[Wien] Hartree-Fock and the Hubbard Model

[tran at theochem.tuwien.ac.at]

Hi,

DFT+U is a cheap but rather rough approximation of HF.
Beside this, there is also the difference that in DFT+U, the Coulomb
operator is attenuated in order to account for the screening due to
correlation. In HF, no correlation is included.
In hybrids, the screening is included by using only ~25% of HF exchange
(and there is also correlation coming from a LDA/GGA correlation term).

Hybrid, onsite-hybrid and DFT+U are more or less the same,
since all of them are one-electron methods and mix HF with LDA/GGA.
As Karel said, they are better than LDA/GGA, but can not reproduce
the experimental observations that are beyond the one-electron methods.
DMFT is better since it is a beyond one-electrons method.

Read that:
https://en.wikipedia.org/wiki/LDA+U

FT

> On Wed, 16 Nov 2016, delamora wrote:
>
> 
> Dear Fabien Tran and Karel Vyborny,
> 
> Thanks for your comments.
> 
> 
> What I want to know is if the Hartree Fock exchange is what the Hubbard U is
> trying to model
> 
> What I know is that for strong intraatomic repulsion, 3d and 4f, the Hubbard
> U gives good results, although the U is a parameter.
> 
> But for intermediate intraatomic repulsion, 4d, 5d, 5f then more expensive
> methods are needed, such as DMFT
> 
> So, if this is the case that the Hartree Fock exchange is what the Hubbard U
> is trying to model then the hybrid functionals would do a better job.
> 
> So, my question is; What are the Hubbard U and DMFT trying to model?
> 
> 
> Cheers
> 
>           Pablo de la Mora