On Dec 7, 2010, at 1:50 PM, Jed Brown wrote: > On Tue, Dec 7, 2010 at 20:45, Barry Smith <bsmith at mcs.anl.gov> wrote: > Why can you not use VecAXPY() on the local Vecs? > > You're absolutely right. I must have been thinking of error control when I > decided this was a problem. You need some sort of norms and don't want to > double-count the ghost entries.
Yes, norms are an issue in the case of DMDA. When using ghosted local vectors norms are fine. BTW: I consider the DMDA Vec just a very special case that gets way to much attention and can sometimes distort the overall design of PETSc in issues that are not important in that non-special case. Barry > > Jed
