Use of Intrepid(2) RealSpaceTools about albany HOT 5 CLOSED

In practice, the setting of the container sizes necessary to use the RST and FST is error prone, and honestly I was tired of trying to debug it. This is related to the way the workset size is computed and the impact it has on the container sizes. The alternative is to implement the same operations locally, often using MiniTensor objects, which is cleaner and more readable, and much less prone to memory and indexing errors.

from albany.

I'm probably biased, but I would think we could take more advantage of Intrepid2. There is a lot of work happening in Itrepid2 and downstream (e.g. Kokkos Kernels) and if we stick to Intrepid2 and properly use their functions we could take advantage of that, especially regarding portability.
Dan, if you want I can help you with the issue you are having right now.

from albany.

@ibaned Just saw your related issue in Trilinos.

from albany.

@mperego could you please take a look at trilinos/Trilinos#1030 ? Thanks !

from albany.

The resolution of trilinos/Trilinos#1030 also resolves this issue in the near term.

from albany.