AI Reveals Milky Way’s Black Hole Spins at Record Speed, Defying Scientific Expectations

Is Our Black Hole Defying Physics? New AI Study Challenges Theories

Astronomers have gained new insights into Sagittarius A* — the supermassive black hole at the center of our galaxy — by combining artificial intelligence with high-throughput computing from the University of Wisconsin-Madison’s Center for High Throughput Computing (CHTC).

Using a neural network trained on millions of simulations, researchers discovered that Sagittarius A* is spinning nearly at its maximum speed, with its axis of rotation pointed toward Earth. This conclusion is drawn from data collected by the Event Horizon Telescope (EHT), offering fresh perspectives on black hole behavior.

Key Findings from the AI Analysis

• The emission near the black hole primarily comes from extremely hot electrons in the accretion disk, not a jet as previously thought.
• Magnetic fields within the disk behave differently from established models.
• The black hole’s spin is near maximal, with its axis aligned toward Earth.

These findings challenge some aspects of prevailing theories about black hole dynamics and emission mechanisms.

High-Throughput Computing Enables Massive Data Processing

This research was made possible by leveraging high-throughput computing, a distributed computing technique pioneered by Miron Livny. This method allowed the team to efficiently process the enormous volume of data required to train the AI model.

Over the past three years, the Event Horizon black hole project completed more than 12 million computing jobs using resources from the NSF-funded Open Science Pool. This pool is operated by PATh and combines computing power from over 80 institutions across the U.S.

Miron Livny, director of the CHTC and lead investigator of PATh, noted, “A workload consisting of millions of simulations is a perfect match for our throughput-oriented capabilities that have been developed and refined over four decades.”

Collaboration and Future Directions

Michael Janssen, lead researcher from Radboud University Nijmegen in the Netherlands, commented on the significance of their approach: “That we are defying the prevailing theory is, of course, exciting. 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.”

Chi-kwan Chan, Associate Astronomer at Steward Observatory and a longtime PATh collaborator, emphasized the technical achievements: “Scaling up to the millions of synthetic data files required to train the model is an impressive achievement. It requires dependable workflow automation and effective workload distribution across storage resources and processing capacity.”

Professor Anthony Gitter, Morgridge Investigator and PATh Co-PI, added, “We are pleased to see EHT leveraging our throughput computing capabilities to bring AI into their science. CHTC’s infrastructure enabled EHT researchers to assemble the quantity and quality of AI-ready data needed to train effective models that facilitate scientific discovery.”

Implications for Research and AI Applications

This study is a clear example of how AI, combined with large-scale computing infrastructure, can push the boundaries of astrophysics research. Training neural networks on millions of simulations opens new paths for interpreting complex astronomical data.

For professionals interested in AI techniques applied to scientific research, exploring high-throughput computing and neural network training is increasingly valuable. To learn more about AI-driven tools and courses relevant to research and data science, visit Complete AI Training.

As future work refines these models and simulations, the relationship between black hole physics and AI will likely deepen, presenting new challenges and opportunities for both fields.