Pressures during NPT equilibration with real units

Pressure is a quantity that fluctuates significantly, especially true for a small system. The structure had large axial stresses at the beginning, indicating something might not be fully correct with the structure or the choice of force field parameters. Notice that there is a ~25% increase in volume. Whether 50,000 steps is long enough to achieve thermodynamic equilibrium is one question. Even If the structure is fully equilibrated, you should look at the averaged pressure instead of the instantaneous pressure of the last step (since pressure/stresses fluctuate).


Hi Ray,

Thanks for your answer. The equilibration ran for 5 ps (50,000 * 0.1 fs), and the averaged pressures in the three directions, from 508.2 to 5000 fs, were approx. 173, -216 and -4 atm (The initial pressures were considerably higher than the rest.) That’s still high, so I’ll run the equilibration longer, for say, 50 ps.

Since minimization left the sample at 0 K, should I specify Tstart=0 K and Tend=298 K, rather than Tstart=Tend=298 K as I’m doing now (‘fix equilibrate all npt temp 298 298 10 iso 0 0 100 drag 0.2’)? By asking LAMMPS to “jack up” the temperature instead of raising it gradually, am I making it harder to achieve equilibrium?

I followed your earlier suggestion about reflective walls, and am using ‘fix wall/harmonic’ lines to generate six walls, well away from the original box limits (5 Angstom away in the x- and y-directions, 10 Angstrom away in the z-direction). This is a sample line:

fix xlowall all wall/harmonic xlo -5.0 30.0 1.0 2.5 units box pbc yes

I chose the limits to be outside the new box limits in the dump file generated at the end of equilibration.