[petsc-users] Convergence trouble with SNES and CG

[Matthew Knepley]

On Mon, Feb 22, 2016 at 6:27 PM, David Knezevic <david.knezevic at akselos.com>
wrote:

> On Mon, Feb 22, 2016 at 7:10 PM, Matthew Knepley <knepley at gmail.com>
> wrote:
>
>> On Mon, Feb 22, 2016 at 6:02 PM, David Knezevic <
>> david.knezevic at akselos.com> wrote:
>>
>>> On Mon, Feb 22, 2016 at 6:57 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
>>>
>>>>
>>>>   Add -ksp_converged_reason
>>>>
>>>
>>>
>>> That gives "DIVERGED_INDEFINITE_PC". The full output is below:
>>>
>>
>> Just for testing, replace CG with GMRES.
>>
>
> With GMRES the first four nonlinear iterations behave much that same as
> before, converging at KSP iteration 150 or so.
>
> On the 5th nonlinear iteration, GMRES doesn't give a KSP convergence
> failure, but the convergence seems to stagnate. It's currently at KSP
> iteration 2000, and tolerance of 6e-3 (and still running).
>

Can you use LU? The idea is to determine is the linear system becomes
(almost) rank deficient or if the nonlinear problem is just hard.

   Matt


> David
>
>
>


-- 
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/20160222/0a892ba1/attachment-0001.html>