/// <summary>
/// Evaluates the fitness.
/// </summary>
private void EvaluateFitness(CCESpecies spec)
{
// For each individual we run fitness with a set
// of best chromosomes, and a random set
//
// BestChromosomeList can be used
// Get random from spec.Population.CurrentGeneration.Chromosomes
// -- also verify that it is not the same as the best
// --- just take any but the first in list (it is ordered)
//
// Use fitness function to get fitness for both sets
// Fitness given is best of the two
try
{
CurrentSpec = spec;
var coll = spec.Offspring;
if (coll == null)
{
coll = spec.Population.CurrentGeneration.Chromosomes;
}
foreach (var individual in coll)
{
if (individual.Fitness.HasValue)
{
continue;
}
List <IChromosome> RandomList = new List <IChromosome>();
for (int o = 0; o < Species.Count; o++)
{
if (Object.ReferenceEquals(Species[o], spec))
{
// this is to keep it in the same order as the species list, always
RandomList.Add(individual);
continue;
}
int randIndex;
if (Species[o].Population.CurrentGeneration.Chromosomes.Count > 1)
{
randIndex = RandomizationProvider.Current.GetInt(0, Species[o].Population.CurrentGeneration.Chromosomes.Count - 1);
}
else
{
randIndex = 0;
}
RandomList.Add(Species[o].Population.CurrentGeneration.Chromosomes[randIndex]);
}
TaskExecutorFit.Add(() =>
{
RunEvaluateFitness(individual, RandomList);
});
}
if (!TaskExecutorFit.Start())
{
throw new TimeoutException("The fitness evaluation rech the {0} timeout.".With(TaskExecutorFit.Timeout));
}
}
finally
{
TaskExecutorFit.Stop();
TaskExecutorFit.Clear();
}
}
/// <summary>
/// Ends the current generation.
/// </summary>
/// <returns><c>true</c>, if current generation was ended, <c>false</c> otherwise.</returns>
private bool EndCurrentGeneration()
{
if (_debugging)
{
Console.WriteLine("Starting fitness evaluation");
timer.Reset();
timer.Start();
}
// Insert multithreading here?
try
{
foreach (var spec in Species)
{
TaskExecutorFitMaster.Add(() => {
EvaluateFitness(spec);
});
}
if (!TaskExecutorFitMaster.Start())
{
throw new TimeoutException("The fitness evaluation rech the {0} timeout.".With(TaskExecutorFitMaster.Timeout));
}
}
finally
{
TaskExecutorFitMaster.Stop();
TaskExecutorFitMaster.Clear();
}
if (_debugging)
{
Console.WriteLine("Evolved a generation. About to calculate fitness and end it. Time at {0}", timer.Elapsed.ToString());
timer.Stop();
}
if (_debugging)
{
Console.WriteLine("Calculating Fitness done. Timer at {0}", timer.Elapsed);
timer.Stop();
timer.Reset();
Console.WriteLine("Starting reinsertion");
timer.Start();
}
foreach (var spec in Species)
{
spec.EndCurrentGeneration();
}
if (_debugging)
{
Console.WriteLine("Reinsertion done. Timer at {0}", timer.Elapsed);
timer.Stop();
}
GenerationsNumber++;
//update uber best set
if (UberBestSet == null || UberBestSet.Count == 0)
{
UberBestSet = new List <IChromosome> ();
foreach (IChromosome c in BestChromosomeSet)
{
UberBestSet.Add(c.Clone());
}
}
else
{
for (int i = 0; i < BestChromosomeSet.Count; i++)
{
if (UberBestSet [i].Fitness < BestChromosomeSet [i].Fitness)
{
UberBestSet [i] = BestChromosomeSet [i].Clone();
}
}
}
if (GenerationRan != null)
{
GenerationRan(this, EventArgs.Empty);
}
if (Termination.HasReached(this))
{
State = GeneticAlgorithmState.TerminationReached;
if (TerminationReached != null)
{
TerminationReached(this, EventArgs.Empty);
}
return(true);
}
if (m_stopRequested)
{
TaskExecutorGen.Stop();
TaskExecutorFit.Stop();
State = GeneticAlgorithmState.Stopped;
}
return(false);
}