public CandidateSolution(MealCollection meals, NutritionConstraints nutritionConstraints, List <bool> isSelected)
{
this.isSelected = new List <bool>(isSelected);
Meals = meals;
NutritionConstraints = nutritionConstraints;
CalcFitness();
}
static void Main(string[] args)
{
Console.Write("Initialising engine...");
GeneticAlgorithEngine ge = new GeneticAlgorithEngine(INITIALPOPULATIONSIZE, CROSSOVERPERCENTAGE, MUTATIONPERCENTAGE);
Console.Write("\nReading data...");
DataReader dataReader = new DataReader();
MealCollection mealCollection = dataReader.GetMealCollectionFromFile();
NutritionConstraints constraints = new NutritionConstraints(PROTEIN, CARBS, FAT, CALORIES, VITAMINA, VITAMINC, CALCIUM, IRON);
Console.Write("\nRunning optimalisation...");
MealCollection solution = ge.FindOptimalItems(mealCollection, constraints);
WriteSolutionToConsole(solution);
}
//generation 0, creates random solution
public CandidateSolution(MealCollection mealCollection, NutritionConstraints nutritionConstraints)
{
NutritionConstraints = nutritionConstraints;
Meals = mealCollection;
var numMeals = Meals.Count;
isSelected = new List <bool>(numMeals);
//randomly initilase membership flag
for (int i = 0; i < numMeals; i++)
{
isSelected.Add(Randomizer.GetDoubleFromZeroToOne() >= 0.5);
}
CalcFitness();
}
private static void WriteSolutionToConsole(MealCollection solution)
{
Meal totalMeal = new Meal();
Console.WriteLine("\nYour meals:");
foreach (var item in solution.ToList())
{
WriteMeal(item);
totalMeal.ServingSize += item.ServingSize;
totalMeal.Calories += item.Calories;
totalMeal.Carbohidrate += item.Carbohidrate;
totalMeal.Fat += item.Fat;
totalMeal.Protein += item.Protein;
totalMeal.Sodium += item.Sodium;
totalMeal.VitaminA += item.VitaminA;
totalMeal.VitaminC += item.VitaminC;
}
Console.WriteLine("============================");
Console.WriteLine("Your total daily consumption:");
WriteMeal(totalMeal);
Console.ReadLine();
}
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());
}
