> template<class Btype>
> void Pa_Preconditioner<Btype>::
> vmult(LA::MPI::BlockVector &dst, const LA::MPI::BlockVector &src) const
> {
> SolverControl cn;
> PetScWrappers::SparseDirectMUMPS solver(cn,mpi_communicator);
> solver.set_symmetric_mode(true);
> solver.solve(B,dst.block(0),src.block(0));
> solver.solve(B,dst.block(1),src.block(1));
> }
>
> Does this look reasonable? I think, this makes program much less general
> than my previous version where I could choose between direct and iterative
> solvers with the template parameter. And also I am not sure that LDLt
> decomposition of B is reused in this case.It looks like you are reconstructing the decomposition every time inside your vmult. Why do you do that? -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
