[Wien] LDA+U for large U value or EECE for large alpha value

[esmail ghasemi]

Dear Wien2k users,
 
I am working on an anti-ferromagnetic crystal using Wien2k_9.2. 
I have changed the spin directions in case.inst file to set up the antiferromagnetic phase and used runsp_lapw script.
Our LDA and/or GGA  calculation yields zero for the total magnetic moment (:MMTOT = 0) as expected.
The :MMTOT also converged to zero applying LDA+U and/or GGA+U successfully.
We calculated the :MMTOT for several U values and plotted the :MMTOT versus U. We found :MMTO = 0 for small U values.
However, at a larger U value we suddenly observe a jump in :MMTOT, which results in a nonzero :MMTOT in an antiferromagnetism calculations. 
We analyzed the result by monitoring the :MMTOT from first to last iterations and found a well converged result without any serious jump each iterations. This mad us sure about the accuracy of our calculations. 
But, we are not sure how it can be possible to reach a nonzero :MMTOT within our initial case.inst with antiferromagnetic ordering.
Can we conclude that for that large U value ferromagnetic phase is more stable than the antiferromagnetic phase? Or we should conclude that the large U value is not suitable value for our case.
 
Similar experiences were observed by the alpha parameter in the EECE calculations. For small alpha values we obtained :MMTOT = 0, but for at a large alpha value we obtained again a nonzero total magnetic moment.
 
Although the LDA+U  and EECE calculations are not constraint calculations like fixed magnetic moment calculations, but it is hard to expect to obtain ferromagnetic phase from an initial antiferromagnetic setup. Should we change the density matrix elements to change the initial value to reach MMTOT=0? Or we should not consider that U or alpha value as a physical parameter?
 
Any comments will be highly appreciated.
 
Esmaeil Ghasemi


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20100602/72db827b/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 21347 bytes
Desc: not available
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20100602/72db827b/attachment-0001.jpeg>