Attention to the setting of light vector

[Quxiao]

Dear friends：
I have a problem. I know that the setting of linear light vector is determined by setting blongdir. For example, the x direction is 100, the vibration direction of light is only in X，which represents eλ，λ= 1，2（polarization mode）,is that correct?

If my understanding is correct, but it is strange that this is not consistent with my group theory transition analysis. Through the weight file, I found that the main exciton contribution is in the 14th band and the 16th band, while the 15th band is the highest valence band, which should be allowed. However, during the calculation, I found that it was banned. I suspect it is the problem of the direction setting of my incident light, and my system is anisotropic，(2D).Are there any other parameters that need extra attention? For example, is longdrx set to 110 direction?I have done GW calculation before BSE calculation. Will longdrd in GW calculation also be affected？

It's strange that I can't find the transition from 15 to 16 states at gamma point, even a dark exciton state，the only explanation I can think of is that the direction of light addition is unreasonable,please give me some help！

Finally, in order to ensure the correctness of light, I also tried to set longdrxs and blongdir to 111, but that is a 2D system, and I'm not sure whether this is reasonable. Of course, I also tried to set it to 110.
Best Wishes
Quxiao
BIT

[Daniele Varsano]

Dear Quxiao,

yes, that's correct setting Blongdir (100) you have light polarized along the x cartesian direction.
Note that this affects the calculation of the dipoles only <v|x|c>, so the strength of the excitation.
If the transition you expect is not present, this should not be ascribed to the light polarization, while its strength is affected, so you should have found a dark exciton.
Have you checked if that transition does participate in higher excitations?

The direction LongdrXs also affects the long-range dielectric screening eps_00(q=0), but using (110) direction is usually a good choice and major difference even. using (100) or (001) are not expected.

Best,
Daniele

[Quxiao]

Dear Daniele：
Thank you for your patience and help！
I know that dipole，so I find something strange when I calculate with the GW +BSE.
I use the ypp to analyse the first bright exciton,and find the contribution due to the index 16 -17 band at index 1 kpoints (gamma),which shown in o-weight file.
However,I run the readdipole.py,in the 16th to 17th States, dipole is 0 at any k point and in any direction.(0,1,2)
The index 16 is the top valence,and the 17 is the bottom conduct.
This is the case for both GW and yambo calculated dipole.This is obviously contrary to the definition that the oscillator strength is composed of dipole, so I don't know what causes it.
readdipole.py:

Code: Select all

from yambopy import *
       # Generic indices for testing
       #ik,ir,ic,iv = [3,1,6,2]
          f = open('1','w')
      # Read lattice info
          ylat = YamboLatticeDB.from_db(filename='./SAVE/ns.db1')
      #for i in [0,160]:
      #    print(ydip.dpoles[i,0,16,17])
     # Read dipole matrix element
        ydip = YamboDipolesDB(ylat,save='gw.out',filename='ndb.dipoles')#or yambo.out
        for i in range(0,160):
             print(ydip.dipoles[i,0,16,17],file = f)

I suspect that indexband starts from 0 and calculates that < 15|r|16 > is also super small at gamma point, so it makes me very confused
Please give me some help to ensure something!
Best wishes!
Quxiao

[Daniele Varsano]

Dear Quxiao,
have you tried to plot the KS wave function in order to check they look as you expect?
is it possible that you have zero dipole due to syimmetries?

Best,
Daniele

[Quxiao]

Dear Daniele:
I tried to draw the exciton projection energy band, and found that it does contribute to 16-17, but the calculated dipole is super small, which makes me feel very contradictory. I have calculated the 16-17 transition forbidden system before,another system, but in fact there should not be such a forbidden system, which makes me feel very confused. I suspect that my symmetry is not forced. I will continue to try to force symmetry, I don't know whether the transition information read by YPP and readdipole.py are the same way.
Is it possible that it is caused by the inconsistency between the GW k-point grid and BSE kpoint grid ?
The symmetry given by the structure I calculated before may not be very correct, but it does not seem to affect the BSE eigenvalue . I also calculated BSE through VASP, and found that the eigenvalue remains unchanged, and the obtained GW energy is almost the same, but the contribution is different. It is preliminarily determined that it may be the reason for symmetry, Or maybe it's caused by the bad addition of polarized light. I don't have a reasonable explanation,according to the existing symmetry analysis, the dipole should not be 0, and the transition of 16-17 here given by YPP command does exist, (I do that with VASP also and find that exist),which is considered as the main contribution.I try to force symmetry before calculation.
Thanks a lot!
Best wishes!
Quxiao

[Daniele Varsano]

Dear Quxiao,
sorry but I'm bit confused.

In the first post you say that the 15th band is the highest valence band. Then, in the second post you pasted an output where the 16th band is occupied. How is this possible?
Next in order to clarify, you can plot all the contribution to an exciton. By default yambo print transition contributing more than 5% but you can change the threshold by setting in input the variable "Weight_treshold" if you set it to zero you will have all the transitions with their weights.

Important, note that the weights are the square modulus of the eigenvectors |A_cvk|^2 and they are not weighted with the dipoles.

Best,
Daniele

[Quxiao]

Dear Daniele：
This is the case. I calculated the two systems separately. The highest occupied electronic state of the two systems is the 16th state, but the distribution of exciton contribution states in the first system is different. The first system is distributed in the 15th and 17th, the exciton in the second system contribution is in 16th and 17th, according to the principle of group theory, both should contribute in 16th to 17th, that I have calculated from VASP code ,so this makes me wonder, I guess it is caused by symmetry , so giving symmetry is still being tested. For the relationship between the electron-hole pair weight and the dipole of the excitons, I did have a blurry image before, thanks for your help!
I will continue to test that!
Thank you very much for your patience！
Best wishes!
Quxiao

[Quxiao]

Dear Daniele：
Thanks for your help!
I have solved my previous problems，that is related to the symmetry，it is now in good agreement with my theoretical prediction.
But I have a new question,"the LongDrXs is not critical as it will affect only one element of the matrix W_oo(q=0)",as you said before, the impact is not great, but the ‘q’, as a photon emission, should have a direction. According to my understanding, the LongDrxs here is consistent with the photon incident direction. For example, if the LongfDrXs is set to 110, the photon is incident in a vertical plane, corresponding to a small disturbance. Is it such a physical image ？
Best wishes！
Quxiao
BIT

[Daniele Varsano]

Dear Quixiao,

if the LongfDrXs is set to 110, the photon is incident in a vertical plane, corresponding to a small disturbance. Is it such a physical image ？

Yes, that's correct if with vertical you mean (001).

Best,
Daniele