index_pair selector::stochastic_roulette(generation& gen)
{
auto chromosome = std::max_element(gen.begin(), gen.end(),
[](auto& c1, auto& c2) { return c1.fitness > c2.fitness; });
double max_fitness = chromosome->fitness;
// double max_fitness = 0.9;
auto chromosome_index = range::distribution_of(gen);
auto select = [&](auto& random)
{
while(true)
{
int index = random(chromosome_index);
double fitness = gen[index].fitness;
// double fitness = 0.2;
if (random(fitness_range) < fitness / max_fitness)
return index;
}
};
return index_pair(select(random), select(random));
}
index_pair selector::ranking(generation& gen)
{
std::sort(gen.begin(), gen.end(), [](auto& c1, auto& c2) { return c1.fitness < c2.fitness; });
int sum = 0;
for (int i = 0; i < gen.size(); ++i)
sum += i;
std::uniform_int_distribution<> dis_sum(0, sum);
auto select = [&](auto& random)
{
int probability = random(dis_sum);
int index = 0;
while (probability > 0)
{
probability -= index + 1;
++index;
}
return index >= gen.size() ? index - 1 : index;
};
return index_pair(select(random), select(random));
}
std::vector<chromosome> selector::elitism(generation& gen)
{
std::sort(gen.begin(), gen.end(), [](auto& c1, auto& c2) { return c1.fitness > c2.fitness; });
std::vector<chromosome> next_gen;
next_gen.reserve(elitism_rate);
for (int i = 0; i < elitism_rate; ++i)
next_gen.push_back(gen[i]);
return next_gen;
}
index_pair selector::roulette(generation& gen)
{
double sum = std::accumulate(gen.begin(), gen.end(), 0.0, [](double prev, auto& c) { return prev + c.fitness; });
// double sum = 30.0;
std::uniform_real_distribution<> dis_sum(0.0, sum);
auto select = [&](auto& random)
{
double probability = random(dis_sum);
int index = 0;
while (probability > 0)
{
probability -= gen[index].fitness;
++index;
}
return index >= gen.size() ? index - 1 : index;
};
return index_pair(select(random), select(random));
}
void gen_initialize(generation &geninit, int &spop, int &apop, int &idnumber, const int &aleaspace, const int &n_param, const vector<parameters> ¶ms, const double &lambda) {
if(aleaspace==0) {
///Populate the space with equidistant values. First generation
int geninit_nindividuals = (int)pow(spop,n_param);
geninit.construct(geninit_nindividuals, n_param, idnumber, lambda);
///Determination of parameters_equally_spaced
mat samples = doe_uniform(spop, n_param, params);
for(int j=0; j<n_param; j++) {
for(int i=0; i<geninit.size(); i++) {
geninit.pop[i].p(j) = samples(i,j);
}
}
}
if(aleaspace==1) {
///Populate the space with equidistant values. First generation
int geninit_nindividuals = (int)pow(spop,n_param);
geninit.construct(geninit_nindividuals, n_param, idnumber, lambda);
///Determination of parameters_equally_spaced
mat samples = doe_uniform_limit(spop, n_param, params);
for(int j=0; j<n_param; j++) {
for(int i=0; i<geninit.size(); i++) {
geninit.pop[i].p(j) = samples(i,j);
}
}
}
else if(aleaspace==2) {
int geninit_nindividuals=apop;
geninit.construct(geninit_nindividuals, n_param, idnumber, lambda);
///Determination of random values
///Determination of parameters_equally_spaced
mat samples = doe_random(geninit_nindividuals, n_param, params);
for(int j=0; j<n_param; j++) {
for(int i=0; i<geninit.size(); i++) {
geninit.pop[i].p(j) = samples(i,j);
}
}
}
else if(aleaspace==3) {
mat samples;
read_gen(apop, samples, n_param);
int geninit_nindividuals=apop;
geninit.construct(geninit_nindividuals, n_param, idnumber, lambda);
for(int j=0; j<n_param; j++) {
for(int i=0; i<geninit.size(); i++) {
geninit.pop[i].p(j) = samples(i,j);
}
}
}
}
