Different Fermi energies between GW and BSE

[nicholas_dimakis1]

Hello

I have a 4x4 WSe2 monolayer and I am doing G0W0 and BSE. BSE gives an unusual fermi energy as follows

LMDQQFR6FC7M:BSE dpm164$ grep fermi OUTCAR
ISMEAR = 0; SIGMA = 0.01 broadening in eV -4-tet -1-fermi 0-gaus
E-fermi : -2.7680

whereas GW shows that

LMDQQFR6FC7M:GW dpm164$ grep fermi OUTCAR
ISMEAR = 0; SIGMA = 0.01 broadening in eV -4-tet -1-fermi 0-gaus
E-fermi : 0.3060

For ALGO= Exact I get

LMDQQFR6FC7M:exact dpm164$ grep fermi OUTCAR
ISMEAR = 0; SIGMA = 0.01 broadening in eV -4-tet -1-fermi 0-gaus
E-fermi : -0.5996 XC(G=0): -5.1845 alpha+bet : -4.9107

So the question is why BSE gives a so different Fermi energy than GW.

Thank you-Nick

[ferenc_karsai]

In the ALGO=exact step the fermi energy is not shifted to the midst of the bandgap.

In the GW step the fermi energy is obtained from the WAVECAR file before the GW calculation. This step should yied the same fermi energy as ALGO=exact, but inn contrast to the ALGO=exact step the fermi energy is shifted to the mid bandgap.

In the BSE step the fermi energy is obtained from the WAVECAR at the beginning of the calculation before the BSE calculations are done. This fermi energy is the one after updating the eigenvalues in GW.