Research & Papers

In-Context System Identification for Nonlinear Dynamics Using Large Language Models

Scientists use AI to automatically find the mathematical rules behind chaotic real-world systems.

Deep Dive

Researchers have created a new method that uses a large language model (LLM) to automatically discover the governing equations of complex, nonlinear systems from data. The AI iteratively refines candidate equations based on error feedback, outperforming the standard method (SINDy) on 63 test systems, including a nuclear reactor model. It finds equations that are more accurate and structurally similar to the true, hidden mathematical rules.

Why It Matters

This could accelerate scientific discovery by automatically revealing the fundamental principles behind everything from climate to engineering.