G0W0 calculations on double grid

[mrefiore]

Dear YAMBO developers and users,

I'm currently trying to run a G0W0 calculation exploiting the double grid feature. I performed an nscf run on a dense grid (32x32x1) and in the coarse grid (9x9x1) SAVE I successfully created the ndb.DoubleGrid database. However, when I run the G0W0 calculation on the coarse grid, even though in the report and logfile I notice that YAMBO uses the fine grid I get a gap result that is extremely close (just 1 or 2 meV away) to the straightforward coarse grid calculation (without double grid).
What am I missing? I would expect a bigger impact of the double grid on the GW calculation. I also tried with a finer coarse grid (14x14x1) but the outcome did not change, the double-grid result is very close to the coarse one.
I'm attaching the input, report and logfiles of my calculations.
Thank you very much for your help!
Best,


Michele

[claudio]

Dear Michele

I think this is due to the no correct treatment of the q=0 term,
that unfortunately the double-grid cannot solve.
You may have a look here:

How dielectric screening in two-dimensional crystals affects the convergence of excited-state calculations: Monolayer MoS2
Falco Hüser, Thomas Olsen, and Kristian S. Thygesen
Phys. Rev. B 88, 245309
https://core.ac.uk/download/pdf/192327286.pdf

maybe Daniele will add more comments on this point.

best
Claudio

[Daniele Varsano]

Hi Michele,
indeed we have very decently developed a new technique to accelerate GW convergence for 2D materials. This development it is not yet merged in the official release, but it will merged soon.

Best,
Daniele

[mrefiore]

Dear Claudio and Daniele,

thank you very much for your answers! I'm looking forward to the new release then!
Best,


Michele