Yambo Community Forum

Dear Daniele,

I did a small test. All the scf/nscf input and out files are attached. This material has four types of atoms, and i considered soc effect in scf and nscf calculations.

Best wishes,
Haozhe

Dear Haozhe,

I was not able to reproduce your error. I did a test reducing a lot the QE parameter to speed up the calculation.
The results I get are reasonable.

What I did in QE:
Reduce the parameter as.
ecutwfc = 20 Ry
use the default diagonalization and not "cg"
k points 5 5 1
in nscf I considered 70 bands.

input and output of Yambo are attached.

I suggest you:
1) be sure to have the last release of Yambo
2) Try to reproduce the minimal test I did and see if you get weird results or not.

Best,
Daniele

Dear Daniele,

Thanks for your suggestions and patient help. I upgraded my yambo to the latest version, and I got the reasonable results.
By the way, back to the polarizability alpha_2D, I noticed that using different number of bands (BSEBands) in the BSE calculation (in my case, i tested 1 valence band+1 conduction band, 2 valence bands+2 conduction bands and 3 valence bands+3 conduction bands) , I got the same exciton energy, which might mean that the calculation reached convergence. But alpha_2D in the o.alpha_q1_diago_bse for those three calculations are very different.
So does this mean that the alpha_2D in BSE calculation only comes from the contribution of the bands (BSEBands) we considered?

Best wishes,
Haozhe

Dear Haozhe,

the real part of the polarisability has a lower convergence of the Imaginary part. You can realise it looking at the Kramers-Kroning relation between real and imaginary part. The static value of the real part depends on all the energy range of the imaginary part.

Best,
Daniele

Dear Daniele,

Thanks a lot for your explanations.
Is there any way we can extract the screened coulomb interaction W （like Extended Data Fig. 1 of paper Nature Nanotechnology volume 15, 367–372 (2020)） using yambo or yambopy?

Best wishes,
Haozhe

Dear Haozhe,

this can be done reading the nbd.em1s and ndb.cutoff databases, and yes can done using yambo-py.
Best,
Daniele