public void Evolve_ManyGenerations_Fast()
{
var selection = new EliteSelection();
var crossover = new UniformCrossover();
var mutation = new UniformMutation(true);
var chromosome = new AutoConfigChromosome();
var targetChromosome = new TspChromosome(10);
var targetFitness = new TspFitness(10, 0, 100, 0, 100);
var fitness = new AutoConfigFitness(targetFitness, targetChromosome);
fitness.PopulationMinSize = 20;
fitness.PopulationMaxSize = 20;
fitness.Termination = new TimeEvolvingTermination(TimeSpan.FromSeconds(5));
var population = new Population(10, 10, chromosome);
var ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation);
ga.TaskExecutor = new SmartThreadPoolTaskExecutor()
{
MinThreads = 10,
MaxThreads = 20
};
ga.Termination = new GenerationNumberTermination(2);
ga.Start();
Assert.NotNull(ga.BestChromosome);
}
private void Start()
{
var fitness = new TspFitness(m_numberOfCities);
var chromosome = new TspChromosome(m_numberOfCities);
// This operators are classic genetic algorithm operators that lead to a good solution on TSP,
// but you can try others combinations and see what result you get.
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var selection = new RouletteWheelSelection();
var population = new Population(50, 100, chromosome);
m_ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation);
m_ga.Termination = new TimeEvolvingTermination(System.TimeSpan.FromHours(1));
// The fitness evaluation of whole population will be running on parallel.
m_ga.TaskExecutor = new ParallelTaskExecutor
{
MinThreads = 100,
MaxThreads = 200
};
// Everty time a generation ends, we log the best solution.
m_ga.GenerationRan += delegate
{
var distance = ((TspChromosome)m_ga.BestChromosome).Distance;
Debug.Log($"Generation: {m_ga.GenerationsNumber} - Distance: ${distance}");
};
DrawCities();
// Starts the genetic algorithm in a separate thread.
m_gaThread = new Thread(() => m_ga.Start());
m_gaThread.Start();
}
public void Evolve_ManyGenerations_Fast()
{
int numberOfCities = 40;
var selection = new EliteSelection();
var crossover = new OrderedCrossover();
var mutation = new TworsMutation();
var chromosome = new TspChromosome(numberOfCities);
var fitness = new TspFitness(numberOfCities, 0, 1000, 0, 1000);
var population = new Population(40, 40, chromosome);
var ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation);
ga.Start();
var firstDistance = ((TspChromosome)ga.Population.BestChromosome).Distance;
ga.Termination = new GenerationNumberTermination(1001);
TimeAssert.LessThan(100000, () =>
{
ga.Start();
});
var lastDistance = ((TspChromosome)ga.Population.BestChromosome).Distance;
Assert.Less(lastDistance, firstDistance);
}
/// <summary>
/// Creates the fitness.
/// </summary>
/// <returns>The fitness.</returns>
public override IFitness CreateFitness()
{
var r = Context.DrawingArea;
m_fitness = new TspFitness(m_numberOfCities, r.Left, r.Right, r.Top, r.Bottom);
return(m_fitness);
}
static GeneticAlgorithm CreateGA(IPopulation population)
{
var selection = new EliteSelection();
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var fitness = new TspFitness(100, 0, 1000, 0, 1000);
return(new GeneticAlgorithm(population, fitness, selection, crossover, mutation));
}
public void Evaluate_ChromosomeWithLowerCities_FitnessDividedByDiff()
{
var target = new TspFitness(10, 0, 10, 0, 10);
var chromosome = new TspChromosome(9);
var actual = target.Evaluate(chromosome);
Assert.AreNotEqual(0, actual);
}
public void Evaluate_FitnessLowerThanZero_Zero()
{
var target = new TspFitness(10, 0, 10000000, 0, 10000000);
var chromosome = new TspChromosome(10);
var actual = target.Evaluate(chromosome);
Assert.AreEqual(0, actual);
}
public override IFitness CreateFitness()
{
var targetChromosome = new TspChromosome(10);
var targetFitness = new TspFitness(10, 0, 100, 0, 100);
var fitness = new AutoConfigFitness(targetFitness, targetChromosome);
fitness.Termination = new FitnessStagnationTermination(500);
fitness.PopulationMinSize = 20;
fitness.PopulationMaxSize = 20;
return(fitness);
}
public void Evaluate_StartFailed_ZeroFitness()
{
var chromosome = new AutoConfigChromosome();
var targetChromosome = Substitute.For <IChromosome>();
targetChromosome.CreateNew().Returns(x => throw new Exception("TEST"));
var targetFitness = new TspFitness(10, 0, 100, 0, 100);
var target = new AutoConfigFitness(targetFitness, targetChromosome);
var actual = target.Evaluate(chromosome);
Assert.AreEqual(0, actual);
}
public void Initialize(int numberOfCities, int areaWidth, int areaHeight)
{
Stop();
Fitness = new TspFitness(numberOfCities, areaWidth, areaHeight);
var chromosome = new TspChromosome(numberOfCities);
// This operators are classic genetic algorithm operators that lead to a good solution on TSP,
// but you can try others combinations and see what result you get.
