def update_external_archive(self):
feasible_solutions = []
for solution in self.solutions:
if is_feasible(solution):
feasible_solutions.append(copy.deepcopy(solution))
if len(feasible_solutions) > 0:
feasible_solutions = feasible_solutions + self.archive
ranking = FastNonDominatedRanking()
ranking.compute_ranking(feasible_solutions)
first_rank_solutions = ranking.get_subfront(0)
if len(first_rank_solutions) <= self.population_size:
self.archive = []
for solution in first_rank_solutions:
self.archive.append(copy.deepcopy(solution))
else:
crowding_distance = CrowdingDistance()
while len(first_rank_solutions) > self.population_size:
crowding_distance.compute_density_estimator(
first_rank_solutions)
first_rank_solutions = sorted(
first_rank_solutions,
key=lambda x: x.attributes['crowding_distance'],
reverse=True)
first_rank_solutions.pop()
self.archive = []
for solution in first_rank_solutions:
self.archive.append(copy.deepcopy(solution))
def execute(self, front: List[S]) -> List[S]:
if front is None:
raise Exception('The front is null')
elif len(front) == 0:
raise Exception('The front is empty')
ranking = FastNonDominatedRanking(self.dominance_comparator)
crowding_distance = CrowdingDistance()
ranking.compute_ranking(front)
ranking_index = 0
new_solution_list = []
while len(new_solution_list) < self.max_population_size:
if len(ranking.get_subfront(ranking_index)) < (self.max_population_size - len(new_solution_list)):
new_solution_list = new_solution_list + ranking.get_subfront(ranking_index)
ranking_index += 1
else:
subfront = ranking.get_subfront(ranking_index)
crowding_distance.compute_density_estimator(subfront)
sorted_subfront = sorted(subfront, key=lambda x: x.attributes['crowding_distance'], reverse=True)
for i in range((self.max_population_size - len(new_solution_list))):
new_solution_list.append(sorted_subfront[i])
return new_solution_list
def __init__(
self,
problem: Problem,
population_size: int,
neighborhood: Neighborhood,
archive: BoundedArchive,
mutation: Mutation,
crossover: Crossover,
selection: Selection = BinaryTournamentSelection(
MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])),
termination_criterion: TerminationCriterion = store.
default_termination_criteria,
population_generator: Generator = store.default_generator,
population_evaluator: Evaluator = store.default_evaluator,
dominance_comparator: Comparator = store.default_comparator,
):
"""
MOCEll implementation as described in:
:param problem: The problem to solve.
:param population_size: Size of the population.
:param mutation: Mutation operator (see :py:mod:`jmetal.operator.mutation`).
:param crossover: Crossover operator (see :py:mod:`jmetal.operator.crossover`).
:param selection: Selection operator (see :py:mod:`jmetal.operator.selection`).
"""
super(MOCell, self).__init__(
problem=problem,
population_size=population_size,
offspring_population_size=1,
mutation=mutation,
crossover=crossover,
selection=selection,
termination_criterion=termination_criterion,
population_evaluator=population_evaluator,
population_generator=population_generator,
)
self.dominance_comparator = dominance_comparator
self.neighborhood = neighborhood
self.archive = archive
self.current_individual = 0
self.current_neighbors = []
self.comparator = MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])
def __init__(
self,
problem: DynamicProblem[S],
population_size: int,
offspring_population_size: int,
mutation: Mutation,
crossover: Crossover,
selection: Selection = BinaryTournamentSelection(
MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])),
termination_criterion: TerminationCriterion = store.
