Black holes are part of the club of astronomical objects that arouse great curiosity in people. We have known for a long time that they are there, in space, devouring everything in their path, but their existence is shrouded in mystery.
As their name implies, they have a dark nature that makes them very difficult to detect with our scientific instruments. In 2019, however, we achieved the unimaginable: we captured that mysterious force in an image.
The first image of a black hole, now improved
After many years of work, a group of scientists “photographed” a black hole located in the Messier 87 galaxy, some 54 million light-years away. The image showed us the accretion disk of the astronomical object, that is, its event horizon.
The problem? Although that event was classified as a scientific milestone, the image did not stand out precisely for its sharpness. In fact, we were seeing something as massive as the deformation of spacetime in a disk with a brighter area and a less bright area.
But we live in times where artificial intelligence seems to be in almost everything and, as you can imagine, also in improving that historical image. So a much sharper and more striking image of Messier 87’s supermassive inhabitant has been released.
As explained in Astrophysical Journal Letters, researchers led by Princeton Institute for Advanced Study astrophysicist Lia Medeiros used a machine learning algorithm to fill in the missing data from the original image.
The original image published in 2019 was the product of the Event Horizon Telescope (EHT), which is not a single telescope, but rather a network of telescopes distributed at different points on the planet. That prodigy gave us the possibility of having something like a plate the size of the Earth.
And we needed such technical capacity to be able to capture the radiation emitted by the matter that rolled around the black hole, since this was extremely tenuous, but this solution also had a problem, and it was precisely that it left “gaps” when presenting the information.
The PRIMO algorithm, which is what the system is called, was trained with thousands of simulations of matter feeding black holes. With all that information, he was ready to fill in the missing EHT data, a task he recently completed and thoroughly evaluated.
After applying various methods of analysis and testing, the researchers concluded that the ” image reconstruction algorithm ” did a very precise job of presenting a sharp, high-resolution image of the black hole.
“Since we cannot study black holes up close, the detail of an image plays a fundamental role in our ability to understand their behavior,” Medeiros explained about the importance of this advance achieved thanks to artificial intelligence.