Question about pair style gauss

Dear lammps developers,

I’m a little confused about signs in gauss and gauss/cut potential.

In gauss:

if (eflag)

evdwl = -(a[itype][jtype]*exp(-b[itype][jtype]*rsq) - offset[itype][jtype]);

In gauss/cut:
ugauss = pgauss[itype][jtype]exp(-0.5rexp*rexp);

if (eflag) {
evdwl = ugauss - offset[itype][jtype];
evdwl *= factor_lj;
}
So, positive A corresponds to negative E in pair style gauss and positive H/sigma corresponds to positive E in pair style gauss/cut.
Is that correct? If yes, formula on doc page need to be change.
Also, I’ve found a small bug on http://lammps.sandia.gov/doc/pair_gauss.html: links to (Lenart) and (Jusufi) are broken.

Regards,

Grigory

The 2 gauss versions were developed independently, so they
may have different sign conventions. I'll let Axel comment.

Steve

Grigory - I agree with you that the doc page formula for pair gauss needs
a minus sign. Just made the change. So the 2 gauss pair styles do
have different sign conventions. I think the code for both thus requires
no change. Do you agree?

Axel, there are 2 citations in the doc text for pair gauss/cut that are not
listed at the bottom of the page ...

Steve

Yes, I agree with you. pair/gauss is useful for creating potential wells, thus minus sign in formula is more convenient. And this is implemented in the current code.

Regards,

Grigory

2012/10/1 Steve Plimpton <[email protected]>