《深入浅出强化学习编程实战》鸳鸯系统（策略迭代）

import random
import time
from testyuanyang import YuanYangEnv
class DP_Policy_Iter:
    def __init__(self,yuanyang):
        self.states=yuanyang.states
        self.actions=yuanyang.actions
        self.v=[0.0 for i in range(len(self.states)+1)]         #值函数
        self.pi=dict()                                          #创建空字典
        self.gamma= yuanyang.gamma
        self.yuanyang=yuanyang
        #初始化策略
        for state in self.states:
            flag1=0
            flag2=0
            flag1=yuanyang.collide(yuanyang.state_to_position(state))
            flag2=yuanyang.find(yuanyang.state_to_position(state))
            if flag1==1 or flag2==1:continue
            self.pi[state]=self.actions[int(random.random()*len(self.actions))]
    def policy_evaluate(self):
        #策略评估计算值函数
        for i in range(100):
            delta=0.0
            for state in self.states:
                flag1=0