Charting the Stellar Streams of the Milky Way

Abstract

With massive datasets, we can now study Galactic halo structures in unprecedented detail, incorporating kinematic information alongside positions and photometry. However, the data are low-signal, noisy, and high-dimensional, making it challenging to detect, characterize, or model these halo structures like stellar streams. In this thesis I tackle these challenges by employing statistical approaches to astrophysical observations and developing novel data-driven methods to characterize and model stellar streams of the Galaxy. I also release open-source software to reproduce the work in this thesis. Stellar streams encode information about their progenitor systems and the Galaxy they orbit. Using Gaia data, including kinematics, I extend the detected Pal 5 stream’s tidal tails to approx- imately 30 degrees, with a newly detected 7-degree segment in the leading arm. This extended detection provides insights into the stream’s interactions with the Galactic bar and the Galaxy’s gravitational potential. Following this detection, I develop novel methods for characterizing stellar streams. One method, using machine learning and time-series analysis, constructs parametric stream paths. This method accounts for measurement errors and data sparsity, is independent of Galactic models, and is applicable to phase-wrapped streams. I develop another method for characterizing observed stellar streams, combining Bayesian models and machine learning. This method simultaneously operates on astrometric and photometric data to detect streams in very low signal-to-noise fields and provides a comprehensive characterization of the stream’s properties, independent of Galaxy models. It handles incomplete phase-space observations. I apply this method to GD-1 and Pal 5 to start building a homogeneous catalog of stellar membership probabilities. Finally, I develop a probabilistic framework to constrain the Galactic potential using the developing catalog of stellar stream observations. By comparing observed stream data to forward models, my methodology aims to identify the underlying parameters of the Galaxy’s potential and its orbiting clusters. This framework already provides informative constraints on components such as halo mass, triaxiality, and Galactic bar rotation, laying the foundation for expanding the catalog to improve constraints and sensitivity to additional Galactic components.