Astronomers detect nearly 200,000 candidate metal-poor stars

Metal-poor stars are rare objects, as only a few thousand stars with iron [Fe/H] abundances below -2 have been discovered so far.

By analyzing data from various astronomical surveys, astronomers have detected 188,002 metal-poor candidate stars. The discovery, reported in a research paper published March 30 on the arXiv preprint server, may help us better understand how the universe chemically evolved. Metal-poor stars are rare objects, as only a few thousand stars with iron [Fe/H] abundances below -2 have been discovered so far. Currently, SMSS J0313–6708, with a metallicity below -7.3, is the most metal-poor star known to date.

Astronomers are interested in expanding the still short list of metal-poor stars, as these objects have the potential to improve our understanding of the chemical evolution of the universe. The early evolution of the universe is believed to depend on the properties of the first generation of metal-free stars.

Low-resolution blue and red photometer spectra (BP/RP or XP spectra) for 210 million stars have recently been published with Data Release 3 (DR3) from ESA’s Gaia satellite. A team of astronomers led by Yupeng Yao of the University of Chicago decided to analyze this data set, which provides an opportunity to greatly increase the number of metal-poor star candidates. Their study was supplemented with data from the Large Area of ​​Sky Multi-Object Fiber Spectroscopic Telescope (LAMOST) survey and the Apache Point Observatory Galactic Evolution Experiment (APOGEE). “In this work, we train XGBoost models to identify metal-poor stars in Gaia DR3. The input to the models are the reddening-free and normalized XP spectral coefficients… We use the reddening-free and normalized XP spectral coefficients together with their corresponding [Fe/H] from LAMOST or APOGEE to compose training and test sets for training the XGBoost model to identify metal-poor stars in Gaia DR3,” the researchers explained.

XGBoost is a powerful and flexible algorithm that has been used in a variety of subfields of astrophysics. Using the XGBoost classification algorithm, the astronomers obtained three corresponding catalogs of metal-poor candidate stars. In total, the team managed to identify 127,096 bright stars and 60,906 faint metal-poor star candidates in the Milky Way galaxy. The researchers noted that the total number of metal-poor candidate stars they obtained is about an order of magnitude higher than in previous studies. According to the article, bulge and halo stars are the dominant objects in the entire sample. The astronomers estimate that around 84,200 stars out of the 188,002 identified candidates are expected to be genuinely metal-poor stars, representing an overall purity of 45%.

source: https://arxiv.org/abs/2303.17676