BSE on the top of PBE/Hybrid without GW

[subrata]

Hi,
Is it possible to run BSE (not mBSE) on the top of the PBE (with scissor) or on the top of the hybrid (without scissor)?
I can see in other codes there is a option to run BSE directly after PBE, without going to GW steps.
Is that possible. I want to avoid GW steps.
Is a small hack in the code is poosible to do this?

Thanks
Subrata

[pedro_melo]

Dear Subrata,

I am not entirely sure what king of calculation you want to run, so I'll need some more details. Please bear with me.

To run a BSE calculation you need both the corrected energies and the dielectric function. The corrections can come from GW, a hybrid functional, or indeed a scissor operator. My question is now from where do you want to get your dielectric function. If it from neither a previous GW calculation or a model, then I guess that you want to perform a TD-DFT calculation. If that is the case then yes, it is possible for VASP to run a BSE-like calculation immediately after a DFT-PBE run, without performing GW.

Could you check if this is what you want to do? My advice is that you also confirm that for your system TD-DFT can accurately compute the macroscopic dielectric function.

Kind regards,
Pedro

[subrata]

Dear Pedro,
Yes, I want to do a BSE-like calculation immediately after a DFT-PBE run, without performing GW.
And yes, the frequency-dependent dielectric constant before BSE should be generated from eigher DFT-PBE or Hybrid.

Could you please provide me with the short scheme to bypass the GW calculation to BSE and perform PBE directly after PBE or Hybrid

At least I do not find any recipe from the VASP wiki

Thanks
Subrata

[subrata]

Another thing is that the BSE-like calculation (mBSE) is not the BSE calculation
If you have Hybrid eigenvalues which is similar accuracy as GW, then one should able to run BSE after that
There is some exciting code that is capable of doing it without doing GW.

[pedro_melo]

Dear Subrata,

From what I understand you seem to wish to perform a TD-DFT Casida run, with the screened interaction kernel coming from the a DFT functional, either PBE or hybrid. This is possible with VASP, please check the page for Time-dependent DFT.

This way you do not need to compute any W*.tmp files from a previous GW calculation. However, please check that this approximation is good enough for you, as TD-DFT notoriously fails to accurately describe excitons in solids or very large molecules.

Kind regards,
Pedro