main concepts scientists sections
SimplePopularAdvanced

Magnetic Bubbles at the Edge of Black Holes ✛ ✦

Original: "Identifying Observational Signatures of Flux Eruption Events in Supermassive Black Hole Accretion Flows with Machine Learning"
· Angelo Ricarte, Erandi Chavez, Franc O, Pavlos Protopapas
arXiv:2606.28603 · 2026-06-26 · CC BY 4.0 · 1 min
High Energy
A neural network revealed a telltale sign that telescopes can use to catch magnetic explosions near black holes.

In the gas disk around a black hole, magnetic fields twist like overwound ropes. When they snap, a gigantic bubble bulges and bursts in the disk. The released energy feeds the jets shooting from the hole's poles.

A century ago, Karl Schwarzschild described a black hole mathematically. John Wheeler gave them their name, and Roger Penrose proved that inside lies a region where physics breaks down.

To spot these fleeting events, scientists simulated tens of thousands of images and trained a neural network to distinguish frames with a bubble. Its accuracy is 95%. Then they found simple clues for real interferometer telescopes (arrays of antennas). It turns out that during an eruption, the bright ring near the event horizon, the point of no return expands by 10–20%. But the most surprising thing is that its light becomes sharply polarized, as if combed: the waves oscillate in a single plane. However, the brightness barely changes. That's why astronomers missed these storms for decades, looking only for brightness outbursts.

Whole graph →

For the black hole at the center of galaxy M87, such an eruption lasts 11 days — faster than the Moon changes phases.

In the movie Interstellar, the characters see the black hole Gargantua. If you could zoom in on its edge, gigantic magnetic bubbles would flicker on and off around the glowing ring.

\phi = \frac{\sqrt{4\pi}}{2\sqrt{\dot{M}_{\bullet,2000}}} \int_{\theta}\int_{\phi} |B^r|_{r=r_H} \sqrt{-g}\, d\phi\, d\theta
Dimensionless magnetic flux threading the horizon; values around 50 correspond to the MAD regime, where the field dominates over accretion.
\mathcal{L} = -\frac{1}{N} \sum_{i=1}^N \left[ y_i \log(p_i) + (1-y_i) \log(1-p_i) \right] w_i
Weighted cross-entropy used in training the neural network to handle class imbalance: rare eruptions accounted for far fewer images than the quiescent disk.
Scientists
Stephen HawkingJacob BekensteinJohn Archibald WheelerFritz ZwickyKarl SchwarzschildSubrahmanyan Chandrasekhar
Concepts
black hole light echo black hole accretion EHT event selection event horizon EEG seizure detection Bekenstein-Hawking entropy quasi periodic eruptions maximum likelihood iron k alpha reverberation Machine Learning accretion disk wind maximum likelihood estimation model comparison prediction error seyfert galaxy symbolic regression neural network regression reverberation mapping astrophysical signal detection