On 12/10/12 2:00 PM, Roy Stogner wrote: > I'm currently looking at the parallel-correctness problems Ataollah > Mesgarnejad and Vikram Garg are having, and it appears that the simple > workaround for both of them involves sticking the right > ghosted_vector.localize(ghosted_vector) call in the right place... > > Except, I can't find the right NumericVector::localize() call! All > our localize-from-NumVec-to-NumVec calls seem to be designed around > the assumption that the source and target vectors are different, and > therefore we need to spend a bunch of time copying local coefficients > around rather than just ghosted coefficients. Am I misinterpreting > our localize() methods, or do we need to add something new for > ghosted vector localization-to-self communications?
If you are trying to re-fill the ghost values (e.g. in a ghosted PETSc Vec), you can use PetscVector::close(). There may be a small cost related to VecAssemblyBegin()/VecAssemblyEnd(). -- Boyce ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
