Researchers have developed a new machine-learning technique that could revolutionize the study of neutron star collisions, allowing astronomers to observe these rare events in real time.
The 2017 detection of gravitational waves from the collision of two neutron stars triggered one of the largest collaborative efforts in astronomy, with over 70 teams racing to study the aftermath. The new algorithm, trained on simulations of data collected by gravitational-wave observatories in the minutes leading up to a neutron star merger, could provide advance warning of such events.
These mergers create kilonovas, events thought to produce some of the heaviest elements in the universe, including gold, platinum, and uranium. The algorithm could alert other astronomers that a collision is imminent, pinpointing the time and location in the sky with 30% greater accuracy than current rapid-response methods.
“The combination of speed and accuracy in the localization presented in this paper is actually fantastic,” says Mansi Kasliwal, an astrophysicist at the California Institute of Technology in Pasadena. The findings were published in the journal Nature on March 5.
Neutron stars are the remnants of massive stars that have collapsed at the end of their lives, compressing more than the mass of the Sun into a dense ball of neutrons about 20 kilometers across. Sometimes, neutron stars orbit each other in pairs, or binaries. When their orbit is sufficiently tight, Einstein’s general theory of relativity takes effect, and the binary starts to produce gravitational waves—ripples in the fabric of space-time. As a result, the stars lose energy and spiral into each other until they merge. To date, researchers have observed only a small handful of neutron-star mergers. The 2017 event was the only instance where the merger was also observed by non-gravitational-wave observatories, such as telescopes and gamma-ray-burst detectors.
