[Wien] LDA+U asking again

[Pavel Novak]

Dear Tulika,

find my answers to your questions below.

Regards
Pavel Novak

On Thu, 1 Jun 2006, Tulika Maitra wrote:

>
> I have a few questions regarding LDA+U which I am giving below and I
> will very thankful to receive an answer.
>
> 1. In LDA+U (FLL) where now Ueff=U-J is implemented which are the terms
> that are neglected? If I want to do a calculation keeping terms
> proportion to J (I mean, as implemented in the old version) can I do
> that? This is because, I am getting strange results for my system with new
> LDA+U (Ueff) version  whereas the old version is giving me physically sensible
> results. But in some other systems I tested, it works fine. So I am
> wondering if it could be that in my case the terms proportional to J could
> be important.
Using Ueff and J=0 means that multipole terms in the expansion of
the Coulomb interaction are neglected. Usually this works, but sometimes
it may lead to incorrect results.

>
> 2. In LDA+U (AMF), how is it implemented? Is it also Ueff=U-J or finite
> U and J like in the old version?
There is no 'new' and 'old' version. If your input has nonzero J then
automatically program takes it.

>
> 3. I am working with a system having inversion symmetry but I still get
> a complex density matrix with the new version, what could be the reason?
> In my last e-mail I have sent the density matrices I got.
Form of the density matrix depends on the local coordinate system. E.g. if
you have NiO with center of inversion and you rotate local coordinate
system of Ni out of the main axes, your density matrix will certainly be
complex.


>
>
> Thanking you very much for your reply.
>
> -best regards
> Tulika Maitra
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>