ERROR: Out of range atoms - cannot compute PPPM (src/OPENMP/pppm_disp_tip4p_omp.cpp:400)

[Mayank7]

LAMMPS version: Aug 2023

Dear LAMMPS users,

I am simulating implicit tip4p water model in NPT ensemble with 500 molecules. The parameters are chosen to compare the simulation results (density distributions close to liquid-gas critical point) to that of GROMACS. Essentially I disabled the long range dispersion corrections and have used the lj/long/tip4p/long pair style. I ran 2 simulations with slightly different state points (T = 705, 710 and P = 163.5, 174.5). Both the simulations crashed, one after 2.7M steps and the other after 26M steps with the error:

...
26074000 737.347254449303 -3305.8866882018 2195.69421429599 -1110.19247390582 872.085388980185 0.490735049435132
26074500 698.458505612482 -3218.14468234431 2079.89016090463 -1138.25452143968 46.6566151953563 0.491028781821433
ERROR: Out of range atoms - cannot compute PPPM (src/OPENMP/pppm_disp_tip4p_omp.cpp:400)
...

After going through the mailing list, I thought that probably the timestep of 1.0 fs is large for my temperature ranges so I restarted the runs with 0.5fs but that again led to crash for both the cases (after 6.9M and 55M steps respectively). I am again running a simulation with tighter thermostat and barostat coupling (50 and 500 instead of 100 and 1000 with 1fs timestep) but I am not sure if something else could be causing the issue. Please let me know if there is something else I can try or the potential errors in my scripts. The input and restart scripts are attached (comments removed) along with their log files.

Thanks in advance,
Mayank

in.TIP4P_Ice_NPT (2.3 KB)
LOG.txt (643.6 KB)

in.restart (1.6 KB)
LOG_restart.txt (1.1 MB)

[Germain]

Hi @Mayank7,

@Mayank7

After going through the mailing list, I thought that probably the timestep of 1.0 fs is large for my temperature ranges

What makes you conclude so? If that was the case, why is your simulation only crashing after several thousands iteration? This hardly makes sense to me.

@Mayank7

I am again running a simulation with tighter thermostat and barostat coupling (50 and 500 instead of 100 and 1000 with 1fs timestep)

Same question. Why would there be any issue with barostat and thermostat only after such a long time?

@Mayank7

I am not sure if something else could be causing the issue.

You modify a lot of kspace command parameters and apparently have large density fluctuations. This is sufficient to cause issues because the grid defined at the beginning of the run is accurate only to a given volume range.

@Mayank7

Please let me know if there is something else I can try or the potential errors in my scripts. […] along with their log files.

Is there any reason to format your log file so poorly with the fix print command when thermo_style and thermo would make it way, way, way easier to read?