[Pw_forum] scf_must_converge

[Paolo Giannozzi]

On Fri, 2012-11-02 at 10:33 +0000, Davide Tiana wrote:

> why are you using such tight threshold? setting such small value 
> and don't care about the convergence (if you can obtain one) makes
> no sense

it actually does in some cases. It often happens during a NEB run 
or a structural optimization that the code stops because it hasn't
achieved convergence at a single step. Sometimes it is convenient 
to just ignore the problem and keep the code running: the problem
may disappear at the next step

P.
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy