Support of NSGA-II

Author: ahmedfgad

pygad/pygad.py

@@ -1683,6 +1683,12 @@ def cal_pop_fitness(self):
                         if self.fitness_batch_size in [1, None]:
                             fitness = self.fitness_func(self, sol, sol_idx)
                             if type(fitness) in GA.supported_int_float_types:
+                                # The fitness function returns a single numeric value.
+                                # This is a single-objective optimization problem.
+                                pass
+                            elif type(fitness) in [list, tuple, numpy.ndarray]:
+                                # The fitness function returns a list/tuple/numpy.ndarray.
+                                # This is a multi-objective optimization problem.
                                 pass
                             else:
                                 raise ValueError(f"The fitness function should return a number but the value {fitness} of type {type(fitness)} found.")
@@ -1718,6 +1724,12 @@ def cal_pop_fitness(self):
 
                         for index, fitness in zip(batch_indices, batch_fitness):
                             if type(fitness) in GA.supported_int_float_types:
+                                # The fitness function returns a single numeric value.
+                                # This is a single-objective optimization problem.
+                                pop_fitness[index] = fitness
+                            elif type(fitness) in [list, tuple, numpy.ndarray]:
+                                # The fitness function returns a list/tuple/numpy.ndarray.
+                                # This is a multi-objective optimization problem.
                                 pop_fitness[index] = fitness
                             else:
                                 raise ValueError(f"The fitness function should return a number but the value {fitness} of type {type(fitness)} found.")
@@ -1779,6 +1791,12 @@ def cal_pop_fitness(self):
                     if self.fitness_batch_size in [1, None]:
                         for index, fitness in zip(solutions_to_submit_indices, executor.map(self.fitness_func, [self]*len(solutions_to_submit_indices), solutions_to_submit, solutions_to_submit_indices)):
                             if type(fitness) in GA.supported_int_float_types:
+                                # The fitness function returns a single numeric value.
+                                # This is a single-objective optimization problem.
+                                pop_fitness[index] = fitness
+                            elif type(fitness) in [list, tuple, numpy.ndarray]:
+                                # The fitness function returns a list/tuple/numpy.ndarray.
+                                # This is a multi-objective optimization problem.
                                 pop_fitness[index] = fitness
                             else:
                                 raise ValueError(f"The fitness function should return a number but the value {fitness} of type {type(fitness)} found.")
@@ -1810,6 +1828,12 @@ def cal_pop_fitness(self):
 
                             for index, fitness in zip(batch_indices, batch_fitness):
                                 if type(fitness) in GA.supported_int_float_types:
+                                    # The fitness function returns a single numeric value.
+                                    # This is a single-objective optimization problem.
+                                    pop_fitness[index] = fitness
+                                elif type(fitness) in [list, tuple, numpy.ndarray]:
+                                    # The fitness function returns a list/tuple/numpy.ndarray.
+                                    # This is a multi-objective optimization problem.
                                     pop_fitness[index] = fitness
                                 else:
                                     raise ValueError(f"The fitness function should return a number but the value ({fitness}) of type {type(fitness)} found.")

pygad/utils/mutation.py

@@ -471,7 +471,13 @@ def adaptive_mutation_population_fitness(self, offspring):
 
         first_idx = len(parents_to_keep)
         last_idx = fitness.shape[0]
-        fitness[first_idx:last_idx] = [0]*(last_idx - first_idx)
+        if len(fitness.shape) > 1:
+            # TODO This is a multi-objective optimization problem.
+            # fitness[first_idx:last_idx] = [0]*(last_idx - first_idx)
+            raise ValueError('Edit adaptive mutation to work with multi-objective optimization problems.')
+        else:
+            # This is a single-objective optimization problem.
+            fitness[first_idx:last_idx] = [0]*(last_idx - first_idx)
 
         if self.fitness_batch_size in [1, None]:
             # Calculate the fitness for each individual solution.
@@ -667,7 +673,7 @@ def adaptive_mutation_by_space(self, offspring):
                                                                                          gene_type=self.gene_type,
                                                                                          num_trials=10)
         return offspring
-        
+
     def adaptive_mutation_randomly(self, offspring):
 
         """

pygad/utils/parent_selection.py

@@ -395,7 +395,9 @@ def nsga2_selection(self,
 
         # The number of remaining parents to be selected.
         num_remaining_parents = num_parents
-    
+
+        # Index of the current parent.
+        current_parent_idx = 0
         # A loop variable holding the index of the current pareto front.
         pareto_front_idx = 0
         while num_remaining_parents != 0 and pareto_front_idx < len(pareto_fronts):
@@ -407,10 +409,12 @@ def nsga2_selection(self,
                 for sol_idx in range(len(current_pareto_front)):
                     selected_solution_idx = current_pareto_front[sol_idx, 0]
                     # Insert the parent into the parents array.
-                    parents[sol_idx, :] = self.population[selected_solution_idx, :].copy()
+                    parents[current_parent_idx, :] = self.population[selected_solution_idx, :].copy()
                     # Insert the index of the selected parent.
                     parents_indices.append(selected_solution_idx)
-    
+                    # Increase the parent index.
+                    current_parent_idx += 1
+
                 # Decrement the number of remaining parents by the length of the pareto front.
                 num_remaining_parents -= len(current_pareto_front)
             else:
@@ -422,9 +426,11 @@ def nsga2_selection(self,
 
                 for selected_solution_idx in crowding_dist_pop_sorted_indices[0:num_remaining_parents]:
                     # Insert the parent into the parents array.
-                    parents[sol_idx, :] = self.population[selected_solution_idx, :].copy()
+                    parents[current_parent_idx, :] = self.population[selected_solution_idx, :].copy()
                     # Insert the index of the selected parent.
                     parents_indices.append(selected_solution_idx)
+                    # Increase the parent index.
+                    current_parent_idx += 1
 
                 # Decrement the number of remaining parents by the number of selected parents.
                 num_remaining_parents -= num_remaining_parents