Projected density of states is greater than total

[alexander_duong]

Hello,

I am running a density of states (DOS) calculation with spin-orbit coupling and with LORBIT=11. When I plot the orbital- and atom-projected DOS, for one of the atoms and for one of its orbitals, the projected DOS is greater than the total DOS in an energy interval of around 0.3-0.4 eV, somewhat close to the Fermi level. I tried doing the equivalent calculation in Quantum Espresso to compare with, and the same issue does not appear. Why is this occurring in VASP? Is this something that could possibly happen due to numerical approximations/errors, or does it signal a deeper underlying problem in my calculations?

Thanks,
Alex

[jonathan_lahnsteiner2]

Dear Alexander Duong,

Please supply your input files according to the vasp forum guidelines https://www.vasp.at/forum/viewtopic.php?t=17928. Otherwise I can not help with this issue.

All the Best Jonathan

[alexander_duong]

Dear Jonathan,

Thanks for your reply. I have attached the files.

Thanks,
Alex

[jonathan_lahnsteiner2]

Dear Alexander Duong,

Sorry for the late response. We are currently looking into this issue. I will keep you updated.

All the best Jonathan

[jonathan_lahnsteiner2]

Dear Alexander Duong,

In vasp it is not expected that the sum over the projected DOS gives the total density of states. The projected density of states is computed within a cutoff radius around each atom in your simulation box. These cutoff spheres around each atom can not fill out your simulation box entirely. You can set the cutoff radius within which the projections are computed with the tag https://www.vasp.at/wiki/index.php/RWIGS. If the tag is not set in the INCAR file then the default value of the POTCAR file will be used. I would recommend to play around with this cutoff radius until you obtain satisfactory results.

All the best Jonathan

[alexander_duong]

Hello Jonathan,

Thanks for your reply. However, I used LORBIT=11 in my calculation, and from what I understand, this means RWIGS is ignored and the projection uses the projector functions that are provided by the PAW method. Could you check?

Thanks,
Alex