//The method the sort the genomes.
#endregion
public void PopulationGeneration()
{
for (int i = 0; i < kInitialPopulation; i++)
{
ListGenome aGenome = new ListGenome(kMin, kMax, kParameterMin, kParameterMax, kParameterStep);
int kLength = kParameterMin.Length;
int crossoverRandomSeed = ListGenome.TheSeed.Next(kLength);
kCrossover = crossoverRandomSeed;
//kCrossover or crossover point was a random number among the length of the genome.
aGenome.SetCrossoverPoint(kCrossover);
aGenome.CalculateFitness(calibrationRoundID, dataDirectory, workDirectory, projectDirectory, genericAlgorithmsDirectory, observationArray, stdArray, kParameterSiteIndex, amplificationFactor, CV, observationDateArray, observationSeperator);
aGenome.OutputString(observationArray);
Genomes.Add(aGenome);
}
Genomes.Sort(myCompareMethod);
int kPopulationLimitOfInitialPopulation = 20 * kPopulationLimit;//The population limit of first generation is 10 times of the limits of other generations.
// kill all the genes above the population limit
if (Genomes.Count > kPopulationLimitOfInitialPopulation)
{
Genomes.RemoveRange(kPopulationLimitOfInitialPopulation, Genomes.Count - kPopulationLimitOfInitialPopulation);
}
CurrentPopulation = Genomes.Count;
}
//Declare ab arrry list to save the model predictions.
public override int CompareTo(object a)
{
ListGenome Gene1 = this;
ListGenome Gene2 = (ListGenome)a;
return(Math.Sign(Gene2.CurrentFitness - Gene1.CurrentFitness));
}
public override void CopyGeneInfo(Genome dest)
{
ListGenome theGene = (ListGenome)dest;
theGene.Length = Length;
theGene.TheMin = TheMin;
theGene.TheMax = TheMax;
}
public void LocalPopulationGeneration( )
{
int parameterNumber = optimalParameterValueArray.Count;
double bestFitness = 0;
int numberOfGenomes = 0;
double[] kLocalParameterValue = new double[parameterNumber];
double[] kLocalParameterMin = new double[parameterNumber];
double[] kLocalParameterMax = new double[parameterNumber];
double[] kLocalParameterStep = new double[parameterNumber];
double[] kLocalParameterStepLimit = new double[parameterNumber];
for (int i = 0; i < optimalParameterValueArray.Count; i++)
{
double tempLocalParameterValue = Convert.ToDouble(optimalParameterValueArray[i]);
kLocalParameterValue[i] = tempLocalParameterValue;
double tempInitialParameterMin = kParameterMin[i];
double tempInitialParameterMax = kParameterMax[i];
kLocalParameterMin[i] = Math.Max((tempLocalParameterValue * (1 - kLocalParameterRatio)), tempInitialParameterMin);
kLocalParameterMax[i] = Math.Min((tempLocalParameterValue * (1 + kLocalParameterRatio)), tempInitialParameterMax); //Restric the local seaching range of the parameters with climbing hill methods. Modified by He, 14/10/2010.
kLocalParameterStep[i] = kParameterStep[i] * (kLocalParameterRatio / kLocalParameterRatioPartition); //Define the local searching spaces.
}
for (int j = 0; j < kLocalPopulation; j++)
{
ListGenome aGenome = new ListGenome(kMin, kMax, kLocalParameterValue, kLocalParameterMin, kLocalParameterMax, kLocalParameterStep, kParameterChangeProbability);
aGenome.CalculateFitness(calibrationRoundID, dataDirectory, workDirectory, projectDirectory, genericAlgorithmsDirectory, observationArray, stdArray, kParameterSiteIndex, amplificationFactor, CV, observationDateArray, observationSeperator);
aGenome.OutputString(observationArray);
if (j == 0)
{
Genomes.Add(aGenome);
}
//Add the first genome.
numberOfGenomes = Genomes.Count;
double tempFitness = aGenome.CurrentFitness;
if (tempFitness > bestFitness)
{
Genomes.RemoveAt((numberOfGenomes - 1));//Remove the last element of Genomes.
Genomes.Add(aGenome);
bestFitness = tempFitness;
}
}
}
public override Genome Crossover(Genome g)
{
ListGenome aGene1 = new ListGenome();
ListGenome aGene2 = new ListGenome();
g.CopyGeneInfo(aGene1);
g.CopyGeneInfo(aGene2);
ListGenome CrossingGene = (ListGenome)g;
for (int i = 0; i < CrossoverPoint; i++)
{
aGene1.TheArray.Add(CrossingGene.TheArray[i]);
aGene2.TheArray.Add(TheArray[i]);
}
for (int j = CrossoverPoint; j < Length; j++)
{
aGene1.TheArray.Add(TheArray[j]);
aGene2.TheArray.Add(CrossingGene.TheArray[j]);
}
// 50/50 chance of returning gene1 or gene2
ListGenome aGene = null;
if (TheSeed.Next(2) == 1)
{
aGene = aGene1;
}
else
{
aGene = aGene2;
}
return(aGene);
}
public void DoCrossover(ArrayList genes)
{
ArrayList GeneMoms = new ArrayList();
GeneMoms.Clear();
ArrayList GeneDads = new ArrayList();
GeneDads.Clear();
//Console.WriteLine(genes.Count);
for (int i = 0; i < genes.Count; i++)
{
// randomly pick the moms and dad's
if (ListGenome.TheSeed.Next(100) % 2 > 0)
{
GeneMoms.Add(genes[i]);
}
else
{
GeneDads.Add(genes[i]);
}
}
// now make them equal
if (GeneMoms.Count > GeneDads.Count)
{
while (GeneMoms.Count > GeneDads.Count)
{
GeneDads.Add(GeneMoms[GeneMoms.Count - 1]);
GeneMoms.RemoveAt(GeneMoms.Count - 1);
}
if (GeneDads.Count > GeneMoms.Count)
{
GeneDads.RemoveAt(GeneDads.Count - 1); // make sure they are equal
}
}
else
{
while (GeneDads.Count > GeneMoms.Count)
{
GeneMoms.Add(GeneDads[GeneDads.Count - 1]);
GeneDads.RemoveAt(GeneDads.Count - 1);
}
if (GeneMoms.Count > GeneDads.Count)
{
GeneMoms.RemoveAt(GeneMoms.Count - 1); // make sure they are equal
}
}
// now cross them over and add them according to fitness
for (int i = 0; i < GeneDads.Count; i++)
{
// pick best 2 from parent and children
ListGenome babyGene1 = (ListGenome)((ListGenome)GeneDads[i]).Crossover((ListGenome)GeneMoms[i]);
ListGenome babyGene2 = (ListGenome)((ListGenome)GeneMoms[i]).Crossover((ListGenome)GeneDads[i]);
GenomeFamily.Clear();
GenomeFamily.Add(GeneDads[i]);
GenomeFamily.Add(GeneMoms[i]);
GenomeFamily.Add(babyGene1);
GenomeFamily.Add(babyGene2);
CalculateFitnessForAll(GenomeFamily);
GenomeFamily.Sort(myCompareMethod);
if (Best2 == true)
{
// if Best2 is true, add top fitness genes
GenomeResults.Add(GenomeFamily[0]);
GenomeResults.Add(GenomeFamily[1]);
}
else
{
GenomeResults.Add(babyGene1);
GenomeResults.Add(babyGene2);
}
}
}
