Knowledge Base Resources

This website is the official documentation for pymdp, a Python library for implementing active inference. Active inference is a computational framework from neuroscience used to model how intelligent agents—like animals or sophisticated AI—perceive, act, and learn. The core idea is that these agents are driven to minimize surprise and uncertainty about their environment, a concept formalized by the Free Energy Principle.

This documentation is a useful resource for anyone looking to build and simulate these agents. It provides a comprehensive guide to using pymdp to create generative models, which are the internal belief systems that agents use to understand and predict their world. You will learn how to design agents that can make decisions, update their beliefs based on new observations, and pursue goals.

The website is organized into several key sections:

Getting Started: This section provides tutorials that walk you through the basics of setting up your first active inference agent.

API Reference: For more advanced users, this section offers a detailed reference to all the classes and functions in the pymdp library.

Examples: A collection of real-world examples and case studies that showcase how pymdp can be used to model specific behaviors and tasks.

Theoretical Background: For those interested in the underlying theory, this section provides in-depth explanations of active inference and the Free Energy Principle.

The documentation is written in a clear and accessible style, with plenty of code examples to help you along the way. Whether you're a student or researcher, this website will help you get started with pymdp and active inference.