[gmx-users] exclusions for a specific model

[Mark Abraham]

On 17/03/2010 9:20 PM, Cristiano De Michele wrote:
> Dear GROMACS users,
> I created a custom topology file to
> define a molecule composed of 5 sites.
> Two sites are real massive sites (but for them
> C6=C12=0 and their charge is 0, hence they do not interact
> with sites of same type on other molecules),
> while remaining three sites are virtual sites.
> Two of these virtual sites carry a charge (but C6=C12=0), while the
> third one has no charge and
> interact with other virtual sites of same type through VdW interactions.
> With this topology real sites do not interact with
> other real sites but they are used only to carry a given mass.
> My question is: is it possible to exclude these real sites from
> non-bonded forces calculations (and neighbors search) to speedup
> simulations? unless having C6=C12=0 and 0 charge for real sites
> they are automatically excluded...

I suspect GROMACS does this check-for-zero already. If you look at the 
FLOP summary at the end of a .log file, it tells you how many flops were 
spent in various inner loops doing some combination of LJ and Coulomb. 
You should see a pattern characteristic of your system.

Mark