AI Science Center Seminar: AI and HPC: Powering the Next Breakthroughs in Cosmology

30.09.2026 10:30 – 12:00

Artificial intelligence and high-performance computing are creating a profound change in discovery methodologies in physical sciences, unlocking possibilities for answering new science questions, which would otherwise be out of reach. The field of large scale structure cosmology is currently undergoing this transformation, with AI-powered simulation-based inference methodologies delivering +100% more precise measurements than classical approaches. Thus, we will be able to differentiate between complex models of dark matter, dark energy, modified gravity, accelerating discoveries from limited astrophysical data. The upcoming mega-telescopes, such as European Space Agency’s Euclid satellite, and the upcoming radio Square Kilometer Array (SKA), will provide images of our universe over 10 billion years of cosmic history, and enable us to study its evolution with unprecedented level of detail. Conversely, new hyperscaler computing centers, such as CSCS Alps, Jülich JUPITER, and Isambard-AI, will enable the production of high-resolution simulated theory prediction spanning large space of cosmological models. In this talk, I will review the current and future applications of HPC and AI in cosmology, focusing on areas where they already play an indispensable role: simulation-based inference, generative AI for simulations. In particular, I will show how these ideas are already delivering results in practice for Dark Energy Survey and Euclid. I will also present a number of yet-unsolved problems in data science, the solutions to which are critical for enabling next generation measurements in cosmology.

Lieu

Bâtiment: Battelle

Room 404, Building A

entrée libre