<html><body><div style="color:#000; background-color:#fff; font-family:HelveticaNeue, Helvetica Neue, Helvetica, Arial, Lucida Grande, sans-serif;font-size:12pt"><div><span>Do you mean using "<font size="3" face="NimbusMonL-Regu"><font size="3" face="NimbusMonL-Regu">-pc_factor_mat_ordering_type wbm"? It does not seem to be in the manual. What exactly is it? I will try it anyway.</font></font></span></div><div><span><font size="3" face="NimbusMonL-Regu"><font size="3" face="NimbusMonL-Regu"></font></font></span> </div><div><span><font size="3" face="NimbusMonL-Regu"><font size="3" face="NimbusMonL-Regu">Many thanks,</font></font></span></div><div><span><font size="3" face="NimbusMonL-Regu"><font size="3" face="NimbusMonL-Regu">Qin</font></font></span></div><font size="3" face="NimbusMonL-Regu"><font size="3" face="NimbusMonL-Regu"><div></div></font><div></div></font><div></div><div style="display: block;" class="yahoo_quoted"> <br> <br> <div
style="font-family: HelveticaNeue, Helvetica Neue, Helvetica, Arial, Lucida Grande, sans-serif; font-size: 12pt;"> <div style="font-family: HelveticaNeue, Helvetica Neue, Helvetica, Arial, Lucida Grande, sans-serif; font-size: 12pt;"> <div dir="ltr"> <font size="2" face="Arial"> On Tuesday, February 18, 2014 10:11 AM, Jed Brown <jed@jedbrown.org> wrote:<br> </font> </div> <div class="y_msg_container">Qin Lu <<a href="mailto:lu_qin_2000@yahoo.com" shape="rect" ymailto="mailto:lu_qin_2000@yahoo.com">lu_qin_2000@yahoo.com</a>> writes:<div id="yqtfd91444" class="yqt5567078419"><br clear="none"><br clear="none">> RCM did not help much for my case. This case represents some strong<br clear="none">> connectivities in channels. I am wondering if PETSc has some<br clear="none">> reordering algorithm or solver that is not based on connectivity<br clear="none">> graph, but is based on connectivity strength (such as percolation type<br
clear="none">> of reordering). Any information is appreciated. Thanks, Qin</div><br clear="none"><br clear="none">There isn't such an ordering, but it would be a welcome contribution.<br clear="none">One challenge is that features like anisotropy are not necessarily<br clear="none">apparent in the matrix entries. In that case, and for vector-valued<br clear="none">problems, you would ideally use a better strength-of-connection measure.<br clear="none">If you computed a strength-of-connection, then thresholded (or used edge<br clear="none">weights as a "priority"), followed by something like an RCM ordering,<br clear="none">you could find something good for low-fill incomplete factorization in<br clear="none">problems with hidden anisotropy.<br><br></div> </div> </div> </div> </div></body></html>