Semantic Evolution over Populations for LLM-Guided Automated Program Repair

TL;DR

EvolRepair introduces a semantic evolutionary algorithm for automated program repair using LLMs, enhancing diversity, repair family reasoning, and structured failure exploitation to improve repair success.

Contribution

It reformulates LLM-based APR as a semantic evolutionary process, replacing syntax-based operators with semantics-aware components powered by LLMs and structured feedback.

Findings

EvolRepair significantly outperforms existing LLM-based APR methods.

The approach effectively maintains repair diversity and explores alternative fix strategies.

Structured failure patterns guide the search towards more promising repair hypotheses.

Abstract

Large language models (LLMs) have recently shown strong potential for automated program repair (APR), particularly through iterative refinement that generates and improves candidate patches. However, state-of-the-art iterative refinement LLM-based APR approaches cannot fully address challenges, including maintaining useful diversity among repair hypotheses, identifying semantically related repair families, composing complementary partial fixes, exploiting structured failure information, and escaping structurally flawed search regions. In this paper, we propose a Population-Based Semantic Evolution framework for APR iterative refinement, called EvolRepair, that formulates LLM-based APR as a semantic evolutionary algorithm. EvolRepair reformulates the search paradigm of classic genetic algorithm for APR, but replaces its syntax-based operators with semantics-aware components powered by LLMs and structured execution feedback. Candidate repairs are organized into behaviorally coherent groups, enabling the algorithm to preserve diversity, reason over repair families, and synthesize stronger candidates by recombining complementary repair insights across the population. By leveraging structured failure patterns to guide search direction, EvolRepair can both refine promising repair strategies and shift toward alternative abstractions when necessary. Our experiments show that EvolRepair substantially improves repair effectiveness over existing LLM-based APR approaches.