AI and the World's Hard Problems · Astronomy & space · arXiv · 2026-10-26
Physics Informed Neural Network for Multi-dimensional Turbulent Self-gravitating Flows
Grid-based simulations of turbulent, self-gravitating molecular clouds become expensive as resolution and dimensionality rise, so the authors applied physics-informed neural networks (PINNs) as a mesh-free alternative. Tested on isothermal clouds in 2D and 3D, subsonic and supersonic, and benchmarked against Athena++, PINNs recovered the global flow evolution over multiple free-fall times. Training cost rose by less than 25% when resolution went from 128^3 to 1024^3.