Language models optimize complex problems using strategy and tools
Think in Strategies, Act with Tools: Traceable Agentic Optimization for Combinatorial Problems
Artificial Intelligence
Summary
Solving complicated puzzles often means trying many solutions and learning from past attempts. The authors found that when AI language models try to solve these puzzles by generating answers, they forget what happened before. To fix this, they made a system that remembers previous tries and plans its next move using that memory. This system also builds special helper tools to improve its guessing. Their tests show this method produces better answers and is easier to follow step-by-step.
What this means in practice
- •For software engineers: Improve algorithms that solve resource allocation and scheduling by adding strategic memory and custom tools to optimization processes.
- •For logistics planners: Refine route and delivery optimization systems by enabling traceable strategy updates that enhance solution quality.
Authors
Zhongnian Li, Shi Xu, Peng Ying, Weikang Gong, Zhixiao Wang, Xinzheng Xu
Abstract
Optimization by PROmpting (OPRO) directly leverages Large Language Models (LLMs) to generate candidate solutions through natural-language prompting. However, OPRO's optimization process is difficult to track, which prevents LLMs from capturing useful information accumulated over previous iterations, thereby limiting solution quality. To address this limitation, we propose a Tool-Augmented Strategic Agent (TASA) which summarizes and exploits trajectory to improve solution quality. Specifically, TASA plans the strategy for each step based on the optimization trajectory, which decomposes the optimization process into traceable strategy selections. Moreover, we leverage LLM to construct problem-specific tools and introduce a strategy-guided adaptive tool planner, which efficiently utilizes the information of trajectory to update solutions. We establish local convergence under persistent exploration and evaluate the complete framework under a shared candidate-evaluation budget. The results support its solution quality and the traceability of strategy-guided search.