[Pw_forum] optimal alpha not found

Paolo Giannozzi giannozz at nest.sns.it
Mon May 9 19:46:40 CEST 2005

On Thursday 05 May 2005 17:43, Eyvaz Isaev wrote:

> Actually, the error means that the Ewald sum (ion-ion
> energy contribution) could not converge for some
> reason. I have to remind that the Ewald sum consists 
> of two parts: erf(\alpha R)/R and erfc(\alpha R)/R.
> The first term has a fast convergence in G-space,
> while  the second term converges in R-space.
> There are also 2 \alpha-dependent contributions,
> excluded from sums.
> Optimum \alpha has to be chosen so that it gives good
> convergence for both G-, and R- sums.

the exact meaning of the error is that the code (PW/ewald.f90)
could not determine a value for alpha giving a converged 
G-space sum for the available set of G-vectors. "Optimal"
is somewhat misleading in this respect. The only reasons
for such error I can think of, iare either a very weird structure,
or an exceedingly low cutoff


Paolo Giannozzi             e-mail:  giannozz at nest.sns.it
Scuola Normale Superiore    Phone:   +39/050-509876, Fax:-563513 
Piazza dei Cavalieri 7      I-56126 Pisa, Italy

More information about the users mailing list