This Innovation Study aims to empower the computation of systems containing up to 10,000 atoms by focusing on Moiré superlattices. Grounded in the standard of GW methods, Exa4GW aims to enhence the capabilities of band structures and alignments at materials interfaces. Exa4GW focuses on reducing the computational prefactor through an efficient treatment of divergent terms at the Γ-point in the Brillouin zone.
The research in this study will prioritize working with leading LBS (Low-scaling and low-Bandwidth Scaling) codes, particularly FHI aims and CP2K. These codes adopt different strategies for evaluating polarizability, driven by considerations of applicability to deep core (FHI-aims) or valenceχ excitations (CP2K). This collaborative approach ensures a comprehensive exploration of diverse materials and electron.
Exa4GW project incorporates a comprehensive GPU offloading. By exploiting the parallel processing capabilities of GPUs, the goal is to significantly accelerate the computation of intricate electronic structures, enhancing speed and efficiency in addressing complex materials science challenges.
Another aspect of the Exa4GW project is the publication of the W-engine as a component of the GreenX library. This ensures that the developments achieved through the project will be accessible to a broader scientific community, supporting collaboration and knowledge exchange.
By addressing key challenges through efficient treatments, GPU offloading, and collaborative efforts with prominent LBS codes, Exa4GW not only advances upscaling of algorithms for the Inno4scale project but also ensures the accessibility of its innovations to the broader research community.