Reinforcement Learning Toolbox

00001 // Copyright (C) 2003
00002 // Gerhard Neumann (gneumann@gmx.net)
00003 // Stephan Neumann (sneumann@gmx.net) 
00004 //                
00005 // This file is part of RL Toolbox.
00006 // http://www.igi.tugraz.at/ril_toolbox
00007 //
00008 // All rights reserved.
00009 // 
00010 // Redistribution and use in source and binary forms, with or without
00011 // modification, are permitted provided that the following conditions
00012 // are met:
00013 // 1. Redistributions of source code must retain the above copyright
00014 //    notice, this list of conditions and the following disclaimer.
00015 // 2. Redistributions in binary form must reproduce the above copyright
00016 //    notice, this list of conditions and the following disclaimer in the
00017 //    documentation and/or other materials provided with the distribution.
00018 // 3. The name of the author may not be used to endorse or promote products
00019 //    derived from this software without specific prior written permission.
00020 // 
00021 // THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
00022 // IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
00023 // OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
00024 // IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
00025 // INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
00026 // NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
00027 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
00028 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
00029 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
00030 // THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
00031 
00032 #ifndef C_VPOLICYFUNCTIONLEARNER__H
00033 #define C_VPOLICYFUNCTIONLEARNER__H
00034 
00035 
00036 #include "cagentlistener.h"
00037 #include "cqfunction.h"
00038 #include "cvfunction.h"
00039 #include "cqetraces.h"
00040 #include "cresiduals.h"
00041 #include "cpolicygradient.h"
00042 #include "cdynamicmodel.h"
00043 #include "cpegasus.h"
00044 #include "ccontinuousactiongradientpolicy.h"
00045 
00046 class CVPolicyLearner : public CSemiMDPRewardListener
00047 {
00048 protected:
00049         typedef std::list<CFeatureList *> CStateGradient;
00050 
00052         CGradientVFunction *vFunction;
00053         CVFunctionInputDerivationCalculator *vFunctionInputDerivation;
00054         
00055         CContinuousActionGradientPolicy *gradientPolicy;
00056         CCAGradientPolicyInputDerivationCalculator *policydInput;
00057 
00058 
00059 //      std::list<CState *> *states;
00060 
00061         ColumnVector *dReward;
00062         ColumnVector *dVFunction;
00063         Matrix *dPolicy;
00064         Matrix *dModelInput;
00065 
00066         CContinuousActionData *data;
00067 
00068         std::list<CStateGradient *> *stateGradients;
00069 
00070         CStateGradient *stateGradient1;
00071         CStateGradient *stateGradient2;
00072         CStateGradient *dModelGradient;
00073         
00074         CStateReward *rewardFunction;
00075         CDynamicModel *dynModel;
00076         CDynamicModelInputDerivationCalculator *dynModeldInput;
00077 
00078 
00079         CStateCollectionImpl *tempStateCol;
00080 
00081         CFeatureList *policyGradient;
00082 
00083         void getDNextState(CStateGradient *stateGradient1, CStateGradient *stateGradient2, CStateCollection *currentState, CContinuousActionData *data);
00084         void multMatrixFeatureList(Matrix *matrix, CFeatureList *features, int index, std::list<CFeatureList *> *newFeatures);
00085 
00086         //CFeatureList *valueGradient;
00087 
00088         int nForwardView;
00089 
00090 public:
00091         CVPolicyLearner(CStateReward *rewardFunction, CDynamicModel *dynModel, CDynamicModelInputDerivationCalculator *dynModeldInput,CGradientVFunction *vFunction, CVFunctionInputDerivationCalculator *vFunctionInputDerivation, CContinuousActionGradientPolicy *gradientPolicy, CCAGradientPolicyInputDerivationCalculator *policydInput, std::list<CStateModifier *> *stateModifiers, int nForwardView);
00092         virtual ~CVPolicyLearner();
00093 
00094         virtual void nextStep(CStateCollection *oldState, CAction *action, double reward, CStateCollection *nextState);
00095 
00096         virtual void newEpisode();
00097 };
00098 
00099 #endif
00100
ctdpolicylearner.h
