'problem in move'

[Emilio Romano-Diaz]

Hi there,

I'm having an issue with RAMSES that I've not found before and I'm not quite sure how to fix it.
I'm running a zoom-in DM+Baryon simulation with ICs created with MUSIC (levels 7-9). The DMO simulation ran without any problem. The same simulation with the inclusion of baryons is giving issues after reading the ICS and creating the AMR. The error comes from the move_fine.f90 subroutine, after having balanced the AMR grid, dumped the 1st output data (the ics) and in the first fine step (I've included the log-file of the run). It seems the error has to do with some cells outside of the refined region, but I'm not sure how this can happen since the ICs have been aligned and forced to be an exact cube. Although the error still persists if instead a mask is read or the region is not a perfect cube.

I've changed the "nexpand" parameter from small to large values (10) or just setting all of them to the same value (1) but to no avail, the code always crashes more or less in the same point. I've also included the RAMES parameter file. 

As extra info, RAMSES was compiled with standard MPI+gfortran flags in our local cluster. 

I wonder if any of you have encountered a similar situation, and if so, how did you solve it?
Any help will be greatly appreciated.
Best regards
Emilio

[Romain Teyssier]

Hi Emilio,

The only problem I see is that you should compile with NVAR=7 since you are using metals and delayed cooling.

Romain

--
You received this message because you are subscribed to the Google Groups "ramses_users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ramses_users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ramses_users/132cf707-5e84-4700-a3a1-a6e61fa0c010%40googlegroups.com.
<run.log><ramses.nml>

[Emilio Romano-Diaz]

Hi Romain,

Thanks for spotting this point. The preliminary test seems to work.

Best

e

On Thursday, July 25, 2019 at 11:51:03 PM UTC+2, Romain Teyssier wrote:

Hi Emilio,

The only problem I see is that you should compile with NVAR=7 since you are using metals and delayed cooling.

Romain

Le 25 juil. 2019 à 20:49, Emilio Romano-Diaz <emil...@astro.uni-bonn.de> a écrit :

Hi there,

I'm having an issue with RAMSES that I've not found before and I'm not quite sure how to fix it.
I'm running a zoom-in DM+Baryon simulation with ICs created with MUSIC (levels 7-9). The DMO simulation ran without any problem. The same simulation with the inclusion of baryons is giving issues after reading the ICS and creating the AMR. The error comes from the move_fine.f90 subroutine, after having balanced the AMR grid, dumped the 1st output data (the ics) and in the first fine step (I've included the log-file of the run). It seems the error has to do with some cells outside of the refined region, but I'm not sure how this can happen since the ICs have been aligned and forced to be an exact cube. Although the error still persists if instead a mask is read or the region is not a perfect cube.

I've changed the "nexpand" parameter from small to large values (10) or just setting all of them to the same value (1) but to no avail, the code always crashes more or less in the same point. I've also included the RAMES parameter file. 

As extra info, RAMSES was compiled with standard MPI+gfortran flags in our local cluster. 

I wonder if any of you have encountered a similar situation, and if so, how did you solve it?
Any help will be greatly appreciated.
Best regards
Emilio

--
You received this message because you are subscribed to the Google Groups "ramses_users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ramses...@googlegroups.com.