/// <summary>
/// Determine the complexity regulation mode that the evolution algorithm search should be in given the
/// provided evolution algorithm statistics object, and set the current mode to that mode.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics.</param>
/// <param name="popStats">Population statistics.</param>
/// <returns>The determined <see cref="ComplexityRegulationMode"/>.</returns>
public ComplexityRegulationMode UpdateMode(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
// This is the null strategy, therefore do nothing.
return(ComplexityRegulationMode.Complexifying);
}
//
// GET: /Population/
public ActionResult Index()
{
var statistics = new PopulationStatistics();
var countries = statistics.GetCountries();
return(View(countries));
}
private ComplexityRegulationMode DetermineMode_WhileSimplifying(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
// Currently simplifying.
// Test if simplification (ongoing reduction in complexity) has stalled.
// We allow simplification to progress for a few generations before testing of it has stalled, this allows
// a lead in time for the effects of simplification to occur.
// In addition we do not switch to complexifying if complexity is above the currently defined ceiling.
if (
((eaStats.Generation - _lastTransitionGeneration) > _minSimplifcationGenerations) &&
(popStats.MeanComplexity < _complexityCeiling) &&
((popStats.MeanComplexityHistory.Mean - _prevMeanMovingAverage) >= 0.0))
{
// Simplification has stalled; switch back to complexification.
_currentMode = ComplexityRegulationMode.Complexifying;
_lastTransitionGeneration = eaStats.Generation;
_prevMeanMovingAverage = 0.0;
}
// else: otherwise remain in simplifying mode.
// Update previous mean moving average complexity value.
_prevMeanMovingAverage = popStats.MeanComplexityHistory.Mean;
// Set a new complexity ceiling, relative to the current population complexity mean.
_complexityCeiling = popStats.MeanComplexity + _relativeComplexityCeiling;
return(_currentMode);
}
/// <summary>
/// Update the graph data.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics object.</param>
/// <param name="popStats">Population statistics object.</param>
public virtual void UpdateData(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
// Note. This method could be defined as abstract, but that would prevent the Window Forms UI designer from working;
// therefore instead it is defined as virtual with no implementation.
}
public ActionResult Details(string id, int?year)
{
var statistics = new PopulationStatistics();
var details = new CountryDetails();
var availableYears =
statistics.GetAvailableYearsForCountry(id);
details.Name = id;
details.AvailableYears =
availableYears.Select(availableYear =>
new SelectListItem
{
Text = availableYear.ToString(),
Value = availableYear.ToString()
});
var selectedYear = year ?? availableYears.First();
details.MalePopulation = statistics.GetMalePopulation(id, selectedYear);
details.FemalePopulation = statistics.GetFemalePopulation(id, selectedYear);
details.TotalPopulation = details.MalePopulation + details.FemalePopulation;
return(View(details));
}
private void Verify(GeneticAlgorithm genetic)
{
this.Verify((Heuristic)genetic);
PopulationStatistics statistics = genetic.PopulationStatistics;
Assert.IsTrue(statistics.Generations <= genetic.Parameters.MaximumGenerations);
Assert.IsTrue(statistics.CoefficientOfVarianceByGeneration.Count == statistics.Generations);
Assert.IsTrue(statistics.MeanAllelesPerLocusByGeneration.Count == statistics.Generations);
Assert.IsTrue(statistics.MeanHeterozygosityByGeneration.Count == statistics.Generations);
Assert.IsTrue(statistics.NewIndividualsByGeneration.Count == statistics.Generations);
Assert.IsTrue(statistics.PolymorphismByGeneration.Count == statistics.Generations);
for (int generationIndex = 0; generationIndex < statistics.Generations; ++generationIndex)
{
float coefficientOfVariation = statistics.CoefficientOfVarianceByGeneration[generationIndex];
float meanAllelesPerLocus = statistics.MeanAllelesPerLocusByGeneration[generationIndex];
float meanHeterozygosity = statistics.MeanHeterozygosityByGeneration[generationIndex];
int newIndividuals = statistics.NewIndividualsByGeneration[generationIndex];
float polymorphism = statistics.PolymorphismByGeneration[generationIndex];
Assert.IsTrue(coefficientOfVariation >= 0.0F);
Assert.IsTrue((meanAllelesPerLocus >= 0.0F) && (meanAllelesPerLocus <= 2.0F)); // assumes HarvestPeriodSelection.All
Assert.IsTrue((meanHeterozygosity >= 0.0F) && (meanHeterozygosity <= 1.0F));
Assert.IsTrue((newIndividuals >= 0) && (newIndividuals <= 2 * genetic.Parameters.PopulationSize)); // two children per breeding
Assert.IsTrue((polymorphism >= 0.0F) && (polymorphism <= 1.0F));
}
}
public void TestTransitionToSimplifying()
{
var strategy = new AbsoluteComplexityRegulationStrategy(10, 10.0);
var eaStats = new EvolutionAlgorithmStatistics();
var popStats = new PopulationStatistics();
ComplexityRegulationMode mode;
// The strategy should initialise to, and remain in, Complexifying mode
// while mean population complexity is below the threshold.
