Reinforcement Learning Toolbox 2.0 last updated:

General Documentation Manual Tutorial Class Reference Master Thesis Examples Related Papers Downloads Links News mailto:webmaster Main Page     Class Hierarchy   Compound List   File List   Compound Members   File Members CSemiMDPLastNRewardFunction Class Reference Reward Function for Behaviours. More... #include <crewardmodel.h> Inheritance diagram for CSemiMDPLastNRewardFunction: List of all members. Public Member Functions   CSemiMDPLastNRewardFunction (CRewardFunction *rewardFunction, double gamma)   Creates the reward function with the discount factor gamma. virtual  ~CSemiMDPLastNRewardFunction () virtual double  getReward (CStateCollection *oldState, CAction *action, CStateCollection *newState)   Calculates the reward for extended actions. Protected Attributes double  gamma   The discount factor. Detailed Description Reward Function for Behaviours. Very often the reward of a behaviour consists of the summud up rewards from the primitiv actions which were executed during the behaviour was activ. The class CSemiMDPLastNRewardFunction does this reward calculation. It is a subclass of CRewardEpisode, so it mantains a reward array containing all rewards from the past. The function getReward(CStateCollection *oldState, CAction *action, CStateCollection *newState). For an action of duration d, the function calculates the reward of the transition by sum_{i=0}^{d-1} gamma^i * r(N-i). The discount factor is needed since the reward from the past has to be weakened by this factor. Since the object has to have access to the past "primitiv" rewards, it has to be added to the listener list of the agent. Constructor & Destructor Documentation CSemiMDPLastNRewardFunction::CSemiMDPLastNRewardFunction (  CRewardFunction *  rewardFunction, double  gamma )    Creates the reward function with the discount factor gamma. virtual CSemiMDPLastNRewardFunction::~CSemiMDPLastNRewardFunction (   )  [virtual] Member Function Documentation virtual double CSemiMDPLastNRewardFunction::getReward (  CStateCollection *  oldState, CAction *  action, CStateCollection *  newState )  [virtual]   Calculates the reward for extended actions. For an action of duration d, the function the reward is calculated by sum_{i=0}^{d-1} gamma^i * r(N-i). The discount factor is needed since the reward from the past has to be weakened by this factor. Implements CRewardFunction. Member Data Documentation double CSemiMDPLastNRewardFunction::gamma [protected]   The discount factor. The documentation for this class was generated from the following file: crewardmodel.h