[CP2K-user] [CP2K:20843] Re: RPA+rSE with gradients

[Frederick Stein]

Dear Benjamin,
Conceptually, this may not work because rSE is a correction from the 
non-diagonal elements of the Fock matrix plus higher-order terms normalized 
by differences in (semi-canonical) orbital energies whereas EXX/HF 
considers diagonal blocks of the occupied elements.
If you want to apply your method nethertheless, you could try the 
ENERGY_CORRECTION section 
(https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/ENERGY_CORRECTION.html), 
probably with ENERGY_FUNCTIONAL DCDFT to extract the EXX part. This skips 
the expensive RPA calculation step. Unfortunately, I have never used this 
method but maybe other people can jump in if you have any questions to 
these methods.
Best,
Frederick

Benjamin Shi schrieb am Freitag, 1. November 2024 um 12:41:16 UTC+1:

> Dear CP2K forum,
>
> My understanding is that RPA+rSE does not have gradients currently 
> available. I was wondering whether in a standard RPA calculation, it would 
> be possible to output the gradients from the EXX component of RPA (or do it 
> in a separate calculation), such that I can subtract these EXX gradients 
> from the total RPA gradients and add in HF gradients to get an effective 
> RPA+rSE with gradients.
>
> Best wishes,
> Benjamin
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/5f2f6a0c-1564-4ee1-9a03-3737437ad7f0n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20241101/8bc482d4/attachment.htm>