PETSc examples refinement

[valeriabarra]

I don't know what's changed on the PETSc side lately, but currently on main our problems that use uniform refinement, e.g., examples/petsc/area and examples/petsc/bpssphere are broken. Our CI didn't catch this because it doesn't test the refinement. It should have. I'm going to add it.

I'm investigating this issue, but I don't know exactly what has changed to cause it. It was fine until a few days ago. This also affects PR #550 .

To reproduce:

./bpssphere -dm_refine 1

Stack trace:

[0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
[0]PETSC ERROR: Nonconforming object sizes
[0]PETSC ERROR: The section point (0) closure size 12 != dual space dimension 9
[0]PETSC ERROR: See https://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting.
[0]PETSC ERROR: Petsc Development GIT revision: v3.13.4-637-gcaa22a8a69  GIT Date: 2020-08-04 04:43:23 +0000
[0]PETSC ERROR: ./bpssphere on a arch-linux2-debug named work by valeria Fri Aug  7 18:24:16 2020
[0]PETSC ERROR: Configure options --download-cmake --download-exodusii --download-fblaslapack=1 --download-hdf5 --download-hypre --download-metis --download-ml --download-mpich --download-mumps --download-netcdf --download-parmetis --download-pnetcdf --download-scalapack --download-superlu --download-superlu_dist --download-zlib --with-cxx-dialect=C++11 --with-debugging=1 --with-x=1 --with-zlib PETSC_ARCH=arch-linux2-debug
[0]PETSC ERROR: #1 DMProjectLocal_Generic_Plex() line 700 in /home/valeria/Research/CEED/petsc/src/dm/impls/plex/plexproject.c
[0]PETSC ERROR: #2 DMProjectFieldLocal_Plex() line 867 in /home/valeria/Research/CEED/petsc/src/dm/impls/plex/plexproject.c
[0]PETSC ERROR: #3 DMProjectFieldLocal() line 8291 in /home/valeria/Research/CEED/petsc/src/dm/interface/dm.c
[0]PETSC ERROR: #4 DMPlexRemapGeometry() line 3124 in /home/valeria/Research/CEED/petsc/src/dm/impls/plex/plexgeometry.c
[0]PETSC ERROR: #5 DMSetFromOptions_Plex() line 2464 in /home/valeria/Research/CEED/petsc/src/dm/impls/plex/plexcreate.c
[0]PETSC ERROR: #6 DMSetFromOptions() line 844 in /home/valeria/Research/CEED/petsc/src/dm/interface/dm.c
[0]PETSC ERROR: #7 main() line 145 in /home/valeria/Research/CEED/libCEED/examples/petsc/bpssphere.c
[0]PETSC ERROR: PETSc Option Table entries:
[0]PETSC ERROR: -dm_refine 1
[0]PETSC ERROR: ----------------End of Error Message -------send entire error message to [email protected]
application called MPI_Abort(MPI_COMM_SELF, 145060) - process 0
[unset]: write_line error; fd=-1 buf=:cmd=abort exitcode=145060
:
system msg for write_line failure : Bad file descriptor

[knepley]

I put a new thing in the sphere mesh that automatically projects to the correct shape when you refine. This says that the closure size is 12, which looks right for a linear quad with 3D coordinates, but the dual space is of size 9, so it looks like it was created for a linear triangle with 3D coordinates.

And now I see the error. It is on line 2202 of plexcreate.c. The PETSC_TRUE should be 'simplex'. Do you guys want to fix it or do you want me to make an MR?

[valeriabarra]

If it's thta simple at the line that you pointed, I can do it. Thanks!

[knepley]

Go ahead and change it so you can get your thing running, but I made a branch already and added tests :)

[valeriabarra]

Ok, so if you started a branch with tests, I think it's better if you start the MR, too. Please link it here when you open it, so I know which one it is. Thanks!

…

On Sat, Aug 8, 2020, 09:40 Matthew Knepley ***@***.***> wrote: Go ahead and change it so you can get your thing running, but I made a branch already and added tests :) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#603 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJQVA7WKYVDNALP6VWD4NOLR7VWWJANCNFSM4PYG5SLQ> .

[knepley]

It is here: https://gitlab.com/petsc/petsc/-/merge_requests/3034

[valeriabarra]

Closed via PETSc's MR !3034 and PR #604