[MMTK] Ewald summation parameters

Konrad Hinsen hinsen@cnrs-orleans.fr
10 Jul 2002 13:36:29 +0200

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

> Where do these estimation expressions come from (could you maybe point
> out a reference)? Do they work for most systems? I'd like to test and

There is a reference, but I don't have it at hand. The parameter
estimation is based on a worst-case scenario, it should be "safe"
for all systems, but is grossly inefficient for most. Hand-tuning for
your system is always a good idea.

> -the bigger beta, the more precise the calculation (or how should beta
> be chosen)

No. If the two cutoffs are big enough, then the result should be
independent of beta. In practice, the finite system size puts an upper
limit on the real-space cutoff and thus on the useful range of beta.

Beta controls the relative importance of the real-space and the
reciprocal-space sum. By increasing beta, the real-space cutoff can be
reduced, but then the reciprocal-space cutoff has to be reduced as

You can find discussions about the "optimal" value of beta in the
literature (unfortunately the assumptions behind the estimates are not
always stated explicitly). Note that beta is also often called alpha.
A general problem with the literature about Ewald summation is that
most presentations are wrong somewhere.

> -the bigger real_cutoff, the more precise the calculation


> -the smaller reciprocal_cutoff, the more precise (you actually write
> that in the manual, I think)


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