[Wien] Cu - convergence

[Laurence Marks]

The oscillations are not due to the mixer, they are due (almost
certainly) to numerical issues due to limited precision which is
preventing full convergence as well as (for instance) neglect of
higher-order LM components and "noise" from the -in1new switch. The
old advice about how to improve the mixing does not apply to the
current one (it is completely wrong). Using -cc 0.000001 is really too
small in my opinion, and I have only ever used -cc 0.00001.

Things will probably be slightly better if you increase the LM's to 8,
but realistically I think you are concerned about nothing important.

On Tue, Nov 25, 2008 at 7:23 AM, Tomas Kana <kana at seznam.cz> wrote:
>  Dear Wien2k users,
>  I am having problem with convergence of Cu.
>  I am using  WIEN2k_08.2 (Release 21/4/2008).
>  My case is fcc copper  with lattice constant
>  6.831170 a.u.  RMT = 2.23 a.u.
>  I am using
> run_lapw -in1new1 -i 98  -cc 0.000001
>  For RMT*KMAX from 7.0 up to 7.5 it the convergence of
>  charge distance was good. Starting from RMT*KMAX = 8.0
>  to 8.5, the total energy still decreases, however, charge distance
>  begins to oscillate.
>  The charge convergence oscilates between 0.0000107 and
> 0.0000016
>  This happens for both low number of k-points
> (120 in the irreducible wedge of Brillouin zone or 4000 in the whole Brillouin zone)
> and for high number of k-points
>  (1540 in the irreducible wedge  or 60 000 in the whole Brillouin zone,respectively).
>   I tried to improve my  case.inm file according to older advice in mailing list:
>  [Wien] How to set the mixing parameters?
>  http://zeus.theochem.tuwien.ac.at/pipermail/wien/2005-May/005277.html
>  I tried to rewrite the MSEC1 with BROYD  and BROY0 in order to have compatible
>  case.inm with
> BROYD  0.0   YES  (BROYD/PRATT, extra charge (+1 for additional e), norm)
> 0.10            mixing FACTOR for BROYD/PRATT scheme
> 0.10  1.00      PW and CLM-scaling factors
> 9999  8         idum, HISTORY
>
> and  I reduced the ratio CLM/PW - scaling. But I did not achieve the convergence
>  0.000001.
>   The  energies in case.in1 suggested by -in1new switch
>  agree with the  Density of states.
>  The only warning :WARN in case.scf
> comes from the QTL-B value
> QTL-B VALUE .EQ.    3.91227   in Band of energy    0.65110   ATOM=    1   L=  2
> Most likely no ghostbands, but adjust Energy-parameters or use -in1new
>       Do you think I should use  charge convergence -cc 0.000002
>  and ignore the inaccuracy in convergence or could there be some more serious fault?
>               Thanks a lot                   Tomas Kana
>
>
>
>
>
> _______________________________________________
> 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 to study the structure of matter.