An artificial network was able to detect rare, super-fast stars through the Milky Way, thanks to an AI that was collecting data from the European Space Agency's Gaia probe. The Gaia mission has been trying to construct detailed 3D space maps by measuring positions of stars in far-off places.
Hypervelocity stars (HVSes), which are far-flung from the Milky Way's galactic center, are said to reach speeds faster than the escape velocity. However, only 20 of them have been discovered to be outbound, most of which are late B-type stars, or stars that are brighter and larger than the Sun.
A team of researchers presented the results of their study regarding HVSes at the European Week of Astronomy Space Science in the Czech Republic. The paper, which they published at the end of May in the Monthly Notices of the Royal Astronomical Society, described the system as a neural network that has five input units for astrometric parameters. They simulated mock data based on real results from the Gaia catalog with inputs describing the coordinates of these stars, including the distance and brightness. Then they calculated the velocity.
The Register noted that the algorithm reduced the data set of nearly 2 million stars to some 20,000 or 1 percent of the catalog. The results showed the complex dynamic systems of the HVSes.
The team found 14 stars to have a total velocity in Galactic rest frame, but up to 50 percent of them has the probability of escaping the Milky Way. Elena Maria Rossi, a co-author of the paper and a researcher at Leiden University, stated the importance of hypervelocity stars, especially with regards to the study of the overall structure of the Milky Way. She said that their density yields crucial information regarding the galaxy's gravitational field, from the center to the outskirts.