Researchers have applied a machine learning technique to uncover unexpected features of the non-reciprocal forces that shape the behavior of a many-body system.
The study, published in PNAS, was conducted by experimental and theoretical physicists at Emory University. It combines a specially designed neural network with laboratory measurements from a dusty plasma, a type of ionized gas that contains interacting particles. Unlike most uses of artificial intelligence in science, which focus on analyzing data or making predictions, this work used AI to help reveal previously unknown 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.”
To read more, click here.