CoupleEvo: Evolving Heuristics for Coupled Optimization Problems Using Large Language Models

Many real-world optimization tasks—like supply chain logistics or scheduling—involve tightly coupled subproblems where solutions must be coordinated. Existing LLM-driven heuristic design methods only handle single-problem settings. Now, researchers Thomas Bömer, Bastian Amberg, Max Disselnmeyer, and Anne Meyer propose CoupleEvo, a novel framework that leverages LLMs to evolve heuristics specifically for coupled optimization problems. The paper, accepted at GECCO 2026, introduces three evolutionary coordination strategies: the sequential strategy evolves heuristics for one subproblem after another; the iterative strategy alternates evolution across subproblems over generations; and the integrated strategy evolves all subproblems simultaneously.

Experimental results on two representative coupled optimization problems reveal that decomposition-based strategies (sequential and iterative) provide more stable convergence and higher solution quality compared to the integrated approach, which suffers from increased search complexity and variability. These findings highlight the critical importance of coordinating evolutionary search across interdependent subproblems. CoupleEvo demonstrates the potential of LLM-driven heuristic design for complex, coupled scenarios, with code publicly available. The work opens new directions for automated algorithm design in logistics, manufacturing, and other domains where subproblems cannot be solved in isolation.