for (int i = 0; i < 11; i++)
{
eaStats.Generation = i;
popStats.MeanComplexity = i;
popStats.MeanComplexityHistory.Enqueue(i);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.AreEqual(ComplexityRegulationMode.Complexifying, mode);
}
// The strategy should switch to simplifying mode when mean complexity
// rises above the threshold.
eaStats.Generation = 11;
popStats.MeanComplexity = 10.01;
popStats.MeanComplexityHistory.Enqueue(10.01);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.AreEqual(ComplexityRegulationMode.Simplifying, mode);
}
public void VerifyPopulationStats()
{
IRandomSource rng = RandomDefaults.CreateRandomSource(0);
// Create a test population.
NeatPopulation <double> pop = CreateNeatPopulation(100, 10.0, rng);
// Modify a few genome fitnesses.
pop.GenomeList[10].FitnessInfo = new FitnessInfo(100.0);
pop.GenomeList[20].FitnessInfo = new FitnessInfo(0.0);
// Initialise the NEAT population species; the wider population stats are dependent on this occurring.
InitialiseSpecies(pop);
// Calc/update stats.
pop.UpdateStats(PrimaryFitnessInfoComparer.Singleton, rng);
// Validate stats.
// Fitness stats.
PopulationStatistics stats = pop.Stats;
Assert.Equal(10, stats.BestGenomeIndex);
Assert.Equal(100.0, stats.BestFitness.PrimaryFitness);
Assert.Equal(10.8, stats.MeanFitness);
Assert.Equal(1, stats.BestFitnessHistory.Length);
Assert.Equal(100.0, stats.BestFitnessHistory.Total);
// Complexity stats.
Assert.Equal(6.0, stats.BestComplexity);
Assert.Equal(6.0, stats.MeanComplexity);
Assert.Equal(1, stats.MeanComplexityHistory.Length);
Assert.Equal(6.0, stats.MeanComplexityHistory.Total);
}
/// <summary>
/// Update the time series data.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics object.</param>
/// <param name="popStats">Population statistics object.</param>
public override void UpdateData(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
_ppl.Add(eaStats.Generation, eaStats.EvaluationsPerSec);
RefreshGraph();
}
/// <summary>
/// Update the time series data.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics object.</param>
/// <param name="popStats">Population statistics object.</param>
public override void UpdateData(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
_bestPpl.Add(eaStats.Generation, popStats.BestComplexity);
_meanPpl.Add(eaStats.Generation, popStats.MeanComplexity);
RefreshGraph();
}
/// <summary>
/// Update the time series data.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics object.</param>
/// <param name="popStats">Population statistics object.</param>
public override void UpdateData(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
_bestPpl.Add(eaStats.Generation, popStats.BestFitness.PrimaryFitness);
_meanPpl.Add(eaStats.Generation, popStats.MeanFitness);
RefreshGraph();
}
/// <summary>
/// Determine the complexity regulation mode that the evolution algorithm search should be in given the
/// provided evolution algorithm statistics object, and set the current mode to that mode.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics.</param>
/// <param name="popStats">Population statistics.</param>
/// <returns>The determined <see cref="ComplexityRegulationMode"/>.</returns>
