[Wien] initial projections with spin-orbit coupling for wien2wannier?

[Elias Assmann]

Dear Wenhu Xu,

SOC is one of those areas that are less well explored with wien2wannier,
so please take everything I say with a grain of salt.

On 05/06/2015 07:34 PM, Xu Wenhu wrote:
> 1. Is it possible to mix spin-up and -dn components in the
> 'case.inwf' file to have the Jeff=1/2 orbitals as the trial function?
> I notice there are only LM components in 'case.inwf' and no options
> for the spin sector.

The A_mn and M_mn matrix elements are summed over spins before they are 
passed to Wannier90.  In this sense, yes, spin-up and -dn are mixed, 
though you cannot give a specific linear combination for your initial 
projections.

> Should I also execute BOTH 'init_w2w -up'  and  'init_w2w -dn' for a
> non-spin polarized calculation?

No.  init_w2w should only be called once for a given directory, and the
only thing that the -up/-dn options are used for in that script is
findbands (i.e., figuring out which band numbers lie within a given
energy interval).

With SO, in any case, init_w2w should be run in “-so mode” (although the
explicit -so switch should not be necessary, i.e. just run ‘init_w2w’);
it will then look at case.outputso for the states (and normally find
twice as many states as you would have for one spin).

> 2. If not possible to mix spin-sectors in 'case.inwf', do I need to
> repeat each LM projector  when using 'init_w2w' so there will be two
> for each of them in 'case.inwf' and 'case.win'?

Yes.  Since you have “twice as many states” you should also give twice 
as many initial projections.  Normally, they will just be repeated.


-- 
Elias Assmann

Wien2Wannier: maximally localized Wannier functions
               from linearized augmented plane waves

<http://www.ifp.tuwien.ac.at/forschung/arbeitsgruppen/cms/software-download/wien2wannier/>