Skip to content

Commit

Permalink
Merge pull request #47 from GuilhermeAGouveia/GuilhermeAGouveia/issue45
Browse files Browse the repository at this point in the history 
# Issue 45 Otimizar parametros do modelo de ilha
  • Loading branch information
@GuilhermeAGouveia
GuilhermeAGouveia authored Nov 11, 2023
2 parents 9766c63 + 9cf29a1 commit 20c08e5
Show file tree
Hide file tree
Showing 200 changed files with 37,720 additions and 2,171 deletions.
155 changes: 59 additions & 96 deletions src/algorithms/aco.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -25,13 +25,13 @@ individuo *candidates;
void set_default_parameters_ant()
{
if (!parameters.num_ants)
parameters.num_ants = 5252;
parameters.num_ants = 2828; // 5252;
if (!parameters.num_generations_per_epoca)
parameters.num_generations_per_epoca = 2000;
parameters.num_generations_per_epoca = (int)(5505098/2828); //423;
if (!parameters.tax_evaporate)
parameters.tax_evaporate = 0.31458;
parameters.tax_evaporate = 0.63855 ;
if (!parameters.num_candidates)
parameters.num_candidates = 9;
parameters.num_candidates = 20;
if (!parameters.p_exploitation)
parameters.p_exploitation = 0.25;
if (!parameters.function_number)
Expand All @@ -42,6 +42,21 @@ void set_default_parameters_ant()
parameters.seed = time(NULL);
if (!parameters.dimension)
parameters.dimension = 10; // 10 or 30
if (!parameters.domain_function.min)
parameters.domain_function.min = -100;
if (!parameters.domain_function.max)
parameters.domain_function.max = 100;
srand(parameters.seed);
}

void reset_parameters_ant()
{
parameters.num_ants = 0;
parameters.num_generations_per_epoca = 0;
parameters.tax_evaporate = 0;
parameters.num_candidates = 0;
parameters.p_exploitation = 0;
parameters.seed = 0;
}

void print_individuos(individuo *individuo, int dimension)
Expand Down Expand Up @@ -87,11 +102,9 @@ void update_pheromones(double **pheromones, individuo *individuos, int n, int d,
for (int j = 0; j < d; j++)
{
pheromones[i][j] = (1 / sqrt(2 * sigma[j] * PI)) * exp(pow((best_individuo->chromosome[j] - individuos[i].chromosome[j]), 2) / (-2 * pow(sigma[j], 2)));
// printf("pheromoneo: %lf\n",pheromones[i][j] );
if (j % 2 == 0)
pheromones[i][j] += random_double(0, 0.3);
}
// printf("\n\n");
}
}

Expand Down Expand Up @@ -147,7 +160,6 @@ void select_next_position2(double **pheromones, individuo *individuos, int d)
{
sum_pheromone[i] = sum_pheromone_dimension(pheromones, i);
}
// printf("Select_Sum_Pheromone:%lf\n", sum_pheromone[1]);

for (int i = 0; i < parameters.num_ants; i++)
{
Expand Down Expand Up @@ -185,8 +197,6 @@ void candiate_calculator_crossover(individuo *individuos, int d, int id_individu
current_dimension_value = individuos[id_individuo].chromosome[j];

candidates[id_candit].chromosome[j] = (current_dimension_value + individuos[individuo2].chromosome[j]) / 2;
// printf("Original: %lf ", current_dimension_value);
// printf("candidato: %lf\n", candidates[id_candit].individuo_chromossome[j]);
if (random_double(0, 1) <= parameters.p_exploitation || !(candidates[id_candit].chromosome[j] <= 100 && candidates[id_candit].chromosome[j] >= -100))
candidates[id_candit].chromosome[j] = random_double(-100, 100);
}
Expand All @@ -204,8 +214,6 @@ void candiate_calculator(individuo *individuos, double **pheromones, double *sum
delta = distance / current_dimension_value;
// Cij = Xi,j + Δij * (m - Xi,j).
candidates[id_candit].chromosome[j] = current_dimension_value + delta * (mean - current_dimension_value);
// printf("Original: %lf ", current_dimension_value);
// printf("candidato: %lf\n", candidates[id_candit].individuo_chromossome[j]);
if (random_double(0, 1) <= parameters.p_exploitation || !(candidates[id_candit].chromosome[j] <= 100 && candidates[id_candit].chromosome[j] >= -100))
candidates[id_candit].chromosome[j] = random_double(-100, 100);
}
Expand Down Expand Up @@ -277,14 +285,9 @@ double sigma_sums(individuo *individuos, int dimension, individuo *best_individu
subtraction_fitness = 0.001;
sum_one += pow(best_individuo->chromosome[dimension] - individuos[i].chromosome[dimension], 2) / subtraction_fitness;
sum_two += 1 / subtraction_fitness;
// printf("sum_one-:%Lf\n", sum_one);
// printf("sum_two-:%Lf\n", sum_two);
// printf("subtract-:%Lf\n", subtraction_fitness);
}
// printf("sum_one-FINAL:%Lf\n", sum_one);
// printf("sum_two-FINAL:%Lf\n", sum_two);

