On Thu, May 7, 2009 at 4:09 PM, Shao-Ching Huang <span dir="ltr"><<a href="mailto:schuang@ats.ucla.edu">schuang@ats.ucla.edu</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi,<br>
<br>
In a parallel PCG solve, is there a way to instruct PETSc to call my own (serial) solver to invert the per-process, local precondition matrix (as in the block-Jacobi preconditioner)?</blockquote><div><br>It sounds like the proper way to do this is to use -pc_type bjacobi -sub_ksp_type preonly and then pull out<br>
then replace the block PC with a PCSHELL with wraps up your preconditioner.<br><br> Matt<br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Is there an example/documentation that I can follow? Which part of code should I start looking at?<br>
<br>
We are trying to determine if there is any value to use the aforementioned customized preconditioner as compared to the general ones in PETSc and to DMMG, for this particular matrix at hand. This is a structured mesh problem.<br>
<br>
Thank you.<br><font color="#888888">
<br>
Shao-Ching<br>
</font></blockquote></div>-- <br>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener<br>