private Individ RightUnitGreegy(int n = 0) //Прямой удельный жадный алгоритм, начинающий поиск с n-ой компании
{
Individ data = new Individ();
int limit = n + numberOfCompany - 1;
int counter = numberOfAttachmentSizes - 1;
int currentPos;
while ((counter > 0) && (n < limit))
{
double maxGain = 0;
int pos = 0;
currentPos = n % numberOfCompany;
for (int j = 1; j < counter + 1; j++)
{
double gain = (double)tableOfOffers[currentPos, j] / sizeOfAttachment[j];
if (gain >= maxGain)
{
maxGain = gain;
pos = j;
}
}
data.Set(currentPos, pos);
counter -= pos;
n++;
}
data.Set(limit % numberOfCompany, counter);
return(data);
}
private Individ[] ProportionalSelection(Individ[] population, Individ[] child, int populationSize, double averageFitness) //Работает не очень, быстро сходится
{
Individ[] newPopulation = new Individ[populationSize];
int counter = 0;
for (int i = 0; i < populationSize && counter < populationSize; i++)
{
if (averageFitness < population[i].GetFitness())
{
newPopulation[counter++] = new Individ(population[i]);
}
}
for (int i = 0; i < populationSize * 2 && counter < populationSize; i++)
{
if (averageFitness < child[i].GetFitness())
{
newPopulation[counter++] = new Individ(child[i]);
}
}
while (counter < populationSize)
{
newPopulation[counter++] = new Individ(child[rnd.Next(populationSize * 2)]);
}
return(newPopulation);
}
private Individ LeftUnitGreedy(int n = 0) //Обратный удельный жадный алгоритм, начинающий поиск с n-ой компании
{
Individ data = new Individ();
int counter = numberOfAttachmentSizes - 1;
n %= numberOfCompany;
int currentPos = (n - 1 + numberOfCompany) % numberOfCompany;
while ((counter > 0) && (currentPos != n))
{
double maxGain = 0;
int pos = 0;
for (int j = 1; j < counter + 1; j++)
{
double gain = (double)tableOfOffers[currentPos, j] / sizeOfAttachment[j];
if (gain >= maxGain)
{
maxGain = gain;
pos = j;
}
}
data.Set(currentPos, pos);
counter -= pos;
currentPos = (currentPos - 1 + numberOfCompany) % numberOfCompany;
}
data.Set(currentPos, counter);
return(data);
}
public Individ(Individ a)
{
genotype = new int[NoC];
for (int i = 0; i < NoC; i++)
{
genotype[i] = a.genotype[i];
}
UpdateFitness();
}
private Individ UniformCrossover(Individ[] population, int populationSize)
{
Individ a = new Individ();
int firstParent = rnd.Next(populationSize);
int secondParent = rnd.Next(populationSize);
a.UniformCrossover(population[firstParent], population[secondParent], rnd);
return(a);
}
public Individ(Individ a)
{
code = new bool[fullLength];
for (int i = 0; i < fullLength; i++)
{
code[i] = a.code[i];
}
profit = a.profit;
resources = a.resources;
fitness = a.fitness;
}
public void GreedProportional()
{
Individ[] solutionPair = new Individ[2];
for (int i = 0; i < 2; i++)
{
solutionPair[i] = new Individ();
}
solutionPair[0] = RightUnitGreegy();
solutionPair[1] = LeftUnitGreedy();
solutionPair[0].Print();
solutionPair[1].Print();
}
public Individ(Individ a, Individ b, int pos) //Crossover
{
genotype = new int[NoC];
for (int i = 0; i < pos; i++)
{
genotype[i] = a.genotype[i];
}
for (int i = pos; i < NoC; i++)
{
genotype[i] = b.genotype[i];
}
UpdateFitness();
}
public void RandSolution()
{
int numberOfSolutions = 5;
Individ[] RandSolution = new Individ[numberOfSolutions];
Console.WriteLine("Случайные решения: ");
for (int i = 0; i < numberOfSolutions; i++)
{
RandSolution[i] = new Individ(rnd, rnd.Next(1, 2));
RandSolution[i].Print();
}
Console.WriteLine("Конец случайных решений!");
}
public void UniformCrossover(Individ firstParent, Individ secondParent, Random rnd)
{
for (int i = 0; i < NoC; i++)
{
if (rnd.Next(2) == 0)
{
genotype[i] = firstParent.genotype[i];
}
else
{
genotype[i] = secondParent.genotype[i];
}
}
UpdateFitness();
