<p dir="ltr">You need --download-superlu  due to licensing restrictions. And you need the development version.</p>
<p dir="ltr">   Matt</p>
<div class="gmail_quote">On Feb 18, 2014 11:25 AM, "Qin Lu" <<a href="mailto:lu_qin_2000@yahoo.com">lu_qin_2000@yahoo.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><div style="font-size:12pt;font-family:HelveticaNeue,Helvetica Neue,Helvetica,Arial,Lucida Grande,sans-serif"><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"> <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 face="Arial"> On Tuesday, February 18, 2014 10:11 AM, Jed Brown <<a href="mailto:jed@jedbrown.org" target="_blank">jed@jedbrown.org</a>> wrote:<br> </font> </div>  <div>Qin Lu <<a href="mailto:lu_qin_2000@yahoo.com" shape="rect" target="_blank">lu_qin_2000@yahoo.com</a>> writes:<div>
<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></div></blockquote></div>