Measuring dark matter’s properties in strong gravitational lenses using machine learning

Dark matter’s microphysical properties are imprinted on the structure of the universe at subgalactic scales. Images of strong lenses, where the light from a distant galaxy is dramatically distorted into a ring by the mass of an intervening galaxy, provide an avenue for probing dark matter clumps at these scales via their purely-gravitational effects. However, the analysis of such images is a daunting task, requiring modeling myriad dark matter clumps, the mass distribution in the lens galaxy, and the detailed light distribution of the source. In this colloquium I will share how new machine learning techniques are poised to unlock this high-dimensional inference problem, paving the way for constraining dark matter’s nature with upcoming observational data.