On Tue, Mar 31, 2009 at 11:40 AM, Roy Stogner <roystgnr@ices.utexas.edu> wrote:
The easiest test case would be to loop over all semilocal nodes,
assign boundary ids that are some hash of node locations, hit the mesh
with some repartitioning steps, then loop over all semilocal nodes
again and make sure the ids still obey the same hash.

I might be able to do something like this... what is the easiest way to get a random repartitioning?