[gpaw-users] Calculations in LCAO mode

[Ask Hjorth Larsen]

Hi

On Thu, 23 Feb 2012, Jens Jørgen Mortensen wrote:
> What I meant was, is the f-functions necessary - does it improve things
> to have it?

I suppose it does for lanthanides and friends.

However it sometimes happens that f-functions are generated as 
polarization functions for transitions metals.  This happens when there 
are semicore s and p states (Ti, Sc, etc.).  The basis generator has no 
way of knowing that it should generate a p-polarization function based on 
the valence 4s state.  All it sees is s, p and d functions, and thus it 
generates an f function.  This can nowadays be overridden by doing:

   gpaw-basis --lpol=1

Then the polarization function will be based on (have roughly the same 
range as) the outermost atomic orbital with l=1.

Regards
Ask