sum_one = sqrt(abs_double(sum_one / sum_two));
// printf("sigma-:%Lf\n", sum_one);
return sum_one;
}

Expand Down Expand Up @@ -322,28 +325,11 @@ void best_individuo_check(individuo *individuos, individuo *best_individuo, doub
}
}

void catch_best_individuo(individuo *individuos, individuo *best_individuot, double **pheromones, int d)
void catch_best_individuo(populacao *population, individuo *best_individuot, double **pheromones, int d)
{
double best_fitness = best_individuot->fitness;
int best_individuo = -1;

for (int i = 1; i < parameters.num_ants; i++)
if (get_best_of_population(*population)->fitness < best_individuot->fitness)
{
if (individuos[i].fitness < best_fitness)
{
best_fitness = individuos[i].fitness;
best_individuo = i;
}
}

if (best_individuo != -1)
{

for (int i = 0; i < d; i++)
{
best_individuot->chromosome[i] = individuos[best_individuo].chromosome[i];
}
best_individuot->fitness = individuos[best_individuo].fitness;
copy_individuo(get_best_of_population(*population), best_individuot, parameters.dimension);
}
}

Expand Down Expand Up @@ -372,9 +358,12 @@ double desvio_padrao_individuo(individuo *individuos, int individuos_size)
populacao *aco(populacao *population)
{
set_default_parameters_ant();
if(population == NULL){
population = generate_island(1,parameters.num_ants, parameters.dimension, parameters.domain_function, parameters.function_number);
// print_parameters(parameters);
if (population == NULL)
{
population = generate_island(1, parameters.num_ants, parameters.dimension, parameters.domain_function, parameters.function_number);
}

individuo *individuos = population->individuos;
int d = 10;
DEBUG(printf("aco\n");)
Expand Down Expand Up @@ -404,86 +393,60 @@ populacao *aco(populacao *population)
}

// Allocate memory for the individuos
//individuo *individuos = (individuo *)malloc(parameters.num_ants * sizeof(individuo));
individuo *best_individuo = (individuo *)malloc(1 * sizeof(individuo));
// individuo *individuos = (individuo *)malloc(parameters.num_ants * sizeof(individuo));
individuo *best_individuo = generate_population(1, 10, parameters.domain_function, 15);
candidates = (individuo *)malloc(parameters.num_candidates * sizeof(individuo));

// Initialize the individuos' positions and fitness valuess
initialize(individuos, parameters.num_ants, d);
initialize(best_individuo, 1, d);
initialize(candidates, parameters.num_candidates, d);

// Find the best individuo and its fitness value
catch_best_individuo(individuos, best_individuo, pheromones, d);
copy_individuo(get_best_of_population(*population), best_individuo, parameters.dimension);

sigma = (double *)malloc(d * sizeof(double));
update_sigma(individuos, d, best_individuo);

// Update the pheromone matrix with the best individuo's path
update_pheromones(pheromones, individuos, parameters.num_ants, d, best_individuo);
// Iterate over the specified number of iterations
int max_inter_add = 100;
int max_inter = 150;
int cont_or_stop = 1;

while (cont_or_stop && difftime(time_now, time_init) < parameters.time_limit)
int generations_count = 0;
while (generations_count < parameters.num_generations_per_epoca && difftime(time_now, time_init) < parameters.time_limit)
{
double best_individuoer = best_individuo->fitness;
for (int iter = 0; iter < max_inter && difftime(time_now, time_init) < parameters.time_limit; iter++)
{
DEBUG(print_individuo(individuos, d, best_individuo););
// printf("Best_fitness: %lf\n", best_individuo->fitness);
// Move each individuo to a new individuo_chromossome
select_next_position(pheromones, individuos, d, best_individuo);
evaporate_pheromones(pheromones, d);

