[Pw_forum] U for Oxygen

Thu Aug 6 11:06:20 CEST 2009

Dear Matteo
In your tutorial about LDA_U, you mentioned that the perturbation should be isolated. and  said that we normally use supercell of increasing size up to when convergence of is reached. I don't understand how do you increase the size of the cell and how do you compute the U in larger supercell.? I mean how do you apply perturbation to bigger size or how to extrapolate the U to converge?
Another question according to prior email for computation of U by common method(not linear response) is : you mentioned that espresso cann't directly change the occupation of special orbital by hand, but it is possible to do by using constraint on total number of electron(total occupation) indirectly . Do you mean that( for example for Ti atom that has 2 electron in d orbital) ,what can i do is change the of number of nelup and neldw  without change in nelec? or I have to do another switch? Would you please tell me in more details.
Thank you very much for your aids
Ali Kazempour

Physics department, Isfahan University of Technology

84156 Isfahan, Iran.            Tel-1:  +98 311 391 3733

Fax: +98 311 391 2376      Tel-2:  +98 311 391 2375

--- On Tue, 8/4/09, Matteo Cococcioni <matteo at umn.edu> wrote:

From: Matteo Cococcioni <matteo at umn.edu>
Subject: Re: [Pw_forum] U for Oxygen
To: "PWSCF Forum" <pw_forum at pwscf.org>
Date: Tuesday, August 4, 2009, 9:52 AM

this is what comes out from the calculation (U(O) > U(Ni)). whether it's 
reasonable to apply U on O or not, I'm not sure.

I think the U shouldn't be used on states that are completely occupied.

Matteo

ali kazempour wrote:
> Dear Matteo
> This means that hubbard parameter for oxygen is large (grater than 
> Ni). Is it correct or originate from another challenge?
> thanks
>
> Ali Kazempour
> Physics department, Isfahan University of Technology
> 84156 Isfahan, Iran. Tel-1: +98 311 391 3733
> Fax: +98 311 391 2376 Tel-2: +98 311 391 2375
>
> --- On *Tue, 8/4/09, Matteo Cococcioni /<matteo at umn.edu>/* wrote:
>
>
>     From: Matteo Cococcioni <matteo at umn.edu>
>     Subject: Re: [Pw_forum] U for Oxygen
>     To: "PWSCF Forum" <pw_forum at pwscf.org>
>     Date: Tuesday, August 4, 2009, 6:59 AM
>
>
>
>     Dear Ali,
>
>
>
>     ali kazempour wrote:
>     > Dear Matteo
>     > I used your comment and followed   your example for NiO and I
>     succeed 
>     > to get U for TiO2. But in file Ur16.out I see three column .
>     >
>
>     I don't have any three column section in any of my files. so if
>     you want
>     me to help you you have to report at least a section of your output
>     file. This is a good habit actually in consideration of the fact that
>     this is not a private conversation (I would have asked you in any
>     case
>     as I don't remember by heart the format of any file) between me
>     and you
>     but a message to the forum and you want other people to understand
>     what you are saying.
>
>     > The first is number of atom in supercell , second is U value for Ni
>     > atom and I don,t know  the third? does the third belong to  U value
>     > for Oxygen or not? Because the third column kas big value?
>     >
>     >
>     > another question is for isolated atom Hubbard parameter. What do
>     you
>     > expect about the value of U for Isolated atom? Is this big value
>     > (greater than 10 eV) or  it would be tiny value?
>     >
>
>     I think it will be meaningless for the reasons I explained in my last
>     email. To obtain a meaningful number for the isolated atom you
>     probably
>     have to abandon the idea to use a perturbation in the potential and
>     change instead the number of electrons directly. This means you will
>     have to do calculations on atoms having fractional charges which
>     requires additional corrections if you use a code designed for
>     periodic
>     systems (crystals). However I have never done this isolated atom
>     calculation of U before so I don't have more details about it.
>     I see (from a private email) you are thinking of computing U for the
>     isolated atom and to obtain a value for the crystal just by
>     dividing by
>     the epsilon_inft. the method is proposed in
>
>     http://link.aps.org/doi/10.1103/PhysRevB.74.045202
>
>     (which I haven't re-read yet).
>
>     I see a couple of problems:
>
>     1) (not important for ZnO). suppose you have Fe3O4 with Fe2+ and
>     Fe3+.
>     these Fe ions likely have different U. how do you obtain
>     a different number from the atomic code? would you do the calculation
>     for isolated Fe2+ and Fe3+?
>
>     2) how do you get epsilon? since it's a linear response property I
>     don't
>     think its calculation is any cheaper than the direct calculation of U.
>
>     regards,
>
>     Matteo
>
>     > should we choose for running the script for isolated atom only
>     gamma
>     > point or we have to choose a fine mesh?
>     > thanks a lot
>     >
>     >
>     > Ali Kazempour
>     > Physics department, Isfahan University of Technology
>     > 84156 Isfahan, Iran. Tel-1: +98 311 391 3733
>     > Fax: +98 311 391 2376 Tel-2: +98 311 391 2375
>     >
>     > --- On *Sun, 8/2/09, Matteo Cococcioni /<matteo at umn.edu
>     </mc/compose?to=matteo at umn.edu>>/* wrote:
>     >
>     >
>     >     From: Matteo Cococcioni <matteo at umn.edu
>     </mc/compose?to=matteo at umn.edu>>
>     >     Subject: Re: [Pw_forum] occupation
>     >     To: "PWSCF Forum" <pw_forum at pwscf.org
>     </mc/compose?to=pw_forum at pwscf.org>>
>     >     Date: Sunday, August 2, 2009, 8:31 AM
>     >
>     >
>     >     Dear Ali,
>     >
>     >     1, -1, 0 are just spin indexes to tell the code what is the
>     relative
>     >     magnetization of the atoms (parallel, antiparallel, or none). if
>     >     your system is non magnetic put all 0. or set magn = .false.
>     and put
>     >     nothing.
>     >     you should look at the examples on Fe and Si as well and
>     learn from
>     >     there by comparison.
>     >
>     >     Matteo
>     >
>     >
>     >
>
>
>     -- 
>     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>     Matteo Cococcioni
>     Department of Chemical Engineering and Materials Science,
>     University of Minnesota
>     421 Washington Av. SE
>     Minneapolis, MN 55455
>     Tel. +1 612 624 9056    Fax +1 612 626 7246
>     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
>     _______________________________________________
>     Pw_forum mailing list
>     Pw_forum at pwscf.org </mc/compose?to=Pw_forum at pwscf.org>
>     http://www.democritos.it/mailman/listinfo/pw_forum
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>   

-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Matteo Cococcioni
Department of Chemical Engineering and Materials Science,
University of Minnesota
421 Washington Av. SE
Minneapolis, MN 55455
Tel. +1 612 624 9056    Fax +1 612 626 7246
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org
http://www.democritos.it/mailman/listinfo/pw_forum

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20090806/7884c208/attachment.html>