Machine Learning Finds a Surprising Early Galaxy

Aug 4, 2020 | Daily Space, Galaxies

Machine Learning Finds a Surprising Early Galaxy
IMAGE: Image of HSC J1631+4426 discovered by the international team with the Subaru Telescope. HSC J1631+4426 broke the record for the lowest oxygen abundance. CREDIT: NAOJ/Kojima et al.)

Sometimes, pleasing science comes from knowing your software works. 

Researchers using the National Astronomical Observatory of Japan’s (NAOJ) Subaru Telescope in Hawaii have taken amazing wide-field images in which they hope to find faint smudges of light from young galaxies. Searching these images by eye is time-consuming and subject to all kinds of human error, so this team has instead trained a neural network to search for them. 

In a new paper, appearing in The Astrophysical Journal, this team, lead by Takashi Kojima, describes a new, ultra metal-poor galaxy their software discovered. Cataloged as HSC J1631+4426, this system is tiny with the mass of only 800,000 Suns and is 430 million light-years away. While not at some great cosmological distance, this young galaxy is made of material largely leftover and unchanged from the earliest times in our universe. It appears to have only 1.6% the heavy element abundance of our Sun, and most of the stars in the galaxy have formed very recently. 

This relatively nearby system is a challenge to observe because of its size, but it can be observed with the largest telescopes in the world, and this data should give us new insight into the formation of galaxies. It’s also hoped that as they observe more of the sky, their neural net will be able to uncover more of these smudges of forming galaxies. 

More Information

Subaru Telescope press release 

Extremely Metal-Poor Representatives Explored by the Subaru Survey (EMPRESS). I. A Successful Machine Learning Selection of Metal-Poor Galaxies and the Discovery of a Galaxy with M*<10^6 M_Sun and 0.016 Z_Sun,” Takashi Kojima et al., 2020 Aug. 3, Astrophysical Journal (Preprint on arxiv.org)

0 Comments

Got Podcast?

365 Days of Astronomy LogoA community podcast.

URL * RSS * iTunes

Astronomy Cast LogoTake a facts-based journey.

URL * RSS * iTunes * YouTube

Visión Cósmica LogoVisión Cósmica

URL * RSS

Escape Velocity Space News LogoEscape Velocity Space News
New website coming soon!
YouTube

Become a Patron!
CosmoQuest and all its programs exist thanks the generous donations of people like you! Become a patron & help plan for the future while getting exclusive content.