Unusual internal cavity / inverted faces when visualising StructuredQuadBox mesh

[Belay-129]

Hi Uw3 team,
When creating and visualizing a simple StructuredQuadBox mesh, the rendered mesh appears to contain an internal hollow region / inverted face, giving the impression of a missing or collapsed interior. This is unexpected for a uniform structured mesh and makes it difficult to visually verify mesh integrity.

The issue appears during visualization rather than mesh construction logic, but it is unclear whether this is a rendering artefact or a deeper mesh connectivity issue.

" import underworld3 as uw
from underworld3 import visualisation as vis

mesh = uw.meshing.StructuredQuadBox(
minCoords=(0.0, 0.0, 0.0),
maxCoords=(1.0, 1.0, 1.0),
elementRes=(20, 20, 20),
)

vis.plot_mesh(mesh)
"

[lmoresi]

Thanks for the report — we're looking into this.

This could be a genuine mesh connectivity issue in the PETSc/DMPlex pipeline for 3D structured hex meshes, rather than just a rendering artifact. The StructuredQuadBox in 3D goes through gmsh and PETSc mesh import, and there may be an issue with how interior faces or element connectivity are handled.

We'll investigate this properly. A couple of things that would help:

1. Does the issue appear at all resolutions, or only at certain element counts?
2. If you run a simple Poisson solve on this mesh, do you get sensible results?
3. Does uw.meshing.UnstructuredSimplexBox with the same extents show any similar issues?

This is on our list to investigate — will update when we have findings.

[lmoresi]

Fixed in PR #62. PETSc DMPlex uses opposite vertex winding to VTK for 3D cells — both hexahedra and tetrahedra. Applied vertex permutations in mesh_to_pv_mesh() so face normals, shading, clipping, and back-face culling all work correctly. Regression test added.