Though if we require "non-coincident nodes", it would be helpful if MeshTools::Private::globally_renumber_nodes_and_elements could detect coincident nodes and throw an error? I haven't looked into the details though so I don't know if that's feasible...?
On 06/20/2013 09:34 AM, David Knezevic wrote: > It appears that the "partition agnostic" renumbering causes problems > on "slit" meshes, i.e. meshes with coincident nodes on unconnected > elements. > > You can see this from the misc_ex9 example that I "pull requested": If > I do equation_systems.write at the end, and then subsequently read in > that solution data, the solution is mixed up on the slit. I guess this > isn't surprising since I gather > MeshTools::Private::globally_renumber_nodes_and_elements works based > on node locations, so coincident nodes presumably cause problems. > > I've attached a figure showing the "mixed up" solution, though I > assume it'll get stripped by the list so email me separately if you > want me to send it directly. > > I resolved this by making a new mesh with a slit of size 1.e-6, and > that works fine. If we want to require "non-coincident nodes" then > that's fine with me, but I just wanted to bring this issue to the > attention of the list. > > David > ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
