[gmx-users] Calculation of LJ potential

[Erik Lindahl]

sadhna wrote:

> Dear Sir,
>           I would like to know how LJ potential is calculated in Gromacs.
> Does the LJ  function exclude the first, second and third bonded
> neighbours of a  particular atom or does it calculate the potential
> between all atoms without bothering about the bond connectivity.

Hi Sadhna,

This depends on the forcefield, so you can actually select how many bonded
neighbors to exclude for each molecule type in the topology. Just to give
you an example, this is the first lines in benzamide.itp from the
distribution:

[ moleculetype ]
; Name  nrexcl
Benz 3

In practice, both the Gromacs and Gromos96 forcefields are designed for 3
neighbor exclusions, so most users never need to worry about it.

For the third neighbor you can use 1-4 interactions for both LJ and coulomb,
either by explicitly specifying the 1-4 parameters in the forcefield (like
Gromos96) or by scaling down the normal LJ and/or electrostatic
interactions. This "fudge factor" is specified in the forcefield; check the
topology section of the manual for details.

Cheers,

Erik