Hello Zhen, this type of question should be asked to the mailing list: there is not enough information to be sure that the issue is generated in fix colvars or in another part of LAMMPS.
I tried running your input with LAMMPS version 11Aug2017 (both in serial and in parallel) and could not reproduce the seg
fault.
Can you reproduce it, and if so, can you provide details on how LAMMPS was built, how you’re running it, and the output before the seg fault?
Giacomo
Dear Giacomo,
Thank you very much for the reply!
I used the following setup to install the lammps on our cray machine. If you turn off the fix colvar line in the in.mixture, the code can successfully run. However, if I used this command
The out.colvars.traj.BAK and out.colvars.traj files are empty, indicating there is no output corresponding to the umbrella sampling calculations.
I used one node (32 cores) to run it. Without colvar command, the code can perform perfectly, while the colvar kill the process.
Besides, the #SBATCH --err=std.err commands output the error message in the std.err
I checked the std.err file, it gives the following error message
srun: error: nid00011: tasks 0,16: Segmentation fault
srun: Terminating job step 6553381.0
slurmstepd: error: *** STEP 6553381.0 ON nid00011 CANCELLED AT 2018-09-30T15:16:26 ***
Please let me know if you need other information to identify the problem.
Best Regards,
Cao, Zhen
#!/bin/bash
Hello Zhen, I repeated the test, this time compiled with the Intel 2017 compiler suite as in your setup. Again, I could not reproduce the seg fault.
Can you try confirming whether the seg fault is reproducible on your cluster, and if so, we can try looking into the problem in more detail.
Again, please attach the Slurm output (both output and error channels).
Giacomo