I am calculating the site projected DOS of SrNbO3 in the ideal perovskite geometry. The program recognises the symmetry as O-h. The K-point grid is a simple Monkhorst-Pack grid with the same number of points in all directions. NSPIN=1. As described in the tutorials, I first do a normal SCF with ICHARG=1, followed by a DOS calculation with, the following parameters:
ALGO = none
ICHARG = 11
RWIGS 1.13 0.78 1.4
LORBIT = 11
NPAR = 1
NEDOS = 602 # Increasing or reducing this doesn't really change things
LMAXMIX = 4
I am using the PBE PAW potentials (Sr_sv, Nb_pv, O; also tried O-h). The site-projected DOS for the three oxygens differs quite a bit. The band edges are the same, but one of them has most of its energy states at the lower end of the conduction band, the others have peaks higher up.
The "vasptips" site recommends just to average over these atoms, but I'm afraid this might only hide a problem.
I am using VASP 4.6. If I had money....
