[Wien] initial projections for wien2wannier with spin orbital coupling

[Elias Assmann]

On 12/12/2014 07:34 AM, Zhu, Jianxin wrote:
> 1/ runsp_c_lapw –so  is one way for spin-orbit coupling but
> non-spin-polarized case.

runsp_c does a *spin-polarized* calculation where up and dn are 
*constrained* to be equal.

> Elias has also suggested the following way after faking the spin
> polarized calculation --
> $ x w2w -so -up && x w2w -so -dn
> $ cp CASE.eig CASE.eigup
> $ cp CASE.eig CASE.eigdn
> $ x wannier90 -so

I think you are referring to this message 
<http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg11327.html>. 
  This is just due to a minor bug in the ‘wannier90’ wrapper script. 
Anyway, it applies to non-SP SO cases, and so should not be relevant here.

> 2/ For the initial projection, the following order may be more
> consistent with the Bloch wavefunction in the spin space (spinor).

Let me note that the order of orbitals is completely up to your 
preference (it will make no difference to the calculation).

> From: 李志 <zhili at hfut.edu.cn <mailto:zhili at hfut.edu.cn>>
>     Firtsly, I do a spin polarization calculation with SOC, and zero
>     local magnetic moment, i.e. runsp_c_lapw -so
>
>     Then, preparew2w_dir wannier
>
>            init_w2w -up
>
>     However, 26 inital projections are requied, and we just 7+5+1=13
>     orbiatls basis.

With SO, it is not possible to treat “up” and “dn” states separately. 
You need to define 13*2=26 Wannier functions; Amn and Mmn will be 
computed for the “up” and “dn” parts and then added, i.e.

Amn = Amn(up) + Amn(dn)
Mmn = Mmn(up) + Mmn(dn)

>     So, I should type into f,d,and s-orbitals two times,i.e.:
>
>     1:f
>
>     1:f
>
>     1:d
>
>     1:d
>
>     1:s
>
>     1:s
>
>     correct?

That looks correct.  For brevity, I would write it as

1:s,p
1:s,p


	Elias