Dear Marcos:

I have confirmed with the developer of that
part (Daniel Sanchez) and his answer is:

Yes, the same Ueff will give the same final
results. The current implementation follows
the work of Dudarev et. al. The plan is to
change it for a more rigorous implementation
in a future, though.

Yours,

--
Jose A. Torres, Ph.D.
Manager of the SIESTA Software
tel: +34 914973805



from    Marcos Veríssimo Alves <[email protected]>
reply-to    [email protected]
to    [email protected]
date    Thu, Jan 7, 2010 at 4:21 PM
subject    [SIESTA-L] Doubt on LDA+U

Hi all,

I have a doubt on the LDA+U implementation in siesta, more
specifically on the definition of U and J.

The siesta LDA+U manual states that "although the input allows to
define independent values of the U and J parameters for each atomic
shell, in the actual calculation the two parameters are combined to
produce an effective Coulomb repulsion U_{eff}=U-J. U_{eff} is the
parameter actually used in the calculation for the time being.". Now,
does that really mean what it seems to imply, namely that any
combination of U and J that produces a given U_{eff} will give me the
same final results? For example, if I want my U_{eff} to be 1.0 eV,
then U=0.6, J=0.4 and U=1.4 and J=-0.4 will give me the same final
results?

Cheers,

Marcos

--


Responder a