rein q6

reinforcement/qlearningAgents.py

@@ -192,12 +192,21 @@ class ApproximateQAgent(PacmanQAgent):
           where * is the dotProduct operator
         """
         "*** YOUR CODE HERE ***"
+        featureVector = self.featExtractor.getFeatures(state, action)
+        qValue = 0
+        for feature in featureVector:
+            qValue += self.weights[feature] * featureVector[feature]
+        return qValue
 
     def update(self, state, action, nextState, reward: float):
         """
            Should update your weights based on transition
         """
         "*** YOUR CODE HERE ***"
+        featureVector = self.featExtractor.getFeatures(state, action)
+        difference = reward + self.discount * self.computeValueFromQValues(nextState) - self.getQValue(state, action)
+        for feature in featureVector:
+            self.weights[feature] += self.alpha * difference * featureVector[feature]
 
 
     def final(self, state):