[Pw_forum] Using a pseudopotential with ghosts

[Paolo Giannozzi]

On Nov 7, 2012, at 18:36 , Pietro Bonfa' wrote:

> The US pseudopotetial works perfectly for all the configurations I
> tested except for the one with an additional electron i.e. 4s2 4p6  
> 4d2.0
> 5s2.0 (the pseudo has semicore s and p in valence) where ghosts  
> appear.

negative ions are typically unbound with local or semilocal functionals
so some funny behavior will be present anyway

> The logarithmic derivative looks good enough in the energy range ± 8
> Ry but a (ghost?) spike appear for ~9 Ry in the p channel where the  
> all
> electron log derivative is almost flat. [...]
>
> So the point is: as long as the pseudopotential is not used in 4d2.0
> 5s2.0 configuration, is it safe to use it?

I would say so. At sufficiently high energies the logaritmic  
derivative of all
PPs differs anyway from the all-electron one. 9Ry is a rather large  
energy
and I don't expect anything physically relevant to happen in that  
range of
energies. Moreover Y is more likely to be in a 3+ state than in a -1  
state.
No warranty (as usual).

P.
---
Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222