Phase factors in PROCAR

[georg]

Dear fellow VASPers,

When running a calculation with LORBIT=12 (PAWs), the PROCAR file that I get is essentially composed out of two blocks for each state: one with one number per atom (same as with LORBIT=11) and one with two numbers per atom.

Now, about the latter: What are these two numbers per atom? Real and Imaginary part? Angle and absolut value? What units? How (if at all) do they relate to the single number per atom in the first block?

I would like to use these numbers to project states from isolated molecules onto adsorbate states. Can anyone outline to me a VASP-specific procedure for doing just that?

Thanks for your help

Georg

[lahaye]

Hi all,

I have searched this forum and found this question a couple
of times, but no replies with answers were present.

I too am curious about the numbers in the PROCAR file.
The information in the VASP Guide is not sufficient.

The top part of my PROCAR file has this (I've trimmed it a bit):

-------------------------------------------------------
PROCAR lm decomposed + phase
# of k-points: 365 # of bands: 10 # of ions: 2

k-point 1 : 0.0 0.0 0.0 weight = 0.00137174
band 1 # energy -20.739 # occ. 2.0

ion s py pz px dxy (...) tot
1 0.321 0.000 0.000 0.000 (...) 0.321
2 0.324 0.000 0.000 0.000 (...) 0.324
tot 0.645 0.000 0.000 0.000 (...) 0.645

ion s py pz px dxy (...)
1 0.008 0.000 0.000 0.000 (...)
1 1.425 0.000 0.000 0.000 (...)
2 0.008 0.000 0.000 0.000 (...)
2 1.431 0.000 0.000 0.000 (...)
(...)
-------------------------------------------------------

I assume that the first table has the spd-projections onto the
sites of the ions. What is the second table? If this is the phase,
then I'm confused: I use LORBIT=11, which should not produce
phase factors (LORBIT=12 should do that, at least according
the VASP GUide).

Anyway, what do all these numbers mean, and how are the two
tables related to each other.

Thank you very much,
Rob.

[Hugo Strand]

I doubt that this will help you guys posting this five years ago, but maybe for new people like me this will help.

From what I understand of the Vasp code and the subroutine SPHPRO_FAST (sphpro.F) in particular,

the fist table of the spd occupancies rho_lm for each ion is calculated from

rho_lm = Sum_nk f_nk Sum_ee' (PhiPS_nk|p_lme) Q_ee' (p_lme'|PhiPS_nk)

where f_nk is the fermi occupancy for each k-point k and band n, |PhiPS_nk) is the Pseudo-Kohn-Sham wave function, |p_lme) is the PAW projector for the augmentation channel with atomic quantum numbers l and m, with the additional twist of e indexing the two reference energies used to construct the two All Electron (AE) local wave functons PsiAE_lme(r) for each lm channel. See Kresse J. Phys. Condens Matter v6, 8245 (1994).

Q_ee' is simply the radial overlap,
int dr PsiAE_lme(r) PsiAE_lme'(r)

that corrects for the non-normalized local pseudo wave functions PsiPS_lm(r).

So what is this good for? Well the second table in PROCAR is just the real and imaginary parts of the PS KS wavefunction and projetor overlap,

2*(p_lm,e=1|PhiPS_nk)

Where for some reason only the projector part from the first reference energy, the atomic eigen energy I guess, e=1 is used. Multiplied by a factor of two instead of including the second reference energy projector e=2, probably for some good reason.

NOTE! I'm not a VASP developer so the above information may be totaly wrong. But maybe this can inspire ADMIN to enlighten us.

Best whishes, Hugo