ABINIT Discussion Forums

I have tried making paw pseudopotential files with atompaw to use in GW calculations. When I did this, I found that the calculated band gap is smaller than when using norm-conserving pseudopotentials (which may be true) and the convergence with the number of bands gives the wrong behavior, almost no change in the gap as the number of bands increases (almost certainly not true). Has anyone seen similar behavior using PAW and GW? I am assuming I'm making some simple error, but I haven't been able to find it so far. I've attached a plot using two different paw pseudopotential input files.

Questions:
1) Which crystal is under consideration?
2) For which electron configuration have been generated your PAWs?
3) Which GW-model do you use?
Best wishes,
Steve.