public static ArtificialChromosome[] GenerateInitialPopulation()
{
ArtificialChromosome[] initialPopulation = new ArtificialChromosome[Config.NumberOfParentsMiu];
var randomGenerator = new RandomDouble();
var braninRcos = new BraninRcos();
for (var i = 0; i < Config.NumberOfParentsMiu; i++)
{
var solution = new ArtificialChromosome();
do
{
solution.X1 = randomGenerator.GetRandomNumber(Config.MinX1, Config.MaxX1);
solution.X2 = randomGenerator.GetRandomNumber(Config.MinX2, Config.MaxX2);
} while (i != 0 &&
Math.Abs(solution.X1 - initialPopulation[i - 1].X1) < 0.000000000000001 &&
Math.Abs(solution.X2 - initialPopulation[i - 1].X2) < 0.000000000000001);
solution.MutationSigmaX1 = Config.InitialSigmaX1;
solution.MutationSigmaX2 = Config.InitialSigmaX2;
solution.ObjectiveFunctionResult = braninRcos.BraninRcosObjectiveFunction(solution.X1, solution.X2);
initialPopulation[i] = solution;
}
return(initialPopulation);
}
private static ArtificialChromosome[] CreateNextGeneration(ArtificialChromosome[] parentPopulation)
{
try
{
ArtificialChromosome[] children = new ArtificialChromosome[Config.NumberOfChildrenLambda];
var braninRcos = new BraninRcos();
var oneFifthTracker = 0;
for (int i = 0; i < children.Length; i++)
{
var child = new ArtificialChromosome();
var randomParentIndex = new Random().Next(parentPopulation.Length); // between 0 and length - 1
var randomParent = parentPopulation[randomParentIndex];
child.X1 = randomParent.X1 + (randomParent.MutationSigmaX1 * GetARandomNormarDistributionNumber(0, 1));
child.X2 = randomParent.X2 + (randomParent.MutationSigmaX2 * GetARandomNormarDistributionNumber(0, 1));
child.ObjectiveFunctionResult = braninRcos.BraninRcosObjectiveFunction(child.X1, child.X2);
child.MutationSigmaX1 = randomParent.MutationSigmaX1;
child.MutationSigmaX2 = randomParent.MutationSigmaX2;
if (randomParent.ObjectiveFunctionResult > child.ObjectiveFunctionResult)
{
oneFifthTracker++;
}
child.MutationSigmaX1 = Config.InitialSigmaX1;
child.MutationSigmaX2 = Config.InitialSigmaX2;
children[i] = child;
}
// adjust sigma
if (oneFifthTracker > Config.NumberOfChildrenLambda / 5)
{
// increase sigma for the next generation
Config.InitialSigmaX1 = Config.InitialSigmaX1 +
Config.OneOverFiveSigmaRuleConstant * Config.InitialSigmaX1;
Config.InitialSigmaX2 = Config.InitialSigmaX2 +
Config.OneOverFiveSigmaRuleConstant * Config.InitialSigmaX2;
}
else if (oneFifthTracker < Config.NumberOfChildrenLambda / 5)
{
// increase sigma for the next generation
Config.InitialSigmaX1 = Config.InitialSigmaX1 -
Config.OneOverFiveSigmaRuleConstant * Config.InitialSigmaX1;
Config.InitialSigmaX2 = Config.InitialSigmaX2 -
Config.OneOverFiveSigmaRuleConstant * Config.InitialSigmaX2;
}
ArtificialChromosome[] parents = TakeBestForNextGenParents(children);
return(parents);
}
catch (Exception e)
{
Console.WriteLine(e);
throw;
}
}
private static ArtificialChromosome GetBestSolution(ArtificialChromosome[] population)
{
ArtificialChromosome bestSolution = new ArtificialChromosome {
ObjectiveFunctionResult = Double.MaxValue
};
for (var i = 0; i < Config.NumberOfParentsMiu; i++)
{
bestSolution = population[i].ObjectiveFunctionResult < bestSolution.ObjectiveFunctionResult ? population[i] : bestSolution;
}
return(bestSolution);
}
public static void DisplayChromosome(ArtificialChromosome solution)
{
DisplayChromosome(solution.X1, solution.MutationSigmaX1, solution.X2, solution.MutationSigmaX2, solution.ObjectiveFunctionResult);
}
