Exemplo n.º 1
0
        // Fonction permettant de faire muter un solver
        static void Mutate(ParametrizeSolver solver, Random r)
        {
            int luck;
            // Probabilité de mutation et degré de mutation
            int probMut = 2;
            int mutRate = 1;

            // Pour chaque paramètre on regarde
            for (int i = 0; i < 3; i++)
            {
                luck = r.Next(10);
                if (luck > probMut)
                {
                    if (luck > 5)
                    {
                        solver.Params[i] += mutRate;
                    }
                    else if (solver.Params[i] > 0)
                    {
                        solver.Params[i] -= mutRate;
                    }
                }
            }
        }
Exemplo n.º 2
0
        static void Main(string[] args)
        {
            // Initialisations
            Setup();
            int nbGen   = 10;  // Nombre de génération
            int nbIndiv = 100; // Nombre d'individus de la population
            int nbCible = 20;  // Nombre d'exemple a résoudre

            Random r = new Random();

            List <ParametrizeSolver> _population = new List <ParametrizeSolver>();
            List <EvaluableBoard>    _targets    = new List <EvaluableBoard>();

            //Initialisation Création de la population
            Console.Write("Initializing population : ");
            for (int _ = 0; _ < nbIndiv; _++)
            {
                int max = 10;
                //ParametrizeSolver current = new ParametrizeSolver(r.Next(0, max), r.Next(0, max), r.Next(0, max), r.Next(0, max), r.Next(0, max));
                ParametrizeSolver current = new ParametrizeSolver(7, 4, 8);//,9,4);
                Mutate(current, r);
                _population.Add(current);
            }
            Console.Write(_population.Count + " solver created.");
            // Initialisation : Creation des cibles
            Console.Write(" - Initializing targets : ");
            for (int _ = 0; _ < nbCible; _++)
            {
                Board b = new Board(5);
                Fill(b);
                EvaluableBoard target = new EvaluableBoard(b);
                _targets.Add(target);
            }
            Console.Write(_targets.Count + " target created. \n \n");

            // Life Loop
            Console.WriteLine("Starting Life Loop :\n");
            for (int i = 0; i < nbGen; i++)
            {
                Console.ForegroundColor = ConsoleColor.DarkBlue;
                Console.Write("Starting generation {0}", i);
                // Mélanger les cibles
                Console.ForegroundColor = ConsoleColor.DarkCyan;
                // Updating targets
                _targets = new List <EvaluableBoard>();
                for (int _ = 0; _ < nbCible; _++)
                {
                    Board b = new Board(5);
                    Fill(b);
                    b = Shuffle(b, r, 100 * (i + 1)); // Difficulté du mélange
                    EvaluableBoard target = new EvaluableBoard(b);
                    _targets.Add(target);
                }
                Console.WriteLine(" - All targets shuffled - Starting :");
                // Calculating performances
                Console.ForegroundColor = ConsoleColor.Black;
                int top = Console.CursorTop;
                foreach (ParametrizeSolver solver in _population)
                {
                    List <TimeSpan> _performances = new List <TimeSpan>();

                    int    progress    = (_population.IndexOf(solver) + 1);
                    string progressStr = new String('=', progress);

                    foreach (EvaluableBoard board in _targets)
                    {
                        Stopwatch chrono = solver.Solve(board);
                        TimeSpan  perf   = chrono.Elapsed;
                        _performances.Add(perf);
                    }
                    // Informations d'avancement
                    Console.SetCursorPosition(0, top);
                    Console.Write("\rProgress [" + progressStr + new String(' ', (nbIndiv - progress)) + "] - ");
                    Console.ForegroundColor = ConsoleColor.Red;
                    Console.Write(progress + "% \n");
                    Console.ForegroundColor = ConsoleColor.Black;
                    // Calcul du temps moyen
                    var averageTimespan = new TimeSpan(Convert.ToInt64(_performances.Average(ts => ts.Ticks)));
                    solver.performance = averageTimespan;
                }
                _population = _population.OrderBy(s => s.performance).ToList();
                ParametrizeSolver _currentBest = _population[0];
                Console.ForegroundColor = ConsoleColor.Green;
                //Console.Write("Result of generation {0} : Best solver mean time is {1}, with p1 = {2} - p2 = {3} - p3 = {4} - p4 = {5} - p5 = {6}\n",i, _currentBest.performance, _currentBest.Params[0], _currentBest.Params[1], _currentBest.Params[2], _currentBest.Params[3], _currentBest.Params[4]);
                Console.Write("Result of generation {0} : Best solver mean time is {1}, with p1 = {2} - p2 = {3} - p3 = {4}\n", i, _currentBest.performance, _currentBest.Params[0], _currentBest.Params[1], _currentBest.Params[2]);

                ParametrizeSolver _currentWorst = _population[nbIndiv - 1];
                Console.ForegroundColor = ConsoleColor.DarkRed;
                //Console.Write("Result of generation {0} : Worst solver mean time is {1}, with p1 = {2} - p2 = {3} - p3 = {4} - p4 = {5} - p5 = {6}\n\n", i, _currentWorst.performance, _currentWorst.Params[0], _currentWorst.Params[1], _currentWorst.Params[2], _currentBest.Params[3], _currentBest.Params[4]);
                Console.Write("Result of generation {0} : Worst solver mean time is {1}, with p1 = {2} - p2 = {3} - p3 = {4}\n\n", i, _currentWorst.performance, _currentWorst.Params[0], _currentWorst.Params[1], _currentWorst.Params[2]);

                Console.ForegroundColor = ConsoleColor.Black;
                // Dealing with generation decendance...

                _population = _population.GetRange(0, nbIndiv / 2);
                // Evolution pour la génération suivante
                while (_population.Count < nbIndiv - nbIndiv / 4)
                {
                    // Choix de parents
                    ParametrizeSolver father = _population[r.Next(0, _population.Count)]; // /2 Pour favoriser les meilleurs pas forcément
                    ParametrizeSolver mother = _population[r.Next(0, _population.Count)]; // La meilleure solution
                    int param1, param2, param3;                                           //, param4, param5;
                    int gene1, gene2, gene3;                                              //, gene4, gene5;
                    gene1 = r.Next(0, 2);
                    gene2 = r.Next(0, 2);
                    gene3 = r.Next(0, 2);
                    //gene4 = r.Next(0, 2);
                    //gene5 = r.Next(0, 2);
                    param1 = gene1 == 0 ? father.Params[0] : mother.Params[0];
                    param2 = gene2 == 0 ? father.Params[1] : mother.Params[1];
                    param3 = gene3 == 0 ? father.Params[2] : mother.Params[2];
                    //param4 = gene4 == 0 ? father.Params[3] : mother.Params[3];
                    //param5 = gene5 == 0 ? father.Params[4] : mother.Params[4];
                    // Création et mutation d'un enfant
                    ParametrizeSolver child = new ParametrizeSolver(param1, param2, param3);//, param4, param5);
                    Mutate(child, r);
                    _population.Add(child);
                }
                // Ajout d'objets aléatoire pour agrandir la diversité
                while (_population.Count < nbIndiv)
                {
                    int max = 10;
                    ParametrizeSolver current = new ParametrizeSolver(r.Next(0, max), r.Next(0, max), r.Next(0, max));//, r.Next(0, max), r.Next(0, max));
                    Mutate(current, r);
                    _population.Add(current);
                }
            }

            Console.Read();
        }