A single AI model simulates the behavior of many substances, from sliding sand to folding proteins

Researchers at MIT, the University of British Columbia, and other institutions have developed a unified deep-learning system called WorldParticle that simulates diverse physical phenomena, from sliding sand to folding proteins. The model represents materials as tiny particles and predicts their motion over time by learning underlying dynamics rather than relying on hand-coded equations. By integrating a particle merging technique and external force corrections, the scalable framework can generalize to unseen materials and environmental conditions. Open article →