Closing the stellar labels gap: Stellar label independent evidence for [$\alpha /M$] information in Gaia BP/RP spectra

Abstract

Data-driven models for stellar spectra which depend on stellar labels suffer from label systematics which decrease model performance: the "stellar labels gap". To close the stellar labels gap, we present a stellar label independent model for Gaia BP/RP spectra. We develop a novel implementation of a variational auto-encoder, which learns to generate an XP spectrum and accompanying 'scatter' without relying on stellar labels. We demonstrate that our model achieves competitive XP spectra reconstructions in comparison to stellar label dependent models. We find that our model learns stellar properties directly from the data itself. We then apply our model to XP/APOGEE giant stars to study the [$\alpha$/M] information in Gaia XP. We provide strong evidence that the XP spectra contain meaningful [$\alpha$/M] information by demonstrating that our model learns the $\alpha$-bimodality, without relying on stellar label correlations for stars with $T_{\mathrm{eff}}<$ 5000 K, while also being sensitive to the anomalous abundances of Gaia-Enceladus stars. We publicly release our trained model, codebase and data. Importantly, our stellar label independent model can be implemented for any/all XP spectra because our model performance scales with training object density, not training label density.

arXiv comment20 pages, 15 figures, published in ApJ. Code available at https://zenodo.org/records/14041979. Data available at https://zenodo.org/records/14041773

References in corpus (30)

• Gaia Data Release 3: Summary of the content and survey properties
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The Apache Point Observatory Galactic Evolution Experiment (APOGEE)
• The merger that led to the formation of the Milky Way's inner stellar halo and thick disk
• Chemical Cartography with APOGEE: Metallicity Distribution Functions and the Chemical Structure of the Milky Way Disk
• The GALAH+ Survey: Third Data Release
• The spatial structure of mono-abundance sub-populations of the Milky Way disk
• The Cannon: A data-driven approach to stellar label determination
• Origin and structure of the Galactic disc(s)
• Gaia FGK Benchmark Stars: Effective temperatures and surface gravities
• The Milky Way has no thick disk
• The Payne: self-consistent ab initio fitting of stellar spectra
• The GALAH survey and Gaia DR2: dissecting the stellar disc's phase space by age, action, chemistry and location
• Gaia Data Release 3: Processing and validation of BP/RP low-resolution spectral data
• Chemodynamics of the Milky Way. I. The first year of APOGEE data
• Tracing chemical evolution over the extent of the Milky Way's Disk with APOGEE Red Clump Stars
• Deep learning of multi-element abundances from high-resolution spectroscopic data
• Gaia Data Release 3: External calibration of BP/RP low-resolution spectroscopic data
• Parameters of 220 million stars from Gaia BP/RP spectra
• The vertical motions of mono-abundance sub-populations in the Milky Way disk
• The AMBRE project: The thick thin disk and thin thick disk of the Milky Way
• Overview of the LAMOST survey in the first decade
• Gaia Data Release 3: The Galaxy in your preferred colours. Synthetic photometry from Gaia low-resolution spectra
• The Gaia-ESO Survey: Characterisation of the [alpha/Fe] sequences in the Milky Way discs
• Towards an astronomical foundation model for stars with a Transformer-based model
• Spectroscopic study of CEMP-(s & r/s) stars- Revisiting classification criteria and formation scenarios, highlighting i-process nucleosynthesis
• Cycle-StarNet: Bridging the gap between theory and data by leveraging large datasets
• Carbon-Enhanced Metal-Poor star candidates from BP/RP Spectra in $Gaia$ DR3
• Information content of BP/RP spectra in Gaia DR3
• Estimating $\left[\alpha /\text{Fe}\right]$ from Gaia low-resolution BP/RP spectra using the ExtraTrees algorithm

Cited by in corpus (3)

• An Unsupervised Machine Learning Approach to Identify Spectral Energy Distribution Outliers: Application to the S-PLUS DR4 data
• Trustworthy scientific inference with generative models
• J-PLUS: Identifying Outlier Spectral Energy Distributions in the Fourth Data Release Using Unsupervised Machine Learning