Not getting degenrate m-resolved DOS for t2g and eg orbitals of SrVO3 in real spherical harmonic basis

Hello users,
I am trying to get m-resolved partial DOS using real spherical harmonics of Vandium in SrVO3. For V, I expect to get m-resolved partial DOS of d sub-shell to be degenerate in t2g and eg orbital. But I am not getting any degeneracy. I am using prtdosm=2 for real spherical harmonics basis(Input file is written below). Please help me in understanding what I am doing wrong.
Trishu Verma

prtdos 3
prtdosm 2
pawprtdos 2
natsph 1
iatsph 1
prtvol 4
pawprtvol 3


nstep 30
ecut 20
pawecutdg 60
tolvrs 1.0d-10
nband 30
occopt 3 tsmear 1200 K


natom 5 ntypat 3 typat 1 2 3 3 3
znucl 23.0 38.0 8.0
xred 0.00 0.00 0.00
0.50 0.50 0.50
0.50 0.00 0.00
0.00 0.50 0.00
0.00 0.00 0.50
acell 3*7.2605
rprim 1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0

== Points k and symetries

kptopt 1
ngkpt 8 8 8


pp_dirpath “./”
pseudos “V.GGA_PBE-JTH.xml, Sr.GGA_PBE-JTH.xml, O.GGA_PBE-JTH.xml” # Name and location of the pseudopotentials

Dear Trishu,

This calculation should give you degenerate t2gs and egs orbitals. Can you show me the resulting figure you get for the DOS so I can try to reproduce?


Dear Olivier,
As I am new user of the forum, I could not directly attach the resulting figure of DOS. Therefore, I uploaded the same (along with DOS raw data I got) on google drive and link is here : For Oliver - Google Drive


Dear Trishu,

I will need some additional information from you, in particular you output file. At first glance, it seems like the eg vs t2g orbitals are in the right places of the spectrum, but the degeneracies inside of these are wrong. So I think you are not converged in kpoints.

In DOS calculations, you need pretty big amounts of kpoints. You should still find the right symmetries, but here you do not specify nshiftk and shiftk, so by default you will have shiftk 0.5 0.5 0.5 which can break the symmetries. So you can try to fix shiftk 0 0 0 and see.

Another thing you can do is plot the band structure. Is it right compared with the literature, or simply compared with the ABINIT tutorial? What about the fatband calculations? I’d like to see that before investigating further.


Dear Olivier,
As per your suggestion I did calculation at pretty k-points,specifically by using ngkpt 20x20x20. I also try using shiftk 0 0 0. But still there is no significant improvement regarding degeneracy.
I also plotted band structure and fatbands. They do look same as ABINIT tutorial.
For your reference I am attaching input, (after making changes you suggested), output, log, m-resolved DOS for l=2, band structure and fatbands. Google drive link: For_Olivier - Google Drive


Hi Trishu,

Thanks for your patience. I think the reason why the degeneracy between your t2g orbitals is not enforced is because you compute your DOS using symmetries. The orbitals transform like non-trivial irreducible representations of the space group, so to recover the degenerate DOS, you need so sample the whole Brillouin Zone. Try setting kptopt 3 for the DOS. Because this is a post-processing step, I suggest you do your regular convergence of the ground state with smaller k-grid and using the full symmetries, and then compute the DOS in the following dataset, with a finer k-grid and no symmetries.

Let me know how that goes,