APOGEE Net: Improving the Derived Spectral Parameters for Young Stars through Deep Learning

Abstract

Machine learning allows efficient extraction of physical properties from stellar spectra that have been obtained by large surveys. The viability of ML approaches has been demonstrated for spectra covering a variety of wavelengths and spectral resolutions, but most often for main sequence or evolved stars, where reliable synthetic spectra provide labels and data for training. Spectral models of young stellar objects (YSOs) and low mass main sequence (MS) stars are less well-matched to their empirical counterparts, however, posing barriers to previous approaches to classify spectra of such stars. In this work we generate labels for YSOs and low mass MS stars through their photometry. We then use these labels to train a deep convolutional neural network to predict log g, T ef f and Fe/H for stars with APOGEE spectra in the DR14 dataset. This "APOGEE Net" has produced reliable predictions of log g for YSOs, with uncertainties of within 0.1 dex and a good agreement with the structure indicated by pre-main sequence evolutionary tracks, and correlate well with independently derived stellar radii. These values will be useful for studying pre-main sequence stellar populations to accurately diagnose membership and ages.

A gravitational and dynamical framework of star formation: The Orion Nebula

“…Such measurement is a more direct method of determining an age of a star, as it is significantly less biased by multiplicity. Recently, Olney et al (2020) have performed analysis of APOGEE spectra to use data-driven techniques to measure calibrated T ef f and log g.…”

APOGEE Net: An expanded spectral model of both low mass and high mass stars

“…APOGEE Net I (Olney et al 2020) attempted to build on the efforts from the Payne, supplementing the available parameters for intermediate mass dwarfs and the red giants with T eff , log g, and [Fe/H] derived from photometric relations and the theoretical isochrones for the M dwarfs and the pre-main sequence stars. This combination of the parameters was used to train a neural network that is capable of deriving stellar properties for APOGEE spectra for all stars with T eff <6700 K in a homogeneous manner.…”

“…Although the model architecture from Olney et al (2020) that operated only on the spectra could perform well on the full data sample, here we explore adding metadata for each star to further improve the performance. This metadata consists of 2MASS and Gaia EDR3 broadband photometry (G, BP, RP, J, H, K), as well as parallax (Cutri et al 2003;Gaia Collaboration et al 2021).…”

“…We generate the uncertainties in the parameters in a manner similar as that described in Olney et al (2020). We used the uncertainties in the spectral flux that is reported in the DR17 apStar file, and created multiple realization of the same spectrum.…”

“…Previously, we have developed a neural network, APOGEE Net, which self-consistently measured T eff , log g, and [Fe/H] for cool (T eff <6500 K) stars, including more exotic and rare objects, such as pre-main sequence stars (Olney et al 2020). In this work, we extend its functionality to estimate these parameters for all stellar objects, ranging from early O to late M types.…”