unstable BSE results and ypp reads the wrong IPA+sc transition energy for one-dimesional structure

[lyzhao]

Dear all,
I run a TDA-BSE calculation for quasi-one-dimensional C60-fullerene-chain.
The quasi-one-dimensional C60-fullerene-chain is constructed by adjusting the distance (d) between two C60-fullerene-chains. When d is small, the system is considered as two-dimensional structure. For this two-dimensional structure, the TDA-BSE result seems to be fine. However, when d is large enough, the system is considered as quasi-one-dimensional C60-fullerene-chain. For this quasi-one-dimensional C60-fullerene-chain, the TDA-BSE calculation without scissor correction produces the negative excitonic transition energy. When the scissor correction (sc) is used, the TDA-BSE calculation does not produce the negative excitonic transition energy. I think the TDA-BSE is unstable, similar to viewtopic.php?t=1261
I try to analysize the TDA-BSE-sc results without negative excitonic transition energy, to my surprise, when I use ypp to analysize the first dark exciton (Es=0.27754515), the major IPA contribution (120_HOMO->121_LUMO) listed in the weights file shows that the IPA+sc transition energy (1.30) is much smaller than the smallest IPA transition energy output by pwscf+sc (1.27+0.84 = 2.118313).
Note that, the quasi-one-dimensional C60-fullerene-chain has a direct band gap.
I also tried to use the Coulomb potential cutoff and RIM, and got the similar results.

Some input and output files are attached.

referfiles.zip

Any help will be appreciated.

Part of o-bse_k0205_blk5Ry_v10c15_X010.exc_qpt1_E_sorted

Code: Select all

#    E [ev]             Strength           Index
#
    0.27754515         0.87041714E-15      1.0000000    
    0.415798336        0.116360305E-6      2.00000000   
    0.489503205        0.178568413E-7      3.00000000   
    0.52257109         0.24699835E-15      4.0000000    
    0.59026492         0.91507294E-13      5.0000000    
    0.66519332         0.41979115E-11      6.0000000    
    0.707085669        0.723122503E-4      7.00000000   
    0.723453581        0.522076746E-7      8.00000000   
     1.0587963         0.72475255E-10      9.0000000    
     1.14108109        0.231848247E-6      10.0000000   

Part of o-bse_k0205_blk5Ry_v10c15_X010_nosc.exc_qpt1_E_sorted

Code: Select all

#    E [ev]             Strength           Index
#
   -0.56455517         0.13588325E-14      1.0000000    
   -0.426303297        0.116355196E-6      2.00000000   
   -0.352599204        0.178404331E-7      3.00000000   
   -0.31952795         0.59441336E-15      4.0000000    
   -0.25183672         0.40022032E-13      5.0000000    
   -0.17690800         0.41128806E-11      6.0000000    
   -0.135013938        0.723130215E-4      7.00000000   
   -0.118647352        0.522816563E-7      8.00000000   
    0.21669467         0.71330108E-10      9.0000000    
    0.298980176        0.231615005E-6      10.0000000   
    0.33745739         0.26135211E-09      11.000000    
    0.35067469         0.64098706E-14      12.000000    
    0.36473015         0.95327324E-09      13.000000    
    0.377417266        0.330991696E-6      14.0000000   
    0.41102618         0.64979502E-11      15.000000    
    0.432607919        0.127243638E-6      16.0000000   

Part of o-bse_k0205_blk5Ry_v10c15_X010.exc_qpt1_weights_at_1

Code: Select all

#    Band_V             Band_C             Kv-q ibz           Symm_kv            Kc q ibz           Symm_kc            Weight             Energy [eV]
#
        120                121                3                  3                  3                  3                  0.112998           1.305158
        120                121                 6                 3                   6                 3                  0.112985           1.305238
        120                121                3                  1                  3                  1                  0.112871           1.305158
        120                121                 6                 1                   6                 1                  0.112858           1.305238

[Daniele Varsano]

Dear Youzhao,

1) when calculating BSE spectrum, to be consistent, QP corrections need to be included (e.g. via scissors), the fact you get negative excitation energies when considering QP correction is not a sign of instability.

2) Looking at your file, specifically pw_nscf.out, you can see that the minimum KS gap is 1.276:

Code: Select all

 highest occupied, lowest unoccupied level (ev):    -3.7854   -2.5091

Looking at the first excitation contribution, it is indicated that come mostly from k point number 3, and the QE result is exactly the one reported by Yambo. Please note that energy reported by Yambo is the KS energy difference (without QP correction).

Best,
Daniele

[lyzhao]

Sorry for my mistake, ypp reads the right IPA transition energy.
But I still need possible help for issue of negative transition energy.

Best regards.

[Daniele Varsano]

Dear Youzhao,
negative excitation energies are obtained when QP corrections are not included, so they are meaningless.
Best,
Daniele