source: branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Policies/GaussianThompsonSamplingPolicy.cs @ 11728

using System;
using System.Diagnostics;
using System.Linq;
using HeuristicLab.Common;

namespace HeuristicLab.Algorithms.Bandits {
  public class GaussianThompsonSamplingPolicy : BanditPolicy {
    private readonly Random random;
    private readonly double[] sumRewards;
    private readonly double[] sumSqrRewards;
    private readonly int[] tries;
    public GaussianThompsonSamplingPolicy(Random random, int numActions)
      : base(numActions) {
      this.random = random;
      this.sumRewards = new double[numActions];
      this.sumSqrRewards = new double[numActions];
      this.tries = new int[numActions];
    }


    public override int SelectAction() {
      Debug.Assert(Actions.Any());
      var maxTheta = double.NegativeInfinity;
      int bestAction = -1;
      foreach (var a in Actions) {
        if (tries[a] == 0) return a;
        var mu = sumRewards[a] / tries[a];
        var stdDev = Math.Sqrt(sumSqrRewards[a] / tries[a] - Math.Pow(mu, 2));
        var theta = Rand.RandNormal(random) * stdDev + mu;
        if (theta > maxTheta) {
          maxTheta = theta;
          bestAction = a;
        }
      }
      return bestAction;
    }

    public override void UpdateReward(int action, double reward) {
      Debug.Assert(Actions.Contains(action));

      sumRewards[action] += reward;
      sumSqrRewards[action] += reward * reward;
      tries[action]++;
    }

    public override void DisableAction(int action) {
      base.DisableAction(action);
      sumRewards[action] = 0;
      sumSqrRewards[action] = 0;
      tries[action] = -1;
    }

    public override void Reset() {
      base.Reset();
      Array.Clear(sumRewards, 0, sumRewards.Length);
      Array.Clear(sumSqrRewards, 0, sumSqrRewards.Length);
      Array.Clear(tries, 0, tries.Length);
    }
  }
}