Doing BSE on top up GW

Hello everyone,
i I want to do BSE on top up GW in exciting, so for doing this I did G0W0 and plot the valence band and conduction band as well as tutorial explained and I plotted DOS and everything worked very well as I predicted for my structure, but when I use in the gw part "taskname=“g0w0” " to skip this part and enter the BSE part, and I plot the “dielectric function”, it’s completely wrong.
my first question is: Should I skip the gw part to do BSE top up GW or not? because after doing skip, I cannot see any evidence of doing GW in the INFOXS.OUT output file that it’s strange?
and is there someone to check my input file and tell me if I am doing something wrong?
I attached the input file and dielectric function figure.
BSE.txt (3.1 KB)

I found the problem and solve it :slight_smile:

Hi Nasrin,

I’m not sure this feature (running BSE on top of GW) was ever tested. I would refrain from using it, unless you’re sure the results look correct.


Hi Alex,
Yes, I am pretty sure, and doing BSE on top of G0W0(single shot of GW)(but I guess,we can run BSE on GW too) is tested already, even in the tutorial, you had a part to how calculate it, and I use this too:
and in so many paper theoretically it calculated.
But, I have a question, after SCF calculation for example for my structure with a bandgap around 3.29, the groundstate calculation is not good enough, cause the bandgap is around 2.4, so after going further and use G0W0 and it gave me 3.12, then I run the BSE and use the quasiparticle and so on for this part, I compared my result with experiment, and they are in ba good agreement, but after calculating BSE, it wides the band gap to 4.48, and my question is why? I need some physical answer to understand it?

What material and what experiment are you comparing to?

GW gives the quasiparticle gap, BSE gives the optical gap. The difference between the optical gap and the QP gap is the binding energy of the lowest exciton state. I would generally expect the optical gap to be smaller than the QP gap in the case of a bound excitonic state.

the material is SiC-4H a hexagonal structure.
I am trying to do BSE on top of GW.
maybe it’s the reason, cause my structure has an indirect band gap and a optical band-gap, it has a vertical interband transition. So it would be bigger as compared QP gap.
I am not sure, please let me know, if I am wrong,