The RLPy API

Base Classes

• Experiment
• Agent
• Policy
• Representation
• MDP Solvers
• Domain

Domains

• Acrobot with Euler Integration
• Acrobot with Runge-Kutta Integration
• Bicycle Balancing
• BlocksWorld
• Linear Chain MDP
• Fifty-State Chain MDP
• FlipBoard
• GridWorld
• HIV Treatment
• Helicopter Hovering
• Intruder Monitoring
• Mountain Car
• Persistent Search and Track Mission
• Pacman
• Pinball
• PuddleWorld
• RCCar
• System Administrator
• Swimmer
• Finite Track CartPole: Balance Task
• Finite Track CartPole: Balance Task (“Original”) per Sutton + Barto Definition
• Finite Track CartPole: Balance Task (“Modern”) with 3 (not 2) possible actions
• Finite Track CartPole: Swing-Up Task
• Finite Track CartPole: Swing-Up Task with Friction
• Infinite Track CartPole: Balance Task
• Infinite Track CartPole: Swing-Up Task

Helper modules

• Tools.run - Running Experiments in Batch
• Tools.results - Analyzing results
• Tools.hypersearch - Optimizing Hyperparameters