simtk.org

Dear all,

I am able to run a simulation on a mesh with max edge size obtained by the 'estimate' button.
When I chance this max edge size however, the simvascular 'finishes' the job in less than a second but when I try to visualise the results in paraview there is nothing in the converted folder.

I checked the 'more details' button and I received this errors:
ERROR opening file [geombc.dat.1].
ERROR opening file [restart.0.1].
ERROR opening file [geombc.dat.2].
ERROR opening file [restart.0.2].
ERROR opening file [geombc.dat.3].
ERROR opening file [restart.0.3].
ERROR opening file [geombc.dat.4].
ERROR opening file [restart.0.4].
/usr/local/sv/svsolver/2017-08-14/bin/../svsolver: line 58: 2865 Segmentation fault: 11 $SV_HOME/bin/svsolver $*
/usr/local/sv/svsolver/2017-08-14/bin/../svsolver: line 58: 2863 Segmentation fault: 11 $SV_HOME/bin/svsolver $*
/usr/local/sv/svsolver/2017-08-14/bin/../svsolver: line 58: 2866 Segmentation fault: 11 $SV_HOME/bin/svsolver $*
/usr/local/sv/svsolver/2017-08-14/bin/../svsolver: line 58: 2864 Segmentation fault: 11 $SV_HOME/bin/svsolver $*

I am running simvascular on os x 10.11 (I had to downgrade my system after simvascular couldn't open anymore after installation of os x 10.13 high sierra).

Thanks in advance.
Kind regards,
Charlotte

Hi Charlotte,

Thank you for the question! From your error message, it looks to me like you are running the simulation using 4 processors. When running simulations on multiple cores, SimVascular will split up your mesh into different pieces to make the computations easier. This is why the solver is looking for the files: geombc.dat.1, geombc.dat.2, geombc.dat.3, and geombc.dat.4 as these are your mesh split into 4 different pieces.

Can you look into your "Simulations" folder and verify that there is a folder called "4-procs_case"? And inside that folder, check to see if the following files are there:

geombc.dat.1, geombc.dat.2, geombc.dat.3, geombc.dat.4
restart.0.1, restart.0.2, restart.0.3, restart.0.4

If they are not all there, try deleting the 4-procs_case folder and try running the simulation again. Alternatively, you can try running the simulation on a single processor and see if that works.

Hi,

thanks for the reply.
I did indeed not find the parts of the mesh for the different cores, so I tried running them on one core, which appeared to work (slowly).

Is there a way to use more than one core?

Kind regards,
Charlotte

Great to hear that the single core case is working. You can try again with the 4 core case and see if that works. Often, users have issues with MPI not being able to launch parallel processes, but it seems like that was not an issue for your case.

I would create a new Simulation job (separate from the single core job) with all the same settings and try again with 4 cores to see if that will work. Sometimes it just takes a fresh job to make sure all the settings and partitioning are done right.

Hi all,

It seems I'm having a similar problem: sometimes when starting a job with mpiexec I get this error: So it's only the "segmentation fault" bit, not the missing files. Without mpi, on a single core, the simulation runs just fine.

At first I thought it might be a problem with the mesh, but in the meantime I've run a case on the exact same mesh, only with different boundary conditions, and didn't have any trouble. I've already tried re-creating the input files, but that didn't help. By creating a new simulation job, do you mean really creating an entirely new one, from scratch?

Many thanks,
Sita

Hi Sita,

I believe what Justin meant was to create another Simulations entry under SV Data Manager Simulations.

It would be easier to just remove the the appropriate directory in your projects Simulations sub-directory. For example, the Cylinder demo project has four sub-directories under Simulations:

steady/ steady_rcr_deformable/ steady_rcr/ and steady_rcr_variable

one for each entry under SV Data Manager Simulations. You can delete any of these directories to start a fresh simulation, just create the simulation files. Don't delete the .sjb files directly under the Simulations sub-directory, they contain all of the simulation parameters set in the GUI.

Cheers,
Dave

Hi Dave,

Thanks, I'll give that a try.

Sita