<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Thu, Sep 11, 2014 at 3:24 PM, Evan Um <span dir="ltr"><<a href="mailto:evanum@gmail.com" target="_blank">evanum@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Dear PETSC and MUMPS users,</div><div><br></div><div>I try to use an iterative refinement option (ICNTL(10)=max # of iterative refinement) of MUMPS in my PETSC application. MUMPS manual says that if the solution is kept distributed (ICNTL(21)=1), the iterative refinement option is disabled. When a problem is solved using KSPSolve() with multiple cores, the solution is automatically kept distributed over the processors. Does this mean that iterative refinement of MUMPS is available from PETSC only when we run our application on a single core? I tried to solve my problem on multiple cores with ICNTL(10)=10 and INCTL(21)=0, but the program crashed. Does anyone know how to use iterative refinement option of MUMPS from PETSC on a parallel computer? In advance, thanks for your comments.</div></div></blockquote><div><br></div><div>Just use Newton to solve your system, which for a linear operator is iterative refinement.</div><div><br></div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Regards,</div><div>Evan</div><div><br></div><div><br></div><div> </div></div>
</blockquote></div><br><br clear="all"><div><br></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
</div></div>