5. QLearning

// R[6][6] = Reward Matrix 보상 행렬
#define MROW 6
#define MLOW 4
/*
double R[MROW][MLOW] = {                // double R[MROW][MLOW] = {        //       -> Action
{ -1, -1, -1, -1, 0, -1 },                //     { -1, -1, 0, -1 },            //   0 1 2 3 4 5 6
{ -1, -1, -1, 0, -1, 100 },                //     { -1, 0, -1, 100 },            // s 1
{ -1, -1, -1, 0, -1, -1 },                //     { -1, 0, -1, -1 },            // t 2
{ -1, 0, 0, -1, 0, -1 },                //     { 0, -1, 0, -1 },            // a 3
{ 0, -1, -1, 0, -1, 100 },                //     { -1, 0, -1, 100 },            // t 4
{ -1, 0, -1, -1, 0, 100 }                //     { -1, -1, 0, 100 }            // e 5
};                                        // };                            //   6
*/
 
double Reward[MROW][MLOW] = {
    { -1, -1, 0, -1 },
    { -1, 0, -1, 100 },
    { -1, 0, -1, -1 },
    { 0, -1, 0, -1 },
    { -1, 0, -1, 100 },
    { -1, -1, 0, 100 }
};
 
// 학습한 결과를 나타낸 Q matrix (행동을 실제적으로 결정해주는 행렬)
double QLearning[MROW][MLOW] = { 0, };
 
void print_matrix(double m[MROW][MLOW], int rows=MROW, int columns=MLOW) {
    for (int i = 0; i < rows; ++i) {
        for (int j = 0; j < columns; ++j) {
            std::cout << m[i][j] << "\t";
        }
        std::cout << std::endl;
    }
}
 
double get_max_q(int state) {
    double temp_max = 0;
    for (int i = 0; i < MLOW; ++i) {
        if ((Reward[state][i] >= 0) && (QLearning[state][i] > temp_max)) {
            temp_max = QLearning[state][i];
        }
    }
    return temp_max;
}
 
void main()
{
    print_matrix(QLearning);
    double gamma = 0.8;
    int _Fcount = 0;
    bool is_goal = true;
    while (is_goal)
    {
        for (int state = 0; state < MROW; state++)
        {
            for (int action = 0; action < MLOW; action++)
            {
                if (Reward[state][action] == -1)
                {
                    continue;
                }
                int next_state = action;
                double max_q = get_max_q(next_state);
                QLearning[state][action] = Reward[state][action] + gamma * max_q;
            }
        }
        _Fcount++;
        if (_Fcount == 100)
        {
            is_goal = false;
        }
    }
    std::cout << std::endl;
    print_matrix(QLearning);
    int iiiii = 0;
}