From: bangerth Date: Wed, 13 Aug 2008 03:08:02 +0000 (+0000) Subject: Add a comment to the place where we skip copying elements into Trilinos. X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=32c997c1a9522a4e516e2f620104ffc2697a688d;p=dealii-svn.git Add a comment to the place where we skip copying elements into Trilinos. git-svn-id: https://svn.dealii.org/trunk@16532 0785d39b-7218-0410-832d-ea1e28bc413d --- diff --git a/deal.II/examples/step-31/step-31.cc b/deal.II/examples/step-31/step-31.cc index df879ac0a4..7e17aa1e92 100644 --- a/deal.II/examples/step-31/step-31.cc +++ b/deal.II/examples/step-31/step-31.cc @@ -177,8 +177,20 @@ TrilinosAmgPreconditioner::TrilinosAmgPreconditioner( vector_valued_problem (VectorValuedProblem) { - // Init Epetra Matrix, avoid - // storing the nonzero elements. + // Init Epetra Matrix. Even though we build + // up the sparsity pattern to only include + // those entries that actually couple, there + // are quite a number of entries in the + // sparse matrix that are actually zero + // (these are at rows or columns that + // correspond to degrees of freedom that are + // either constrained by hanging nodes or by + // boundary values). We avoid copying these + // nonzero elements to the Trilinos matrix to + // keep the preconditioner as compact as + // possible. While this increases the number + // of outer iterations by a tiny bit but + // makes the overall scheme faster. { Map.reset (new Epetra_Map(n_u, 0, communicator));