/// <summary>
/// This is used to crossover two parents to make a child
/// </summary>
/// <param name="parentA"></param>
/// <param name="parentB"></param>
private Genome(EvolutionConfig config, Genome parentA, Genome parentB)
: this(config, false)
{
Genome moreFitParent = parentA.fitness > parentB.fitness ? parentA : parentB;
Genome lessFitParent = parentA.fitness > parentB.fitness ? parentB : parentA;
// Take all similar genes from the most fit parent, and randomly select disjoint genes
IEnumerable<Gene> childGenes = moreFitParent.Genes.Where(item => lessFitParent.Genes.Exists(item2 => item2.InnovationNumber == item.InnovationNumber) || config.rng.NextDouble() > 0.50)
.Union(lessFitParent.Genes.Where(item => !moreFitParent.Genes.Exists(item2 => item2.InnovationNumber == item.InnovationNumber) && config.rng.NextDouble() > 0.50));
// Set mutation rates to the most fit parent
mutationRates = moreFitParent.mutationRates;
// Create unique copies of genes
foreach (Gene g in childGenes)
Genes.Add((Gene)g.Clone());
// Copy parent A's start values
startValues = new Dictionary<int,double>(moreFitParent.startValues);
// Throw in parent B's extra start values
foreach (var pair in lessFitParent.startValues) {
if (!startValues.ContainsKey(pair.Key))
startValues.Add(pair.Key, pair.Value);
}
}
/// <summary>
/// Main constructor for evolution. Requires a configuration instance
/// </summary>
/// <param name="config"></param>
public Evolution(EvolutionConfig config)
{
// Setup new lists
genomes = new List<Genome>();
species = new List<Species>();
// Reset basic variables
evaluationIndex = 0;
speciesFitnessSum = 0;
generationNumber = 0;
topScore = 0;
// Set the configuration
evoConfig = config;
// Create new random genomes
for (int i = 0; i < config.PopulationSize; i++)
genomes.Add(new Genome(config));
// Mutate each genenome, add add them to a species
foreach (Genome g in genomes) {
g.Mutate();
AddGenomeToSpecies(g);
}
// Setup netwrok
evaluationNetwork = new NeuralNetwork(genomes[0]);
}
/// <summary>
/// Call this function to set the default gene values (still needs an inputNode and outputNode)
/// </summary>
/// <param name="input"></param>
/// <param name="output"></param>
public Gene(EvolutionConfig config, int input, int output)
{
// Defaults
NodeInput = input;
NodeOutput = output;
ConnectionEnabled = true;
ConnectionWeight = 1.0;
InnovationNumber = Innovation.NewInnovationNumber;
}
/// <summary>
/// This breeds to parents to make a new child from a species offspring has a chance of being a clone of the parent
/// </summary>
/// <param name="config"></param>
/// <param name="species"></param>
/// <returns></returns>
public static Genome BreedChild(EvolutionConfig config, Species species)
{
Genome child;
// If the specie only contains one child, the species can't crossover
if (species.genomes.Count == 1)
child = (Genome) species.genomes.First().Clone();
else {
Genome parentA = species.genomes.ElementAt(rng.Next(species.genomes.Count));
// Non-crossover chance
if (rng.NextDouble() >= config.CrossOverRate)
child = (Genome) parentA.Clone();
else { // Crossover
Genome parentB = species.genomes.Where(g => g != parentA).ElementAt(rng.Next(species.genomes.Count - 1));
child = new Genome(config, parentA, parentB);
}
}
// Mutate
child.Mutate();
return child;
}
public static bool IsOutputNode(EvolutionConfig config, int n)
{
return (n >= config.InputNodeCount + 1 && IsInputOutputNode(config, n));
}
public static bool IsInputOutputNode(EvolutionConfig config, int n)
{
return (n < config.InputNodeCount + 1 + config.OutputNodeCount);
}
