public void Set(QLearning q)
{
var points = new Point[Size * Size];
for (int row = 0; row < Size; row++)
{
for (int col = 0; col < Size; col++)
{
points[row * Size + col] = new Point(row, col);
}
}
State ToState(Point p) => new State(p.Row, p.Col);
void Fill(Dictionary <State, Dictionary <State, float> > dictionary, float value)
{
foreach (var statePoint in points)
{
var state = ToState(statePoint);
var actions = dictionary[state] = new Dictionary <State, float>();
foreach (var actionPoint in points)
{
actions.Add(ToState(actionPoint), value);
}
}
}
Fill(q.Q, 0);
Fill(q.R, -1);
foreach (var from in points)
{
foreach (var to in points)
{
if (CanMove(from, to))
{
float value = 0;
if (Blocks.Contains(to))
{
value = -10;
}
if (Start.Equals(to))
{
value = 0;
}
if (Cheese.Equals(to))
{
value = 10;
}
q.R[ToState(from)][ToState(to)] = value;
}
}
}
}
static void Main(string[] args)
{
const float Gamma = 0.8f;
const int Size = 4;
const int Episodes = 1000;
const int Blocks = 5;
Random r = new Random(42);
FindTheCheese findTheCheese = FindTheCheese.Generate(r, Size, Blocks);
QLearning qLearning = new QLearning(Gamma);
findTheCheese.Set(qLearning);
ConsoleExtensions.WriteMatrix(qLearning.R, "R");
qLearning.Learn(findTheCheese.Start, findTheCheese.Cheese, Episodes, Size, r);
ConsoleExtensions.WriteMatrix(qLearning.Q, "Final Q");
qLearning.WritePath(findTheCheese.Start, findTheCheese.Cheese);
}