<div class="gmail_quote">On Tue, Dec 21, 2010 at 21:16, Vijay S. Mahadevan <span dir="ltr"><<a href="mailto:vijay.m@gmail.com">vijay.m@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 id=":1cz">Jed, I ask because after the restart, the residual changes 10 orders<br>
of magnitude and a-priori, it is quite hard to decide the restart<br>
number. Yes in the test case I presented, the residual gets close<br>
enough to the tolerance and I can afford few more vector storage but<br>
for a much refined problem, this might not be the case and so it<br>
worries me.<br></div></blockquote><div><br></div><div>What happens if you run with -ksp_gmres_modifiedgramschmidt? This is slow in parallel, but provides insight into what is causing the problem.</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div id=":1cz">
My initial tests with bcgs were not satisfactory (very bad convergence<br>
as compared to gmres) but I tried GCR just now and it seems to<br>
converge correctly to the right solution, monotonically for the same<br>
problem.</div></blockquote><div><br></div><div>GCR provides a cheap way to access the solution, see what it does with monitor_true_residual.</div><div><br></div></div>