public Ind <G> Run(int steps, float goalFit)
{
// Aux. variable
Ind <G>[] aux;
// Run the genetic algorithm until a maximum number of steps or
// a goal fitness is reached
for (int i = 0; i < steps && best.Fit < goalFit; step++, i++)
{
// Select the next generation of individuals
select.Invoke(popCurrent, popNext);
// Clone the selected individuals
for (int j = 0; j < popSize; j++)
{
popNext[j] = popNext[j].Copy();
}
// Crossover (mate)
for (int j = 0; j < popSize; j += 2)
{
if (random.NextDouble() < crossoverProb)
{
mate.Invoke(popNext[j], popNext[j + 1]);
popNext[j].Fit = null;
}
}
// Mutation
foreach (Ind <G> ind in popNext)
{
if (random.NextDouble() < mutateProb)
{
mutate.Invoke(ind);
ind.Fit = null;
}
}
// Evaluate new generation
foreach (Ind <G> ind in popNext)
{
if (!ind.Fit.HasValue)
{
evaluate.Invoke(ind);
if (ind.Fit > best.Fit)
{
best = ind;
}
}
}
// Update population
aux = popCurrent;
popCurrent = popNext;
popNext = aux;
}
// Return best solution found so far
return(best);
}
public void Mutate(Ind <G> ind)
{
for (int i = 0; i < ind.GeneCount; i++)
{
if (random.NextDouble() < flipProb)
{
G changed;
do
{
changed = geneValues[random.Next(geneValues.Length)];
} while (changed.Equals(ind[i]));
ind[i] = changed;
}
}
}
public void Init()
{
step = 0;
// Generate a new population and determine each individual's fitness
for (int i = 0; i < popSize; i++)
{
Ind <G> ind = generate.Invoke();
evaluate.Invoke(ind);
popCurrent[i] = ind;
if (i == 0 || ind.Fit > best.Fit)
{
best = popCurrent[i];
}
}
}
public void Mate(Ind <G> ind1, Ind <G> ind2)
{
int split = random.Next(ind1.GeneCount);
for (int i = 0; i < split; i++)
{
G aux = ind1[i];
ind1[i] = ind2[i];
ind2[i] = aux;
}
for (int i = split; i < ind1.GeneCount; i++)
{
G aux = ind2[i];
ind2[i] = ind1[i];
ind1[i] = aux;
}
}
