Automated Reformulation of Robust Optimization via Memory-Augmented Large Language Models

2026-05-12

AI Summary

This work addresses the labor-intensive and expertise-dependent process of manually reformulating robust optimization models with uncertainty into tractable deterministic equivalents, which hinders their broad adoption. To overcome this limitation, the authors propose AutoREM, a fine-tuning-free, memory-augmented framework that leverages structured textual experience memory to guide large language models in performing accurate, multi-step reformulations automatically. The key contributions include AutoRO-Bench, the first benchmark specifically designed for evaluating robust optimization reformulation, and a novel experience memory mechanism that requires neither expert knowledge nor model parameter updates and is transferable across different base models. Experimental results demonstrate that AutoREM significantly improves both accuracy and efficiency in reformulating robust optimization problems, consistently across in-distribution and out-of-distribution instances and various foundational large language models.

Abstract

Robust optimization (RO) provides a principled framework for decision-making under uncertainty, but its practical use is often limited by the need to manually reformulate uncertain optimization models into tractable deterministic counterparts. Recent large language models (LLMs) have been shown promising for automating optimization formulation, yet RO reformulation remains challenging because it requires precise multi-step reasoning and mathematically consistent transformations. To facilitate systematic evaluation of LLM-based reformulation, for which no dedicated benchmark currently exists, we develop AutoRO-Bench, a benchmark featuring an automated data generation pipeline for the core RO reformulation task and a curated dataset for the RO application task. To address the reformulation challenge, we propose Automated Reformulation with Experience Memory (AutoREM), a tuning-free memory-augmented framework that autonomously builds a structured textual experience memory by reflecting on past failed trajectories through a tailored offline adaptation procedure. AutoREM requires neither domain-specific expert knowledge nor parameter updates, and the resulting memory readily transfers across different base LLMs. Experimental results show that AutoREM consistently improves the accuracy and efficiency of RO reformulation across in-distribution datasets, out-of-distribution datasets, and diverse base LLMs.

Problem

• Robust Optimization
• Automated Reformulation
• Large Language Models
• Uncertainty
• Optimization Modeling

Innovation

• memory-augmented LLMs
• robust optimization reformulation
• AutoREM
• AutoRO-Bench
• tuning-free framework