强化学习基础：环境、状态、动作、奖励与 MDP

G_t = r_t + γr_(t+1) + γ²r_(t+2) + ...

python -m pip install "gymnasium[classic-control]>=1,<2"

import gymnasium as gym


def discounted_return(rewards: list[float], gamma: float) -> float:
    value = 0.0
    for reward in reversed(rewards):
        value = reward + gamma * value
    return value


env = gym.make("CartPole-v1")
observation, info = env.reset(seed=42)
env.action_space.seed(42)
rewards = []
terminated = truncated = False

while not (terminated or truncated):
    action = env.action_space.sample()
    observation, reward, terminated, truncated, info = env.step(action)
    rewards.append(float(reward))

print({
    "steps": len(rewards),
    "return_gamma_1": discounted_return(rewards, 1.0),
    "return_gamma_099": round(discounted_return(rewards, 0.99), 3),
    "terminated": terminated,
    "truncated": truncated,
})
env.close()