Something massive lurks at the center of our galaxy. It weighs about four million times more than our Sun, and it bends space and time around itself like a whirlpool warping water. For years, scientists have studied it through fuzzy images and indirect signals, piecing together clues about its nature from radio waves collected by telescopes scattered across the planet. But until recently, some of its most basic properties remained a mystery.

Now, a team of astronomers has pulled back the curtain on Sagittarius A*, and what they found raises questions nobody quite expected.

New research published in Astronomy & Astrophysics used artificial intelligence trained on millions of simulated black hole datasets to decode real observations from the Event Horizon Telescope. Led by Michael Janssen at Radboud University in the Netherlands, the study cracked open information buried deep inside noisy, blurry data from the galactic center. And the picture that emerged rewrites much of what we assumed about the engine at the heart of the Milky Way.

How a Self-Learning AI Cracked Open Black Hole Data

Previous Event Horizon Telescope studies compared observations against only a handful of simulated datasets. Imagine trying to identify a suspect from a police lineup of three people when the real pool numbers in the millions. Janssen’s team changed the game entirely by feeding roughly 962,000 synthetic black hole datasets into a Bayesian neural network, a type of AI that can measure its own uncertainty.

Each synthetic dataset mimicked what the Event Horizon Telescope would see under different conditions, with different black hole spins, magnetic field strengths, viewing angles, and accretion behaviors. By training on such a massive library, the neural network learned to recognize subtle patterns in the data that human analysis alone could never isolate.

“The ability to scale up to millions of synthetic data files is an impressive achievement,” said co-researcher Jordy Davelaar of Princeton University. “You need storage capacity, a supercomputer, a software pipeline, and a program that distributes the work.”

And distributed it was. Data lived on CyVerse storage systems. Computing ran through the Open Science Grid. Pegasus software managed the workflow. Germany’s Max Planck Computing and Data Facility handled the neural network training. Software tools, including TensorFlow and Horovod, stitched it all together. No single institution could have done it alone.

Once trained, the AI went to work on actual Event Horizon Telescope observations of Sagittarius A* captured in April 2017. What it returned surprised everyone.

Spinning at Near Maximum Speed

Sagittarius A* appears to be rotating at nearly the fastest rate physics will allow. General relativity sets a hard ceiling on how fast a black hole can spin. Push past that limit, and the math says the event horizon, that boundary beyond which nothing escapes, would theoretically vanish. Sagittarius A* sits right up against that ceiling, spinning with a dimensionless spin parameter somewhere around 0.8 to 0.9 out of a maximum of 1.

At that rate, the black hole drags the very fabric of space and time around itself at astonishing speed. Matter, light, and even the geometry of spacetime get pulled along like leaves caught in a tornado. For a black hole of four million solar masses, each tick of its gravitational clock lasts about 20 seconds, and within that time, the entire inner region of swirling gas rearranges itself.

Scientists had suspected some spin but never had enough data resolution to pin it down with confidence. Only with the AI-driven approach, working through millions of comparisons rather than a few dozen, did the answer become clear.

Pointed Straight at Us

Here is where things get stranger. Sagittarius A*’s spin axis appears oriented almost directly toward Earth. Our line of sight falls at an inclination of roughly 18 to 29 degrees from the spin axis, depending on which of Janssen’s two equally strong neural network models you consult. In simpler terms, we are looking nearly straight down the barrel.

In the sky, the position angle of that axis sits between about 106 and 137 degrees east of north, a finding that matches independent measurements from instruments like GRAVITY and from ALMA light curve analyses. Several separate lines of evidence converge on the same geometry, which lends strong support to the result.

What does this alignment mean? Because we peer nearly along the spin axis, we get a remarkably clean view of the black hole’s light-bending silhouette and the hot matter spiraling around it. A different angle might show us a messier, more confused picture. Our vantage point, purely by cosmic accident, happens to be one of the best seats in the house.

Hot Electrons, Not a Jet

Many supermassive black holes blast powerful jets of charged particles outward at nearly the speed of light. M87*, the black hole at the center of galaxy Messier 87 and the first ever directly imaged in 2019, does exactly that. Scientists reasonably wondered whether Sagittarius A* might too. It does not, or at least not in any obvious way.

