]> https://gitweb.dealii.org/ - dealii-svn.git/commitdiff
Some more changes in the AMG settings. The option Uncoupled seems fine up to very...
authorkronbichler <kronbichler@0785d39b-7218-0410-832d-ea1e28bc413d>
Fri, 9 Jan 2009 19:53:11 +0000 (19:53 +0000)
committerkronbichler <kronbichler@0785d39b-7218-0410-832d-ea1e28bc413d>
Fri, 9 Jan 2009 19:53:11 +0000 (19:53 +0000)
git-svn-id: https://svn.dealii.org/trunk@18162 0785d39b-7218-0410-832d-ea1e28bc413d

deal.II/lac/source/trilinos_precondition.cc

index f423d1932fada0d12152f3086207ee623385eab9..ea3cc0db6d734ffdfecf8291807d49868e9efa14 100755 (executable)
@@ -568,27 +568,30 @@ namespace TrilinosWrappers
        ML_Epetra::SetDefaults("SA",parameter_list);
        parameter_list.set("smoother: type", "Chebyshev");
 
-                                  // uncoupled mode gives a lot of warnings
-                                  // when there are too many entries per
-                                  // row and aggreggation gets complicated,
-                                  // but MIS does not work if too few
+                                  // uncoupled mode can give a lot of
+                                  // warnings or even fail when there
+                                  // are too many entries per row and
+                                  // aggreggation gets complicated, but
+                                  // MIS does not work if too few
                                   // elements are located on one
-                                  // processor. work around these warnings
-                                  // by choosing the different strategies
-                                  // in different situations: for low
-                                  // order, always use the standard choice
-                                  // uncoupled. if higher order, use
-                                  // Uncoupled with few dofs and MIS
-                                  // with many dofs per processor
+                                  // processor. work around these
+                                  // warnings by choosing the different
+                                  // strategies in different
+                                  // situations: for low order, always
+                                  // use the standard choice
+                                  // uncoupled. if higher order, right
+                                  // now we also just use Uncoupled,
+                                  // but we should be aware that maybe
+                                  // MIS might be needed
                                   // 
-                                  // TODO: Maybe there are some better
-                                  // options?
+                                  // TODO: Maybe there are any
+                                  // other/better options?
        if (additional_data.higher_order_elements)
          {
-           if (matrix.m()/matrix.matrix->Comm().NumProc() < 5000)
-             parameter_list.set("aggregation: type", "Uncoupled-MIS");
-           else
-             parameter_list.set("aggregation: type", "MIS");
+           //if (matrix.m()/matrix.matrix->Comm().NumProc() < 50000)
+           parameter_list.set("aggregation: type", "Uncoupled");
+           //else
+           //parameter_list.set("aggregation: type", "MIS");
          }
       }
     else

In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move.

Douglas Adams


Typeset in Trocchi and Trocchi Bold Sans Serif.