add javadoc

src/main/java/core/Action.java

@@ -1,4 +0,0 @@
-package core;
-
-public interface Action {
-}

src/main/java/core/DeterministicStateActionTable.java

@@ -17,13 +17,17 @@ public class DeterministicStateActionTable<A extends Enum> implements StateActio
         this.discreteActionSpace = discreteActionSpace;
     }
 
-    /*
-       If the state is not present in the table at the time of
-       calling this method the DEFAULT_VALUE gets returned BUT
-       no the missing state is not inserted into the table!
-
-       Inserting of missing states is ONLY done in "setValue()"
-       method.
+    /**
+     * If the state is not present in the table at the time of
+     * calling this method the DEFAULT_VALUE gets returned BUT
+     * no the missing state is not inserted into the table!
+     *
+     * Inserting of missing states is ONLY done in "setValue()"
+     * method.
+     *
+     * @param state given state
+     * @param action given action
+     * @return estimate value of state-action pair
      */
     @Override
     public double getValue(State state, A action) {
@@ -34,11 +38,15 @@ public class DeterministicStateActionTable<A extends Enum> implements StateActio
         return DEFAULT_VALUE;
     }
 
-    /*
-       Update the value of an action for a specific state.
-       If the state is not present in the table yet,
-       it will get stored in combination with every action
-       from the action space initialized with the default value.
+    /**
+     * Update the value of an action for a specific state.
+     * If the state is not present in the table yet,
+     * it will get stored in combination with every action
+     * from the action space initialized with the default value.
+     *
+     * @param state given state
+     * @param action given action
+     * @param value new estimate of the state-action pair
      */
     @Override
     public void setValue(State state, A action, double value) {
@@ -52,6 +60,10 @@ public class DeterministicStateActionTable<A extends Enum> implements StateActio
         actionValues.put(action, value);
     }
 
+    /**
+     * @param state given state
+     * @return all available action in given state and their corresponding estimated values
+     */
     @Override
     public Map<A, Double> getActionValues(State state) {
         if(table.get(state) == null){
@@ -60,6 +72,9 @@ public class DeterministicStateActionTable<A extends Enum> implements StateActio
         return table.get(state);
     }
 
+    /**
+     * @return Map with initial values for every available action
+     */
     private Map<A, Double> createDefaultActionValues(){
         final Map<A, Double> defaultActionValues = new LinkedHashMap<>();
         for(A action: discreteActionSpace){

src/main/java/core/DiscreteActionSpace.java

@@ -1,5 +1,14 @@
 package core;
 
+/**
+ * Collection of all available actions for a specific
+ * environment. It defines the actions the agent is able
+ * to choose from at every timestamp T.
+ * Extending from <interface>Iterable</interface> for easy
+ * "unmutable" iteration of the action space.
+ *
+ * @param <A> Actions as defined in an <Enum> based class
+ */
 public interface DiscreteActionSpace<A extends Enum> extends Iterable<A> {
     int getNumberOfActions();
     void addAction(A a);

src/main/java/core/Environment.java

@@ -1,5 +1,17 @@
 package core;
 
+/**
+ * Interface of the environment as defined in the RL framework.
+ * An agent is able to interact with its environment, submitting
+ * an action (.step()) every timestamp T and receiving a reward and a new
+ * observation.
+ * <class>StepResultEnvironment</class> is the datatype to combine all the
+ * receiving information.
+ * After each episode the environment is reset to its original state,
+ * returning the starting state.
+ *
+ * @param <A> related <Enum> which defines the available action for this environment
+ */
 public interface Environment<A extends Enum> {
     StepResultEnvironment step(A action);
     State reset();

src/main/java/core/ListDiscreteActionSpace.java

@@ -3,6 +3,13 @@ package core;
 import java.io.Serializable;
 import java.util.*;
 
