Dear lammps users,
I want to simulate the expansion of gasphase LJ particle in a nozzle from one gcmc reservoir into vacuum (another gcmc) reservoir.
At timestep 2105874 I get a segmentation fault:
[snoopy:01402] *** Process received signal ***
[snoopy:01402] Signal: Segmentation fault (11)
[snoopy:01402] Signal code: Address not mapped (1)
[snoopy:01402] Failing at address: 0x5dc002c
[snoopy:01402] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10340) [0x7fb88c061340]
[snoopy:01402] [ 1] lammps-daily() [0x7e649a]
I tried it with lammps-daily Version:2~git+8029+18~ubuntu14.04.1 and
a build of lammps from git (2015-03-13), same segmentation fault error for both.
First simulation I executed with mpirun -np 8 on 8 cores and afterwards on a single core. Crash also happens on a single core. I tried to make a restart file as close as possible on the crash. The restart files really close are getting corrupted (frestart.2105800). Restarting them leads to:
ERROR: Did not assign all restart atoms correctly (…/read_restart.cpp:500)
I’m using the boundary setting:
boundary p p p
Disabling gcmc on a restart at time step 2105700 does not solve the problem ether, (thought this problem is caused by gcmc) still there is the Segmentation fault also with the fix gcmc disabled.
Disabling my dump and dump_modify in my input file will lead to an earlier segmentation fault crash at timestep 2105740.
I will attach a link to the files necessary to reproduce this error and hope somebody has time to look at it. The dump is disabled in the attached input file.
https://www.dropbox.com/sh/2p9r5aivtnbkc7t/AADnVhKCxKNDteaN5JuKmb3Ba?dl=0
Thank you for your time!
Regards,
Helmut Ortmayer