An AI just started cracking math problems about the laws of physics that have basically been bullying scientists for centuries.
arXiv · March 17, 2026 · 2603.14353
The Takeaway
While humans usually need supercomputers to approximate how things like air or heat flow, this AI found a way to derive the exact formulas for these processes autonomously. It successfully solved 100 classic physics problems and even discovered 'hidden' solutions that had never been recorded in scientific literature.
From the abstract
Partial differential equations (PDEs) encode fundamental physical laws, yet closed-form analytical solutions for many important equations remain unknown and typically require substantial human insight to derive. Existing numerical, physics-informed, and data-driven approaches approximate solutions from data rather than systematically deriving symbolic expressions directly from governing equations. Here we introduce LawMind, a law-driven symbolic discovery framework that autonomously constructs c