[gmx-users] Meaningless pressure values: NVT and NVE ensembles

[Francesco Mambretti]

Dear all, I am using Gromacs 5.0.1 to simulate a simple atomic system of 108 solid LJ argon particles in the void, pbc applied.
Using g_energy, I find - both after NVT runs and NVE runs - ok values for energies and temperatures, but absurd values for pressures (something like 10^3 bar or -10^3 bar). I read lots about it on the web, included an old post on http://article.gmane.org/gmane.science.biology.gromacs.user/6696/match=negative+pressure; what I am not still able to understand is if output values for pressure computed by Gromacs are meaningful.
I am using the right input parameters (density, temperature and so on) - with other simple MD programs they work well.
Moreover, if I understand well, a negative pressure means that the system wants to collapse, but this doesn't make sense for my system, if it is already at the right density!
So, should I trust pressure output values from g_energy? Does Gromacs compute pressure in a way not compatible with NVE and NVT?
If not, how can I check the pressure of my microcanonical or canonical system?

I am applying long-range dispersion corrections, I generate velocities from the temperature; timestep = 1 fs, no pressure coupling, potential has a cutoff and it is shifted to 0 in correspondence of cutoff radius.

Thanks,


Francesco Mambretti