public ComplexityRegulationMode UpdateMode(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
return(_currentMode switch
{
ComplexityRegulationMode.Complexifying => DetermineMode_WhileComplexifying(eaStats, popStats),
ComplexityRegulationMode.Simplifying => DetermineMode_WhileSimplifying(eaStats, popStats),
_ => throw new InvalidOperationException("Unexpected complexity regulation mode."),
});
protected void Page_Load(object sender, EventArgs e)
{
if (!IsPostBack)
{
_stats = new PopulationStatistics();
BindHighestMalePopulations();
BindLowestTotalPopulations();
BindFemalePopulationIncrease();
}
}
public void Population()
{
Population binaryPopulation = new Population(2, 2, 0.5F, 5);
binaryPopulation.IndividualFitness[0] = 0.0F;
binaryPopulation.IndividualFitness[1] = 1.0F;
binaryPopulation.IndividualTreeSelections[0] = new int[] { 0, 0, 0, 0, 0 };
binaryPopulation.IndividualTreeSelections[1] = new int[] { 1, 1, 1, 1, 1 };
Population clones = new Population(2, 2, 0.5F, 5);
clones.IndividualFitness[0] = 0.0F;
clones.IndividualFitness[1] = 0.0F;
clones.IndividualTreeSelections[0] = new int[] { 0, 0, 0, 0, 0 };
clones.IndividualTreeSelections[1] = new int[] { 0, 0, 0, 0, 0 };
Population heterozygousPopulation = new Population(2, 2, 0.5F, 5);
heterozygousPopulation.IndividualFitness[0] = 0.4F;
heterozygousPopulation.IndividualFitness[1] = 0.6F;
heterozygousPopulation.IndividualTreeSelections[0] = new int[] { 1, 0, 0, 1, 0 };
heterozygousPopulation.IndividualTreeSelections[1] = new int[] { 1, 0, 1, 0, 1 };
PopulationStatistics statistics = new PopulationStatistics();
statistics.AddGeneration(binaryPopulation);
statistics.AddGeneration(clones);
statistics.AddGeneration(heterozygousPopulation);
Assert.IsTrue(statistics.Generations == 3);
Assert.IsTrue(statistics.CoefficientOfVarianceByGeneration[0] == 1.0F);
Assert.IsTrue(statistics.MaximumFitnessByGeneration[0] == 1.0F);
Assert.IsTrue(statistics.MeanFitnessByGeneration[0] == 0.5F);
Assert.IsTrue(statistics.MinimumFitnessByGeneration[0] == 0.0F);
Assert.IsTrue(statistics.MeanHeterozygosityByGeneration[0] == 0.5F);
Assert.IsTrue(statistics.NewIndividualsByGeneration[0] == 0);
Assert.IsTrue(statistics.PolymorphismByGeneration[0] == 1.0F);
Assert.IsTrue(statistics.CoefficientOfVarianceByGeneration[1] == 0.0F);
Assert.IsTrue(statistics.MaximumFitnessByGeneration[1] == 0.0F);
Assert.IsTrue(statistics.MeanFitnessByGeneration[1] == 0.0F);
Assert.IsTrue(statistics.MinimumFitnessByGeneration[1] == 0.0F);
Assert.IsTrue(statistics.MeanHeterozygosityByGeneration[1] == 0.0F);
Assert.IsTrue(statistics.NewIndividualsByGeneration[1] == 0);
Assert.IsTrue(statistics.PolymorphismByGeneration[1] == 0.0F);
Assert.IsTrue(MathF.Round(statistics.CoefficientOfVarianceByGeneration[2], 6) == 0.2F);
Assert.IsTrue(statistics.MaximumFitnessByGeneration[2] == 0.6F);
Assert.IsTrue(statistics.MeanFitnessByGeneration[2] == 0.5F);
Assert.IsTrue(statistics.MinimumFitnessByGeneration[2] == 0.4F);
Assert.IsTrue(statistics.MeanHeterozygosityByGeneration[2] == 0.3F);
Assert.IsTrue(statistics.NewIndividualsByGeneration[2] == 0);
Assert.IsTrue(statistics.PolymorphismByGeneration[2] == 0.6F);
}
/// <summary>
/// Construct the population with the provided list of genomes that make up the initial population.
