[Wien] lda+u calculation

Fecher, Gerhard fecher at uni-mainz.de
Fri Feb 17 09:03:46 CET 2012 

Try the following, make the NiO non-magnetic before starting the LDA+U calculation (with U from the very beginning)
most probably the solution will stay non-magnetic (indeed also if you do that without U for Fe as an example).

For LSDA you use an additional term sigma B in the potential, if this is once Zero, you don't have a magnetic solution anymore
and it will not come back if you use U.

For LSDA+U the U will split certain states (in youre case  Ni d) and keeps the magnetic splitting during all SCF cycles
(and thus it keeps the B term non-Zero but only if it was not Zero, the states have to be split already)
The result is: with LDA+U you can make rather everything "magnetic".

This is in very short what you observed.

Sometimes, the calculations run into a "wrong" state just because the initial density when starting the calculation is a "bad" choice.
In that cases it may help to start with a fixed moment calculation (option fsm) until the density is improved. This will haphazardlyx shift the
spin up and dn densities but may lead to improved densities. If you continue without fsm then the calculation should converge into the "correct" magnetic state.
However, this are cases where one realy has to think about what one is doing. 



Ciao
Gerhard


====================================
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
________________________________________
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at zeus.theochem.tuwien.ac.at]" im Auftrag von "Yundi Quan [quan at ms.physics.ucdavis.edu]
Gesendet: Freitag, 17. Februar 2012 08:31
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] lda+u calculation

Hi,
I'm calculating NiO which is a canonical example of using LDA+U. Your speculation is right. The non-magnetic case is the one where I started from the converged LDA calculation. Is there a reason for that?



Yundi



On Thu, Feb 16, 2012 at 11:27 PM, Fecher, Gerhard <fecher at uni-mainz.de<mailto:fecher at uni-mainz.de>> wrote:
I guess the non-magnetic solution is the one where you started from the converged LDA calculation
and the magnetic one is the one where you started with LDA+U from scratch.

You did not tell much about the system you are calculating and why you used LDA+U
without knowing this, any thing for explanation just stays speculation.

Ciao
Gerhard



====================================
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
________________________________________
Von: wien-bounces at zeus.theochem.tuwien.ac.at<mailto:wien-bounces at zeus.theochem.tuwien.ac.at> [wien-bounces at zeus.theochem.tuwien.ac.at<mailto:wien-bounces at zeus.theochem.tuwien.ac.at>]&quot; im Auftrag von &quot;Yundi Quan [quanyundi at gmail.com<mailto:quanyundi at gmail.com>]
Gesendet: Freitag, 17. Februar 2012 03:34
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] lda+u calculation

Both cases are very converged. One gives magnetic solution. The other gives non-magnetic solution. The magnetic solution is slightly lower in energy.




On Thu, Feb 16, 2012 at 4:48 PM, Laurence Marks <L-marks at northwestern.edu<mailto:L-marks at northwestern.edu><mailto:L-marks at northwestern.edu<mailto:L-marks at northwestern.edu>>> wrote:
Maybe, or maybe not! If both cases are well converged, that means that
for your approximations there are two locally stable spin states.
Maybe this is real, maybe not, that is hard to say.

2012/2/16 Yundi Quan <quan at ms.physics.ucdavis.edu<mailto:quan at ms.physics.ucdavis.edu><mailto:quan at ms.physics.ucdavis.edu<mailto:quan at ms.physics.ucdavis.edu>>>:
> You are right. The spin state is different in two cases. Applying U after a
> scf calculation seems to tend to reduce the magnetic moments.
>
>
>
>
>
>
>
>
> On Thu, Feb 16, 2012 at 3:44 PM, Laurence Marks <L-marks at northwestern.edu<mailto:L-marks at northwestern.edu><mailto:L-marks at northwestern.edu<mailto:L-marks at northwestern.edu>>>
> wrote:
>>
>> If all other things are the same, the lowest energy is correct
>> (variational principle). Probably the spin state is different in the
>> two cases.
>>
>> But...please ensure that you really have both converged. Assuming that
>> you have a recent version, do a "grep -e :MV case.scf" (with case
>> replaced as appropriate). The % value should be small for true
>> convergence; if it is not you do not have real convergence
>> (technically it is called a "trap").
>>
>> 2012/2/16 Yundi Quan <quanyundi at gmail.com<mailto:quanyundi at gmail.com><mailto:quanyundi at gmail.com<mailto:quanyundi at gmail.com>>>:
>> > Dear Sir/Madam,
>> > When using LSDA+U, it is recommended to first do an LSDA caculation and
>> > then
>> > turn on U. I first did an LSDA calculation to convergence. Then, I
>> > turned on
>> > U. I also did a calculation by turning on U from the very beginning. I
>> > compared the two results, and found that the later is lower in energy.
>> > Which
>> > one is correct?
>> >
>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > Wien mailing list
>> > Wien at zeus.theochem.tuwien.ac.at<mailto:Wien at zeus.theochem.tuwien.ac.at><mailto:Wien at zeus.theochem.tuwien.ac.at<mailto:Wien at zeus.theochem.tuwien.ac.at>>
>> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>> >
>>
>>
>>
>> --
>> Professor Laurence Marks
>> Department of Materials Science and Engineering
>> Northwestern University
>> www.numis.northwestern.edu<http://www.numis.northwestern.edu><http://www.numis.northwestern.edu> 1-847-491-3996<tel:1-847-491-3996><tel:1-847-491-3996<tel:1-847-491-3996>>
>> "Research is to see what everybody else has seen, and to think what
>> nobody else has thought"
>> Albert Szent-Gyorgi
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at<mailto:Wien at zeus.theochem.tuwien.ac.at><mailto:Wien at zeus.theochem.tuwien.ac.at<mailto: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<mailto:Wien at zeus.theochem.tuwien.ac.at><mailto:Wien at zeus.theochem.tuwien.ac.at<mailto:Wien at zeus.theochem.tuwien.ac.at>>
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>



--
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu<http://www.numis.northwestern.edu><http://www.numis.northwestern.edu> 1-847-491-3996<tel:1-847-491-3996><tel:1-847-491-3996<tel:1-847-491-3996>>
"Research is to see what everybody else has seen, and to think what
nobody else has thought"
Albert Szent-Gyorgi
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at<mailto:Wien at zeus.theochem.tuwien.ac.at><mailto:Wien at zeus.theochem.tuwien.ac.at<mailto: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<mailto:Wien at zeus.theochem.tuwien.ac.at>
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

More information about the Wien mailing list