The researchers used machine learning to improve the first image of a black hole

The researchers used machine learning to sharpen a previously released image of the black hole. As a result, the image of the black hole at the center of the Messier 87 galaxy, more than 53 million light-years from Earth, shows an even thinner ring of light and matter surrounding its center in Report published today in Astrophysical Journal Letters.

The original images were taken in 2017 by the Event Horizon Telescope (EHT), a network of radio telescopes around Earth that combine to serve as a planet-sized superimaging instrument. The initial image looked like a “fuzzy donut”. described by NPR, but the researchers used a new method called PRIMO to reconstruct a more accurate image. PRIMO is a “new dictionary-based learning algorithm” that learns to “restore high-resolution images even in the presence of sparse coverage” by training on simulations created of more than 30,000 black holes. In other words, it uses machine learning data based on what we know about the physical laws of the universe – and black holes specifically – to produce a better-looking and more accurate snapshot than the raw data captured in 2017.

Black holes are mysterious and strange regions in space where gravity is so strong that nothing can escape from them. They form when dying stars collapse in on themselves under their own gravity. As a result, the collapse compresses the star’s mass into a small space. The boundary between a black hole and the mass around it is called the event horizon, and it is the point of no return where nothing (whether it be light, matter, or Matthew McConaughey) will cross it again.

See also  MIT physicists discover a family of 'magic' superconducting graphene structures

“What we’re really doing is we’re learning the correlations between different parts of the image. And so we do that by analyzing tens of thousands of high-resolution images generated from simulations,” said astrophysicist and paper author Lia Medeiros of the Institute for Advanced Study in Princeton, New Jersey , NPR. “If you have an image, the pixels close to any given pixel will not be completely uncorrelated. It’s not that each pixel does completely independent things.”

The researchers say the new image is consistent with Albert Einstein’s predictions. However, they expect that further research into machine learning and telescope hardware will lead to additional reviews. “In 20 years, the picture may not be the one I show you today,” Medeiros said. “It might be better.”

Leave a Reply

Your email address will not be published. Required fields are marked *