Yambo Community Forum

Dear developers,

I performed GW calculations for a two-dimensional magnetic material with both ONCV and DOJO pseudopotentials. The DFT band structure and QP corrections were apparently the same for both pseudopotentials (figure 1,2). But when I did the GW-BSE calculations, the absorption spectrum obtained from the ONCV pseudopotential was completely correct, while the absorption spectrum obtained from the DOJO pseudopotential was wrong (figure 3). However, when I performed GW-BSE calculations with scissor for DOJO pseudopotentials, the correct absorption spectrum was obtained (figure 4).
Is there a way to solve this problem? Because I want to perform GW-BSE calculations for a material whose scf calculations do not converge with ONCV pseudopotentials, while they converge easily with DOJO pseudopotentials.
The r-setups of GW-BSE calculations with full GW for both pseudopotentials are attached.

Thanks and Regards,
Mitra Helmi

Dear Mitra,

at first glance there are some differences between the two calculations:
1) Different number of k points
2) Slightly different lattice constants
3) Different number of QP corrections
4) Different QP gap (about 0.1 eV of difference)

As a first test, I would compare the two calculations using the same parameters.
Looking at your QP corrections, it seems that the scissor operator approximation here does not apply.

Best,

Daniele