AI Finds New Plasma Law
- Emory University physicists reported that a physics-tailored neural network learned force laws from laboratory dusty-plasma experiments, then exposed interaction rules that standard models had missed in this charged-particle system. - The PNAS study said the model inferred non-reciprocal particle forces from three-dimensional trajectories with R2 above 0.99, and the team used two independent mass estimates to check the result. - Dusty plasma appears in space and industrial settings, and the authors say the same method could help infer laws in other many-body systems. (pnas.org)
Plasma is a gas with its atoms stripped into charged pieces, and dusty plasma adds tiny grains that push and pull on each other in hard-to-measure ways. Emory University researchers said a neural network trained on those motions uncovered force rules that standard theory had missed. (pnas.org) (news.emory.edu) The team, led by Wentao Yu, Eslam Abdelaleem, Ilya Nemenman and Justin C. Burton, published the work in the Proceedings of the National Academy of Sciences on July 31, 2025. Their model learned from three-dimensional trajectories of particles moving inside a laboratory dusty plasma. (pnas.org) (pmc.ncbi.nlm.nih.gov) Those particles do not always obey the usual equal-and-opposite intuition from school physics. In dusty plasma, one particle can affect another differently than it is affected back, a one-way interaction physicists call non-reciprocal. (pmc.ncbi.nlm.nih.gov) (sciencedaily.com) The Emory group built those asymmetries into the network instead of treating artificial intelligence as a black box. The paper said the system accounted for symmetries, non-identical particles and external forces while inferring interparticle forces directly from the data. (pnas.org) (pmc.ncbi.nlm.nih.gov) The model reproduced the effective non-reciprocal forces with R2 greater than 0.99. The researchers also said they validated the approach by inferring particle masses in two independent ways that agreed with each other. (pmc.ncbi.nlm.nih.gov) (news.emory.edu) That accuracy let the team test assumptions that plasma physicists often use as shortcuts. The paper said the measurements found large deviations from common expectations about particle charge and screening length, the distance scale over which electric forces fade in a plasma. (pmc.ncbi.nlm.nih.gov) (sciencedaily.com) Dusty plasma is not just a lab curiosity. The paper said it appears in planetary and space environments, and Emory’s release linked related plasma conditions to settings ranging from interstellar dust to wildfire smoke and industrial materials. (pmc.ncbi.nlm.nih.gov) (sciencedaily.com) The authors argued the method could travel beyond plasma physics because many-body systems all share the same basic problem: too many interacting pieces moving at once. They pointed to possible uses in colloids such as paint and ink, and even in clusters of living cells. (news.emory.edu) (pmc.ncbi.nlm.nih.gov) Burton said the point was not just better prediction, but readable physics. The result leaves researchers with a new set of force laws for dusty plasma and a template for using machine learning to extract laws from real experiments. (news.emory.edu) (pnas.org)
