MIT and UBC researchers develop WorldParticle, one AI model for many physical simulations
AIResearchers at MIT, UBC and collaborators have developed WorldParticle, a deep-learning system that simulates many physical processes with one architecture. It represents materials as particles and predicts their motion one step at a time, and it was demonstrated on six phenomena, including deformable objects, granular materials and proteins.
AIWhy it matters.
The source contrasts WorldParticle with the separate, hand-coded solvers scientists use today, showing what a single learned model would replace.
Source details
MIT News · AI · Adam Zewe | MIT News · news.mit.eduPublished · added here
Original title: A single AI model simulates the behavior of many substances, from sliding sand to folding proteins