private void CrossOverParents(CandidateSolution parent1, CandidateSolution parent2, out CandidateSolution child1, out CandidateSolution child2)
{
child1 = parent1.DeepClone();
child2 = parent2.DeepClone();
//use exact copies or do crossover
if (Randomizer.GetDoubleFromZeroToOne() < crossoverRate)
{
int numItems = parent1.Meals.Count;
int crossoverPoint = Randomizer.IntLessThan(numItems);
for (int i = 0; i < crossoverPoint; i++)
{
child1.SetSelected(i, parent1.IsSelected(i));
child2.SetSelected(i, parent2.IsSelected(i));
}
for (int i = crossoverPoint; i < numItems; i++)
{
child1.SetSelected(i, parent2.IsSelected(i));
child2.SetSelected(i, parent1.IsSelected(i));
}
}
}
public MealCollection FindOptimalItems(MealCollection meals, NutritionConstraints constraints)
{
currentGeneration = new List <CandidateSolution>(populationSize);
for (int i = 0; i < populationSize; i++)
{
currentGeneration.Add(new CandidateSolution(meals, constraints));
}
generationNumber = 1;
//main loop
while (true)
{
float bestFitnessScoreThisGeneration = System.Int32.MaxValue;
CandidateSolution bestSolutionThisGeneration = null;
foreach (var candidate in currentGeneration)
{
candidate.Repair();
float fitness = candidate.Fitness;
//sum up fitness scores for the roulette wheel selection
totalFitnessThisGeneration += fitness;
totalInverseFitnessThisGeneration += 1 / (double)fitness;
if (fitness < bestFitnessScoreThisGeneration)
{
bestFitnessScoreThisGeneration = fitness;
bestSolutionThisGeneration = candidate;
}
}
Debug.WriteLine("Iteration count {0}, best fitness: {1}", generationNumber, bestFitnessScoreAllTime);
//compare this generation's best to to the best we had so far
if (bestFitnessScoreThisGeneration < bestFitnessScoreAllTime)
{
//save the best score
bestFitnessScoreAllTime = bestFitnessScoreThisGeneration;
//and save possible solution
bestSolution = bestSolutionThisGeneration.DeepClone();
bestSolutionGenerationNumber = generationNumber;
}
else
{
if ((generationNumber - bestSolutionGenerationNumber) > MaxGenerationsWithNoChange)
{
break;
}
}
List <CandidateSolution> nextGeneration = new List <CandidateSolution>();
while (nextGeneration.Count < populationSize)
{
//Select two parents(the lower the fitness, the higher the chance of selection)
var parent1 = SelectCandidate();
var parent2 = SelectCandidate();
//cross them over to generate two new children
CandidateSolution child1, child2;
CrossOverParents(parent1, parent2, out child1, out child2);
//appy mutation if needed
child1.AddPossibleMutation(mutationRate);
child2.AddPossibleMutation(mutationRate);
//add them to next generation
nextGeneration.Add(child1);
nextGeneration.Add(child2);
}
currentGeneration = nextGeneration;
generationNumber++;
}
return(bestSolution.GetSelectedItems());
}
internal CandidateSolution DeepClone()
{
var clone = new CandidateSolution(Meals, NutritionConstraints, isSelected);
return(clone);
}
