[Decision] New FEM code(s) to integrate

[niklassiemer]

replace fenics

Our current support for fenics is relying on a fairly outdated package. Moreover the fenics code itself is not actively maintained anymore, so we have to decide on a new FEM "engine". Candidates identified in our last infrastruture meeting

• dolfinx: framework for fenicsx (fenics successor) -> @Tara-Lakshmipathy already works on the integration. Also, have a look at the tutorial
• elmerFEM, e.g. used here.
• coupled DDD and XFEM https://www-cast3m.cea.fr
• Eriks paper including a pedestrian coupling scheme to make atomistics-informed FEM calculations: https://iopscience.iop.org/article/10.1088/0965-0393/21/5/055011
• deal.II
• (maybe: abaqus as a widely used, but proprietary FEM code)

[mbruns91]

I'd say integrating dolfinx is agreed upon. We should somewhen in the future

• discuss differences, advantages and disadvantages of the other packages/ codes
• try to derive a nice, in-depth use-case for FEM integration, ideally from projects/ IUCs from the digitalization initiatives PMD and MatWerk.

Because he seemed to be better connected to people actually doing FEM than the rest of us, Erik should be included in those discussions.

[samwaseda]

This PR is still on going but the decision has been basically made, so I'm gonna close this one.