Chapter 2, Reinforcement Learning and Deep Reinforcement Learning, prepped you with the basics of reinforcement learning. With that theoretical background, we will be able to implement some cool algorithms. Before that, we will make sure we have the required tools and libraries at our disposal.
We can actually write cool reinforcement learning algorithms in Python without using any higher-level libraries. However, when we start to use function approximators for the value functions or the policy, and especially if we use deep neural networks as the function approximators, it is better to use highly optimized deep learning libraries instead of writing our own routines. A deep learning library is the major tool/library that we will need to install. There are different libraries out there today: PyTorch, TensorFlow...