Answer project 3 question 3.

p3_rl/analysis.py

@@ -20,44 +20,45 @@
 # value iteration.
 
 def question2():
+    """
+    Changing the discount value doesn't change the behavior of the agent because
+    falling down cause too high of a penalty. Changing the noise to only one percent
+    failure (noise=0.01) makese it safe enough to cross the bridge even with 0.8
+    discount factor.
+    """
     answerDiscount = 0.9
-    answerNoise = 0.2
+    answerNoise = 0.01
     return answerDiscount, answerNoise
 
 def question3a():
-    answerDiscount = None
-    answerNoise = None
-    answerLivingReward = None
+    answerDiscount = 0.2
+    answerNoise = 0
+    answerLivingReward = 0
     return answerDiscount, answerNoise, answerLivingReward
-    # If not possible, return 'NOT POSSIBLE'
 
 def question3b():
-    answerDiscount = None
-    answerNoise = None
-    answerLivingReward = None
+    answerDiscount = 0.31622776601683794
+    answerNoise = 0.2
+    answerLivingReward = 0
     return answerDiscount, answerNoise, answerLivingReward
-    # If not possible, return 'NOT POSSIBLE'
 
 def question3c():
-    answerDiscount = None
-    answerNoise = None
-    answerLivingReward = None
+    answerDiscount = 0.9
+    answerNoise = 0
+    answerLivingReward = 0
     return answerDiscount, answerNoise, answerLivingReward
-    # If not possible, return 'NOT POSSIBLE'
 
 def question3d():
-    answerDiscount = None
-    answerNoise = None
-    answerLivingReward = None
+    answerDiscount = 0.9
+    answerNoise = 0.2
+    answerLivingReward = 0
     return answerDiscount, answerNoise, answerLivingReward
-    # If not possible, return 'NOT POSSIBLE'
 
 def question3e():
-    answerDiscount = None
-    answerNoise = None
-    answerLivingReward = None
+    answerDiscount = 0
+    answerNoise = 0
+    answerLivingReward = 1
     return answerDiscount, answerNoise, answerLivingReward
-    # If not possible, return 'NOT POSSIBLE'
 
 def question6():
     answerEpsilon = None