Changeset 11730 for branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Bandits/BernoulliBandit.cs

Timestamp:

01/02/15 16:08:21 (9 years ago)

Author:

gkronber

Message:

#2283: several major extensions for grammatical optimization

File:

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

branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Bandits/BernoulliBandit.cs

@@ -6 +6 @@
 
 namespace HeuristicLab.Algorithms.Bandits {
-  public class BernoulliBandit {
+  public class BernoulliBandit : IBandit {
     public int NumArms { get; private set; }
     public double OptimalExpectedReward { get; private set; } // reward of the best arm, for calculating regret
+    public int OptimalExpectedRewardArm { get; private set; }
+    // the arm with highest expected reward also has the highest probability of return a reward of 1.0
+    public int OptimalMaximalRewardArm { get { return OptimalExpectedRewardArm; } }
+
     private readonly Random random;
     private readonly double[] expReward;
@@ -19 +23 @@
       for (int i = 0; i < nArms; i++) {
         expReward[i] = random.NextDouble();
-        if (expReward[i] > OptimalExpectedReward) OptimalExpectedReward = expReward[i];
+        if (expReward[i] > OptimalExpectedReward) {
+          OptimalExpectedReward = expReward[i];
+          OptimalExpectedRewardArm = i;
+        }
       }
     }