AI Discovers Dusty Plasma's Hidden Law

Emory University researchers used a physics-informed neural network to uncover the hidden rules of particle interactions in dusty plasma. The key point is that instead of relying on massive datasets, the model learned the relational equations using only fundamental physical laws and small-scale experimental data.

• Researchers floated ionized microplastic particles in a vacuum chamber filled with gas, then tracked their 3D trajectories using high-speed cameras and lasers.
• The model reconstructed non-reciprocal force, where the force one particle exerts on another is not equal to the reaction force, and it matched actual movements with over 99% agreement.
• It also helped correct existing textbook errors regarding the relationship between particle size and charge.

This framework could extend to other many-body systems, such as industrial materials like paint, cell colonies, and fish schools or bird flocks. In this approach, human experts establish the rules and handle interpretation, while AI uncovers the mathematical structure of complex interactions.