allow_duplicate_genes=False has no effect in 2.17.0

[sebastianardelean]

Hi, I tested the pygad library version 2.17.0 for solving the knapsack problem and I observed that even if I set allow_duplicate_genes=False, I still have duplicated genes in the solution.

As shown in the following solution, It can be seen that gene -5 is duplicated, -25 also.

 [28, -3, 12, -21, 18, 24, -22, 24, -30, 2, 28, -27, 21, -5, -5, 5, -1, 7, 12, 23, -9, -15, -21, -16, -16, -23, 26, -14, -25, -25]

I instantiate pygad as following

ga_instance = pygad.GA(num_generations=METAHEURISTIC_NO_OF_GENERATIONS,
                           num_parents_mating=METAHEURISTIC_NUM_PARENS_MATING,
                           gene_type=int,
                           fitness_func=fitness_func,
                           sol_per_pop=METAHEURISTIC_POPULATION_SIZE,
                           num_genes=METAHEURISTIC_GENE_LENGTH,
                           init_range_low=METAHEURISTIC_GENE_LOW,
                           init_range_high=METAHEURISTIC_GENE_HIGH,
                           parent_selection_type=METAHEURISTIC_SELECTION_TYPE,
                           keep_parents=METAHEURISTIC_KEEP_PARENTS,
                           crossover_type=METAHEURISTIC_CROSSOVER_TYPE,
                           crossover_probability=METAHEURISTIC_CROSSOVER_RATE,
                           mutation_type=METAHEURISTIC_MUTATION_TYPE,
                           mutation_probability = METAHEURISTIC_MUTATION_RATE,
                           #                       mutation_probability = METAHEURISTIC_MUTATION_PROBABILITY,
                           #mutation_percent_genes=METAHEURISTIC_MUTATION_PERCENT_GENES,
                           stop_criteria="saturate_75",
                           allow_duplicate_genes=False,
                           parallel_processing=['process', 10]
                           )

and the fitness function is defined as:

def fitness_func(solution, solution_idx):
    # print(solution)
    calculated_mass = 0
    calculated_value = 0
    output = 0

    unique_elements = check_duplicates(solution)
    if len(unique_elements) < len(solution):
        return 1/sys.maxsize

    for gene in solution:
        position = abs(gene) - 1
        if gene > 0:
            calculated_mass += items[position].m
            calculated_value += items[position].v
    output = fitness_function(calculated_mass, calculated_value)
    if calculated_mass > MAX_ALLOWED_MASS:
        return 1/sys.maxsize
    return output

[ahmedfgad]

Hi @sebastianardelean,

Please refer to this issue: https://github.com/ahmedfgad/GeneticAlgorithmPython/issues/39

Does this solve your problem?

[ardeleanasm]

Yes, it solves. Thank you!