Different Solutions Sizes Every Generation

[Christian-Nordentoft]

Hello, I am using PyGAD for my University project, and I encountered a problem today that I did not find earlier. If anyone can help me figure out the problem, thanks in advance :)

The problem I have is that the number of solutions, in each generation is different. When printing the solutions, I can see I am getting 18 solutions shown one generation, 17 in a different generation etc. Furthermore, it repeats the solutions within the generation.

The parameters I have chosen are based on wanting a convergence for two design variables within a search space, where parents are kept for the next generation and selecting parents based on which ones have the best fitness evaluation results. The code that I have used to decide the parameters of PyGAD:

num_generations = 100
sol_per_pop = 12
num_parents_mating = 4
mutation_probability = 0.5
keep_parents = 2
num_genes = 2
parent_selection_type = "sss"
crossover_type = "uniform"
mutation_type = "random"
initial_population = np.array([
[10, 1],
[100, 2],
[200, 3],
[300, 4],
[400, 5],
[500, 1],
[570, 2],
[250, 3],
[150, 4],
[350, 5],
[450, 2],
[550, 3],
])
gene_space=[range(10, 571), range(1, 6)]

def fitness_func(ga_instance, solution, solution_idx):
Objective_Function = User_Definition(Inputs)
solution_fitness = Objective_Function
print(ga_instance.generations_completed, solution_idx, solution_fitness, solution)
return solution_fitness

def callback_generation(ga_instance):
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(' ---> End of generation', ga_instance.generations_completed, ":", solution_fitness)
print(' ---> End of generation', ga_instance.generations_completed, ":", solution)

ga_instance = pygad.GA(num_generations=num_generations,
num_parents_mating=num_parents_mating,
fitness_func=fitness_func,
sol_per_pop=sol_per_pop,
num_genes=num_genes,
parent_selection_type=parent_selection_type,
crossover_type=crossover_type,
mutation_type=mutation_type,
mutation_probability=mutation_probability,
on_generation=callback_generation,
initial_population=initial_population,
gene_space=gene_space)