[Pw_forum] A question about projwfc.x and pseudopotential generation

[Paolo Giannozzi]

On Mon, 2012-03-19 at 15:19 +0100, fabrizio gala wrote:

> I generated a pseudo-potential for Calcium with the valence 
> configuration 3d^{0} 4s^{2} 4p^{0}, with an energy of +0.05 a.u. 
> for the unbound d and p states (...); however after a scf+projwfc.x
> run, there are no projections on such states.

projections are calculated only for bound atomic states. I think that
unbound states are not even saved to the UPF files in the PP_PSWFC
section. Not sure that projecting over unbound states makes any sense.

> Can I solve the problem generating a Ca pseudo-potential with a valence 
> configuration of the form:
> 3d^{\alpha}4s^{2-2\alpha}4p^{\alpha} with \alpha a small fraction of 
> electron charge, or have I to generate the Ca pseudo-potential with a 
> negative pseudo-energy for them? 

you have to produce in a way or another bound 3d and 4p states. This
can be done by changing the reference electron configuration. You
may need to use an ionic configuration, though. Another option is
to keep you PP but generate bound 3d and 4p states from an ionic
configuration, replace the atomic orbitals of the PP_PSWFC section 
(they are not used in actual calculations, except to generate 
starting wavefunctions and projections over atomic states, which 
are ill-defined anyway).

PG
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy