4
$\begingroup$

After I run an SCF calculation with PBE0, I want to use calculation= 'nscf'/ (‘bands’) with PBE0, but this message occurs:

Error in routine setup (1):
     hybrid XC not allowed in non-scf calculations

Does any one know how to fix it? Or how can I calculate the hybrid function band structure and DOS in the right way

$\endgroup$

1 Answer 1

3
$\begingroup$

There is no problem with your input (if that's the only error). Quantum ESPRESSO doesn't support nscf calculation with hybrid functional. But if your goal is to obtain the DOS and band structure only, then I think it's possible to obtain them without nscf calculation.

So, what we usually do:

  • (step1) a calculation='scf' with rough k-grid (pw.x)
  • (step2) a calculation='nscf' with fine k-grid and occupations='tetrahedra' (pw.x)
  • (step3) post-processing to obtain the density of states (dos.x)
  • (step4) a calculation='bands' by not setting occupations='tetrahedra' and choosing the K_POINTS as crystal_b or tpiba_b according to our crystal's BZ (pw.x)
  • (step5) post-processing to obtain the band structure (bands.x)

Instead, for hybrid functional, consider the following path:

For DOS:

  • (step1) remains the same as before but instead of a rough grid, use the densest k-point grid you can afford. Try occupations='tetrahedra' if that works. There is a caution mentioned in QE's input description: BEWARE: only grids having the full symmetry of the crystal work with tetrahedra. Some grids with offset may not work.
  • (step2) post-processing to obtain the density of states (dos.x), i.e., we are skipping previous (step2). Or if you want the projected DOS, you can also do projwfc.x post-processing right after the scf calculation. Skipping nscf will not create a problem here since we already used a denser grid in our scf calculation in the previous step.

For Band structure:

We cannot perform calculation='bands' which is also basically an nscf type calculation. But there are a few workarounds. See this SE question - Calculating HSE06 band structures on Quantum ESPRESSO. It has accepted answer. Also, in this blog - Band structure calculations in QE using hybrid functionals - the procedure is described with examples.

In order to follow some background on why such calculations cannot be done and all the possible workarounds, you might want to check the following QE user forum threads: this, this, this.

$\endgroup$
0

You must log in to answer this question.

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