/// </summary>
/// <param name="genomeList">Genome list.</param>
public Population(List <TGenome> genomeList)
{
this.GenomeList = genomeList ?? throw new ArgumentNullException(nameof(genomeList));
if (genomeList.Count == 0)
{
throw new ArgumentException("Empty genome list. The initial population cannot be empty.", nameof(genomeList));
}
this.PopulationSize = genomeList.Count;
this.Stats = CreatePopulatonStats();
}
public void TestTransitionToComplexifying()
{
var strategy = new RelativeComplexityRegulationStrategy(10, 10.0);
var eaStats = new EvolutionAlgorithmStatistics();
var popStats = new PopulationStatistics();
ComplexityRegulationMode mode;
// Cause an immediate switch to into simplifying mode.
int generation = 0;
eaStats.Generation = generation++;
popStats.MeanComplexity = 11.0;
popStats.MeanComplexityHistory.Enqueue(11.0);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.Equal(ComplexityRegulationMode.Simplifying, mode);
// Reset the buffer that the moving average is calculated from;
// This allows us to change the mean to below the threshold and for the
// moving average to start rising immediately; that would ordinarily cause
// an immediate switch back to complexifying mode, but that is prevented by
// {minSimplifcationGenerations} being set to 10.
popStats.MeanComplexityHistory.Clear();
for (int i = 0; i < 10; i++)
{
eaStats.Generation = generation++;
popStats.MeanComplexity = 2.0;
popStats.MeanComplexityHistory.Enqueue(2.0);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.Equal(ComplexityRegulationMode.Simplifying, mode);
}
// Now that {minSimplifcationGenerations} have passed, the strategy should switch
// back to complexifying mode.
eaStats.Generation = generation++;
popStats.MeanComplexity = 2.0;
popStats.MeanComplexityHistory.Enqueue(2.0);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.Equal(ComplexityRegulationMode.Complexifying, mode);
// The threshold should have been set relative to the popStats.MeanComplexity (to 12).
eaStats.Generation = generation++;
popStats.MeanComplexity = 11.9;
popStats.MeanComplexityHistory.Enqueue(11.9);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.Equal(ComplexityRegulationMode.Complexifying, mode);
eaStats.Generation = generation++;
popStats.MeanComplexity = 12.01;
popStats.MeanComplexityHistory.Enqueue(12.01);
mode = strategy.UpdateMode(eaStats, popStats);
Assert.Equal(ComplexityRegulationMode.Simplifying, mode);
}
public static void Run()
{
var config = new EAConfiguration
{
PopulationSize = 13,
OverproductionFactor = 1.5,
MaximumGenerations = 10000,
CrossoverType = CrossoverType.Uniform,
AdultSelectionType = AdultSelectionType.Overproduction,
ParentSelectionType = ParentSelectionType.Tournament,
CrossoverRate = 0.21,
MutationRate = 0.98,
TournamentSize = 7,
TournamentProbability = 0.895,
TargetFitness = 1.0,
Mode = EAMode.MaximizeFitness,
Elites = 0,
CalculateStatistics = true
};
var lolzea = new LOLZEA(config, new DefaultRandomNumberGenerator());
var stopwatchtot = new Stopwatch();
var stopwatchgen = new Stopwatch();
PopulationStatistics currentStats = new PopulationStatistics();
lolzea.PopulationStatisticsCalculated += (stats) =>
{
currentStats = stats;
};
lolzea.NewGenerationEvent += (gen) => {
//PrintProgressBar(gen, config.MaximumGenerations);
var progress = (gen / (double)config.MaximumGenerations) * 100.0;
var totruntime = stopwatchtot.Elapsed;
var genruntime = stopwatchgen.Elapsed;
Console.WriteLine();
Console.WriteLine(string.Format("G# {0} best_f: {1:F3} avg_f: {2:F3} SD: {3:F3} Progress: {4,5:F2} Gen: {5} Tot: {6}", gen, lolzea.Best.Fitness, currentStats.AverageFitness, currentStats.StandardDeviationFitness, progress, genruntime, totruntime));
Console.WriteLine($"Generation winner: {lolzea.Best.ToString()}");
stopwatchgen.Restart();
};
stopwatchtot.Start();
stopwatchgen.Start();
var res = lolzea.Evolve();
