maze = Maze()
π = create_policy_s(maze)
# Value Iteration
Θ = 0.00000000001
V = create_value_funtion(maze)
iterations_number = 0
while True:
iterations_number += 1
Δ = 0
for state in maze.all_states():
if not maze.is_terminal(state):
v = V[state]
actions = maze._actions[state]
p = 1 / len(actions)
for action in actions:
maze.set_state(state)
r = maze.move(action)
calculation = p * (r + γ * V[maze.current_state()])
if V[state] < calculation:
V[state] = calculation
Δ = max(Δ, abs(v - V[state]))
if Δ < Θ:
print("Δ:",Δ)
