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
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
