Changeset 11727 for branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Policies/BanditPolicy.cs

Timestamp:

12/29/14 11:02:36 (8 years ago)

Author:

gkronber

Message:

#2283: worked on grammatical optimization problem solvers (simple MCTS done)

File:

• branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Policies/BanditPolicy.cs (modified) (2 diffs)

branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Policies/BanditPolicy.cs

@@ -1 +1 @@
 using System;
 using System.Collections.Generic;
+using System.Diagnostics;
 using System.Linq;
 using System.Text;
@@ -7 +8 @@
 namespace HeuristicLab.Algorithms.Bandits {
   public abstract class BanditPolicy : IPolicy {
-    public int NumActions { get; private set; }
-    public BanditPolicy(int numActions) {
-      this.NumActions = numActions;
+    public IEnumerable<int> Actions { get; private set; }
+    private readonly int numInitialActions;
+
+    protected BanditPolicy(int numActions) {
+      this.numInitialActions = numActions;
+      Actions = Enumerable.Range(0, numActions).ToArray();
     }
 
     public abstract int SelectAction();
     public abstract void UpdateReward(int action, double reward);
-    public abstract void Reset();
+
+    public virtual void DisableAction(int action) {
+      Debug.Assert(Actions.Contains(action));
+
+      Actions = Actions.Where(a => a != action).ToArray();
+    }
+
+    public virtual void Reset() {
+      Actions = Enumerable.Range(0, numInitialActions).ToArray();
+    }
   }
 }