Has the ability to run GROMACS in a ‘vacuum’ (non-periodic) state been recovered yet? I know that in the 2020 version update the ‘group’ cut-off scheme was removed, and this was the only method for running pbc=‘no’. It’s been about 3 years now, so I wanted to check in to see if the ability to run vacuum simulations had been restored. I see from the MDP options that ‘group’ cut-off is still deprecated, but is there perhaps some other combination of MDP options that gets this simulation type to run?
There is still no way to do a non-PBC simulation in GROMACS. There is an open GitLab issue about this as it is really important for force field comparisons but it has not been a high priority (or is for some reason particularly challenging to implement).
Got it. So, what would be the most ‘canonical’ way to run a ‘vacuum’ simulation in current versions? I have been using an older version of gromacs for these simulations in my workflow (i.e. gromacs 2019), but its going to become unsustainable to keep doing that as gromacs changes, so I’m exploring alternatives.
I saw this comment in the mdp options for ‘pbc’=‘no’:
To simulate without cut-offs, set all cut-offs and nstlist to 0. For best performance without cut-offs on a single MPI rank, set nstlist to zero and ns-type=simple .
So, if I set both the cutoff and nstlist to 0, will the system compute all pairwise interactions between all particles in the system directly in real space?
Naturally one could mimic a vacuum box with pbc=‘xyz’ by running in a very large box (e.g. 10 nm). The challenge with this is that the simulations become quite slow, even though its a trivial calculation. The advantage of running with pbc=‘no’ is that its an essentially instantaneous calculation on a single core. Is there a way to accomplish this more or less using pbc=‘xyz’ along with the right combination of other mdp options? For example, if I set ‘vdwtype’=‘Cut-off’ and ‘coulombtype’=‘Cut-off’, will that result in only realspace interactions with no long-range corrections? Will that make the simulation fast even if there is a box?
