Scientists used artificial intelligence to build a three-dimensional model of the bursts of energy, or flares, that occur around the Milky Way's central black hole, Sagittarius A* (Sgr A*). This 3D model could help scientists get a clearer picture of the overall noisy environment that forms around supermassive black holes.
The material swirling around Sgr A* exists in a flat structure called an “accretion disk,” which can flare periodically. These flares occur across a variety of wavelengths of light, from high-energy X-rays to low-energy infrared and radio waves.
Supercomputer simulations show that the flare observed by the Atacama Large Millimeter/Submillimeter Array (ALMA) on April 11, 2017 originated from two bright spots of dense material in the accretion disk of Sgr A*. This suggests that both faced the Earth. These bright spots swirl around a supermassive black hole, about 4.2 million times more massive than the Sun, and are only about half the distance between Earth and the Sun. This equates to approximately 47 million miles (75 million kilometers).
Related: A new view of the supermassive black hole at the center of the Milky Way galaxy suggests exciting hidden features
Reconstructing these flares in 3D from observational data is no mean feat. To address this, a team led by Caltech scientist Aviad Levis has proposed a new imaging technique called orbital polarization tomography. This method is no different from medical computed tomography (CT) scans performed in hospitals around the world.
“The compact regions around galactic centers are extreme locations where hot, magnetized gas orbits supermassive black holes at relativistic speeds. [speeds approaching that of light]. “This unique environment produces high-energy eruptions known as flares that leave observational signatures in X-ray, infrared, and radio wavelengths,” Levis told Space.com. It passes through a very bright and compact region that suddenly forms within the accretion disk. ”
He added that an important outcome of this study is the ability to reconstruct what the 3D structure of the radio brightness around Sgr A* looks like immediately after flare detection.
Build a black hole from a single pixel
“Sgr A* is located at the center of our Milky Way galaxy and is the closest supermassive black hole, making it a prime candidate for studying such flares,” Levis said. . “To do that effectively, we also need an element of luck when ALMA observations coincide with flares.”
He explained that he had been observing Sgr A* on April 11, 2017, shortly after a violent eruption captured by ALMA in X-rays. Radio data obtained by ALMA contained periodic signals consistent with what would be expected in orbit around Sgr A*.
“This prompted the development of a computational approach that can extract 3D structure from the time series data observed by ALMA,” Leavis added. “In contrast to his 2D images of Sgr A*'s Event Horizon Telescope (EHT), we are interested in recovering the 3D volume, and to do so, how light curves within a strong gravitational field. It relied on physical modeling of how black holes move along orbits.
To achieve their results, scientists looked at physics derived from Albert Einstein's 1915 theory of gravity, General Relativity, and applied concepts about supermassive black holes to neural networks. This network was then used to create the Sgr A* model.
“This research is a unique collaboration of astronomers and computer scientists advancing state-of-the-art computational tools from both the fields of AI and gravitational physics, each revealing the 3D radio emission structure around Sgr A. '*' Levis said. “The result is not a photograph in the usual sense, but rather a neural network that explains the expected physics of how gas orbits a black hole and how synchrotron radiation is emitted within a black hole. It is a computational 3D image extracted from time series observations by constraining the process. “
He explained that the team computationally placed a 3D “emission” in orbit around Sgr A*, starting with an arbitrary structure. Through ray tracing, a graphical simulation of the physical behavior of light, Levis and his colleagues were able to model how ALMA would see the structure around Sgr A* in the future. These models started 10 minutes after him in the flare, then he followed after 20 minutes, 30 minutes, and so on.
“Neural radiation fields and general relativistic ray tracing techniques provide a way to start modifying the 3D structure until the model matches observations,” Levis added.
The researchers say their results lead to conclusions about Sgr A*'s surrounding environment, which is indeed predicted by theory, and show that the brightness is concentrated in a few small regions within the accretion disk. I discovered that. Still, some aspects of this work came as a surprise to Levi and the rest of the team.
“The biggest surprise was that we were able to recover a 3D structure from light curve observations…essentially a video of a single flickering pixel,” the researchers said. “Think about it. If we said that he could recover a video from just one pixel, you'd think it's virtually impossible. The point is, we're not recovering any video. about it.
“We are reconstructing the 3D structure of the emission around the black hole, and we can use the expected gravitational and emission physics to constrain the reconstruction.”
Levis added that the fact that ALMA measures not only the intensity of the light but also its polarization has given the team a very informative signal, including clues about the 3D structure of the flare around Sgr A*. Ta.
Going forward, Levis said he and his team will run simulations varying the physical parameters used to constrain the AI.
“These results are an exciting first step and are based on the idea that Sgr A* is a black hole whose environment follows a given gravity and emission model. The accuracy of the results depends on these assumptions. It depends on relevance,” Levi concluded. “In the future, we would like to relax these constraints to allow for deviations from the expected physics.
“Our approach, which harnesses the synergy of physics and AI, opens the door to new and exciting questions, the answers of which will continue to advance our understanding of black holes and the universe.”
The research team's study was published in the academic journal Nature Astronomy on Monday (April 22).
