[Wien] How to set the mixing parameters?

[L. D. Marks]

These are not close to appropriate. What matters are:
1) The mixing Factor
2) The ratio PW/CLM scaling
3) The size of the angle output
4) How well you have defined the problem.

A general method is:
	a) Set the mixing to some reasonable value (if you are having
problems), e.g. 0.1. A value of .001 is meaningless and in fact ignored in
large part by the code.
	b) Set the ratio PW/CLM to 0.1, i.e.

BROYD  0.0   YES  (BROYD/PRATT, extra charge (+1 for additional e), norm)
0.1            mixing FACTOR for BROYD/PRATT scheme
0.1  1.0     PW and CLM-scaling factors

	c) Run, and look at the angle. If it is large (often > 60)
stop, remove the broyden files, reduce the ratio then continue
	d) If you are still having problems, check if you have correctly
posed the problem. For instance, if you try and run NiO without spin
polarization the convergence will be very poor. The better the
Hamiltonian, the more convex (i.e. convergent) the algorithm will be.
Convergence problems can be due to bad RMT's, KMAX etc.

If these fail (and only if) then repost.

On Sat, 14 May 2005, liyh wrote:

> Dear WIEN users,
>
>       This is my case.inm file:
> BROYD  0.0   YES  (BROYD/PRATT, extra charge (+1 for additional e), norm)
> 0.001            mixing FACTOR for BROYD/PRATT scheme
> 0.001  0.003     PW and CLM-scaling factors
>
> _______________________________________________
> 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
-----------------------------------------------