Co-evolving Agent Architectures and Interpretable Reasoning for Automated Optimization

2026-04-19

AI Summary

This work addresses the limitations of existing large language model (LLM)-driven approaches to operations research (OR) optimization, which rely on handcrafted reasoning-execution pipelines and struggle to dynamically coordinate problem understanding, modeling, solving, and debugging in complex tasks. To overcome this, the authors propose EvoOR-Agent, a novel framework that jointly models agent architecture and reasoning trajectories as evolvable objects. The approach explicitly represents workflow topology via an active-edge network and introduces graph-guided path-condition recombination alongside multi-granularity semantic mutation mechanisms to enable adaptive and interpretable optimization automation. Experimental results demonstrate that EvoOR-Agent significantly outperforms zero-shot LLMs, fixed-pipeline agents, and representative evolutionary frameworks on heterogeneous OR benchmarks, achieving simultaneous improvements in solution quality and structural interpretability.

Technology Category

• Search and Optimization: Evolutionary Computation
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Automating operations research (OR) with large language models (LLMs) remains limited by hand-crafted reasoning--execution workflows. Complex OR tasks require adaptive coordination among problem interpretation, mathematical formulation, solver selection, code generation, and iterative debugging. To address this limitation, we propose EvoOR-Agent, a co-evolutionary framework for automated optimization. The framework represents agent workflows as activity-on-edge (AOE)-style networks, making workflow topology, execution dependencies, and alternative reasoning paths explicit. On this representation, the framework maintains an architecture graph and evolves a population of reasoning individuals through graph-mediated path-conditioned recombination, multi-granularity semantic mutation, and elitist population update. A knowledge-base-assisted experience-acquisition module further injects reusable OR practices into initialization and semantic variation. Empirical results on heterogeneous OR benchmarks show that the proposed framework consistently improves over zero-shot LLMs, fixed-pipeline OR agents, and representative evolutionary agent frameworks. Case studies and ablation analyses further indicate that explicit architecture evolution and graph-supported reasoning-trajectory search contribute to both performance improvement and structural interpretability. These results suggest that treating agent architectures and reasoning trajectories as evolvable objects provides an effective route toward adaptive and interpretable automated optimization.

Problem

• automated optimization
• large language models
• agent architectures
• interpretable reasoning
• operations research

Innovation

• co-evolutionary framework
• interpretable reasoning
• agent architecture evolution
• graph-mediated recombination
• automated optimization