// Update the best individuo and its fitness value
catch_best_individuo(individuos, best_individuo, pheromones, d);
// Verifica se a melhor posição não foi perdida
best_individuo_check(individuos, best_individuo, pheromones, d);
// Update the pheromone matrix with the best individuo's path
update_pheromones(pheromones, individuos, parameters.num_ants, d, best_individuo);
update_sigma(individuos, d, best_individuo);
time(&time_now);
}
DEBUG(print_individuo(individuos, d, best_individuo););
// Move each individuo to a new individuo_chromossome
select_next_position(pheromones, individuos, d, best_individuo);
evaporate_pheromones(pheromones, d);

// double desv = desvio_padrao_individuo(individuos, parameters.num_ants);
// printf("Desvio: %lf\n", desv);

if (doubleEqual(best_individuoer, best_individuo->fitness, 2))
cont_or_stop = 0;
// Update the best individuo and its fitness value
catch_best_individuo(population, best_individuo, pheromones, d);
// Verifica se a melhor posição não foi perdida
best_individuo_check(individuos, best_individuo, pheromones, d);
// Update the pheromone matrix with the best individuo's path
update_pheromones(pheromones, individuos, parameters.num_ants, d, best_individuo);
update_sigma(individuos, d, best_individuo);
time(&time_now);
generations_count++;
}

population->individuos = individuos;
population->size = parameters.num_ants;
copy_individuo(best_individuo, &population->individuos[0], d);

// Print the best solution found
// print_individuos(individuos, d);
// printf("Best fitness value: %lf\n", best_individuo->fitness);
// printf("Best individuo_chromossome:");
// for (int i = 0; i < d; i++)
// {
// printf(" %lf", best_individuo->chromosome[i]);
// }
// printf("\n");
// printf("Best %lf\n", best_individuo->fitness);

// Free memory
for (int i = 0; i < parameters.num_ants; i++)
{
free(pheromones[i]);
// free(individuos[i].chromosome);
}
// for (int i = 0; i < parameters.num_ants; i++)
//{
// free(pheromones[i]);
// free(individuos[i].chromosome);
// }

for (int i = 0; i < parameters.num_candidates; i++)
{
free(pheromones_candidates[i]);
free(candidates[i].chromosome);
}
// for (int i = 0; i < parameters.num_candidates; i++)
//{
// free(pheromones_candidates[i]);
// free(candidates[i].chromosome);
//}
free(pheromones);
free(pheromones_candidates);
free(candidates);
// free(candidates);
free(sigma);
free(best_individuo);
reset_parameters_ant();
return population;
}
70 changes: 33 additions & 37 deletions src/algorithms/clonalg.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -24,19 +24,28 @@ void set_default_parameters_clonalg()
if (!parameters.time_limit)
parameters.time_limit = 10; // seconds
if (!parameters.population_size)
parameters.population_size = 124;
parameters.population_size = 46;
if (!parameters.dimension)
parameters.dimension = 10; // 10 or 30
if (!parameters.domain_function.min)
parameters.domain_function.min = -100;
if (!parameters.domain_function.max)
parameters.domain_function.max = 100;
if (!parameters.num_generations_per_epoca)
parameters.num_generations_per_epoca = 300;
parameters.num_generations_per_epoca = (int)(5505098/46);//615;
if (!parameters.clones)
parameters.clones = 36;
parameters.clones = 72;
if (!parameters.seed)
parameters.seed = time(NULL);
srand(parameters.seed);
}

void reset_parameters_clonalg()
{
parameters.population_size = 0;
parameters.num_generations_per_epoca = 0;
parameters.clones = 0;
parameters.seed = 0;
}

void mutation_n_genes(populacao *populacao_clones, int n_genes, int dimension, domain domain_function)
Expand Down Expand Up @@ -129,11 +138,13 @@ void union_populacao_clones_and_main(populacao *populacao_clones, populacao *pop
populacao *clonalg(populacao *population)
{
set_default_parameters_clonalg();
if(population == NULL){
population = generate_island(1,parameters.population_size, parameters.dimension, parameters.domain_function, parameters.function_number);
// print_parameters(parameters);
if (population == NULL)
{
population = generate_island(1, parameters.population_size, parameters.dimension, parameters.domain_function, parameters.function_number);
}
DEBUG(printf("\nevolution\n"););
//populacao *population_main = generate_island(1, parameters.population_size, parameters.dimension, parameters.domain_function, parameters.function_number);
// populacao *population_main = generate_island(1, parameters.population_size, parameters.dimension, parameters.domain_function, parameters.function_number);
populacao *population_main = population;
qsort(population_main->individuos, population_main->size, sizeof(individuo), comparador_individuo);
populacao *populacao_clones;
Expand All @@ -146,39 +157,24 @@ populacao *clonalg(populacao *population)
// while (difftime(time_now, time_init) < parameters.time_limit)
// {