default_termination_criteria,
population_generator: Generator = store.default_generator,
population_evaluator: Evaluator = store.default_evaluator,
dominance_comparator: DominanceComparator = DominanceComparator()):
super(DynamicNSGAII, self).__init__(
problem=problem,
population_size=population_size,
offspring_population_size=offspring_population_size,
mutation=mutation,
crossover=crossover,
selection=selection,
population_evaluator=population_evaluator,
population_generator=population_generator,
termination_criterion=termination_criterion,
dominance_comparator=dominance_comparator)
self.completed_iterations = 0
self.start_computing_time = 0
self.total_computing_time = 0
def replacement(self, population: List[S], offspring_population: List[S]) -> List[List[S]]:
result = self.dominance_comparator.compare(population[self.current_individual], offspring_population[0])
if result == 1: # the offspring individual dominates the current one
population[self.current_individual] = offspring_population[0]
self.archive.add(offspring_population[0])
elif result == 0: # the offspring and current individuals are non-dominated
new_individual = offspring_population[0]
self.current_neighbors.append(new_individual)
ranking: Ranking = FastNonDominatedRanking()
ranking.compute_ranking(self.current_neighbors)
density_estimator: DensityEstimator = CrowdingDistance()
for i in range(ranking.get_number_of_subfronts()):
density_estimator.compute_density_estimator(ranking.get_subfront(i))
self.current_neighbors.sort(key=cmp_to_key(self.comparator.compare))
worst_solution = self.current_neighbors[-1]
self.archive.add(new_individual)
if worst_solution != new_individual:
population[self.current_individual] = new_individual
return population
def __init__(
self,
problem: Problem,
population_size: int,
mutation: Mutation,
crossover: Crossover,
number_of_cores: int,
client,
selection: Selection = BinaryTournamentSelection(
MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])),
termination_criterion: TerminationCriterion = store.
default_termination_criteria,
dominance_comparator: DominanceComparator = DominanceComparator()):
super(DistributedNSGAII, self).__init__()
self.problem = problem
self.population_size = population_size
self.mutation_operator = mutation
self.crossover_operator = crossover
self.selection_operator = selection
self.dominance_comparator = dominance_comparator
self.termination_criterion = termination_criterion
self.observable.register(termination_criterion)
self.number_of_cores = number_of_cores
self.client = client
def __init__(
self,
problem: Problem,
dominance_comparator: Comparator = store.default_comparator,
max_evaluations: int = 250,
individual_population_size: int = 100,
report_interval: int = 100,
dataDirectory: str = "./decmopy/decmo2/weigths",
):
super().__init__()
self.problem = problem
self.population_size = individual_population_size
self.max_evaluations = max_evaluations
self.report_interval = report_interval
self.dataDirectory = dataDirectory
self.mix_interval = 1
"""Replacement"""
ranking = FastNonDominatedRanking(dominance_comparator)
density_estimator = CrowdingDistance()
self.r = RankingAndDensityEstimatorReplacement(
ranking, density_estimator, RemovalPolicyType.SEQUENTIAL)
self.MIN_VALUES = 0
self.MAX_VALUES = 1
min_values: List[float] = []
max_values: List[float] = []
for _ in range(problem.number_of_objectives):
min_values.append(sys.float_info.max)
max_values.append(sys.float_info.min)
self.extreme_values: List[List[float]] = []
self.extreme_values.append(min_values)
self.extreme_values.append(max_values)
def __init__(self, maximum_size: int, reference_point: List[float]):
super(CrowdingDistanceArchiveWithReferencePoint, self).__init__(
maximum_size=maximum_size,
reference_point=reference_point,
comparator=SolutionAttributeComparator("crowding_distance",
lowest_is_best=False),
density_estimator=CrowdingDistance())
def replacement(self, population: List[S],
offspring_population: List[S]) -> List[List[S]]:
""" This method joins the current and offspring populations to produce the population of the next generation
by applying the ranking and crowding distance selection.
:param population: Parent population.
:param offspring_population: Offspring population.
:return: New population after ranking and crowding distance selection is applied.
