VASP GW with Silicon

[Redorb]

Hello,

I am attempting to preform a GW calculation with VASP 5.2.8. I preform a static calculation and then do a bandstructure calculation ICHARG=11 in a new directory from the CHGCAR that was generated by the static run. This gives me a bandgap of 0.5412 eV. NEXT. I replace the INCAR file in the static run's directory by another INCAR file that will do the GW calculations. I then take the CHGCAR from this directory and run a bandstructure ICHARG=11 calculation on it. This leads to a bandgap of 0.5413 eV. The question is what is wrong? Why am I only getting an bandgap increase of 0.0001 eV? In my static and GW calculations I am using an automatically generated kpoint-grid of 3X3X3 gamma. The bandstructure from the static run and the gw run look identical when plotted against one another.

The POTCAR is the PBE Si POTCAR that comes with VASP 5.2

The POSCAR for these runs is (This is the primitive cell of Si):
Basis file from gf2k2vasp
3.8395898218430
0.707106781186547 0.707106781186547 0.000000000000000
0.000000000000000 0.707106781186547 0.707106781186547
0.707106781186547 0.000000000000000 0.707106781186547
2
Cartesian
0.000000000000000 0.000000000000000 0.000000000000000
0.353553390593274 0.353553390593274 0.353553390593274

INCAR for this static run is:
System = Si
NBANDS = 96
ISMEAR = 0; SIGMA = 0.05
LOPTICS = .TRUE.

INCAR for the GW run is:
System = Si
NBANDS = 96
ISMEAR = 0; SIGMA = 0.05
ALGO = GW0 ; NOMEGA = 60
NELM=4
LSPECTRAL=.TRUE.

KPOINTS file for the GW and static run is:
Gamma 3X3X3 kpoints (useful for hexagonal lattices)
0
Gamma
3 3 3
0 0 0

Thank you in advance for any advice you can give!

[tilak]

The first thing is that this 0.0001 eV is not any more increment of gap due to GW calculations than the normal static calculated gap!


Now what i find out in you post that:

"I then take the CHGCAR from this directory and run a bandstructure ICHARG=11 calculation on it."

it means you asked for the GW calcualtions but do not allow for updating of the self-energy correction on the normal DFT calculated eigenvalues !!!! So finally you do not have the GW calculated gap.