gym-RLcourse

Environments used in the RL-course at Uppsala University. Implementing some examples from "Reinforcement Learning - An introduction" by Sutton and Barto, and some other very simple environments for educational purposes.

Installation

Requirements: gymnasium and numpy

git clone https://github.com/magni84/gym_RLcourse.git
cd gym_RLcourse
pip install -e .

Attribution

The code for GridWorldEnv is based on the FrozenLake environment licensed under an MIT license.

Environments

• MultiArmedBandits-v0 - Implements the multi-armed bandits of Section 2.3 in Sutton and Barto.
• GridWorld-v0 - Implements Example 4.1 in Sutton and Barto.
• GridWorld-AB-v0 - Implements Example 3.5 in Sutton and Barto.
• GridWorld-Windy-v0 - Implements Example 6.5 in Sutton and Barto.
• GridWorld-WindyKing-v0 - Same as GridWorld-Windy-v0 but with king moves.
• GridWorld-Maze-v0 - Implement mazes from Chapter 8 in Sutton and Barto.