private TAction SelectRandomAction()
{
IList <TAction> allActions = MDP.GetAllActions();
return(allActions[0]);
// return Util.selectRandomlyFromList(allActions);
}
public QLearningAgent(MDP <TState, TAction> mdp) : base(mdp)
{
Q = new Dictionary <Pair <TState, TAction>, double>();
qTable = new QTable <TState, TAction>(mdp.GetAllActions());
stateActionCount = new FrequencyCounter <Pair <TState, TAction> >();
actionCounter = 0;
}
private TAction ActionMaximizingLearningFunction()
{
TAction maxAct = null;
double maxValue = double.NegativeInfinity;
foreach (TAction action in MDP.GetAllActions())
{
double qValue = qTable.GetQValue(CurrentState, action);
var lfv = this.LearningFunction(qValue);
if (lfv > maxValue)
{
maxValue = lfv;
maxAct = action;
}
}
return(maxAct);
}