On 14/12/15 14:40, Lawrence Mitchell wrote:
Dear Henrik,
On 14/12/15 14:24, Buesing, Henrik wrote:
Dear all,
I tried to create a simple mesh in gmsh with a hole in it and would like to import this into firedrake. When I try to use this mesh with
mymesh = Mesh("test1.msh")
elt = FiniteElement("CG", "triangle", 1)
CG1 = FunctionSpace(mymesh, elt)
I get the error below.
Any ideas why? I attached the mesh. Thank you!
This looks like a bug, either in our mesh reading, or somewhere a bit further down the line.
Could you please file a github issue for the problem so we can track it sensibly.
Hmm, on closer inspection it looks like you have two "rogue" points in your mesh file that are not part of any cell (at the center of the mesh in the hole). If I remove those points then I can read the mesh successfully. What happens is that we compute a numbering of vertices (and edges) in the mesh by traversing the cells in order and numbering all the vertices we see attached to each cell. But your mesh has some vertices not attached to any cells, so we didn't number them, and then you get a nasty error later on. We could handle this case more gracefully, so if you can still report the issue, that would be great. Cheers, Lawrence