Reference direction - lm decomposed DOS

[forsdan]

When determining the lm-decomposed density of states for collinear calculations with the lm-ordering:

s, p_y, p_z, p_x, d_xy, d_yz, d_z2-r2, d_xz, d_x2-y2.

does the z-component (e.g. p_z) correspond to the orbital orientation along the carteesian vector (0, 0, 1) or which reference direction is considered? The interpretation of the projected charge densities depend on if the z-axis corresponds to (0, 0, 1) or (1, 0, 0).

Best regards,
/Dan Fors


EDITED:
Resolved the issue by looking at a diatomic molecule and concluded the directions judging from the symmetry. Thank you all anyway.


<span class='smallblacktext'>[ Edited ]</span>

[admin]

dear Dan Fors,
you are correct, the principal axis is taken to be the z-axis of the lattice (box) in VASP, it is not aligned along the binding axis of the molecule (as you might suppose for a molecule which is not oriented along z)
actually, the l,m splitting is (l,-m),... l,0, (l,m), hence the correspondence to the labels px...of course only holds for certain lattice types