public object Clone()
{
FoodSource fs = new FoodSource();
if (this.solutionTree != null)
{
fs.solutionTree = (Node)this.solutionTree.Clone();
}
fs.rawFitness = this.rawFitness;
fs.trialCounter = this.trialCounter;
return(fs);
}
private void button1_Click(object sender, EventArgs e)
{
int limit = Convert.ToInt32(textBox2.Text);
int colonySize = Convert.ToInt32(textBox1.Text);
int maxNumberOfCycles = Convert.ToInt32(textBox3.Text);
int numberOfFoodSources = colonySize / 2;
int numberOfRuns = Convert.ToInt32(textBox4.Text);
int initMaxDepth = Convert.ToInt32(textBox5.Text);
int maxDepth = Convert.ToInt32(textBox6.Text);
int numberOfFitnessCases = Convert.ToInt32(textBox7.Text);
double lowerB = Convert.ToDouble(textBox8.Text);
double upperB = Convert.ToDouble(textBox9.Text);
double Pip = 0.9;
//functionAndTerminalSet will contain the items of both terminal and function sets.
functionAndTerminalSet = new string[terminalSet.Length + functionSet.Length];
for (int i = 0; i < terminalSet.Length; i++)
{
functionAndTerminalSet[i] = terminalSet[i];
}
int itemIndex = terminalSet.Length;
for (int i = 0; i < functionSet.Length; i++)
{
functionAndTerminalSet[itemIndex] = functionSet[i];
itemIndex++;
}
FoodSource[] runBests = new FoodSource[numberOfRuns];
for (int r = 0; r < numberOfRuns; r++)
{
//run the ABCP
runBests[r] = RunABCP(numberOfFoodSources, initMaxDepth, maxNumberOfCycles, maxDepth, limit, Pip, numberOfFitnessCases, lowerB, upperB);
}
double meanOfRawFitnesses = 0;
for (int r = 0; r < numberOfRuns; r++)
{
meanOfRawFitnesses += runBests[r].rawFitness;
}
meanOfRawFitnesses = meanOfRawFitnesses / numberOfRuns;
string results = "Mean of the raw fitness values: " + meanOfRawFitnesses.ToString() + "\r\n ********************";
for (int r = 0; r < numberOfRuns; r++)
{
WriteTheFunction(runBests[r].solutionTree);
results += "\r\n " + "Run " + r.ToString() + "\r\n Raw fitness value:" + String.Format("{0:0.000000}", runBests[r].rawFitness) + "\r\n" + functionString;
functionString = null;
}
richTextBox1.Text += results;
}
public static FoodSource FindBestSource(FoodSource[] population)
{
FoodSource best = population[0];
for (int k = 1; k < population.Length; k++)
{
if (best.rawFitness > population[k].rawFitness)
{
best = population[k];
}
}
return(best);
}
public static bool ControlDifferenceForScout(FoodSource[] population, FoodSource sourceOfScout)
{
int counterD = 0;
bool isDifferentFromWholePopulation = false;
for (int h = 0; h < population.Length; h++)
{
bool isDifferent = false;
ControlDifferenceOfTwoTrees(population[h].solutionTree, sourceOfScout.solutionTree, ref isDifferent);
if (isDifferent == true)
{
counterD++;
}
}
if (counterD == population.Length)
{
isDifferentFromWholePopulation = true;
}
return(isDifferentFromWholePopulation);
}
public FoodSource[] InitializePopulationWithRampedHaH(int numberOfFoodSources, int initMaxDepth)
{
FoodSource[] population = new FoodSource[numberOfFoodSources];
int numberOfTreePerDepthLevel = population.Length / (initMaxDepth - 1);
int minDepthLevel = 2;
for (int i = minDepthLevel; i < initMaxDepth + 1; i++)
{
for (int j = (i - 2) * numberOfTreePerDepthLevel; j < (i - 1) * numberOfTreePerDepthLevel; j++)
{
Node tree = new Node();
population[j] = new FoodSource();
if (j < (((i - 2) * numberOfTreePerDepthLevel) + numberOfTreePerDepthLevel / 2))
{
if (j != 0)
{
bool isDifferent = false;
while (isDifferent == false)
{
population[j].solutionTree = CreateTreeWithGrowMethod(tree, 0, i);
isDifferent = Methods.ControlDifference(population, j);
}
}
else
{
population[j].solutionTree = CreateTreeWithGrowMethod(tree, 0, i);
}
}
else
{
bool isDifferent = false;
while (isDifferent == false)
{
population[j].solutionTree = CreateTreeWithFullMethod(tree, 0, i);
isDifferent = Methods.ControlDifference(population, j);
}
}
}
}
return(population);
}
public FoodSource RunABCP(int numberOfFoodSources, int initMaxDepth, int maxCycleNumber, int maxDepth, int limit, double Pip, int numberOfFitnessCases, double lowerB, double upperB)
{
//Initialize the population
FoodSource[] population = InitializePopulationWithRampedHaH(numberOfFoodSources, initMaxDepth);
//Evaluate the solutions.
