source: branches/HeuristicLab.Problems.GrammaticalOptimization-gkr/HeuristicLab.Problems.Bandits/Bandit.cs @ 13338

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;

namespace HeuristicLab.Algorithms.Bandits {
  public class Bandit : IBandit {
    public int NumArms { get { return distributions.Length; } }
    public int OptimalExpectedRewardArm { get; private set; }
    public int OptimalMaximalRewardArm { get; private set; }

    private readonly IModel[] distributions;
    private readonly Random random;
    public Bandit(Random random, IEnumerable<IModel> distributions, int bestExpRewardArm, int bestMaxRewardArm) {
      this.random = random;
      this.distributions = distributions.ToArray();

      OptimalExpectedRewardArm = bestExpRewardArm;
      OptimalMaximalRewardArm = bestMaxRewardArm;
    }

    // pulling an arm results in a bernoulli distributed reward 
    // with mean expReward[i]
    public double Pull(int arm) {
      return distributions[arm].Sample(random);
    }
  }
}