Reinforcement_learning_FrozenLake.py

# The OpenAI-gym- for environment
# Python 3.11: Coursera Lab exercise. 

# check readme file for more details

import numpy as np
env = gym.make('FrozenLake-v0')

#Initialize table with all zeros to be uniform
Q = np.zeros([env.observation_space.n, env.action_space.n])

# Learning parameters
alpha = 0.1
gamma = 0.95
num_episodes = 2000

# array of reward for each episode
rs = np.zeros([num_episodes])

for i in range(num_episodes):
    # Set total reward and time to zero, done to False
    r_sum_i = 0
    t = 0
    done = False
    
    #Reset environment and get first new observation
    s = env.reset()
    
    while not done:
        # Choose an action by greedily (with noise) from Q table
        a = np.argmax(Q[s,:] + np.random.randn(1, env.action_space.n)*(1./(i/10+1)))
        
        # Get new state and reward from environment
        s1, r, done, _ = env.step(a)
        
        # Update Q-Table with new knowledge
        Q[s,a] = (1 - alpha)*Q[s,a] + alpha*(r + gamma*np.max(Q[s1,:]))
        
        # Add reward to episode total
        r_sum_i += r*gamma**t
        
        # Update state and time
        s = s1
        t += 1
    rs[i] = r_sum_i

## Plot reward vs episodes
# Sliding window average
r_cumsum = np.cumsum(np.insert(rs, 0, 0)) 
r_cumsum = (r_cumsum[50:] - r_cumsum[:-50]) / 50

# Plot
plt.plot(r_cumsum)
plt.show()

# Print number of times the goal was reached
N = len(rs)//10
num_Gs = np.zeros(10)

for i in range(10):
    num_Gs[i] = np.sum(rs[i*N:(i+1)*N] > 0)
    
print("Rewards: {0}".format(num_Gs))