Effect of quasiparticle self-consistent schemes on the GW method with Bloch Gaussian orbitals

The GW method is one of the simplest, yet non-trivial, approximations to the self-energy of a many-electron system, which has been shown to be successfull on a wide variety of systems. For example, it can accurately predict the band gaps of many weakly correlated systems (solids and molecules) and is often chosen as the low-level theory for more sophisticiated embedding methods such as the dynamical mean-field theory, or the self-energy embedding theory. However, the result in a GW calculation can be very sensitive to the choice of basis set, whether or not relativistic effects are to be considered, as well as the type of self-consistency or quasiparticle approximations that are used (or if they are used). In this talk, we will compare band-gaps for insulators and semi-conductors using various self- consistent quasiparticle GW schemes against the one-shot G0W0 and fully self-consistent results. Interestingly, we find that the trends in the accuracy of the quasiparticle schemes that have been observed in previous studies (using plane waves or linear augmented plane wave basis sets) does not necessary translate to Gaussian orbitals. We will also touch upon the effect of other factors, such as relativity, temperature, etc., on these findings.