Select the penalty for forbidden states:

Note: Changing this will affect the results in Optimality, Q-Learning, and TD Linear modules.

Click on cells to cycle actions (这个修改只对当前页面下方的 "Evaluate Policy"（策略评估）按钮生效):

Convergence Analysis (PI vs Truncated PI)

Step-by-Step Visualization

Note: 傅里叶变换阶数越高，结果越精确，但计算量也越大。此处展示初始策略为随机的预计算结果。