High throughput GW

Many technologically important material properties directly relate to their electronic structure. Computational screening studies searching for new materials with specific prop- erties hence rely on methods that accurately predict electronic structure and do so in automatic manner. For solids the GW-method is currently one of the most reliable ab ini- tio approaches to predict electronic structure. It is mostly used to perturbatively correct density functional theory results from fundamental shortcomings and errors introduced by the approximate exchange correlation functional. It is however computationally demanding and requires much expertise to perform accurately. For its applications in high throughput screenings hence sophisticated algorithms for setting up the calculations and determining convergence are needed. In this work we develop such a method and, as a first application, use it to validate the accuracy of G0W0 using the PBE starting point (G0W0@PBE) and the Godby-Needs plasmon pole model on a set of about 100 solids. The automatic conver- gence screening utilized in this works provides valuable insights, e.g., correlations between computational parameters, that can be used to further improve the automatization of GW calculations. The large scale validation of G0W0@PBE shows a correlation between the PBE and the G0W0@PBE gaps that is much stronger than that between GW and exper- imental gaps. With this work we hence show that GW can be made automatic but that for accurate predictive results for a broad class of materials an improved starting point or form of self-consistency is necessary