On Nov 9, 2011, at 5:06 PM, Jed Brown wrote: > On Wed, Nov 9, 2011 at 16:58, Barry Smith <bsmith at mcs.anl.gov> wrote: > Error in parallel not sequential. > > Okay, just note that the concept of a local space that is different from a > global space is useful even in serial (e.g. ghost points for boundary > conditions or periodic). The local form is just the domain of the > LocalToGlobalMapping.
Understood, thus there is the possiblity of user error if they are ghosting but process a non-ghosted vector. I think the fact that basic stuff runs unchanged on one process with seq (for non-periodic stuff) out ways the risk of this problem. Barry