Can one use shake/rattle with pimd/langevin? I am trying to do that but the job dies quietly without any hint about possible issues. Thanks for any help.
When reporting issues, you should always also report which LAMMPS version you are using and on which platform.
I doubt that that is the case. It is probably that there is some error message, but it is buffered by MPI and that buffer is not flushed when it aborts.
Please try running again but add the -nb or -nonbuf command-line flag. That should disable the MPI buffering and you may see any additional output.
Hello everyone,
From a quick look at the sources, I did not see any obvious exclusion between those two fixes.
Did you manage to figure it out? If not, could you send us a simple script to reproduce the issue?
All the best,
Julien.
Hello,
please find below the script you requested. If you comment the fix shake, everything work like a charm (including langevin/npt).
I have extensively tried to modify time scales parameters, nothing worked, atoms are always lost at some point.
Thanks for helping.
Stefano
–
LAMMPS (19 Nov 2024)
mpiexec.openmpi -n 4 lmp_mpi -nb -sc none -in ./input.pimd -p 4x1
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 20.04.6 LTS
Release: 20.04
Codename: focal
pimd.tar.gz (274.6 KB)
I am not surprised that fix shake will mess up the dynamics (just look at the evolution of the temperature) since applying the constraint will change positions of the beads and thus completely mess up the dynamics for the path-integral part.
Have you tried using very stiff/large force constants for the bonds/angles instead? That should get you close to what real constraints do.
Yes I did consider stiff force constants, this is what you get if you just comment the fix shake. It seems to work, in the sense that the trajectory is stable to very long times. I am now trying to figure out if what I get is correct, by looking at what happens by changing the beads number and, probably, trying to recover the classical limit for \h_bar → 0.
Thanks for your previous hint on the buffered error messages.
The force constants that you are using are not particularly stiff.
@Yifan_Li do you see a (simple?) way how we can make the two PIMD fixes in LAMMPS compatible with SHAKE/RATTLE?
I am contemplating adding a test that will stop any calculation that has fix shake/rattle active while a PIMD fix is also in use.