"""
ranking = FastNonDominatedRanking(self.dominance_comparator)
density_estimator = CrowdingDistance()
r = RankingAndDensityEstimatorReplacement(ranking, density_estimator,
RemovalPolicyType.ONE_SHOT)
solutions = r.replace(population, offspring_population)
front = self.get_result()
if type(front) is not list:
solutions = [front]
for solution in front:
print(solution.variables[0])
for solution in front:
print(str(front.index(solution)) + ": ",
sep=' ',
end='',
flush=True)
print(solution.objectives, sep=' ', end='', flush=True)
print()
return solutions
def __init__(self,
reference_directions,
problem: Problem,
mutation: Mutation,
crossover: Crossover,
population_size: int = None,
selection: Selection = BinaryTournamentSelection(
MultiComparator([FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()])),
termination_criterion: TerminationCriterion = store.default_termination_criteria,
population_generator: Generator = store.default_generator,
population_evaluator: Evaluator = store.default_evaluator,
dominance_comparator: Comparator = store.default_comparator):
self.reference_directions = reference_directions.compute()
if not population_size:
population_size = len(self.reference_directions)
if self.reference_directions.shape[1] != problem.number_of_objectives:
raise Exception('Dimensionality of reference points must be equal to the number of objectives')
super(NSGAIII, self).__init__(
problem=problem,
population_size=population_size,
offspring_population_size=population_size,
mutation=mutation,
crossover=crossover,
selection=selection,
termination_criterion=termination_criterion,
population_evaluator=population_evaluator,
population_generator=population_generator,
dominance_comparator=dominance_comparator
)
self.extreme_points = None
self.ideal_point = np.full(self.problem.number_of_objectives, np.inf)
self.worst_point = np.full(self.problem.number_of_objectives, -np.inf)
def __init__(self,
problem: Problem,
population_size: int,
offspring_population_size: int,
mutation: Mutation,
crossover: Crossover,
selection: Selection = BinaryTournamentSelection(
MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])),
termination_criterion: TerminationCriterion = store.
default_termination_criteria,
population_generator: Generator = store.default_generator,
population_evaluator: Evaluator = store.default_evaluator,
dominance_comparator: Comparator = store.default_comparator,
target_value_threshold: List[float] = None,
target_pattern: List[int] = None):
"""
NSGA-II implementation as described in
* K. Deb, A. Pratap, S. Agarwal and T. Meyarivan, "A fast and elitist
multiobjective genetic algorithm: NSGA-II," in IEEE Transactions on Evolutionary Computation,
vol. 6, no. 2, pp. 182-197, Apr 2002. doi: 10.1109/4235.996017
NSGA-II is a genetic algorithm (GA), i.e. it belongs to the evolutionary algorithms (EAs)
family. The implementation of NSGA-II provided in jMetalPy follows the evolutionary
algorithm template described in the algorithm module (:py:mod:`jmetal.core.algorithm`).
.. note:: A steady-state version of this algorithm can be run by setting the offspring size to 1.
:param problem: The problem to solve.
:param population_size: Size of the population.
:param mutation: Mutation operator (see :py:mod:`jmetal.operator.mutation`).
:param crossover: Crossover operator (see :py:mod:`jmetal.operator.crossover`).
:param selection: Selection operator (see :py:mod:`jmetal.operator.selection`).
"""
super(NSGAII, self).__init__(
problem=problem,
population_size=population_size,
offspring_population_size=offspring_population_size,
mutation=mutation,
crossover=crossover,
selection=selection,
termination_criterion=termination_criterion,
population_evaluator=population_evaluator,
population_generator=population_generator)
self.dominance_comparator = dominance_comparator
self.generation = 0
self.target_pattern = target_pattern
self.problem_solved = False
self.target_value_threshold = target_value_threshold
self.file_pareto_front = os.getcwd() + '/' + str(
time.strftime("%Y_%m_%d")) + '_PARETO_'
if not os.path.exists(self.file_pareto_front):
os.mkdir(self.file_pareto_front)
def replacement(self, population: List[S], offspring_population: List[S]) -> List[List[S]]:
ranking = FastNonDominatedRanking(self.dominance_comparator)
density_estimator = CrowdingDistance()
r = RankingAndDensityEstimatorReplacement(ranking, density_estimator, RemovalPolicyType.ONE_SHOT)
solutions = r.replace(population, offspring_population)
return solutions
def replacement(self, population: List[S],
offspring_population: List[S]) -> List[List[S]]:
""" This method joins the current and offspring populations to produce the population of the next generation
by applying the ranking and crowding distance selection.
:param population: Parent population.
