[Wannier] Orbital magnetization of individual band

[David Vanderbilt]

Tomasz,

I would add that the contribution of an individual band
may not be very meaningful if it is one of a group of closely
bunched occupied bands.  Avoided crossings between two
such occupied bands generate large, but equal and opposite,
contributions that cancel.  This cancellation is automatically
taken into account in the trace over the occupied group,
which is part of the reason the code adopts this strategy.

On the other hand, if you have separated bands or groups of
bands, for example an isolated group of occupied t2g bands,
then I can see that it makes sense to get the contribution from
this group separately.

David

On Mon, 5 Nov 2018, Ivo Souza wrote:

> Dear Tomasz,
>
> On Fri, 2 Nov 2018, Tomasz Wo?niak wrote:
>
>>  Dear Wannier90 Users,
>>
>>  According to eq. 12.20 in User Guide, the orbital magnetization at given k
>>  point is calculated in the berry module as a sum over all bands. Is it
>>  possible to calculate the orbital magnetization of a single individual
>>  band?
>
> The orbital magnetization is calculated by tracing over the occupied bands. 
> Thus, all you need to do is to trick the code into thinking that the 
> (isolated) band of interest is the only one that is occupied.
>
> The occupancies are set in the routine pw90common_get_occ in the file 
> src/postw90/postw90_common.F90. I suggest setting there the occupancy of the 
> band of interest to one, and the occupancies of the other num_wann-1 
> wannierized bands to zero.
>
>  Best,
>
>  Ivo
>