Can I consider that each processor have all the mesh data, such that looping the whole mesh does not cause extra communication?
On Wed, Apr 16, 2014 at 9:41 PM, Kirk, Benjamin (JSC-EG311) < [email protected]> wrote: > Correct. If you only iterate over local elements and modify the mesh it > will become inconsistent across processors; and you'd need to fix that with > communication. So it's cleaner to loop over all elements. > > On Apr 16, 2014, at 9:35 PM, "walter kou" <[email protected]> wrote: > > So I guess, If I want to modify the subdomain_id, I will also need to > use mesh.elements_{begin,end}, > > Right? > > > On Wed, Apr 16, 2014 at 9:33 PM, walter kou <[email protected]> wrote: > >> It works with mesh.elements_{begin,end}(). >> >> Thanks, >> >> >> >> >> On Wed, Apr 16, 2014 at 8:42 PM, Kirk, Benjamin (JSC-EG311) < >> [email protected]> wrote: >> >>> You're modifying the mesh, so you need to be sure to do it the same on >>> all processors. In this case, the culprit I think is looping over the >>> local elements. What happens when you loop over >>> mesh.elements_{begin,end}(); instead? >>> >>> -Ben >>> >>> On Apr 16, 2014, at 8:35 PM, walter kou <[email protected]> wrote: >>> >>> > 1) Given a 3D mesh, I try to write a separate code to add boundary id, >>> > using: >>> > >>> > Mesh::element_iterator it_el = >>> > mesh.active_local_elements_begin(); >>> > const Mesh::element_iterator it_last_el = >>> > mesh.active_local_elements_end(); >>> > for ( ; it_el != it_last_el ; ++it_el) { >>> > Elem* elem = *it_el; >>> > // some code to detect null-neighbour >>> > mesh.boundary_info->add_side(elem->id(), side, bc_code); >>> > >>> > equation_systems.init(); >>> >>> >> > ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/NeoTech _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
