public Population(int count, IPopulationGenerator <T> generator, IFitnessEvaluator <T> evaluator)
{
if (count <= 0)
{
throw new ArgumentException($"{nameof(count)} must be greater than 0. Actual: {count}.");
}
if (generator == null)
{
throw new System.ArgumentNullException(nameof(generator));
}
if (evaluator == null)
{
throw new System.ArgumentNullException(nameof(evaluator));
}
TargetSize = count;
// TODO: This really shouldn't be here. We should instead pass the
// population
for (var i = 0; i < count; i++)
{
var value = generator.Generate();
var fitness = evaluator.ComputeFitness(value);
var chromosome = new Chromosome <T>(value, fitness);
Chromosomes.Add(chromosome);
}
}
public Population(int count, IPopulationGenerator <T> generator, IFitnessEvaluator <T> evaluator)
{
Contract.Requires <ArgumentOutOfRangeException>(count > 0);
Contract.Requires <ArgumentNullException>(generator != null);
Contract.Requires <ArgumentNullException>(evaluator != null);
TargetSize = count;
Evaluator = evaluator;
// TODO: This really shouldn't be here. We should instead pass the
// population
for (var i = 0; i < count; i++)
{
var value = generator.Generate();
var chromosome = new Chromosome <T>(value);
chromosome.Fitness = evaluator.ComputeFitness(chromosome.Genes);
Chromosomes.Add(chromosome);
}
}
// TODO: Write results to InnerResult.
public InnerResult start(int population_size, int max_generations = int.MaxValue, bool silent = false)
{
int last_index = population_size - 1;
int convergence_hit = -1;
current_generation = 0;
int first_goal_hit = -1;
population = new List <T>();
//Console.WriteLine("Going to run algorithm. Max generations: {0}, Min fitness: {1}", goal.MaxGenerations, goal.MinFitness);
var stopwatch = System.Diagnostics.Stopwatch.StartNew();
population = generator.Generate(population_size, genome_size);
population[0].FunctionEvaluations = 0;
do
{
shufflePopulation();
procreatePopulation(population_size);
//sortPopulation();
//selectPopulation(population_size);
population = selector.DoSelection(population, population_size).ToList();
if (!silent)
{
printPopulation();
}
//Dit moet nog even netter worden gemaakt eigenlijk met Goal enzo...:
if (population[0].Fitness == 2)
{
break;
}
current_generation++;
if (first_goal_hit < 0 && population[0].Fitness >= goal.MinFitness)
{
first_goal_hit = current_generation;
}
// No new genes entered the population (whatever the genome as, the last one should have shifted.
// Comparing pointers here
if ((IGenome)previous_last == (IGenome)population[last_index])
{
convergence_hit = current_generation;
break;
}
// Or if all genomes are the same, possibly costly operation.
if (converged(last_index))
{
convergence_hit = current_generation;
break;
}
if (current_generation >= max_generations)
{
break;
}
} while (true);
stopwatch.Stop();
Console.WriteLine(results(population_size));
InnerResult res = new InnerResult();
res.FirstHitGeneration = first_goal_hit;
res.Success = goal.AchievedFitnessGoal(population[0].Fitness);
res.ConvergenceGeneration = convergence_hit;
res.CPUTime = stopwatch.ElapsedTicks;
res.FunctionEvaluations = population[0].FunctionEvaluations;
res.BestScore = population[0].Fitness;
res.BestSolution = population[0].Data;
if (best_result == null || population[0].Fitness > best_result.Fitness)
{
best_result = population[0];
}
return(res);
}
