[gmx-users] how can I prevent gmx from adding the LR-forces ?
mic0404 at yahoo.com
Sun Dec 14 16:51:02 CET 2003
I run a sim using tables with a generalized reaction
field potential for the Coulomb interations (and the
built-in gromacs routines for VdW, this works)
the Coulomb potential and its second derivative given in the
table are zero beyond a cut-off of 1.2 nm. I use
coulomb_type=cut-off, rcoulomb=1.4 and rlist=1.2.
Thus hoping to prevent any artefacts due to interactions
between atom-pairs crossing the rlist distance
between two neighbour-searches.
The surprising part of the output: the Coulomb-LR
term is non-zero! I thought this term contained
only the interactions calculated for pair-distances
bewteen rlist and rcoulomb (This was also confirmed
in a recent mail by David, wasn't it ?).
With the set-up described here this Coulomb-LR term
should be ZERO, by definition.
I plotted the ctab.xvg file as obtained with the
-debug flag and the potential and forces are
indeed ZERO between 1.2 and 1.4 nm.
Apparently gmx uses the full 1/x interaction when
calculating the LR part of the potential and forces.
This would be no problem was it not for the fact
that during NS also the forces are calculated if
nstlist>1. In the time between two neighbour
searches these forces are added to the SR forces.
I don't want to have these additional forces
since they modify the (physically well justified)
GRF potential resulting in a potential that
goes to zero at rlist and then between
rlist and rcoulomb becomes finite again ...
looking roughly like ...
.. rlist rcoulomb
... | |
........ . .
0 ___________________ ..... ...... _______
With the two steps in the potential function the resulting forces
will, of course, look awkward ...
Can I prevent this from happening (without setting nstlist=1) ??
If not, where in the code are these LR-forces added ?
It'd probably suffice to comment out a couple of lines
in the source code to get the desired effect.
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
More information about the gromacs.org_gmx-users