public void Setup()
{
_chromosomeService = new ChromosomeService();
_populationService = new PopulationService(_chromosomeService);
_matrixService = new MatrixService(new GraphConsistentService());
_sut = new EvolutionService(_chromosomeService);
}
public EvolutionController(IValidatorFactory validatorFactory,
IEvolutionService evolutionService)
{
_validatorFactory = validatorFactory;
_evolutionService = evolutionService;
}
public Algorithm(IEvolutionService evolutionService, IEnvironmentService environmentService)
{
var population = environmentService.GenerateInitialisedPopulation(75);
// Evolve our population until we reach an near-optimal solution
var generationCount = 0;
var maxGenerations = 100000;
var globalHighestScore = 0;
var generationsWithNoChangeToScore = 0;
var maxGenerationsWithNoChangeToScore = 400;
// Simulate 'maxGenerations' amount of generations, unless the score hasn't changed for 'maxGenerationsWithNoChangeToScore'.
while (generationCount < maxGenerations &&
generationsWithNoChangeToScore < maxGenerationsWithNoChangeToScore)
{
generationCount++;
var scoreOfFittestIndividual = population.MostSuitableIndividualToProblem.SuitabilityScore;
// If the score hasn't improved since the last generation ->
if (scoreOfFittestIndividual == globalHighestScore)
{
// -> Increment 'generationsWithNoChangeToScore' by 1.
// TODO: Check how many parts were high value assignments,
// to influence the overwriting of better assignments
generationsWithNoChangeToScore++;
}
else
{
// Reset 'generationsWithNoChangeToScore' to 0.
generationsWithNoChangeToScore = 0;
}
// If the 'scoreOfFittestIndividual' is more than the 'globalHighestScore' ->
if (scoreOfFittestIndividual > globalHighestScore)
{
// -> update the 'globalHighestScore' with the 'scoreOfFittestIndividual'.
globalHighestScore = scoreOfFittestIndividual;
}
// [ Purpose: To log out data less frequently.
// How: Sub-sample the data by logging an update every 'subSampleRate' generations.
var subSampleRate = 10;
if (generationCount % subSampleRate == 0)
{
Console.WriteLine($"Generation [{generationCount}] ~ Highest Fitness [{scoreOfFittestIndividual}]");
}
population = evolutionService.EvolvePopulation(population);
environmentService.UpdatePopulationSuitability(population);
// ]
}
Console.WriteLine("\nSolution found!");
if (generationsWithNoChangeToScore == maxGenerationsWithNoChangeToScore)
{
Console.WriteLine(
$"The score hasn't changed in {maxGenerationsWithNoChangeToScore} generations, so it has been considered an optimal solution.");
}
Console.WriteLine("\n\nGeneration: " + generationCount);
Console.WriteLine("\n\nSolution\n--------");
Console.WriteLine(population.MostSuitableIndividualToProblem);
Console.WriteLine("\nFitness: " + population.MostSuitableIndividualToProblem.SuitabilityScore);
}