:param offspring_population: Offspring population.
:return: New population after ranking and crowding distance selection is applied.
"""
ranking = FastNonDominatedRanking(self.dominance_comparator)
density_estimator = CrowdingDistance()
r = RankingAndDensityEstimatorReplacement(ranking, density_estimator,
RemovalPolicyType.ONE_SHOT)
solutions = r.replace(population, offspring_population)
return solutions
def __init__(self,
problem: MultiTransfer.MultiTransferProblem,
population_size: int,
offspring_population_size: int,
mutation: Mutation,
crossover: Crossover,
selection: Selection = BinaryTournamentSelection(
MultiComparator([FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()])),
termination_criterion: TerminationCriterion = store.default_termination_criteria,
population_generator: Generator = store.default_generator,
population_evaluator: Evaluator = store.default_evaluator,
dominance_comparator: Comparator = store.default_comparator):
"""
NSGA-II implementation as described in
* K. Deb, A. Pratap, S. Agarwal and T. Meyarivan, "A fast and elitist
multiobjective genetic algorithm: NSGA-II," in IEEE Transactions on Evolutionary Computation,
vol. 6, no. 2, pp. 182-197, Apr 2002. doi: 10.1109/4235.996017
NSGA-II is a genetic algorithm (GA), i.e. it belongs to the evolutionary algorithms (EAs)
family. The implementation of NSGA-II provided in jMetalPy follows the evolutionary
algorithm template described in the algorithm module (:py:mod:`jmetal.core.algorithm`).
.. note:: A steady-state version of this algorithm can be run by setting the offspring size to 1.
:param problem: The problem to solve.
:param population_size: Size of the population.
:param mutation: Mutation operator (see :py:mod:`jmetal.operator.mutation`).
:param crossover: Crossover operator (see :py:mod:`jmetal.operator.crossover`).
:param selection: Selection operator (see :py:mod:`jmetal.operator.selection`).
"""
super(NSGAII_MEDA, self).__init__(
problem=problem,
population_size=population_size,
offspring_population_size=offspring_population_size,
mutation=mutation,
crossover=crossover,
selection=selection,
termination_criterion=termination_criterion,
population_evaluator=population_evaluator,
population_generator=population_generator
)
self.dominance_comparator = dominance_comparator
def __init__(
self,
problem: Problem,
dominance_comparator: Comparator = store.default_comparator,
max_iterations: int = 250,
individual_population_size: int = 100,
report_interval: int = 100,
):
super().__init__()
self.problem = problem
self.population_size = individual_population_size
self.max_iterations = max_iterations
self.report_interval = report_interval
self.mix_interval = self.population_size / 10
"""Replacement"""
ranking = FastNonDominatedRanking(dominance_comparator)
density_estimator = CrowdingDistance()
self.r = RankingAndDensityEstimatorReplacement(
ranking, density_estimator, RemovalPolicyType.SEQUENTIAL)
def construct(self, hyperparameters: Mapping, scenario: Mapping,
warm_startup_info: Mapping) -> None:
# Constructing meta-heuristics initialization arguments (components + simple hyperparameters)
init_args = dict(copy(hyperparameters))
if "crossover_type" in init_args:
import jmetal.operator.crossover as crossover
crossover_class = JMetalPyWrapper._get_class_from_module(
name=init_args.pop("crossover_type"), module=crossover)
init_args["crossover"] = crossover_class(
init_args.pop("crossover_probability"))
if "mutation_type" in init_args:
import jmetal.operator.mutation as mutation
mutation_class = JMetalPyWrapper._get_class_from_module(
name=init_args.pop("mutation_type"), module=mutation)
init_args["mutation"] = mutation_class(
init_args.pop("mutation_probability"))
if "selection_type" in init_args:
selection_type = init_args.pop('selection_type')
if selection_type == 'ReuletteWheelSelection':
from jmetal.util.comparator import MultiComparator
from jmetal.util.density_estimator import CrowdingDistance
from jmetal.util.ranking import FastNonDominatedRanking
init_args["selection"] = MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])
else:
import jmetal.operator.selection as selection
selection_class = JMetalPyWrapper._get_class_from_module(
name=selection_type, module=selection)
init_args["selection"] = selection_class()
# Add all non-component Metaheuristic parameters
if "offspring_population_size" in init_args:
offsp_population = init_args.pop("offspring_population_size")
# offspring_population_size should be even
offsp_population += offsp_population % 2
init_args['offspring_population_size'] = offsp_population
# elitist should be bool
if "elitist" in init_args:
init_args['elitist'] = True if init_args.pop(
'elitist') == "True" else False
if "population_size" in init_args:
init_args["population_size"] = init_args.pop("population_size")
termination_class = JMetalPyWrapper._get_class_from_module(
name=scenario["Budget"]["Type"], module=termination)
init_args["termination_criterion"] = termination_class(
scenario["Budget"]["Amount"])
# making dict hashable enables using LRU cache
problem_init_params = HashableDict(
scenario['problem_initialization_parameters'])
problem = JMetalPyWrapper._get_problem(
problem_name=scenario['Problem'], init_params=problem_init_params)
