def __init_window(self):
init()
self.__surface = display.set_mode(self.__screen_size)
display.set_caption(
'Game of Life --- SPACE - Toggle evolution, R - reset generation')
model = Population(PygameCellFactory(), self.__screen_size)
self.__board_view = PygameBoard('Population', model, self.__surface)
model.add_observer(self.__board_view)
self.controller.model = model
self.controller.view = self.__board_view
while True:
for e in event.get():
if e.type == QUIT:
quit()
exit()
if e.type == KEYDOWN:
if e.key == K_SPACE:
model.toggle_evolution()
if e.key == K_r:
model.reset_population()
self.controller.handle()
self.__pyg_clock.tick(40)
display.flip()
def selection(population, iteration, breakPoint=50, k=3):
shape = population.individuals.shape[0]
new_population = Population(None)
new_population.individuals = np.empty(shape=shape, dtype=object)
for i in range(shape):
new_population.individuals[i] = roulette(population) if (
iteration < breakPoint) else tournament(population)
return new_population
def crossover(population, crossover_ratio=0.70):
shape = population.individuals.shape[0]
newPopulation = Population(None)
newPopulation.individuals = np.empty(shape=shape, dtype=object)
for i in range(shape):
newPopulation.individuals[i] = one_point_crossover(
population.individuals[i],
population.individuals[(i + 1) % (shape - 1)])
return newPopulation
def main():
args_parser = build_args_parser()
args = args_parser.parse_args()
results_dir_path = args.output_path
if not os.path.exists(results_dir_path):
os.makedirs(results_dir_path)
for function_type in [f.value for f in FunctionType]:
function = build_function(function_type)
best_fitness_history_by_topology_type = {}
for topology_type in [t.value for t in TopologyType]:
best_fitness_history = []
for i in range(args.num_simulations):
start_time = time.time()
population = Population(args.population_size, args.dimension,
function.lower_limit,
function.upper_limit, topology_type)
if args.use_clerc:
best_fitness = pso.optimize_with_clerc(
function, population, args.num_iterations,
args.cognitive_coeff, args.social_coeff)
else:
best_fitness = pso.optimize(function, population,
args.num_iterations,
args.initial_inertia_coeff,
args.final_inertia_coeff,
args.cognitive_coeff,
args.social_coeff)
elapsed_time = time.time() - start_time
print(function_type + ", " + topology_type + ", simulação: " +
str(i + 1) + ", fitness: " +
str(round(best_fitness, 2)) + "(" +
str(round(elapsed_time, 2)) + " s)")
best_fitness_history.append(best_fitness)
best_fitness_history_by_topology_type[
topology_type] = best_fitness_history
plt.clf()
plt.boxplot(list(best_fitness_history_by_topology_type.values()),
labels=list(best_fitness_history_by_topology_type.keys()))
plt.title(function_type)
plt.savefig(os.path.join(results_dir_path,
function_type + "_box_plot.png"),
bbox_inches='tight')
def __init__(self, size, individualSize):
self._size = size
self._individualSize = individualSize
self._initialPopulation = Population(size, individualSize)
self._graph = [self._initialPopulation]
def evolvePopulation(self):
newPopulation = self._graph[-1].iterationEA()
new = Population(self._size, self._individualSize)
new.setPopulation(newPopulation)
self._graph.append(deepcopy(new))
if args.gpu == -1:
supernet = torch.nn.DataParallel(supernet)
# optimizer
supernetOptimizer = optim.SGD(supernet.parameters(),
lr=args.init_learning_rate,
momentum=args.momentum,
weight_decay=args.weight_decay)
# scheduler
scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(
supernetOptimizer,
args.warmup_epoch_first + decision_number * args.warmup_epoch_others)
# define the population for search
population = Population()
logging.info('Population 0: %d', len(population.archList))
# begin search
for geneOpIndex in range(decision_number):
# The number of train epochs of the first decision is greator than that of the others.
if geneOpIndex == 0:
tmp_warmup_epoch = args.warmup_epoch_first
else:
tmp_warmup_epoch = args.warmup_epoch_others
# If you have already trained the models after first warming up, you can reload it to avoid too much time consuming.
if args.reload_model and os.path.exists(
pretrainModelCheckPointPath) and geneOpIndex == 0:
logging.info('==> Resuming from checkpoint: %s',
os.path.abspath(pretrainModelCheckPointPath))