/// <summary>
/// Evolve the the population to the next generation.
/// </summary>
public void Evolve()
{
foreach (T chromosome in _population)
{
chromosome.FitnessValue = _fitnessFunction.Invoke(chromosome);
}
GetFittestChromsome();
IList <T> selection = _selection.ApplySelection(_population).Select(x => x.Clone()).ToArray();
foreach (IEvolveFunction function in _eveolveFunctions)
{
function.Execute(selection, this._population);
}
EnsureSurvival();
foreach (T chromosome in _population)
{
chromosome.OnEvolved();
}
this.CurrentGeneration++;
}
public OptimizationResult Optimize(InputDataEntity inputDataEntity)
{
InputDataEntity = inputDataEntity;
NumberOfBitsInChromosome = PopulationService.CalculateNumberOfBitsInChromosome(inputDataEntity.DataRange, inputDataEntity.Accuracy);
BinaryChromosomes = PopulationService.GeneratePopulation(inputDataEntity.PopulationAmount, NumberOfBitsInChromosome);
// Method selection
ISelection selectionMethod = GetSelectionMethod(InputDataEntity.SelectionMethod, inputDataEntity);
ICross crossMethod = GetCrossMethod(InputDataEntity.CrossMethod);
IMutation mutationMethod = GetMutationMethod(InputDataEntity.MutationMethod);
var bestOfEachEpoch = new List <FunctionResult>();
List <BinaryChromosome> tmpBest = new List <BinaryChromosome>();
List <double> means = new List <double>();
List <double> standardDeviation = new List <double>();
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
for (int i = 0; i < InputDataEntity.EpochsAmount; i++)
{
List <BinaryChromosome> newPopulation = selectionMethod.ApplySelection(BinaryChromosomes, inputDataEntity.DataRange, NumberOfBitsInChromosome);
if (i != 0)
{
double newValue = GetFunctionResultForChromosomes(newPopulation[0], newPopulation[1], inputDataEntity.DataRange).FunctionValue;
double oldValue = GetFunctionResultForChromosomes(tmpBest[0], tmpBest[1], inputDataEntity.DataRange).FunctionValue;
if ((inputDataEntity.Maximize == false && newValue > oldValue) || (inputDataEntity.Maximize == true && newValue < oldValue))
{
newPopulation.InsertRange(0, tmpBest);
}
else
{
newPopulation.InsertRange(newPopulation.Count - 1, tmpBest);
}
}
bestOfEachEpoch.Add(GetFunctionResultForChromosomes(newPopulation[0], newPopulation[1], inputDataEntity.DataRange));
tmpBest.Clear();
for (int j = 0; j < inputDataEntity.EliteStrategyAmount * 2; j++)
{
tmpBest.Add(newPopulation[j]);
}
means.Add(CalculateMean(newPopulation, inputDataEntity.DataRange));
standardDeviation.Add(CalculateDeviation(newPopulation, inputDataEntity.DataRange, means[i]));
List <BinaryChromosome> crossedPopulation = crossMethod.Cross(newPopulation, InputDataEntity.CrossProbability);
List <BinaryChromosome> mutatedPopulation = mutationMethod.Mutate(crossedPopulation, InputDataEntity.MutationProbability);
List <BinaryChromosome> populationAfterInversion = Inversion.PerformInversion(mutatedPopulation, InputDataEntity.InversionProbability);
BinaryChromosomes = populationAfterInversion;
if (BinaryChromosomes.Count <= 2)
{
break;
}
}
var bestIndividuals = selectionMethod.ApplySelection(BinaryChromosomes, inputDataEntity.DataRange, NumberOfBitsInChromosome).Take(2).ToArray();
double x1 = BinaryUtils.BinaryToDecimalRepresentation(bestIndividuals[0].BinaryRepresentation, inputDataEntity.DataRange, NumberOfBitsInChromosome);
double x2 = BinaryUtils.BinaryToDecimalRepresentation(bestIndividuals[1].BinaryRepresentation, inputDataEntity.DataRange, NumberOfBitsInChromosome);
double extremum = Function.Compute(x1, x2);
stopwatch.Stop();
return(new OptimizationResult
{
ExtremeValue = extremum,
X1 = x1,
X2 = x2,
BestFromPreviousEpochs = bestOfEachEpoch,
MeanFromPreviousEpochs = means,
StandardDeviation = standardDeviation,
TimeElapsed = stopwatch.ElapsedMilliseconds
});
}