FudgeLJ in ffamber - Was: [gmx-users] Re: fudgeQQ (again ...)

[Jones de Andrade]

Hi Erik.

Well, first of all thanks a lot for the reply. I hope I do not take
too much time from you or space on the list with this discussion.

I would like to exemplify it, and see if I'm doing it right, or where
there is a mistake. In the worst scenario, there will be an example
for anyone with the same doubt recorded forever in the list.

What I'm doing is:

1 - In a top file, have the lines like this:
[ defaults ]
; nbfunc      com-rule    gen-pairs      fudgeLJ  fudgeQ
    1            2          yes           0.5000  0.8333
#include "ff_moleculeA.itp"
#include "moleculeA.itp"
[ system ]
Pure A
[ molecules ]
;molecule name   nr.
moleculeA           256

2 - In the ff_moleculeA.itp file, it has the [atomtypes], [bondtypes],
[angletypes] and [dihedraltypes] directives, in the standard way and
with all potential parameters.

3 - In the moleculeA.itp file, I included:
[ moleculetype ]
; Name                     nrexcl
moleculeA                        3
[atoms] directive, with atomic masses and charges included;
[bonds], [angles] and [dihedrals] directives, *without* the potential
parameters, just the atom indexes;
And the [pairs] directive, with the "funct" included but *no* sigma or
epsilon parameters. The following as an example:
[ pairs ]
;  ai    al funct     sigma (nm)  epsilon (kJ/mol)
    9     5     1
   10     5     1
   11     5     1
...

Now, the question is: is it right, in a way that will use both the
different valued fudgeLJ of 0.5, and the fudgeQQ of 0.8333 for 1-4
interactions, or should I make any kind of correction still?

Thanks a lot in advance for everything,

Sincerally yours,

Jones

On 2/13/07, Erik Lindahl <lindahl at cbr.su.se> wrote:
> Hi,
>
> On Feb 13, 2007, at 6:52 AM, Jones de Andrade wrote:
>
> > Hi guys.
> >
> > Well, I'm not "exactly" trying t reopen an old discussion. I'm
> > trying to see a different aspect from it, as the subject indicates.
> >
> > Since the message reproduced below is the last of the "fudges"
> > questions, I'm worried about the fact of the fudgeLJ meaning
> > nothing for the simulation.
> >
> > First, is it true, and there is no way around to get a working
> > fudge LJ of 0.5 AND a fudgeQQ of 0.8333 in gromacs simulations?
> >
> No, you can certainly use fudge factors with Gromacs.
>
> Essentially:
>
> 1. If you do not generate 1,4 ("pair") interactions they will be zero
> unless you specify them explicitly.
> 2. If you do generate them, standard combination rules of the force
> field will be applied, combined with the fudge factor.
> 3. If you generate pairs and still specify some interactions, those
> will override the generated values.
>
> Cheers,
>
> Erik