for (int i = 0; i < population.Length; i++)
{
population[i].CalculateRawFitness(rn, lowerB, upperB, numberOfFitnessCases);
}
//Find the best solution.
FoodSource best = (FoodSource)Methods.FindBestSource(population).Clone();
int c = 0;
while (c < maxCycleNumber)
{
//Employed bee phase
for (int i = 0; i < population.Length; i++)
{
FoodSource candidate = new FoodSource();
candidate.solutionTree = population[i].ProduceCandidate(population, i, rn, maxDepth, Pip);
candidate.CalculateRawFitness(rn, lowerB, upperB, numberOfFitnessCases);
if (candidate.rawFitness < population[i].rawFitness)
{
population[i] = candidate;
}
else
{
population[i].trialCounter++;
}
}
//Calculate probabilities for the onlookers
FoodSource populationBest = Methods.FindBestSource(population);
double qualityOfPopulationBest = 1 / (1 + populationBest.rawFitness);
double[] probabilies = new double[population.Length];
for (int pr = 0; pr < probabilies.Length; pr++)
{
double quality = 1 / (1 + population[pr].rawFitness);
probabilies[pr] = (((quality * 0.9) / qualityOfPopulationBest) + 0.1);
}
//Onlooker bee phase
int t = 0;
int m = 0;
while (t < population.Length)
{
if (rn.NextDouble() < probabilies[m])
{
t++;
FoodSource candidate = new FoodSource();
candidate.solutionTree = population[m].ProduceCandidate(population, m, rn, maxDepth, Pip);
candidate.CalculateRawFitness(rn, lowerB, upperB, numberOfFitnessCases);
if (candidate.rawFitness < population[m].rawFitness)
{
population[m] = candidate;
}
else
{
population[m].trialCounter++;
}
}
m++;
if (m == (population.Length))
{
m = 0;
}
}
//Update the best solution achieved so far.
FoodSource currentPopulationBest = Methods.FindBestSource(population);
if ((currentPopulationBest.rawFitness < best.rawFitness))
{
best = (FoodSource)currentPopulationBest.Clone();
}
//Scout bee phase
int k = Methods.FindMaxTrialCounterValuedSource(population);
if (population[k].trialCounter > limit)
{
FoodSource sourceOfScout = new FoodSource();
sourceOfScout.solutionTree = new Node();
bool isDifferent = false;
while (isDifferent == false)
{
sourceOfScout.solutionTree = CreateTreeWithGrowMethod(sourceOfScout.solutionTree, 0, maxDepth);
isDifferent = Methods.ControlDifferenceForScout(population, sourceOfScout);
}
sourceOfScout.CalculateRawFitness(rn, lowerB, upperB, numberOfFitnessCases);
population[k] = sourceOfScout;
}
c++;
}
return(best);
}