd like an additional reviewer on this PR since it touches...

Apologies for the rebase. I ran into a few small bugs for the `Multigrid` changes in testing so I fixed them and squashed the history. I will avoid making any...

Hi all, just a follow up for review to try to get this approved and merged. Thank you!

Thanks @psocratis! Indeed, originally this PR included #3081, #3082, and #3083 which is really where the multiple RHS functionality is useful. Useful for MG also if the coarse solve is...

@tzanio @dylan-copeland I originally thought it would be more natural to make the interface use `DenseMatrix` instead of `Array`, where each RHS could be individually accessed by columns of the...

Hi @drzisga @jandrej, any chance you could give this a final review when you have a moment? Sorry for the ping but this would be nice to move on from...

This should resolve https://github.com/mfem/mfem/issues/3463 also.