exciton binding energy

[Daniele Varsano]

Dear Samaneh,
I have already answered you in a previous post.
Calculate the dynamical dielectric function at the level of theory you desire (RPA, TDDFT, HF, BSE), and look at the zero energy value.
Daniele

[samaneh]

Dear Daniele,
Yes I found it.
Thanks

[Reza_Reza]

Dear Daniele,

I computed the optical absorption for a 1D system. However, the energy of the peaks on my plot of the optical absorption is wrong. ( plot "o-ev_GW_3.alpha_q1_diago_bse_TEST" u 1:2 w l)
when i plot with (plot "o-ev_GW_3.alpha_q1_diago_bse_TEST" u 1:4) it seems correct.


I attached the report and output files. (Since we can add three attachments, I write the excitonic energies here.)

# Maximum Residual Value = 0.38730E+01
#
# E [ev] Strength Index
#
-0.124014705 0.999999940 1.00000000
0.15619701 0.33847322E-09 2.0000000
0.258908361 0.122013018E-1 3.00000000
0.47375387 0.92644096E-11 4.0000000
0.502913177 0.894569885E-2 5.00000000
0.74984646 0.16381771E-11 6.0000000
0.778735518 0.653948821E-2 7.00000000
0.81744349 0.54038078E-12 8.0000000
1.0005046 0.85885682E-11 9.0000000
1.02878881 0.522387633E-2 10.0000000
1.1728863 0.41521029E-12 11.000000
1.1996268 0.93699465E-10 12.000000
1.2272360 0.25179446E-12 13.000000
1.2286303 0.96458583E-12 14.000000
1.25555646 0.440468034E-2 15.0000000
1.2989506 0.13916714E-11 16.000000
1.3741199 0.52030075E-10 17.000000
1.4155179 0.37175178E-12 18.000000
1.4333268 0.28858803E-10 19.000000
1.4381497 0.11025027E-10 20.000000
1.46215904 0.393986190E-2 21.0000000
1.4740620 0.11227375E-11 22.000000
1.5333049 0.88010932E-10 23.000000
1.6097236 0.26518143E-11 24.000000
1.6208855 0.90077832E-11 25.000000
1.6253181 0.28864687E-10 26.000000
1.65039110 0.400217436E-2 27.0000000
1.6764745 0.59737519E-11 28.000000
1.7230855 0.52400878E-09 29.000000
1.7903579 0.21673523E-09 30.000000
1.8104832 0.97735570E-10 31.000000


Best,

Reza

[Daniele Varsano]

Dear Reza,

what do you mean by "wrong" and "right"?
The fourth column is the independent particle spectrum, with excitations corresponding to Ec-Ev and the onset in the position of the gap.
In the second column you have the result of the BSE and you get a binding energy larger than the gap, resulting in a negative excitation (instability), the spectrum is negative for positive energy because of the time-ordering. It is up to you to check if it is physical or not. I suggest you to check the convergence parameter both in GW and in BSE, in particular the k point sampling.


Best,
Daniele

[Reza_Reza]

Dear Daniele,

I did my calculation using larger k-points. However, the optical absorption shape is strange and not compatible with the exciton energy range. I attached the needed files for analysis.
When I increase the k points from 50 to 70, the size of the band gap is similar in both cases.

Best,

Reza

[Daniele Varsano]

Dear Reza,

the shape is due to the fact you have a negative excitation energy which hinder an instability of your system. You should carefully check your convergence both at GW and BSE level.

Anyway, I noticed in your report you have used database coming from different calculations (a different SAVE directory) and this is dangerous and error-prone. See the many warning in your report:

Code: Select all

 [WARNING] Incompatible serial number for ./job_00_first_run/

I strongly discourage to do this.

Next, you are using the WS cutoff, this should work, but it is not extensively tested, you can try to use a different cutoff as the "box x" together with the rim_cut to check in case this is giving some problem.

Other parameters to check are:
BndsRnXp: considering you have 66 occupied bands, 100 bands seem to be rather low
BSENGexx: this can be converged, but can be veirified, increasing its value does not cost much.

Best,
Daniele

[Reza_Reza]

Dear Daniele,

Many thanks for your considerable help, I think I performed p2y two times for GW and screening computation separately. I have to check it and do it once for both calculations.
In the case of BndsRnXp, in my screeching calculation, I used BndsRnXs instead. Do I have to change it to BndsRnXp in the case of the GW correction in the BSE calculation?

cat > 01_screening.in << EOF
screen # [R] Inverse Dielectric/Response Matrix
em1s # [R][Xs] Statically Screened Interaction
dipoles # [R] Oscillator strenghts (or dipoles)
Chimod= "HARTREE" # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
% BndsRnXs ------ > BndsRnXp ?
1 | 100 | # [Xs] Polarization function bands
%
NGsBlkXs= 8 -----> NGsBlkXp ? Ry # [Xs] Response block size
% LongDrXs
1.000000 | 0.000000 | 0.000000 | # [Xs] [cc] Electric Field
%
EOF

Best,

Reza