Nathaniel Starkman

Date: Tue. February 24th, 2026, 11:30 am-12:30 pm
Location: Foldy room

Title: Computationally Unscrambling EGGS (Extra-Galactic Gravity from Streams)

Abstract:

Wide-field surveys like Euclid mark a new era of extragalactic stellar stream studies, with exciting applications in measuring baryon and dark matter distributions, and connecting galaxies to their cosmological context. In this talk, I present Extra-Galactic Gravity from Streams (EGGS), a computational program that uses projected stream morphologies to infer the shape, orientation, and barycenter of extragalactic potentials on sub-virial scales. We first introduce PhaseCurveFit, a machine-learning framework that reconstructs smooth, dynamically consistent stream tracks from noisy imaging data, providing the fundamental observable for dynamical inference. We then describe a new analytic method, which links stream orientation and convexity to constraints on halo flattening and triaxiality, enabling a cosmological test of dark matter models. Using further methodological advances, in a Euclid Key Project pilot study combining Euclid imaging with citizen science classifications, we demonstrate both stacked population-level constraints and individual-halo inference, with early results consistent with CDM expectations and complementary to weak lensing. Finally, we develop a physics-informed machine-learning framework that fuses analytic potentials with neural residual fields to achieve high-fidelity, interpretable reconstructions of galaxy potentials and their time evolution. Together, these advances establish stellar streams as promising dynamical probes of baryonic structure and dark matter beyond the Milky Way.