Fast radio bursts (FRBs) are bursts of radio waves, each lasting only a few milliseconds, from far outside the Milky Way. They are among the most puzzling astronomical phenomena. FRBs are difficult to study because most are singular in nature, emitting only one unprompted blast. But one FRB, called FRB 121102, frequently shoots out radio waves from a galaxy 3 billion light-years from Earth. Now, astronomers have begun using machine-learning technology to spot more bursts than ever before. Their results have been accepted for publication in The Astrophysical Journal.
In August 2017, researchers at Breakthrough Listen, which scans the skies for signs of extraterrestrial communications in association with the Search for Extraterrestrial Intelligence (SETI), took an indepth look at FRB 121102. During an observing session using the Green Bank Telescope in West Virginia, they detected 21 FRBs from the galaxy in one hour. The data indicated that the radio waves switched between periods of extreme and zero activity. But because the standard computer algorithms used to find the 21 bursts have limitations in their ability to recognize real patterns amid background noise, University of California, Berkeley, Ph.D. student Gerry Zhang created a more powerful machine-learning algorithm to scan the dataset for additional FRBs. That new algorithm revealed 72 new FRBs from the galaxy lurking in the data, the largest number detected in a single observation period.
After analysis, Zhang and his colleagues found no patterns in the arrival times of the 93 total FRBs shorter than 10 milliseconds. This information will help constrain models to better pin down the size and type of object giving off the bursts. With additional observations, Zhang and his team hope to further track signal arrival times and radio frequencies, ultimately pinpointing the source of these mysterious phenomena.