[petsc-users] rtol meaning

[Matthew Knepley]

On Thu, May 29, 2014 at 11:31 AM, Danny Lathouwers - TNW <
D.Lathouwers at tudelft.nl> wrote:

>  Dear users,
>
>
>
> I have a problem where I step time and repeatedly solve a system with
> differing rhs.
>
> At some time step petsc solver returns that initial solution is good
> enough (converged reason = 2 with 0 iterations done).
>
> I do not expect this behaviour. I use rtol = 0.001, atol=0 and dtol =large
> number.
>
>
>
> The manual seems to suggest the criterion is: rnorm < MAX (rtol * rnorm_0, abstol) (probably based on preconditioned residual).
>
> How could this lead to zero iterations being done? Or is the criterion based on rnorm/bnorm instead (which I found in some reference on the internet concerning petsc and would explain the observed behaviour).
>
> Its ||b||, not ||r_0||. You can change it to get the other behavior, as
detailed here


http://www.mcs.anl.gov/petsc/petsc-dev/src/ksp/ksp/interface/iterativ.c.html#KSPConvergedDefault

     Matt


> Thanks,
>
> Danny.
>
>
>
>
>
>
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140529/fa9f6a24/attachment.html>