maymayw

import numpy as np
from rl.lib.envs.gridworld import GridworldEnv

env = GridworldEnv()

def valiter(env, dr=1.0, theta = 0.0001):
    def onestepahead(s, V):
        q = np.zeros(env.nA)
        for a in np.arange(env.nA):
            [(prob, nexts, reward, done)] = env.P[s][a]