sudden decrease in density

Dear Lammps Users,

I have a system where I randomly position the atoms inside the box, and I ran nve for the first 50000 steps (timestep = 0.1fs using ReaxFF). The simulation did not show any error (no lost atoms or no sudden increase in volume). When I tried to run npt with temp 5000 5000 100 press 0 0 1000 for 1ns, the system showed a sudden decrease in density after 70000 steps. I have attached the log file. Any suggestions are welcome. ( I am using lammps 12 Dec 2018 version).

logfile.out (10.1 KB)

​Your pressure fluctuations are on the order of a GPa. 1000 atoms is quite small. Try replicating 2 2 2 or 3 3 3 or more. Given the size of the density fluctuations at the beginning, maybe you should increase your pressure damping coefficient. You should calculate the average pressure at the end of your constant volume run, and barostat slowly from that pressure to your target pressure. Periodically invoking velocity all create can remove ballistic motion (once your atoms escape the cutoff distances, ​the system will just keep expanding). Also, specifying 1.0 atm as the pressure instead of 0.0 can be helpful. All of these things could stabilize this run.

Good luck,


Thanks, Keith. This is helpful.