init_args["problem"] = problem
# Attach initial solutions.
self.load_initial_solutions(warm_startup_info, problem)
llh_name = init_args.pop("low level heuristic")
if llh_name == "jMetalPy.SimulatedAnnealing":
init_args["solution_generator"] = self._solution_generator
else:
init_args['population_generator'] = self._solution_generator
# Instantiate Metaheuristic object
self._llh_algorithm = JMetalPyWrapper._get_algorithm_class(llh_name)(
**init_args)
def __init__(self, maximum_size: int):
super(CrowdingDistanceArchive, self).__init__(
maximum_size=maximum_size,
comparator=SolutionAttributeComparator("crowding_distance",
lowest_is_best=False),
density_estimator=CrowdingDistance())
from jmetal.util.termination_criterion import StoppingByEvaluations
if __name__ == '__main__':
problem = TSP(instance='../../resources/TSP_instances/kroA100.tsp')
print('Cities: ', problem.number_of_variables)
algorithm = GeneticAlgorithm(
problem=problem,
population_size=100,
offspring_population_size=100,
mutation=PermutationSwapMutation(1.0 / problem.number_of_variables),
crossover=PMXCrossover(0.8),
selection=BinaryTournamentSelection(
MultiComparator([
FastNonDominatedRanking.get_comparator(),
CrowdingDistance.get_comparator()
])),
termination_criterion=StoppingByEvaluations(max=2500000))
algorithm.observable.register(observer=PrintObjectivesObserver(1000))
algorithm.run()
result = algorithm.get_result()
print('Algorithm: {}'.format(algorithm.get_name()))
print('Problem: {}'.format(problem.get_name()))
print('Solution: {}'.format(result.variables))
print('Fitness: {}'.format(result.objectives[0]))
print('Computing time: {}'.format(algorithm.total_computing_time))
def run(self):
""" Execute the algorithm. """
self.start_computing_time = time.time()
create_solution = dask.delayed(self.problem.create_solution)
evaluate_solution = dask.delayed(self.problem.evaluate)
task_pool = as_completed([], with_results=True)
for _ in range(self.number_of_cores):
new_solution = create_solution()
new_evaluated_solution = evaluate_solution(new_solution)
future = self.client.compute(new_evaluated_solution)
task_pool.add(future)
batches = task_pool.batches()
auxiliar_population = []
while len(auxiliar_population) < self.population_size:
batch = next(batches)
for _, received_solution in batch:
auxiliar_population.append(received_solution)
if len(auxiliar_population) < self.population_size:
break
# submit as many new tasks as we collected
for _ in batch:
new_solution = create_solution()
new_evaluated_solution = evaluate_solution(new_solution)
future = self.client.compute(new_evaluated_solution)
task_pool.add(future)
self.init_progress()
# perform an algorithm step to create a new solution to be evaluated
while not self.stopping_condition_is_met():
batch = next(batches)
for _, received_solution in batch:
offspring_population = [received_solution]
# replacement
ranking = FastNonDominatedRanking(self.dominance_comparator)
density_estimator = CrowdingDistance()
r = RankingAndDensityEstimatorReplacement(
ranking, density_estimator, RemovalPolicyType.ONE_SHOT)
auxiliar_population = r.replace(auxiliar_population,
offspring_population)
# selection
mating_population = []
for _ in range(2):
solution = self.selection_operator.execute(
auxiliar_population)
mating_population.append(solution)
