<div style="line-height:1.7;color:#000000;font-size:14px;font-family:arial"><DIV>At 2013-01-16 03:02:32,"Barry Smith" <bsmith@mcs.anl.gov> wrote:<BR>><BR>>> 1.701448294063e+04 > 1.e4*1.145582415879e+00 hence it declares divergence<BR></DIV>
<DIV> Hello, Barry</DIV>
<DIV> I made some tests and it is true. Thanks.</DIV>
<DIV> </DIV>
<DIV> But in the mannual, both version 3.2 and 3.3, the rule is:||rk||>dtol*||b||. It is ||b||, not r_0.</DIV>
<DIV>My misunderstanding? Or the error in the mannual?</DIV>
<DIV> Besides, in the mannual, both version 3.2 and 3.3, the default dtol=1.0E+5. But from the results</DIV>
<DIV>in the example, it is 1.0E+4. I donot know the reason. My petsc version is 3.2-p7.</DIV>
<DIV> Thanks. Jim<BR></DIV>
<DIV> </DIV>
<DIV><BR>>Note that at iteration 171 the preconditioned residual is 9.348832909193e-13 < 1.e-12 * 1.145582415879e+00 very good convergence.<BR>><BR>>You seem to have set an unreasonably tight convergence criteria. In double precision you can never realistically expect to use a rtol smaller than e-12. In fact normally it is not reasonable to use more than like 1.e-8. Those extra digits don't mean anything.<BR>><BR>> Barry<BR>><BR>><BR></DIV></div><br><br><span title="neteasefooter"><span id="netease_mail_footer"></span></span>