Does svPre correctly reorder element connectivity

[ktbolt]

svPre checks for correct element ordering using the check_node_order function in helpers.cxx.

Element connectivity is correctly reordered no matter what the input mesh ordering is.

Note that there is a .svpre flag named cvsolver_node_order that if set does not reorder elements. The correct reordering is still checked and modified though.

[ktbolt]

@mrp089 I can't find anything wrong with svPre, at least with the tests I'm using. Can you point me to a directory that has data that was causing you problems?

[mrp089]

@ktbolt I uploaded two simulation directories (all meshes, svPre input, svPre output, svSolver input) for the same model, once with positive and once with negative Jacobian: https://drive.google.com/file/d/1B0Cqm6_052IzQoHS10kVze5TisT-yGbV

To check the Jacobian, I use the "Mesh Quality" Filter in ParaView (select "Jacobian" for "Tet Quality"). The simulation with negative Jacobian blows up after ~100 time steps.

[ktbolt]

@mrp089 Looking at the meshes in 0075_1001_jac_negative I see that the boundary surface files (mesh-complete.exterior.vtp, mesh-complete/mesh-surfaces/inflow.vtp, etc.) don't have the same orientation as the mesh-complete/mesh-complete.mesh.vtu surface. The mesh-complete/mesh-complete.mesh.vtu surface triangles connectivity is reordered but the other surfaces are not. Of course!

One solution is to add a flag to svPost to reorder the element connectivity before writing out the .vtu and .vtp results files.

But I think that it would be better to make reorder the element connectivity the default behavior, want the VTK files to have the correct ordering.