LDA+U CoO insulator problem
Posted: Tue Jul 01, 2008 3:38 pm
Dear Forum,
We have been using the Dudarev implementation of the LDA+U on CoO bulk, which should be an insulator.
We cannot seem to target the correct ground-state. A publication by Wdowik and Parlinski (PRB 75, 104306 (2007)) shows that VASP is able to do it.
Another publication describes the problem in being in the initial spin-density distribution.
There is very little info in the Wdowik and Parlinski paper on how they perturbed the initial spin-density distribution so that they could get to the correct lower energy state. We suspect that it is to do with the charge and spin mixing parameters in VASP (AMIX and AMIX_MAG), which they also vaguely refer to.
Would that be a reasonable guess (?)
We are trying a linear mixing scheme and are changing the AMIX and AMIX_MAG values.
Do you have any further advice on this ?
We would very much appreciate your insight.
Thank you!
We have been using the Dudarev implementation of the LDA+U on CoO bulk, which should be an insulator.
We cannot seem to target the correct ground-state. A publication by Wdowik and Parlinski (PRB 75, 104306 (2007)) shows that VASP is able to do it.
Another publication describes the problem in being in the initial spin-density distribution.
There is very little info in the Wdowik and Parlinski paper on how they perturbed the initial spin-density distribution so that they could get to the correct lower energy state. We suspect that it is to do with the charge and spin mixing parameters in VASP (AMIX and AMIX_MAG), which they also vaguely refer to.
Would that be a reasonable guess (?)
We are trying a linear mixing scheme and are changing the AMIX and AMIX_MAG values.
Do you have any further advice on this ?
We would very much appreciate your insight.
Thank you!