Cell numbering: inconsistent gmsh <=> dolfinx numbering

[dokken]

@pa-beaufort

here it is the numbering of cells instead

I specifically mention cell indices:

@dokken

Secondly, we do not associate the original_cell_index with the index it had in GMSH, as in the msh format, there might be cells that are not part of the actual mesh.
Following is a code that highlights that the vertices have the correct global numbering, but it then also shows you that for DOLFINx to read a msh file, we extract cells by their physical group, and order them in an array (which the input global index is relative to):

and I even show how to extract the cells in the correct way using GMSH:

@dokken

import gmsh
gmsh.initialize()
gmsh.model.add("Mesh from file")
gmsh.merge("mh104.msh")

# Extract cells in original order
topologies_2 = []
entities = gmsh.model.getEntities()
entities_sorted = np.sort([e[1] for e in entities])
for obj in entities_sorted:
    (entity_types, entity_tags, entity_topologies) = gmsh.model.mesh.getElements(2, tag=obj)
    assert len(entity_types) == 1
    properties = gmsh.model.mesh.getElementProperties(entity_types[0])
    name, dim, _, num_nodes, _, _ = properties
    topology = entity_topologies[0].reshape(-1, num_nodes)
    topologies_2.append(topology)

topologies_2 = np.vstack(topologies_2)
print(topologies_2[1115])