[Wien] equivalent-core approximation

[L. D. Marks]

I strongly suggest that rather than using a very small mixing factor, you
do a grep on :DIRB (later mixers only) in case.scf and look at the value
of ANGLE. If this is large (e.g. > 90 DEGREES) you need to reduce the PW
scaling in case.inm. Reducing the mixing factor is often not the correct
approach.

N.B., this normally works, but I know of one case where it did not and the
only solution was to reduce the mixing and be patient.

On Thu, 24 Feb 2005, Michael Gurnett wrote:

> Thank you Peter for your very indepth reply. This method you have is what I
> originally tried. Howerver, once I made all the changes I found that the SCF
> would oscilate even with a mixing factor of 0.01 and would eventually crash.
> If this is the best method I will try it again, and with any luck I can
> continue on from there.
>
> Once again thank you for your answer.
>
> Michael
>
>
>
> ----- Original Message -----
> From: "Peter Blaha" <pblaha at zeus.theochem.tuwien.ac.at>
> To: <wien at zeus.theochem.tuwien.ac.at>
> Sent: Thursday, February 24, 2005 3:50 PM
> Subject: Re: [Wien] equivalent-core approximation
>
>
> > I'm not an expert in these core level calculations, but for me the z+1
> > approximation always seemed a quite poor approximation and I would not use
> > it if I have another chance.
> >
> > For your problem I propose two methods within WIEN:
> >
> > First of all you should always have a supercell (unless you have a big
> > cell
> > anyway) with broken symmetry so that you can deal with a core hole on a
> > single Ge atom (screened by the environment).
> > In addition, try to make the Ge sphere "as large" as possible (but still
> > meaningfull - details depend on your structure).
> >
> > Firt do a regular WIEN calculation for this supercell to verify that you
> > get
> > identical results as before.
> >
> > Then you have to do some tricks:
> >
> > Edit case.inc and put the Ge 3d electrons into the core for  all Ge atoms
> > (I'm not sure about this, but I guess it is more cosistent than just for
> > the single Ge atom) ?
> >
> > edit case.in2(c) and reduce NE by 10*(number of Ge atoms)
> >
> > edit case.in1(c) (make a backup first) and remove the Ge-3d LO
> > (around -1.5 Ry). Do not forget to
> > change also the "number of exceptions" (see UG description of case.in1)
> >
> > Do an scf calculation. Check the results, i.e. core-leakage and the effect
> > on the Ge-3d eigenvalues (compare the 3d core with the 3d bands with
> > respect
> > to EF). Of course the bands will have a certain bandwidth which is not
> > present in the core (but there you have SO splitting).
> >
> > Once you can estimate the "error" introduced by the core-like treatment
> > you may
> > remove 1 or 1/2 core electron from case.inc and add it to case.in2. Do
> > another
> > scf calculation and estimate your core levels (E-tot or Slater...).
> >
> > At this position, your 3d-core with the hole should be quite separated
> > from
> > the other Ge. You could now try a "2-window calculation" using a case.in2s
> > file.
> > Remove all 3d states from the inc file.
> > Take the original case.in1 file (so that you have again 3d LOs)
> > cp case.in2 case.in2s
> > edit case.in2s and specify NE=9 (just for the Ge 3d with hole)
> > edit case.in2 with NE="all electrons+1"; but set EMIN such that the
> > 3d-states
> > of the Ge-hole atoms are cut-off.
> > cp case.in1 case.in1s
> >
> > Finally you will have to edit   run_lapw
> > Look for     lapw1s:
> > replace the line
> >     total exec lapw1 -sc $para
> > by   cp $file.vector $file.vectors
> >
> > Now run an scf cycle. You should have the Ge-3d states in the valence, but
> > have introduced a hole in the Ge-3d states of one particular atom.
> >
> > It's not simple and maybe I've made an error in the description, but I'd
> > expect it should work. (Of course, if the core leakage is not that bad
> > you might not need the second step).
> >
> >
> >
> >
> >> Ge 3d electrons (approx. -1.5 ryd). So far I have tried manually placing
> >> these
> >> in core and reducing the number of valence electrons. However, this too
> >> was
> >> unstable. Another method I have heard about is simply using the Slater
> >
> > What should that mean ? I'm pretty sure it works.
> >
> >
> >                                      P.Blaha
> > --------------------------------------------------------------------------
> > Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
> > Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
> > Email: blaha at theochem.tuwien.ac.at    WWW:
> > http://info.tuwien.ac.at/theochem/
> > --------------------------------------------------------------------------
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> >
>
>
> _______________________________________________
> 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
-----------------------------------------------