+/**
+ * Implementation of a discrete action space.
+ * "Discrete" because actions are exclusively defined as Enum,
+ * realized as generic "A".
+ *
+ * @param <A> Enum class of actions in the specific environment
+ */
 public class ListDiscreteActionSpace<A extends Enum> implements DiscreteActionSpace<A>, Serializable{
     private static final long serialVersionUID = 1L;
     private List<A> actions;

src/main/java/core/Observation.java

@@ -1,4 +0,0 @@
-package core;
-
-public interface Observation {
-}

src/main/java/core/RNG.java

@@ -3,6 +3,15 @@ package core;
 import java.security.SecureRandom;
 import java.util.Random;
 
+/**
+ * To ensure deterministic behaviour of repeating program executions,
+ * this class is used for all random number generation methods.
+ * Do not use Math.random()!
+ * It is not necessary to set a seed explicit, because a default one
+ * "123" is defined. Nonetheless a set-method is exposed which should
+ * ONLY be called in the very beginning of the program. (Do not reseed while
+ * execution)
+ */
 public class RNG {
     private static SecureRandom rng;
     private static int seed = 123;

src/main/java/core/SaveState.java

@@ -5,6 +5,15 @@ import lombok.Getter;
 
 import java.io.Serializable;
 
+/**
+ * Small datatype to combine needed information to save and recover
+ * the learning progress. Essentially, only the Q-Table needs to be saved
+ * for all tabular methods because they all try to estimate
+ * the action values until convergence.
+ * For episodic method the number of episodes so far is also saved.
+ *
+ * @param <A> enum class of action for a specific environment
+ */
 @AllArgsConstructor
 @Getter
 public class SaveState<A extends Enum> implements Serializable {

src/main/java/core/State.java

@@ -1,5 +1,8 @@
 package core;
 
+/**
+ *
+ */
 public interface State {
     String toString();
     int hashCode();

src/main/java/core/StateActionTable.java

@@ -2,6 +2,12 @@ package core;
 
 import java.util.Map;
 
+/**
+ * Q-Table which saves all seen states, all available actions for each state
+ * and their value (state-action values/ action values).
+ *
+ * @param <A>
+ */
 public interface StateActionTable<A extends Enum> {
     double DEFAULT_VALUE = 0.0;
 

src/main/java/core/StepResult.java

@@ -3,6 +3,14 @@ package core;
 import lombok.AllArgsConstructor;
 import lombok.Getter;
 
+/**
+ * Almost the same as the datatype <class>StepResultEnvironment</class>
+ * but includes the last action taken as well.
+ * The environment does not return the last action taken in its result
+ * <class>StepResultEnvironment</class> but is needed for the prediction problem.
+ *
+ * @param <A> Enum class of last action taken
+ */
 @AllArgsConstructor
 @Getter
 public class StepResult<A extends Enum> {

src/main/java/core/StepResultEnvironment.java

@@ -4,6 +4,13 @@ import lombok.AllArgsConstructor;
 import lombok.Getter;
 import lombok.Setter;
 
+/**
+ * After each timestamp the environment returns a reward
+ * for the previous action (still the same timestamp t), the resulting
+ * observation/state (environment is in charge to process the observation
+ * and build a markov state) and the information whether or not the episode
+ * has ended.
+ */
 @Getter
 @Setter
 @AllArgsConstructor

src/main/java/core/Util.java

@@ -1,6 +1,15 @@
 package core;
 
+
 public class Util {
+    /**
+     * Checks whether or not the given string input is a number or not.
+     * Used for example to check the input field to trigger additional
+     * episodes.
+     *
+     * @param strNum string input from user
+     * @return true if numeric and parsable, false if not
+     */
     public static boolean isNumeric(String strNum) {
         if (strNum == null) {
             return false;

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

@@ -14,6 +14,10 @@ import java.util.List;
 import java.util.Set;
 import java.util.concurrent.CopyOnWriteArrayList;
 
+/**
+ *
+ * @param <A> discrete action type for a specific environment
+ */
 @Getter
 public abstract class Learning<A extends Enum>{
     protected Policy<A> policy;

src/main/java/core/algo/Method.java

@@ -1,5 +1,9 @@
 package core.algo;
 
+/**
+ * Instead of reflections this enum is used to determine
+ * which RL-algorithm should be used.
+ */
 public enum Method {
     MC_ONPOLICY_EGREEDY, TD_ONPOLICY
 }

src/main/java/core/gui/LearningView.java

@@ -2,6 +2,9 @@ package core.gui;
 
 import java.util.List;
 
