[Wien] symmetry reduction and density matrix inverted calculation

[Rocquefelte]

Dear Colleague,

Le 28/04/2011 04:59, 王静 a écrit :
> Dear,
>   Recently I have calculated a system in cubic crystal structure (No.225, Fm-3m). The compound contains two magnetic ions A (0 0 0) and B (0.5 0.5 0.5). The experimental reported magnetic structure is each of the magnetic moments of A and B orders in the type I arrangement while the moment of A and B are antiparallel with each other. 
> My first question is: when I design the magnetic structure as the experiment reported, the crystal structure reduced to be tetragonal (No.123, P4mmm). Will this symmetry reduction in calculation influence the results? 

No, except that you impose a given symetry to the spin density. Thus the
spin density will relax in this symmetry during the SCF cycle.

> Since my A ion is rare earth elements and the calculation from VASP (use the P1 symmetry) gets no magnetic moment for it. The earlier literatures have reported the magnetic moments of rear earth ions may be closely relative to the local crystal filed. Therefore I wonder whether the magnetic moment for the rare earth in my case from wien is reliable. 

You should differenciate the local symetry (local crystal field) and the
space group. The local field around A is always the same, whatever is
the space group you are using to describe the system (Fm-3m, P4mmm or P1).
So the magnetic moment will not be affected by the choice of space
group, except if it leads to a different magnetic order than the
experimental one.

Concerning your VASP calculation, to have a magnetic moment on A you
need to use a adequate pseudopotential (in which the f states are in the
valence) and GGA+U. It should lead to similar results than WIEN2k.

> My second question is: The magnetic moment of  A ions converged to be parallel from GGA calculation (The spins have been inverted in case.inst file). To get the antiparallel magnetic moment of A ions. I do two GGA+U calculations. For the first one I do it based on the result from GGA calculation. The magnetic moment of A ions from this GGA+U calculation is sure to be parallel. For the other one I also use the GGA results. The difference is I use the case.dmat files from the first GGA+U calculation but invert the density matrix of the A ion whose spin has been inverted in case.inst file. Then I use “-orbc” option. I did get antiparallel magnetic moment of A ions in the second calculation and the total energy is much lower than the first GGA+U calculation (about 3 Ry/f.u.). It that OK, or any problem? 

I would restart the calculation after using -orb to finalize the
calculation.
Just for information, did you simply try to invert the magnetic moments
in the case.inst file and directly run GGA+U?


Regards

Xavier

> Any words will be appreciated!
> Best regards!
>
>
>
> 　　　　　　　　wangjingjing at ciac.jl.cn
> 　　　　　　　　　　2011-04-28
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien