rename MC class and improve specific analysis of antGame examples

src/main/java/core/algo/mc/MonteCarloControlEGreedy.java

@@ -12,19 +12,19 @@ import java.io.ObjectOutputStream;
 import java.util.*;
 
 /**
- * Includes both variants of Monte-Carlo methods
+ * Includes both! variants of Monte-Carlo methods
  * Default method is First-Visit.
  * Change to Every-Visit by setting flag "useEveryVisit" in the constructor to true.
  * @param <A>
  */
-public class MonteCarloControlFirstVisitEGreedy<A extends Enum> extends EpisodicLearning<A> {
+public class MonteCarloControlEGreedy<A extends Enum> extends EpisodicLearning<A> {
 
     private Map<Pair<State, A>, Double> returnSum;
     private Map<Pair<State, A>, Integer> returnCount;
     private boolean isEveryVisit;
 
 
-    public MonteCarloControlFirstVisitEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, float discountFactor, float epsilon, int delay, boolean useEveryVisit) {
+    public MonteCarloControlEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, float discountFactor, float epsilon, int delay, boolean useEveryVisit) {
         super(environment, actionSpace, discountFactor, delay);
         isEveryVisit = useEveryVisit;
         this.policy = new EpsilonGreedyPolicy<>(epsilon);
@@ -33,11 +33,11 @@ public class MonteCarloControlFirstVisitEGreedy<A extends Enum> extends Episodic
         returnCount = new HashMap<>();
     }
 
-    public MonteCarloControlFirstVisitEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, float discountFactor, float epsilon, int delay) {
+    public MonteCarloControlEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, float discountFactor, float epsilon, int delay) {
         this(environment, actionSpace, discountFactor, epsilon, delay, false);
     }
 
-    public MonteCarloControlFirstVisitEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, int delay) {
+    public MonteCarloControlEGreedy(Environment<A> environment, DiscreteActionSpace<A> actionSpace, int delay) {
         this(environment, actionSpace, LearningConfig.DEFAULT_DISCOUNT_FACTOR, LearningConfig.DEFAULT_EPSILON, delay);
     }
 

src/main/java/core/algo/td/QLearningOffPolicyTDControl.java

@@ -45,9 +45,7 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
 
         sumOfRewards = 0;
         int timestampTilFood = 0;
-        int rewardsPer1000 = 0;
         int foodCollected = 0;
-        int iterations = 0;
         int foodTimestampsTotal= 0;
         while(envResult == null || !envResult.isDone()) {
             actionValues = stateActionTable.getActionValues(state);
@@ -58,20 +56,8 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
             double reward = envResult.getReward();
             State nextState = envResult.getState();
             sumOfRewards += reward;
-            rewardsPer1000+=reward;
             timestampTilFood++;
 
-          /*  if(iterations == 100){
-                File file = new File(ContinuousAnt.FILE_NAME);
-                try {
-                    Files.writeString(Path.of(file.getPath()),  "\n", StandardOpenOption.APPEND);
-                } catch (IOException e) {
-                    e.printStackTrace();
-                }
-                return;
-            }*/
-
-
             if(reward == Reward.FOOD_DROP_DOWN_SUCCESS) {
                 foodCollected++;
                 foodTimestampsTotal += timestampTilFood;
@@ -95,7 +81,7 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
                     ((EpsilonGreedyPolicy<A>) this.policy).setEpsilon(0.05f);
                 }
                 if(foodCollected == 4000){
-                    System.out.println("final 0 expl");
+                    System.out.println("Reached 0 exploration");
                     ((EpsilonGreedyPolicy<A>) this.policy).setEpsilon(0.00f);
                 }
                 if(foodCollected == 15000){
@@ -106,7 +92,6 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
                     }
                     return;
                 }
-                iterations++;
                 timestampTilFood = 0;
             }
 

src/main/java/core/controller/RLController.java

@@ -7,7 +7,7 @@ import core.ListDiscreteActionSpace;
 import core.algo.EpisodicLearning;
 import core.algo.Learning;
 import core.algo.Method;
-import core.algo.mc.MonteCarloControlFirstVisitEGreedy;
+import core.algo.mc.MonteCarloControlEGreedy;
 import core.algo.td.QLearningOffPolicyTDControl;
 import core.algo.td.SARSA;
 import core.listener.LearningListener;
@@ -49,10 +49,10 @@ public class RLController<A extends Enum> implements LearningListener {
     public void start() {
         switch(method) {
             case MC_CONTROL_FIRST_VISIT:
-                learning = new MonteCarloControlFirstVisitEGreedy<>(environment, discreteActionSpace, discountFactor, epsilon, delay);
+                learning = new MonteCarloControlEGreedy<>(environment, discreteActionSpace, discountFactor, epsilon, delay);
                 break;
             case MC_CONTROL_EVERY_VISIT:
-                learning = new MonteCarloControlFirstVisitEGreedy<>(environment, discreteActionSpace, discountFactor, epsilon, delay, true);
+                learning = new MonteCarloControlEGreedy<>(environment, discreteActionSpace, discountFactor, epsilon, delay, true);
                 break;
 
             case SARSA_ON_POLICY_CONTROL: