def test_raises_value_errors(self, params):
print(params)
# Assert raises ValueError when parameters are incorrect.
with self.assertRaises(ValueError):
genetic.GeneticAlgorithm(self.agent,
random_state=self.random_state,
**params).search()
def test_genetic_similarity_raises_error(self):
ga = genetic.GeneticAlgorithm(
self.agent, mutation_factor=.5, mutation_probability=1,
max_evolution_duration=4, min_genetic_similarity=.5,
random_state=self.random_state)
with self.assertRaises(RuntimeError):
ga.genetic_similarity()
def test_breed(self):
ga = genetic.GeneticAlgorithm(self.agent,
population_size=100, n_selected=100)
(ga.search_start().generate_population().cycle_start()
.select_for_breeding().breed())
self.assertEqual(len(ga.population_), 100)
self.assertEqual(len(ga.offspring_), 50)
def test_select_for_breeding(self, method):
ga = genetic.GeneticAlgorithm(self.agent,
n_selected=20,
breeding_selection=method,
max_evolution_cycles=1)
(ga.search_start().generate_population().cycle_start()
.select_for_breeding())
self.assertEqual(len(ga.selected_), 20)
def test_search_duration_constraint(self, params, acceptable_elapsed):
ga = genetic.GeneticAlgorithm(self.agent,
random_state=self.random_state,
**params)
elapsed = time.time()
ga.search()
elapsed = time.time() - elapsed
# Assert that the duration constraint was respected.
self.assertLess(elapsed, acceptable_elapsed)
self.assertIsNotNone(ga.solution_candidate_)
def test_generate_population(self):
ga = genetic.GeneticAlgorithm(self.agent, max_evolution_cycles=1,
random_state=self.random_state)
ga.search()
self.assertEqual(ga.population_size_, 1000)
# Assert that the arrays necessary for the search were disposed.
self.assertIsNone(ga.population_)
self.assertIsNone(ga.selected_)
self.assertIsNone(ga.offspring_)
ga = genetic.GeneticAlgorithm(self.agent, population_size=20,
max_evolution_cycles=1,
random_state=self.random_state)
ga.search()
self.assertEqual(ga.population_size_, 20)
ga = genetic.GeneticAlgorithm(self.agent,
max_evolution_cycles=10,
max_evolution_duration=1,
n_jobs=1,
random_state=self.random_state)
ga.search()
self.assertGreater(ga.population_size_, 100)
def test_search(self, params, expected):
ga = genetic.GeneticAlgorithm(self.agent,
random_state=self.random_state, **params)
solution = ga.search().solution_candidate_
# Attributes were set as expected.
for key, value in expected.items():
self.assertEqual(getattr(ga, key), value)
# Assert clean-up was made.
self.assertIsNone(ga.offspring_)
self.assertIsNone(ga.selected_)
# Assert it eventually finds a solution.
self.assertIsNotNone(solution)
self.assertEqual(solution.data, 'hello world')
def test_preemption_by_genetic_similarity(self):
expected_variability = .4
a = _A(search=genetic.GeneticAlgorithm, environment=_E2())
ga = genetic.GeneticAlgorithm(
a, max_evolution_duration=60,
min_genetic_similarity=expected_variability,
population_size=50,
mutation_probability=0,
random_state=self.random_state,
debug=True).search()
# Assert that the last population's variability is smaller
# than the `min_genetic_similarity` parameter passed.
self.assertLessEqual(ga.variability_, expected_variability)
self.assertIsNotNone(ga.solution_candidate_)
self.assertGreaterEqual(a.utility(ga.solution_candidate_), 7)
def test_sanity(self):
ga = genetic.GeneticAlgorithm(self.agent,
random_state=self.random_state)
self.assertIsNotNone(ga)