The US National Science Foundation’s Laser Interferometer Gravitational-wave Observatory (LIGO) has long been hailed as the most precise ruler in the world. Capable of measuring motions smaller than 1/10,000 the width of a proton, LIGO detects gravitational waves—ripples in space-time caused by cosmic events like black hole collisions. Now, artificial intelligence is set to enhance this groundbreaking observatory’s capabilities even further.
LIGO, which operates two facilities in Washington and Louisiana, first made headlines in 2015 with the direct detection of gravitational waves, a discovery that earned three of its founders the Nobel Prize in Physics in 2017. Since then, improvements to LIGO’s interferometers have enabled it to detect an average of one black hole merger every three days during its current science run. Together with the Virgo detector in Italy and KAGRA in Japan, LIGO has identified hundreds of black hole merger candidates, including some involving neutron stars.
AI Enhancements: A New Era for LIGO
Researchers are now seeking to further enhance LIGO’s capabilities to detect a wider variety of black hole mergers, including those that might belong to a hypothesized intermediate-mass class. These enhancements would also allow LIGO to identify black holes with eccentric orbits and detect mergers earlier in their coalescing process.
To achieve these goals, scientists at Caltech and the Gran Sasso Science Institute in Italy have collaborated with Google DeepMind to develop an AI method called Deep Loop Shaping. This innovative approach aims to reduce unwanted noise in LIGO’s detectors—noise that can mask gravitational-wave signals by causing tiny vibrations in LIGO’s giant mirrors.
“We were already at the forefront of innovation, making the most precise measurements in the world, but with AI, we can boost LIGO’s performance to detect bigger black holes,” says Rana Adhikari, professor of physics at Caltech.
The Science Behind the Silence
LIGO’s facilities are shaped like enormous “L’s,” with each arm containing a vacuum tube that houses advanced laser technology. Lasers bounce between giant suspended mirrors, detecting minute changes in length caused by gravitational waves. However, background noise, such as ocean waves, can cause unwanted vibrations in these mirrors.
Christopher Wipf, a gravitational-wave interferometer research scientist at Caltech, explains, “It’s as if the LIGO detectors are sitting at the beach. Water is sloshing around on Earth, and the ocean waves create low-frequency vibrations that disturb both LIGO facilities.”
The AI method works similarly to noise-canceling headphones, counteracting these vibrations. Traditional feedback control systems already manage this noise, but they introduce a higher-frequency quiver, akin to a faint hiss in headphones. The new AI approach aims to reduce this self-inflicted noise, improving LIGO’s sensitivity to gravitational waves in the lower-frequency range.
AI’s Broader Implications
Jan Harms, a professor at Gran Sasso Science Institute, initiated the collaboration with Google DeepMind to explore AI’s potential in controlling vibrations in LIGO’s mirrors. The team employed reinforcement learning, training the AI algorithm to manage noise by simulating multiple LIGOs in parallel.
“This method requires a lot of training,” Adhikari says. “We supplied the training data, and Google DeepMind ran the simulations. The result is beautiful—the algorithm works to suppress mirror noise.”
Richard Murray, a professor at Caltech, notes that AI can exploit features in systems that classical methods might overlook. This approach could revolutionize technologies beyond LIGO, impacting fields like aerospace, robotics, and structural engineering.
Looking to the Future
The new AI method has been tested on LIGO for only an hour, but the team plans to conduct longer tests and implement the method across multiple systems. “This is a tool that changes how we think about what ground-based detectors are capable of,” Wipf says. “It makes an incredibly challenging problem less daunting.”
The study, titled “Improving cosmological reach of LIGO using Deep Loop Shaping,” was published in Science and supported by the National Science Foundation. As LIGO continues to push the boundaries of gravitational-wave astronomy, AI promises to play a crucial role in its future discoveries.
