Physicists at the Austrian Academy of Sciences (ÖAW) are deploying artificial intelligence to sift through gravitational wave data in hopes of uncovering a long-sought class of cosmic objects: intermediate-mass black holes.
The initiative aims to address fundamental questions about how black holes form and how the universe evolves. By re-analysing gravitational wave signals with advanced AI tools, researchers believe they may be able to detect black holes that have so far escaped observation.
The focus is on intermediate-mass black holes, with masses ranging roughly from 100 to 100,000 times that of the Sun. These objects are heavier than typical stellar black holes but far smaller than the supermassive black holes at the centres of galaxies. Scientists suspect they may represent a missing evolutionary link — yet direct evidence remains scarce.
“We know that small black holes can grow through mergers,” said Gianluca Inguglia of the Marietta Blau Institute for Particle Physics at ÖAW. “But whether that process ultimately leads to supermassive black holes is still not fully understood.” Intermediate-mass black holes could hold the key to that puzzle.
Clues lie in gravitational waves — ripples in spacetime produced when massive objects collide. When black holes merge, they emit distinctive signals that can be detected by highly sensitive instruments. However, mergers involving intermediate-mass black holes produce extremely short-lived signals that are difficult to separate from background noise.
In a study published in Physics Letters B, Inguglia and his team describe how artificial intelligence could enhance detection efforts at the planned Einstein Telescope, Europe’s next-generation gravitational wave observatory. Expected to begin operations in the 2030s, the facility will be significantly more sensitive than current detectors and generate vast quantities of data.
Traditional detection methods compare observed signals to theoretical templates or rely on multiple detectors to confirm events. While effective for well-characterised phenomena, these approaches can miss signals if scientists do not know precisely what to look for.
“With intermediate-mass black holes, many properties are still uncertain,” Inguglia said. “That increases the risk of overlooking or misinterpreting signals.”
The team’s AI-driven method takes a different approach. Instead of searching for predefined patterns, the system first learns what normal background noise looks like in a detector. It is trained to reproduce this noise with high accuracy. When an unfamiliar pattern appears — potentially a real merger signal — the AI fails to replicate the data precisely, triggering an alert.
“In this case, the model’s failure is a success,” Inguglia explained. “It tells us that something unusual — and possibly real — is present.”
The researchers tested the method using simulated data from the Einstein Telescope Mock Data Challenge. The AI successfully identified short merger signals, even when analysing data from a single detector. After further training, the system detected all test signals, regardless of their strength or mass.
One key advantage of the approach is its flexibility. Because it does not depend on detailed assumptions about the source, it can potentially uncover unexpected phenomena. “Fifteen years ago, few would have predicted that we would routinely detect black holes with 70 or 80 solar masses,” Inguglia noted. “Today, we see them regularly. Future discoveries may surprise us again.”
As gravitational wave observatories become more powerful and data volumes grow, researchers expect conventional analysis methods to reach their limits. AI-based techniques could become essential tools in navigating what Inguglia describes as the “astronomical data jungle.”
Ultimately, the team hopes their work will shed light on the broader history of black holes — and, by extension, galaxies and the universe itself.
“We want to know whether there is a continuous evolutionary path from small black holes to the largest ones,” Inguglia said. “If we can answer that, we will better understand how cosmic structures formed and evolved over time.”
Sources : ÖAW
