Executable programs are under samples/.
| Environment | Algorithms |
|---|---|
| CartPole | A2C, DQN, PPO, Q-learning, REINFORCE, Sarsa |
| Maze | A2C, DQN, PPO, Q-learning, Sarsa |
| MountainCar | A2C, DQN, PPO, Q-learning, Sarsa |
| ContinuousMountainCar | A2C, DDPG, PPO+gSDE, SAC+gSDE |
| Pendulum | A2C, DDPG, PPO+gSDE, SAC+gSDE |
Some CartPole, Maze, ContinuousMountainCar, and Pendulum programs include image observation variants with feature layers.
Run a sample from the repository root so relative model and graphics paths are resolved correctly:
php samples/cartpole-dqn.php
MountainCar samples expose alternative reward transforms for experiments. Raw
evaluation reward remains the criterion for solving the original environment.
Review each sample’s constants and RL_* environment-variable helpers before
launching a long run.