LAMMPS Getting Stuck Prior to Thermo Output

All,

I’ve been having this issue where a simulation starts on our HPCC, but it hangs before the thermo output section of the log files. It stays like that until the job is deleted. In the past I have had some success solving this by explicitly defining the numbers of processors in the LAMMPS script e.g. “processors 2 4 4”, but it’s not working now. I’m not sure what else I can try to get past this.

The output is attached. Thank you very much for your help in advance.

Happy Holidays,

Will Pisani

CE1212RxEqS0.in.sh.o419233 (1.48 KB)

All,

I've been having this issue where a simulation starts on our HPCC, but it
hangs before the thermo output section of the log files. It stays like that
until the job is deleted. In the past I have had some success solving this
by explicitly defining the numbers of processors in the LAMMPS script e.g.
"processors 2 4 4", but it's not working now. I'm not sure what else I can
try to get past this.

The output is attached. Thank you very much for your help in advance.

​there is not enough information provided to make any specific statement.
most likely your simulation is stuck in some iterative procedure somewhere,
possibly the charge equilibration.
you have some troubling warnings in your output. are you sure they can be
safely ignored?

axel.

Are there any ways to avoid getting stuck in charge equilibration?

Thanks,

Will

If those warnings help you diagnose why your simulation gets stuck somewhere, it’s probably smart to not ignore them.

Are there any ways to avoid getting stuck in charge equilibration?

​as i already mentioned before, you are not providing any helpful
information about your system and simulation, so it is impossible to
diagnose what is going on and provide specific and meaningful advice.
when the charge equilibration cannot converge or worse, that usually is the
consequence of bad choice of simulation parameters or a bad geometry.

since santa declined my request for a crystal ball again, there is not much
else that i can advise.

axel.​

Axel,

I think I’ve solved this problem. I’ve been having this same issue (where it seems like the simulation gets stuck in a charge equilibration abyss) with three different polymer systems in PCFF. The common theme between these four cases was the use of the gzipped dump file. Once I removed the .gz extension from all dump file commands and reran the simulations, this issue disappeared. I’m not sure why a gzipped dump file was causing this issue. Perhaps it’s a bug?

Thanks,
Will

unlikely. this feature is used a *lot*. however, there are known
issues on some parallel machines with doing the compression through a
pipe, which requires the fork() library call. there are dump style
variants in the COMPRESS package, that do not need a pipe, but use the
zlib library instead. from the dump documentation:

Axel,

I changed the dump file command to custom/gz and everything is working as it should now. Thank you very much for your help. I really appreciate it.

Thanks,
Will