[Pw_forum] LDA+U character
Matteo Cococcioni
matteo at umn.edu
Tue Jul 6 21:49:17 CEST 2010
Hi Ali,
it's difficult to predict what is going to happen during the
calculation. Maybe with a finite U the system prefers to occupy a state
that was originally empty
and viceversa. or this could be a local minimum. There are few things
you could try:
1) start the LDA+U run from an LDA one (e.g. reading potential and
wavefunctions) instead of beginning from scratch
2) try to constrain the eigenvalue of the occupation matrices to be 1 or
0 or equal to the value you want and see if the system "likes" the solution
3) as in 1 or 2 keeping occupation fixed for a few iterations and see if
the system readjusts tself in the solution you suggest.
Of course at the end you will need to compare the total energies of all
these solutions.
unfortunately with lda+U it's not uncommon that the system gets trapped
in local minima of the energy.
good luck!
Matteo
ali kazempour wrote:
> Dear Matteo
> Thanks many for your reply. But my special case is for the defect
> states (has d-character )in TiO2 that is in resonance with conduction
> band and is filled with 2 electrons. When I increased the U from 1 to
> 4 I saw that the defect state bieng pushed up instead lowering of this
> state as we expected from theory of LDA+U. Is it true? Moreover for
> another defect state that is empty and is above conduction band I saw
> it being peshed down in contrast with LDA+U expection. But when I
> perform GW@ GGA+U I see that the behavior is reversed in which empty
> states being pushed up while filled state is pushing down. what does
> it mean? is it rational?
> thanks a lot
>
> Ali Kazempour
>
>
>
> Fritz-Haber-Institut fax : ++49-30-8413 4701
> der Max-Planck-Gesellschaft
> Faradayweg 4-6 e-mail: kazempou at fhi-berlin.mpg.de
> D-14 195 Berlin-Dahlem / German
>
> --- On *Sun, 7/4/10, Matteo Cococcioni /<matteo at umn.edu>/* wrote:
>
>
> From: Matteo Cococcioni <matteo at umn.edu>
> Subject: Re: [Pw_forum] LDA+U character
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Date: Sunday, July 4, 2010, 8:29 PM
>
>
> dear Ali,
>
> I will try to answer your questions
>
>
>
> ali kazempour wrote:
> > Dear All
> > I have some questions about LDA+U nature. Does U treate like a
> > perturbation( I know that GW( many body peturbation method) is
> > modified version of LDA+U? if yes why we are allowed to take
> > U=0,1,2,........ up to large values? and why we solve system
> > self-consistent?
> >
>
> no, U is not treated as a perturbation. it is more a correction to
> the
> "standard" DFT energy functional. you can check the relationship
> between
> +U and GW in the following
> reference (probably there are many more):
>
> Anisimov et al, Journal of Physics: Condensed Matter 9, 767 (1997)
>
>
> > Does U comute with KS-hamiltonian in every U values? I mean
> does the
> > order of states remains intact with respect to LDA or they can
> change?
> > and final question:
> >
>
> For sure the +U correction to the potential commutes with the
> translation and other symmetries of the crystal.
> I think it doesn't commute with the rest of the Hamiltonian (the
> unperturbed DFT one). So if you think to switch on the U from a
> U=0 ground state, unless the KS states are already bloch sums of
> atomic
> orbitals (that would be a very unlikely accident) the +U correction
> introduces interactions between them if allowed by symmetry and
> you are
> going to get a non trivial mixing of them besides a shift in the
> eigenvalue. Probably the most significant modifications are going to
> take place for states around the Fermi level that may get more
> localized.
> The ones that are deeper in energy are probably closer to be
> atomic-like
> already and they will, at most, get a shift in the corresponding
> eigenvalues.
>
> > What is the effect of U on
> > 1-empty states in conduction band (with d-character)
> >
>
> pushed up
>
> > 2-half filled states in conduction band(with d-character)
> >
>
> depends: see above
>
> > 3-filled- states in conduction band (with d-character)
> >
>
> pushed down
>
> > Thanks a lot
> >
> >
> > Ali Kazempour
> >
> >
>
>
> hope this helps.
>
> Matteo
>
>
> >
> > Fritz-Haber-Institut fax : ++49-30-8413 4701
> > der Max-Planck-Gesellschaft
> > Faradayweg 4-6 e-mail: kazempou at fhi-berlin.mpg.de
> <http://us.mc1144.mail.yahoo.com/mc/compose?to=kazempou@fhi-berlin.mpg.de>
> > D-14 195 Berlin-Dahlem / German
> >
> >
> >
> ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> <http://us.mc1144.mail.yahoo.com/mc/compose?to=Pw_forum@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://us.mc1144.mail.yahoo.com/mc/compose?to=Pw_forum@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
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
More information about the Pw_forum
mailing list