Unsupervised Kinematic Dissection of the Solar Neighborhood: Identifying Stellar Moving Groups with Gaia DR3

TL;DR

This paper uses Gaia DR3 data and unsupervised clustering to identify and analyze stellar moving groups in the solar neighborhood, revealing their kinematic structures without prior assumptions.

Contribution

It introduces an unsupervised machine learning method to detect stellar moving groups using full phase space data, advancing the mapping of local Milky Way kinematic substructures.

Findings

Successfully recovers major local moving groups like Hyades and Pleiades

Demonstrates robustness of clustering algorithms in dynamical history analysis

Provides a catalog of high-probability stellar members for future studies

Abstract

We present a comprehensive kinematic analysis of the solar neighborhood (d < 50 pc) using high-precision astrometric data from the third Gaia Data Release (DR3). By leveraging the full six dimensional phase space information (positions and velocities), we apply the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to blindly identify stellar overdensities in the Galactocentric Cartesian velocity space (U, V, W ). Our unsupervised machine learning approach successfully recovers the kinematic cores of major local moving groups, including the Hyades and Pleiades streams, without prior assumptions regarding their membership or spatial distribution. We analyze the velocity dispersion and structural properties of these associations, demonstrating that automated clustering algorithms are robust tools for mapping the complex dynamical history of the local Milky Way disk. These results confirm the hierarchical nature of stellar kinematic substructures and provide a catalog of high-probability members for future spectroscopic follow-up.