Initial commit

Divers/renforcement2/MountainCar_predict.py

@@ -0,0 +1,19 @@
+import gym
+import numpy as np
+import MountainCar_common
+
+env=gym.make("MountainCar-v0")
+
+q_table=np.load("MountainCar_qtable.npy")
+
+for epoch in range(1000):
+    state = env.reset()
+    while True:
+        env.render()
+        discrete_state=MountainCar_common.discretise(state)
+        action=np.argmax(q_table[discrete_state])
+        state, reward, done, info=env.step(action)
+        if done:
+            print("Essai {:05d}: {}".format(epoch, "OK" if state[0]>=env.goal_position else "raté ..."))
+            break
+env.close()