Is there an efficient way to preserve the GMSH boundary numbering scheme on submeshes generated from the global mesh?

Question

I want to find a way to efficiently transfer the number scheme for the boundaries I have specified in GMSH to two submeshes I've generated using the following code:

submesh1 = SubMesh(mesh, subdomains, 1)
subdomain1 = CellFunction("size_t", submesh1)
boundaries1 = FacetFunction("size_t", submesh1)

V1 = FunctionSpace(submesh1, "CG", 1)
dx1 = Measure("dx", domain=submesh1)

submesh2 = SubMesh(mesh, subdomains, 1)
subdomain2 = CellFunction("size_t", submesh2)
boundaries2 = FacetFunction("size_t", submesh2)

V2 = FunctionSpace(submesh2, "CG", 1)
dx2 = Measure("dx", domain=submesh2)

If it helps, I have generated a global mesh in GMSH and I have identified several boundaries--in particular an interfacial boundary between subdomain1 and subdomain2 and two Dirichlet boundaries I want to specify on subdomain 1. Is there an efficient way to do this?

Thanks!

Answer 1

I've been trying to address this question on and off recently, and I'm now wondering if statements like

boundaries1 = FacetFunction("size_t", submesh1)
parent_IDs = boundaries1.data().array('parent_vertex_indices', 0)
boundaries1 = [rename_boundaries.array()[i] for i in range(len(parent_IDs))]

or something like this could be used to obtain the mapping between a FacetFunction() that identifies boundary surface labels on the main or "global" mesh and the surfaces of a submesh.

For example, a smaller cube immersed in a larger cube. If I use GMSH to explicitly label two faces of the larger cube, then I make a submesh of each domain (the larger cube sans the smaller one for the first submesh and the smaller cube for the second), how do I identify the labels from the main mesh to the submesh in order to apply them during my solution process?