Exascale Lensing Engine for Gravitational Astrophysics and Numerical Cosmology Investigations – ELEGANCI

The proposed research aims at designing and implementing a high-performance raytracing engine for General Relativity, addressing the computational needs of upcoming astronomical surveys. By leveraging massively parallel GPU architectures, the engine will enable precise simulations of relativistic effects including gravitational lensing and redshift perturbations, and offer new insights into the large scale structure of the Universe. The project represents a significant interdisciplinary effort to bring together cosmology, high-performance computing, programming language research, and software architecture in order to create a scalable, efficient, and versatile foundational tool to shape the future of observational astronomy. For the first time, the raytracing engine will make it possible to accurately simulate the propagation of photons through large scale cosmic structures, enabling the fine analysis of their trajectories, and allowing the production of realistic images of simulated skies to feed observational analysis workflows and machine-learning algorithms. This undertaking will be of critical importance for Euclid and the Vera Rubin Observatory that are going online this year, and whose collaborations are currently relying on convoluted and approximated methods to compensate for the lack of out-of-the box algorithms to integrate Einstein's equations directly. Beyond its scientific impact, the project will serve as a starting point for the standardization of differential equation solvers for the C++ language, and has a lot of potential for public outreach in terms of visualization.