DynaPlex 2 is a Python library for solving Markov Decision Problems and similar models (POMDP, HMM). It supports deep reinforcement learning, approximate dynamic programming, classical parameterized policies, and exact methods based on policy and value iteration. Models in DynaPlex 2 are written in Python, and exposed via a generic and vectorized interface.

DynaPlex 2 focuses on solving problems arising in Operations Management: Supply Chain, Transportation and Logistics, Manufacturing, etc.

Note

If you are new to MDPs, you might benefit from first reading the introduction to MDPs and going through the step-by-step tutorial, starting with the MDP formulation pages.

Contents¶

Getting started

Installation
A brief introduction to MDPs

Tutorial

Airplane Ticket Selling MDP formulation
- The Components of the MDP
- Policy
Airplane Ticket Selling MDP Python Code
Bin Packing MDP formulation
- The Components of the MDP
- Policy
Bin Packing MDP Python Code

Getting help and Contributing

Getting help