public Chromossome SelectIndividual(Chromossome p_differentFromThis)
{
float __totalFitness = 0;
float __selectedFitness = 0;
int __count = 0;
Chromossome __selectedChromossome = p_differentFromThis;
for (int i = 0; i < populationSize; i++)
{
__totalFitness += population[i].fitness;
}
while (__selectedChromossome == p_differentFromThis && __count < 10)
{
__selectedFitness = Random.value * __totalFitness;
for (int i = 0; i < populationSize; i++)
{
if (population[i].fitness >= __selectedFitness)
{
__selectedChromossome = population[i];
break;
}
else
{
__selectedFitness -= population[i].fitness;
}
}
__count++;
}
return(__selectedChromossome);
}
public void Crossover(Chromossome c1, Chromossome c2)
{
for (int i = 0; i < m_Length; i++)
{
m_Genes[i] = Random.Range(0.0f, 1.0f) < 0.5f ? c1.m_Genes[i] : c2.m_Genes[i];
}
}
public Chromossome SelectIndividual()
{
float __totalFitness = 0;
float __selectedFitness = 0;
Chromossome __selectedChromossome = new Chromossome(mapSize, mapSize);
for (int i = 0; i < populationSize; i++)
{
__totalFitness += population[i].fitness;
}
__selectedFitness = Random.value * __totalFitness;
for (int i = 0; i < populationSize; i++)
{
if (population[i].fitness >= __selectedFitness)
{
__selectedChromossome = population[i];
break;
}
else
{
__selectedFitness -= population[i].fitness;
}
}
return(__selectedChromossome);
}
private void NewPopulation()
{
Evaluate();
List <Brain> newPopulation = new List <Brain>();
for (int i = 0; i < m_PopulationSize; i++)
{
Chromossome parent1 = TournamentSelection().m_Chromosome;
Chromossome parent2 = TournamentSelection().m_Chromosome;
GameObject offspring = Instantiate(m_Prefab, m_Begin.position, m_Begin.rotation);
Brain brain = offspring.GetComponent <Brain>();
brain.Initialize();
brain.m_Chromosome.Crossover(parent1, parent2);
if (Random.Range(0.0f, 1.0f) < m_MutationRate)
{
brain.m_Chromosome.Mutate();
}
newPopulation.Add(brain);
}
for (int i = 0; i < m_PopulationSize; i++)
{
Destroy(m_Population[i].gameObject);
}
m_Population = newPopulation;
m_Generation++;
}
public void Initialize()
{
// Gene 0 - Forward
// Gene 1 - Angle
m_Chromosome = new Chromossome(2, 90);
m_StartPosition = transform.position;
m_Alive = true;
}
public void EvaluateFitness()
{
//DebugLogPopulation();
//testPop = new List<Chromossome>();
/**
* for (int i = 0; i < population.Count; i++)
* {
* testPop.Add(population[i]);
* }
* /**/
for (int i = 0; i < population.Count; i++)
{
float __tempFitness = 0;
Chromossome __tempChromossome = population[i];
int[][] __simulatedMap = CellularAutomata.SimulateMap(__tempChromossome.GetMap(), 5);
__simulatedMap = CellularAutomata.FillExclaves(__simulatedMap); //by doing this we end up having only the largest cave area
//first we calculate the fitness from Fill Ratio
float __fill = CellularAutomata.GetFilledPercentage(__simulatedMap);
__tempFitness += (1 - (Mathf.Abs(fillTarget - __fill))) * fitnessBaseValue;
//now we calculate the fitness based on points
for (int j = 0; j < fitnessPoints.Length; j++)
{
if (__simulatedMap[(int)fitnessPoints[j].y][(int)fitnessPoints[j].x] == 0)
{
__tempFitness += fitnessBaseValue * fitnessPointsWeight;
}
}
__tempChromossome.fitness = __tempFitness;
population[i] = __tempChromossome;
}
/**
* for (int i = 0; i < population.Count; i++)
* {
* Debug.Log(testPop[i] == population[i]);
* }
* /**/
//DebugLogPopulation();
}
public void ChangeGeneration()
{
//DebugLogPopulation();
List <Chromossome> __newPopulation = new List <Chromossome>();
for (int i = 0; i < eliteSize; i++)
{
__newPopulation.Add(population[i]);
//Debug.Log("Generation: " + generation + " Checking Elite " + i + ": " + __newPopulation[i].fitness);
}
for (int i = eliteSize; i < populationSize; i += 2)
{
Chromossome __parent1 = SelectIndividual();
Chromossome __parent2 = SelectIndividual(__parent1);
//Debug.Log("Are parents equal? " + (__parent1 == __parent2));
Chromossome[] __tempChromossome = Crossover(__parent1, __parent2);
__newPopulation.Add(__tempChromossome[0]);
if (i < populationSize - 1)
{
__newPopulation.Add(__tempChromossome[1]);
}
}
/**
* for (int i = 0; i < population.Count ; i++)
* {
* Debug.Log("Are populations equal? " + (population[i] == __newPopulation[i]));
* }
* /**/
//Debug.Log(population.Count + " " + __newPopulation.Count);
population = __newPopulation;
generation++;
//DebugLogPopulation();
}
public void Initialize()
{
population = new List <Chromossome>();
for (int i = 0; i < populationSize; i++)
{
Chromossome __temp = new Chromossome(mapSize, mapSize);
__temp.InitializeRandom();
population.Add(__temp);
}
EvaluateFitness();
OrderPopulation();
//Debug.Log("Generation: " + generation + " Best Fitness: " + population[0].fitness);
/**
* for (int i = 0; i < population.Count; i++)
* {
* Debug.Log(population[i].fitness);
* }
* /**/
}
public Chromossome[] Crossover(Chromossome p_parent1, Chromossome p_parent2)
{
int __lenght = p_parent1.genes.Length;
bool __swap = false;
Chromossome[] __children = new Chromossome[2];
//initializing
__children[0] = new Chromossome(mapSize, mapSize);
__children[1] = new Chromossome(mapSize, mapSize);
//Debug.Log("Are parents equal? " + (p_parent1 == p_parent2));
//We are using "Uniform Crossover"
for (int i = 0; i < __lenght; i++)
{
if (Random.value < 0.5f)
{
__swap = !__swap;
}
__children[0].genes[i] = __swap ? p_parent2.genes[i] : p_parent1.genes[i];
__children[1].genes[i] = __swap ? p_parent1.genes[i] : p_parent2.genes[i];
if (Random.value <= mutationRate)
{
__children[0].genes[i] = __children[0].genes[i] == 0 ? 1 : 0;
//Debug.Log("Mutation");
}
if (Random.value <= mutationRate)
{
__children[1].genes[i] = __children[1].genes[i] == 0 ? 1 : 0;
//Debug.Log("Mutation");
}
}
return(__children);
}
