An AI makes impressive discoveries on the central black hole of the Milky Way
Illustration of a network of neurons connecting observations (left) to models (right). © EHT Collaboration/Janssen et al.
The images of the Event Horizon Telescope (EHT) went around the world in 2019 and 2022 when, for the first time in the history of astronomy, scientists were able to produce a “real” image of two black holes.
First there was M87*, Powehi of his little nickname, a pretty baby of 6 billion solar masses housed in the center of the M87 galaxy. Two years later, we were able to contemplate the image of the black hole of our own Milky Way, responding to the somewhat obscure name of Sgr A*, more modest black ogre of 4 million solar masses.
The central black hole of the Milky Way would almost run at the speed of light
Originally, the EHT collaboration had to produce the image of the black hole in our galaxy before that of M87, but the task had been deemed too complicated. The three new studies published by Michael Jansen of the University of Radboud (Netherlands) and his colleagues confirm what we supposed. Sgr A* is very turbulent! According to their analysis, it could turn at a speed close to that of light.
Neural networks to study millions of data
Polarization of SGR A*magnetic fields. © EHT
Astronomers have formed a network of neurons with millions of data sets on black holes, collected by the Network of EHT radio stakes, all distributed on the surface of the globe. This sum of data is colossal: just for the first observation campaign that led to the image of M87*, five petacts (5000 teraoctets) had been collected!
The hard drives of the EHT did not lack data and that made Katie Bouman happy. © EHT
Michael Jansen’s team arrives at several fairly spectacular conclusions. Thus, Sgr would have turned almost at the speed of light, which would explain the difficulties in producing the famous shot. Similarly, the earth would almost be in the alignment of the axis of rotation of the supermassive black hole and the nearby emission would be due to hot electrons in the accretion disc (dust and enveloping gas Sgr A*), not to a jet of material. Finally, the magnetic fields of the accretion disk seem to present a different behavior of theoretical predictions.
“The fact that we challenge the dominant theory is of course excitingsays Michael Janssen. However, I consider our approach to AI and automatic learning mainly as a first step. Then, we will improve and widen the associated models and simulations. And when the Africa Millimetter telescope [en construction, NDLR] will be part of data collection, we will obtain even better information to validate the general theory of relativity for compact supermassive objects with great precision. ”
By studying with their neural network the data of M87*, these astrophysicists also think that the latter runs according to a contractive gas direction which falls into it, a peculiarity which could be due to an old major galactic merger.
Want to save even more? Discover Our promo codes Selected for you.
