[cig-commits] [commit] master: Add some commentary. (ef6253b)

[cig_noreply at geodynamics.org]

Repository : https://github.com/geodynamics/aspect

On branch  : master
Link       : https://github.com/geodynamics/aspect/compare/36e4420c87fda8026b8a3d1f639e798f6c59f3d0...01f19ef36177d8c77f84b59ece4918ba5f3e7661

>---------------------------------------------------------------

commit ef6253b6f66e87e30ad13995832e1511756ebf15
Author: Wolfgang Bangerth <bangerth at math.tamu.edu>
Date:   Mon Dec 8 19:45:26 2014 -0600

    Add some commentary.


>---------------------------------------------------------------

ef6253b6f66e87e30ad13995832e1511756ebf15
 source/simulator/assembly.cc | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/source/simulator/assembly.cc b/source/simulator/assembly.cc
index 06be596..3c427dc 100644
--- a/source/simulator/assembly.cc
+++ b/source/simulator/assembly.cc
@@ -1027,6 +1027,16 @@ namespace aspect
     Amg_data.constant_modes = constant_modes;
     Amg_data.elliptic = true;
     Amg_data.higher_order_elements = true;
+
+    // set the AMG parameters in a way that minimizes the run
+    // time. compared to some of the deal.II tutorial programs, we
+    // found that it pays off to set the aggregration threshold to
+    // zero, especially for ill-conditioned problems with large
+    // variations in the viscosity
+    //
+    // for extensive benchmarking of various settings of these
+    // parameters and others, see
+    // https://github.com/geodynamics/aspect/pull/234
     Amg_data.smoother_sweeps = 2;
     Amg_data.aggregation_threshold = 0.001;
 #endif