[Wien] min_lapw / *** false convergence ***

L. D. Marks L-marks at northwestern.edu
Tue Mar 29 18:01:16 CEST 2005


This is not in the FAQ, so let me give a slightly long(winded?)
explanation.

The mini code will optimize your structure using your conditions (e.g.
RMT, RKMAX) and the forces. It is not clever enough to correct your
conditions, so if you get them wrong, GIGO. It often happens that the
forces and/or energies are not that accurate (e.g. 0.1 mRyd/au or worse
for the forces) so the mini code will keep trying to minimize the
(inconsistent) energy/forces by taking smaller and smaller steps (reducing
the trust region radius). Eventually it will give up and print out
"*** FALSE CONVERGENCE***". Provided that your forces truly are small,
this is not a problem; you probably just have too small a value for the
force tolerance in case.inM. You should check, however, that you don't
have touching or nearly touching atoms or unreasonable values of RKMAX,
GMAX etc. If you don't, ignore the "*** FALSE CONVERGENCE***".

On Fri, 25 Mar 2005, Olivier Delaire wrote:

> Dear Wien-users,
>
> I was wondering about the meaning of the "*** FALSE CONVERGENCE***"
> which sometimes appears in the case.outputM file.
> I get this message in case.outputM , but  I still get small forces at
> the end of the case.outputM file, and an apparently trouble-free
> termination. The message shows up after a series of Cauchy steps of size
> becoming smaller and smaller and showing up as "0.000" and a decreasing
> value of "radius" all the way down to 0.438E-12.
> Also, the forces reported in case.scf_mini are different from those in
> case.outputM and those are VERY large.
> Could someone give me some explanations as to what may be going on?
> thanks and regards,
> Olivier.
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

-----------------------------------------------
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60201, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L - marks @ northwestern . edu
http://www.numis.northwestern.edu
-----------------------------------------------




More information about the Wien mailing list