Scientists say they have built an artificial intelligence (AI) tool that can successfully identify and confirm supernovas.
A supernova is an “extremely bright, super-powerful explosion of a star,” the American space agency NASA reports. The explosion takes place at the end of a star’s life.
The current process for discovering supernovas usually involves both machines and people. Telescopes repeatedly collect images of the night sky and compare these images to identify differences in observed objects. These differences can suggest the presence of supernovas.
Robotic telescopes can present a list of possible supernovas to human scientists. This data is then examined by human scientists whose job it is to identify and confirm supernova candidates. Now, for the first time, researchers say they have successfully used an AI tool to complete all steps in this process.
An international team of scientists developed the AI-based system. Researchers from Northwestern University in Chicago, Illinois led the team. The team’s chief was Adam Miller, an assistant professor of Physics and Astronomy at Northwestern.
Miller said his team recently informed the international astronomical community of its success with the new AI system. The tool is called the Bright Transient Survey Bot, or BTSbot.
The robot was used to find, identify and classify a supernova for the first time, Miller said. He noted that researchers had spent more than 2,000 hours over the past six years looking at telescope data in order to find and classify supernova candidates.
“For the first time ever, a series of robots and AI algorithms has observed, then identified, then communicated with another telescope to finally confirm the discovery of a supernova,” Miller said.
The researchers say the new system can take over much of the work humans now do looking for and classifying supernovas. This can free up scientists to do more in-depth examinations of the supernovas themselves and the reasons they were created.
In order to confirm a supernova, scientists have to examine its spectrum. The spectrum represents the object’s dispersed light, which shows elements that were present during the explosion of a supernova.
The study of supernovas has high scientific value because it can help astronomers learn more about the universe and how it began. For example, NASA says scientists have learned a lot about the behaviors and chemical processes of stars from supernovas.
Miller said the new tool can help researchers “analyze their observations and develop new hypotheses” to explain why the supernova explosions happened.
Nabeel Rehemtulla co-led the research and development of the BTSbot. He said the tool greatly “streamlines large studies of supernovas.” This helps scientists better understand the life cycles of stars and helps to identify elements supernovas create, such as carbon, iron and gold, Rehemtulla noted.
The research team said testing of the BTSbot involved looking at a newly discovered supernova candidate called SN2023tyk. The AI tool identified the supernova on October 5.
After the identification, the BTSbot requested additional data on its own from the California-based Palomar Observatory. Then, another telescope, known as the SED Machine, performed its own “in-depth observations” to get information on the object’s spectrum.
The SED Machine then sent the spectrum data to the California Institute of Technology (Caltech). Caltech uses an AI algorithm to classify the kind of supernova. This can either be a thermonuclear explosion of a white dwarf or the collapse of a star’s center, or core.
The algorithm classified the supernova a Type 1a, a kind of star explosion in which a white dwarf in a star system fully explodes.
Rehemtulla said the success of the BTSbot will ease supernova identifications, while freeing up human scientists to do more detailed study of the newly discovered objects.
“The beauty of it is that, once everything is turned on and working properly, we don’t actually do anything,” Rehemtulla noted. “We go to sleep at night, and, in the morning, we see that BTSbot, and these other AIs, unwaveringly do their jobs.”
I’m Bryan Lynn.
