An Enhanced Sample of Galactic Red Supergiants Reveals Spiral Structures

TL;DR

Addressing the scarcity of Galactic red supergiants (RSGs), the study leverages Gaia XP spectra and a feedforward neural network to assign probabilities $P(\mathrm{RSG})$ to each spectrum. Ten independent training runs across randomly split data produce catalogs, with true RSGs defined as objects appearing in at least eight catalogs and requiring a threshold of $P(\mathrm{RSG}) \ge 0.9$, yielding 2,436 RSGs. The resulting sample exhibits a clear spatial correlation with OB stars and traces the Milky Way's spiral arms, validating the method and highlighting RSGs as robust tracers of Galactic structure. This work provides a large, extinction-resilient RSG catalog for comprehensive studies of Galactic morphology and the physics of massive stars.

Abstract

Red supergiants (RSGs), representing a kind of massive young stellar population, have rarely been used to probe the structure of the Milky Way, mainly due to the long-standing scarcity of Galactic RSG samples. The Gaia BP/RP spectra (hereafter XP), which cover a broad wavelength range, provide a powerful tool for identifying RSGs. In this work, we develop a feedforward neural network classifier that assigns to each XP spectrum a probability of being an RSG, denoted as $\mathrm{P(RSG)}$. We perform ten independent runs with randomly divided training and validation sets, and apply each run to all XP spectra of stars with $G < 12$ mag. By selecting sources with $\mathrm{P(RSG)} \geq 0.9$, ten high-confidence candidate samples are obtained. A star is considered a ture Galactic RSG only if it appears in at least eight of these samples, yielding a final catalog of 2,436 objects. These RSGs show a clear spatial correlation with OB stars and trace the Galactic spiral arms well, confirming the reliability of our classification, and highlighting their potential to serve as powerful tracers of the Milky Way's structure.

Figures (10)

• Figure 1: The Gaia CMD within 2 kpc. All the stars are marked by density, and the black dotted line represents the boundary of RSGs. The blue, yellow and red solid lines in the left panel represent the stellar evolutionary tracks of the MIST model, which are used to distinguish BSGs, YSGs and RSGs. Open triangles 2005ApJ...628..973L, squares 2019AJ....158...20M and circles 2023AA...671A.148M on the right panel represent the known Galactic RSGs from the literatures.
• Figure 2: The overall architecture of the proposed network. This plot is modified from the original image contributed by srvmshr for ML Visuals.
• Figure 3: Evolution of the loss (blue, left axis) and recall (red, right axis) for the training (dashed lines) and validation (solid lines) sets as a function of epoch for ten independent runs.
• Figure 4: Mean confusion matrix of the validation set over ten runs. Each cell shows the average count and its 1$\sigma$ uncertainty.
• Figure 5: Precision as a function of $\mathrm{P(RSG)}$. Blue lines represent ten independent runs, and the red line and shaded region show the mean and 1$\sigma$ dispersion, respectively.