[gmx-users] neighbor list

[m b]

Anton Feenstra wrote:

> There is a 'grace distance' beyond the cut-off to prevent atoms that are
> within the cutoff to be 'forgotten' due to charge group being outside cutoff.

This is confusing! does that mean that atom pairs belonging
to different charge groups that are separated by more than the cut-off
can possibly be included when gromacs calculates the energy 
(as opposed to ALL atom pairs between the two groups) ?
this would invalidate the whole point of charge groups the
purpose of which is to make the coulomb interaction short(er) ranged ...

which cut-off are you referring to here, rlist or rvdw (rcoulomb)?

I believe there is such a thing as the "grace distance" you mention 
here, but that does not apply to single atoms but to charge groups.
Its purpose is to prevent (to some extent) the miscounting of 
interactions between two groups whose distance has "crossed" the 
cut-off between two neighbour searches (as far as i understand it)

I find the way the neighbour-list is implemented (or at least 
documentd) in Gromacs generally rather confusing and there are a 
few contradicting statements about it to be found in this list.
It might be helpful if the developers included a clear diagram
into the documentation at some point ...

> (don't remember how large that is, though).

anybody else in the know ?

cheers,
Michael


__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus