Multi-level discretization + High order Gauss-Lobatto for SEM

[labesse40]

Multilevel Discretization + High order Gauss-Lobatto for SEM High order Gauss-Lobatto for SEM.pdf

• [x] Multi-level discretization

• [x] Field initialization: Done in https://github.com/GEOSX/GEOSX/pull/1963

• [ ] Make deep copies (for example: face normal) for the new mesh level https://github.com/GEOSX/GEOSX/commit/60477266a20a2a6815a58e250c3f2e9747168e55

• [ ] preprocessor guard computation of gradients: memory usage on dNdX variable for CUDA

[rrsettgast]

This crashes when trying to set the sourceNodeIds here:

https://github.com/GEOSX/GEOSX/blob/5dd4586e5050d66451e661793f280b367c6c4927/src/coreComponents/physicsSolvers/wavePropagation/AcousticWaveEquationSEMKernel.hpp#L163

The number of nodes is now 64, but the allocation is hard coded here: https://github.com/GEOSX/GEOSX/blob/5dd4586e5050d66451e661793f280b367c6c4927/src/coreComponents/physicsSolvers/wavePropagation/AcousticWaveEquationSEM.cpp#L149

So how do you want to handle the sources? I would assume we should remove the hard coded 8 and replace with the order of the discretization.

[XL64]

Sorry for the accidental force push, I restored the branch.

[sframba]

A rebaseline will be needed, due to the change in the meshBodies structure (Group has a child XXXX in the file to compare but not the baseline file.). Most tests are affected.