public void WriteToStream_Should_Write_All_Population_DNA_To_Given_Stream()
{
int individuals = 5;
int genesEach = 20;
MockRandomEnvironment.Setup(m => m.GetNextDouble(0, It.IsAny <double>())).Returns(0.1f);
PopulationManager populationManagerTested = GetPopulationManager();
populationManagerTested.GeneratePopulation(individuals, genesEach);
double fitness = 0;
populationManagerTested.Population.ForEach(dna => dna.Fitness = fitness++);
MemoryStream stream = new MemoryStream();
populationManagerTested.WriteToStream(stream);
stream.Position = 0;
using (StreamReader sr = new StreamReader(stream))
{
int currentIndividual = 0;
do
{
string line = sr.ReadLine();
string expectedLine = $"{populationManagerTested.Population[currentIndividual].Fitness}:{populationManagerTested.Population[currentIndividual].ToString()}";
line.Should().Be(expectedLine);
currentIndividual++;
}while (currentIndividual < individuals);
}
}
private App(MySqlConnection dbConn, Dictionary <string, string> cities)
{
//1. init the populationManager with connection to the database
populationManager = new PopulationManager(cities);
//2. Generate the first population
population = populationManager.GeneratePopulation();
}
public void GeneratePopulation_Should_Create_Given_Number_Of_Individuals_With_Given_Number_Of_Genes_Each_Between_One_And_Minus_One()
{
int individuals = 10;
int genesEach = 200;
PopulationManager populationManagerTested = GetPopulationManager();
populationManagerTested.GeneratePopulation(individuals, genesEach);
populationManagerTested.Population.Should().HaveCount(individuals);
populationManagerTested.Population.ForEach(p => p.Genes.Should().HaveCount(genesEach));
populationManagerTested.Population.ForEach(p => p.Genes.ForEach(g => g.Should().BeInRange(-1, 1)));
}
public void NextGeneration_Should_Set_Fitness_Of_Population_To_Zero()
{
int individuals = 5;
int genesEach = 20;
MockRandomEnvironment.Setup(m => m.GetNextDouble(0, It.IsAny <double>())).Returns(0.1f);
PopulationManager populationManagerTested = GetPopulationManager();
populationManagerTested.GeneratePopulation(individuals, genesEach);
double fitness = 0;
populationManagerTested.Population.ForEach(dna => dna.Fitness = fitness++);
populationManagerTested.NextGeneration();
populationManagerTested.Population.Select(i => i.Fitness).Should().OnlyContain(i => i == 0.0f);
}
public void NextGeneration_Should_Complete_Population_To_Reach_Same_Size_By_Combining_Best()
{
int individuals = 5;
int genesEach = 20;
MockRandomEnvironment.Setup(m => m.GetNextDouble(0, It.IsAny <double>())).Returns(0.1f);
PopulationManager populationManagerTested = GetPopulationManager();
populationManagerTested.GeneratePopulation(individuals, genesEach);
double fitness = 0;
populationManagerTested.Population.ForEach(dna => dna.Fitness = fitness++);
populationManagerTested.NextGeneration();
populationManagerTested.Population.Should().HaveCount(individuals);
}
public void NextGeneration_Should_Copy_At_Least_The_Individual_With_Best_Fitness()
{
int individuals = 5;
int genesEach = 20;
PopulationManager populationManagerTested = GetPopulationManager();
populationManagerTested.GeneratePopulation(individuals, genesEach);
double fitness = 0;
populationManagerTested.Population.ForEach(dna => dna.Fitness = fitness++);
var oldPopulation = populationManagerTested.Population.ToList();
var best = oldPopulation.OrderByDescending(dna => dna.Fitness).First();
populationManagerTested.NextGeneration();
populationManagerTested.Population.Should().Contain(best);
}
private static PopulationManager GeneratePopulation(int inputCount, int outputCount, int depth, int hiddenNeuronsPerLayer)
{
PopulationManager pop = new PopulationManager(new RandomEnvironment());
if (File.Exists(PopulationFileToLoad))
{
FileStream stream = new FileStream(PopulationFileToLoad, FileMode.Open);
pop.LoadFromStream(stream);
Console.WriteLine($"Loaded from file {PopulationFileToLoad}");
}
else
{
Console.WriteLine($"File {PopulationFileToLoad} not found, generating population at random");
pop.GeneratePopulation(30, hiddenNeuronsPerLayer * (inputCount + depth * hiddenNeuronsPerLayer + outputCount));
}
return(pop);
}
