Error norm between two nonmatching meshes dolfinx

dolfinx
1

I want to calculate the convergence rate for a problem where the exact solution is not known. As a result, I need to calculate the error norm between two solutions from different sized meshes. In dolfin this was possible using the function errornorm(u8, u16) but the dolfinx equivalent

np.sqrt(domain8.comm.allreduce(fem.assemble_scalar(fem.form(ufl.inner(u8 - u16, u8 - u16) * ufl.dx)), op=MPI.SUM))

gives the error “TypeError: ‘<’ not supported between instances of ‘Mesh’ and ‘Mesh’”

Here u8 and u16 are the solutions to the example Poisson equation where u8 is the solution in an 8x8x8 mesh (defined by mesh.create_unit_cube(MPI.COMM_WORLD, 8, 8, 8, mesh.CellType.tetrahedron)) and u16 is the solution in a 16x16x16 mesh (defined by mesh.create_unit_cube(MPI.COMM_WORLD, 16, 16, 16, mesh.CellType.tetrahedron)).

Is there a way to find the error norm between these two different solutions?
Thanks.

2

See L2 projection of fine mesh solution to a coarse mesh - #6 by dokken Which you can adapt to your needs