Frozen Lake Environment (OpenAI Gym) Solution using Random Policy

FrozenEnv_RandomPolicy.py

import numpy as np
import gym
import time

def generate_random_policy():
	# Generates a vector of shape (16,) with an action between 0 and 3 (inclusive)
	return np.random.choice(4, size=((16)))

def run_episode(env, policy, n_episodes=100, render=False):
	total_reward = 0

	# Reset the environment after each episode
	s = env.reset()

	for i in range(n_episodes):
		if render:
			env.render()
		s, reward, done, _ = env.step(policy[s])
		total_reward += reward
		if done:
			break
	
	return total_reward

def evaluate_policy(env, policy, n_episodes=100):
	total_rewards = 0.0
	for _ in range(n_episodes):
		total_rewards += run_episode(env, policy)
	return total_rewards / n_episodes

if __name__ == '__main__':
	env = gym.make('FrozenLake-v0')
	n_policies = 2000
	start = time.time()
	policy_set = [generate_random_policy() for _ in range(n_policies)]
	policy_score = [evaluate_policy(env, p) for p in policy_set]
	end = time.time()

	print("Best score = %0.2f. Time taken = %4.4f seconds" %(np.max(policy_score) , end - start))