[gmx-users] New force fields

[Ram'on Garc'ia Fern'andez]

On Fri, May 03, 2002 at 02:34:51PM +0200, David van der Spoel wrote:
> On Fri, 3 May 2002, Ram'on Garc'ia Fern'andez wrote:
> 
> >This is necessary for generating startup configurations of polymer melts.
> >In that case, I use a soft potential. This soft potential is different for
> >different pairs of atom types.
> Do you also use different potentials for doing production runs? In that 
> case, what do they look like?

No. Only for generating starting configurations. However, the "starting" might
take lot of time.

> For generating starting conformations I would think LJ with softcore would 
> be quite pain-free. 

But at present, it is almost imposible to do LJ with softcore.

The only method reasonably documented is to use a table with a custom potential.
But this method is quite painful. You can only customize the interaction for each
pair of atom types by three coefficients, and the energy is a linear combination
of fixed functions by that coefficients. In my case, I had two types of atoms, both
charged. And this method of customizing the potential ensued too restrictive. I had
to do much trial and error until I found three functions that I could combine and
have the desired result.

Gromacs should document how to add a custom potential by formula to
the source code, and allow with a reasonable amount of paramenters (such as
four). It was straightforward with DLPOLY, I just modified forgen.f.

Ramon