Remove elemag in combination with DiscretizationHDG?

[davidrudlstorfer]

Based on the recent PR #333 it turned out that there are some problematic things happening in elemag (most importantly that its built upon scatra). Additionally, it is currently used by no one. Can we remove it?

If elemag can be removed we could also think about removing the DiscretizationHDG - it's mostly used my elemag and also somehow in fluid. Its removal would allow us to further improve the modularization (move the ProblemType out of core).

[sebproell]

Elemag looks like a clear candidate for removal to me. As far as I know, this is unused.

Not sure about DiscretizationHDG. It seems to be used in fluid and scatra, so maybe the domain experts can comment on current or future plans.

From a modularization perspective, removal of the two would remove ProblemType hacks from core as reported by @davidrudlstorfer.

[c-p-schmidt]

I agree that a removal of elemag makes sense.

From my side, it is also not necessary to keep the scatra part, but maybe @amgebauer knows more, as it is mainly used in the context of cardiac simulations.

[amgebauer]

It's about this paper here: https://onlinelibrary.wiley.com/doi/abs/10.1002/cnm.2959

I had a look at the implementation and the part where we use the ProblemType in the discretization. It is only used by elemag and can therefore be deleted if we delete elemag.

[sebproell]

That's good news, thanks for checking!

[maxfirmbach]

I'm heavily in favor of deleting DiscretizationHDG ... To me 4C is a multiphysics code purely based on a standard finite element approach.

[amgebauer]

No, 4C is more than just boring standard FEM. Ask for example your colleagues who want to use NURBS 😉

[davidrudlstorfer]

I just wanted to ask if there's an update to this one month later? Has this already been asked to the PMMs?

During the discussions it emerged that elemag can probably be deleted and HDG stays. Is there a final decision so I can continue with the DBC check?

[amgebauer]

yes, elemag can be removed, so we can transform this into an issue.

[mayrmt]

This discussion has resulted in the issue #493. Hence, closing this discussion.