int max_inter_add = 50;
int max_inter = 100;
int cont_or_stop = 1;
while (cont_or_stop && difftime(time_now, time_init) < parameters.time_limit)
while (generation_count < parameters.num_generations_per_epoca && difftime(time_now, time_init) < parameters.time_limit)
{
double best_anter = population_main->individuos[population_main->size - 1].fitness;
while (generation_count < max_inter && difftime(time_now, time_init) < parameters.time_limit)
{
STATISTICS(print_coords(&population_main->individuos[population_main->size - 1], 1, generation_count, parameters.num_generations););

DEBUG(printf("Populacao principal na geracao %d:\n", generation_count););
DEBUG(print_population(population_main->individuos, population_main->size, parameters.dimension, 1););
populacao_clones = generate_clones(population_main, parameters.clones, parameters.dimension, parameters.domain_function);
populacao_clones_mutated = mutation_clones(populacao_clones, parameters.population_size, parameters.dimension, parameters.domain_function);
union_populacao_clones_and_main(populacao_clones_mutated, population_main, parameters.population_size);
free_population(populacao_clones, parameters.population_size);
qsort(population_main->individuos, population_main->size, sizeof(individuo), comparador_individuo);
generation_count++;
time(&time_now);
}
// double desv = desvio_padrao(population_main->individuos, parameters.population_size);
// printf("Desvio_P: %lf\n", desv);
// printf("Anterior: %lf, Atual_best:%lf\n", best_anter, population_main->individuos[population_main->size - 1].fitness);
STATISTICS(print_coords(&population_main->individuos[population_main->size - 1], 1, generation_count, parameters.num_generations););

if (doubleEqual(best_anter, population_main->individuos[population_main->size - 1].fitness, 2))
{
cont_or_stop = 0;
}

else
{
max_inter += max_inter_add;
}
DEBUG(printf("Populacao principal na geracao %d:\n", generation_count););
DEBUG(print_population(population_main->individuos, population_main->size, parameters.dimension, 1););
populacao_clones = generate_clones(population_main, parameters.clones, parameters.dimension, parameters.domain_function);
populacao_clones_mutated = mutation_clones(populacao_clones, parameters.population_size, parameters.dimension, parameters.domain_function);
union_populacao_clones_and_main(populacao_clones_mutated, population_main, parameters.population_size);
destroy_island(populacao_clones, parameters.population_size);
qsort(population_main->individuos, population_main->size, sizeof(individuo), comparador_individuo);
generation_count++;
time(&time_now);
}
// double desv = desvio_padrao(population_main->individuos, parameters.population_size);
// printf("Desvio_P: %lf\n", desv);
// printf("Anterior: %lf, Atual_best:%lf\n", best_anter, population_main->individuos[population_main->size - 1].fitness);

reset_parameters_clonalg();
return population_main;
}
11 changes: 10 additions & 1 deletion src/algorithms/commom.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@

#include "../libs/log.h"
#define STATISTICS(x)
#define DEBUG(x)
#define DEBUG(x)
#define LOG(x)

void fitness(individuo *individuo, int dimension, int function_number)
Expand All @@ -26,6 +26,14 @@ void fitness(individuo *individuo, int dimension, int function_number)
// printf("fitness: %f\n", individuo->fitness);
}

int comparador_proporcoes_alg(const void *a, const void *b)
{
proporcao_alg *v1 = (proporcao_alg *)a;
proporcao_alg *v2 = (proporcao_alg *)b;

return v1->proporcao < v2->proporcao;
}

int comparador_individuo(const void *a, const void *b)
{
individuo *v1 = (individuo *)a;
Expand Down Expand Up @@ -96,6 +104,7 @@ void destroy_population(individuo *population, int n_individuos)
for (int i = 0; i < n_individuos; i++)
{
free(population[i].chromosome);
free(population[i].velocidade);
}
free(population);
}
Expand Down
Loading

0 comments on commit 20c08e5

Please sign in to comment.