Yambo Community Forum

Good evening,

I am doing GW band gap calculations using Quantum Espresso. I noticed that the energies of the bands that I consider for calculating the GW gap are different when using Yambo+Abinit and Yambo+Quantum Espresso.
Before moving to Yambo, I checked the value of the gap at the DFT level which was the same for both Abinit and QE. When moving to Yambo, I also checked the gap at the DFT level which again is the same. So, the only thing that changes are the band energies. Is this due to some differences in p2y and a2y? Or withing the DFT codes?

Sincerely,
Laura

Dear Laura,

So, the only thing that changes are the band energies. Is this due to some differences in p2y and a2y? Or withing the DFT codes?

What do you mean exactly by band energies?
p2y and a2y are just meant to read the ground state electronic structure for both the codes and rewrite in yambo format. Of course, if the overall electronic structure is different, apart from the gap, you can find different results. Anyway it seems to me quite odd that you find different electronic structure with the two codes if you use the same set up, e.g. pseudo potentials and functionals.

Best,
Daniele

Dear Laura,
Is (1) a rigid shift of the bands compatible with what you observe ?
or is (2) the shape of the bands actually modified ?

(1) is expected, since yambo shifts the DFT eigenvalues such that the KS-HOMO (or the Fermi level) is set to zero.
(2) instead would be a bug.

take care
Andrea

Dear Daniele,

I mean like this. With Quantum Espresso:

# K-point Band Eo [eV]
#
1 36 -0.754478
1 37 0.790687

With Abinit:
1 36 0.000000
1 37 1.551236

The gap is quite the same (I used a higher cut off in one of the two codes, that's why one is only slightly higher). However, the values of the energies of the bands are different, as a shift has happened.

Dear Andrea,

Sorry maybe I was answering in the meantime that you were replying. As I said, yes it is like a shift and the HOMO value is set to zero only within the Abinit framework.

Regarding the shape of the bands, you mean by checking the Density of States/band structure in DFT framework?

Hi Laura,

no worries then,
In the case of QE (at least) yambo applies a rigid shift as discussed above.
It could well be that in the case of Abinit this does not happen, or happens differently.

A

Dear Andrea,

In my case, it seems like it is the inverse (QE does not apply it and Abinit does). But however, as you said, it is a rigid shift so, if I understand, it should be fine.

Thanks for your support

Dear Laura,
the VBM in Yambo is set to zero.
Are you using the same k point grid (number and shifts) in the two codes?
Is it possible that the k-point index 1 does not correspond to the same point?
you can check in the report the corresponding kpoint and also see at what k point the direct gap is found.

Best,
Daniele