[MMTK] Small energy differences

Konrad Hinsen hinsen@cnrs-orleans.fr
10 Jul 2002 11:45:59 +0200

Berit Hinnemann <Berit.Hinnemann@fysik.dtu.dk> writes:

> great that MMTK 2.2 is out. I noticed that minimizing a particular
> configuration (two glycine chains in periodic universe with Amber94)
> that I get slightly different energies with different versions of MMTK
> (2.1b6 and 2.1b9, now I am trying 2.2). The differences are always
> smaller than 0.5 kJ/mol, which seems pretty small to me. Is that
> something you would expect because of minor changes, or should I look
> into this?

Do you get different energies for the same conformation, or different
conformations as a result of the minimization procedure? In the former
case, energy differences should be negligible (1e-5 or less). In the
latter case, it depends very much on the system.

There have never been beta releases of 2.1, so either you are comparing
to 2.1.6 or to 2.2b6. In either case, the energy evaluation code
has not changed between then and 2.2. The minimization code hasn't
changed either. However, the installation procedure has changed and
uses somewhat different compiler options now, which can lead to very
small differences in energies, and in the case of minimization they
can be amplified to lead to different conformations.

Between 2.1.2 and 2.1.3, some parameters for the Amber force field
have been corrected, which of course causes different energies.
See Doc/CHANGELOG for details.

Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-
Rue Charles Sadron                       | Fax:  +33-
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais