def setUp(self):
self.neighborhood = Mock()
self.performance = Mock()
self.tolerance = Mock()
self.mutation_probability = Mock()
self.mutation_probability.get_relational_probability.return_value = 0.2
epsilon = 1
self.selection_size = 1
self.mutator = Mutator(self.neighborhood, self.performance, self.tolerance, self.mutation_probability, epsilon)
self.tolerance.get_tolerance.return_value = 2**-5
class TestMutator(unittest.TestCase):
def setUp(self):
self.neighborhood = Mock()
self.performance = Mock()
self.tolerance = Mock()
self.mutation_probability = Mock()
self.mutation_probability.get_relational_probability.return_value = 0.2
epsilon = 1
self.selection_size = 1
self.mutator = Mutator(self.neighborhood, self.performance, self.tolerance, self.mutation_probability, epsilon)
self.tolerance.get_tolerance.return_value = 2**-5
def set_neighborhood(self):
neighborhood = set()
neighborhood.add(self.expected_rep)
neighborhood.add((1, 1, 0, 0))
neighborhood.add((1, 0, 0, 1))
self.neighborhood.get_neighborhood_of_representation.return_value = neighborhood
def test_mutator__one_in_bene(self):
representation = (1, 1, 0, 1)
perf_of_representation = 0.2
self.expected_rep = (1, 1, 1, 1)
perf_of_expected_representation = 0.2 + self.tolerance.get_tolerance.return_value
def perf_returns(rep):
if rep == representation:
return perf_of_representation
if rep == self.expected_rep:
return perf_of_expected_representation
return 0
self.set_neighborhood()
self.performance.get_estimated_performance.side_effect = perf_returns
self.assertEqual(self.expected_rep, self.mutator.evolutionary_step(representation))
def test_mutator__two_in_neut(self):
representation = (1, 1, 0, 1)
self.expected_rep = set()
self.expected_rep.add((1, 1, 1, 1))
self.expected_rep.add((0, 0, 0, 0))
self.performance.get_estimated_performance.return_value = 0
self.neighborhood.get_neighborhood_of_representation.return_value = self.expected_rep
self.assertIn(self.mutator.evolutionary_step(representation), self.expected_rep)
def learn_single_time__classic_model(common_classes):
length, epsilon, mutation_neighborhood, tolerance = common_classes.get_common_classes(
)
mutation_probability = common_classes.get_mutation_probability()
concept_class = MonotoneConjunction(length)
performance = common_classes.get_perf_without_precomp(concept_class)
mutator = Mutator(mutation_neighborhood, performance, tolerance,
mutation_probability, epsilon)
algorithm = ConjunctionEvolvabilityAlgorithm(mutator, length, epsilon,
performance)
return algorithm.learn_ideal_function_until_match()
def main():
common_classes = CommonClassesCreator(False)
length, epsilon, mutation_neighborhood, tolerance = common_classes.get_common_classes()
concept_class = MonotoneConjunction(length)
performance = common_classes.get_perf_without_precomp(concept_class)
mutation_probability = common_classes.get_mutation_probability()
mutator = Mutator(mutation_neighborhood, performance, tolerance, mutation_probability, epsilon)
algorithm = ConjunctionEvolvabilityAlgorithm(mutator, length, epsilon, performance)
hypo = algorithm.learn_ideal_function(epsilon)
print "HYPO IS: " + str(hypo)
def learn_DNF__basic_model():
dnf = DNF()
length = 58
epsilon = (2**-53)
tau = (epsilon / length)**3 * log(1 / epsilon)
tolerance = ConjunctionTolerance(length, epsilon)
conjunction_performance_oracle = OneSidedPerformanceOracleWithTolerance(
dnf, tau)
performance_oracle = DNFOneSidePerformanceOracleWithTolerance(
dnf, tau, epsilon, conjunction_performance_oracle)
mutation_neighborhood = MonotoneConjunctionNeighborhood()
mutation_probability = ConjunctionMutationProbability(
mutation_neighborhood)
mutator = Mutator(mutation_neighborhood, performance_oracle, tolerance,
mutation_probability, epsilon)
algorithm = ConjunctionEvolvabilityAlgorithm(mutator, length, epsilon,
performance_oracle)
print algorithm.learn_ideal_function_until_match()