[Wien] GGA+U

[Pavel Novak]

my present (incomplete) understanding is that it does not matter whether
LDA+U or GGA+U is used. Peter Blaha found in HTC cuprates that these are
(almost) equivalent providing somewhat smaller U is used in GGA+U
(P. Blaha, K.Schwarz, P.Novak, Int. J. Quant. Chem., in print). We found
recently that the same is true for iron in magnetite and hexaferrites.
Simple, maybe too simple, explanation is that GGA 'contains more
correlation' than LSDA.

Regards
Pavel


On Mon, 22 Nov 2004, Stefaan Cottenier wrote:

>
>
> > there in fact is a paper of Savrasov and Kotliar, PRL 84,3670 (2000) in
> > which GGA+U is used. GGA+U is also used in one of A.S.Shick papers (I do
> > not remember the reference) and we submited recently two papers using
> > GGA+U.
>
> Hello Pavel,
>
> Could you tell us wether there is a common reason in all these papers to use
> GGA+U rather than LDA+U? Do they deal with materials which are significantly
> better described by GGA than by LDA, such that one wants to start with the U
> from this better description? (and if true, couldn't it be that LDA+'small
> U' is more or less equivalent to GGA also for these materials, such that in
> the end it doesn't really matter whether LDA/GGA+U is used)
>
> Stefaan
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>