[gmx-users] The meaning of rlist, rvdm and rcolumn

Berk Hess gmx3 at hotmail.com
Wed Jul 28 15:12:32 CEST 2004




>From: Anton Feenstra <feenstra at chem.vu.nl>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Subject: Re: [gmx-users] The meaning of rlist, rvdm and rcolumn
>Date: Wed, 28 Jul 2004 10:56:40 +0200
>
>Hi Erik, David and/or Berk,
>
>
>Since Li and I can't seem to agree on what the parameters mean, and
>although I'm pretty sure my explanation is correct, could one of you
>set us straight here?

I guess there is confusion here because of shifted or switched potential.
When using cutoff VdW/electrostatics interactions there are two
options in gromacs (I will use vdw, but the same holds for coulomb):
rlist = rvdw: this option is clear.
rlist > rvdw: twin-range/twin-cutoff method, forces beyond rlist are only 
updated every nstlist steps.

When using shifted or switched interactions you want to have:
rlist > rvdw
such that pairs of atoms do not move beyond rvdw while the neighborlist is 
not updated.
Before 3.2 Gromacs would not allow rlist > rvdw, but that has been fixed.

> > So in v3.2, this has been solved. And now the rlist is actually the
> > neighbor list search cutoff, means rlist=rvdm+shell, rcdm and rcoul are
> > the real cut-off. This is the common method that MD simulation use.

I would not say that that is the common method.
It is used often for VdW interactions, but less often for electrostatic,
it should not be used with for instance PME.
The Gromos forcefield uses a 1.4 nm cut-off for the VdW interactions,
in which case there is no need for switching or shifting, the errors are 
negligible.
Mostly a 0.9/1.4 nm twin-range cut-off is used with this forcefield.

Berk.

_________________________________________________________________
MSN Search, for accurate results! http://search.msn.nl




More information about the gromacs.org_gmx-users mailing list