Hunt for Dark Subhalos in the Galactic Stellar Field Using Computer Vision

The nature of dark matter remains uncertain despite several decades of dedicated experimental searches. The lack of tangible evidence for its non-gravitational interactions with ordinary matter gives good motivation for exploring new avenues of inferring its properties through purely gravitational probes. In particular, addressing its small-scale distribution could provide valuable new insights into its particle nature, either confirming the predictions of cold dark matter hypothesis or favouring models with suppressed small-scale matter power spectrum. In this work a machine learning technique for constraining the abundance of DM subhalos through the analysis of galactic stellar field is proposed. While the study is the first of its kind and hence applied only to a simplified synthetic datasets, the obtained results show promising potential for addressing the amount of DM substructure present within Milky Way. Using accurate astrometric observations, which became available only recently and are expected to rapidly improve in the near future, there is good hope to reach sensitivity needed for detecting DM subhalos with masses down to $10^7 M_\odot$.