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var selection = new RouletteWheelSelection();
var population = new Population(50, 100, chromosome);
_ga = new GeneticAlgorithm(population, Fitness, selection, crossover, mutation);
}
public void Evaluate_StartOk_Fitness()
{
var chromosome = new AutoConfigChromosome();
var targetChromosome = new TspChromosome(10);
var targetFitness = new TspFitness(10, 0, 100, 0, 100);
var target = new AutoConfigFitness(targetFitness, targetChromosome);
target.PopulationMinSize = 20;
target.PopulationMaxSize = 20;
target.Termination = new FitnessThresholdTermination(0.1f);
var actual = target.Evaluate(chromosome);
Assert.AreNotEqual(0, actual);
}
public void Evolve_CompareToSingleChromosome_Evolved()
{
int numberOfCities = 30;
var selection = new EliteSelection();
var crossover = new UniformCrossover();
var mutation = new TworsMutation();
// Given enough generations, the Multiple Chromosome should start exhibiting convergence
// we compare TSP /25 gen with 3*TSP / 500 gen
IChromosome chromosome = new TspChromosome(numberOfCities);
IFitness fitness = new TspFitness(numberOfCities, 0, 1000, 0, 1000);
var population = new Population(30, 30, chromosome);
var ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = new GenerationNumberTermination(26)
};
ga.Start();
var simpleChromosomeDistance = ((TspChromosome)ga.Population.BestChromosome).Distance;
chromosome = new MultipleChromosome((i) => new TspChromosome(numberOfCities), 3);
//MultiChromosome should create 3 TSP chromosomes and store them in the corresponding property
Assert.AreEqual(((MultipleChromosome)chromosome).Chromosomes.Count, 3);
var tempMultiFitness = ((MultipleChromosome)chromosome).Chromosomes.Sum(c => fitness.Evaluate(c));
fitness = new MultipleFitness(fitness);
//Multi fitness should sum over the fitnesses of individual chromosomes
Assert.AreEqual(tempMultiFitness, fitness.Evaluate(chromosome));
population = new Population(30, 30, chromosome);
ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = new GenerationNumberTermination(501)
};
ga.Start();
var bestTSPChromosome = (TspChromosome)((MultipleChromosome)ga.Population.BestChromosome).Chromosomes
.OrderByDescending(c => c.Fitness).First();
var multiChromosomesDistance = bestTSPChromosome.Distance;
Assert.Less(multiChromosomesDistance, simpleChromosomeDistance);
}
public void GA_WithAlternatingPositionCrossover_Evolve()
{
var chromosome = new TspChromosome(50);
var population = new Population(50, 50, chromosome)
{
GenerationStrategy = new TrackingGenerationStrategy()
};
var fitness = new TspFitness(chromosome.Length, 0, 1000, 0, 1000);
var crossover = new AlternatingPositionCrossover();
var ga = new GeneticAlgorithm(population, fitness, new EliteSelection(), crossover, new ReverseSequenceMutation())
{
Termination = new GenerationNumberTermination(100)
};
ga.Start();
Assert.Less(
population.Generations.First().BestChromosome.Fitness.Value,
population.Generations.Last().BestChromosome.Fitness.Value);
}
static void Main(string[] args)
{
var fitness = new TspFitness(m_numberOfCities);
var chromosome = new TspChromosome(m_numberOfCities);
// This operators are classic genetic algorithm operators that lead to a good solution on TSP,
// but you can try others combinations and see what result you get.
var crossover = new GeneticSharp.Domain.Crossovers.OrderedCrossover();
var mutation = new GeneticSharp.Domain.Mutations.ReverseSequenceMutation();
var selection = new GeneticSharp.Domain.Selections.RouletteWheelSelection();
var population = new GeneticSharp.Domain.Populations.Population(500, 1000, chromosome);
m_ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation);
m_ga.Termination = new GeneticSharp.Domain.Terminations.TimeEvolvingTermination(System.TimeSpan.FromHours(1));
// The fitness evaluation of whole population will be running on parallel.