Console.WriteLine("\n\nDone!");
Console.WriteLine($"Winner: {res.Winner}");
Console.Read();
}
public static void Run()
{
var config = new EAConfiguration
{
PopulationSize = 100,
OverproductionFactor = 2.0,
MaximumGenerations = 100,
CrossoverType = CrossoverType.OnePoint,
AdultSelectionType = AdultSelectionType.GenerationalMixing,
ParentSelectionType = ParentSelectionType.Tournament,
CrossoverRate = 0.9,
MutationRate = 0.01,
TournamentSize = 10,
TournamentProbability = 0.8,
TargetFitness = 1.0,
Mode = EAMode.MaximizeFitness,
Elites = 1,
CalculateStatistics = true
};
var onemaxEA = new OneMaxEA(config, new DefaultRandomNumberGenerator());
var stopwatchtot = new Stopwatch();
var stopwatchgen = new Stopwatch();
PopulationStatistics currentStats = new PopulationStatistics();
onemaxEA.PopulationStatisticsCalculated += (stats) =>
{
currentStats = stats;
};
onemaxEA.NewGenerationEvent += (gen) => {
//PrintProgressBar(gen, config.MaximumGenerations);
double progress = (gen / (double)config.MaximumGenerations) * 100.0;
var totruntime = stopwatchtot.Elapsed;
var genruntime = stopwatchgen.Elapsed;
Console.WriteLine();
Console.WriteLine(string.Format("G# {0} best_f: {1:F3} avg_f: {2:F3} SD: {3:F3} Progress: {4,5:F2} Gen: {5} Tot: {6}", gen, onemaxEA.Best.Fitness, currentStats.AverageFitness, currentStats.StandardDeviationFitness, progress, genruntime, totruntime));
Console.WriteLine("Generation winner: " + ((BinaryGenotype)onemaxEA.Best?.Genotype).ToBitString());
stopwatchgen.Restart();
};
stopwatchtot.Start();
stopwatchgen.Start();
var res = onemaxEA.Evolve();
Console.WriteLine("\n\nDone!");
Console.WriteLine($"Winner: {res.Winner}");
Console.Read();
}
/// <summary>
/// Determine the complexity regulation mode that the evolution algorithm search should be in given the
/// provided evolution algorithm statistics object.
/// </summary>
/// <param name="eaStats">Evolution algorithm statistics.</param>
/// <param name="popStats">Population statistics.</param>
public ComplexityRegulationMode UpdateMode(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
switch (_currentMode)
{
case ComplexityRegulationMode.Complexifying:
return(DetermineMode_WhileComplexifying(eaStats, popStats));
case ComplexityRegulationMode.Simplifying:
return(DetermineMode_WhileSimplifying(eaStats, popStats));
}
throw new InvalidOperationException("Unexpected complexity regulation mode.");
}
public void TestInitialisation()
{
var strategy = new AbsoluteComplexityRegulationStrategy(10, 10.0);
var eaStats = new EvolutionAlgorithmStatistics();
var popStats = new PopulationStatistics();
// The strategy should initialise to, and remain in, Complexifying mode.
for (int i = 0; i < 100; i++)
{
eaStats.Generation = i;
ComplexityRegulationMode mode = strategy.UpdateMode(eaStats, popStats);
Assert.AreEqual(ComplexityRegulationMode.Complexifying, mode);
}
}
private ComplexityRegulationMode DetermineMode_WhileComplexifying(
EvolutionAlgorithmStatistics eaStats,
PopulationStatistics popStats)
{
// Currently complexifying.
// Test if the complexity ceiling has been reached.
if (popStats.MeanComplexity > _complexityCeiling)
{
// Switch to simplifying mode.
_currentMode = ComplexityRegulationMode.Simplifying;
_lastTransitionGeneration = eaStats.Generation;
_prevMeanMovingAverage = popStats.MeanComplexityHistory.Mean;
}
return(_currentMode);
}
/// <summary>
/// Update the population statistics object.
/// </summary>
/// <param name="fitnessComparer">A genome fitness comparer.</param>
/// <param name="rng">Random source.</param>
public override void UpdateStats(
IComparer <FitnessInfo> fitnessComparer,
IRandomSource rng)
{
// Calculate some population-wide stats; these are non-NEAT specific.
CalcPopulationStats(out double primaryFitnessSum, out double complexitySum, out double maxComplexity);
// Calculate NEAT specific and species based stats.
