<div dir="ltr">yes, I am repeatedly calling KSPSolve. I set the operator once at the beginning, but I change some of the elements of the operator between solves without changing the nonzero pattern.</div><div class="gmail_extra">
<br><br><div class="gmail_quote">On Tue, Jan 28, 2014 at 2:58 PM, Jed Brown <span dir="ltr"><<a href="mailto:jed@jedbrown.org" target="_blank">jed@jedbrown.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">David Liu <<a href="mailto:daveliu@mit.edu">daveliu@mit.edu</a>> writes:<br>
<br>
> Hi, I'm writing an application that solves a sparse matrix many times using<br>
> Pastix. I notice that the first solves takes a very long time, while the<br>
> subsequent solves are very fast. I don't fully understand what's going on<br>
> behind the curtains, but I'm guessing it's because the very first solve has<br>
> to read in the non-zero structure for the LU factorization, while the<br>
> subsequent solves are faster because the nonzero structure doesn't change.<br>
<br>
</div>How are you doing this? Repeated calls to KSPSolve? Are you changing<br>
the operator between solves?<br>
</blockquote></div><br></div>