[gmx-users] Energy conservation issue

[Berk Hess]

>From: Yujie Wu <yujie.wu at hec.utah.edu>
>Reply-To: wu at hec.utah.edu,Discussion list for GROMACS users 
><gmx-users at gromacs.org>
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Subject: Re: [gmx-users] Energy conservation issue
>Date: Thu, 27 Oct 2005 15:07:40 -0600
>
>
>Thanks, Berk.
>
><snipped>
> > I would say that the LJ cut-off is the largest source of errors.
> > I don't know how DL_POLY handles this.
>
>I checked its code; it is a simple truncation (without smoothing
>whatsoever).

I just ran your system in 3.3 for 1 ns.
in single precision I get -70 kJ/mol drift
in double precision I get +30 kJ/mol drift

With a plain cut-off one of the biggest sources of error is the
charge group based neighborlist.
When I change the water topology from one charge group to
three (one for each atom) or when I used a shift function for the LJ
I get 0 drift in double precision (300 ps) and about -85 kJ/mol/ns
for single precision.

One can not expect that single precision does as good as double
precision. But the -85 kJ/mol/ns is already very good and comparable
to other errors introduced by the simulation setup, such a charge group
based neighborlists and PME approximations.

Only for very accurate energy conservation one needs double precision.
However, I would say in constant temperature simulations errors of this
magnitude have no measurable influence on the ensemble.

That said, it is still interesting to know what the main source of errors
is in single precision.

Berk.