}
public Individ(Individ a, Individ b, int pos) //Конструктор особи при кроссовере
{
code = new bool[fullLength];
int codePos = pos * shortLength;
for (int i = 0; i < codePos; i++)
{
code[i] = a.code[i];
}
for (int i = codePos; i < fullLength; i++)
{
code[i] = b.code[i];
}
UpdateFitness();
}
public DataCase(int a, int b, int r)
{
rnd = new Random();
if (r % (b - 1) == 0)
{
MemoryAllocation(a, b, r);
RandInitValues();
}
else
{
Console.WriteLine("Недопустимые значения, инициализация по умолчанию!\n");
MemoryAllocation();
BaseInitValues(); //Метод инициализации проверочными значениями
}
Individ.StaticInit(numberOfCompany, numberOfAttachmentSizes, resources, tableOfOffers, sizeOfAttachment);
}
public void GreedyAlg()
{
Individ[] solutionPair = new Individ[2];
for (int i = 0; i < 2; i++)
{
solutionPair[i] = new Individ();
}
int counter = numberOfAttachmentSizes;
for (int i = 0; i < numberOfCompany; i++)
{
int maxGain = -sizeOfAttachment[numberOfAttachmentSizes - 1];
int pos = -1;
for (int j = 0; j < counter; j++)
{
int gain = tableOfOffers[i, j] - sizeOfAttachment[j];
if (gain >= maxGain)
{
maxGain = gain;
pos = j;
}
}
solutionPair[0].Set(i, pos);
counter -= pos;
}
counter = numberOfAttachmentSizes;
for (int i = numberOfCompany - 1; i >= 0; i--)
{
int maxGain = -sizeOfAttachment[numberOfAttachmentSizes - 1];
int pos = -1;
for (int j = 0; j < counter; j++)
{
int gain = tableOfOffers[i, j] - sizeOfAttachment[j];
if (gain >= maxGain)
{
maxGain = gain;
pos = j;
}
}
solutionPair[1].Set(i, pos);
counter -= pos;
}
solutionPair[0].Print();
solutionPair[1].Print();
}
private Individ[] WheelRotation(Individ[] population, int populationSize, int sumFitness) //Уже лучше, но не идеал
{
Individ[] newPopulation = new Individ[populationSize];
for (int i = 0; i < populationSize; i++)
{
int rangePoint = rnd.Next(sumFitness); //Не нравится поиск, переписать, основываясь на лабе по ген алгоритмам
double topRange = 0; //Конкретно не нравится интовый поиск по даблам
for (int j = 0; j < populationSize * 3; j++)
{
topRange += population[j].GetFitness();
if (rangePoint < topRange)
{
newPopulation[i] = new Individ(population[j]);
break;
}
}
}
return(newPopulation);
}
private Individ[] WheelRotation(Individ[] population, int populationSize, int sumFitness)
{
Individ[] newPopulation = new Individ[populationSize];
for (int i = 0; i < populationSize; i++)
{
int rangePoint = rnd.Next(sumFitness);
double topRange = 0;
for (int j = 0; j < populationSize * 3; j++)
{
topRange += population[j].GetFitness();
if (rangePoint < topRange)
{
newPopulation[i] = new Individ(population[j]);
break;
}
}
}
return(newPopulation);
}
private Individ SinglepointCrossover(Individ[] population, int populationSize)
{
Individ a = new Individ(population[rnd.Next(populationSize)], population[rnd.Next(populationSize)], rnd.Next(1, numberOfCompany));
return(a);
}
public void GeneticAlg() //Отвратительно работает, когда количество компаний больше количества предложений, т.е. когда в оптимальном генотипе есть множество нулей
{
int populationSize = numberOfAttachmentSizes * numberOfCompany;
Individ.StaticInit(numberOfCompany, numberOfAttachmentSizes, resources, tableOfOffers, sizeOfAttachment);
Console.WriteLine("Стартовая популяция: ");
Individ[] population = new Individ[populationSize];
for (int i = 0; i < populationSize; i++)
{
population[i] = new Individ(rnd, rnd.Next(2)); //Случайная генерация стартовой популяции
//population[i].Print();
}
Console.WriteLine("Конец стартовой популяции");
int counter = 0;
while (counter++ < 100)
{
double sumFitness = 0;
Individ[] child = new Individ[populationSize * 3];
for (int i = 0; i < populationSize; i++)
{
child[i] = new Individ(population[i]);
