One AI model simulates sand, fluids and proteins
Ongoing: AI Hot
Researchers at MIT, the University of British Columbia (UBC) and other institutions built WorldParticle, a single deep-learning system that simulates many materials. It treats each material as tiny particles and predicts how they move, one step at a time.
Peter Yichen Chen, an assistant professor at UBC and a former MIT postdoc, co-led the work with MIT postdoc Minghao Guo. The team says the model learns how matter moves, so it can handle materials and forces it did not see in training. Today, scientists need separate tools for each process. Chen calls this a first step towards a general simulation model. The paper is open access and will be presented at a computer graphics conference.