Rlist, nstlist tuning and verlet-buffer-tolerance; rvdw and rcoulomb

User discussions
1

GROMACS version: Any version
GROMACS modification: No

Hi everyone;

very recently I am going through basic stuff in gromacs which I see now I did not understand completely.

  1. Can I ask how verlet-buffer-tolerance tunes rlist and nstlist whenever the first is not -1 ? I tried to go through some reference and of course the manual, but did not get it in detail, which is exactly what I am looking for. Moreover, if using PME, does rlist gets fixed to rcoulomb, meaning that the verlet-buffer-tolerance parameter does not tune it anymore (again, without considering the case of -1 value)? ;

  2. Honestly, I admit I don’t get what the rlist is used for, if the pair interactions are explicitly computed as pairs according to rcoulomb and rvdw. If it is not useful and used as some kind of bookkeeping purposes, then consequently I don’t get what veret-buffer-tolerance is used for, too;

  3. Let’s consider a FF parametrized with rvdw = 0.8 nm and rcoulomb = 0.8 nm: is it safe to enlarge them, claiming a better accuracy in doing so? I thought about this looking at this paper ( DOI : 10.1063/1.4893965 ), even if it covers solely the VdWs. Being on the side of the purists, given a FF publication, I would instead respect the publication itself. But here’s the thing: the authors suggest that an higher rvdw is suggested for anisotropic systems (solute embedded in protein matrix, and also protein matrix surrounded by lipids, of course), hinting the reader that he should push it to larger values and thus ending up changing it with respect to the publication.

  4. DispCorr = Ener or EnerPres ? I know it is related to vdws, and it is suggested to go from “no” to “Ener” and to “EnerPres” (in the order) the smaller the rvdw is. Ok, but what if the authors don’t state this in the paper? It is rather usual to find it omitted in publications. I would like to have a rule of thumb, like knowing how to decide between (i) “Ener” and “EnerPres” in case of rvdw = 0.8 nm, as well as (ii) between “Ener” and “no” if rvdw is 1.2 nm;

  5. Limiting to VdWs, I did not get why I see cases in which:
    vdwtype = cutoff
    vdw-modifier = potential-switch
    rvdw-switch = 1.0
    rvdw = whatever, put larger than 1.0
    I mean: what’s the point a potential-switch added at 1.0 up to rvdw? Would’t it still introduce a discontinuity in the potential, but this time at 1.0 (even if recovering that at the cutoff, of course)? I only understand the sense of a potential switch if rvdw-switch = 0, i.e. it’s default value.

Kindly thank you for any help, I really appreciate as I think all these details are subtle, even if maybe it’s just a limit of mine.

Bests,
Jacopo

2

Sorry for the 5th point, that was me reading “shift” instead of “switch”. Of course, that very point is thus readily clarified