Scientists believe that the environment immediately surrounding a black hole is turbulent, with hot magnetized gas swirling inside the disk at breakneck speeds and temperatures. Astronomical observations have shown that mysterious flares occur several times a day within such disks, briefly becoming brighter and then disappearing. Now, a team led by scientists at the California Institute of Technology has used telescope data and artificial intelligence (AI) computer vision techniques to detect just such a flare around the constellation Sagittarius A* (Sgr A*, pronounced sa-gee-eye). We have restored his first 3D video showing what it looks like. -star), the supermassive black hole at the center of our Milky Way galaxy.
The 3D flare structure features two bright, compact features located approximately 75 million kilometers from the black hole's center (half the distance between Earth and the Sun). This is based on data collected by Chile's Atacama Large Millimeter Array (ALMA) over his 100 minutes, immediately after the eruption confirmed by his X-ray data on April 11, 2017.
“This is the first three-dimensional reconstruction of rotating gas near a black hole,” says Katie Bauman, assistant professor of computing and mathematical sciences, electrical engineering and astronomy at the California Institute of Technology. His group is leading the effort described in a new 2016 paper. natural astronomy.
Aviad Levis, a postdoctoral researcher in Bowman's group and lead author of the new paper, emphasizes that while the video is not a simulation, it is also not a direct record of what happened. “This is a reconstruction based on our models of black hole physics. There are still a lot of uncertainties associated with it because it depends on whether these models are accurate or not,” he said. say.
Understand possible 3D structures using physics-based AI
To reconstruct the 3D images, the team needed to develop new computer imaging tools that can account for the bending of light due to the curvature of space-time around massive gravitational objects, such as black holes, for example.
The multidisciplinary team first considered whether it was possible to create 3D videos of flares around black holes in June 2021. The Event Horizon Telescope (EHT) collaboration, of which Bowman and Levi are members, had already released its first images. was trying to do the same with his EHT data from Sgr A*, a supermassive black hole at the center of a distant galaxy called M87. Google Research's Pratul Srinivasan, a co-author of the new paper, was visiting the Caltech team at the time. He helped develop a technique known as neural radiation fields (NeRF), which at the time was just beginning to be used by researchers. It has had a huge impact on computer graphics ever since. NeRF uses deep learning to create his 3D representation of a scene based on 2D images. This provides a way to observe the scene from different angles, even if only a limited view of the scene is available.
Building on these recent developments in neural network representation, the research team wondered if it might be possible to reconstruct the 3D environment around a black hole. Their big challenge is that, like everywhere else, Earth only provides a single view of the black hole.
Because gas behaves in a somewhat predictable way as it moves around a black hole, the researchers thought they might be able to overcome this problem. Consider the analogy of trying to capture a 3D image of a child wearing an inner tube around her waist. Capturing such images with traditional her NeRF techniques requires photos taken from multiple angles while the child remains stationary. But in theory, you could have your child rotate while the photographer stands still and takes the photo. Timed snapshots, combined with information about the child's rotation speed, can be used to reconstruct 3D scenes similarly well. Similarly, by leveraging knowledge about how gas moves at different distances from a black hole, researchers are using measurements taken from Earth over long periods of time to solve the 3D flare reconstruction problem. We aimed to solve the problem.
With this insight, the team built a version of NeRF that takes into account how gas moves around a black hole. But they also had to consider how light bends around massive objects like black holes. Under the guidance of co-author Andrew Chael of Princeton University, the team developed a computer model to simulate this bending, also known as gravitational lensing.
With these considerations in place, the new version of NeRF was able to recover the structure of bright features orbiting around the black hole's event horizon. In fact, the first proof of concept showed promising results on synthetic data.
Flare around Sgr A*, the research target
But the team needed real data. Then the ALMA telescope appeared. EHT's now famous image of Sgr A* is based on data collected between April 6 and 7, 2017, when the environment around the black hole was relatively calm. But just a few days later, on April 11, astronomers detected an explosive and sudden brightening in the vicinity. When Macek Wiergus, a team member at Germany's Max Planck Institute for Radio Astronomy, looked back at that day's ALMA data, he noticed a signal. This period corresponds to the time it takes for a bright spot in the disk to go around his Sgr A*. The research team set out to reconstruct the 3D structure of the brightening around Sgr A*.
ALMA is one of the most powerful radio telescopes in the world. However, because the galactic center is very far away (more than 26,000 light years), even ALMA does not have the resolution to see the immediate surroundings of Sgr A*. What ALMA measures is a light curve, which is essentially a video of a single flickering pixel, created by collecting all the radio wavelength light detected by the telescope at each moment of observation. .
It may seem impossible to recover a 3D volume from a single pixel video. However, by leveraging additional information about the expected physics of the disk around the black hole, the research team was able to circumvent the lack of spatial information in the ALMA data.
The strong polarization from the flare provided a clue.
ALMA does more than just capture a single light curve. In fact, each observation provides several such “videos”, since the telescope records data related to different polarization states of light. Like wavelength and intensity, polarization is a fundamental property of light that describes the orientation of the electrical component of a light wave relative to the wave's general direction of travel. “What you get from ALMA is two polarized, single-pixel videos,” says Bowman, a Rosenberg Scholar and Heritage Medical Research Institute investigator. “That polarization is actually really, really beneficial.”
Recent theoretical studies suggest that hot spots that form within the gas are strongly polarized. This means that the light waves coming from these hotspots have a clear preferred orientation direction. This is in contrast to the rest of the gases, which have a more random or scrambled orientation. By collecting various polarization measurements, ALMA data provided scientists with information that helped them determine where in 3D space the emission was coming from.
Introduction of orbital polarimeter tomography
To unravel the 3D structure that likely explains the observations, the researchers studied the physics of the bending and dynamics of light around the black hole, as well as the polarization expected at hot spots orbiting the black hole. We have developed an updated version of the method that also incorporates radiation. In this method, each potential flare structure is represented as a continuous volume using a neural network. This allows researchers to computationally advance his initial 3D structure over time as the hotspot orbits the black hole, creating an entire light curve. We were then able to find an optimal initial 3D structure that matches ALMA observations over time according to black hole physics.
The result was a video showing two compact bright regions moving clockwise following a path around the black hole. “This is very exciting,” Bowman said. “We didn't have to end up like this; we could have scattered arbitrary brightness throughout the volume. The fact that this closely resembles the flares predicted by his computer simulations of black holes is , very exciting.”
Levis said the study was uniquely interdisciplinary. “There is a partnership between computer scientists and astrophysicists that has a unique synergy. Together, we have developed cutting-edge things in both fields. This is the development of a numerical code to model the propagation of black holes into their surroundings.''This is the computational imaging research we conducted on black holes. ”
Scientists point out that this is just the beginning of this exciting technology. “This is a really interesting application that shows how AI and physics can work together to reveal things that are otherwise invisible,” Levis says. “We hope astronomers can use it on other rich time series data to uncover the complex dynamics of other such events and draw new conclusions.”
The title of the new paper is “Orbital optical rotation tomography of flares near the Sagittarius A* supermassive black hole.” This research was supported by funding from the National Science Foundation, California Institute of Technology's Carver-Mead New Adventures Fund, the Princeton Gravity Initiative, and the European Research Council.
