An international team of astronomers, led by experts from Northwestern University, has developed an artificial intelligence (AI) system to detect supernova for the first time.
Bright Transient Survey Bot (BTSbot) automates the search for supernovae, eliminating the possibility of human error, allowing you to speed up the analysis and classification of new candidate objects. To develop it, scientists trained the neural network on a sample of more than 1.4 million images from nearly 16,000 sources, including confirmed supernovae, transient bright stars, periodically transforming stars and dazzling galaxies. To test BTSbot, astronomers used data from supernova candidate SN2023tyk, discovered on October 3 by the Zwicky Transient Facility (ZTF) robotic observatory. On October 5, BTSbot examined the ZTF records in real time and found the source, then requested data from the Palomar Observatory, where another robotic telescope, the SED, obtained the spectra momentarily. This spectrum was then studied using Caltech’s SNIascore machine learning algorithm to determine the type of supernova.
The results showed that the candidate is a type Ia supernova, which occurs when a white dwarf star explodes in a binary star system. The temporary nameserver lists its redshift as 0.05, meaning it is more than 660 million light-years away. It was previously reported that the James Webb Space Telescope discovered a Type Ia supernova in the PLCK galaxy cluster G165.7+67.0 (G165). The flare, known as SN H0pe, originated in Arc 2, a galaxy whose brightness is enhanced by gravitational lensing created by G165.