def pareto_front_population():
size_of_list = 6
population = []
for i in range(size_of_list + 1):
values = [False] * (size_of_list - i) + [True] * i
indv = MultipleValueChromosome(values)
indv.genetic_age = i
population.append(indv)
return population
def test_fitness_is_not_inherited_crossover():
crossover = SinglePointCrossover()
parent1 = MultipleValueChromosome(
[np.random.choice([True, False]) for _ in range(10)])
parent2 = MultipleValueChromosome(
[np.random.choice([True, False]) for _ in range(10)])
child1, child2 = crossover(parent1, parent2)
assert not child1.fit_set
assert not child2.fit_set
def test_adding_trainers_to_predictor_fitness_function(
predictor_fitness_function):
for i in range(10):
trainer = MultipleValueChromosome([i])
predictor_fitness_function.add_trainer(trainer)
assert trainer not in predictor_fitness_function._trainers
assert np.any([[i] == trainer.values
for trainer in predictor_fitness_function._trainers])
def non_dominated_population():
young_weak = MultipleValueChromosome([False, False])
middle_average = MultipleValueChromosome([False, True])
middle_average.genetic_age = 1
old_fit = MultipleValueChromosome([True, True])
old_fit.genetic_age = 2
return [young_weak, middle_average, old_fit]
def test_predicted_fitness_for_trainer(predictor_fitness_function,
sample_population):
predictor = MultipleValueChromosome([0, 9])
for i, trainer in enumerate(sample_population):
prediction = \
predictor_fitness_function.predict_fitness_for_trainer(predictor,
trainer)
expected_prediction = i + 4.5
np.testing.assert_almost_equal(prediction, expected_prediction)
def test_mutation_is_single_point():
mutator = SinglePointMutation(mutation_onemax_specific)
parent = MultipleValueChromosome(
[np.random.choice([True, False]) for _ in range(10)])
child = mutator(parent)
discrepancies = 0
for i in range(len(parent.values)):
if child.values[i] != parent.values[i]:
discrepancies += 1
assert discrepancies <= 1
def selected_indiviudals(pareto_front_population):
list_size = len(pareto_front_population[0].values)
list_one = [False] * int(list_size / 2) + [True] * int((list_size + 1) / 2)
list_two = [False] * int((list_size + 1) / 2) + [True] * int(list_size / 2)
selected_indv_one = MultipleValueChromosome(list_one)
selected_indv_two = MultipleValueChromosome(list_two)
selected_indv_one.genetic_age = list_size
selected_indv_two.genetic_age = list_size + 1
return [selected_indv_one, selected_indv_two]
def test_genetic_age_is_oldest_parent():
crossover = SinglePointCrossover()
parent1 = MultipleValueChromosome([np.random.choice([True, False])
for _ in range(10)])
parent2 = MultipleValueChromosome([np.random.choice([True, False])
for _ in range(10)])
parent1.genetic_age = 8
parent2.genetic_age = 4
child1, child2 = crossover(parent1, parent2)
assert child1.genetic_age == 8
assert child2.genetic_age == 8
def sample_bool_list_chromosome():
chromosome = MultipleValueChromosome(
[np.random.choice([True, False]) for _ in range(10)])
return chromosome
def test_fitness_predictor_fitness_function_call(predictor_fitness_function,
predictor_values):
predictor = MultipleValueChromosome(predictor_values)
expected_fitness = abs(4.5 - np.mean(predictor_values))
fitness = predictor_fitness_function(predictor)
np.testing.assert_almost_equal(fitness, expected_fitness)
def sample_population():
return [MultipleValueChromosome(list(range(i, i + 10))) for i in range(10)]
def sample_int_list_chromosome():
chromosome = MultipleValueChromosome(
[np.random.choice([1, 0]) for _ in range(10)])
return chromosome
def test_fitness_is_not_inherited_mutation():
mutator = SinglePointMutation(mutation_onemax_specific)
parent = MultipleValueChromosome(
[np.random.choice([True, False]) for _ in range(10)])
child = mutator(parent)
assert not child.fit_set