CalcNeatPopulationStats(
fitnessComparer, rng,
out double sumMeanFitness,
out double sumBestFitness,
out int bestGenomeIdx,
out int bestGenomeSpeciesIdx);
// Update PopulationStatistics object.
PopulationStatistics stats = this.Stats;
int genomeCount = this.GenomeList.Count;
var bestGenome = this.GenomeList[bestGenomeIdx];
// Update fitness stats.
stats.BestGenomeIndex = bestGenomeIdx;
stats.BestFitness = bestGenome.FitnessInfo;
stats.MeanFitness = primaryFitnessSum / genomeCount;
stats.BestFitnessHistory.Enqueue(bestGenome.FitnessInfo.PrimaryFitness);
// Update complexity stats.
stats.BestComplexity = bestGenome.Complexity;
double meanComplexity = complexitySum / genomeCount;
stats.MeanComplexity = meanComplexity;
stats.MeanComplexityHistory.Enqueue(meanComplexity);
stats.MaxComplexity = maxComplexity;
// Update NeatPopulationStatistics object.
this.NeatPopulationStats.BestGenomeSpeciesIdx = bestGenomeSpeciesIdx;
this.NeatPopulationStats.SumSpeciesMeanFitness = sumMeanFitness;
this.NeatPopulationStats.AverageSpeciesBestFitness = sumBestFitness / SpeciesArray !.Length;
}
protected override void ProcessRecord()
{
if (this.Runs !.Count < 1)
{
throw new ArgumentOutOfRangeException(nameof(this.Runs));
}
using StreamWriter writer = this.GetWriter();
StringBuilder line = new StringBuilder();
if (this.ShouldWriteHeader())
{
if (this.Runs[0].HighestHeuristicParameters == null)
{
throw new NotSupportedException("Cannot generate header because first run is missing highest solution parameters");
}
line.Append("stand,heuristic," + this.Runs[0].HighestHeuristicParameters !.GetCsvHeader() + ",thin age,rotation,generation,highest min,highest mean,highest max,highest cov,highest alleles,highest heterozygosity,highest individuals,highest polymorphism,lowest min,lowest mean,lowest max,lowest cov,lowest alleles,lowest heterozygosity,lowest individuals,lowest polymorphism");
writer.WriteLine(line);
}
for (int runIndex = 0; runIndex < this.Runs.Count; ++runIndex)
{
HeuristicSolutionDistribution distribution = this.Runs[runIndex];
if ((distribution.HighestSolution == null) || (distribution.LowestSolution == null) || (distribution.HighestHeuristicParameters == null))
{
throw new NotSupportedException("Run " + runIndex + " is missing a highest solution, lowest solution, or highest solution parameters");
}
GeneticAlgorithm highestHeuristic = (GeneticAlgorithm)distribution.HighestSolution;
GeneticAlgorithm lowestHeuristic = (GeneticAlgorithm)distribution.LowestSolution;
StandTrajectory highestTrajectory = highestHeuristic.BestTrajectory;
string linePrefix = highestTrajectory.Name + "," + highestHeuristic.GetName() + "," + distribution.HighestHeuristicParameters.GetCsvValues() + "," + highestTrajectory.GetFirstHarvestAge() + "," + highestTrajectory.GetRotationLength();
PopulationStatistics highestStatistics = highestHeuristic.PopulationStatistics;
PopulationStatistics lowestStatistics = lowestHeuristic.PopulationStatistics;
int maxGenerations = Math.Max(highestStatistics.Generations, lowestStatistics.Generations);
for (int generationIndex = 0; generationIndex < maxGenerations; ++generationIndex)
{
line.Clear();
string?highestMinimumFitness = null;
string?highestMeanFitness = null;
string?highestMaximumFitness = null;
string?highestCoefficientOfVariance = null;
string?highestMeanAlleles = null;
string?highestMeanHeterozygosity = null;
string?highestNewIndividuals = null;
string?highestPolymorphism = null;
if (highestStatistics.Generations > generationIndex)
{
highestMinimumFitness = highestStatistics.MinimumFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestMeanFitness = highestStatistics.MeanFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestMaximumFitness = highestStatistics.MaximumFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestCoefficientOfVariance = highestStatistics.CoefficientOfVarianceByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestMeanAlleles = highestStatistics.MeanAllelesPerLocusByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestMeanHeterozygosity = highestStatistics.MeanHeterozygosityByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestNewIndividuals = highestStatistics.NewIndividualsByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
