[Wannier] Spin-polarized calculation

[Jonathan Yates]

On 2 Jul 2014, at 20:26, Ghosh, Soham Subhra <ssg09d at my.fsu.edu> wrote:

> Dear Wannier90 users, 
> 
> I am facing the following issues.
> 
> 1. I am trying to form a few-band model of magnetism in a system that has a few bands crossing the Fermi level.  I understand that orbitals are strongly localized only in insulators. Does it make any sense to use Wannier90 in my system?

yes. Read about wannier functions from entangled bands in
http://journals.aps.org/rmp/abstract/10.1103/RevModPhys.84.1419

> 2. I have the VASP generated DFT groundstate of a spin-polarized system. I want to target a band which is a up-spin band. The  corresponding down-spin band is displaced by about 0.5 eV and I do not want to include that in my calculation.  However, the 'spinors' tag in Wannier90 expects to find twice as many initial projections. Is there a way to form a Wannier orbital out of an orbital with a spin index, and if there is, how do I specify the initial projection for that band, since projections do not carry a spin-index?

Run two separate calculations. One for spin up, the other for spin down.
  See example 8 in the wannier90 tutorials. I know these are for pwscf - but I expect it can be easily adapted to vasp.

> 3. Could you kindly tell me why Wannier functions should be mostly real when they are well-localized? I failed to find such a constraint in the derivation of MLWFs. 

It’s not a constraint. There was an initial conjecture that MLWF should be real - and this has been found empirically to be true. There is a “proof” of this (for certain conditions) in the papers on exponential localisation of Wannier Functions - see the relevant section (and cited articles therein) of the above RMP article.

 Yours
   Jonathan




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