[lammps-users] Np_zAT simulation of liquid/water interface

Dear LAMMPS users,

I use the 3Mar20 version of the package.

I am performing the Np_zAT simulations for the rough estimation

of melting pressure of the TIP4P/Ice model.

I have arranged the simulation box so that there are approximately
50% of ice and water molecules in the cell that is elongated in “z” direction.
The unit cell of ice is congruent with the desired pressure and temperature.

Everything works fine in the preliminary runs in NVT ensemble which has been run for 5 ns,
however the issue begins when I want to switch to the Np_zAT simulations.
The simulation box shrinks in the z direction (which I am fine with) and after approx. 0.1 ns

I obtain the following error;
ERROR: Out of range atoms - cannot compute PPPM (…/pppm_tip4p.cpp:106)

I use the following commands for the neighbor list build:

neighbor 2.0 bin
neigh_modify every 1 delay 0 check yes one 10000

The latter parameter, i.e. “one 10000” has been set, because I thought it might be

the issue with not enough number of possible neighbors for each water molecule
when the box shrinks, however it does not seem to help.

Enclosed please find all the necessary files for my simulation.
Thanks in advance for your help.

Best wishes,

in_npzAT_long.in (2.84 KB)

temperature (106 Bytes)

interactions (326 Bytes)

due to the excessive use of variables, your input is difficult to read (one always has to jump back and forth to see which settings are really applied and what values are used).
due to the lack of a restart (better would be a corresponding data file) it is not possible to re-run and due to a lack of a log file it is impossible to get an idea of what is happening.

the usual recommendation is to visualize the resulting trajectory. it may be needed to break your simulation into chunks and switch to writing a dump and thermo info more frequently when getting close to the point where the error happens, so that it will be better to see what happens. most commonly this error is a consequence of atoms getting too close and then getting very strongly repelled. this can be due to problems with the force field or too large a time step for the given conditions.

axel.

Dear Axel,

Enclosed please find the configuration file and the log file for the simulation
that gives this error. The system’s conditions should be T=262 K and p=800 atm.
From the calculation of average pressure it shows that the average exceeds

for some reason nearly twice the target pressure.

Thanks a lot for your help.

Best wishes,

conf.tar.xz (715 KB)

Dear Axel,

Enclosed please find the configuration file and the log file for the simulation
that gives this error. The system’s conditions should be T=262 K and p=800 atm.
From the calculation of average pressure it shows that the average exceeds

for some reason nearly twice the target pressure.

the fact that the pressure is higher than expected could be an indication that you may need to reduce your time step to get a stable time integration at the desired conditions.
also, you are using a huge number of MPI ranks for the system size (having about 250 atoms per subdomain is a quite small number).
you may get better efficiency from using fewer MPI ranks or using a hybrid parallelization approach or both (depending on whether you are running on a machine with hyper-threading enabled).