highestPolymorphism = highestStatistics.PolymorphismByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
}
string?lowestMinimumFitness = null;
string?lowestMeanFitness = null;
string?lowestMaximumFitness = null;
string?lowestCoefficientOfVariance = null;
string?lowestMeanAlleles = null;
string?lowestMeanHeterozygosity = null;
string?lowestNewIndividuals = null;
string?lowestPolymorphism = null;
if (lowestStatistics.Generations > generationIndex)
{
lowestMinimumFitness = lowestStatistics.MinimumFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestMeanFitness = lowestStatistics.MeanFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestMaximumFitness = lowestStatistics.MaximumFitnessByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestCoefficientOfVariance = lowestStatistics.CoefficientOfVarianceByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestMeanAlleles = lowestStatistics.MeanAllelesPerLocusByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestMeanHeterozygosity = lowestStatistics.MeanHeterozygosityByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestNewIndividuals = lowestStatistics.NewIndividualsByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
lowestPolymorphism = lowestStatistics.PolymorphismByGeneration[generationIndex].ToString(CultureInfo.InvariantCulture);
}
line.Append(linePrefix + "," +
generationIndex + "," +
highestMinimumFitness + "," +
highestMeanFitness + "," +
highestMaximumFitness + "," +
highestCoefficientOfVariance + "," +
highestMeanAlleles + "," +
highestMeanHeterozygosity + "," +
highestNewIndividuals + "," +
highestPolymorphism + "," +
lowestMinimumFitness + "," +
lowestMeanFitness + "," +
lowestMaximumFitness + "," +
lowestCoefficientOfVariance + "," +
lowestMeanAlleles + "," +
lowestMeanHeterozygosity + "," +
lowestNewIndividuals + "," +
lowestPolymorphism);
writer.WriteLine(line);
}
}
}
public static void Run()
{
var config = new EAConfiguration
{
PopulationSize = 10,
OverproductionFactor = 1.0,
MaximumGenerations = 100000,
CrossoverType = CrossoverType.OnePoint,
AdultSelectionType = AdultSelectionType.GenerationalReplacement,
ParentSelectionType = ParentSelectionType.Tournament,
CrossoverRate = 0.18,
MutationRate = 0.99,
TournamentSize = 19,
TournamentProbability = 0.77,
TargetFitness = 0.0,
Mode = EAMode.MinimizeFitness,
Elites = 1,
CalculateStatistics = true,
SnapshotFilename = "snapshot2.bin",
SnapshotGenerationInterval = 100
};
var hammingEA = new HammingEA(config, new DefaultRandomNumberGenerator());
var stopwatchtot = new Stopwatch();
var stopwatchgen = new Stopwatch();
PopulationStatistics currentStats = new PopulationStatistics();
hammingEA.PopulationStatisticsCalculated += (stats) =>
{
currentStats = stats;
};
hammingEA.NewBestFitnessEvent += (pheno, gen) => {
double progress = (gen / (double)config.MaximumGenerations) * 100.0;
var totruntime = stopwatchtot.Elapsed;
var genruntime = stopwatchgen.Elapsed;
Console.WriteLine();
Console.WriteLine(string.Format("G# {0} best_f: {1:F2} avg_f: {2:F2} SD: {3:F2} Progress: {4,5:F2} Gen: {5} Tot: {6}", gen, hammingEA.Best.Fitness, currentStats.AverageFitness, currentStats.StandardDeviationFitness, progress, genruntime, totruntime));
Console.WriteLine("Generation winner: " + ((StringGenotype)hammingEA.Best?.Genotype));
stopwatchgen.Restart();
};
hammingEA.TerminationEvent += (r) =>
{
if (r == TerminationReason.FitnessLimitReached)
{
Console.WriteLine($"Fitness limit reached: {hammingEA.Best.Fitness}");
}
};
stopwatchtot.Start();
stopwatchgen.Start();
var res = hammingEA.Evolve();
Console.WriteLine("\n\nDone!");
Console.WriteLine($"Winner: {res.Winner}");
WriteResultToFile(hammingEA.PopulationStatistics);
Console.Read();
}
