private GenomePair GetChromoRoulette()
{
//generate a random number between 0 & total fitness count
double Slice = AIUtils.Random(_totalFitnessForChromoRoulette);
//this will be set to the chosen chromosome
GenomePair TheChosenOne = null;
//go through the chromosones adding up the fitness so far
double FitnessSoFar = 0;
foreach (GenomePair genpair in _population)
{
FitnessSoFar += genpair.Genome.Fitness;
if (FitnessSoFar >= Slice)
{
TheChosenOne = genpair;
break;
}
}
return(TheChosenOne);
}
public List <GenomePair> Epoch(List <GenomePair> old_population)
{
_generation++;
_population = old_population;
_totalFitness = 0;
_bestFitness = 0;
WorstFitness = double.MaxValue;
_averageFitness = 0;
// this sort is requiered for further scale and fintess calculations!
_population.Sort(GenomePair.GenomePairComparison);
CalculateFitnesses();
ScaleFitnesses();
List <GenomePair> new_population = new List <GenomePair>();
if (AIParams.NumCopiesElite * AIParams.NumElite % 2 == 0)
{
GrabNBest(AIParams.NumElite, AIParams.NumCopiesElite, ref new_population);
}
while (new_population.Count < _populationSize)
{
GenomePair mum = GetChromoRoulette();
GenomePair dad = GetChromoRoulette();
if (mum == null)
{
mum = new_population[0];
}
if (dad == null)
{
dad = new_population[2];
}
List <double> baby1 = new List <double>();
List <double> baby2 = new List <double>();
try
{
if (AIParams.MultiPointCrossover)
{
CrossoverSplitPoints(mum.Genome.Weights, dad.Genome.Weights, ref baby1, ref baby2);
}
else
{
Crossover(mum.Genome.Weights, dad.Genome.Weights, ref baby1, ref baby2);
}
Mutate(baby1);
Mutate(baby2);
new_population.Add(new GenomePair("AIObjMut " + (++_populationCounter).ToString(), new Genome(baby1, 0)));
new_population.Add(new GenomePair("AIObjMut " + (++_populationCounter).ToString(), new Genome(baby2, 0)));
}
catch
{
new_population.Add(new_population[0]);
new_population.Add(new_population[2]);
}
}
_population = new_population;
return(_population);
}
