You can see in the info output what is happening. The test is<br><br> rnorm >= ksp->divtol*ksp->rnorm0<br><br>Here ||r_0|| = ||b|| = 0.00331536 and ||r|| = 513.979, and thus it results<br>in divergence. It appears that ILU(0) is very bad for this problem.<br>
<br> Matt<br><br><div class="gmail_quote">On Mon, Dec 14, 2009 at 9:07 AM, <span dir="ltr"><<a href="mailto:jarunan@ascomp.ch">jarunan@ascomp.ch</a>></span> wrote:<br><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>
Thank you for your advice. With this mail I attach the output file run.log.<br>
My grid is 2-level grid. The coarse grid is computed first then the fine grid. The problem occur in the fine grid. You can see at lines 3098, 4552 and 6127 where it is diverged.<br>
<br>
Regards,<br>
Jarunan<div><div></div><div class="h5"><br>
<br>
<br>
<br>
<br>
Quoting Matthew Knepley <<a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a>>:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Run with -ksp_view -ksp_monitor -info and send us all the output.<br>
<br>
Matt<br>
<br>
On Mon, Dec 14, 2009 at 8:36 AM, <<a href="mailto:jarunan@ascomp.ch" target="_blank">jarunan@ascomp.ch</a>> wrote:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Dear Petsc Team,<br>
<br>
There is something not clear for me about divergence tolerance. Firstly, my<br>
computation diverges with the reason of dtol (-4), which was set by default<br>
(10000). When I increased dtol to 1.0e+8, it became converged. However, the<br>
relative residual where it was diverged is less than the default dtol. I<br>
would like to know, how it is exactly calculated.<br>
<br>
----------I describe my problem more in detail here.-------------------<br>
As I set the tolerances for the solver as below, my problem diverges with<br>
converged reason -4 (from divergence tolerance)<br>
<br>
call KSPSetTolerances(ksp,1.0e-06,1.0e-20,<br>
PETSC_DEFAULT_DOUBLE_PRECISION,10000,ierr)<br>
<br>
I monitor the residual. it printed the information below and quit the<br>
computation.(I delete the true residual norm and Ax/Ae. Please tell me if<br>
you need them.)<br>
<br>
0 KSP preconditioned resid norm 5.139790011051e+02<br>
<br>
With -ksp_view, it shows that default value of dtol is 10000. I increased<br>
dtol to 1.0e+8, then the solver solved the problem. With<br>
-ksp_monitor_true_residual, the first 5 iterations read:<br>
<br>
0 KSP preconditioned resid norm 5.139790011051e+02<br>
1 KSP preconditioned resid norm 1.840751662590e+01<br>
2 KSP preconditioned resid norm 3.653339750400e+00<br>
3 KSP preconditioned resid norm 2.538794826933e+00<br>
4 KSP preconditioned resid norm 2.299844715275e-01<br>
<br>
Here, from iteration 0 to iteration 1,the relative residual is 0.03580,<br>
much less than 1.0e+4 the default dtol value. So, why was it diverged at the<br>
first time?<br>
<br>
----------------------------------------------------------------------------<br>
<br>
Regards,<br>
Jarunan<br>
<br>
<br>
<br>
<br>
--<br>
Jarunan Panyasantisuk<br>
Development Engineer<br>
ASCOMP GmbH, Technoparkstr. 1<br>
CH-8005 Zurich, Switzerland<br>
Phone : +41 44 445 4072<br>
Fax : +41 44 445 4075<br>
E-mail: <a href="mailto:jarunan@ascomp.ch" target="_blank">jarunan@ascomp.ch</a><br>
<a href="http://www.ascomp.ch" target="_blank">www.ascomp.ch</a><br>
<br>
</blockquote>
<br>
<br>
<br>
--<br>
What most experimenters take for granted before they begin their experiments<br>
is infinitely more interesting than any results to which their experiments<br>
lead.<br>
-- Norbert Wiener<br>
<br>
</blockquote>
<br>
<br>
<br>
-- <br>
Jarunan Panyasantisuk<br>
Development Engineer<br>
ASCOMP GmbH, Technoparkstr. 1<br>
CH-8005 Zurich, Switzerland<br>
Phone : +41 44 445 4072<br>
Fax : +41 44 445 4075<br>
E-mail: <a href="mailto:jarunan@ascomp.ch" target="_blank">jarunan@ascomp.ch</a><br>
<a href="http://www.ascomp.ch" target="_blank">www.ascomp.ch</a><br>
</div></div></blockquote></div><br><br clear="all"><br>-- <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>