This repository is a curated list of research papers on Foundation Models (FMs) for Combinatorial Optimization (CO). It covers two main areas: leveraging existing Large Language Models (LLMs) for CO tasks and developing domain-specific FMs for CO. The collection is valuable for researchers and practitioners exploring the intersection of AI and optimization.

How It Works

The repository categorizes papers based on their approach: using LLMs to generate/improve solutions or algorithms, interpret solver behavior, automate problem formulation, or simplify tool usage. It also tracks research on building unified architectures or representations for domain-specific FMs capable of solving a wide range of COPs.

Quick Start & Requirements

This is a curated list of research papers, not a software package. No installation or execution is required.

Highlighted Details

• Comprehensive coverage of recent advancements (2022-2025) in LLMs for CO.
• Papers span various CO problems including TSP, VRP, MILP, SAT, and more.
• Categorization includes solution generation, algorithm design, problem formulation, and interpretability.
• Tracks the development of domain-specific FMs for CO.

Maintenance & Community

This is a static list of research papers. The primary contributor is ai4co. No community links or roadmap are provided.

Licensing & Compatibility

The repository itself is likely under a permissive license (e.g., MIT, Apache 2.0) as it's a collection of links. However, the linked research papers are subject to their respective publication licenses and copyright.

Limitations & Caveats

This repository is a bibliography and does not provide code, datasets, or executable models. Users must access and evaluate the individual research papers independently. The field is rapidly evolving, and new papers are published frequently.