[Pw_forum] Berry phase electric field calculations with empty bands

[Alexander Urban]

Thank you for your response, Paolo Giannozzi.


On 11/01/2011 04:11 PM, Paolo Giannozzi wrote:
> based on my little knowledge of the subject, I would say (1) yes,
> (2) likely no.
That is what I thought...

> It should be relatively straightforward to generalized
> the current algorithms to work even in presence of unoccupied
> bands (it is sufficient to consider the correct weight), but
> i) it hasn't been done (or it hasn't been completed), and
> ii) it must be done in a proper way, avoiding to increase
> the complexity of the code.
I do consider to introduce the necessary modifications.

Maybe someone more familiar with the PWSCF code could point me into the 
right direction?

My thoughts:

- The electric enthalpy SCF wavefunctions are wrong (with empty bands)
   because the electric polarization P is wrong.
- The reason for the error in P is that the sums in the computation
   of P in `c_phase_field()' run over all nbnd bands where it should
   only be the occupied ones (i.e. the occupation numbers have to be used
   as weights).
- The only subroutine that has to be modified is `c_phase_field()'.

If this reasoning is correct, I would take a closer look at 
`c_phase_field()' and estimate the effort to introduce the changes.

Thanks,

Alex

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

Alexander Urban (PhD student)
Interdisciplinary Center for Molecular Materials (ICMM)
Computer Chemistry Center (CCC)
University of Erlangen-Nuremberg
Germany