[Pw_forum] LDA+U calculations to get U and J

[Dmitry Korotin]

Dear Gianluca,

> i would like to calculate the U and J for a given material Fe based.
> I can do LDA+U calculations without problems is such system using PW.

If you are going to use U and J within original QE code you should
calculate U and J for atomic states as Matteo Cococcioni suggests.

> In the current distribution of PW in the directory PP there is a file
> wannier_ham.f90.
> I believe it is referring to the paper:
>
> http://xxx.lanl.gov/pdf/0801.3500
>
> The subroutines are reading wfs and eigenvalues of a given self-consistent
> calculation and they calculate the Hamiltonian in Wannier basis set.
> This gives the possibility to calculate on-site energy of given d states.
> If now we change the occupancy at a given site by the flag Hubbard_alpha()
> and we recalculate the Hamiltonian in Wannier basis set for such
> self-consistent calculation, the variation of eigenvalues respect to the
> occupations should give the U and J parameters.
> Is this correct?
> Is this the procedure used in the mentioned paper?

In the mentioned paper was used very similar but another procedure.
With use of the flag Hubbard_alpha() you are able to change occupancy
of a pure atomic (pseudoatomic) orbital. If Wannier function differs
noticeably from pure atomic state standart procedure is not suitable.

-- 
Best regards,
Dmitry Korotin

Ph. D. Student,
Institute of Metal Physics
S. Kovalevskaya, 18
620041 Ekaterinburg GSP-170
Russia