BSE calculate

[Abeshipour]

Dear Yambo developers,

I'm trying to calculate the dielectric function of NaBO2 bulk with BSE according to the "Calculating optical spectra including excitonic effects: a step-by-step guide' tutorial on the website, however...

I first calculated optical spectra at the independent particle level for bulk NaBO2 according to "First steps... step4" and find that The highest peak of the imaginary part of the dielectric function of this material is about 6 eV.
But when I do calculations BSE for this system and plot the imaginary part of the dielectric function according to the output file, o-3D_BSE.eps_q1_diago_bse in the same direction x, the highest peak of this part is about
1 eV. That is, it gets smaller.
It is also non-convergent, meaning that for every value I give to the input data, it is drawn the same and does not change...

I tried very hard to understand my mistake but I could not. can you help me?

I put the input files of BSE in this section.

[Daniele Varsano]

Dear Abeshipur,
please sign your posts with your full name and affiliation. This is a rule of the forum, you can do once for all by filling your signature in you forum profile.
Next please, post your question in the appropriate subform, your question is not related with initialisation.

Coming to your question, can you explain better what is the problem?
What I can see is that you applied a scissor-shift of this kind:

Code: Select all

% KfnQP_E 
 2.50000 | 1.40000 | 0.950000 |        # [EXTQP BSK BSS] E parameters  (c/v) eV|adim|adim
% 

How did you choose these parameters?
The peak position of the spectrum depends on the scissor and stretching you choose as you are applying a rigid shit to the bands plus a stretching which is rather large.

Best,
Daniele

[Abeshipour]

Dear Daniele

Thanks for your reply, I Will correct my profile info ASAP.



I've asked that, in my BSE calculation, the absorption peak of dielectric function is smaller than the peak which calculated by RPA.

Is this correct or not?
If it's not correct, how can I fix it?


About your second question, I have to say:


I got the way for choosing KfnQP_E from this link

"viewtopic.php?t=1378"

according to that, I found out that to select this parameter, I should calculate the slope of the linearly QP energies with respect to KS energies.
Which they are column 3+ column 4, and column 3 in the o.qp output file.


I put Figure 1 again for two approximations. for better comparison

I also put the QP energies with respect to KS energies. To express your opinion about parameter KfnQP_E

[Daniele Varsano]

Dear Abeshipour,
your results sound strange, are the RPA and BSE calculated with the same number of bands?
Can you post your BSE output and report as well as the output/report of your GW calculations?

Best,
Daniele

[Abeshipour]

Dear Daniele

Thanks for quick reply,

Yes, I do not know where the problem is
Please help me...

I put the outputs.

Thank you

[Daniele Varsano]

Dear Href,
the spectra when compared with the Independent particle absorption looks ok.
The intensity comparison is most probably due to the different number of bands included in the two calculations.

In the GW calculation I would check the convergence of bands and blocks in the screening as a total of 70 bands with 48 occupied states seems to me it is very low.

Best,
Daniele

[Abeshipour]

Dear Daniele

Thanks for your helpful comments.

I'll check the GW calculations again. It would be great if let me know your insights about the GW calculations and it's problem.

Is there any necessity to increase the number of bands or blocks in order to achieve a better convergence?


Best regards
Aref

[Daniele Varsano]

Dear Aref,

convergence parameters are system dependent and should carefully checked.
There is not a precise rule to know in advance how many empty band are needed, but at a first glance the proportion of the bands you used with respect the occupied bands seems to me very low.

Best,
Daniele

[Abeshipour]

Dear Daniele

Thanks to you and your colleagues for all of these insightful comments/answers/recommendations, which help me to learn Yambo very well.

However, I am often doubtful about the convergence of the "BSEBands" in the BSE part.
I have calculated the "BSEBands" up to several intervals. Based on the attached convergence plot it would be great if you answer the question below:

Is the selection of "BSEBands" in the interval of (34-64) is an appropriate selection or should be charged to a larger one?

Best regards
Aref

[Daniele Varsano]

Dear Aref,
this depends on which part of the spectrum you are interested in and to what level of accuracy you need.
The first peak is perfectly converged, the second intense peak is is probably also converged, I suggest you to use more frequencies (BEnSteps) in order to better judge the quality of the convergence as it does not add computational effort.

Best,

Daniele