<div class="gmail_quote">On Thu, Jan 5, 2012 at 21:34, Mark F. Adams <span dir="ltr"><<a href="mailto:mark.adams@columbia.edu">mark.adams@columbia.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
This is scary having two communicators running around but the processors that are dropped out of the new communicator have no rows -- that is why they are dropped out.</blockquote></div><br><div>Okay, now I see the "if (ncrs0 != 0)". I also recall that this is only running on a sub-communicator because ParMetis happens to crash if you give it zero vertices on a given proc. I'm inclined to move this communicator shuffle into MatPartitioningApply_Parmetis() so that user code doesn't have to mess with it.</div>