[Pw_forum] More pseudopotential troubles

[Paolo Giannozzi]

On Oct 11, 2006, at 12:15 , Miguel Martínez Canales wrote:

> Although your previous comments on Ca pseudopotential were very  
> helpful,
> now I'm finding another problem. I may be wrong to think that the  
> input attached
> generates an US pseudopotential with semicore states, using l=2 as the
> pseudization channel and two reference energies for the s and p  
> channels.

you should first try something simple before trying something difficult.
A US pseudopotential is not easy to generate. The "atomic/" code
uses Rabe-Rape-Kaxiras-Joannopoulos pseudization: very good when
it works, but it is not easy to find a set of parameters for which it  
works.
You can use Martins-Troullier pseudization, though. Moreover the
"atomic/" code does not accept pseudostates with nodes, so you cannot
use two bound states to pseudize if you have semicore states. You have
to select a bound state, plus an unbound state having an energy that is
not too far away from the energy of the nodeless (when pseudized)
bound state

Paolo