[gmx-users] gmx 3.1.4 vs 3.2.1

[Marcos Villarreal]

On Friday 05 November 2004 11:25, Berk Hess wrote:
> From: Marcos Villarreal <arloa at mail.fcq.unc.edu.ar>
>
> >Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
> >To: gmx-users at gromacs.org
> >Subject: [gmx-users] gmx 3.1.4 vs 3.2.1
> >Date: Fri, 5 Nov 2004 10:53:50 -0300
> >
> >Hello,
> >
> >I am testing gmx 3.1.4 vs 3.2.1. For several  systems, I have calculated
> >the
> >potential energy using the same pdb, topology, compiler, etc.
> >When comparing the results,  the Long range part of the LJ energy is the
> >only
> >term that varies by a significant amount (about 10%) between the two
> >versions. The LR-LJ energies calculated with 3.1.4 are always smaller than
> >the calculated with 3.2.1.
> >When you use a twin-range cutoff of 0.9-1.2 and compare with the results
> >made
> >with a single 1.2 nm cutoff, the sum of SR-LJ + LR-LJ of the twin-range
> >only
> >coincides with the result of the single cutoff when you use 3.2.1.
> >
> >Is this indicating that the LR-LJ term is not correctly calculated in gmx
> >3.1.4?
>
> I for your twin-range in 3.1.4 rcoulomb < rvdw?
>
> The changes pages states for 3.1 to 3.2"
>
> A bug has been fixed which could cause some LJ interactions at distances
> close to rvdw to be neglected when rvdw > rcoulomb (in practice this would
> only occur with GROMOS forcefields combined with PME).
>

Yes, I was using GROMOS with PME
Thanks, Marcos.