[gmx-users] another attempt to clarify rlist, rvdw, ...

[David]

On Sun, 2003-12-07 at 19:20, m b wrote:
> dear gmx-users,
> 

> 
> If I understand that correctly this means that
> if nstlist>1 the difference between 
> max(rvdw,rcoulomb) and rlist should alsways be 
> greater than zero (the manual suggests 0.25 nm)
> otherwise one runs the risk of gmx occasionally
> not considering interactions between some atoms 
> that are separated by a distance of less than rlist.
> The approprate value for this distance will depend 
> on the parameter nstlist (and on the system 
> geometry, temperature, etc).
> I wonder if it wouldn't be a good idea for gmx
> to check (at best continiously during a simulation)
> whether the relation between the parameter nstlist 
> and the difference between max(rvdw,rcoulomb) 
> and rlist is reasonable? AFAIK this is not
> implemented in gmx, in fact it doesn't even
> give a warning if I set nstlist to a high
> value and at the same time have
> max(rvdw,rcoulomb)=rlist.

Much depends on how one treats coulomb interactions.
If one use cut-off, then a *charge-group* based list is made at step 0
and used for nstlist steps. Making rvdw,rcoulomb > rlist will trigger
the same list as having rvdw,rcoulomb = rlist, but additionaly the
forces due to charge groups being between rlist and rcoulomb/rvdw are
computed during neighborsearching (NS), stored and added to the forces
at every step, until new NS is done. With shifts and reaction field it
is the same. With PME the Coulomb is almost correct, the only error
remaining being that atoms drifting together between NS are not taken
into account until NS is done again.



-- 
David.
________________________________________________________________________
David van der Spoel, PhD, Assist. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,  	75124 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++