<div dir="ltr"><div dir="ltr">On Fri, Jun 5, 2020 at 6:12 AM Y. Shidi <<a href="mailto:ys453@cam.ac.uk">ys453@cam.ac.uk</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Dear developers,<br>
<br>
We are using filed splitting method solve saddle point problem.<br>
Below is our ksp configuration:<br>
<br>
   PetscOptionsSetValue(NULL,"-ksp_type","gmres");<br>
   PetscOptionsSetValue(NULL,"-ksp_initial_guess_nonzero","");<br>
   PetscOptionsSetValue(NULL,"-ksp_rtol","1e-20");<br></blockquote><div><br></div><div>This algebraic tolerance is strange. Why put it below unit roundoff?</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
   PetscOptionsSetValue(NULL,"-ksp_converged_reason", "");<br>
   PetscOptionsSetValue(NULL,"-pc_type", "fieldsplit");<br>
   PetscOptionsSetValue(NULL,"-pc_fieldsplit_type", "schur");<br>
   PetscOptionsSetValue(NULL,"-pc_fieldsplit_schur_fact_type", "lower");<br>
   PetscOptionsSetValue(NULL,"-pc_fieldsplit_schur_precondition", <br>
"selfp");<br>
   PetscOptionsSetValue(NULL,"-fieldsplit_0_ksp_type", "preonly");<br>
   PetscOptionsSetValue(NULL,"-fieldsplit_0_pc_type", "hypre");<br>
   PetscOptionsSetValue(NULL,"-fieldsplit_1_ksp_type", "preonly");<br>
   PetscOptionsSetValue(NULL,"-fieldsplit_1_pc_type", "hypre");<br>
<br>
It can be converged. However, when we turn on the ksp monitor:<br>
<br>
   PetscOptionsSetValue(NULL,"-ksp_monitor_true_residual", "");<br>
<br>
It won't converge. It seems very odd, I am wondering why it happens.<br></blockquote><div><br></div><div>Can you try it with a relative tolerance of 1e-10, and send the output (with -ksp_view as well).</div><div><br></div><div>  Thanks,</div><div><br></div><div>     Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
Thank you for your time and help.<br>
<br>
Kind regards,<br>
Shidi<br>
</blockquote></div><br clear="all"><div><br></div>-- <br><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div>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><br></div><div><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank">https://www.cse.buffalo.edu/~knepley/</a><br></div></div></div></div></div></div></div></div>