Abstract: Upcoming astronomical surveys, such as the ESA Euclid mission, will provide us with unprecedentedly large cosmological datasets, potentially capable of revolutionizing our understanding of the Universe. However, the challenges posed by the size of these datasets dangerously hinder the feasibility of their analysis within a rigorous statistical framework for uncertainty propagation, such as the one provided by Bayesian inference.
In my talk, I will present COSMOPOWER, an open-source Python framework for Bayesian inference from next-generation Cosmic Microwave Background (CMB) and Large-Scale Structure (LSS) surveys. COSMOPOWER provides orders-of-magnitude acceleration to the Bayesian inference pipeline by training Deep Learning emulators of key cosmological quantities. I will show how these emulators meet the accuracy requirements for application to both currently available cosmological data, such as from the Kilo-Degree Survey (KiDS), as well as to simulated, next-generation data from e.g. a Euclid-like survey. The emulators always recover the fiducial cosmological constraints, while providing a speed-up factor up to O(10^4) to the complete inference pipeline. Bayesian parameter contours can thus be recovered in just a few seconds on a common laptop, as opposed to the many hours, days or months of runtime on computer clusters required by standard methods. I will also show a recent application of COSMOPOWER to derive constraints on an interacting dark energy model from the latest weak lensing data release from the KiDS survey. I will conclude with an outlook on extensions of COSMOPOWER that are currently being developed to solve other long-standing problems in the analysis of current and future cosmological data.