“Strong gravitational lensing: a machine-learning perspective
Abstract
We are experiencing a period of deep and crucial changes in astronomy. As the first light of next-generation facilities approaches, the need to efficiently analyse unprecedented amounts of data becomes more and more compelling. Strong gravitational lensing represents a perfect example of this ongoing process. After decades in which the low number of discovered lenses has strongly affected the potential of lensing-based science, the ESA’s Euclid telescope and the Vera Rubin Observatory are expected to observe more than 100.000 strongly lensed systems. But – now – new questions arise. How can we search for rare lensing configurations in datasets including billions of galaxies? And how can we efficiently model hundreds of thousands of lenses when classical techniques can require up to several hours to analyse a single lensed system? In brief, are we ready to exploit the full scientific potential of these new instruments? This talk will focus on how machine-learning techniques can represent a promising answer to these key questions. I will present our latest results in the field of the automatic research and analysis of strong gravitational lenses, the different algorithms that we developed, and the future perspectives of our ongoing projects.”