Skip to content

Update run_tests to run with different MCC3, geometry, and solver options

Kumar, Shikhar requested to merge run_tests_updates into pygriffin_initial

This MR updates the tests in test_pygriffin_run.py to include loose convergence criteria and input geometries of 2 rings to run these tests more quickly. A variety of MCC3, geometry, and solver options are also tested in these "pygriffin_run" tests. The Sample 17 ABTR problem is also updated to run with DFEM_SN CMFD, which finishes running in around half an hour with MPI enabled (80 procs on VTR).

A few notes:

  • Currently, there needs to be some more logic enabled for PyGriffin to use a pre-existing mesh file, since currently there is no good way right now to make the material block mapping from a pre-existing mesh file. A placeholder unit test is included to test this (see test_PyGriffinConnect_run_geometry_test1_skip in test_pygriffin_run.py), but it is currently commented out
  • PyGriffin needs a more sophisticated method to determine the MPI executable, since right now it just runs mpirun -n [N_PROCS] [GRIFFIN_COMMAND], but there is no guarantee that this MPI executable is the same one used to compile Griffin. I have gotten around this by directly specifying the path to the MPI executable within PyGriffin, but unless this is done in a more flexible way, all MPI-enabled runs with PyGriffin are likely to crash unless the MPI executable is in the user's PATH.

Merge request reports