m_ga.TaskExecutor = new ParallelTaskExecutor
{
MinThreads = 100,
MaxThreads = 200
};
// Every time a generation ends, we log the best solution.
m_ga.GenerationRan += delegate
{
var distance = ((TspChromosome)m_ga.BestChromosome).Distance;
Console.WriteLine($"Generation: {m_ga.GenerationsNumber} - Distance: ${distance}");
};
// Starts the genetic algorithm in a separate thread.
m_gaThread = new Thread(() => m_ga.Start());
m_gaThread.Start();
Console.ReadLine();
// When the script is destroyed we stop the genetic algorithm and abort its thread too.
m_ga.Stop();
m_gaThread.Abort();
}
protected override GeneticAlgorithm CreateGA()
{
var size = (int)Camera.main.orthographicSize - 1;
m_fitness = new TspFitness(m_numberOfCities, -size, size, -size, size);
var chromosome = new TspChromosome(m_numberOfCities);
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var selection = new RouletteWheelSelection();
var population = new Population(50, 100, chromosome);
var ga = new GeneticAlgorithm(population, m_fitness, selection, crossover, mutation);
ga.Termination = new TimeEvolvingTermination(System.TimeSpan.FromDays(1));
ga.TaskExecutor = new ParallelTaskExecutor
{
MinThreads = 100,
MaxThreads = 200
};
return(ga);
}
private GeneticAlgorithm CreateGA(Func <TspChromosome, Population> createPopulation = null)
{
var selection = new EliteSelection();
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var chromosome = new TspChromosome(_numberOfCities);
var fitness = new TspFitness(_numberOfCities, 0, 1000, 0, 1000);
var population = createPopulation == null
? new Population(_minPopulationSize, _minPopulationSize, chromosome)
: createPopulation(chromosome);
population.GenerationStrategy = new PerformanceGenerationStrategy();
var ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = new GenerationNumberTermination(_generations)
};
return(ga);
}
/// <summary>
/// GeneticSharp TSP Console Application template.
/// <see href="https://github.com/giacomelli/GeneticSharp"/>
/// </summary>
static void Main(string[] args)
{
var selection = new EliteSelection();
var crossover = new OrderedCrossover();
var mutation = new TworsMutation();
var fitness = new TspFitness(20, 0, 1000, 0, 1000);
var chromosome = new TspChromosome(fitness.Cities.Count);
var population = new Population(50, 70, chromosome);
var ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = new FitnessStagnationTermination(100)
};
ga.GenerationRan += (s, e) =>
{
Console.Clear();
Console.WriteLine($"Generation: {ga.GenerationsNumber}");
var c = ga.BestChromosome as TspChromosome;
Console.WriteLine("Cities: {0:n0}", c.Length);
Console.WriteLine("Distance: {0:n2}", c.Distance);
var cities = c.GetGenes().Select(g => g.Value.ToString()).ToArray();
Console.WriteLine("City tour: {0}", string.Join(", ", cities));
};
Console.WriteLine("GA running...");
ga.Start();
Console.WriteLine();
Console.WriteLine($"Elapsed time: {ga.TimeEvolving}");
Console.WriteLine("Done.");
Console.ReadKey();
}
public void TSPSolve(int numberOfPoints, List <Point> stationList)
{
var fitness = new TspFitness(stationList);
var chromosome = new TspChromosome(numberOfPoints);
var crossover = new OrderedCrossover();
var mutation = new ReverseSequenceMutation();
var selection = new RouletteWheelSelection();
var population = new Population(50, 100, chromosome);
m_ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation);
m_ga.Termination = new FitnessStagnationTermination(100);
m_ga.TaskExecutor = new ParallelTaskExecutor {
MinThreads = 100,
MaxThreads = 200
};
m_ga.GenerationRan += delegate {
var distance = ((TspChromosome)m_ga.BestChromosome).Distance;
};
m_ga.TerminationReached += delegate {
var c = m_ga.Population.CurrentGeneration.BestChromosome as TspChromosome;
var genes = c.GetGenes();
var points = ((TspFitness)m_ga.Fitness).Points;
finalPoints = new List <Point>();
for (int i = 0; i < genes.Length; i++)
{
finalPoints.Add(points[(int)genes[i].Value]);
}
tspSolved();
};
m_gaThread = new Thread(() => m_ga.Start());
m_gaThread.Start();
}
public IFitness CreateFitness()
{
m_fitness = new TspFitness(m_numberOfCities, 0, 1000, 0, 1000);
return(m_fitness);
}
