"Mapping the Milky Way’s Chemistry with The Cannon: Applications to Gaia RVS, GALAH DR4, and Gaia-ESO"

Abstract

Modern spectroscopic surveys—ranging from millions of stars in LAMOST to billions in Gaia XP—are transforming our view of the Milky Way, yet deriving precise stellar parameters from low- to moderate-resolution spectra remains challenging. The Gaia RVS dataset, eventually expected to cover over 200 million stars, offers an unprecedented opportunity to map Galactic structure with precise stellar labels. We applied The Cannon, a supervised, data-driven algorithm, to transfer parameters and abundances from high-resolution GALAH DR4 spectra to lower-resolution RVS spectra, without relying on predefined line lists or stellar models. For over 800,000 RVS stars, we determined Teff, log g, [Fe/H], and multiple α-element abundances, recovering the high-α/low-α bimodality in the Milky Way disc. Extending to neutron-capture elements, we also developed a framework using The Cannon for ~350,000 RVS giants with a chemical-based Membership Probability Function that identifies likely members of accreted populations such as Gaia-Sausage-Enceladus. We further apply The Cannon to the Gaia-ESO Survey, using high-resolution UVES spectra to train the model and infer labels for lower-resolution FLAMES/GIRAFFE spectra, enabling consistent multi-element abundance scales across instruments and paving the way towards a unified, data-driven chemical map of the Milky Way.