<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Fri, Nov 6, 2015 at 9:15 AM, Denis Davydov <span dir="ltr"><<a href="mailto:davydden@gmail.com" target="_blank">davydden@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Hong,<br>
<span class=""><br>
> On 6 Nov 2015, at 16:09, Hong <<a href="mailto:hzhang@mcs.anl.gov">hzhang@mcs.anl.gov</a>> wrote:<br>
><br>
> Denis:<br>
> Do you use shift-and-invert method for solving eigenvalue problem?<br>
</span>no, it’s just shift with zero value. So for GHEP one inverts B-matrix.<br>
<span class=""><br>
> If so, the linear problems would be extremely ill-conditioned, for which the direct solver, such LU or Cholesky are usually the only working option.<br>
</span>Depends on the shift, i would say.<br>
In any case the same problem works with jacobi preconditioner no no other changes,<br>
so i would not relate it to any settings on SLEPc part.</blockquote><div><br></div><div>Is it possible that the matrix is rank deficient? Jacobi will just chug along and sometimes work, but</div><div>AMG will fail spectacularly in that case.</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"><span class=""><br>
> You may run your petsc/slepc code with option '-ksp_monitor' to observe convergence behavior.<br>
</span>Will do, thanks.<br>
<br>
Regards,<br>
Denis.<br>
<br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">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></div>