5
$\begingroup$

I wish to compute the projected density of states for a system by employing hybrid functionals. I am aware that the band structure can be computed using either the Wannier90 approach or using Boltztrap2. Is it possible to modify one of these procedures to obtain the projected DOS, or any other work around to determine the same?

Any inputs would be highly appreciated.

$\endgroup$
5
  • $\begingroup$ I have never tried this but what's the problem of running projwfc.x right after the hybrid scf? $\endgroup$ Commented Apr 5 at 8:28
  • $\begingroup$ Since I'm using ONCV pseudopotentials for my calculations, the code stops saying, "Cannot project on zero atomic wavefunctions" @AbdulMuhaymin $\endgroup$
    – ansonthms
    Commented Apr 5 at 13:06
  • 1
    $\begingroup$ This error does not have anything to do with hybrid functionals. This means that the ONCV pseudopotential you are using do not have <PP_PSWFC> </PP_PSWFC> section or that section is empty. I guess you are using the pseudopotentials from SG15 database. This database has this this problem. There are several workarounds for this but first confirm if this is really the case. Could you find the abovementioned section in your PP file? If yes, then is it empty? $\endgroup$ Commented Apr 5 at 13:46
  • $\begingroup$ Yes, you are correct @AbdulMuhaymin. This is the case. I am using the ONCV pseudos from the SG15 database, and the PP_PSWFC section is empty. $\endgroup$
    – ansonthms
    Commented Apr 6 at 4:21
  • $\begingroup$ You have modified the question but as I mentioned earlier that the problem is not with hybrid functional in your case. It is about the SG15 database, the pseudopotentials that you are using. You cannot get PDOS from them be it hybrid or a regular functional $\endgroup$ Commented Apr 25 at 4:02

1 Answer 1

1
$\begingroup$

As resolved in the comment, this particular error does not have anything to do with the functional being hybrid. This error occurs due to the pseudopotential choice. Not all pseudopotential files contain atomic wavefunctions. In the UPF format (used by Quantum ESPRESSO), the atomic projection information is contained in the <PP_PSWFC> ... </PP_PSWFC> section. In case this section does not exist in the pseudopotential or is empty, then that means one cannot obtain atomic projections using that pseudopotential. So whether it's a hybrid functional or a regular one, it's not possible to run projwfc.x with such pseudopotential.

One such example is the SG15 Optimized Norm-Conserving Vanderbilt Pseudopotentials database. One can always regenerate the pseudopotentials as suggested here. Fortunately, someone already regenerated this set of pseudopotentials that contains the <PP_PSWFC> ... </PP_PSWFC> section and is available here in GitHub.

$\endgroup$
2
  • $\begingroup$ Thank you for the answer @AbdulMuhaymin for the answer. In my case, however, only the ONCV pseudopotentials from the SG15 database gives the correct value of band gap. Is there a work around in which one can obtain the projected dos using this database itself? $\endgroup$
    – ansonthms
    Commented Apr 10 at 4:56
  • $\begingroup$ @ansonthms Not that I know of but you can ask this as a new question. Maybe it is possible since you can generate the same pseudopotential with PP_PSWFC and then copy this section to your existing pseudopotential and run projwfc.x. But I am not sure how good it will be or even if it's gonna work. $\endgroup$ Commented Apr 15 at 12:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .