[Pw_forum] Optimal use of bfgs_ndim?

[Guido Fratesi]

Dear PW users,

has anyone made some study on the optimal choice of bfgs_ndim for a pw.x 
structural relaxation (calculation='relax')?

Increasing its value above the default (which is 1) can reduce the 
number of BFGS steps, and I'm wondering if there could be some intuition 
to guess a reasonable choice of bfgs_ndim, e.g., avoiding to take too 
large values.

If that helps, I'm mostly concerned about optimizing molecular 
coordinates at surfaces.

Many thanks,
Guido

-- 
Guido Fratesi

Dipartimento di Scienza dei Materiali
Universita` degli Studi di Milano-Bicocca (Italy)