Getting Started

Tutorial goes here…

Downloading ERA5 forecasts

Generating Synthetic Forecasts

Running a simulation in Gym

The RL-HAB simulation environment is built on the Gym API. This API is a standard API for most reinforcement learning applications, and can be easily coupled with out-of-the-box RL implementations such as StableBaselines3. There are two types of custom Gym environments to choose from: RLHAB_gym_SINGLE.py and RLHAB_gym_DUAL.py. - What are the two types of custom Gym envrionments and what do each mean - Requirements for every environment (ERA5 + Synth) - Modify the env_config.py file

  • What do each of the variables here mean? Which ones are open for modification? - Should this link to the API?

  • Dynamics noise

  • Example call of the environment with random actions

Training an agent with DQN

  • StableBaselines3 DQN explanation
    • Simple training example usage

  • Optuna explanation + usage
    • Optuna dashboard

  • Weights and Biases explanation + usage
    • Screenshot of W&Biases

Evaluating an Agent