Nuclear astrophysics studies have shown that elements, including those heavier than carbon, form in stars. But the first stars born shortly after the Big Bang did not contain the heavy elements that astronomers call “metals.” The next generation of stars contained only a small amount of the heavy elements that the first stars formed. To understand the universe in its infancy, researchers need to study these metal-poor stars.

Fortunately, these metal-poor second-generation stars are observed in our Milky Way galaxy and have been studied by a team affiliated with the Kavli Institute for Physics and Mathematics (Kavli IPMU) to approximate the physical properties of the first stars. Universe.

The team led by Kavli IPMU Visiting Scientist and University of Tokyo Institute of Intelligence Physics Associate Professor Tilman Hartwig, Japan National Astronomical Observatory Visiting Scientist and Associate Professor Miho Ishigaki, University of Hertfordshire Visiting Senior Scientist and University of Hertfordshire Professor Chiaki Kobayashi, Japan National Astronomical Observatory Guest Humane Professor Nozomu Tominaga and University of Tokyo Visiting Senior Scientist and Professor Emeritus Kenichi Nomoto used artificial intelligence to analyze the element abundances in more than 450 stars with extremely low metal abundances observed to date.

Based on a newly developed supervised machine learning algorithm trained on theoretical models of supernova nucleosynthesis, they found that 68% of the extremely metal-poor stars observed had a chemical signature consistent with enrichment from several previous supernovae. The team’s results provide the first quantitative constraint based on observations of the abundance of early stars.

“Until now, the abundance of the first stars was estimated only by numerical simulations, and until now there was no way to test the theoretical prediction observationally,” said lead author Hartwig. “Our results suggest that many of the first stars formed in small clusters, so that large numbers of supernovae could contribute to metal enrichment of the early interstellar medium,” he said.

“Our new algorithm provides an excellent tool for interpreting the big data we will gain from current and future astronomical research around the world over the next decade,” said Kobayashi, who is also a Leverhulme scientist.

“Currently, available data on ancient stars is the tip of the iceberg around the Sun. The Prime Focus Spectrograph, an advanced multi-object spectrograph on the Subaru Telescope developed by the international community led by Kavli IPMU, is based on the ancient Milky Way’s outer regions far beyond the solar neighborhood. It’s the best tool for detecting stars.” said Ishigaki.

The new algorithm invented in this work opens the door to making the most of the various chemical signatures in metal-poor stars detected by the Prime Focus spectrograph.

“The early stellar theory tells us that the first stars must have been more massive than the Sun. The natural expectation was that the first star would have been born in a gas cloud with a mass of one million times the Sun. But our new discovery is that the first stars were not born individually, but rather a cluster of stars. “It also means that we can expect gravitational waves from the first binary stars shortly after the Big Bang that can be detected by future missions in space or to the Moon.” said.