[Wien] GGA+U

[Stefaan Cottenier]

> I have a question about the LDA+U method.
> In the LDA+U method the orbital dependent potential  is added to the LDA
xc
> potential  and then the double-counting  formula is substructed.
> Since the LDA part remains unvaried is it possible to consider a GGA xc
> potential   and add to it the orbital dependent potential  and thus have a
> GGA+U method ?

Yes, this is possible. From what I've heard, LDA+U1 = GGA+U2 with U2 a bit
(0.1-0.2 Ry) smaller than U1, reflecting the amount of correlation that is
contained in GGA but not in LDA.

Stefaan