Scientists who simulate physical processes usually need a separate, hand-coded solver for each job: one for flowing water, another for how a protein folds. Researchers at MIT, the University of British Columbia and elsewhere have built a single deep-learning system, WorldParticle, that handles many of them.
WorldParticle represents any material as tiny particles and predicts where they will move one step at a time. "It is a similar architecture to ChatGPT," said Peter Yichen Chen. "But instead of predicting one word at a time... we are predicting how these particles will move." Because it learns the underlying dynamics instead of relying on one equation, it can adapt to materials, forces and environments it never saw in training.
The team tested it on six types of physical phenomena, including deformable objects, granular materials and proteins. It modeled changes in protein structure about 100 times faster than numerical simulators. The work will be presented at SIGGRAPH Asia.
Source: MIT News, November 9, 2026 — https://news.mit.edu/2026/single-ai-model-simulates-many-substances-behavior-1109
