The Gaia-ESO Survey: Chemical evolution of Mg and Al in the Milky Way with machine learning^⋆

¹ Institute of Theoretical Physics and Astronomy, Vilnius University, Saulėtekio Av. 3, 10257 Vilnius, Lithuania
e-mail: markus.ambrosch@ff.vu.lt
² Max Planck Institute for Astronomy, Königstuhl 17, 69117 Heidelberg, Germany
³ Leibniz-Institut für Astrophysik Potsdam (AIP), An der Sternwarte 16, 14482 Potsdam, Germany
⁴ Institut für Physik und Astronomie, Universität Potsdam, Karl-Liebknecht-Str. 24/25, 14476 Potsdam, Germany
⁵ Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK
⁶ INAF – Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Florence, Italy
⁷ Lund Observatory, Division of Astrophysics, Department of Physics, Lund University, Box 43 22100 Lund, Sweden
⁸ European Southern Observatory, Karl Schwarzschild-Straße 2, 85748 Garching bei München, Germany
⁹ Instituto de Física y Astronomía, Facultad de Ciencias, Universidad de Valparaíso, Av. Gran Bretaña 1111, Valparaíso, Chile
¹⁰ Niels Bohr International Academy, Niels Bohr Institute, University of Copenhagen Blegdamsvej 17, 2100 Copenhagen, Denmark
¹¹ Nicolaus Copernicus Astronomical Center, Polish Academy of Sciences, ul. Bartycka 18, 00-716 Warsaw, Poland
¹² INAF – Padova Observatory, Vicolo dell’Osservatorio 5, 35122 Padova, Italy
¹³ Núcleo de Astronomía, Universidad Diego Portales, Ejército 441, Santiago, Chile
¹⁴ Departmento de Astrofísica, Centro de Astrobiología (CSIC-INTA), ESAC Campus, Camino Bajo del Castillo s/n, 28692 Villanueva de la Canada, Spain

Received: 18 August 2022
Accepted: 25 January 2023

Abstract

Context. To take full advantage of upcoming large-scale spectroscopic surveys, it will be necessary to parameterize millions of stellar spectra in an efficient way. Machine learning methods, especially convolutional neural networks (CNNs), will be among the main tools geared at achieving this task.

Aims. We aim to prepare the groundwork for machine learning techniques for the next generation of spectroscopic surveys, such as 4MOST and WEAVE. Our goal is to show that CNNs can predict accurate stellar labels from relevant spectral features in a physically meaningful way. The predicted labels can be used to investigate properties of the Milky Way galaxy.

Methods. We built a neural network and trained it on GIRAFFE spectra with their associated stellar labels from the sixth internal Gaia-ESO data release. Our network architecture contains several convolutional layers that allow the network to identify absorption features in the input spectra. The internal uncertainty was estimated from multiple network models. We used the t-distributed stochastic neighbor embedding tool to remove bad spectra from our training sample.

Results. Our neural network is able to predict the atmospheric parameters T_eff and log(g) as well as the chemical abundances [Mg/Fe], [Al/Fe], and [Fe/H] for 36 904 stellar spectra. The training precision is 37 K for T_eff, 0.06 dex for log(g), 0.05 dex for [Mg/Fe], 0.08 dex for [Al/Fe], and 0.04 dex for [Fe/H]. Network gradients reveal that the network is inferring the labels in a physically meaningful way from spectral features. We validated our methodology using benchmark stars and recovered the properties of different stellar populations in the Milky Way galaxy.

Conclusions. Such a study provides very good insights into the application of machine learning for the analysis of large-scale spectroscopic surveys, such as WEAVE and 4MOST Milky Way disk and bulge low- and high-resolution (4MIDABLE-LR and -HR). The community will have to put substantial efforts into building proactive training sets for machine learning methods to minimize any possible systematics.