GROMACS version: 2025.2
GROMACS modification: No
I am running some tests to make sure that my understanding of LJ-PME is correct before using it for a somewhat special system. Long story short: I need the long-range LJ force over a considerable distance, ideally infinite, and I think LJ-PME will do this for me.
I created a periodic slab of 20x20 argon particles in the xy-plane, with a distance 1xsigma between the particles in a square pattern, and one lone argon particle sitting 10 x sigma over the slab in the z-direction. The box size was 20xsigma in x and y, and 30xsigma in z. As such, I expect a force on the lone particle in the z-direction (negative), as it is closer to the slab than its periodic image, and no lateral forces, because it is perfectly lined up with one of the slab particles at x=y=10xsigma.
I made a single-timestep simulation and got the forces from the trr-file with gmx dump. I see the z-force, but am surprised over the forces in the x and y directions. Sure they are smaller than the z-force, but only by one order of magnitude. Is this just a numerical artefact that I can probably neglect, or am I missing something fundamental?
gmx dump -f slab.trr says
f[ 400]={ 9.37551e-04, 9.37775e-04, -1.07467e-02}
The input gro file I made looks as follows:
Made with make_slab.py
401
1AR AR 1 0.000 0.000 0.000 0.000 0.000 0.000
2AR AR 2 0.000 0.382 0.000 0.000 0.000 0.000
3AR AR 3 0.000 0.763 0.000 0.000 0.000 0.000
4AR AR 4 0.000 1.145 0.000 0.000 0.000 0.000
5AR AR 5 0.000 1.527 0.000 0.000 0.000 0.000
…
398AR AR 398 7.253 6.490 0.000 0.000 0.000 0.000
399AR AR 399 7.253 6.871 0.000 0.000 0.000 0.000
400AR AR 400 7.253 7.253 0.000 0.000 0.000 0.000
401AR AR 401 3.817 3.817 1.909 0.000 0.000 0.000
7.634987 7.634987 7.634987
For relevant mdp parameters I used:
integrator = md
cutoff-scheme = verlet
pbc = xyz
coulombtype = Cut-off ; because there are no partial charges.
vdw-type = pme
vdw-modifier = none
DispCorr = EnerPres