source: branches/HeuristicLab.Problems.GrammaticalOptimization-gkr/HeuristicLab.Algorithms.Bandits/ActionInfos/DefaultPolicyActionInfo.cs @ 13777

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

namespace HeuristicLab.Algorithms.Bandits.BanditPolicies {
  // stores information that is relevant for most of the policies
  public class DefaultPolicyActionInfo : IBanditPolicyActionInfo {
    public double SumReward { get; private set; }
    public int Tries { get; private set; }
    public double MaxReward { get; private set; }
    private double avgValue = 0.0;
    public double Value {
      get {
        return Tries > 0 ? avgValue : double.PositiveInfinity;
      }
    }
    public DefaultPolicyActionInfo() {
      MaxReward = 0.0;
    }


    public void UpdateReward(double reward)
    {
      MaxReward = Math.Max(MaxReward, reward);
      Tries++;
      SumReward += reward;
      var delta = reward - avgValue;
      double alpha = 1.0 / Tries;
      avgValue = avgValue + alpha * delta;
    }

    public void Reset() {
      SumReward = 0.0;
      Tries = 0;
      MaxReward = double.NegativeInfinity;
      avgValue = 0.0;
    }

    public override string ToString() {
      return string.Format("{0:F3} {1:F3} {2}", Value, MaxReward, Tries);
    }
  }
}