[Pw_forum] Generating a pseudo with semicore states

[Paolo Giannozzi]

Lorenzo Paulatto wrote:

> It is a bit complicated. The configuration in the test section is (can  
> be?) used for descreening the potential. I think it was useful for some  
> kind of trick, used to generate semi-core pseudopotentials before the  
> multi-projector form was invented.
> 
> I would advise you to remove it from the input file completely, because if  
> it is present and it has a different configuration than the generation  
> one, it is likely to mess up the pseudopotential.

unless something has been changed recently, I don't think the code
will automatically recognize the correct number of valence electrons,
even with multiple projectors, if you have electrons both in shell N
and in shell N+1. This is why the test configuration is used for
descreening. It would if you could specify both sehll N and shel
N+1 as reference states for pseudization, but usually you cannot

P.
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy