def test_createMutantNetwork():
pop_gen = PopulationGenerator(10, 53, 45)
weights = WalkingNetwork.generateRandomWeights()
nw = WalkingNetwork(weights)
ten_networks = []
for _i in range(0, 10):
ten_networks.append(pop_gen.createMutantNetwork(nw))
assert True == False, 'Test not implemented'
def test_mutate5050():
pop_gen = PopulationGenerator(10, 53, 45)
weights = WalkingNetwork.generateRandomWeights()
nw = WalkingNetwork(weights)
ten_networks = []
for _i in range(0, 10):
ten_networks.append(pop_gen.mutate5050(nw))
assert nw in ten_networks, 'Always mutation (fails with 0.1%)'
atleastOneChanged = False
for i in range(0, 10):
if (nw is not ten_networks[i]):
atleastOneChanged = True
assert atleastOneChanged, 'Never mutated (fails with 0.1%)'
def test_getRandomIndexBetterPreferred():
pop_gen = PopulationGenerator(10, 53, 45)
init_population = pop_gen.initPopulation()
ten_choosen_indexes = []
for _i in range(0, 10):
ten_choosen_indexes.append(
pop_gen.getRandomIndexBetterPreferred(init_population))
individual_indexes = []
for element in ten_choosen_indexes:
if element not in individual_indexes:
individual_indexes.append(element)
assert len(ten_choosen_indexes) >= len(
individual_indexes
), 'no index chosen double, can happen but unlikely' # TODO how unlikely
assert init_population[
0] in ten_choosen_indexes, 'first index not chosen, can happen but unlikely' # TODO how unlikely
def test_crossoverNetwork():
pop_gen = PopulationGenerator(10, 43, 45)
weights = WalkingNetwork.generateRandomWeights()
nw1 = WalkingNetwork(weights)
weights = WalkingNetwork.generateRandomWeights()
nw2 = WalkingNetwork(weights)
crossover_network = pop_gen.crossoverNetwork(nw1, nw2)
crossedOver = False
for i in range(0, crossover_network.getNumberOfWeights()):
if (crossover_network.getWeightAt(i) == nw1.getWeightAt(i)):
assert not crossedOver, 'Switched back to nw1 (or weirder) (or really bad luck)'
else:
crossedOver = True
if (crossedOver):
assert crossover_network.getWeightAt(i) == nw2.getWeightAt(
i), 'Weights from neither nw1 nor nw2 taken'
def test_createNextGeneration():
pop_gen = PopulationGenerator(10, 53, 45)
init_population = pop_gen.initPopulation()
new_population = pop_gen.createNextGeneration(init_population)
assert init_population[0] is new_population[
0], 'Good networks not kept alive'
assert init_population[1] is new_population[
1], 'Good networks not kept alive'
for i in range(2, 10):
assert init_population[i] is not new_population[
i], 'Unchanged WalkingNetwork'
changedWeights = False
for i in range(0, new_population[3].getNumberOfWeights()):
if (init_population[0].getWeightAt(i) !=
new_population[3].getWeightAt(i)):
changedWeights = True
assert changedWeights, 'No changed Networkweights on nw in position 3'
assert len(new_population
) == 10, 'new population not the same size as old population'
def initPopGen(self, popsize, mutation_rate, crossover_rate):
self.pop_generator = PopulationGenerator(popsize, mutation_rate,
crossover_rate)
def test_initPopulation():
pop_gen = PopulationGenerator(10, 53, 45)
init_population = pop_gen.initPopulation()
assert init_population[0] is not None, 'Generated WalkingNetwork is None'
assert len(init_population
) == 10, 'wrong amount of individuals in the init-population'