Caltech astronomers used machine learning algorithms to classify 1,000 supernovae fully autonomously. The algorithm was applied to data acquired by his Zwicky Transient Facility (ZTF), a sky survey instrument located at Caltech’s Palomar Observatory.
“We needed help. We knew that training the computer to get the job done would relieve a lot of the strain on our backs,” says staff member Christoffer Fremling. Astronomer at Caltech and the mastermind behind it new algorithm,dubbing SNI score“SNIascore classified its first supernova in April 2021, and a year and a half later, we have reached the impressive milestone of 1,000 supernovae.”
ZTF scans the night sky every night looking for changes called transient events. This includes everything from migrating asteroids, to black holes that have just eaten a star, to stellar explosions known as supernovae. ZTF sends hundreds of thousands of alerts per night. Astronomer Notify the world of these temporary events. Astronomers then use other telescopes to track the changing nature of the object. So far, ZTF data have led to the discovery of thousands of supernovae.
However, with the constant stream of data flowing in every night, the ZTF team members cannot organize all the data themselves.
“The traditional notion of an astronomer sitting in an observatory and sifting through telescope images contains a lot of romanticism, but it’s also far from reality,” said ZTF project scientist, says Matthew Graham, a research professor of astronomy at the California Institute of Technology.
Instead, the team developed a machine-learning algorithm to help search. They developed SNIascore for the task of classifying supernova candidates. Supernovae fall into two broad classes he of type I and type II. Type I supernovae do not contain hydrogen, whereas type II supernovae contain abundant hydrogen. The most common Type I supernova occurs when a massive star strips matter from a neighboring star, causing a thermonuclear explosion. A Type II supernova occurs when a massive star collapses under its own gravity.
SNIascore can now classify Type Ia supernovae, or what are known as the “standard candles” in the sky. These are dying stars that explode in thermonuclear explosions of consistent intensity. Type Ia supernovae allow astronomers to measure the expansion rate of the universe. Fremling and his colleagues are now algorithm To classify other types of supernovae in the near future.
Each night, after the ZTF catches a possible supernova flash in the sky, it sends the data to Palomar’s spectrograph, called an SEDM (spectral energy distribution machine), housed in a dome just a few hundred meters away. increase. SNIascore works with SEDM to classify likely type Ia supernovae. As a result, the ZTF team is rapidly building a more reliable supernova data set for astronomers to further explore and ultimately learn the physics of powerful stellar explosions.
“SNIascore is very accurate. supernovawe’ve seen how algorithms work in the real world,” says Fremling.
Ashish Mahabal, principal computational scientist and data scientist at the Center for Data-Driven Discovery at Caltech, who leads machine learning efforts at ZTF, adds:
California Institute of Technology
Quote: Machine Learning Tool Autonomously Classifies 1,000 Supernovae (November 23, 2022) https://phys.org/news/2022-11-machine-tools-autonomously-supernovae.html
This document is subject to copyright. No part may be reproduced without written permission, except in fair trade for personal research or research purposes. Content is provided for informational purposes only.