[Wien] HoMnO3

[Laurence Marks]

For more iterations, use "-i XYZ -NI" to continue.

One important comment: the "mixing factor". Use a mixing factor of
0.1-0.3, except in extreme cases never reduce it below 0.1. If the
calculation is not converging this is probably because you have the
physical model wrong, which I think can easily happen with
f-electrons.

Conventional "wisdom" is that one reduces the mixing factor to achieve
convergence, but this is only for algorithms such as a classic Pratt
or Good/Bad Broyden not for the multisecant in MSEC1. In fact reducing
the mixing factor does not do what conventional wisdom says. More
correct is to consider it similar to a Wolfe condition
(http://en.wikipedia.org/wiki/Wolfe_conditions); you need to take a
sufficient step along the unpredicted direction otherwise quasi-Newton
methods (which all these algorithms are) will not converge.

The mixing factor controls how greedy the algorithm is (and I intend
to change the name in the next version). Too large a greed is bad and
leads to instabilities; too small a greed and the algorithm starves to
death.

On Tue, May 25, 2010 at 9:12 AM, John Rundgren <jru at kth.se> wrote:
> Dear W2k experts,
> HoMnO3 (30 atoms/unit cell and f-electrons) stops with charge not
> converged after 40 iterations.
>
> A lower mixing factor can be set in case.inm.
>
> Where can the max number of iterations be redefined? The dayfile
> starting with the remark "lapw0 (40/99 to go)" seems to indicate that
> the iterations terminate in two ways.
>
> Grateful for hints,
> John, KTH
>
>
> _______________________________________________
> 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 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Chair, Commission on Electron Crystallography of IUCR
www.numis.northwestern.edu/
Electron crystallography is the branch of science that uses electron
scattering and imaging to study the structure of matter.