Moderators: Global Moderator, Moderator
Dear all,
I came across some questions when I use RS-DDH as functional to calculate the imaginary part of the dielectric function. I found that my calculated results couldn't agree with the experimental results very well. So I want to know whether the RS-DDH is a little inaccurate, and how large the disagreement with experiments can be accepted. In addition, what can I do to lessen the disagreement?
Looking forward to your reply. Thanks!
Best wishes.
Dear Jike Wang,
Thank you for your contribution to the VASP forum.
I moved your post to the From users for users part of the forum, because it does not contain a specific question about VASP usage, but is a very broad question about the accuracy of dielectric-dependent hybrid functionals.
I hope for a lively discussion with other users in this part of the forum, and suggest starting from this publication:
Assessing model-dielectric-dependent hybrid functionals on the antiferromagnetic transition-metal monoxides MnO, FeO, CoO, and NiO
The Using VASP part of the forum is reserved for technical questions about the VASP software. If you have a more specific question on the topic, or any other topic, you are more than welcome to make another post there. Please consult the posting guidelines for the using VASP forum for more information.
Cheers, Michael
Dear Jike.Wang,
We wrote a paper where we analysed the performance of time-dependent-RSDDH: Tal et al., Phys. Rev. Research, 032019 (2020).
In short: this approach shows a very good performance for the calculation of the dielectric function. But it makes a number of approximations compared to the full BSE-GW calculations. Hence, one should not expect it to perform better than BSE-GW. At best TD-RSDDH can reproduce the BSE results. Hence, if in your system the BSE-GW spectrum doesn't agree well with experiment, one should not expect that TD-RSDDH would do better. For example, the lack of the phonons in the calculation can be responsible for a large discrepancy in BSE-GW and experiment.
Furthermore, in some systems the GW band gap and the RS-DDH band gaps show a substantial discrepancy. Hence, the peaks in the dielectric function will be shifted for BSE-GW vs TD-RSDDH. But the overall shape of the spectrum calculation with these two methods might be in a very good agreement.
In your document in particular, it seem that the overall shape of the spectrum is reproduced reasonable well. However, it looks to me that the spectra of Si and LiF are not well converged with the number of k-points, which is especially important when the results are compared to experiment.
I hope that helps.
Best wishes,
Alexey