This repository contains the design and engine for the Lux AI Challenge Season 1, a competitive programming event focused on multi-variable optimization, resource gathering, and allocation in a 1v1 scenario. It targets AI developers and researchers seeking to build intelligent agents capable of strategic planning and opponent analysis. The primary benefit is a robust simulation environment for developing and testing complex AI strategies.

How It Works

The challenge simulates a turn-based strategy game where agents must manage resources, expand territory, and generate light to survive an encroaching darkness. The engine, built using the Dimensions package, facilitates 1v1 agent matches, allowing for detailed logging, replay generation, and local leaderboard evaluation using systems like TrueSkill.

Quick Start & Requirements

• Install via npm: npm install -g @lux-ai/2021-challenge@latest
• Requires Node.js version 12 or above.
• Starter kits are available for Python, Javascript, Rust, C++, Java, Typescript, and Kotlin.
• Official documentation and starter kits: https://github.com/Lux-AI-Challenge/Lux-Design-2021/tree/master/kits
• Visualizer: https://github.com/Lux-AI-Challenge/LuxViewer2021

Highlighted Details

• Supports local tournament simulation with TrueSkill ranking.
• Offers stateful and action-based replay generation.
• Provides a CLI tool for match execution and management.
• Includes community-developed tools like a faster Python Gym and visualizers.

Maintenance & Community

• Competition ran until December 6th, 2021.
• Community Discord available: https://discord.gg/aWJt3UAcgn
• Sponsors included QuantCo, J Ventures, and QAImera.

Licensing & Compatibility

• The repository itself is not explicitly licensed in the README. However, the competition was hosted on Kaggle, implying a focus on the competition rules rather than open-source licensing for the engine itself.

Limitations & Caveats

The competition has concluded, meaning active development and support for this specific season's engine may be limited. The primary use case is now for historical analysis or as a reference for future AI challenges.