My Community

I am trying to calculate the DOS for a system which has 3 atoms equivalent to each other by symmetry. The program identifies the correct symmetry at start, while calculating the angular momentum density of states as well as the integrated charge (written at the end of the OUTCAR if RWIGS is given) the program gives them to be inequivalent. The INCAR and POSCAR that I am using is given below. Should some additional flags be given?
----------------------------------------- POSCAR ----------------------------------------------------
System
3.84000000000000
0.9778911174852863 0.0000000000000000 0.0000000000000000
0.0000000000000000 0.9778911174852865 0.0000000000000000
0.0000000000000000 0.0000000000000000 0.9778911174852865
1 1 3
Direct
0.0000000000000000 0.0000000000000000 0.0000000000000000
0.5000000000000000 0.5000000000000000 0.5000000000000000
0.5000000000000000 0.5000000000000000 0.0000000000000000
0.0000000000000000 0.5000000000000000 0.5000000000000000
0.5000000000000000 0.0000000000000000 0.5000000000000000

--------------------------------------- INCAR --------------------------------------------------------
ISMEAR = 0
SIGMA=0.1
PREC=ACC

ISTART=1
EDIFF=0.1E-5

NBANDS=120

EMIN= -20.0
EMAX= 15.0
NPTS=700

RWIGS=0.9 0.9 0.9

AMIX=0.2

the reason of this behaviour is that the
partial (and ONLY the partial ) magnetizations and
charge densities are not symmetrized before they are written to
OUTCAR. (the total charge density is, of course)
Therefore, the difference in the output stems from the difference
in the k-meshes which are generated:
if ISYM is switched off, the irreducible part of the Brillouin
zone is not determined, hence the k-points are spread all over
the 1st BZ, and the partial densities look symmetric, as they in fact are.
if ISYM is switched on, only the irreducible part of the BZ is filled
with k-points, but for the patrial charges no a-posteriori symmetrisation
is done (as should be and is in fact done for the charge densities).
Therefore, these partial quantities may differ slightly,
even for symmetry-equivalent atoms in a unit cell.

I would only add that the projected quantities can be very different (not just slightly). For a simple example, see http://cst-www.nrl.navy.mil/~erwin/vasptips/#pdos.

I have LORBIT=1 tag in the INCAR file, therefore I see the site projected DOS for each atom in the DOSCAR file. However I have 10 columns data on each atom. I know the first columm is energy, and if column 2-4 is s p d DOS, what are the rest 6 columns? (I have a spin unpolarized calculation, therefore there should be 4 columns in total, isn't it?)

columns 2-9 give the l,m decomposed local partial DOS, ordered
s, p_y, p_z, p_x, d_xy, d_yz, d_z2-r2, d_xz, d_x2-y2,
i.e. (l,-m, ..... l,0, ..... l,+m)