[FLASH-USERS] Orbit Test Problem crash during runtime

[Anshu Dubey]

Ryan,
This is not a workaround, it is the right thing to do. The way particles
migrate between blocks and processors is simpler with time advance than
with regridding. With time advance they only move to the nearest neighbor,
whereas during regridding blocks themselves change identity so the
association between particles and blocks has to be redone, and so the
migration uses a different method.
Anshu

On Wed, Nov 8, 2017 at 2:08 PM, Ryan Farber <rjfarber at umich.edu> wrote:

> Hi FLASH-users,
>
> I am trying to perform the "Orbit" test problem that comes with FLASH4.2.2
> to test Poisson solvers in simulations including particles.
>
> I used the following setup command:
> ./setup Orbit -auto -3d
>
> It compiled fine but crashed during runtime. The message dumped to
> stdout/stderr was:
>
> DRIVER_ABORT: src == gr_meshMe, and we still have unmatched neData, bad
>
> while the orbit.log file had:
>
>  [ 11-07-2017  15:38:20.889 ] step: n=440 t=1.756000E+03 dt=2.000000E+00
>  [ 11-07-2017  15:38:20.894 ] [DRIVER_ABORT]: Driver_abort() called by
> PE           0
>  [ 11-07-2017  15:38:20.895 ] abort_message: src == gr_meshMe, and we
> still have unmatched neData, bad
>
> I managed to fix the issue by setting the logical parameter "regrid" to
> .false. in Particles_advance.F90 (as reminded by looking at a previous
> FLASH-USERS email), but I am concerned if this is an acceptable workaround.
>
> Shouldn't particles that move off the block owned by its source processor
> need to have regrid=.true.? Would anyone be able to explain in more detail
> the purpose of having regrid=.true. or regrid=.false.
>
> Thanks in advance for your help.
>
> Best,
> --------
> Ryan Farber
> Graduate Student
> University of Michigan, Ann Arbor
>
> P.S. Thanks Klaus for responding to my last question! I greatly appreciate
> the help!
>



-- 
********************************************************
Anshu Dubey
630 252 3313 (work)
312 420 0033 (mobile)
Argonne National Laboratory
********************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20171110/d31f841f/attachment-0001.htm>