That is indeed the case. Petsc allows you to overestimate with no major issue, so it's faster to build the graph that way. Others require the full graph, which is more expensive to build.
> On Oct 13, 2014, at 5:17 PM, David Knezevic <dkneze...@seas.harvard.edu> > wrote: > > I was looking into the sparsity pattern code a bit, and I was wondering if > the !need_full_sparsity_pattern branch in > SparsityPattern::Build::parallel_sync() over-estimates n_nz and n_oz? It > doesn't do the equivalent of the my_row.erase operation that we have in the > need_full_sparsity_pattern branch, and hence it presumably can double-count > some dofs, no? > > This isn't a big deal for me, but I was just curious if this is indeed the > case. > > David > > > ------------------------------------------------------------------------------ > Comprehensive Server Monitoring with Site24x7. > Monitor 10 servers for $9/Month. > Get alerted through email, SMS, voice calls or mobile push notifications. > Take corrective actions from your mobile device. > http://p.sf.net/sfu/Zoho > _______________________________________________ > Libmesh-devel mailing list > Libmesh-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/libmesh-devel ------------------------------------------------------------------------------ Comprehensive Server Monitoring with Site24x7. Monitor 10 servers for $9/Month. Get alerted through email, SMS, voice calls or mobile push notifications. Take corrective actions from your mobile device. http://p.sf.net/sfu/Zoho _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
