Page 1 of 1

Does known issue #113 affect metallic RPAR/ACFDTR calculations without WAVEDER?

Posted: Tue Nov 17, 2026 7:48 am
by sc_han

Dear VASP developers,

Could you clarify whether known issue #113 affects metallic RPA calculations with ALGO = RPAR or ACFDTR when LOPTICS is left unset and no WAVEDER file is provided, particularly with LFINITE_TEMPERATURE = .TRUE.?

The RPA wiki recommends omitting long-wavelength contributions for metals by disabling LOPTICS and removing WAVEDER. However, issue #113 does not explicitly distinguish this case from calculations that include those contributions.

Is it correct that the Drude-term bug has no effect when the long-wavelength head is omitted? Would fixing #113 change the correlation energies under these conditions, or only when a nonzero Drude contribution is present?

Thank you for your clarification.


Re: Does known issue #113 affect metallic RPAR/ACFDTR calculations without WAVEDER?

Posted: Tue Nov 17, 2026 9:25 am
by manuel_engel1

Hi,

Thanks for posting on the forum. Your understanding is correct on both points. Just to reiterate:

  • Yes. The Drude term scales with the plasma frequency, which VASP reads from WAVEDER. Without WAVEDER the long-wavelength head is zero, no Drude term is added, and the bug has no effect at any temperature.

  • The fix changes results only when a nonzero Drude contribution is present. In your setup, it does not change the correlation energies.

There is one exception: VASP still includes the head if a WAVEDER from an earlier LOPTICS run is in the directory. The same happens with LH5=True if a LOPTICS run wrote vaspwave.h5. In these cases you are affected.

You can check this yourself in the OUTCAR file. Search for "HEAD OF MICROSCOPIC DIELECTRIC TENSOR (INDEPENDENT PARTICLE)". If it is 1 on the diagonal and 0 elsewhere at every frequency, then the head was left out and your results are unaffected.

I hope that I was able to answer your questions.