[Wien] about the LDA+U

[Yushan Wang]

hi, what surprise me is that in this non-magnetic system that has
well-balanced spin-up and spin-dn state, the LDA+U method gives the
approxinatly same result as the experimental data, unlike the commonly
used LSA, which usually underestimates the band gap.

yusahn

On Tue, 2 Nov 2004, jan kunes wrote:

> I do not argue that LDA+U gives the same results as LDA. I wanted to say
> that in cases like this LDA+U is not supposed to provide reasonble
> description of reality. This, I believe, is true for systems with strong
> correlations e.g. TM oxides where the filling of the d shell is close to
> integer. I do not know much about how LDA+U is supposed to work in systems
> with less strong correlation.
>
> Jan
>
> On Tue, 2 Nov 2004, Yushan Wang wrote:
>
> > but, I really find the results obtained by LDA+U are different from those
> > by LDA only, although this system has a balanced spin-up and spin-dn
> > structure.
> >
> > yushan
> >
> > On Tue, 2 Nov 2004, jan kunes wrote:
> >
> > > Dear yushan,
> > >
> > >
> > > > Dear users,
> > > > does the LDA+U apply to the light transition metal. such as Ti, V, Cr and
> > > > so on?
> > >
> > > It depends on the compound. It should work reasonably in systems, where
> > > the d electrons are localized, e.g. oxides.
> > >
> > > , and if the spin-up and spin-dn balance well, is the LDA+U
> > > > full of physical significance?
> > >
> > > No, LDA+U applies to ordered phase. If you have local moments as is
> > > typical for localized electrons, LDA+U provides a reasonable description
> > > of the ordered phase below T_c. Above T_c you still have local moments LM
> > > (Curie-Weiss susceptibility), but the fluctuate in time, i.e. this is not
> > > a non-magnetic NM state described by LDA or LDA+U. If you look at the
> > > occupation numbers <n> there is no difference between LM and NM, in
> > > both cases up=dn. The difference will be obvious from <N_up N_dn>. In NM
> > > <n_up n_dn>=<n_up><n_dn> while in LM <n_up n_dn> << <n_up n_dn> due to
> > > Coulomb repulsion (LDA+U however can not capture this behavior since it is
> > > static method).
> > >
> > > Jan
> > >
> > >
> > >  > > yushan
> > > >
> > > > _______________________________________________
> > > > Wien mailing list
> > > > Wien at zeus.theochem.tuwien.ac.at
> > > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > > >
> > >
> > > _______________________________________________
> > > Wien mailing list
> > > Wien at zeus.theochem.tuwien.ac.at
> > > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> > >
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>