RxEnvironments.jl

A Julia package for designing Active Inference environments

Project Overview

RxEnvironments.jl is a Julia package containing all boilerplate code to create reactive environments for control. The package is designed with Active Inference in mind, and therefore models all interaction between agents and environments as communication through a Markov Blanket. Since this communication abstracts away from the actual implementation of agents and environments and instead only focuses on Markov Blankets, we support multi-agent environments out of the box.

Key Features

• Reactive programming paradigm for environment design
• Built-in support for multi-agent environments
• Markov Blanket-based communication model
• Active Inference-oriented design
• Alternative to OpenAI Gym with unique features

Advantages Over OpenAI Gym

While OpenAI Gym is more established, RxEnvironments offers several unique advantages:

• Native support for multi-agent scenarios
• Reactive programming paradigm for more flexible control
• Built specifically for Active Inference applications
• Clean abstraction through Markov Blankets

Project Status

The package is feature-complete and stable. It does what it was designed to do, and I use it as the simulation backbone for my Active Inference research. Bug reports and contributions are welcome through the GitHub issues page.