The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear James,

In ABINIT there is EITHER the spin-orbit treatment, in the
non-spin-polarized case (lifting of degeneracies
for band structures), OR magnetic treatment (collinear or non-collinear)
Unfortunately, it is NOT possible, at present, to treat
spin-orbit for a spin-polarized material.
The modifications are likely very simple to do, but it is not
done.

One more detail, about Matthieu's mail :
spin-orbit is not implemented with perturbation theory,
the full treatment of spin-orbit is implemented, thanks to pseudopotentials
that have a L.S part.

Best regards,
Xavier

---------------------------------------------------------------------


Hi James,

You're absolutely right, even if the wavefunctions are full spinors you
can always calculate the average of the spin operator and get a spin up
and spin down value. For the moment this isn't in abinit, but it should be
fairly easy to add (I had thought about it a while).

I expect (could someone confirm?) that you just need to calculate
|psi_up|^2 and |psi_down|^2 separately. The sum of the two is 1 by
normalization, and with the current implementation of SO (perturbation
theory) the two spinor bits should correspond to spin up and down
(Xavier?).

This might even already be in the code, when the total orthogonalization
and normalization is done. The loops over spinor component are always
innermost after sums over plane waves. You'd just need to output the
numbers then.

If you keep the numbers for each band and kpoint in a big array, and pass
it up to scfcv, you can call tetrahedron with the spin weights in
dos_fractions(:,:,:) and get a DOS weighted accordingly (make a new
partial_dos_flag value).

cool.

Matthieu



On Tue, 14 Sep 2004, James Raynolds wrote:

> To all,
>
> I need to get a spin-projected density of states for a magnetic system
> including spin-orbit.  With spin-orbit one is forced to use spinor
> wavefunctions (nspinor=2) which mix the spin-up and spin-down components.
> In a magnetic system, however, it is still resonable to compute the
> expectation value of the spin operator to get a spin-projected density of
> states.
>
> Is this possible in abinit as it stands or do I need to write some code of
> my own to get it?
>
> Thanks for any help anyone can give me.
>
> Sincerely,
> Jim Raynolds

--
===================================================================
Matthieu Verstraete               mailto:verstraete@pcpm.ucl.ac.be
PCPM, Boltzmann, pl. Croix du Sud, 1            tel: 010/ 47 33 59
B-1348 Louvain-la-Neuve  Belgium                fax: 010/ 47 34 52