Cosmology today is facing two fundamental problems: we do not know for certain what is the present expansion rate of the Universe, or why that expansion seems to be accelerating. In order to address these challenges, new telescopes and instruments are becoming available. One of these instruments is J-PAS, which at the end of 2023 started to map the Universe using 56 narrow optical bands, detecting $\sim 4. \times 10^4$ extragalactic objects in each square degree. Massive new data sets such as J-PAS can be quite challenging to digest, and Machine Learning (ML) has become central in our efforts to detect, classify and to extract properties of astrophysical sources. I will discuss how ML has allowed us to advance in source classification, determining which ones are stars, galaxies or quasars, paving the way for J-PAS to become one of the most complete galaxy surveys at $z \lesssim 1$, and the best quasar survey at $z \lesssim 4$. I will also show that, on the theoretical side, numerical simulations together with ML techniques have allowed us to reproduce the intricate relationships between halos and galaxies with unprecedented accuracy, which is what will ensure that we can realize the potential of these amazing new observations in terms of a new understanding about the evolution of our Universe.

Zoom:  https://rediris.zoom.us/j/94907029675?pwd=amVxUmxRSldweXZLaGFkYWdMdU9JUT09

ID: 949 0702 9675

Password:  254951

Youtube: https://youtube.com/live/JE-rCBMhpVQ?feature=share