CooperativeQLearning.hpp

Go to the documentation of this file.

#ifndef AI_TOOLBOX_FACTORED_MDP_COOPERATIVE_QLEARNING_HEADER_FILE
#define AI_TOOLBOX_FACTORED_MDP_COOPERATIVE_QLEARNING_HEADER_FILE
 
#include <AIToolbox/Factored/MDP/Types.hpp>
 
#include <AIToolbox/Factored/Utils/BayesianNetwork.hpp>
#include <AIToolbox/Factored/MDP/Policies/QGreedyPolicy.hpp>
 
namespace AIToolbox::Factored::MDP {
    class CooperativeQLearning {
        public:
            // SparseCooperativeQLearning(State S, Action A, double discount, double alpha);
            CooperativeQLearning(const DDNGraph & g, const std::vector<std::vector<size_t>> & basisDomains, double discount, double alpha);
 
            Action stepUpdateQ(const State & s, const Action & a, const State & s1, const Rewards & rew);
 
            void setLearningRate(double a);
 
            double getLearningRate() const;
 
            void setDiscount(double d);
 
            double getDiscount() const;
 
            const DDNGraph & getGraph() const;
 
            const State & getS() const;
 
            const Action & getA() const;
 
            const FactoredMatrix2D & getQFunction() const;
 
            void setQFunction(double val);
 
        private:
            const DDNGraph & graph_;
            double discount_, alpha_;
            FactoredMatrix2D q_;
            QGreedyPolicy<> policy_;
            // Helper
            Vector agentNormRews_;
    };
}
 
#endif