<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">Hello Markus, one more question: since I have complex values matrices,<div><br></div><div>i.e. <span class="Apple-style-span" style="color: rgb(5, 97, 66); font-family: Monaco; font-size: 11px; "><span style="color: #000000">Dune::</span></span><span class="Apple-style-span" style="color: rgb(5, 97, 66); font-family: Monaco; font-size: 11px; ">BCRSMatrix</span><span class="Apple-style-span" style="color: rgb(5, 97, 66); font-family: Monaco; font-size: 11px; "><span style="color: #000000"><</span></span><span class="Apple-style-span" style="color: rgb(5, 97, 66); font-family: Monaco; font-size: 11px; ">FieldMatrix<span style="color: #000000"><</span>ComplexValueType<span style="color: #000000">,1,1> </span></span><span class="Apple-style-span" style="font-family: Monaco; font-size: 11px; ">></span></div><div><br></div><div>how will Dune's SuperLU connection react to this ?</div><div><br></div><div>Will it automatically choose the complex variant of SuperLU ?</div><div><br></div><div>Thanks and greetings, Benedikt</div><div><br></div><div><br><div><div>On Nov 22, 2011, at 8:02 PM, Markus Blatt wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div>Hi Benedikt,<br><br><br>On Tue, Nov 22, 2011 at 04:44:31PM +0100, Benedikt Oswald wrote:<br><blockquote type="cite">Dear Dune,<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">I need to instantiate the AMG preconditioner so that I can specify a specific coarseSolver<br></blockquote><blockquote type="cite">since my problem does not work with the default behavior of AMG, i.e.<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">"Construct an AMG with an inexact coarse solver based on the smoother.<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">As coarse solver a preconditioned CG method with the smoother as preconditioner will be used. The matrix hierarchy is built automatically."<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">Thus, I would like to use this constructor<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">Dune::Amg::AMG< M, X, S, PI, A >::AMG<span class="Apple-tab-span" style="white-space:pre"> </span>(<span class="Apple-tab-span" style="white-space:pre"> </span>const OperatorHierarchy & <span class="Apple-tab-span" style="white-space:pre"> </span>matrices,<br></blockquote><blockquote type="cite">CoarseSolver & <span class="Apple-tab-span" style="white-space:pre"> </span>coarseSolver,<br></blockquote><blockquote type="cite">const SmootherArgs & <span class="Apple-tab-span" style="white-space:pre"> </span>smootherArgs,<br></blockquote><blockquote type="cite">const Parameters & <span class="Apple-tab-span" style="white-space:pre"> </span>parms<span class="Apple-tab-span" style="white-space:pre"> </span> <br></blockquote><blockquote type="cite">)<br></blockquote><br>May I ask why you cannot live with the default behaviour?<br><br>If you have parmetis and SuperLU installed, the default behaviour is<br>to agglomerate the data onto fewer processor until eventualy all data<br>on the coarsest level is on only one processor. To solve this coarse<br>level system SuperLU is solved.<br><br>I always used this approach.<br><br>If that is not sufficient, I need to take a look at the code for<br>specifying the coarsesolver, which I fear has not been tested lately.<br><br>Cheers,<br><br>Markus<br><br>-- <br>Do you need more support with DUNE or HPC in general? <br><br>Dr. Markus Blatt - HPC-Simulation-Software & Services <a href="http://www.dr-blatt.de">http://www.dr-blatt.de</a><br>Rappoltsweilerstr. 5, 68229 Mannheim, Germany<br>Tel.: +49 (0) 160 97590858 Fax: +49 (0)322 1108991658 <br><br>_______________________________________________<br>Dune mailing list<br><a href="mailto:Dune@dune-project.org">Dune@dune-project.org</a><br>http://lists.dune-project.org/mailman/listinfo/dune<br></div></blockquote></div><br><div apple-content-edited="true">
<div style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><div>--------------------------------------------------------------------------------------------------------------------------------------------------------------------</div><div>Benedikt Oswald, Dr. sc. techn., dipl. El. Ing. ETH, <a href="http://www.psi.ch">www.psi.ch</a>, Computational Accelerator Scientist, <a href="http://amas.web.psi.ch">http://amas.web.psi.ch</a></div><div>Paul Scherrer Institute (PSI), CH-5232 Villigen, Suisse, <a href="mailto:benedikt.oswald@psi.ch">benedikt.oswald@psi.ch</a>, +41(0)56 310 32 12</div><div><br></div><div>"Passion is required for any great work, and for the revolution passion and audacity are required in big doses.",<br> Ernesto 'Che' Guevara, Letter to his parents.<br>--------------------------------------------------------------------------------------------------------------------------------------------------------------------</div></div></div>
</div>
<br></div></body></html>