public Gene crossover(Gene p_crossover_object)
{
Gene new_gene = new Gene(m_genes.Length, new Range <float>(0, 0), m_mutation_iterations, m_mutation_range);
for (int i = 0; i < new_gene.m_genes.Length; i++)
{
new_gene.m_genes[i] = BoolCalc.random() ? m_genes[i] : p_crossover_object.m_genes[i];
}
return(new_gene);
}
public override NeuralNetDNA <T> mutate()
{
NeuralNetDNA <T> mutated = Clone();
for (int i = 0; i < m_weights.Length; i++)
{
bool test = BoolCalc.random();
Range <float> mut = test ? m_mutation_multiplier : new Range <float>(-m_mutation_multiplier.Min, -m_mutation_multiplier.Max);
mutated.m_weights[i] = MatrixCalc.elementwiseRandomMultiply(m_weights[i], mut);
}
return(mutated);
}
public DecisionTreeDNANode crossover(DecisionTreeDNANode p_crossover_object)
{
int cross_input_index = BoolCalc.random() ? m_input_index : p_crossover_object.m_input_index;
float cross_min_input = BoolCalc.random() ? m_min_input : p_crossover_object.m_min_input;
float cross_max_input = BoolCalc.random() ? m_max_input : p_crossover_object.m_max_input;
if (cross_max_input < cross_min_input)
{
cross_max_input = cross_min_input;
}
float[] cross_output_values = ArrayCalc.crossover(m_output_values, p_crossover_object.m_output_values);
return(new DecisionTreeDNANode(cross_input_index, cross_min_input, cross_max_input, cross_output_values));
}
private void recRandom(DecisionTreeDNANode p_current_node, int p_current_depth, int p_max_depth, Range <float> p_threshold_range, Range <float> p_outputs_range)
{
if (p_current_depth == p_max_depth)
{
return;
}
if (BoolCalc.random())
{
p_current_node.addChild(BinaryDirection.LEFT, new DecisionTreeDNANode(m_inputs.Length, p_threshold_range, m_outputs.Length, p_outputs_range), p_current_node);
recRandom(p_current_node.LeftChild, p_current_depth + 1, p_max_depth, p_threshold_range, p_outputs_range);
}
if (BoolCalc.random())
{
p_current_node.addChild(BinaryDirection.RIGHT, new DecisionTreeDNANode(m_inputs.Length, p_threshold_range, m_outputs.Length, p_outputs_range), p_current_node);
recRandom(p_current_node.RightChild, p_current_depth + 1, p_max_depth, p_threshold_range, p_outputs_range);
}
}
private DecisionTreeDNANode recCrossover(DecisionTreeDNANode p_node1, DecisionTreeDNANode p_node2)
{
DecisionTreeDNANode crossedover = p_node1.crossover(p_node2);
if (p_node1.hasLeft && p_node2.hasLeft)
{
crossedover.addChild(BinaryDirection.LEFT, recCrossover(p_node1.LeftChild, p_node2.LeftChild), crossedover);
}
else if (!p_node1.hasLeft && !p_node2.hasLeft)
{
}
else
{
DecisionTreeDNANode only_existing = p_node1.hasLeft ? p_node1.LeftChild : p_node2.LeftChild;
if (BoolCalc.random())
{
crossedover.addChild(BinaryDirection.LEFT, recClone(only_existing), crossedover);
}
}
if (p_node1.hasRight && p_node2.hasRight)
{
crossedover.addChild(BinaryDirection.RIGHT, recCrossover(p_node1.RightChild, p_node2.RightChild), crossedover);
}
else if (!p_node1.hasRight && !p_node2.hasRight)
{
}
else
{
DecisionTreeDNANode only_existing = p_node1.hasRight ? p_node1.RightChild : p_node2.RightChild;
if (BoolCalc.random())
{
crossedover.addChild(BinaryDirection.RIGHT, recClone(only_existing), crossedover);
}
}
return(crossedover);
}
public void add(string p_type, float p_priority, DPriorityDelegate p_callback)
{
if (!m_list.ContainsKey(p_type))
{
m_list.Add(p_type, new PriorityElement(p_priority, p_callback));
return;
}
if (m_list[p_type].Priority > p_priority)
{
return;
}
if (m_list[p_type].Priority == p_priority)
{
if (!BoolCalc.random())
{
return;
}
}
m_list.Remove(p_type);
m_list.Add(p_type, new PriorityElement(p_priority, p_callback));
}
