def test_mutation_parent_selection():
""" Tests mutation_parent_selection function """
gp_par = gp.GpParameters()
gp_par.n_population = 8
gp_par.f_mutation = 0.5
gp_par.parent_selection = gp.SelectionMethods.ELITISM
gp_par.mutate_co_parents = True
gp_par.mutate_co_offspring = True
population = gp.create_population(gp_par.n_population, 5)
crossover_parents = [5, 6, 7]
crossover_offspring = gp.create_population(2, 5)
fitness = [0, 1, 2, 1, 2, 1, 1, 2, 2, 0]
parents = gp.mutation_parent_selection(population, fitness,
crossover_parents,
crossover_offspring, gp_par)
assert parents == [8, 7, 4, 2]
gp_par.mutate_co_parents = False
gp_par.mutate_co_offspring = False
parents = gp.mutation_parent_selection(population, fitness,
crossover_parents,
crossover_offspring, gp_par)
assert parents == [4, 2, 3, 1]
gp_par.n_population = 4
gp_par.f_mutation = 0.1
fitness = [0, 1, 2, 1]
parents = gp.mutation_parent_selection(population, fitness, [], [], gp_par)
assert parents == []
def test_mutation():
""" Tests mutation function """
gp_par = gp.GpParameters()
gp_par.n_population = 5
gp_par.n_offspring_mutation = 1
gp_par.mutation_p_add = 0.3
gp_par.mutation_p_delete = 0.3
gp_par.allow_identical = False
for _ in range(5): #Test a number of random possibilities
population = gp.create_population(gp_par.n_population, 5)
mutated_population = gp.mutation(population, range(len(population)),
gp_par)
assert len(mutated_population
) == gp_par.n_offspring_mutation * gp_par.n_population
for i in range(len(mutated_population)):
assert mutated_population[i] not in population
for j in range(len(mutated_population)):
if i != j:
assert mutated_population[i] != mutated_population[j]
gp_par.n_offspring_mutation = 5 * gp_par.n_population + 1
gp_par.allow_identical = True
for _ in range(5): #Test a number of random possibilities
population = gp.create_population(gp_par.n_population, 5)
mutated_population = gp.mutation(population, range(len(population)),
gp_par)
assert len(mutated_population
) == gp_par.n_offspring_mutation * gp_par.n_population
#Test with too limited mutation possibilities
gp_par.n_population = 5
gp_par.n_offspring_mutation = 1
gp_par.mutation_p_add = 0.0
gp_par.mutation_p_delete = 1.0
gp_par.allow_identical = False
parents = gp.create_population(gp_par.n_population, 2)
blockers = gp.create_population(7, 1)
population = parents + blockers
mutated_population = gp.mutation(population, range(len(parents)), gp_par)
assert len(mutated_population
) != gp_par.n_offspring_mutation * gp_par.n_population
for i in range(len(mutated_population)):
assert mutated_population[i] not in population
for j in range(len(mutated_population)):
if i != j:
assert mutated_population[i] != mutated_population[j]
def test_crossover():
""" Tests crossover function """
pop_size = 10
gp_par = gp.GpParameters()
gp_par.n_population = pop_size
gp_par.n_offspring_crossover = 1
gp_par.allow_identical = False
for _ in range(5): #Test a number of random possibilities
population = gp.create_population(pop_size, 5)
crossover_population = gp.crossover(population, range(len(population)),
gp_par)
for i in range(len(crossover_population)):
assert crossover_population[i] not in population
for j in range(len(crossover_population)):
if i != j:
assert crossover_population[i] != crossover_population[j]
gp_par.n_offspring_crossover = 3 * pop_size
gp_par.allow_identical = True
for _ in range(5): #Test a number of random possibilities
population = gp.create_population(pop_size, 5)
crossover_population = gp.crossover(population, range(len(population)),
gp_par)
assert len(crossover_population
) == gp_par.n_offspring_crossover * gp_par.n_population
#Test what happens if no valid crossovers can be found
parents = [['s(', 'a0', ')'], ['s(', 'a1', ')']]
gp_par.n_offspring_crossover = 1
gp_par.allow_identical = False
gp_par.n_population = 2
blockers = [['s(', 'a1', 'a0', ')'], ['s(', 'a0', 'a1', ')']]
population = parents + blockers
crossover_population = gp.crossover(population, range(len(parents)),
gp_par)
assert len(crossover_population
) == gp_par.n_offspring_crossover * gp_par.n_population
for i in range(len(crossover_population)):
assert crossover_population[i] not in population
for j in range(len(crossover_population)):
if i != j:
assert crossover_population[i] != crossover_population[j]
#Number of parents not factor of two
with pytest.raises(ValueError):
crossover_population = gp.crossover(population, range(3), gp_par)
def test_create_population():
""" Tests create_population function """
pop_size = 5
genome_length = 5
for _ in range(5): #Test a number of random possibilities
population = gp.create_population(population_size=pop_size,
genome_length=genome_length)
for i in range(pop_size):
assert len(population[i]) >= genome_length
for j in range(genome_length):
assert population[i][j] != []
def test_crossover_parent_selection():
""" Tests crossover_parent_selection function """
gp_par = gp.GpParameters()
gp_par.n_population = 10
gp_par.f_crossover = 0.4
gp_par.parent_selection = gp.SelectionMethods.ELITISM
population = gp.create_population(gp_par.n_population, 5)
fitness = [0, 1, 2, 1, 2, 1, 1, 2, 2, 0]
parents = gp.crossover_parent_selection(population, fitness, gp_par)
assert parents == [8, 7, 4, 2]
gp_par.n_population = 4
gp_par.f_crossover = 0.1
fitness = [0, 1, 2, 1]
parents = gp.crossover_parent_selection(population, fitness, gp_par)
assert parents == []