This is a web version of my bachelor thesis. This github repository is mainly here for reproducibility purposes. The book is available at the page
To reproduce the experiments in the last sections, simply open the main.py file, change the parameters to the desired parameter and run! This will generate a folder inside the log folder with the date as folder name. Inside is a text file with the hyperparameters used, and a csv file with the evolution of the parameters after each episode, as well as the mean reward each episode.
The book was rendered and published with quarto 1.5.57, using a python 3.11.7 (see requirements.txt), Jupyter 5.7.2.
Keywords: reinforcement learning, policy gradient, REINFORCE, Runge-Kutta method, differential equation, scientific machine learning