Data-driven discovery of governing differential equations across physical systems
A new review paper proposes a problem-oriented perspective on data-driven differential equation discovery, a field that uses AI to infer governing laws from data. The paper introduces a phase diagram to organize discovery problems by complexity and a Representation-Evaluation-Optimization (REO) framework to abstract the discovery process. This approach aims to shift focus from individual algorithms to fundamental principles of discoverability, with applications across various scientific domains. AI
IMPACT Provides a structured framework for advancing AI-driven scientific discovery in differential equations.