Making gravitational waves easier to hear
Mathyn van Dael defended his PhD thesis at the Department of Mechanical Engineering on January 23rd.
In 2015, scientists achieved something extraordinary: they detected gravitational waves for the very first time. Almost a century after Albert Einstein predicted their existence, these tiny ripples in spacetime were finally observed, confirming a cornerstone of modern physics. More importantly, the discovery opened an entirely new way of exploring the universe. Instead of only observing the cosmos through light, scientists can now listen to it, capturing signals from extreme events such as colliding black holes and neutron stars鈥攅vents that would otherwise remain hidden from view.
Gravitational waves travel across the universe largely undisturbed, carrying unique information about their origins and about the fundamental laws governing space and time. Some of these signals may even contain traces from the earliest moments after the Big Bang. Detecting them on Earth, however, is an immense challenge: the distortions they cause in spacetime are unimaginably small and require instruments of unprecedented precision.
Precision at an unthinkable scale
Gravitational-wave detectors such as LIGO and Virgo use kilometer-long laser interferometers to measure changes in length that are smaller than an atomic nucleus. Even the slightest disturbance鈥攆rom seismic activity to thermal fluctuations鈥攃an mask a gravitational-wave signal. Maintaining the stability required for these measurements is therefore a continuous and demanding task.
This is where control systems come in. Acting as highly sophisticated autopilots, they constantly adjust mirrors, lasers, and suspension systems to counteract disturbances and keep the detector in its optimal operating state.
When traditional control reaches its limits
As detectors become more sensitive, their control becomes increasingly complex. Many subsystems interact across different physical domains, their dynamics change over time, and experimenting with the system is costly. Traditional model-based control approaches struggle to remain flexible and robust under these conditions.
To support the next leap in detector performance, new control strategies are needed鈥攐nes that can adapt to complexity rather than trying to eliminate it.
Learning to control by using data
In his research, Mathyn van Dael demonstrated how data-driven control methods can learn directly from the vast amounts of data produced by gravitational-wave detectors. By adapting to time-varying behavior and complex couplings within the system, these techniques improve noise suppression and stability without requiring extensive downtime for re-tuning.
The result is a detector that not only hears weaker signals, but is also available for scientific observations more often.
From research to reality
Several of the developed methods have been successfully implemented on the Virgo detector in Pisa, Italy, where they were used during a recent observing run. Their deployment shows that data-driven control is not merely an academic concept, but a practical solution with real impact on detector performance and reliability.
Enabling the next generation of detectors
Future observatories, such as the Einstein Telescope, aim to push sensitivity even further, potentially detecting signals from the earliest phases of the universe. Achieving these ambitious goals will require advances not only in physics and hardware, but also in control engineering.
This PhD research highlights that data-driven control techniques are an essential building block for the next generation of gravitational-wave detectors, helping humanity listen more clearly鈥攁nd for longer鈥攖o the deepest signals of the cosmos.
Title of PhD thesis: . Supervisors: Prof. Tom Oomen, Dr. Gert Witvoet, and Dr. B.L. Swinkels.