According to the neural network analysis, the glow we detect from Sagittarius A* comes primarily from extremely hot electrons spiraling within its accretion disk rather than from a focused jet. A relatively low electron temperature coupling factor supports this reading, consistent with earlier models of how Sagittarius A*’s faint radio and X-ray emissions are produced.

No instrument has ever directly detected a jet from Sagittarius A*. And these new results explain why. Without a fully developed magnetically arrested accretion state, the conditions needed to launch a powerful jet simply may not exist at the galactic center.

Magnetic Fields Breaking From Script

Perhaps the most puzzling result concerns the magnetic fields around Sagittarius A*. Standard models of black hole accretion disks predict two broad categories of magnetic behavior. In one, enormous magnetic flux piles up near the event horizon until it erupts, creating a magnetically arrested disk, or MAD. In the other, a calmer configuration called SANE allows gas to flow inward with far less magnetic drama. Sagittarius A* does not fit neatly into either category.

“That we are defying the prevailing theory is of course exciting,” said lead researcher Michael Janssen. “However, I see our AI and machine learning approach primarily as a first step. Next, we will improve and extend the associated models and simulations.”

Something between or beyond MAD and SANE appears to be happening, and current models cannot fully explain it. Magnetic fields in the accretion disk behave differently from what either standard theory predicts. For theorists, that gap between expectation and observation is where new physics often hides.

M87*’s Black Hole Tells a Different Story

When the same neural network analyzed M87*, the results painted a sharply different portrait. M87* also spins fast, somewhere between half and near maximum speed, but it rotates in the opposite direction to its infalling gas. Astronomers suspect a past galaxy merger caused that counter-rotation, a plausible scenario for a giant elliptical galaxy like Messier 87, which likely absorbed smaller galaxies over billions of years.

Unlike Sagittarius A*, M87* strongly favors a magnetically arrested disk with powerful jet emission. Its electrons radiate heavily from the jet region rather than from the disk. In nearly every respect, M87* and Sagittarius A* represent opposite ends of how supermassive black holes can behave, even though both sit at the centers of large galaxies.

A New Telescope in Namibia Could Change Everything

Better data is on the way. An African Millimetre Telescope is currently under construction, with a likely location on the Gamsberg mountain in Namibia. When it joins the Event Horizon Telescope array, it will add new baseline coverage in directions that current stations cannot reach, improving northeast and southwest resolution.

According to the study, adding just one well-placed dish will reduce parameter inference errors by a factor of three for models testing alternatives to general relativity. Longer monitoring campaigns will also help, since many black hole properties only reveal themselves through changes measured over months or years.

Looking further ahead, a next-generation Event Horizon Telescope with more than 20 stations could create images with dramatically better detail. “We have achieved something presumed to be impossible just a generation ago,” EHT project director Sheperd S. Doeleman once said about the first black hole image. Each new station and each new algorithm push that boundary further.

What a Spinning Giant Tells Us About Being Human

A black hole four million times heavier than our Sun spins at the absolute limit of known physics and happens to aim its axis in our direction. We did not simply stumble onto that information. Scientists built a planet-sized virtual telescope, trained artificial intelligence on nearly a million simulations, and coordinated across continents and computing systems to wring meaning from impossibly faint signals.

Every boundary we assumed was fixed, from the resolution limits of telescopes to the interpretive limits of human analysis, turned out to be movable. Life on Earth has always existed inside the gravitational influence of Sagittarius A*, orbiting the galactic center every 225 million years, even if we only now understand the ferocity of its spin.

Knowing that the heart of our galaxy operates at its physical extreme invites a larger question about how we measure our own limits. Nature, it turns out, already runs at full speed at scales we can barely observe. If the central engine of the Milky Way pushes right up against the boundaries of what physics permits, perhaps the boundaries we set for ourselves also deserve a second look. Every generation of astronomers inherits tools that their predecessors called impossible. And every generation proves, once again, that the only real limit is the willingness to keep looking.

Loading...