On Sep 1, 2010, at 4:36 PM, Roy Stogner wrote: > Initialize the vector e.g. as GHOSTED with size 0 and an empty ghost > index set?
Really? You would really prefer that? Seems like a hack to me. We store other info about vectors this same way... like whether or not to project them. Should we put that information into NumericVector as well? BTW - is it ok if I add_vector() solution and current_local_solution other than in the constructor? Are there pieces of code that actually depend on solution and current_local_solution being built but not initialized in the constructor like they are now? It makes it hard to carry out this plan because there isn't any way to set default_parallel_vector_type before the constructor.... I'm actually thinking of abandoning this whole idea. These are quite a lot of changes to libMesh for something that is kind of auxiliary to the actual problem of needing ghosted vectors in the newton update check callback. I'm actually thinking now of just creating my own ghosted petsc vectors and vec copying into them using those vectors for the duration of the check and then destroying them (or holding onto them and just vec copying into them each time... either way). Derek ------------------------------------------------------------------------------ This SF.net Dev2Dev email is sponsored by: Show off your parallel programming skills. Enter the Intel(R) Threading Challenge 2010. http://p.sf.net/sfu/intel-thread-sfd _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
