[FLASH-USERS] Some issue running the Blast Wave with Thermal Conduction test

[Klaus Weide]

On Fri, 3 Oct 2014, Ernesto Zurbriggen wrote:

> I started running the supplied "Blast Wave with Thermal Conduction" test
> problem in the flash4.2.2 release. The code seems to run well but it prints
> out a sort of warning between each time step (see below for the first 10 time
> steps).
> This seems to be an issue with the hypre-related subroutine "gr_hypreSolve".
>

Ernesto,

I confirm this problem, I have reproduced it, for the setup line that you 
gave (same as in Users Guide), with our latest code.

> Does anybody know if things are going well in spite of this warning?

I don't think so. There seem to be NaN's and Inf's in the checkpoint 
files...

> I followed exactly the setup given by the manual's page 452.
> 
> This is: setup -auto ReinickeMeyer -1d +pm4dev +spherical
> -parfile=flash_SPH1D.par

I will try to investigate next week what goes wrong with the
spherical setup. In the meantime, here is a variant of the
ReinickeMeyer setup that DOES work (we are testing like this
daily):


  ./setup ReinickeMeyer -2d -auto +hdf5typeio +uhd -parfile=coldstart_pm_2p_2d.par


> I also ran this test problem with flash4.0 release, same setup, and it did not
> appear this issue.

It could be the case that FLASH 4.2.2 just does more checking than FLASH 
4.0.


Klaus