[lammps-users] lennard-jones fluid simulation

Hi Lammps users,

I’m running into some questionable results from simulations of a lennard-jones fluid. In particular, I’m testing a N=256 particle system at a constant reduced density of 0.8 and a constant reduced temperature of T=3.2. (On the LJ phase diagram this is above the critical point. ) It’s a single-component LJ fluid with a cutoff of 2.5 sigma. I use a langevin thermostat via fix langevin.

I have run multiple simulations of the above system. The problem is that for some simulations the total potential energy is predominantly positive. Naively I would expect it to be negative, since for a density=0.8 the average particle spacing is larger than sigma, and thus the average pair interaction energy would be negative.

Even stranger still, some of the simulations have an overal NEGATIVE potential energy. Is there some ultra-long relaxation time I am forgetting about? Am I really in the supercritical region?

Any insight would be very helpful.