Open jorgensd opened 2 years ago
I suspect this issue might be fixed by https://github.com/FEniCS/dolfinx/pull/2386
This is not fixed yet. There are issues with the sub-spaces, i.e.
V.sub(i)
does not work.
Is the issue related to this by any chance? i.e. does specifying dim=mesh.geometry.dim
when creating the vector function space fix the problem?
Is the issue related to this by any chance? i.e. does specifying
dim=mesh.geometry.dim
when creating the vector function space fix the problem?
I've opened a PR in Basix to fix that and maybe fix this
@mscroggs is this resolved?
@mscroggs, @jorgensd is this fixed now?
No, this is still not resolved.
You get the issue by adding V_v.sub(i)
i
being 0, 1 or 2 to the example above. You then get
issues when creating the sub function space, as the sub elements are using the wrong cell type
Is there a sensible test we can add to DOLFINx that will check that we don't break this in future?
This may have been fixed by https://github.com/FEniCS/basix/pull/617
Is there a sensible test we can add to DOLFINx that will check that we don't break this in future?
Good question, maybe we need to solve a PDE on a manifold, or do something similar to what we do above. I guess we could use locate_dofs_topological on a subspace to compare it to what one gets on the collapsed sub space.
This may have been fixed by FEniCS/basix#617
Is this issue now fixed?
Broken by PR #2301 Reported at: https://fenicsproject.discourse.group/t/triangle-elements-in-3d-broken-in-dolfinx/9426 MWE: XDMFFile:
and code
yields error: