This project aims to address a major gap in ab initio calculations by enabling reliable simulations of excited electronic states (GW method) using the Projector Augmented-Wave (PAW) approach. These advances will be integrated into the open-source software ABINIT, a recognized international collaborative project. The GW approximation is considered the gold standard for determining electronic energy levels in condensed matter, correcting the underestimated band gap in DFT. The PAW method, on the other hand, offers precision and flexibility and is widely used for ground state and material response calculations.
However, the combined GW+PAW approach encounters difficulties in some well-identified cases (e.g., zinc oxide), with underlying reasons understood but not yet fully resolved. Low-energy excited states are well described, but high-energy states remain problematic. The current debate focuses on the need to perform complete (but computationally expensive) calculations, to neglect certain terms (with complex error control), or to modify the PAW method (at the cost of reduced efficiency).
The project aims to adapt the PAW formalism to the GW approach, to develop a fast and accurate numerical scheme, and to clarify the current, somewhat confusing situation. The CEA team is a leading developer of ABINIT for PAW and GW and will ensure access to large computing resources. The postdoctoral objectives include theoretical development, implementation in ABINIT, and improving electronic properties for realistic solid systems (surfaces, semiconductor junctions, etc.).