import java.util.Random;
import java.util.Arrays;

/** Driver.java - a simple example illustrating the mechanics of genetic algorithms
 *  We seek to create an "optimal" list of 10 numbers.
 *  But only the fitness function knows what constraint we are aiming for!
 *
 *  The program will create 20 individuals.
 *  We also need to set the number of iterations (next generations).
 */

public class Driver
{
  public static final int ITERS = 300;
  public static final double MUTATION_THRESHOLD = 0.001;

  public static void main(String [] args)
  {
    // Initialize the population.
    Population p = new Population();
    
    // A loop for the generations.
    for (int i = 0; i < ITERS; ++i)
    {
      // Calculate the fitness of each individual.
      for (int j = 0; j < Population.SIZE; ++j)
        p.individual[j].findFitness();
      
      // Output
      System.out.printf("Iteration %d, Population = \n%s\n", i, p.toString());
      
      // Ready to create a new generation.
      // Sort the population by fitness.
      // p is the population object, whose "individual" attribute is the array.
      Arrays.sort(p.individual, new FitnessComparator());
      
      // For i = 1 to n, select 2 parents to mate based on high fitness.
      // Choose them randomly from the upper 1/2 of the population.
      Random gen = new Random(System.nanoTime());
      
      Individual [] newGeneration = new Individual[Population.SIZE];
      
      for (int j = 0; j < Population.SIZE; ++j)
      {
        int parent1 = gen.nextInt(Population.SIZE / 2);
        int parent2 = gen.nextInt(Population.SIZE / 2);
        
        // Create a child individual by randomly selecting an allele from
        // either parent.
        
        Individual child = new Individual();
        for (int gene = 0; gene < Individual.SIZE; ++gene)
        {
          // Flip a coin, a number either 1 or 2.
          int coin = gen.nextInt(2) + 1;
          
          // Based on the coin flip, set child's allele to the parent's.
          // Be careful with the syntax.
          if (coin == 1)
            child.a[gene] = p.individual[parent1].a[gene];
          else
            child.a[gene] = p.individual[parent2].a[gene];
        }
        
        // Allow for a mutation in the child.
        double prob = gen.nextDouble();
        if (prob < MUTATION_THRESHOLD)
        {
          // Arbitrarily change the first gene to a random allele value.
          child.a[0] = Individual.randomAlleleValue();
        }
        
        // Put the child into the new generation.
        newGeneration[j] = child;
      }
      
      // We now have a new generation.  Replace the old one.
      for (int j = 0; j < Population.SIZE; ++j)
        p.individual[j] = newGeneration[j];
    }
  }
}