Physicists have used a machine learning approach to reveal unexpected details about how particles interact in complex systems. Their work focuses on non-reciprocal forces, where one particle influences another differently than it is influenced in return.
The findings, published in PNAS, come from a collaboration between experimental and theoretical physicists at Emory University. By combining a custom neural network with laboratory data from a dusty plasma, the team showed that artificial intelligence can do more than analyze data or make predictions. It can help uncover entirely new physical laws.
“We showed that we can use AI to discover new physics,” says Justin Burton, an Emory professor of experimental physics and senior co-author of the paper. “Our AI method is not a black box: we understand how and why it works. The framework it provides is also universal. It could potentially be applied to other many-body systems to open new routes to discovery.”
High-Precision Insights Into Dusty Plasma Forces
The study offers one of the most detailed descriptions to date of the physics governing dusty plasma. This system consists of ionized gas filled with interacting charged particles, including tiny grains of dust.
Using their AI model, the researchers were able to describe non-reciprocal forces with more than 99% accuracy. These forces are notoriously difficult to measure and model.
“We can describe these forces with an accuracy of more than 99%,” says Ilya Nemenman, an Emory professor of theoretical physics and co-senior author of the paper. “What’s even more interesting is that we show that some common theoretical assumptions about these forces are not quite accurate. We’re able to correct these inaccuracies because we can now see what’s occurring in such exquisite detail.”
The team believes this method could be applied broadly to systems made up of many interacting components. These range from industrial materials such as paint and ink to groups of living cells.
The study’s first author is Wentao Yu, who worked on the project as an Emory PhD student and is now a postdoctoral fellow at the California Institute of Technology. Co-author Eslam Abdelaleem also contributed as an Emory graduate student and is now a postdoctoral fellow at Georgia Tech.
The research was primarily supported by the National Science Foundation, with additional funding from the Simons Foundation.
“This project serves as a great example of an interdisciplinary collaboration where the development of new knowledge in plasma physics and AI may lead to further advances in the study of living systems,” says Vyacheslav (Slava) Lukin, program director for the NSF Plasma Physics program. “The dynamics of these complex systems is dominated by collective interactions that emerging AI techniques may help us to better describe, recognize, understand and even control.”
The Fourth State of Matter Explained
Plasma is often called the fourth state of matter. In this state, gas becomes ionized, meaning electrons and ions move freely and create unique properties such as electrical conductivity. Plasma makes up about 99.9% of the visible universe, from the solar wind streaming from the Sun to lightning strikes on Earth.
Dusty plasma includes additional charged dust particles and appears in many environments, from the rings of Saturn to the Earth’s ionosphere.
On the Moon, weak gravity allows charged dust to hover above the surface. “That’s why when astronauts walk on the moon their suits get covered in dust,” Burton explains.
On Earth, dusty plasma can form during wildfires when soot mixes with smoke. These charged particles can disrupt radio signals, making communication more difficult for firefighters.
Tracking Particle Motion in 3D
Burton’s lab studies dusty plasma and similar materials by recreating them in controlled experiments. Researchers suspend tiny plastic particles in a plasma-filled vacuum chamber to simulate more complex systems. By adjusting gas pressure, they can mimic real-world conditions and observe how particles respond to different forces.
For this project, Burton and Yu developed a tomographic imaging method to capture the three dimensional (3D) motion of particles. A laser sheet moves through the chamber while a high speed camera records images. These snapshots are then combined to reconstruct the positions of dozens of particles over time, allowing researchers to track their motion in detail.
Using AI To Understand Collective Motion
Nemenman, a theoretical biophysicist, studies how complex systems emerge from simple interactions. He is especially interested in collective motion, such as how cells move within the human body.
“General questions of how a whole system arises from interactions of tiny parts are very important,” Nemenman explains. “In cancer, for instance, you want to understand how the interaction of cells may relate to some of them breaking away from a tumor and moving to a new place, becoming metastatic.”
Compared to living systems, dusty plasma offers a simpler environment for testing new ideas. This made it an ideal case for exploring whether AI could uncover new physical principles.
“For all the talk about how AI is revolutionizing science, there are very few examples where something fundamentally new has been found directly by an AI system,” Nemenman says.
Designing a Neural Network for Discovery
Building the AI model required careful planning. Unlike systems trained on massive datasets, this project had limited experimental data.
“When you’re probing something new, you don’t have a lot of data to train AI,” Nemenman explains. “That meant we would have to design a neural network that could be trained with a small amount of data and still learn something new.”
The team spent more than a year refining the design through weekly meetings.
“We needed to structure the network to follow the necessary rules while still allowing it to explore and infer unknown physics,” Burton explains.
“It took us more than a year of back-and-forth discussions in these weekly meetings,” Nemenman adds. “Once we came up with the correct structure of the network to train, it turned out to be fairly simple.”
The final model separated particle motion into three main influences: drag from velocity, environmental forces such as gravity, and forces between particles.
Surprising Results and New Insights
After training on 3D particle trajectories, the AI successfully captured complex interactions, including asymmetrical forces between particles.
The researchers compare this behavior to two boats moving across a lake. Each boat creates waves that affect the other. Depending on their positions, these waves can push or pull the boats differently.
“In a dusty plasma, we described how a leading particle attracts the trailing particle, but the trailing particle always repels the leading one,” Nemenman explains. “This phenomenon was expected by some but now we have a precise approximation for it which didn’t exist previously.”
The results also challenge earlier theories. One long standing idea suggested that a particle’s electric charge increases in direct proportion to its size. The new findings show that while larger particles do carry more charge, the relationship is more complex and depends on factors such as plasma density and temperature.
Another assumption held that forces between particles decrease exponentially with distance in a way that does not depend on particle size. The AI model revealed that particle size does affect how quickly these forces weaken.
The team confirmed these conclusions through additional experiments.
A New Tool for Exploring Complex Systems
The researchers developed a physics based neural network that can run on a standard desktop computer. They believe it offers a flexible framework for studying many-body systems across different fields.
Nemenman will soon teach at the Konstanz School of Collective Behavior in Germany, where scientists study systems ranging from flocks of birds to human crowds.
“I’ll be teaching students from all over the world how to use AI to infer the physics of collective motion — not within a dusty plasma but within a living system,” he says.
Even with these advances, human expertise remains essential. Scientists must design the models carefully and interpret the results.
“It takes critical thinking to develop and use AI tools in ways that make real advances in science, technology and the humanities,” Burton says.
He remains optimistic about the future.
“I think of it like the Star Trek motto, to boldly go where no one has before,” Burton says. “Used properly, AI can open doors to whole new realms to explore.”