See the highest-resolution footage of a black hole jet ever taken, thanks to AI and 27 years of observations

Astronomers have unveiled a groundbreaking 27-year-long video of a black hole jet, the highest-definition animation ever captured. This remarkable video, showcasing the jet from a distant galaxy, was created using 116 images taken between 1995 and 2022. The subject of this mesmerizing footage is a blazar called 3C 345, located in the constellation Hercules.

Blazars are a type of quasar, which are incredibly bright and powered by supermassive black holes located at the centers of distant galaxies. These jets of gas ejected at nearly the speed of light are charged with intense X-rays and gamma rays. In a study published in Nature on August 26, researchers gathered images of 3C 345 using observations from the Very Long Baseline Array (VLBA), a network of 10 telescopes spread across the United States.

The team utilized an AI neural network called Kine to process these images, achieving a resolution four times higher than any individual image. This enabled the researchers to map the speed of the black hole jet with unprecedented precision. By analyzing the video, they made a surprising discovery: the jet's brightest components were traveling at 10 to 13 times the speed of light, while the surrounding gas was moving at about nine to 12 times the speed of light.

This finding defies the conventional understanding, which asserts that these bright components should move faster than the surrounding plasma due to shock perturbations. However, the researchers' work does not entirely dismiss the shock model but rather raises questions about its validity in this specific case. The VLBA's use of radio antennae across the continental U.S. provided a wide view of the universe, but its limited pixel resolution due to the array's 10 telescopes proved to be a constraint.

To overcome this limitation, the researchers employed Kine, an AI model designed to create videos of observations over time, focusing on astronomical sources with variable brightness. Kine, a neural network that learns from a dataset using layers of neurons, processes observations and leverages spatio-temporal correlations in the data. The team hopes that this innovative approach will revolutionize the study of jet dynamics through precise measurements of projected velocity at any point within the jet.