> On Sep 16, 2015, at 8:28 AM, Alexander Lindsay <[email protected]> wrote:
> 
> Hi all,
> 
> I've built a MOOSE application and the problem I'm studying solves great 
> using serial LU. However, when I try to solve in parallel using super LU, I 
> encounter many DIVERGED_LINE_SEARCH errors using the default bt line search. 
> If I switch to line_search=none, then instead of DIVERGED_LINE_SEARCH errors 
> I get DIVERGED_FNORM_NAN errors. I do not have a lot of debugging experience, 
> but I am slowly gaining more. Is there a good way to get more information 
> about what's causing the NAN's? I.e. a specific residual or jacobian 
> statement?
> 

   It would be useful to have full output with -snes_monitor 
-snes_converged_reason -snes_linesearch_monitor -ksp_monitor_true_residual 
-ksp_converged_reason

   The original DIVERGED_LINE_SEARCH is almost always an indication that the 
step direction is not a Newton step. This needs to be debugged see 
http://www.mcs.anl.gov/petsc/documentation/faq.html#newton   Switching to no 
line search is not a useful thing to do here.

  Barry


> This may not even be an appropriate list to post this problem in (could be 
> more on the MOOSE side); apologies if that's the case.
> 
> Alex

Reply via email to