[Wien] Questions about "-orbc" calculations

[Hongbin Zhang]

Dear Prof. Blaha,

Thank you for your reply!

I suppose by "assumed magnetic order and also on the direction of the
magnetization" you mean equivalent magnetic ordering/magnetization
directions for a given structure. But I do not see the reason why this
is relevant. For instance, the system I am considering now is Sr2IrO4,
where the magnetic structure has determined experimentally and fixed
due to the rotation of IrO6 octahedra. Yes there are many different
ways to chose the unit cell, and I think once the structure is chosen,
the magnetization configuration is chosen too. Maybe I should try to
force the magnetization direction to be aligned along different
directions to see whether it will converge (to a "correct" solution).

Yes, with large U I could get convergence, but the physics is different
from expected.

Regards,

Hongbin
On Fri, 8 Mar 2013, Peter Blaha wrote:

> I have done once some test calculations on some iridates.
>
> As far as I remember, everything depends on the assumed magnetic order
> and also on the direction of the magnetization (with SO). Some magnetic 
> structures are easy to converge, others will not (or very hard) converge.
>
> With LDA+U and -orbc you can often enforce a particular magnetic/orbital 
> state, but of course not always. If the system feels it is even with an 
> external constrained potential not favorable, it won't go into this state, 
> (unless you specify a really large U value ?)
>
> And remember, LDA+U was made for real insulators, for "correlated metals" 
> only DMFT may help (wien2wannier + DMFT, see eg. unsupported software).
>
>
>
> On 03/06/2013 02:55 PM, Hongbin Zhang wrote:
>> 
>> Dear Prof. Blaha and wien users,
>> 
>> Recently I am trying to do calculations on J=1/2 insulation iridates using
>> LDA+U+SOC as implemented in WIEN2k. However, I found that it is very hard
>> to converge to the desired insulating states with U smaller than 3 eV
>> (J=0). Moreover, the ratio of spin and orbital moments is one if a large
>> U, for instance, U=4 eV is used. This deviates also from what is expected
>> that the ratio between spin and orbital moments should be 1/2.
>> 
>> So my first questionsis , does any of you have ever done such calculations
>> on J=1/2 iridates? I could not find any calculations done by wien2k in the
>> literature. Is there any pitfalls that it is not doable?
>> 
>> Since I could not converge it within normal self-consistent routines, I
>> tried to make some density matrices up with desired spin/orbital moments
>> and also occupation numbers. Then I do x orb -up/dn and afterwards self-
>> consistent calculations with -orbc -so. After it converges, the orbital
>> and spin moments are totally different from what is expected/designed.
>> And also the density matrices in the case.scfdmup are quite different from
>> my inputs.
>> 
>> Here is my second question that how can one enforce the code to
>> converge to a desired density matrix? Or one can only generate a rigid
>> orbital potential and hope that the system will converge to the desired
>> one?
>> 
>> Any suggestion is welcome and appreciated!
>> 
>> Regards,
>> 
>> Hongbin
>> 
>> 
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
> -- 
>
>                                      P.Blaha
> --------------------------------------------------------------------------
> Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
> Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
> Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
> --------------------------------------------------------------------------
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>