+/**
+ * Switched out Views have to implement this interface.
+ */
 public interface LearningView {
     void repaintEnvironment();
     void updateLearningInfoPanel();

src/main/java/core/gui/View.java

@@ -61,7 +61,7 @@ public class View<A extends Enum> implements LearningView{
         fileMenu.add(new JMenuItem(new AbstractAction("Save") {
             @Override
             public void actionPerformed(ActionEvent e) {
-                String fileName = JOptionPane.showInputDialog("Enter file name", "save");
+                String fileName = JOptionPane.showInputDialog("Enter file name", "path/to/file");
                 if(fileName != null){
                     viewListener.onSaveState(fileName);
                 }

src/main/java/core/gui/Visualizable.java

@@ -2,6 +2,13 @@ package core.gui;
 
 import javax.swing.*;
 
+/**
+ * Classes that implement this interface are able to create a visual component
+ * that can be utilized and displayed by the view. It is optional for an environment
+ * to defined a GUI (View is checking it via "instance of").
+ * Furthermore a state implement this interface, so it can be displayed from the
+ * state-action-table.
+ */
 public interface Visualizable {
     JComponent visualize();
 }

src/main/java/core/listener/LearningListener.java

@@ -2,6 +2,9 @@ package core.listener;
 
 import java.util.List;
 
+/**
+ * Methods that gets triggered to inform about the current learning process.
+ */
 public interface LearningListener{
     void onLearningStart();
     void onLearningEnd();

src/main/java/core/listener/ViewListener.java

@@ -1,5 +1,10 @@
 package core.listener;
 
+/**
+ * Interface the controller is implementing and gets passed to
+ * the View. (Preventing the controller from adding all
+ * ActionListeners to view elements)
+ */
 public interface ViewListener {
     void onEpsilonChange(float epsilon);
     void onDelayChange(int delay);

src/main/java/core/policy/EpsilonPolicy.java

@@ -1,5 +1,11 @@
 package core.policy;
 
+/**
+ * Chooses the action with the highest values with possibility: 1-Ɛ + Ɛ/|A|
+ * With possibility of Ɛ, a random action is taken (highest values option included).
+ *
+ * @param <A> Enum class of available action in specific environment
+ */
 public interface EpsilonPolicy<A extends Enum> extends Policy<A> {
     float getEpsilon();
     void setEpsilon(float epsilon);

src/main/java/core/policy/GreedyPolicy.java

@@ -6,6 +6,12 @@ import java.util.ArrayList;
 import java.util.List;
 import java.util.Map;
 
+/**
+ * Always chooses the action with the highest value
+ * with ties broken arbitrarily.
+ *
+ * @param <A> Enum class of available action in specific environment
+ */
 public class GreedyPolicy<A extends Enum> implements Policy<A> {
 
     @Override

src/main/java/core/policy/Policy.java

@@ -2,6 +2,12 @@ package core.policy;
 
 import java.util.Map;
 
+/**
+ * Strategy to choose a specific action available in the agent's current
+ * state.
+ *
+ * @param <A> Enum class of available action in specific environment
+ */
 public interface Policy<A extends Enum> {
     A chooseAction(Map<A, Double> actionValues);
 }

src/main/java/core/policy/RandomPolicy.java

@@ -4,6 +4,11 @@ import core.RNG;
 
 import java.util.Map;
 
+/**
+ * Chooses an action arbitrarily.
+ *
+ * @param <A> Enum class of available action in specific environment
+ */
 public class RandomPolicy<A extends Enum> implements Policy<A>{
     @Override
     public A chooseAction(Map<A, Double> actionValues) {

src/main/java/evironment/antGame/AntObservation.java

@@ -1,6 +1,5 @@
 package evironment.antGame;
 
-import core.Observation;
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
 import lombok.Getter;
@@ -12,7 +11,7 @@ import java.awt.*;
 @AllArgsConstructor
 @Getter
 @Setter
-public class AntObservation implements Observation {
+public class AntObservation {
       private Cell cell;
       private Point pos;