2025 workshop program
CLiMBing the Galactic Ladder: Unveiling Hidden Structures with Semi-supervised Clustering Algorithms
The work
Abstract
RR Lyrae stars, as ancient and well-characterized stellar populations, serve as excellent tracers for unraveling the formation and evolution of the Milky Way. They help uncover old dynamical substructures and reveal the galaxy’s formation history. We introduce CLiMB (CLustering in Multiphase Boundaries), a Python-based framework designed to detect dynamical substructures within the Galactic halo by applying unsupervised clustering algorithms to integrals of motion.
CLiMB employs a two-phase clustering approach: an initial constrained phase to identify known Galactic components, followed by an exploratory phase utilizing density-based clustering algorithms to uncover novel groupings. The framework supports data from both Gaia and the upcoming LSST, enabling the analysis of large-scale stellar datasets.
In preliminary applications to RR Lyrae data, CLiMB successfully recovered the Galactic disk and identified two additional clusters, potentially corresponding to the hypothesized proto-galactic fragments Shakti and Shiva in the inner Milky Way. These findings highlight CLiMB’s potential to automate and enhance the reproducibility of Galactic substructure analyses.