[Wien] A question about RMT and AIM

[Laurence Marks]

We need more information:
1) How many clm's do you have in case.in2? I suggest that you expand
these to 10.
2) Did you re-run the calculation to consistency after changing to
lapw? Also, have you checked that the linearization energies are OK?
3) Did you change both NAPWL and NAPW to zero?
4) Are you using current versions of aim & lapw5 -- there were some
bugs in the first 6.2 release.

If these don't help and you are computer competent enough to tweak
some of the scripts I have a small utility which will check the
discontinuities that I can send you directly -- too big for the
mailing list (125k).

On 12/5/06, Dr. Grigori V. Vajenine <G.Vajenine at fkf.mpg.de> wrote:
> Dear All,
>
> A quick question on discontinuities in the calculated electron density:
> I see that in my cases and I tried to get rid of these by switching to
> LAPW as mentioned below. For that I changed the NAPWL switch in case.in1
> from 1 to 0 for all orbitals, but I still get the steps in the computed
> electron density at the RMTs (the el. dens. is higher outside of the
> spheres). What could be the reason? What did I miss?
>
> Thanks in advance,
> Grigori Vajenine
> MPI-FKF
> Stuttgart
>
> A
>
> Laurence Marks wrote:
> > Aim is a hard calculation to get right because there is not currently
> > any really good algorithm for finding the BCP and surface, and it is
> > done by a brute force search method. What can certainly create
> > problems are:
> >
> > 1) The discontinuities at the RMT's. You need to use very high clm's
> > (in case.in2) to reduce this; for a small calculation like you have
> > you may want to use an lapw basis set (rather than APW+lo) since this
> > has no discontinuity. (I have a small utility to calculate the
> > discontinuities.)
> >
> > 2) Insufficient sampling. At the end of case.outputaim there are some
> > error guesstimates which are approximately correct (within a factor of
> > 2). If these are large you need to increase the sampling in
> > case.inaim.
> >
> > 3) Using too large an initial R and/or too large steps.
> >
> > In general, calculations with smaller RMT's should be more reliable
> > albeit slower.
> >
> > On 11/27/06, G.C.ZHOU <gc_zhou1982 at 163.com> wrote:
> >
> >>
> >>  Dear Professor and users:
> >>  I am doing the AIM calculation of ZnO, it has three
> >> structure:wurtzite,zinc-blende and rocksalt.
> >>  when I do the calculation of BCP,the other setting are all same,I find RMT
> >> affect the result very greatly.
> >>
> >> the structure is wurtzite,
> >>    and set R(Zn)=1.75, R(O)=1.85, the Laplace of BCP is 5.003;
> >>    when set R(Zn)=1.85, R(O)=1.90, the Laplace of BCP is 0.580,and the
> >> position of BCP is not correct.
> >>
> >> the structure is zinc-blende,
> >>   and set R(Zn)=1.75, R(O)=1.85, the Laplace of BCP is 4.993;
> >>   when set R(Zn)=1.85, R(O)=1.90, I can't find the BCP.
> >>
> >> the structure is rocksalt,
> >>   and set R(Zn)=1.75, R(O)=1.85, the Laplace of BCP is 4.511;
> >>   when set R(Zn)=1.94, R(O)=2.10, the Laplace of BCP is 3.256;
> >>
> >> I am puzzled with this results, but the two results of evety structure are
> >> too different with only the RMT is changing. And the Laplace of rocksalt's
> >> BCP tells that, in the rocksalt structure, the covanlent of Zn-O bond turns
> >> more then wurtzite, it's contrary with our earlier cognition, which is the
> >> bonding became more ionic in this structure.
> >> even worse, when I reduce the RMT, the laplace will turns smaller yet.
> >>
> >> I set RKmax=9.0, LM=10, and set Ecut=-6.5Ry, there is no electron leak out.
> >> and use x lapw -c to do the program.
> >>
> >> Can tell me why the BCP change so much in different RMT, and which
> >> result should I take as the right one.
> >>
> >>  Any advices is welcome!
> >>
> >> Thanks.
> >>
> >>  ________________________________
> >>
> >> G.C.ZHOU
> >> 2006-11-27
> >> _______________________________________________
> >> 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 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
http://www.numis.northwestern.edu