hbn tutorial and the "Coulomb Cutoff"

[chwolf]

Dear all,

I have a probably stupid question: the hbn tutorial has a section about layer separation and Coulomb cutoff; I tried to converge the BSE spectra wrt. the layer distance but increasing the layer separation (I went up to 120 a.u.) still does not "converge" the BSE spectrum. As a matter of fact it looks like it is shifted by a constant just as I increase the layer separation.

Note that the dielectric converges around 65 a.u. layer separation.

Is there any reason why the gap keeps opening by increasing the layer separation or is this just a case of (most likely) underconverged BSE wrt. k-points etc?

Thanks for your help!
Chris

[Daniele Varsano]

Dear Chris,
many-body calculations are hard to converge with respect to the vacuum and the coulomb cutoff is meant to mitigate this slow convergence.
Let me ask:
1) You BSE spectra are made on top on different GW calculations (i.e. the GW gap is calculated for each of the considered supercell volume or it is kept fixed?)
2) Are you using the box cutoff or Wigner-Seitz cutoff? If you are using the box you can have a try with the WS which is better defined for the head of the potential.
Note that there is a dependence with the k point sampling.

Best,
Daniele