How does the mutation operation work, when a discrete gene space is given?

[TiMo3654]

My implementation looks like this:

num_generations         = 100
num_parents_mating      = 2

sol_per_pop             = 5000
num_genes               = 10

init_range_low          = 0
init_range_high         = 1

parent_selection_type   = "tournament"
keep_elitism            = 50

crossover_type          = "single_point"

mutation_type           = "random"
mutation_percent_genes  = 20

continious_gene_space   = {'low' :  0, 'high':   1}
discrete_gene_space_1   = np.linspace(0.3, 2.4, 8)
discrete_gene_space_2   = np.linspace(3, 12, 4)


def check_for_termination(ga_instance):  
     if ga_instance.best_solution(pop_fitness=ga_instance.last_generation_fitness)[1] >= np.float64(2.0):
        return "stop"
    

def genetic_optimizer(fun):

    ga_instance = pygad.GA(num_generations      = num_generations,
                       num_parents_mating       = num_parents_mating,
                       fitness_func             = fun,
                       sol_per_pop              = sol_per_pop,
                       num_genes                = num_genes,
                       init_range_low           = init_range_low,
                       init_range_high          = init_range_high,
                       parent_selection_type    = parent_selection_type,
                       keep_elitism             = keep_elitism,
                       crossover_type           = crossover_type,
                       mutation_type            = mutation_type,
                       mutation_percent_genes   = mutation_percent_genes,
                       gene_space               = [continious_gene_space, continious_gene_space, continious_gene_space, continious_gene_space, continious_gene_space, continious_gene_space, continious_gene_space, continious_gene_space, discrete_gene_space_1, discrete_gene_space_2],
                       on_generation            = check_for_termination,
                       mutation_by_replacement  = False,
                       random_mutation_min_val  = -1,
                       random_mutation_max_val  = 1,
                       save_best_solutions      = True)
    
    ga_instance.run()

    solution            = ga_instance.best_solutions[-1]
    solution_fitness    = ga_instance.best_solutions_fitness[-1]

    num_completed_generations = ga_instance.generations_completed

    print('Completed Generation: ' + str(num_completed_generations))

    return solution_fitness, solution

I was not able to find a statement in the documentation, on how the mutation effects discrete genes.

Are the genes mutated and then rounded to the next discrete value or is there any other mechanism?

Is it furthermore possible, to define an individual mutation range for each gene similar to the gene_space attribute?

[ahmedfgad]

When gene_space is used, then mutation happens by replacing the gene value by a value randomly selected from the gene space. This happens for both int and float data types.

Defining a mutation space is a good feature to be supported. Thanks for your suggestion.

[ahmedfgad]

Update: Custom mutation range feature is implemented. It will be supported in the next release. 31a3975