public void executeTrial(Randomizer r)
{
currentState = mdp.getInitialState();
currentReward = mdp.getRewardFor(mdp.getInitialState());
previousState = default(STATE_TYPE);
previousAction = default(ACTION_TYPE);
MDPPerception <STATE_TYPE> perception = new MDPPerception <STATE_TYPE>(
currentState, currentReward);
ACTION_TYPE action = default(ACTION_TYPE);
do
{
action = decideAction(perception);
if (action != null)
{
perception = execute(action, r);
}
} while (action != null);
}
public MDPAgent(MDP <STATE_TYPE, ACTION_TYPE> mdp)
{
this.mdp = mdp;
this.currentState = mdp.getInitialState();
this.currentReward = mdp.getRewardFor(currentState);
}