Julian Andrej
Julian Andrej
> @jandrej, could I suggest a couple small changes that I noticed in the DFEM code? > > Return unique pointer instead of shared > You can always assign the...
I have never seen output like this either, unless I specifically told cmake to provide it. Do you mind sharing the complete environment configuration so we can judge what's going...
I would try invoking nvcc by passing the options `-Xcompiler -std=gnu++14`, which will set the host compiler c++'s dialect version to match nvcc's toolchain. Or just set the std to...
Doesn't https://github.com/mfem/mfem/issues/2976 and https://github.com/mfem/mfem/pull/3004 take care of that?
@thelfer related, you might want to have a look at https://github.com/mfem/mfem/blob/dfem-coefficient/examples/dfem/dfem.hpp#L1719
> @jandrej Thanks for the pointer. Have you published anything about this, just to get an overview of the project ? Not yet, but feel free to ping me if...
> For me, a much cleaner solution would be to make some virtual method of `IterativeSolver`, let us say `CheckConvergence()`, which would do the classical check of the norm, but...
> @jandrej I prefer the composition approach of using `IterativeSolverMonitor` as currently proposed vs a virtual convergence check. Then we just have to maintain one custom class instead of a...
Please review the current changes and refrain from commits/merges that aren't moderated.