ALGO=None and VASP.5.4.4

[IBRAHIM]

Dear VASP Admin,

For the DOS calculation using HSE06, we try ALGO=None and ISMEAR=-5 after pre-convergence with ALGO=All and ISMEAR=0 in VASP.5.4.4.

With ALGO=None, the run finishes quickly, however, we noticed that the exchange energy for HF is not included, and hence the total energy is different.
- Is this acceptable, and can we proceed with the DOS figure? The band gap remains the same for both runs.

From our experience, this is not the case with VASP.6.3.2, and the total energy is the same for both runs.

Many thanks in advance,
Ibrahim

[fabien_tran1]

Hi,

The energies of the orbitals printed in OUTCAR of the second calculation are identical to the ones from the first calculation. This should indicate that everything is fine and that the DOS should correspond to HSE06. In any case, this is not clear why the HF exchange energy is not printed.

Actually, if you have also access to VASP.6.3.2, you could compare the two HSE06 DOSs obtained with VASP.5.4.4 and VASP.6.3.2 for a small system like MgO. They should be the same.

[IBRAHIM]

Thanks for your reply.

In our current case, the run with ALGO=None finished quickly using VASP5.4.4, and the HF exchange energy was not included.

However, when checking old files with ALGO=None in VASP6.3.2, the run took a long time due to the hybrid functional.

If the DOS corresponds to HSE06 in VASP5.4.4, this will save a lot of time.

-Can we get more information about this issue?

Many thanks in advance,
Ibrahim

[fabien_tran1]

I could reproduce this different behavior between VASP.5.4.4 and a recent version.

With VASP.5.4.4, the calculation of the HF exchange energy is indeed skipped when ALGO=None, but this does not matter for the DOS because it is calculated from the WAVECAR of the previous calculation and the total energy is of no interest.

With recent VASP, the HF exchange energy is calculated which is expensive and useless for the DOS. This can be avoided by disabling the HF part in INCAR, which does not affect the DOS.