[Pw_forum] Question about potential (pp tool)

Lilia Boeri L.Boeri at fkf.mpg.de
Wed Oct 4 18:04:33 CEST 2006


dear Stefano,

on the contrary it is very useful, thanks a lot.
One last question: is there any way to visualize what the other (i.e. 
non-local) part of the scf potential look like in PWSCF?
thanks,
Lilia

On Wed, 4 Oct 2006, degironc wrote:

> If you are reading a psedopotential in the OLD fomat (not the UPF format)
> there is an array Vloc and then a number of
> DeltaV_l, and then there is also a variable called lloc that specify which
> l-component is considered as local.
> Inside the PW code then  Vloc_at(r) = Vloc(r) + DeltaV_lloc(r) [this is the
> quantity that enters in Vbare referred to in pp.x] and the NON-LOCAL part is
> built  from DeltaV_l - DeltaV_lloc.
> If the pseudopotential were used in the semilocal form (not in the KB form)
> different choices for  lloc would all coincide for
> 0<= l <=lmax ( max ang component in the pseudo) and only angular component
> correspondig to l > lmax would be affected... this should not have large
> effects (this form is not anymore implemented in PW).
> In the KB fully non-local form different choices of lloc therefore produce
> slightly different pseudopotentials that  all gives the same results on the
> atomic configuration used for the KB projection and should behave similarly
> provided they do not have GHOST states.  Usually lloc=lmax (or lmax+1) would
> be the most "physical" choice but this is not always possible due to the
> ghosts problem above.
> 
> Using UPF format this little algebra has alredy been done in the
> pseudopotential file and Vloc_at above is given direclty in the PP_LOCAL field
> while the KB projections = DeltaV*Psi_atomic are given in the PP_BETA fields
> and lloc is not needed anymore.
> 
> Sorry if  this explanation confuses you again...
> 
> stefano



More information about the users mailing list