[Wien] LDA+U

[Gerhard Fecher]

Sorry but I have to intervene another time.
Please apologize this hard words, but this is distribution of half-knowledge, please stop that kind of discussion, it will bring more confusion than answers.

For example, there is by no means a direct relation telling if U is zero then electrons are delocalised.

Correlation is already included in LDA thats why one has an exchange-correlation functional. However, as its name (LDA) tells it is an Aproximation.
I cannot repeat here the complete theory and all details of DFT and LDA with its various extensions, therfore I strongly suggest that you use some good textbooks as a starting point (e.g.: Martin, Electronic structure, Cambridge).

You may be partially right, but not that way. There is no experiment that is able to measure U and there is no method to calculate U ab-initio! There are only methods to estimate U from both experiment and calculation by some approximations.

Ciao
Gerhard


-----Ursprüngliche Nachricht-----
Von: wien-bounces at zeus.theochem.tuwien.ac.at im Auftrag von Saeid Jalali
Gesendet: Mi 26.07.2006 00:20
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] LDA+U
 
    Performing careful experiment can answer if a system, i.e.  in your case a pure metal, is highly correlated or not at a given thermodynamic  state. If this were (not) so and you know it from experiment LDA+U is (not)  necessary. If you do not know it and (or) would theoretically investigate  (confirm) it, then performing LDA+U assuming U as an variable seems to be an  appropriate tool. Since if you find zero value for U comparing your result with  experiment, then you can confirm that it is not strongly correlated system. One  can investigate utilizing the constraint method of calculating U value the  degree of localization. If U is calculated to be zero then the system is  predicted purely theoretically to be delocalized. Screening is a good point  that you have touched it. But I suggest to be careful about your interpretation  taking correct model into account:
      The d/f-electrons interact either indirectly as described  by the RKKY model within the conduction electrons, where the spins of the d/f  electrons are coupled together by the magnetization induced in the conduction  electrons, or directly with the other delocalized conduction electrons in the  vicinity of the Fermi level as described by the Kondo model, where spin-flip  scattering between magnetic d/f electrons and nonmagnetic conduction electrons  play more important role than the indirect RKKY interactions of d/f moments.
      If you would correctly interpreter your result you may know  about these two model Hamiltonian reading many-body text books.
  
  Say hello to everybody there.
  
    hamad at fhi-berlin.mpg.de wrote:  Thanks Saeid,
         but as we know that LDA+U is needed for systems with strong
electron-electron interaction, this appears in transition metal
oxides that causes the wrong prediction of metallic type rather
than insulator "Mott-Hubbard insulators".
So what about pure transition-metal systems, I think in this case the
screening effect reduces the electron-electron interaction and in this
case LDA+U is not needed. Am I right??
Thanks
Bothina


>     Yes, it is meaningful. LDA+U is intended to improve L(S)DA  weakness
> of treating strongly correlated f/d density of states. Magnetic
> systems will be non-magnetic  in higher temperature. In higher
> temperatur,
  those systems again have their d/f states which need
> (sometimes) improvement including  LDA+U. However, in practice as
> mentioned in the textbook it is necessary to be treated magnetically
> even non-magnetic systems.  This can be done as  discussed in the
> usersguide and frequently in the  list using runsp_c_lapw script
> (instead of run_lapw) to constrain to be zero the magnetic moment:
>         http://zeus.theochem.tuwien.ac.at/pipermail/wien/2006-March/006863.html
>   http://zeus.theochem.tuwien.ac.at/pipermail/wien/2005-October/005994.html
>   http://zeus.theochem.tuwien.ac.at/pipermail/wien/2004-April/002425.html
>   http://zeus.theochem.tuwien.ac.at/pipermail/wien/2005-December/006304.html
>
> Bothina Hamad  wrote:  Dear Wien users,
>                  I want to ask if is it possible to use LDA+U method in
> a nonspin-polarized calculations? does it have a meaning?
> Best regards
> Bothina
> --
> -----------------------------------------------------------------------
> Bothina Hamad             Phone:  (+49 30) 8413-4851
> Fritz-Haber-Institut      Fax:    (+49 30) 8413-4701
> Van't-Hoff-Straße 19      WWW:    www.fhi-berlin.mpg.de
> D-14195 Berlin (Germany)  E-Mail: hamad at fhi-berlin.mpg.de
> -----------------------------------------------------------------------
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
>
> Sincerely yours,
> S. Jalali.
> /_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> Saeid Jalali Asadabadi,
> Department of Physics, Faculty of Science,
> University of Isfahan (UI), Hezar Gerib Avenue,
> 81744 Isfahan, Iran.
> Phones:
> Dep. of Phys.      :+98-0311-793 2435
> Office             :+98-0311-793 2430
> Fax No.            :+98-0311-793 2409
> E-mail             :s_jalali_a at yahoo.com
> www                :http://www.ui.ac.ir
> /_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
>
> ---------------------------------
> Groups are talking. We´re listening. Check out the handy changes to
> Yahoo! Groups. _______________________________________________
> 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



Sincerely yours,
S. Jalali.
/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
Saeid Jalali Asadabadi,
Department of Physics, Faculty of Science,
University of Isfahan (UI), Hezar Gerib Avenue,
81744 Isfahan, Iran.
Phones:
Dep. of Phys.      :+98-0311-793 2435
Office             :+98-0311-793 2430
Fax No.            :+98-0311-793 2409
E-mail             :s_jalali_a at yahoo.com
www                :http://www.ui.ac.ir
/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
 __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5513 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20060726/75bcc677/attachment.bin