[Pw_forum] Hubbard U approach in PWSCF

[Matteo Cococcioni]

Dear Hanghui,

curiously this topic was discussed just few days ago in this forum.
presently in pwscf only the simpler scheme is implemented (U_eff = U - 
J). At the beginning we thought that having J was not
so important and so we opted for the most readable and simplest 
implementation. Some papers appeared recently, however,
seem to suggest that J has some role.
A prehistoric version of the code exists that has U and J exists but 
that needs to be re-implemented and put to work in the latest version.

regards,

Matteo



hanghui chen wrote:
> Dear PWSCF users,
>       Maybe this question is too technical but I would like to give a try.
>       There are two LSDA+U approaches in VASP. One is the so-called 
> 'fully localized limit' developed by Anisimov, et al. In this approach 
> you need to specify both U and J (two adjusting parameters). The other 
> approach is a simplifed one developed by Dudarev et al (PRB 57 1505, 
> 1998) in which there is only one parameter Ueff = U - J. 
>       In PWSCF we only have one adjusting parameter Hubbard_U 
> (Hubbard_alpha is something else). Does that mean PWSCF implements the 
> second approach and Hubbard_U = Ueff? However from the menu 
> INPUT_PW.txt, the references you give are the papers from Anisimov 
> et al. Which approach do you implement?
>       If  I want to reproduce some results from VASP calculation, 
> shall I turn on Hubbard_U = U-J where U and J are the input parameters 
> in the VASP?
>      Thank you very much.
>  
> Hanghui Chen
> Department of Physics
> Yale University
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>