VASP has a tag NEDOS, which helps specify the # of points on which the DoS is evaluated. Other than that, we have EMIN and EMAX, which help decide the range of energy in which DoS is evaluated at NEDOS evenly spaced grid-points. This procedure misses the exact fermi energy point, at least in all of my calculations. Is there a way to make the code explicitly calculate DoS at the Fermi energy? I wish to avoid interpolation due to certain concerns as relayed here.


1 Answer 1


Let me suggest a simple answer, although I am not convinced you will get anything sufficently different for the fermi level than just using a converged grid.

  • Consider a range of energies such as 10 eV.
  • Set EMIN=Fermi-5 and EMAX=Fermi+5
  • Set NEDOS=Anything Odd

As a rule, you should always hit the middle point doing this and the middle point is the Fermi level since you defined it that way. This should work in essentially any DFT calculator.

  • $\begingroup$ +1 for the quick answer, we haven't had a lot of hot network questions in a while! I've just edited the typesetting of DFT in case that's what you meant: You can revert the edit otherwise. $\endgroup$ Oct 21, 2020 at 12:17
  • 1
    $\begingroup$ There are lots of more optimized grids you could use as well to hit the fermi level, but this was the quickest and easiest to remember solution I could think of. Let me know if this makes any actual difference as I am curious. $\endgroup$ Oct 21, 2020 at 14:02
  • $\begingroup$ @TristanMaxson, got the DoS at $E_F$. Other values match with the ones from the calculation without this workaround. The integrated DoS matches exactly. I was surprised because I expected to see the value of integrated DoS start from the DoS at EMIN, but it included all the DoS values. $\endgroup$ Oct 21, 2020 at 21:40
  • $\begingroup$ Maybe at this point you can answer your own other question if you have an idea of how all of this makes a difference. $\endgroup$ Oct 22, 2020 at 2:04

You must log in to answer this question.

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