Unclear ypp output

[gsethi]

Hi everyone,

I ran a BSE calculation on Yambo and used ypp to get the weights and amplitudes of first 10 excitonic states.

I am not getting any contribution for my first 5 excitonic states while the latter states have same energy as the first exciton. To be more clear the first exciton energy in sorted_E output file is ~0.18 eV. The o*.exc_weights_at_1 is all commented (attached). Same is the case for weight files up to *_weights_at_6. The next weight file starting 7(attached) has a lot of contributions but I am unable to figure out what each contribution mean. But as can be seen the energy is ~0.18eV. Below are first few sorted exciton energies. The 7th state has a different energy than what is being reported in it's weight file.

# E [ev] Strength Index
#
0.1858992130 0.2196121786E-3 1.000000000
0.1881001443 0.1584975143E-4 2.000000000
0.2346948087 0.5123577648E-4 3.000000000
0.237847984 0.353519297E-09 4.00000000
0.24073363841 0.01839883439 5.00000000000
0.2416909635 0.1722074558E-5 6.000000000
0.2512641847 0.1120866422E-7 7.000000000
0.2513728738 0.8134040982E-3 8.000000000
0.2524401248 0.1903054248E-7 9.000000000
0.2530377507 0.3779506544E-2 10.00000000
0.2531914115 0.1715678809E-4 11.00000000
0.2545221448 0.3859438635E-7 12.00000000

I am quite confused with this. Also there are commented lines in the weight files specifying eh pair weights at different k's. I am not sure how they are related to uncommented lines below. I am also attaching the output I got when I ran ypp(image). I am a little concerned about what merged states 1 -> 2 mean. Any suggestions or insight would be really helpful.

Thanks a lot.

~Sethi

[Daniele Varsano]

Dear gsethi,

please sign your post with your name and affiliation, this is a rule of the forum and you can do once for all by filling the signature filed in your user profile.

1) As you can see from your output file, the output reports the transitions having a weight larger than 5% of the total:

Code: Select all

Electron-Hole pairs that contributes to Excitonic State 1 for iq=1 more than  5.000000%

If the output is empty it means that none of the transitions is higher than the threshold. You can lower the threshold by changing the threshold using the variable:

Code: Select all

Weight_treshold=  0.05000  

This variable appears in the input by adding verbosity in the input file (ypp -e ...-V all)

2) The output file contain the weight (7th column) and the respective transition indicated by index (bands, k points in the BZ). The k points in the BZ is indicated as k point in the irreducible BZ and its associated symmetry. Note that you have K_v = K_c as the BSE is done for Q=0, the ypp utility it is also meant for finite q but the solution of the finite q BSE has not been yet released in the GPL version. The last column is the Kohn-Sham energy of the transition E_c-E_v.

3) The commented part are the weights integrated over bands at each k points (also here there is a threshold).

4) Degenerate excitons are summed up and a linear combination is considered, this is done because if you want to plot excitonic wave function in real space you need to consider degenerate excitons together in order to preserve the symmetry (if any), anyway this can be removed by setting the variable:

Code: Select all

Degen_Step=   0.0100   eV  

to zero, in this case excitons are not merged anymore. Also this variable will appear setting the -V all verbosity.

Best,
Daniele

[gsethi]

Hi Daniele,

Thank you so much for your reply. I apologize for not including my signature earlier. I updated it now.

Your suggestions and explanations helped me a lot. I have another problem which came up while I was trying to do a triplet excitonic calculation for the same system. I used BSEmod = 'coupling' , uncommented '#WehCpl', and set BSENGexx = 0 RL. I ran the BSE calculation again using previous database.

Following are the issues I am running into -
1. I used ypp to sort eigenvalues. Usually the eigenvalues index is based on energy values but this is not the case here. I have negative energies which I am expecting but in the E_sorted file even with sorted energies the indexes start from a higher value. I am not sure if this implies an error somewhere.

2. I tried to get the weight and amplitude of the exciton I was interested in specifically the one with highest intensity and negative energy. I received an error attached in the screenshot below.

3. I went ahead and tried to use yambopy in order to plot excitonic wavefunctions but ended up getting another error for the above excitonic state which is in the next screenshot.

These could be because of negative energies I am getting but like I mentioned I am expecting negative energies for which I need to plot the wavefunctions and calculate other excitonic properties. Also it could be the way I am running the triplet exciton calculations which I am not sure about. I found these tags in previous discussions on this forum.

I would really appreciate any help or suggestions.

Thanks a lot

~Sethi

[Daniele Varsano]

Dear Sethi,
what you observe is not expected: can you attach your input, report of the BSE calculation and the E_sorted and o.weights files?
In order to attach them in the post you need to rename the files adding an allowed suffix (e.g. .txt) or zip them.

Best,
Daniele

[gsethi]

Hi Daniele,

Definitely. Thanks so much!

I am attaching my BSE report file, E_sorted, and .weight file for the state I am interested in.

~Sethi

[gsethi]

The BSE input file is attached here. Thank you so much for your help.

~Sethi

[Daniele Varsano]

Dear Sethi,
in the E_sorted file the energy are correctly reported in ascending order. The index are not ordered but this is expected as you are considering the coupling terms in the excitonic matrix. Also the weight file (index 15) it has been generated, so in which case you encounter the problem, for the index 19 (bright exciton?).

I any case, you have positive and negative eigenvalues due to the coupling and the excitation energies seems to me they are symmetric wrt the zero energy, so it means that the coupling here is not really relevant, and you are essentially looking at the resonant and antiresonant spectrum (v->c and c->v transitions). Most probably you can avoid to insert the coupling (at least I suggest you to see how much change the positive eigenvalue spectrum when the coupling is neglected) and look at the positive energy part of the spectrum.

About the excitonic plot, I cannot help on yambopy, what happen if you calculate it running directly ypp?

Best,
Daniele


PS: note you are considering BSENGexx= 1 RL, which does not produce a triplet exciton, you should use BSENGexx= 0 eV instead. Note also that BSENGBlk= 4 RL could be hardly under convergence.