Researchers at the California Institute of Technology have used machine learning to classify 1,000 supernovae caused by exploding dying stars. Space.com reports.
Astronomers, using the SNIascore algorithm, created a catalog from data gathered by the Zwicky Transient Facility (ZTF) attached to the Samuel Oschin Telescope at the institute’s Palomar Observatory.
According to the scientists, ZTF scans the night sky and collects a vast amount of data that is difficult to process manually. To tackle this, they developed the SNIascore algorithm.
“We knew that once we trained our computers to do this work, they would lift a great burden from us,” said Christopher Fremling, a California Institute of Technology employee and one of the project’s authors.
Since 2017, ZTF has identified thousands of supernovae, which can be divided into two classes:
- hydrogen-poor Type I;
- hydrogen-rich Type II.
Type I supernovae most often arise due to accretion of matter from a neighboring donor star, which falls onto its surface and triggers a thermonuclear explosion. Type II forms when massive stars run out of the fuel needed for nuclear fusion and can no longer resist gravitational collapse.
SNIascore classifies a special type of supernovae called Type Ia. They occur when a dying star explodes and form a homogeneous light output, known as “standard candles.” Astronomers use information about them to measure distances in the universe and determine the rate of expansion of the cosmos.
Each night, after finishing the search for transient events and objects in the sky with the help of ZTF, the AI begins classifying Type Ia supernovae.
“SNIascore found its first supernova in April 2021, and a year and a half later we reached a milestone of 1,000 objects,” said Fremling.
According to him, the algorithm works virtually without error. In the future, scientists plan to deploy the same model with other observing facilities.
Astronomers also intend to apply SNIascore to classify other types of supernovae. Even before these achievements occur, the machine-learning tool is changing astronomy and demonstrating the evolving face of this scientific field.
As reported in June 2021, artificial intelligence discovered hidden connections between the Milky Way and Andromeda.
In July, scientists introduced an algorithm, classifying thousands of galaxies per second.
In December, astronomers catalogued 170 rogue exoplanets discovered by artificial intelligence.
Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!