[Wien] Q: symmetso - B type symmetries

[jan kunes]

Dear Eeuwe,

I do not use symmetso myself, but can say something about how the files 
shouls look like. The main problem with the case you mentioned 
(SOC+polarization-inversion) is that the k-space and real space point 
groups are not identical. Therefore when you generate your klist you need 
different set of symmetry operations (case.struct) than that you use in
lapw1, lapw2, lapw2. The procedure is as follows. 
(1) You find the crystallographic symmetry (x symmetry)
(2) You find how the symmetry is reduced by the presence of uniaxial
field (this depends on the direction of the magnetization definned in 
case.inso)-at this point you do not distiguish between time-even (e.g. 
uniaxial pressure) and time-odd (magnetic, exchange) fields. This comes in 
the next step.
(3) Now, because the exchange field (unlike uniaxial pressure) has also 
orientation the symmetry operations which change its sign are not 
symmetries of the system. However, if you add time inversion (complex 
conjugation) to these rotations the composite operation is the symmetry of 
the hamiltonian. This is taken care of in places where necesary in the 
codes (e.g. qtl, optic or lapwdm) and has not to be specified explicitely 
in the inputs.
(4) If you apply complex conjugation to a Bloch function you change the 
sign of the corresponding k vector, i.e. in k-space the time inversion is 
represented by inversion. This means that to all the rotations, to which 
you have added complex conjugation, inversion should be added when 
k-space symmetry is involved. In the wien code this happens only when you 
generate the klist.

I hope this could help a bit although I know that it is not very 
comprehensive.

Jan

On Wed, 8 Sep 2004, Eeuwe Zijlstra wrote:

> Dear all:
> 
> I am trying to do a spin-polarized calculation with spin-orbit coupling
> for a system without inversion symmetry. I noticed that "symmetso" keeps
> the B-type symmetries, which it should not, according to the Userguide
> (Ch. 9.1).
> 
> The same behavior has already been noted before in this mailing list on
> Monday September 23, 2002 by Torsten Andersen, who wrote: "I still believe
> there is an undocumented feature (or bug) in symmetso for this CXY case.
> It also did not remove the B types, as it should according to the manual,
> p. 129, for a system without inversion symmetry such as this one."
> 
> Unfortunately, nobody has addressed Torsten Andersen's latter point in
> this mailing list (at least, I couldn't find it).
> 
> I would like to know, whether the above-mentioned behavior of symmetso is
> a bug or the correct behavior (in which case I somehow misinterpreted Ch.
> 9.1 of the Userguide).
> 
> Any additional information will be highly appreciated!
> 
> Sincerely,
> 
> 
> Eeuwe Sieds Zijlstra
> Postdoctoral Fellow
> Department of Physics
> Brock University
> St. Catharines, Ontario
> Canada
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>