Tamm-Dancoff Approximation

[nrbsh]

Dear Yambo developer,

Using GW0+BSE in yambo-4.0.2-rev.90, I am studying optical properties of some nano-structure systems.

I calculate the dielectric function in TDA and beyond TDA ( using BSEmod= "xc_coupling" BSKmod= "SEX" BSSmod= "d" WehCpl)
But there is not any difference between tow spectra (in both of imaginary and in real part).

In literature, it is mentioned that TDA is not a good approximation for nano structures and now I am worried about my results.

I also attached the report file of BSC.

I really appreciate your help in advance.

With best regards,
Zara
Tehran University

[Daniele Varsano]

Dear Zara,
I had a look to your report and I can see that you have set:

Code: Select all

| BSEmod= "coupling" 

and you BSE kernel reads:

BSK|Identifier : 5802
|Dimension : 1536
|Bands : 20 35
|Exchange [res]:yes
|Correlation [res]:yes
|Kernel`s coupling :yes
|Exchange [cpl]:yes
|W interaction is bare :no
|Correlation [cpl]:no
|ALDA kernel in R-space :no
|RL vectors [exchange]: 2713
|RL vectors [correlation]: 709
|E/h energy range [ev]:-1.000000 -1.000000
|Coupling range [o/o]: 100.0000 100.0000

In order to go beyond the TDA you need to add also the correlation in the coupling part: this is achieved by setting:

Code: Select all

BSEmod= "xc_coupling"

as you wrote, but it looks that there was some problem in your input as it was not recognized.
Best,
Daniele

PS:

n literature, it is mentioned that TDA is not a good approximation for nano structures

True, but often it also happen that TDA results compare agrees with the experiment better than the full BSE.

[nrbsh]

Dear Daniele,
Thank you for your quick reply.

I have checked different options as BSEmod, also used xc_coupling but did not find any different between different options.

I attached my input file and the reports file for xc_coupling here.
Thank you

Best regards
Zara
Tehran University

[Daniele Varsano]

Dear Zara,
I can't see anything wrong in your report.
I can see that you are dealing with a 1D system. In my experience, the 1D systems I studied (polymers, nanotubes) the TDA approximation was reliable, very small difference observed when going beyond the TDA.
You obtain similar results or *exactly the same results* when adding couplings?

Best,
Daniele

[nrbsh]

Dear Daniele,
They are very similar to each other but not exactly same.
I attached the diagram of the imaginary part of the dielectric function with and without TDA.

Best regards,
Zara
Tehran university

[Daniele Varsano]

Dear Zara,
Most probably I think it indicates that the TDA is avery good approximation for your system.
As usual checks and convergence studies are mandatory.

Best,
Daniele

[nrbsh]

Thank you very much Daniele.

Zahra
Tehran University

[haseebphysics1]

Dear Daniele, in the latest version of Yambo (4.5.1),
we have three options for BSEmod

BSEmod= "retarded" # [BSE] resonant/retarded/coupling
It is apparent that coupling will be full BSE (coupling in the kernel also)? But what about the other two options? Retarded was not present in earlier versions! Is it the same as TDA? And when we should use resonant? I have no idea how this will affect the results of the BSE! I had chosen the default (retarded) and exchange and cooling was not included in the kernel (as you can see in the below report)

And is there any preference of usage of these options for the dimensions of materials (like bulk, 1D etc)?

Code: Select all

  BSK|Identifier              :4135
#      |Dimension               :36432
#      |Bands                   :138 - 182
#      |Exchange           [res]: yes
#      |Correlation        [res]: yes
#      |Kernel`s coupling       : no
#      |Exchange           [cpl]: no
#      |Correlation        [cpl]: no
#      |W interaction is bare   : no
#      |ALDA kernel in R-space  : no
#      |RL vectors    [exchange]:1895
#      |RL vectors [correlation]:695
#      |E/h energy range    [ev]:-1.000000 - -1.000000
#      |Coupling range     [o/o]: 100.0000 -  100.0000
#    W |Interaction is diagonal : no
#      |Matrix size             :695
#      |Bands                   :50 - 350
#      |e/h energy range    [ev]:-1.000000 - -1.000000
#      |Poles              [o/o]: 100.0000
#      |RL vectors in the sum   :5353
#      |[r,Vnl] included        : yes
#      |Field direction         :0.5774E-5 0.5774E-5 0.5774E-5
#      |Coulomb Cutoff          :none
#      |xc-Kernel               :none
#   RIM|RL components      [col]:0
#      |Random points      [col]:0

Thanks,

[Daniele Varsano]

Dear Haseeb,

both resonant and retarded relies on Tamm-Dancoff approximation (the retarded also consider the anti-resonant part to build the response function) and it is the default for TDA calculations.

and exchange and cooling was not included in the kernel (as you can see in the below report)

Yes, they are included in the resonant part, they are not in the coupling as you are in TDA and there is no coupling.
If you have some hint, or doubt the TDA can fail you can activate the coupling (the dimension of the matrix will be doubled).
By default only exchange is considered in the coupling terms, in case you want also to consider the eh interaction (W) you need to uncomment the following variable:
#WehCpl

Best,
Daniele

[haseebphysics1]

Dear Daniele, thanks for the guidance, I further want to ask,

in case you want also to consider the eh interaction (W) you need to uncomment the following variable:
#WehCpl

Can we activate the eh interaction (WehCpl) while in TDA (retarded) or do we need to go to full bse mode (coupling)?

Does BSE naturally not include the eh-interaction by default? I was of the view that in BSE eh-interaction always present and that's why I was doubtful about the BSEmod variable!

Thanking you,