[Wannier] spin information of Wannier functions

[Jonathan Yates]

On 2 Apr 2017, at 10:36, Chong Wang <ch-wang at outlook.com> wrote:

> Hi Shu-Ting and everyone else:
> 
> I think I figured out what happened. 
> 
> If we do not use spinor feature in wannier90 but include spin orbit coupling in ab initio calculations, Steepest-descent minimization does not mix spin up and spin down. Thus, the orbitals keep their spin information.

 This is not correct.

   The spinor keyword in wannier90 only changes how the projections are written (and also in some of the postw90 routines for deciding how to occupy the bands i.e. it sets the value of the variable num_elec_per_state). It has no effect on the wannierisation procedure.
 What is likely is that the minimisation does not change much the spin character of the WF from the initial projections. So your observation is roughly true. But this not imposed anywhere - and without checking you cannot be sure.

 We did once try an alternative procedure of first dividing the bands into two subspaces (roughly corresponding to spin up and spin down), and then minimising each space separately. The results were very close to minimising all the bands together. For this reason we didn’t include this in the main code.  see https://doi.org/10.1103/PhysRevB.74.195118 There might be cases in which this approach would be useful.

 Jonathan



-- 
Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK
tel: +44 (0)1865 612797                http://users.ox.ac.uk/~oums0549/