Large Lemma Miners: Can LLMs do Induction Proofs for Hardware?

TL;DR

This paper demonstrates that large language models, with proper prompting, can generate inductive proofs for hardware verification, potentially reducing manual effort in formal verification processes.

Contribution

It introduces a neurosymbolic approach with prompting frameworks enabling LLMs to produce correct inductive invariants for hardware designs.

Findings

LLMs can generate inductive proofs for open-source RTL designs

At least one prompt setup succeeded in 84% of cases

Proper reprompting improves proof generation success rate

Abstract

Large Language Models (LLMs) have shown potential for solving mathematical tasks. We show that LLMs can be utilized to generate proofs by induction for hardware verification and thereby replace some of the manual work done by Formal Verification engineers and deliver industrial value. We present a neurosymbolic approach that includes two prompting frameworks to generate candidate invariants, which are checked using a formal, symbolic tool. Our results indicate that with sufficient reprompting, LLMs are able to generate inductive arguments for mid-size open-source RTL designs. For 84% of our problem set, at least one of the prompt setups succeeded in producing a provably correct inductive argument.