add constant for default reward

.idea/misc.xml

@@ -11,7 +11,7 @@
       </list>
     </option>
   </component>
-  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" default="false" project-jdk-name="11.0.5" project-jdk-type="JavaSDK">
+  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" default="false" project-jdk-name="11.0.3" project-jdk-type="JavaSDK">
     <output url="file://$PROJECT_DIR$/out" />
   </component>
 </project>

src/main/java/core/algo/Learning.java

@@ -16,8 +16,6 @@ import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
 import java.util.concurrent.CopyOnWriteArrayList;
-import java.util.concurrent.ExecutorService;
-import java.util.concurrent.Executors;
 
 /**
  *
@@ -99,7 +97,7 @@ public abstract class Learning<A extends Enum>{
 
     public void save(ObjectOutputStream oos) throws IOException {
         oos.writeObject(rewardHistory);
-        oos.writeObject(stateActionTable);
+        // oos.writeObject(stateActionTable);
     }
 
     public void load(ObjectInputStream ois) throws IOException, ClassNotFoundException {

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

@@ -7,7 +7,6 @@ import core.policy.GreedyPolicy;
 import core.policy.Policy;
 import evironment.antGame.Reward;
 import example.ContinuousAnt;
-import example.DinoSampling;
 
 import java.io.File;
 import java.io.IOException;
@@ -77,7 +76,7 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
                 foodCollected++;
                 foodTimestampsTotal += timestampTilFood;
                 if(foodCollected % 1000 == 0){
-                    System.out.println(foodTimestampsTotal/1000f);
+                    System.out.println(foodTimestampsTotal / 1000f + " " + timestampCurrentEpisode);
                     File file = new File(ContinuousAnt.FILE_NAME);
                     try {
                         Files.writeString(Path.of(file.getPath()),  foodTimestampsTotal/1000f +",", StandardOpenOption.APPEND);

src/main/java/evironment/antGame/AntWorld.java

@@ -54,7 +54,7 @@ public class AntWorld implements Environment<AntAction>, Visualizable {
 
     protected StepCalculation processStep(AntAction action) {
         StepCalculation sc = new StepCalculation();
-        sc.reward = -1;
+        sc.reward = Reward.DEFAULT_REWARD;
         sc.info = "";
         sc.done = false;
         Cell currentCell = grid.getCell(myAnt.getPos());

src/main/java/evironment/antGame/Reward.java

@@ -1,6 +1,7 @@
 package evironment.antGame;
 
 public class Reward {
+    public static final double DEFAULT_REWARD = -1;
     public static final double FOOD_PICK_UP_SUCCESS = 0;
     public static final double FOOD_PICK_UP_FAIL_NO_FOOD = -1;
     public static final double FOOD_PICK_UP_FAIL_HAS_FOOD_ALREADY = -1;

src/main/java/evironment/jumpingDino/DinoWorld.java

@@ -44,7 +44,7 @@ public class DinoWorld implements Environment<DinoAction>, Visualizable {
     @Override
     public StepResultEnvironment step(DinoAction action) {
         boolean done = false;
-        int reward = 0;
+        int reward = 1;
 
         if(action == DinoAction.JUMP){
             dino.jump();
@@ -68,7 +68,7 @@ public class DinoWorld implements Environment<DinoAction>, Visualizable {
             spawnNewObstacle();
         }
         if(ranIntoObstacle()) {
-            reward = -1;
+            reward = 0;
             done = true;
         }
 

src/main/java/example/ContinuousAnt.java

@@ -3,10 +3,8 @@ package example;
 import core.RNG;
 import core.algo.Method;
 import core.controller.RLController;
-import core.controller.RLControllerGUI;
 import evironment.antGame.AntAction;
 import evironment.antGame.AntWorldContinuous;
-import evironment.antGame.AntWorldContinuousOriginalState;
 
 import java.io.File;
 import java.io.IOException;
@@ -31,7 +29,7 @@ public class ContinuousAnt {
         rl.setNrOfEpisodes(1);
         //0.99 0.9 0.5
         //0.99 0.95 0.9 0.7 0.5 0.3 0.1
-        rl.setDiscountFactor(0.05f);
+        rl.setDiscountFactor(0.1f);
         // 0.1, 0.3, 0.5, 0.7 0.9
         rl.setLearningRate(0.9f);
         rl.setEpsilon(0.2f);

src/main/java/example/DinoSampling.java

@@ -6,7 +6,6 @@ import core.controller.RLController;
 import core.controller.RLControllerGUI;
 import evironment.jumpingDino.DinoAction;
 import evironment.jumpingDino.DinoWorld;
-import evironment.jumpingDino.DinoWorldAdvanced;
 
 import java.io.File;
 import java.io.IOException;
@@ -34,13 +33,13 @@ public class DinoSampling {
                 System.out.println("seed: " + i * 13);
                 RNG.setSeed(i * 13);
 
-                RLController<DinoAction> rl = new RLController<>(
-                        new DinoWorldAdvanced(),
-                        Method.Q_LEARNING_OFF_POLICY_CONTROL,
+                RLController<DinoAction> rl = new RLControllerGUI<>(
+                        new DinoWorld(),
+                        Method.MC_CONTROL_FIRST_VISIT,
                         DinoAction.values());
-                rl.setDelay(0);
-                rl.setDiscountFactor(0.99f);
-                rl.setEpsilon(f);
+                rl.setDelay(300);
+                rl.setDiscountFactor(1f);
+                rl.setEpsilon(0.5f);
                 rl.setLearningRate(0.9f);
                 rl.setNrOfEpisodes(400000);
                 rl.start();