\n<\/p>\n
An image of Sagittarius A*, the black hole at the centre of the Milky Way galaxy<\/p>\n
EHT<\/p>\n<\/div>\n<\/figcaption><\/figure>\n<\/p>\n
Our galaxy\u2019s centre may contain an exceptional cosmic spinning top \u2013 a black hole that seems to be spinning almost as fast as possible.<\/p>\n
Michael Janssen at Radboud University in the Netherlands and his colleagues were studying the black hole at the Milky Way\u2019s centre, Sagittarius A*, using the data gathered by a network of observatories collectively known as the Event Horizon Telescope (EHT). To deal with the complexity of the data, they turned to artificial intelligence.<\/p>\n
<\/p>\n First, they used well-known mathematical models to simulate about a million black holes \u2013 which was itself a computational feat that required millions of hours of supercomputer time. Then they used these simulations to train a type of AI called a neural network, enabling it to determine a black hole\u2019s traits based on observational data. Finally, they fed the AI the data about Sagittarius A* that EHT had collected throughout 2017.<\/p>\n The AI indicated that Sagittarius A* is spinning at between 80 and 90 per cent of its highest possible speed. It also alerted the researchers that none of their magnetic field models fit our black hole particularly well \u2013 so more mathematical work is necessary. Janssen says that past studies had narrowed down the range of properties that Sagittarius A* could have, such as how fast it\u2019s spinning and what kind of magnetic fields surround it, but this new approach pinned them down more precisely.<\/p>\n Dimitrios Psaltis at the Georgia Institute of Technology in Atlanta says he found some of these findings counterintuitive. Previous analyses couldn\u2019t even offer clarity on whether black hole spin could be determined this precisely from EHT data, he says.<\/p>\n Some past work did indicate that Sagittarius A* could be spinning very quickly, says Yosuke Mizuno at Shanghai Jiao Tong University in China. But he notes that the computer models used in the new study have room for improvement. \u201cOur theoretical model is still not perfect,\u201d he says.<\/p>\n But both Mizuno and Psaltis say that AI is becoming an integral part of we how learn about exotic cosmic objects like black holes. \u201cWe are in a situation where we have a lot of data and we have a lot of models, and we need modern ways of combining the two,\u201d says Psaltis. \u201cThis is where machine learning makes a big, big difference.\u201d<\/p>\n At the same time, this brings its own challenges, as AI\u2019s work must be double-checked and subsequent analysis proofed against possible hallucinations.<\/p>\n Janssen and his team have already performed many such checks, such as testing the AI with specifically designed simulation data. More tests will happen as they analyse data from other years of EHT\u2019s operation and, ultimately, data from new observatories, he says.<\/p>\n Topics:<\/p>\n<\/section><\/div>\n