[gmx-users] Influence from images (PBC)

[Mark Abraham]

Yinghong wrote:
> Dear users:
>  
> By using periodic boundary condition, images are existing. If I am 
> not totally wrong, images should have some influence on our simulation, 
> right?

Yes, of course. If they had no effect, there would be no purpose to
having them.

> Now, in my simulation, a solute is solvated in the water with a small 
> box. I doubt that images have certain unexpected influence due to the 
> small size of simulated box. My question is how to eliminate the 
> influence from images?

In the literal sense, you can't. In general and non-exhaustively, in MM
you can either have explicit solvent with a finite size effect and
boundary effects (a droplet), or explicit solvent with a finite size
effect and periodicity artefacts (PBC), or implicit solvent and neither
of the above. One reasonable approximation to the ideal case of infinite
extent with explicit solvent is the second option with a very large box.
Ordinarily, this is too costly for people to want to use it. From these
questions, it sounds like you should read some more of the available
introductory material on molecular mechanics/dynamics/whatever.

> As known, we can set mdp file with *pbc=no* to represent the case 
> without periodic boundary. Somebody can tell me what is the detailed 
> meaning of pbc=no, and what should be the values for cutoff  rvdws, 
> rcoulomb and rlist?

Please try reading the manual for this - there's a reason someone wrote
300+ pages for it. Values for these parameters should be system,
application and resource dependent anyway, so advice on numbers is a bit
pointless.

> Besides, what is the difference between pbc=xyz and pbc=full except that 
> the later denote the case of infinite molecule?

I'd look for the answer to this in the manual. Normally I'd suggest you
try there before posting here, but it isn't there. Searching the
archives of this list is also a good idea - luckily for you, there's a
explanation given already -
http://www.gromacs.org/pipermail/gmx-users/2005-September/017080.html

Mark