def test_perform_crossover(self):
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
pop = Population([chromosome1, chromosome2])
crossover_pop = pop._perform_crossover()
expected_len = 3
self.assertTrue(len(crossover_pop) == expected_len)
def test_selection(self):
chromosome2 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome1 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
pop = Population([chromosome1, chromosome2, chromosome3])
self.assertEqual(pop, pop.selection())
def test_perform_crossover_with_seed(self):
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
pop = Population([chromosome1, chromosome2])
crossover_chromosome = pop._perform_crossover(1)
expected_crossover_chromosome = Chromosome(
[Gene(2, 1, True),
Gene(3, 4, True),
Gene(5, 6, True)])
self.assertEqual(crossover_chromosome, expected_crossover_chromosome)
def test_selection_with_different_fitness_func(self):
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
pop = Population([chromosome1, chromosome2, chromosome3], max_weight=9)
expected_pop = Population([chromosome3, chromosome1], max_weight=9)
self.assertEqual(expected_pop, pop.selection(maximum_selection=2))
def test_selection_with_less_chromosomes(self):
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
pop = Population([chromosome1, chromosome2, chromosome3])
expected_pop = Population([chromosome3, chromosome1])
self.assertEqual(expected_pop, pop.selection(maximum_selection=2))
def _initialize_first_population(self, gene_pool: List[Gene]):
"""
Helper method that creates a new population with randomly generated gene states.
:param gene_pool: The pool of genes that we can use for each chromosome.
:return: A population of chromosomes composed of genes.
"""
from genetic_algorithm.Population import Population
population = list()
population.append(Chromosome(gene_pool))
for _ in range(self.pop_limit):
new_chromosome = []
for gene in gene_pool:
toss = bool(random.getrandbits(1))
new_chromosome.append(Gene(gene.value, gene.weight, toss))
population.append(Chromosome(new_chromosome))
return Population(population, population_limit=self.pop_limit)
def parameterized_fitness_func(chromosome: Chromosome) -> int:
"""
Fitness function that calculates the fitness for a given chromosome.
:param chromosome: The given chromosome.
:return: An integer denoting the fitness of the chromosome.
"""
profits, weights = chromosome.calculate_active_values_and_weights()
return profits - weights if weights > MAX_WEIGHT else profits
def test_crossover_does_not_return_empty_chromosomes(self):
population_limit = 10
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
pop = Population([chromosome1, chromosome2, chromosome3],
population_limit=population_limit)
new_pop = pop.crossover()
for chrom in new_pop.chromosomes:
self.assertTrue(len(chrom.genes) > 0)
def test_mutation_does_not_alter_chromosome_length(self):
population_limit = 10
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
expected_chromosome_length = 3
pop = Population([chromosome1, chromosome2, chromosome3],
population_limit=population_limit)
pop.mutate(0.2)
for chromosome in pop.chromosomes:
self.assertTrue(len(chromosome.genes), expected_chromosome_length)
def test_crossover_returns_same_population_size(self):
population_limit = 10
chromosome1 = Chromosome(
[Gene(2, 1, True),
Gene(3, 1, True),
Gene(4, 1, True)])
chromosome2 = Chromosome(
[Gene(1, 2, True),
Gene(3, 4, True),
Gene(5, 6, True)])
chromosome3 = Chromosome(
[Gene(5, 1, True),
Gene(8, 2, True),
Gene(3, 1, True)])
pop = Population([chromosome1, chromosome2, chromosome3],
population_limit=population_limit)
new_pop = pop.crossover()
expected_pop_size = population_limit
for chrom in new_pop.chromosomes:
self.assertTrue(len(chrom.genes) > 0)
self.assertTrue(len(new_pop.chromosomes) == expected_pop_size)
def test_fitness_positive_fitness_high_values(self):
genes = [Gene(6, 1, True), Gene(8, 2, True), Gene(4, 3, True)]
chromosome = Chromosome(genes)
score = chromosome.fitness_score(
BoundedKnapsackGA.fitness_func(BoundedKnapsackGA.MAX_WEIGHT))
other_genes = [Gene(2, 1, True), Gene(3, 2, True), Gene(4, 3, True)]
other_chromosome = Chromosome(other_genes)
other_score = other_chromosome.fitness_score(
BoundedKnapsackGA.fitness_func(BoundedKnapsackGA.MAX_WEIGHT))
self.assertTrue(score > 0 and score > other_score)
def _perform_crossover(self, random_seed=None):
"""
Helper function to perform crossover. Chooses randomly two chromosomes and mixes half and half of each.
:param random_seed: An optional random seed to use.
:return: A new chromosome resulting from the crossover of two parent chromosomes.
"""
if not random_seed:
rng = random.Random()
else:
rng = random.Random(random_seed)
# Choose two parents randomly
chain = []
parents = rng.sample(self.chromosomes, 2)
parent1 = parents[0]
parent2 = parents[1]
chromosome_length = len(parent1.genes) // 2
chain.extend(parent1.genes[:chromosome_length])
chain.extend(parent2.genes[chromosome_length:])
# Return unique genes. This is a 0-1 knapsack after all.
return Chromosome(list(chain))
def test_mutation_alters_chromosome_length(self):
genes = [Gene(6, 1, True), Gene(8, 2, True), Gene(4, 3, True)]
expected_length = 3
chromosome = Chromosome(genes)
chromosome.mutate(1)
self.assertTrue(len(chromosome) == expected_length)
def initPopulation(self, size):
for i in range(size):
self.population.append(Chromosome(10))
def test_fitness_func_bad_fitness(self):
genes = [Gene(1, 5, True), Gene(2, 8, True), Gene(3, 2, True)]
chromosome = Chromosome(genes)
score = chromosome.fitness_score(
BoundedKnapsackGA.fitness_func(BoundedKnapsackGA.MAX_WEIGHT))
self.assertTrue(score == 0)
def test_fitness_func_with_reward(self):
genes = [Gene(2, 1, True), Gene(3, 2, True), Gene(4, 3, True)]
chromosome = Chromosome(genes)
score = chromosome.fitness_score(
BoundedKnapsackGA.fitness_func(BoundedKnapsackGA.MAX_WEIGHT))
self.assertTrue(score > 0)
def test_inequality_between_chromosomes(self):
first_genes = [Gene(1, 2, True), Gene(3, 4, True)]
second_genes = [Gene(3, 4, True), Gene(1, 2, True)]
first_chromosome = Chromosome(first_genes)
second_chromosome = Chromosome(second_genes)
self.assertFalse(first_chromosome == second_chromosome)
def test_equality_between_chromosomes(self):
genes = [Gene(1, 2, True), Gene(3, 4, True)]
first_chromosome = Chromosome(genes)
second_chromosome = Chromosome(genes)
self.assertTrue(first_chromosome == second_chromosome)
