Local-MIP

Open-Source Local Search for Mixed Integer Programming

Quick Start Software Source Code Tutorials Examples

Project Leader Shaowei Cai Developer Peng Lin

🏆 CP 2024 Best Paper Award Peng Lin, Mengchuan Zou, Shaowei Cai 📄 Artificial Intelligence 2025 Peng Lin, Shaowei Cai, Mengchuan Zou, Jinkun Lin

Customizable Modules

Initialization Restart Strategies Neighbor Operators Weighting Schemes Scoring Functions

Highlights

Fast Feasible Solutions

Local-MIP is designed to find high-quality feasible solutions quickly, especially under short time budgets.

Flexible APIs and Callbacks

C++ and Python APIs support file-based solving, in-memory modeling, and callback-based customization of the search.

Ready to Embed

Use Local-MIP as a standalone solver or integrate it into larger systems as a warm-start primal heuristic.

Quick Start

Step 1

Get the Code

Start with a release archive, the source repository, or an archived version for paper reproduction.

Open Software →
Step 2

Choose a Path

Use Local-MIP from the command line, embed it from C++, call it from Python, or customize the search with callbacks.

Browse Tutorials →
Step 3

Build and Run

Use the solver README for the core workflow and the Python bindings README for Python setup and demos.

Full Quick Start →

Software using Local-MIP

Integrated Solver

NVIDIA cuOpt

cuOpt uses Local-MIP as part of its GPU-accelerated optimization stack.

Official Site Technical Reference
Integrated Solver

Exact

Exact includes Local-MIP in its pseudo-Boolean solving workflow.

Source Code Solver Description

Documentation & Resources

Start Here

Getting Started

Use these pages when you want the shortest path from downloading Local-MIP to a first successful run.

Quick Start Software
Interfaces

Technical Documentation

Learn the main interfaces for file-based solving, in-memory modeling, parameter control, and callback customization.

Tutorials Command-Line Usage Modeling API Parameters Reference Callback System Examples
Results

Performance and Research

See benchmark records, research papers, and the broader context behind Local-MIP.

MIPLIB Records Papers
Latest Release: GitHub Releases License: MIT License