Testing Dark Matter with Generative Models for Extragalactic Stellar Streams (2508.02666v1)

Abstract: Upcoming ground and space-based surveys are poised to illuminate low surface brightness tidal features, providing a new observable connection to dark matter physics. From imaging of tidal debris, the morphology of stellar streams can be used to infer the geometry of dark matter halos. In this paper, we develop a generative approach, X-Stream, which translates stream imaging into constraints on the radial density profile of dark matter halos--from the inner region out to the virial radius. Using the GPU-accelerated code streamsculptor, we generate thousands of stream realizations in trial gravitational potentials and apply nested sampling with a custom objective function to explore viable regions of parameter space. We find that multiple stellar streams can be used to constrain the entire radial density profile of a halo, including both its inner and outer density slopes. These constraints provide a test for alternatives to cold dark matter, such as self-interacting dark matter, which predicts cored density profiles. From cosmological simulations, the outer density slope is expected to correlate with merger histories though remains underexplored observationally. With ongoing and upcoming missions such as Euclid, the Rubin Observatory, ARRAKIHS, and the Nancy Grace Roman Space Telescope, X-Stream will enable detailed mapping of dark matter for thousands of galaxies across a wide range of redshifts and halo masses.