[CP2K-user] [CP2K:19104] Re: No agreement between the LR and RT absorption spectra

[Natalia K]

Dear Beliz,

thank you for your answer. As I said, I tried changing many different 
parameters in TDA-CP2K. For instance, MT Poisson solver changes the 
spectrum by a tiny bit, as does the TZP basis set. The parameter change 
affects more the higher frequency and not the low frequencies. The XC 
functional does affect a lot the results, but still, the difference remains 
huge when comparing to RTP+Fourier transform (the latter agrees with the 
literature, so I trust it) with the same parameters.

I do not perform Gaussian calculations myself, I just compare with the 
literature that employs Gaussian, so I cannot check the TD vs TDA.

(Just a side note that with SIESTA, I do get agreement between the linear 
response and real-time propagation with delta kick for the absorption 
spectra of the same system. So, is it the Tamm Dankoff approximation that 
does something weird??)

Natalia

On Monday, July 10, 2023 at 4:43:45 PM UTC+2 Beliz Sertcan wrote:

>
> Hello Natalia,
>
> Regarding the mismatch between CP2K and Gaussian TDDFT results, once 
> should use the TDA keyword in Gaussian instead of TD to get an exact match 
> as CP2K employs the Tamm-Dancoff approximation. Furthermore, I would highly 
> recommend turning on the Poisson section and setting the solver to MT 
> instead of analytical.
>
>     &POISSON
>
>        PERIODIC NONE
>
>        POISSON_SOLVER MT
>
>     &END
>
> Since you are working with a heavy element I would also consider using a 
> bigger basis such as TZVP maybe even TZV2P. With these additions I think 
> you will be able to match the RTP spectrum better, but I would not expect a 
> perfect match. In my opinion what you have with the positions of the sticks 
> is not so bad.
>
> Best,
> Beliz
>
>
> On Monday, 10 July 2023 at 10:55:39 UTC+2 Natalia K wrote:
>
> Dear community,
>
> I am calculating the absorption spectra for a silver chain using both 
> linear response (TDDFPT) and real-time propagation implemented in CP2K 9.1. 
> Within the RT propagation, I apply a delta kick in all three directions and 
> then take the Fourier transform of the dipole moment and get the average 
> (over three directions) spectrum that is in a very good agreement with the 
> literature. In particular, with another GGA linear-response calculation 
> with the Gaussian code (image attached, top - my results, bottom - 
> reference). Within the TDDFPT approach, I calculate the oscillator strength 
> which should give the same frequencies as the real-time approach, but it 
> doesn't. In the figure attached, the oscillator strength is plotted with 
> impulses. I tested many different parameters but LR never agrees with RT. I 
> also tested a simpler system, a water molecule, and got a better agreement 
> between the two methods, but still not as perfect as here, for instance: 
> https://nwchemgit.github.io/RT-TDDFT.html#absorption-spectrum-of-water
>
> I would appreciate any help or suggestion. I attach both RT and LR input 
> files used to obtain the results on the figure.
>
> Best regards,
>
> Natalia
>
>

-- 
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 on the web visit https://groups.google.com/d/msgid/cp2k/aee1a43f-8a19-44fb-a9ac-a1dbd71d7fc6n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20230711/c0ccf46e/attachment.htm>