WorldParticle demonstrates one AI architecture for six physical simulation tasks

Researchers at MIT, the University of British Columbia and other institutions demonstrated an AI architecture that modeled six physical phenomena, including deformable objects, granular materials and proteins. MIT says WorldParticle achieved higher accuracy than numerical simulation methods; researcher Minghao Guo says it modeled changes in protein structures about 100 times faster than numerical simulators. WorldParticle represents materials as particles and predicts their next positions and properties. It separates external forces before each prediction and incorporates them afterward, allowing it to handle unfamiliar materials and conditions. A particle-merging technique reduces computation while retaining fine detail. The researchers plan further work toward a full general-purpose simulation pipeline.