As part of the SETI project, scientists from the University of California at Berkeley, based on computer simulations, were able to immediately identify 72 new fast radio signals from a mysterious source located three billion light-years from Earth. The research results are described in a new article accepted for publication in The Astrophysical Journal.
Fast radio signals are bright pulses of radio emission, millisecond duration, believed to come from distant galaxies. However, the source of these signals is still unclear. Theories range from highly magnetized neutron stars blown up by gas streams from a nearby supermassive black hole to suggestions that the signals are man-made and sent by an intelligent civilization.
“This work is exciting not only because it helps us understand the dynamic behavior of fast radio signals in more detail, but also because of the perspective we see when trained computers use classical algorithms to find these signals themselves,” said Andrew Simion. Director of the Berkeley SETI Research Center and Principal Investigator of the Breakthrough Listen program, dedicated to the search for intelligent life in the universe.
As part of this program, researchers have successfully used a machine learning algorithm to search for new types of signals that may come from extraterrestrial civilizations.
While the fastest radio signals are one-off in nature, the detected source, named FRB 121102, is unique in that it shows a whole complex of signals. This behavior has attracted the attention of many astronomers, hoping to reveal the cause and the extreme physics involved in the nature of such a phenomenon.
Artificial intelligence detected radio signals in the database, in a five-hour interval of observations on August 26, 2017 with the Green Bank telescope in West Virginia. An earlier analysis of 400 terabytes of data used standard computer algorithms to identify 21 radio bursts during this period. They were all sighted within one hour, and the source was assumed to alternate between periods of rest and frenzied activity, at least as noted by Berkeley SETI researcher Ph.D. Vishal Gajjar.
Fellow SETI study author Jerry Zhang and his collaborators subsequently developed a powerful new machine learning algorithm and reanalyzed the 2017 data, finding 72 more spikes that were not originally detected. In the end, observers were amazed to conclude that the total number of detected bursts from FRB 121102 is about 300 since the object was discovered in 2012.
“This work is just the beginning of using powerful new techniques to find transient radio signals,” Zhang said. “We hope our success can inspire other serious organizations to apply machine learning to radio astronomy.”
Zhang's team used the same techniques that internet technologists use to optimize search results and classify images. They developed an algorithm, known as a convolutional neural network, that recognizes radio bursts found by the classical search method used by Gajar and co-workers, and then locate them in the database, finding bursts that were missed in the classical search approach.
The results helped to establish new limits on the frequency of pulses from FRB 121102 and indicated that the pulses are irregular in the event that the period of this pattern is greater than about 10 milliseconds. Just as pulsar pulse models helped astronomers limit computer models of extreme physical conditions in such objects, the new FRB measurements, scientists say, will help clarify the nature of mysterious new sources.
“Regardless of whether the signals from the FRB are ultimately a sign of extraterrestrial technology, Breakthrough Listen is helping to push the boundaries of a new and rapidly growing area of our understanding of the universe around us,” Jan concluded.
