On Tue, 2 Aug 2016, Harriet Li wrote: > MeshRefinement::refine_elements() to refine them, like in Misc Ex 4. > However, when I call EquationSystems::reinit(), I get > " > Assertion `!libmesh_assert_pass' failed. > " > with the backtrace > " > #0 libMesh::MeshRefinement::test_level_one (this=0x7fffffffc950, > > It appears that somewhere, MeshRefinement::_face_level_mismatch_limit > becomes 1
> I'm I missing something (a flag somewhere?), or does projecting on reinit > require the mesh satisfy the "level one" rule? This is a bug in libMesh. We're creating a new internal MeshRefinement object inside the EquationSystems so we can use it to do refinement and coarsening separately, but we're leaving its conformity rules at their defaults rather than trusting that the user already got their flags correct to their own satisfaction via their own MeshRefinement settings. Try setting face_level_mismatch_limit() to unlimited on the MeshRefinement object in EquationSystems::reinit(). If that works for you, then I'll do the same with overrefined_boundary_limit and underrefined_boundary_limit and I'll make up a PR. Thanks for catching this and sorry about the hassle, --- Roy ------------------------------------------------------------------------------ _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