sumFitness += child[i].GetFitness();
}
for (int i = populationSize; i < populationSize * 3; i++) //Создание потомков путем одноточечного кроссовера
{
child[i] = Crossover(population, populationSize); //Можно вставить формулу, но так красивее
sumFitness += child[i].GetFitness();
}
for (int i = 0; i < populationSize / 10; i++)
{
sumFitness += child[rnd.Next(populationSize * 3)].Mutation(rnd);
}
double averageFitness = sumFitness / (populationSize * 3);
population = WheelRotation(child, populationSize, Convert.ToInt32(Math.Ceiling(sumFitness)));
//Console.WriteLine("Потомки:");
//for (int i = 0; i < populationSize * 2; i++)
// child[i].Print();
//population = ProportionalSelection(population, child, populationSize, averageFitness);
//Console.WriteLine("Суммарная приспособленность равна: " + sumFitness);
//Console.WriteLine("Средняя приспособленность: " + averageFitness); //Округляет вверх всегда
//for (int i = 0; i < populationSize; i++)
// population[i].Print();
}
//for (int i = 0; i < populationSize; i++)
// population[i].Print();
Console.WriteLine("Наилучшее найденное решение: ");
population[FindBest(population, populationSize)].Print();
}
public void GeneticAlg(System.Diagnostics.Stopwatch record) //Отвратительно работает, когда количество компаний больше количества предложений, т.е. когда в оптимальном генотипе есть множество нулей
{
System.Diagnostics.Stopwatch stopwatch = new System.Diagnostics.Stopwatch();
stopwatch.Start();
int populationSize = numberOfAttachmentSizes;
Individ[] population = new Individ[populationSize];
for (int i = 0; i < populationSize; i++)
{
population[i] = new Individ();
}
//RandPopulation(population, populationSize); //Случайная генерация стартовой популяции
GreedyPopulation(population, populationSize); //Генерация стартовой популяции жадными алгоритмами со сдвигом
//for (int i = 0; i < populationSize; i++)
// population[i].Print(1);
//int testcount = 0;
int iterationCounter = 0;
while ((stopwatch.ElapsedMilliseconds < record.ElapsedMilliseconds / 2) /*| testcount++ < 100*/)
{
double sumFitness = 0;
Individ[] child = new Individ[populationSize * 3];
for (int i = 0; i < populationSize; i++)
{
child[i] = new Individ(population[i]);
sumFitness += child[i].GetFitness();
}
for (int i = populationSize; i < populationSize * 3; i++) //Создание потомков путем одноточечного кроссовера
{
if (rnd.Next(2) == 0)
{
child[i] = SinglepointCrossover(population, populationSize);
}
else
{
child[i] = UniformCrossover(population, populationSize);
}
sumFitness += child[i].GetFitness();
}
for (int i = 0; i < populationSize / 10; i++)
{
sumFitness += child[rnd.Next(populationSize, populationSize * 3)].Mutation(rnd); //Мутации производятся только среди потомков
}
double averageFitness = sumFitness / (populationSize * 3);
//Console.WriteLine(sumFitness);
//population = WheelRotation(child, populationSize, sumFitness);
//population = WheelRotation(child, populationSize, Convert.ToInt32(Math.Ceiling(sumFitness))); //Старая версия
//Console.WriteLine("Потомки:");
//for (int i = 0; i < populationSize * 2; i++)
// child[i].Print();
population = ProportionalSelection(population, child, populationSize, averageFitness);
//Console.WriteLine("Суммарная приспособленность равна: " + sumFitness);
//Console.WriteLine("Средняя приспособленность: " + averageFitness); //Округляет вверх всегда
//for (int i = 0; i < populationSize; i++)
// population[i].Print();
iterationCounter++;
}
stopwatch.Stop();
Console.WriteLine("Количество итераций: {0}\nНаилучшее найденное решение: ", iterationCounter);
population[FindBest(population, populationSize)].Print();
//for (int i = 0; i < populationSize; i++)
// Console.WriteLine(population[i].GetFitness());
}
