Formal Safety Verification for Nonlinear Systems with Generative Barrier Certificate

Abstract

Safety verification is a fundamental problem in control theory. Barrier certificates (BCs) provide a powerful formal mechanism, yet deriving BCs is computationally intensive. This paper introduces a generative framework that leverages large language models (LLMs) to synthesize BCs through reasoning. Based on the classical Sum-of-Squares (SOS) approach, we train a domain-specific LLM capable of generating high-quality BC candidates for nonlinear systems. Then, the LLM-generated BCs transform the intractable Bilinear Matrix Inequality (BMI) solving problems into convex Linear Matrix Inequality (LMI) feasibility test, significantly improving efficiency while preserving correctness. Experimental results show that our generative method achieves several orders of magnitude speedup over traditional numerical BC approaches and, perhaps surprisingly, surpasses the state-of-the-art dedicated neural BC model. These findings mark a substantive step toward integrating generative AI with formal safety verification for dynamical systems.

Disclosure

“erages large language models (LLMs) to synthesize BCs through reasoning. Based on the classical Sum-of-Squares (SOS) approach, we train a domain-specific LLM capable of generating high-quality BC candidates for nonlinear systems. Then, the LLM-generated BCs transform the intractable Bilinear Matrix Inequality (BMI) solving problems into convex Linear Matrix Inequality (LMI) feasibility test, significantly improving efficiency while preserving correctness. Experimental results sh”

arXiv metadata: abstract

Classification

Substantial mathematical content or result generation

Multiplier

10

Verified

2026-08-23T18:51:38.655215Z

Structural counts