# Reproduction and evaluation
new_task = self.client.submit(reproduction, mating_population,
self.problem,
self.crossover_operator,
self.mutation_operator)
task_pool.add(new_task)
# update progress
self.evaluations += 1
self.solutions = auxiliar_population
self.update_progress()
if self.stopping_condition_is_met():
break
self.total_computing_time = time.time() - self.start_computing_time
# at this point, computation is done
for future, _ in task_pool:
future.cancel()
def setUp(self):
self.crowding = CrowdingDistance()
class CrowdingDistanceTestCases(unittest.TestCase):
def setUp(self):
self.crowding = CrowdingDistance()
def test_should_the_crowding_distance_of_an_empty_set_do_nothing(self):
solution_list = []
self.crowding.compute_density_estimator(solution_list)
def test_should_the_crowding_distance_of_single_solution_be_infinity(self):
solution = Solution(3, 3)
solution_list = [solution]
self.crowding.compute_density_estimator(solution_list)
value = solution_list[0].attributes["crowding_distance"]
self.assertEqual(float("inf"), value)
def test_should_the_crowding_distance_of_two_solutions_be_infinity(self):
solution1 = Solution(3, 3)
solution2 = Solution(3, 3)
solution_list = [solution1, solution2]
self.crowding.compute_density_estimator(solution_list)
value_from_solution1 = solution_list[0].attributes["crowding_distance"]
value_from_solution2 = solution_list[1].attributes["crowding_distance"]
self.assertEqual(float("inf"), value_from_solution1)
self.assertEqual(float("inf"), value_from_solution2)
def test_should_the_crowding_distance_of_three_solutions_correctly_assigned(self):
solution1 = Solution(2, 2)
solution2 = Solution(2, 2)
solution3 = Solution(2, 2)
solution1.objectives[0] = 0.0
solution1.objectives[1] = 1.0
solution2.objectives[0] = 1.0
solution2.objectives[1] = 0.0
solution3.objectives[0] = 0.5
solution3.objectives[1] = 0.5
solution_list = [solution1, solution2, solution3]
self.crowding.compute_density_estimator(solution_list)
value_from_solution1 = solution_list[0].attributes["crowding_distance"]
value_from_solution2 = solution_list[1].attributes["crowding_distance"]
value_from_solution3 = solution_list[2].attributes["crowding_distance"]
self.assertEqual(float("inf"), value_from_solution1)
self.assertEqual(float("inf"), value_from_solution2)
self.assertEqual(2.0, value_from_solution3)
def test_should_the_crowding_distance_of_four_solutions_correctly_assigned(self):
solution1 = Solution(2, 2)
solution2 = Solution(2, 2)
solution3 = Solution(2, 2)
solution4 = Solution(2, 2)
solution1.objectives[0] = 0.0
solution1.objectives[1] = 1.0
solution2.objectives[0] = 1.0
solution2.objectives[1] = 0.0
solution3.objectives[0] = 0.5
solution3.objectives[1] = 0.5
solution4.objectives[0] = 0.75
solution4.objectives[1] = 0.75
solution_list = [solution1, solution2, solution3, solution4]
self.crowding.compute_density_estimator(solution_list)
value_from_solution1 = solution_list[0].attributes["crowding_distance"]
value_from_solution2 = solution_list[1].attributes["crowding_distance"]
value_from_solution3 = solution_list[2].attributes["crowding_distance"]
value_from_solution4 = solution_list[3].attributes["crowding_distance"]
self.assertEqual(float("inf"), value_from_solution1)
self.assertEqual(float("inf"), value_from_solution2)
self.assertGreater(value_from_solution3, value_from_solution4)
