On Fri, 6 Nov 2015, Miguel Angel Salazar de Troya wrote: >> I found out I can use elem->id () == mesh.processor_id() instead of >> elem->is_remote().
This is what the _local_ iterators (that John recommended, IIRC) do for you. >> However, there's a problem when I iterate over all the >> elements and it's that not all processors iterate over all the elements. Are you using SerialMesh? If so, then your mesh has become inconsistent. That's not something we can protect against (we don't do anything mesh-breaking in the library, but we can't stop you from doing so in user code). You'll probably have to use some of the dbg-mode MeshTools tests (or your own code snippet) to locate the problem. Are you instead using ParallelMesh? If so, then you've just discovered the whole point of ParallelMesh: processors don't have to store every element in the mesh, only those they own and a layer of "ghosts" around them. --- Roy ------------------------------------------------------------------------------ _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
