Jun 4 2020
Over 2000 large protostars—new stars that are still emerging and may hold clues to the genesis of the stars in the Milky Way galaxy—have now been identified by an artificial intelligence (AI) system. This discovery was made by the AI system while it was examining data from the Gaia space telescope.
This is an artist’s impression of the Gaia space telescope in orbit. Image Credit: ESA/D. Ducros, 2013.
Earlier, researchers had cataloged only 100 protostars, and exploration of these celestial objects had yielded most of the knowledge that underpinned studies relating to the formation of stars.
Miguel Vioque, a PhD researcher at the University of Leeds, headed the project. The study titled “New catalogue of Herbig AE/BE and classical Be stars: A machine learning approach to Gaia DR2” was published in the Astronomy & Astrophysics journal.
According to Vioque, exploring these recently discovered stars could change the researchers’ interpretation of the formation of giant stars and their approach to analyzing the galaxy.
Mr Vioque and his collaborators were fascinated in the so-called Herbig Ae/Be stars—stars that are still developing and whose mass is at least twofold that of the Sun. The Herbig Ae/Be stars also contribute to the emergence of other stars.
The scientists took the large amount of data being compiled by the Gaia space-borne telescope as it plots the galaxy. Introduced in 2013, the data compiled by the Gaia space-borne telescope has allowed researchers to determine the distances for approximately one billion stars—around 1% of the total is believed to be present in the galaxy.
After cleaning that data, the scientists decreased it to a subset of 4.1 million stars that may probably contain the required protostars. The AI tool sorted out the data and eventually produced a list of 2,226 stars that has about 85% chance of being a Herbig Ae/Be protostar.
There is a huge amount of data being produced by Gaia—and AI tools are needed to help scientists make sense of it. We are combining new technologies in the way researchers survey and map the galaxy with ways of interrogating the mountain of data produced by the telescope—and it is revolutionising our understanding of the galaxy. This approach is opening an exciting, new chapter in astronomy.
Mr Miguel Vioque, PhD Researcher, School of Physics and Astronomy, University of Leeds
Mr Vioque and his collaborators subsequently verified the findings of the AI system by analyzing 145 of the stars detected by the AI tool at ground observatories in Chile and Spain, where they effectively quantified the light—recorded as spectra—emerging from the stars.
The results from the ground-based observatories show that the AI tool made very accurate predictions about stars that were likely to fall into the Herbig Ae/Be classification.
Mr Miguel Vioque, PhD Researcher, School of Physics and Astronomy, University of Leeds
One of the target stars is called Gaia DR2 428909457258627200. The star has a mass that is 2.3 times that of the Sun and is located 8,500 light-years away. The surface temperature of Gaia DR2 is 9,400 °C—the sun is around 5,500 °C—and its radius is double that of the Sun. The star has existed in space for about six million years, which in astronomical context renders it a new star that is still developing.
This research is an excellent example of how the analysis of the Big Data collected by modern scientific instruments, such as the Gaia telescope, will shape the future of astrophysics. AI systems are able to identify patterns in vast quantities of data—and it is likely that in those patterns, scientists will find clues that will lead to new discoveries and fresh understanding.
René Oudmaijer, Professor, School of Physics and Astronomy, University of Leeds
Professor Oudmaijer supervised the study.
The study was financially supported by the European Union’s Horizon 2020 research and innovation program, under the STARRY project.
Journal Reference:
Vioque, M., et al. (2020) Catalogue of new Herbig Ae/Be and classical Be stars—A machine learning approach to Gaia DR2. Astronomy & Astrophysics. doi.org/10.1051/0004-6361/202037731.