LLM as a Physicist: Autonomously Discovering Equations of Motion
CatAstro_Piyush · x · 2026-07-19
Maximilian Nägele and Florian Marquardt have proposed a general AI physicist method, handing over the complete physical discovery loop to a single LLM agent.
- Core Mechanism: The LLM (e.g., GPT-5) is given two general tools (executing arbitrary Python code, generating and viewing plots), a memory system, and a prompt to "act like a rigorous scientist."
- Autonomous Exploration: Without fine-tuning or task-specific tools, the agent actively selects initial conditions, observes system dynamics, and deduces unknown physical models (like equations of motion and Hamiltonians) by designing numerical experiments.
More from Research
- Datology webinar: Curation Studio matched top data baselines with under 20% compute — josh_wills · 2026-10-27
- KANLab: an interactive browser playground for Kolmogorov-Arnold Networks — Lukas Niehaus · 2026-10-27
- CMU's LLM2Cert synthesizes stability certificates verified via SMT and Lean — CarnegieMellonU · 2026-10-27
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- RSNA 2026 Deep Learning Labs return: eight hands-on AI classes over four days in Chicago — PRouzrokh · 2026-10-27
- Steering two orthogonal concept neurons makes a model compose "gnome" — Sauers_ · 2026-10-27